Journal of Cereal Science

Journal of Cereal Science 

1. 

Volume 49, Issue 1, Pages 1-162 (January 2009) 

IFC - Editorial Board 

Page IFC 

Rapid Communication 

2. 

Degradation of cereal beta-glucan by ascorbic acid induced oxygen radicals 

Pages 1-3 

Reetta Kivelä, Fred Gates, Tuula Sontag-Strohm 

Research Papers 

3. 

4. 

5. 

6. 

7. 

8. 

9. 

Variation in Granule Bound Starch Synthase I (GBSSI) loci amongst Australian wild cereal 

relatives (Poaceae) 

Pages 4-11 

F.M. Shapter, P. Eggler, L.S. Lee, R.J. Henry 

Effect of high temperature on albumin and globulin accumulation in the endosperm 

proteome of the developing wheat grain 

Pages 12-23 

William J. Hurkman, William H. Vensel, Charlene K. Tanaka, Linda Whitehand, Susan B. Altenbach 

Accumulation of mixed linkage (1 → 3) (1 → 4)-β-d-glucan during grain filling in barley: A 

vibrational spectroscopy study 

Pages 24-31 

Helene Fast Seefeldt, Andreas Blennow, Birthe Møller Jespersen, Bernd Wollenweber, Søren 

Balling Engelsen 

Mechanism of gas cell stabilization in bread making. I. The primary gluten–starch matrix 

Pages 32-40 

Baninder S. Sroan, Scott R. Bean, Finlay MacRitchie 

Mechanism of gas cell stabilization in breadmaking. II. The secondary liquid lamellae 

Pages 41-46 

Baninder S. Sroan, Finlay MacRitchie 

Expression of globulin-2, a member of the cupin superfamily of proteins with similarity to 

known food allergens, is increased under high temperature regimens during wheat grain 

development 

Pages 47-54 

Susan B. Altenbach, Charlene K. Tanaka, William J. Hurkman, William H. Vensel 

Biochemical markers: Efficient tools for the assessment of wheat grain tissue proportions in 

milling fractions 

Pages 55-64 

Youna Hemery, Valérie Lullien-Pellerin, Xavier Rouau, Joël Abecassis, Marie-Françoise Samson, 

Per Åman, Walter von Reding, Cäcilia Spoerndli, Cécile Barron

10. 

11. 

12. 

13. 

14. 

15. 

16. 

17. 

18. 

19. 

20. 

21. 

Lateral growth of a wheat dough disk under various growth conditions 

Pages 65-72 

Amy Penner, Leaelaf Hailemariam, Martin Okos, Osvaldo Campanella 

Digestibility of protein and starch from sorghum (Sorghum bicolor) is linked to biochemical 

and structural features of grain endosperm 

Pages 73-82 

Joshua H. Wong, Tsang Lau, Nick Cai, Jaswinder Singh, Jeffrey F. Pedersen, William H. Vensel, 

William J. Hurkman, Jeff D. Wilson, Peggy G. Lemaux, Bob B. Buchanan 

Fundamental study on protein changes taking place during malting of oats 

Pages 83-91 

Christina Klose, Beatus D. Schehl, Elke K. Arendt 

Effect of milling, pasta making and cooking on minerals in durum wheat 

Pages 92-97 

Francesco Cubadda, Federica Aureli, Andrea Raggi, Marina Carcea 

Starch granule size distribution of hard red winter and hard red spring wheat: Its effects on 

mixing and breadmaking quality 

Pages 98-105 

Seok-Ho Park, Jeff D. Wilson, Bradford W. Seabourn 

Total phenolics, flavonoids, antioxidant capacity in rice grain and their relations to grain 

color, size and weight 

Pages 106-111 

Yun Shen, Liang Jin, Peng Xiao, Yan Lu, Jinsong Bao 

Nucleotide polymorphisms in the waxy gene of NaN3-induced waxy rice mutants 

Pages 112-116 

Toong Long Jeng, Chang Sheng Wang, Tung Hai Tseng, Min Tze Wu, Jih Min Sung 

Amaranth (Amaranthus hypochondriacus) as an alternative crop for sustainable food 

production: Phenolic acids and flavonoids with potential impact on its nutraceutical quality 

Pages 117-121 

A.P. Barba de la Rosa, Inge S. Fomsgaard, Bente Laursen, Anne G. Mortensen, L. Olvera- 

Martínez, C. Silva-Sánchez, A. Mendoza-Herrera, J. González-Castañeda, A. De León-Rodríguez 

Relationship of milled grain percentages and flowering-related traits in rice 

Pages 122-127 

Rodante E. Tabien, Stanley Omar P.B. Samonte, Emmanuel R. Tiongco 

Texture, processing and organoleptic properties of chickpea-fortified spaghetti with insights 

to the underlying mechanisms of traditional durum pasta quality 

Pages 128-133 

Jennifer Ann Wood 

Impact of re-grinding on hydration properties and surface composition of wheat flour 

Pages 134-140 

M. Mohamad Saad, C. Gaiani, J. Scher, B. Cuq, J.J. Ehrhardt, S. Desobry 

Identification of novel haze-active beer proteins by proteome analysis 

Pages 141-147 

Takashi Iimure, Nami Nankaku, Megumi Watanabe-Sugimoto, Naohiko Hirota, Zhou Tiansu, 

Makoto Kihara, Katsuhiro Hayashi, Kazutoshi Ito, Kazuhiro Sato

22. 

23. 

Mixing properties and dough functionality of transgenic lines of a commercial wheat 

cultivar expressing the 1Ax1, 1Dx5 and 1Dy10 HMW glutenin subunit genes 

Pages 148-156 

Elena León, Santiago Marín, María J. Giménez, Fernando Piston, Marta Rodríguez-Quijano, Peter 

R. Shewry, Francisco Barro 

Wheat glutenin proteins assemble into a nanostructure with unusual structural features 

Pages 157-162 

Sarah H. Mackintosh, Susie J. Meade, Jackie P. Healy, Kevin H. Sutton, Nigel G. Larsen, Adam 

M. Squires, Juliet A. Gerrard


Aims and Scope 

The Journal of Cereal Science was established in 1983 to provide an international forum for the publication of original research papers of high standing covering all 

aspects of cereal science related to the functional and nutritional quality of cereal grains and their products. 

The journal also publishes concise and critical review articles appraising the status and future directions of specific areas of cereal science and short rapid communications 

that present news of important advances in research. The journal aims at topicality and at providing comprehensive coverage of progress in the field. 

Research areas include: 

Composition and analysis of cereal grains in relation to quality in end use 

Morphology, biochemistry, and biophysics of cereal grains relevant to functional and nutritional characteristics 

Structure and physicochemical properties of functionally and nutritionally important components of cereal grains such as polysaccharides, proteins, oils, 

enzymes, vitamins, and minerals 

Storage of cereal grains and derivatives and effects on nutritional and functional quality 

Genetics, agronomy, and pathology of cereal crops if there is a substantive relationship to end-use properties of cereal grains 

Functional and nutritional aspects of cereal-based foods and beverages, whether baked, fermented, or extruded 

Industrial products (e.g. starch derivatives, syrups, protein concentrates, and isolates) from cereal grains, and their technology 

J. Dexter 

Grain Research Laboratory 

Canadian Grain Commission 

Winnipeg, Canada 

Editorial Board 

N.-G. Asp 

University of Lund, Sweden 

D. E. Briggs 

University of Birmingham, UK 

G. B. Fincher 

Waite Agricultural Research 

Institute, 

University of Adelaide, 

Australia 

P. Colonna 

INRA, Laboratoire de Biochimie et Technologies 

des Glucides Nantes, France 

K. Denyer 

John Innes Centre, Norwich, UK 

P. J. Frazier 

Dalgety Food Technology Centre, Cambridge, UK 

R. A. Graybosch 

University of Nebraska, Lincoln, Nebraska, USA 

Zhonghu He 

International Maize and Wheat Improvement 

Center, Chinese Academy of Agricultural Sciences, 

Beijing, China 

Founding Editors 

T. Galliard and J. D. Schofield 

Editor-in-Chief 

F. MacRitchie 

Grain Science and Industry Department 

Kansas State University 

Manhattan, Kansas, USA 

R. J. Hamer 

Wageningen Centre for 

Food Sciences 

Wageningen, 

The Netherlands 

G. E. Inglett 

USDA, National Center for Agricultural 

Utilization Research, Peoria, 

Illinois, USA 

B. O. Juliano 

Philippines Rice Research Institute, Laguna, 

Philippines 

J. Kokini 

Cook College, Rutgers University, 

New Brunswick, 

New Jersey, USA 

P. Linko 

Helsinki University of Technology, 

Espoo, Finland 

A. W. MacGregor 

Livingston, Scotland, UK 

Editors 

Reviews Editor 

P. R. Shewry 

Rothamsted Research 

Harpenden, UK 

D. Lafiandra 

Dipartimento di Agrobiologia 

e Agrochimica 

University of Tuscia 

Viterbo, Italy 

J. R. N. Taylor 

University of Pretoria, 

South Africa 

Y. Popineau 

INRA, Unite de Recherche sur les 

Proteines Vegetales et leurs Interactions, 

Nantes, France 

K. Poutanen 

VTT Biotechnology, 

Espoo, Finland 

S. O. Serna Saldivar 

ITESM, Monterrey, Mexico 

B. A. Stone 

Department of Biochemistry 

La Trobe University 

Bundoora, Victoria, Australia 

B. Svensson 

BioCentrum-DTU, 

The Technical University of 

Denmark, Kgs. 

Lyngby, Denmark 

Author enquiries 

For enquiries relating to the submission of articles (including electronic submission where available) please visit this journal’s homepage at http://www.elsevier.com/locate/jcs. 

You can track accepted articles at http://www.elsevier.com/trackarticle and set up e-mail alerts to inform you of when an article’s status has changed. Also accessible from 

here is information on copyright, frequently asked questions and more. 

Contact details for questions arising after acceptance of an article, especially those relating to proofs, will be provided by the publisher. 

For a full and complete Guide for Authors, please go to: http://www.elsevier.com/locate/issn/0733-5210/authorinstructions

Rapid Communication 

Degradation of cereal beta-glucan by ascorbic acid induced oxygen radicals 

Reetta Kivelä *, Fred Gates 1 , Tuula Sontag-Strohm 

Department of Food Technology, University of Helsinki, Agnes Sjöberg Street 2, P.O. Box 66, FIN-00014 Helsinki, Finland 

article info 

Article history: 

Received 24 April 2008 

Received in revised form 15 August 2008 

Accepted 5 September 2008 

Keywords: 

Beta-glucan 

Degradation 

Oxygen radicals 

Ascorbic acid 

abstract 

We report degradation of cereal beta-glucan in aqueous solutions 

as a result of a treatment with ascorbic acid and its oxidation 

product dehydroascorbic acid. It was found that (1 / 3), (1 / 4)beta-D-glucan 

was depolymerised when treated with ascorbic acid 

or dehydroascorbic acid and FeSO4 in the absence of enzymatic 

activity or conditions that could cause acid hydrolysis. The degradation 

was inhibited by both glucose addition and by displacing 

oxygen with nitrogen. Non-enzymatic, ascorbate induced degradation 

of cell-wall polysaccharides is established in vitro and suggested 

to occur in plant cells during its elongation phase and 

ripening (Fry, 1998; Fry et al., 2002; Schopfer, 2001). However, this 

mechanism appears to be neglected in food technology in spite of 

its potentially crucial role in reducing the physiological efficacy, and 

hence the validity of health claims, of high beta-glucan foods as 

well as influencing properties such as rheology and physical 

stability. 

Macromolecules are typically susceptible to degradation 

induced by reactive oxygen species (ROS). Particularly OH-radicals, 

one specific species of ROS, have been established to scission 

polysaccharides (Fry, 1998). One way to produce OH-radicals nonenzymatically 

is a Fenton type reaction (Reaction (1)). The key 

feature of Fenton chemistry is the catalytic nature of reduced 

Abbreviations: AH2, ascorbic acid; A, dehydroascorbic acid; BBG, beta-glucan 

isolate from barley; HPSEC, high performance size exclusion chromatography; Mw, 

weight average molecular mass; OBG, beta-glucan isolate from oat; ROS, reactive 

oxygen species, i.e. all kinds of chemically reactive molecules derived from ground 

state dioxygen ( 3 O2). 

* Corresponding author. Tel.: þ358 9 19158540; fax: þ358 9 19158460. 

E-mail address: reetta.kivela@helsinki.fi (R. Kivelä). 

1 

Present address: Department of Physics, University of Helsinki, P.O. Box 64, 

FIN-00014 Helsinki, Finland. 

0733-5210/$ – see front matter Ó 2008 Elsevier Ltd. All rights reserved. 

doi:10.1016/j.jcs.2008.09.003 

Journal of Cereal Science 49 (2009) 1–3 

Contents lists available at ScienceDirect 


journal homepage: www.elsevier.com/locate/jcs 

Degradation of cereal beta-glucan is usually attributed to enzymes or acid hydrolysis. However, there is 

evidence that polysaccharides are also susceptible to OH-radical induced depolymerisation, and that 

these radicals can be produced in cereal food systems. The role of Fenton type oxidation was demonstrated 

in pure beta-glucan solution (0.6%). An addition of ascorbic acid (10 mM) or its oxidation product, 

dehydroascorbic acid, in the presence of iron sulphate resulted in a significant decrease of the solution 

viscosity and molecular degradation of beta-glucan. The viscosity decrease was inhibited by introducing 

a OH-scavenger (glucose) in the solution or limiting oxygen level in the sample solution. This demonstrates 

the role of OH-radicals in beta-glucan scission and suggests oxidative cleavage to be a potential 

threat for the stability of beta-glucan in certain fibre enriched products. 

Ó 2008 Elsevier Ltd. All rights reserved. 

metals (Fe 2þ ,Cu þ ,Zn þ ), which can produce aggressive OH-radicals 

from the modestly reactive H2O2 (Wardman and Candeias, 1996). A 

reducing agent is needed to reduce the catalysing metal to the 

active state and the dissolved oxygen to H2O2, thus it initiates the 

Fenton type reaction. All these reagents (dissolved oxygen, transition 

metal and compounds with reduction potential) are typically 

present in the aqueous phases of cereal foods, which also contains 

the soluble fibres (Slavin et al., 1999). In addition, beta-glucan is 

known to form complexes with metals (Platt and Clydesdale, 1984), 

which makes it a specific target for OH-radical induced degradation, 

as the short living radicals are produced in immediate vicinity 

of beta-glucan molecules (Chevion, 1988). 

Cu þ =Fe 2þ þ H 2O 2/ $ OH þ OH þ Cu 2þ =Fe 3þ 

Ascorbic acid (AH2) is an excellent reducing agent. It readily 

undergoes a two-step oxidation process forming dehydroascorbic 

acid (A) and ascorbate radical ( AH) as an intermediate. The 

ascorbate radical is a relatively stable free radical that has a capability 

of inhibiting oxidation chain reactions and thus can act as 

a secondary antioxidant. However, ascorbic acid is known to have 

a dual nature: in certain conditions it behaves as a pro-oxidant 

instead of antioxidant initiating the Fenton reaction due to the 

reducing capacity (Reactions (2a) and (2b)). The dual effect is 

considered to be concentration dependent, so that the rate of 

antioxidant chain reactions is lowered and the pro-oxidant nature 

dominates at lower concentrations (Buettner and Jurkiewicz, 1996; 

Wardman and Candeias, 1996). Ascorbic acid is a common food 

constituent and commonly added as an antioxidant; it is also used 

as a flour improver which makes Fenton chemistry even more 

interesting in cereal foods. 

(1)

2 

AH 2 þ 2Cu 2þ =Fe 3þ /A þ 2H þ þ 2Cu þ =Fe 2þ 

AH 2 þ O 2/A þ H 2O 2 

(2a) 

(2b) 

Degradation of cereal beta-glucan by the chemical radical 

reactions may explain some of the controversial results obtained in 

various studies. It is not uncommon that a reduction in viscosity of 

beta-glucan (i.e. molecular size reduction) or molecular degradation 

is observed in food systems such as bakery products and 

beverages (Åman et al., 2004; Beer et al., 1997). These changes have 

generally been attributed to enzymatic or acid hydrolysis or to high 

shear forces. However, in some cases enzymes seem to have been 

inactivated in heating processes or the conditions are not suitable 

for acid hydrolysis to occur, which requires a very low pH (1–2) and 

simultaneous heating (Johansson et al., 2006). In addition, there is 

no clear evidence to link the acid hydrolysis of beta-glucan to the 

organic acids present in foods. 

Technologically the degradation of beta-glucan is serious. In 

aqueous systems the breakdown of the polymer easily leads to 

product structural breakdown and instability. The degradation of 

beta-glucan is also nutritionally crucial, since the health benefits of 

beta-glucan are closely related to its rheology. In particular, it is 

well established that the ability of beta-glucan to attenuate postprandial 

blood glucose and insulin levels is related to the viscosity 

and molecular weight of beta-glucan (Wood et al., 2000). The 

structure of food matrices, in addition to the rheology, is proposed 

to affect cholesterol lowering of beta-glucan and liquid beta-glucan 

rich foods have been established to be especially effective (Kerckhoffs 

et al., 2003; Naumann et al., 2006). There is thus an 

increasing need to properly understand the threats for instability of 

native beta-glucan in aqueous environments. 

In the present study effects of ascorbic acid (10 mM) and 

dehydroascorbic acid (10 mM) were examined in pure 0.6% betaglucan 

solutions when added ionic iron (0.01 mM FeSO4$H2O) was 

present. Also the roles of glucose (1 M, 18 w/w-%) and oxygen were 

studied. The beta-glucan solutions were prepared from high 

viscosity barley beta-glucan (BBG, Megazyme, purity >97%) or 

high viscosity oat beta-glucan (OBG, Megazyme, purity >97%) by 

wetting dry beta-glucan with ethanol and dispersing the blend in 

water. The dispersion was first boiled under stirring for 10 min and 

then continued for 3 h at 80 C before adjusting the volume in 

a volumetric flask. Viscosities of the stock solutions were 

controlled up to three weeks and were observed to maintain 

constant viscosity. Sodium azide (0.02 w/w-%) was added to 

prevent microbial enzyme activity. At this concentration, sodium 

azide was not shown to affect the studied reactions. All the 

reagents were added as solid compounds to the stock solution. The 

pH was adjusted to 4.8 0.1 to represent acidic conditions of 

fermented and beverage-like foods and to enable ascorbic acid to 

exist in dissociated form (pKaAH2 ¼ 4.2) and the sample volumes 

were adjusted after pH adjustment. Viscosity measurements and 

oscillation tests were performed using a rheometer (Thermo- 

Haake RheoStress 600, Thermo Electron GmbH, Dreieich, Germany). 

Flow curves were obtained over a shear rate range of 0.3– 

300–0.3 s 1 using a double gap geometry (DG41). The mean 

apparent viscosities of three independent measurements 

(mean SD) at a shear rate of 10 s 1 as a function of time are 

shown. Dynamic rheological behaviour was characterised by 

a cone and plate geometry (35 mm/2 ) using a frequency sweep of 

0.01–10 Hz and a stress of 0.02 Pa. All the rheological experiments 

were performed at þ10 0.1 C. Prior to measurement the 

samples were stored at þ6 0.1 C. The molecular weight of betaglucan 

was analysed by HPSEC with calcofluor-fluoresence and 

dual angle light scattering detection in 0.1 N sodium hydroxide 

solution (Suortti, 1993). 

R. Kivelä et al. / Journal of Cereal Science 49 (2009) 1–3 

The rheological studies showed that 10 mM AH2 added together 

with 0.01 mM FeSO4$H2O caused an approximately 50% drop in 

viscosity of the barley beta-glucan (BBG) solution in 1 h, while the 

viscosity of BBG treated with FeSO4 alone remained unchanged 

(Fig. 1). The results of BBG solution alone are not shown, because 

the ionic iron treatment did not affect it. Consistent results were 

obtained with the oat beta-glucan solution (OBG, results not 

shown). The viscosity behaviour of these solutions changed from 

shear thinning to Newtonian in 2 h suggesting that the critical 

concentration, at which the entanglements between molecules or 

aggregates are lost, had been reached as a result of molecular size 

reduction (Fig. 2). HPSEC-studies showed that this reduction in 

molecule size was depolymerisation instead of other changes in 

physicochemical properties, since the weight average molecular 

mass (Mw) of beta-glucan was reduced from 520 ( 5) to 35 

( 2) 10 3 g/mol as a result of the ascorbic acid and iron treatment. 

The effect of the third reagent, oxygen, was demonstrated by 

limiting the amount of dissolved oxygen with nitrogen supplement 

during sample preparation and testing. After 4 h of N2-treatment, 

air was mixed into the nitrogen treated ascorbic acid-BBG sample in 

the measuring cap of the rheometer and this caused a drop in 

viscosity (Fig. 1). Thus all the reagents of Fenton chemistry, i.e. 

transition metal, dissolved oxygen and a reducing agent (Reactions 

(1)–(2b)), were shown to be needed for the depolymerisation of 

beta-glucan in aqueous solution. 

Dehydroascorbic acid (10 mM) caused a 30% viscosity reduction 

when added with ionic iron to the beta-glucan solution (Fig. 1). The 

total decrease during the measuring time was appr. 50%, while 

ascorbic acid decreased the viscosity over 90%. The lower viscosity 

reduction, i.e. the rate of oxidative cleavage compared to the action 

of ascorbic acid agrees with the results of Fry (1998) with xyloglucan. 

Dehydroascorbic acid is likely to occur in food systems, 

since ascorbic acid undergoes a reversible oxidation reaction 

induced by light, heat and metal first forming semidehydroascorbic 

Apparent viscosity at10 1/s (mPas) 

180 

160 

140 

120 

100 

80 

60 

40 

20 

0 

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 

Time / h 

+1M glucose 

Control 

+10 mM AH2+ 

1M glucose 

+10 mM A 

+10 mM AH2, 

limited oxygen 

+10 mM AH2 

Fig. 1. The effect of ascorbic acid (AH2) and dehydroascorbic acid (A) with ionic iron on 

the viscosity of beta-glucan (control, BBG, 0.6%) and the inhibition effects of oxygen 

limitation and glucose addition (1 M). The arrow shows the point of air introduction 

into the nitrogen treated sample. All the samples contained 0.01 mM FeSO4$7H2O.

Viscosity Pas 

1 

0,1 

Control 

0.5 h 

1 h 

2 h 

0,01 

0,001 

4 h 

6 h 

8 h 

10 h 

18 h 

0,1 1 10 

Shear rate 1/s 

100 1000 

Fig. 2. Flow curves of ascorbic acid (10 mM) treated beta-glucan solution (0.6%, BBG) 

from 0 h to 18 h and the control at 18 h. Both samples contained 0.01 mM FeSO4$7H2O. 

(AH ) acid and further dehydroascorbic acid (A). It has been 

reported that, like ascorbic acid, dehydroascorbic acid can also keep 

metals such as copper in their reduced form, but it cannot reduce 

oxygen to H2O2 (Fry, 1998). Yet as endogenic H2O2 usually occurs in 

water solutions and especially in natural solutions such as food 

systems, dehydroascorbic acid may also participate in the oxidative 

cleavage of beta-glucan. 

Glucose efficiently inhibited the ascorbic acid induced viscosity 

reduction of BBG solution (Fig. 1). Glucose was introduced to the 

study as a competitive OH-radical scavenger (Buxton et al., 1988). In 

the solutions with high concentration of glucose (1 M, appr. 20 w/ 

w-%), the majority of indiscriminant OH-radicals will react with the 

glucose rather than with the high molecular weight polysaccharide 

(beta-glucan, 0.6 w/w-%). After the reaction with OH, glucose itself 

becomes a relatively stable free radical. In addition, the glucose 

radical may regenerate itself with a hydrogen atom from the 

solution, thus enhancing its stability and efficiency as a scavenger 

(Morelli et al., 2003). The viscosity of glucose–BBG solution was 17% 

higher than the viscosity of BBG solution alone, which is in accordance 

with that reported before (Autio et al., 1987), and it decreased 

20% due to the ascorbic acid treatment (Fig. 1). However, the 

viscosity of ascorbic acid–glucose–BBG solution remained 

unchanged as a function of time indicating that the proposed OHinduced 

degradation of beta-glucan was inhibited by glucose 

addition. 

The flow curves of ascorbic acid treated BBG solution showed 

hysteresis and noise after 5 h (Fig. 2). This may both indicate novel 

networks and particle formation. Gel formation was observed as 

increased turbidity in the relatively dilute ascorbic acid–BBG 

solution (0.6%) after four days storage, while the control BBG 

solution stayed clear. The gelation was shown by dynamic rheology, 

the elastic modulus (G 0 ) being nearly frequency-independent (ca. 

10 Pa at 1 Hz) and the loss modulus (G 00 ) being low (appr. 1 Pa at 

1 Hz) at frequencies 1–10 Hz. It is known that gelation rate as well 

as gel strength increase with decreasing molecular weight so that 

molecules with an Mw above 250 000 g/mol do not practically form 

gels (Lazaridou et al., 2003; Tosh et al., 2004). Such oxidative 

cleavage may bring new aspects for the phenomenon of gelation 

during food processing, since it can occur in enzyme and acid 

hydrolysis free conditions and the cleavage may result in different 

compounds as found in hydrolysis (Fry, 1998; Fry et al., 2002). 

In conclusion, it was found that ROS represents a threat to the 

stability of cereal beta-glucan in an aqueous environment, and that 

Fenton chemistry results in depolymerisation of beta-glucan. All 

R. Kivelä et al. / Journal of Cereal Science 49 (2009) 1–3 3 

the reagents, i.e. a reducing agent (ascorbic acid), transition metal 

(iron) and oxygen, were shown to have a role in the depolymerisation 

of beta-glucan. The viscosity loss was inhibited by an oxygen 

radical scavenger (glucose), which supports the role of OH-radicals 

in the non-enzymatic cleavage of cereal beta-glucan. Common food 

additives as well as the minor compounds of the soluble fibre 

phases may be harmful for the stability of aqueous beta-glucan 

enriched foods. In addition, their role in the research of health 

benefits of beta-glucan in different processed foods needs proper 

understanding. 

Acknowledgements 

The authors thank Dr. Tapani Suortti (VTT, Finland) for performing 

the molecular weight measurement and data analysis and 

Professor Hannu Salovaara and Dr. Laura Nyström for their important 

support in the process. 

References 

Autio, K., Myllymäki, O., Mälkki, Y., 1987. Flow properties of solutions of oat bglucans. 

Journal of Food Science 52, 1364–1366. 

Beer, M.U., Wood, P.J., Weisz, J., Fillion, N., 1997. Effect of cooking and storage on the 

amount and molecular weight of (1 / 3), (1 / 4)-b-D-glucan extracted from 

oat products by an in vitro digestion system. Cereal Chemistry 74, 705–709. 

Buettner, G.R., Jurkiewicz, B.A., 1996. Catalytic metals, ascorbate and free radicals: 

combinations to avoid. Radiation Research 145, 532–541. 

Buxton, G.V., Greenstock, C.L., Helman, W.P., Ross, A.B., 1988. Critical review of rate 

constants for reactions of hydrated electrons, hydrogen atoms and hydroxyl 

radicals ( OH/ O ) in aqueous solution. Journal of Physical and Chemical 

Reference Data 17, 513–769. 

Chevion, M., 1988. A site-specific mechanism for free radical induced biological 

damage: the essential role of redox-active transition metals. Free Radical 

Biology 5, 27–37. 

Fry, S.C., 1998. Oxidative scission of plant cell wall polysaccharides by ascorbateinduced 

hydroxyl radicals. Biochemical Journal 332, 507–515. 

Fry, S.C., Miller, J.G., Dumville, J.C., 2002. A proposed role for copper ions in cell wall 

loosening. Plant and Soil 247, 57–67. 

Johansson, L., Virkki, L., Anttila, H., Esselström, H., Tuomainen, P., Sontag-Strohm, T., 

2006. Hydrolysis of b-glucan. Food Chemistry 97, 71–79. 

Kerckhoffs, D., Hornstra, G., Mensink, R.P., 2003. Cholesterol-lowering effect of betaglucan 

from oat bran in mildly hypercholesterolemic subjects may decrease 

when beta-glucan is incorporated into bread and cookies. The American Journal 

of Clinical Nutrition 78, 221–227. 

Lazaridou, A., Biliaderis, C.G., Izydorczyk, M.S., 2003. Molecular size effects on 

rheological properties of oat beta-glucans in solution and gels. Food Hydrocolloids 

17, 693–712. 

Morelli, R., Russo-Volpe, S., Bruno, N., Lo Scalzo, R., 2003. Fenton-dependent 

damage to carbohydrates: free radical scavenging activity of some simple 

sugars. Journal of Agricultural and Food Chemistry 51, 7418–7425. 

Naumann, E., van Rees, A.B., Onning, G., Oste, R., Wydra, M., Mensink, R.P., 2006. 

Beta-glucan incorporated into a fruit drink effectively lowers serum LDLcholesterol 

concentrations. The American Journal of Clinical Nutrition 83, 601– 

605. 

Platt, S.R., Clydesdale, F.M., 1984. Binding of iron by cellulose, lignin, sodium phytate 

and beta-glucan, alone and in combination, under simulated gastrointestinal 

pH conditions. Journal of Food Science 49, 531–535. 

Schopfer, P., 2001. Hydroxyl radical-induced cell-wall loosening in vitro and in vivo: 

implications for the control of elongation growth. The Plant Journal 28, 679– 

688. 

Slavin, J.L., Martini, M.C., Jacobs, D.R., Marquart, L., 1999. Plausible mechanisms for 

the protectiveness of whole grains. The American Journal of Clinical Nutrition 

70, 459–463. 

Suortti, T., 1993. Size-exclusion chromatographic determination of beta-glucan with 

postcolumn reaction detection. Journal of Chromatography 19, 105–110. 

Tosh, S.M., Wood, P.J., Wang, Q., Weisz, J., 2004. Structural characteristics and 

rheological properties of partially hydrolyzed oat b-glucan: the effects of 

molecular weight and hydrolysis method. Carbohydrate Polymers 55, 425–436. 

Wardman, P., Candeias, L.P., 1996. Fenton chemistry: an introduction. Radiation 

Research 145, 523–531. 

Wood, P.J., Beer, M.U., Butler, G., 2000. Evaluation of role of concentration and 

molecular weight of oat beta-glucan in determining effect of viscosity on 

plasma glucose and insulin following an oral glucose load. The British Journal of 

Nutrition 84, 19–23. 

Åman, P., Rimsten, L., Andersson, R., 2004. Molecular weight distribution of bglucan 

in oat-based foods. Cereal Chemistry 81, 356–360.

Variation in Granule Bound Starch Synthase I (GBSSI) loci amongst 

Australian wild cereal relatives (Poaceae) 

F.M. Shapter *, P. Eggler, L.S. Lee 1 , R.J. Henry 

Grain Foods CRC, Centre for Plant Conservation Genetics, Southern Cross University, PO Box 157, Lismore, NSW 2480, Australia 

article info 


Received 30 July 2007 

Received in revised form 12 November 2007 

Accepted 20 June 2008 

Keywords: 

Australian native grasses 

Starch 

Crop wild relatives 

Cereal 

Phylogenetic 

1. Introduction 

abstract 

Granule Bound Starch Synthase I (GBSSI), encoded by the Waxy 

gene, is responsible for the accumulation of amylose during the 

development of starch granules in cereal endosperm. GBSSI has 

been reported with changing enzyme nomenclature over the last 

two decades. The most recent, EC 2.4.1.242, clearly delineates this 

enzyme from previously recorded enzymes EC 2.4.1.11 (Taira et al., 

1995) and EC 2.4.1.21 (Baldwin, 2001; Demeke et al., 1999), as being 

able to use both UDP-glucose and ADP-glucose as a substrate for 

starch synthesis. Although GBSSI and II share the same enzyme 

nomenclature and approximately 69% sequence identity, suggesting 

that they are homologous, the genes encoding them are 

situated at independent loci. The gene encoding GBSSI is predominantly 

expressed in endosperm and pollen whereas GBSSII is 

expressed in the leaf, culm and pericarp (Vrinten and Nakamura, 

2000). Equally well reported is that cereal species with a null allele 

at all Waxy loci will produce an amylose free endosperm. The 

* Corresponding author. Tel.: þ61 2 6620 3466; fax: þ61 2 6622 2080. 

E-mail address: frances.shapter@scu.edu.au (F.M. Shapter). 

1 Present address: Centre for Tropical Crops and Biocommodities, Queensland 

University of Technology, GPO Box 2434, Brisbane, Queensland 4001, Australia. 


doi:10.1016/j.jcs.2008.06.013 





A complex cascade of enzymes is responsible for the development of starch granules in grain endosperm. 

Granule Bound Starch Synthase I (GBSSI), encoded by the Waxy gene, is a key enzyme of starch synthesis 

and determines the accumulation of amylose in the starch granules. The complete genomic GBSSI 

sequence was ascertained for eight Australian cereal wild relatives (CWR) to determine diversity within 

the gene. A phylogeny derived from the coding sequence of the entire Waxy gene was compared to 

established phylogenetic relationships. Starch granule morphology observed in conjunction with this 

phylogeny suggests that small polygonal starch granules arranged as compound granules are the 

ancestral state, evolving subsequently to bimodal starch granules and to larger simple granules. Genomic 

sequence length varied within the species from 2800 to 3572 bp. Most variation occurred within the 

intron sequences, the largest insertion showing strong homology to a retrotransposon. One wild species 

was determined to have a deletion in the 3 0 -end of exon 1 resulting in a putatively non-functional allele. 

Alignment of the amino acid sequence showed strong homology throughout the central fragments of the 

gene but broad variation in the transit peptides. All putative functional alleles maintained the reported 

active sites for glycogen synthesis, though with variations in other highly conserved areas of the gene. 

These variations within the wild relatives of cultivated cereals may provide novel sources of genetic 

diversity for future cereal improvement programs. 


amylose status of cereal starches affects their functional properties 

and therefore their end uses (Domon et al., 2002; Gaines et al., 

2000; Kim et al., 2003) and a simple, rapid iodine test for the 

presence/absence of amylose in endosperm has been recently 

optimised (Pedersen et al., 2004). 

Within cultivated cereals Waxy mutants are common from 

induced (Nakamura et al., 1995) and naturally occurring origins 

(Domon et al., 2002), with mutations remaining conserved in the 

genome over generations. The concentration of Waxy mutants 

occurring in cereal wild relatives (CWR) is as yet undetermined. Of 

the waxy and low-amylose cereal species sequenced to date, it 

appears that there is no specific polymorphism across all the 

species which results in the waxy phenotype. Rather, any disruption 

to the production or regulation of this enzyme is critical in 

cereal species (Mangalika et al., 2003; Rahman et al., 2000; Shure 

et al., 1983; Yan et al., 2000). Within any one cereal species, 

multiple alleles of the GBSSI coding gene are common and are 

denoted by a subscript or chromosomal reference which is species 

specific (Mangalika et al., 2003; Pedersen et al., 2007). In the case of 

polyploid species such as wheat, the number of functional GBSSI 

coding alleles is correlated with the concentration of amylose in the 

grain (Mangalika et al., 2003). 

The phylogenetic utility of the GBSSI encoding gene has been 

repeatedly reported (Evans and Campbell, 2002; Mason-Gamer

et al., 1998; Small et al., 2004). Phylogenetic relationships between 

species provide an understanding of the breadth of their genetic 

diversity and have been utilised in crop-breeding programs to 

determine relationships between cultivated species and their wild 

cereal relatives (Williams et al., 2006). Initial phylogenetic analyses 

across this subset of Poaceae utilising a small central fragment of 

the gene was insufficient to resolve the basal bifurcations and 

interspecies relationships (data not shown). As a result, more 

sequence was captured to enhance phylogenetic analysis. In order 

to ensure that phylogenetic predictions are resolving species relationships 

and not gene evolution, analysing more than one gene 

concurrently is prudent (Small et al., 2004). However, functional 

genes evolve under selective pressure and therefore their evolutionary 

histories can be correlated with their functional evolution 

(Ochieng et al., 2007). In this instance establishing the diversity and 

evolution of GBSSI is the primary objective. 

Following crop improvements made by plant breeders in the 

last century, cereal-breeding programs are increasingly targeting 

cereal wild relatives as a source of novel germplasm (Hajjar and 

Hodgkin, 2007). This is occurring in a time of increasing human 

populations and climatic instability. It has been proposed that 

meeting global food requirements by 2025 will require a 25% 

increase in grain production (Khush, 2001). The use of CWR in plant 

breeding programs will become increasingly important because of 

their environmental adaptations (Ashikari and Matsuoka, 2006). 

Previously, hybridisation of CWR from the primary gene pool of the 

respective cultivated cereals has been used to improve agronomic 

traits such as growth habit, and pest and disease resistance. Recent 

advances in molecular and plant breeding techniques have shifted 

the focus from primary to secondary and/or tertiary gene pool 

species as sources of novel allelic variation for genes of importance 

(Hajjar and Hodgkin, 2007; Rao et al., 2003). Characterisation of the 

full coding region of GBSSI in selected native species may identify 

variants of the GBSSI coding gene. In this study GBSSI gene 

sequence is reported, and used as an indicator of the breadth of 

genetic diversity within and between Australian CWR and cereals. 

Furthermore, alignment of the amino acid sequences derived from 

the genomic data generated identifies novel putative GBSSI amino 

acid sequences in a subset of the Australian CWR. 

2. Materials and methods 

2.1. Plant material and DNA extraction 

Representative species of the four major clades of Poaceae found 

in Australia were selected for their close relationship to a cereal 

species, large seed size, broad geographic distribution and/or, in the 

case of Microlaena stipoides and Astrebla lappacea, their proven 

potential to be cultivated. Seed was germinated and leaf material 

harvested for DNA extractions. Species identity was confirmed by 

a herbarium and voucher details are listed in Table 1. Field collection 

was undertaken in Northern NSW for A. lappacea and M. 

F.M. Shapter et al. / Journal of Cereal Science 49 (2009) 4–11 5 

stipoides. Elymus scaber and Austrostipa aristiglumis were sourced 

from Native Seeds Pty Ltd Australia (http://www.nativeseeds.com. 

au). Native Oryza and Sorghum seeds were sourced from the 

Australian Tropical Crops and Forages Collection, Queensland 

Department of Primary Industries and Fisheries, Biloela, Australia 

(www.dpi.gov.au/auspgris/). DNA was extracted using a Qiagen 

DNeasy Plant Maxi Kit to ensure sufficient high-quality/highconcentration 

DNA from a single extraction. DNA quality was 

confirmed by gel electrophoresis and quantified using a NanoDrop Ò 

Technologies ND-1000 spectrophotometer. 

2.2. Presence of amylose 

The presence of amylose was determined by Pederson’s iodine 

staining technique (Pedersen et al., 2004). Controls for waxy 

(Sorghum bicolor cv. Batsuto Red) and non-waxy (S. bicolor cv. 

LR2409) endosperm (Pedersen et al., 2007) were used. 

2.3. Sequence alignment 

GBSSI coding DNA sequence of Oryza sativa (X65183), Triticum 

aestivum (AB019623), Hordeum vulgare (AB089162), Zea mays 

(X03935), Setaria italica (AB089141) and S. bicolor (AF488412.1) was 

retrieved from the NCBI Genbank Database (http://www.ncbi.nlm. 

nih.gov/). Sequence alignments were performed using ClustalW 

(EMBL – European Bioinformatics Institute; http://www.ebi.ac.uk/ 

clustalw/). 

2.4. PCR amplification and DNA sequencing 

Universal primer pair C and E from McIntosh et al. (2005) was 

used in accordance with the described method, to amplify a small 

central fragment (CE) of the GBSSI gene in the wild grasses. Small 

fragment PCR products were purified using a Qiagen Gel Extraction 

Kit prior to DNA sequencing. Full gene amplification was obtained 

by PCR using AccuprimeÔ Hi-Fi (Invitrogen) according to the 

manufacturer’s guidelines. To increase specificity, DMSO and 50% 

glycerol were added to the PCR reaction at 3 and 6%, respectively, 

for some of the species (Table 2). 

Heteroduplexes can be induced by PCR of multi-copy loci, followed 

by mismatch repair during cloning (Jansen and Ledley, 1990; 

Jensen and Straus, 1993; Longeri et al., 2002). To ensure sequence 

integrity, reconditioning PCRs were performed prior to cloning of 

the full gene fragments (Thompson et al., 2002). Reconditioning 

PCRs were spiked with 10% volume of the primary PCR product as 

template; otherwise PCR conditions were identical to that of the 

original PCR. Sequencing reactions utilised ABI PRISM Ò Big DyeÔ 

Terminator (BDT) V3.1 chemistry at one-eighth concentration 

reactions and the resulting amplicons analysed by capillary 

electrophoresis on an ABI 3730 Genetic Analyzer by Southern Cross 

Plant Genomics (SCPG, Lismore, Australia). Sequence data were 

Table 1 

Classification, collection record, herbarium voucher and Genbank accession number of eight native grass species included in this study 

Species Tribe Collection record a 

Herbarium voucher Genbank accession 

Astrebla lappacea Cynodonteae AC04-1003495 BRIAQ 751244 EF600041 

Austrostipa aristiglumis Stipeae AC04-1003487 BRIAQ 751222 EF600042 

Elymus scaber Triticeae AC04-1003502 BRIAQ 751227 EF600043 

Microlaena stipoides Erharteae AC04-1003504 BRIAQ 751226 EF600044 

Oryza australiensis Oryzeae AusTRCF310676 BRIAQ723945 EF600036 

Oryza rufipogon Oryzeae AusTRCF 309313 BRIAQ723944 EF600037 

Sorghum leiocladum Andropogoneae AusTRCF 300187 DNAD0155695 EF600038 

Sorghum nitidum Andropogoneae AusTRCF 302543 CANB479881 50 EF600039; 30 EF600040 

a 

Collection numbers prefixed with AC04 can be located through the Australian Plant DNA Bank (www.dnabank.com.au) and those prefixed with AusTRCF through the 

Australian Plant Genetic Resources Information Service (www.dpi.qld.gov.au/auspgris/).

6 

Table 2 

Full length gene primers and PCR conditions for A. aristiglumis (Aa), A. lappacea (Al), E. scaber (Es), S. leiocladum (Le), M. stipoides (Ms), O. australiensis (Oa) and O. rufipogon (Or) 

aligned and edited using SequencherÔ V4.5 (Gene Codes 

Corporation, Ann Arbor, USA). 

2.5. Genome walking 

Species-specific primers were designed using Primer 

PremierÔ V5.0 (Premier Biosoft International, Palo Alto, USA). To 

amplify the gene from the central CE fragment out to the 5 0 - and 

3 0 -UTR, respectively, the BD GenomeWalkerÔ Universal Kit (BD 

Biosciences, San Jose, USA) was used, as per the manufacturer’s 

instructions, except that all PCR amplification mixes were 

prepared at half volume. A mixture of 3% DMSO and 6% glycerol 

was added to the PCR mix and five cycles added to the standard 

PCR program, as per the manufacturer’s trouble-shooting guide. 

2.6. Cloning and DNA sequencing 

GenomeWalkerÔ and full gene PCR products were run on a 1% 

low-melt agarose gel and the individual PCR product bands of 

interest excised. Immediately prior to ligation into the vector, the 

gel slice was melted in a thermocycler at 72 C for 1 min and then 

held at 37 C. Three microliter of molten gel was included in the 

standard ligation reaction mix of the pGEM Ò -T Easy Vector II 

System (Promega Corporation, Madison, USA) following the manufacturer’s 

recommended protocol. Clones were screened for size 

by PCR using Roche Taq (Roche Products Pty Limited, Australia) in 

the manufacturer’s recommended reaction mix. Target clones were 

isolated, then amplified using the TempliphiÔ System (GE 

Healthcare) as per the manufacturer’s instructions. Standard BDT 

V3.1 one-eighth sequencing reactions used 5 ml of diluted Templiphi 

as template. 

2.7. Exon/intron boundaries 

The putative location of each exon/intron boundary was determined 

by alignment with cDNA sequence from Genbank accessions 

of the CWR’s most closely related available cDNA. The start of the 

introns was then assigned as being the nearest GT to that position 

and ending on the most distal AG prior to the following exon, 

whilst maintaining the open reading frame (Murai et al., 1999; 

Ramakrishna et al., 2002). 

2.8. Species comparisons 

Neighbour-joining phylogenetic analyses were conducted in 

PAUP*Version 4.0b10 (Swofford, 2001), using Kimura-2 parameter, 

F.M. Shapter et al. / Journal of Cereal Science 49 (2009) 4–11 

Primer name Sequence 50 –30 Tm ( C) Size 

GC (%) PCR Step 1 

(bp/mer) 

a 

PCR Step 2 

(cycle #/temp. C) 

a 

Polymerase 

(cycle #/temp. C) 

GBSSAaF1 AGTGCAGTCATCTTTCACCA 53.1 20 45.0 8/58–50 30/50 Accuprime DG 

GBSSAaR1 CAGAACCTCAAACTTATTAGCC 53.6 22 40.9 

GBSSAlF3 ACCGCAGCTACGTACTCCGCC 60.0 21 67.7 10/62–53 30/52 Accuprime DG 

GBSSAlR2 CACCGATCCCCCTTGTTCT 59.3 19 57.9 

GBSSEsF3 TGAGGATTCATGCTCTAGGGTTA 58.6 23 43.5 12/60–48 30/49 Accuprime 

GBSSEsR3 AAGAGGCAAGCGGCACA 57.8 17 58.8 

GBSSLeF5 GCACATGTCTTTTCTTGATGC 55.7 21 42.9 12/60–48 30/49 Accuprime 

GBSSLeR5 TGATCCATCATCCTTGTGCTA 56.1 21 42.9 

GBSSMsF1 AAAATTCTGTTATCCCAACCC 55.4 21 39.1 8/8–50 30/50 Accuprime DG 

GBSSMsR2 ATAATGGCTACGACAGACTCAC 53.9 22 45.5 

GBSSOaF1 ACAGCAAGAGCTAGACAACCG 57.4 21 52.4 8/58–50 30/50 Accuprime DG 

GBSSOaR2 AAACCACTGCTTCATTCGTCT 56.6 21 42.9 

GBSSOrF1 ATCTTCCACAGCAACAGCTA 53.1 20 45.0 10/60–50 30/50 Accuprime 

GBSSOrR2 ATCACTCCATATAGATCTCAGGC 54.8 23 43.5 

a 

Two step PCR was utilised: the first step a touchdown cycle, each cycle the annealing temperature decreasing 1 C within the given range, the second step at a constant 

annealing temperature for the given number of cycles. 

distance method and maximum parsimony settings. Support for 

branches was obtained from 1000 pseudoreplicates for 1920 of 

1921 characters, using the TBR branch swapping algorithm. 

3. Results 

3.1. Gene sequence 

Initial alignment of the GBSSI coding sequence from Genbank 

determined that sites for universal primer design at the 5 0 - and 3 0 - 

ends of the gene were not available due to poor homology between 

the species in these regions. Direct sequencing analysis of purified 

PCR amplicons of a small central fragment (CE) of the GBSSI gene 

spanning exons 7 and 8 (McIntosh et al., 2005) produced poor 

sequence data for some species with ambiguous base calls indicating 

multiple alleles. This was expected as, apart from the two 

wild Oryza species, the CWR are polyploids and hence sequence 

data could only be resolved by cloning the PCR products prior to 

sequencing. 

Fragment CE sequence was used as a platform for the design of 

bidirectional species-specific primers for use in the BD Genomewalking 

protocol. The first set of gene/species-specific primers 

produced gene fragments that did not reach the start and stop 

codons. Up to three sets of gene/species-specific primers were 

designed in the exon closest to the respective 5 0 - and 3 0 -ends of the 

sequences obtained in the current round of genome walking 

(Fig. 1). Addition of the recommended concentrations of DMSO and 

glycerol in the second and third rounds of genome walking resulted 

in improved amplification specificity. 

Of the eight species selected for full gene characterisation, seven 

were successfully amplified with full gene primers located in the 5 0 - 

and 3 0 -UTR. Due to the polyploid status of most of these species it 

was necessary to amplify the entire gene sequence in a single PCR 

fragment, then clone the amplicon. This ensured the sequence was 

obtained from a single genome. 

In the eighth species, Sorghum nitidum, genome walking 

produced two major gene fragments, overlapping at intron 4 and 

the start of exon 5. Polymorphisms between the overlapping fragments 

determined that these were two different alleles; indicating 

more than one genome may have been amplified. Although more 

than 20 primer pairs were utilised under various PCR conditions, 

full gene amplification was unsuccessful for this species. Consequently, 

S. nitidum has been entered into Genbank as two separate 

(5 0 and 3 0 ) sequences (Table 1). To determine a putative coding 

sequence for S. nitidum, the gene sequence was spliced, in silico, at

exon 5 and degenerate base calls inserted for the three polymorphic 

positions in the overlapping fragment. 

A. lappacea appeared to be missing the 3 0 portion of exon 1 and 

missing its expected exon 1/intron 1 boundary when compared to 

all other species examined. Exons 2 and 3 were as expected; 

however, the 3 0 boundary of exon 4 occurred 1 base earlier, causing 

a frame shift when translated. Other than exons 5 and 11, the exon 

sizes varied by between 1 and 29 bases, causing repeated frame 

shifts and encoding multiple stop codons commencing with the 

first in exon 5. 

Amongst all the species examined, the genomic base count 

ranged from 2760 bases (E. scaber) through to 3572 bases (Oryza 

rufipogon). This large variation in nucleotide number can be 

accounted for in the Triticeae by the absence of introns 4 and 7. The 

insertion of over 400 bases in intron 11 of O. rufipogon is partially 

responsible for its larger size (Fig. 1). The number of putative coding 

bases varies from 1812 base pairs (bp) in A. aristiglumis up to 

1830 bp for both Oryza australiensis and O. rufipogon, with the four 

cereal species investigated falling within this range. This translates 

to an addition/deletion of up to 14 amino acids. 

3.2. Transposable elements 

In Sorghum leiocladum the first set of genome walking gene/ 

species-specific primers produced a 1300 bp fragment. Nine 

hundred bases of this fragment were homologous to exons 3 to 7 of 

the O. sativa GBSSI coding sequence. At the 5 0 -end of the fragment 

the first 400 bases after the GenomeWalkerÔ AP2 primer sequence 

had no homology to the exons of any GBSSI coding gene in Genbank. 

These 400 bases were isolated and submitted to a BlastN 

search which returned a strong identity to a retrotransposon isolated 

from S. bicolor. Similarly, PCR amplification within the O. 

australiensis GenomeWalkerÔ library produced a fragment which 

had the GenomeWalkerÔ AP2 primer sequence at both ends of the 

fragment. The sequence showed no homology to the cultivated rice 

GBSSI exons, and was determined by BlastN search to be homologous 

to an O. australiensis retrotransposon. The 400 base insert in 

intron 12 of O. rufipogon did not return any matches when used as 

a BlastN query sequence. 


Fig. 1. Schematic diagram showing/illustrating the variations observed across eight Australian grass species and their cultivated cereal relatives. The location of central fragment 

(CE) and the genome-walking gene/species-specific primers are indicated for each species. 

3.3. Granule Bound Starch Synthase I 

Comparison of the putative amino acid sequences derived in this 

study with those for the cereals retrieved from Genbank showed 

little conservation within the 79 amino acid consensus positions of 

the transit peptide (Fig. 2). The consensus of the mature protein was 

540 amino acids and averaged 75% identity. The conserved domain 

database (Marchler-Bauer et al., 2007) indicates two conserved 

domains exist in GBSSI; the glycosyl transferase groups five and 

one. Group five has 79.5% identity and group one maintains 71.4% 

identity. The C-terminus has 74.7% identity within the amino acid 

sequence. The ADPG active binding site near the N-terminus 

(Baldwin, 2001; Denyer et al., 2001) was identical in all species 

except A. lappacea and S. nitidum. 

A single amino acid change occurred in the spliced S. nitidum 

sequence. The deletion of part of exon 1 in the A. lappacea sequence 

translated to the loss of the entire binding site. Of the two 

conserved motifs near the C-terminus (Baldwin, 2001), Motif II had 

a proline substituted for the threonine in the two Oryza species. 

Motif III was conserved across all species. 

3.4. Phylogenetic outcomes 

Phylograms were developed containing all the possible 

boundary variations for the A. lappacea sequence (data not 

shown), including a sequence with no intron 1. All 16 permutations 

were monophyletic and remained a sister group to the 

remainder of the Poaceae examined. The representative coding 

sequence of A. lappacea used in the phylogenetic analyses was 

determined by having the least variation in exon sizes and the 

least amino acid sequence disruptions. Preliminary inclusion of 

the spliced S. nitidum sequence resulted in its position within the 

topology changing when analyses between the second half of the 

gene sequence were compared to the full coding sequence. Using 

the entire sequence S. nitidum was placed between Z. mays and S. 

italica. When only the second half of the gene was analysed S. 

nitidum was clustered with S. leiocladum with stronger statistical 

support. As S. nitidum sequence data was not derived from

8 

a single genome it was excluded from the phylogenetic analysis 

presented. 

No suitable outgroup was available (from Genbank) for the 

phylogenetic analysis so all trees were unrooted. Neighbourjoining 

and maximum parsimony analyses of the full coding 

sequence produced trees with A. lappacea as an outgroup to the 


Fig. 2. Amino acid alignment of GBSSI and associated conserved domains. 

rest of the Poaceae examined (Fig. 3). Analyses produced a single 

congruent phylogeny with good bootstrap support except at the 

bifurcation of A. aristiglumis and M. stipoides which had lower 

bootstrap values. The basal bifurcation of the trees separates 

the BEP (Bambusoideae, Erhartoideae and Pooideae) from the 

PACCAD (Panicoideae, Arundinoideae, Chloridoideae,

63 

66 

100 

100 

61 

60 

50 changes 

51 

53 

Centothecoideae, Aristidoideae and Danthoniodeae) clades. 

Previous phylogenetic analyses utilising the Waxy gene did not 

incorporate the entire coding region and therefore a second tree 

was produced excluding characters 1–986 which resulted in no 

changes to the tree topology. 

3.5. Iodine test 

The iodine test for the presence or absence of amylose determined 

that all eight of the CWR had amylose present in their 

endosperm (Fig. 4). 

4. Discussion 

92 

91 

100 

100 

Alappacea 

100 

100 

100 

100 

Sitalica 

Oaustraliensis 

Orufipogon 

Osativa 

Mstipoides 

Aaristiglumis 

Zmays 

100 

100 

71 

75 

Sleiocladum 

Sbicolor 

Escaber 

T.aestivum 

Hvulgare 

Compound granules 

Bimodal granules 

Simple granules 

Compound granules 

Fig. 3. Neighbour-joining Tree of GBSSI complete coding sequence. Bootstrap values 

are calculated with 1000 replicates with neighbour-joining values presented above the 

branches and maximum parsimony values below. The type of starch granule 

morphology reported for each species is noted on the right of the phylogram. 

GBSSI was successfully amplified in seven of the eight species 

examined. For the eighth species, S. nitidum, screening multiple 

genome walker libraries was unable to isolate the same locus for 

the 5 0 - and 3 0 -gene fragments. As such the two overlapping sections 

of the gene that were sequenced did not provide matching UTR 

Fig. 4. Iodine test for the presence or absence of amylose. A magenta colour indicates 

a waxy phenotype (BR) and blue indicates the presence of amylose (LR). Aa – A. 

aristiglumis, Al–A. lappacea, Es–E. scaber, Ms–M. stipoides, Le–S. leiocladum, Ni–S. 

nitidum, Oa–O. australiensis, Or–O. rufipogon. 


sequence for full gene primers to be successfully applied. 

Comparison of the UTR of the other CWR and cereals sequenced 

showed no homology in this region even amongst closely related 

species. It is therefore not surprising that the sequence differentiation 

between genomes in this region of the gene is diverse enough 

to prevent allele specific primers from amplifying. 

The GBSSI homolog amplified in A. lappacea was missing 

a significant portion of exon 1 which translated to a putative active 

site in the protein. There was little conservation of the exon/intron 

boundaries throughout the gene and once translated, multiple stop 

codons were identified throughout the transcript. It is proposed 

that this sequence was derived from a non-functional ortholog. 

Null-alleles of this gene in diploids result in a waxy or amylose free 

phenotype. For example, an indigenous barley mutant with a large 

deletion including exon 1 was determined to have a waxy phenotype 

(Domon et al., 2002). Testing for amylose indicated that all 

eight CWR contained some amylose (Fig. 4) in their endosperm. The 

ploidy of A. lappacea is not clearly stated in the literature, though its 

base chromosome number of 2n ¼ 40 (Jozwik, 1969) would suggest 

some level of polyploidy. It can be assumed then, that A. lappacea 

has another GBSSI locus yet to be characterised, that is functional 

and being expressed in the endosperm. Wheat lines with multiple 

GBSSI loci still produced amylose in the endosperm in lower 

concentrations than normal when null-alleles did not occur 

concurrently at all GBSSI loci (Mangalika et al., 2003). 

Transposable elements have been previously identified in the 

intron sequence of the Waxy gene (Mason-Gamer et al., 1998; Wang 

et al., 1994). In the case of S. italica, insertion of transposable 

elements in the coding region is responsible for a waxy phenotype 

(Kawase et al., 2005). Multiple attempts to design genome-walking 

primers to amplify past the retrotransposon in the 5 0 -fragment of S. 

leiocladum were undertaken, though remained unsuccessful. It was 

hypothesised that the retrotransposon may be too large to amplify 

across in a single genome walking reaction. An alternative genomewalking 

library was used to amplify the 5 0 -end of the gene from 

another genome in which no retrotransposon was present in intron 

3. Retrotransposons have been reported as being specific within 

a genus or tribe (Bennetzen et al., 2004). Finding retrotransposable 

elements within the CWR genomes is to be expected as these 

transposable elements can comprise up to 50% of nuclear DNA 

content (Kumar and Bennetzen, 1999). The inclusion of transposable 

elements has also been associated with the larger genome size 

of O. australiensis (Vaughan et al., 2003). 

Sequence alignment of the complete gene sequence from 

multiple species identified hot spots of coding sequence divergence 

between the species. The first 200 and final 100 bp of the GBSSI 

coding gene were particularly poorly conserved within Poaceae. 

Amongst the putative functional alleles, variable intron sizes (Fig. 1) 

indicating insertions/deletions were observed, across the species 

with especially large variations within intron 12. Phylogenetic 

investigation used these indels to identify homoeologous copies of 

the Waxy gene (Fortune et al., 2007). The gene variation observed in 

the current study concurs with comparative studies of other 

homologous grass genes, where exon/intron boundaries remain 

conserved, but the size of the introns varies (Feuillet and Keller, 

2002). The coding sequence also showed considerable diversity 

across the taxa sampled. 

Early protein studies determined that the molecular weight of 

GBSSI varied from 58 to 60 kDa between species (Taira et al., 1995). 

Amongst the CWR, coding sequence diversity translates to the 

omission of 14 amino acids over the length of the protein and 

multiple polymorphisms in the amino acid sequence. Any changes 

to the amino acid sequence have the potential to cause novel 

phenotypic expression of this gene. Previous studies of diploid 

mutants containing a null Waxy allele revealed GBSSI to be 

responsible for amylose synthesis in the storage tissues of plants

10 

(Vrinten and Nakamura, 2000). Sequence variations, which translate 

to amino acid changes, may lead to the synthesis of novel 

starches in planta, thus creating the potential for plant breeders to 

manipulate starch synthesis through conventional breeding or 

transgenic modification of cereal crops. The production of cereals 

containing novel starches will reduce or eliminate the need for 

costly post-harvest modifications (Davis et al., 2003). Introgression 

of novel genes from CWR has already been responsible for major 

cereal production gains (Hajjar and Hodgkin, 2007; Langridge et al., 

2001). The allelic variation observed in this small subset of 

Australia’s native grasses shows that considerable functional 

diversity may exist in the native grass gene pool, which could be 

harvested for grain improvement programs. 

The basal bifurcation of the tree separates the BEP and PACCAD 

clades, as was reported from previous phylogenetic studies of the 

Poaceae (Kellogg, 2002; Preston and Kellogg, 2006; Tomlinson 

and Denyer, 2003). Mason-Gamer et al. (1998) concluded that 

sequencing through the introns of GBSSI was unnecessary to 

resolve relationships between genera, tribes and subfamilies within 

Poaceae. Sufficient resolution was obtained from cDNA data and the 

results from the current study support their findings. Similarly, 

Mason-Gamer et al. (1998) utilised just the 3 0 half of the gene, 

which can be universally amplified for analysis. Comparisons of 

phylogenies developed from this smaller region of the gene 

compared to the full coding sequence did not result in any change 

in topology. This analysis indicates that time consuming and costly 

acquisition of sequence data from non-conserved areas of the Waxy 

gene does not provide any clearer resolution of phylogenetic 

inferences for the Poaceae. 

A previous scanning electron microscopy study (Shapter et al., 

2008) identified the starch granule morphology of the eight species 

and their cereal relatives. Since these trees were developed using 

the entire coding region of the Waxy gene, and given the Waxy 

gene’s role in starch formation, comparative analysis of the two 

data sets was undertaken (Fig. 3). There are three main starch 

granule morphologies within the cereals: (1) compound granulesdsmall 

rigidly polygonal granules arranged in large spherical 

compound granules typical of rice; (2) simple granulesdindividual 

large spherical/lenticellular to polygonal granules found in maize 

and sorghum; (3) bimodal granule arrangement with a mix of small 

spherical (B-type) and large lenticellular (A-type) granules typical 

of wheat (Tomlinson and Denyer, 2003). S. italica is the earliest 

diverging lineage within the PACCAD clade (Fig. 3). It is also the only 

member of this clade from this study with compound granules. 

Early SEM investigations of starch granule morphology of other 

members of this clade outside the Andropogoneae, revealed they 

too had compound granules within their endosperm, and it was 

proposed that this phenotype is the ancestral state (Shapter et al., 

2008; Tateoka, 1962). This hypothesis would appear to be correct as 

both simple and bimodal granules occur only in the more recently 

diverged lineages. 

While the association between starch granule morphology and 

Waxy phylogeny appears to indicate a possible link between GBSSI 

and starch granule formation, it seems unlikely as it has previously 

been reported that a loss of function in this gene causes no 

apparent change to starch granule appearance (Buleon et al., 1998; 

Kim et al., 2003). Conversely, lesions in the isoamylase gene in 

barley were correlated to a lack of A and B-type granules and the 

appearance of compound granules in the endosperm (Burton et al., 

2002). Screening of the amino acid alignment of all the species 

examined in this study observed several amino acid changes in the 

conserved domains of the gene which also correlated with the 

starch granule morphology (Fig. 2). These separated either all three 

types of granule (V), just the species with simple granules (U) or 

just the bimodal granules (x). While it is likely that these amino acid 

associations are due to the evolutionary history of the taxa, rather 


than a direct genetic effect of the expression of a particular ortholog 

of GBSSI, it would be uncommon for an enzyme to have a single 

function (determination of amylose) with no other biochemical 

consequences. Perhaps as more is discovered about the role of 

GBSSI in endosperm development, these associations will be found 

to have more than just an evolutionary connection. 


Seed for this study was supplied by Australian Tropical Crops 

and Forages Collection, Queensland Department of Primary 

Industries and Fisheries, www.dpi.qld.gov.au/auspgris/and Native 

Seeds Pty Ltd Australia http://www.nativeseeds.com.au. Funding 

was provided by the Grain Foods Cooperative Research Centre. 

References 

Ashikari, M., Matsuoka, M., 2006. Identification, isolation and pyramiding of 

quantitative trait loci for rice breeding. Trends in Plant Science 11, 344–350. 

Baldwin, P., 2001. Starch granule-associated proteins and polypeptides: a review. 

Starke. (Starch) 53, 475–503. 

Bennetzen, J.L., Coleman, C., Liu, R.Y., Ma, J.X., Ramakrishna, W., 2004. Consistent 

over-estimation of gene number in complex plant genomes. Current Opinion in 

Plant Biology 7, 732–736. 

Buleon, A., Colonna, P., Planchot, V., Ball, S., 1998. Starch granules: structure and 

biosynthesis. International Journal of Biological Macromolecules 23, 85–112. 

Burton, R.A., Jenner, H., Carrangis, L., Fahy, B., Fincher, G.B., Hylton, C., Laurie, D.A., 

Parker, M., Waite, D., van Wegen, S., Verhoeven, T., Denyer, K., 2002. Starch 

granule initiation and growth are altered in barley mutants that lack isoamylase 

activity. The Plant Journal 31, 97–112. 

Davis, J.P., Supatcharee, N., Khandelwal, R.L., Chibbar, R.N., 2003. Synthesis of novel 

starches in planta: opportunities and challenges. Starke. (Starch) 55, 107–120. 

Demeke, T., Huel, P., Abdel-Aal, E.S.M., Baga, M., Chibbar, R.N., 1999. Biochemical 

characterization of the wheat waxy A protein and its effect on starch properties. 

Cereal Chemistry 76, 694–698. 

Denyer, K., Johnson, P., Zeeman, S., Smith, A.M., 2001. The control of amylose 

synthesis. Journal of Plant Physiology 158, 479–487. 

Domon, E., Saito, A., Takeda, K., 2002. Comparison of the waxy locus sequence from 

a non-waxy strain and two waxy mutants of spontaneous and artificial origins 

in barley. Genes & Genetic Systems 77, 351–359. 

Evans, R.C., Campbell, C.S., 2002. The origin of the apple subfamily (Maloideae; 

Rosaceae) is clarified by DNA sequence data from duplicated GBSSI genes. 

American Journal of Botany 89, 1478–1484. 

Feuillet, C., Keller, B., 2002. Comparative genomics in the grass family: molecular 

characterization of grass genome structure and evolution. Annals of Botany 89, 

3–10. 

Fortune, P.M., Schierenbeck, K.A., Ainouche, A.K., Jacquemin, J., Wendel, J.F., 

Ainouche, M.L., 2007. Evolutionary dynamics of Waxy and the origin of hexaploid 

Spartina species (Poaceae). Molecular Phylogenetics and Evolution 43, 

1040–1055. 

Gaines, C.S., Raeker, M.O., Tilley, M., Finney, P.L., Wilson, J.D., Betchel, D.B., Martin, R.J., 

Seib, P.A., Lookhart, G.L., Donelson, T., 2000. Associations of starch gel hardness, 

granule size, waxy allelic expression, thermal pasting, milling quality, and kernel 

texture of 12 soft wheat cultivars. Cereal Chemistry 77, 163–186. 

Hajjar, R., Hodgkin, T., 2007. The use of wild relatives in crop improvement: a survey 

of developments over the last 20 years. Euphytica 156, 1–13. 

Jansen, R., Ledley, F.D., 1990. Disruption of phase during PCR amplification and 

cloning of heterozygous target sequences. Nucleic Acids Research 18, 5153–5156. 

Jensen, M., Straus, N., 1993. Effect of PCR conditions on the formation of heteroduplex 

and single-stranded DNA products in the amplification of bacterial 

ribosomal DNA spacer regions. PCR Methods Application 3, 186–194. 

Jozwik, F.X., 1969. Some systematic aspects of Mitchell grasses. Australian Journal of 

Botany 17, 359–374. 

Kawase, M., Fukunaga, K., Kato, K., 2005. Diverse origins of waxy foxtail millet crops 

in east and southeast Asia mediated by multiple transposable element insertions. 

Molecular Genetics and Genomics 274, 131–140. 

Kellogg, E.A., 2002. Classification of the grass family. In: Mallett, K., Orchard, A. 

(Eds.), Flora of Australia. ABRS/CSIRO Australia, Melbourne, pp. 19–36. 

Khush, G.S., 2001. Challenges for meeting the global food and nutrient needs in the 

new millennium. Proceedings of the Nutrition Society 60, 15–26. 

Kim, W., Johnson, J.W., Graybosch, R.A., Gaines, C.S., 2003. Physicochemical properties 

and end-use quality of wheat starch as a function of waxy protein alleles. 

Journal of Cereal Science 37, 195–204. 

Kumar, A., Bennetzen, J., 1999. Plant retrotransposons. Annual Reviews in Genetics 

33, 479–532. 

Langridge, P., Lagudah, E.S., Holton, T.A., Appels, R., Sharp, P.J., Chalmers, K.J., 2001. 

Trends in genetic and genome analyses in wheat: a review. Australian Journal of 

Agricultural Research 52, 1043–1077. 

Longeri, M., Zanotti, M., Damiani, G., 2002. Recombinant DRB sequences produced 

by mismatch repair of heteroduplexes during cloning in Escherichia coli. European 

Journal of Immunogenetics 29, 517–523.

Mangalika, W.H.A., Miura, H., Yamauchi, H., Noda, T., 2003. Properties of starches 

from near-isogenic wheat lines with different Wx protein deficiencies. Cereal 

Chemistry 80, 662–666. 

Marchler-Bauer, A., Anderson, J.B., Derbyshire, M.K., DeWeese-Scott, C., 

Gonzales, N.R., Gwadz, M., Hao, L., He, S., Hurwitz, D.I., Jackson, J.D., Ke, Z., 

Krylov, D., Lanczycki, C.J., Liebert, C.A., Liu, C., Lu, F., Lu, S., Marchler, G.H., 

Mullokandov, M., Song, J.S., Thanki, N., Yamashita, R.A., Yin, J.J., Zhang, D., 

Bryant, S.H., 2007. CDD: a conserved domain database for interactive domain 

family analysis. Nucleic Acids Research 35, D237–D240. 

Mason-Gamer, R.J., Weil, C.F., Kellogg, E.A., 1998. Granule-bound starch synthase: 

structure, function, and phylogenetic utility. Molecular Biology and Evolution 

15, 1658–1673. 

McIntosh, S.R., Pacey-Millar, T., Henry, R.J., 2005. A universal protocol for identification 

of cereals. Journal of Cereal Science 41, 37–46. 

Murai, J., Taira, T., Ohta, D., 1999. Isolation and characterization of the three Waxy 

genes encoding the granule-bound starch synthase in hexaploid wheat. Gene 

234, 71–79. 

Nakamura, T., Yamamori, M., Hirano, H., Hidaka, S., Nagamine, T., 1995. Production 

of waxy (amylose-free) wheats. Molecular and General Genetics 248, 

253–259. 

Ochieng, J.W., Muigai, A.W.T., Ude, G.N., 2007. Phylogenetics in plant biotechnology: 

principles, obstacles and opportunities for the resource poor. African Journal of 

Biotechnology 6, 639–649. 

Pedersen, J.F., Bean, S.R., Funnell, D.L., Graybosch, R.A., 2004. Rapid iodine staining 

techniques for identifying the waxy phenotype in sorghum grain and waxy 

genotype in sorghum pollen. Crop Science 44, 764–767. 

Pedersen, J.F., Graybosch, R.A., Funnell, D.L., 2007. Occurrence of the waxy alleles 

wxa and wxb in waxy sorghum plant introductions and their effect on starch 

thermal properties. Crop Science 37. 

Preston, J.C., Kellogg, E.A., 2006. Reconstructing the evolutionary history of 

paralogous APETALA1/FRUITFULL-like genes in grasses (Poaceae). Genetics 174, 

421–437. 

Rahman, S., Li, Z., Batey, I., Cochrane, M.P., Appels, R., Morell, M., 2000. Genetic 

alteration of starch functionality in wheat. Journal of Cereal Science 31, 

91–110. 

Ramakrishna, W., Ma, J.X., SanMiguel, P., Emberton, J., Dubcovsky, J., Shiloff, B.A., 

Jiang, Z.Y., Rostoks, N., Busso, C.S., Ogden, M., Linton, E., Kleinhofs, A., 

Devos, K.M., Messing, J., Bennetzen, J.L., 2002. Frequent genic rearrangements in 


two regions of grass genomes identified by comparative sequence analysis. 

Comparative and Functional Genomics 3, 165–166. 

Rao, N.K., Reddy, L.J., Bramel, P.J., 2003. Potential of wild species for genetic 

enhancement of some semi-arid food crops. Genetic Resources and Crop 

Evolution 50, 707–721. 

Shapter, F.M., Lee, L.S., Henry, R.J., 2008. Endosperm and starch granule morphology 

in wild cereal relatives. Plant Genetic Resources 6, 85–97. 

Shure, M., Wessler, S., Fedoroff, N., 1983. Molecular identification and isolation of 

the Waxy locus in maize. Cell 35, 225–233. 

Small, R.L., Cronn, R.C., Wendel, F., 2004. L.A.S. Johnson Review No. 2: use of nuclear 

genes for phylogeny reconstruction in plants. Australian Systematic Botany 17, 

145–170. 

Swofford, D.L., 2001. PAUP: Phylogenetic Analysis Using Parsimony (*and Other 

Methods), version 4.0b10. Sinauer Associates, USA. 

Taira, T., Fujita, N., Takaoka, K., Uematsu, M., Wadano, A., Kozaki, S., Okabe, S., 1995. 

Variation in the primary structure of waxy proteins (granule-bound starch 

synthase) in diploid cereals. Biochemical genetics 33, 269–281. 

Tateoka, T., 1962. Starch grains of endosperm in grass systematics. Botany Magazine 

Tokyo 75, 377–383. 

Thompson, J.R., Marcelino, L.A., Polz, M.F., 2002. Heteroduplexes in mixed-template 

amplifications: formation, consequence and elimination by ‘reconditioning 

PCR’. Nucleic Acids Research 30, 2083–2088. 

Tomlinson, K., Denyer, K., 2003. Starch synthesis in cereal grains. Advances in 

Botanical Research 40, 1–61. 

Vaughan, D.A., Morishma, H., Kadowaki, K., 2003. Diversity in the Oryza genus. 

Current Opinion in Plant Biology 6, 139–146. 

Vrinten, P.L., Nakamura, T., 2000. Wheat granule-bound starch synthase l and ll are 

encoded by separate genes that are expressed in different tissues. Plant 

Physiology 122, 255–263. 

Wang, Z.Y., Zheng, F.Q., Gao, J.P., Wang, X.Q., Wu, M., Zhang, J.L., Hong, M.M., 1994. 

Identification of two transposon-like elements in rice Wx gene. Science in 

China 37, 437–447. 

Williams, W., Verry, I.M., Ellison, N.W., 2006. A phylogenetic approach to germplasm 

use in clover breeding. In: Woodfield, D. (Ed.), 13th Australasian Plant 

Breeders Conference, Christchurch, NZ. 

Yan, L., Bhave, M., Fairclough, R., Konik, C., Rahman, S., Appels, R., 2000. The genes 

encoding granule-bound starch synthase at the waxy loci of the A, B and D 

progenitors of common wheat. Genome 43, 264–272.

Effect of high temperature on albumin and globulin accumulation in the 

endosperm proteome of the developing wheat grain q 

William J. Hurkman *, William H. Vensel, Charlene K. Tanaka, Linda Whitehand, Susan B. Altenbach 

U.S. Department of Agriculture, Agricultural Research Service, Western Regional Research Center, 800 Buchanan Street, Albany, CA 94710, USA 

article info 


Received 14 December 2007 

Received in revised form 6 June 2008 


Keywords: 

Albumins 

Globulins 

Grain fill 

High temperature 


abstract 

High temperature during grain fill is one of the more significant 

environmental factors that affects wheat yield and flour quality 

(Skylas et al., 2002 and references therein). Although major 

processes in grain fill have been described (Laudencia-Chingcuanco 

et al., 2007; McIntosh et al., 2007; Vensel et al., 2005), little is 

known about the effects of high temperature on this developmental 

program. High temperature shortens the duration of grain fill and 

decreases the time to apoptosis and harvest maturity (Altenbach 

et al., 2003). Consistent with these events, transcripts for a-, g-, and 

u-gliadins, high molecular weight glutenin subunits (HMW-GS), 

and low molecular weight glutenin subunits (LMW-GS) accumulate 

and disappear earlier (Altenbach et al., 2002; Altenbach and 

Kothari, 2004). Transcripts for genes functioning in protein 

synthesis, starch synthesis, stress/defense, as well as storage 

accumulate earlier in response to high temperature (Altenbach and 

Abbreviations: dpa, days post-anthesis; HMW-GS, high molecular weight glutenin 

subunits; LMW-GS, low molecular weight glutenin subunits. 

q Disclaimer: The mention of a trademark or proprietary product does not 

constitute a guarantee or warranty of the product by the United States Department 

of Agriculture and does not imply its approval to the exclusion of other products 

that may be suitable. 


E-mail address: william.hurkman@ars.usda.gov (W.J. Hurkman). 

0733-5210/$ – see front matter Published by Elsevier Ltd. 

doi:10.1016/j.jcs.2008.06.014 





The accumulation of KCl-soluble/methanol-insoluble albumins and globulins was investigated in the 

endosperm of developing wheat (Triticum aestivum, L. cv. Butte 86) grain produced under a moderate 

(24 C/17 C, day/night) or a high temperature regimen (37 C/28 C) imposed from 10 or 20 days postanthesis 

(dpa) until maturity. Proteins were separated by 2-DE and developmental profiles for nearly 200 

proteins were analyzed by hierarchical clustering. Comparison of protein profiles across physiologically 

equivalent stages of grain fill revealed that high temperature shortened, but did not substantially alter, 

the developmental program. Accumulation of proteins shifted from those active in biosynthesis and 

metabolism to those with roles in storage and protection against biotic and abiotic stresses. Few proteins 

responded transiently when plants were transferred to the high temperature regimens, but levels of 

a number of proteins were altered during late stages of grain development. Specific protein responses 

depended on whether the high temperature regimens were initiated early or mid development. Some of 

the heat responsive proteins have been implicated in gas bubble stabilization in bread dough and others 

are suspected food allergens. 

Published by Elsevier Ltd. 

Kothari, 2004; Hurkman et al., 2003). High temperatures during 

grain fill influence gluten protein accumulation levels. The relative 

amounts of certain a-gliadins and HMW-GS were higher and those 

for certain LMW-GS were lower in grain produced under a 37 C/ 

28 C day/night regimen (DuPont et al., 2006b). In addition, accumulation 

rates increased more for the a-gliadins and HMW-GS than 

the LMW-GS in grain produced under this regimen (DuPont et al., 

2006a). High temperature during grain fill also affects the accumulation 

levels of stress/defense proteins. Skylas et al. (2002) 

reported that heat shock, a shift from a 24 C/18 Ctoa40 C/25 C 

day/night regimen from 15–17 dpa, increased the number of small 

heat shock protein (HSP) isoforms in total protein extracts from 

endosperm. Majoul et al. (2003) reported that levels of HSPs as well 

as proteins that defend against reactive oxygen species (ROS) and 

desiccation increased in flour when developing grain was exposed 

to an extended high temperature day/night regimen of 34 C/10 C 

from anthesis to maturity. Majoul et al. (2004) isolated an albumin 

and globulin fraction from mature grain and found that the levels of 

a number of proteins, including HSPs as well as enzymes involved 

in starch biosynthesis, carbohydrate metabolism, and ATP 

synthesis, responded to the high temperature regimen. 

In this study, we utilize proteome maps to develop a comprehensive 

picture of endosperm development and identify the effects 

of high temperature on protein composition during this crucial 

reproductive stage. Plants were grown under three temperature 

conditions – a moderate temperature regimen maintained

throughout grain development and a high temperature regimen 

initiated early or mid development and maintained until maturity. 

The high temperature regimens are more severe than might be 

encountered under field conditions, but were selected to accentuate 

protein responses. KCl-soluble/methanol-insoluble albumins 

and globulins were isolated from the endosperm at selected time 

points during grain development and separated by 2-DE. These 

proteins function in a wide range of cellular processes (Vensel et al., 

2005), including the synthesis of gluten proteins and starch, major 

determinants of wheat yield and grain quality. Protein accumulation 

profiles and functions were compared across physiologically 

equivalent stages of grain fill to define major events of endosperm 

development and identify specific responses to high temperature. 

Protein levels in mature grain produced under the three temperature 

regimens were also compared to discover changes in protein 

composition related to grain quality. 


2.1. Plant material 

Triticum aestivum, L. cv. Butte 86, was grown in climatecontrolled 

greenhouses under moderate and high temperature 

regimens (Altenbach et al., 2007b). For the moderate temperature 

regimen, plants were grown at a maximum daytime temperature of 

24 C and a minimum nighttime temperature of 17 C. For the high 

temperature regimen, plants were transferred at 10 or 20 days postanthesis 

(dpa) to a second greenhouse and grown at a maximum 

daytime temperature of 37 C and a minimum nighttime temperature 

of 28 C. Water and fertilizer (Plantex 20-20-20, 300 mg/day) 

were applied by drip irrigation. Natural light was supplemented 

with 100 W high-pressure sodium lamps to maintain a day length of 

16 h. Heads were collected at selected time points during grain 

development. Under the moderate temperature regimen, heads 

were collected at 4-day intervals from 10 to 38 dpa and at 40 dpa. 

Because high temperature shortened the grain developmental 

program, heads were collected at 2-day intervals from 10 to 26 dpa 

for the high temperature regimen initiated at 10 dpa and from 20 to 

28 dpa for the high temperature regimen initiated at 20 dpa. The 

end point for each regimen was the oldest age that endosperm could 

be squeezed or scraped from the grain. Grains were removed from 

the heads and the region containing the embryo was excised with 

a razor blade. The endosperm was squeezed through the resultant 

opening in the pericarp/testa. Endosperm was transferred immediately 

into tubes cooled in liquid nitrogen and stored at 80 C. 

Endosperm was also collected from plants grown in three additional 

experiments in which a 37 C/28 Cora37 C/17 C high temperature 

regimen with 300 mg per day fertilizer was initiated at anthesis 

and a 37 C/28 C high temperature regimen with 150 mg per day 

fertilizer was initiated at 15 dpa. 

2.2. 2-DE analysis 

A KCl-soluble/methanol-insoluble albumin and globulin fraction 

was prepared from the endosperm and triplicate 2-D gels loaded with 

equal protein amounts from each time point were run as described 

previously (Hurkman and Tanaka, 2004; Vensel et al., 2005). Gels 

were digitized with a calibrated scanner (UMAX Powerlook III; Dallas, 

TX) at 300 dpi using the same settings for all gels. Computer software 

(Progenesis PG240 Ver. 2006; Nonlinear Dynamics Limited, Newcastle 

uponTyne, UK) was used to match proteinpatterns and analyze 

the quantitative and qualitative differences across time points for the 

three temperature regimens; matching was validated manually for 

all spots. Protein accumulation profiles were constructed from 

the normalized spot volume (individual spot volume/total spot 

volume 100) data and validated manually. 

W.J. Hurkman et al. / Journal of Cereal Science 49 (2009) 12–23 13 

2.3. Protein identification 

Proteins contained in 2-D gel spots were digested with trypsin 

using a DigestPro gel-spot-processing robot (Intavis, Lagenfeld, 

Germany). Peptides from the tryptic digests were identified using 

a QSTAR PULSAR i quadrupole time-of-flight (TOF) mass spectrometer 

(Applied Biosystems/MDS SCIEX, Toronto, Canada) as 

previously described (Vensel et al., 2005). The resulting AnalystQS 

wiff files were converted to MGF files using Mascot Daemon (http:// 

www.matrixscience.com/) and data analysis of the MS/MS files was 

carried out using a locally installed copy of the Global Proteome 

Machine Organization (GPM) software (http://www.thegpm.org). 

X!Tandem (version 2005.06.01.2), the spectrum modeler in the 

GPM software, was used to match MS/MS fragmentation data to 

peptide sequences. MGF files were submitted in batch mode to 

X!Tandem using a script provided by Jayson Faulkner (University of 

Michigan). X!Tandem was configured to search a file containing 

695,000 protein sequence entries from all proteins in the 

HarvEST:Wheat version 1.04 database (http://harvest.ucr.edu/ 

HWheat104.exe), NCBI nonredundant green plant database, NCBI 

Triticum aestivum: UniGene Build #37, and wEST Database (http:// 

wheat.pw.usda.gov/wEST). A local installation of the GPM and the 

GPMDB (Craig et al., 2004) was used to analyze, store, and display 

the data. The protein identification and mass spectrometry data are 

available as XML files (http://wheat.pw.usda.gov/pubs/2008). To 

access the data, click on Hurkman, scroll down to Compressed 

directory of individual XML files for gel spots, click to download, 

and then unzip the files. To view the files, access the GPM viewer 

(http://h777.thegpm.org/tandem/thegpm_upview.html), click 

Browse, select the file for the spot number of interest, and click 

View Models. Click Protein to see the protein sequence and identified 

peptides. 

2.4. Statistical analyses 

The protein accumulation data was analyzed by hierarchical 

cluster analysis to determine similarities and differences across the 

developmental time courses for the moderate and high temperature 

regimens. The data set consisted of accumulation profiles for 

193 proteins. Each profile consisted of 23 time points across the 

three temperature regimens. In order to discover patterns independent 

of protein abundance, the data for each protein were 

converted to a percent scale (normalized volume of a given time 

point/total normalized volume for the 23 time points 100). 

Hierarchical cluster analysis was performed with SAS software, 

PROC CLUSTER, using the average method. Statistical algorithms, 

including CCC, pseudo t, and pseudo F were used to choose the final 

number of clusters (SAS Institute Inc., 2004). 

To determine reproducibility of the data for effects of high 

temperature on endosperm protein levels late in grain development, 

correlations were computed for the data from the high 

temperature regimens initiated at anthesis, 10, 15, and 20 dpa. To 

ensure that the correlations were not unduly influenced by large 

observations, the Kendall correlation method, based on ranks 

across all proteins, was performed using the SAS software PROC 

CORR (SAS Institute Inc., 2004). In addition Cronbach’s alpha was 

computed to estimate consistency among regimens. 

3. Results 

3.1. Endosperm proteome maps 

In a previous study (Vensel et al., 2005), we developed proteome 

maps for albumins and globulins isolated from the endosperm of 

grain grown under moderate temperatures (24 C/17 C, day/ 

night). The maps, which contain 254 proteins, were established

14 

using two developmental time points, 10 and 36 dpa, to maximize 

proteome coverage. In the present study, we mapped and identified 

35 additional endosperm proteins (Fig. S1). Spot numbers, accession 

numbers, percent coverage, expectation values, and peptide 

sequence data for these proteins are listed in Table S1. Of the 289 

proteins in the updated proteome maps, 193 were selected for 

profile analysis. Spots that contained more than one protein, 90 in 

all, were eliminated from the experimental set, because normalized 

volumes for the individual proteins within these spots could not be 

determined. Since this study focuses on the albumins and globulins, 

the 6 spots containing gliadins and glutenins that co-purified 

with this fraction (Vensel et al., 2005) also were eliminated from 

the data set. 

3.2. Hierarchical cluster analysis 

Accumulation profiles based on normalized spot volumes were 

established for proteins in endosperm harvested from grain grown 

under the moderate and high temperature regimens (Fig. 1). The 

data set for the high temperature regimen initiated at 20 dpa 

includes spot volumes from the three initial time points (10, 14, and 

18 dpa) of the moderate temperature regimen under which the 

plants were grown before transfer to this high temperature regimen. 

When normalized protein volumes were summed for all time points 

across the three temperature regimens for each protein, the totals 

spanned four orders of magnitude, ranging from 0.13 for an isoform 

of globulin-2 (#561) to 118 for protein disulfide isomerase (PDI, 

#118). Because of this wide dynamic range, the normalized volume 

data were converted to percentages so that hierarchical cluster 

analysis results would be independent of protein amount. In Fig. 1, 

the protein profile data were converted to colors: white represents 

the lowest percentage range (0–0.75), shades of blue the intermediate 

percentage ranges (>0.75–40), and black the highest 

percentage range (>40). The identities of the proteins corresponding 

to the profiles depicted in Fig. 1 and their functions are listed in 

Table 1 by the order and cluster numbers determined by the hierarchical 

cluster analysis (Fig. S2). The order numbers in Table 1 may 

be used to identify proteins shown in Fig. 1. The spot numbers in 

Table 1 may be used to locate the spots on the proteome maps in 

Fig. S1 of this paper and Figs. 1 and 2 of Vensel et al. (2005). 

The hierarchical cluster analysis separated the 193 protein 

accumulation profiles into 39 clusters (Fig. 1 and Fig. S2), but nearly 

65% of the proteins in the analysis were members of four clusters (1, 

2, 3, and 5). Cluster 1 with 58 members was the largest cluster. The 

majority of these proteins functioned in carbohydrate metabolism, 

protein synthesis/assembly, and nitrogen metabolism. Cluster 5 

with 40 members was the second largest cluster. The majority of 

the proteins in this cluster functioned in stress/defense, although 

some also functioned in carbohydrate and nitrogen metabolism. 

The next largest clusters were 2 and 3 with 16 and 13 members, 

respectively. In contrast to cluster 1, the majority of the proteins 

in these clusters functioned in transcription/translation. The 

remaining clusters, 4 and 6–39, were smaller, containing 1–10 

members, and the majority of these proteins functioned in carbohydrate 

metabolism, protein synthesis/assembly, storage and 

stress/defense. The most abundant proteins (total normalized 

volume >15, indicated by purple boxes in Fig. 1 and an H in Table 1) 

were members of the four largest clusters (1, 2, 3, and 5). The least 

abundant proteins (total normalized volume


Fig. 1. Accumulation patterns of KCl-soluble/methanol-insoluble albumins and globulins in wheat endosperm. Proteins were isolated from grain produced under moderate (24 C/ 

17 C) and high temperature conditions (37 C/28 C from 10 or 20 dpa). Profiles are shown as percent of total normalized volume. Hierarchical clusters (1–39) and accompanying 

protein order were determined by SAS analysis (see Fig. S1). Protein identifications are in Table 1. The most abundant and least abundant proteins, based on total normalized spot 

volume for the three treatments, are indicated in the ‘Abundance’ column. Differences in profiles between the 24 C/17 C and the 37 C/28 C from 10 or 20 dpa regimens are 

indicated in the ‘Changes with Heat’ columns.

16 

W.J. Hurkman et al. / Journal of Cereal Science 49 (2009) 12–23 

Table 1 

Identities of endosperm proteins corresponding to the accumulation profiles shown in Fig. 1 

Order no. Cluster no. Spot no. Name Function a Peak timing b 

24/17 37/28 10 dpa 37/28 20 dpa 

1 1 264 Glyceraldehyde 3-P dehydrogenase (NAD) CM E E E H 

2 1 294 Malate dehydrogenase (NAD) CM E E E H 

3 1 262 Glyceraldehyde 3-P dehydrogenase (NAD) CM E E E H 

4 1 112 PPi-fructose 6-P 1-phosphotransferase b-subunit CM E E E 

5 1 252 Aldolase CM E E E H 

6 1 383 Cyclophilin A-2 PS E E E H 

7 1 97 Fructokinase CM E E E 

8 1 118 Protein disulfide isomerase PS E E E H 

9 1 176 Alanine amino transferase 2 NM E E E H 

10 1 138 Ketol-reductoisomerase NM E E E 

11 1 154 Aldehyde dehydrogenase CM E E E 

12 1 250 Reversibly glycosylated polypeptide CM E E E 

13 1 354 Triosephosphate isomerase CM E E E 

14 1 6 Carbamoyl phosphate synthetase NM E E E L 

15 1 69 Poly(A)-binding protein TT E E E 

16 1 72 10-Formyltetrahydrofolate synthetase NM E E E 

17 1 159 ADP-glucose PPase, SS CM E E E 

18 1 433 Gycine-rich RNA-binding protein TT E E E 

19 1 65 Heat shock protein 70 PS E E E 

20 1 145 Selenium binding protein S E E E 

21 1 103 Phosphoglycerate mutase, 2,3-bisphosphoglycerate-independent CM E E E 

22 1 104 Phosphoglycerate mutase, 2,3-bisphosphoglycerate-independent CM E E E 

23 1 96 Phosphoglucomutase CM E E E 

24 1 173 Alanine amino transferase 2 NM E E E 

25 1 135 Leucine amino peptidase PT E E E 

26 1 295 Malate dehydrogenase (NAD) CM E E E 

27 1 48 Methionine synthase NM E E E 

28 1 301 Guanine nucleotide-binding protein b subunit-like protein ST E E E 

29 1 415 Ubiquitin-protein ligase PT E E E 

30 1 444 40S Ribosomal protein S21 PS E E E 

31 1 157 UDP-glucose PPase CM E E E 

32 1 158 UDP-glucose PPPase CM E E E 

33 1 411 Gycine-rich RNA-binding protein TT E E E 


35 1 153 Aldehyde dehydrogenase S E E E 

36 1 655 S-Adenosylmethionine synthetase 2 NM E E E 

37 1 98 Acetohydroxyacid synthase NM E E E 

38 1 25 Aconitase CM E E E 

39 1 27 Aconitase CM E E E 

40 1 179 Enolase CM E E E 

41 1 185 Enolase CM E E E H 

42 1 161 Dihydrolipoamide dehydrogenase CM E E E 

43 1 82 Stress-induced protein, sti1-like S E E E 

44 1 117 Chaperonin 60 kDa b-subunit PS E E E 

45 1 1004 Fructose bisphosphate aldolase CM E E E 

46 1 373 20S Proteasome a-subunit B PT E E E 

47 1 278 TGF-b receptor-interacting protein 1 ST E E E 

48 1 306 Aldolase CM E E E 

49 1 163 Heat shock associated protein PS E E E 

50 1 74 DNAK-type molecular chaperone HSP70 PS E E E 

51 1 149 Heat shock associated protein PS E E E 

52 1 285 Legumin-like protein SP E E E 

53 1 61 Poly(A)-binding protein TT E E E L 

54 1 378 GSH-dependent dehydroascorbate reductase 1 S E E E 

55 1 447 Polyubiquitin 6 PT E E E 

56 1 49 Heat shock protein 80-2 PS E E E 

57 1 471 DNAK-type molecular chaperone HSP70 PT E E E 

58 2 342 Ascorbate peroxidase S E E E 


60 2 119 Phosphoglycerate dehydrogenase-like protein CM E E E 

61 2 193 6-Phosphogluconate dehydrogenase CM E E E 

62 2 657 S-Adenosylmethionine synthetase 1 NM E E E 


64 2 178 Enolase CM E E E 






70 2 31 Pyruvate Pi dikinase CM E E E 


72 2 493 UDP-glucose dehydrogenase PS E E E 


74 3 361 20S Proteasome a-subunit PT E E E 

Abundance c

Table 1 (continued ) 



24/17 37/28 10 dpa 37/28 20 dpa 

Abundance c 

75 3 399 Translationally controlled tumor protein TT E E E 

76 3 362 Elongation factor 1-b PS E E E 

77 3 94 RNA-binding protein, similarity TT E E E L 

78 3 182 Tubulin a-3 chain CD E E E 

79 3 412 Glycine-rich RNA-binding protein TT E E E 

80 3 201 SGT1 S E E E 



83 3 330 14-3-3 protein ST E E E 

84 3 640 Phosphoglycerate kinase CM E E E 

85 3 279 Adenosine kinase A E E E 

86 3 93 RNA-binding protein, similarity TT E E E 

87 4 42 Heat shock protein, 82 K PS E E E 

88 5 350 Small ras-related GTP-binding protein T M M M 

89 5 382 Dehydroascorbate reductase S M M M H 

90 5 356 Triosephosphate isomerase CM M M M 

91 5 110 2-Isopropylmalate synthase NM M M M 

92 5 205 26S Proteasome regulatory particle triple-A ATPase subunit 4 PT M M M 

93 5 232 Aspartate amino transferase NM M M M 

94 5 175 Globulin-2 SP M M M 

95 5 353 Avenin SP M M L H 

96 5 417 a-Amylase/trypsin inhibitor, CM3 S M M L H 

97 5 43 Late embryogenesis abundant protein-like S M L L L 

98 5 379 Expressed protein U M M M 

99 5 430 a-Amylase inhibitor 0.19 S M M M H 

100 5 111 PPi-fructose-6-P 1-phosphotransferase CM M M M 

101 5 436 a-Amylase inhibitor 0.53 S M M M 

102 5 384 Avenin N9 SP M L M H 

103 5 440 a-Amylase inhibitor Ima1, monomeric S M M M 

104 5 381 a-Amylase/subtilisin inhibitor S M L L H 

105 5 386 F10K1.21/F7A7_100 protein, similarity U M L L H 

106 5 245 Serpin WZS3 S M L M 

107 5 369 Avenin N9 SP E M E 

108 5 434 a-Amylase inhibitor 0.53 S M M L 

109 5 233 Serpin S E E M 

110 5 236 Aspartate amino transferase NM E E M 

111 5 231 Serpin WZS2 S E E M H 

112 5 246 Serpin S L M M H 

113 5 237 Serpin WZS3 S L M M 

114 5 33 Elongation factor 2 PS L M L 

115 5 305 Glyoxalase I S M L L 

116 5 128 ADP-glucose PPase, LS CM M M M 



119 5 148 Catalase isozyme 1 S M M M 

120 5 244 Aldolase CM M M M 

121 5 251 Formate dehydrogenase NM M M M 

122 5 249 Formate dehydrogenase NM M M M 

123 5 206 Eukaryotic initiation factor 4A TT M E M 

124 5 312 Purple acid phosphatase S L E M 

125 5 137 Leucine amino peptidase PT L L L 

126 5 418 Superoxide dismutase [Cu–Zn] S M L L 

127 5 102 Protein disulfide isomerase-like protein PS L M L 

128 6 141 Elongation factor 1-a PS L L L 

129 6 337 Seed globulin SP L M L 

130 6 336 Seed globulin SP L M L 

131 6 327 Seed globulin SP L M M 

132 6 335 Seed globulin SP L M M 

133 7 1009 Peroxidase 1 S L M M 

134 8 199 Triticin SP M L M 

135 9 255 Peroxidase 1 S L L L 

136 9 259 Peroxidase 1 S L L L 

137 9 263 Peroxidase BP1 S L L L 

138 9 371 Peroxiredoxin S L L L 

139 9 314 OSJNBb0118P14.5 U L L L 

140 9 324 OSJNBb0118P14.5 U L L L L 

141 9 168 Dihydrolipoamide dehydrogenase CM L L L 

142 9 212 DNAK-type molecular chaperone HSP70 PS L L L 

143 9 441 Avenin N9 SP L L L L 

144 9 639 Serpin S L L L 

145 10 407 Heat shock protein 16.9C PS M M M 


147 11 338 DNA-binding protein HEXBP TT E E E 


149 13 341 Chitinase-c S L L L 

(continued on next page)

18 

Table 1 (continued ) 


24/17 37/28 10 dpa 37/28 20 dpa 

150 13 351 Chitinase-a S L L L 

151 13 282 Chitinase-a S L L L 

152 13 842 Late embryogenesis abundant protein-like S L L L L 

153 13 1001 Unknown protein [similar to late embryogenesis abundant protein-like] S L L L 

154 13 557 Embryo-specific protein U L L L L 

155 13 551 Late embryogenesis abundant protein-like S L L L 

156 13 427 Pathogenesis-related protein 4 S L L L 

157 14 318 Xylanase inhibitor I S L L L 

158 14 550 Sucrose synthase type 2 CM L L L 

159 14 633 Nonspecific lipid-transfer protein T L L L 

160 15 355 Seed globulin SP L L L 

161 15 591 Glucose and ribitol dehydrogenase CM L L L 

162 15 275 Glyceraldehyde 3-P dehydrogenase (NAD) CM L L L 

163 15 347 20S Proteasome a-subunit PT L L L 

164 15 420 Nucleoside diphosphate kinase I A L L L 

165 15 281 Globulin Beg 1 SP E L E 

166 16 147 Globulin-2 SP L L L 

167 17 376 Superoxide dismutase [Mn] S L L L 


169 18 620 Globulin Beg 1 SP L L L 

170 19 631 Globulin-like protein SP L L L L 

171 20 1012 Peroxidase 1 S L A M 

172 20 1014 Peroxidase 1 S L A M 

173 21 310 Glucose and ribitol dehydrogenase CM L M E 

174 22 283 TGF-b receptor-interacting protein 1 ST E A E 

175 23 359 Peroxidase 1 S E M E 

176 24 1010 Peroxidase 1 S M L M L 

177 25 419 Glycine-rich RNA-binding protein TT M E E 

178 26 672 Reversibly glycosylated polypeptide CM E E E 

179 27 357 Triosephosphate isomerase CM E E E 

180 28 869 60S Ribosomal protein L12 PS L L L 

181 29 125 b-Amylase CM L L L 

182 30 1007 Glyceraldehyde 3-P dehydrogenase (NAD) CM L A L 


184 31 851 Globulin-2 SP L L L L 

185 32 852 Globulin-2 SP L L A L 

186 33 871 Barwin S L L L 

187 34 561 Globulin-2 SP L L A L 

188 35 1022 Embryo globulin SP A L A L 

189 36 1002 Embryo globulin SP L L L L 

190 36 1003 Embryo globulin SP L L L 

191 37 311 Initiation factor 3g TT E E E L 

192 38 530 Cyclophilin A-1 PS E E E 

193 39 364 Ascorbate peroxidase S E E E L 

analysis of physiologically equivalent stages shows that the high 

temperature regimens did not substantially alter the overall pattern 

of protein accumulation during grain development. 

3.4. Transient effects of high temperature on protein profiles during 

grain development 

Developmental profiles were examined for transient responses 

following transfer of plants to the high temperature regimens 

initiated at 10 and 20 dpa. Proteins that increased (yellow boxes in 

Fig. 1) or decreased (purple boxes in Fig. 1) transiently in response 

to the high temperature regimens are listed in Table 2. Under the 

high temperature regimen initiated at 10 dpa, two HSP70 isoforms 

(225, 471), legumin-like protein (285), and triosephosphate isomerase 

(354) increased transiently early in development. HSP16.9 

(407), four globulin isoforms (327, 335–337), and a peroxidase 

(1009) increased while three isoforms of ADP-glucose PPase large 

subunit (128, 130, 132) decreased transiently during mid development 

under this regimen. Under the high temperature regimen 

initiated at 20 dpa, only two transient changes were detected. 


Abundance c 

Protein clusters were determined using SAS as outlined in Section 2. Spot numbers correspond to 2-DE map locations in Fig. S1 and Figs. 1 and 2 in Vensel et al. (2005). 

a 

A, ATP interconversion; CD, cell division; CM, carbohydrate metabolism; NM, nitrogen metabolism; PS, protein synthesis/assembly; PT, protein turnover; S, stress/defense; 

SP, storage protein; ST, signal transduction; T, transport; TT, transcription/translation; U, unknown. 

b 

E, early; M, middle; L, late stages. See text for time points included for each of the temperature regimens. 

c 

H, high and L, low abundance proteins. 

HSP16.9 (407) increased and triticin (199) decreased transiently 

during mid development. 

Because the proteome maps were developed for endosperm 

proteins isolated from grain produced under the moderate 

temperature regimen, spots listed in Table 2 were excised from 2-D 

gels of endosperm proteins extracted from grain produced under 

the high temperature regimens and re-identified by MS/MS. The 

original protein identifications were confirmed for all spots. The 

peptide sequence data, accession numbers, percent coverage, and 

expectation values are listed in Table S2. 

3.5. Effects of high temperature regimens on endosperm protein 

levels late in grain development 

The high temperature regimens altered the accumulation levels 

of a number of proteins in the mature grain. Proteins that increased 

or decreased 1.8-fold or more relative to the moderate temperature 

regimen are listed in Table 3. Under the high temperature regimen 

initiated at 10 dpa, 31 proteins increased, the majority of which 

functioned in stress/defense and storage. The stress/defense

Average fresh weight/kernel (mg) 

100 

80 

60 

40 

20 

0 

0 

10 20 30 40 50 

Days Post Anthesis (DPA) 

Fig. 2. Effect of high temperature on grain fresh weight. Wheat plants were grown 

under three day/night temperature regimens: -, 24 C/17 C; C, 37 C/28 C initiated 

at 10 dpa; and B, 37 C/28 C initiated at 20 dpa. 

proteins included a-amylase/subtilisin inhibitor (381), barwin/PR-4 

protein (427, 871), chitinase (282, 341, 351), late embryogenesis 

abundant (LEA) protein (551, 1001), and xylanase inhibitor protein 

(318). It should be noted that lipid-transfer protein (LTP, 633), 

which is classified as a transport protein, may also have a role in 

stress/defense (Jung et al., 2003; Wu et al., 2004). Nearly all of the 

storage proteins that increased in response to high temperature 

were isoforms of globulin-2 (147, 155, 559, 561, 851, 852) or globulin 

Beg 1 (281, 620, 1002, 1003). In addition, 32 proteins decreased 

Number of Proteins 

40 

30 

20 

10 

0 

30 

20 

10 

0 

30 

20 

10 

0 

Carb. Metab. 

N Metab. 

Pro. Syn./Assem. 

Pro. Turnover 

EARLY 

MID 

LATE 

Stress/Defense 

Function 

Storage Pro. 

Signal Trans. 

Transcrip./Transl. 

Fig. 3. Effect of high temperature during grain fill on the timing of biochemical 

processes in wheat endosperm. See text for specific time points included in the early 

(A), mid (B), and late (C) stages for each of the three temperature regimens. ,, 24 C/ 

17 C; ,37 C/28 C initiated at 10 dpa; and -, 37 C/28 C initiated at 20 dpa. 


A 

B 

C 

under this high temperature regimen and the majority of these 

function in carbohydrate metabolism, including starch biosynthesis, 

and protein synthesis. 

Under the high temperature regimen initiated at 20 dpa, only 10 

proteins increased and, like the regimen initiated at 10 dpa, the 

majority functioned in stress/defense and storage (Table 3). Eight of 

these proteins also increased during the high temperature regimen 

initiated at 10 dpa, but the increases in sucrose synthase type 2 

(550), aspartate amino acid transferase (236), and Cu–Zn superoxide 

dismutase (418) were specific for the regimen initiated at 

20 dpa. In addition, 15 proteins decreased, including 7 that also 

decreased under the high temperature regimen initiated at 10 dpa. 

However, decreases in 40S ribosomal protein S12 (422) and LEA 

protein (842) were specific for the 20 dpa regimen. Six proteins, 

including five globulins (147, 561, 620, 631, 852) and a 40S ribosomal 

protein S21 (444), that increased under the high temperature 

regimen initiated at 10 dpa, decreased under the 20 dpa regimen. 

Like the proteins that responded transiently to high temperature, 

proteins whose levels were altered late in development were also reidentified 

by MS/MS (Table S2). The original protein identifications 

were confirmed for all but spot 69. This spot was initially identified as 

a poly(A)-binding protein, but was re-identified as formate tetrahydrofolate 

ligase. This is likely due to slight differences in the 2-DE 

pattern in this region of the gel, illustrating the importance of confirming 

identifications of proteins in the high temperature proteomes. 

Nine of the re-identified spots contained peptides from one 

or two additional proteins (superscript a in Table 3). Because the 

proteins in these spots comigrate in the 2-D gels, it is not possible to 

know which proteins are responding to high temperature. Nonetheless, 

these identifications are valuable because they provide 

additional candidates for heat responsive endosperm proteins. 

3.6. Protein responses to high temperature regimens initiated at 

anthesis and 15 dpa 

Accumulation profiles of albumins and globulins were analyzed 

in independent experiments in which grain was produced under 

a24 C/17 C regimen and one of three additional high temperature 

regimens: 37 C/17 C initiated at anthesis, 37 C/28 C initiated 

at anthesis, or 37 C/28 C initiated at 15 dpa. Cronbach’s 

alpha was computed for the control data from the last time point 

in these three experiments and the original experiment where 

heat was applied from 10 and 20 dpa. For this analysis, proteins 

that were absent at the last time point in one or more experiments 

were deleted, resulting in a data set of 163 proteins for 

which the calculated Cronbach’s alpha was 0.96 out of a maximum 

of 1.0. This high level of consistency among the controls justified 

a comparison of the data obtained for the four high temperature 

experiments. At the last time point for each protein, the ratio of 

the normalized volume under the high temperature regimens to 

the moderate temperature regimen was calculated. Proteins that 

were absent at the last time point for one or more experiments 

were deleted, resulting in a data set containing 159 proteins, and 

the ratios were used to assign ranks to the proteins. Cronbach’s 

alpha was computed for these ranks at the last time point for the 

four high temperature experiments. The calculated Cronbach’s 

alpha of 0.60 showed a high degree of association among the four 

experiments. Individual correlations based on Kendall’s Tau rank 

correlation method showed that the highest correlation was 

between the 37 C/28 C regimens initiated at anthesis and 10 dpa 

(0.31 at p < 0.0001). The 37 C/28 C regimen initiated at 10 dpa 

was also highly correlated with the 37 C/17 C regimen initiated 

at anthesis (0.18 at p < 0.0007), but less correlated with the 37 C/ 

28 C regimens initiated at 15 (0.12 at p < 0.022) and 20 dpa (0.19 

at p < 0.001). The 37 C/28 C regimens initiated at 15 and 20 dpa 

showed a high correlation with each other (0.22, significant at

20 

Table 2 

Proteins that undergo transient changes in accumulation levels in response to high temperatures during grain fill 

Spot no. Name Function a 

Order no. b 

Cluster no. Increase c 

Decrease 

10 dpa 20 dpa 10 dpa 20 dpa 

128 ADP-glucose PPase, LS CM 116 5 M 

130 ADP0-glucose PPase, LS CM 118 5 M 

132 ADP-glucose PPase, LS CM 117 5 M 

199 Triticin SP 134 8 M 

225 Heat shock protein 70 PS 146 11 E 

285 Legumin-like protein SP 52 1 E 

327 Seed globulin SP 131 9 M 




354 Triosephosphate isomerase CM 13 1 E 

407 Heat shock protein 16.9C PS 145 10 M M 

471 DNAK-type molecular chaperone HSP70 PT 57 1 E E 

1009 Peroxidase 1 S 133 7 M 

Identities of these proteins contained in these spots were confirmed by mass spectrometry. Accession numbers, percent coverage, expectation values, and peptide sequence 

data are listed in Table S2. XML files that contain the mass spectrometry data are available online (see Section 2.3). 

a 

Abbreviations for functional groups are: CM, carbohydrate metabolism; PS, protein synthesis/assembly; PT, protein turnover; S, stress/defense; SP, storage protein. 

b 

Order and cluster numbers were determined by hierarchal clustering. See Fig. 1 for accumulation profiles. 

c 

E, transient change early in development; M, transient change in mid development. See text for specific time points. 

p < 0.0001), lesser correlation with regimens initiated at 10 dpa as 

described above and even less with the 37 C/28 C regimen 

initiated at anthesis (0.10 and 0.16 at p < 0.056 and 0.004). Regimens 

initiated at 15 and 20 dpa were not related to the 37 C/ 

17 C regimen initiated at anthesis. 

Like the 37 C/28 C regimen initiated at 10 dpa, a relatively 

large number of proteins responded to the high temperature regimens 

initiated at anthesis (Table 3). During the 37 C/28 C regimen 

28 proteins increased and 29 decreased and during the 37 C/17 C 

regimen 24 proteins increased and 22 decreased. A comparison of 

the changes accompanying the high temperature regimens initiated 

at anthesis and 10 dpa revealed that 18 proteins increased and 

17 decreased under all three regimens. The majority of the proteins 

that increased functioned in storage and stress defense. Storage 

proteins included two globulin Beg 1 isoforms (281, 620), seven 

globulin-2 isoforms (147, 155, 559, 851, 852, 561, 631), and a seed 

globulin (355). The stress defense proteins were barwin/PR-4 

protein 4 (427), chitinase (351), LEA (551), and LTP (633). The 

majority of proteins that decreased functioned in carbohydrate 

metabolism and stress defense. Proteins with roles in carbohydrate 

metabolism were ADP-glucose PPase (130, 132), aldehyde dehydrogenase 

(154), and pyruvate Pi dikinase (31). Stress defense 

proteins were dehydroascorbate reductase (382), peroxidase 

(1009), purple acid phosphatase (312), and serpin (231). Like the 

37 C/28 C regimen initiated at 20 dpa, relatively few proteins 

responded when this regimen was initiated at 15 dpa (Table 3). Of 

the 11 proteins that increased and 17 that decreased under the 

regimen, only 4 proteins increased and 6 decreased under the 

37 C/28 C regimens initiated at 15 and 20 dpa. The proteins that 

increased have roles in stress/defense (341, chitinase; 427, barwin/ 

PR-4 protein 4; 633, LTP) and storage (1002, globulin Beg 1). The 

proteins that decreased are involved in carbohydrate metabolism 

(193, 6-phosphogluconate dehydrogenase; 310, glucose and ribitol 

dehydrogenase), storage (631 and 852, globulin-2), protein 

synthesis (361, 20S proteosome a-subunit), and stress/defense 

(312, purple acid phosphatase). 

4. Discussion 

4.1. Endosperm development under the moderate 

temperature regimen 

The protein accumulation profiles developed in this study 

provide a dynamic picture of biological events that occur during 

wheat grain fill under a moderate temperature regimen. During 


early grain fill, endosperm proteins have roles in seven functional 

processes: carbohydrate metabolism, nitrogen metabolism, protein 

synthesis/assembly, protein turnover, stress/defense, storage, 

signal transport, and transcription/translation. Based on the 

number of proteins, carbohydrate metabolism, protein synthesis/ 

assembly, stress/defense, and transcription/translation are the 

principal functions at this stage of development. During mid grain 

fill, the number of stress/defense and storage proteins remains at 

earlier levels, but the number of proteins functioning in carbohydrate 

metabolism, nitrogen metabolism, protein synthesis/ 

assembly, protein turnover, and transcription/translation 

decreases. Late in grain fill, the number of proteins that function in 

stress/defense and storage increase and proteins involved in 

nitrogen metabolism, signal transduction, and transcription/ 

translation decrease to undetectable levels. This developmental 

program ensures that the endosperm synthesizes the requisite 

reserves for germination and seedling growth as well as proteins 

that protect these reserves from abiotic stresses and biotic invasion 

during grain desiccation and dormancy. Transcript analysis using 

micro arrays (Laudencia-Chingcuanco et al., 2007) and serial analysis 

of gene expression (SAGE) (McIntosh et al., 2007) revealed 

a similar sequence of events during wheat grain development. 

A more detailed picture of grain development emerges when 

the levels of specific proteins within the functional categories are 

compared. Early in grain fill, proteins involved in carbohydrate 

metabolism include those that function in glycolysis (glyceraldehyde-3-P 

dehydrogenase, phosphoglycerate kinase, phosphoglycerate 

mutase, enolase, pyruvate Pi dikinase), the citric acid cycle 

(malate dehydrogenase, aconitase), and starch biosynthesis (ADPand 

UDP-glucose PPase, phosphoglucomutase). During mid grain 

fill, enzymes functioning in glycolysis and the citric acid cycle were 

proportionately less abundant. ADP glucose PPase is present at 

lower levels, but UDP-glucose PPase and phosphoglucomutase 

decrease to undetectable levels. These decreases coincide with the 

decline in starch accumulation that accompanies this stage of grain 

development. Late in grain fill, proteins involved in carbohydrate 

metabolism function in starch (sucrose synthase) and glucose (bamylase, 

glucose and ribitol dehydrogenase) degradation rather 

than starch biosynthesis. Isoforms of glyceraldehyde-3-P dehydrogenase 

are again present. Since the wheat grain is undergoing 

desiccation at this developmental stage, this result is consistent 

with an earlier report that dehydration strongly increases the level 

of glyceraldehyde-3-P dehydrogenase in leaves of the resurrection 

plant, Craterostigma plantagineum, a species capable of withstanding 

severe desiccation (Velasco et al., 1994). However,


Table 3 

Proteins that increased or decreased 1.8-fold or more in grain produced under high temperature regimens initiated at anthesis, 10, 15, or 20 dpa 

Spot Protein name Function Order Cluster 37 C/28 C 37 C/28 C 37 C/17 C 37 C/28 C 

no. 

no. no. 

Anthesis Anthesis 15 dpa 

10 dpa 20 dpa 

420 a 

Nucleoside diphosphate kinase A 164 15 2.09 B b 

275 a 

Glyceraldehyde 3-P dehydrogenase (NAD) CM 162 15 5.15 2.25 C C 

591 Glucose and ribitol dehydrogenase CM 161 15 2.79 C C 

110 2-Isopropylmalate synthase NM 91 5 2.57 B B 

444 40S Ribosomal protein S21 PS 30 1 2.58 2.33 , , 

869 60S Ribosomal protein L12 PS 180 28 1.81 C 

163 Heat shock associated protein PS 49 1 1.83 1.83 C 

347 20S Proteasome a-subunit [proteasome subunit a type 6] PT 163 15 2.74 2.52 C 

381 a-Amylase/subtilisin inhibitor S 104 5 1.87 C 

871 Barwin [pathogenesis-related protein 4] S 186 33 8.29 - B C 

351 Chitinase-a [chitinase-c] S 150 13 2.39 C C C 

282 Chitinase-a [chitinase-c] S 151 13 2.21 C C 

341 Chitinase-c S 149 13 2.54 1.85 C C 

551 Late embryogenesis abundant protein-like S 155 13 3.00 C C 

427 Pathogenesis-related protein 4 [Barwin] S 156 13 4.22 1.82 C C C 

1001 Unknown protein [similar to late embryogenesis abundant 

protein] 

S 153 13 1.92 C C 

318 Xylanase inhibitor protein I S 157 14 1.86 C C 

1002 Globulin Beg 1 SP 189 36 8.33 1.83 C 

1003 Globulin Beg 1 SP 190 36 5.41 - B C 

281 Globulin Beg 1 [embryo globulin] SP 165 15 8.46 C C C 

620 Globulin Beg 1 [embryo globulin] SP 169 18 3.83 2.59 C C 

147 Globulin-2 SP 166 16 1.85 2.75 C C 

155 Globulin-2 SP 168 18 4.08 C C 

559 Globulin-2 SP 183 31 3.35 C C C 

851 Globulin-2 SP 184 31 2.92 2.38 - C 

852 Globulin-2 SP 185 32 3.86 - C , 

561 Globulin-2 SP 187 34 3.78 - C 

631 Globulin-like protein [globulin-2] SP 170 19 2.07 6.43 - C B 

355 a 

Seed globulin SP 160 15 3.23 C C 

633 Nonspecific lipid-transfer protein T 159 14 2.53 2.54 C C C 

557 a 

Embryo-specific protein U 154 13 2.41 - C 

193 6-Phosphogluconate dehydrogenase CM 61 2 3.74 , , 

132 ADP-glucose PPase, LS CM 117 5 2.20 B B B 

130 ADP-glucose PPase, LS CM 118 5 2.63 B B B 

154 a 

Aldehyde dehydrogenase CM 11 1 2.47 B , 

1004 Aldolase CM 45 1 2.26 

310 Glucose and ribitol dehydrogenase CM 173 21 4.92 C C B 

31 Pyruvate Pi dikinase CM 70 2 3.34 B B 

158 UDP-glucose PPase CM 32 1 1.99 B B 

72 10-Formyltetrahydrofolate synthetase [formate tetrahydrofolate NM 

ligase] 

16 1 4.36 B B 

98 Acetohydroxyacid synthase NM 37 1 2.52 B , 

138 a 

Ketol-reductoisomerase NM 10 1 3.25 B B B 

362 Elongation factor 1-b PS 76 3 7.30 , B B 

33 Elongation factor 2 PS 114 5 2.85 B B 

149 Heat shock associated protein PS 51 1 2.79 B B 

225 Heat shock protein 70 PS 146 11 2.01 

65 Heat shock protein 70 [DNAK-type molecular chaperone HSC70] PS 19 1 5.20 , B B 

118 Protein disulfide isomerase PS 8 1 2.08 B 

361 20S Proteasome a-subunit PT 74 3 4.83 1.81 , , B 

205 26S Proteasome PT 92 5 1.93 , C 

382 a 

Dehydroascorbate reductase S 89 5 2.09 B B 

1009 Peroxidase S 133 7 5.66 B , 

312 a 

Purple acid phosphatase S 124 5 5.34 2.85 B B B 

246 Serpin S 112 5 2.27 

639 a 

Serpin S 144 9 2.98 2.14 B 

231 Serpin WZS2 S 111 5 1.92 B B 

441 Avenin N9 SP 143 9 1.85 1.95 , - 

285 Legumin-like protein SP 52 1 2.41 

330 14-3-3 Protein ST 83 3 2.50 , B 

301 Guanine nucleotide-binding protein ST 28 1 3.97 B B B 

60 Poly(A)-binding protein TT 68 2 2.80 1.87 , , 

69 Poly(A)-binding protein [formate tetrahydrofolate ligase] TT 15 1 2.88 B 

550 Sucrose synthase type 2 CM 158 14 2.03 

236 Aspartate amino transferase NM 110 5 1.85 

422 40S Ribosomal protein S12 PS 34 1 4.35 , B 

418 Superoxide dismutase [Cu–Zn] S 126 5 2.32 C B 

842 Late embryogenesis abundant protein-like S 152 13 2.11 C C 

Identities of the proteins in these spots were confirmed by MS. Accession numbers, percent coverage, expectation values, and peptide sequence data are listed in Table S2. 

Brackets denote current database names for proteins identified in Vensel et al. (2005). Spot 69, previoulsy identified as poly(A)-binding protein, was identified as formate 

tetrahydrofolate ligase in this study. XML files that contain the mass spectrometry data are available online (see Section 2.3). 

a When reanalyzed by MS/MS, these spots contained peptides from one or two additional proteins, see Table S2. 

b Symbol key: B, decrease 1.8-fold or more; C, increase 1.8-fold or more; ,, absent under high temperature regimen; -, absent under moderate temperature regimen.

22 

a relationship between increased levels of this enzyme and drought 

tolerance has not been established. 

The major wheat storage proteins, the gliadins and glutenins, 

comprise more than 80% of endosperm protein and are found in the 

KCl-insoluble fraction. However, the KCl-soluble fraction contains 

a number of proteins with best matches to a variety of storage 

proteins, including avenin, triticin, legumin-like protein, seed 

globulin/19 kDa globulin, embryo globulin, globulin-2, or globulin 

Beg1. Avenin is a protein in oats that is similar to wheat gliadin. 

Triticin is a protein in wheat that is similar to pea legumin, an 11S 

globulin. Globulin-2 and globulin Beg1 are so named because they 

are similar to 7S globulins present in maize and barley, respectively. 

Like the gliadins and glutenins (DuPont et al., 2006a,b), the 

majority of these non-gluten storage proteins are most abundant 

late in grain fill. Eighteen globulins (one isoform each of avenin and 

globulin-like protein; two isoforms of gobulin Beg1, three of 

embryo globulins, six of globulin-2, five of seed globulin) were 

detected during late development while only two globulins (one 

isoform each of avenin and legumin-like protein) were detected 

during early development and five (three isoforms of avenin and 

one each of globulin-2 and triticin) during mid development. The 

complement of globulin storage proteins is very complex and 

warrants further analysis, particularly since some of these proteins 

increase in response to high temperature. 

A range of stress/defense proteins is present throughout endosperm 

development, a finding in accord with transcript profile 

analyses (McIntosh et al., 2007). A number of these proteins protect 

cells from reactive oxygen species (ROS) that are produced under 

normal growth conditions and can cause oxidative damage to 

proteins, DNA, and lipids. Thus, ascorbate peroxidase, peroxidase and 

GSH-dependent dehydroascorbate reductase accumulate early in 

grain fill. SGT1, a component of R-gene triggered disease resistance, 

and serpin, a serine protease inhibitor, are also present and may 

protect the developing grain against various pathogens. During 

mid grain fill, ROS-scavenging enzymes include dehydroascorbate 

reductase, peroxidase, catalase, and glyoxalase. Glyoxalase protects 

cells against reactive 2-oxoaldehydes generated by the triosephosphate 

isomerase reaction in glycolysis. Additional stress/ 

defense proteins present at this stage of development include the 

a-amylase inhibitors and bifunctional a-amylase/trypsin inhibitors, 

which guard against digestive enzymes of insects and fungi. The 

levels of these inhibitors are actually much higher than those 

reported in this study, because many of these proteins partition into 

the KCl-soluble/methanol-soluble fraction (Wong et al., 2004). Late 

in grain fill, ROS-scavenging enzymes include peroxidase and 

superoxide dismutase. Pathogen resistance proteins present at this 

stage include serpin, chitinase, which hydrolyzes the structural 

carbohydrate of fungal cell walls, barwin/PR-4 protein, which is 

induced by fungal pathogens and binds chitin, and xylanase inhibitor 

protein, which inhibits a fungal enzyme that degrades plant cell 

walls. As the developing grain matures it undergoes desiccation, an 

event accompanied by the accumulation of LEA proteins, proteins 

known to increase in response to drought stress. 

4.2. Effect of high temperature on endosperm development 

The duration of grain fill is shortened by high temperature. Grain 

maturity is achieved earlier and is accompanied by corresponding 

decreases in kernel weight. These observations are in agreement 

with earlier studies inwhich maximum fresh weight, dry weight, and 

protein and starch content were lower in grain produced under high 

temperature regimens (37 C/17 Cor37 C/28 C) imposed from 

anthesis (Altenbach et al., 2002, 2003; DuPont et al., 2006b; Hurkman 

et al., 2003). Comparison of protein profiles revealed that 

although the high temperature regimens compressed the developmental 

program, they did not substantially alter it. Analysis of protein 


profiles across physiologically equivalent stages under the three 

temperature regimens showed similar shifts in the timing of cellular 

functions during the transition from early to late development. The 

results in this study suggest that specific protein responses depend 

on when the high temperature regimens are imposed during 

development. Fewer proteins responded when the high temperature 

regimens were initiated during mid development (15 or 20 dpa) than 

during early development (anthesis or 10 dpa). 

Transient increases and decreases in protein levels were 

observed following transfer of plants to the high temperature 

regimens. These transient changes occurred over a period of days 

rather than hours as is typically observed in transcript studies. 

Under the regimen initiated at 10 dpa, HSP16.9, HSP70, peroxidase, 

triosephosphate isomerase, and globulins (legumin-like protein, 

seed globulin), increased transiently. The responses of HSP16.9 and 

HSP70 were also observed in previous studies on the effect of high 

temperature on wheat grain composition (Majoul et al., 2003, 

2004; Skylas et al., 2002). Since ROS production is elevated during 

abiotic and biotic stresses (reviewed, Suzuki and Mittler, 2006), an 

increase in peroxidase would be expected. The increase in triosephosphate 

isomerase agrees with findings that transcript 

(Minhas and Grover, 1999) and protein levels (Yan et al., 2005) for 

this enzyme increase in rice seedlings exposed to high temperature. 

The transient increase in globulins is interesting because it suggests 

that these proteins may have functional roles other than storage. 

One isoform of ADP-glucose PPase decreased transiently under the 

high temperature regimen initiated at 10 dpa, but several isoforms 

of the large subunit of this enzyme decreased and remained at low 

levels following initiation of this high temperature regimen. These 

results are in keeping with the decrease in transcript levels for this 

enzyme and the reduction of starch accumulation that accompanies 

high temperatures during grain fill (Altenbach et al., 2003; 

Hurkman et al., 2003). Under the high temperature regimen initiated 

at 20 dpa, the only protein that increased transiently was an 

isoform of HSP16.9. This finding agrees with the results of Skylas 

et al. (2002), where a number of HSP16.9 isoforms that increased 

during heat shock were absent in the mature grain. 

Among the proteins that responded to the high temperature 

regimens, seven were identified previously (Wong et al., 2003) as 

potential targets of thioredoxin, a widely distributed small disulfide 

protein that functions in redox regulation. These targets function in 

stress/defense (a-amylase/subtilisin inhibitor 381, serpin 231, 246, 

639, and dehydroascorbate reductase 382), storage (globulin Beg1 

620), and carbohydrate metabolism (pyruvate Pidikinase 31). A 

previous study showed that a number of potential thioredoxin 

targets responded to drought stress, including serpin and dehydroascorbate 

reductase (Hajheidari et al., 2007). 

4.3. High temperature and flour quality 

A striking feature of the high temperature regimens is the 

increase in the levels of stress/defense and globulin storage proteins 

in the endosperm of the mature grain. The stress/defense proteins 

include those known to respond to desiccation (glyceraldehyde-3-P 

dehydrogenase, LEA protein) and pathogen invasion (chitinase, 

xylanase inhibitor protein, a-amylase/subtilisin inhibitor, barwin/ 

PR-4 protein, LTP, polyubiquitin, 20S proteosome a-subunit). Altenbach 

et al. (2007a,b) also reported that expression of transcripts for 

two of these genes, LTP and the PR-4 protein wheatwin, was 

enhanced under the 37 C/28 C regimen. The protein changes that 

occur late in grain development are of particular interest, because 

they may impact the functional properties of the resultant flour. 

Some of the proteins that increase in response to high temperature 

are present in dough liquor, a soluble fraction of wheat dough, or 

foams prepared from dough liquor (Salt et al., 2005) and may have 

roles in gas bubble stabilization in dough and crumb structure of

ead. These include LTP, a-amylase inhibitor, xylanase inhibitor, 

chitinase, serpin, HSP70, DNAK-type chaperone, peroxidase, glyceraldehyde-3-P 

dehydrogenase, nucleoside diphosphate kinase, and 

globulin-2. Some of these proteins, notably a-amylase inhibitor 

(Posch et al., 1995; Weiss et al., 1997), glyceraldehyde-3-P dehydrogenase, 

triosephosphate dehydrogenase, and serpin (Sander 

et al., 2001), also react with sera from patients with Baker’s asthma. 

LTP has been shown to react with sera from patients with immunoglobulin 

E-mediated food allergies, including wheat-dependent 

atopic dermatitis (Battais et al., 2005). In addition, many of the KClsoluble 

globulins and stress/defense proteins have structural similarities 

to food allergens (Jenkins et al., 2005). Further studies are 

warranted on the possible roles of these proteins in flour quality and 

their allergenic potential. 


The authors thank Drs. F.M. DuPont and R. Thilmony for critical 

reading of the manuscript. 

Appendix. Supplementary material 

Supplementary data associated with this article can be found in 

the online version, at doi:10.1016/j.jcs.2008.06.014. 

References 

Altenbach, S.B., DuPont, F.M., Kothari, K.M., Chan, R., Johnson, E.L., Lieu, D., 2003. 

Temperature, water and fertilizer influence the timing of key events during 

grain development in a US spring wheat. Journal of Cereal Science 37, 9–20. 

Altenbach, S.B., Kothari, K.M., 2004. Transcript profiles of genes expressed in 

endosperm tissue are altered by high temperature during wheat grain development. 

Cereal Science 40, 115–126. 

Altenbach, S.B., Kothari, K.M., Lieu, D., 2002. Environmental conditions during 

wheat grain development alter temporal regulation of major gluten protein 

genes. Cereal Chemistry 79, 279–285. 

Altenbach, S.B., Kothari, K.M., Tanaka, C.K., Hurkman, W.J., 2007a. Expression of 

9 kDa lipid transfer protein genes in developing wheat grains is enhanced by 

high temperature but not by post-anthesis fertilizer. Journal of Cereal Science, 

doi:10.1016/j.jcs.2007.03.010. 

Altenbach, S.B., Kothari, K.M., Tanaka, C.K., Hurkman, W.J., 2007b. Genes encoding 

the PR-4 protein wheatwin are developmentally regulated in wheat grains and 

respond to high temperature during grainfill. Plant Science 173, 135–143. 

Battais, F., Courcoux, P., Poplineau, Y., Kanny, G., Moneret-Vautrin, D.A., Denery- 

Papini, S., 2005. Food allergy to wheat: differences in immunoglobulin Ebinding 

proteins as a function of age and symptoms. Journal of Cereal Science 

42, 109–117. 

Craig, R., Cortens, J.P., Beavis, R.C., 2004. An open source system for analyzing, 

validating and storing protein identification data. Journal of Proteome Research 

3, 1234–1242. 

DuPont, F.M., Hurkman, W.J., Vensel, W.H., Chan, R., Lopez, R., Tanaka, C.K., 

Altenbach, S.B., 2006a. Differential accumulation of sulfur-rich and sulfur-poor 

wheat flour proteins is affected by temperature and mineral nutrition during 

grain development. Journal of Cereal Science 44, 101–112. 

DuPont, F.M., Hurkman, W.J., Vensel, W.H., Tanaka, C.K., Kothari, K.M., Chung, O.K., 

Altenbach, S.B., 2006b. Protein accumulation and composition in wheat grains: 

effects of mineral nutrients and high temperature. European Journal of 

Agronomy 25, 96–107. 

Hajheidari, M., Elivazi, A., Buchanan, B.B., Wong, J.H., Majidi, I., Salekdeh, G.H., 2007. 

Proteomics uncovers a role for redox in drought tolerance in wheat. Journal of 

Proteome Research 6, 1451–1460. 


Hurkman, W.J., McCue, K.F., Altenbach, S.B., Korn, A., Tanaka, C.K., Kothari, K.M., 

Johnson, E.L., Bechtel, D.B., Wilson, J.D., Anderson, O.D., DuPont, F.M., 2003. 

Effect of temperature on expression of genes encoding enzymes for starch 

biosynthesis in developing wheat endosperm. Plant Science 164, 873–881. 

Hurkman, W.J., Tanaka, C.K., 2004. A simplified method for the isolation of wheat 

endosperm proteins for proteome analysis. Journal of Cereal Science 40, 

294–299. 

Jenkins, J.A., Griffiths-Jones, S., Shewry, P.R., Breiteneder, H., et al., 2005. Structural 

relatedness of plant food allergens with specific reference to cross-reactive 

allergens: an in silico analysis. Journal of Allergy and Clinical Immunology 115, 

163–170. 

Jung, H.W., Kim, W., Hwang, B.K., 2003. Three pathogen-inducible genes 

encoding lipid transfer protein from pepper are differentially activated by 

pathogens, abiotic, and environmental stresses. Plant, Cell, and Environment 

26, 915–928. 

Laudencia-Chingcuanco, D.L., Stamova, B.S., You, F.M., Lazo, G.R., Beckles, D.M., 

Anderson, O.D., 2007. Transcriptional profiling of wheat caryopsis development 

using cDNA microarrays. Plant Moleular Biology 63, 651–668. 

Majoul, T., Bancel, E., Triboi, E., Ben Hamida, J., Branlard, G., 2003. Proteomic analysis 

of the effect of heat stress on hexaploid wheat grain: characterization of heatresponsive 

proteins from total endosperm. Proteomics 3, 175–183. 

Majoul, T., Bancel, E., Triboi, E., Ben Hamida, J., Branlard, G., 2004. Proteomic analysis 

of heat stress on hexaploid wheat grain: characterization of heat-responsive 

proteins from non-prolamins fraction. Proteomics 3, 505–513. 

McIntosh, S., Watson, L., Bundock, P., Crawford, A., White, J., Cordeiro, G., Barbary, D., 

Rooke, L., Henry, R., 2007. SAGE of the developing wheat caryopsis. Plant 

Biotechnology Journal 5, 69–83. 

Minhas, D., Grover, A., 1999. Transcript levels of genes encoding various glycolytic 

and fermentation enzymes change in response to abiotic stresses. Plant Science 

146, 41–51. 

Posch, A., Weill, W., Wheeler, C., Dunn, M.J., Görg, A., 1995. Sequence analysis of 

wheat grain allergens separated by two-dimensional electrophoresis with 

immobilized pH gradients. Electrophoresis 16, 1115–1119. 

Salt, L.J., Robertson, J.A., Jenkins, J.A., Mulholland, J., Clare Mills, E.N., 2005. The 

identification of foam-forming soluble proteins from wheat (Triticum aestivum) 

dough. Proteomics 5, 1612–1623. 

Sander, I., Flagge, A., Merget, R., Halder, T.M., Meyer, H.E., Baur, X., 2001. Identification 

of wheat flour allergens by means of 2-dimensional immunoblotting. 

Journal of Allergy and Clinical Immunology 107, 907–913. 

SAS Institute Inc., 2004. SAS OnlineDoc Ò 9.1.3. SAS Institute Inc., Cary, NC. 

Skylas, D.J., Cordwell, S.J., Hains, P.G., Larsen, M.R., Basseal, D.J., Walsh, B.J., 

Blumenthal, C., Rathmell, W., Copeland, L., Wrigley, C.W., 2002. Heat shock of 

wheat during grain filling: proteins associated with heat-intolerance. Journal of 


Suzuki, N., Mittler, R., 2006. Reactive oxygen species and temperature stresses: 

a delicate balance between signaling and destruction. Physiologia Plantarum 

126, 45–51. 

Velasco, R., Salamini, F., Bartels, D., 1994. Dehydration and ABA increase mRNA 

levels and enzyme activity of cytosolic GAPDH in the resurrection plant Craterostigma 

plantagineum. Plant Molecular Biology 26, 541–546. 

Vensel, W.H., Tanaka, C.K., Cai, N., Wong, J.H., Buchanan, B.B., Hurkman, W.J., 2005. 

Developmental changes in the metabolic protein profiles of wheat endosperm. 

Proteomics 5, 1594–1611. 

Weiss, W., Huber, G., Engel, K.-H., Pethran, A., Dunn, M.J., Gooley, A.A., Görg, A., 1997. 

Identification and characterization of wheat grain albumin/globulin allergens. 

Electrophoresis 18, 826–833. 

Wong, J.H., Balmer, Y., Cai, N., Tanaka, C.K., Vensel, W.H., Hurkman, W.J., 

Buchanan, B.B., 2003. Unraveling thioredoxin-linked metabolic processes of 

cereal starchy endosperm using proteomics. FEBS Letters 547, 151–156. 

Wong, J.H., Cai, N., Tanaka, C.K., Vensel, W.H., Hurkman, W.J., Buchanan, B.B., 2004. 

Thioredoxin reduction alters the solubility of proteins of wheat starch 

endosperm: an early event in cereal germination. Plant Cell Physiology 45, 

407–415. 

Wu, G., Robertson, A.J., Liu, X., Zheng, P., Wilen, R.W., Nesbitt, N.T., Gusta, L.V., 2004. 

A lipid transfer protein gene BG-14 is differentially regulated by abiotic stress. 

Journal of Plant Physiology 161, 449–458. 

Yan, S., Tang, Z., Su, W., Sun, W., 2005. Proteomic analysis of salt stress-responsive 

proteins in rice root. Proteomics 5, 235–244. 

Zadoks, J.C., Chang, T.T., Konzak, C.F., 1974. A decimal code for the growth stages of 

cereals. Weed Research 14, 415–421.

Accumulation of mixed linkage (1 / 3) (1 / 4)-b-D-glucan during grain 

filling in barley: A vibrational spectroscopy study 

Helene Fast Seefeldt a,b , Andreas Blennow c , Birthe Møller Jespersen b , Bernd Wollenweber a , 

Søren Balling Engelsen b, * 

a University of Aarhus, Faculty of Agricultural Sciences, Department of Genetics and Biotechnology, Forsøgsvej 1, DK-4200 Slagelse, Denmark 

b University of Copenhagen, Faculty of Life Sciences, Department of Food Science, Quality & Technology, DK-1958 Frederiksberg C, Denmark 

c University of Copenhagen, Faculty of Life Sciences, Department of Plant Biology and Biotechnology, VKR research centre Pro-Active Plants, 

DK-1871 Frederiksberg C, Denmark 

article info 


Received 20 November 2007 



Keywords: 

Infrared 

IR 

Near-infrared 

NIR 

(1 / 3) (1 / 4)-b-D-glucan 

Grain filling 


abstract 

In the quest for optimising cereals for soluble fibres and other 

health-promoting components, there is great interest in studying 

the metabolic changes during grain filling, such as cell wall fibre 

development in barley mutants differing in mixed linkage (1 / 3) 

(1 / 4)-b-D-glucan (BG) using high throughput methods. Inexpensive, 

spectroscopic methods based on vibrational spectroscopy 

offer the possibility of fast and flexible analysis of a large number of 

genotypes (Osborne, 2006). Near-infrared (NIR) and infrared (IR) 

spectroscopy measure the vibrations of molecular covalent bonds. 

The NIR region 14,300–4000 cm 1 (780–2500 nm) mainly gives 

Abbreviations: ADP-glucose, adenosine 5 0 diphosphate glucose; ATR, attenuated 

total reflection; BG, Mixed linkage (1 / 3) (1 / 4)-b-D-glucan; DAF, days after 

flowering; EISC, extended inverted signal correction; FT-IR, Fourier transform 

infrared; iPLS, interval partial least squares regression; MSC, multiplicative scatter 

correction; NIR, near-infrared reflectance; PC, principal component; PCA, principal 

component analysis; PLS, partial least squares regression; RMSECV, root mean 

square error of cross validation; SECV, standard error of cross validation; SEE, 

standard error of estimate; VIS, visual part of the NIR spectra. 


E-mail address: se@life.ku.dk (S.B. Engelsen). 


doi:10.1016/j.jcs.2008.06.012 





The accumulation of mixed linkage barley (1 / 3) (1 / 4)-b-D-glucan (BG) during grain filling at eight 

stages was studied using standard reference methods and infrared spectroscopy. Two mutant barley 

genotypes having higher (starch mutant lys5f) and lower (high lysine mutant lys3a) BG content than the 

normal control Cork were studied. The Cork and lys3a genotypes showed a linear BG accumulation 

throughout the grain filling to reach a maximum of approximately 6 and 4% BG (w/w) dry matter, 

respectively. However, lys5f mutant exhibited an exponential increase in BG synthesis to a maximum of 

approximately 18% BG (w/w) dry matter 30 days after flowering (DAF), seemingly compensating for 

a decreased synthesis of starch. 

The spectral information of the barley flour was compared to pure BG spectra and partial least squares 

regression (PLS) models were constructed for calibration to BG content. Informative regions in the nearinfrared 

(NIR) and the infrared (IR) spectra were identified for separation of temporal and genetic 

differences. Interval PLS yielded good calibration models to BG (R 2 ¼ 0.94 for NIR in the region 

1194–1240 nm, whereas the global PLS gave correlations with BG with R 2 ¼ 0.92 for IR). 


information about overtones and combination tones (stretching 

and bending) involving anharmonic bonds primarily to hydrogen, 

whereas the IR region 4000–200 cm 1 (2500–50,000 nm) gives 

information about the fundamental vibrations. NIR spectroscopy is 

already well established for at/on-line quality control in the food 

and food ingredients industries (Zachariassen et al., 2005) and is 

able to provide information about chemical parameters such as 

water, protein and starch content as well as about physical 

parameters such as particle size and temperature. In cereals, NIR 

transmittance spectra of single seeds have proven very informative 

for the determination of different quality traits such as protein and 

fat content with high accuracy (Delwiche 1995; Pedersen et al., 

2002). In fact, a new high capacity single-seed TriQ NIR sorting 

system (Bomill AB, Lund, Sweden) (Munck, 2007) has been developed 

utilising NIR spectra to diversify heterogeneous bulk lots of 

wheat with regard to multivariate complex quality traits such as 

dough performance and baking value (Munck, 2007; Tønning, 

2008). Furthermore, NIR can be used to evaluate quality traits such 

as the fibre fraction of cereal cell walls (Blakeney and Flinn, 2005) 

and genetics (Jacobsen et al., 2005; Munck et al., 2004). 

While NIR spectroscopy has primarily been used in quantitative 

analysis of bulk components (in spite of its documented ability to 

predict complex qualitative traits like baking and malting quality

(Munck, 2007)), IR spectroscopy has mainly been used to study 

well-defined components such as plant cell wall polysaccharides 

(Chen et al., 1998; Kacurakova and Wilson, 2001; McCann et al., 

1992; Robert et al., 2005; Séné et al., 1994). Recently, micro Fourier 

transformed infrared spectroscopy (FT-IR) was used to study the 

deposition of cell wall polysaccharides in wheat endosperm during 

grain development with emphasis on BG and arabinoxylans (Philippe 

et al., 2006). Also, NIR spectroscopy has proven valuable in 

monitoring plant physiological processes such as carbohydrate 

accumulation during grain filling (Gergely and Salgo, 2005). NIR 

screenings of normal barley and mutants have resulted in the 

discovery of genotypes with a strongly increased content of soluble 

BG fibres (Munck et al., 2004). The high BG barley mutant line lys5f 

is a structural (enzyme functional) low starch mutant (Munck et al., 

2004; Munck and Møller, 2005) unable to transport ADP–glucose 

across the plastid envelope due to an inactive ADP–glucose transporter 

(Patron et al., 2004). Lys5f is thus disabled in efficient 

synthesis of starch, but compensates via an extremely high content 

(approximately 17% d.m.) of soluble and insoluble BG (Munck et al., 

2004). The lys3a mutant is a regulatory mutant (Jacobsen et al., 

2005) that primarily inhibits the synthesis of the hordein, but 

increases the synthesis of soluble proteins. It has approximately 2– 

3% BG compared to 3–4% in normal barley like Cork when grown in 

green houses (Munck et al., 2004). 

The aim of this study was thus to characterise and investigate 

two extreme recessive barley mutants with respect to BG and 

starch content during grain filling using classical reference methods 

as well as spectroscopic fingerprinting methods. The carbohydrate 

mutant lys5f with high BG content and the protein lys3a with low 

BG content were compared to the variety ‘Cork’ that has a normal 

content of BG. 

2. Experimental 


Three genotypes of barley were included in the study: a malt 

barley (Hordeum vulgare cv. Cork) and a barley mutant lys3a with 

alterations in the lys3 locus on chromosome five, tightly linked to 

adjacent BG synthesis suppressing genes (Munck et al., 2004). The 

mutant lys5f has a mutation in chromosome six. A ‘semifield’ pot 

experiment (72 pots, 16.5 cm diameter, 13 cm height) was carried 

out from April to August 2005 at the University of Aarhus, Research 

Centre Flakkebjerg, Denmark. Each pot was filled with 10 l of 

sphagnum with 15% Perlite (Perlite, Denmark) added. Ten seeds of 

each genotype were sown on April 20th and thinned to three 

seedlings per pot on May 31st. All pots were drip-watered 

throughout the experiment and standard pest control was performed 

against mildew and aphids when needed. Flowering was 

judged visually when 50% of the spikes showed clear pollen release, 

which occurred between the 26th and 27th of June. Spikes were 

harvested at eight time points during grain filling: 9, 13, 16, 20, 23, 

30, 39 and 47 days after flowering (DAF). 

2.2. Plant analysis 

The spike on the main tiller and the first spike of the side tillers 

were cut and immediately frozen in liquid nitrogen. After freezing, 

the kernels were detached from the spike, counted and weighed. 

Kernels were transferred to 80 C and freeze-dried within 3 weeks 

after harvest. One sample consisted of the seeds from two spikes 

from the same individual plant. The freeze-dried grains were milled 

(0.5 mm, Cyclotec 1093, Foss Tecator AB, Högenas, Sweden). A total 

of 91 samples were analysed (Cork, 31; lys5f, 31; and lys3a, 29). The 

total sample set consisted of three genotypes at eight temporal 

harvest points– two replicate spectra of each genotype from 

H.F. Seefeldt et al. / Journal of Cereal Science 49 (2009) 24–31 25 

harvests 1–4 and six replicate spectra of each genotype from 

harvests 5–8 were analysed. This experimental design yielded 

a total of 96 samples (3 genotypes 4 (1–4 harvests) 2 replicates 

þ 3 genotypes 4 (5–8 harvests) 6 replicates). Due to 

a minor flaw in the experiment, three replicates of lys3a from 

harvests 1, 5 and 7, and two replicates of Cork harvest 5 and lys5f 

harvest 5 were lost. The ground material was stored in sealed 

plastic bottles at room temperature until analysis. 

2.3. Chemical analysis 

The content of soluble BG was analysed by fluorimetry (Calcoflour 

reagent type II, Scandinavian Brewery Laboratory, Frederiksberg, 

Denmark) (Munck et al., 1989). The high BG content flours 

showed major deviations in values when determined by Calcoflour, 

and hence the values from the last four time points in lys5f were 

double-checked with an enzymatic kit (Megazyme, Wicklow, Ireland) 

specific for mixed linkage BG. 

The total content of starch was analysed as follows: freeze-dried 

flour (10 mg) was washed three times with 1-ml aliquots of 80% 

ethanol in screw-cap Eppendorf tubes. Starch in the washed flour 

was gelatinised by the addition of KOH (400 ml, 0.2 M) and incubation 

at 95 C for 1 h. After cooling, 140 ml of 1 M acetic acid was 

added, the samples mixed and diluted 20-fold with water. The 

diluted sample (10 ml) was mixed with an equal volume of amyloglucosidase 

(10 U/ml, Fluka) solubilised in 50 mM Na acetate pH 5.0 

and starch was hydrolysed by incubation at 37 C for 2 h. A 210-ml 

aliquot of 50 mM Mops/KOH pH 7.3, 5 mM MgCl2, 1 mM EDTA, 

1 mM ATP, 1 mM NAD and 4 U/ml hexokinase was added and 

absorbance at 340 nm was registered. One microlitre of 500 U/ml 

glucose-6-phosphate dehydrogenase was added and the production 

of NADH was followed until steady from the absorbance at 

340 nm. Starch content in the original flour was calculated using 

glucose as standard. Amylose in the flour was determined by iodine 

complexation, as described by Bay-Smidt et al. (1999). The amylopectin 

chain length distribution of the starch was analysed by high 

performance anionic exchange chromatography with pulsed 

amperometric detection (HPAEC-PAD) as described by Blennow 

et al. (1998). 

FT-IR and NIR spectra were obtained from the pure substances: 

Cellulose (9004d34-6, Sigma-Aldrich Chemie, Steinheim, Germany), 

BG (Barley– Medium Viscosity, Megazyme, Wicklow, Ireland) 

and wheat starch with normal and high content of amylose 

(Ritmo, Sejet Plantbreeding, Horsens, Denmark). The starches were 

purified according to Blennow et al. (1998). 

2.4. FT-IR measurements 

All FT-IR spectra were acquired at room temperature using the 

Arid-Zone MB100 FT-IR spectrometer (ABB Bomen Inc., Quebec, PQ, 

Canada). The sampling was performed using an Attenuated Total 

Reflection (ATR) device with a diamond crystal (ZnSe, TR-plate, ARK 

0055-603, Spectra-Tech Inc., CT, USA) operating in the range 4000– 

750 cm 1 . Measurements were obtained using 64 scans at 4 cm 1 

resolution. Background scans were obtained using 128 scans. The 

scans were averaged and ratioed against a single-beam spectrum of 

the clean ATR crystal and converted into absorbance units. 

2.5. NIR measurements 

A near-infrared instrument (Foss NIRSystems 6500, USA) 

was used in reflectance mode in the range of 400–2500 nm. All 

measurements were acquired at room temperature using a 

commercial ’’small sample cup’’ containing approximately 0.5 g of 

flour. The spectra were recorded in 2-nm steps using a spinning 

sample module. Each sample was measured using 32 scans which

26 

were ratioed against 16 reference scans using a ceramic sample. The 

results were averaged as a log 1/R spectrum. 

2.6. Data analysis 

The spectra were analysed using the chemometric software 

LatentiX 1.0 (http://www.latentix.com, Latent5, Copenhagen, Denmark) 

both for visual inspection of the spectra and calculation of 

principal component analysis (PCA) (Wold et al., 1987) and partial 

least squares (PLS) regressions. The PCA was performed to visualise 

systematic spectral variation. Calibrations and predictions of BG 

and amylose based on the spectral information were made using 

PLS. Interval partial least squares regression (iPLS) (Nørgaard et al., 

2000) was performed in order to reveal the most important spectral 

interval correlated with the calibration parameters. Prior to 

analysis, the spectral data was scatter-corrected using the extended 

inverted signal correction (EISC) (Martens et al., 2003; Pedersen 

et al., 2002) and all models are mean-centred. The iPLS was carried 

out using the iTOOLBOX (http://www.models.life.ku.dk) as 

a routine for MATLAB 6.0 (The Mathworks Inc., Natick, USA). All 

reported models are based on 30 intervals and are validated using 

full cross validation. 

3. Results and discussion 

The seeds showed normal fresh weight and water content 

during development (Fig. 1a,b) during grain filling (Gergely and 

Salgo, 2003; Jennings and Morton, 1962): Three distinct stages of 

the grain filling process could be determined by analysis of the 

fresh grain weight (Fig. 1a). Until 20 DAF a rapid increase in grain 

weight occurred, followed by a lag phase with stable grain weight. 

After 30 DAF the grain weight decreased due to the drying and 

maturation of the seed. The first phase is characterised by a rapid 

influx of water and cell enlargement and continues until 10–15 DAF. 

During the second phase the seeds take up nutrients and start to 

synthesise starch and protein (Jenner et al., 1991). In the third phase 

the fresh weight decreases due to loss of water as indicated in 

Fig. 1b. 

3.1. Accumulation of BG during grain filling – chemical analysis 

Fig. 1c shows the development of BG synthesis between genotypes 

during grain filling. The analysis of the BG content of the three 

barley genotypes (Fig. 1c) revealed that lys5f from 16 to 30 DAF 

exhibited a very strong increase in BG accumulation to reach an 

extreme maximum BG content of approximately 18% at 30 DAF. 

During the same period lys3a and Cork exhibited slow linear 

accumulation of BG, which continued throughout the grain filling 

period until maturity at 47 DAF. The mutant lys3a had lower BG 

content as compared to both Cork and lys5f, reaching only 

approximately 4% at harvest. Cork displayed a normal content of 

BG, reaching a maximum level of approximately 6% BG at maturity. 

During grain filling, Cork and lys3a did not start to accumulate BG 

until 16–20 DAF, whereas lys5f already started to accumulate BG 

between 9 and 13 DAF (Fig. 1c). 

In a previous study of three normal barley varieties (Coles, 

1979), a large increase in accumulation of BG starting at 19 DAF was 

found. In wheat, the cell walls mainly consist of arabinoxylans, 

whereas BG constitutes only 25%, but BG was accumulating from 

early grain filling until 10 DAF after which arabinoxylans were 

dominating (Philippe et al., 2006). These observations led to the 

hypothesis that BG could function as a structural element of 

growing cell walls as well as storage material that could be 

hydrolysed during germination of the grain (Philippe et al., 2006). 

The starch synthesis in Cork and lys3a (Fig. 1d) followed the 

same pattern having a rapid increase from 9 DAF until 20 DAF and 

H.F. Seefeldt et al. / Journal of Cereal Science 49 (2009) 24–31 

reaching a maximum of approximately 47% starch remaining at this 

level until 39 DAF. It should be noted that the levels of grain starch 

at maturation in our study were somewhat lower than generally 

found in barley grain as a result of a decrease in starch content of 

approximately 10% the last 17 DAF. The mutant lys5f displayed 

a much slower increase throughout the entire grain filling period 

until 39 DAF to reach a maximum at 23% starch. A low final level of 

starch in lys5f was previously reported (Munck et al., 2004). 

The high content of BG in lys5f can hence be interpreted as a new 

mechanism for compensating starch synthesis with BG synthesis 

(Munck et al., 2004). Interestingly, the total accumulation of starch 

and BG reveals three similar patterns for the genotypes with just an 

offset in total accumulation. Thus, the rapid accumulation rate of 

BG in lys5f corresponds largely to the rapid accumulation of starch 

in the non-starch genotypes lys3a and Cork. This confirms the 

observation of the starch and BG levels found in mature seed 

described by (Munck et al., 2004). The higher starch content found 

in Cork was followed by a high dry matter weight (Fig. 1e). The lys5f 

mutant had a markedly lower ratio of amylose/amylopectin (Fig. 1f) 

which is interesting from a starch functionality viewpoint (Tester 

et al., 2004). The chain length of amylopectin did, however, not 

differ between the three genotypes and it did not differ markedly 

over the developmental stages (data not shown). 

3.2. Near-infrared analysis of the barley grain filling 

The EISC pre-processed VIS/NIR spectra (400–2498 nm) of the 

91 barley flour samples are displayed in Fig. 2a. The first three 

harvests (9–16 DAF) represented by the light blue colours clearly 

differ in their NIR spectral profiles from the rest of the harvests (20– 

47 DAF, shown as dark red colours). The first three harvests are less 

intensive in the first overtone region from 1400–1850 nm and the 

O-H combination tone centred at 1940 nm. In general, the temporal 

differences influenced the offset in the spectra, while the genetic 

differences influenced the shape of the spectra. In the visible part of 

the spectrum, the grain filling is clearly evidenced by the gradual 

elimination of the chlorophyll peak at 672 nm (Fig. 2b), which 

reflects that the barley grains were visibly green until 30 DAF (the 

sixth harvest). Fig. 2c displays a score plot from a PCA of the 91 NIR 

spectra of the barley flour, explaining 92% of the variance. In the 

score plot, the temporal variation is separated into four welldefined 

groups. The first three harvests (9–16 DAF) are markedly 

different from the later harvests and located in the upper left corner 

of the PCA. The fourth and the fifth harvests (20 and 23 DAF), in 

which the BG is also primarily synthesised (Fig. 1c), are clustered 

with the lowest scores in the score plot, the sixth harvest (30 DAF) 

located in the lower right part of the spectra and the last two 

harvests (39–47 DAF) in the upper right corner of the PCA. By 

inspection of the loadings, the primary reason for the clear separation 

of the temporal grain fillings was found to be the chlorophyll 

peak at 672 nm (Table 1). As observed in Fig. 2c, the actual time of 

flowering is difficult to determine– in the figure a few samples 

reside in their previous group (marked with stars in Fig. 2c), 

presumably due to inaccuracy in the DAF determination. 

Munck et al. (2004) found that the combination tone region 

2280–2360 nm contains unique information about the lys3 

mutants and Szczodrak et al. (1992) found that two wavelengths in 

the region between 2260 and 2380 nm correlate the best with BG. 

Fig. 2d shows that the NIR spectra of lys5f have a peak in the NIR 

spectra around 2345 nm (the second bold arrow). This peak is also 

present in the spectra of lys3a, but as a shoulder, whereas the 

spectra of Cork only have a very weak shoulder. On the other hand, 

the spectra of Cork have a characteristic shoulder at 2280 nm, 

indicating starch (first bold arrow in Fig. 3d, Table 1), according to 

Munck (2007). This shoulder is absent in the mutants lys5f and 

lys3a. As Cork and lys3a have almost identical starch contents, it

mg seed-1 

% dm 

mg seed-1 

110.0 

95.0 

80.0 

65.0 

50.0 

35.0 

22.0 

20.0 

18.0 

16.0 

14.0 

12.0 

10.0 

8.0 

6.0 

4.0 

2.0 

0.0 

60.0 

50.0 

40.0 

30.0 

20.0 

10.0 

0.0 

A 80.0 

B 

C 

9 13 16 20 23 30 39 47 

Fresh weight 

9 13 16 20 23 30 39 47 

Beta-glucan 

9 13 16 20 23 30 39 47 

Dry matter 

could be questioned whether this peak corresponds to starch only. 

When comparing the average barley flour NIR spectra with the NIR 

spectra of pure cellulose, starch, BG and amylose (Fig. 3a), it is 

evident that a high degree of similarity exists with the starch 

spectrum except for the region 2200–2400 nm where the spectra 

show very distinct profiles including genotype specific patterns. 

From the lines indicated in Fig. 2d it can be observed that the 

three genotypes exhibit similar spectral properties during the first 

harvest (9 DAF) as well as during the second and third harvests. In 

order to avoid the dominance of the chlorophyll information, the 

visual part of the spectra was discarded and the NIR region 1100– 

2498 nm was examined separately (data not shown). The three 

genotypes are clustered together until 16 DAF. From the fourth 

harvest (20 DAF) and onwards, the three genotypes are separated 

into three distinct clusters. This is in excellent agreement with the 


% 

% dm 

60.0 

40.0 

20.0 

60.0 

45.0 

30.0 

15.0 

0.0 

40.0 

E F 

% starch 

35.0 

30.0 

25.0 

20.0 

15.0 

10.0 

D 

9 13 16 20 23 30 39 47 

Water content 

9 13 16 20 23 30 39 47 

Starch 

9 13 16 20 23 30 39 47 

Amylose 

Fig. 1. (A) Fresh seed weight (mg) (B) Water content (%). (C) BG content (% d.m.). (D) Starch content (% d.m.). (E) Seed dry weight (mg). (F) Amylose (% in starch). All values are 

denoted in relation to DAF (days after flowering). Solid, Cork; dotted, lys5f; broken, lys3a. 

rapid increase in the BG synthesis from 16 DAF to 23 DAF for lys5f 

(Fig. 1c) and the concomitant rapid increase in starch seen for lys3a 

and Cork. The loadings associated with the PCA reveal that the first 

PC is mainly spanned by information from moisture and starch (O- 

H stretching vibrations, 1440 and 1940 nm) (Table 1), which is in 

good accordance with the separation of grain filling time points 

along PC1. The third PC separates the genotypes into a Cork/lys3a 

cluster and a lys5f cluster. The corresponding loadings are starchrelated 

peaks (Munck et al., 2004; Szczodrak et al., 1992), which 

manifest the different syntheses in the three genotypes. 

3.3. Infrared investigation of barley grain filling 

The EISC-treated FT-IR spectra (1900–750 cm 1 ) of the 91 barley 

flour samples are shown in Fig. 4a. As was the case for the NIR

28 

Log 1/R 

Scores PC#2 (14.%) 

0.55 

0.5 

0.45 

0.4 

0.35 

0.3 

0.25 

0.2 

0.15 

0.10 

0.05 

0.00 

-0.05 

-0.10 

-0.15 

-0.20 

A 

400 800 1200 1600 

nm 

2000 2400 

1 2 3 4 5 6 7 8 

Levels of Harvest 

C 

1 

2 

2 

2 2 

1 

1 

232 

3 

3 

3 

78 

8 

87 

7 7 

8 

8878 

87 8 

7 

8 

7 

85 8 

87 

7 

7788 

87 

6 

6 

6 

6 

7 6 6 

6 

6 6 6 

6 6 

66 

6 

5 

4 5 5 5 5 

4 4 5 

5 

5 4 65 

4 5 5 

-0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 

Scores PC#1 (78.%) 

spectra, a clear temporal dependence is observed in the spectra 

(data not shown), but the difference between genotypes appears 

more pronounced than in the NIR spectra. 

The barley flour spectra represent typical broad absorption 

bands of polysaccharides in the region 1200–750 cm 1 (Kacurakova 

and Wilson, 2001) with a maximum absorption around 

1021 cm 1 . The glycosidic absorption band at 1160 cm 1 reported 

by Robert et al. (2005) and at 1150 cm 1 (Philippe et al., 2006) is 

found at 1152 cm 1 in our samples. Fig. 4c (raw spectra) shows the 

enlargement of the maximum absorption region in which a clear 

1 


Log 1/R 

0.36 

0.34 

0.32 

0.30 

0.28 

0.26 

0.24 

1log/R 

B 

D 

580 620 660 

nm 

700 740 

9 DAF 

2290 2310 2330 2350 

nm 

Fig. 2. The complete NIR spectra (400–2498 nm) of the three genotypes and eight harvest times. (A) The full, EISC-treated spectra. Squares indicate enlargements shown in B and D. 

(B) Magnification of the chlorophyll peak (672 nm). (C) PCA model of the full NIR region. Blue, first three harvests; Purple, fourth and fifth harvests; cerise, sixth harvest; pink, 

seventh and eighth harvests. (D) The region 2280–2360 nm. Open arrows indicate spectral features of interest. Black arrow indicates spectra of 9 DAF. 

Cork 

lys5f 

lys3a 

difference between the three genotypes is observed. At 1002 cm 1 

the high BG mutant lys5f has a different intensity compared to Cork 

and lys3a. Moreover, lys5f has a lower intensity of the shoulder 

around 1070 cm 1 , whereas lys3a and Cork have well-defined 

peaks at 1078 cm 1 . With regard to the temporal changes, a shift 

from 1036 cm 1 for 9 DAF towards 1021 cm 1 for the end of the 

grain filling is observed in the spectra (Fig. 4c). The anomer specific 

peaks in the range from 800 to 950 cm 1 contain information about 

BG as well as other polysaccharide cell wall components (Philippe 

et al., 2006). Fig. 4d displays an enlargement of this region as 

Table 1 

The most informative peaks and their putative alignments as determined from loading plots corresponding to PCAs made on the defined regions 

Region 

NIR method 

PC Peak 1 Peak 2 Peak 3 Peak 4 Alignment 

500–750 1 672 þ400 þ1440 þ1930 672: chlorophyll a, 1440: C–H, O–H in starch,1940 H2O 

500–750 2 672 400 þ500 þ560 

1100–2498 1 þ1444 1856 þ1930 1670 1440 O–H starch, H2O, 1940 H2O 

1100–2498 3 þ2250 2477 þ2355 þ2384 2252, 2461 O–H starch, 2353 C–H cellulose, 2380: R–OH 

2280–2360 1 2308 2348 2310 CH2, 2347 HC¼CHCH2 

2280–2360 2 þ2288 þ2314 2360 2280: CH3, 2310 C–H CH2, 2352: C–H cellulose 

IR method 

1900–750 1 þ1010 960 900–990: aromatic, 1000–1260: C–O alcohol, 

1900–750 2 1049 þ1147 þ1002 þ1094 1002 ring stretching (arabinoxylans a ), 1160–1210:C–C(O)–C ester 

The peaks with the highest intensity are listed first. þ indicates a positive loading, while indicates a negative loading. Units for region: NIR, nm; IR, cm 1 . 

a From (Philippe et al., 2006). All other alignments from (Osborne et al., 1993).

logR/1 

0.45 

0.4 

0.35 

0.3 

0.25 

0.2 

0.15 

0.1 

A B 

Average NIR flour spectra 

Betaglucan 

Cellulose 

Wheat starch 

Wheat amylose 

1400 1600 1800 2000 2200 2400 

nm 

Absorbance 

0.16 

0.14 

0.12 

0.1 

0.08 

0.06 

0.04 

0.02 

BG 

peak 

Average IR flour spectra 

Betaglucan 

Cellulose 

Wheat starch 

Wheat amylose 

800 900 1000 1100 1200 1300 1400 1500 1600 1700 

cm-1 Fig. 3. IR and NIR spectra of pure seed carbohydrates compared to an averaged IR and NIR flour spectrum. (A) IR spectra. The arrow indicates the BG and cellulose specific peak. (B) 

NIR spectra. 

Absorbance Absorbance 

0.11 

0.1 

0.09 

0.08 

0.07 

0.06 

0.05 

0.04 

0.03 

0.02 

0.01 

0.11 

0.10 

0.09 

0.08 

0.07 

0.06 

0.05 

A Cork 

0.10 B 

lys5f 

0.08 

lys3a 

0.06 

800 1000 1200 1400 1600 1800 

C 

9-13 DAF 

cm -1 

Cork 

lys5f 

lys3a 

cm-1 980 1000 1020 1040 1060 1080 1100 1120 


Scores PC#2 (1.4 %) 

0.04 

0.02 

0 

-0.02 

-0.04 

-0.06 

2.derivative 

7 

4 

8 

8 628 2 

67 

7 78 

6 5 

53 

8 

6 

3 3 

78 

6 4 7 

356 74 

7 786 

5 5 

8 

878 

4 

86 7 

54 

6 5 

6 

5 5 

7 

78 

7 

8 

78 2 

66 8 

665 

56 

5 3 

63 

2 

24 

-0.36 -0.34 -0.32 -0.3 -0.28 0.26 -0.24 -0.22 -0.2 

D 

Cork 

lys5f 

lys3a 

Scores PC#1 (97 %) 

57 

1 

1 

Cork 

+ lys5f 

lys3a 

cm-1 820 840 860 880 900 920 940 

Fig. 4. FT-IR spectra of the region 750–1900 cm 1 . (A) EISC-treated spectra. Squares are indicated magnified in C and D. (B) PCA of the second derivative of the spectra (not meancentred). 

(C) EISC-pre-processed spectra of the region of maximum absorption. Arrows indicate 9 and 13 DAF. (D) The second-derivative spectra of the region from 800 to 940 cm 1 .

30 

RMSECV 

6 

5 

4 

3 

2 

1 

314568775617645654625615486531 

1098 1290 1482 1672 1856 2040 2224 2408 

Wavelength(nm) 

second-derivative spectra which reveal that lys5f has less intensive 

peaks compared to Cork and lys3a. Cork and lys3a display a relatively 

strong a-anomer band at 855 cm 1 , whereas the high BG lys5f 

and the early stages of grain filling show a much weaker a-glucan 

peak. This is in good accordance with the starch levels in Cork and 

lys3a compared to lys5f. In the case of the b-anomer sensitive peak 

normally present at approximately 890 cm 1 (Fig. 4d), we observed 

higher background intensity in lys5f than in the other two genotypes. 

From approximately 23 DAF, the intensity is the highest 

corresponding to the highest level of BG. Pure barley BG shows 

a peak at 895 cm 1 , not present in the average grain spectra (see 

Fig. 3b). 

A PCA was applied to the full second-derivative IR spectra 

(Fig. 4b) and the first PC explains 97% of the variance. Both the first 

and the second PC separate the genotypes. The corresponding 

loading plot (Table 1) revealed that PC1 is mainly spanned by the 

peak at 1010 cm 1 and PC2 at 1049 cm 1 , which are the dominating 

peaks of starch in the second-derivative spectra of the pure 

substances (Fig. 3b). Thus BG seems to play a less pivotal role in 

discriminating the genotypes compared to starch. 

3.4. Relationships between the chemical and spectroscopic methods 

In order to study the relationships between the spectral data 

and grain filling, a range of calibration models were investigated. 

Global PLS models based on the entire data set were used, as well as 

iPLS models, where sub-regions are tested against the global 

model. Only a few studies have attempted to construct an NIR 

calibration to BG content. The quality of the predictions has varied 

greatly. A correlation coefficient of 0.69 and an RMSE of 0.557% was 

achieved in the BG range 5.8–8.4% (Czuchajowska et al., 1992). In 

another study, an 1-VR value that is a determination for cross 

validation and measure of goodness of fit was found to be 0.92 and 

an SECV of 0.45 in the BG range 0.09–5.12% (Blakeney and Flinn, 

2005). One major problem has been the limited BG range for the 

calibrations. However, Szczodrak et al. (1992) succeeded in developing 

calibration models to BG with a correlation coefficient of 

0.871 and an SEE (standard error of estimate) of 0.677 using the 

calibration range 2.7–9.5%. 

In this study, the BG variation (calibration range) was further 

expanded in order to establish a quantitative model between the 

NIR spectra and the BG content. The global NIR model predicts BG 

with a squared correlation coefficient of 0.84 and an RMSECV of 

2.60 using two PLS components with three outliers removed (lys5f 

from 9 DAF and two from 30 DAF). However, application of iPLS 


A B 

Predicted (BG%) 

20 

15 

10 

5 

0 

r = 0.97 

RMSECV = 1.57 

revealed that the region 1194–1240 nm yielded an R 2 of 0.94 and 

RMSECV of 1.6% using four PLS components (Fig. 5a,b). This subregion 

covers second overtones of the C-H stretching and shows 

that BG can be predicted well in this region. 

PLS calibrations to the FT-IR region and BG were also studied. 

The global FT-IR models predicted BG slightly better than the global 

NIR model with a squared correlation coefficient of 0.92 and 

a RMSECV value of 1.83 using six PLS components and removing 

two outliers (lys5f from 9 DAF and from 30 DAF) from the model. 

Subsequently an iPLS was applied to the IR spectra but no intervals 

were able to improve the calibration performance when compared 

to the global model. It thus appear that the information regarding 

BG is hidden across the IR spectra from 1900 to 750 cm 1 , and not 

confined to the beta specific peak at approx 890 cm 1 , whereas the 

BG information in the NIR spectra can with advantage be correlated 

with the region of C-H stretching at 1194–1240 nm. 

Correlations with the amylose content were predicted with 

a squared correlation coefficient of 0.81 and an RMSECV of 2.46 for 

the FT-IR and 0.77 and an RMSECV of 2.71 for NIR. The poorer 

correlations with amylose probably reflect the uncertainty of the 

amylose determinations more than the ability of the spectra to 

predict amylose or starch content. 

4. Conclusions 

0 5 10 15 20 

Measured (BG%) 

Fig. 5. RMSECV correlation coefficients of iPLS models of NIR and FT-IR spectra. (A) The iPLS performed on 30 intervals superimposed on the NIR spectra. Many intervals are better 

calibrators (smaller RMSECV values) for BG than the global model using two PLS components. Numbers on bars indicate PLS components used. The red marked interval 1194– 

1240 nm is the best calibrator. (B) Measured versus predicted plot of the model in the interval 1194–1240 nm and with three outliers expelled: two replicates of lys5f, 30 DAF and 

one lys5f at 9 DAF. 

A rapid increase in grain fresh weight during grain filling is 

concomitant with the increase of starch and BG. The high BG 

mutant lys5f shows an exponential increase of BG until 23 DAF and 

reaches a final content of 18%. This is in contrast to the low BG 

mutant lys3a and the conventional malting barley Cork having 

linear phases of BG deposition and reaching maxima of only 4 and 

6%, respectively. Starch deposition in these latter genotypes was 

exponential, reaching maxima of 47% compared to a linear increase 

in lys5f leading to a maximum of 23% starch. 

The NIR and FT-IR spectra are complete chemical fingerprints of 

the grain and thus more complex to resolve than single chemical 

analysis of grains. Data extraction relies on the use of chemometrics 

to reveal hidden information. In the case of NIR, the genetic variation 

was mainly resolved from the spectral region between 2260 

and 2380 nm. Moreover, the NIR spectra showed mostly temporal 

differences primarily due to complex changes in moisture. In the 

NIR region 2260–2380 nm, all the barley flour spectra differed and 

the average grain spectrum was influenced by both the starch/ 

amylose and the BG contents, which underlines the importance of 

this NIR region for breeding purposes.

Both NIR and IR spectra gave good correlations with BG. The NIR 

sub-region 1194–1240 nm gave the best correlation to BG with an 

R 2 of 0.94 which is promising for using NIR to predict BG in 

heterogeneous samples such as seeds and flour with a wide span in 

BG content. 

Fast and non-destructive spectroscopic fingerprinting is obviously 

advantageous for studying physiological processes such as 

grain filling, as the techniques are non-destructive, fast and sensitive 

and it is possible to do real-time analysis. 


The DIAS Competence Fund and the Ministry of Food Agriculture 

and Fisheries are greatly acknowledged for financial support to the 

project (FFS05-9: Build Your Food). Betina Sørensen (University of 

Aarhus) and Lisbeth T. Hansen and Louise Nancke (Faculty of Life 

Sciences, University of Copenhagen) are acknowledged for technical 

support in semi-field experiments and chemical analysis. 

Gilda Kischinovsky is acknowledged for proofreading. The authors 

wish to thank Lars Munck for valuable comments and for sharing 

his unique barley mutant collection. 

References 

Bay-Smidt, A.M., Blennow, A., Bojko, M., Møller, B.L., 1999. The effect of phosphate 

and amylopectin molecular size on peak viscosity of starch pastes. Annual 

Transactions of the Nordic Rheology Society 7, 31–38. 

Blakeney, A.B., Flinn, P.C., 2005. Determination of non-starch polysaccharides in 

cereal grains with near-infrared reflectance spectroscopy. Molecular Nutrition & 

Food Research 49, 546–550. 

Blennow, A., Bay-Smidt, A.M., Wischmann, B., Olsen, C.E., Møller, B.L., 1998. The 

degree of starch phosphorylation is related to the chain length distribution of 

the neutral and the phosphorylated chains of amylopectin. Carbohydrate 

Research 307, 45–54. 

Chen, L., Carpita, N.C., Reiter, W.-D., Wilson, R.H., Jeffries, C., McCann, M.C., 1998. A 

rapid method to screen for cell-wall mutants using discriminant analysis of 

Fourier transform infrared spectra. The Plant Journal 16, 385–392. 

Coles, G., 1979. Relationship of mixed-link beta-glucan accumulation to accumulation 

of free sugars and other glucans in the developing barley endosperm. 

Carlsberg Research Communications 44, 439–453. 

Czuchajowska, Z., Szczodrak, J., Pomeranz, Y., 1992. Characterization and estimation 

of barley polysaccharides by near-infrared spectroscopy.1. Barleys, starches, and 

betaDglucans. Cereal Chemistry 69, 413–418. 

Delwiche, S.R., 1995. Single wheat kernel analysis by near-infrared transmittance: 

protein content. Cereal Chemistry 72, 11–16. 

Gergely, S., Salgo, A., 2003. Changes in moisture content during wheat maturation – 

what is measured by near infrared spectroscopy? Journal of Near Infrared 

Spectroscopy 11, 17–26. 

Gergely, S., Salgo, A., 2005. Changes in carbohydrate content during wheat maturation 

– what is measured by near infrared spectroscopy? Journal of Near 

Infrared Spectroscopy 13, 9–17. 

Jacobsen, S., Søndergaard, I., Møller, B., Desler, T., Munck, L., 2005. A chemometric 

evaluation of the underlying physical and chemical patterns that support near 

infrared spectroscopy of barley seeds as a tool for explorative classification of 

endoperm genes and gene combinations. Journal of Cereal Science 42, 281–299. 

Jenner, C.F., Ugalde, T.D., Aspinall, D., 1991. The physiology of starch and protein 

deposition in the endosperm of wheat. Australian Journal of Plant Physiology 

18, 211–226. 


Jennings, A.C., Morton, R.K., 1962. Changes in carbohydrate, protein and non-protein 

nitrogenous compounds of developing wheat grain. Australian Journal of Biological 

Sciences 16, 318–331. 

Kacurakova, M., Wilson, R.H., 2001. Developments in mid-infrared FT-IR spectroscopy 

of selected carbohydrates. Carbohydrate Polymers 44, 291–303. 

Martens, H., Nielsen, J.P., Engelsen, S.B., 2003. Light scattering and light absorbance 

separated by extended multiplicative signal correction. Application to nearinfrared 

transmission analysis of powder mixtures. Analytical Chemistry 75, 

394–404. 

McCann, M.C., Hammouri, M., Wilson, R., Belton, P., Roberts, K., 1992. Fourier 

transform infrared microspectroscopy is a new way to look at plant cell walls. 

Plant Physiology 100, 1940–1947. 

Munck, L., 2007. Breeding for Quality Traits in Cereals – A Revised Outlook on Old 

and New Tools for Integrated Breeding. Cereals. Springer Verlag, New York. 

Munck, L., Møller, B., 2005. Principal component analysis of near infrared spectra as 

a tool of endosperm mutant characterisation and in barley breeding for quality. 

Czech Journal of Genetics and Plant Breeding 41, 89–95. 

Munck, L., Jørgensen, K.G., Ruudhansen, J., Hansen, K.T., 1989. The EBC methods for 

determination of high molecular-weight beta-glucan in barley, malt, wort and 

beer. Journal of the Institute of Brewing 95, 79–82. 

Munck, L., Møller, B., Jacobsen, S., Søndergaard, I., 2004. Near infrared spectra 

indicate specific mutant endosperm genes and reveal a new mechanism for 

substituting starch with (1-3,1-4) betaglucan in barley. Journal of Cereal Science 

40, 213–222. 

Nørgaard, L., Saudland, A., Wagner, J., Nielsen, J.P., Munck, L., Engelsen, S.B., 2000. 

Interval partial least-squares regression (iPLS). A comparative chemometric 

study with an example from near-infrared spectroscopy. Applied Spectroscopy 

54, 413–419. 

Osborne, B.G., 2006. Applications of near infrared spectroscopy in quality screening 

of early-generation material in cereal breeding programmes. Journal of Infrared 

Spectroscopy 14, 93–101. 

Osborne, B.G., Fearn, T., Hindle, P.H., 1993. Theory of near infrared spectrophotometry. 

Practical NIR Spectroscopy with Applications in Food and Beverage 

Analysis, second ed. Longman Scientific and Technical, Harlow, UK, pp. 13-35. 

Patron, N.J., Greber, B., Fahy, B.F., Laurie, D.A., Parker, M.L., Denyer, K., 2004. The lys5 

mutations of barley reveal the nature and importance of plastidial ADP-Glc 

transporters for starch synthesis in cereal endosperm. Plant Physiology 135, 

2088–2097. 

Pedersen, D.K., Martens, H., Nielsen, J.P., Engelsen, S.B., 2002. Near-infrared 

absorption and scattering separated by extended inverted signal correction 

(EISC): analysis of near-infrared transmittance spectra of single wheat seeds. 

Applied Spectroscopy 56, 1206–1214. 

Philippe, S., Robert, P., Barron, C., Saulnier, L., Guillon, F., 2006. Deposition of cell 

wall polysaccharides in wheat endosperm during grain development: Fourier 

transform-infrared microspectroscopy study. Journal of Agricultural and Food 

Chemistry 54, 2303–2308. 

Robert, P., Marquis, M., Barron, C., Guillon, F., Saulnier, L., 2005. FT-IR investigation 

of cell wall polysaccharides from cereal grains. Arabinoxylan infrared assignment. 

Journal of Agricultural and Food Chemistry 53, 7014–7018. 

Séné, C.F.B., McCann, M.C., Wilson, R.H., Grinter, R., 1994. Fourier-transform raman 

and Fourier-transform infrared spectroscopy – an investigation of five higher 

plant cell walls and their components. Plant Physiology 106, 1623–1631. 

Szczodrak, J., Czuchajowska, Z., Pomeranz, Y., 1992. Characterization and estimation 

of barley polysaccharides by near-infrared spectroscopy. 2. Estimation of total 

betaDglucans. Cereal Chemistry 69, 419–423. 

Tester, R.F., Karkalas, J., Qi, X., 2004. Starch – composition, fine structure and 

architecture. Journal of Cereal Science 39, 151–165. 

Tønning, E., 2008. Wheat baking quality in a process analytical perspective. 

Sampling, diversification, prediction and chemometric method development. 

PhD thesis, University of Copenhagen. ISBN: 87-91949-23-8. 

Wold, S., Esbensen, K.H., Geladi, P., 1987. Principal component analysis. Chemometrics 

and Intelligent Laboratory Systems 2, 37–52. 

Zachariassen, C.B., Larsen, J., van den Berg, F., Engelsen, S.B., 2005. Use of NIR 

spectroscopy and chemometrics for on-line process monitoring of ammonia in 

low methoxylated amidated pectin production. Chemometrics and Intelligent 

Laboratory Systems 76, 149–161.

Mechanism of gas cell stabilization in bread making. I. The primary 

gluten–starch matrix q 

Baninder S. Sroan a, *, Scott R. Bean b , Finlay MacRitchie a 

a Department of Grain Science and Industry, Kansas State University, Manhattan, KS 66506, USA 

b USDA-ARS, Grain Marketing and Production Research Center, Manhattan, KS 66502, USA 

article info 




Accepted 8 July 2008 

Keywords: 

Bread making 

Gas cell stability 

Liquid lamella 

Gluten–starch matrix 

Dough rheology 


abstract 

During mixing, gas is occluded and concentrated in the liquid 

phase of dough in the form of small nuclei. No further occlusion of 

Abbreviations: MALLS, multiangle laser light scattering; MDDT, mixograph 

dough development time; MT, threshold molecular weight; MW, molecular weight; 

MW, weight average molecular weight; MWD, molecular weight distribution; Rmax, 

extensograph maximum resistance to extension; SDS, sodium dodecyl sulfate; SE- 

HPLC, size-exclusion high-performance liquid chromatography; UPP, unextractable 

polymeric protein. 

q Mention of firm names or trade products does not constitute endorsement by 

the U.S. Department of Agriculture over others not mentioned. 

* Corresponding author. Present address: FritoLay, 7701 Legacy Dr., 3T-124, Plano, 

TX 75024, USA. Tel.: þ1 972 334 4526; fax: þ1 972 334 2329. 

E-mail address: baninder@k-state.edu (B.S. Sroan). 


doi:10.1016/j.jcs.2008.07.003 





Expansion of dough and hence bread making performance is postulated to depend on a dual mechanism 

for stabilization of inflating gas bubbles. Two flours were used in this study, one from the wheat variety 

Jagger (Jagger) and the other from a composite of soft wheat varieties (Soft). Thin liquid lamellae (films), 

stabilized by adsorbed surface active compounds, act as an auxiliary to the primary gluten–starch matrix 

in stabilizing expanding gas cells and this mechanism operates when discontinuities begin to appear in 

the gluten–starch matrix during later proving and early baking stages. Contributions of the liquid 

lamellae stability to dough expansion were assessed using flours varying in their lipid content. Incremental 

addition of natural lipids back into defatted flour caused bread volume to decrease, and, after 

reaching a minimum, to increase. Strain hardening is a key rheological property responsible for stabilizing 

the primary gluten–starch matrix. Jagger gave higher test-bake loaf volume than Soft and higher 

strain hardening index for dough. The different lipid treatments were found to have negligible effects on 

strain hardening index. Image analysis of crumb grain revealed that differences in number of gas cells 

and average cell elongation with different lipid treatments were insignificant. The evidence agrees with 

a dual mechanism to stabilize the gas cells in bread dough. To understand dough rheology at a molecular 

level, rheological properties of doughs were varied by addition of flour protein fractions prepared by pH 

fractionation. Fractions were characterized by SE-HPLC and MALLS. The molecular weight distribution 

(MWD) of fractions progressively shifted to higher values as the pH of fractionation decreased. Mixograph 

dough development time paralleled the MWD. However, the strain hardening index and the testbake 

loaf volume increased with increasing MWD up to a point (optimum), after which they declined. At 

a given strain rate, the behavior at the optimum is thought to result from slippage of the maximum 

number of statistical segments between entanglements, without disrupting the entangled network of 

polymeric proteins. Shift of MWD to molecular weight higher than the optimum results in a stronger 

network with reduced slippage through entanglement nodes, whereas a shift to lower molecular weights 

will decrease the strength of the network due to a lesser number of entanglements per chain. 


gas occurs in succeeding stages (sheeting, molding, etc.) of the 

bread making process (Baker and Mize, 1946). However, these 

subsequent stages (punching, sheeting and molding) cause subdivision 

of already existing gas cells, thus improving their number 

and size distribution. Gas nuclei expand during fermentation due to 

release of fermentation gases, and during baking due to expansion 

of these gases as temperature increases. 

Wheat flour dough, due to its unique visco-elastic properties, is 

capable of stabilizing the expanding gas cells. The classical view has 

emphasized only the gluten–starch matrix as the cell membrane 

that stabilizes the expanding gas cells (Bloksma, 1990; Hoseney, 

1992). However, scanning electron micrographs of dough at the end 

of the proving stage appear to show the existence of intact gas cells 

even when discontinuities in the gluten–starch matrix begin to 

appear (Gan et al., 1990). Based on these electron micrographs of

the dough, Gan and co-workers (Gan et al., 1990, 1995) proposed 

that the expanding gas cells appear to be stabilized against coalescence 

and disproportionation by a primary gluten–starch matrix 

with a secondary liquid lamella on its inner side, enveloping the gas 

cell. The dual film hypothesis seems plausible in view of evidence 

from previous studies, which show presence of a continuous liquid 

phase in dough (MacRitchie, 1976a) and the effect of surface active 

components on bread loaf volume (DeStefanis and Ponte, 1976; 

MacRitchie, 1976a,b; MacRitchie and Gras, 1973). In such cases, 

where there is an independent mechanism of gas cell stabilization 

apart from the gluten–starch matrix, the dough rheology should 

not be affected by variations in natural lipid levels. MacRitchie and 

Gras (1973) reported that Alveograms are not affected by variations 

in the natural lipid fraction of the flour. A Stable Micro Systems 

dough inflation system mounted on a texture analyzer (TAXT 

2plus), used in this study to understand biaxial extensional 

rheology of wheat flour dough, works on a principle similar to the 

Alveograph, with the advantage of inflating bubbles at constant 

strain rates. The amounts of lipid in the study by MacRitchie and 

Gras (1973) were as low as 1–1.5% and had significant effects on 

baking performance. It is probably due to their surface activity that 

these compounds (lipids and proteins) are adsorbed at the gas– 

liquid interface of the liquid lamellae and affect stability of the gas 

cells (Mills et al., 2003; Paternotte et al., 1993; Ross and MacRitchie, 

1995). Analogous variations in foaming properties of the dough 

liquor, and bread loaf volume and crumb structure (MacRitchie, 

1976a) provide reasonable evidence for their action at the interface. 

Loaf volume can be defined as the extent of dough expansion 

(Gandikota and MacRitchie, 2005), which depends upon how thin 

the gluten–starch matrix can be stretched before reaching its 

expansion limit or point of rupture. Rheology of the gluten–starch 

matrix is important in the bread making process as this determines 

extensibility and strength. The proving and baking stages of bread 

making are characterized by fast biaxial expansion of gas cells, 

expanding at strain rates of 0.001–0.0001/s and 0.01–0.001/s, 

respectively (Dobraszczyk, 1997). During expansion, the gluten– 

starch matrix around gas cells expands biaxially to large strains 

(>100%) due to excess pressure produced in the gas cells by 

diffusion of carbon dioxide during proving, and by thermal 

expansion of gases during baking. This causes thinning of gas cell 

walls and, if a gas cell continues to expand along this thin region, it 

may rupture. However, if the stress in the thin region increases 

more than proportionally to strain, the thin region of a cell wall or 

(gluten–starch matrix) will resist further deformation and the gas 

cell will continue to expand along thicker parts of the cell wall. This 

localized increase of stress in response to strain, preventing failure 

of the gas cell walls, is called strain hardening (Dobraszczyk and 

Roberts, 1994; van Vliet et al., 1992), and is a necessary rheological 

property for obtaining good bread volume. 

A power law relation between stress and Hencky strain, as 

described by the equation below, was used by Dobraszczyk and 

Roberts (1994) to determine strain hardening. 

s [ K3 n 

where, s is the true stress, K is a power law constant, 3 is strain and 

n is a strain hardening index. The index n must be greater than 1 in 

order to have a curved relation between stress and Hencky strain 

(Fig. 1). Doughs with a strain hardening index of 1 or higher have 

the potential to give good loaf volume by allowing gas cells to 

expand without undergoing disproportionation and coalescence 

(Dobraszczyk and Roberts, 1994). In later studies, Dobraszczyk and 

co-workers (Dobraszczyk and Salmanowicz, 2008; Dobraszczyk 

et al., 2003) found that an exponential relationship shows better fit 

to data, particularly in the case of doughs for which bubbles inflate 

B.S. Sroan et al. / Journal of Cereal Science 49 (2009) 32–40 33 

Fig. 1. Stress vs. strain (Hencky) curves of Soft and Jagger wheat flours showing 

differences in their tendencies to strain harden. The Soft wheat flour with poor bread 

making potential fails at relatively lower strains. 

to strains greater than 2. Chin and Campbell (2005) also used an 

exponential relationship to study dough rheology by bubble inflation. 

Strain hardening index is a reliable criterion for differentiating 

flours based on bread making potential (Dobraszczyk and Roberts, 

1994; Dobraszczyk and Salmanowicz, 2008; Dobraszczyk et al., 

2003; Tronsmo et al., 2003;). This is illustrated in Fig. 1 for the two 

flours used in this study. 

At a molecular level, strain hardening can be explained by a well 

established polymer entanglement network theory (MacRitchie 

and Lafiandra, 1997; Singh and MacRitchie, 2001). Strain hardening 

is believed to originate from entanglement coupling of large glutenin 

molecules (Singh and MacRitchie, 2001), with molecular 

weight (MW) greater than a threshold MW (MT), MT being the 

minimum molecular weight at which stable entanglements are 

formed (Bersted and Anderson, 1990). Various studies substantiate 

this theory. Dough strength in terms of Rmax (extensograph 

maximum resistance to extension) shows positive correlation of 

the glutenin fraction with MW greater than MT, which is estimated 

to be approximately 250,000 Da (Bangur et al., 1997). The shift in 

MW distribution (MWD) of this fraction towards higher MW or 

increase in relative proportion of this fraction will increase the 

number of entanglements per chain and reduce the MW between 

entanglements (Termonia and Smith, 1988). This will lead to an 

increase in strength (Gupta et al., 1990). A similar phenomenon is 

observed in wheat flour doughs, where the strain hardening index 

(n) seems to have a positive curvilinear relation with failure strain. 

As n approaches higher limits, the failure strain seems to become 

constant (Dobraszczyk and Roberts, 1994; Dobraszczyk et al., 2003). 

Successive addition of glutenin fractions with increasing molecular 

weight to a base flour at constant protein level causes loaf volume 

to increase, and, after reaching a maximum, to decrease (Lundh and 

MacRitchie, 1989; MacRitchie, 1987; MacRitchie et al., 1991). The 

optimum here possibly indicates a balance between strength and 

extensibility, beyond which high entanglement network density, 

resulting in decrease in failure strain, might cause lower loaf 

volumes (Singh and MacRitchie, 2001; Termonia and Smith, 1988). 

The objectives of this study were: firstly, to investigate and seek 

evidence for the presence of liquid lamellae and their ability to 

stabilize gas cells; i.e. whether or not the dual film hypothesis is 

plausible; secondly, to understand the rheology of the gluten– 

starch matrix required to stabilize the expanding gas cells and to 

understand the underlying molecular structure–function relationship 

as explained by polymer entanglement network theory 

(MacRitchie and Lafiandra, 1997; Singh and MacRitchie, 2001).

34 


2.1. Materials 

Jagger flour and soft wheat flour (a blend of soft wheat varieties) 

were two untreated and unbleached flours used in this study. The 

flours were milled in a Buhler mill (73% milling extraction rate) and 

were evaluated for certain quality characteristics which are shown 

in Table 1. These flours were stored at 20 C until use. For the sake 

of brevity, Jagger wheat flour and soft wheat flour have been 

referred to as Jagger and Soft, respectively. Reagents were 

purchased from Sigma–Aldrich, USA. Chloroform was of HPLC 

grade, whereas all other chemicals used were of ACS grade. Distilled 

deionized water, sterilized in an autoclave, was used in all stages of 

the experiments. 

2.2. Analytical procedures 

Moisture content was determined as per AACC method 44-15 A 

(AACC, 2001). Protein content was determined by the nitrogen 

combustion method using a LECO FP-2000 nitrogen/protein 

analyzer with a factor of 5.7 to convert N to protein. Lipid content 

was determined by cold extraction with chloroform (HPLC grade) 

using the procedure of MacRitchie and Gras (1973). 

2.3. Dough mixing properties 

Mixing properties were evaluated using a 10 g Mixograph 

(National Manufacturing Co., Lincoln, NE). Mixing parameters (peak 

development time and weakening angle) were used for comparison. 

The procedure used was similar to AACC method 55-40 (AACC, 

2001) except that sodium chloride (1.5% w/w on a flour weight 

basis) was added. 

2.4. Preparation of gluten proteins 

Gluten protein was prepared from defatted Jagger flour 

following the procedure used by MacRitchie (1985). The wet gluten 

mass was freeze-dried and ground to a particle size less than 

150 microns. Powdered gluten was stored at a temperature of 

20 C until its use. 

2.5. pH fractionation of gluten protein 

Powdered gluten prepared from Jagger was subjected to pH 

fractionation, using the procedure of Gupta et al. (1990). Gluten was 

stirred in water at pH 5.3 (35 g/1000 ml) for 10 min and centrifuged 

at 2300 g for 10 min. The residue, after collection of supernatant, 

was subjected to two more extractions at the same pH. The total 

Table 1 

Physico-chemical analysis and SE-HPLC relative composition of polymeric proteins 

of Jagger and Soft wheat flours 

Jagger wheat 

flour 

Soft wheat 

flour 

Protein (%) (14% moisture basis) 10.4 0.02 9.2 0.03 

Lipid Content (%) (14% moisture basis) 0.89 0.03 0.93 0.01 

Mixograph midline peak development time 

(min) 

7.21 0.06 9.25 0.00 

Mixograph absorption (%) 63 60 

Area (%) under chromatogram curve 

TPP 36.4 0.01 37.0 0.05 

EPP 43.6 0.02 39.0 0.11 

UPP 56.4 2.77 61.1 3.40 

TPPdtotal polymeric protein; EPPdextractable polymeric protein; UPPdunextractable 

polymeric protein. 

B.S. Sroan et al. / Journal of Cereal Science 49 (2009) 32–40 

combined supernatant was named as pH 5.3 gluten fraction. The 

residue was subjected to further fractionation in a similar manner 

by lowering of pH to 4.9 by addition of HCl. The gluten protein was 

thus fractionated into 6 fractions by sequential lowering of pH. 

Fractions were obtained at pH 5.3, 4.9, 4.1, 3.5 and 3.1, obtaining two 

fractions at pH 3.1 i.e. supernatant of pH 3.1 and residue of pH 3.1. 

All fractions were then freeze-dried. Freeze-dried fractions were 

ground to particle sizes less than 150 microns. The fractions were 

stored in polyethylene bags in containers with desiccant at 

a temperature of 20 C until their use. 

Fractions were analyzed for percentage protein and moisture. 

Protein size characterization of the fractions was done using 

SE-HPLC and MALLS. Fractions retained their functional properties 

(dough mixing properties) when reconstituted in proportions 

relative to their yield, with starch (Midsol-50, MGP Ingredients, 

Inc., Hutchison, KS) to form a flour with protein content equal to 

that of the original flour (Jagger). 

2.6. Addition of protein fractions to flour 

Addition of protein fractions to defatted base flour (Jagger and 

Soft) was done at a level of 1% (dry protein) on a total flour weight 

basis. 

2.7. Size-exclusion high-performance liquid chromatography 

(SE-HPLC) and multiangle laser light scattering (MALLS) 

Size characterization of gluten proteins was done using SE-HPLC 

(Hewlett-Packard 1100 system) in conjunction with MALLS (Multiangle 

light scattering detector DAWN EOS of Wyatt Technology 

Corp., Santa Barbara, CA). The basic procedure for SE-HPLC is 

described elsewhere (Batey et al., 1991). Proteins were fractionated 

with a Biosep SEC-4000 column (Phenomenex, Torrance, CA) using; 

50 mM disodium orthophosphate buffer (NaPhos), pH 7.0, containing 

1% sodium dodecyl sulfate (SDS), as mobile phase at a flow 

rate of 0.5 ml/min. Proteins were detected at 214 nm. Injection 

volume of samples for total and extractable protein analysis was 

80 ml and that for unextractable protein analysis was 40 ml. Therefore, 

results obtained from chromatograms of unextractable 

proteins were multiplied by a factor of 2. Reduced injection volume 

for unextractable protein analysis was used to avoid issues in 

MALLS analysis due to the high concentration of very large glutenin 

protein. 

SEC-HPLC data was analyzed using software program Chem- 

Station (Agilent Technologies, USA). MALLS data was analyzed with 

software program ASTRA 4.50 (Wyatt Technology Corp.) using 0.31 

as the dn/dc value, as per procedure by Bean and Lookhart (2001). 

2.8. Sample preparation for SE-HPLC 

The basic procedure for SE-HPLC is based on the method of 

(Batey et al., 1991. Samples were prepared to analyze total, 

extractable and unextractable polymeric protein (Gupta et al., 

1993). Samples for total and extractable protein analysis were 

weighed in microfuge tubes. Sample size was determined based on 

protein content of each sample such that protein content in all 

samples was kept constant, using Jagger flour as standard and the 

quantity for Jagger being 10.0 mg 0.1 mg. Weighed samples were 

suspended in 50 mM NaPhos, pH 6.9, þ0.5% SDS and solubilized by 

vortexing for 10 min. For total protein analysis, to achieve solubilization 

of the largest molecular size fraction, samples were sonicated 

(Singh et al., 1990) at room temperature at an output of 6 W 

for 15 s. No sonication step was included for extractable protein 

analysis. The sonicator probe was placed at 1/3 distance from the 

bottom of the microfuge tube. Microfuge tubes with protein 

suspensions were then centrifuged at 12,000 g for 20 min. The

supernatant was decanted in HPLC vials and sealed. To ensure 

stability of prepared samples, the vials with supernatant were heat 

treated in a water bath at 85 C for 5 min to inhibit any intrinsic 

proteolytic activity. After heat treatment, vials were cooled with 

crushed ice and analyzed by SE-HPLC. Residue from extractable 

protein was used for unextractable protein analysis. The same 

procedure was followed, except that suspensions were sonicated 

for 25 s at an output of 6 W. 

2.9. Lipid extraction from flour 

Flour lipids were extracted using three batch extractions with 

chloroform in a glass beaker, followed by Buchner filtration through 

Whatman No. 1 filter paper (MacRitchie and Gras, 1973). 200 g of 

flour and 400 ml of chloroform were used for each extraction. The 

defatted flour was spread out on a flat glass tray in a fume hood for 

12 h to allow evaporation of solvent. 

2.10. Addition of lipids to defatted flour 

For incorporation of intact natural flour lipids into defatted flour, 

the parent flour was mixed with defatted flour in different 

proportions, for both the flours, to give different flour lipid levels. 

2.11. Test baking 

Test loaves (35 g flour) were baked using a modified rapid bake 

test (MacRitchie and Gras, 1973). A lean formulation was used with 

no added shortening; flour (100%), sugar (6%), sodium chloride 

(1.5%), instant yeast (2.7%), potassium bromate (30 ppm), water and 

mix time (as optimized from Mixograph analysis). Loaf volumes 

were measured by rapeseed displacement after cooling for 20 min. 

2.12. Image analysis 

Image analysis of crumb grain of baked loaves was carried out 

12 h after baking, with a C-Cell, an image analyzing software and 

equipment (Calibre Control International Ltd., UK). Loaves were 

sliced using a rotary disc blade (unserrated Graef Ò blade) cutter. 

Image analysis was performed on central slices of 15 mm thickness, 

as soon as possible to avoid any shrinkage of crumb grain. Image 

analysis parameters (number of cells and average cell elongation) 

were used for comparison between different treatments. 

2.13. Biaxial extensional rheology 

Biaxial extensional rheological properties of the doughs were 

measured with a Stable Micro Systems dough inflation system 

mounted on a texture analyzer (TAXT 2plus) by means of the 

procedure established by Dobraszczyk (1997). Doughs for rheological 

testing were mixed in the same mixer as used for bake tests, 

using the same water absorption, mixing times and sodium chloride 

addition. After mixing, dough pieces were squashed by hand 

on a sheeting board without putting too much stress on the dough, 

and then allowed to relax for 5 min. They were then sheeted, rolled 

out slowly with several passes and rotated by 90 degrees after each 

pass. Sheeting was done for 5 min with relaxation of 10 s between 

each pass. Sheeting in all directions prevents anisotropic effects 

during dough inflation, allowing dough pieces to expand uniformly 

into spherical shapes. After sheeting, dough pieces were relaxed for 

20 min. They were then cut into circular discs using a 55 mm cookie 

cutter, squashed to a height of 2.67 mm for 20 s. Sample dough 

pieces (in pots) were then proved at 35 C for 25 min. During 

sample preparation, dough pieces were protected against loss of 

moisture using a fine coating of mineral oil (Saybold viscosity 335/ 

358) and covering with shrink wrap film. Mineral oil of lower 


viscosity seems to penetrate dough pieces and may affect rheological 

measurements. 

Dough pieces were inflated at a flow rate of 500 cm 3 /min at 

a strain rate of 0.1/s. Rheological parameters (peak stress, failure 

strain and strain hardening index) were used to compare between 

different treatments. Strain hardening index was calculated by 

fitting an exponential curve to the stress–strain (Hencky) curve, 

after transferring data to Microsoft Excel. 

2.14. Statistical analysis 

Results were analyzed using analysis of variance (ANOVA). 

ANOVA was performed using a general linear model procedure to 

determine significant differences and interactions for the various 

treatments. Means were compared by using Fishers LSD procedure 

(a ¼ 0.05). Statistical analysis was performed using proc GLM in SAS 

(version 9.1; SAS Institute Inc., Cary, NC) software. Duplicates were 

prepared for each treatment and the order of treatment was not 

significant. 


This section is further divided in two sub-sections in accordance 

with objectives of the study. 

3.1. General mechanism of gas cell stability 

(investigating dual film hypothesis) 

3.1.1. Physico-chemical analysis of flours 

Two wheat flours, Jagger and Soft, used for this study were 

analyzed for their chemical composition and physical dough mixing 

properties. Moisture (wet weight basis), protein (14% moisture 

basis) and lipid (14% moisture basis) contents of the flours are given 

in Table 1. Jagger (10.4%) was nearly 1% higher in protein content 

than Soft (9.2%). However, Soft gave higher mixing time (9.25 min) 

compared to Jagger (7.21 min). Higher percentage of unextractable 

polymeric protein (UPP) (Table 1) in Soft explains the higher mixing 

requirements for this flour. This is not typical of soft wheat flours, 

which normally are considered weak. However, some soft varieties 

like Caldwell are relatively stronger than others. Soft wheat flour 

was milled from a composite of soft wheat varieties grown in 

Missouri during 2004–2005 and obtained from grain elevators at 

Kansas City, MO. Increase in mixing requirements in terms of 

Mixograph dough development time (MDDT) with increase in 

percent UPP has been reported previously by Gupta et al. (1993). 

Data on polymeric proteins was obtained by SE-HPLC analysis of 

the flours. Lipid content of Jagger and Soft was 0.89% and 0.93%, 

respectively. 

3.1.2. Effect of natural flour lipid level variation on baking 

performance 

3.1.2.1. Bread making. Significant differences (P < 0.0001) in loaf 

volumes of the two flours were observed in response to varying 

levels of flour lipids (Fig. 2). Incremental addition of flour lipids 

back into defatted parent flour caused bread volume to decrease, 

and after reaching a minimum, to increase. At all levels of lipid 

addition, relatively lower volumes were observed for Soft 

compared to Jagger. For Soft, minimum volume is reached at 

a relatively lower percentage of natural flour lipids (w30%) in 

comparison to Jagger (w50%) (Fig. 2). 

The results of loaf volume variations in response to natural flour 

lipid level variations agreed qualitatively with those of MacRitchie 

and co-workers (MacRitchie, 1976a; MacRitchie and Gras, 1973; 

McCormack et al., 1991). Loaf volume seems to be governed by 

expansion capacity of the gas cells. The expansion capacity of the 

gas cells can be defined as the extent to which gas cells can expand

36 

Fig. 2. Loaf volume vs. intact natural flour lipids, added to defatted Jagger (- - -) and 

Soft (d) wheat flours as percentage of natural flour lipids. 

without failure; i.e. the value reached after which the loaf does not 

increase in volume. This maximum for doughs giving lower loaf 

volumes is achieved during the late proving stage, and for those 

giving higher loaf volumes is achieved during the early baking 

stage. This was determined by measuring oven spring (data not 

shown). Oven spring was observed only in doughs giving higher 

loaf volume. 

Scanning electron micrographs by Gan et al. (1990) appeared to 

show the presence of intact gas cells with discontinuities in gluten– 

starch matrix at advanced stages of proving. This means that the 

expansion capacity of the gas cells is not just controlled by the 

gluten–starch matrix and there is possibly a secondary factor 

contributing to it. This secondary factor as hypothesized by Gan 

et al. (1990) is a liquid lamella present on the inner side of the gas 

cell. When liquid lamellae also fail, the presence of discontinuities 

in the gluten–starch matrix leads to coalescence of gas cells, thus 

decreasing volume. It is quite possible that lipids, due to their 

surface action, will be affecting stability of liquid lamellae, thus 

causing variation in loaf volume. However, this needs to be investigated 

further in order to make sure that the action of flour lipids is 

independent of the rheological properties of the gluten–starch 

matrix. 

Relatively lower loaf volumes at all levels of lipid addition in 

Soft, could be attributed to inherent differences in gluten quality of 

the two flours, as previously suggested by MacRitchie (1978). The 

differences in the two flours are clear from SEC-MALLS data (Table 

5). A new desirable relative proportion of polymeric proteins 

greater than 250,000 Da in Jagger is probably responsible for better 

loaf volumes as explained in Section 3.2.2.3. 

3.1.2.2. Crumb structure. Image analysis of bread crumb (Table 2) 

showed that addition of different levels of flour lipids resulted in 

insignificant variations in number of gas cells (P ¼ 0.02) and 

average cell elongation (P ¼ 0.94 for Jagger and P ¼ 0.12 for Soft). 

These non-significant differences in number of gas cells further 

make it evident that differences in loaf volumes are due to differences 

in expansion capacity of the gas cells (Table 2), and factors 

like gas cell concentration are not involved. Negligible variations in 

average cell elongation indicated that lipids might not be causing 

any variations in rheology of the gluten–starch matrix, since cell 

elongation is thought to be associated with dough rheology (Gandikota 

and MacRitchie, 2005). 

3.1.3. Effect of natural flour lipid level variation on biaxial 

extensional rheology of gluten–starch matrix 

In order to investigate possible independent action of lipids in 

causing loaf volume variations, to indicate presence of liquid 


Table 2 

Crumb structure responses of Jagger and Soft wheat flour breads to natural flour 

lipid levels 

%of 

natural 

flour 

lipids 

Jagger wheat flour Soft wheat flour 

Number of cells Average cell 

elongation 

(C-cell score) 

Number of cells Average cell 

elongation 

(C-cell score) 

0 2000.0 11.3 a 1.58 0.01 b 2544.5 187.4 a,b 1.69 0.01 a,b 

20 1935.0 222.03 a,b 1.59 0.05 b 2521.0 110.3 a,b,c 1.68 0.01 a,b,c 

40 1880.5 78.5 a,b 1.64 0.06 a,b 2210.5 115.3 d 1.65 0.03 b,c 

50 1898.0 59.4 a,b 1.68 0.04 a 2209.0 79.2 d 1.65 0.02 c 

60 1788.5 81.3 a,b 1.67 0.05 a,b 2347.5 2.1 b,c,d 1.70 0.02 a 

80 1907.5 2.1 a,b 1.62 0.01 a,b 2196.0 86.3 d 1.66 0.00 a,b,c 

100 1749.0 41.0 b 1.63 0.01 a,b 2229.0 114.6 d 1.69 0.03 a,b 

Values represent mean standard deviation for duplicate determinations. 

Means with the same letter within columns are not significantly different (p > 0.05). 

lamellae, biaxial extensional rheology tests were conducted on 

defatted Jagger and Soft doughs, and at flour lipid levels where 

maximum and minimum loaf volumes were observed. 

Biaxial extensional rheological parameters (maximum stress, 

failure strain and strain hardening index) were higher for Jagger 

doughs (Table 3) in comparison to Soft doughs. Different lipid 

treatments did not influence these parameters within doughs 

prepared from each flour type (Table 3). Though minor differences 

were observed for strain hardening indices and failure strain of the 

doughs at some lipid levels, no specific trend was observed, thus 

attributing the variation to slight experimental scatter. 

Higher values of biaxial extensional rheological parameters for 

Jagger verify that, rheologically, Jagger is a better bread making 

flour. Higher strain hardening index and failure strains ensure that 

gas cells in Jagger doughs expand to a greater extent than Soft 

doughs. Similar results that differentiate between bread making 

potential of flours have been reported by Dobraszczyk and coworkers 

(Dobraszczyk and Roberts, 1994; Dobraszczyk and 

Salmanowicz, 2008; Dobraszczyk et al., 2003;). On the other hand, 

variations in natural lipid levels in a particular flour that produce 

significant variations in loaf volume (Fig. 2) made no difference to 

biaxial extensional rheology of the dough from that flour. This 

provides clear evidence of the role of lipids as surface active 

compounds stabilizing liquid lamellae, a phenomenon independent 

of dough rheology. 

This study, for the first time provides clear evidence on the 

presence of liquid lamellae as an independent stabilizing mechanism 

working auxiliary to the gluten–starch matrix in stabilizing 

expanding gas cells during bread making. The study also demonstrates 

that lipids at their natural levels do not affect biaxial 

extensional rheology as determined by dough inflation. 

Table 3 

Mean bubble inflation rheological response of Jagger and Soft wheat flour doughs to 

natural flour lipid levels 

% of natural flour 

lipid 

Max. stress (Kpa) Failure strain 

(Hencky) 

Strain hardening 

index 

Jagger wheat flour 

0 571.62 305.45 a 2.58 0.18 a 2.21 0.12 b,c 

60 491.68 237.21 a 2.60 0.12 a 2.38 0.064 a,b 

100 598.62 112.05 a 2.66 0.04 a 2.44 0.06 a 

Soft wheat flour 

0 120.80 12.76 a 1.79 0.00 a 1.76 0.01 a 

40 125.42 12.14 a 1.85 0.02 a 1.74 0.01 a 

100 102.40 5.45 a 1.66 0.05 b 1.64 0.04 c 


Means with the same letter within columns are not significantly different (p > 0.05).

3.2. Mechanism of stability of the gluten–starch matrix 

3.2.1. Gluten fractionation 

Yield (%) and protein content (%) of each fraction from the pH 

fractionation are given in Table 4. Cumulative percentage of total 

polymeric protein extracted is plotted as a function of pH of solution 

in Fig. 3. This was calculated using yield data and percentage 

total polymeric protein of each fraction as given in Table 4. 

Approximately 35% of total polymeric proteins and most of the 

gliadins were solubilized and extracted at pH 5.3. The relative 

proportion of total polymeric proteins and unextractable polymeric 

proteins increased in subsequent fractions (Table 4), with the 

highest values in the pH 3.1 residue. SE-HPLC analyses of individual 

fractions (Fig. 4) also show the subsequent shift in MWD of the 

fractions to higher MW. This is clear in Table 5, where addition of 

these fractions to the base flours causes similar variations in 

percentage of polymeric protein greater than 250,000 Da. The 

solubility of gluten proteins depends on their molecular weight 

(Singh and MacRitchie, 2001). As the pH is lowered, relatively larger 

sized gluten proteins become soluble, and can be separated into 

different fractions by stepwise reduction of pH (Gupta et al., 1990; 

MacRitchie, 1985). 

3.2.2. Reconstitution studies 

Extracted fractions were added to defatted base flours (Jagger 

and Soft) at a level of 1% (dry protein) on a total flour weight basis. 

MacRitchie (1987) found significant rheological differences at these 

levels of addition of protein fractions. The reconstituted flours were 

analyzed for their MWD, dough mixing properties, baking performance 

and biaxial extensional rheology. 

3.2.2.1. SE-HPLC and MALLS (SEC-MALLS) analysis. Cut-off analysis, 

as described elsewhere (Bangur et al., 1997), of SE-HPLC chromatograms 

overlaid with MALLS signal, of all Jagger and Soft wheat 

flours (base and reconstituted) at 4 min intervals, helped to determine 

the elution time at which the polymers with MW of 

250,000 Da were being eluted. The chromatograms were integrated 

to determine the monomeric (mainly gliadins) to polymeric 

(mainly glutenins) ratio, and relative proportion of polymeric 

proteins greater than 250,000 Da (Table 5). Addition of the gliadin 

rich (pH 5.3) fraction caused a decrease in the relative proportion of 

polymeric proteins greater than 250,000 Da. However, with addition 

of later fractions (pH4.9 to pH 3.1 residue), progressive increase 

in the proportion of polymeric proteins greater than 250,000 Da 

was observed. Use of the relative proportion of polymeric proteins 

greater than MW of 250,000 is based on a previous estimate by 

Bangur et al. (1997), which also seems to hold true in this study and 

is explained in Section 3.2.2.3. 

Table 4 

Yield, moisture and protein content of gluten protein fractions extracted by pH 

fractionation, and their SE-HPLC relative composition of polymeric proteins (all 

numbers in percentages) 

Fractions Yield 

(%) 

Moisture 

content 

(% wet basis) 

Protein 

content 

(%) (14% m.b.) 

Area (%) under 

chromatogram curve 

TPP UPP 

pH 5.3 50.4 4.5 0.08 83.8 0.05 31.6 1.59 f 23.9 0.13 d 

pH 4.9 20.5 4.6 0.16 78.0 0.08 52.6 0.08 e 72.6 1.48 c 

pH 4.1 10.6 4.1 0.15 79.6 0.12 64.1 0.10 d 78.3 2.85 b 

pH 3.5 4.2 4.2 0.37 78.5 0.07 65.5 0.01 c 77.6 2.31 b 

pH 3.1 

supernatant 

3.5 4.3 0.09 73.2 0.14 68.0 0.21 b 83.2 0.93 a 

pH 3.1 residue 10.8 4.9 0.26 45.8 0.18 69.5 0.49 a 86.6 0.08 a 

TPPdtotal polymeric protein; UPPdunextractable polymeric protein 


Means with the same letter within columns are not significantly different 

(p > 0.05). 


Fig. 3. Cumulative percentage of total polymeric protein extracted as a function of final 

pH of supernatant of Jagger gluten. 

Elution of proteins through a SE-HPLC column is based on their 

hydrodynamic radii. Proteins with similar MW may vary in the 

conformation and thus the hydrodynamic radii in solution. Such 

proteins will not elute at the same time through the SE-HPLC 

column. An overlaid MALLS signal provides a true measure of MW at 

different elution times (Bean and Lookhart, 2001), providing nearly 

precise estimation of the relative proportion of polymeric proteins 

greater than 250,000 Da. 

3.2.2.2. Dough mixing properties. To observe variations in mixing 

properties of the flours on addition of fractions, Mixograph analysis 

was performed with constant water absorption; i.e. 63% for Jagger 

wheat flour and 60% for Soft wheat flour. Mixograph dough 

development time (MDDT) was recorded from Mixograph traces of 

Jagger and Soft wheat flours. MDDT is plotted for each fraction as 

a function of the total protein that has been extracted (Fig. 5). 

Addition of the first 35% (approximately) of total polymeric protein 

extracted from Jagger gluten (corresponding to the pH 5.3 fraction) 

caused a decrease in mixing requirements. In contrast, addition of 

subsequent fractions caused an increase in mixing requirements. 

These trends for MDDTs can be explained on the basis of SEC- 

MALLS data (Table 5). As expected, the decrease in the monomeric 

to polymeric ratio and increase in relative proportion of polymeric 

proteins greater than 250,000 Da lead to an increase in MDDT and 

corresponding decrease in breakdown. At a molecular level, mixing 

Fig. 4. SE-HPLC chromatograms of total protein of Jagger gluten protein fraction. As pH 

is reduced, the percentage of polymeric protein increases and MWD shifts towards 

higher molecular weight.

38 

Table 5 

Parameters calculated from SEC-MALLS chromatograms of Jagger wheat flour with 

addition of gluten protein fractions, added at 1% (dry protein) level (on flour weight 

basis) 

Fraction added Jagger wheat flour Soft wheat flour 

MON/POL a 

% 

POL > 250,000 b 

MON/POL a 

% 

POL > 250,000 b 

None (Control) 1.47 0.00 a 65.9 0.01 d 1.41 0.01 a 76.8 0.07 d 

pH 5.3 1.48 0.01 a 35.0 0.05 f 1.43 0.03 a 46.7 0.10 f 

pH 4.9 1.36 0.01 b,c 66.9 0.18 c 1.27 0.07 b 68.1 0.22 e 

pH 4.1 1.34 0.00 c,d 58.0 0.04 e 1.28 0.01 b 99.7 0.14 b,c 

pH 3.5 1.37 0.00 b 67.1 0.02 c 1.30 0.01 b 100.0 0.00 a 

pH 3.1 

supernatant 

1.33 0.01 d 85.1 0.12 b 1.16 0.00 c 96.1 0.04 c 

pH 3.1 residue 1.32 0.00 d 100 0.00 a 1.16 0.02 c 100.0 0.00 a 



a 

Monomeric to polymeric protein ratio. 

b 

Percentage of polymeric proteins 250,000 Da. 

is characterized by extension (stretching) of glutenin polymers and 

entangling of these stretched polymers (MacRitchie, 1986). This is 

achieved by work input and mixing intensity that must be above 

certain minimum critical levels (Kilborn and Tipples, 1972; Tipples 

and Kilborn, 1975). At a given work input and intensity of mixing, 

MDDT probably relates to the extension and entangling of the 

largest glutenin molecules (Singh and MacRitchie, 2001). The work 

input and mixing intensity of the flour is therefore a function of its 

MWD. It has been observed that when the percentage of UPP 

increases and/or the MWD of the polymeric fraction is shifted to 

higher MWs, the MDDT increases, as requirements for work input 

increase (Gupta et al., 1993). 

3.2.2.3. Baking performance. Loaf volumes (Fig. 6) decreased on 

addition of the pH 5.3 fraction. However, on addition of subsequent 

fractions that lead to increase in MDDT, increase in loaf volume was 

observed, which after reaching a maximum, again decreased. This 

maximum was achieved in the case of Jagger on addition of total 

polymeric proteins that were extracted after the first 60% and 

before the first 80% corresponding to the pH 4.1 fraction. In the case 

of Soft, the maximum occurred on addition of total polymeric 

proteins that were extracted after the first 35% and before the first 

70%, corresponding to the pH 4.9 fraction (Figs. 3 and 6). It seems 

that the optimum balance of strength and extensibility was achieved 

on addition of these fractions to base flours, thus giving best 

baking performance. 

Bangur et al. (1997) reported that the ratio of polymeric proteins 

with MWs above the threshold molecular weight (MT) to those (i.e. 

Fig. 5. Effect on Mixograph peak times of Jagger (- - -) and Soft (d) wheat flours on 

addition of gluten protein fractions at a 1% (dry protein) level (on flour weight basis). 


Fig. 6. Effect on loaf volumes (d) and strain hardening index (- - -) of Jagger (:) and 

Soft (-) wheat flours on addition of gluten protein fractions at a 1% (dry protein) level 

(on flour weight basis). 

polymeric proteins) with MWs below MT was approximately 60:40. 

The MT for polymeric proteins was estimated to be 250,000 Da. 

Polymeric proteins greater in molecular weight than 250,000 Da 

confer strength to the entangled gluten protein network and those 

with molecular weight less than this may counter the strength by 

acting as diluents (or plasticizers). This 60:40 proportion is nearly 

achieved on addition of the pH 4.1 fraction in Jagger wheat flour 

and pH 4.9 fraction in Soft wheat flour (Table 5) giving maximum 

loaf volumes and strain hardening index. Any shift in the proportion 

shows deleterious effects on baking performance (Fig. 6). 

3.2.2.4. Biaxial extensional rheology. Significant differences 

(P < 0.0001) in biaxial extensional rheological parameters (strain 

hardening index and failure strain) were recorded for addition of 

fractions to the base flours. Variations in strain hardening index 

paralleled those of loaf volume, for both flours (Fig. 6). Analogous 

variations in strain hardening index and loaf volume show that this 

parameter is a good determinant of rheological requirements for 

the gluten–starch matrix in terms of bread making. As reported in 

some previous studies (Dobraszczyk and Roberts, 1994; Dobraszczyk 

et al., 2003; Tronsmo et al., 2003), in this case also, strain 

hardening shows a high degree of correlation with loaf volume and 

failure strain (Fig. 7). The Mixograph parameter (MDDT), on the 

other hand, correlates well with MWD but does not predict baking 

performance equally well. 

The concept of strain hardening explains the requirements for 

maximum inflatability of gas cells in bread making without failure, 

to give good loaf volume and crumb grain. In a gluten–starch 

matrix, gluten proteins form a continuous network and the 

rheology of the dough is that of a continuous gluten protein 

network. It is only the fraction of glutenins with MW greater than 

MT that confers strength, as suggested by Bangur et al. (1997), 

whereas smaller ones act as diluents, preventing additional physical 

constraints or entanglements and enhancing extensibility 

(Bersted and Anderson, 1990; Singh and MacRitchie, 2001). To get 

maximum inflatability of the gas cells, the gluten–starch matrix 

around them must stretch to its maximum extensibility without 

breaking, i.e. the glutenins/polymeric proteins (mainly with 

MW > MT; i.e. 250,000 Da) must be stretched to their maximum 

length through entanglements, as described by entanglement 

network theory (MacRitchie and Lafiandra, 1997; Singh and 

MacRitchie, 2001). Addition of the gliadin rich fraction, i.e. the pH 

5.3 fraction (Fig. 4), significantly increases diluent concentration 

that will lead to disentanglement of the gluten protein network 

without sufficient elongation when subjected to extensional forces, 

thus leading to earlier failure of the gas cells. As MWD shifts

towards larger glutenins, the strength increases (Gupta et al., 1993). 

On doing so, it passes through maximum extensibility or strain 

(Hencky), beyond which the balance shifts towards strength and 

the gluten protein network is no longer as extensible. This leads to 

decrease in strain hardening index as well as loaf volume upon 

addition of the latter extracted fractions (Fig. 6). 

4. Conclusion 

Results of this study provide clear evidence for the presence of 

liquid lamellae, thus supporting the dual film hypothesis. The liquid 

lamellae act as a secondary stabilizing mechanism and is on the 

inner side of the gluten–starch matrix enveloping the gas cells (Gan 

et al., 1995). 

Stability of the gluten–starch matrix, which is the primary 

stabilizing factor for expanding gas cells against disproportionation 

and coalescence, depends on its tendency to strain harden. The 

phenomenon of strain hardening appears to depend on the balance 

between strength and extensibility of the entangled network of 

polymeric proteins of wheat flour. Extensibility ensures slippage of 

the maximum number of statistical segments between entanglements 

(Singh and MacRitchie, 2001), whereas strength prevents 

disruption of the entangled network of polymeric proteins. Thus, to 

ensure stability of gas cells, the dough needs to be sufficiently 

extensible to respond to gas pressure but also strong enough to 

resist collapse. Differences in gluten quality, as demonstrated in 

this study can significantly affect the bread making potential. 

Strength is conferred by the fraction of polymeric proteins having 

molecular weight greater or equivalent to MT (250,000), and the 

fraction of gluten protein smaller than MT may counter the strength 

by acting as diluents. The optimum balance seems to exist when the 

relative proportions of polymeric proteins greater and smaller than 

MT are roughly 60:40. Shift in the balance to either side will 

decrease loaf volume. Increase in smaller proteins (less than MT) 

may decrease stability of the gluten–starch matrix due to a lesser 

number of entanglements per chain. On the other hand, increase in 

strength conferring proteins may prevent sufficient expansion of 

the gluten–starch matrix required to increase loaf volume due to 

reduced slippage of gluten polymers through entanglement nodes 

as a result of increase in number of entanglements per chain. 


Fig. 7. Scatter plots of strain hardening index vs. failure strain and loaf volume (cc) for different Jagger and Soft wheat flour doughs exhibiting high degree of correlation. 

The secondary stabilizing mechanism involves thin liquid 

lamellae stabilized by adsorbed surface active compounds (lipids 

and proteins) at the gas–liquid interface. Liquid lamellae prevent 

coalescence and disproportionation of gas cells when they come in 

close contact with each other during the late proving and early 

baking stages of bread making i.e. when discontinuities begin to 

appear in the gluten–starch matrix. The study demonstrates that 

the flour lipids at their natural levels do not influence rheological 

properties of the gluten–starch matrix surrounding the gas cells, as 

measured by the dough inflation system. Nevertheless, the small 

amounts in which these lipids are naturally present are sufficient to 

influence surface properties. The exact mechanism by which these 

lipids and other surface active components (proteins) stabilize 

liquid lamellae is discussed in the succeeding paper (Part II). 

References 

AACC, 2001. Approved Methods of the AACC, tenth ed. American Association of 

Cereal Chemists, St. Paul, Minnesota, USA. 

Baker, J.C., Mize, M.D., 1946. Gas occlusion during dough mixing. Cereal Chemistry 

23, 39–51. 

Bangur, R., Batey, I.L., McKenzie, E., MacRitchie, F., 1997. Dependence of extensograph 

parameters on wheat protein composition measured by SE-HPLC. Journal 

of Cereal Science 25, 237–241. 

Batey, I.L., Gupta, R.B., MacRitchie, F., 1991. Use of size-exclusion high-performance 

liquid chromatography in the study of wheat flour proteins: an improved 

chromatographic procedure. Cereal Chemistry 68, 207–209. 

Bean, S.R., Lookhart, G.L., 2001. Factors influencing the characterization of gluten 

proteins by size-exclusion chromatography and multiangle laser light scattering 

(SEC-MALLS). Cereal Chemistry 78, 608–618. 

Bersted, B.H., Anderson, T.G., 1990. Influence of molecular weight and molecular 

weight distribution on the tensile properties of amorphous polymers. Journal of 

Applied Polymer Science 39, 499–514. 

Bloksma, A.H., 1990. Dough structure, rheology, and baking quality. Cereal Foods 

World 35, 237–244. 

Chin, N.L., Campbell, G.M., 2005. Dough aeration and rheology. II. Effects of flour 

type, mixing speed and total work input on aeration and rheology of bread 

dough. Journal of the Science of Food and Agriculture 85, 2194–2202. 

DeStefanis, V.A., Ponte Jr., J.G., 1976. Studies on breadmaking properties of wheatflour 

nonpolar lipids. Cereal Chemistry 53, 636–642. 

Dobraszczyk, B.J., 1997. Development of a new dough inflation system to evaluate 

doughs. Cereal Foods World 42, 516–519. 

Dobraszczyk, B.J., Roberts, C.A., 1994. Strain hardening and dough gas cell-wall 

failure in biaxial extension. Journal of Cereal Science 20, 265–274. 

Dobraszczyk, B.J., Salmanowicz, B.P., 2008. Comparison and predictions of baking 

volume using large deformation rheological properties. Journal of Cereal 

Science 47, 292–301.

40 

Dobraszczyk, B.J., Smewing, J., Albertini, M., Maesmans, G., Schofield, J.D., 2003. 

Extensional rheology and stability of gas cell walls in bread doughs at elevated 

temperatures in relation to breadmaking performance. Cereal Chemistry 80, 

218–224. 

Gan, Z., Angold, R.E., Williams, M.R., Ellis, P.R., Vaughan, J.G., Galliard, T., 1990. The 

microstructure of gas retention of bread dough. Journal of Cereal Science 12, 

15–24. 

Gan, Z., Ellis, P.R., Schofield, J.D., 1995. Gas cell stabilization and gas retention in 

wheat bread dough. Journal of Cereal Science 21, 215–230. 

Gandikota, S., MacRitchie, F., 2005. Expansion capacity of doughs: methodology and 

applications. Journal of Cereal Science 42, 157–163. 

Gupta, R.B., Bekes, F., Gras, P.W., MacRitchie, F., 1990. Functionality of glutenin, 

gliadin and secalin fractions as measured by extensograph and mixograph. In: 

Gluten Proteins. Association of Cereal Research, Detmold, pp 550–559. 

Gupta, R.B., Khan, K., MacRitchie, F., 1993. Biochemical basis of flour properties in 

bread wheats. I. Effects of variation in the quality and size distribution of 

polymeric protein. Journal of Cereal Science 18, 23–41. 

Hoseney, R.C., 1992. Physical chemistry of bread dough. In: Schwartzberg, H.G., 

Hartel, R.W. (Eds.), Physical Chemistry of Foods. Marcel Dekker, Inc., New York, 

USA, pp. 443–457. 

Kilborn, R.H., Tipples, K.H., 1972. Factors affecting mechanical dough development. 

I. Effect of mixing intensity and work input. Cereal Chemistry 49, 34–47. 

Lundh, G., MacRitchie, F., 1989. Size exclusion HPLC characterization of gluten 

protein fractions varying in breadmaking potential. Journal of Cereal Science 10, 

247–253. 

MacRitchie, F., 1976a. The liquid phase of dough and its role in baking. Cereal 

Chemistry 53, 318–326. 

MacRitchie, F., 1976b. Monolayer compression barrier in emulsion and foam 

stability. Journal of Colloid and Interface Science 56, 53–56. 

MacRitchie, F., 1978. Differences in baking quality between wheat flours. Journal of 

Food Technology 13, 187–194. 

MacRitchie, F., 1985. Studies of the methodology for fractionation and reconstitution 

of wheat flours. Journal of Cereal Science 3, 221–230. 

MacRitchie, F., 1986. Physicochemical processes in mixing. In: Blanshard, J.M.V., 

Frazier, P.J., Galliard, T. (Eds.), Chemistry and Physic of Baking. Royal Society of 

Chemistry, London, UK, pp. 132–146. 


MacRitchie, F., 1987. Evaluation of contributions from wheat protein fractions to 

dough mixing and breadmaking. Journal of Cereal Science 6, 259–268. 

MacRitchie, F., Gras, P.W., 1973. The role of flour lipids in baking. Cereal Chemistry 

50, 292–302. 

MacRitchie, F., Kasarda, D.D., Kuzmicky, D.D., 1991. Characterization of wheat 

protein fractions differing in contributions to breadmaking quality. Cereal 

Chemistry 68, 122–130. 

MacRitchie, F., Lafiandra, D., 1997. Structure–functionality relationships of wheat 

proteins. In: Damodaran, S., Paraf, A. (Eds.), Food Proteins and Their Applications. 

Marcel Dekker, Inc., New York, USA, pp. 293–324. 

McCormack, G., Panozzo, J., MacRitchie, F., 1991. Contributions to breadmaking of 

inherent variations in lipid content and composition of wheat cultivars. II. 

Fractionation and reconstitution studies. Journal of Cereal Science 13, 263–274. 

Mills, E.N.C., Wilde, P.J., Salt, L.J., Skeggs, P., 2003. Bubble formation and stabilization 

in bread dough. Food and Bioproducts Processing 81, 189–193. 

Paternotte, T.A., Orsel, R., Hamer, R.J., 1993. Interactions between flour proteins and 

flour lipids at the liquid/air interface. In: Gluten Proteins. Association of Cereal 

Research, Detmold, Germany, pp. 207–217. 

Ross, A., MacRitchie, F., 1995. Interactions of wheat proteins, carbohydrates and 

lipids. In: Gaonkar, A.G. (Ed.), Ingredient Interactions Effects on Food Quality. 

Marcel Deker, New York, USA, pp. 321–356. 

Singh, N.K., Donovan, G.R., Batey, I.L., MacRitchie, F., 1990. Use of sonication and 

size-exclusion high-performance liquid chromatography in the study of wheat 

flour proteins. I. Dissolution of total proteins in the absence of reducing agents. 


Singh, H., MacRitchie, F., 2001. Application of polymer science to properties of 

gluten. Journal of Cereal Science 33, 231–243. 

Termonia, Y., Smith, P., 1988. Kinetic model for tensile deformation of polymers. 2. 

Effect of entanglement spacing. Macromolecules 21, 2184–2189. 

Tronsmo, K.M., Færgestad, E.M., Schofield, J.D., Magnus, E.M., 2003. Wheat protein 

quality in relation to baking performance evaluated by the Chorleywood bread 

process and a hearth bread baking test. Journal of Cereal Science 38, 205–215. 

Tipples, K.H., Kilborn, R.H., 1975. Unmixingdthe disorientation of developed bread 

doughs by slow speed mixing. Cereal Chemistry 52, 248–262. 

van Vliet, T., Janssen, A.M., Bloksma, A.H., Walstra, P.,1992. Strain hardening of dough 

as a requirement for gas retention. Journal of Texture Studies 23, 439–460.

Mechanism of gas cell stabilization in breadmaking. II. The secondary liquid 

lamellae 

Baninder S. Sroan *, Finlay MacRitchie 

Department of Grain Science and Industry, Kansas State University, Manhattan, KS 66506, USA 

article info 





Keywords: 

Breadmaking 

Gas cell stability 

Liquid lamella 

Gluten–starch matrix 

Coalescence 

Disproportionation 


abstract 

Concentration of uniformly sized gas bubbles in the liquid phase 

of dough during mixing is essential for their sufficient expansion 

without failure during the breadmaking process for good loaf 

volume and crumb grain. There is evidence that the liquid lamellae 

surrounding expanding gas cells act as a secondary protection 

together with the primary gluten–starch matrix, thus preventing 

their coalescence and disproportionation as outlined in Part-I. The 

secondary stabilizing mechanism of thin liquid lamellae (films) 

operates when gas bubbles come in close contact during later 

proving and early baking when discontinuities begin to appear in 

the gluten–starch matrix. Stability of the liquid lamellae depends 

on surface active compounds (proteins and lipids) (Baker et al., 

1946; MacRitchie, 1976a; Mills et al., 2003), that are adsorbed at the 

gas–liquid interface. Analogous variations in foaming properties of 

the dough liquor, and bread loaf volume and crumb structure 

(MacRitchie, 1976a) provide plausible evidence of surface action of 

these compounds at the gas–liquid interface. Contributions of 

liquid film stability to dough expansion can be assessed from 

different studies. It has been shown that flour lipids, though 

present in very small amounts (w1–1.5%) have significant effects on 

* Corresponding author. Present address: FritoLay, 7701 Legacy Dr., 3T-124, Plano, 

TX 75024, USA. Tel.: þ1 972 334 4526; fax: þ1 972 334 2329. 

E-mail address: baninder@k-state.edu (B.S. Sroan). 


doi:10.1016/j.jcs.2008.07.004 





The study for the first time demonstrates that flour lipids at their natural levels do not affect dough 

rheology as measured by bubble inflation, thus indicating the presence of liquid lamellae as an independent 

secondary gas cell stabilizing mechanism in bread dough. The liquid lamellae, stabilized by 

adsorbed surface active compounds, plays its role during the later proving and early baking stage, when 

discontinuities occur in the gluten–starch matrix surrounding gas bubbles. To study this secondary 

stabilizing mechanism, different lipid fractions were added incrementally to the defatted flours. No 

effects were observed on the rheological properties of the dough. However, large effects on the loaf 

volume were measured. The additives used were the total flour lipid and its polar and non-polar fractions 

and the fatty acids palmitic, linoleic and myristic. Polar lipids and palmitic acid had positive or little effect 

on loaf volume, respectively. Non-polar lipid, linoleic and myristic acids had negative effects on loaf 

volume. The different effects of the lipid fractions are thought to be related to the type of monolayer that 

is formed. Polar lipid and palmitic acid form condensed monolayers at the air/water interface whereas 

non-polar lipid, linoleic and myristic acids form expanded monolayers. 


baking performance in terms of volume and crumb structure 

(MacRitchie and Gras, 1973; Mills et al., 2003). Polar lipids improve 

bread loaf volume and crumb grain, whereas non-polar lipids have 

the opposite effects (MacRitchie and Gras, 1973; Ponte and DeStefanis, 

1969). DeStefanis and Ponte (1976) demonstrated that an 

unsaturated free fatty acid (linoleic acid), a component of the nonpolar 

lipid fraction is responsible for detrimental effects on loaf 

volume. Mills et al. (2003) reviewed some of the more recent 

studies on effects of surface active components of wheat flour on 

breadmaking performance. 

The most probable mechanism by which these surface active 

components control stability of gas cells is through formation of 

monolayers at the gas–liquid interface (Mills et al., 2003; Ross and 

MacRitchie, 1995). Saturated free fatty acids, like palmitic and 

stearic, form condensed monolayers, which are highly incompressible 

with high surface elasticity and these molecules do not 

desorb easily from the monolayers. These condensed monolayers 

generate substantial elastic restoring forces, which resist destabilization 

of liquid lamellae when changes in interfacial area occur 

(MacRitchie, 1976b). Molecules like digalactosyl diglyceride 

(DGDG) with strong polar head groups are expected to orient 

themselves strongly at the interface as condensed monolayers. 

Unsaturated free fatty acids like linoleic acid give expanded 

monolayers, which are relatively compressible and slightly soluble 

in the subphase, leading to instability of the liquid lamellae 

(MacRitchie, 1976b). Rich in surface active compounds (lipids,

42 

proteins, arabinoxylans, added surfactants, etc.), the liquid phase of 

dough is a very competitive environment in terms of adsorption of 

these compounds at the gas–liquid interface (Mills et al., 2003; Ross 

and MacRitchie, 1995). Proteins (soluble or insoluble) in wheat flour 

are capable of forming condensed monolayers. Pure protein 

monolayers possess high surface elasticity. When lipids are introduced 

in the system, surface elasticity drops as lipids compete with 

proteins for the gas–liquid interface (Paternotte et al., 1993; Salt 

et al., 2006). It seems that the type of monolayer formed will either 

stabilize or destabilize the freshly occluded gas cells, and any 

changes in monolayer state will influence the gas cell stability at all 

stages of breadmaking. 

Therefore, the stability of liquid lamellae determines the ease 

with which gas cells are concentrated during mixing and ensure 

their permanence during the entire breadmaking process. 

The objective of this part of the study was to understand the 

possible mechanism by which liquid lamellae are stabilized. This 

part of the study provides further evidence in support of independent 

existence of liquid lamellae as a secondary stabilizing 

mechanism supporting expanding gas cells. It shows for the first 

time that lipids, at their natural levels, have no effect on the 

rheological properties of the starch/protein (gluten–starch) matrix 

as measured by bubble inflation and their effects are therefore 

independent of the stabilizing mechanism discussed in Part-I. 


Materials and various procedures (analytical, dough mixing 

properties, test baking, image analysis, biaxial extensional rheology 

and statistical analysis) used in the study are explained in detail in 

Part-I. 

2.1. Free fatty acids 

Palmitic acid (99%), linoleic acid (60%), and myristic acid (99%) 

were purchased from Sigma-Aldrich, USA. Values in brackets are of 

minimum purity by gas chromatography as per certificate of analysis 

by manufacturer. Linolenic acid was the main impurity in 

linoleic acid. 

2.2. Lipid extraction from flour 

Natural flour lipids were extracted using three batch extractions 

with chloroform in a glass beaker, followed by Buchner filtration 

through Whatman No. 1 filter paper (MacRitchie and Gras, 1973). 

200 g of flour and 400 ml of chloroform were used for each 

extraction. The defatted flour was spread out on a flat glass tray in 

a fume hood for 12 h to allow evaporation of solvent. 

2.3. Lipid fractionation 

Natural lipids of both flours in chloroform were collected 

together and concentrated under vacuum using a rotary evaporator 

at a temperature

Fig. 2. Loaf volume vs. different lipid types, added to defatted Jagger wheat flour as 

percentage of natural flour lipids. 

depression for small additions followed by increase in loaf volume. 

This increase in volume was appreciably larger than what was 

observed for the same levels of natural flour lipids. Addition of 

increasing levels of the non-polar lipids and linoleic acid caused 

a continuous decrease in loaf volume for both the flours. 

Incremental addition of palmitic acid produced no change in loaf 

volume of Jagger flour and slight depression in soft wheat flour. 

Incremental addition of myristic acid caused depression in loaf 

volume at initial levels, however, at considerably higher levels of 

addition, some reversal was observed. This increase, in the case 

of myristic acid, leveled off at the volumes equivalent to those of 

Fig. 3. Loaf volume vs. different lipid types, added to defatted soft wheat flour as 

percentage of natural flour lipids. 

B.S. Sroan, F. MacRitchie / Journal of Cereal Science 49 (2009) 41–46 43 

intact flours (100% of natural flour lipids). Maximum addition of 

myristic acid in Jagger was 400% of natural flour lipid content. This 

was a very high level of addition in comparison to other lipid types 

and free fatty acids used in the study; this was done to have a clear 

picture of the loaf volume trend in this case. 

The results for natural flour lipids and their polar and non-polar 

fractions agreed qualitatively with those of MacRitchie and coworkers 

(McCormack et al., 1991; MacRitchie, 1977; MacRitchie, 

1978; MacRitchie and Gras, 1973), and those of unsaturated (linoleic) 

and saturated (palmitic) free fatty acids concurred with results 

of DeStefanis and Ponte (1976). Effects of different lipid types on 

loaf volume and crumb structure correlated with results from film 

balance studies reported by various workers (Gaines, 1966; 

MacRitchie, 1990). This substantiated our earlier results (Part-I), 

providing evidence for the presence of liquid lamellae around 

expanding gas cells. The results also indicated that the most 

probable mechanism by which different lipid types influenced the 

gas cell stability was through their adsorption as monolayers at the 

gas–liquid interface. 

The monolayers formed are either condensed or expanded 

(Gaines, 1966; MacRitchie, 1990). Condensed monolayers are 

characterized by close packing of surface active molecules, with 

polar head groups oriented towards water and non-polar hydrocarbon 

chains towards air. Expanded monolayers are formed due to 

loose packing of surface active molecules, occupying a much higher 

area per molecule. Condensed monolayers are relatively incompressible 

and elastic compared to expanded ones and are less easily 

desorbed than expanded monolayers (MacRitchie, 1976b). On 

expansion of gas cells, condensed monolayers provide elastic 

restoring forces, contributing to resistance to collapse of liquid 

lamellae with change of interfacial area (MacRitchie, 1976b). 

In defatted flours, proteins are the only surface active components 

forming monolayers. This results in a higher loaf volume and 

a finer and more uniform crumb structure as proteins provide quite 

stable monolayers and do not desorb easily from the gas–liquid 

interface. This has been attributed to suitable configurations 

adopted by protein molecules at the interface and to their large 

sizes (Larsson et al., 2006; MacRitchie, 1990). Addition of different 

lipid types may lead to the formation of mixed monolayers (Mills 

et al., 2003; Ross and MacRitchie, 1995; Salt et al., 2006). The 

occurrence of minima in loaf volume on addition of natural flour 

lipid and its polar fraction is analogous to the decrease in stability of 

mixed films reported by Paternotte et al. (1993) on passage from 

a pure protein film to a lipid (monogalactosyl monoglyceride)– 

protein stabilized film. Variations in percentage of natural flour 

lipids at which the minimum volume is reached could be attributed 

to differences in protein composition of the two flours. A study by 

Salt et al. (2006) also indicated that mixed protein–lipid interfaces 

are relatively less elastic and therefore more unstable compared to 

pure protein ones, which possess high surface elasticity. Palmitic 

acid, a saturated free fatty acid, and polar lipids having digalactosyl 

diglyceride (DGDG), due to their tendency to form condensed 

monolayers (Gaines, 1966; MacRitchie, 1990), had either no effect 

or positive effect on loaf volumes, respectively. On the other hand, 

the probable formation of an expanded monolayer by linoleic acid 

resulted in detrimental effects on loaf volume. Myristic acid is 

a saturated free fatty acid but, due to its shorter chain length, forms 

expanded monolayers (MacRitchie, 1990), thus decreasing loaf 

volume. At higher levels of myristic acid addition, a reversal of the 

effect is observed (Figs. 2 and 3). This could be attributed to the 

presence of impurities in myristic acid, which is w99% pure by gas 

chromatographic analysis. However, such purity analysis does not 

indicate surface chemical purity of surface active compounds like 

myristic acid (MacRitchie, 1990). The presence of impurities, even 

in minor amounts, can influence surface properties. This might 

have been the case at higher levels of addition of myristic acid in

44 

B.S. Sroan, F. MacRitchie / Journal of Cereal Science 49 (2009) 41–46 

Table 1 

Crumb structure responses of Jagger and soft wheat flour breads to different lipid types and levels, added to defatted flours as percentage of natural flour lipids 

% Of natural flour lipids Jagger wheat flour Soft wheat flour 

Number of cells Average cell elongation (C-cell score) Number of cells Average cell elongation (C-cell score) 

Natural flour lipids 

0% 2000.0 11.3 a 1.58 0.01 b 2544.5 187.4 a,b 1.69 0.01 a,b 

20% 1935.0 222.03 a,b 1.59 0.05 b 2521.0 110.3 a,b,c 1.68 0.01 a,b,c 

40% 1880.5 78.5 a,b 1.64 0.06 a,b 2210.5 115.3 d 1.65 0.03 b,c 

50% 1898.0 59.4 a,b 1.68 0.04 a 2209.0 79.2 d 1.65 0.02 c 

60% 1788.5 81.3 a,b 1.67 0.05 a,b 2347.5 2.1 b,c,d 1.70 0.02 a 

80% 1907.5 2.1 a,b 1.62 0.01 a,b 2196.0 86.3 d 1.66 0.00 a,b,c 

100% 

Flour polar lipids 

1749.0 41.0 b 1.63 0.01 a,b 2229.0 114.6 d 1.69 0.03 a,b 

0% 2361.5 7.8 a 1.56 0.03 c,d 2107.0 26.9 b 1.60 0.02 c 

20% 2307.0 157.0 a 1.57 0.01 b,c,d 1876.5 37.5 c 1.66 0.02 a,b 

40% 2001.0 24.0 b 1.60 0.04 a,b,c 2389.0 0.0 a 1.64 0.00 b 

66% 1985.0 0.0 b 1.62 0.00 a 2183.0 0.0 b 1.67 0.00 a,b 

132% 2285.0 0.0 a 1.54 0.00 d 2180.5 23.3 b 1.69 0.03 a 

200% 

Flour non-polar lipids 

2385.0 0.0 a 1.61 0.00 a,b 2092.0 216.4 b 1.67 0.01 a,b 

0% 2436.0 113.1 a 1.57 0.01 b 2110.0 22.6 a 1.60 0.02 a 

15% 2306.5 126.6 a 1.57 0.00 b 2036.5 95.5 a,b 1.59 0.01 a 

30% 1989.0 69.3 b,c 1.60 0.04 b 1832.0 144.2 b 1.68 0.05 a 

60% 2202.5 187.4 a,b 1.61 0.01 a,b 1950.0 104.7 a,b 1.66 0.05 a 

132% 1937.5 47.4 b,c 1.65 0.01 a 2053.0 14.1 a,b 1.64 0.01 a 

200% 

Linoleic acid 

1761.5 0.0 c 1.61 0.00 a,b – – 

0% 2452.5 26.2 a 1.63 0.06 a 2060.5 29.0 a,b 1.60 0.00 c 

66% 1803.0 220.6 b 1.68 0.02 a 2290.0 158.4 a 1.68 0.02 b 

132% 1796.5 67.2 b 1.72 0.04 a 1926.0 116.0 b 1.69 0.02 a,b 

200% 1473.5 169.0 b 1.70 0.04 a 2019.0 144.2 a,b 1.73 0.01 a 

250% 

Palmitic acid 

1505.5 13.4 b 1.72 0.00 a 1994.0 120.2 a,b 1.71 0.02 a 

0% 2477.5 139.3 a 1.61 0.04 a 2060.5 29.0 a 1.60 0.00 b 

66% 2438.5 103.9 a 1.64 0.02 a 2043.0 69.3 a 1.64 0.01 a,b 

132% 2522.5 171.8 a 1.63 0.06 a 1940.5 68.6 a 1.66 0.06 a,b 

200% 2331.0 14.1 a 1.58 0.00 a 1880.5 17.7 a 1.72 0.04 a 

250% 

Myristic acid 

2285.5 126.5 a 1.59 0.01 a 1829.5 204.4 a 1.65 0.03 b 

0% 2328.0 63.6 a 1.62 0.01 a 2100.0 135.8 a 1.68 0.01 b 

66% 2368.0 190.9 a 1.62 0.02 a 1954.0 24.0 a 1.68 0.01 b 

132% 2404.5 54.5 a 1.61 0.01 a 2263.5 54.5 a 1.68 0.01 b 

200% 2447.5 150.6 a 1.64 0.00 a 2270.5 147.8 a 1.68 0.01 b 

250% 2244.5 17.7 a 1.65 0.00 a 2156.5 188.8 a 1.68 0.01 b 

330% 2324.0 84.9 a 1.66 0.00 a – – 

400% 2632.0 35.4 a 1.65 0.00 a – – 



Fig. 4. C-cell images of Jagger wheat flour bread slices showing differences in gas cell expansion at different polar lipid levels (represented as percentages) added to defatted flour as 

percentage of natural flour lipids.

Fig. 5. C-cell images of Jagger wheat flour bread slices showing differences in gas cell 

expansion at different non-polar lipid levels (represented as percentages) added to 

defatted flour as percentage of natural flour lipids. 

baking studies. Therefore analysis of surface chemical purity 

(Lunkenheimer and Miller, 1987) of added surfactants, though not 

performed in this study, will be of great advantage in future studies. 

3.1.2. Crumb structure 

Image analysis of bread crumb showed that addition of different 

lipid types and variations in their levels resulted in negligible 

differences in number of gas cells and average cell elongation (Table 

1). These insignificant differences in the number of gas cells for 

a particular flour when treated with different levels of the same 

lipid type and free fatty acid, suggest that higher loaf volumes were 

due to an increase in expansion capacity of gas cells and not due to 

their number. That is, the film stabilizers (polar lipids and palmitic 

acid) allowed gas cells to expand, increasing the volume (Fig. 4). 

With improvement in expansion capacity, the size of gas cells 

increased and crumb structure changed from fine to more 

uniformly distributed larger gas cells. The inability of gas cells to 

expand (Fig. 5) also explains the earlier observations of relatively 

fine crumb appearance with addition of non-polar lipids reported 

by Ponte and DeStefanis (1969). 

However, an exception was observed in the case of linoleic acid 

addition to Jagger flour. Incremental additions of linoleic acid to 

Jagger flour caused small but continuous decrease in number of gas 

cells (Table 1). Linoleic acid might also be detrimental to initial 

concentration and stability of gas cells during mixing, thus 

hindering the ease with which gas cells are occluded into the liquid 

phase of dough. This initial adsorption depends upon the diffusion 

coefficient of surface active compounds (lipids and proteins) in the 

liquid phase of the dough (MacRitchie, 1990), and will vary from 

flour to flour. The nature of surface active compounds which has 

been adsorbed initially at the interface to form a monolayer will 

stabilize or destabilize freshly occluded gas cells, thus affecting the 

ease with which they are concentrated (Larsson et al., 2006). 

Negligible variations in average cell elongation indicated that 

different lipid types and free fatty acids at the levels of addition 

used might not be influencing rheology of the gluten–starch 

matrix, since cell elongation is thought to be associated with dough 

rheological properties (Gandikota and MacRitchie, 2005). 

3.2. Effect of variations in lipid types and free fatty acids, and their 

levels on biaxial extensional rheology of gluten–starch matrix 

Biaxial extensional rheological tests were performed on Jagger 

and soft doughs with different lipid types and their levels, in order 

B.S. Sroan, F. MacRitchie / Journal of Cereal Science 49 (2009) 41–46 45 

Table 2 

Mean bubble inflation rheological responses of Jagger wheat flour doughs to 

different lipid types and levels, added to defatted flours as percentage of natural 

flour lipids 

% Of natural flour 

lipid 

to investigate possible independent action of lipids on baking 

performance (loaf volume and crumb structure). Different lipid 

types and free fatty acids at the levels used did not influence biaxial 

extensional rheological parameters (maximum stress, failure strain 

and strain hardening index) of the doughs (Tables 2 and 3). 

However, higher values were observed for Jagger than soft wheat 

flour. Though minor differences were observed for strain hardening 

indices within a particular flour at some treatments, no specific 

trend was observed, thus attributing the variation to slight experimental 

scatter. 

As for the case in Part-I of this study, higher values of biaxial 

extensional rheological parameters were observed in the case of 

Jagger flour, indicating its superior breadmaking potential 

(Dobraszczyk et al., 2003). Inability of different lipid types and free 

fatty acids to cause any effect on biaxial extensional rheology, 

demonstrates that their action is independent of rheology of the 

Table 3 

Mean bubble inflation rheological responses of soft wheat flour doughs to different 

lipid types and levels, added to defatted flours as percentage of natural flour lipids 

% Of natural flour 

lipid 

Max. stress 

(kPa) 

Max. stress 

(kPa) 

Failure strain 

(Hencky) 

Failure strain 

(Hencky) 


index 


0% 571.62 305.45 a 2.58 0.18 a 2.21 0.12 b,c 

60% 491.68 237.21 a 2.60 0.12 a 2.38 0.064 a,b 

100% 598.62 112.05 a 2.66 0.04 a 2.44 0.06 a 


40% 480.85 67.26 a 2.56 0.04 a 2.18 0.03 c 

200% 660.22 130.74 a 2.64 0.05 a 2.22 0.04 b,c 


60% 936.93 560.68 a 2.74 0.16 a 2.44 0.09 a 

Linoleic acid 

132% 650.21 18.07 a 2.66 0.04 a 2.36 0.10 a,b,c 

Palmitic acid 

132% 598.55 239.23 a 2.62 0.12 a 2.23 0.09 b,c 

Myristic acid 

250% 613.71 90.70 a 2.65 0.03 a 2.32 0.10 b,c 

Over all average 623.58 236.59 2.63 0.10 2.31 0.12 




index 


0% 120.80 12.76 a 1.79 0.00 a 1.76 0.01 a 

40% 125.42 12.14 a 1.85 0.02 a 1.74 0.01 a 

100% 102.40 5.45 a 1.66 0.05 b 1.64 0.04 c 


20% 105.49 3.52 a 1.68 0.02 b 1.66 0.00 b,c 

200% 130.17 1.61 a 1.84 0.02 a 1.72 0.02 a,b 


60% 108.71 27.51 a 1.83 0.02 a 1.73 0.01 a,b 

Linoleic acid 

132% 104.35 12.49 a 1.81 0.06 a 1.75 0.07 a 

Palmitic acid 

132% 116.21 0.52 a 1.82 0.01 a 1.79 0.01 a 

Myristic acid 

250% 115.44 23.14 a 1.78 0.02 a 1.72 0.03 a 

Over all average 114.19 13.69 1.79 0.07 1.72 0.05 


Means with the same letter within columns are not significantly different (p > 0.05).

46 

gluten–starch matrix and that they act purely as surface active 

compounds at these levels of addition. 

4. Conclusion 

This study for the first time demonstrates that the lipids at their 

natural levels do not affect dough rheology and gas cell stabilization 

by the gluten–starch matrix. Nevertheless, their ability to modify 

loaf volume and crumb structure supports the dual film hypothesis 

of Gan et al. (Gan et al., 1990, 1995). It suggests the presence of 

liquid lamellae, providing an independent mechanism of gas cell 

stabilization. The effects of different surface active components 

may be explained by the type of monolayer that they form. 

References 

Baker, J.C., Parker, H.K., Mize, M.D., 1946. Supercentrifugates from dough. Cereal 

Chemistry 23, 539–544. 

DeStefanis, V.A., Ponte Jr., J.G., 1976. Studies on breadmaking properties of wheatflour 

nonpolar lipids. Cereal Chemistry 53, 636–642. 

Dobraszczyk, B.J., Smewing, J., Albertini, M., Maesmans, G., Schofield, J.D., 2003. 

Extensional rheology and stability of gas cell walls in bread doughs at elevated 

temperatures in relation to breadmaking performance. Cereal Chemistry 80, 

218–224. 

Gaines, G.L., 1966. Insoluble monolayers at liquid–gas interfaces. In: Prigogine, I. 

(Ed.), Interscience Monographs on Physical Chemistry. John Wiley and Sons, 

Inc., New York, USA. 

Gan, Z., Angold, R.E., Williams, M.R., Ellis, P.R., Vaughan, J.G., Galliard, T., 1990. The 

microstructure of gas retention of bread dough. Journal of Cereal Science 12,15–24. 

Gan, Z., Ellis, P.R., Schofield, J.D., 1995. Gas cell stabilization and gas retention in 

wheat bread dough. Journal of Cereal Science 21, 215–230. 

B.S. Sroan, F. MacRitchie / Journal of Cereal Science 49 (2009) 41–46 

Gandikota, S., MacRitchie, F., 2005. Expansion capacity of doughs: methodology and 

applications. Journal of Cereal Science 42, 157–163. 

Larsson, K., Quinn, P., Sato, K., Tiberg, F., 2006. Lipids: Structure, Physical Properties 

and Functionality. The Oily Press, Bridgwater, England. 

Lunkenheimer, K., Miller, R., 1987. A criterion for judging the purity of adsorbed 

surfactant layers. Journal of Colloid Interface Science 120, 176–183. 

McCormack, G., Panozzo, J., MacRitchie, F., 1991. Contributions to breadmaking of 

inherent variations in lipid content and composition of wheat cultivars. II. 

Fractionation and reconstitution studies. Journal of Cereal Science 13, 

263–274. 

MacRitchie, F., 1976a. The liquid phase of dough and its role in baking. Cereal 

Chemistry 53, 318–326. 

MacRitchie, F., 1976b. Monolayer compression barrier in emulsion and foam 

stability. Journal of Colloid and Interface Science 56, 53–56. 

MacRitchie, F., 1977. Flour lipids and their effects in baking. Journal of the Science of 

Food and Agriculture 28, 53–58. 

MacRitchie, F., 1978. Differences in baking quality between wheat flours. Journal of 

Food Technology 13, 187–194. 

MacRitchie, F., 1990. Chemistry at Interfaces. Academic Press, Inc., San Diego, CA, 

USA. 

MacRitchie, F., Gras, P.W., 1973. The role of flour lipids in baking. Cereal Chemistry 

50, 292–302. 

Mills, E.N.C., Wilde, P.J., Salt, L.J., Skeggs, P., 2003. Bubble formation and stabilization 

in bread dough. Food and Bioproducts Processing 81, 189–193. 

Paternotte, T.A., Orsel, R., Hamer, R.J., 1993. Interactions between flour proteins and 

flour lipids at the liquid/air interface. In: Gluten Proteins. Association of Cereal 

Research, Detmold, Germany, pp. 207–217. 

Ponte Jr., J.G., DeStefanis, V.A., 1969. Note on separation and baking properties of 

polar and nonpolar wheat flour lipids. Cereal Chemistry 46, 325–329. 

Ross, A., MacRitchie, F., 1995. Interactions of wheat proteins, carbohydrates and 

lipids. In: Gaonkar, A.G. (Ed.), Ingredient Interactions Effects on Food Quality. 

Marcel Deker, New York, USA, pp. 321–356. 

Salt, L.J., Wilde, P.J., Georget, D., Wellner, N., Skeggs, P.K., Mills, E.N.C., 2006. 

Composition and surface properties of dough liquor. Journal of Cereal Science 

43, 284–292.

Expression of globulin-2, a member of the cupin superfamily of proteins with 

similarity to known food allergens, is increased under high temperature 

regimens during wheat grain development 

Susan B. Altenbach *, Charlene K. Tanaka, William J. Hurkman, William H. Vensel 

USDA-ARS Western Regional Research Center, 800 Buchanan Street, Albany, CA 94710, USA 

article info 


Received 27 February 2008 



Keywords: 

Flour quality 

Gene expression 

Proteomics 

Quantitative RT-PCR 


abstract 

Wheat is one of the major crops grown throughout the world 

with a primary use in human nutrition. In 2005, wheat was grown 

on more than 200 million hectares and more than 600 million tons 

of grain were produced worldwide (http://faostat.fao.org/). A large 

variety of breads, noodles and baked goods are made from wheat 

flour because of the unique viscoelastic properties that result when 

the flour is mixed with water. These viscoelastic properties are 

determined largely by the gluten proteins, the major storage 

proteins that comprise 80–85% of wheat flour protein. The gluten 

proteins, consisting of gliadins and glutenin subunits, are characterized 

by unusually high levels of proline and glutamine and have 

very low solubilities in water or salt solutions. These proteins have 

been studied extensively because of their importance in flour 

functionality. The remaining 15–20% of the endosperm protein is 

Abbreviations: 2-DE, two-dimensional gel electrophoresis; 2-DE/MS, twodimensional 

gel electrophoresis/mass spectrometry; DPA, days’ post-anthesis; 

ESTs, expressed sequence tags; MS/MS, tandem mass spectrometry; NPK, 

nitrogen–phosphorous–potassium; qRT-PCR, quantitative reverse-transcriptase 

polymerase chain reaction; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel 

electrophoresis. 


E-mail address: altnbach@pw.usda.gov (S.B. Altenbach). 


doi:10.1016/j.jcs.2008.07.005 





Twenty-three expressed sequence tags (ESTs) from the US spring wheat Butte 86 were identified that 

encode proteins similar to a globulin-2 protein from maize embryos. The ESTs assembled into three 

contigs, two of which include the entire coding region for the mature protein. The encoded proteins 

contain two cupin domains and show significant identities with 7S seed proteins from other species that 

are known or putative food allergens. Quantitative reverse-transcriptase polymerase chain reaction (qRT- 

PCR) with primers specific for two of the sequences demonstrated that the globulin-2 genes are 

expressed late in grain development and that transcript levels increase when grain is produced under 

high temperature conditions. Transcripts were detected in both whole grain and endosperm, but levels 

were significantly higher in whole grain and highest in embryo. In wheat flour, at least 17 protein spots 

that differ in both size and pI were identified as globulin-2 by 2-DE/MS. Seven of the spots increased 

more than 2-fold in relative proportion when grain was produced under high temperature regimens. The 

data suggest that both transcriptional and post-translational mechanisms are involved in the response of 

globulin-2 to high temperatures. 


a heterogeneous mixture of proteins, most of which are soluble in 

water or dilute salt solutions. Within the salt-soluble protein fraction 

are a number of proteins related to the predominant storage 

proteins characterized in seeds from dicotyledonous plants. Some 

of these proteins have been shown to have sedimentation values 

similar to the 7S vicilins from pea and to share antigenic determinants 

with the pea vicilins (Robert et al., 1985). A 7S þ 3S globulin 

fraction characterized by Robert et al. (1985) contained major 

proteins of w75, 55, 36–40, 24,17–20 and 10 kDa when analyzed by 

SDS-PAGE. 

Recently, a 2-DE/MS approach was used to identify more than 

200 proteins contained in a salt-soluble fraction of wheat endosperm 

from the US spring wheat Butte 86 (Vensel et al., 2005). 

Seven spots in the 55 kDa region of the 2-D gels were determined to 

be most similar to a 7S vicilin-like protein characterized from maize 

embryos, referred to as globulin-2 (Wallace and Kriz, 1991). These 

proteins were found in wheat endosperm at 36 DPA but not at 10 

DPA. In a subsequent proteomic analysis of endosperm from 

developing grain, the globulin-2 proteins were shown to accumulate 

during late stages of wheat grain development. Interestingly, 

these proteins also were among a group that increased in relative 

proportion when grain development occurred under a high 

temperature regimen of 37/28 C (day/night) rather than under a 

moderate temperature regimen of 24/17 C(Hurkman et al., 2009). 

Since these data suggest biological roles for the globulin-2 proteins

48 

in the response of the grain to abiotic stress, we surveyed databases 

of wheat expressed sequence tags (ESTs) to identify the complement 

of globulin-2 genes from the US spring wheat Butte 86. 

Additionally, we examined the expression of two of the globulin-2 

genes in developing grain and endosperm under different 

controlled environmental regimens using quantitative RT-PCR and 

evaluated the relative proportions of globulin-2 proteins in flour 

produced from grain subjected to moderate and high temperature 

regimens. 


2.1. Growth of plants and tissue collection 

The US hard red spring wheat Triticum aestivum cv. Butte 86 was 

grown in a climate-controlled greenhouse as described by Altenbach 

et al. (2007). During grain development, plants were placed 

under either a moderate (24/17 C day/night) or a high (37/28 C 

day/night) temperature regimen. Levels of fertilizer (Plantex 

20–20–20 NPK) were adjusted at anthesis so that plants received 

300 mg per day (1 ), 150 mg per day (0.5 )orno(0 ) NPK during 

grain development. Developing grain and endosperm were 

collected at various intervals after anthesis under both temperature 

regimens. Embryo tissue was harvested from 30 DPA grain from 

plants grown under the 24/17 C regimen. Awns, glumes, stems and 

leaves were obtained during middle stages of grain development 

under the moderate regimen. Roots were harvested from seedlings 

grown on water-saturated sand at room temperature for 20 days. 

All tissues were frozen in liquid nitrogen and stored at 80 C 

until use. 

2.2. RNA preparation and qRT-PCR 

Total RNA was isolated from all tissues as described previously 

for endosperm (Altenbach, 1998), treated with RQ1 RNAse-free 

DNAse (Promega, Madison, WI), and further purified by extraction 

with phenol:chloroform:isoamylalcohol (24:24:1) and ethanol 

precipitation. RNA was reverse-transcribed using the QuantiTect 

Reverse Transcription Kit (QIAGEN, Valencia, CA) according to the 

manufacturer’s directions. Amplification reactions were carried out 

in a volume of 25 ml containing cDNA, 0.3 mM of forward and 

reverse primers and SYBR Green Supermix (Biorad Laboratories, 

Hercules, CA) using a BioRad iCycler with an initial denaturation 

step of 95 C for 3 min, followed by 40 cycles of 95 C for 10 s and 

55 C for 45 s. At the end of the PCR cycles, a melting curve was 

generated and analyzed by the iCycler software. 

Oligonucleotide primers for globulin-2 genes were based on 

expressed sequence tags (ESTs) from developing grain of Butte 86 

available from NCBI. Primers were selected using Beacon Designer 

4.0 (Premier Biosoft International, Palo Alto, CA) and synthesized by 

Operon (Huntsville, AL). Forward primer QF123 has the sequence 5 0 

GGTCCAAATGGCTCCGC 3 0 and reverse primer QR123 has the 

sequence 5 0 CAAGAAGGACTGCGGGC 3 0 . Forward primer QF126 has 

the sequence 5 0 GGTCGTCATGCTCCTCAAC 3 0 and reverse primer 

QR126 has the sequence 5 0 GACGCTGAAGAAGGACTGTG 3 0 . Primers 

for an 18S rRNA used as a reference were reported in Altenbach 

et al. (2007). Amplification efficiencies for each primer pair were 

calculated from standard curves generated in three independent 

experiments using the iCycler software. Each standard curve had 

a minimum of five points and R values greater than 0.99. Amplification 

products were evaluated by gel electrophoresis on 4% 

Metaphor agarose gels (VWR International, Brisbane, CA). 

For quantification of transcripts, PCR reactions were carried out 

in triplicate using cDNA from the equivalent of 10 ng RNA from each 

time point. The 18S rRNA served as a reference RNA and was 

amplified in parallel with target genes in all experiments. The Ct 

S.B. Altenbach et al. / Journal of Cereal Science 49 (2009) 47–54 

value was determined for each RNA sample and primer pair by the 

iCycler software. Mean normalized expression and standard errors 

were determined for each target gene relative to the 18S reference 

using the Q-Gene Core Module (available at http://www.genequantification.de/download.html#qgene). 

Amplification efficiencies 

of each primer pair were taken into account for all calculations 

using equation #3 of Muller et al. (2002). Mean normalized 

expression was plotted as a function of chronological age for each 

experiment. 

2.3. Protein extraction, 2-DE and MS/MS analyses 

Endosperm was collected from developing grain and a KClsoluble/methanol-insoluble 

protein fraction was isolated and 

quantified as described previously (Hurkman and Tanaka, 2004; 

Vensel et al., 2005). Equal amounts of protein (18 mg) from each 

time point were separated by 2-DE in triplicate. Gels were digitized 

with a calibrated scanner (UMAX PowerLook III, Dallas, TX) at 

300 dpi using the same setting for all gels and analyzed using 

a computerized image analysis system (Progenesis PG240 version 

2006, Non-Linear Dynamics Limited, Newcastle upon Tyne, UK). 

Normalized volumes (individual spot volume/total spot volume 

100) were determined for each spot, averaged over the three 

gels, and plotted against the chronological age of the grain in DPA. 

Grain produced under different temperature regimens was 

tempered to 15% moisture and milled to flour by the USDA-ARS 

Wheat Quality Laboratory (Manhattan, KS) using a Brabender 

Quadrumat Jr. mill (Hackensack, NJ). Methods for extraction and 

analysis of proteins from flour were the same as those used for 

endosperm. 

For identification of proteins by tandem mass spectrometry 

(MS/MS), spots were excised from a representative gel and reduction, 

alkylation, reagent removal, and tryptic digestion were carried 

out automatically by a DigestPro xyz robot (INTAVIS Bioanalytical 

Instruments AG, Bergisch Gladbach, Germany). The DigestPro 

collection tray containing the tryptic peptides was placed in an 

autosampler that was interfaced with a QSTAR pulsar I hybrid 

quadrupole-TOF instrument (Applied Biosystems/MDX Sciex, 

Toronto, Canada) configured with an ESI source. Acquisition of 

tandem mass spectrometry data was as follows. From an initial 

survey scan of mass range m/z 400–2000, the most abundant 

doubly or triply charged ion above a threshold of 20 counts was 

selected for fragmentation. Collision induced dissociation of the 

mass-selected ion was carried out using UHP nitrogen. Following 

the 3S MS/MS fragmentation period, the MS survey scan was 

repeated until another MS/MS period was triggered. Wiff data files 

were created for each sample by the Analyst QS version 1.1 

software, converted to MGF text files using Mascot Daemon (http:// 

www/matrixscience.com/) and submitted in batch mode to 

a locally installed copy of X!Tandem (Craig and Beavis, 2004) using 

a script provided by Jayson Faulkner (University of Michigan). 

X!Tandem was configured to search files containing sequences 

from the NCBI non-redundant collection: nr-Arabidopsis-thaliana.fasta, 

nr-other-Viridiplantae.fasta, nr-Oryza-sativa.fasta, in addition 

to entries from all proteins in the HarvEST:Wheat version 1.04 

database (http://harvest.ucr.edu/HWheat104.exe), NCBI Triticum 

aestivum: UniGene Build #37, and wEST Database (http://wheat.pw. 

usda.gov/wEST). The results were visualized using a locally 

installed copy of the Global Proteome Machine (GPM) (http:// 

thegpm.org/). Trypsin was selected as the cleavage enzyme. The 

results were searched with a fragment ion mass tolerance of 

1.00 Da and a parent ion tolerance of 0.40 Da. Oxidation of methionine 

was specified as a variable modification (see also: Vensel 

et al., 2002, 2005). Identifications accepted as valid had log e values 

less than 4.0.

3. Results 

3.1. Accumulation of globulin-2 in developing endosperm under 

different temperature regimens 

A number of protein spots were identified as globulin-2 in 

proteomic maps of KCl-soluble/MeOH-insoluble proteins from 

developing endosperm from Butte 86 (Vensel et al., 2005). Six spots 

(referred to as 147, 155, 559, 561, 851, and 852 in Vensel et al., 2005) 

ranged from about 54.8 to 56.1 kDa by 2-DE and had pIs between 

6.2 and 7.0. Another spot (631 in Vensel et al., 2005) was about 

15 kDa and had a pI of about 5.7. Accumulation profiles for these 

seven globulin-2 proteins during grain development under 

moderate and high temperature regimens were summarized as 

part of a large-scale proteomic analysis by Hurkman et al. (2009) 

and are shown in more detail in Fig. 1. Under moderate temperatures 

(24/17 C), these proteins accumulated very late in grain 

development, between 32 and 34 DPA, and each protein spot 

S.B. Altenbach et al. / Journal of Cereal Science 49 (2009) 47–54 49 

encompassed less than 0.2% of the normalized volumes of all spots 

in the KCl-soluble/MeOH-insoluble fraction. When a high temperature 

regimen of 37/28 C was applied from 10 DPA until maturity, 

the proteins accumulated earlier, between 22 and 24 DPA. In 

addition, maximum normalized spot volumes were 1.9- to 4.1-fold 

higher than in grain produced under the 24/17 C regimen (Fig. 1 

and Hurkman et al., 2009). 

3.2. Identification of ESTs encoding globulin-2 from Butte 86 

A collection of contigs assembled from more than 116,000 wheat 

ESTs by Chao et al. (2006) was surveyed to identify sequences that 

encode the wheat globulin-2 proteins. The EST collection includes 

3639 ESTs from Butte 86 developing grain. Information about both 

contigs and ESTs is available at http://wheat.pw.usda.gov/westsql/. 

Three contigs, NSFT03P2_Contig18428, NSFT03P2_Contig17295 and 

NSFT03P2_Contig17366, encode proteins similar to a globulin-2 

characterized previously in maize (Wallace and Kriz, 1991). Amino 

Fig. 1. Accumulation of globulin-2 proteins in endosperm from grain produced under moderate and high temperatures. In each panel the solid lines denote endosperm produced 

under the 24/17 C regimen and the dashed lines denote endosperm produced under the 37/28 C regimen. The high temperature regimen was imposed from 10 DPA to maturity 

and plants were supplied with 0.5 NPK. Bars indicate the standard error for each time point among triplicate gels. Accumulation data for spots 155, 559, 851, 147, 631, 852 and 561 

from Vensel et al. (2005) are presented in panels A, B, C, D, E, F and G, respectively.

50 

acid sequences of the proteins encoded by the contigs are shown in 

Fig. 2. Contig 18428 includes nine ESTs from Butte 86 (BQ806323, 

BQ807065, BQ806087, BQ805228, BQ806420, BQ805507, 

BQ806904, BQ804552, BQ804667) that represent the entire coding 

region. Contig 17295 includes four ESTs from Butte 86 (BQ805639, 

BQ839104, BQ805390, BQ807159) that cover the 5 0 portion of the 

coding region and one EST (BQ805520) that represents the 3 0 end. 

Nine ESTs from Butte 86 were included in contig 17366 (BQ839060, 

BQ805920, BQ807178, BQ838786, BQ839055, BQ838678, BQ807089, 

BQ806653, BQ805350) and all nine represent the 5 0 half of the 

coding region. It is likely that this contig is missing the 3 0 end of the 

coding region. 

When evaluated with the Signal P algorithm (http://www.cbs. 

dtu.dk/services/SignalP/), proteins encoded by the three contigs 

are predicted to contain signal peptide cleavage sites between the 

alanine and the serine residues at positions 26 and 27 in contigs 

17295 and 17366 and positions 20 and 21 in contig 18428 (Bendtsen 

et al., 2004), suggesting that these proteins enter the secretory 

pathway (Fig. 2). Contigs 18428 and 17295 encode proteins with 

predicted molecular weights of 53,495 and 53,773, respectively, 

excluding the signal peptide. Both proteins contain abundant 

arginine, glycine and glutamine residues. An InterProScan (http:// 

www.ebi.ac.uk/InterProScan/) showed that each protein contains 

two cupin domains (Fig. 2) that are able to form barrel-like 

structures typical of the cupin superfamily of proteins. 

Since many plant proteins within the cupin superfamily are food 

allergens (Mills et al., 2002), the protein encoded by contig 17295 

was used to search the Food Allergy Research and Resource Program 

(FARRP) Protein AllergenOnline Database version 8.0 (http://www. 

allergenonline.com/databasefasta.asp). A sliding 80-mer search 

revealed matches of >35% identity between this globulin-2 and 18 

7S proteins that are known or putative allergens (Table 1). These 

include proteins from sesame, various tree nuts (English walnut, 

black walnut, hazelnut, cashew) and legumes (lentil, pea, soybean, 

peanut). Current Codex Alimentarius guidelines indicate that 

proteins with identities to known allergens that are >35% over 80 

amino acid regions may be allergenic (Goodman, 2006). 


3.3. Selection of primers for qRT-PCR and analysis of gene 

expression 

ESTs BQ806323, BQ805639 and BQ839060 from cv Butte 86 

represent the first half of the coding sequences of contigs 18428, 

17295 and 17366, respectively, and were used for the design of 

gene-specific primers. Primer QF123 was a perfect match with 

BQ806323, but had only 13/17 nucleotides identical to BQ839060 

and BQ805639 (Fig. 3). Primer QR123 was a reverse complement of 

BQ806323, but had only 13/17 nucleotides that complemented 

either BQ839060 or BQ805639. QF123/QR123 amplified a 110 bp 

fragment that was verified to correspond to BQ839060 by digestion 

with AlwI, HaeII, and HpaII. The average amplification efficiency of 

QF123/QR123 was 96.7%. Primer QF126 was a perfect match with 

both BQ805639 and BQ839060 but had only 16/19 nucleotides in 

common with BQ806323 (Fig. 3). Primer QR126 was a reverse 

complement of BQ805639, but had only 18 and 17 of 20 residues 

that complemented BQ806323 and BQ839060, respectively. QF126/ 

QR126 amplified a 100 bp fragment that was verified to correspond 

to BQ805639 by digestion with HpaII, MboII, AciI, AlwI and HaeII. 

The average amplification efficiency of QF126/QR126 was 93.7%. 

Transcript accumulation for the two globulin-2 genes was 

assessed in developing grain produced under either moderate or 

high temperature regimens from anthesis to maturity (Fig. 4). In the 

absence of post-anthesis NPK (Fig. 4A, B), transcripts corresponding 

to both genes were first detected at 11 DPA under moderate 

temperatures and reached maximum levels at 26 DPA. Under the 

high temperature regimen, transcripts began to accumulate earlier 

in grain development, by 6 DPA, and reached maximum levels at 22 

DPA. Changes in the timing of globulin-2 transcript accumulation 

were consistent with changes in the timing of developmental 

events due to high temperature (Altenbach et al., 2003) and with 

changes in the timing of expression of other genes (Altenbach and 

Kothari, 2004; Altenbach et al., 2007, 2008). Maximum levels of 

transcripts corresponding to BQ806323 and BQ805639 were 4.1and 

4.9-fold higher, respectively, under high temperatures than 

under moderate temperatures. In a second growth experiment, 

Fig. 2. Comparison of protein sequences encoded by NSFT03P2_Contig18428, NSFT03P2_Contig17295 and NSFT03P2_Contig17366. Identical amino acid residues in each protein are 

enclosed in boxes. Brackets denote portions of the coding regions in each contig that correspond to ESTs from Butte 86. The arrow indicates putative signal peptide cleavage sites. 

Cupin domains in each protein are shown in bold type and putative N-linked glycosylation sites are indicated by asterisks.

Table 1 

Comparison of globulin-2 encoded by NSFT03P2_Contig17295 with known and 

putative allergens in the FARRP Protein AllergenOnline Database Version 8.0 using 

a sliding 80-mer search 

GenBank 

accession # 

Source Best % 

identity 

13183177 Sesame indicum (sesame) 51.90 285 

6580762 Juglans regia (English walnut) 48.10 269 

31321944 Juglans nigra (black walnut) 47.50 251 

19338630 Corylus avellana (hazelnut) 47.50 204 

21666498 Anacardium occidental (cashew) 43.80 173 

21914823 Anacardium occidental (cashew) 43.80 179 

29539109 Lens culinaris (lentil) 42.51 112 

46560474 Arachis hypogaea (peanut) 41.29 62 

42414627 Pisum sativum (pea) 41.24 97 

42414629 Pisum sativum (pea) 41.24 105 





29539111 Lens culinaris (lentil) 40.00 123 

256427 Glycine max (soybean) 38.79 45 



# Hits > 35% a 

Proteins with >35% identity over 80 amino acids to wheat globulin-2 are shown. 

a of 426 80-mers. 

plants were supplied with post-anthesis NPK (Fig. 4C, D). In this 

experiment, transcripts for BQ806323 were first detected by 12 DPA 

and reached maximum levels at 34 DPA under moderate temperatures 

(Fig. 4C) while transcripts for BQ805639 were detectable at 

low levels at 7 DPA and reached maximum levels at 34 DPA 

(Fig. 4D). Under the high temperature regimen, transcripts for both 

genes were first detected at 7 DPA and reached maximum levels at 

18 DPA, the last time point sampled. Maximum levels of BQ806323 

and BQ805639 transcripts were 4.1- and 2.1-fold higher, respectively, 

under the high temperature regimen than under moderate 

temperatures. 

The accumulation of globulin-2 transcripts also was assessed in 

developing endosperm (Fig. 4E, F). Transcripts for both BQ806323 

and BQ805639 were detected at low levels at both 7 and 14 DPA 

under moderate temperatures and reached maximum levels at 34 

DPA. When high temperatures were applied from anthesis to 

maturity, transcripts for both genes were detected at the first time 

point sampled at 5 DPA and increased to maximum levels by 20 

DPA. Transcript levels at the 20 DPA time point under the high 

temperature regimen were 3.9- and 5.3-fold higher for BQ806323 

Fig. 3. Comparison of the nucleotide sequences of BQ806323, BQ839060 and 

BQ805639 in regions used for gene-specific primers. QF123 is identical to nucleotides 

481–497 of BQ806323 and QR123 is the reverse complement of nucleotides 573–590 

of BQ806323. QF126 is identical to nucleotides 532–550 of BQ805639 and QR126 is the 

reverse complement of nucleotides 612–631 of BQ805639. Nucleotides that differ in 

other ESTs are underlined. 


and BQ805639, respectively, than at the 34 DPA time point under 

moderate temperatures. 

In the experiments shown in Fig. 4, globulin-2 transcript levels 

relative to the 18S reference RNA were higher in whole grain than in 

endosperm. Fig. 5 shows a direct comparison of globulin-2 transcript 

levels in whole grain at 30 DPA and in embryo and endosperm 

dissected from 30 DPA grain produced under the moderate 

temperature regimen. In this analysis, transcripts corresponding to 

BQ806323 and BQ805639 were 10.1- and 7.8-fold higher, respectively, 

in whole grain than in endosperm. BQ806323 and BQ805639 

transcripts were most abundant in embryo, about 2.5- and 2.6-fold 

higher, respectively, in embryo than in whole grain. Consistent with 

these findings is the fact that a significant proportion of the ESTs 

that comprise the globulin-2 contigs were derived from an embryo 

cDNA library, including 9/22 for contig 18428, 4/11 for contig 17295 

and 2/11 for contig 17366. The collection of ESTs assembled into 

contigs by Chao et al. (2006) included 4147 sequences from endosperm 

and 1916 sequences from embryo in addition to the 3639 

from Butte 86 whole grain. RNA prepared from awns, glumes, 

stems, leaves and roots also was analyzed in the same experiment. 

Transcripts corresponding to BQ805639 were detected at very low 

levels in roots (Fig. 5B), but were not detectable in awns, glumes, 

stems or leaves. BQ806323 was not expressed in any of the nongrain 

samples (data not shown). 

3.4. Globulin-2 proteins in flour produced under moderate 

and high temperatures 

A proteomic analysis was undertaken to determine the levels of 

globulin-2 proteins in white flour milled from grain produced 

under moderate and high temperatures with and without postanthesis 

NPK. This analysis revealed a complex collection of globulin-2 

proteins in the KCl-soluble/MeOH-insoluble fraction 

(Fig. 6A–C). Seventeen spots in the 52–55 kDa region of the 2-D gels 

were identified as globulin-2. Spots 219, 209, 199, 200, 193, 190 and 

258 in Fig. 6C correspond to spots 147, 155, 559, 561, 851, 852 and 

175 identified in Vensel et al. (2005). MS/MS data used to identify 

the remaining spots (192, 202, 205, 221, 222, 228, 236, 239, 320, 

and 210) as globulin-2 proteins are summarized in Table S1. One of 

the spots (210) contained more than one protein and was excluded 

from further analysis. Collectively, the 16 remaining spots comprise 

about 2.9 and 3.1% of the total protein in the fraction when grain 

was produced under a moderate temperature regimen with or 

without post-anthesis NPK, respectively. When grain was produced 

under a high temperature regimen from anthesis to maturity, the 

16 spots comprised a larger percentage of the total protein in the 

fraction, 5.2% when plants were supplied with NPK and 5.0% when 

plants did not receive post-anthesis NPK. Of the 16 spots, nine 

showed statistically significant increases with high temperatures 

(P < 0.05) under at least one of the fertilizer regimens. Seven spots 

increased more than 2-fold in response to high temperature under 

both fertilizer regimens (192, 193, 199, 202, 205, 209 and 320) 

(Fig. 6C, D). The largest change with high temperature was a 8.3fold 

increase detected for spot 199 when grain was produced 

without NPK. When grain was produced with NPK, the largest 

response was a 4.8-fold increase for spot 320. Four spots (200, 221, 

228 and 239) showed small decreases with high temperatures. The 

other three spots identified as globulin-2 proteins encompassed 

similar proportions of the KCl-soluble/MeOH-insoluble flour fraction 

under all four environmental regimens (222, 236 and 258). 

Five of the seven spots that increased more than 2-fold (193, 199, 

202, 209 and 320) also increased in flour produced under high 

temperatures in a second growth experiment in which plants were 

supplied with 0.5 NPK (data not shown).

52 

Fig. 4. Accumulation of transcripts corresponding to BQ806323 (A, C, E) and BQ805639 (B, D, F) in developing grain (A–D) or endosperm (E, F) produced under a 24/17 C regimen 

(solid lines) or a 37/28 C regimen (dashed lines). Plants in A and B did not receive post-anthesis NPK, while plants in C and D received 1 NPK and plants in E and F received 0.5 

NPK. Bars indicated the standard error among triplicate reactions. 

4. Discussion 

In dicots, the 7S globulins are embryo proteins that serve 

a storage function and are the primary source of amino acids for the 

developing seedling. In wheat, the 7S globulins make relatively 

small contributions to the storage reserves of the grain and the 

primary storage proteins are the gliadins and glutenin subunits. 

While Halford and Shewry (2007) suggested that globulin proteins 


in cereals have no biological role other than storage, data presented 

here suggests that some of the wheat globulin-2 proteins may be 

involved in the response of the developing grain to high temperature 

stress. Transcript profiles demonstrate that two closely-related 

globulin-2 genes are expressed in whole grain and endosperm late 

in grain development and that levels of transcripts increase in 

response to high temperatures. Globulin-2 proteins also encompass 

a greater proportion of the total KCl-soluble/MeOH-insoluble 

Fig. 5. Accumulation of transcripts corresponding to BQ806323 (A) and BQ805639 (B) in embryo, endosperm and whole grain, all at 30 DPA, and in root. Bars indicate the standard 

error among triplicate reactions.

protein in endosperm and flour from grain produced under high 

temperature conditions. 

These findings are consistent with previous work on the 

homologous globulin-2 protein from maize. In an early study, Kriz 

and Wallace (1991) reported maize variants lacking globulin-2 and 

suggested that this protein was not essential for seed development, 

maturation or germination. More recent studies using proteomic 

approaches have shown that globulin-2 proteins respond to stress 

in maize. Kollipara et al. (2002) found that globulin-2 was among 

the proteins up-regulated in desiccation-tolerant maize hybrids 

and Labra et al. (2006) noted that a number of proteins identified as 

globulin-2 were up-regulated in maize plantlets germinated in the 

presence of potassium dichromate. 

Analysis of wheat ESTs indicates that the wheat globulin-2 

proteins are encoded by a simple gene family consisting of three 

members. Yet at least 17 protein spots have been identified as 

globulin-2 in wheat flour using 2-DE/MS, suggesting that the 

proteins undergo post-translational modifications that result in 

changes in size and pI. It is likely that the wheat globulin-2 proteins, 

like 7S seed storage proteins from legumes, are glycosylated. Robert 

et al. (1985) noted charge heterogeneity among proteins isolated 

from a 7S þ 3S fraction of globulins from wheat grain in the 55 kDa 

region of 2-D gels and observed that 55 kDa proteins also were 

bound to a concanavalin A sepharose column. Additionally, proteins 

encoded by contigs 18428, 17295 and 17366 contain putative sites 

for N-linked glycosylation, defined as N–X–T, where X can be any 

amino acid except P (Fig. 2). Using affinity blotting, Kriz (1989) also 

noted that the homologous globulin-2 protein from maize was able 


Fig. 6. 2-DE analysis of proteins from flour produced under a 24/17 C (A) and a 37/28 C (B) regimen with 1 NPK. Boxes indicate regions of each gel that contain spots identified 

by MS/MS as globulin-2. The boxed region of the gel from panel B is enlarged in panel C and spots identified as globulin-2 are numbered. Numbers that are underlined refer to spots 

that increase in relative proportion under high temperature conditions. Panel D compares the average normalized volumes for each globulin-2 spot from plants subjected to the 24/ 

17 C and 37/28 C regimens, with and without post-anthesis NPK. The solid and open bars denote the 24/17 C regimens with or without NPK, respectively. The hatched and 

stippled bars denote the 37/28 C regimen, with or without NPK, respectively. Asterisks denote changes with high temperature under each NPK regimen that are statistically 

significant (P < 0.05). 

to bind Con A. Some 7S globulins also undergo proteolytic processing 

(reviewed by Müntz, 1996). It is interesting that not all of 

the globulin-2 protein spots identified by 2-DE increased in 

response to high temperatures. In fact, decreases in the levels of 

several globulin-2 proteins were detected in flour. Thus, both 

transcriptional and post-translational mechanisms may be 

involved in the response of globulin-2 genes to high temperatures 

in the developing grain. 

Transcripts for globulin-2 were present at lower levels in the 

endosperm than in whole grain. Halford and Shewry (2007) noted 

that 7S globulins from cereals are expressed predominantly in the 

embryo and in the aleurone layer of the endosperm. Our analyses 

did not distinguish the aleurone tissue, which is a single cell layer in 

wheat, from the starchy endosperm tissue. Nonetheless, transcript 

levels clearly increased in both the whole grain and the endosperm 

samples. Globulin-2 proteins were present in endosperm tissue 

extruded from grain from which the embryo had been removed. In 

addition, globulin-2 proteins were present in milled white flour 

that generally contains minimal amounts of germ and bran. 

Increases in the relative proportion of globulin-2 proteins in 

flour produced under high temperatures may have important 

implications given the potential allergenicity of these proteins. The 

KCl-soluble/MeOH-insoluble proteins account for about 11% of the 

total flour protein (Hurkman and Tanaka, 2004) and globulin-2 

proteins comprise as much as 6% of this fraction under high 

temperature conditions. Wheat is among the top eight foods 

responsible for human food allergies. Allergies to wheat are characterized 

by a broad spectrum of symptoms including urticaria,

54 

gastrointestinal distress, asthma and anaphylaxis and are due to 

sensitivities to a variety of proteins in both the gluten protein and 

albumin/globulin fractions (Battais et al., 2005). Since the wheat 

globulin-2 shows significant identity to other proteins that are 

known or putative allergens, further studies on the allergenicity of 

these proteins are warranted. IgE binding studies using sera from 

patients with characterized wheat allergies would be key to 

determining whether the globulin-2 proteins should be classified 

as food allergens. 

An additional question is whether changes in the relative 

proportions of the globulin-2 proteins might alter the functional 

properties of the flour. It is generally accepted that the gluten 

proteins contribute elasticity and extensibility properties essential 

for the formation of wheat flour doughs. Interactions between the 

gluten protein matrix and other flour components also may affect 

rheological properties but have been little studied. Recently, the 

globulin-2 proteins were identified in a proteomic study of a dough 

liquor fraction prepared from wheat flour (Salt et al., 2005). It is 

interesting that the levels of the globulin-2 proteins in the dough 

liquor increased when salt or salt plus ascorbate was included in 

the dough mixture. Salt is frequently included in commercial 

processes to improve breadmaking properties and has been shown 

to influence water absorption, dough development time, dough 

stability and loaf volume. It is not known whether the globulin-2 

proteins are associated with the gluten matrix in the absence of 

salt. Further study is required to determine whether these proteins 

play roles in flour functionality. 


The authors thank Drs. Frances DuPont and Ann Blechl for 

critical review of the manuscript. Mention of a specific product 

does not constitute an endorsement and does not imply a recommendation 

over other suitable products. 

Appendix. Supplementary data 

Supplementary data associated with this article can be found, in 

the online version, at doi:10.1016/j.jcs.2008.07.005. 

References 

Altenbach, S.B., 1998. Quantification of individual low-molecular-weight glutenin 

subunit transcripts in developing wheat grains by competitive RT-PCR. 

Theoretical and Applied Genetics 97, 413–421. 



grain development in a US spring wheat. Journal of Cereal Science 37, 9–20. 

Altenbach, S.B., Kothari, K.M., 2004. Transcript profiles of genes expressed in 

endosperm tissue are altered by high temperature during wheat grain 

development. Journal of Cereal Science 40, 115–126. 


Altenbach, S.B., Kothari, K.M., Tanaka, C.K., Hurkman, W.J., 2007. Genes encoding the 

PR-4 protein wheatwin are developmentally regulated in wheat grains and 

respond to high temperatures during grainfill. Plant Science 173, 135–143. 

Altenbach, S.B., Kothari, K.M., Tanaka, C.K., Hurkman, W.J., 2008. Expression of 9kDa 

non-specific lipid transfer protein genes in developing wheat grains is 

enhanced by high temperatures but not by post-anthesis fertilizer. Journal of 


Battais, F., Courcoux, P., Popineau, Y., Kanny, G., Moneret-Vautrin, D.A., Denery- 

Papini, S., 2005. Food allergy to wheat: differences in immunoglobin Ebinding 

proteins as a function of age or symptoms. Journal of Cereal Science 

42, 109–117. 

Bendtsen, J.D., Nielsen, H., von Heijne, G., Brunak, S., 2004. Improved prediction of 

signal peptides: SignalP 3.0. Journal of Molecular Biology 340, 783–795. 

Chao, S., Lazo, G.R., You, F., Crossman, C.C., Hummel, D.D., Lui, N., Laudencia- 

Chingcuanco, D., Anderson, J.A., Close, T.J., Dubcovsky, J., Gill, B.S., Gill, K.S., 

Gustafson, J.P., Kianian, S.F., Lapitan, N.L.V., Nguyen, H.T., Sorrells, M.E., 

McGuire, P.E., Qualset, C.O., Anderson, O.D., 2006. Use of a large-scale Triticeae 

expressed sequence tag resource to reveal gene expression profiles in hexaploid 

wheat (Triticum aestivum L.). Genome 49, 531–544. 

Craig, R., Beavis, R.C., 2004. TANDEM: matching proteins with mass spectra. 

Bioinformatics 20, 1466–1467. 

Goodman, R.E., 2006. Practical and predictive bioinformatics methods for the 

identification of potentially cross-reactive protein matches. Molecular Nutrition 

and Food Research 50, 655–660. 

Halford, N.G., Shewry, P.R., 2007. The structure and expression of cereal storage 

protein genes. Plant Cell Monographs 8, 195–218. 

Hurkman, W.J., Tanaka, C.K., 2004. Improved methods for separation of wheat 

endosperm proteins and analysis by two-dimensional gel electrophoresis. 


Hurkman, W.J., Vensel, W.H., Tanaka, C.K., Whitehand, L., Altenbach, S.B., 2009. 

Effect of high temperature on albumin and globulin accumulation in the 

endosperm proteome of the developing wheat grain. Journal of Cereal Science 

49, 12–23. 

Kollipara, K.P., Saab, I.N., Wych, R.D., Lauer, M.J., Singletary, G.W., 2002. Expression 

profiling of reciprocal maize hybrids divergent for cold germination and 

desiccation tolerance. Plant Physiology 129, 974–992. 

Kriz, A.L., 1989. Characterization of embryo globulins encoded by the maize Glb 

genes. Biochemical Genetics 27, 239–251. 

Kriz, A.L., Wallace, N.H., 1991. Characterization of the maize globulin-2 gene and 

analysis of two null alleles. Biochemical Genetics 29, 241–254. 

Labra, M., Gianazza, E., Waitt, R., Eberini, I., Sozzi, A., Regondi, S., Grassi, F., Agradi, E., 

2006. Zea mays L. protein changes in response to potassium dichromate 

treatments. Chemosphere 62, 1234–1244. 

Mills, E.N.C., Jenkins, J., Marigheto, N., Belton, P.S., Gunning, A.P., Morris, V.J., 2002. 

Allergens of the cupin superfamily. Biochemical Society Transactions 30, 

925–929. 

Müntz, K., 1996. Proteases and proteolytic cleavage of storage proteins in developing 

and germinating dicotyledonous seed. Journal of Experimental Botany 47, 

605–622. 

Muller, P.Y., Janovjak, H., Miserez, A.R., Dobbie, Z., 2002. Processing of gene 

expression data generated by quantitative real-time RT-PCR. BioTechnology 32, 

1372–1379. 

Robert, L.S., Adeli, K., Altosaar, I., 1985. Homology among 3S and 7S globulins from 

cereals and pea. Plant Physiology 78, 812–816. 

Salt, L.J., Robertson, J.A., Jenkins, J.A., Mulholland, F., Mills, E.N.C., 2005. The identification 

of foam-forming soluble proteins from wheat (Triticum aestivum) 

dough. Proteomics 5, 1612–1623. 

Vensel, W.H., Harden, L., Tanaka, C.K., Hurkman, W.J., Haddon, W.F., 2002. Identification 

of wheat endosperm proteins by MALDI mass spectrometry and LC-MS/ 

MS. Journal of Biomolecular Techniques 13, 95–100. 

Vensel, W.H., Tanaka, C.K., Cai, N., Wong, J.H., Buchanan, B.B., Hurkman, W.J., 2005. 

Developmental changes in the metabolic protein profiles of wheat endosperm. 

Proteomics 5, 1594–1611. 

Wallace, N.H., Kriz, A.L., 1991. Nucleotide sequence of a cDNA clone corresponding 

to the maize globulin-2 gene. Plant Physiology 95, 973–975.

Biochemical markers: Efficient tools for the assessment of wheat grain tissue 

proportions in milling fractions 

Youna Hemery a , Valérie Lullien-Pellerin a , Xavier Rouau a , Joël Abecassis a , Marie-Françoise Samson a , 

Per Åman b , Walter von Reding c ,Cäcilia Spoerndli c ,Cécile Barron a, * 

a 

INRA, UMR 1208 ‘‘Agropolymers Engineering and Emerging Technologies’’, INRA-CIRAD-UMII-Supagro, F-34000 Montpellier, France 

b 

Department of Food Science, Swedish University of Agriculture Science (SLU), Box 7051, S-750 07 Uppsala, Sweden 

c 

Bühler AG, CH-9240 Uzwil, Switzerland 

article info 



Received in revised form 3 July 2008 


Keywords: 

Wheat 

Grain 

Bran 

Aleurone 

Outer layers 

Fractionation 

Milling 

Processes 


abstract 

Numerous epidemiological studies have demonstrated the 

health benefits of consuming more whole-grain foods (Jacobs and 

Steffen, 2003; Liu, 2007). However, all the wheat grain parts are not 

health-promoting, for example the outermost parts have been 

Abbreviations: ARs, alkylresorcinols; Cr, crousty cultivar; FAt, 4-O-8 0 , 5 0 -5 00 

dehydrotriferulic acid (ferulic acid trimer); P2O5, phosphorus pentaoxyde; p-CA, 

para-coumaric acid; Tg, Tiger cultivar; WGA, wheat germ agglutinin. 

* Corresponding author. Tel.: þ33 4 99 61 31 04; fax: þ33 4 99 61 30 76. 

E-mail address: barron@supagro.inra.fr (C. Barron). 


doi:10.1016/j.jcs.2008.07.006 





To produce safe and healthy whole wheat food products, various grain or bran dry fractionation 

processes have been developed recently. In order to control the quality of the products and to adapt these 

processes, it is important to be able to monitor the grain tissue proportions in the different milling 

fractions produced. Accordingly, a quantitative method based on biochemical markers has been developed 

for the assessment of grain tissue proportions in grain fractions. Grain tissues that were quantified 

were the outer pericarp, an intermediate layer (including the outer pericarp, the testa and the hyaline 

layer), the aleurone cell walls, the aleurone cell contents, the endosperm and the germ, for two grain 

cultivars (Tiger and Crousty). Grain tissues were dissected by hand and analysed. Biochemical markers 

chosen were ferulic acid trimer, alkylresorcinols, para-coumaric acid, phytic acid, starch and wheat germ 

agglutinin, for outer pericarp, intermediate layer, aleurone cell walls, aleurone cell contents, endosperm 

and germ respectively. The results of tissue quantification by hand dissection and by calculation were 

compared and the sensitivity of the method was regarded as good (mean relative errors of 4% and 8% for 

Crousty and Tiger outer layers respectively). The impact of the analytical variability (maximum 13% 

relative error on coarse bran) was also regarded as acceptable. Wheat germ agglutinin seems to be 

a promising marker of wheat germ: even if the quantification method was not able to quantify the germ 

proportions in milling fractions, it was able to classify these fractions according to their germ content. 

The efficiency of this method was tested, by assessing the grain tissue proportions of fractions exhibiting 

very different compositions such as flour, bran and aleurone-rich fractions obtained from three different 

grain or bran dry fractionation processes (conventional milling, debranning process, production of 

aleurone-rich fractions from coarse bran). By calculation of the composition of the different products 

generated, it was possible to study the distribution of the different tissues among fractions resulting from 

the different fractionation processes. This quantitative method is thus a useful tool for the monitoring 

and improvement of processes, and allows the effects of these processes to be understood and their 

adaption to reach the objectives. 


shown to concentrate the majority of the grain contaminants, like 

microorganisms, mycotoxins, pesticide residues and heavy metals 

(Aureli and D’Egidio, 2007; Fleurat-Lessard et al., 2007; Laca et al., 

2006). On the other hand, the wheat aleurone layer has been shown 

to have great nutritional interest, and to concentrate most of the 

minerals and vitamins of the wheat grain (Pomeranz, 1988). Buri 

et al. (2004) reported that wheat aleurone layer contains interesting 

proportions of proteins, b-glucans, phenolic compounds and 

other phytochemicals (lignans, sterols). Zhou et al. (2004) pointed 

out that antioxidant, including phenolic acids, are concentrated in 

the aleurone layer of wheat bran, and Mateo Anson et al. (2008) 

showed that the higher the proportion of aleurone material in 

wheat fractions, the higher the antioxidant capacity observed for

56 

these fractions. Amrein et al. (2003) found that aleurone-rich 

fractions exhibited better in vitro digestibility and colonic fermentability 

than wheat bran, and Bach Knudsen et al. (1995) 

observed that the digestibility of minerals, protein and non-starch 

polysaccharides is much higher in bran fractions rich in aleurone 

than in fractions rich in pericarp and testa. These results suggest 

that it could be interesting to produce aleurone-rich fractions for 

use as food ingredients. As a consequence, new processes are 

developed in order to exploit all the nutritional benefits of whole 

grain and to produce new wheat foods and wheat-based ingredients 

with enhanced nutritional quality (Hemery et al., 2007). For 

example, depending on the desired product, a process can aim at 

discarding the pericarp to obtain whole grains containing less 

contaminants, while other processes may be developed in order to 

produce bran fractions highly concentrated in aleurone material 

(Buri et al., 2004). It is often difficult to exactly monitor the 

distribution of the different grain tissues among fractions during 

processing, as no simple method exists to quantify the respective 

proportions of these tissues in fractions. However, the monitoring 

of tissue proportions in the different fractions is essential, as it 

allows to control the quality of the products and to consequently 

adapt the processes. Therefore quantitative tools are needed. 

Different compounds can be measured to evaluate bran 

contamination in wheat flours and fractions. Ash content and 

measurement of flour colour are widely used in the milling industry 

as indicators of flour purity. The amino acid composition of the 

various tissues was also studied but did not allow the quantification 

of the grain tissues in milling fractions (Jensen and Martens, 1983). 

Some authors (Lempereur et al., 1998; Pussayanawin et al., 1988) 

suggested using the concentration of ferulic acid to quantify bran in 

flours and semolinas, and alkylresorcinols have more recently been 

shown to be good markers of wheat bran content in foods (Chen 

et al., 2004). Such analyses may be useful to evaluate the total outer 

layer content but they do not allow to distinguish between the 

different outer tissues. Another way to evaluate the different grain 

tissues in wheat fractions consists of the use of specific fluorescence 

properties of the outer layers. Indeed, the aleurone cell walls 

display blue fluorescence under UV-light, due to the presence of 

ferulic acid, whereas the pericarp shows green fluorescence under 

blue light (Fulcher et al., 1972; Jensen et al., 1982; Symons and 

Dexter, 1996). Based on these fluorescence properties, commercial 

equipment has been developed to determine the amount of aleurone 

in flours. Multispectral fluorescence image analysis of grain 

sections coupled with classification techniques have also been 

developed to more precisely quantify the proportions of the 

different parts of the grain (Baldwin et al., 1997; Courcoux et al., 

2002), but have not yet been applied to powdery samples. These 

imaging methods can be good tools for on-line use, but their main 

disadvantage would be their lack of specificity, as they do not allow 

to quantify other tissues than aleurone and pericarp. Moreover, all 

these methods allow determination of bran proportions in flours 

during milling, but they may not be adaptable to other fractionation 

systems (such as progressive abrasion or bran fractionation). 

Peyron et al. (2002) and Antoine et al. (2004) carried out 

biochemical analyses of isolated wheat grain tissues and used the 

differences in chemical composition between these tissues to 

assess the histological composition of technological fractions. They 

used compounds such as phenolic acids, phytic acid, and starch as 

biochemical markers. Indeed, these compounds were shown to be 

either present exclusively in one part of the grain (starch in starchy 

endosperm), or present in greater amounts in one particular tissue 

(phytic acid in aleurone cell contents and some phenolic acids in 

the cell walls). These biochemical markers were used to determine 

the amounts of aleurone layer and pericarp in flours and other 

milling fractions (Peyron et al., 2002), and in the samples obtained 

from a wheat bran fractionation process (Antoine et al., 2004). 

Y. Hemery et al. / Journal of Cereal Science 49 (2009) 55–64 

Antoine et al. (2004) used different markers for aleurone cell walls 

and aleurone cell contents to assess the histological composition of 

bran fractions and to evaluate the dissociation and the accessibility 

of aleurone cellular components. This method was reported to 

provide accurate quantification of the histological composition of 

samples and to be versatile, as it can be refined and adapted 

depending on the type of sample analysed, from either bran or 

whole-grain fractionation. However, it did not allow the quantification 

of testa (this tissue was deduced by subtraction and thus was 

perhaps overestimated), and it neither allowed the detection of the 

germ. Having a marker for wheat germ would nevertheless be very 

useful as it is either a part of the grain that needs to be excluded to 

avoid lipid oxidation and rancidity, or a nutritionally interesting byproduct 

that could be followed during fractionation processes in 

order to get germ-rich fractions. 

The aim of this study was first to improve the accuracy of the 

biochemical markers method by introducing new markers of testa 

and germ and by developing a better quantification of aleurone cell 

walls and cell contents, to evaluate the sensitivity of the method, 

and then to validate its use as a tool to assess the grain tissue 

proportions in technological fractions exhibiting very distinct 

compositions, obtained by different grain or bran fractionation 

processes. 


2.1. Preparation of grain tissues and fractions 

2.1.1. Wheat samples 

Two common wheat (Triticum aestivum L.) cultivars differing in 

kernel hardness (hard wheat: cv. Tiger, and soft wheat: cv. Crousty), 

harvested in 2005 in Germany (Tiger) and France (Crousty) were 

used in this study. 

2.1.2. Isolation of wheat grain tissues for biochemical analyses 

The structure and composition of the different wheat grain 

tissues have already been described (Barron et al., 2007; Hemery 

et al., 2007; Surget and Barron, 2005). To obtain isolated grain 

tissues, the grains ends (germ and brush) were removed with 

a razor blade and the remaining parts were immersed in distilled 

water for 12–16 h at 20 C. A crease incision was made and the 

endosperm was removed using a scalpel. The crease area was 

removed and three different strips comprising different tissues 

were separated from these ‘‘whole outer layers’’ using a scalpel. 

Antoine et al. (2003) showed that the outer strip corresponds to 

the outer pericarp (epidermis and hypodermis), the inner strip 

corresponds to the aleurone layer, and the intermediate layer is 

a composite of several tissues (inner pericarp, testa and nucellar 

tissue). Scutellum and embryonic axis were also dissected. Nonadhering 

outer tissues surrounding the embryonic axis were 

removed and then the embryonic axis was separated from dry 

grains with a needle. The scutellum was removed with a scalpel 

after soaking the grain in distilled water for 12–16 h at 20 C. 

These hand-isolated tissues were then used for the biochemical 

analyses. 

2.1.3. Determination of the relative amount of tissues within the 

grain 

Similar hand dissections of 10 grains were performed in triplicate 

for each cultivar, to quantify the relative amounts of the 

different tissues, as described by Barron et al. (2007). All recovered 

tissues were weighed after drying at 25 C over phosphorus pentaoxide 

(P2O5). Outer layer tissues proportions were deduced from 

the combined weights of the aleurone, intermediate layer and 

outer pericarp, dissected from whole outer layers without the 

crease.

2.1.4. Production of the conventional milling fractions 

White flour (0.55% ash content), whole meal flour, fine bran and 

coarse bran fractions were produced using a conventional milling 

process, carried out on a Test-mill in the Department of Safety & 

Quality of Cereals, Federal Research Centre for Nutrition and Food 

(BFEL), Germany. 

2.1.5. Production of bran and flour fractions by peeling, pearling 

and milling 

Eight different bran and flour fractions were provided by Bühler 

A.G. Uzwil, Switzerland. These fractions were obtained by two 

different debranning processes before milling (Fig. 1), by friction 

(peeling) and by abrasion (pearling), according to the published 

Bühler A.G. patent applications (Eugster and Gerschwiler, 2006). 

Peeling fractions were produced using a Bühler DC Peeler taking off 

3–3.5% of the wheat kernel. Peeled wheat kernels were used as raw 

material for the pearling processing, using a Bühler stone polishing 

machine taking off approximately 3% more of the grain, resulting in 

pearled kernels and the pearling fractions. Peeled and pearled 

wheat kernels were tempered for 12–15 h to a moisture content of 

15.5%, and then milled on a Bühler laboratory mill to 76% extraction 

rate flour (based on whole kernels before peeling and pearling) to 

give the 76% flours after peeling and after pearling, and the bran 

fractions after peeling and after pearling. For the production of 

reconstituted ‘whole-grain’ flour (100% extraction of the remaining 

kernel) from peeled and pearled kernels, coarse bran as well as 

bottom flour were reduced to particle sizes 180 mm and Aleurone 1 180 mm 

and Aleurone 2 < 180 mm) using electrostatic separation. The Byproduct 

1 and By-product 2 fractions correspond to the by-products 

of the Aleurone 1 and Aleurone 2 fractions, respectively. 

Peeling fraction 

Milling 

100% flour 

after peeling 

Wheat grains 

Peeling (friction) 

Y. Hemery et al. / Journal of Cereal Science 49 (2009) 55–64 57 

Grains after peeling 

76% flour 

after peeling 

Milling 

Bran after peeling 

2.2. Biochemical analyses 

Dissected tissues were dried at 25 C over P2O5, and then ground 

under cryogenic conditions for 4 min (with a Spex CertiPrep 6750 

lab impact grinder), and dried again before chemical analyses. 

Milling fractions (except flours) were ground in a ball mill (Dangoumeau, 

France) for 4 min. 

2.2.1. Isolation of aleurone cell wall material 

The amount of cell walls in Tiger and Crousty hand-isolated 

aleurone layers was assessed by gravimetric determination of 

insoluble cell wall material after proteolysis, using an adaptation of 

the method described by Brillouet et al. (1988). The dried and 

ground sample was suspended in hexane and centrifuged, the 

supernatant was discarded, these steps were repeated once, and 

the final pellet was air dried. The dried pellet was suspended in 

water with 1.5% SDS and 5 mM Na-bisulfite, a pronase solution was 

added (0.5 ml, 1 mg pronase/ml water), the mixture was agitated 

for 1 h at room temperature and then centrifuged. The pellet was 

washed extensively with water with intermittent centrifugation, 

and then successively suspended in 80% ethanol, absolute ethanol, 

acetone and ether, with centrifugation between two successive 

suspensions. The final pellet corresponding to the aleurone cell 

wall material was dried at 25 C over P2O5 for 72 h, and weighed 

with 0.1 mg accuracy. This protocol was only used to evaluate the 

cell wall content in aleurone layers, and not for biochemical 

analyses. 

2.2.2. Starch 

Total starch content of fractions was measured using Megazyme 

kits (Megazyme International Ireland Ltd., Ireland) according to 

approved AACC method 76-13 (AACC, 2000). Samples were analysed 

in duplicate with c.v.

58 

2.2.4. Phenolic acids 

Ester-linked phenolic acids were saponified under Argon 

(oxygen-free) at 35 C in 2 N sodium hydroxide. An internal standard 

(2,3,5-trimethoxy-(E)-cinnamic acid (TMCA), T-4002, Sigma 

Chemical Co., St Louis, USA) was added before adjusting pH to 2. 

Phenolic acids were then extracted with diethylether and quantified 

by RP-HPLC as described by Antoine et al. (2003). The response 

factors of the para-coumaric acid (p-CA) and the 4-O-8 0 , 5 0 -5 00 

dehydrotriferulic acid (ferulic acid trimer, FAt) relative to the 

internal standard were determined at 320 nm with pure 

compounds. All analyses were performed at least in duplicate, with 

c.v. 12 times more concentrated in the aleurone layer than in the 

outer pericarp, intermediate layer and germ (including scutellum). 

Thus, as p-CA is known to be bound to cell wall polysaccharides 

(Fincher and Stone, 1986; Grabber et al., 2004; Rhodes et al., 2002), 

this compound can be used as a marker of aleurone cell walls. 

However, as the amounts of p-CA found in the outer pericarp and 

the intermediate layer are not negligible, these amounts have to be 

taken into account and subtracted from the total p-CA content of 

a fraction in order not to bias the calculation of the proportion of 

aleurone cell walls in this fraction. 

Antoine et al. (2004) assumed that the sum of the outer pericarp, 

intermediate layer, aleurone and endosperm proportions in 

the studied fractions was equal to 100% (i.e. they supposed that the 

bran fractions did not contain germ), and then deduced the intermediate 

layer proportion by subtraction. If fractions containing germ 

were analysed using this method, the results may be biased and the 

intermediate layer proportion would be overestimated. A specific 

marker is thus needed to allow an accurate quantification of the 

intermediate layer proportion in fractions. Landberg et al. (2008) 

Table 1 

Composition of hand-isolated outer layers and peripheral tissues (reference values) 

Tissues Cultivar FAt 

(mg/g) 

Total ARs 

(mg/g) 

p-CA 

(mg/g) 

Phytic acid 

(mg/g) 

Maximum c.v. 10% 10% 6% 9% 6% 

Outer pericarp Tiger 1.35 0.08 0.01 – – 

Crousty 1.15 0.10 0.02 – – 

Intermediate 

layer a 

Tiger 0.06 16.4 0.11 – – 

Crousty 0.04 16.2 0.09 – – 

Aleurone layer Tiger 0.12 0.03 0.29 152.3 – 

Crousty 0.10 nd 0.22 156.3 – 

Starch 

(% w/w) 

Endosperm Tiger nd 0.02 nd – 78.3 

Crousty nd 0.01 nd – 78.1 

Total outer 

layers 

Tiger 0.34 4.07 0.16 70.7 – 

Crousty 0.28 3.93 0.13 78.4 – 

Whole grains Tiger 0.04 0.47 0.03 13.7 644.4 

Crousty 0.03 0.42 0.02 11.8 656.5 

nd: non-detected; FAt: ferulic acid dehydrotrimer; ARs: alkylresorcinols; p-CA: paracoumaric 

acid. 

a Composed of the hyaline layer þ testa þ inner pericarp.

studied the location of alkylresorcinols in cereal grains and found 

that more than 99% of wheat grain total ARs are concentrated in 

the hand isolated so-called intermediate layer, and more precisely 

in the cuticle of the testa, showing that these phenolic lipids could 

be used as specific biochemical markers. The total amount of 

ARs was therefore chosen as a new marker of intermediate layer. 

Wheat germ agglutinin is a lectin found exclusively in wheat germ 

(Miller and Bowles, 1982). It was therefore selected as a potential 

marker of wheat germ in milling fractions and was used as a qualitative 

indicator to classify milling fractions according to their germ 

content. 

3.1.2. Quantification of tissue proportions and validation by 

comparison with hand-dissection 

After selection of the biochemical markers and assessment 

of the biochemical composition of isolated tissues and fractions, 

the grain tissue proportions can be estimated. By combining the 

biochemical marker contents of hand-isolated grain tissues and 

the marker contents of the different technological fractions, the 

following relations allow calculation of the grain tissue proportions 

within fractions. 

%p ¼½FAtŠF =½FAtŠp 100 

%i ¼½ARsŠF =½ARsŠi 100 

%aw ¼ ½p-CAŠF %p ½p-CAŠp %i ½p-CAŠi %ac ¼ ½Phytic acidŠF =½Phytic acidŠac 100 

%e ¼½StarchŠF =½StarchŠe 100 

with: 

½p-CAŠ aw ¼½p-CAŠ a =½awŠ a 

½Phytic acidŠ ac ¼½Phytic acidŠ a =½acŠ a 

. 

½p-CAŠ aw 100 

where: 

(%p), (%i), (%ac), (%aw), and (%e): proportion of outer pericarp, 

intermediate layer, aleurone cell content, aleurone cell walls, and 

starchy endosperm in studied fractions. 

[FAt]: ferulic acid dehydrotrimer content; [ARs]: total alkylresorcinols 

content; [p-CA]: para-coumaric acid content; [phytic 

acid]: phytic acid content; [starch]: starch content, in studied 

fractions (F), outer pericarp (p), intermediate layer (i), total aleurone 

(a), aleurone cell walls (aw), aleurone cell content (ac), and 

starchy endosperm (e). 

[aw]a and [ac]a: concentration of cell walls and cell contents in 

the aleurone layer. 

The approximate proportions of germ in fractions (%g) was 

calculated as follows: 

%g ¼½WGAŠ F =½WGAŠ grain 

½germŠ grain 

where [WGA]F and [WGA]grain are the WGA content in fractions and 

ground whole grain respectively, and [germ]grain is the proportion 

of germ (scutellum plus embryonic axis) obtained by hand 

dissection of grains. 

In order to separately quantify the proportions of aleurone cell 

walls and cell contents in the different fractions, the results have to 

be calculated in relation to the amount of markers found in aleurone 

cell walls and aleurone cell contents (expressed in mgp-CA/ 

galeurone cell walls and in mgphytates/galeurone cell contents), and not in 

relation to the amount of markers found in the whole aleurone 

layer (expressed in mgmarker/galeurone), otherwise these proportions 

would be over-estimated. Thus, the amount of aleurone cell walls 

was assessed by gravimetric determination of insoluble cell wall 

material. Crousty and Tiger aleurone layers were found to contain 

42.4% 2.5% and 48.0% 2.6% cell walls, respectively. This high 

proportion of cell walls in the aleurone layer corresponds to 


microscopy observations carried out in previous studies (Antoine 

et al., 2003; Bacic and Stone, 1981; Evers and Reed, 1988; Rhodes 

et al., 2002), that showed thick aleurone cell walls accounting for an 

important proportion of aleurone cell total volume. 

To test the efficiency of the described tissue proportion calculation 

method, the composition of whole grains and whole outer 

layers was calculated and compared to the composition assessed by 

hand dissection and weighing. The comparison of the results 

obtained for Crousty and Tiger grains by both methods (markers 

and dissection) is shown in Table 2. Almost no difference was 

found between the endosperm amounts, and the total outer layer 

proportions obtained by both methods were very close (0.6% and 

0.1% absolute error for Crousty and Tiger grains respectively). The 

proportions of peripheral tissues were found to be a little more 

variable (from 0.4% to 1.4% absolute error) depending on the 

method used, but these variations in peripheral tissue proportions 

are due to the fact that these tissues are present in small amounts in 

the ground grains analysed, and that biochemical analyses are less 

precise on small amounts. These results show that the biochemical 

markers method allows the distinct quantification of tissues 

proportions in wheat fractions that contain very few peripheral 

layers, with an absolute error of 1.5%. About 2.5% of the wheat 

grain was not quantified with this method. This proportion is 

comparable to the proportion of germ found in wheat grain by 

hand-dissection. Fig. 2 presents the tissue proportions obtained for 

Crousty and Tiger whole outer layers. The tissues proportions 

calculated using the biochemical markers are very close to those 

assessed by hand dissection. The mean relative errors are 4% and 8% 

for Crousty and Tiger respectively, and the total of the tissue 

proportions are 98% and 96%. These results show that this method 

allows effective quantification, and suggest that it may be very well 

adapted for the assessment of grain tissue proportions in fractions 

rich in bran tissues. 

3.1.3. Impact of analytical variability and possible simplification of 

the method 

This method is based on the measurement of 5 different 

compounds (FAt, ARs, p-CA, phytic acid and starch) on two sample 

sets (i.e. the hand-isolated reference tissues and the technological 

fractions), it implies therefore 10 quantitative determinations. The 

random analytical error on the tissue proportions was calculated 

for Tiger and Crousty coarse bran, by taking into account the 

reproducibility (i.e. the mean c.v. of the dosage) of each measurement. 

Coarse bran was chosen for the calculation of variability 

because this fraction contains all of the grain tissues. If these 

calculations had been carried out for other samples, the results 

could have been different as the random analytical error depends 

on the composition of the studied fraction, due to the fact that the 

reproducibility varies from one dosage to another. The effects on 

Table 2 

Comparison of Crousty and Tiger grains composition (%), using either biochemical 

markers or hand-dissection 

Grains tissues Crousty Tiger 

Markers Dissection Markers Dissection 

Total outer layers 13.7 13.1 14.7 14.6 

Outer pericarp 2.7 3.5 3.1 4.3 

Intermediate layer 2.6 3.2 2.9 3.3 

Aleurone cell walls 4.1 2.7 4.1 3.4 

Aleurone cell contents 4.4 3.7 4.7 3.6 

Starchy endosperm 84.1 84.1 82.4 82.7 

Embryo þ scutellum – 2.8 – 2.7 

Total of grain tissues 97.8 100 97.1 100 

Unknown fraction 2.2 – 2.9 – 

The ‘‘unknown fractions’’ is the difference between 100% and the sum of all the grain 

tissues proportions.

60 

100% 

0% 

98% 

29% 28% 24% 

21% 

24% 

100% 96% 

100% 

25% 

21% 22% 23% 

24% 25% 

24% 27% 25% 30% 

Markers Dissection Markers Dissection 

CROUSTY TIGER 

outer pericarp intermediate layer 

aleurone cell walls aleurone cell contents 

22% 

Fig. 2. Comparison of Crousty and Tiger outer layers composition (%), using either 

biochemical markers or hand-dissection. 

under-estimates or over-estimates of the biochemical markers 

content in reference hand-isolated tissues and/or in coarse bran 

were calculated. The mean, low (mean value 1 standard deviation) 

and high (mean value þ 1 standard deviation) values of each 

biochemical marker were calculated for each grain tissue and for 

coarse bran. Then, the grain tissue proportions resulting from the 

combinations of all these values were calculated. The lowest variability 

was observed for the aleurone cell walls (4%) and the highest 

for the intermediate layer (22%). The maximum relative error (i.e. 

random analytical error) observed for coarse bran was 13% (similar 

results were obtained for both cultivars). The impact of analytical 

variability on tissue proportion calculation can thus be considered 

as acceptable. 

To determine the tissue distribution in fractions obtained from 

wheat grain processing, four different biochemical assays need to 

be carried out, i.e. a rapid and simple colorimetric method to estimate 

the phytic acid content, an enzymatic assay to measure starch 

content which is standardized as a commercial kit, and two more 

complex and time consuming assays to analyse either the phenolic 

acids or the alkylresorcinols composition and content. However, if 

analysis of the composition in different phenolic acids appeared 

necessary in order to distinguish between distinct tissues as the 

pericarp and the aleurone layer, only the total content of alkylresorcinols 

is required as all of these compounds were found to be 

concentrated in the intermediate layer of wheat grains (Landberg 

et al., 2008). Thus, in order to simplify and reduce the time for 

biochemical analyses, we tested the potential of a colorimetric 

method for the determination of alkylresorcinols using Fast Blue B 

reagent as described by Tluscik et al. (1981). Choice of the most 

effective solvent and of the extraction conditions were adapted to 

fractionation samples after preliminary studies (cf. Section 2). A 

very good correlation was observed between determination of total 

AR content using either the colorimetric method or the sum of each 

of the distinct AR molecules separated by gas chromatography 

(colorimetric method ¼ 0.91 GC method, R 2 ¼ 0.98). This result 

agrees with previous studies (Andersson et al., in press) and shows 

that it is possible to develop a simpler and quicker assay for 

determination of wheat grain tissue proportions in fractions 

obtained from processing. Therefore, only the analysis of phenolic 

acids appears as the limiting factor when trying to simplify the 

assays. However it seems possible, in future, to develop specific 


immunological tests in order to further simplify the determination 

of the biochemical markers. 

3.2. Biochemical markers as a tool to monitor fractionation 

processes 

The method was then used to assess the histological proportion 

of fractions exhibiting very different compositions, obtained by 

three different grain or bran dry fractionation processes, in order to 

verify if this method effectively allows study of the distribution of 

the different tissues among fractions, and to evaluate if this method 

can be a useful tool. 

3.2.1. Conventional milling process: whole-grain fractionation 

The calculated grain tissue composition of whole grains, whole 

meal flours, white flours, fine brans and coarse brans is shown in 

Table 3. The composition of the fractions made of both cultivars was 

quite similar, except for fine brans. Indeed, Tiger fine bran contains 

twice as much endosperm as Crousty fine bran, may be due to 

differences in bran brushing steps. Whole grain and whole meal 

flour exhibit very similar composition, and the white flour (0.55% 

ash content) contains only 1–2% of non-endosperm tissues. For 

most of the fractions, the sum of the tissue proportion is superior to 

85%, but for fine brans, totals of 73% and 67% are observed for Tiger 

and Crousty respectively, suggesting that a non-negligible part of 

these fractions was not quantified. In conventional milling 

processes, wheat germs are often found in high amounts in the fine 

bran fractions (Posner and Li, 1991). Fig. 3 shows that there is 

a relation between the approximate germ proportion and the ‘nonquantified’ 

proportion in fractions, suggesting that the ‘non-quantified’ 

30% in fine brans may effectively be composed of germ. Fig. 4 

presents the proportions of each of the peripheral layers relative to 

the total amount of outer layers in bran fractions. Crousty and Tiger 

coarse brans exhibit tissue proportions similar to those found for 

whole hand-isolated outer layers, except for the outer pericarp 

proportion (about 19% outer pericarp in coarse bran compared to 

25% in hand-isolated outer layers). This difference may be due to 

the high friability of outer pericarp, which is a brittle material that 

can be easily removed from the other layers during milling and 

fragmented into finer particles not recovered in coarse brans 

(Antoine et al., 2003, 2004). Fig. 4 also shows that the proportions 

of cell walls in the total aleurone layer were almost the same for 

coarse and fine bran as for whole hand-isolated outer layers, 

showing that the aleurone cell walls and cell contents were not 

dissociated in these fractions (i.e. the aleurone cells were not 

damaged). 

Table 3 

Grain tissues proportions (%) in fractions produced by a conventional milling process 

(Tiger and Crousty cv.) 

Whole 

grain 

Whole 

meal 

flour 

White 

flour 

Outer Intermediate Aleurone layer Starchy Total 

pericarp layer 

Total Cell Cell 

endosperm 

walls content 

Tiger 3.1 2.9 8.8 4.1 4.7 82.4 97 

Crousty 2.7 2.6 8.5 4.1 4.4 84.1 98 

Tiger 3.0 3.2 8.1 3.9 4.2 69.9 84 

Crousty 2.4 2.7 8.5 3.7 4.8 75.0 89 

Tiger 0.1 0.1 1.2 0.5 0.7 99.9 101 

Crousty 0.2 0.1 1.5 0.5 1.0 97.0 99 

Fine bran Tiger 6.1 6.6 17.4 7.9 9.5 42.9 73 

Crousty 7.1 8.0 26.6 12.6 14.0 25.3 67 

Coarse 

bran 

Tiger 15.9 23.0 42.5 20.9 21.6 8.0 89 

Crousty 15.1 21.9 46.3 21.3 24.9 8.0 91

Germ proportion (%) 

45 

40 

35 

30 

25 

20 

15 

10 

5 

0 

y = 0,81x 

R 2 = 0,74 

0 5 10 15 20 25 30 35 40 45 

Unknown proportion (%) 

Fig. 3. Comparison of the approximate germ proportions with the ‘unknown 

proportions’ in milling fractions. The ‘unknown proportion’ in a fraction is the difference 

between 100% and the sum of all the calculated grain tissues proportions. 

3.2.2. Whole-grain fractionation by peeling, pearling, and milling 

This process, by sequential removal of grain outer layers by 

debranning prior to milling, aims at producing less contaminated 

wheat grains and whole-grain flours, by removing the outermost 

layers of the grain. The first debranning step carried out by friction 

removes 3–4% of the grain by weight. As shown in Table 4, this 

removed fraction (i.e. the peeling fraction) contains 53% and 60% of 

outer pericarp for Tiger and Crousty respectively. It also contains 

various proportions of all the other grain tissues, including endosperm, 

probably due to the presence of broken grains in the peeling 

fractions and to the fact that the grain ends are more worn away 

during debranning. This first debranning step by peeling removes 

the major part of the outer pericarp of the grain (up to 60%). Due to 

the shape of wheat grains, removing all the outer pericarp by 

peeling does not seem possible as about 25% of this tissue is located 

in the crease (Evers and Millar, 2002). The pearling fractions, 

that correspond to the fraction removed off the grain during the 

second debranning step (about 3% additional debranning), contains 

30–35% aleurone material and 25–30% starchy endosperm, 

showing that all the peripheral layers have been removed on some 

parts of the grains: the grain ends are more eroded and the grains 

are rounded by pearling. These pearling fractions also contain 7–13% 

of germ. As a large part of the outer pericarp was removed during 

the friction step, the other bran fractions and the flour fractions 

obtained afterwards do not contain much of this tissue (Table 4). As 

a consequence, the 100% flours from peeling contain less outer 

layer particles than the whole meal flours, and the 100% flours from 

pearling, which are made from grains that have undergone an 

additional debranning step (pearling), contain even less outer 

layers. All these 100% flours contain 3–4% germ, that is close to the 

amount of germ quantified in whole grains (Table 2). The 76% flours 

from peeling and pearling are both almost only composed of starchy 

endosperm (only 2% of non-endosperm particles). 

Fig. 4 presents the proportions of each of the outer layers relative 

to the total amount of peripheral layers in bran fractions. The 

pearling fractions contain the most important proportion of intermediate 

layer (8–10% more than in hand-isolated whole outer 

layers), due to the removal of outer pericarp by peeling and to the 

fact that a part of the aleurone layer probably remains attached to 

the grains and is found in the bran fraction after pearling. 

Surprisingly, bran fractions after peeling and after pearling exhibit 

quite similar tissues proportions. The differences between these 

two fractions are probably reduced because they both contain the 

bran present in the crease area. This crease bran represents a nonnegligible 

proportion of the grain total outer layers and is not easily 

removed neither by peeling nor by pearling (Dexter and Wood, 

1996). Fig. 4. shows that for almost all the bran fractions, the ratio 

aleurone cell walls/aleurone cell contents is very close to that 

observed for coarse bran and hand-isolated whole outer layers, 

suggesting that aleurone material is not dissociated and that the 

aleurone cells are not broken. Only the peeling fractions exhibit 

a surprisingly high proportion of aleurone cell walls compared to 

their proportion of aleurone cell contents. This may be explained by 

the fact that the cell walls, being linked to the other outer layers, are 

torn off the grain by peeling, leaving the cell contents on the grain. 

But this high proportion of aleurone cell walls in peeling fractions 

seems quite excessive when compared to the proportions of 

intermediate layers in these fractions, suggesting that it may have 

been over-estimated due to the high content of para-coumaric acid 

in these fractions. If the proportions of total aleurone layer is 

calculated by taking into account only the phytic acid content and 

not the p-CA content of the peeling fractions, the total aleurone 

content in these fractions is found to be 3–4%. This could be more 

25 29 31 34 27 30 26 27 32 38 

29 27 33 35 

23 22 

26 

30 

26 26 

8 4 

26 

26 

24 27 

27 29 

26 

26 

25 

24 

22 

20 

19 

17 

28 

20 

26 

18 

63 68 

32 

9 

30 

6 

35 

11 

34 

11 

30 

10 

28 

8 

Tg Cr Tg Cr Tg Cr Tg Cr Tg Cr Tg Cr Tg Cr 

Hand-isolated 

whole bran 


Fine bran Coarse bran peeling 

fraction 

bran fraction 

after peeling 

pearling 

fraction 

outer pericarp intermediate layer aleurone cell walls aleurone cell contents 

bran fraction 

after pearling 

Fig. 4. Study of the dissociation of the outer layers in the different Tiger and Crousty bran fractions: histological composition of the particles of outer tissues present in these 

fractions (i.e. endosperm excluded). The proportions of outer pericarp, intermediate layer, and aleurone layer are re-calculated relatively to the total amount of these tissues in bran 

fractions.

62 

Table 4 

Grain tissues proportions (%) in fractions produced by a whole-grain fractionation process including peeling, pearling, and milling steps (Tiger and Crousty cv.) 

plausible, as these fractions are also supposed to contain 9–17% of 

germ. The high amount of p-CA quantified in the peeling fractions 

might be due to an external contamination. Indeed, contaminating 

materials such as straw and chaff can be found in the debranning 

fraction, and several authors showed that wheat straw is very rich 

in para-coumaric acid (Benoit et al., 2006; Sun et al., 2001; Tapin 

et al., 2006). 

3.2.3. Bran fractionation process: production of aleurone-rich 

fractions 

The wheat aleurone layer has been shown to have great nutritional 

interest, and to concentrate most of the micronutrients 

(minerals and vitamins), antioxidants (mainly phenolic acids) and 

other phytochemicals of the wheat grain (Hemery et al., 2007). 

Thus, it can be interesting to produce bran fractions highly 

concentrated in aleurone material, to transform a by-product of 

flour into a high nutritional value food ingredient. In the present 

study, various aleurone-rich bran fractions were studied (Table 5). 

The Aleurone 2 180 mm Tiger 22.9 13.6 55.3 28.3 27.0 2.0 94 

Crousty 19.1 13.1 64.7 31.0 33.7 2.4 99 

Aleurone 1 180 mm Tiger 9.1 12.9 74.2 32.7 41.5 2.6 99 

Crousty 11.2 12.7 83.2 38.9 44.3 2.7 110 

Aleurone 2 < 180 mm Tiger 7.8 11.1 79.3 37.3 42.0 2.9 101 

Crousty 8.5 12.0 89.1 42.4 46.7 3.0 113 

By-product 1 Tiger 30.9 18.9 29.7 20.5 9.2 3.6 83 

Crousty 36.6 18.6 29.4 19.9 9.6 4.4 89 

By-product 2 Tiger 45.6 11.0 27.6 15.9 11.7 1.6 86 

Crousty 33.2 12.7 32.9 20.2 12.7 2.0 81

fractions (compared to Aleurone 1 fractions) after the second 

fractionation step. 

For almost all the aleurone fractions, the ratio aleurone cell 

walls/aleurone cell contents is really close to the ratio observed for 

hand-isolated aleurone layers, suggesting that aleurone material is 

not dissociated (i.e. that the aleurone cells are not broken). This 

observation are in agreement with microscopy observations 

(Amrein et al., 2003; Buri et al., 2004) showing that these aleuronerich 

fractions are composed of clusters of intact cells. All these 

observations are in agreement with the observations made by 

several authors. Indeed, the higher protein and beta-glucan 

contents observed by Amrein et al. (2003), the higher phenolic 

acids content observed by Zhou et al. (2004), and the higher 

minerals and phytic acid content observed by Buri et al. (2004) for 

Aleurone 2 than for Aleurone 1, can be explained by the enrichment 

in aleurone material of Aleurone 2 compared to Aleurone 1. 

4. Conclusion 

The evaluation of the method described here, in comparison 

with tissue quantification by hand dissection, has shown its efficacy 

(mean relative errors of 4% and 8% for Crousty and Tiger outer layers 

respectively), and the impact of the analytical variability 

(maximum 13% relative error on coarse bran) was regarded as 

acceptable. Some improvements can be made to the current 

method. For example, a better quantification of the germ content is 

needed, as the germ proportion is often not negligible in bran 

factions and fractions coming from debranning processes. Wheat 

germ agglutinin seems to be a promising marker of germ. The 

quantification method does not currently enable the accurate 

quantification of the germ proportions in milling fractions, but it 

gives qualitative information and allows classification of these 

fractions according to their germ content (Fig. 3). This method 

should be improved to add WGA as another marker to complete the 

current method. 

In the milling industry, batches of different grain cultivars are 

frequently used as starting material for dry fractionation processes. 

In this study, the tissue proportions have been deduced from 

reference values assessed on exactly the same wheat cultivars from 

which the fractions were produced: the genetic and agronomic 

variability was thus not taken into account. Some of the constituents 

chosen as biochemical markers have specific functions in 

wheat grain. Therefore, the natural variability of compounds such 

as phytic acid, phenolic acids, and alkylresorcinols among a wheat 

population could be very high. For example, Barrier-Guillot et al. 

(1996) showed that the amount of Phytic P can vary from 0.92 to 

2.8 mg/g dm for common wheat with various agronomic and 

genetic characteristics. Within the European HEALTHGRAIN project 

(Poutanen et al., 2008), other authors analysed the numerous 

wheat varieties in the diversity screen of the project and showed 

that the total phenolic acid content varies from 326 to 1171 mg/g dm 

in the 130 winter wheat lines and from 456 to 892 mg/g dm in the 

20 spring wheat lines (Li et al., in press), while the total alkylresorcinol 

content varies from 220 to 652 mg/g dm in these winter 

wheat lines and from 254 to 537 mg/g dm in the spring wheat lines 

(Andersson et al., in press). To try to evaluate the impact of genetic 

variability on grain tissue proportions using biochemical markers, 

tissue proportions of Tiger coarse bran were calculated with 

Crousty reference values (and vice versa). The mean relative error 

observed was 10%, and this variability was found to be less than the 

maximum analytical variability. This study should be carried out on 

other wheat cultivars. If this observation were confirmed, it could 

then be possible to use ‘‘standard reference values’’ (i.e. the means 

of the values obtained for the grain tissues of different cultivars) to 

quantify tissues proportion in fractions, to limit grain tissues 


dissection, and to apply this quantification method to batches of 

grains of various origins. 

Several biochemical analyses have to be carried out to assess the 

composition of fractions using the described method. Therefore, it 

would be difficult to use it for controls on the production line, but 

this method can be useful in quality control laboratories of firms, 

and can also be used as a reference for the calibration of equipment 

based on physical properties (i.e. fluorescence). This biochemical 

marker method has been proved to be an efficient tool for assessment 

of grain tissue proportions in various milling fractions 

exhibiting contrasting composition, such as flours, brans, or aleurone-rich 

fractions. The present study showed that this quantitative 

method allows, by calculation of the composition of the different 

products generated by grain or bran fractionation processes, an 

understanding of the effects of these processes and monitoring and 

adapting of them to reach the objective (i.e. efficient debranning, 

production of specific fractions). 


The authors thank the Federal Research Centre for Nutrition and 

Food (BFEL), Germany, for the production of the conventional 

milling fractions used in this study. We also gratefully thank T.-M. 

Lasserre for her work on the biochemical analyses, and S. Chay and 

A. Putois for the manual dissection of grain tissues. This publication 

is financially supported by the European Commission in the 

Communities 6th Framework Program, Project HEALTHGRAIN 

(FOOD-CT-2005-514008). It reflects the author’s views and the 

Community is not liable for any use that may be made of the 

information contained in this publication. 

References 

AACC, 2000. American Association of Cereal Chemists Approved Methods, 10th ed. 

The Association, St Paul, Minnesota. 

Amrein, T.M., Gränicher, P., Arrigon, E., Amadò, R., 2003. In vitro digestibility and 

colonic fermentability of aleurone isolated from wheat bran. Lebensmittel- 

Wissenschaft und Technology 36, 451–460. 

Andersson, A.M.A., Kamal-Eldin, A., Fras, A., Boros, D., Åman, P. Alkylresocinols in 

wheat varieties in the HEALTHGRAIN diversity screen. Journal of Agricultural 

and Food Chemistry, in press. 

Antoine, C., Peyron, S., Mabille, F., Lapierre, C., Bouchet, B., Abecassis, J., Rouau, X., 

2003. Individual contribution of grain outer layers and their cell wall structure 

to the mechanical properties of wheat bran. Journal of Agricultural and Food 

Chemistry 51, 2026–2033. 

Antoine, C., Peyron, S., Lullien-Pellerin, V., Abecassis, J., Rouau, X., 2004. Wheat bran 

tissue fractionation using biochemical markers. Journal of Cereal Science 39, 

387–393. 

Aureli, G., D’Egidio, M.G., 2007. Efficacy of debranning on lowering of deoxynivalenol 

(DON) level in manufacturing processes of durum wheat. Tecnica 

Molitoria 58, 729–733. 

Bach Knudsen, K.E., Steenfeldt, S., Borsting, C.F., Eggum, B.O., 1995. The nutritive 

value of decorticated mill fractions of wheat. 1. Chemical composition of raw 

and enzyme treated fractions and balance experiments with rats. Animal Feed 

Science and Technology 52, 205–225. 

Bacic, A., Stone, B.A., 1981. Isolation and ultrastructure of aleurone cell walls from 

wheat and barley. Australian Journal of Plant Physiology 8, 453–474. 

Baldwin, P.M., Bertrand, D., Novales, B., Bouchet, B., Collobert, G., Gallant, D.J., 1997. 

Chemometric labeling of cereal tissues in multichannel fluorescence microscopy 

images using discriminant analysis. Analytical Chemistry 69, 4339–4348. 

Barrier-Guillot, B., Casado, P., Maupetit, P., Jondreville, C., Gatel, F., Larbier, M., 1996. 

Wheat phosphorus availability. 1. In vitro study; factors affecting endogenous 

phytasic activity and phytic phosphorus content. Journal of the Science of Food 

and Agriculture 70, 62–68. 

Barron, C., Surget, A., Rouau, X., 2007. Relative amounts of tissues in mature wheat 

(Triticum aestivum L.) grain and their carbohydrate and phenolic acid composition. 


Benoit, I., Navarro, D., Marnet, N., Rakotomanomana, N., Lesage-Meessen, L., 

Sigoillot, J.C., Asther, M., Asther, M., 2006. Feruloyl esterases as a tool for the 

release of phenolic compounds from agro-industrial by-products. Carbohydrate 

Research 341, 1820–1827. 

Bohm, A., Bogoni, C., Behrens, R., Otto, T., 2003. Method for the extraction of 

aleurone from bran. US Patent Application Publication US 2003/0175384 A1. 

Applicant: Buhler AG. 

Bohm, A., Kratzer, A., 2005. Method for isolating aleurone particles. US Patent 

Application Publication US 2005/0103907 A1. Applicant: Buhler AG.

64 

Brillouet, J.M., Rouau, X., Hoebler, C., Barry, J.L., Carre, B., Lorta, E., 1988. A new 

method for determination of insoluble cell-walls and soluble nonstarchy 

polysaccharides from plant materials. Journal of Agricultural and Food Chemistry 

36, 969–979. 

Buri, R.C., von Reding, W., Gavin, M.H., 2004. Description and characterization of 

wheat aleurone. Cereal Foods World 49, 274–282. 

Chen, Y., Ross, A.B., Aman, P., Kamal-Eldin, A., 2004. Alkylresorcinols as markers of 

whole grain wheat and rye in cereal products. Journal of Agricultural and Food 

Chemistry 52, 8242–8246. 

Courcoux, P., Devaux, M.F., Bouchet, B., 2002. Simultaneous decomposition of 

multivariate images using three-way data analysis. Application to the 

comparison of cereal grains by congocal laser scanning microscopy. Chemometrics 

and Intelligent Laboratory Systems 62, 103–113. 

Dexter, J.E., Wood, P.J., 1996. Recent applications of debranning of wheat before 

milling. Trends in Food Science & Technology 7, 35–41. 

Eugster, W., Gerschwiler, O., 2006. Method and installation for cleaning cereal. US 

Patent Application Publication US 2006/0147591 A1. Applicant: Bühler AG. 

Evers, A.D., Reed, M., 1988. Some novel observations by scanning electron microscopy 

on the seed coat and nucellus of the mature wheat grain. Cereal Chemistry 

65, 81–85. 

Evers, A.D., Millar, S., 2002. Cereal grain structure and development: some implications 

for quality. Journal of Cereal Science 36, 261–284. 

Fincher, G.B., Stone, B.A., 1986. Cell walls and their components in cereal grain 

technology. In: Pomeranz, Y. (Ed.), Advances in Cereal Science and Technology, 

vol. 8. American Association of Cereal Chemists, St Paul, MN, pp. 207–295. 

Fleurat-Lessard, F., Chaurand, M., Marchegay, G., Abecassis, J., 2007. Effects of processing 

on the distribution of pirimiphos-methyl residues in milling fractions of 

durum wheat. Journal of Stored Products Research 43, 384–395. 

Fulcher, R.G., O’Brien, T.P., Lee, J.W., 1972. Studies on the aleurone layer. I. 

Conventional and fluorescence microscopy of the cell wall with emphasis on 

phenol-carbohydrate complexes in wheat. Australian Journal of Biological 

Science 25, 23–34. 

Grabber, J.H., Ralph, J., Lapierre, C., Barriere, Y., 2004. Genetic and molecular basis of 

grass cell-wall degradability. I. Lignin-cell wall matrix interactions. Comptes 

Rendus Biologies 327, 455–465. 

Greffeuille, V., Abecassis, J., Bar L’Helgouac’h, C., Lullien-Pellerin, V., 2005. Differences 

in the aleurone layer fate between hard and soft common wheats at grain 

milling. Cereal Chemistry 82, 138–143. 

Hemery, Y., Rouau, X., Lullien-Pellerin, V., Barron, C., Abecassis, J., 2007. Dry 

processes to develop wheat fractions and products with enhanced nutritional 

quality. Journal of Cereal Science 46, 327–347. 

Jacobs, D.R., Steffen, L.M., 2003. Nutrients, foods, and dietary patterns as exposures 

in research: a framework for food synergy. American Journal of Clinical Nutrition 

78, 508S–513S. 

Jensen, S.A., Munck, L., Martens, H., 1982. The botanical constituents of wheat and 

wheat milling fractions. 1. Quantification by autofluorescence. Cereal Chemistry 

59, 477–484. 

Jensen, S.A., Martens, H.,1983. The botanical constituents of wheat and wheat milling 

fractions. 2. Quantification by amino-acids. Cereal Chemistry 60, 172–177. 

Laca, A., Mousia, Z., Diaz, M., Webb, C., Pandiella, S.S., 2006. Distribution of 

microbial contamination within cereal grains. Journal of Food Engineering 72, 

332–338. 

Landberg, R., Kamal-Eldin, A., Salmenkallio-Marttila, M., Rouau, X., Aman, P., 2008. 

Localization of alkylresorcinols in barley, wheat and rye kernels. Journal of 


Latta, M., Eskin, M., 1980. A simple and rapid colorimetric method for phytate 

determination. Journal of Agricultural and Food Chemistry 28, 1313–1315. 

Lempereur, I., Surget, A., Rouau, X., 1998. Variability in dehydrodiferulic acid 

composition of durum wheat (Triticum durum Desf.) and distribution in milling 

fractions. Journal of Cereal Science 28, 251–258. 


Li, L., Shewry, P.R., Ward, J.L. Phenolic acids in wheat varieties in the HEALTHGRAIN 

diversity screen. Journal of Agricultural and Food Chemistry, in press. 

Liu, R.H., 2007. Whole grain phytochemicals and health. Journal of Cereal Science 

46, 207–219. 

Mateo Anson, N., Van den Berg, R., Havenaar, R., Bast, A., Haenen, G., 2008. 

Ferulic acid from aleurone determines the antioxidant potency of wheat 

grain (Triticum aestivum L.). Journal of Agricultural and Food Chemistry 56, 

5589–5594. 

Miller, R.C., Bowles, D.J., 1982. A comparative-study of the localization of wheatgerm-agglutinin 

and its potential receptors in wheat grains. Biochemical 

Journal 206, 571–576. 

O’Dell, B.L., De Boland, A.R., Koirtyohann, S.R., 1972. Distribution of phytate and 

nutritionally important elements among the morphological components of 

cereal grains. Journal of Agricultural and Food Chemistry 20, 718–723. 

Peyron, S., Surget, A., Mabille, F., Autran, J.C., Rouau, X., Abecassis, J., 2002. 

Evaluation of tissue dissociation of durum wheat grain (Triticum durum 

desf.) generated by the milling process. Journal of Cereal Science 36, 

199–208. 

Pomeranz, Y., 1988. Chemical composition of kernel structures. In: Pomeranz, Y. 

(Ed.), Wheat: Chemistry and Technoloy, vol. 4. AACC, St Paul, pp. 97–158. 

Posner, E.S., Li, Y.Z., 1991. A technique for separation of wheat germ by impacting 

and subsequent grinding. Journal of Cereal Science 13, 49–70. 

Poutanen, K., Shepherd, R., Shewry, P., Delcour, J.A., Björck, I., van der Kamp, J.W., 

2008. Beyond whole grain: the European HEALTHGRAIN project aims at 

healthier cereal foods. Cereal Foods World 53, 32–35. 

Pussayanawin, V., Wetzel, D.L., Fulcher, R.G., 1988. Fluorescence detection and 

measurement of ferulic acid in wheat milling fractions by microscopy and HPLC. 


Rhodes, D.I., Sadek, M., Stone, B.A., 2002. Hydroxycinnamic acids in walls of wheat 

Aleurone cells. Journal of Cereal Science 36, 67–81. 

Ross, A.B., Shepherd, M.J., Schupphaus, M., Sinclair, V., Alfaro, B., Kamal-Eldin, A., 

Aman, P., 2003. Alkylresorcinols in cereals and cereal products. Journal of 

Agricultural and Food Chemistry 51, 4111–4118. 

Sun, R.C., Sun, X.F., Zhang, S.H., 2001. Quantitative determination of hydroxycinnamic 

acids in wheat, rice, rye, and barley straws, maize stems, oil palm frond 

fiber, and fast-growing poplar wood. Journal of Agricultural and Food Chemistry 

49, 5122–5129. 

Surget, A., Barron, C., 2005. Histologie du grain de blé. Industrie des Céréales 145, 

3–7. 

Symons, S.J., Dexter, J.E., 1996. Aleurone and pericarp fluorescence as estimators of 

mill stream refinement for various canadian wheat classes. Journal of Cereal 


Tapin, S., Sigoillot, J.C., Asther, M., Petit-Conil, M., 2006. Feruloyl esterase utilization 

for simultaneous processing of nonwood plants into phenolic compounds and 

pulp fibers. Journal of Agricultural and Food Chemistry 54, 3697–3703. 

Tluscik, F., Kozubek, A., Mejbaum-Katzenellenbogen, W., 1981. Alkylresorcinols in 

rye (Secale-Cereale L) grains. 6. Colorimetric micromethod for the determination 

of alkylresorcinols with the use of diazonium salt, fast blue-B. Acta Societatis 

Botanicorum Poloniae 50, 645–651. 

Vaintraub, I.A., Lapteva, N.A., 1988. Colorimetric determination of phytate in 

unpurified extracts of seeds and the products of their processing. Analytical 

Biochemistry 175, 227–230. 

Vincenzi, S., Zoccatelli, G., Perbellini, F., Rizzi, C., Chignola, R., Curioni, A., 

Peruffo, A.D.B., 2002. Quantitative determination of dietary lectin activities by 

enzyme-linked immunosorbent assay using specific glycoproteins immobilized 

on microtiter plates. Journal of Agricultural and Food Chemistry 50, 

6266–6270. 

Zhou, K., Laux, J.J., Yu, L., 2004. Comparison of swiss red wheat grain and fractions 

for their antioxidant properties. Journal of Agricultural and Food Chemistry 52, 

1118–1123.

Lateral growth of a wheat dough disk under various growth conditions 

Amy Penner a , Leaelaf Hailemariam b , Martin Okos a,b , Osvaldo Campanella a, * 

a School of Agricultural and Biological Engineering, Purdue University, West Lafayette, IN 47907-2093, USA 

b School of Chemical Engineering, Purdue University, West Lafayette, IN 47907-2100, USA 

article info 


Received 17 October 2007 



Keywords: 

Viscoelasticity 

Dough 

Pressure release 

Expansion 


abstract 

In the food industry, bubble formation is used to create structure 

and texture which are primarily responsible for the product attributes 

(Aguilera and Germain, 2007; Moraru and Kokini, 2003; 

Trater et al., 2005). Bubbles are generated through the evolution 

and expansion of gas, either through phase change (flash evaporation), 

or chemical reaction (yeast or chemical leavening). Gas that 

does not escape into the atmosphere is bound in nucleating sites 

which act as growth sites for bubbles in the expanding material 

(Baker and Mize, 1941). The quantity of bubbles created and 

sustained is the main factor in determining the final product 

quality, texture, and volume (Cauvain et al., 1999; Elmehdi et al., 

2003; Fox,2004;Strybulevych et al., 2007; Whitworth and Alava, 

1999). Thus, an understanding of bubble growth mechanics is 

crucial for control and product optimization. 

Previous work on mathematical modeling of bubble growth has 

been done by Alavi et al. (2003), Chen and Rizvi (2006), de Cindio 

and Correra (1995), Fan et al. (1994, 1999), Schwartzberg et al. 

(1995), Shimiya and Yano (1988), Singh and Bhattacharya (2005), 

and Wang et al. (2005). Recently, Hailemariam et al. (2007) 

presented a model that included the effect of multi-scale mass 


E-mail address: campa@purdue.edu (O. Campanella). 


doi:10.1016/j.jcs.2008.07.007 





Numerous studies have tried to understand and model bubble growth inside dough. Experimental 

studies are inconvenienced by the methods’ inability to capture the dynamic phenomena. In this paper, 

a versatile experimental method was developed to allow for macroscopic expansion of wheat dough. The 

study evaluates expansion of a dough disk under varying: moisture content (40, 41, 42, 43, and 44% wb), 

leavening acid concentration (30, 40, and 50% db), pressure schemas, pressurizing gas (compressed air 

and CO2), and lubrication (Teflon Ò film coating and Pam Ò aerosol lubricant). Dough expansion increased 

22.6% by increasing moisture content from 40 to 44%. Increased baking powder formulation (40% db) was 

used to enhance initial growth conditions and CO2 production. ‘Pressure pulse’ and ‘pressure vacuum 

methods’ added pressurization alternatively with full vacuum. The former method included a rest period 

before vacuum application, and increased expansion by 10.8%. Teflon Ò and Pam Ò reduced friction 

between the dough and acrylic plate and increased the final expansion by 14.7% compared to no 

lubricant following the ‘standard pressurization method’. ‘Pressure pulse’ and ‘pressure vacuum’ 

experiments decreased expansion by 28.4 and 38.2%, respectively compared to ‘standard pressurization’ 

while using Teflon Ò and Pam Ò . 


transfer, surface tension, inertia and viscoelasticity in a multibubble 

environment. However, a major challenge with these 

models has been the experimental verification of bubble dynamics. 

Bell et al. (1975) described bubble growth in dough during baking 

using television microscopy of a thin film of dough baking on 

a small slide. Kumagai et al. (1991) used a similar approach, 

spreading a thin layer of dough on a slide heated by a thermostat. 

de Cindio and Correra (1995) microscopically analyzed bubble size 

distributions of dough samples which were immersed in liquid 

nitrogen and cut in thin slices. Similar analysis was done by 

Campbell et al. (1998) and Schwartzberg et al. (1995). Alavi et al. 

(1999) cut extrudate samples into thin sections which were dried in 

high temperature ovens before electron microscopy analysis. A 

similar approach was taken by Chen and Rizvi (2006) and Singh and 

Bhattacharya (2005). 

With the exception of the ‘dough smear’ methods of Bell et al. 

(1975), the experimental methods used thus far have yielded only 

the final bubble size distribution, which is of little help in studying 

bubble dynamics. The ‘dough smear’ methods do not allow for 

control of the surrounding environment and are difficult to scale 

up. 

Measuring the dough’s properties while it is undergoing physical 

and chemical changes is difficult due to the dynamic and fragile 

nature of dough (Aguilera and Germain, 2007). Invasion during the 

expansion process could cause gas cell leakage which would reduce 

expansion (Wagner et al., 2006). Older techniques used interval

66 

measurements, but this yielded inconsistent, time intensive results 

that were not fully accurate or representative of on-line changes 

(Fox et al., 2004). Modern microscopic measuring techniques may 

be applied for dynamic size measurements, but these are not 

without their own hurdles. These methods include electron 

microscopy (Aguilera and Germain, 2007; Trater et al. 2005; 

Whitmore and Alava, 1999), X-ray tomography (Babin et al., 2006, 

2007; Bellido et al., 2006; Trater et al., 2005; Whitmore and Alava, 

1999), Magnetic resonance imaging (MRI) (Wagner et al., 2006), 

and ultrasound (Elmhedi et al., 2003; Fox et al., 2004; Strybulevych 

et al., 2007). The reader is asked to consult these references for 

greater details. 

In summary, one may remark that no single experimental 

method is free of challenges. The ease of experimentation, the size 

range of the bubbles, and the versatility of the methods should be 

considered in using or developing an experimental method. The 

bubble size range in the food industry is 10–1000 mm(Fyrillas et al., 

2000). Current detection methods are applied to determine gas 

bubble distribution, cell wall size, and cell wall thickness. The 

microscopic bubble size poses a challenge for the development of 

a dynamic method of measuring bubble size because the material 

structure is fixed either through heat or freezing before analysis. 

This may destroy the structure, the entity to be measured, and is 

not fully representative of the dynamic system (Babin et al., 2007; 

Trader et al., 2005; Wagner et al., 2006). 

There is a need for a versatile experimental method that facilitates 

the dynamic observation of the bubble size history and 

distribution. To overcome the challenge of measuring bubble 

growth in situ, the authors developed an experimental method to 

study microstructure growth through macroscopic investigation. A 

sample of wheat dough was constrained to grow laterally in a thin 

gap between parallel plates which allowed the maximum change in 

density with time due to decreased surface area for gas escape. 

Some work has been done on lateral growth with a mass of 

dough, including looking at cookie diameters (Lee and Inglett, 

2006; Swanson et al., 1999; Zoulias et al., 2002). Taki et al. (2003) 

developed an experimental method for analyzing the growth of 

bubbles in polymers through examination of an expanding disk of 

fluid. Recently, Elmehdi et al. (2007) described a method for 

measuring dough density between two polyacrylic plates at 

atmospheric conditions where radial growth was the only dimension 

encountered. Digital imaging was used to capture dough 

expansion every 2 min. In this paper, we extend that method and 

describe a method that allows for dynamic tracking of overall 

growth under different pressure schema with increased measurement 

recordings. This work concentrates on the study of rapid 

dough growth initiated by sudden pressure and vacuum changes. 

Analysis of the lateral growth in a dough disk may be imagined to 

be a cylindrical section of the expanding material. It was hypothesized 

that pressurization during the initial phase of growth would 

increase the final expansion due to prevention of early gas escape. 

To test this idea, different pressurization schemas were implemented 

to evaluate the effects of pressurization and vacuum 

application to aid in increasing expansion. 

The work describes a simple dough system consisting of flour, 

water, and baking powder to observe expansion in experimental 

testing. Moisture content and baking powder concentration work 

to drive dough expansion in this system and thus warranted further 

study to generate optimal testing conditions. Moisture content 

affects the release of CO2 from the baking powder. It also affects the 

ability of the dough matrix to expand and retain the gas that is 

produced. Baking powder concentration was varied to determine 

its effect on expansion. Use of external CO2 as the pressurizing gas 

was investigated for its potential effects in the final system equilibrium. 

Dough expansion is based on internal CO2 production and 

thus it was necessary to compare CO2 with compressed air as the 

A. Penner et al. / Journal of Cereal Science 49 (2009) 65–72 

pressurizing gas. It was also hypothesized that expansion was 

limited by friction at the dough/acrylic plate interface, and thus 

a reduction in friction was investigated using Teflon Ò film and 

Pam Ò aerosol lubrication. 

2. Experiment description 

The pressure chamber consists of a clear acrylic cylinder with 

removable top and bottom clear acrylic plates. The plates sealed the 

chamber using a rubber gasket at each end and each end was 

securely bolted to maintain constant pressure once the sample was 

placed inside. The cylinder measured 10 inches vertically and had 

a 5 inch inner diameter. A schematic representation of the experimental 

setup is shown in Fig. 1. A sample of dough was placed 

between two 5-mm thick clear acrylic plates which were clamped in 

place with a fixed 2 mm gap. This ensured only lateral expansion 

which is shown in Fig. 2. The acrylic plate was marked with x and y 

major and minor axes in the horizontal plane using a mm length 

scale to determine the diameter of dough sample with time. All 

experiments were performed in triplicate and averaged results are 

reported. For all experiments temperature was held at 20 3 C. 

The following parameters were evaluated throughout this work: 

i. Moisture content: 40, 41, 42, 43, and 44% (wb). 

ii. Baking powder concentration: 30, 40, and 50% (db). 

iii. Pressure schema: combinations of increased pressure 

(2 bars), and/or full vacuum application; see Section 3.2 for 

full description. 

iv. Pressurizing gas: compressed air or CO2. 

v. Lubrication: Teflon Ò film coating and/or aerosol lubrication. 

Fig. 1. Expansion chamber: (a) top-schematic; and (b) bottom-laboratory setup.

Fig. 2. Parallel plate: (a) schematic; and (b) laboratory setup. 


3.1. Dough preparation 

Flour was determined to have 11.5% moisture content (AOAC 

method 925.10) and baking powders to have 6% moisture content 

(http://www.nutritiondata.com). The wheat dough was prepared 

from 10.00 0.02 g all purpose bleached flour (Gold Medal Ò 

General Mills, Minneapolis, MN, USA), 4.00 0.02 g double acting 

baking powder (Clabber Girl Ò , Terre Haute, IN, USA). A representative 

formula describes baking powder by mass as starch (40.07%), 

sodium bicarbonate (26.73%), sodium aluminum sulfate (19.92%) 

and monocalcium phosphate (13.28%) (Matz, 1992). Room 

temperature (22 2 C) distilled water was added to produce 40, 

41, 42, 43, and 44% moisture percentages on a wet basis. In this 

work, we used 40% baking powder (db) which is much higher than 

industrial practices such as 4–6.25% for baking powder biscuits 

(Matz, 1992). This was done to accelerate the production of CO2 and 

enhance expansion in experiments. 

A. Penner et al. / Journal of Cereal Science 49 (2009) 65–72 67 

Flour and baking powder were kept in sealed containers in 

a freezer at 4 C to prevent moisture absorption from the air. They 

were taken out of the freezer immediately before weighing and 

stored in the freezer immediately after the sample was measured. 

Flour and baking powder were blended by stirring 20 times by 

hand. Water was added all at once and immediately mixed for 

2 min at 86 rpm in a 100-g bowl pin mixer (National Manufacturing 

Company, Lincoln, NE, USA). 

3.2. Measurement of lateral expansion 

Upon completion of mixing, a dough sample (2–3 g) was placed 

between the two acrylic plates. Clamps were added to the sides to 

ensure a constant 2 mm distance between the plates. This also 

eliminated the vertical expansion component as a variable in 

experiments. The plates holding the dough were placed inside the 

pressure chamber and sealed. Periodic photographs show the 

dough disk expanding as seen in Fig. 3. Measurements were taken 

of the horizontal major diameters (x and y radial diameters) every 

minute for 30 min starting 4 min after mixing was completed. The 

4 min window was used to mount the sample in the chamber 

between the clear acrylic plates. 

Density was determined from the mass of dough (assumed 

constant) divided by the changing volume of dough. 

rðtÞ ¼ m 

TA ¼ 

Where: 

m 

Tp D hþDv 

4 

2 

A is the area of the dough (mm 2 ) r is density of dough as 

a function of time 

m is mass of dough in grams 

T is sample thickness, constant at 2 mm 

Dh is the horizontal distance occupied by the dough at each time 

measurement (mm) 

Dv is the vertical distance occupied by the dough at each time 

measurement (mm) 

This method of computing area was validated using Origin 6.1 

Scientific Graphing and Data Analysis Software (OriginLab Corporation, 

Northampton, MA, USA). Images of the dough disk were 

digitized, and the area was computed by using numerical integration 

through Origin 6.1 software. Area measurements from on-line 

experiments were hand calculated and the results were in close 

agreement. For example, a hand calculated sample area was 

13.50 mm 2 while the corresponding software calculated the area at 

13.52 mm 2 . 

The experiments were monitored, varying: moisture content, 

baking powder concentration, pressure and vacuum cycles in the 

Fig. 3. Change in dough disk dimensions through the pressure scheme: (a) before compression (1 bar); (b) pressurization (2 bar); (c) after release (1 bar). Dotted circle added to 

show changes in dough dimensions.

68 

chamber, pressurizing gas, and the application of lubricants on the 

acrylic plates. All the experiments, except for those done to study 

the effect of the pressure profile (discussed later), were done with 

the ‘standard pressurization method’: 

Minutes 0–3: The chamber pressure was maintained at 1 bar to 

allow the dough to rest in the expansion chamber. This allowed 

for initial CO2 production and temperature equilibration. 

Minutes 4–8: Pressure was increased to 2 bars by CO2 or 

compressed air application to the chamber. 

Minutes 9–30: Pressure was immediately released and maintained 

at 1 bar. 

Such a pressure profile is necessary because it allows the 

observer to simulate conditions during rapid expansion or 

compression of the dough. The diameter was recorded throughout 

the procedure at 1 min intervals and the dimensionless ratio of 

density at any time to the initial density was compared. To study 

pressure effects, the ‘pressure pulse method’ was examined: 

Minutes 0–3: Chamber pressure was maintained at 1 bar to 

allow the dough to rest in the expansion chamber. 

Minutes 4–8: Pressure was increased to 2 bars by CO2 application 

to the chamber. 

Minutes 9–13: Pressure was released and the chamber pressure 

was maintained at 1 bar. 

Minutes 14–19: A full vacuum was applied to the chamber. 

Minutes 20–30: Vacuum was released and chamber pressure 


The final pressurization schema used to study pressure effects 

was the ‘pressure vacuum method’: 

Minutes 0–3: Chamber pressure was maintained at 1 bar to 

allow the dough to rest in the expansion chamber. 

Minutes 4–8: Pressure was increased to 2 bars by CO2 application 

to the chamber. 

Minutes 9–13: Pressure was released into a full vacuum 

environment in the chamber and thereby eliminated the 

re-equilibration phase after pressurization. 


Minutes 14–30: Vacuum was released and chamber pressure 


4. Results 

4.1. Effect of moisture content (40, 41, 42, 43, and 44%) 

During experimentation, after pressure was applied the density 

dramatically increased due to compression of the dough. As 

moisture content increased from 40 to 44%, greater compression 

and total expansion were produced. Under the standard pressurization 

method the high moisture dough (44% wb) yielded a 26% 

greater expansion compared to the low moisture dough (40% wb) 

with only 4% expansion as seen in Fig. 4. It was noted that 40 and 

41% moisture content doughs were dry and non-cohesive. This was 

in strong contrast to the 44% dough which was highly viscous. The 

dough made using 43% moisture content was chosen as the balance 

between the above extremes and was used in all other 

experiments. 

4.2. Effect of baking powder concentration 

Formulation consisting of 40% baking powder is significantly 

higher than current industrial standards of 4.25–6% (db) for baking 

powder products such as biscuits (Matz, 1992). However, this value 

produced rapid CO2 generation for monitoring dough expansion. 

Using 6% baking powder (db) in the dough, produced low expansion 

that also produced poor repeatability when monitoring 

expansion. The growth effects were seen more easily with 40% (db) 

baking powder formulation. This concentration was varied to 

investigate 30, 40, and 50% (db) baking powder in the formula; 

however there was no significant difference in the final expansion. 

The authors were most concerned with initial conditions, and 

therefore used 40% baking powder concentration with all reactions 

because the author had used this with many previous experiments 

and sought to standardize the concentration. 

4.3. Effect of pressure change 

Lateral expansion was measured under three pressure schemas 

that were a variation of the ‘standard pressurization method’, as 

Fig. 4. Change in dough density with varying moisture content where dough followed the standard pressurization method: 1 bar in region A, 2 bar in region B, and 1 bar in region C.

described above. The dough was placed directly on the acrylic 

plates as the effects of friction were explored later. For the ‘atmospheric 

pressure method’, the dough sample was at 1 bar the entire 

measurement period. This showed the dough disk expanding for 

15 min. When the standard pressurization method was applied, the 

density quickly increased as the dough contracted, and upon 

release the dough showed gradual expansion. Pressure was introduced 

into the system to keep the CO2 generated within the 

compressed dough and demonstrate increased expansion after 

pressure release. With application of vacuum, the bubble volume 

was expected to be enhanced. To evaluate the effects of a reduced 

pressure environment, the ‘standard vacuum method’ was applied. 

This followed the standard pressurization method, but instead of 

applying 2 bars of pressure during minutes 4–8, a full vacuum was 

applied. The dough expanded with the vacuum, but upon release, it 

collapsed and density increased. The use of vacuum produced the 

least overall expansion with 15.1%. Atmospheric pressure and 

standard pressurization method gave 17.7, and 19.6% expansion 

respectively, as represented in Fig. 5. The vertical lines in the figures 

denote regions of different pressure. 

4.3.1. Pressure pulse (with/without vacuum) 

The pressure pulse experiments were performed by subjecting 

the dough to large pressure differentials using a combination of 

2 bars of pressure or a full vacuum. In all cases, 5 ml Fluoro-Grip Ò -F 

Optically Clear (HD-FEP) Teflon Ò coating (Integument Technologies, 

Inc., Tonawanda, NY, USA) was applied on the acrylic plates as 

a lubricant to reduce friction and enhance the effects of multiple 

changes in pressure. The effect of lubrication is discussed in a later 

section. The first and second test case followed the ‘pressure pulse 

method’ as described above with the addition of Pam Ò aerosol 

lubricant (Pam Ò , International Home FoodsÓ, Parsippany, NJ, USA) 

to the second case. The third test case followed the ‘pressure 

vacuum’ profile where vacuum was applied immediately after 

pressurization. The intermediate rest period that occurred during 

minutes 9–13 in case 1 and 2 led to a greater expansion than case 3 

Fig. 5. The effect of sudden pressure release and sudden increase on dough expansion 

where dough followed the standard pressurization method: 1 bar in region A, 2 bar in 

region B, and 1 bar in region C; atmospheric pressure: 1 bar in region A, B and C; 

standard vacuum method: 1 bar in region A, full vacuum in region B, and 1 bar in 

region C. 


with 20.7, 11.7, and 1.9%, respectively. Case 3 experienced a continuous 

drop in pressure which led to a significantly lower final 

expansion. The addition of a pressurization period increased 

expansion by the potential for distributing gas cells that had begun 

growth during the initial incubation phase. This is seen in Fig. 6, 

where vertical lines denote changes in pressure. The repeated 

change in pressure was performed to study the expansion behavior 

of the dough after significant deformation. 

4.4. Effect of pressurizing gas (air or CO2) 

Compressed air and CO2 gas were compared as the pressurizing 

media to determine if the added environmental CO-2 affected the 

chemical reaction which produced CO2. Each gas was used to 

pressurize the chamber to 2 bars following the standard 

pressurization 

4.4.1. Method 

Using CO2 for pressurization produced a 5.8% greater increase in 

the final dough density as seen in Fig. 7. 

The explanation for this enhancement is that CO2 has a higher 

solubility in the dough than air. CO2 is produced internally from the 

baking powder (sodium bicarbonate component) releasing the CO2 

from the chemical reaction. Environmental CO2 must be solubilized 

and transferred to the gaseous phase in the existing gas cells before 

it can aid in expansion (Eliasson and Larson, 1993). This presents 

the possibility of diffusion effects when multiple gases, e.g. air and 

CO2 are present. 

Fig. 6. The effect of different pressure schema on dough expansion through the 

pressure pulse method: 1 bar in region A, 2 bar in region B, 1 bar in region C, full 

vacuum in region D, and 1 bar in region E; and pressure vacuum method: 1 bar in 

region A, 2 bar in region B, full vacuum in region C, 1 bar in region D and E.

70 

Fig. 7. Effect of CO2 versus air as the pressurizing agent on dough expansion where 

dough followed the standard pressurization method: 1 bar in region A, 2 bar in region 

B, and 1 bar in region C. 

4.5. Effect of lubrication 

The authors were concerned that lateral expansion was inhibited 

by friction between the acrylic plate and the dough. Reducing 

this friction using lubricants (Teflon Ò and Pam Ò ) was done to 

evaluate the role of friction in expansion. The following four cases 

were studied with the standard pressurization method: expansion 

without lubrication, with Teflon Ò , with Pam Ò and with both. As 

expected, during the pressure increase to 2 bars the dough disk 

compressed the greatest with both Teflon Ò and Pam Ò at 30.9%, 

followed by Pam Ò with 19.0%, Teflon Ò with 9.9%, and without 

Fig. 8. Effect of lubrication on dough expansion where dough followed the standard 

pressurization method: 1 bar in region A, 2 bar in region B, and 1 bar in region C. 


lubrication at 4.9%. Following the pressure release, the greatest 

expansion was with Teflon Ò and Pam Ò which expanded 40.1% 

compared to Pam Ò at 30.2% or Teflon Ò at 20.9%. It is interesting to 

note that no lubrication (adding neither Teflon Ò nor Pam) gave 

25.4% expansion which is greater than Teflon Ò alone as shown in 

Fig. 8. However, the difference in using Teflon Ò and no lubrication 

was not significant. It is of worth to note that in Fig. 6, Teflon Ò alone 

was compared to both Teflon and Pam Ò using the pressure pulse 

method. The combination of Teflon Ò and Pam Ò led to an overall 

expansion of 11.7% compared to Teflon Ò alone which was at 20.7%. 

The author would suggest that the friction between the dough and 

Teflon Ò surface restricted the dough’s movement after the full 

vacuum was released. 

The effect of lubrication was an indirect study of the viscosity of 

the dough through the interaction with the stationary plates. The 

combination of dough and Pam Ò can be considered a two-liquid 

flow, while the Teflon Ò is considered stationary. The difference 

between the observed behaviors in the two cases would translate 

into the effect of wall roughness versus a liquid film. 

During expansion, bubbles were seen emerging at the Pam Ò - 

dough interface. Monitoring 1 s intervals at the interface, allows 

one to notice the emergence of bubbles that become entrapped 

within the Pam Ò liquid. This progression of bubble growth is 

highlighted with arrows in Fig. 9 which shows the same bubble 

increasing. The gas released from the dough matrix is believed to be 

the driving force for creation of the bubbles along the plates. 

5. Discussion 

A series of experiments was performed to study the lateral 

expansion of dough during rapid expansion and compression. 

Experiments under varying moisture contents (40, 41, 42, 43, and 

44%) showed a decrease in density with increase in the moisture 

content of the dough. This is consistent with the work of Elmhedi 

et al. (2007) who showed that dough density decreased as a function 

of fermentation time for yeast dough using digital imaging. Lai 

et al. (1989) reported that the addition of sodium bicarbonate 

contributed a weakened structure in wheat dough. 

One can observe the dramatic difference in the final expansion 

through the change in the pressure schema. It is expected that CO2 

is removed more quickly using a full vacuum which reduced the 

amount of retained gas which would have potentially formed 

bubbles. The presence of a short residence period of 1 bar pressure 

following pressurization in the pressure pulse experiments falls in 

line with this hypothesis: the rapid shift to vacuum would help 

deprive the dough of CO2. Application of vacuum after an initial 

growth period was used to understand dough behavior in 

a reduced pressure environment. 

The viscoelastic properties of dough may account for the 

expansion after removal of the deformation induced by pressure or 

vacuum. The elasticity of the dough is related to the release of 

potential energy stored during the deformation by either extension 

or compression forces (Eliasson and Larsson, 1993). The growth 

phenomena under vacuum and pressure applications appear to 

qualitatively mirror each other. A comparison of pressurization by 

compressed air and CO2 showed very similar behavior, with CO2 

showing slightly more expansion and compression tendencies. 

These effects were minor and would not indicate a significant mass 

transfer. One possible occurrence for the small discrepancy could 

be the ability of increased CO2 in the pressurized atmosphere to be 

incorporated into the dough matrix. 

The authors were concerned about potential reduction in the 

dough’s final expansion due to friction at the dough-plate interface. 

To overcome this challenge, Elmhedi et al. (2007) applied mineral 

oil during expansion testing to reduce air bubble entrapment at the 

interface of the measuring vessel and the dough. Teflon Ò film

Fig. 9. Rapid growth of air bubbles through entrapment at the dough-Pam Ò interface: 

(a) top at 9:22.00 min; (b) middle at 9:22.30 min; (c) bottom 9:23.00 min. 

coating and Pam Ò aerosol spray were used to reduce friction for 

these experiments. These additions did not appear to affect the 

qualitative expansion behavior, but they did enhance the expansion 

and compression effects observed. The compression was significantly 

different for pressure pulse experiments when the 


combination of Teflon Ò coating with Pam Ò lubrication was 

compared to only Teflon Ò coating as seen in Fig. 6. As expected, 

combining Teflon Ò and Pam Ò yielded the greatest expansion, but it 

was interesting to note that Teflon Ò had little effect by itself. 

6. Summary and future work 

Experimental studies for bubble growth are inconvenienced by 

the methods’ inability to capture the dynamic phenomena. In this 

paper, a versatile experimental method is developed to allow for 

macroscopic expansion of wheat dough. The study evaluated 

expansion of a dough disk under varying: moisture content (40, 41, 

42, 43, and 44% wb), leavening acid concentration (30, 40, and 50% 

db), pressure schemas, pressurizing gas composition (compressed 

air and CO2), and lubrication (Teflon Ò film coating and Pam Ò 

aerosol lubricant). Dough expansion increased 22.6% by increasing 

moisture content from 40 to 44% (wb). At 44% moisture (wb), this 

dough was highly viscous and 43% moisture content was chosen for 

all experiments. Baking powder was used at an increased level, 40% 

(db) to enhance initial growth conditions and facilitate rapid CO2 

production. ‘Pressure pulse’ and ‘pressure vacuum methods’ added 

pressurization alternatively with full vacuum. The former method 

included a rest period before vacuum application which increased 

expansion by 10.8%. The addition of a pressurization period aided 

expansion through potential for the promotion of distributing gas 

cells that had begun growth during the initial incubation phase. 

Teflon Ò and Pam Ò reduced friction between the dough and acrylic 

plate to increase the final expansion by 14.7% compared to no 

lubricant (‘standard pressurization method’). ‘Pressure pulse’ and 

‘pressure vacuum’ experiments decreased expansion by 28.4 and 

38.2%, respectively compared to ‘standard pressurization’ while 

using Teflon Ò and Pam Ò . 

Several extensions of this work are possible. Work is currently 

underway to model the reaction rate of CO2 production within 

a dough matrix. In addition to this, other upcoming extensions are 

to include the determination of whether the improved expansion 

for higher moisture content was a result of improved consistency or 

increased reaction rate. The incorporation of lateral mass transfer 

could be studied by drilling holes through the acrylic plates. Finally, 

work is to be done on the study of air entrapment and its effect on 

lubrication with an emphasis on possible creation of an air cushion 

between the dough and the acrylic surface. 


The authors would like to thank Apri Melinda, Airine Airine, 

Reinaldo Kusliawan and Louis Nelson III for their invaluable help in 

developing the methodology. We would also like to thank the SURF 

program and NASA-SBIR grant for providing the funding. 

References 

Aguilera, J.M., Germain, J.C., 2007. Advances in image analysis for the study of food 

microstructure. In: McClements, D.J. (Ed.), Understanding and Controlling the 

Microstructure of Complex Food. Woodhead Publishing Limited, Cambridge, pp. 

261–287. 

Alavi, S.H., Gogoi, B.K., Khan, M., Bowman, B.J., Rizvi, S.S.H., 1999. Structural properties 

of protein-stabilized starch-based supercritical fluid extrudates. Food 

Research International 32, 107–118. 

Alavi, S.H., Rizvi, S.S.H., Harriot, P., 2003. Process dynamics of starch based microcellular 

foams produced by supercritical fluid extrusion I: model development 

and II. Numerical simulation and experimental evaluation. Food Research 

International 6, 309–330. 

AOAC International, 2000. Official Methods of Analysis, sixteenth ed. AOAC International, 

Gaithersburg, MD. 

Babin, P., Della Valle, G., Chiron, H., Cloetens, P., Hoszowska, J., Pernot, P., 

Reguerre, A.L., Salvo, L., Dendievel, R., 2006. Fast X-ray tomography analysis of 

bubble growth and foam setting during breadmaking. Journal of Cereal Science 

43 (3), 393–397.

72 

Babin, P., Della Valle, G., Dendievel, R., Lourdin, D., Salvo, L., 2007. X-ray tomography 

study of the cellular structure of extruded starches and its relation with 

expansion phenomenon and foam mechanical properties. Carbohydrate Polymers 

68 (2), 329–340. 

Baker, J.C., Mize, M.D., 1941. The origin of the gas cell in bread dough. Cereal 

Chemistry 18, 19–34. 

Bell, A.V., Berger, K.G., Russo, J.V., White, G.W., Weathers, T.L., 1975. A study of the 

micro-baking of sponges and cakes using cine and television microscopy. 

Journal of Food Technology 10, 147–156. 

Bellido, G.G., Scanlon, M.G., Page, J.H., Hallgrimsson, B., 2006. The bubble size 

distribution in wheat flour dough. Food Research International 39 (10), 

1058–1066. 

Campbell, G.M., Reilly, C.D., Fryer, P.J., Sadd, P.A., 1998. Aeration of bread dough 

during mixing: effect of mixing dough at reduced temperature. Cereal Food 

World 433, 163–167. 

Cauvain, S.P., Whitworth, M.B., Alava, J.M., 1999. The evolution of bubble structure 

in bread doughs and its effect on bread structure. In: Campbell, G.M., Webb, C., 

Pandiella, S.S., Niranjan, K. (Eds.), Bubbles in Foods. Eagan Press, St. Paul, MN, 

pp. 85–88. 

Chen, K.H., Rizvi, S.S.H., 2006. Rheology and expansion of starch-water-CO2 

mixtures with controlled gelatinization by supercritical fluid extrusion. International 

Journal of Food Properties 9, 863–876. 

de Cindio, B., Correra, S., 1995. Mathematical modeling of leavened cereal goods. 

Journal of Food Engineering 24, 379–403. 

Eliasson, A.C., Larson, K., 1993. Cereals in Breadmaking – A Molecular Colloidal 

Approach. Marcel Dekker, Inc, New York. 

Elmehdi, H.M., Page, J.H., Scanlon, M.G., 2003. Using ultrasound to investigate the 

cellular structure of bread crumb. Journal of Cereal Science 38, 33–42. 

Elmehdi, H.M., Page, J.H., Scanlon, M.G., 2007. Evaluating dough density changes 

during fermentation by different techniques. Journal of Cereal Chemistry 84, 

2007. 

Fan, J., Mitchell, J.R., Blanchard, J., 1994. A computer simulation of the dynamics of 

bubble growth and shrinkage during extrudate expansion. Journal of Food 

Engineering 23, 337–356. 

Fan, J., Mitchell, J.R., Blanchard, J.M.V., 1999. A model for the oven rise of dough 

during baking. Journal of Food Engineering 41, 69–77. 

Fox, P., Smith, P.P., Sahi, S., 2004. Ultrasound measurements to monitor the specific 

gravity of food batters. Journal of Food Engineering 65 (3), 317–324. 

Fyrillas, M.M., Kloek, W., van Vliet, T., Mellema, J., 2000. Factors determining the 

stability of a gas cell in an elastic medium. Langmuir 16, 1014–1019. 

Hailemariam, L., Okos, M., Campanella, O., 2007. A mathematical model for the 

isothermal growth of bubbles in wheat dough. Journal of Food Engineering 82, 

466–477. 

http://www.nutritiondata.com/facts-C00001-01c21Ke.html Retrieved on 8/28/ 

2007. 


Kumagai, H., Lee, B.H., Kumagai, H., Yano, T., 1991. Critical radius and time course of 

expansion of an isolated bubble in wheat flour dough under temperature rise. 

Agricultural and Biological Chemistry 55 (4), 1081–1087. 

Lai, C.S., Guetzlaff, J., Hoseney, R.C., 1989. Role of sodium bicarbonate and trapped 

air in extrusion. Journal of Cereal Chemistry 66, 69–73. 

Lee, S., Inglett, G.E., 2006. Rheological and physical evaluation of jet-cooked oat 

bran in low calorie cookies. International Journal of Food Science and Technology 

41, 553–559. 

Matz, S.A., 1992. Bakery Technology and Engineering, third ed. Van Nostrand 

Reinhold, New York. 

Moraru, C.I., Kokini, J.L., 2003. Nucleation and expansion during extrusion and 

microwave heating of cereal foods. Comprehensive Reviews in Food Science 

and Food Safety 2, 120–138. 

Schwartzberg, H.G., Wu, J., Nussinovitch, A., Mugerwa, J., 1995. Modeling deformation 

and flow during vapor-induced puffing. Journal of Food Engineering 25, 

329–372. 

Shimiya, Y., Yano, T., 1988. Rates of shrinkage and growth of air bubbles entrained 

in wheat flour dough. Agricultural and Biological Chemistry 52 (11), 2879– 

2883. 

Singh, A.P., Bhattacharya, M., 2005. Development of dynamic modulus and cell 

opening of dough during baking. Journal of Texture Studies 36, 44–67. 

Strybulevych, A., Leroy, V., Scanlon, M.G., Page, J.H., 2007. Characterizing a model 

food gel containing bubbles and solid inclusions using ultrasound. Soft Matter 3 

(11), 1388–1394. 

Swanson, R.B., Carden, L.A., Parks, S.S., 1999. Effect of a carbohydrate-based fat 

substitute and emulsifying agents on reduced-fat peanut butter cookies. Journal 

of Food Quality 22, 19–29. 

Taki, K., Nakayama, T., Yatsuzuka, T., Ohshima, M., 2003. Visual observations of batch 

and continuous foaming processes. Journal of Cellular Plastics 39, 155–169. 

Trater, A.M., Alavi, S., Rizvi, S.S.H., 2005. Use of non-invasive X-ray microtomography 

for characterizing microstructure of extruded biopolymer foams. 

Food Research International 38 (6), 709–719. 

Wagner, M., Zhang, L., Quellec, S., Doursat, C., Flick, D., Trystram, G., Lucas, T., 2006. 

Role of the crust formation on local expansion during bread baking. In: 

Campbell, G.M., Pyle, L., Scanlon, M. (Eds.), Proceedings of Bubbles in Food: 

Novelty, Health and Luxury, September 10–14, Windermere (UK). 

Wang, L., Ganjyal, G.M., Jones, D.D., Weller, C.L., Hanna, M.A., 2005. Modeling of 

bubble growth dynamics and nonisothermal expansion in starch-based foams 

during extrusion. Advances in Polymer Technology 24 (1), 29–45. 

Whitworth, M.B., Alava, J.M., 1999. The imaging and measurement of bubbles in 

bread doughs. In: Campbell, G.M., Webb, C., Pandiella, S.S., Niranjan, K. (Eds.), 

Bubbles in Foods. Eagan Press, St. Paul, MN, pp. 221–231. 

Zoulias, E., Oreopoulou, V., Kounalaki, E., 2002. Effect of fat and sugar replacement 

on cookie properties. Journal of the Science of Food and Agriculture 82, 1637– 

1644.

Digestibility of protein and starch from sorghum (Sorghum bicolor) 

is linked to biochemical and structural features of grain endosperm 

Joshua H. Wong a , Tsang Lau a , Nick Cai a , Jaswinder Singh a,2 , Jeffrey F. Pedersen b , William H. Vensel c , 

William J. Hurkman c , Jeff D. Wilson d , Peggy G. Lemaux a, *, Bob B. Buchanan a,1 

a Department of Plant and Microbial Biology, University of California, Berkeley, CA 94720, USA 

b United States Department of Agriculture, Agricultural Research Service, Lincoln, NE 68583, USA 

c United States Department of Agriculture, Agricultural Research Service, Western Regional Research Center, Albany, CA 94710, USA 

d United States Department of Agriculture, Agricultural Research Service, Manhattan, KS 66502, USA 

article info 





Keywords: 

Disulfide proteins 

Starch–protein interface 

Granule-bound starch synthase I 

Sorghum 


abstract 

Grain sorghum (Sorghum bicolor L. Moench) ranks fifth in 

worldwide production among cereal crops, after wheat, rice, corn 

and barley. The U.S. is the number two producer and the number 

one exporter of sorghum, primarily to Mexico for use as animal feed 

(U.S. Grains Council, 2006). Popularity of sorghum is due in part to 

its ability to produce reasonable yields in warmer, drier regions. 

Because subsistence farmers in Africa and Asia cultivate sorghum 

widely as a staple food for home consumption, the crop is 

Abbreviations: DTT, dithiothreitol; GBSS, granule-bound starch synthase; 

IVDMD, in vitro dry matter disappearance; ME, 2-mercaptoethanol; Mr, relative 

molecular mass; PAGE, polyacrylamide gel electrophoresis; SDS, sodium dodecyl 

sulfate; MOPS, 3-(N-morpholino) propane sulfonic acid. 

* Corresponding author. Tel.: þ1 5106421589; fax: þ1 5106427356. 

E-mail addresses: viewmont@nature.berkeley.edu (J.H. Wong), shout_at_neko@ 

hotmail.com (T. Lau), monto@nature.berkeley.edu (N. Cai), jaswinder.singh@ 

mcgill.ca (J. Singh), jeff.pedersen@ars.usda.gov (J.F. Pedersen), vensel@pw.usda.gov 

(W.H. Vensel), bhurkman@pw.usda.gov (W.J. Hurkman), jeff.d.wilson@ars.usda.gov 

(J.D. Wilson), lemauxpg@nature.berkeley.edu (P.G. Lemaux), view@nature. 

berkeley.edu (B.B. Buchanan). 

1 

Submitting author. 

2 

Present address: McGill University, Macdonald Campus, 21,111 Lakeshore Road, 

Ste-Anne-de-Bellevue, Quebec H9X 3V9, Canada. 


doi:10.1016/j.jcs.2008.07.013 





Although a principal source of energy and protein for millions of the world’s poorest people, the 

nutritional value of sorghum (Sorghum bicolor L. Moench) is diminished because of low digestibility of 

grain protein and starch. To address this problem, we analyzed the properties of two sorghum lines that 

have a common pedigree but differ in digestibility. Consistent with results based on a ruminal fluid assay, 

the protein and starch of one line (KS48) was more thoroughly digested than that of the other (KS51) 

using in vitro assays based on pepsin and a-amylase. The indigestibility of KS51 relative to KS48 was 

shown to be due to (i) a greater abundance of disulfide-bonded proteins; (ii) presence in KS51 of nonwaxy 

starch and the accompanying granule-bound starch synthase; and (iii) the differing nature of the 

protein matrix and its interaction with starch. The current findings suggest that each of these factors 

should be considered in efforts to enhance the nutritional value of sorghum grain. 


a principal source of energy and protein for millions of the world’s 

poorest (Klopfenstein and Hoseney, 1995). 

Seed proteins of sorghum are less digestible than those of other 

cereals and digestibility is exacerbated by wet cooking the meal or 

flour, which results in significant nutritional losses (Hamaker et al., 

1986; MacLean et al., 1981; Mertz et al., 1984). Numerous factors 

contribute to the digestibility problem [for review, see (Duodu 

et al., 2003)]. Exogenous aspects include the interaction of protein 

with non-protein components, such as polyphenols, starch, nonstarch 

polysaccharides, phytates and lipids. Endogenous factors 

arise from the nature of the proteins themselves and their organization 

within the grain (Belton et al., 2006; Duodu et al., 2003; 

Ezeogu et al., 2005, 2008; Hamaker and Bugusu, 2003). As in other 

cereals, the low amount of protein relative to starch, i.e., approximately 

10% protein, vs. 70–80% starch, based on grain dry weight 

(Rooney and Pflugfelder, 1986), affects the functional properties of 

starch, such as gelatinization and digestion rate, to a greater extent 

in sorghum than in other cereals (Chandrashekar and Kirleis, 1988; 

Duodu et al., 2002; Ezeogu et al., 2005, 2008). 

The proteins of the sorghum grain are classically divided, based on 

solubility in different solvents (Jambunathan et al., 1975; Landry and 

Moureaux, 1970): albumins (water-soluble), globulins (salt-soluble), 

kafirins (prolamins, aqueous alcohol-soluble), cross-linked kafirins 

(aqueous alcohol þ reducing agent-soluble), cross-linked glutelins 

(detergent þ reducing agent þ alkaline pH-soluble) and unextracted

74 

structural protein residue. A newer and more simplified classification 

scheme for sorghum proteins has been proposed that divides them 

into two groups, kafirins and non-kafirins. This scheme is based on 

the homogeneous nature and varied origin of the kafirin storage 

prolamins relative to the heterogeneous nature of the non-kafirin 

proteins (i.e., albumins, globulins and glutelins) that are involved in 

cellular functions (Hamaker and Bugusu, 2003; Hamaker et al.,1995). 

Despite a high degree of sequence homology to maize zeins, sorghum 

storage proteins contain a higher proportion of cross-linked fractions 

and are more hydrophobic, explaining their greater propensity to 

form intermolecular disulfide-cross linkages and possibly additional 

protein aggregates that could facilitate the formation of more covalent 

bonds compared to maize zeins (Belton et al., 2006; Hamaker 

and Bugusu, 2003). 

Kafirins, comprising 70–80% of the protein in whole grain 

sorghum flour (Hamaker et al., 1995), are synthesized and translocated 

into the lumen of the endoplasmic reticulum where they 

form protein bodies (Taylor et al., 1985). They are subclassified as: 

a- (23 and 25 kDa), b- (20 kDa) and g- (28 kDa) types based on 

molecular weight, extractability, structure and cross-reactivity with 

sera against analogous maize zeins (Belton et al., 2006; Mazhar 

et al.,1993; Shull et al., 1991). Comprising w80% of total kafirins, the 

a-type is considered the principal storage protein, followed by the 

g- (w15%) and b- (w5%) members. Recently, a DNA-derived 

sequence for a d-kafirin was shown to have high homology with the 

Mr 10,000 d-zein, except for absence of part of the methionine-rich 

region (Belton et al., 2006; Izquierdo and Godwin, 2005). a-Kafirins 

have low levels of cysteine relative to the b- and g-types (5 and 

7 mol%, respectively), whereas d-kafirins are rich in methionine 

(16–18 mol%) but lack cysteine. Electron microscopy of internal 

protein body structure reveals that g-kafirins, and to a lesser extent 

b-kafirins, encapsulate the more digestible a-kafirins in a disulfidebond 

polymer network (Oria et al., 2000; Shull et al., 1992), thereby 

impeding exposure to proteases. 

In the corneous endosperm, non-kafirins (albumins, globulins, 

glutelins) form around protein bodies, effectively ‘‘gluing’’ the 

bodies into a matrix surrounding the starch granules (Hamaker and 

Bugusu, 2003; Shull et al., 1990; Taylor et al., 1984). This protein 

matrix appears to act as a barrier to starch gelatinization and 

digestibility (Chandrashekar and Kirleis, 1988; Duodu et al., 2002; 

Ezeogu et al., 2005, 2008) due to cross-linking between g- and bkafirins 

and matrix proteins (Duodu et al., 2001; Hamaker and 

Bugusu, 2003; Oria et al., 1995a). Cooking reduces digestibility by 

effecting a conformational change in proteins that could facilitate 

formation of disulfide-linked polymers (Axtell et al., 1981; Duodu 

et al., 2002, 2003; Hamaker et al., 1987; Oria et al., 1995b). The 

negative impact of cooking on protein digestibility was mitigated 

by addition of 2-mercaptoethanol (ME) or other reducing agents 

(Elkhalifa et al., 1999; Hamaker et al., 1987). Sorghum grains rich in 

kafirin-containing protein bodies also have a lower capacity for 

starch gelatinization (Chandrashekar and Kirleis, 1988; Ezeogu 

et al., 2005, 2008)dan observation consistent with the finding that 

adding ME during cooking to cleave disulfide bonds within the 

protein matrix increased the degree of starch gelatinization and 

digestion (Elkhalifa et al., 1999; Ezeogu et al., 2005, 2008; Zhang 

and Hamaker, 1998). 

The protein barrier surrounding the starch granule may also 

reduce the hydrolysis of native and processed starch by amylolytic 

enzymes (Rooney and Pflugfelder, 1986). Thus, the addition of 

pronase to hydrolyze the protein matrix significantly enhanced in 

vitro rates of starch hydrolysis by increasing surface area and 

enabling starch to interact with a-amylase and amyloglucosidase 

(Rooney and Pflugfelder, 1986). Similar effects were observed when 

sorghum flour was either treated with pepsin before cooking or 

cooked in the presence of dithiothreitol (DTT) or other reductants 

(Zhang and Hamaker, 1998). Abundance of starch granules may also 

J.H. Wong et al. / Journal of Cereal Science 49 (2009) 73–82 

decrease proteolysis by limiting accessibility of proteolytic 

enzymes, especially when gelatinized during cooking (Duodu et al., 

2002). Uniqueness of the protein matrix and its interaction with 

starch that affect the rate of starch digestion are key differences 

between the feed quality of sorghum and corn (Rooney and Miller, 

1982; Rooney and Pflugfelder, 1986). In addition, endosperm 

texture and cooking conditions have been shown to have a significant 

effect on in vitro digestibility of starch and protein in these two 

cereals (Duodu et al., 2002, 2003; Ezeogu et al., 2005, 2008). 

Sorghum displays significant variation in rates of starch disappearance 

(Wester et al., 1992). Based on a 12-h incubation in the in 

vitro dry matter disappearance (IVDMD) assay, loss of dry matter in 

sorghum correlated closely with rates of starch, but not necessarily 

protein digestion (Pedersen et al., 2000). 

Collectively, these findings suggest that properties of starch and 

protein in sorghum could affect their mutual digestibility. The 

majority of published references deal with digestibility of the 

proteins and their impact on starch, but evidence for the reverse is 

scant (Duodu et al., 2002). To pursue this issue and assess the 

importance of relevant factors in a single study, we investigated 

how protein and starch influence the breakdown of one another by 

comparing two sorghum lines having a common pedigree but 

differing in digestibility. 


2.1. Grain and preparation of materials 

The sorghum lines KS48 and KS51 were selected for these 

experiments because of previously determined differences in 

digestibility but the identical pedigree, Texioca-63 Short Kaura 

(Casady, 1972). Both have a clear pericarp and yellow endosperm. 

Field-grown seeds were produced at the University of Nebraska 

Agricultural Research Development Center, Ithaca, NE; additional 

seeds were generated from greenhouse-grown plants at the 

University of California, Berkeley. Seed size and hardness of KS48 

and KS51 were assessed using a Perten SKCS 4100 Single Kernel 

Characterization System instrument. The starch content of KS48 

and KS51 was 64.7% and 67.8%, respectively, on a dry weight basis. 

For 12-h IVDMD analysis, mature dry seeds were ground in a Wiley 

mill to pass a 20-mesh screen. For all other analyses, seeds were 

ground in a Wiley mill to pass a 40-mesh screen. Ground meal was 

stored at 25 C until used in digestion assays. Pepsin (porcine 

stomach mucosa, P-7000) and three types of a-amylases – bacterial 

(A-3403, Type XII-A), porcine pancreas (A-3176, Type VI-B) and 

human saliva (A-1031, Type XIII-A) – were used in digestibility 

assays (Sigma, St Louis, MO). 

2.2. Microscopy studies 

For microscopic observations, seeds were sectioned with a razor 

blade and stained with iodine (0.3 g I2, 1.0 g KI in 100 ml water). 

Sectioned seeds were used for electron microscopic analyses, using 

a Hitachi TM 1000 environmental scanning electron microscope 

(Hitachi High Technologies America, Inc., Pleasanton, CA). Prior to 

analysis, sections were coated with gold-palladium to 1 mm using 

a Tousimis sputter machine (Rockville, MD). 

2.3. Twelve-hour IVDMD of sorghum meal 

Sorghum meal was digested for 12 h in rumen fluid inoculum 

obtained from a ruminally fistulated steer using previously 

described methods (Pedersen et al., 2000).

2.4. Sequential protein extraction 

Sorghum meal (2 g) was sequentially extracted with the six 

solvents listed below in 20 ml lots at 25 C and centrifuged 

(18,900g, 10 min) at 4 C after each step [modified from (Hamaker 

et al., 1986)]. Initially, meal was extracted with 0.5 M NaCl by 

shaking for 60 min. Resulting NaCl-soluble albumins and globulins 

(fraction 1) were retained, and the pellet sequentially extracted 

yielding the following supernates: double distilled H2O (ddH2O) for 

20 min (water-wash fraction, fraction 2); 60% 2-propanol (v/v) for 

4 h (kafirins, fraction 3); 0.1 M borate buffer, pH 10.8, for 4 h (glutelin-like 

protein, fraction 4); 60% 2-propanol with 1% dithiothreitol 

(DTT) for 4 h (cross-linked kafirins, fraction 5); and 0.1 M borate 

buffer, pH 10.8, containing 1% DTT and 1% sodium dodecyl sulfate 

(SDS) for 18 h at 4 C (cross-linked glutelins, fraction 6). Protein 

fractions were stored at 20 C. 

2.5. Disulfide protein content of selected fractions 

Equal amounts of protein in fractions extracted without 

reducing agent were reduced with 1 mM DTT; protein sulfhydryl 

groups were labeled with monobromobimane (mBBr) and analyzed 

using SDS-PAGE (Wong et al., 2004). Relative intensity of fluorescence 

of labeled proteins was equated to relative amount of protein 

disulfides. 

2.6. Rapid in vitro pepsin digestion 

Time course of in vitro pepsin digestion of sorghum meal and 

processing of the resulting undigested residue were as described 

(Aboubacar et al., 2001; Nunes et al., 2004). 

2.7. Extraction of protein undigested by pepsin 

Protein in pellets, not digested by pepsin, was extracted with 

0.0125 M borate buffer, pH 10 containing 1% SDS and 2% 2-ME as 

described (Aboubacar et al., 2001). The extraction was modified 

further to include extraction of a granule-bound starch synthase I 

(GBSSI) from the borate–SDS–ME extracted residue (see Section 2.8). 

2.8. Extraction of granule-bound starch synthase 

To determine if unextracted protein remained in residues after 

extraction with borate–SDS–ME (0.0125 M borate, 1% SDS, 2% ME, 

pH 10), the washed pellets (in 0.5 ml of the same buffer) were 

mixed with 1 ml borate–SDS–ME buffer and extracted for 5–10 min 

in a boiling water bath and cooled in tap water. Gelatinized starch 

was pelleted by centrifugation (20,800g, 10 min) and the clear 

supernate was saved for SDS-PAGE analysis. 

2.9. SDS-PAGE 

A Criterion Pre-Cast Gel System using Tris–HCl gels (Bio-Rad, 

Hercules, CA) was used for SDS-PAGE analysis (Wong et al., 2004). 

Equal amounts of protein were used for protein distribution analyses 

and estimation of sulfhydryl content. For protein digestion 

analysis, equal volumes (w5 ml) with a specified amount of 

extracted protein were used. 

2.10. SDS-PAGE of kafirins and glutelins undigested by pepsin 

NuPAGE Novex Bis–Tris gels (Invitrogen, Carlsbad, CA), with 

a neutral pH to minimize protein modifications, were used to 

obtain optimal resolution of small- to medium-sized proteins. 

Aliquots (w5 ml) of specific amounts of extracted protein were 

mixed with 1/4 volume of 4 NuPAGE sample buffer plus 0.4% ME, 

J.H. Wong et al. / Journal of Cereal Science 49 (2009) 73–82 75 

boiled 5 min, centrifuged and run in 12% Bis–Tris gels with MOPS 

buffer for 1 h 20 min at 150 V. Fluorescent imaging and protein 

staining were as described (Section 2.11). 

2.11. Quantification of fluorescent proteins 

To determine redox status, fluorescent (unsaturated and saturated) 

images of reduced proteins were captured from gels on a Gel 

Doc 1000 imager using Quantity One software (Bio-Rad, Hercules, 

CA) with 365 nm UV light either immediately or after a 15 min 

water-wash; images were saved in TIFF format. Gels were stained 

for protein with colloidal Coomassie Blue G-250 overnight at 25 C, 

destained in ddH2O and scanned with a UMAX PowerLook 1100 

scanner using UMAX MagicScan version 4.4 within Adobe Photoshop, 

v.6 (Adobe Systems, San Jose, CA). Amount of protein, 

expressed as volume (intensity area) of fluorescent bands and/or 

undigested kafirin bands, was quantified with Quantity One software. 

Higher protein band volume indicates more protein in the 

gel, and thus lower digestibility (Aboubacar et al., 2001). 

2.12. 2-D gel electrophoresis 

Protein contained in equal volumes of extracts was precipitated 

by addition of 5 vol of cold ( 20 C) acetone and incubation 

at 20 C overnight. Proteins were recovered by centrifugation and 

separated by 2-D gel electrophoresis (Wong et al., 2004). 

2.13. Western blot analysis 

After electrophoresis, proteins were transferred to nitrocellulose 

membranes at 4 C for 75 min at 50 V constant voltage (Wong et al., 

2004) and cross-reactivity analyzed with waxy protein antibody 

(Terada et al., 2000). 

2.14. In vitro starch digestion 

Digestibility of starch in uncooked sorghum meal was measured 

using a modified hydrolysis method (Zhang and Hamaker, 1998) 

with bacterial a-amylase (A-3403, Type XII-A, Sigma Chemical Co., 

St. Louis, MO). 

2.15. Sequential addition of a-amylase/pepsin or pepsin/a-amylase 

2.15.1. a-Amylase followed by pepsin 

Meal (50 mg) was treated with 50 mg a-amylase (human saliva 

form, Sigma A-1031, Type XIII-A) in 0.3 ml 1 mM Na glycerophosphate–HCl 

buffer (25 mM NaCl, 5 mM CaCl2, 0.02% w/v azide), 

pH 6.9, at 37 C for 1 h with shaking. Pepsin solution (1 ml at 20 mg/ 

ml, pH 2.0) was added, the mixture incubated for 2 h at 37 C and 

neutralized with 0.1 ml 2 N NaOH. Supernate containing a-amylaseand 

pepsin-digested material was collected by centrifugation 

(20,800g, 10 min) for reducing sugar determination. Pellet was 

washed with 1 ml 0.1 M KH2PO4 buffer, pH 7.0 and then 1 ml ddH2O 

and extracted for remaining kafirins and GBSSI as above. Controls 

were: (A) a-amylase alone, (B) pepsin alone and (C) buffer alone. 

2.15.2. Pepsin followed by a-amylase 

Meal (50 mg) was incubated with pepsin (1 ml at 20 mg/ml, pH 

2.0) for 2 h at 37 C with shaking and then neutralized (0.1 ml 2 N 

NaOH to pH 7.0). a-Amylase (50 mg) in 0.3 ml Na glycerophosphate 

buffer as above, was added and the mixture incubated for 1 h at 37 C. 

Reaction was stopped by acidification and the supernate containing 

pepsin- and a-amylase-digestible materials was saved for reducing 

sugar determination. Pellets were washed 2 with 1 ml ddH2O and 

extracted for kafirins and GBSSI (Sections 2.7 and 2.8). Controls were: 

(A) pepsin alone, (B) a-amylase alone and (C) buffer alone.

76 

2.16. Effect of DTT reduction on digestion of cross-linked kafirins 

and glutelins by pepsin 

Meal (50 mg) was incubated 5 mM DTT in 0.5 ml of 0.1 M 

KH2PO4, pH 7.0, for 1 h at 37 C and supernate removed by centrifugation 

(20,800g, 10 min). The DTT-treated pellet was washed 

with 1 ml (A) ddH2O and (B) 0.1 M KH2PO4 buffer, pH 2.0. The 

washed residue was incubated with pepsin (1 ml at 20 mg/ml in pH 

2.0 KH2PO4 buffer) at 37 C for 2 h with shaking and digestion was 

stopped with 0.1 ml 2 N NaOH. Pepsin-treated residue was 

extracted for remaining kafirins and glutelins and analyzed by 

NuPAGE as above. 

2.17. Granule-bound starch synthase (GBSSI) identification 

Coomassie blue-stained spots were excised from 2-D gels and 

placed in a 96-well reaction tray. After proteins were destained, 

reduced, alkylated and cleaved with trypsin by a robotic sample 

handler, they were analyzed by tandem mass spectrometry and 

identified as described (Balmer et al., 2006). 

2.18. Other analytical procedures 

Protein was estimated in buffer-soluble extracts by the dyebinding 

method and those in the presence of SDS and ME or DTT by 

the Non-Interfering Protein Assay (Geno Technology, Inc., St. Louis, 

MO) as described (Wong et al., 2004). The ratio of amylose and 

amylopectin was determined using a MegazymeÔ Amylose/ 

Amylopectin Assay Kit (Megazyme, Bray, Ireland). 


The 12-h IVDMD assay estimates the in vitro rate of starch 

digestion, which is highly correlated to feed efficiency or gain/feed 

ratio (r ¼ 0.94) (Stock et al., 1987). Because it mimics biochemical 

conditions in cattle, use of ruminal fluid containing hydrolytic 

enzymes produced by the ruminant and its microbial flora is relevant 

to studies of this type. The 12-h IVDMD applied to the two 

sorghum lines, KS48 and KS51, gave digestibility values of 33.0% 

and 22.1%, respectively. The results revealed a difference in dry 

matter disappearance and, hence, in the rates of digestion of starch 

as well as protein. The difference revealed by IVDMD, plus the 

common pedigree of these two lines, led to their choice for further 

study of digestibility. In the present work, grain from the two lines 

are subjected to in vitro pepsin and a-amylase assays that are 

generally considered to reflect activities of the human stomach and 

small intestine, respectively (Astwood et al., 1996; Zhang and 

Hamaker, 1998). 

Table 1 

Protein distribution of two sorghum lines with a common pedigree differing in digestibility 

KS48 a 

3.1. Protein distribution of KS48 and KS51 

The distribution of protein in fractions extracted with the indicated 

solvents (Section 2.4) suggested that, in general, the two 

sorghum lines yielded approximately the same amount of total 

extractable protein (Table 1). However, differences in distribution 

of protein between the two lines suggested that the more digestible 

KS48 contains more protein than the less digestible KS51 in the first 

five fractions extracteddddH2O wash, albumin and globulin (saltsoluble), 

kafirin (60% 2-propanol), glutelin-like protein (borate 

soluble), and cross-linked kafirin (2-propanol þ DTT). By contrast, 

the least soluble (last-extracted) fraction, cross-linked glutelin 

(borate þ DTT þ SDS), represented a considerably greater fraction 

in KS51 than in KS48. To determine whether these differences in 

protein distribution affect digestibility, we examined the composition 

and redox status of protein in the six fractions (Fig. 1). 

Disulfide bonds between cysteine residues, a common feature of 

most cereal seed proteins, have long been known to stabilize 

various molecular structures (Wall, 1971). Many proteins with 

disulfide bonds resist digestion by proteases (Astwood et al., 1996; 

Opstvedt et al., 1984; del Val et al., 1999). Since a large percentage of 

sorghum kafirin storage proteins exist in polymeric forms linked by 

disulfide bonds in their native state (Duodu et al., 2002, 2003; El 

Nour et al., 1998; Hamaker et al., 1987; Oria et al., 1995b), differences 

in content of fractions rich in insoluble disulfide proteins, i.e., 

2-propanol þ DTT (cross-linked kafirin) and borate þ DTT þ SDS 

(cross-linked glutelin) could contribute to protein digestibility 

differences. Accordingly, although KS48 had more 2-propanol 

þ DTT extractable proteins (cross-linked kafirins), the content 

of borate þ DTT þ SDS-extractable glutelins was appreciably 

greater in KS51 than in KS48 – a reflection of the relative abundance 

of intermolecular S–S groups in this fraction (Table 1). The latter 

could be a factor in the lower digestibility of KS51. 

3.2. Estimation of disulfide protein status 

Protein disulfide content was estimated by determining the 

sulfhydryl groups resulting from DTT reduction, using a thiolspecific 

fluorescent probe, mBBr (Wong et al., 2004). Four fractions 

(nos. 1–4 in Table 1), extracted without reducing agent, were 

monitored for DTT-induced sulfhydryl changes. KS51 showed more 

intense fluorescence relative to KS48 in fraction 1 salt (albumin and 

globulin) and fraction 4 borate (glutelin-like) (Fig. 1B). The more 

pronounced increase in proteins with disulfide bonds in fraction 6 

(borate þ DTT þ SDS) could also contribute to the lower digestibility 

of KS51, as could the markedly different distribution of 

proteins in that fraction (Fig. 1A). For reasons not yet understood, 

the fluorescence intensity of the kafirin (propanol) fraction showed 

larger relative differences between the two lines in untreated vs. 

DTT-treated samples (Fig. 1B, lanes 9 vs. 11; 10 vs. 12). In summary, 

Fraction Extraction solvent Total mg protein extracted % Total protein extracted Total mg protein extracted % Total protein extracted D, % 

1 NaCl, 0.5 M 28.3 0.5 19.0 27.6 0.4 18.8 0.2 

2 H2O-wash 0.4 1.0 0.3 0.4 1.4 0.3 0.0 

3 2-Propanol, 60% 17.3 1.0 11.6 12.4 1.3 8.5 3.1 

4 Borate, 0.1 M pH 10.8 5.7 0.7 3.8 4.4 1.2 3.0 0.8 

5 2-Propanol, 60% þ 1% DTT 30.9 2.8 20.8 27.8 4.8 19.0 1.8 

6 Borate, 0.1 M pH 10.8 þ 1% 

DTT þ 1% SDS 

66.3 3.1 44.5 73.8 4.3 50.4 5.9 

Total 148.9 1.5 100 146.4 2.2 100 – 

a Mean of 5 experiments. 


KS51 a 

KS48–KS51

the relative abundance of intermolecular S–S groups in the crosslinked 

glutelin fraction in KS51 suggests that the non-kafirins, 

which make up the protein matrix, have a strong influence on the 

digestibility of KS51 relative to KS48. 

3.3. Time course of in vitro pepsin digestion 

The gel-based system used to analyze protein digestion by 

pepsin (Aboubacar et al., 2001; Nunes et al., 2004) requires less 

material, is faster and can accommodate more samples than 

a commonly used batch procedure (Mertz et al., 1984). The gel 

procedure not only reveals the types of undigested kafirins, but also 


Fig. 1. Effect of DTT on redox status of protein fractions extracted from grain of KS48 and KS51. (A): Protein stained with Coomassie blue; (B): protein labeled with mBBr visualized 

by UV absorption. Tris–HCl gel was used. Individual protein fractions are identified above (A). 

allows an estimate of the percentage digested over time. This gel 

system has been successfully used to separate sorghum proteins 

(Bean, 2003) and wheat glutenins (Kasarda et al., 1998). In our 

study, we observed that separation of different kafirin types is 

improved in the neutral pH environment of the NuPAGE Novex Bis– 

Tris gels relative to that seen with the conventional Tris–HCl gels 

(cf. Fig. 2Avs.Fig. 1A). A NuPAGE gel image (Fig. 2A) shows that high 

Mr components of the insoluble protein fraction, including glutelins 

(35–100 kDa), were readily digested, while the smaller kafirins (Mr 

18–27 kDa) were more resistant. Further, the time course of the in 

vitro digestion of this fraction demonstrated that kafirins in KS48 

were digested more rapidly by pepsin than those in KS51 (Fig. 2A). 

Fig. 2. Time course of in vitro digestion of storage protein and starch in KS48 and KS51. (A): BT-NuPAGE gel showing digestion of glutelins and kafirins over time; (B): estimation of 

glutelin and kafirin digestion rate with the NI protein assay. Results were calculated by linear regression and expressed as % of control without pepsin; (C): starch digestion by 

bacterial a-amylase. Corn starch is shown as a control.

78 

Gel densitometry measurements revealed that various kafirins, a, 

b and g (El Nour et al., 1998), were digested at different rates: 

b > g > a (data not shown). Furthermore, data on the disappearance 

with time of insoluble proteins (kafirins and glutelins) (Fig. 2B) 

indicated that KS48 was digested more rapidly than KS51 – negative 

slope of 0.643 (SE 0.010) for KS48 vs. 0.530 (SE 0.023) for 

KS51, mean of three determinations. These values, which represent 

a 22% rate difference, positively correlate with the corresponding 

IVDMD difference of 36%. The lower value obtained with the in vitro 

pepsin method could be due to use of a single protease vs. a mixture 

of proteolytic and amylolytic enzymes present in ruminal fluid used 

in the IVDMD procedure. 

3.4. Time course of in vitro starch digestion by a-amylase 

The finding that the 12-h IVDMD value was higher for KS48 than 

KS51 suggested that starch digestion alone might follow the same 

trend, and this was confirmed (Fig. 2C). The average of three timecourse 

experiments for in vitro digestion of meal from KS48 and 

KS51 by bacterial a-amylase gave slopes of 1.74 10 3 (SE 

1.69 10 4 ) and 1.41 10 3 (SE 5.01 10 5 ), respectively. These 

results represent a 23% difference in starch digestion rate between 

the two lines – a value similar to that obtained with the in vitro 

pepsin (Fig. 2B) and IVDMD (Pedersen et al., 2000) procedures. 

More significantly, the three independent in vitro techniques used 

to assess digestibility showed a positive correlation. This is the first 

time the three approaches have been compared in a single study. 

Ezeogu et al. (2005) made significant contributions to the idea 

that in vitro digestibility of starch in sorghum and maize flours is 

related to endosperm texture and cooking conditions. These 

researchers found that starch digestion was significantly higher in 

floury sorghum endosperm than in vitreous endosperm, while 

their counterparts in maize were similar in this regard. Cooking 

with ME increased starch digestion in both sorghum and maize, but 

more so with sorghum, and with vitreous endosperm flours. This is 

consistent with the fact that more polymeric kafirins, formed by 

intermolecular disulfide bonds (Chandrashekar and Mazhar, 1999; 

Oria et al., 1995a), occurred in the vitreous endosperm fraction 

(Ezeogu et al., 2005; Kumari and Chandrashekar, 1994), probably 

because the cysteine-rich g- and b-kafirins abound in that part of 

the sorghum grain (Chandrashekar and Mazhar, 1999; Ezeogu et al., 

2005, 2008; Mazhar and Chandrashekar, 1995). Unlike their findings, 

which were based on similar types of starch – i.e., a 75:25 

amylopectin/amylose ratio from different parts of the sorghum and 

maize endosperm – our results are based on different types of 


starch, i.e., an amylopectin/amylose ratio of 97:3 and 70:30, for 

KS48 and KS51, respectively (see below). 

3.5. Starch digestibility differences in starch in KS48 and KS51 

In the original rapid protein digestion assay (Aboubacar et al., 

2001), the pellet remaining after extraction of kafirins and glutelins 

with borate–SDS–ME was not further analyzed. In this study, we 

attempted to fill this gap by analyzing the residual pellet that still 

contained between 2.2–2.5% (pepsin-treated) and 4.7–5.0% (buffer 

control) of protein nitrogen (data not shown). After boiling the 

residual pellet with excess borate–SDS–ME buffer, we unexpectedly 

observed a marked difference in the gelatinization properties 

of starch in KS48 and KS51. The KS48 pellet yielded a transparent, 

soft starch gel that swelled to a greater extent than the KS51 

counterpart. Further, when equal amounts of extraction buffer 

were added to the pellets before boiling, only a small amount of 

supernate was recovered after centrifuging the cooled gel KS48 

preparation, whereas KS51 yielded a larger amount of clear 

supernate that separated easily from the harder opaque gel (data 

not shown). The gelatinization differences in KS48 and KS51 starch 

resemble those observed for waxy vs. non-waxy starch. 

In addition to obvious swelling differences in gelatinized starch, 

supernates from KS48 and KS51 contained different proteins. Onedimensional 

SDS-PAGE analysis revealed a major protein band of 

Mr w 58 kDa in KS51, not detected in KS48 (Fig. 3A). Because it was 

released from starch granules only after gelatinization, the protein 

did not appear to change after treating the parent meal with pepsin 

(data not shown). Two-dimensional gel analysis gave information 

on the nature of the unknown band (Fig. 3C, D). Two major 61 kDa 

spots with acidic pI’s, uniquely observed in KS51 (note arrows, 

Fig. 3D), were identified as granule-bound starch synthase I by 

mass spectrometry [GBSSI, chloroplast (sorghum) accession No. 

Q43134]. Identity was further confirmed by Western blot analysis 

using an antibody against maize waxy protein (Fig. 3B). In KS51, 

a highly antibody-reactive band was seen at 58 kDa and a less 

reactive one at 61 kDa, corresponding to two isoforms of GBSSI: 

a major form in the endosperm (58 kDa) and a minor form in the 

pericarp (61 kDa) (Nakamura et al., 1998). While lower levels of the 

pericarp isoform were seen in KS48 relative to KS51, the endosperm 

component was not detected. The granule-bound isoform of starch 

synthase, GBSSI, also referred to as the waxy protein encoded in the 

waxy (Wx) locus of cereals, functions specifically to elongate amylose 

(Denyer et al., 2001) by adding glucose residues in a-1,4-glycosidic 

Fig. 3. Starch-bound proteins extracted from KS48 and KS51. The pellet remaining after removing glutelins and kafirins was boiled with borate–SDS–2-ME buffer and the proteins 

solubilized were analyzed by one-dimensional SDS-PAGE. (A): Protein; (B): Western blot analysis using waxy protein antibody; (C and D): two-dimensional gel analysis of starchbound 

proteins extracted in (A).

linkage. Recent genetic experiments have revealed that sorghum 

contains at least two GBSS genes (Pedersen et al., 2005). 

3.6. Starch of waxy and non-waxy sorghum 

Normal or non-waxy sorghum starch has approximately 75% 

amylopectin and 25% amylose, while waxy starch contains nearly 

100% amylopectin (Ezeogu et al., 2005; Rooney and Miller, 1982; 

Rooney and Pflugfelder, 1986). The apparent absence of GBSSI 

protein in the endosperm of KS48 (Fig. 3A, B) indicates that its 

starch should be low in amylose and contain mainly amylopectin. 

Waxy and non-waxy polymers of starch can be distinguished by 

iodine staining (Denyer et al., 2001; Pedersen et al., 2004). Starch of 

non-waxy lines, containing amylose and amylopectin, forms blueblack 

complexes with iodine, while starch from waxy lines, lacking 

amylose, stains reddish brown. Since the waxy nature of KS48 and 

KS51 was not indicated in the USDA-ARS, Genetic Resource Information 

Network (Pedersen et al., 2004), grains from the two lines 

were stained with iodine. Results confirmed KS48 has a waxy 

(brownish red) and KS51 a non-waxy (dark blue) type of endosperm 

(see below). Biochemical analyses confirmed this observation: 

KS48 contains 3% and KS51 30% amylose. 

Numerous studies have shown that starch is more readily 

digested in waxy sorghum than in non-waxy types [for review, see 

(Rooney and Pflugfelder, 1986)]. Our results are consistent with this 

conclusion, not only for starch but also for protein (KS48 > KS51). 

Digestibility of protein, however, may not necessarily be consistent 

with waxy vs. non-waxy sorghum lines. The nature of the protein 

matrix and the extent of embedded starch in the endosperm are 

proposed to account for this inconsistency (Chandrashekar and 

Kirleis, 1988; Duodu et al., 2003; Rooney and Miller, 1982; Rooney 

and Pflugfelder, 1986; Shull et al., 1990). To gain insight into this 

point, we conducted experiments with our waxy and non-waxy 

lines as described below. 

3.7. Sequential addition of a-amylase and pepsin 

Digestion protocols, using sequential addition of pepsin followed 

by a-amylase, and vice versa, were designed to study the 

inter-relationship of starch and protein in KS48 and KS51. Prior 

digestion with pepsin, followed by treatment with human saliva aamylase 

(designated P/A in Table 2), enhanced to a greater extent 

starch hydrolysis, relative to treatment with a-amylase followed by 

pepsin (A/P), in KS48 compared to KS51 (1.8- vs. 1.5-fold, respectively). 

This observation suggests that KS51 starch granules were 

intrinsically harder to digest than those of KS48, even if the protein 

matrix was partially removed by prior pepsin digestion. This 

explanation agrees with the time course of in vitro starch digestion 

of meal (Fig. 2C), as well as with the relative ease of digestion of 

isolated waxy vs. non-waxy starch (Rooney and Miller, 1982; Rooney 

and Pflugfelder, 1986). 

Reversal of this protocoldremoval of starch prior to pepsin 

digestiondshowed the opposite effect; proteolysis was enhanced 

to a greater extent with KS51 than KS48 (2.2- vs. 1.6-fold) (Table 3). 

Table 2 

Starch digestion with and without prior removal of protein by pepsin 

Sorghum line With pepsin 

pre-treatment 

(pepsin/a-amylase a ) 

Without pepsin 

pre-treatment 

(a-amylase/pepsin a ) 

Ratio 

P/A:A/P 

mg reducing sugar/h mg reducing sugar/h 

KS48 84.8 4.2 47.6 2.1 1.8 

KS51 59.1 2.7 39.3 1.3 1.5 



Table 3 

Protein digestion by pepsin without and with prior removal of starch by a-amylase 

Sorghum 

line 

With a-amylase 

pre-treatment 

(a-amylase/pepsin a ) 

Without a-amylase 

pre-treatment 

(pepsin/a-amylase a ) 

This unexpected result suggests that, in addition to the intrinsic 

properties of the starch, spatial arrangement between the granules 

and protein bodies differs in the two lines. Further, starch in KS51 

appears not only to be harder to digest with a-amylase, but also 

more capable of impeding protein digestion by pepsin than in KS48. 

Once starch-imposed restrictions are removed, protein in KS51 

appears more exposed and thus more accessible to pepsin. A 

question arises as to whether the observed difference could simply 

be due to the physical properties of the two grains, i.e., variation in 

the hardness of the two grains, or different ratios of hard (closed) to 

soft (open) endosperm, as reported by Ezeogu et al., 2005. Hardness 

data for KS48 and KS51 samples using the single kernel characterization 

system (SKCS) showed that they are practically and 

statistically the same, with a mean for seed hardness of 69.12 and 

69.83, respectively. Therefore, our interpretation is that the difference 

observed above is mostly due to the properties of the starch 

and proteins. 

3.8. Effect of DTT reduction on digestion of kafirins and 

glutelins by pepsin 

Ratio 

A/P:P/A 

Protein remaining, volume, % Protein remaining, volume, % 

KS48 65.4 2.0 40.9 2.7 1.6 

KS51 75.3 1.5 34.0 3.8 2.2 


Reducing agents, such as 2-ME (Elkhalifa et al., 1999; Ezeogu 

et al., 2005, 2008; Hamaker et al., 1987) and DTT or thioredoxin (del 

Val et al., 1999), were shown to improve protein digestibility, thus 

highlighting the importance of components with disulfide bonds 

(Hamaker et al., 1987). In assessing DTT effects, we observed that 

42% and 50% of the kafirins of KS48 and KS51, respectively, were not 

digested by pepsin in control samples. Reduction with DTT 

Fig. 4. Effect of redox status on digestion of the combined glutelin and kafirin fractions 

of KS48 and KS51 by pepsin. D þ P ¼ DTT þ pepsin; P ¼ pepsin; D ¼ DTT; C ¼ control 

(minus DTT and pepsin). * ¼ Kafirin region.

80 

enhanced digestibility by 3.5-fold and 2.0-fold, with undigested 

protein decreasing from 42% to 12% for KS48 and from 50% to 25% 

for KS51 (Fig. 4, * designates kafirin region). To achieve a level of 

kafirin digestion with KS51 equivalent to that of KS48, either 

a longer incubation period or more reductant was required (data 

not shown). 

3.9. Microscopy studies 

Our results thus far show that a greater abundance of disulfide 

proteins and the non-waxy type of starch contribute to the low 

digestibility of KS51 vs. KS48. We have performed experiments 

with additional waxy and non-waxy lines of sorghum and corn 

(3 pairs of sorghum lines, 1 pair of corn lines). Digestion of waxy 

starch was consistently faster than that of non-waxy starch in all 

samples tested but protein digestibility was mixed (data not 

shown). Thus, this data also show the importance of the interactions 

between starch and protein to the digestibility of both 


components. However, they provide no information on the nature 

of the differences observed in these interactions between the two 

lines. 

To study this point, we examined iodine-stained sections of 

grains microscopically and found that the endosperm in KS51 was 

more highly organized than that of KS48di.e., there was a distinct 

demarcation between floury and corneous endosperm in KS51, not 

present in KS48 (Fig. 5A). Taken together, the analytical data presented 

above and the iodine staining results confirmed the earlier 

classification of the lines as waxy (KS48) and non-waxy (KS51) 

(Pedersen et al., 2004, 2005). Electron micrographs revealed 

organizational differences between the two. While floury endosperm 

was similar (Fig. 5C), corneous endosperm differed markedly 

in the two lines (Fig. 5B). Protein bodies were more numerous and 

more tightly associated with starch granules in KS51 than in KS48 

(Fig. 5B). This type of interaction between the two components 

could contribute to the low digestibility of protein and starch in 

KS51. 

Fig. 5. Light and electron microscopy of KS48 and KS51 seeds. A. Cross section of seeds stained with iodine showing floury (F) and corneous (C) endosperm. Magnification: 32 .B. 

Electron micrograph of corneous endosperm of KS48 and KS51, 5000 . C. Electron micrograph of floury endosperm of KS48 and KS51, 5000 .SG¼ starch granule; PB ¼ protein 

body; M ¼ protein matrix.

In summary, our results highlight the importance of disulfide 

proteins and GBSSI in conferring the chemical and structural 

properties that influence the digestibility of stored reserves, both 

protein and starch. These findings also add insight into the role of 

the protein matrix and its interaction with starch. Comparison of 

two sorghum lines with a common pedigree that differed in 

digestibility made it possible to demonstrate how protein–starch 

interactions in the seed influence the digestibility of each. Attempts 

to improve the digestibility of grain by classical breeding and 

genetic engineering should further our understanding of these 

factors and their relative contribution to digestibility. These efforts 

may also provide answers to related fundamental questions, such 

as what role the key redox protein, thioredoxin, plays in forming 

and maintaining the protein matrix (Buchanan and Balmer, 2005) 

and the importance of this unique packaging of stored reserves to 

the structure and physiology of seed. Answers to these questions 

will in all likelihood be relevant to human and animal nutrition and 

to the use of sorghum as a biofuel. 


This study was supported by the U.S. Agency for International 

Development and the Bill and Melinda Gates Foundation. PGL was 

supported by the USDA Cooperative Extension Service and BBB by 

the Agricultural Experiment Station through UC Berkeley. We thank 

Drs. K. McDonald and G. Min (Electron Microscope Laboratory) and 

D. Schichnes (Biological Imaging Laboratory) for assistance with the 

microscopy studies and Dr. S.R. Wessler for the gift of waxy protein 

antibody. The late Dr. K. Kobrehel helped launch this study during 

a sabbatical visit to Berkeley in the early 1990s. Numerous undergraduate 

students also made contributions, especially D. Bergmann, 

P.-H. Ren and P. Yu-Man Kei. 

References 

Aboubacar, A., Axtell, J.D., Huang, C.-P., Hamaker, B.R., 2001. A rapid protein 

digestibility assay for identifying highly digestible sorghum lines. Cereal 

Chemistry 78. 

Astwood, J.D., Leach, J.N., Fuchs, R.L., 1996. Stability of food allergens to digestion in 

vitro. Nature Biotechnology 14, 1269–1273. 

Axtell, J.D., Kirleis, A., Hassen, M.M., D’Croz-Mason, N., Mertz, E.T., Munck, L., 1981. 

Digestibility of sorghum proteins. Proceedings of the National Academy of 

Sciences of the United States of America 78, 1333–1335. 

Balmer, Y., Vensel, W.H., DuPont, F.M., Buchanan, B.B., Hurkman, W.J., 2006. Proteome 

of amyloplasts isolated from developing wheat endosperm presents 

evidence of broad metabolic capability. Journal of Experimental Botany 57, 

1591–1602. 

Bean, S.R., 2003. Evaluation of novel pre-cast gradient gels for the separation of 

sorghum proteins. Cereal Chemistry 80, 500–504. 

Belton, P.S., Delgadillo, I., Halford, N.G., Shewry, P.R., 2006. Kafirin structure and 

functionality. Journal of Cereal Science 44, 272–286. 

Buchanan, B.B., Balmer, Y., 2005. Redox regulation: a broadening horizon. Annual 

Reviews of Plant Biology 56, 187–220. 

Casady, A.J., 1972. 10 Sorghum parental lines. Crop Science 12, 722. 

Chandrashekar, A., Kirleis, A.W., 1988. Influence of protein on starch gelatinization 

in sorghum. Cereal Chemistry 65, 457–462. 

Chandrashekar, A., Mazhar, H., 1999. The biochemical basis and implications of 

grain strength in sorghum and maize. Journal of Cereal Science 30, 193–207. 

Denyer, K., Johnson, P., Zeeman, S., Smith, A.M., 2001. The control of amylose 

synthesis. Journal of Plant Physiology 158, 479–487. 

Duodu, K.G., Nunes, A., Delgadillo, I., Parker, M.L., Mills, E.N.C., Belton, P.S., 

Taylor, J.R.N., 2002. Effect of grain structure and cooking on sorghum and maize 

in vitro protein digestibility. Journal of Cereal Science 35, 161–174. 

Duodu, K.G., Tang, H., Grant, A., Wellner, N., Belton, P.S., Taylor, J.R.N., 2001. FTIR and 

solid state 13 C spectroscopy of proteins of wet cooked and popped sorghum and 

maize. Journal of Cereal Science 33, 261–269. 

Duodu, K.G., Taylor, J.R.N., Belton, P.S., Hamaker, B.R., 2003. Factors affecting 

sorghum protein digestibility. Journal of Cereal Science 38, 117–131. 

El Nour, I.N.A., Peruffo, A.D.B., Curioni, A., 1998. Characterisation of sorghum kafirins 

in relation to their cross-linking behaviour. Journal of Cereal Science 28, 

197–207. 

Elkhalifa, A.E.O., Chandrashekar, A., Mohamed, B.E., El Tinay, A.H., 1999. Effect of 

reducing agents on the in vitro protein and starch digestibilities of cooked 

sorghum. Food Chemistry 66, 323–326. 


Ezeogu, L.I., Duodu, K.G., Taylor, J.R.N., 2005. Effects of endosperm texture and 

cooking conditions on the in vitro starch digestibility of sorghum and maize 

flours. Journal of Cereal Science 42, 33–44. 

Ezeogu, L.I., Duodu, K.G., Emmanbux, M.N., Taylor, J.R.N., 2008. Influence of cooking 

conditions on the protein matrix of sorghum and maize endosperm flours. 


Hamaker, B.R., Bugusu, B.A., 2003. Overview: sorghum proteins and food quality. 

Paper presented at: Workshop on the proteins of sorghum and millets: 

enhancing nutritional and functional properties for Africa [CD] (Pretoria: South 

Africa). 

Hamaker, B.R., Kirleis, A.W., Butler, L.G., Axtell, J.D., Mertz, E.T., 1987. Improving the 

in vitro protein digestibility of sorghum with reducing agents. Proceedings of 

the National Academy of Sciences of the United States of America 84, 626–628. 

Hamaker, B.R., Kirleis, A.W., Mertz, E.T., Axtell, J.D., 1986. Effect of cooking on the 

protein profiles and in vitro digestibility of sorghum and maize. Journal of 

Agricultural and Food Chemistry 34, 647. 

Hamaker, B.R., Mohamed, A.A., Habben, J.E., Huang, C.-P., Larkins, B.A., 1995. An efficient 

procedure for extracting maize and sorghum kernel proteins reveals higher 

prolamin contents than the conventional method. Cereal Chemistry 72, 583–588. 

Izquierdo, L., Godwin, I.D., 2005. Molecular characterization of novel methioninerich 

delta-kafirin seed storage protein gene in sorghum (Sorghum bicolor L.). 

Cereal Chemistry 82, 706. 

Jambunathan, R., Mertz, E.T., Axtell, J.D., 1975. Fractionation of soluble proteins of 

high-lysine and normal sorghum grain. Cereal Chemistry 52, 119–121. 

Kasarda, D.D., Woodard, K.M., Elva, A.A., 1998. Resolution of high molecular weight 

glutenin subunits by a new SDS-PAGE system incorporating a neutral pH buffer. 


Klopfenstein, C.F., Hoseney, R.C., 1995. Nutritional properties of sorghum and the 

millets. In: Dendy, D.A.V. (Ed.), Sorghum and Millets: Chemistry and Technology. 

American Association of Cereal Chemists, St. Paul, MN, pp. 125–168. 

Kumari, S.R., Chandrashekar, A., 1994. Relationships between grain hardness and 

the contents of prolamins and three anti-fungal proteins in sorghum. Journal of 


Landry, J., Moureaux, T., 1970. Heterogeneite des gluenines du grain de mais: 

Extraction slelective et composition en acides amines des trios fractions isolees. 

Bulletin de la Societe de Chimie Bioologique 52, 1021–1031. 

MacLean, W.C., de Romana, G.L., Placko, R.P., Graham, G.G., 1981. Protein quality and 

digestibility of sorghum in preschool children: balance studies and plasma free 

amino acids. Journal of Nutrition 111, 1928–1936. 

Mazhar, H., Chandrashekar, A., 1995. Quantitation and distribution of kafirins in the 

kernels of sorghum cultivars varying in endosperm vitreousness. Journal of 


Mazhar, H., Chandrashekar, A., Shetty, H.S., 1993. Isolation and immunochemical 

characterization of alcohol-extractable proteins (kafirins) of Sorghum bicolor (L.) 

Moench. Journal of Cereal Science 17, 83–93. 

Mertz, E.T., Hassen, M.M., Cairn-Whittern, C., Kirleis, A.W., Tu, L., Axtell, J.D., 1984. 

Pepsin digestibility of proteins in sorghum and other major cereals. Proceedings 

of the National Academy of Sciences of the United States of America 81, 1–2. 

Nakamura, T., Vrinten, P., Hayakawa, K., Ikeda, J., 1998. Characterization of 

a granule-bound starch synthase isoform found in pericarp of wheat. Plant 


Nunes, A., Correia, I., Barros, A., Delgadillo, I., 2004. Sequential in vitro pepsin 

digestion of uncooked and cooked sorghum and maize samples. Journal of 


Opstvedt, J., Miller, R., Hardy, R.W., Spinelli, J., 1984. Heat-induced changes in 

sulfhydryl groups and disulfide bonds in fish protein and their effect on protein 

and amino acid digestibility in rainbow trout (Salmo gairdneri). Journal of 


Oria, M.P., Hamaker, B.R., Axtell, J.D., Huang, C.-P., 2000. A highly digestible sorghum 

mutant cultivar exhibits a unique folded structure of endosperm protein bodies. 

Proceedings of the National Academy of Sciences of the United States of 

America 97, 5065–5070. 

Oria, M.P., Hamaker, B.R., Shull, J.M., 1995a. In vitro protein digestibility of developing 

and mature sorghum grain in relation to a-, b- and g-kafirin disulfide 

crosslinking. Journal of Cereal Science 22, 85–93. 

Oria, M.P., Hamaker, B.R., Shull, J.M., 1995b. Resistance of sorghum a-, b- and gkafirins 

to pepsin digestion. Journal of Agricultural and Food Chemistry 43, 

2148–2152. 

Pedersen, J.F., Bean, S.R., Funnell, D.L., Graybosch, R.A., 2004. Rapid iodine staining 

techniques for identifying the waxy phenotype in sorghum grain and waxy 

genotype in sorghum pollen. Crop Science 44, 764–767. 

Pedersen, J.F., Bean, S.R., Graybosch, R.A., Park, S.H., Tilley, M., 2005. Characterization 

of waxy grain sorghum lines in relation to granule-bound starch synthase. 

Euphytica 144, 151–156. 

Pedersen, J.F., Milton, T., Mass, R.A., 2000. A twelve-hour in vitro procedure for 

sorghum grain feed quality assessment. Crop Science 40, 204–208. 

Rooney, L.W., Miller, F.R., 1982. Variation in the structure and kernel characteristics 

of sorghum. Paper presented at: International Symposium on Sorghum Grain 

Quality (ICRISAT Center, Patancheru, India). 

Rooney, L.W., Pflugfelder, R.L., 1986. Factors affecting starch digestibility with special 

emphasis on sorghum and corn. Journal of Animal Science 63, 1607–1623. 

Shull, J.M., Chandrashekar, A., Kirleis, A.W., Ejeta, G., 1990. Development of sorghum 

[Sorghum bicolor (L.) Moench] endosperm in varieties of varying hardness. Food 

Structure 9, 253–267. 

Shull, J.M., Watterson, J.J., Kirleis, A.W., 1991. Proposed nomenclature for the 

alcohol-soluble proteins (kafirins) of Sorghum bicolor (L. Moench) based on

82 

molecular weight, solubility, and structure. Journal of Agricultural and Food 


Shull, J.M., Watterson, J.J., Kirleis, A.W., 1992. Purification and immunocytochemical 

localization of kafirins in Sorghum bicolor (L. Moench) endosperm. Protoplasma 

171, 64–74. 

Stock, R.A., Brink, D.R., Kreikemeier, K.K., Smith, K.K., 1987. Evaluation of earlyharvested 

and reconstituted grain sorghum in finishing steers. Journal of 

Animal Science 65, 548–556. 

Taylor, J.R.N., Novellie, L., Liebenberg, N.V.D.W., 1984. Sorghum protein body 

composition and ultrastructure. Cereal Chemistry 61, 69. 

Taylor, J.R.N., Schussler, L., Liebenberg, N.V.D.W., 1985. Protein body formation in the 

starchy endosperm of developing Sorghum bicolor (L.) Moench seeds. S-Afr 

Tydskr Plantk 51, 81–86. 

Terada, R., Nakajima, M., Isshiki, M., Okaagaki, R.J., Wessler, S.R., Shimamoto, K., 

2000. Antisense waxy genes with highly active promoters effectively suppress 

waxy gene expression in transgenic rice. Plant and Cell Physiology 41, 881–888. 


U.S. Grains Council, 2006. Sorghum, grain sorghum: any of various plants of the 

genus sorghum family Poaceae, a cereal grain. 

del Val, G., Yee, B.C., Lozano, R.M., Buchanan, B.B., Ermel, R.W., Lee, Y.M., Frick, O.L., 

1999. Thioredoxin treatment increases digestibility and lowers allergenicity of 

milk. Journal of Allergy and Clinical Immunology 1103, 690–697. 

Wall, J.S., 1971. Disulfide bonds: determination, location and influence on molecular 

properties of proteins. Journal of Agricultural and Food Chemistry 19, 619. 

Wester, T.J., Gramlich, S.M., Britton, R.A., Stock, R.A., 1992. Effect of grain sorghum 

hybrid on in vitro rate of starch disappearance and finishing performance of 

ruminants. Journal of Animal Science 70, 2866–2876. 

Wong, J.H., Cai, N., Tanaka, C.K., Vensel, W.H., Hurkman, W.J., Buchanan, B.B., 2004. 

Thioredoxin reduction alters the solubility of proteins of wheat starchy endosperm: 

an early event in cereal germination. Plant and Cell Physiology 45, 

407–415. 

Zhang, G., Hamaker, B.R., 1998. Low a-amylase starch digestibility of cooked 

sorghum flours and the effect of protein. Cereal Chemistry 75, 710–713.

Fundamental study on protein changes taking place during malting of oats 

Christina Klose, Beatus D. Schehl, Elke K. Arendt * 

Department of Food and Nutritional Sciences, National University of Ireland, University College Cork, College Road, Cork, Ireland 

article info 





Keywords: 

Oats 

Malting 

Proteins 

Lab-on-a-Chip analysis 


abstract 

Oats (Avena sativa) is one of the most popular cereals for human 

consumption and they have received increased interest because of 

their excellent health-related properties, such as high contents of 

dietary fibre, especially b-glucan, as well as minerals and antioxidants. 

The most abundant antioxidants are vitamin E (tocols), 

phytic acid, phenolic compounds and avenanthramides; these are 

concentrated in the outer layers of the kernel (Peterson, 2001). The 

major component of the endosperm cell walls of the oat grain or 

caryopsis is a (1 / 3), (1 / 4)-b-D-glucan, known as oat b-glucan 

(Ren et al., 2003). The early interest in oat b-glucan arose from 

knowledge of its highly viscous properties and therefore potential 

commercial value as a thickening agent in food formulations and 

other industrial applications, which is interesting for healthy 

gluten-free products. Whether oats is gluten-free is not yet clear. 

However, oats can be tolerated by most but not all people with 

coeliac disease and in Finland 69% of patients were reported to use 

oat products (Peraaho et al., 2004). For this purpose, a gluten-free 

beer out of malted oats could also be attractive for coeliacs. 

Malted oats has been widely used as an ingredient for beer 

production in medieval times and before. Nowadays oat malt is 

used in the brewing industry mainly as a flavour adjunct for the 


E-mail address: e.arendt@ucc.ie (E.K. Arendt). 


doi:10.1016/j.jcs.2008.07.014 





During the malting process, storage proteins are degraded by proteolytic enzymes into small peptides 

and amino acids. The activity of these enzymes was measured during malting of oats and was found to be 

increased. To quantify proteolytic degradation, proteins of unmalted, germinating and malted grains 

were fractionated. After extracting the oat proteins (Osborne fractionation), protein fractions were 

analysed using a Lab-on-a-Chip technique, which separates the proteins – based on their molecular 

weight – by capillary electrophoresis. This new technique for the analysis of proteins was supported by 

using two-dimensional gel electrophoresis. In addition, amino acid analysis was carried out. In general 

a degradation of proteins to small peptides and amino acids could be observed in the globulin, prolamin 

and glutelin fractions. In the albumin fraction a protein increase was observed, which is due to the fact 

that this fraction contains the majority of the metabolically active proteins. Amino acid analysis supported 

the observation of increased protein amount in the albumin fraction and decreased protein 

amounts in the other fractions. Some proteins, which have not been described in the literature, were 

detected in the albumin and glutelin fraction, since Lab-on-a-Chip technique allows detection of proteins 

with low molecular weights of 4.5 kDa. 


production of special lagers, ales and stouts. Malting is the initial 

step in beer production and strongly defines type and quality of the 

beer. The main purpose of malting is to produce enzymes and to 

breakdown cell walls surrounding starch granules. One of the most 

important physical–chemical changes that occur during malting is 

the degradation of the proteinaceous matrix that surrounds the 

starch granules within the cells of the endosperm and their 

conversion into soluble peptides and amino acids to provide 

substrates for the synthesis of proteins in the growing embryo 

(Briggs et al., 1981). Malt proteins have a high impact on the 

brewing process and the subsequent quality of beer. Protein 

content and size distribution are of particular interest in terms of 

filtering, fermentability, foam and haze stability. Compared to the 

commonly used grain for malting, barley and oats has not only 

a unique protein composition, but also a high protein content of 

11–15 %, of which about a third passes into the final beer. A number 

of authors (Lapvetelainen et al., 1995; Ma and Harwalkar, 1984; 

Robbins et al., 1971; Robert et al., 1985, 1983a; Shewry, 1995; Wu, 

1983; Zarkadas, 1982) have reported on the protein composition of 

oats. In general cereal proteins have been classified based on their 

solubility (Osborne fractionation), albumins (water soluble), globulins 

(salt water soluble), prolamins (soluble in dilute alcohol 

solutions) and glutelins (soluble in acids or bases). The difference 

between oats and other cereal grains in the structure of the 

proteinaceous components is consistent with the differences in the 

distribution of the protein fractions. Oats lacks the protein matrix, 

which is characteristic of wheat and some other cereals. In wheat

84 

and some other cereals, the storage proteins are insoluble in salt 

solutions. In terms of oats, a large portion of the salt water-soluble 

globulins also belong to the storage proteins of the endosperm. 

Although the literary data varies widely, oats contains a relatively 

low quantity of prolamins and a high amount of globulins, up to 

80% of the total oat protein (Lásztity, 1996). The prolamins (avenins) 

of oats account for about 15% of the total protein and have, similar 

to other cereal prolamins, mainly a storage function (Robert et al., 

1983b). Glutelins have typically been difficult to completely solubilise 

and therefore values from 5 to 66% have been reported, 

depending on the extraction solvent and concentration (Robert 

et al., 1985). Most of the metabolically active proteins of oats are in 

the water-soluble albumin fraction. Nevertheless, it cannot be 

excluded that the globulin fraction and probably the glutelin fraction 

contain this type of protein. Oat albumins account for 1–12% of 

the total protein (Lásztity, 1996). Oat protein distribution affects 

also the amino acid composition of oats. The generally higher lysine 

content of albumins and globulins causes the relatively high lysine 

content in oats, compared to other cereals. In addition, glutamic 

acid and proline contents are relatively lower (Lásztity, 1996). 

The aim of this study was to evaluate the changes in protein and 

amino acid composition from the raw oats through germination to 

the final malt. For this purpose, a novel analysis technique, the Labon-a-Chip 

capillary electrophoresis, was used and was supported 

by using two-dimensional gel electrophoresis. This is the first 

report on a detailed analysis of the protein changes during malting 

of oats. 



The oat grain was harvested in 2005 and obtained from Raisio, 

Finland. Malting was carried out using a micro malting machine 

(Joe White Malting Systems, Perth, Australia). For this purpose, 8 kg 

of oat seeds were steeped three times, each stage containing a wet 

stage at 13 C (6 h, 4 h, 3 h) and an air rest stage at 18 C (10 h, 7 h, 

1 h). After a total of 31 h of steeping, seeds were germinated for 5 

days and 6 h at three stages (17 C, 15 C and 13 C) and kilned in six 

stages for 23 h using a schedule that started at 55 C and increased 

to 85 C. 

2.2. Total protein analysis of oats 

Samples were taken daily throughout the malting process. 

Preliminary analysis of all samples with the Lab-on-a-Chip system 

revealed degradation of all proteins during the process, whereas 

largest changes could be observed during the germination process 

and only minor changes after germination (before kilning). Thus, 

unmalted, 3-day-germinating and kilned (malted) samples were 

taken and freeze-dried (VirTus Benchtop 4k, Biopharma Process 

Systems, Winchester, UK). Roots were removed and samples were 

milled with a Bühler Miag laboratory-scale disc mill (Bühler GmbH, 

Braunschweig, Germany) set at a fine setting of 0.05 mm and stored 

at 80 C. 

The proteolytic enzyme activity during the malting process was 

measured according to the method of Brijs et al. (2002), where 

haemoglobin was used as substrate. After incubation, the reaction 

was stopped by the addition of 10% (w/v) trichloroacetic acid. The 

free a-amino nitrogen levels of the supernatants were assayed with 

trinitrobenzene-sulfonic acid reagent (0.3%, v/v, in 0.2 M sodium 

phosphate buffer, pH 8.0) using L-leucine as standard. For this 

purpose, the supernatant and trinitrobenzene-sulfonic acid reagent 

were incubated and reaction was stopped with 0.2 M HCl. The 

absorbance was measured at 340 nm. One unit of activity corresponds 

to the enzyme activity that liberated 1 mg of L-leucine/h 

C. Klose et al. / Journal of Cereal Science 49 (2009) 83–91 

under the assay conditions. Total nitrogen contents of the unmalted, 

germinating and malted samples were determined by 

combustion (according to Analytica-EBC, method 4.3.2) using 

a Leco FP-528 nitrogen determinator (St Joseph, MI). 

2.3. Modified Osborne fractionation of oat proteins 

The extraction procedure was carried out once, where first, 

unmalted, germinating and malted oats (10 g of each sample) were 

defatted with 100 mL hexane using a Soxhlet extractor. The defatted 

samples were then extracted twice with 100 mL of distilled 

water (albumin fraction). After water extraction the residue was 

extracted twice with 100 mL of 5.0% NaCl (globulin fraction). The 

remaining flour was then extracted three times with 150 ml of 55% 

1-propanol þ 1% DTT (prolamin fraction) and the glutelin fraction 

together with the residue was extracted three times with a solvent 

containing 6 M urea, 2% SDS and 1% DTT. Each extraction was 

carried out at room temperature for 10 min and centrifuged afterwards 

at 10,000 rpm for 10 min. All supernatants were collected, 

further dialyzed against distilled water for 24 h, freeze-dried and 

stored at 80 C. 

2.4. Lab-on-a-Chip analysis of total protein composition 

To evaluate protein changes during the malting process, a novel 

analysis technique, the Lab-on-a-Chip capillary electrophoresis, 

was used. The principles of these electrophoretic assays are based 

on traditional gel electrophoresis principles that have been transferred 

to a chip format. The chip accommodates sample wells, gel 

wells and a well for an external standard (ladder). Micro-channels 

are fabricated in glass to create interconnected networks among 

these wells. During chip preparation, the micro-channels are filled 

with a sieving polymer and fluorescence dye. Once the wells and 

channels are filled, the chip becomes an integrated electrical circuit. 

Charged biomolecules like proteins are electrophoretically driven 

by a voltage gradient, similar to slab gel electrophoresis. Because of 

a constant mass-to-charge ratio and the presence of a sieving 

polymer matrix, the molecules are separated by size. Smaller 

fragments are migrating faster than larger ones. Dye molecules 

intercalate into protein micelles. These complexes are detected by 

laser-induced fluorescence. Data is translated into gel-like images 

(bands) and electropherograms (peaks). With the help of a ladder 

that contains components of known sizes, a standard curve of 

migration time versus fragments size is plotted. From the migration 

times measured for each fragment in the sample, the size is 

calculated. Two markers are run with each of the samples bracketing 

the overall sizing range. The ‘‘lower’’ and ‘‘upper’’ markers are 

internal standards used to align the ladder data with data from the 

sample wells. This is necessary to compensate for drift effects that 

may occur during the course of a chip run (Agilent Technologies, 

2005). 

For analysis of the total protein content, 40 mg of milled oat 

samples were extracted at room temperature for 5 min with 400 mL 

2 M urea solution containing glycerol, Tris–HCl and 0.1 M DTT and 

centrifuged for 15 min. Four microliters of each supernatant were 

denatured using 2 mL of Agilent denaturing solution and heated for 

5 min at 100 C. After dilution with deionised water, 6 mL were 

applied to the Protein 230 þ LabChip and the Protein 80 þ LabChip 

for analysis in the Agilent 2100 Bioanalyzer (Agilent Technologies, 

Palo Alto, CA). 

2.5. Lab-on-a-Chip analysis of extracted protein fractions 

Ten micrograms of each extracted protein fraction were dissolved 

in 1 mL of their extraction solvents and applied to the Agilent 

2100 Bioanalyzer, as described above. For the 230 þ LabChip,

each run included a ladder comprising reference proteins of 7, 15, 

28, 46, 63, 95, 150 kDa plus an upper marker of 240 kDa and a lower 

marker of 4.5 kDa. Each sample contained an internal standard 

comprising the upper and lower marker as well. Any peak detected 

below 14 kDa is termed a system peak and is not included in 

analysis. The detection performance was also carried out in 

a molecular weight range between 4.5 and 95 kDa using the Protein 

80 þ LabChip. For this analysis the ladder consisted of reference 

proteins of 3.5, 6.5, 15, 28, 46, 63 kDa plus the upper and the lower 

markers of 95 and 1.6 kDa. According to the Agilent manual any 

peak detected below 5 kDa is named a system peak and is not 

included in analysis. Analyses were carried out in triplicate on each 

LabChip and a relative standard deviation of 8% was considered. 

2.6. Two-dimensional gel electrophoresis of extracted protein 

fractions 

Initially oat protein fractions of unmalted, germinating and 

malted samples were defrosted. All samples were then dissolved in 

a solubilisation buffer containing 9 M urea, 4% CHAPS, 0.05% Triton 

X100 and 65 mM DTT. Protein solution (125 mL) was applied to each 

strip. Isoelectrofocusing (IEF) was carried out using 7-cm IPG 3–10 

strips (ReadyStripÔ, BioRad) and a BioRad PROTEAN IEF cell with 

controlled cell temperature of 20 C. The running conditions were 

as follows: passive rehydration: 8 h, active rehydration (50 V): 8 h, 

rapid (300 V): 30 min, linear (4000 V): 20,000 V-h, rapid (300 V): 

99 h. After IEF was completed, the IPG strips were equilibrated for 

15 min in a buffer containing 50 mM Tris–HCl pH 8.8, 6 M urea, 30% 

glycerol, 2% SDS and130 mM DTT. After that, the strips were 

equilibrated for another 15 min in the same solution except that 

DTT was replaced with 130 mM iodoacetamide and traces of bromophenol 

blue. SDS-PAGE was performed in a BioRad Criterion 

Dodeca cell with gels of total acrylamide concentration of 12.5% at 

20 C, sealed with Tris/glycine/SDS running buffer. The gels were 

run at 100 V until the tracking dye reached the bottom of the gel. To 

visualize the proteins, the gels were first incubated in a fixation 

solution, containing ethanol and phosphoric acid and then in an 

incubation solution containing methanol, phosphoric acid and 

ammonium sulphate dissolved in distilled water. The staining was 

carried out with the incubation solution containing Coomassie 

brilliant blue for 4 days. Analysis was carried out in triplicate. 

2.7. Amino acid analysis 

The free amino acid profile of the milled samples was measured 

in triplicate using a Jeol JLC-500/V amino acid analyser (Jeol (UK) 

Ltd., Garden City, Herts, UK) fitted with a Jeol Na þ high performance 

cation exchange column. Prior to analysis, samples were extracted 

with a 12% (w/v) trichloracetic acid solution. This procedure was 

carried out three times. For the total amino acid profile, milled 

samples as well as each protein fraction were analysed three times 

according to the method of Moore and Stein (1963). Due to this 

method, tryptophan, which is not stable during hydrolysis, could 

not be identified, and glutamine and asparagine were converted to 

their acids during hydrolysis, so that they were detected with their 

acid. 

3. Results 

3.1. Proteolytic activity 

Total proteolytic enzyme activity was analysed using haemoglobin 

as substrate and absorption was measured against L-leucine 

as standard. During the malting process an 8.6-fold increase in 

proteolytic activity could be observed. The results revealed an 

increase in the proteolytic activity level also during steeping (from 

C. Klose et al. / Journal of Cereal Science 49 (2009) 83–91 85 

2.4 to 4.5 mg of L-leucine/h/g), during 3 days of germination (from 

4.5 to 16.3 mg of leucine/h/g) and during the last steps of malting 

(from 16.3 to 20.3 mg of leucine/h/g). The highest increase could be 

detected during the first 3 days of germination. 


The results of total protein analysis using the Protein 230 þ 

LabChip showed several protein peaks in the area of 23–30 kDa 

(area 1), as well as in the area of 38–52 kDa (area 2) in all three 

samples (results not shown). Peak heights in area 1 demonstrate 

that these proteins were degraded by 47%, from about 500 FU 

(Fluorescence Units) to about 200 FU during malting. In area 2, peak 

heights decreased by 74% after malting. Identical protein patterns 

were obtained with the Protein 80 þ LabChip (results not shown). 

Both peak areas decreased by 44% (1) or by 71% (2) during the 

malting process. However, the Protein 80 þ LabChip reveals more 

protein peaks of low molecular sizes between 6 and 19 kDa, where 

not only a decrease, but also an increase of a protein peak with the 

molecular size of 10 kDa can be seen. 

Total nitrogen analysis revealed nitrogen levels of 1.734% for the 

unmalted, 1.725% for the germinating and 1.750% for the malted 

oats. According to Wu (1983), the total nitrogen content of oat 

kernels decreased at first during germination, which is probably 

due to the loss of water-soluble nitrogen. After kilning the 

percentage of total nitrogen was increased. 

3.3. Protein fractions 

Total protein analysis alone did not give a detailed picture of the 

changes occurring during the malting process. Proteins were fractioned 

according to Osborne to further investigate how oat protein 

alters during malting. 

3.3.1. Albumin fraction 

The electropherogram of the water-soluble fraction of unmalted 

and malted oat samples using the Protein 80 þ LabChip is shown in 

Fig. 1A. Results indicate important changes among the overall 

malting process. After malting, new or increased protein peaks can 

be seen. The peak area reaching from 6 to 9 kDa and the peak at 

61 kDa were not present before, but after malting. Increased 

protein peaks could be found at 13, 20 and 50 kDa. Proteins with 

molecular weights of 35–44, 15 and 76 kDa were decreased during 

the process. 

Results from the Lab-on-a-Chip capillary electrophoresis agreed 

with the effect, found when two-dimensional gel electrophoresis 

(Fig. 1B) was applied. An increase in intensity of protein spots as 

well as new spots could be detected in the malted sample. Nevertheless, 

results of Lab-on-a-Chip analysis appear clearer and more 

precisely. 

3.3.2. Globulin fraction 

Fig. 2A shows the patterns of the unmalted, germinating and 

malted oat globulin samples on the Protein 230 þ LabChip. Results 

reveal a reduction of the intensity of the group of bands located 

between 42 and 51 kDa (2) over the malting process. In the area of 

the molecular weights between 20 and 28 kDa (1), the band with 

the size of 25 kDa disappears. These patterns are clearly depicted in 

Fig. 2B, where the molecular weight distribution of the unmalted, 

germinating and malted globulin fraction can be seen in the electropherogram. 

A decrease in the protein peaks reaching from 42 to 

51 kDa (2) was also detected. In this molecular weight range, 

protein peaks were degraded by 67% during the first 3 days of 

germination and by 76% during the overall malting process, 

calculated based on the peak area. The electropherogram shows 

that the 25 kDa protein peak decreases in peak height by 30%, but

86 

a decrease of the concentration of all the 20–28 kDa proteins (1), 

calculated based on the peak area, could not be observed. 

Results of two-dimensional gel electrophoresis are shown in 

Fig. 2C. A large number of overlaying spots in the area of 40–50 kDa 

in the unmalted sample (picture a) make the identification difficult. 

However, a decrease of these proteins after the malting process 

(picture c) can be seen. This fact concurs with the results from the 

Lab-on-a-Chip analysis, where a decrease in peak area 2 (42– 

51 kDa) could be detected. The lower molecular weight protein 

spots ranging from 20 to 30 kDa seem to be changed only slightly 

during the malting process, whereas the protein spots with pH of 

5–6 decrease and the spot with pH 7 seems not to be most affected. 

This is in accordance with results of Lab-on-a-Chip analysis. 

3.3.3. Prolamin (avenin) fraction 

The results from the Protein 230 þ LabChip of prolamin degradation 

during the malting process are illustrated in Fig. 3A. The 

electropherogram displays a low molecular weight peak area 

extending from 15 to 23 kDa (1), a protein peak with the size of 

31 kDa (2) and an area between the molecular weights from 44 to 

51 kDa (3). All prolamins appear to be decreased during the 

process. In area 1 a general decrease by about 90% could be 

observed. The peak height of the 31 kDa peak (2) was reduced 

during 3 days of germination from 188 to 107 FU and to 52 FU after 

kilning, what resembles a decline of 72%. The 44–51 kDa part (3) 

was reduced from a peak area of 68–37 FU during germination and 

to 13 FU after the entire malting process, which equals a decrease 

by 46 and 81%, respectively. 

Fig. 3B shows the gels of unmalted (a), germinating (b) and 

malted (c) prolamin samples. Protein spot areas of 15, 30–35 and 

40–45 kDa could be observed. Results correspond approximately 

with the peak areas of the electropherogram. During germination 

the spot intensity decreases and, after kilning, almost all spots 

disappeared. 

3.3.4. Glutelin fraction 

Glutelin proteins decreased during the malting process (Fig. 4A). 

A protein peak with molecular weight of 9 kDa could be detected, 


Unmalted albumins 

Malted albumins 

Fig. 1. (A) Electropherogram of the albumin fraction of unmalted and malted samples (Protein 80 þ LabChip). (B) 2D gels of unmalted (a) and malted (b) albumin fraction. 

A 

90 

kDa 

10 

kDa 

90 

kDa 

10 

kDa 

pH 3 

which seems to increase slightly by 10% during malting. Peak areas 

ranging from 22 to 27 kDa and from 46 to 51 kDa were detected as 

well. A moderate decrease by 40% during the process could be 

measured at the protein peaks ranging from 20 to 27 kDa. The 

peaks between 46 and 51 kDa appear to be entirely degraded 

during the malting procedure. 

The acrylamide gels of the two-dimensional gel electrophoresis 

showed an overall decrease of intensity of spots after the malting 

process (Fig. 4B). 

3.4. Amino acid analysis 

pH 10 

3.4.1. Free and total amino acid composition of unmalted and 

malted oats 

In Table 1 the free amino acid composition of unmalted and 

malted oats is specified. Glutamic acid (254.1 mg/g oats) was found 

to be the major free amino acid in unmalted oats. In addition, 

aspartic acid (103.7 mg/g oats) and arginine (94.1 mg/g oats) were 

also present in high amounts in the unmalted oat kernel. Furthermore, 

no free proline could be detected. Results clearly show an 

increase in all free amino acids in oats after the malting process. 

Arginine (735.5 mg/g oats) was present in the highest concentration 

of malted oats, followed by proline (664.4 mg/g oats), threonine 

(654.3 mg/g oats), glutamic acid (481.6 mg/g oats) and phenylalanine 

(477.5 mg/g oats). Especially the free proline content increased from 

0 to 664.4 mg/g oats during malting. And also tyrosine, threonine 

and phenylalanine increased significantly by 98, 97, and 95%. 

Table 2 shows the total amino acid composition of unmalted and 

malted oats. Glutamic acid is by far the amino acid with the highest 

amount (20.61 mg/g) in unmalted oats. Asparitc acid, followed by 

arginine and leucine are also present in relatively high amounts of 

8.27, 7.81 and 7.72 mg/g, respectively. Also the essential amino 

acids valine, phenylalanine and lysine could be found in high 

amounts of 5.86, 5.53 and 4.38 mg/g. Considering a maximum 5% 

error in the method used, the following amino acid levels significantly 

decrease during malting: tyrosine (by 57%), arginine (by 

16%), proline (by 14%), phenylalanine (by 14%) and glycine (by 13%). 

Only the amino acid histidin appears to increase during the process 

B 

a 

b

1 

by 20%. However, in total, a value of 102.23 in the unmalted and 

a slightly lower level of 94.39 mg/g in the malted sample could be 

detected. 

3.4.2. Total amino acid composition of unmalted, germinating and 

malted oat protein fractions 

The total amino acid composition of oat protein fractions shows 

significant differences (Table 3). On the one hand, the amino acid 

composition changes during the malting process and on the other 

hand, amino acids occur in different amounts in the four fractions. 

In the albumin fraction, a total of 289.65 g amino acids/kg albumin 

could be found in the unmalted sample, whereas glutamic acid has 

the highest concentration (39.00 g/kg albumin), followed by 

glycine (24.96 g/kg albumin) and aspartic acid (24.09 g/kg 

albumin). All these amino acids were increased after malting by 12, 

27 and 35%, respectively. These increases led to a total amino acid 

content in the malted sample of 357.60 g amino acids/kg albumin. 

For the unmalted globulin sample, a value of 587.18, for the 

germinating sample of 554.38 and, for the malted sample, 503.11 g 

amino acids/kg globulin was found. In the globulin fraction as well, 

glutamic acid has the highest, but in a 3.3-fold increased concentration 

(127.37 g/kg globulin). Aspartic acid (55.52 g/kg globulin), 

2 

A 


2 

1 

Unmalted globulins 

Germinating globulins 

Malted globulins 

90 

kDa 

10 

kDa 

90 

kDa 

10 

kDa 

90 

kDa 

10 

kDa 

pH 

3 

40-50 kDa 

20-30 kDa 

Fig. 2. (A) SDS-PAGE image of the globulin fraction of unmalted, germinating and malted samples. (B) Electropherogram of the globulin fraction of unmalted, germinating and 

malted samples (Protein 230 þ LabChip). (C) 2D gels of unmalted (a), germinating (b) and malted (c) globulin fraction. 

B 

pH 

10 

arginine (54.27 g/kg globulin), leucine (44.03 g/kg globulin) and 

phenylalanine (34.30 g/kg globulin) appeared also in high 

concentrations. Except for lysine, which increased by 14%, all other 

amino acids decreased over the process. Glutamic acid was 

decreased by 28%, as well as tyrosine (by 22%), valine (by 20%) and 

serine (by 19%). For the prolamin fraction, a total of 354.07 g/kg 

prolamin was determined in the unmalted sample. The total 

concentration was decreased during germination to 311.69 g/kg 

prolamin and further decreased after kilning to 224.83 g/kg 

prolamin. Again, glutamic acid was found in the highest concentration 

(125.23 g/kg prolamin), but was decreased markedly during 

malting by 43%. Values of the other amino acids were a lot lower 

than glutamic acid concentration, but were degraded as well; 

leucine by 37%, valine by 38%, phenylalanine by 34% and tyrosine by 

37%. The amino acid concentration of the glutelin fraction was 

(192.38 g/kg glutelin) in the unmalted sample and it was decreased 

during the process to 169.42 g/kg glutelin and further to 134.67 g/ 

kg glutelin. As already found in the other fractions, glutamic acid 

(37.76 g/kg glutelin) and aspartic acid (18.07 g/kg glutelin) were 

apparent in the highest concentrations, followed by arginine 

(16.46 g/kg glutelin), leucine (13.31 g/kg glutelin), valine (12.14 g/ 

kg glutelin) and phenylalanine (11.4 g/kg glutelin). All amino acids 

C 

a 

b 

c

88 

were degraded markedly, especially proline (by 56%), tyrosine (by 

41%) and valine (by 41%). 

4. Discussion 


1 

2 

The total nitrogen content of oat samples during malting was 

analysed using the combustion method. In this study a total 

nitrogen content of 1.734, 1.725 and 1.750% was found for the 

unmalted, germinating and malted sample. The protein content 

was calculated by multiplying the nitrogen content by the 

conversion factor 5.4 according to Mariotti et al. (2008). The 

calculated protein content is 9.36, 9.32 and 9.45% for the three 

samples. According to the literature (Lásztity, 1996) the nitrogen 

content of oats ranges from 11 to 15 %, probably determined by 

using conversion factors of 6.25 or 5.83. According to Mossé (1990) 

and Mariotti et al. (2008) this factor leads to overestimated protein 

content. Thus, the corrected conversion factor for oats of 5.4 was 

used for calculating the protein content of the oats used, which 

leads to the relatively low protein content of the analysed oat grain. 

A comparison of the free amino acid content in unmalted and 

malted oats (Table 1) revealed an increase in all amino acids during 

malting. This is due to hydrolysis of the native protein, which gives 

both high molecular weight and low molecular weight protein 

breakdown products (peptides and free amino acids) (Kunze, 1996). 

Quantitatively, glutamic acid and aspartic acid were most present 

in the unmalted sample. In the malted sample, instead arginine, 

3 


Unmalted prolamins 

Germinating prolamins 

Malted prolamins 

A 

90 

kDa 

10 

kDa 

90 

kDa 

10 

kDa 

90 

kDa 

10 

kDa 

proline and threonine had the highest concentrations, which was 

also reported by Chittenden et al. (1987), where free proline was 

found to increase during germination of wheat. The largest increase 

could be observed in the concentration of free proline, which was 

not present in the unmalted sample and thus was increased by 

100%. In particular, the essential amino acids isoleucine, leucine, 

lysine, threonine, valine and phenylalanine were highly increased 

by about 90% each. 

Total amino acid composition of unmalted oats (Table 2) 

corresponds to that reported by Lásztity (1996). There are no great 

differences during malting. Total amino acid concentration of 

unmalted oats is 102.23 and of malted grain is 94.39 mg/g. 

According to Briggs et al. (1981), there is no net loss or gain of 

nitrogen in barley grain during malting, apart from substances 

leached during steeping. However, tyrosine was decreased by 57% 

and histidine is the only amino acid, which was increased by 20% 

during malting. Before and after malting, glutamic acid and aspartic 

acid had the highest concentrations, which can be attributed to the 

fact that during hydrolysis glutamine and asparagine were converted 

into their acids. 

4.2. Protein fractions 

pH 3 

pH 10 

4.2.1. Albumin fraction 

The albumin fraction of other cereals, such as barley, contains 

metabolically active proteins (Lásztity, 1996); thus, most of the 

enzymes of oats are probably found in the water-soluble fraction as 

well. Since the main purpose of malting is the production of 

B 

a 

ca. 40-45 kDa 

ca. 30-35 kDa 

ca. 15 kDa 

Fig. 3. (A) Electropherogram of the prolamin fraction of unmalted, germinating and malted samples (Protein 230 þ LabChip). (B) 2D gels of unmalted (a), germinating (b) and malted 

(c) prolamin fraction. 

b 

c

enzymes, an increase in the albumin fraction was expected. As 

shown in Fig. 1 and Table 3, an increase in several proteins could be 

observed. Proteins with molecular weights of 14–17, 20–27 and 36– 

47 kDa have been described in the literature (Lásztity, 1996) and 

could be detected with the Protein 80 þ LabChip, whereas the 36– 

47 kDa protein is decreasing during malting. The other proteins (13, 

20 and 23–26 kDa) were increased during the process, which can 

be explained by the increasing enzyme activity. In addition, 

a smaller protein peak area (6–9 kDa) was found with LabChip, 

which cannot be detected with gel electrophoresis, because the 

commonly used protein analysis technique detects proteins above 

14 kDa only. These small proteins and the protein with the size of 

61 kDa are probably protein breakdown products, which are 

formed during malting and are water soluble. The formation of 

water-soluble protein breakdown products is well known for barley 

proteins (Kunze, 1996). The isoelectric points of oat albumins, 

detected with two-dimensional gel electrophoresis (Fig. 1B), range 

from pH 5 to 8. Similar values have been reported by Ma and 

Harwalkar (1984). Amino acid analysis of oat albumins revealed an 

Table 1 

Free amino acid composition of the unmalted and malted oat grain (mg/g [dry wt]) 

Amino Acid 

Essential 

Unmalted Malted 

Isoleucine 15.0 0.7 245.3 5.0 

Leucine 18.5 0.9 316.5 9.0 

Lysine 42.3 2.0 356.5 9.5 

Threonine 21.2 1.0 654.3 12.0 

Valine 46.2 2.1 430.3 10.1 

Phenylalanine 22.1 1.1 477.5 9.9 

Histidin 58.0 2.8 322.5 8.9 

Arginine 94.1 4.0 735.5 13.2 

Nonessential 

Alanine 55.0 2.2 364.5 9.3 

Aspartic acid 103.7 5.0 189.1 4.5 

Glutamic acid 254.1 5.2 481.6 10.5 

Glycine 21.4 1.0 67.2 2.7 

Proline 0 664.4 12.1 

Serine 25.4 1.2 277.3 6.4 

Tyrosine 8.0 0.3 336.9 6.9 

1 

2 


Unmalted glutelins 

Malted glutelins 

3 

A 

90 

kDa 

10 

kDa 

90 

kDa 

10 

kDa 

increase in all amino acids, with the exception of arginine and 

histidine. The increase of proline by 48% is notable. The high 

increase in albumin–proline probably contributes to the strong 

increase in free proline as mentioned in Section 4.1. But also 

phenylalanine, aspartic acid and tyrosine were increased markedly. 

This effect concurs with results reported by Wu (1983), where an 

increase in phenylalanine plus tyrosine was reported during 

germination. 

4.2.2. Globulin fraction 

The salt water-soluble protein of oats is known to be the major 

protein fraction. This fraction functions as storage proteins and 

consists of 3S, 7S and 12S globulins (Lásztity, 1996; Shewry and 

Halford, 2002). Peterson (1978) showed that the 12S globulin has 

a molecular weight of 322 kDa and consists of two subunits with 

molecular weights of about 32 and 22 kDa, called the a- and bsubunits. 

As shown in Fig. 2, these subunits could be detected with 

the Protein 230 þ LabChip, covered in area 1. This type of protein 

seems to be affected only slightly by the degradation process 

B 

pH 3 pH 10 

Fig. 4. (A) Electropherogram of the glutelin fraction of unmalted and malted samples (Protein 80 þ LabChip). (B) 2D gels of unmalted (a) and malted (b) glutelin fraction. 

Table 2 

Total amino acid composition of the unmalted and malted oat grain (mg/g [dry wt]) 

Amino Acid 

Essential 

Unmalted Malted 

Isoleucine 3.88 0.2 3.79 0.1 

Leucine 7.72 0.4 7.16 0.3 

Lysine 4.38 0.2 4.36 0.2 

Threonine 3.22 0.1 3.34 0.1 

Valine 5.86 0.2 5.17 0.2 

Phenylalanine 5.53 0.2 4.77 0.2 

Histidin 3.00 0.1 3.59 0.1 

Arginine 7.81 0.4 6.57 0.3 

Nonessential 

Alanine 5.31 0.2 4.79 0.2 

Aspartic acid 8.27 0.4 8.89 0.4 

Glutamic acid 20.61 1.0 19.38 0.9 

Glycine 5.56 0.2 4.85 0.2 

Proline 5.94 0.3 5.09 0.2 

Serine 4.43 0.2 4.44 0.2 

Tyrosine 2.87 0.1 1.25 0.0 

a 

b

90 

Table 3 

Average amino acid compositions of the different protein fractions extracted from oats grown in Finland (mg/g protein fraction) 

Albumin Globulin 

1a 2a 3a 1b 2b 3b 

Isoleucine 8.33 0.4 7.80 0.3 10.03 0.5 26.49 1.3 26.25 1.3 23.21 1.1 

Leucine 17.45 0.8 16.45 0.8 20.74 1.0 44.03 2.2 42.09 2.1 38.97 1.9 

Lysine 13.91 0.6 11.88 0.5 15.41 0.7 18.89 0.9 19.40 0.9 21.53 1.0 

Threonine 11.54 0.5 11.11 0.5 14.59 0.7 20.85 1.0 19.66 0.9 19.04 0.9 

Valine 12.13 0.6 11.18 0.5 14.61 0.7 33.34 1.6 29.71 1.4 26.68 1.3 

Phenylalanine 9.73 0.4 11.68 0.5 13.18 0.6 34.30 1.7 31.50 1.5 29.25 1.4 

Histidin 8.15 0.4 6.92 0.3 8.00 0.4 18.16 0.9 17.92 0.8 17.34 0.8 

Arginine 16.88 0.8 13.33 0.6 15.73 0.7 54.27 2.7 52.31 2.6 44.96 2.2 

Alanine 17.52 0.8 15.41 0.7 20.82 1.0 27.64 1.3 27.08 1.3 27.53 1.3 

Aspartic acid 24.09 1.2 24.07 1.2 32.46 1.6 55.52 2.7 52.93 2.6 50.55 2.5 

Glutamic acid 39.00 1.9 35.07 1.7 43.84 2.1 127.37 6.3 114.37 5.7 91.51 4.5 

Glycine 24.96 1.2 23.90 1.1 31.57 1.5 28.81 1.4 28.63 1.4 26.83 1.3 

Proline 13.45 0.6 14.77 0.7 19.92 0.9 23.09 1.1 24.46 1.2 21.39 1.0 

Serine 20.75 1.0 20.60 1.0 26.74 1.3 29.97 1.4 26.93 1.3 24.32 1.2 

Tyrosine 12.44 0.6 12.65 0.6 16.64 0.8 18.84 0.9 16.85 0.8 14.72 0.7 

Total 289.65 276.88 357.60 587.18 554.38 503.11 

Prolamin Glutelin 

1c 2c 3c 1d 2d 3d 

Isoleucine 11.44 0.5 10.48 0.5 7.48 0.3 8.96 0.4 8.63 0.4 6.21 0.3 

Leucine 33.34 1.6 29.87 1.4 21.15 1.0 13.31 0.6 13.43 0.6 10.37 0.5 

Lysine 4.84 0.2 5.01 0.2 4.03 0.2 7.50 0.3 3.74 0.1 4.79 0.2 

Threonine 7.51 0.3 7.02 0.3 5.21 0.2 6.55 0.3 5.85 0.2 4.65 0.2 

Valine 18.66 0.9 16.69 0.8 11.54 0.5 12.14 0.6 11.00 0.5 7.21 0.3 

Phenylalanine 21.73 1.0 18.53 0.9 14.33 0.7 11.40 0.5 10.53 0.5 9.81 0.4 

Histidin 9.51 0.4 8.77 0.4 7.54 0.3 7.93 0.3 6.67 0.3 7.37 0.3 

Arginine 17.13 0.8 16.36 0.8 11.23 0.5 16.46 0.8 15.00 0.7 12.08 0.6 

Alanine 14.19 0.7 13.04 0.6 9.81 0.4 9.05 0.4 8.55 0.4 7.36 0.3 

Aspartic acid 12.72 0.6 12.17 0.6 10.91 0.5 18.07 0.9 17.14 0.8 12.31 0.6 

Glutamic acid 125.23 4.2 104.82 5.0 71.79 3.5 37.76 1.8 32.00 1.5 27.48 1.3 

Glycine 8.26 0.4 7.93 3.9 6.24 0.3 8.82 0.4 8.03 0.4 6.17 0.3 

Proline 22.51 1.1 19.97 0.9 15.35 0.7 6.76 0.3 5.81 0.2 2.98 0.1 

Serine 9.18 0.4 8.31 0.4 6.30 0.3 8.89 0.4 7.48 0.3 5.97 0.2 

Tyrosine 4.60 0.2 4.06 0.2 2.89 0.1 6.58 0.3 5.79 0.2 3.85 0.1 

Total 354.07 311.69 224.83 192.38 169.42 134.67 

Sample 1a–1d ¼ unmalted; sample 2a–2d ¼ germinating; sample 3a–3d ¼ malted. 

taking place during malting. In contrast to the 12S globulin, the 7S 

globulin, which mainly consists of a 55 kDa polypeptide and is 

found in area 2 in Fig. 2, is decreasing during the malting process. 

This area probably represents only the 7S globulin, although the 

disulfide-bonded dimer, formed by the a- and b-subunit has 

a molecular weight of 54 kDa. Due to the fact that analysis was 

performed under reducing conditions, no disulfide-bond 

remained. The 3S fraction was described in the literature as 

a protein consisting of a 15 and 21 kDa subunit (Lásztity, 1996). 

The 15 kDa polypeptide could not clearly be detected with the 

LabChip, but the 21 kDa component was probably found in area 1 

(Fig 2B). Amino acid analysis revealed decreasing proteins as well, 

since the total amino acid amount of the globulin fraction is 

decreasing, which concurs with the degradation in the globulin 

fraction during malting. In particular the lysine content was found 

to be increased during the process, which is remarkable from 

a nutritional point of view and was also found by Wu (1983), who 

found an increasing lysine content in germinating oats. This can be 

attributed to the fact that the globulin fraction, which is high in 

lysine, was degraded during the malting process. This probably 

revealed free lysine, which is water soluble and thus, was 

extracted with the albumin and globulin fraction. The amino acid 

composition of globulins corresponds to that reported from 

Lásztity (1996). Glutamic acid is the most frequently found amino 

acid and a decrease by 28% of this amino acid was detected, which 

is probably due to decreases in the 12S globulin subunit, since the 

12S subunit has been described to contain a higher amount of 

glutamic acid than the 7S globulin. 


4.2.3. Prolamin fraction 

Since oats main storage protein fraction is the globulin fraction, 

oats prolamin fraction only accounts for 15% of total oat protein 

(Lásztity, 1996), which is different compared to other cereals, such 

as wheat, barley or rye. However, two-dimensional gel electrophoresis 

revealed more than 20 components (Fig. 3B, a). The Labon-a-Chip 

analysis disclosed three main peak areas in the same 

molecular weight range (Fig. 3A). The pH of isoelectric focusing was 

found to range from pH 6 to 9. These results concur with those 

reported by Lásztity (1996). During the malting process the 

prolamin fraction was degraded almost entirely, which is in 

agreement with other germinating cereals such as barley, reported 

by Celus et al. (2006). The degradation by endoproteases is due to 

the fact that storage proteins are supporting the embryo during the 

first stages of development. The oats prolamin fraction is like many 

other cereals also rich in the amino acids proline and glutamine 

(Simpson, 2001). Thus, prolamin proteins most likely function as 

storage proteins, because the aim of the glutamine- and prolinerich 

polypeptides is to supply the embryo with amino acids and 

nitrogen during germination. 

4.2.4. Glutelin fraction 

The proteins remaining after removing the water-soluble albumins, 

the salt water-soluble globulins and the alcohol soluble 

prolamins are called glutelins. Unfortunately, these extractions are 

generally incomplete and some nitrogen remains in the residue. 

The quantity of glutelins observed is often directly related to the 

efficiency of the preceding extractions of the albumins, globulins

and prolamins (Shewry et al., 1978). Thus, it is likely to extract not 

only glutelins with the last extraction step, but also remaining 

proteins from the water, salt water and alcohol soluble fractions, 

especially the globulin fraction (Lásztity, 1996). Robert et al. (1985) 

reported two-dimensional gel electrophoresis of oats proteins, 

where they found a few proteins spots, which probably are true 

candidates for the glutelin fraction, along with proteins spots, 

which are most likely a- and b-globulin subunits. A similar effect 

can be seen in our results of the glutelin fraction (Fig. 4). Compared 

to the gels from the globulin fraction (Fig. 2C) it seems likely that 

not all globulins had been extracted with the 5.0% NaCl solution and 

remained until the extraction with urea, SDS and DTT, since the 

molecular weights of 22–27 and 42–51 kDa could be found in both 

fractions. The proteins ranging from 22 to 27 kDa are probably bsubunit 

proteins from the 12S globulin. However, the LabChip 

revealed a so far unknown polypeptide with molecular weight of 

9 kDa, which is most likely a glutelin protein, that seems to be 

unaffected by the malting process. 

In conclusion, this study reveals an understanding about the 

protein changes taking place during malting of oats. In general 

a degradation of the proteins to small peptides and amino acids could 

be observed in any fraction except in the albumin fraction, in which 

some proteins increased in amount. This is due to the fact that these 

represent most likely metabolically active proteins. Amino acid 

analysis supported the observation of increased protein amount in 

the albumin fraction and decreased protein amounts in the other 

fractions. In addition, so far unknown proteins could be detected in 

the albumin and glutelin fraction with the Lab-on-a-Chip analysis. 

This technique was found to be appropriate for analysis of degrading 

or increasing proteins, as it revealed repeatable and reliable results, 

which could be validated by using common protein analysis techniques 

such as two-dimensional gel electrophoresis. 


This project was supported by the Irish Government under the 

National Development Plan, 2006–2013. The authors would like to 

thank Mrs. Paula O’Connor for her support in amino acid analysis. 

References 

Agilent Technologies, 2005. Agilent 2100 Bioanalyzer - 2100 Expert User’s Guide. 

Agilent Technologies, Waldbronn, Germany. 

Briggs, D.E., Hough, J.S., Stevens, R., Young, T.W., 1981. Malting and Brewing Science, 

second ed. Chapman & Hall, London, U.K. 

Brijs, K., Trogh, I., Jones, B.L., Delcour, J.A., 2002. Proteolytic enzymes in germinating 

rye grains. Cereal Chemistry 79, 423–428. 

Celus, I., Brijs, K., Delcour, J.A., 2006. The effects of malting and mashing on barley 

protein extractability. Journal of Cereal Science 44, 203–211. 


Chittenden, C.G., Laidman, D.L., Ahmad, N., Wyn Jones, R.G., 1987. Amino acids and 

quaternary nitrogen compounds in the germinating wheat grain. Phytochemistry 

17, 1209–1216. 

Kunze, W., 1996. Technology Brewing and Malting, seventh ed. Versuchs- und 

Lehranstalt für Brauerei, Berlin, Germany. 

Lapvetelainen, A., Bietz, J.A., Huebner, F.R., 1995. Reversed-phase high-performance 

liquid chromatography of oat proteins: application to cultivar comparison and 

analysis of the effect of wet processing. Cereal Chemistry 72, 259–264. 

Lásztity, R., 1996. The Chemistry of Cereal Proteins, second ed. CRC Press, Boca 

Raton, Florida. 

Ma, C.-Y., Harwalkar, V.R., 1984. Chemical characterisation and functionality 

assessment of oat protein fractions. Journal of Agricultural and Food Chemistry 

32, 144–149. 

Mariotti, F., Tome, D., Mirand, P.P., 2008. Converting nitrogen into protein – beyond 

6.25 and Jones’ factors. Critical Reviews in Food Science and Nutrition 48, 

177–184. 

Moore, S., Stein, W.H., 1963. Chromatographic determination of amino acids by 

the use of automatic recording equipment. Methods in Enzymology 6, 

819–831. 

Mossé, J., 1990. Nitrogen to protein conversion factor for ten cerals and six legumes 

or oilseeds. A reappraisal of its definition and determination. Variation 

according to species and to seed protein. Journal of Agricultural and Food 


Peraaho, M., Kaukinen, K., Mustalahti, K., Vuolteenaho, N., Maki, M., Laippala, P., 

Collin, P., 2004. Effect of an oats-containing gluten-free diet on symptoms and 

quality of life in coeliac disease. A randomized study. Scandinavian Journal of 

Gastroenterology 39, 27–31. 

Peterson, D.M., 1978. Subunit structure and composition of oat seed globulin. Plant 


Peterson, D.M., 2001. Oat antioxidants – mini review. Journal of Cereal Science 33, 

115–129. 

Ren, Y., Ellis, P.R., Ross-Murphy, S.B., Wang, Q., Wood, P.J., 2003. Dilute and semidilute 

solution properties of (1 / 3), (1 / 4)-b-D-glucan, the endosperm cell 

wall polysaccharide of oats (Avena sativa L.). Carbohydrate Polymers 53, 

401–408. 

Robbins, G.S., Pomeranz, Y., Briggle, L.W., 1971. Amino acid composition of oat 

groats. Journal of Agricultural and Food Chemistry 19, 536. 

Robert, L.S., Nozzolillo, C., Altosaar, I., 1985. Characterization of oat (Avena sativa L.) 

residual proteins. Cereal Chemistry 62, 276–279. 

Robert, L.S., Matlashewski, G.J., Adeli, K., Nozzolillo, C., Altosaar, I., 1983a. Electrophoretic 

and developmental characterization of oat (Avena sativa L.) 

globulins in cultivars of different protein content. Cereal Chemistry 60, 

231–234. 

Robert, L.S., Nozzolillo, C., Altosaar, I., 1983b. Molecular weight and charge 

heterogeneity of prolamins (avenins) from nine oat (Avena sativa L.) cultivars of 

different protein content and from developing seeds. Cereal Chemistry 60, 

438–442. 

Shewry, P.R., 1995. Plant storage proteins. Biological Reviews 70, 375–426. 

Shewry, P.R., Halford, N.G., 2002. Cereal seed storage proteins: structures, properties 

and role in grain utilization. Journal of Experimental Botany 53, 

947–958. 

Shewry, P.R., Hill, J.M., Pratt, H.M., Leggatt, M.M., Miflin, B.J., 1978. An evaluation of 

techniques for the extraction of hordein and glutelin from barley seed and 

a comparison of the protein composition of Bomi and Risø 1580. Journal of 

Experimental Botany 29, 677–692. 

Simpson, D.J., 2001. Proteolytic degradation of cereal prolamins – the problem with 

proline. Plant Science 161, 825–838. 

Wu, Y.V., 1983. Effect of germination on oats and oat protein. Cereal Chemistry 60, 

418–420. 

Zarkadas, C.G., 1982. A comparison of the amino acid composition of two 

commercial oats groats. Cereal Chemistry 59, 323–327.

Effect of milling, pasta making and cooking on minerals in durum wheat 

Francesco Cubadda a, *, Federica Aureli a , Andrea Raggi a , Marina Carcea b 

a Istituto Superiore di Sanità, Viale Regina Elena 299, 00161 Rome, Italy 

b Istituto Nazionale di Ricerca per gli Alimenti e la Nutrizione, Via Ardeatina 546, 00178 Rome, Italy 

article info 





Keywords: 

Durum wheat 

Pasta 

Minerals 

Trace elements 

Milling 

Processing 

Cooking 


abstract 

Cereals and derived products are among the major dietary 

sources of essential elements for humans. The contribution of 

cereal products to the estimated dietary intake of several minerals 

and nutritionally beneficial trace elements is about 20–30% of the 

total intake in Western countries (Carcea et al., 2007). In the case of 

iron and manganese the contribution is as high as 40–50% (Carcea 

et al., 2007). These figures are even higher in other regions of the 

world (Choi and Kim, 2007; Hattori et al., 2004) and especially in 

developing countries (Hussein and Brüggemann, 1997). 

Wheat is the main cereal crop used for human consumption in 

many areas worldwide. Common wheat (Triticum aestivum L.) is 

widely used for breadmaking, whereas durum wheat, i.e., Triticum 

turgidum L. subsp. durum (Desf.) Husn., is mainly employed in the 

production of other food items, pasta being the most popular. In 

Italy, the main pasta producer, pasta is a staple food. However pasta 

Abbreviations: RDA, daily recommended dietary allowance. 


E-mail address: francesco.cubadda@iss.it (F. Cubadda). 


doi:10.1016/j.jcs.2008.07.008 





The effect of technological processing on the contents of eight minerals – i.e., calcium, copper, iron, 

magnesium, phosphorous, potassium, selenium, and zinc – was investigated in pasta making. Milling of 

durum wheat as well as pasta making were carried out in a pilot plant by using three different grain 

samples. Pasta samples purchased on the market were also surveyed to gain information on the mineral 

content of commercial products. The effect of cooking was also investigated in order to determine the 

retention of the selected elements in the final ‘ready-to-eat’ product. Analyte concentrations in whole 

grains, semolina, pasta and cooked pasta were determined by inductively coupled plasma-mass 

spectrometry. 

Conventional roller milling significantly reduced the content of each mineral in durum wheat grains. 

However concentration losses as a consequence of milling widely differed among elements, from 16% 

for Se to 66% for Mg and Zn on a dry weight basis. Retention of elements after milling followed the order 

Se > Ca > Cu > P z K > Fe > Mg z Zn. Pasta making had little effect on element concentrations in 

semolina. Cooking caused an increase in the calcium content of pasta whereas the concentrations of the 

other elements were either unchanged or slightly reduced (0–18% on a dry weight basis) except 

potassium, which showed a decrease of 74%. 

Commercial pasta samples showed concentrations of minerals similar to those of the experimental 

samples, except selenium which was higher due to the use of imported wheat with higher levels of 

selenium in industrial semolina production. Overall, pasta appears to be a valuable source of several 

minerals, especially selenium, copper, magnesium, and zinc. 


consumption has become widespread in several countries (UNIPI, 

2007). Notwithstanding its popularity, the mineral profile of pasta 

as influenced by processing (i.e., milling and pasta making) and 

cooking has been the subject of systematic investigations mainly in 

the 1980s (Albrecht et al., 1987; Brondi et al., 1984; Ranhotra et al., 

1984, 1985; Toepfer et al., 1972; Yaseen, 1993). The introduction of 

new cultivars as well as alterations in agronomic practices and 

changes in the processing conditions do require updated studies. 

These studies should employ state-of-the-art techniques and 

include analytical quality control schemes allowing, for instance, 

precise and accurate determinations of essential microelements 

such as selenium. 

Milling is the critical process affecting the concentrations of 

inorganic elements in wheat-derived food products. As the outer 

parts of the kernel, especially the aleurone layer and the germ, are 

richer in minerals when compared to the starchy endosperm, 

conventional milling reduces their content in flour (semolina, in the 

case of durum wheat) and concentrates them in the milling residues 

(Brondi et al., 1984). However, it has been reported that 

differences in the mineral content likely exist even between the 

inner endosperm and the outer endosperm (Pomeranz, 1988). The 

grain shape and texture (which both depend also on cultivar) and

the technical conditions of milling, primarily the extraction rate, are 

important in determining the extent of mineral loss. However, 

when all these variables are fixed, the distribution of the mineral in 

the various milling fractions ultimately depends on how the 

element is unevenly distributed within the kernel. Therefore, it will 

vary on an element-specific basis. 

There is little information on the effect of milling on durum 

wheat minerals (Brondi et al., 1984; Toepfer et al., 1972). This 

information is needed to assess the nutritional significance of 

mineral loss as a consequence of conventional roller milling and to 

provide basic knowledge in order to establish, for instance, whether 

there is a need for fortification, which minerals should be supplemented 

(in bioaccessible form), and at what levels. Limited and 

sometimes contradictory data are available on the effect of further 

processing of semolina into pasta as well as pasta cooking (Albrecht 

et al., 1987; Ranhotra et al., 1984, 1985; Yaseen, 1993). 

The present study was undertaken to determine the effect of 

conventional roller milling, pasta making and cooking on eight 

nutritionally important minerals (namely calcium, copper, iron, 

magnesium, phosphorous, potassium, selenium, and zinc) in durum 

wheat. Milling and pasta making pilot plants were employed for the 

manufacture of semolina and long (‘spaghetti’ type) pasta in order to 

be able to control the whole processing chain. Commercial pasta 

samples from major Italian brands were purchased on the market 

and included in this study for comparison. The effect of cooking was 

investigated in order to gain information on the retention of the 

selected elements in the final ‘ready-to-eat’ product. 


2.1. Milling and pasta making 

Three samples of durum wheat grain of about 10 kg each were 

included in the study. The samples belonged to three different 

cultivars, namely, Appio, Duilio, and Simeto (hereafter indicated as 

samples 1, 2, and 3) grown in Southern Italy and widely used in Italy 

for industrial pasta production. 

Semolina was obtained from each sample by cleaning and 

tempering the grains for 36–40 h to 16.5% moisture and then 

milling them in a pilot mill (Model MLU 202, Bühler, Uzwill, 

Switzerland) equipped with three breaks, three reduction rolls, six 

steel screens and coupled with a purifier to give standard grade 

semolina, following Approved Methods 26-10 A and 26-41 (AACC, 

2000). Semolina yield (ash

94 

(model 510) and an ADX-500 autodilutor (both from CETAC Technologies, 

Omaha, NE, USA), was used for quantification of the 

analytes. Details of the instrumentation and the operating conditions 

are reported elsewhere (Cubadda et al., 2002). The analytical 

masses were 43 Ca, 63 Cu, 57 Fe, 39 K, 24 Mg, 31 P, 77þ82 Se, and 64 Zn. 

Multielemental calibration standards were prepared from 

1000 mg L 1 stock solutions of individual elements (BDH, Poole, 

England) by dilution with 3% v/v ultrapure concentrated HNO3. In 

all measurements, rhodium (20 mg L 1 ) was selected as internal 

standard for correction of matrix effects and instrumental drift. 

Correction of 44 Ca 16 OH interference on 57 Fe was accomplished by 

application of a mathematical equation calculated daily as reported 

previously (Cubadda et al., 2002). 

Randomly selected samples were analysed on different days to 

verify the precision of results. Samples of the two reference 

materials were analysed in each analytical run to check accuracy of 

measurements. Digestion blanks were run together with samples 

belonging to the same analytical batch and their signal was subtracted 

from that of the sample when calculating analyte 

concentrations. 


The existence of significant differences in the element concentrations 

(on a dry weight basis) among the studied matrixes on 

account of the various treatments (i.e., milling, pasta making, 

cooking) was determined by analysis of variance. The test resulted 

significant for all metals (p ¼ 0.05) and thus a multiple comparison 

test (Scheffé test) was performed to identify which treatments 

caused significant variations (p ¼ 0.05). 


The results obtained for the reference materials analysed for 

quality control purposes are summarized in Table 1. Good agreement 

was observed with the certified or best estimated values of 

each element, indicating effective recovery of analytes after 

digestion and subsequent accurate detection. Precision expressed 

as the coefficient of variation (CV) was, on average, 6% for calcium, 

3% for copper, 2% for iron, 1% for potassium, 1% for magnesium, 2% 

for phosphorous, 2% for selenium, and 4% for zinc. 

Tables 2–4 show the element concentrations measured in the 

three grain samples selected for this study and in their derived 

products, including cooked pasta. Each result is the average of four 

experimental replicates, on which duplicate subsampling was 

carried out for analyses. Data are reported both on a dry and on 

a fresh weight basis. Average moisture content of samples was 9.8%, 

Table 1 

Results obtained for the reference materials analysed (N ¼ 6) 

Element CRM 189 (wholemeal flour) RM 8436 (durum wheat flour) 

Ca b 

Cu b 

Fe b 

K d 

Mg d 

P d 

Se b 

Zn b 

Found Certified Found Best estimated 

Mean (c.i.) a 

Mean (c.i.) a 

Mean (c.i.) a Mean (c.i.) a 

540 (22) [520] c 

276 (14) 278 (26) 

6.36 (0.13) 6.4 (0.2) 4.28 (0.10) 4.30 (0.69) 

69.2 (0.6) 68.3 (1.9) 41.1 (1.2) 41.5 (4.0) 

6.23 (0.09) [6.3] c 

3.20 (0.02) 3.18 (0.14) 

1.92 (0.01) [1.9] c 

1.12 (0.02) 1.07 (0.08) 

5.49 (0.03) [5.3] c 

2.91 (0.06) 2.90 (0.22) 

0.137 (0.002) 0.132 (0.010) 1.25 (0.02) 1.23 (0.09) 

56.6 (1.1) 56.5 (1.7) 22.3 (1.1) 22.2 (1.7) 

a 

Uncertainty as half-width of the 95% confidence interval of the mean. 

b 1 

Concentrations in mg g . 

c 

Indicative value. 

d 1 

Concentrations in mg g . 

F. Cubadda et al. / Journal of Cereal Science 49 (2009) 92–97 

12.6%, 10.1% and 61.1% for grain, semolina, pasta and cooked pasta, 

respectively. 

In order to allow easy recognition of the effects of processing 

and cooking, the overall percent variation in element concentrations 

caused by the different treatments in relation to their original 

levels in grain (¼100%) is shown in Tables 2–4, together with the 

percent variation of the element concentrations in each type of 

sample as a consequence of a specific treatment. 

3.1. Effect of milling 

As expected, milling significantly reduced the concentrations of 

all elements (p < 0.01), with average losses ranging from 16% for 

selenium to 66% for magnesium and zinc on a dry weight basis 

(Tables 2–4). 

A modest reduction of selenium concentration upon milling 

(8%–20%) has also been found in studies on common wheat (Lyons 

et al., 2005; Toepfer et al., 1972). Similar to sulphur, selenium, 

which occurs predominantly as selenomethionine in wheat grains, 

is mostly protein-bound and more evenly distributed throughout 

the kernel when compared to other minerals (Lyons et al., 2005). 

Therefore, due to a higher proportion being stored in the endosperm, 

lower proportions of Se are removed in the milling process. 

Calcium showed an average loss of 41% in concentration terms, 

which is in agreement with the results obtained in earlier work on 

durum wheat (Toepfer et al., 1972). Studies on the distribution of 

minerals within the kernel of common wheat reported that the 

endosperm contains about 50% of calcium whereas about 25% is 

found in the aleurone layer (Pomeranz, 1988). For other elements, 

such as magnesium and zinc, only a minor proportion is found in 

the endosperm, whereas 70% and w50%, respectively, is in the 

aleurone (Bock, 2000; Pieczonka and Rosopulo, 1985; Pomeranz, 

1988). Accordingly, calcium was found to follow an entirely 

different pattern from that of magnesium, zinc and iron in industrially 

milled wheat flours (Lorenz et al., 1980). The level of calcium 

depended mainly on the calcium content of the wheat, whereas 

magnesium, zinc and iron depended more on milling variables 

(Lorenz et al., 1980). 

Next to calcium, copper showed a concentration decrease of 47% 

in this study. This is similar to the results obtained in earlier work 

on durum wheat (Brondi et al., 1984; Toepfer et al., 1972) and 

matches studies showing that about 45% of copper is in the endosperm 

of common wheat kernels (O’Dell et al., 1972; Pieczonka and 

Rosopulo, 1985). 

The concentration of phosphorous and potassium in semolina 

was 56% lower than that in parent grains. Phosphorous mainly 

occurs in wheat kernels as phytic acid and its salts. Potassium is 

associated with phosphorous as it forms, together with magnesium, 

a major part of the phytates found in the kernels (Bock, 2000). 

A sizeable reduction of phytic acid in semolina is achieved upon 

milling of durum wheat grains so that a major proportion of 

phosphorous is present as nucleoprotein, lipid, and inorganic 

phosphorous (Pomeranz, 1988). Since phytic acid is a metalchelating 

agent which can lower the absorption of several essential 

metals (including iron, zinc, calcium and magnesium) its removal 

into milling by-products is nutritionally beneficial (Bock, 2000; 

O’Dell et al., 1972). Recently, low phytic acid mutants of wheat have 

been isolated with the view of improving the nutritional quality of 

wheat by reducing the major storage form of phosphorous and 

increasing the level of inorganic phosphorous, which is more 

readily absorbed by humans and other monogastric animals (Guttieri 

et al., 2006). 

A group of three elements showed a major decrease in 

concentration following milling, i.e., iron (63%), magnesium and 

zinc (66%). These metals are particularly abundant in the aleurone 

(O’Dell et al., 1972; Pieczonka and Rosopulo, 1985). Zinc is present

Table 2 

Variations of potassium, phosphorous, and magnesium concentration in durum wheat as a consequence of processing and cooking 

Sample On a fresh weight basis On a dry weight basis 

Grain Semolina Pasta Cooked pasta Grain Semolina Pasta Cooked pasta 

Potassium 

1 a 

4.37 0.04 1.99 0.02 2.06 0.02 0.22 0.01 4.85 0.04 2.28 0.02 2.30 0.02 0.58 0.01 

2 a 

4.32 0.04 1.80 0.01 1.87 0.01 0.20 0.01 4.79 0.04 2.06 0.02 2.08 0.02 0.52 0.01 

3 a 

5.08 0.04 2.07 0.02 2.14 0.02 0.25 0.01 5.62 0.05 2.37 0.02 2.37 0.02 0.64 0.01 

% vs prec b 

100 43 (41–46) 104 (103–104) 11 (11–12) 100 44 (42–47) 101 (100–101) 26 (25–27) 

% vs grain c 

100 43 (41–46) 44 (42–47) 5 (5–5) 100 44 (42–47) 44 (42–47) 11 (11–12) 

Phosphorous 

1 a 

2 a 

3 a 

% vs prec b 

% vs grain c 

Magnesium 

1 a 

2 a 

3 a 

% vs prec b 

% vs grain c 

4.04 0.07 1.70 0.03 1.76 0.03 0.65 0.01 4.48 0.07 1.95 0.03 1.96 0.03 1.68 0.03 

3.69 0.06 1.68 0.03 1.73 0.03 0.61 0.01 4.10 0.07 1.92 0.03 1.92 0.03 1.57 0.03 

4.07 0.07 1.69 0.03 1.74 0.03 0.59 0.01 4.50 0.07 1.93 0.03 1.94 0.03 1.52 0.02 

100 43 (42–45) 103 (103–103) 36 (34–37) 100 44 (43–47) 100 (100–100) 82 (79–86) 

100 43 (42–45) 44 (43–47) 16 (15–17) 100 44 (43–47) 45 (43–47) 37 (34–38) 

1.20 0.01 0.39 0.01 0.40 0.01 0.18 0.01 1.33 0.01 0.44 0.01 0.45 0.01 0.45 0.01 

1.04 0.01 0.36 0.01 0.38 0.01 0.16 0.01 1.15 0.01 0.42 0.01 0.43 0.01 0.42 0.01 

1.20 0.01 0.40 0.01 0.41 0.01 0.18 0.01 1.33 0.01 0.45 0.01 0.46 0.01 0.46 0.01 

100 33 (32–35) 104 (104–105) 43 (43–44) 100 34 (33–36) 101 (101–102) 100 (99–101) 

100 33 (32–35) 35 (34–37) 15 (15–16) 100 34 (33–36) 35 (34–37) 35 (34–37) 

a 1 

Concentration in mg g 95% confidence interval of the mean. 

b 

Percent variation versus preceding item: average of the three samples (range). 

c 

Percent variation versus grain: average of the three samples (range). 

at a relatively high concentration in the germ as well, this part 

accounting for over 10% of the zinc burden in the caryopsis 

compared to, e.g., 5% of copper (Pieczonka and Rosopulo, 1985). As 

a consequence, a major proportion of iron, magnesium and zinc in 

the grain is removed in the conventional roller milling process. 

Overall, the retention of elements after milling followed the 

order Se > Ca > Cu > P z K > Fe > Mg z Zn. The sequence 

Ca > Cu > (P, K, Fe, Mg, Zn) was obtained in previous investigations 

on durum wheat by Toepfer et al. (1972) (in this case with an 

inversion between phosphorous and copper) and Brondi et al. 

Table 3 

Variations of calcium, iron, zinc, and copper concentration in durum wheat as a consequence of processing and cooking 

(1984). The same order results from studies on common wheat 

(Lorenz et al., 1980; Pomeranz and Dikeman, 1983; Toepfer et al., 

1972), even though some investigations found a higher proportion 

of either potassium (Brondi et al., 1984; Lyons et al., 2005; Zhang 

et al., 1997) oriron(Brondi et al., 1984; Brüggemann and Kumpulainen, 

1995; Rao and Deosthale, 1981; Zhang et al., 1997) in flour. In 

most studies the percentage retentions of copper and phosphorous 

in flour were found to be similar (Pomeranz and Dikeman, 1983; 

Rao and Deosthale, 1981; Toepfer et al., 1972; Zhang et al., 1997), 

although generally slightly higher for copper. Deviations from the 



Calcium 

1 a 

373 18 210 10 228 11 169 8 413 20 241 12 254 12 436 21 

2 a 

412 20 224 11 248 12 193 9 456 22 256 12 276 13 496 24 

3 a 

346 17 206 10 229 11 180 9 383 18 235 11 254 12 462 22 

% vs prec b 

100 57 (54–60) 110 (108–111) 77 (74–79) 100 59 (56–61) 107 (105–108) 178 (172–182) 

% vs grain c 

100 57 (54–60) 63 (60–66) 48 (45–52) 100 59 (56–61) 63 (61–66) 112 (105–121) 

Iron 

1 a 

2 a 

3 a 

% vs prec b 

% vs grain c 

Zinc 

1 a 

2 a 

3 a 

% vs prec b 

% vs grain c 

Copper 

1 a 

2 a 

3 a 

% vs prec b 

% vs grain c 

42.0 0.7 14.7 0.2 15.7 0.3 6.3 0.1 46.6 0.7 16.9 0.3 17.5 0.3 16.3 0.3 

33.7 0.5 12.3 0.2 13.4 0.2 5.1 0.1 37.4 0.6 14.1 0.2 14.9 0.2 13.0 0.2 

37.0 0.6 13.2 0.2 14.5 0.2 5.4 0.1 41.0 0.7 15.1 0.2 16.1 0.3 13.9 0.2 

100 36 (35–36) 109 (107–110) 38 (37–40) 100 37 (36–38) 106 (104–107) 89 (86–93) 

100 36 (35–36) 39 (37–40) 15 (15–15) 100 37 (36–38) 39 (38–40) 35 (34–35) 

39.5 1.3 13.3 0.4 13.7 0.4 5.2 0.2 43.8 1.4 15.3 0.5 15.3 0.5 13.3 0.4 

33.6 1.1 10.3 0.3 10.7 0.3 4.5 0.1 37.3 1.2 11.8 0.4 11.9 0.4 11.6 0.4 

29.8 1.0 10.3 0.3 10.8 0.3 4.4 0.1 33.0 1.1 11.8 0.4 12.0 0.4 11.3 0.4 

100 33 (31–35) 104 (103–104) 40 (38–42) 100 34 (32–36) 101 (100–101) 93 (87–97) 

100 33 (31–35) 34 (32–36) 14 (13–15) 100 34 (32–36) 34 (32–36) 32 (30–34) 

5.54 0.13 2.92 0.07 3.06 0.07 1.13 0.03 6.15 0.15 3.35 0.08 3.41 0.08 2.92 0.07 

5.30 0.13 2.62 0.06 2.71 0.07 0.98 0.02 5.88 0.14 3.00 0.07 3.02 0.07 2.52 0.06 

5.47 0.13 2.87 0.07 3.09 0.07 1.08 0.03 6.06 0.15 3.28 0.08 3.43 0.08 2.76 0.07 

100 52 (49–53) 105 (104–107) 36 (35–37) 100 53 (51–55) 102 (101–104) 83 (80–86) 

100 52 (49–53) 54 (51–56) 20 (18–20) 100 53 (51–55) 54 (51–57) 45 (43–47) 

a 1 

Concentration in mg g 95% confidence interval of the mean. 

b 


c 


F. Cubadda et al. / Journal of Cereal Science 49 (2009) 92–97 95

96 

Table 4 

Variations of selenium concentration in durum wheat as a consequence of processing and cooking 



1 a 

85.8 1.4 72.8 1.2 74.5 1.2 31.2 0.5 95.2 1.5 83.5 1.3 83.1 1.3 80.3 1.3 

2 a 

99.1 1.6 81.2 1.3 83.5 1.3 35.8 0.6 109.8 1.8 93.0 1.5 93.0 1.5 92.0 1.5 

3 a 

110.6 1.8 85.2 1.4 87.5 1.4 36.2 0.6 122.5 2.0 97.3 1.6 97.2 1.6 92.8 1.5 

% vs prec b 

100 81 (77–85) 103 (102–103) 42 (41–43) 100 84 (79–88) 100 (99–100) 97 (95–99) 

% vs grain c 

100 81 (77–85) 83 (79–87) 35 (33–36) 100 84 (79–88) 84 (79–87) 81 (76–84) 

a 1 

Concentration in ng g 95% confidence interval of the mean. 

b 


c 


common pattern occasionally observed for potassium, iron, and 

phosphorous may be due to varying milling conditions among 

studies, primarily extraction rate (Rao and Deosthale, 1981), and to 

an uneven release of some elements from milling equipment, 

especially iron (Cubadda et al., 2005). Furthermore, the proportion 

of each element in the morphological sections of the wheat kernel 

is dependent on the genotype and varies among cultivars (Lyons 

et al., 2005), and this can further explain the slightly different 

patterns from one study to another. 

3.2. Effect of pasta making 

Overall, pasta making had little effect on element concentrations 

in parent semolina. Slight, non-statistically significant 

enrichments of calcium (107%), iron (106%) and copper (102%) were 

observed. The concentrations of these elements in the water used 

for dough preparation were 52 1, 0.10 0.01, and 

0.041 0.02 mg L 1 , respectively, and explained the major part of 

the calcium increase, a low proportion of that of copper, and 1% of 

that of iron. Therefore, the release from pieces of equipment used in 

the pasta making process appeared to be the cause of the 

concentration increase of iron and, to a lesser extent, of copper. 

The elemental concentrations found in the commercial pasta 

samples are summarized in Table 5. These concentrations closely 

match those of the experimental samples. The only differences 

were a slightly higher magnesium concentration (p ¼ 0.014) and an 

almost double concentration of selenium (p < 0.001). Lower 

concentrations of selenium in the experimental samples compared 

to commercial samples are the consequence of the relatively low 

levels of plant-available selenium in the areas of Southern Italy 

from which experimental grain samples originated (Spadoni et al., 

2007). The widespread use of wheat imported from selenium-rich 

areas in industrial semolina production results in a higher level of 

selenium in commercial pasta. 

It was investigated as to whether differences in elemental levels 

between long pasta (spaghetti and related types) and short pasta 

(macaroni and related types) existed in commercial samples. 

Differences, indeed, turned out to be negligible, even though 

a slightly significantly higher content of copper and zinc in long 

pasta was detected (0.01 

Table 5 

Element concentrations in commercial pasta samples (N ¼ 12) a 

Ca 

(mg g 1 Cu 

) (mg g 1 Fe 

) (mg g 1 K 

) (mg g 1 Mg 

) (mg g 1 P 

) (mg g 1 Se 

) (ng g 1 Zn 

) (mg g 1 ) 

Median 266 3.30 12.9 2.07 430 1.66 141 12.7 

Mean 260 3.21 13.0 2.07 430 1.67 154 12.9 

Min 200 2.76 11.3 1.87 398 1.51 84 10.3 

Max 314 3.56 14.7 2.29 457 1.82 227 15.4 

CV(%) 11 9 9 6 5 6 34 14 

a 

Fresh weight basis (average water content for conversion to dry wt basis is 

11.0%). 

F. Cubadda et al. / Journal of Cereal Science 49 (2009) 92–97 

Perhaps of more importance is the variability of concentrations 

in commercial pasta, which differed widely among minerals. 

Potassium, magnesium and phosphorous exhibited the lowest CV 

(5–6%), followed by copper and iron (9%), calcium (11%), zinc (14%) 

and selenium (34%). The large spread of selenium levels is not 

surprising since this element is not essential for plants and its 

concentration in grains and derived cereal products correlates with 

plant-available selenium in the soil (Spadoni et al., 2007). 

The mineral concentrations in pasta in this study are generally 

similar to those found in recent surveys elsewhere (USDA, 2007), 

even though more confidence can be attributed to the data presented 

here due to a more representative sampling. However, in 

earlier studies, substantially lower levels of calcium (Albrecht et al., 

1987; Ranhotra et al., 1984, 1985; Yaseen, 1993), copper (Albrecht 

et al., 1987), potassium (Yaseen, 1993), magnesium (Yaseen, 1993), 

and phosphorous (Ranhotra et al., 1984; Yaseen, 1993) were reported. 

It is unclear whether the lower concentrations measured in 

these earlier studies are the result of analytical shortcomings or 

reflect actual differences due to, e.g., the introduction of new cultivars, 

modifications in agronomic practices, or changes in processing. 

3.3. Effect of cooking 

The effect of cooking can be best assessed considering data 

expressed on a dry weight basis in Tables 2–4. Cooking markedly 

increased calcium concentration of pasta (p < 0.001), which can be 

ascribed to the calcium content of the cooking water (26 1mg 

L 1 ). Except for potassium, the concentrations of the other 

elements remained unchanged (magnesium) or showed a slight 

reduction in the range 79–97%, which turned out to be significant 

only for copper (p < 0.001). Potassium showed a sizeable decrease 

(p < 0.001), its concentration after cooking being reduced to about 

1/4 of that in dry pasta. 

The results obtained in this study can not be easily compared to 

those of previous investigations. When expressed on a dry weight 

basis, the retention values calculated in this study correspond to the 

‘apparent retention’ of nutrients in cooked foods as defined by 

Murphy et al. (1975). Apparent retentions after pasta cooking were 

determined in an earlier study, where slightly lower values were 

found for all minerals (70–82%), including calcium, whereas 

potassium again displayed the highest loss (Yaseen, 1993). 

However, in this latter study, pasta was cooked in distilled deionized 

water without addition of salt, which does not allow easy 

comparison with the results of the present study. A similar 

approach was used by Ranhotra et al. (1984), but in this case no 

distilled water was used and calcium turned out to have the highest 

retention. Other studies (Albrecht et al., 1987; Ranhotra et al., 1985) 

took into account the loss of solids during cooking and calculated 

the so-called ‘true retention’ (Murphy et al., 1975). Albrecht et al. 

(1987) compared the effect of salt addition when pasta is cooked 

either in distilled or tap water. Calcium true retention increased 

from 89% to about 100% when salt was added to distilled water.

Calcium retention in unsalted tap water was about 129% (average of 

macaroni and spaghetti) and increased to 158% upon salt addition, 

with negligible differences following subsequent rinsing. In 

agreement with the other studies, potassium was found to have the 

lowest retention and salt addition decreased it, which probably 

explains the low retention value for this element obtained in the 

present study. 

Studies on different cooking approaches may give insight into 

the effect of each specific medium on mineral retention and suggest 

explanations for different retention patterns. However, in this study 

pasta was cooked in the customary way, i.e., in salted tap water, in 

order to identify the actual mineral content of the final product 

when prepared according to common household practice. 

Using the retention factors obtained in this study, the amount of 

each mineral provided by a standard serving of pasta (80 g of the 

uncooked product) containing each mineral at the average 

concentration found in the commercial samples (Table 5) was 

calculated. This amount was compared with the Italian daily recommended 

dietary allowance (RDA) for the adult population (SINU, 

1996). It resulted that the proportion of the RDA provided by a daily 

serving of pasta is on average 22% for selenium, 18% for copper, 10%– 

14% for zinc (for males and females, respectively), 11% for phosphorous, 

9%–5% for iron (for males and females, respectively), 4% 

for calcium, and 1% for potassium. For magnesium, only a range of 

intakes is established instead of a RDA; if a value one-third above 

the lower level of this range is chosen as a reference, the proportion 

of such an amount provided by a daily serving of pasta turns out to 

be 17%. Overall, pasta appears to be a valuable source of several 

minerals of importance in human nutrition and well-being. 


This study led to a better understanding of the effect of durum 

wheat processing on the levels of eight minerals, namely, calcium, 

copper, iron, magnesium, phosphorous, potassium, selenium, and 

zinc. For selenium, no data on the changes induced by durum wheat 

processing were available so far, notwithstanding the ever 

increasing awareness of the importance of this element to human 

health. Furthermore, the effect of cooking was investigated in order 

to determine the retention factors to be used for the estimation of 

the content of each element in the final product (as consumed) 

when the element concentration in the uncooked product is 

known. 

Milling was the most important processing step in the production 

of conventional pasta in regard to the change in content of 

minerals originally present in the durum wheat grains. At least six 

groups of elements could be distinguished on the basis of their 

concentration decrease upon milling. Selenium had the highest 

retention with concentrations in semolina equal to 77%–85% of that 

in grain (dry weight basis), followed by calcium (54%–60%), copper 

(49%–53%), potassium and phosphorous (42%–47%), iron (36%– 

38%), magnesium and zinc (32%–36%). 

Pasta making had little effect on element concentrations in 

semolina whereas cooking caused negligible to small losses of 

elements, except for calcium and potassium which greatly 

increased and decreased their concentration, respectively. Using 

the retention factors determined by the cooking experiments and 

the average concentrations ascertained in the commercial pasta 

samples it was assessed that pasta can provide nutritionally 

important amounts of several minerals, especially selenium, 

copper, magnesium, and zinc. 


The skilled technical help of Mr. L. Bartoli in the milling of grains 

and in the pasta making is acknowledged. 

F. Cubadda et al. / Journal of Cereal Science 49 (2009) 92–97 97 

References 

AACC International, 2000. Approved Methods of the American Association of Cereal 

Chemists, 10th Ed. The Association, St. Paul, MN, USA. 

Albrecht, J.A., Asp, E.H., Buzzard, I.M., 1987. Contents and retentions of sodium and 

other minerals in pasta cooked in unsalted or salted water. Cereal Chemistry 64, 

106–109. 

Bock, M.A., 2000. Minor constituents of cereals. In: Kulp, K., Ponte, J.G. (Eds.), 

Handbook of Cereal Science and Technology, second ed. Marcel Dekker, New 

York, pp. 479–504. 

Brondi, M., Ciardi, A.M., Cubadda, R., Dall’Aglio, M., Mastino, G.G., Michetti, I., 

Santaroni, G.P., 1984. Trasferimento di elementi traccia dall’ambiente alla catena 

alimentare: livelli in grani e loro prodotti. La Rivista della Società Italiana di 

Scienza dell’Alimentazione 13, 27–38. 

Brüggemann, J., Kumpulainen, J., 1995. The status of trace elements in staple foods. 

II. Some effects of cereal and potato processing. Zeitschrift für Lebensmittel- 

Untersuchung und-Forschung 201, 7–11. 

Carcea, M., Aureli, F., Cubadda, F., 2007. Minerals and trace elements in the Italian 

wheat and products. Tecnica Molitoria International 58, 129–139. 

Choi, M.K., Kim, E.Y., 2007. Evaluation of dietary manganese intake in Korean men 

and women over 20 years old. The Journal of the Korean Society of Food Science 

and Nutrition 36, 447–452. 

Cubadda, F., Baldini, M., Carcea, M., Pasqui, L.A., Raggi, A., Stacchini, P., 2001. 

Influence of laboratory homogenization procedures on trace element content of 

food samples: an ICP-MS study on soft and durum wheat. Food Additives and 

Contaminants 18, 778–787. 

Cubadda, F., Raggi, A., Testoni, A., Zanasi, F., 2002. Multielemental analysis of food 

and agricultural matrixes by inductively coupled plasma-mass spectrometry. 

Journal of AOAC International 85, 113–121. 

Cubadda, F., Raggi, A., Marconi, E., 2005. Effects of processing on five selected metals 

in the durum wheat food chain. Microchemical Journal 79, 97–102. 

Guttieri, M.J., Peterson, K.M., Souza, E.J., 2006. Mineral distributions in milling 

fractions of low phytic acid wheat. Crop Science 46, 2692–2698. 

Hattori, H., Ashida, A., Ito, C., Yoshida, M., 2004. Determination of molybdenum in 

foods and human milk, and an estimate of average molybdenum intake in the 

Japanese population. Journal of Nutritional Science & Vitaminology 50, 404–409. 

Hussein, L., Brüggemann, J., 1997. Zinc analysis of Egyptian foods and estimated 

daily intakes among an urban population group. Food Chemistry 58, 391–398. 

Lorenz, K., Loewe, R., Weadon, D., Wolf, W., 1980. Natural Levels of nutrients in 

commercially milled wheat flours. III. Mineral Analysis. Cereal Chemistry 57, 

65–69. 

Lyons, G.H., Genc, Y., Stangoulis, J.C.R., Palmer, L.T., Graham, R.D., 2005. Selenium 

distribution in wheat grain, and the effect of postharvest processing on wheat 

selenium content. Biological Trace Element Research 103, 155–168. 

Murphy, E.W., Criner, P.E., Gray, B.C., 1975. Comparisons of methods for calculating 

retentions of nutrients in cooked foods. Journal of Agricultural and Food 

Chemistry 23, 1153–1157. 

O’Dell, B.L., de Boland, A., Koirtyohann, S.R., 1972. Distribution of phytate and 

nutritionally important elements among morphological components of cereal 

grains. Journal of Agricultural and Food Chemistry 20, 718–721. 

Pieczonka, K., Rosopulo, A., 1985. Distribution of cadmium, copper, and zinc in the 

caryopsis of wheat (Triticum aestivum L.). Fresenius’ Journal of Analytical 

Chemistry 322, 697–699. 

Pomeranz, Y., 1988. In: Pomeranz, Y. (Ed.), Wheat Chemistry and Technology. 

Chemical composition of kernel structures, Vol. I. American Association of 

Cereal Chemists, St. Paul, MN, USA, pp. 97–158. 

Pomeranz, Y., Dikeman, E., 1983. Minerals and protein contents in hard red winter 

wheat flours. Cereal Chemistry 60, 80–82. 

Ranhotra, G.S., Gelroth, J.A., Novak, F.A., Bock, M.A., Winterringer, G.L., 

Matthews, R.H., 1984. Nutritive value of selected variety breads and pastas. 

Journal of the American Dietetic Association 84, 322–327. 

Ranhotra, G.S., Gelroth, J.A., Novak, F.A., Bock, M.A., Matthews, R.H., 1985. Retention 

of selected minerals in enriched pasta products during cooking. Cereal Chemistry 

62, 117–119. 

Rao, D.S.S., Deosthale, Y.G., 1981. Mineral and trace element composition of wheat 

and wheat flours of different extraction rates. Journal of Plant Foods 3, 251–257. 

SINU, 1996. Livelli di Assunzione Raccomandati di Energia e Nutrienti per la 

Popolazione Italiana (Recommended dietary allowances for energy and nutrients 

for the Italian population). Società Italiana di Nutrizione Umana, Rome. 

Spadoni, M., Voltaggio, M., Carcea, M., Coni, E., Raggi, A., Cubadda, F., 2007. Bioaccesible 

selenium in Italian agricultural soils: comparison of the biogeochemical 

approach with a regression model based on geochemical and 

pedoclimatic variables. The Science of the Total Environment 376, 160–177. 

Toepfer, E.W., Polansky, M.M., Heart, J.F., Slover, H.T., Morris, E.R., Hepburn, F.N., 

Quackenbush, F.W., 1972. Nutrient composition of selected wheats and wheat 

products. XI. Summary. Cereal Chemistry 49, 173–186. 

UNIPI, 2007. Consumo di pasta alimentare nei diversi paesi (Consumption of 

alimentary pasta in the different countries). Web site: http://www.unipi-pasta. 

it/dati/PDF/pdf%202006/TAB31.pdf Tab. 31. 

USDA, 2007. National Nutrient Database for Standard Reference, Release 20, NBD 

No. 20420. Web site: http://nal.usda.gov/fnic/foodcomp/cgi-bin/list_nut_edit.pl. 

Yaseen, A.A.E., 1993. Effect of processing conditions and cooking on retention of 

minerals in macaroni. Die Nahrung 37, 449–455. 

Zhang, Z.W., Moon, C.S., Watanabe, T., Shimbo, S., Ikeda, M., 1997. Contents of 

pollutant and nutrient elements in rice and wheat grown on the neighbouring 

fields. Biological Trace Element Research 57, 39–50.

Starch granule size distribution of hard red winter and hard red spring wheat: 

Its effects on mixing and breadmaking quality 

Seok-Ho Park, Jeff D. Wilson *, Bradford W. Seabourn 

Grain Marketing and Production Research Center, Agricultural Research Service, Department of Agriculture (USDA), 1515 College Avenue, Manhattan, KS 66502, USA 

article info 





Keywords: 

Starch granule size distribution 

HRW 

HRS 

Mixing property 

Breadmaking quality 

Crumb grain score 

Optimum range of B-granules 


abstract 

Starch is an important part of wheat endosperm, not only 

because starch accounts for 65–73% of dry flour mass when milling 

extraction is

Nomenclature 

A-granules larger than 10 mm in diameter 

B-granules smaller than 10 mm in diameter 

HRS hard red spring 

HRW hard red winter 

r simple correlation coefficient 

the Federal Grain Inspection Service (FGIS) Technical Center, 

(Kansas City, MO), Grain Inspection, Packers, and Stockyards 

Administration (GIPSA), U.S. Department of Agriculture. Detailed 

information about the samples has been previously reported 

(Maghirang et al., 2006). 

2.2. Starch isolation and determination of granule size distribution 

Starch was isolated by enzymatic digestion using pepsin A 

(P7012, Sigma, St. Louis, MO), hemicellulase 90 (90,000 U/g activity, 

Amano Enzyme U.S.A., Lombard, IL), and cleaned further by 

a detergent mix (5% SDS, 5% Triton X-100, 5% Tween 40, and 5% 

Triton X-15) (Bechtel and Wilson, 2000). 

The size distribution of isolated starch granules was measured 

using a single wavelength Beckman Coulter LS 13 320 Particle Size 

Analyzer (Beckman Coulter, Miami, FL) with the Universal Liquid 

Module for liquid-based measurements. Each starch sample was 

slurried with 1.0 mL of water and vortexed before analysis. The 

standard refractive indices used were 1.31 for water and 1.52 for 

starch, which is within the sample concentration range of the 

instrument’s specifications. Volumes of all starch granules were 

calculated on the assumption that all granules were spherical in 

shape. 

2.3. Evaluation of wheat and flour properties 

The following properties of wheat were analyzed: test weight 

(lb/bu) by American Association of Cereal Chemists (AACC, 2000) 

Approved Method 55-10; protein content using near-infrared 

reflectance (AACC Approved Method 39-25); single kernel hardness 

(AACC Approved Method 55-31), weight, and size using the 

SKCS 4100 (Perten, Springfield, IL); and ash content (AACC 

Approved Method 08-01). Wheat was milled using a Brabender 

Quadrumat Sr. experimental mill (AACC Approved Method 26-10A). 

Flour protein and ash content were measured using AACC 

Approved Method 39-11 and 08-01, respectively. Flour color (L * , a * , 

and b * ) was determined using a colorimeter (CR-300, Minolta, 

Osaka, Japan). Mixing characteristics of flour were evaluated using 

Mixograph (AACC Approved Method 54-40A) and Farinograph 

(AACC Approved Method 54-21). A modified optimized straightdough 

breadmaking method (AACC Approved Method 10-10B) was 

used for evaluation of experimental breadmaking properties of 

flours. The detailed baking method and description of crumb grain 

score were reported previously (Park et al., 2004). All tests were 

conducted with at least 2 duplicates. 


A complete randomized experimental design was used. The 

difference between starch granule size and volume distributions of 

HRW and HRS was analyzed using the General Linear Models 

procedure of the Statistical Analysis System (SAS Institute, Cary, 

NC). Statistical abbreviations were simple correlation coefficient (r), 

coefficient of determinant (R 2 ), P < 0.01 (*), P < 0.001 (**), 

P < 0.0001 (***), and standard error of the mean (SEM). 

S.-H. Park et al. / Journal of Cereal Science 49 (2009) 98–105 99 

* 

coefficient of determinant 

P < 0.01 

** P < 0.001 

*** P < 0.0001 

SEM standard error of the mean 

SKCS single kernel characterization system 

R 2 


3.1. Starch granule size distributions of HRW and HRS 

Fig. 1 shows a typical bimodal size distribution of wheat starch 

granules, plotting proportions by volume % of granules in equal 

diameter intervals against diameter. Point 1 and 3 represent the 

most frequent differential volume (%) of the small and large granules, 

and Point 2 is differential volume (%) between small and large 

granules that sets apart small and large starch granules. The 

respective integrating areas (%) (A, B, and C) represent proportion of 

volume distribution (%) split by each Point. 

Starch granule size distributions of HRW (n ¼ 98) and HRS 

(n ¼ 99) showed significant differences in mean values of different 

aspects of granule size distribution including differential volume, 

diameter, and volume distribution (Table 1). Table 1 shows that 

HRW has less B-granules as indicated by lower differential volume 

compared to HRS wheat (2.25% vs. 2.80%, respectively). Differential 

volumes of HRW and HRS wheats at Point 2, however, were not 

significantly different. At Point 3, differential volume for HRW 

wheat was higher (6.55%) than that of HRS (6.03%.), and is due to 

the fact that differential volume (%) represents proportionality. 

HRW and HRS wheats have similar starch granule size distribution 

range from less than 1 mm to about 40 mm, thus, a lower % in one 

proportion results in higher % proportion elsewhere. Fig. 2 shows 

typical starch granule size distribution curves of HRW and HRS 

wheats, demonstrating that HRS wheat had higher and lower 

differential volumes at Point 1 and 3, respectively, compared with 

HRW wheat. It should be noted that even though these curves 

(differential volume) were obtained from calculations using real 

numbers of starch granules at a specific size, the differential volume 

represents the proportion of starch granule size distribution in the 

kernel and not the actual volume. 

The mean values of starch granule diameter at each point, 

compared to differential volume, represent actual size. HRW had 

significantly smaller size of B-granules (4.32 mm) compared with 

Differential volume (%) 

8 

7 

6 

5 

4 

3 

2 

1 

Point 1 Point 2 

0 

0 1 2 6 16 40 

Size (diameter, µm) 

Point 3 

Fig. 1. Starch granule size distribution indicating high and low points (Point 1, 2, and 3) 

of differential volume and their volume percents (Area A, B, C).

100 

Table 1 

Starch granule size distribution in differential volume (%), diameter (mm), and volume percent distribution (%) of HRW and HRS a 

HRW HRS 

Size Distribution b 

Mean Minimum Maximum SEM c 

Mean Minimum Maximum SEM c 


Point 1 2.25*** 1.30 3.15 0.025 2.80*** 1.91 4.12 0.029 

Point 2 1.53 0.91 1.96 0.015 1.56 0.73 2.14 0.022 

Point 3 6.55*** 5.17 7.91 0.029 6.03*** 4.74 7.20 0.034 

Diameter (mm) 

Point 1 4.32*** 3.52 5.11 0.030 4.49*** 3.86 5.11 0.029 

Point 2 8.72*** 5.61 10.78 0.063 9.46*** 8.15 10.78 0.043 

Point 3 21.49 18.86 24.95 0.087 21.46 18.86 24.95 0.088 

Volume distribution (%) 

A 23.53*** 14.94 30.87 0.245 29.35*** 20.06 38.24 0.254 

A þ B 38.01*** 12.57 49.07 0.402 46.88*** 34.03 57.81 0.308 

A þ BþC 75.78*** 57.59 89.00 0.300 80.05*** 73.34 87.30 0.180 

a 

*** ¼ Mean values of HRW and HRS in the same row are significantly different at P ¼ 0.0001. 

b 

Explanation of specific points and area are given in Fig. 1. 

c 

SEM ¼ standard error mean. 

HRS (4.49 mm) at Point 1 (Table 1). The observation that HRW 

wheat had a smaller size of B-granules compared with HRS wheat 

was confirmed by the fact that the threshold diameters at Point 2 

were 8.72 mm for HRW wheat and 9.46 mm for HRS wheat. The 

mean diameter of A-granules, however, was not significantly 

different at Point 3 (21.49 and 21.46 mm, respectively). 

The HRW and HRS wheats showed a wide range of volume 

distributions in terms of A- and B-granules. The volume distribution 

of HRW wheat B-granules ranged from 12.57% to 49.07%, while 

HRS wheat B-granules ranged from 34.03% to 57.81% (Table 1, 

Fig. 1). The mean volume distributions of HRW wheat B-granules 

were significantly lower (23.53% and 38.01% for area A and A þ B, 

respectively) than that of HRS (29.35% and 46.88% for area A and 

A þ B, respectively). It is obvious that HRW wheat had smaller 

B-granules in size and number. 

The volume distributions of subdivided ranges of starch granule 

size distribution are shown in Table 2. This data confirms that HRW 

wheat has significantly smaller proportions of B-granules (less than 

10 mm) compared with HRS wheat in our study. 

Specific volume ranges for A- and B-granules derived from HRW 

and HRS wheats have not been reported in the literature, and most 


7 

6 

5 

4 

3 

2 

1 

HRS 

HRW 

0 

0 1 2 6 16 40 

Size (diameter, µm) 

Fig. 2. Starch granule size distribution of HRW (GIPSA identification number: 

03027680) and HRS (GIPSA identification number: 03036243). 

S.-H. Park et al. / Journal of Cereal Science 49 (2009) 98–105 

previous reports on the proportion of B-granules in total starch 

were on a weight basis (Brocklehurst and Evers, 1977; D’Appolonia 

and Gilles, 1971; Dengate and Meredith, 1984; Evers, 1973; Evers 

and Lindley, 1977; Hughes and Briarty, 1976; Meredith, 1981; Park 

et al., 2004; Soulaka and Morrison, 1985a). Our data, however, can 

still be compared with previous reports because the difference in 

density of A- and B-starch granules is negligible (Dengate et al., 

1979; Meredith, 1981). Our volume data on the range of B-granules 

was higher than most previous reports. D’Appolonia and Gilles 

(1971) reported less than 10% B-granules by weight in total starch 

from 12 HRS flours, whereas others reported 30% B-granules (Evers, 

1973; Evers and Lindley, 1977; Hughes and Briarty, 1976). Park et al. 

(2004) observed 15.7–27.0% B-granules from 12 HRW wheat flours, 

and Soulaka and Morrison (1985a) found 13–35%. Stoddard (1999) 

reported a similar range of B-granule volume (%), showing 17–50% 

of B-granules using laser diffraction sizing methods from 130 

Australian hexaploid wheat cultivars and Landraces from Asia. 

Bechtel et al. (1990) found 48% of B- and C-type granules (less than 

15.9 mm) from mature HRW wheat, but this discrimination was 

larger than the commonly used 10-mm range. These different 

observations on the proportion of B-granules in total starch were 

most likely due to different cultivars, growing conditions, and 

starch extraction and analysis methods. 

The smaller size and number of B-granules in HRW wheat 

compared with HRS wheat could be a characteristic of wheat class, 

or it could be caused by higher temperature during the grain-filing 

period. Dengate and Meredith (1984) found that drought affected 

granule size distribution (weight %), decreasing B-granules 

(4–10 mm) and increasing smaller A-granules (10–20 mm). The 

volume % of B-granules decreased when growth temperature was 

increased from 15 Cto40 C during grain-filling period (Shi et al., 

1994). High temperature also seemed to be associated with 

a reduction in the number of B-granules (Bhullar and Jenner, 1985). 

Barley, which also has a bimodal starch granule size distribution, 

showed a similar low ratio in number of B-granules induced by high 

temperature (MacLeod and Duffus, 1988; Tester et al., 1991). These 

environmental effects could cause a difference in the grain-filling 

pattern between A- and B-granules. The A-granule starts to form in 

the amyloplast at about 4–5 days after anthesis (Bechtel et al., 1990; 

Parker, 1985), and continues to increase in size until reaching 

a maximum diameter of 25–50 mm at physiological maturity 

(Bechtel et al., 1990; Dengate and Meredith, 1984; Simmonds and 

O’Brien, 1981). The final number of A-amyloplasts, however, is 

achieved much earlier at approximately seven days post-anthesis 

when cell division ceases (Briarty et al., 1979). On the other hand,

Table 2 

Volume percent distribution (%) on different ranges of granule size of HRW and HRS a 

HRW HRS 

Range (mm) Mean Minimum Maximum SEM b 

Mean Minimum Maximum SEM b 

30 9.6*** 2.2 22.9 0.24 6.5*** 3.3 13.3 0.13 

102 

Table 3 

Correlations between starch granule size distribution of HRW and HRS and wheat and flour properties 

Differential volume (%) Diameter (mm) Area range (%) 

Point 1 Point 2 Point 3 Point 1 Point 2 Point 3 A A þ B AþBþC HRW 

Wheat protein (%) 0.39 *** 0.40 *** 0.49 *** 0.47 *** 

Test weight (bu/lb) 0.36 ** 0.32 * 

Kernel size distribution (%) 

Large kernel 0.56 *** 0.32 * 0.40 *** 0.33 ** 0.61 *** 0.28 * 0.40 *** 0.41 *** 

Medium kernel 0.56 *** 0.33 ** 0.39 *** 0.31 * 0.60 *** 0.28 * 0.39 *** 0.41 *** 

Small kernel 0.28 * 0.28 * 0.36 ** 0.27 * 

Single kernel character 

Weight (mg) 0.40 *** 0.38 *** 0.48 *** 0.34 ** 0.29 * 

Size (mm) 0.49 *** 0.45 *** 0.30 * 0.53 *** 0.30 * 0.33 ** 0.36 ** 

Hardness 0.29 * 0.30 * 0.31 * 

Flour protein (%) 0.35 ** 0.41 *** 0.44 *** 0.45 *** 

Flour color: L 0.32 * 0.41 *** 0.32 * 

HRS 

Wheat protein (%) 0.72 *** 0.48 *** 0.62 *** 0.52 *** 0.67 *** 0.68 *** 0.66 *** 0.30 * 

Test weight (bu/lb) 0.72 *** 0.41 *** 0.67 *** 0.45 *** 0.64 *** 0.70 *** 0.71 *** 0.26 * 

Kernel size distribution (%) 

Large kernel 0.68 *** 0.50 *** 0.60 *** 0.53 *** 0.70 *** 0.30 * 0.67 *** 0.63 *** 0.29 * 

Medium kernel 0.68 *** 0.50 *** 0.60 *** 0.53 *** 0.70 *** 0.30 * 0.67 *** 0.64 *** 0.29 * 

Small kernel 0.29 * 0.28 * 0.29 * 0.28 * 

Single kernel character 

Weight (mg) 0.59 *** 0.49 *** 0.53 *** 0.56 *** 0.48 *** 0.27 * 0.59 *** 0.45 *** 

Size (mm) 0.63 *** 0.44 *** 0.58 *** 0.57 *** 0.54 *** 0.63 *** 0.53 *** 

Hardness 0.29 * 0.27 * 

Flour protein (%) 0.72 *** 0.51 *** 0.59 *** 0.54 *** 0.70 *** 0.29 * 0.68 *** 0.64 *** 0.28 * 

Flour color: L 0.53 *** 0.33 ** 0.50 *** 0.44 *** 0.52 *** 0.54 *** 0.50 *** 0.31 * 

*, **, and *** ¼ significant correlations at P < 0.01, P < 0.001, and P < 0.0001, respectively. 

at Points 1 and 2 (r ¼ 0.53***, 0.65***, respectively), and area 

range A and A þ B(r ¼ 0.64***, 0.59***, respectively) showed 

significant inverse correlations with mixing absorption. This 

observation is most likely due to mixing absorption being generally 

positively correlated to protein content (Ohm and Chung, 1999; 

Park et al., 2006). The B-granules were reported to have higher 

Table 4 

Correlations between starch granule size distribution and mixing properties 


swelling power (intragranular plus entrained intergranular water), 

which was due to higher water absorption capability, and was 

triggered only when temperature was much higher than normal 

mixing temperature, above 90 C(Kulp, 1973; Park et al., 2004; 

Seib, 1994; Wong and Lelievre, 1982). Haraszi et al. (2004) observed 

a decrease in water absorption when protein content decreased 



Mixograph 

Absorption 

Mix time 

0.30 * 0.42 *** 0.40 *** 0.42 *** 

Mix tolerance 0.26 * 

Farinograph 

Absorption 0.35 ** 0.30 * 0.36 ** 

Development time 0.32 * 0.32 * 0.28 * 0.31 * 0.39 *** 

Stability 

Tolerance 0.34 ** 

Breakdown 0.29 * 0.36 ** 0.34 ** 

HRS 

Mixograph 

Absorption 0.67 *** 0.47 *** 0.56 *** 0.53 *** 0.65 *** 0.30 * 0.64 *** 0.59 *** 0.27 * 

Mix time 0.59 *** 0.29 * 0.53 *** 0.36 ** 0.45 *** 0.58 *** 0.57 *** 0.35 ** 

Mix tolerance 0.55 *** 0.35 ** 0.47 *** 0.41 *** 0.39 *** 0.55 *** 0.49 *** 0.32 * 

Farinograph 

Absorption 0.29 * 0.38 *** 0.29 * 0.35 ** 

Development time 0.59 *** 0.33 ** 0.49 *** 0.37 ** 0.54 *** 0.59 *** 0.61 *** 0.31 * 

Stability 0.48 *** 0.38 *** 0.41 *** 0.46 *** 0.48 *** 0.48 *** 0.40 *** 

Tolerance 0.33 ** 0.39 *** 0.34 ** 0.27 * 0.29 ** 

Breakdown 0.62 *** 0.37 ** 0.52 *** 0.43 *** 0.60 *** 0.61 *** 0.61 *** 0.29 * 

*, **, and *** ¼ significant correlations at P < 0.01, P < 0.001, and P < 0.0001, respectively.

with starch addition. Also, mixing absorption is highly influenced 

by protein subclass, 50% 1-propanol insoluble polymeric protein 

content in flour (Park et al., 2006). HRW wheat showed similar 

trends, but with rather weak and insignificant relationships. 

Mixograph mixing time in HRS wheat showed negative correlations 

with B-granules, whereas in HRW wheat, a positive correlation 

was observed between mix time and differential volume at 

Point 1 (Table 4). Park et al. (2004) found that mix time did not 

show significant correlations to protein content with 49 HRW 

wheat flours due to contrasting effects among protein subclass on 

mix time. However, they suggested that mix time would be shorter 

as protein content increases because they also found that 50% 

1-propanol insoluble polymeric protein based on total protein was 

positively correlated with mix time and tended to decrease with 

increasing total protein content. Consequently, considering significant 

inverse correlations between protein content and parameters 

of B-granules, the relationship between mix time and parameters 

of B-granules was expected to be positive. The positive correlation 

(r ¼ 0.33**), however, was obtained only from HRW wheat. The 

inverse correlations between HRS wheat and parameters of 

B-granules were probably due in part to a positive correlation 

between protein content and mix time (r ¼ 0.44***, data not 

shown) in HRS wheat. In HRW wheat, there was no significant 

correlation between protein content and mix time. It has also been 

reported that B-granules require shorter optimum mix time 

compared with A-granules (D’Appolonia and Gilles, 1971; Petrofsky 

and Hoseney, 1995). Chiotelli and Le Meste (2002) also reported 

that B-granules had a higher affinity for water than the A-granules 

at room temperature, resulting in faster hydration. Therefore, the 

present work suggests that a higher proportion of B-granules in 

HRS wheat may partially account for the inverse relationship 

between mix time and parameters of B-granules. 

Farinograph absorption showed weak but significant inverse 

correlations to parameters of B-granules in HRS wheat, whereas in 

HRW wheat, diameter at Point 2 showed significant inverse 

correlation (Table 4). A similar reason, as given for Mixograph mix 

absorption, could explain these relationships. Farinograph 

absorption was positively correlated with wheat protein 

(r ¼ 0.71***, data not shown), and parameters of B-granules were 

inversely correlated to protein content, resulting in negative 

correlations between Farinograph absorption and parameters of 

B-granules. Soh et al. (2006) observed a significant increase in 

Farinograph absorption at constant protein content (17.4–17.7%) as 

% B-granules increased from 17% to 32.4%. Therefore, it appears that 

starch granule size distribution may affect Farinograph absorption 

when protein content and quality are invariable. Our data suggests 

that starch granule size distribution may not be an important 

variable for mix absorption when protein content varies. The 

protein content in the HRS wheat flour varied widely from 10.6% to 

17.8% (data not shown), and the protein composition (quality) was 

altered as the protein content increased (Park et al., 2006). 

The relationship between Farinograph development time and 

starch granule size distribution was similar to the relationship 

between Mixograph mix time and starch granule size distribution. 

The same explanation would be applicable for both relationships. 

Farinograph stability in HRS wheat also showed inverse 

correlation with parameters of B-granules. Several authors have 

reported the effects of starch on rheological properties. B- and nonwheat 

starch granules have been reported to result in large 

rheological differences using a constant vital gluten ratio in the 

reconstituted dough (Petrofsky and Hoseney, 1995). They suggested 

that there were strong interactions between vital gluten and nonwheat 

starches and wheat B-granules, resulting in less extensibility. 

Miller and Hoseney (1999) found that starch isolated from Kansas 

grown wheat gave significantly lower elastic modulus (G 0 ) and 

viscous modulus (G 00 ) than starches from other Kansas and strong 


Canadian wheats when dough was prepared from the same gluten 

source. Thus, it appears that starch from different cultivars and 

different size distributions makes a difference in determining the 

rheological properties of dough when a constant gluten source is 

used. The inverse relationships between Farinograph stability and 

parameters of B-granules, however, are hard to explain from 

previous reports (Miller and Hoseney, 1999; Petrofsky and Hoseney, 

1995) for two reasons. First, the present study used actual HRS 

wheat flours, and not blends of starches and constant gluten, as did 

previous researches, which had various components that could 

affect mixing properties. Second, there is limited research to 

connect rheological parameters of B-granules to Farinograph 

stability. It is generally recognized that rheological development in 

dough is triggered by the formation of a continuous gluten network 

(MacRitchie, 1992), with starch granules embedded in the network. 

In addition, previous studies reported that breakdown (loss of 

stability) during over-mixing was caused by ‘‘depolymerization’’ of 

the protein network (Danno and Hoseney,1982; Skerritt et al.,1999; 

Tanaka and Bushuk, 1973; Weegels et al., 1996). Therefore, with 

large variation in flour quality including protein content (10.6– 

17.8%, 14% moist base), Farinograph absorption (59.8–73.1%), 

development time (5.2–44.5 min), and loaf volume (803–1238 cm 3 ) 

(data not shown), inverse relationships between Farinograph 

stability and parameters of B-granules could be explained by the 

positive correlations between protein content and Farinograph 

stability (r ¼ 0.46***, data not shown). As previously discussed, 

protein content was inversely correlated to parameters of B-granules, 

possibly giving the inverse correlation between Farinograph 

stability and parameters of B-granules. Farinograph breakdown 

showed a similar relationship with stability, and a similar explanation 

could be applicable. Protein content and breakdown were 

highly positively correlated (r ¼ 0.79***, data not shown). The HRW 

wheat showed a similar trend in these relationships, but was 

weaker with many non-significant correlations. 

3.4. Relationships with breadmaking parameters 

The breadmaking parameters showed significant correlations to 

starch granule size distribution (Table 5). A greater number of the 

parameters of HRS wheat, again, were significantly correlated. The 

trends in relationships were similar for HRW and HRS wheat, so our 

discussion will focus on HRS wheat. 

Baking absorption and mix time were positively correlated to 

Mixograph absorption (r ¼ 0.85***), mix time (r ¼ 0.93***), Farinograph 

absorption (r ¼ 0.73***) and development time (r ¼ 0.77***) 

(data not shown). 

Loaf volume was inversely correlated with parameters of 

B-granules. Differential volume at Point 1 showed the greatest 

inverse correlation (r ¼ 0.68***), followed by area range A 

(r ¼ 0.66***), A þ B (r ¼ 0.62***) and diameter at Point 1 

(r ¼ 0.52***). Park et al. (2005) summarized previous studies 

using four different concepts concerning the effects of small starch 

granule size on breadmaking properties: ‘‘beneficial’’ (Hayman 

et al., 1998; Sahlstrom et al., 1998; Van Vliet et al., 1992); ‘‘detrimental’’(D’Appolonia 

and Gilles, 1971; Kulp, 1973); ‘‘little effect’’ 

(Hoseney et al., 1971); and ‘‘optimum ratio’’ of A- and B-granules 

(Lelievre et al., 1987; Park et al., 2004; Soulaka and Morrison, 

1985b). More recently, Park et al. (2005) confirmed their previous 

observation that there seemed to be an optimum weight ratio of Aand 

B-granules for crumb grain score, and there was no benefit of 

small starch granules to loaf volume and crumb grain score when 

used with a constant source of gluten. Considering previous 

controversial results, it is difficult to state that B-granules affect loaf 

volume negatively. It is more appropriate to admit that the relationships 

were obtained due to a positive correlation between 

protein content and loaf volume (r ¼ 0.91***, data not shown), and

104 

Table 5 

Correlations between starch granule size distribution and straight-dough breadmaking parameters 



Breadmaking 

Absorption 

Mix time 

0.35 ** 0.28 * 0.26 * 

Proof height 0.46 *** 0.38 *** 0.31 * 0.45 *** 0.51 *** 0.26 * 0.28 * 0.27 * 

Crumb grain score 0.35 ** 

Loaf volume 0.35 ** 0.43 *** 0.28 * 0.42 *** 0.40 *** 

HRS 

Breadmaking 

Absorption 0.57 *** 0.30 * 0.52 *** 0.40 *** 0.57 *** 0.28 * 0.56 *** 0.58 *** 0.37 ** 

Mix time 0.52 *** 0.46 *** 0.34 ** 0.43 *** 0.52 *** 0.53 *** 0.36 ** 

Proof height 0.56 *** 0.34 ** 0.51 *** 0.37 ** 0.47 *** 0.54 *** 0.52 *** 

Crumb grain score 0.28 * 0.27 * 0.41 *** 0.27 * 

Loaf volume 0.68 *** 0.45 *** 0.60 *** 0.52 *** 0.67 *** 0.28 * 0.66 *** 0.62 *** 0.28 * 

*, **, and *** ¼ significant correlations at P < 0.01, P < 0.001, and P < 0.0001, respectively. 

All correlations are significant in this table at P ¼ 0.01 (>r ¼j 0.261j), P ¼ 0.001 (>r ¼j 0.326j), and P ¼ 0.0001 (>r ¼j 0.374j). 

protein content was inversely correlated with parameters of 

B-granules. 

Crumb grain score showed a generally weak but significant 

inverse linear correlation to differential volume at Point 1 

(r ¼ 0.28*), diameter at Point 1 (r ¼ 0.41***), and area range of A 

(r ¼ 0.27*). These inverse relationships agreed with results from 

Park et al. (2005) where the authors found the lowest value of 

crumb grain score and fineness from the bread baked with 100% 

B-granules. In this study, crumb grain score was not significantly 

correlated to protein, consequently relationships were independent 

of protein content. It should be pointed out that there were no 

other significant correlations to crumb grain score from 68 wheat 

and flour quality parameters, except wheat kernel weight and size 

(r ¼ 0.36***, respectively, data not shown), which were positively 

correlated to parameters of B-granules. So, even though those 

correlation values were low, we believe it has significance. 

3.5. Polynomial relationships between B-granules 

and crumb grain score 

Several authors reported that there seemed to be an optimum 

weight % ratio of A- and B-granules for breadmaking (Lelievre et al., 

1987; Park et al., 2004, 2005; Soulaka and Morrison, 1985b) and 

spaghetti production (Soh et al., 2006). A polynomial relationship 

was applied between B-granules and crumb grain score, resulting 

in improved correlations for both HRW and HRS wheats (Table 6). 

Correlation values of HRS wheat increased from 0.073* to 0.154*** 

for area A, and 0.013 to 0.088* for area A þ B. In addition, we found 

that the linear relationships improved when the ratio of volume % 

of B-granules to flour protein content was used. The correlation 

Table 6 

Linear and polynomial correlations of B-type granule volume % and ratio with 

protein contents to crumb grain score 

Crumb grain score vs. Linear 

Polynomial 

correlation 

correlation 

R 2 

R 2 

HRW Area A 0.004 0.018 

Area A þ B 0.005 0.039 

Area A/flour protein 0.130 ** 0.141 *** 

Area A þ B/flour protein 0.136 ** 0.158 *** 

HRS Area A 0.073 * 0.154 *** 

Area A þ B 0.013 0.088 * 

Area A/flour protein 0.089 * 0.222 *** 

Area A þ B/flour protein 0.049 0.164 *** 

*, **, and *** ¼ significant correlations at P < 0.01, P < 0.001, and P < 0.0001, 

respectively. 


improvements were more obvious for HRW wheat, which 

improved from 0.004 to 0.130** for area range A, and from 0.005 to 

0.136** for area A þ B. The linear correlations between volume % of 

B-granules and crumb grain score were improved from 0.004 to 

0.141*** and from 0.005 to 0.158*** for HRW wheat, and from 

0.073* to 0.222*** and from 0.013 to 0.164*** for HRS wheat after 

obtaining the ratios with protein content and applying polynomial 

correlation. Therefore, it appears that there is an optimum range of 

volume % B-granules, and the optimum range could vary depending 

on protein content. Lelievre et al. (1987) found different optimum 

starch size fractions for different protein concentrations when used 

for breadmaking. Park et al. (2005) suggested that high water 

absorption during baking and/or high surface area of B-granules 

could be responsible for gas cell stabilization and the resultant 

crumb grain score. It has been shown that protein–starch interactions 

produce different rheological properties depending on variety 

and granule size (Miller and Hoseney, 1999; Petrofsky and Hoseney, 

1995). Also, bulk rheological properties of dough could affect 

overall gas cell stability (Van Vliet et al., 1992). Consequently, the 

reason for different optimum weight % of B-starch granules for 

different protein contents may be due to rheological properties 

imparted by protein–starch interactions. The higher water 

absorbing properties of B-granules could pull water from the 

attached protein matrix and liquid film in the dough system during 

baking. Gan et al. (1995) proposed that gas cells are stabilized by 

a continuous liquid film on the protein–starch matrix. Depending 

on how extreme the situations are, e.g. low protein content with 

high content of B-granules (too stiff) vs. high protein content with 

low content of B-granules (too viscous), the overall viscoelastic 

properties and gas cell stability could be changed, resulting in 

different crumb structures. 

4. Conclusion 

Starch granule size distributions of HRW and HRS wheats 

showed significant differences in differential volume, diameter, and 

volume distribution. HRW has smaller size and proportion of 

B-granules than HRS wheat. The reason is not clear in this study, 

but an explanation could be due to hot and dry growing conditions 

during grain filling. Parameters of B-granules showed many 

significant correlations with wheat and flour properties, partly due 

to the inverse correlation between protein content and parameters 

of B-granules. There appears to be different optimum ranges of Bgranule 

weight % for flours with different protein contents. It seems 

that starch granule size distribution is a unique property that

affects physicochemical properties of wheat, flour, and breadmaking 

properties in conjunction with its counterpart, protein. 

References 

AACC, 2000. Approved Methods of the American Association of Cereal Chemists, 

10th ed. AACC, St. Paul, MN, USA. 



grain development in a U.S. spring wheat. Journal of Cereal Science 37, 9–20. 

Bechtel, D.B., Wilson, J.D., 2000. Variability in a starch isolation method and automated 

digital image analysis system used for the study of starch size distributions 

in wheat flour. Cereal Chemistry 77, 401–405. 

Bechtel, D.B., Zayas, I., Kaleikau, L., Pomeranz, Y., 1990. Size-distribution of wheat 

starch granules during endosperm development. Cereal Chemistry 67, 59–63. 

Bhullar, S.S., Jenner, C.F., 1985. Differential responses to high temperatures of starch 

and nitrogen accumulation in the grain of four cultivars of wheat. Australian 


Briarty, L.G., Hughes, C.E., Evers, A.D., 1979. The developing endosperm of wheat – 

a stereological analysis. Annals of Botany 44, 641–658. 

Brocklehurst, P.A., Evers, A.D., 1977. The size distribution of starch granules in 

endosperm of different sized kernels of the wheat cultivar Maris Huntsman. 

Journal of the Science of Food and Agriculture 28, 1084–1089. 

Chiotelli, E., Le Meste, M., 2002. Effect of small and large wheat starch granules on 

thermomechanical behavior of starch. Cereal Chemistry 79, 286–293. 

Danno, G., Hoseney, R.C., 1982. Effects of dough mixing and rheologically active 

compounds on relative viscosity of wheat proteins. Cereal Chemistry 59,196–198. 

D’Appolonia, B.L., Gilles, K.A., 1971. Effect of various starches in baking. Cereal 

Chemistry 48, 625–636. 

Dengate, H., Meredith, P., 1984. Variation in size distribution of starch granules from 

wheat grain. Journal of Cereal Science 3, 83–90. 

Dengate, H.N., Baruch, D.W., McKenzie, J., Simmons, L.D., Meredith, P.W., 

Morrison, R., 1979. Fractionation of wheat starch on the basis of hydrated 

density. Starch 31, 40–44. 

Dronzek, B.L., Hwang, P., Bushuk, W., 1972. Scanning electron microscopy of starch 

from sprouted wheat. Cereal Chemistry 49, 232–239. 

Eliasson, A.C., 1989. Characterization of wheat starch and gluten as related to enduse 

properties. In: Pomeranz, Y. (Ed.), Wheat is Unique. AACC International, St. 

Paul, MN, pp. 177–192. 

Evers, A.D., 1973. The size distribution among starch granules in wheat endosperm. 

Starch 25, 303–304. 

Evers, A.D., Lindley, J., 1977. The particle-size distribution in wheat endosperm 

starch. Journal of Science and Food Agriculture 28, 98–102. 

Gan, Z., Ellis, P.R., Schofield, J.D., 1995. Mini review. Gas cell stabilization and gas 

retention in wheat bread dough. Journal of Cereal Science 21, 215–230. 

Gibson, L.R., McCluskey, P.J., Tilley, K.A., Paulsen, G.M., 1998. Quality of hard red 

winter wheat grown under high temperature conditions during maturation and 

ripening. Cereal Chemistry 75, 421–427. 

Gibson, L.R., Paulsen, G.M., 1999. Yield components of wheat grown under high 

temperature stress during reproductive growth. Crop Science 39, 1841–1846. 

Haraszi, R., Gras, P.W., Tömösközi, S., Salgó, A., Békés, F., 2004. Application of a micro 

Z-arm mixer to characterize mixing properties and water absorption of wheat 

flour. Cereal Chemistry 81, 555–560. 

Hayman, D., Sipes, K., Hoseney, R.C., Faubion, J.M., 1998. Factors controlling gas cell 

failure in bread dough. Cereal Chemistry 75, 585–589. 

Hoseney, R.C., Finney, K.F., Pomeranz, Y., Shogren, M.D., 1971. Functional (breadmaking) 

and biochemical properties of wheat flour components. VII. Starch. 


Hughes, C.E., Briarty, L.G., 1976. Stereological analysis of the contribution to mature 

wheat endosperm starch by large and small granules. Starch 28, 336–337. 

Karlsson, R., Olered, R., Eliasson, A.-C., 1983. Changes in starch granule size distribution 

and starch gelatinization properties during development and maturation 

of wheat, barley and rye. Starch 35, 335–340. 

Kulp, K., 1973. Characteristics of small-granule starch of flour and wheat. Cereal 

Chemistry 50, 666–679. 

Lelievre, J., Lorenz, K., Meredith, P., Baruch, D.W., 1987. Effects of starch particle size 

and protein concentration on breadmaking performance. Starch 39, 347–352. 

Lineback, D.R., 1984. The starch granule organization and properties. Bakers Digest 

58 (2), 16. 18–21. 

MacLeod, L.C., Duffus, C.M., 1988. Temperature effects on starch granules in 

developing barley grain. Journal of Cereal Science 8, 29–37. 

MacRitchie, F., 1992. Physicochemical properties of wheat proteins in relation to 

functionality. Advances in Food Research 36, 1–85. 

Maghirang, E.B., Lookhart, G.L., Bean, S.R., Pierce, R.O., Xie, F., Caley, M.S., 

Wilson, J.D., Seabourn, B.W., Ram, M.S., Park, S.H., Chung, O.K., Dowell, F.E., 

2006. Comparison of quality characteristics and breadmaking functionality of 

hard red winter and hard red spring wheat. Cereal Chemistry 83, 520–528. 

Meredith, P., 1981. Large and small starch granules in wheat – are they really 

different? Starch 33, 40–44. 

Miller, K.A., Hoseney, R.C., 1999. Dynamic rheological properties of wheat starch– 

gluten doughs. Cereal Chemistry 76, 105–109. 

Morrison, W.R., Gadan, H., 1987. The amylose and lipid contents of starch granule in 

developing wheat endosperm. Journal of Cereal Science 5, 263–275. 


Morrison, W.R., Scott, D.C., 1986. Measurements of the dimensions of wheat starch 

granule populations using a Coulter Counter with 100-channel analyzer. Journal 

of Cereal Science 4, 13–21. 

Nikuni, Z., 1978. Studies on starch granules. Starch 30, 105–111. 

Ohm, J.B., Chung, O.K., 1999. Gluten, pasting, and mixograph parameters of hard 

winter wheat flours in relation to breadmaking. Cereal Chemistry 76, 

606–613. 

Ohm, J.B., Chung, O.K., Deyoe, C.W., 1998. Single kernel characteristics of hard 

winter wheats in relation to milling and baking quality. Cereal Chemistry 75, 

156–161. 

Park, S.H., Chung, O.K., Seib, P.A., 2004. Size distribution and properties of wheat 

starch granules in relation to crumb grain score of pup-loaf bread. Cereal 

Chemistry 81, 699–704. 

Park, S.H., Chung, O.K., Seib, P.A., 2005. Effects of varying weight ratios of large and 

small wheat starch granules on experimental straight-dough bread. Cereal 

Chemistry 82, 166–172. 

Park, S.H., Chung, O.K., Seib, P.A., 2006. Levels of protein and protein composition in 

hard winter wheat flours and the relationship to breadmaking. Cereal Chemistry 

83, 418–423. 

Parker, M.L., 1985. The relationship between A-type and B-type starch granules in 

the developing endosperm of wheat. Journal of Cereal Science 3, 271–278. 

Peng, M., Gao, M., Abdel-Aal, E.-S.M., Hucl, P., Chibbar, R.N., 1999. Separation and 

characterization of A- and B-type starch granules in wheat endosperm. Cereal 

Chemistry 76, 375–379. 

Petrofsky, K.E., Hoseney, R.C., 1995. Rheological properties of dough made with 

starch and gluten from several sources. Cereal Chemistry 72, 53–58. 

Pomeranz, Y., 1988. Chemical composition of kernel structures. In: Pomeranz, Y. 

(Ed.), Wheat Chemistry and Technology, vol. 1. AACC International, St. Paul, MN, 

pp. 97–158. 

Raeker, M.O., Gaines, C.S., Finney, P.L., Donelson, T., 1998. Granule size distribution 

and chemical composition of starches from 12 soft wheat cultivars. Cereal 

Chemistry 75, 721–728. 

Sandstedt, R.M., 1961. The function of starch in the baking of bread. Bakers Digest 35 

(3), 36–42. 44. 

Sahlstrom, S., Brathen, E., Lea, P., Autio, K., 1998. Influence of starch granule size 

distribution on bread characteristics. Journal of Cereal Science 28, 157–164. 

Seib, P.A., 1994. Wheat starch: isolation, structure, and properties. Oyo Toshitsu 

Kagaku 41, 49–69. 

Shi, Y.C., Seib, P.A., Bernardin, J.E., 1994. Effects of temperature during grain-filling 

on starches from six wheat cultivars. Cereal Chemistry 71, 369–383. 

Skerritt, J.H., Hac, L., Lindsay, M.P., Bekes, F., 1999. Depolymerization of the gluten 

macropolymer during mixing: II. Differences in retention of specific glutenin 

subunits. Cereal Chemistry 76, 402–409. 

Simmonds, D.H., O’Brien, T.P., 1981. Morphological and biochemical development of 

wheat endosperm. Advances in Cereal Science and Technology 4, 5–14. 

Soh, H.N., Sissons, M.J., Turner, M.A., 2006. Effects of starch granule size distribution 

and elevated amylose content on durum dough rheology and spaghetti cooking 

quality. Cereal Chemistry 83, 513–519. 

Sollars, W.F., Rubenthaler, G.L., 1971. Performance of wheat and other starches in 

reconstituted flours. Cereal Chemistry 48, 397–410. 

Soulaka, A.B., Morrison, W.R., 1985a. The amylose and lipid contents, dimensions, 

and gelatinization characteristics of some wheat starches and their A- and Bgranule 

fractions. Journal of the Science of Food and Agriculture 36, 709–718. 

Soulaka, A.B., Morrison, W.R., 1985b. The bread baking quality of six wheat starches 

differing in composition and physical properties. Journal of the Science of Food 

and Agriculture 36, 719–727. 

Stoddard, F.L., 1999. Survey of starch particle-size distribution in wheat and related 

species. Cereal Chemistry 76, 145–149. 

Stone, P.J., Nicolas, M.E., 1994. Wheat cultivars vary widely in their responses of 

grain yield and quality to short periods of post-anthesis heat stress. Australian 


Tanaka, K., Bushuk, W., 1973. Changes in flour proteins during dough mixing. II. Gel 

filtration and electrophoresis results. Cereal Chemistry 50, 597–605. 

Tashiro, T., Wardlaw, I.F., 1990. The effects of high temperature at different stages of 

ripening on grain set, grain weight, and grain dimension in the semi-dwarf 

wheat ‘Banks’. Annals of Botany 65, 51–61. 

Tester, R.F., South, J.B., Morrison, W.R., Ellis, R.P., 1991. The effects of ambient 

temperature during the grain-filling period on the composition and properties 

of starch from four barley genotypes. Journal of Cereal Science 13, 113–127. 

U.S. Wheat Associate, 2002. Harvest Report – October 8, 2002 – final. http://www. 

uswheat.org/harvestReports/doc/3DCEBA32BAAE4ACF85256F8100572BE1?Open 

Document# (accessed April 24, 2008). 

Van Vliet, T., Janssen, A.M., Bloksma, A.H., Walstra, P., 1992. Strain hardening of 

dough as a requirement for gas retention. Journal of Texture Studies 23, 

439–460. 

Weegels, P.L., van de Pijpekamp, A.M., Graveland, A., Hamer, R.J., Schofield, J.D., 

1996. Depolymerisation and repolymerisation of wheat glutenin during dough 

processing. I. Relationships between glutenin macropolymer content and 

quality parameters. Journal of Cereal Science 23, 103–111. 

Wilson, J.D., Bechtel, T.C., Seib, P.A., 2006. Measurement of wheat starch granule size 

distribution using image analysis and laser diffraction technology. Cereal 

Chemistry 83, 259–268. 

Wong, R.B.K., Lelievre, J., 1982. Comparison of the crystallinities of wheat starches 

with different swelling capacities. Starch 34, 159–161.

Total phenolics, flavonoids, antioxidant capacity in rice grain and 

their relations to grain color, size and weight 

Yun Shen, Liang Jin, Peng Xiao, Yan Lu, Jinsong Bao * 

Institute of Nuclear Agricultural Sciences, Key Laboratory of Chinese Ministry of Agriculture and Zhejiang Province for Nuclear-Agricultural Sciences, College of Agriculture and 

Biotechnology, Zhejiang University, Hua Jiachi Campus, Hangzhou 310029, People’s Republic of China 

article info 


Received 2 March 2008 



Keywords: 

Antioxidant capacity 

Flavonoid 

Phenolics 

Rice 


abstract 

Rice is a staple food being consumed by nearly half of the world 

population. Nutritional quality of rice has received more attention 

in the developing countries, where monotonous consumption of 

rice may lead to deficiencies of essential minerals, vitamins, and 

other nutritional compositions (Bouis et al., 2003). This is not 

caused by nutritional deficiency in rice grain itself, but due to it 

being traditionally eaten in the form of the milled white kernel. 

Milling of the brown rice to obtain milled rice removes bran layers 

Abbreviations: ABTS, 2,2-azino-bis-(3-ehylbenzothiazoline-6-sulphonic acid) 

diammonium salt; GAE, gallic acid equivalent; RE, rutin equivalent; TEAC, trolox 

equivalent antioxidant capacity. 


E-mail address: jsbao@zju.edu.cn (J. Bao). 


doi:10.1016/j.jcs.2008.07.010 





Total phenolics, flavonoid contents and antioxidant capacity from a wide collection of rice germplasm 

were measured, and their relations to grain color, grain size and 100-grain weight were investigated. 

Highly significant genotypic differences were observed in total phenolics, flavonoid contents and 2,2azino-bis-(3-ehylbenzothiazoline-6-sulphonic 

acid) diammonium salt (ABTS) radical cation antioxidant 

capacity. They displayed an increasing order in the white rice, red rice and black rice, yet several white 

rice had higher phenolics and flavonoids contents than the red rice. Significant positive pair-wise 

correlations were found among the phenolics, flavonoid contents and antioxidant capacity, and the 

coefficient between the phenolic contents and antioxidant capacity was extremely high (r ¼ 0.96). 

Among all rice accessions, the grain color parameters had negative correlations with the phenolics, 

flavonoid contents and antioxidant capacity (p < 0.001). The negative correlation between a* and antioxidant 

capacity, and the positive correlation between H and antioxidant capacity were consistent 

within the respective white rice and red rice groups. Flavonoid contents had positive correlation with 

grain length and length to width ratio, and had negative correlation with the 100-grain weight among all 

rice accessions. It was also found that 100-grain weight still had negative correlations with phenolics, 

flavonoid contents and antioxidant capacity within the white rice genotypes. These relationships may 

serve as indexes to indirectly select breeding lines high in the phenolics, flavonoids and antioxidant 

capacity. Principal component analysis including the information for phenolics, flavonoids, antioxidant 

capacity, grain color parameters, grain size and 100-grain weight extracted five principal components 

that explained 83.7% of the total variances. The results of this study may provide new opportunities for 

rice breeders and eventually commercial rice growers to promote the production of rice with enhanced 

nutritional quality. 


that are rich in protein, fiber, oil, minerals, vitamins, and other 

phytochemicals (Orthoefer and Eastman, 2004; Yokoyama, 2004), 

leading to loss of most of the nutritional components of the rice 

grain. Numerous studies have shown that the essential phytochemicals 

in fruits, vegetables and cereal grains, including rice, are 

significantly associated with reduced risk of developing chronic 

diseases such as cardiovascular disease, type 2 diabetes, and some 

cancers (Liu, 2007; Yawadio et al., 2007; Yokoyama, 2004). 

Biofortification of rice to improve nutritional quality and to 

combat nutritional deficiency by a transgenic engineering 

approach has several successful examples (Bouis et al., 2003). Ye 

et al. (2000) introduced the b-carotene synthesis pathway to 

rice endosperm by genetic engineering to obtain the golden rice 

that produced 1.6 mg/g of b-carotene in the grain. Storozhenko 

et al. (2007) reported biofortification of folate content in rice 

grain by over expression of two Arabidopsis genes encoding GTP 

cyclohydrolase I and aminodeoxychorismate synthase under the

control of strong endosperm-specific promoters. The transgenic 

rice grain contained up to 100-fold higher folate levels 

compared to the wild type. It should be noted that conventional 

breeding is still possible to improve the nutritional components 

in grains. For example, Harjes et al. (2008) reported that variations 

at the lycopene epsilon cyclase (lcyE) locus alter flux down 

a-carotene versus b-carotene branches of the carotenoid pathways 

in the maize grain, and four lcyE natural polymorphism 

explained 58% of the variation in these two branches and 

a threefold difference in provitamin A compounds. Thus, selection 

of favorable lcyE alleles with inexpensive molecular markers 

enables developing-country breeders to more effectively 

improve provitamin A levels. However, effort towards biofortification 

of rice grain to improve nutritional quality by 

conventional breeding has been scarcely reported (Bouis et al., 

2003). As a primary step to achieve this goal, it is necessary to 

investigate the genotypic diversity in the phytochemicals among 

diverse rice accessions, so as to find a way to enrich these 

compositions by breeding. 

The genotypic diversity of some phytochemicals in rice bran 

layers has been widely characterized (Dykes and Rooney, 2007; Liu, 

2007). For example, Bergman and Xu (2003) reported genotypic 

and environmental effects on tocopherol, tocotrienol, and g-oryzanol 

contents of rice, and Miller and Engel (2006) also reported the 

contents of g-oryzanol in brown rice. Jiang et al. (2007) reported 

minerals contents and their correlation with other quality traits of 

rice. However, some phytochemicals, including phenolics and 

flavonoids, have not received as much attention as other compositions 

in rice grains and the phytochemicals in other cereals, fruits 

and vegetables (Liu, 2007). 

Phenolics are compounds possessing one or more aromatic 

rings with one or more hydroxyl groups (Liu, 2007). Phenolic 

compounds in diet may provide health benefits associated with 

reduced risk of chronic disease (Liu, 2007). Chinese medicinal 

plants have high levels of phenolics and potent antioxidant 

capacity, which might contribute to the protective effects against 

cancer (Cai et al., 2004). In rice, Goffman and Bergman (2004) 

studied the genotypic and environmental effects of the kernel 

phenolic content, and found that bran color was highly statistically 

significant for bran phenolic contents. Flavonoids are one 

group of phenolics, which consists of two aromatic rings linked 

by 3 carbons that are usually in an oxygenated heterocycle ring 

(Liu, 2004). Anthocyanins are a group of reddish to purple watersoluble 

flavonoids that are the primary pigments in the red and 

black grains, and have been widely identified and characterized 

in cereal grains (Abdel-Aal et al., 2006). The major components 

of anthocyanidins in colored rice are cyaniding-3-O-b-glucoside 

and peonidin-3-O-b-glucoside (Abdel-Aal et al., 2006; Yawadio 

et al., 2007). There have been few reports on characterization of 

other flavonoids such as flavonols, flavones, flavanols, and 

flavanones. 

The phenolic compounds are also known as antioxidants 

(Abdel-Aal et al., 2006; Adom and Liu, 2002; Hu et al., 2003). 

Antioxidants have long been recognized to have protective functions 

against oxidative damage, and are associated with reduced 

risk of chronic diseases (Adom and Liu, 2002; Liu, 2007). Other 

phytochemicals such as carotenoids, tocols and g-oryzanols are also 

antioxidants (Aguilar-Garcia et al., 2007; Choi et al., 2007; Xu et al., 

2001). 

The objective of this study was to evaluate total phenolics, 

flavonoids and antioxidant capacity of a large number of rice 

genotypes (481 accessions) and to analyze their relationships with 

grain color, size and 100-grain weight. The results of this study 

could provide rice breeders and eventually commercial rice 

growers new opportunities to promote the production of rice with 

enhanced levels of the bioactive compounds. 

Y. Shen et al. / Journal of Cereal Science 49 (2009) 106–111 107 



A total of 481 rice accessions including 423 white rice, 52 red 

rice and 6 black rice, were employed in this study. All the rice was 

grown in the Zhejiang University farm in 2006. They were sown in 

late May, transplanted on June 20, and harvested in October, and 

the field management followed conventional practices. Rice grains 

were air-dried and stored at room temperature for three months. 

Then they were dehusked on a Satake Rice Machine (Satake Co., 

Japan), and ground to pass through a 100-mesh sieve on a Cyclone 

Sample Mill (UDY Corporation, Fort Collins, Colorado, USA). 2,2- 

Azino-bis(3-ethylbenzothiazoline-6-sulfonic acid) diammonium 

salt (ABTS), potassium persulfate and gallic acid were purchased 

from BBI (Ontario, Canada), 6-hydroxy-2,5,7,8-tetramethylchromane-2-carboxylic 

acid (Trolox) from Sigma/Aldrich (St. Louis, 

MO), Folin–Ciocalteu reagent from Fluka Chemie AG (Buchs, Switzerland), 

and rutin from Oukang Company (Chengdu, China). 

2.2. Extraction 

Wholemeal flours (1 g) of each accession were extracted with 

25 mL of methanol containing 1% HCl for 24 h at 24 C. The 

procedure was repeated twice. The methanolic extracts were 

centrifuged at w4000g for 15 min and the supernatants were 

pooled and stored at 4 C. 

2.3. Color of rice grain 

The color of rice grain sample was measured with a TC-PIIG 

automatic color difference meter (Beijing Optical Instrument 

Factory, Beijing, China). Color measurements were expressed as 

tristimulus parameters, L*, a*, and b*. L* indicates lightness 

(100 ¼ white and 0 ¼ black). a* indicates redness–greenness and b* 

indicates yellowness–blueness. In addition, the chroma (C) value 

indicates color intensity or saturation, calculated as 

C ¼ða* 2 þ b* 2 Þ 1=2 , and Hue angle was calculated as H ¼ tan 1 (b*/ 

a*) (Bao et al., 2005). 

2.4. Total phenolics 

Total phenolic content was assayed by the Folin–Ciocalteu 

colorimetric method with slight modification (Bao et al., 2005; Cai 

Table 1 

Variations in phenolics, flavonoids contents and antioxidant capacity among white 

(n ¼ 423), red (n ¼ 52) and black (n ¼ 6) rice genotypes 

Phenolics a 

Flavonoids a 

Antioxidant capacity a 

Total rice 

Mean SD 197.5 144.8 134.7 19.8 0.413 0.696 

CV (%) 73.3 14.7 168.63 

Range 108.1–1244.9 88.6–286.3 0.012–5.533 

White rice 

Mean SD 151.8 19.5 131.6 14.2 0.196 0.073 

CV (%) 12.9 10.8 37.33 

Range 108.1–251.4 88.6–170.7 0.012–0.413 

Red rice 

Mean SD 470.1 107.2 147.2 18.0 1.705 0.600 

CV (%) 22.8 12.3 35.22 

Range 165.8–731.8 108.7–190.3 0.291–2.963 

Black rice 

Mean SD 1055.7 176.2 240.6 38.1 4.484 1.095 

CV (%) 16.7 15.8 24.41 

Range 841.0–1244.9 187.6–286.3 2.527–5.533 

a Phenolics content was expressed as mg GAE/100 g, flavonoids content was 

expressed as mg RE/100 g, and antioxidant capacity was expressed as mM TAEC.

108 

No. of accessions 



250 

200 

150 

100 

50 

0 

300 

250 

200 

150 

100 

50 

0 

450 

400 

350 

300 

250 

200 

150 

100 

50 

0 

600 

Phenolics content (mg GAE / 100g) 

180 

Flavonoids content (mg RE / 100g) 

3.0 

Antioxidant capacity (mM TAEC) 

black 

red 

white 

black 

red 

white 

black 

red 

white 

Fig. 1. Mean distributions of phenolics, flavonoids contents and antioxidant capacity 

among white rice (423), red rice (52) and black rice (6) accessions. 

et al., 2004). Briefly, aliquots (1.0 mL) of appropriately diluted 

extracts or standard solutions were mixed with 0.5 mL 0.5 N 

Folin–Ciocalteu reagent, then the reaction was neutralized with 

saturated sodium carbonate (75 g/L). The absorbance of the 

resulting blue color was recorded using a spectrophotometer after 

incubation for 2 h at 23 C. A calibration curve was prepared using 

gallic acid solution. Total phenolics contents were expressed as 

milligrams of gallic acid equivalent (mg GAE) per 100 g of dry 

weight. 

Y. Shen et al. / Journal of Cereal Science 49 (2009) 106–111 

2.5. Total flavonoids 

Total flavonoid content was determined by a colorimetric 

method (Bao et al., 2005) with minor modification. Aliquots 

(0.5 mL) of appropriately diluted extracts or standard solutions 

were pipetted into 15-mL polypropylene conical tubes containing 

2 mL double distilled H2O and mixed with 0.15 mL 5% NaNO2. After 

5 min, 0.15 mL 10% AlCl3$6H2O solution was added, and the mixture 

was allowed to stand for another 5 min, and then 1 mL 1 M NaOH 

was added. The reaction solution was well mixed, kept for 15 min, 

and the absorbance was determined at 415 nm. Total flavonoid 

content was calculated using the standard rutin curve, and 

expressed as mg rutin equivalent (mg RE) per 100 g of dry weight. 

2.6. Radical cation ABTS þ scavenging activity 

Total antioxidant capacity of rice extracts was carried out 

using a spectrophotometer by the improved 2,2-azino-bis- 

(3-ehylbenzothiazoline-6-sulphonic acid) diammonium salt 

(ABTS) radical cation method as described (Bao et al., 2005; Cai 

et al., 2004). ABTS þ solution (3.9 mL, absorbance of 0.700) was 

added to 0.1 mL of the extracts and mixed thoroughly. The 

reaction mixture was kept at room temperature for 6 min and the 

absorbance was immediately recorded at 734 nm. Trolox standard 

solution in 80% ethanol was prepared and assayed under the 

same conditions. Results were expressed in terms of Trolox 

equivalent antioxidant capacity (TEAC, mM Trolox equivalents per 

100 g dry weight). 


All the analyses were carried out at least in duplicate and in 

randomized order with mean values being reported. Analysis of 

variance (ANOVA), correlation analysis and principal component 

analysis of the results were performed in SAS (Software Version 9.1. 

SAS Institute Inc., Cary, NC). 

3. Results 

3.1. Total phenolics, flavonoid contents and antioxidant capacity 

There were wide range of variations in the total phenolics in rice 

grain (Table 1, Fig. 1). Among all the rice accessions, total phenolic 

content ranged from 108.1 to 1244.9 mg GAE/100 g, with the lower 

values coming from the white rice, while the higher values were 

from red and black rice (Fig. 1). Variations were still found within 

the white rice and red rice, ranging from 108 to 251 mg GAE/100 g 

and from 165.8 to 731.8 mg GAE/100 g for white and red rice, 

respectively (Table 1). However, several red rice accessions still had 

lower total phenolic contents than the white rice (Fig. 1). 

Flavonoid contents in all the rice ranged from 88.6 to 

286.3 mg RE/100 g. The mean flavonoid contents among the white, 

red and black rice were 131.6, 147.2 and 240.6 mg RE/100 g, 

respectively. Even though red rice had average higher levels of 

flavonoids than the white rice, some red rice accessions still had 

Table 2 

Pair-wise correlations among phenolics, flavonoids contents and antioxidant 

capacity (ABTS) among the total rice, white and red rice genotypes 

Total rice White rice Red rice 

Flavonoids ABTS Flavonoids ABTS Flavonoids ABTS 

Phenolics 0.681*** 0.962*** 0.703*** 0.231*** 0.461*** 0.777*** 

Flavonoids 0.612*** 0.101* 0.342* 

* , ** and *** were significant at 0.05, 0.01 and 0.001 probability level, respectively.

70 

60 

50 

40 

30 

20 

10 

0 

-10 

-20 

-30 

white rice red rice black rice 

Fig. 2. Mean color parameters of white, red and black rice. 

lower contents than white rice, whereas most of red rice had lower 

contents than the black rice (Fig. 1). 

The total antioxidant capacity was measured using the ABTS 

assay. It was varied to a great extent, averaged 0.413 mM TEAC, 

ranging from 0.012 to 5.533 mM TEAC among the total rice 

accessions (Table 1, Fig. 1). Among the white rice, the mean 

ABTS was 0.196 mM TEAC, ranging from 0.012 to 0.413 mM TEAC, 

whereas among the red rice, it averaged 1.705 mM TEAC, ranging 

from 0.291 to 2.963 mM TEAC. The six black rice samples had 

average antioxidant capacity of 4.484 mM TEAC, around three 

times of that of the red rice (Table 1). 

3.2. Correlations among total phenolics, flavonoids and 

antioxidant capacity 

Pair-wise correlations between the three parameters were 

positive (P < 0.001) among all rice groups, which were still true 

within the respective white rice and red rice accessions (Table 2). 

The phenolic contents were highly positively correlated with the 

antioxidant capacity (r ¼ 0.962) among all the rice. However, the 

correlation coefficient within the white rice accessions (r ¼ 0.231) 

was much smaller than that within the red rice accessions 

(r ¼ 0.777). The correlation coefficient between phenolic contents 

and flavonoid contents was higher within white rice groups 

(r ¼ 0.703) than that within the red rice accessions (r ¼ 0.461). The 

correlation coefficient between flavonoid content and antioxidant 

capacity was much smaller within the white and red rice accessions 

(Table 2) than that within the total rice groups (r ¼ 0.612), which 

might be due to the smaller variation in each group when 

compared to the total rice accessions (Table 2). 

3.3. Relationships between phenolics, flavonoids, antioxidant 

capacity and grain color 

Apparently, the red rice had L*, a*, and b* values of color 

parameters smaller than the white rice, but larger than the black 

rice, though the color parameters still differed within the white and 

L* 

a* 

b* 

red rice groups (Fig. 2). Among total rice accessions, the five color 

parameters (L*, a*, b*, C and H ) were significantly negatively 

correlated with phenolics, flavonoid contents, and antioxidant 

capacity (Table 3). Within the white rice accessions, some correlations 

were still significant (Table 3), but only the L* vs phenolic 

contents, a* vs antioxidant capacity, H vs phenolic contents and 

flavonoid contents were negatively correlated, whereas antioxidant 

capacity vs L*, b* and H , and a* vs phenolics and flavonoid contents 

were positive, which implied that the correlation from total rice 

accessions could not be applied to the individual rice groups. 

Within the red rice group, a* was still negatively correlated with 

phenolic contents and antioxidant capacity, C was also negatively 

correlated with antioxidant capacity, but the H was positively 

correlated with phenolic contents and antioxidant capacity 

(Table 3). It was worthy to note that the negative correlation 

between a* and antioxidant capacity was consistent in white and 

red rice groups, and so was the positive correlation between H and 

antioxidant capacity (Table 3), suggesting that these correlations 

could be used as indirect indexes to select rice breeding lines with 

high antioxidant capacity. 

3.4. Relationships between phenolics, flavonoids, antioxidant 

capacity and grain size and 100-grain weight 

Among all rice accessions, only flavonoid contents were poorly 

positively correlated with grain length, length/width ratio and 100grain 

weight (Table 4). However, it seems that the antioxidant 

capacity was negatively correlated with, whereas the flavonoids 

contents were positively correlated with grain length and length/ 

width ratio within the white rice group (Table 4). The 100-grain 

weight was negatively correlated with phenolics, flavonoid 

contents and antioxidant capacity. Interestingly, the flavonoid 

contents still had positive correlation with grain length within the 

red rice group (r ¼ 0.325). Contrast to the relationship within the 

total rice and white rice, the 100-grain weight was positively 

correlated with the flavonoid contents within the red rice accessions 

(Table 4). 

3.5. Principal component analysis 

Principal component analysis was performed on the ten variables 

including phenolic contents, flavonoid content, antioxidant 

capacity, color parameters, grain size and 100-grain weight (Tables 

5 and 6). All the twelve principal components and their corresponding 

eigenvalues and variances are listed in Table 5. The results 

indicated that the first five principal components could explain 

83.7% of total variance. The first principal component (PC1) was the 

most important one, explaining 26.1% of the total variance (Table 5). 

The PC1 represented the phenolics content, C and H of color 

parameters (Table 6). The second principal component (PC2) 

accounted for an additional 19.2% of the total variances, which was 

mainly attributed to flavonoid contents, L* and a* of color parameters. 

The third principal component (PC3) accounted for 16.2% of 

the total variance; the variation was mainly contributed by 

Table 3 

Correlation coefficients between phenolics, flavonoids contents, antioxidant capacity (ABTS) and grain color parameters among the total rice, white and red rice genotypes 


Phenolics Flavonoids ABTS Phenolics Flavonoids ABTS Phenolics Flavonoids ABTS 

L* 0.774*** 0.463*** 0.738*** 0.103* 0.065 0.333*** 0.166 0.014 0.128 

a* 0.349*** 0.277*** 0.398*** 0.170*** 0.136** 0.460*** 0.292* 0.165 0.335* 

b* 0.592*** 0.481*** 0.579*** 0.066 0.018 0.212*** 0.161 0.101 0.261 

C 0.271*** 0.145** 0.311*** 0.156** 0.089 0.001 0.191 0.089 0.307* 

H 0.589*** 0.529*** 0.582*** 0.152** 0.126* 0.477*** 0.326* 0.195 0.332* 

* , ** and *** were significant at 0.05, 0.01 and 0.001 probability level, respectively. 

Y. Shen et al. / Journal of Cereal Science 49 (2009) 106–111 109

110 

Table 4 

Correlation coefficients between phenolics, flavonoids contents, antioxidant capacity (ABTS) and grain size and 100-grain weight among the total rice, white and red rice 

genotypes 


Phenolics Flavonoids ABTS Phenolics Flavonoids ABTS Phenolics Flavonoids ABTS 

Length 0.005 0.169*** 0.036 0.064 0.177*** 0.178*** 0.072 0.325* 0.106 

Width 0.011 0.030 0.007 0.042 0.052 0.054 0.021 0.155 0.068 

Length/Width ratio 0.031 0.106* 0.042 0.014 0.132** 0.097* 0.064 0.083 0.113 

100 grain weight 0.067 0.110* 0.065 0.146** 0.117* 0.160** 0.126 0.322* 0.168 

* , ** and *** were significant at 0.05, 0.01 and 0.001 probability level, respectively. 

antioxidant capacity, grain length and grain width. The fourth 

principal component mainly representing the grain length to width 

ratio and 100-grain weight explained 13.2% of the total variance. 

The fifth principal component representing the b* of color parameter 

with eigenvalue of 1.095, explained an additional 9.1% of the 

total variance (Tables 5 and 6). 

4. Discussion 

The role of phenolics as natural antioxidants has attracted 

considerable interest due to their pharmacological functions. 

Increased consumption of phenolic compounds has been associated 

with the reduced risk of cardiovascular diseases and certain 

cancers (Liu, 2004, 2007; Dykes and Rooney, 2007). The whole rice 

grain had phenolic contents ranging from 108.1 to 1244.9 mg GAE/ 

100 g (Table 1), depending on the color of grain (Choi et al., 2007; 

Goffman and Bergman, 2004). Goffman and Bergman (2004) 

reported that the phenolic contents in the white, red and purple 

rice ranged from 25 to 246, 34 to 424, 69 to 535 mg GAE/100 g, 

which was a little lower than this study (Table 1). Three reasons 

may explain the differences. First, they used rice materials grown in 

two years, which means that material harvested from the first year 

had to be stored. Storage of rice grain results in a decrease of 

phenolic content (Zhou et al., 2004), thus their data was lower than 

ours. Second, rice phenolic compounds exist in free, esterified and 

insoluble-bound forms, and insoluble-bound phenolics may be 

released by base, acid or enzymatic treatment of samples prior to 

extraction (Adom and Liu, 2002; Choi et al., 2007; Zhou et al., 2004). 

Our extraction solution including 1% HCL may cause release of at 

least part of the bound phenolics, thus leading to higher phenolic 

contents. Third, more rice accessions were used in this study may 

provide more wide diversity in the phenolic contents. Flavonoids 

have potent antioxidant and anticancer activities (Adom and Liu, 

2002; Dykes and Rooney, 2007; Hu et al., 2003). Coloration of rice 

may be derived from accumulation of anthocyanins (Furukawa 

et al., 2007). Thus, it could be expected that the white rice had 

mean flavonoid content (131.6 mg RE/100 g) lower than those of 

red rice (147.2 mg RE/100 g) and black rice (240 mg RE/100 g) 

Table 5 

Principal component analysis for all the ten parameters 

Component Eigenvalue Variance (%) Cumulative variance (%) 

1 3.126 26.05 26.05 

2 2.300 19.17 45.22 

3 1.941 16.18 61.40 

4 1.579 13.16 74.56 

5 1.095 9.12 83.68 

6 0.726 6.05 89.73 

7 0.691 5.76 95.49 

8 0.263 2.19 97.68 

9 0.215 1.79 99.47 

10 0.037 0.30 99.78 

11 0.023 0.19 99.97 

12 0.004 0.03 100.00 

Y. Shen et al. / Journal of Cereal Science 49 (2009) 106–111 

(Table 1). It should be noted that some red rice accessions were 

lower in the flavonoid contents than white rice (Fig. 1). It was 

possible that other flavonoid compositions (e.g. flavonols, flavanones) 

rather than anthocyanins in the white rice were higher than 

those in the red rice. The antioxidant capacity of rice grain was 

contributed not only by phenolic compounds, but also by other 

phytochemicals, such as carotenoids, tocols and g-oryzanols (Choi 

et al., 2007; Xu et al., 2001). However, phenolic contents were 

reportedly positively correlated with TABS antioxidant capacity 

(r ¼ 0.962), which was consistent with nearly all other studies 

(Adom and Liu, 2002; Choi et al., 2007) using samples from 

different plant origins (Cai et al., 2004). 

Biofortification by conventional breeding is one of the strategies 

to improve nutritional quality of rice grain. It is apparent that 

phenolics, flavonoids and antioxidant capacity differ from the grain 

color (Table 1), yet these components and antioxidant capacity still 

differ among the white color rice (Table 1). Therefore, breeding 

efforts for better nutritional quality may be applied for white rice, if 

the whole grain is used for food product development. For ease of 

breeding, indirect selection approaches are often used in the 

breeding program. One of indirect selection methods is to use the 

color parameters which can be easily measured in an automatic 

color difference meter. The color parameters L*, b* and H are 

positively, while a* is negatively correlated with antioxidant 

capacity (Table 3). Thus, breeding lines high in antioxidant could be 

indirectly selected by selecting grain with larger L*, b* and H , but 

smaller a*. 

It is well known that most of the phytochemicals are rich in the 

bran layers which make up approximately 8–10% of the rough rice 

weight (Yokoyama, 2004), though some of them are still present in 

the milled rice. It is general knowledge that, in a given weight of 

rice, the larger the grain size, the smaller surface area per weight of 

grain. Smaller surface area might imply lower phytochemicals in 

the bran layers. However, whether the grain size negatively 

correlated with phenolics, flavonoids and antioxidant capacity was 

not tested before. It is found that the 100-grain weight was truly 

negatively correlated with flavonoid contents among the total rice 

accessions, and the correlations were still true among white rice, 

but not true among red rice accessions (Table 4). The grain size was 

Table 6 

Sources of variation for the first five principal components (PC) 

PC1 PC2 PC3 PC4 PC5 

Phenolics 0.481 0.065 0.092 0.075 0.215 

Flavonoids 0.143 0.591 0.187 0.136 0.164 

ABTS 0.323 0.237 0.433 0.029 0.333 

L* 0.224 0.479 0.320 0.094 0.240 

a* 0.173 0.548 0.255 0.135 0.210 

b* 0.274 0.098 0.087 0.018 0.564 

C 0.485 0.011 0.297 0.111 0.109 

H 0.464 0.062 0.313 0.095 0.062 

Length 0.076 0.079 0.492 0.305 0.455 

Width 0.128 0.128 0.394 0.364 0.413 

Length/width ratio 0.106 0.101 0.095 0.522 0.030 

100 grain weight 0.073 0.123 0.049 0.655 0.053

not necessarily negatively correlated with flavonoids, but inversely 

the relationship was positive. The antioxidant capacity was negatively 

correlated with grain length, length/width ratio and 100grain 

weight (Table 4), so in the white rice breeding practice, high 

antioxidant capacity of breeding lines could be indirectly selected 

with short grain length and smaller 100-grain weight in breeding 

programs. 

In conclusion, this study found wide diversity in the phenolics, 

flavonoid contents and antioxidant capacity in the whole rice grain. 

These data provide opportunities to further increase the content of 

phenolics, flavonoids and antioxidant capacity by breeding, especially 

in white rice. Their relationships with grain color, grain size 

and 100-grain weight could serve as indexes to indirectly select rice 

breeding lines high in phenolics, flavonoids and antioxidant 

capacity. The results could provide rice breeders and eventually 

commercial rice producers, with new opportunities to promote the 

production of rice with enhanced levels of the phytochemicals. Rice 

genotypes rich in phytochemicals may be incorporated into functional 

foods. 

Acknowledgement 

The authors thank Mr. Junquan Yu and Dr. Jianliang Lu for their 

assistance in chemical analysis and the color parameters test, 

respectively. We also thank the anonymous reviewers for their 

constructive comments. Financial support for this work was 

provided in part by National High Technology Development Project, 

National Natural Science Foundation of China and the Science and 

Technology Department of Zhejiang Province. 

References 

Abdel-Aal, E.M., Young, J.C., Rabalski, I., 2006. Anthocyanin composition in black, 

blue, pink, purple, and red cereal grains. Journal of Agricultural and Food 

Chemistry 54, 4696–4704. 

Adom, K.K., Liu, R.H., 2002. Antioxidant activity of grains. Journal of Agricultural and 

Food Chemistry 50, 6182–6187. 

Aguilar-Garcia, C., Gavino, G., Baragano-Mosqueda, M., Hevia, P., Gavino, V.C., 2007. 

Correlation of tocopherol, tocotrienol, g-oryzanol and total polyphenol content 

in rice bran with different antioxidant capacity assays. Food Chemistry 102, 

1228–1232. 

Bao, J.S., Cai, Y., Sun, M., Wang, G.Y., Corke, H., 2005. Anthocyanins, flavonols, and 

free radical scavenging activity of Chinese bayberry (Myrica rubra) extracts and 

their color properties and stability. Journal of Agricultural and Food Chemistry 

53, 2327–2332. 

Y. Shen et al. / Journal of Cereal Science 49 (2009) 106–111 111 

Bergman, C.J., Xu, Z., 2003. Genotype and environment effects on tocopherol, 

tocotrienol, and g-oryzanol contents of Southern U.S. rice. Cereal Chemistry 80, 

446–449. 

Bouis, H.E., Chassy, B.M., Ochanda, J.O., 2003. Genetically modified food crops and 

their contribution to human nutrition and food quality. Trends in Food Science 

and Technology 14, 191–209. 

Cai, Y.Z., Luo, Q., Sun, M., Corke, H., 2004. Antioxidant activity and phenolic 

compounds of 112 traditional Chinese medicinal plants associated with anticancer. 

Life Science 74, 2157–2184. 

Choi, Y., Jeong, H.S., Lee, J., 2007. Antioxidant activity of methanolic extracts from 

some grains consumed in Korea. Food Chemistry 103, 130–138. 

Dykes, L., Rooney, L.W., 2007. Phenolic compounds in cereal grains and their health 

benefits. Cereal Foods World 52, 105–111. 

Furukawa, T., Maekawa, M., Oki, T., Suda, I., Lida, S., Shimada, H., Takamure, I., 

Kadowaki, K.I., 2007. The Rc and Rd genes are involved in proanthocyanidin 

synthesis in rice pericarp. Plant Journal 49, 91–102. 

Goffman, F.D., Bergman, C.J., 2004. Rice kernel phenolic content and its relationship 

with antiradical efficiency. Journal of the Science of Food and Agriculture 84, 

1235–1240. 

Harjes, C.E., Rocheford, T.R., Bai, L., Brutnell, T.P., Kandianis, C.B., Sowinski, S.G., et al., 

2008. Natural genetic variation in Lycopene Epsilon Cyclase tapped for maize 

biofortification. Science 319, 330–333. 

Hu, C., Zawistowski, J., Ling, W., Kitts, D.D., 2003. Black rice (Oryza sativa L. indica) 

pigmented fraction suppresses both reactive oxygen species and nitric oxide in 

chemical and biological model systems. Journal of Agricultural and Food 

Chemistry 51, 5271–5277. 

Jiang, S.L., Wu, J.G., Feng, Y., Yang, X.E., Shi, C.H., 2007. Correlation analysis of 

mineral element contents and quality traits in milled rice (Oryza sativa L.). 


Liu, R.H., 2004. Potential synergy of phytochemicals in cancer prevention: mechanism 

of action. Journal of Nutrition 134, 3479S–3485S. 

Liu, R.H., 2007. Whole grain phytochemicals and health. Journal of Cereal Science 

46, 207–219. 

Miller, A., Engel, K.H., 2006. Content of g-oryzanol and composition of steryl ferulates 

in brown rice (Oryza sativa L.) of European origin. Journal of Agricultural 

and Food Chemistry 54, 8127–8133. 

Orthoefer, F.T., Eastman, J., 2004. Rice bran and oil. In: Champagne, E.T. (Ed.), Rice 

Chemistry and Technology. AACC, St. Paul, Minnesota, U.S.A., pp. 569–593. 

Storozhenko, S., Brouwer, V.E., Volckaert, M., Navarrete, O., Blancquaert, D., 

Zhang, G.F., et al., 2007. Folate fortification of rice by metabolic engineering. 

Nature Biotechnology 25, 1277–1279. 

Xu, Z., Hua, N., Godber, J.S., 2001. Antioxidant activity of tocopherols, tocotrienols, 

and g-oryzanol components from rice bran against cholesterol oxidation 

accelerated by 2,2 0 -azobis(2-methylpropionamidine) dihydrochloride. Journal 

of Agricultural and Food Chemistry 49, 2077–2081. 

Yawadio, R., Tanimori, S., Morita, N., 2007. Identification of phenolic compounds 

isolated from pigmented rices and their aldose reductase inhibitory activities. 


Ye, X., Al-Babili, S., Kloti, A., Zhang, J., Lucca, P., Beyer, P., Potrokus, I., 2000. Engineering 

the pro-vitamin A (beta-carotene) biosynthetic pathway into (carotenoid-free) 

rice endosperm. Science 287, 303–305. 

Yokoyama, W., 2004. Nutritional properties of rice and rice bran. In: 

Champagne, E.T. (Ed.), Rice Chemistry and Technology. AACC, St. Paul, Minnesota, 

U.S.A., pp. 595–609. 

Zhou, Z.K., Robards, K., Helliwell, S., Blanchard, C., 2004. The distribution of phenolic 

acids in rice. Food Chemistry 87, 401–406.

Nucleotide polymorphisms in the waxy gene of NaN3-induced 

waxy rice mutants 

Toong Long Jeng a , Chang Sheng Wang b , Tung Hai Tseng a , Min Tze Wu a , Jih Min Sung c, * 

a Division of Biotechnology, Taiwan Agricultural Research Institute, Wu Fong, Taichung 413, Taiwan, ROC 

b Department of Agronomy, National Chung Hsing University, Taichung 402, Taiwan, ROC 

c Department of Food Science & Nutrition, Hungkuang University, Shalu, Taichung 433, Taiwan, ROC 

article info 





Keywords: 

bp duplication 

G-to-T substitution 

Microsatellite 

Polymorphism 

Rice 

Waxy mutation 


abstract 

Starch is the most abundant storage reserves in rice (Oryza 

sativa L.) grains. The starch reserves of rice grains generally contain 

17–30% amylose, which plays an important role in palatability and 

processing qualities (Ramesh et al., 1999; Zhou et al., 2002). 

Granule bound starch synthase (GBSS) is the key enzyme responsible 

for amylose synthesis in rice grain (James et al., 2003; Vandeputte 

and Delcour, 2004). Six loci are involved in the amylose 

synthesis of rice grains, with the dominant Wx gene that encodes 

GBSS being the major locus (Itoch et al., 2003). Mutation of the 

dominant Wx locus into recessive wx causes inactivation or absence 

of GBSS (Han et al., 2004), and subsequently results in rice grain of 

opaque endosperm that contains little or no amylose (Wang and 

Wang, 2002). 

Waxy (wx) mutants have been observed in both japonica and 

indica rice genotypes by spontaneous and induced mutation (Han 

et al., 2004; Hori et al., 2007; Sato and Nishio, 2003). Several 

functional molecular markers associated with Wx gene are 

detectable in waxy rice genotypes. An alternative splicing at the 5 0 

* Correspondence author. Tel.: þ886 4 26318652x5015. 

E-mail address: sungjm@sunrise.hk.edu.tw (J.M. Sung). 


doi:10.1016/j.jcs.2008.07.009 





Spontaneous and induced waxy phenotype, associated with endosperm containing little or no amylose, 

has been recognized in rice (Oryza sativa L.). Mutation of a dominant gene Wx into a recessive gene wx, 

which causes the inactivation or absence of granule bound starch synthase, is believed to be responsible 

for the change in endosperm starch leading to the waxy grain. In the present study, the nucleotide 

polymorphism in the Wx gene of rice genotype Tainung 67 (wild type) and its 35 NaN3-induced wx 

mutants were examined. Iodine staining confirmed that all the mutants had waxy grain trait. The G-to-T 

single base substitution analysis indicated that the wild type genotype Tainung 67 and its waxy mutants 

carried Wx b allele. Moreover, 23-bp duplication in exon 2 was detected in all the waxy mutants. 

Microsatellite polymorphism (CT)n was also detectable on the Wx gene of the tested genotypes and 

mutants, with at least 5 classes of (CT)n microsatellites identified at the Wx locus. Electrophoretic 

analyses also confirmed the observed nucleotide polymorphsim. Thus, nucleotide polymorphsim exist 

among NaN3-induced waxy mutants in rice. However, only the 23-bp duplication in exon 2 may be used 

as a molecular marker to characterize waxy grain trait in rice genotypes. 


splice site of intron 1 caused by a single base substitution (AGGT to 

AGTT) generally exists in waxy rice genotypes (Han et al., 2004; 

Isshiki et al., 1998; Itoch et al., 2003; Yamanaka et al., 2004). A 

duplication of 23-bp in the second exon, which causes the inactivation 

or loss of GBSS, is detectable only in waxy rice genotypes 

(Hori et al., 2007; Jeng et al., 2007; Wanchana et al., 2003). Polymorphic 

cytosine and thymidine (CT)n microsatellites in the 5 0 - 

untranslated region are also noted in waxy rice genotypes, but with 

the number of alleles (4 alleles) consistently identified much less 

than non-waxy rice genotypes (10 alleles) (Bao et al., 2006). 

Many gamma ray- or chemical mutagen-induced waxy mutant 

lines have been obtained in rice, and the point mutations in Wx 

gene have been identified (Isshiki et al., 2001; Sato and Nishio, 

2003). In this study, DNA polymorphisms of the Wx gene among 

Tainung 67 and its NaN3-induced wx mutant lines (Wang et al., 

2002) were investigated. The G-to-T single base substitution and 

23-bp duplication, which were generally used to differentiate waxy 

and non-waxy rice varieties on a molecular level, were used to 

identify the possible variations in the Wx locus of the tested 

genotypes and mutants. CT-microsatellite polymorphism was also 

employed to compare the possible differences among the tested 

mutants. The results of this study are useful for further study in 

identifying and breeding the waxy rice genotype with desirable 

agronomic traits.


A total of 38 rice (O. sativa L.) accessions, including wild type 

genotype Tainung 67 and its 35 NaN3-induced waxy mutants, and 2 

naturally mutated waxy genotypes TKW 1 and TCSW 1, were 

obtained from the Taiwan Agricultural Research Institute (Table 1). 

Additionally, a NaN3-induced non-waxy recovery of SA 419 (a NaN3induced 

waxy mutant of genotype Tainung 67) was also used to 

identify the possible effect of 23-bp duplication on the expression of 

wx gene. Seeds were sown in the nursery plots. The 3-leaf stage 

seedlings were transplanted to experimental plots (3 6 m) at a hill 

spacing of 30 15 cm with 3 seedlings per hill. Each plot received 

a basal application of fertilizer before transplanting (24 kg N ha 1 , 

36 kg P2O5 ha 1 and 24 kg K2Oha 1 ) and 3 top-dressings of fertilizer 

at 20 d (6 kg N ha 1 , 9 kg P2O5 ha 1 and 6 kg K2Oha 1 ), 40 d 

(9 kg N ha 1 , 13.5 kg P2O5 ha 1 and 9 kg K2Oha 1 ) and 60 d 

(9 kg N ha 1 , 13.5 kg P2O5 ha 1 and 9 kg K2Oha 1 ) after transplanting 

(Jeng et al., 2003). All plants were grown to maturity. 

Mature dry seeds were hand-cut with a razor blade, and the halfseeds 

without embryo were soaked in distilled water at 4 Covernight. 

After the treatment, the half-seeds were sliced and stained 

with potassium iodide for visual examination (Itoch et al., 2003). 

Genomic DNA was extracted from leaves of 30-day-old seedlings 

of each genotype or line (Doyle and Doyle, 1990). For sequence 

analysis of the Wx alleles, a fragment was amplified by PCR (Invitrogen 

Co., USA) using two primer pairs (i.e., 5 0 -CAAGCTGGAAA 

AGCAAAAG-3 0 and 5 0 -TTGACCAACTCGGCTACTAA, 5 0 -ATGTCTCTCG 

CCACTGGA-3 0 and 5 0 -CTCAGCCACAACGCTGGTAT-3 0 ). PCR amplification 

was carried out by using 10 ng genomic DNA, 2 mM of each 

primer and 1 U rTag DNA polymerase (TaKaRa Co., Japan) at 56.4 C 

with 32 cycles. DNA sequencing was performed on an ABI 373 

automated sequencer following the manufacturer’s instruction 

(Applied Biosystems, Inc., USA) using Dye Terminator Cycle 

Sequence Ready Reaction Kit (Perkin Elmer Co., USA). 

The G-to-T substitution at the 5 0 leader intron splice donor site 

of the Wx alleles was examined using a derived cleaved amplified 

polymorphic sequence (dCAPS) technique developed by Yamanaka 

et al. (2004). The forward 5 0 -TGTTGTTCATCAGGAAGAACATCTC 

CAAG-3 0 and reverse primers 5 0 -TTAATTTCCAGCCCAACACC-3 0 

generate a unique EcoT14I restriction site characteristic of the Wx a 

allele. Total DNA was isolated from leaf, and the DNA fragments at 

the splice donor site of the first intron were amplified by PCR 

(Invitrogen Co., USA). PCR amplification was carried out by using 

100 ng genomic DNA, 0.4 mM of each primer and 1 U rTag DNA 

polymerase at 56 C with 32 cycles. Five microliters of each PCR 

product was digested with EcoT14I in a total volume of 20 mL at 

37 C overnight. After digestion, each digest was electrophoresed in 

a30gkg 1 Nusieve 3:1 agarose gel with ethidium bromide staining. 

The 23-bp duplication was detected using the technique 

detailed by Wanchana et al. (2003). The forward 5 0 -TGCAGA 

GATCTTCCACAGCA-3 0 and reverse primers 5 0 -GCTGGTCGTCAC 

GCTGAG-3 0 were used to generate amplicons specific for Wx alleles. 

Table 1 

The 35 NaN3-induced rice mutants of genotype Tainung 67 

Code Mutant Code Mutant Code Mutant Code Mutant 

1 SA 419 10 SA 1472 19 SA 572 28 SA 967 

2 SA 402 11 SA 491 20 SA 573 29 SA 1047 

3 SA 420 12 SA 492 21 SA 580 30 SA 1200 

4 SA 472 13 SA 495 22 SA 588 31 SA 1220 

5 SA 893 14 SA 497 23 SA 649 32 SA 1314 

6 SA 897 15 SA 502 24 SA 674 33 SA 1324 

7 SA 1917 16 SA 523 25 SA 860 34 SA 1398 

8 SA 1170 17 SA 526 26 SA 861 35 SA 1399 

9 SA 1445 18 SA 543 27 SA 908 

The code numbers are used to identify the tested mutants presented on Figs. 1–3. 

T.L. Jeng et al. / Journal of Cereal Science 49 (2009) 112–116 113 

The forward primer was marked by IRDye700 (MWG-Biotech Co., 

Germany). A reaction mixture (20 mL) containing 20 ng of genomic 

DNA as the template, 0.2 mM of dNTPs, 0.2 mM of each primer, 0.5 U 

of Taq DNA polymerase, 50 mM HCl, 2.0 mM MgCl2, and 10 mM 

Tris-HCl (pH 8.3), and deionized H2O were added to make 20 mL 

final volume. The amplification reactions were carried out by the 

following profile: 94 C pre-denaturation for 2 min followed by 32 

cycles of (94 C denaturation for 30 s, 60 C annealing for 30 s and 

72 C for 1 min) and the final extension at 72 C for 5 min. The PCR 

products were then electrophoresed in 70 g kg 1 polyacrylamide 

gel with LICOR 4300 DNA Analyzer (Licor Co., Germany). Similar 

PCR conditions were also used to detect CT-microsatellite polymorphism 

by using the forward 5 0 -CTTTGTCTATCTCAAGACAC-3 0 

and reverse primers 5 0 -TTGCAGATGTTCTTCCTGATG-3 0 to generate 

amplicons specific for Wx alleles. 


Notable differences in appearance of de-hulled grains were 

observed among the tested genotypes and NaN3-induced mutants 

(Fig. 1). All the tested mutants had chalky endosperm. Of these 

samples, the majority of mutants had white pericarp. Only 2 

mutants had dark blue color pericarp (SA 526 and SA 572) and 3 

mutants had brown color pericarp (SA 897, SA 674 and SA 1399). In 

Southeast Asia, waxy rice is grown mainly as a staple food (Wanchana 

et al., 2003). In Taiwan, waxy rice, particularly the waxy rice 

with colored pericarp, is generally reserved for use in festival foods 

and desserts (Jeng et al., 2007). Jeng et al. (2006) previously 

reported that, under adverse environmental conditions, the grain 

yield of NaN3-induced low amylose mutant SA 419 (code number 1 

in Table 1 and Figs. 1–3) was better than its wild type genotype 

Tainung 67, and therefore was recommended to waxy rice growers 

and food manufacturers. In this regard, these 5 mutants with dark- 

Fig. 1. The de-hulled grains of rice genotype Tainung 67 (TNG67) and its 35 NaN3induced 

mutants as well as two naturally occurred waxy mutants TKW 1 (japonica 

type) and TCSW 1 (indica type). The numbers marked in the figure match the code 

numbers listed on Table 1.

114 

Fig. 2. The iodine-stained endosperms of rice genotype Tainung 67 (TNG67) and its 35 

NaN3-induced mutants as well as two naturally occurred waxy mutants TKW 1 

(japonica type) and TCSW 1 (indica type). The numbers marked in the figure match the 

code numbers listed on Table 1. 

colored pericarp could also be recommended to growers and food 

manufacturers, if their palatability, processing qualities and grain 

yield were proved promising. 

Rice genotype is considered as waxy with 0–20 g kg 1 of grain 

amylose (Jeng et al., 2006). The opaque endosperm is generally 

regarded as a phenotypic marker for waxy rice (Bao et al., 2006; 

Han et al., 2004). This grain trait can be easily detected using iodine 

staining, which is a very sensitive technique to detect amylose in 

various tissues (Itoch et al., 2003). In the present study, the dark 

blue color of iodine staining suggested that the Wx gene was 

expressed in cultivar Tainung 67 (200 g kg 1 amylose) (Jeng et al., 

2006) (Fig. 2). On the other hand, all the waxy mutants (genotype 

TCSW 1, genotype TKW 1 and 35 NaN3-induced mutants) had 

orange color of stained endosperm, suggesting that they either 

have no or severely reduced amylose (Fig. 2). 

Several nucleotide polymorphisms are associated with the Wx 

gene, namely single nucleotide polymorphism (G/T) in the intron 1, 

T.L. Jeng et al. / Journal of Cereal Science 49 (2009) 112–116 

23-bp duplication in the exon 2 and CT-microsatellites (Bao et al., 

2002; Wanchana et al., 2003; Yamanaka et al., 2004). Extensive 

research has indicated that two functional Wx alleles exist in 

cultivated rice genotypes, with Wx a having AGGT sequence and Wx b 

having an alternative AGTT sequence at 5 0 splice site in intron 1 

(Han et al., 2004; Isshiki et al., 1998; Itoch et al., 2003; Yamanaka 

et al., 2004). A mutation from AGGT sequence to AGTT sequence at 

5 0 splice site of Wx intron 1 leads to incomplete post-transcriptional 

processing of Wx pre-mRNA. Thus, the expression of Wx a is higher 

than Wx b at RNA, GBSS and amylose levels (Isshiki et al., 1998). 

Waxy rice genotypes generally have no detectable levels of spliced 

mRNA as a result of this mutation (Wang et al., 1995). They are 

characterized as the Wx b allele based on the G-to-T base substitution 

(Wang et al., 1995) and have very low or no amylose in mature 

grains (Wang and Wang, 2002). Nevertheless, Yamanaka et al. 

(2004) recently examined 353 waxy rice strains collected from 

various regions of Asia, and found that the collected waxy strains 

had both Wx a (AGGT)- and Wx b (AGTT)-derived alleles. They further 

indicated that the Wx b (AGTT)-derived allele was predominant, and 

Wx a (AGGT) allele was detected in only a minority of samples. Thus, 

G-to-T base substitution alone might not be a reliable marker for 

differentiating waxy and non-waxy rice varieties at a molecular 

level. In the present study, nucleotide sequence was also analyzed 

in genotype Tainung 67 and its 9 NaN3-induced waxy mutants as 

well as 2 waxy genotypes TKW 1 (japonica type waxy rice) and 

TCSW 1 (indica type waxy rice) (Table 2). Two genotypes Nipponbare 

(japonica genotype with intermediate amylose level) and 93- 

11 (indica genotype with high amylose level), in which their 

nucleotide sequence had been analyzed and released on internet 

http://www.ncbi.nlm.nih.gov/, were also added for comparison 

(Table 2). The results showed that only indica genotype 93-11 had 

AGGT sequence at putative 5 0 leader intron. All the other tested 

accessions had AGTT sequence, regardless of their origin (indica or 

japonica). These results suggest that all the tested waxy mutants 

carry the Wx b allele. On the other hand, both japonica type 

genotypes Tainung 67 and Nipponbare (both with intermediate 

amylose level) have AGTT sequence. These results seem to indicate 

that the G-to-T polymorphism might be a useful molecular marker 

for identifying the rice genotype with grain amylose below 

intermediate level, but it is not sufficient to identify the waxy 

starch grain trait. 

Fig. 3. Nucleotide polymorphisms of (A) 23-bp duplication, (B) G-to-T substitution and (C) (CT)n repeats on Wx alleles of rice genotype Tainung 67 (TNG67) and its 35 NaN3-induced 

mutants as well as two naturally occurred waxy mutants TKW 1 (japonica type) and TCSW 1 (indica type). A high amylose genotype TN 1 is added for comparison (a is wild type 

genotype Tainung 67, b is high amylose genotype TN 1, c is genotype TCSW 1, d is genotype TKW 1, M is marker). The numbers marked in figures match the code numbers listed on 

Table 1.

Table 2 

Comparisons of nucleotide sequences of Wx alleles of rice genotype Tainung 67 

(TNG67) and its NaN3-induced mutants 

.exon 1.intron 1.exon 2. 

TNG 67 

1.50.142.1389. 

.(CT)18.T.(ACGGGTTCCAGGGCCTCAAGCCC)1 

Nipponbare .(CT)18.T.(ACGGGTTCCAGGGCCTCAAGCCC)1 

93-11 .(CT)18.G.(ACGGGTTCCAGGGCCTCAAGCCC)1 

TKW 1 .(CT)18.T.(ACGGGTTCCAGGGCCTCAAGCCC)2 

TCSW 1 .(CT)16.T.(ACGGGTTCCAGGGCCTCAAGCCC)2 

SA 419 .(CT)15.T.(ACGGGTTCCAGGGCCTCAAGCCC)2 









Two naturally mutated waxy genotypes TKW 1 (japonica type) and TCSW 1 (indica 

type) were added for comparison. Additionally, the nucleotide sequences of genotypes 

Nipponbare (japonica type) and 93-11 (indica type) are collected from internet 

http://www.ncbi.nlm.nih.gov/. 

While apparently required for the mutation, the presence of the 

single base substitution (G-to-T) at the 5 0 splice site of intron 1 

alone does not ensure that waxy mutation is expressed (Prathepha 

and Baimai, 2004; Yamanaka et al., 2004). Wanchana et al. (2003) 

identified a unique duplication of 23-bp in the second exon, which 

was present only in the waxy rice genotypes. The 23-bp duplication 

in exon 2 would result in a frame shift and subsequently cause 

a premature stop codon in the second exon and bring about 

degradation of mRNA by nonsense-mediated decay (Isshiki et al., 

2001). Hori et al. (2007) examined 29 waxy genotypes and a 23-bp 

duplication at the second exon was detected in 27 genotypes 

(including paddy and upland waxy genotypes), and the two other 

waxy genotypes were considered to have a mutation (insertion of 

7764 bp in exon 9) different from the 23-bp duplication. We also 

analyzed TKW 1, TCSW 1 and 9 NaN3-induced waxy mutants 

derived from Tainung 67 and detected 23-bp duplication in all the 

tested waxy mutants (Table 2). As expected, no 23-bp insertion was 

detected on non-waxy japonica genotypes Tainung 67, Nipponbare 

and indica genotype 93-11. Thus, the findings of Wanchana et al. 

(2003) are still valid even though the waxy mutation is induced by 

chemical mutagen NaN3, at least for these NaN3-induced waxy 

mutants derived from genotype Tainung 67. 

Polymorphic cytosine and thymidine (CT)n microsatellites are 

noted in the 5 0 -untranslated region of the Wx gene (Bergmaan et al., 

2001; Prathepha and Baimai, 2004), which seems to correlate well 

with the various amylose contents in non-waxy genotypes (Bao 

et al., 2002). Bao et al. (2002) examined a set of waxy rice genotypes, 

and found 4 microsatellite alleles, (CT)16, (CT)17, (CT)18, (CT)19 

at the wx locus, of which (CT)17 was the most frequent. A similar 

conclusion was also reported from other studies (Han et al., 2004; 

Prathepha and Baimai, 2004). In the present study, a total of 5 

classes of (CT)n microsatellites, located at 55-bp upstream of the 

putative 5 0 leader intron splice-junction were detected. Among the 

tested waxy mutants, 5 waxy mutants carried (CT)17 allele, 2 

mutants contained (CT)15 allele, 2 mutants contained (CT)16 allele, 

a mutant had a (CT)18 allele and a mutant had (CT)19 allele (Table 2). 

Similar (CT)18 allele were also detected on non-waxy genotypes 

TNG 67, Nipponbare and 93-11. It should be pointed out that the 

(CT)15 allele was the only allele not detected in other studies (Bao 

et al., 2002; Han et al., 2004; Prathepha and Baimai, 2004). CT 

repeats are present in the 5 0 -untranslated region and variations of 

the number of the CT repeats from 8 to 20 have been report in nonwaxy 

genotype (Olsen and Purugganan, 2002). Therefore, CT 

T.L. Jeng et al. / Journal of Cereal Science 49 (2009) 112–116 115 

repeats is not considered to be the cause of the waxy trait (Hori 

et al., 2007). However, CT repeats might potentially serve as 

a molecular marker for other starch qualities of waxy rice, such as 

G/T-related thermal properties as well as some of the pasting 

viscosity parameters (Bao et al., 2002). 

Direct nucleotide sequencing is labor intensive and costly to 

analyze a large sample set. Electrophoretic analysis is known to be 

an accurate technique to analyze large numbers of samples quickly, 

without the need for direct sequencing determination (Yamanaka 

et al., 2004). In the present study, all the 35 waxy mutants 

(including the accessions used for nucleotide sequence analysis) 

and their wild type genotype Tainung 67 plus 2 waxy genotypes 

TKW 1 and TCSW 1 that were differentiated by iodine staining 

(Fig. 2) were used to identify 23-bp duplication polymorphism 

through electrophoresis. A well-known non-waxy indica type 

genotype TN 1 (Taichung Native 1) was also added to serve as 

a comparison. The electrophoretic results (Fig. 3A) showed that no 

23-bp duplication was detected in genotypes Tainung 67 and TN1, 

which contain intermediate and high grain amylose, respectively. 

As expected, all the waxy accessions (including cultivars TKW 1 and 

TCSW 1) had the 23-bp insertion in the second exon (Fig. 3A). This 

premature stop codon due to the 23-bp insersion would cause the 

loss of function of GBSS encoded by the Wx gene (Sato and Nishio, 

2003). For further verifying the functional role of the 23-bp 

duplication, an electrophoresis on a NaN3-induced non-waxy 

recovery line of SA 419 was conducted and the results indicated 

that no 23-bp duplication at the second exon was found for this 

non-waxy recovery line of SA 419 (Fig. 4). This result clearly indicates 

that the 23-bp duplication at the second exon is essential for 

NaN3-induced waxy mutation, at least for these mutants derived 

from genotype Tainung 67. 

Electrophoresis of G-to-T polymorphism indicated that only 

indica genotype TN 1 having high grain amylose content had AGGT 

sequence on intron 1 splice site (Fig. 3B). On the other hand, all the 

tested waxy mutants, including TKW 1 and TCS 1, as well as their 

wild type TNG 67 contained the AGTT sequence at the 5 0 splice site 

of the first intron (Fig. 3B). Moreover, variations in the length of CTmicrosatellite 

in the 5 0 -untranslated region of the wx gene (Bao 

et al., 2002; Prathepha and Baimai, 2004) was also obtainable in the 

tested accessions (Fig. 3C). 

In conclusion, the present results indicate that significant 

differences in grain appearance are observed among the tested 

genotypes and NaN3-induced waxy mutants. Nucleotide polymorphisms 

also exist among the tested mutants. G/T polymorphism 

analysis indicates that all the NaN3-induced waxy 

mutants carry Wx b allele. However, the presence of the single base 

substitution (G-to-T) at the 5 0 splice site of intron 1 alone does not 

ensure that waxy mutation is expressed, because both wild type 

genotype Tainung 67 and non-waxy genotype Nipponbare (both 

are genotypes with intermediate grain amylose level) also shows 

same G/T substitution at 5 0 splice site of intron 1. Additionally, at 

least 5 classes of (CT)n microsatellites at the Wx locus are identified 

among the tested waxy mutants. However, the CT repeats are also 

not the cause of the waxy trait. On the other hand, a 23-bp 

Fig. 4. Nucleotide polymorphisms of 23-bp duplication on Wx alleles of (1) mutant SA 

419 and its (2) non-waxy recovery.

116 

duplication in the second exon is present in all the NaN3-induced 

waxy mutants. This 23-bp duplication appears to be necessary for 

the waxy trait in genotype Tainung 67. Thus, the 23-bp duplication 

in exon 2 may be used as a molecular marker to characterize waxy 

grain trait in rice breeding program. 


We gratefully acknowledge financial support from the National 

Science Council of ROC. 

References 

Bao, J.S., Corke, H., Sun, M., 2002. Microsatellites in starch-synthesizing genes in 

relation to starch physicochemical properties in waxy rice (Oryza sativa L.). 


Bao, J.S., Corke, H., Sun, M., 2006. Microsatellites, single nucleotide polymorphisms 

and a sequence tagged site in starch-synthesizing genes in relation to starch 

physicochemical properties in non-waxy rice (Oryza sativa L.). Theoretical and 

Applied Genetics 113, 1185–1196. 

Bergmaan, C.J., Delgado, J.T., McClung, A.M., Fjellstrom, R.G., 2001. An improved 

method for using a microsatellite sequence closely linked to the waxy gene to 

determine amylose class. Cereal Chemistry 78, 257–260. 

Doyle, J.J., Doyle, J.L., 1990. Isolation of plant DNA from fresh tissue. Focus 12, 13–15. 

Han, Y., Xu, M., Liu, X., Yan, C., Korban, S.S., Chen, X., Gu, M., 2004. Genes coding for 

starch branching enzymes are major contributors to starch viscosity characteristics 

in waxy rice (Oryza sativa L.). Plant Science 166, 357–364. 

Hori, Y., Fujimoto, R., Sato, Y., Nishio, T., 2007. A novel wx mutation caused by 

insertion of a retrotransposon-like sequence in a glutinous cultivar of rice 

(Oryza sativa). Theoretical and Applied Genetics 115, 217–224. 

Isshiki, M., Yamamoto, Y., Satoh, H., Shimamoto, K., 2001. Nonsense-mediated decay 

of mutant waxy mRNA inrice. Plant Physiology 125, 1388–1395. 

Isshiki, M., Morino, K., Nakajima, M., Okagaki, R.J., Wessier, S.R., Izawa, T., 

Shimamoto, K., 1998. A naturally occurring functional allele of the rice waxy 

locus has a GT to TT mutation at the 5 0 splice site of the first intron. Plant Journal 

15, 133–138. 

T.L. Jeng et al. / Journal of Cereal Science 49 (2009) 112–116 

Itoch, K., Ozaki, H., Okada, K., Hori, H., Takeda, Y., Mitsui, Y., 2003. Introduction of 

Wx transgene into rice wx mutants lead to both high- and low-amylose rice. 

Plant and Cell Physiology 44, 473–480. 

James, M.G., Denyer, K., Meyers, A.M., 2003. Starch synthesis in the cereal 

endosperm. Current Opinions in Plant Biology 6, 215–222. 

Jeng, T.L., Wang, C.S., Yseng, T.H., Sung, J.M., 2007. Expression of granule-bound 

starch synthase in developing rice grain. Journal of the Science of Food and 

Agriculture 87, 2456–2463. 

Jeng, T.L., Tseng, T.H., Wang, C.S., Chen, C.L., Sung, J.M., 2003. Starch biosynthesizing 

enzymes in developing grains of rice cultivar Tainung 67 and its sodium-azideinduced 

rice mutant. Field Crops Research 84, 261–266. 

Jeng, T.L., Tseng, T.H., Wang, C.S., Chen, C.L., Sung, J.M., 2006. Yield and grain 

uniformity in contrasting rice genotypes suitable for different growth environment. 

Field Crops Research 99, 59–66. 

Olsen, K.M., Purugganan, M.D., 2002. Molecular evidence on the origin and 

evolution of glutinuous rice. Genetics 162, 941–950. 

Prathepha, P., Baimai, V., 2004. Variation of Wx microsatellite allele, waxy allele 

distribution and differentiation of chloroplast DNA i n a collection of Thai rice 

(Oryza sativa L. Euphytica 140, 231–237. 

Ramesh, M., Ali, S.Z., Bhattacharya, K.R., 1999. Structure of rice starch and its 

relation to cooked-rice texture. Carbohydrate Polymers 38, 337–347. 

Sato, Y., Nishio, T., 2003. Mutation detection in rice waxy mutants by PCR-RF-SSCP. 


Vandeputte, G.E., Delcour, J.A., 2004. From sucrose to starch granule to starch 

physical behaviour: a focus on rice starch. Carbohydrate Polymers 58, 245–266. 

Wanchana, S., Toojinda, T., Tragoonrung, S., Vanavichit, A., 2003. Duplicated coding 

sequence in the waxy allele of tropical glutinuous rice (Oryza sativa L.). Plant 


Wang, Y.J., Wang, L., 2002. Structures of four waxy rice starches in relation to 

thermal, pasting and textural properties. Cereal Chemistry 79, 252–256. 

Wang, C.-T., Tseng, T.-H., Lin, C.-Y., 2002. Rice biotech research at the Taiwan 

Agricultural Research Institute. Asia Pacific Biotechnology News 6, 950–956. 

Wang, Z.-Y., Zheng, F.-Q., Shen, G.-Z., Gao, J.-P., Snutad, D.P., Li, M.-G., Zhang, J.-L., 

Hong, M.-M., 1995. The amylose content in rice endosperm is related to the 

post-transcriptional regulation of the waxy gene. Plant Journal 7, 613–622. 

Yamanaka, S., Nakamura, I., Watanabe, K.N., Sato, Y.-I., 2004. Identification of SNPs 

in the waxy gene among glutinuous rice cultivars and their evolutionary 

significance during the domestication process of rice. Theoretical and Applied 

Genetics 108, 1200–1204. 

Zhou, Z., Robards, K., Helliwell, S., Blanchard, C., 2002. Ageing of stored rice: changes 

in chemical and physical attributes. Journal of Cereal Science 35, 65–78.

Amaranth (Amaranthus hypochondriacus) as an alternative crop for sustainable 

food production: Phenolic acids and flavonoids with potential impact on its 

nutraceutical quality 

A.P. Barba de la Rosa a, *, Inge S. Fomsgaard b , Bente Laursen b , Anne G. Mortensen b , L. Olvera-Martínez c , 

C. Silva-Sánchez a , A. Mendoza-Herrera d , J. González-Castañeda e ,A.DeLeón-Rodríguez a 

a 

Institute for Scientific and Technological Research at San Luis Potosi, Molecular Biology Division, Camino a La Presa San Jose No. 2055, 

Lomas 4 a sección, San Luis Potosí, SLP, Mexico 

b 

Department of Integrated Pest Management, Faculty of Agricultural Sciences, Aarhus University, Denmark 

c 

CBETa 196 Villa de Pozos, San Luis Potosí, SLP, Mexico 

d 

Horticultural Science Department, HFSB Room 202, 2133 TAMU, College Station, United States 

e 

Instituto de Ciencias Agrícolas, Universidad de Guanajuato, Irapuato, Gto, Mexico 

article info 


Received 7 May 2008 



Keywords: 

Crop yield 

LC/MS/MS 

Protein content 

Phytochemicals 

RAPD 


abstract 

The challenge for agricultural practices to increase food 

production to obtain food security still persists after 40 years of the 

Green Revolution (Hobbs, 2007). The first Millennium Development 

Goal is to reduce hunger and poverty by 2015 (Dixon et al., 

2006). The demand for food is increasing, not only because of the 

growing population, but also to provide more nutritious food with 

high protein quality and nutraceutical compounds. 

Water resources, especially surface and ground water will be 

more limited as domestic and industrial needs increase just as it is 

limited in semi-desert zones with low precipitation, and so future 

crops must be more suited to low water use. Amaranth (Amaranthus 

hypochondriacus) is a crop naturally resistant to water deficit 

and is a good source of nutritious seeds; the seeds have high 

* Corresponding author. Tel.: þ52 4448342000; fax: þ52 4448342010. 

E-mail address: apbarba@ipicyt.edu.mx (A.P. Barba de la Rosa). 


doi:10.1016/j.jcs.2008.07.012 





The demand for food is increasing, not only to meet food security for growing populations, but also to 

provide more nutritious food, rich in good quality proteins and nutraceutical compounds. The amaranth 

(Amaranthus hypochondriacus) plant, in addition to its high nutritive and nutraceutical characteristics, 

has excellent agronomic features. The objective of the present study was to analyze some physical and 

proximal-nutritional properties of amaranth seeds obtained from different varieties grown in arid zones 

and characterize their phenolic acids and flavonoids. Two commercial (Tulyehualco and Nutrisol) and 

two new (DGETA and Gabriela) varieties of A. hypochondriacus were grown at the Mexican Highlands 

zone. Tulyehualco and DGETA varieties had higher seed yield of 1475 and 1422 kg ha 1 , respectively, 

comparable to corn and soybean production in agricultural areas. Gabriela had the highest protein 

content of 17.3%, but all varieties had an adequate balance of essential amino acids. Polyphenols as rutin 

(4.0–10.2 mgg 1 flour) and nicotiflorin (7.2–4.8 mgg 1 flour) were detected. Amaranth can be cultivated 

in arid zones where commercial crops cannot be grown; the seeds besides their well known nutritive 

characteristics could be a source of phenolic compounds of high antioxidant properties. 


amounts of protein containing essential amino acid such as lysine 

and methionine; also significant levels of squalene, an important 

precursor for all steroids (He et al., 2002), were reported in 

amaranth oil. Since the beginning of the 1980s, amaranth has been 

rediscovered and several reports have tried to promote it as a basic 

crop (Kauffman, 1992). In addition to nutritional characteristics, 

amaranth plants have agronomic features identifying it as an 

alternative crop where cereals and vegetables cannot be grown (dry 

soils, high altitudes and high temperatures) (Omami et al., 2006). In 

general, the selection of promising genotype in a breeding program 

is based on various criteria, with the most important being final 

crop yield and seed quality (Kozak et al., 2008). 

The aim of the research presented in this paper was to analyze 

four amaranth varieties and select the most promising for growing 

in arid zones like the Mexican Highlands or arid zones of Southern 

Europe. Variety selection was carried out on the basis of seed yield 

and proximal quality. The presence of a range of phytochemical 

compounds that contribute to nutraceutical properties of the seeds 

was determined as well.

118 



Two commercial varieties (Tulyehualco and Nutrisol) and two 

new varieties (DGETA and Gabriela) of A. hypochondriacus were 

cultivated at the CBTa 196 Villa de Pozos in San Luis Potosi, under 

low watering conditions, precipitations between 25 and 35 mm. 

The experiment was in randomized complete blocks with five 

blocks per variety. Seeds were drilled manually in four rows, each 

6 m long, 80 cm apart and with 30 cm spacing in the row 

(Henderson et al., 2000). Plants were watered at 35% of field 

capacity, at the beginning of soil preparation (April 2002) and in 

August 2002. After stem elongation was complete (November 

2002), hand-harvest was carried out at maturity by cutting plants 

at soil level. 

2.2. Physical analysis 

Measurements of morphologic parameters were taken from 10 

randomly chosen plants. Plant height was measured from the 

ground to the top of the inflorescence head. Spike length was 

measured from the base to the top of inflorescence. For seed yield, 

all plants from each block were hand-harvest, seeds were cleaned 

and weighted, and yield was calculated and extrapolated to kg ha 1 . 

A sample of 100 seeds per triplicate was weighted and extrapolated 

to obtain the weight of 1000 seeds. The seeds were milled to obtain 

the flour able to pass through a 100-mesh screen and flour yield 

was determined as the percentage of flour recovery. The flours 

were defatted with hexane at a flour/hexane ratio of 1:10 (w/v) 

under continuous stirring for 4 h at 4 C, the slurry was centrifuged 

at 9000 g for 20 min and the flour was air-dried at room temperature 

and stored at 4 C(Barba de la Rosa et al., 1992). 

2.3. Chemical analysis 

Total nitrogen (method 954.01), fat (920.39), ash (923.03), crude 

fiber (962.09), and moisture (925.09) contents of the flours were 

determined according to AOAC (1990). Nitrogen was determined 

with a Kjeltec system (Tecator, Sweden). Protein was calculated 

from total nitrogen using a factor of 5.85 (Barba de la Rosa et al., 

1992). Fat amount was obtained from 4 h hexane extraction. Ash 

was calculated from the weight remaining after heating the sample 

at 550 C for 2 h. Moisture measurement was determined from 

sample weight loss after oven drying at 110 C for 4 h. All samples 

were analyzed in triplicate. 

2.4. Essential amino acid analysis 

Amino acid analysis of amaranth flours was performed by 

reversed-phase high performance liquid chromatography (RP-HPLC) 

using a HPLC Waters 600 with a fluorescence detector 2475 (Waters) 

and performed by AccQ-Tag amino acid analysis kit and following 

the manual’s instructions (Waters). Samples of 5 mg of amaranth 

flours were hydrolyzed with 200 mL of 6 N HCl for 24 h at 110 C in 

vacuo system; one or two phenol crystals were added as an oxygen 

scavenger (Barba de la Rosa et al., 1992). The samples were 

neutralized with 1.2 N NaOH and dehydrated. Subsequently, samples 

were derivatized with 6-aminoquinolyl-N-hydroxysuccinimidyl 

carbamate. A 5 mL sample was applied to the column AccQ-Tag C18 

(Waters). All samples were analyzed in triplicate. 

2.5. Phytochemical analysis 

The freeze-dried amaranth flour samples were crushed and 

homogenized before phytochemicals were extracted using an 

A.P. Barba de la Rosa et al. / Journal of Cereal Science 49 (2009) 117–121 

Accelerated Solvent Extraction 200 system (Dionex) (ASE), 

following the protocol reported by Carlsen et al. (2008). Briefly, 

0.1 g of the freeze-dried and homogenized sample was transferred 

to the 33 mL extraction cells followed by 5-g of chemically inert 

Ottawa sand (20–30 mesh particle size, Fisher Chemicals) after 

heating at 400 C. A filter was placed on top of the sample, and the 

extraction cell was filled with glass balls that had been heated at 

400 C. The eluent was 70% methanol, 30% water (v/v). The protocol 

for the ASE extraction was the following: preheat for 5 min, heat for 

5 min, static for 3 min, flush 80%, purge for 60 s, 4 cycles, pressure 

1500 psi, and temperature 80 C. Phytochemical extracts were 

collected in vials, which were filled with eluent to maintain 46.1 g 

( ¼ 50 mL) weight for all extracts, and stored at 20 C until 

chemical analysis. The extracts were filtered on a Sartorius SRP 15 

0.45-m filter (PTFE membrane) and diluted with water in a ratio 

of 1:1. 

Phytochemical analysis was carried out using an Applied 

Biosystems MDS Sciex API3200 Liquid chromatography-Ion Trap 

Quadrupole Mass Spectrometer (LC/MS/MS) with turbo electrospray 

ionization in a negative multiple reaction monitoring (MRM) 

mode. Pure reference compounds were used for identification of 

the phenolic acids and polyphenols based on a comparison of 

fragmentation pattern and retention times. Standard curves for 

commercially obtained pure standards were prepared in 50% 

methanol and 50% water (v/v) at the following concentrations: 3.13, 

6.25, 12.5, 25.0, 50.0, 100, 200, 400, and 800-mgL 1 . The standard 

curve was used for the quantitative determination. The analyzed 

compounds were three polyphenols, rutin (CAS no. 153-18-4, 

Extrasynthese, France), isoquercitrin (CAS no. 482-35-9, Fluka, 

Germany), and nicotiflorin (CAS no. 17650-84-9, Carl Roth, 

Germany), and three phenolic acids, vanillic acid (CAS no. 121-34-6, 

Fluka, Germany), 4-hydroxybenzoic acid (CAS no. 99-96-7, Fluka, 

Germany), and syringic acid (CAS no. 530-57-4, Sigma, Germany). 

2.6. Electrophoretic patterns 

Protein fractions were obtained by following the procedure 

described elsewhere with some modifications (Barba de la Rosa 

et al., 1992). Briefly, for albumin fraction extraction, a suspension of 

flour/distilled water (1:10 w/v) was prepared and stirred for 1 h at 

4 C and centrifuged at 9000 g for 20 min at 4 C. The resulting 

pellet was re-suspended in 10 mM Na2HPO4, pH 7.5, 1 mM EDTA, 

0.1 M NaCl, stirred and centrifuged as before, the supernatant was 

named as globulins 7S. The pellet was again re-suspended in 

phosphate buffer but containing 0.8 M NaCl; the soluble fraction 

was named globulins 11S. The glutelin fraction was extracted with 

0.1 M NaOH. The protein patterns were analyzed by SDS-PAGE at 

15% w/v polyacrylamide in a Mini-Protean III system (Bio-Rad). 

Samples of 1 mg mL 1 were dissolved in Laemmli sample buffer 

(Laemmli, 1970). Electrophoresis was conducted at a constant 

current of 20 mA per gel for 2–3 h. After electrophoresis, the gel 

was stained with Coomassie Brilliant Blue G250 at a final concentration 

of 0.25%. Destaining was achieved by washing the gel for 

2–4 h with acetic acid/methanol/water (4.5:4.5:1 v/v/v). 

2.7. RAPD analysis 

DNA of the four cultivars was extracted from fresh young leaves 

by a standard procedure. DNA quantification was by measurement 

of absorbance at 260/280 nm. The extracted DNA was dissolved in 

double distilled water (Dellaporta et al., 1983). For amplification, 

the pair of oligonucleotides denominated as 23: 5 0 -CCC GCC TCC 

C-3 0 and 43: 5 0 -AAA ACC GGG C-3 0 were used as reported by Chan 

and Sun (1997) and following the protocol described by those 

authors: 3 mM MgCl2, 0.1% Triton X-100, 75 ng primer, and 30 ng 

genomic DNA was used in 25 mL reaction mix. PCR amplification

Table 1 

Physical properties of amaranth seeds 

Variety Plant height 

(cm) 

Tulyehualco 159.0 a 

DGETA 154.2 b 

Gabriela 149.1 b 

Nutrisol 161.3 a 

was performed in a iCycler (Bio-Rad) using the following cycle 

profile: 1 cycle at 94 C for 2 min followed by 45 cycles at 94 C for 

1 min, 35 C for 2 min, and 72 C for 2 min; and the final cycle at 

72 C for 7 min. The amplification products were electrophoresed 

on 1.4% agarose gels and visualized using ethidium bromide 

staining. Negative controls were routinely used to check for 

possible contamination. For RAPD analysis, the banding patterns 

were recorded in a GelDoc photodocumentator (Bio-Rad) and gels 

analyzed with Quantity-one software (Bio-Rad). The bands with the 

same molecular weight and mobility were treated as identical 

fragments; the presence of a band was coded as 1, and absence 

coded as 0. The data matrices were analyzed with the Squared 

Euclidean nearest neighbor method (Statgraphics Plus v5.0 

software, Statistical Graphics Corp). Dendrograms were produced 

from the results. 


Spike length 

(cm) 

Statistical analysis was performed and Tukey’s test was used to 

determine significance of differences among means. Significance 

was when compared means differed at P 0.05 (Reyes-Castañeda, 

1983). 


Seed yield 

(kg ha 1 ) 

68.0 a 

1475 a 

54.5 a 

1422 a 

28.2 b 

1204 b 

nd 1121 b 

3.1. Physical properties of amaranth seeds 

Weight of Flour yield 

1000 seeds (%) 

(g) 

0.6 a 

87.3 a 

0.7 a 

89.1 a 

0.7 a 

87.6 a 

0.6 a 

88.2 a 

Means of three replicates within in the same column (physical property) with 

different letter are significantly different at P 0.05. 

The physical properties of amaranth plant and seeds are shown 

in Table 1. Tulyehualco and Nutrisol plants were the highest (161.3 

and 159.0 cm, respectively), but Tulyehualco had the longest spike 

(68 cm). Tulyehualco and DGETA had the highest seed yields of 

1475 and 1422 kg ha 1 , respectively. When grown in drier environments, 

amaranth seeds yields of 1050 and 410 kg ha 1 were 

reported (Henderson et al., 2000). The amaranth seed yields 

obtained in our work compare favorably to commercially important 

crops like soybean (1532 kg ha 1 ) and maize (2315 kg ha 1 ) 

(SAGARPA, 2002). There were no differences in the weight of 1000 

seeds among varieties. Flour yield is an important quality parameter 

of wheat, and since flour is an alternative use of amaranth 

seeds, this parameter was also measured. There were no differences 

among the varieties with a mean value of 88% (Table 1), values that 

are similar to wheat flour yields (Paredes-López et al., 1990). In 

Central Mexico, amaranth grain is paid at double the price of maize 

Table 2 

Proximal composition of amaranth seeds (% dry basis) 

Variety Protein* Lipids Ash Fiber 

Tulyehualco 15.0 b 

8.1 a 

3.3 a 

2.2 a 

DGETA 14.8 b/c 

7.9 a/b 

3.3 a 

1.9 b 

Gabriela 17.3 a 

8.9 a 

3.9 a 

2.5 a 

Nutrisol 15.3 b 

8.6 a 

3.5 a 

2.0 a/b 

*N 5.85. Means of triplicates in the same column with different letter are significantly 

different at P 0.05. 

A.P. Barba de la Rosa et al. / Journal of Cereal Science 49 (2009) 117–121 119 

Table 3 

Essential amino acid composition of different amaranth variety flours (g of amino 

acid 100 g 1 of crude protein) 

Amino acid Tulyehualco DGETA Gabriela Nutrisol FAO/WHO/UNU (1986) 

Ile 3.2 2.2 3.1 3.5 1.3 

Leu 5.8 5.0 6.5 6.6 1.9 

Lys 6.9 5.2 7.5 6.7 1.6 

Met þ Cys a 

0.3 1.7 6.5 7.2 1.7 

Phe þ Tyr b 

7.9 3.7 8.3 7.3 1.9 

Thr 3.8 3.3 4.8 4.6 0.9 

Val 3.4 2.6 3.6 3.9 1.3 

a 

Requirements for methionine þ cysteine. 

b 

Requirements for phenylalanine þ tyrosine. 

grain. In this sense, farmers could have double benefits, seeds of 

high nutritional quality and higher incomes for the rural families 

(Islas-Gutiérrez and Islas-Gutiérrez, 2001). 

3.2. Proximal composition of amaranth 

Amaranth protein content (Table 2) was higher than values 

reported for cereals (Paredes-López et al., 1990). The variety 

Gabriela had the highest seed protein content (17.3%) followed by 

Tulyehualco, Nutrisol and DGETA (15.0%, 15.3%, and 14.8%, respectively). 

The fat contents were similar among the varieties; however, 

at higher proportions (7–9%), than for cereals such as wheat (2.1%) 

and maize (4.5%). There were no differences in ash contents 

between varieties, but Gabriela had the highest fiber content, 2.5% 

(Table 2). Amaranth has been considered as a new source of fiber 

and methods for fractionation to get high-fiber amaranth products 

have been reported (Tosi et al., 2001). The essential amino acids in 

amaranth seeds (Table 3) are ideal according to FAO requirements 

for adults (FAO/WHO/UNU, 1986). 

3.3. Phytochemicals in amaranth seed flours 

It has been reported that amaranth seed flour contains polyphenols 

(flavonoids) with relatively high antioxidant status. For 

this reason, amaranth has been recommended for use in balanced 

diets (Gorinstein et al., 2007). However, to the best of our knowledge, 

no results have ever been published before on the types of 

polyphenols present in amaranth seed flours. In this study, three 

polyphenols (rutin, isoquercitrin, and nicotiflorin), were identified 

and quantified (Table 4). Rutin was present at higher concentrations 

(10.1 mg g 1 flour) in Tulyehualco seed flour, while the 

highest amount of nicotiflorin (7.2 mg g 1 flour) was found in the 

Gabriela variety (Table 4). Rutin has been reported to be present in 

amaranth leaves (Suryavanshi et al., 2007), but this is the first time 

it has been reported in seed flours. 

Polyphenols in which sugar groups are beta-linked, as in the 

case of the three polyphenols found in this study, are easily 

degraded in the intestine of humans and animals due to the 

abundance of beta-glucosidase enzyme that liberates the aglyconic 

Table 4 

Phenolic acids and flavonoids present in different amaranth variety flour 

(mg metabolite g 1 flour) a 

Metabolite Tulyehualco DGETA Gabriela Nutrisol 

Isoquercitrin 0.5 0.5 0.3 nd 

Nicotiflorin 5.5 5.6 7.2 4.8 

Rutin 10.1 5.8 4.0 4.7 

4-Hydroxybenzoic acid 1.7 2.0 2.2 1.9 

Syringic acid 0.8 0.7 nd nd 

Vanillic acid 1.8 1.7 1.8 1.5 

a Mean of two replicates, nd ¼ not detected.

120 

moiety of the molecules. Several health effects have been published, 

which generally concern the uptake of the aglyconic groups, 

quercetin and kaempferol (Donovan et al., 2007). Rutin and its 

metabolites may effectively modulate advanced glycation end 

product (AGE) formation, which is associated with numerous 

pathologies. Clinical diseases possibly accelerated by AGEs include 

neuropathy, nephropathy, retinopathy, joint stiffness, senile cataracts, 

Alzheimer’s disease, and cardiovascular disease (Cervantes- 

Laurean et al., 2006). Polyphenols such as quercetin have been 

shown to serve as a protective defense against oxidative damage in 

vivo (Meyers et al., 2008). Nicotiflorin has been claimed to have 

protective effects on reducing memory dysfunction (Huang et al., 

2007); recent results have proved a strong pharmacological basis 

for its potential therapeutic role in cerebral ischemic illness (Li 

et al., 2006). Phenolic compounds were proved to have antioxidant 

activity, where rutin, vanillic acid, ferulic acid, and quercetin have 

been reported in rice wine samples and correlated with antioxidant 

activities (Que et al., 2006). 

In relation to phenolic acids, one paper was published in which 

several phenolic acids were identified and quantified in amaranth 

seeds using HPLC-UV analysis (Klimczak et al., 2002). In our study, 

using LC/MS/MS, three phenolic acids; vanillic acid; 4-hydrozybenzoic 

acid, and syringic acid, were identified and quantified (Table 4). The 

levels of syringic acid and vanillic acid are similar to those reported in 

rye where syringic acid has been related with their bitterness (Heiniö 

et al., 2008). 

3.4. Electrophoretic patterns of seed storage amaranth proteins 

Electrophoretic patterns of the different protein fractions in 

amaranth seeds were obtained under denaturing conditions. The 

albumin fraction was the main fraction in amaranth proteins; all 

varieties had a characteristic band at 34 kDa (Fig. 1A) as reported by 

Barba de la Rosa et al. (1992). The albumin fraction was characterized 

by a group of molecular weight around 18 kDa; this group of 

proteins is known as the MRPs (methionine rich proteins), proteins 

that contain 16–18% of methionine (Segura-Nieto et al., 1994). 

Another group of proteins in the albumins fraction was located at 

40–80 kDa. Albumins are comparable with egg-white proteins, and 

this protein fraction has been used as an egg substitute to prepare 

bread (Silva-Sánchez et al., 2004). 

There are two groups of globulin seed storage proteins, the 7S 

globulins were extracted with 0.1 M NaCl (Fig. 1B) and had a main 

band around 38 kDa. Globulin fractions extracted with 0.8 M NaCl 

(Fig. 1C) showed the characteristic pattern of 11S-like globulins, the 

group of acid polypeptides (35–38 kDa) and the group of basic 

polypeptides (22–25 kDa). The band of high molecular weight 

(55 kDa) was previously reported as a globulin precursor (Barba de 

la Rosa et al., 1992). Recently we reported that the 11S globulin 

A.P. Barba de la Rosa et al. / Journal of Cereal Science 49 (2009) 117–121 

Fig. 1. Electrophoretic patterns of amaranth proteins fractions. A) albumins, B) 7S globulins, C) 11S globulins, D) glutelins. Lane S ¼ molecular weight standard; lane 1 ¼ Tulyehualco; 

lane 2 ¼ DGETA; lane 3 ¼ Gabriela; lane 4 ¼ Nutrisol. The arrow shows the differential band found in glutelin fraction. 

fraction is rich in peptides of angiotensin converting enzyme 

inhibitor (Silva-Sánchez et al., 2008). 

The amaranth glutelin pattern (Fig. 1D) had high similarity with 

11S globulins. There are three main polypeptides groups, one at 

22–25 kDa, the second at 35–38 kDa, and the third around 55 kDa. 

In this fraction, there was an additional band at 65 kDa in the 

Tulyehualco and DGETA varieties; these varieties were also those 

with the higher seed yields. Abugouch et al. (2003) suggested that 

glutelins are globulin-like proteins with hexameric structure of 

approximately 300 kDa, and it is known that the amaranth globulin 

sequence has high homology with the Cupin domain that has been 

associated with resistance to extremes of environment, as heat, 

sulfur nutrition, water stress (Higashi et al., 2006; Khuri et al., 

2001). Former studies by LC/MS/MS analysis showed that the 

glutelin fraction is rich in antihypertensive, and also contains the 

anticarcinogenic lunasin-like peptide (Silva-Sánchez et al., 2008). 

3.5. RADP analysis 

Morphological variations cause confusion or misidentification of 

varieties of the same species so that genetic analyses could be 

necessary for correct identification. In comparative studies using 

RAPD, RFLP and/or allozyme markers, the most valuable tool for 

correct identification of genetic variations was RAPD (Chan and 

Sun, 1997). Four different varieties of amaranth were studied in this 

work; the amplified fragments were analyzed and were grouped, 

generating a dendogram (Fig. 2). DGETA and Gabriela are the closer 

varieties and probably both derived from Nutrisol. The data indicate 

Fig. 2. Dendrograms of Amaranth hypochondriacus varieties based on RAPD’s analysis 

using the Squared Euclidean nearest neighbor method from Statgraphics software. 

Lane 1 ¼ Tulyehualco; lane 2 ¼ DGETA; lane 3 ¼ Gabriela; lane 4 ¼ Nutrisol; lane 

5 ¼ control.

that Tulyehualco variety is the one most distantly related to the 

other three. 


As interest in amaranth cultivation has increased and breeders 

have produced a large number of new varieties adapted to different 

zones (Gimplinger et al., 2008; Henderson et al., 2000). However, 

some of these new varieties are only new names for old varieties. 

RAPD analysis showed that the old variety, Tulyehualco, could be 

the progenitor of the other three varieties used in this work; this 

variety also had the highest seed yield and has a good balance of 

amino acids. Here, for the first time, some secondary metabolites 

are reported to be present in amaranth seed flours. These metabolites, 

namely rutin, isoquercitrin, and nicotiflorin are compounds 

that could have implications in the prevention of several illnesses. 

The highest rutin content was found in Tulyehualco, while nicotiflorin 

was found in highest concentration in the Gabriela variety. 

Tulyehualco, because of the high seed yield, high protein and rutin 

content, seems to be an excellent amaranth variety to grow in arid 

areas worldwide. The knowledge of amaranth as a source of 

phytochemicals will increase their importance as a potential source 

of those compounds in the human diet. Amaranth therefore has 

great potential as a sustainable crop that could improve household 

food security and farm incomes. In arid areas where commercial 

crops such as beans, maize or rice cannot be grown, amaranth 

cultivation could contribute to the first Millennium Development 

Goal of reducing hunger and poverty. 

Acknowledgments 

Thanks are due AG Alpuche Solis for his help in plant harvesting, 

and to Fundación Produce San Luis Potosi for financial support. This 

work was partially supported financially by the European 

Commission 6th Framework Programme, AMARANTH:FUTURE- 

FOOD, Contract No. 032263. Thanks to Dagmar Janovská for 

reviewing the manuscript. 

References 

Abugouch, L.E., Martinez, E.N., Añon, M.C., 2003. Influence of the extracting solvent 

upon the structural properties of amaranth (Amaranthus hypochondriacus) 

glutelin. Journal of Agricultural and Food Chemistry 51, 4060–4065. 

AOAC, 1990. Official Methods of Analysis, 15th ed. Association of Official Analytical 

Chemists, Washington, DC. 

Barba de la Rosa, A.P., Gueguen, J., Paredes-López, O., Viroben, G., 1992. Fractionation 

procedures, electrophoretic characterization and amino acid composition 

of amaranth seed protein. Journal of Agricultural and Food Chemistry 40, 

931–936. 

Carlsen, S.C.K., Mortensen, A.G., Oleszek, W., Piacente, S., Stochmal, A., 

Fomsgaard, I.S., 2008. Variation in Flavonoids in Leaves, Stems and Flowers of 

White Clover Cultivars. Natural Products Communications 3, 1299–1306. 

Cervantes-Laurean, D., Schramm, D.D., Jacobson, E.L., Halaweish, I., Bruckner, G.G., 

Biossonneault, G.A., 2006. Inhibition of advanced glycation end product 

formation on collagen by rutin and its metabolites. The Journal of Nutritional 


Chan, K.F., Sun, M., 1997. Genetic diversity and relationships detected by isozyme 

and RAPD analysis of crop and wild species of Amaranthus. Theoretical and 


Dellaporta, S.L., Wood, J., Hicks, J.B., 1983. A plant DNA minipreparation. Plant 

Molecular Biology Reporter 1, 19–22. 

Dixon, J., Nalley, L., Kosina, P., La Rovere, R., Hellin, J., Aquino, P., 2006. Adoption and 

economic impact of improved wheat varieties in the developing world. Journal 

of Agricultural Science 144, 489–502. 

Donovan, J.L., Manach, C., Faulks, R.M., Kroon, P.A., 2007. Absorption and metabolism 

of dietary plant secondary metabolites. In: Crozier, A., Clifford, M.N., 

Ashihara, H. (Eds.), Plant Secondary Metabolites: Occurrence, Structure and 

Role in the Human Diet. Blackwell Publishing, London, pp. 303–351. 

A.P. Barba de la Rosa et al. / Journal of Cereal Science 49 (2009) 117–121 121 

FAO/WHO/UNU, 1986. Special report. Energy and protein requirements. Cereal 

Foods World 3, 694–695. 

Gimplinger, D.M., Schulte auf ´ m Erley, G., Dobos, G., Kaul, H.P., 2008. Optimum crop 

densities for potential yield and harvestable yield of grain amaranth are conflicting. 

European Journal of Agronomy 28, 119–125. 

Gorinstein, S., Medina-Vargas, O.J., Jaramillo, N.O., Arnao-Salas, I., Martínez- 

Ayala, A.L., Arancibia-Avila, P., Toledo, F., Katrich, E., Trakhtenberg, S., 2007. The 

total polyphenols and the antioxidant potentials of some selected cereals and 

pseudocereals. European Food Research and Technology 225, 321–328. 

He, H.P., Cai, Y., Sun, M., Corke, H., 2002. Extraction and purification of squelene 

from amaranthus grain. Journal of Agricultural and Food Chemistry 50, 

368–372. 

Heiniö, R.-L., Liukkonen, K.-H., Myllymäki, O., Pihlava, J.-M., Adlercreutz, H., 

Heinonen, S.-M., Poutanen, K., 2008. Quantities of phenolic compounds and 

their impacts on the perceived flavor attributes of rye grain. Journal of Cereal 


Henderson, T.L., Joson, B.L., Schneiter, A.A., 2000. Row spacing, plant population, 

and cultivar effects on grain amaranth in the northern great plains. Agronomy 


Higashi, Y., Hirai, M.Y., Fujiwara, T., Naito, S., Noji, M., Saito, K., 2006. Proteomic and 

transcriptomic analysis of Arabidopsis seeds: molecular evidence for successive 

processing of seed proteins and its implication in the stress response to sulfur 

nutrition. Plant Journal 48, 557–571. 

Hobbs, P.R., 2007. Conservation agriculture: what is it and why is it important for 

future sustainable food production. Journal of Agricultural Science 145, 

127–137. 

Huang, J.L., Fu, S.T., Jiang, Y.Y., Cao, Y.B., Guo, M.L., Wang, Y., Xu, Z., 2007. Protective 

effects of Nicotiflorin on reducing memory dysfunction, energy metabolism 

failure and oxidative stress in multi-infarct dementia model rats. Pharmacology, 

Biochemistry and Behavior 86, 741–748. 

Islas-Gutiérrez, J., Islas-Gutiérrez, F., 2001. Rentabilidad de los cultivos de amaranto 

y maíz para la zona central de México. Agricultura Técnica de México 27, 

143–151. 

Kauffman, C., 1992. The status of grain amaranth for the 1990s. Food Reviews 

International 8, 165–192. 

Khuri, S., Bakker, F.T., Dunwell, J.M., 2001. Phylogeny, function and evolution of the 

Cupins, a structurally conserved, functionally diverse superfamily of proteins. 

Molecular Biology and Evolution 18, 593–605. 

Klimczak, I., Malecka, M., Pacholek, B., 2002. Antioxidant activity of ethanolic 

extracts of amaranth seeds. Nahrung-Food 46, 184–186. 

Kozak, M., Bocianowski, J., Rybinski, W., 2008. Selection of promising genotypes 

based on path and cluster analyses. Journal of Agricultural Science 146, 85–92. 

Laemmli, U.K., 1970. Cleavage of structural protein during the assembly of head of 

bacteriophage T4. Nature 227, 680–685. 

Li, R.P., Guo, M.L., Zhang, G., Xu, X.F., Li, Q., 2006. Neuroprotection of nicotiflorin in 

permanent focal cerebral ischemia and in neuronal cultures. Biological and 

Pharmaceutical Bulletin 29, 1868–1872. 

Meyers, K.J., Rudolf, J.L., Mitchell, A.E., 2008. Influence of dietary quercetin on 

glutathione redox status in mice. Journal of Agricultural and Food Chemistry 56, 

830–836. 

Omami, E.N., Hammes, P.S., Robbertse, P.J., 2006. Differences in salinity tolerance for 

growth and water-use efficiency in some amaranth (Amaranthus spp.) genotypes. 

New Zealand Journal of Crop and Horticultural Science 34, 11–22. 

Paredes-López, O., Barba de la Rosa, A.P., Hernández-López, D., Cárabez-Trejo, A., 

1990. Amaranto: características y aprovechamiento agroindustrial. Programa 

Regional de Desarrollo Científico y Tecnológico, Washington, DC. 

Que, F., Mao, L., Pan, X., 2006. Antioxidant activities of five chinese rice wines and 

the involvement of phenolic compounds. Food Research International 39, 

581–587. 

Reyes-Castañeda, P. (Ed.), 1983. Bioestadística aplicada, Agronomía, Biología, 

Química. Trillas, México, DF. 

SAGARPA, 2002. Sistema de información agropecuaria de consulta. SIACON, 

México, DF. 

Segura-Nieto, M., Barba de la Rosa, A.P., Paredes-López, O., 1994. Biochemistry of 

amaranth proteins. In: Paredes-López, O. (Ed.), Amaranth: Biology, Chemistry, 

and Technology. CRC Press, Boca Raton, FL, pp. 75–106. 

Silva-Sánchez, C., Barba de la Rosa, A.P., León-GalvánF, de Lumen, B.O., De León- 

Rodríguez, A., González de Mejía, E., 2008. Bioactive peptides in amaranth 

(Amaranthus hypochondriacus) seed storage proteins. Journal of Agricultural and 


Silva-Sánchez, C., González-Castañeda, J., De León Rodríguez, A., Barba de la Rosa, 

A.P., 2004. Functional and rheological properties of amaranth albumins 

extracted from two mexican varieties. Plant Foods for Human Nutrition 59, 

169–174. 

Suryavanshi, V.L., Sathe, P.A., Baing, M.M., Singh, G.R., Lakshmi, S.N., 2007. Determination 

of Rutin in Amaranthus spinosus Linn. Whole Plant Powder by HPTLC. 

Chromatography 65, 767–769. 

Tosi, E.A., Ré, E., Lucero, H., Masciarelli, R., 2001. Dietary fiber obtained from 

amaranth (Amaranthus cruentus) grain by differential milling. Food Chemistry 

73, 441–443.

Relationship of milled grain percentages and flowering-related traits in rice 

Rodante E. Tabien a, *, Stanley Omar P.B. Samonte a , Emmanuel R. Tiongco b 

a Texas A&M University System, AgriLife Research and Extension Center, 1509 Aggie Drive, Beaumont, TX 77713, USA 

b Philippine Rice Research Institute, Science City of Muñoz, Nueva Ecija 3119, Philippines 

article info 





Keywords: 

Rice 

Flowering duration 

Milling qualities 

Flowering traits 


abstract 

The economic value of harvested rice is determined by the grain 

yield and the percentages of head rice (at least 3/4 the length of 

a head or kernel) and total milled rice. The high value of head rice 

relative to broken grains influences the cultivar choice and the preand 

post-harvest handling practices to maximize farm income. 

These milling traits had shown significant positive direct effects on 

gross income, thus these should be considered in selecting cultivars 

to plant and in developing a high income-grossing genotype 

(Samonte et al., 2006). 

Milling traits are influenced by the environment, cultural practices, 

milling processes, drying, and genotype. Environmental 

factors reported to affect head rice percentages include meteorological 

conditions such as relative humidity, air temperature, and 

rainfall (Banaszek and Siebenmorgen, 1990; Jodari and Linscombe, 

1996; Thompson and Mutters, 2006). Cultural practices include 

time of draining (Counce et al., 1990; McCauley and Way, 2002), 

harvesting (McCauley and Way, 2002) and nitrogen fertilization 

(Jongkaewwattana et al., 1993; Leesawatwong et al., 2005; Seetanum 

and De Datta, 1973). Milling processes that influence grain 

milling traits include the type of mill (Bautista and Siebenmorgen, 

2002), milling cylinder speeds (Dilday, 1989), degree of milling 

(Reid et al., 1998), temperature and and duration (Cnossen et al., 

* Corresponding author. Tel.: þ1 409 752 2741. 

E-mail address: retabien@ag.tamu.edu (R.E. Tabien). 


doi:10.1016/j.jcs.2008.07.015 





The economic value of harvested rice is determined by the grain yield and the percentages of head rice 

(at least 3/4 the length of a head or kernel) and total milled rice. This study was conducted to determine 

the effects of flowering-related traits such as duration of flowering, rate of flowering, heading, and 

duration from heading to maturity on head rice and total milling percentages of rice. Flowering data, 

gathered for two years from 105 long grain rice genotypes grown in Beaumont, Texas were analyzed for 

their effects on and relationship with milling traits. A positive linear relationship was obtained for rate of 

flowering and the duration from heading to maturity but negative for duration of flowering and days to 

heading. Genotypes with early heading had relatively shorter flowering durations, and genotypes with 

shorter flowering duration had higher head rice and total milled rice. A faster rate in attaining 100% 

flowering and more days from heading to maturity were favorable in increasing head rice and total 

milled grains. The duration from the start of flowering to heading or to 100% flowering can be used in the 

evaluation and selection for high head rice and total milled rice percentages in rice. 


2003; Schluterman and Siebenmorgen, 2007), cooling methods and 

milling procedures (Pan et al., 2005), and head rice separation 

methods (Lloyd et al., 2001). Drying, as it relates to grain moisture 

content, is critical in getting high percentage of head rice (Banaszek 

and Siebenmorgen, 1990; Dilday, 1989; Siebenmorgen and Jindal, 

1986; Siebenmorgen et al., 1992). Grain moisture content (MC) of 

15–18% has been found ideal for milling grain rice (Jodari and 

Linscombe, 1996), but the most recent study indicated that the 

optimal harvest moisture content (HMC) at which milling traits 

peaked for long grain cultivar was 18–22% while medium grains 

needed 22–24% HMC (Siebenmorgen et al., 2007). 

Cultivars differed significantly in their milled rice percentages 

(Dilday, 1989). Genotypic traits reported to affect head rice include 

tillering and kernel weights at lower seeding rates (Gravois and 

Helms, 1996), kernel thickness (Lu and Siebenmorgen, 1995), 

panicle type and grain weight (Wang et al., 2007), and panicle 

length and maturity (Jongkaewwattana and Geng, 2002). Nitrogen 

fertilization has been shown to increase head rice percentage 

(Jongkaewwattana et al., 1993), but this increase was dependent on 

genotype (Perez et al., 1996). The optimum N rates and harvest 

grain moisture to obtain the maximum head rice percentage of 

different cultivars were also determined by Leesawatwong et al. 

(2005). Siebenmorgen et al. (2007) found that the magnitude of 

head rice reduction at different moisture content was dependent 

on cultivar. 

In estimating the number of days from emergence to heading 

(50% of the panicles are flowering) for each breeding line, rice 

breeders have observed that flowering duration from the onset to

100% flowering and maturity vary across lines. Chau and Kunze 

(1982) reported large variation in grain maturity and consequently 

MC within a field (13–43%). There was a large MC variation in 

individual kernels in panicle and in plant, and among plants 

(Bautista and Siebenmorgen, 2005; Holloway et al., 1995; Kocher 

et al., 1990). Differences in grain weights and quality within panicle 

were found to be dependent on variety (Cheng et al., 2007; Liu 

et al., 2005). 

Rice flowers asynchronously. The flowers at the top of the 

panicles open first and the bottom florets open last, and the flowering 

duration may last up to 15 days (Holloway et al., 1995; Luh 

and Luh, 1991). These variations have resulted in multi-modal MC 

distribution among rice kernels and to a variation in amylose 

content (Cheng et al., 2007). 

Grain structure was also found to be related to grain filling 

which had an impact on head rice (Jongkaewwattana and Geng, 

2001). Decreasing kernel dimensions, volume and density, and 

amylose content from top to lower part of the panicle was observed 

in cultivars with different maturities (Cheng et al., 2007; Jongkaewwattana 

and Geng, 2002) and this non-uniformity influenced 

head rice recovery. It is hypothesized that flowering-related traits 

such as duration of flowering, rate of flowering, heading, and 

duration from heading to harvest can affect head rice and total 

milling percentages. Hence, this study determined these 

relationships. 

2. Experiment 

The Uniform Regional Rice Nursery (URRN) located at the Texas 

A&M University System, AgriLife Research and Extension Center, 

Beaumont, TX (29 57 0 N, 94 30 0 W), is part of a multi-state yield 

trials. The nursery is composed of elite genotypes from various rice 

breeding programs in the USA, and it has 200 entries per year. The 

105 long grain genotypes that were common across the 2005 and 

2006 URRN were used in this study. The soil in the research center 

is Beaumont, an Entic Pelludert (fine, montmorillonitic, and 

thermic). The planting dates of the breeding materials in both years 

were mid-April. Each cropping year, the field experiments were 

fertilized with 225 kg N ha 1 equally split into three and applied at 

planting, one month after planting, and at panicle differentiation. 

The entries were laid out in randomized complete block design 

(RCBD) with two to four replications for entries at the preliminary 

and the advanced trials, respectively, but only two replications 

were used for data gathering in this study. Each plot had six 3-m 

long rows spaced at 25 cm. 

The field data obtained on a plot basis for each line from two 

replications included: days from emergence to onset (start), 25, 50 

(heading), 75, and 100% flowering, and days to harvest. Milling 

samples were obtained at maturity from rice grown in a 1-sq m 

area in each plot. These were threshed and dried right after harvest. 

The head rice and total milled rice percentages were obtained 

following the standard procedures (Fan et al., 2000). 

To produce brown rice, the 125-g rough rice samples were 

hulled by a single pass through a rubber roll huller (Satake Engineering 

Co., LTD., Tokyo, Japan). The brown rice was then milled for 

54 s using a McGill No. 2 mill (Rapsilver Supply Co Inc., Brookshire, 

TX). The generated milled rice was separated into head rice and 

broken fractions with a shaker/separator (#12 screen ¼ 4.76 mm; 

Seedburo Equipment Co., Chicago, IL); The weight of the grain 

recovered was used to calculate the percentage of total milled rice 

and head rice based on 125 g of rough rice. 

From field data, the following flowering durations from seedling 

emergence were estimated on a plot basis: start of flowering 

(onset) to 50% flowering (FD0–50), 50% flowering to 100% flowering 

(FD50–100), start of flowering to 100% flowering (FD0–100), and rate of 

flowering (slope of the number of days to 25, 50, 75, and 100%). 

Date of harvest was used in estimating the number of days from 

50% flowering to maturity. 

Analysis of variance was used to evaluate the effect of genotype, 

year and its interaction to days from emergence to 50% flowering 

(heading), duration of flowering (FD0–50, FD50–100, FD0–100), rate of 

flowering, days from heading to maturity, and milling traits 

(percent head rice and percent total milled rice). Regression and 

correlation analyses were conducted to determine the relationships 

between flowering-related traits (flowering duration, rate of 

flowering, heading, days from heading to maturity) and the 

percentages of head rice and total milled rice, and the days to 

heading with flowering duration. Using the relationships derived, 

estimated values for the flowering-related traits at 70 and 55% total 

and head rice percentages, respectively, were generated. 


3.1. Year, genotype, and genotype year effects on flowering and 

milling traits 

In breeding trials across years or locations, the number of days 

from emergence to heading was significantly affected by genotype 

and genotype year (Table 1). The number of days to heading 

ranged from 82.0 days (recorded in line RU0401084, cvs. ‘Spring’ 

and ‘Trenasse’ (known as very early rice cultivars), to 96.5 days (line 

RU0503126) in 2005 and 72.0 days (cv. Trenasse) to 90.0 days (line 

RU0501139) in 2006. Mean number of days to heading was 88.1 

days in 2005 and 79.9 days in 2006. Although it was much shorter 

for the genotypes to have 50% flowering in 2006, the variations 

were not enough to indicate significance between the two years. 

Earlier genotypes or later genotypes in 2005 showed the same 

flowering response in 2006 indicating a strong genetic control for 

this trait. A QTL in chromosome 8 that explains more than 94.9% of 

variation in heading date was reported recently (Zhang et al., 2006). 

The same locus was identified in several different populations (Lin 

et al., 2003; Xiong et al., 1999; Zhuang et al., 1997); thus this QTL 

could be present in some of the 105 genotypes evaluated in this 

study. 

Table 1 

Means squares from the analysis of variance for flowering-related traits of 105 long grain genotypes grown at Beaumont, Texas in 2005 and 2006 

Source of 

variation 

DF Days to 

heading 

Duration 

0–50% 

flowering 

Duration 

50–100% 

flowering 

Duration 

0–100% 

flowering 

Flowering 

rate 

Duration 

from 

heading to 

maturity 

Whole 

milled rice 

percentage 

Total milled 

rice 

percentage 

Replication 1 21.94 0.09 0.68 0.38 13.30 21.94 0.92 0.19 

Year (Y) 1 7109.49 85.95 0.68 103.09 889.76 237.75 5637.87* 1323.77* 

Error 1 60.95 9.75 3.62 24.38 341.23 60.95 11.55 0.97 

Genotype (G) 104 36.11** 1.84** 0.68** 3.24** 32.24** 47.92** 72.48** 17.65** 

Y G interaction 104 4.72** 1.04 0.52* 1.66 18.17 13.51** 36.58** 6.31** 

Error b 208 1.60 1.16 0.37 1.50 16.86 1.60 8.48 1.20 

*, ** Significant at the 5 and 1% level of probability, respectively. 

R.E. Tabien et al. / Journal of Cereal Science 49 (2009) 122–127 123

124 

Genotype significantly affected FD0–50, FD50–100, and FD0–100, 

while genotype year significantly affected FD50–100 (Table 1). 

These indicate that there was enough variation in the duration of 

flowering among 105 genotypes. Genotype FD0–50 (averaged across 

years) ranged from 2.0 days (line RU0401145 and cv. Spring) to 5.5 

days (line RU0503098), FD0–100 ranged from 3.5 days (cv. Spring) to 

9.5 days (line RU0503098), while FD50–100 ranged from 1.0 days (cv. 

Trenasse) to 4.0 days (line RU0503098). Mean FD0–50, FD0–100, and 

FD50–100 were 3.3, 5.5, and 2.2 days, respectively. 

Shorter duration of flowering (0–50%, 50–100%, or 0–100%) was 

mostly obtained on very early season varieties like cv. Spring and 

Trenasse. Duration of flowering, however, was not taken in yield 

trials, unlike heading date or the number of days from emergence 

to 50% flowering. The highly significant variation obtained and 

interaction noted suggests that these traits may be important in 

obtaining high quantity and quality harvests. Long flowering 

duration may increase variation in kernels at harvest. Since the first 

floret will be at a different stage relative to the late flowering 

florets, there is likely to be high variability in size, degree of grain 

filling, and moisture content. Reports indicated significant variation 

in kernel qualities within the panicle, plant or hill, and fields 

(Bautista et al., 2007; Jongkaewwattana and Geng, 2002), and these 

were related to head rice recovery. 

The rate of flowering (% days 1 ) was significantly affected by 

genotype but not by year, and by the genotype year interaction 

(Table 1). The rate of flowering across years ranged from 9.9% 

days 1 (line RU0503098) to 25.08% days 1 (line RU0502177) in 

2005 and 10.81% days 1 (line RU0503098) to 29.33% days 1 (line 

RU0203032) in 2006. The mean across genotypes for the rate of 

flowering was 17.65% days 1 in 2005 and 20.56% days 1 in 2006. 

The average rate of flowering in two years was slowest in 

RU0503098 (10.36 % days 1 ) and fastest in cultivar Spring (25.66 % 

days 1 ). Among the released cultivars, early maturing cultivars, 

such as Spring and Trenasse, the hybrid XP 723, and the newly 

released Presidio were consistently fast to reach 100% flowering in 

two years. The wide variation observed in two years among 

genotypes suggests that some genotypes had much faster or slower 

rate to attain 100% flowering than the rest of the entries. Similar to 

duration of flowering, the rate of flowering may affect the uniformity 

of the grains. The grains will be more uniform for fast flowering 

genotypes than for slow flowering genotypes. 

The number of days from heading to maturity was significantly 

affected by genotype, and by genotype year interaction, but not 

by year (Table 1). In 2005, the shortest duration from heading to 

harvest was 29 days (line RU0503126) and the longest was 47 days 

(line RU0503181). In 2006, the shortest duration was 26 days (line 

RU0401179) and the longest was 49 days (line RU0501102). Average 

duration to harvest per year was 38 days in 2005 and 40 days in 

2006. Among the released cultivars, Cocodrie, Trenasse, and hybrid 

XP 723 matured longest at 45, 44 and 42 days, respectively. The 

shortest duration among cultivars was from Banks at 34 days followed 

by Saber and Spring at 35 days. The variation among genotypes 

again reflects potential differences in the duration of grain 

filling that can affect milling traits. 

Genotype, genotype year, and year had highly significant 

effect on both head rice and total milled rice percentages (Table 1). 

Head rice percentages (averaged across years) ranged from 37.1% 

(line RU0501096) to 64.3% (line RU0504198) in 2005 and from 

42.2% (cv. ‘Priscilla’) to 67.5% (cv. ‘Cybonnet’) in 2006. Mean (averaged 

across genotypes) head rice percentage was 51.5% in 2005 and 

58.9% in 2006. Total milled rice percentages (averaged across years) 

ranged from 63.0% (line RU0301188) to 72.9% (cv. ‘XP723’) in 2005, 

and from 63.0% (line RU0501139) to 76.9% (line RU0502068) in 

2006. Mean (averaged across genotypes) total milled rice 

percentage was 68.0% in 2005 and 71.5% in 2006. The results clearly 

indicate that better milling traits were obtained in 2006 compared 

R.E. Tabien et al. / Journal of Cereal Science 49 (2009) 122–127 

to 2005, and this can be due to differences in environmental factors 

such as temperature, rainfall, and relative humidity (Banaszek and 

Siebenmorgen, 1990; Cooper et al., 2006; Jodari and Linscombe, 

1996; Thompson and Mutters, 2006). The reported significant 

interaction of year and genotype (Gravois et al., 1991) was considered 

an important source of variation for head rice. 

3.2. Days to heading and milling percentages 

Regression analysis shown in Fig. 1 revealed a linear relationship 

of days to heading with head rice and total milled rice. Although the 

amount of variation that can be explained by this trait was low in 

2005 and much higher at 30–38% in 2006, the same negative and 

highly significant relationships were obtained in both years. The 

head rice and total milled rice percentages decreased with longer 

duration to heading. A reduction of 0.74–0.80% in head rice and 

0.35–0.44% for total milled rice is possible per day increase in days 

to heading. 

Based on the 2006 linear relationship, a desirable total milled 

rice percentage of 70% and head rice percentage of 55% should have 

83.44 days and 85.23 days to heading, respectively. Longer periods 

than these estimates will mean lower milling percentages. Very 

early heading genotypes are usually not selected in most rice 

breeding programs. In most cases, plants that head too early had 

fewer tillers and lighter biomass, thereby reducing grain yield (Xiao 

et al., 1998); these also had prolonged vegetative stage resulting in 

poor grain filling and lower grain yield. 

Grain yield and days to heading were positively correlated (data 

not shown); thus there should be a compromise to obtain both high 

yield and high total and head rice percentages. Using the two year 

data, correlation of heading to total and head rice percentages was 

very high, with r-values of 0.70 and 0.75, respectively. This indicates 

that heading may be useful for the indirect selection of high head 

rice and total milled rice percentages. However, one must consider 

its interaction with year and the potential impact of the environment. 

Actual use of this trait in selection will verify its effectiveness 

in improving milling traits. 

The overlapping of values for total and head rice as shown in 

Fig. 1, for instance, indicates variability if this trait will be used as 

selection index. These variations can be due to timing at which the 

50% heading has occurred. Using historical data of two cultivars, 

average daily low temperature (high night time temperature) at 

50% heading was shown to significantly affect head rice (Cooper 

et al., 2006). High night time temperature was reported to decrease 

Fig. 1. Relationship of number of days from emergence to heading, and head and total 

milled rice percentages in 105 long grain genotypes grown at Beaumont, Texas in 2005 

and 2006.

grain dimensions, increase number of chalky grains, lower amylose 

content, and increase length of amylopectin chains, thereby 

affecting milling qualities (Cooper et al., 2008; Counce et al., 2005). 

3.3. Flowering duration and milling percentages 

Flowering duration to 50% and 100%, and 50 to 100% flowering 

were negatively correlated with milling traits (Table 2). The highest 

correlation was found between FD0–100 and percentage head grain 

(r ¼ 0.464) followed by FD0–50 and head rice percentage 

(r ¼ 0.433). The FD50–100 was also significantly and negatively 

correlated with head rice and total milled rice percentages, but 

these correlation coefficients were lower than both FD0–50 and 

FD0–100. These relationships indicate that the duration to reach 50% 

or 100% flowering is important in obtaining high total and head rice 

percentages than the duration from 50–100%. Moreover, these 

indicate the need for shorter duration of flowering to insure better 

total and head grain percentages. The negative relationship 

between flowering duration and milled rice percentages is shown 

in Fig. 2. There was a 2.4% and 1.1% decrease in head rice and total 

milled, respectively for every one day increase in flower duration. 

Grains from the rice panicle or from the entire plant, develop 

and mature asynchronously. At harvest, the distribution of MC of 

grains is multi-modal (Bautista and Siebenmorgen, 2005; Kocher 

et al., 1990). The range of grain MC decreases as the MC at harvest 

decreases suggesting that the ranges of MC are directly related to 

the maturity of the grains (Bautista and Siebenmorgen, 2005). 

These ranges of variation will be more at prolonged flowering 

duration since the maturity of the grains will be different. The 

differences in flowering result in wider variation in grain development, 

and the final weight and quality of the grains (Wang et al., 

2006). Any non-uniformity expressed in variable grain size and 

shape, grain filling, and maturity has negative effects on rice milling 

traits (Jongkaewwattana and Geng, 2001). 

Rice breeders commonly estimate the number of days from 

emergence to heading of their breeding lines. Adding days to start 

of flowering to the list of evaluation criteria will enable the estimation 

of FD0–50 for indirect selection criterion for relatively higher 

head rice and total milled rice percentage. The mean of 105 genotypes 

for FD0–50 and FD0–100, were 3.3 and 5.5 days, respectively, 

and the mean head rice and total milled rice percentages (averaged 

across years and genotypes) were 55.2 and 69.8%, respectively. 

Based on the regression equations obtained, the FD0–50 and FD0–100 

should be at less than 5.5 days in order to produce higher than 

average head rice and total milled rice percentages. 

Unaffected by year and the genotype year interaction, the trait 

of flowering duration flowering can be used in indirect selection 

relative to the number of days to heading. Although the correlation 

was lower compared to heading date, this trait can be a more 

reliable indicator of milling potentials. Actual use in the breeding 

program will prove its efficiency in improving milling traits. 

Table 2 

Correlation between flowering-related traits and milling percentages of 105 long 

grain rice genotypes grown at Beaumont, Texas in 2005 and 2006 

Flowering duration Correlation coefficient (r) 

Whole milled rice Total milled rice 

Start of flowering to 50% flowering 0.43** 0.43** 

50–100% flowering 0.27** 0.190** 

Start to 100% flowering 0.46** 0.42** 

Heading date 0.70** 0.75** 

Flowering rate þ0.42** þ0.40** 

Duration for heading to maturity 

** Significant at the 1% probability level. 

þ0.18** þ0.43** 

R.E. Tabien et al. / Journal of Cereal Science 49 (2009) 122–127 125 

Fig. 2. Relationship of flowering duration and whole and total milled rice percentages 

in 105 long grain genotypes grown at Beaumont, Texas in 2005 and 2006. 

3.4. Rate of flowering and milling percentages 

The rate of flowering, similar to the duration of flowering, was 

linearly and positively related to both head rice and total milled rice 

percentages as shown in Fig. 3. Increasing rate of flowering was 

favorable in improving milling traits. This result indicates that 

a genotype should reach 100% flowering at a faster rate in order to 

produce high quality grain for milling. The relationship of rate of 

flowering and milling can be explained by the uniformity of grains 

at harvest. If the genotype has a fast rate of flowering, the grains 

will be filled nearly at the same time and will be in a near-uniform 

stage during harvest. Synchronous grain development related to 

grain filling and maturity has a positive impact on milling qualities 

(Jongkaewwattana and Geng, 2001). Correlation of flowering rate 

to milling percentages was significant with r-values comparable to 

that of duration of flowering, thus indicating the former’s potential 

for indirect selection. Estimation of flowering rate, however, needs 

several data points compared to the simpler gathering of data for 

duration of flowering. 

3.5. Duration of heading to maturity and milling percentages 

The number of days from heading to maturity did not have 

a significant relationship with head rice percentages in 2005 and 

2006. However, there were significant linear relationships between 

Fig. 3. Relationship of flowering rate and whole and total milled rice percentages in 

105 long grain genotypes grown at Beaumont, Texas in 2005.

126 

the number of days from heading to maturity and total milled rice 

percentages in 2005 and 2006 (Fig. 4). For each day that the 

number of days from heading to maturity increased, total milled 

rice percentage increased by 0.24% in 2005 and 0.29% in 2006. The 

increase in percent total grain at longer duration from flowering to 

harvest can be attributed to better grain filling. It was observed that 

grain filling affected grain traits such as weight and density, shape 

and size, which, in turn, affected milling traits (Siebenmorgen et al., 

1992). Moreover, at 40 days after flowering, most of the grains had 

20% moisture content, and maximum total and head rice grain was 

obtained at 12–22%. An opposite relationship was noted in Californian 

cultivars with different maturities. There was a quadratic 

relationship between maturity and head rice grain, and the 

percentage head rice increased from early and intermediate to late 

and very late maturing cultivars (Jongkaewwattana and Geng, 

2002). This difference can be attributed to grain shape and size. 

Most of the Californian cultivars were medium grain, but this 

study focused on long grains that are popular in the Gulf Coast. 

Variation in grain size and shape contributed to the variation in rate 

and duration of grain filling (Jongkaewwattana and Geng, 2001) 

and ultimately milling quality. It was reported that late-heading 

entries had larger panicle, and the date of heading was found to be 

correlated with the number of grains per panicle (0.52) and 

spikelets per panicle (0.42) (Zhang et al., 2006). These results 

further support increased grain variations in late heading 

genotypes. 

Number of days to heading and maturity are important for rice 

adaptation to certain cultivation areas and to cropping seasons in 

most rice growing countries. Shorter maturity or early flowering is 

needed in rainfed and upland areas and during the rainy season. 

However, a relatively longer maturity is needed to have higher 

grain yield in irrigated areas during the dry season planting. 

Therefore, it is important to consider the number of days to heading 

and maturity in selecting for higher total and head rice percentages, 

and high grain yield. 

3.6. Duration of flowering and the number of days from emergence 

to 50% flowering 

The number of days from emergence to heading had a significant 

linear relationship with flowering duration (Fig. 5). The 

number of days from the start of flowering to 50% was longer at 

prolonged duration to heading (50% flowering). Across 2005 and 

2006, the daily flowering duration increased by 0.12 days as the 

number of days from emergence to heading increased. This 

Fig. 4. Relationship of number of days from heading and maturity on total milled rice 

percentages in 105 long grain genotypes grown at Beaumont, Texas in 2005 and 2006. 

R.E. Tabien et al. / Journal of Cereal Science 49 (2009) 122–127 

Fig. 5. Relationship of number of days from emergence to start of flowering and the 

flowering duration in 105 long grain genotypes grown at Beaumont, Texas in 2005 and 

2006. 

relationship suggests that late flowering genotypes should be 

avoided to have genotypes with shorter duration of flowering. Early 

flowering genotypes should give better total and head grain 

percentages considering the positive relationship of heading date 

and the duration of flowering, and the negative relationship of 

milling traits and the duration of flowering. Breeders are not 

focusing on very early genotypes because of lower grain yields but 

since earliness is related to duration of flowering and the duration 

of flowering is correlated with milling traits, a trade-off is necessary. 

Based on the linear relationships obtained, a line should 

flower between 80 and 85 days. The US-released varieties included 

in the trials had 72–86 days to 50% flowering and the majority of 

cultivars were within the estimated range (data not shown). 

Based on these results, flowering-related traits such as days to 

heading, rate of flowering, duration of flowering, and duration from 

heading to maturity can impact head rice and total milled rice 

percentages. These milling traits were positively related with rate 

of flowering, and duration from heading to maturity but negatively 

related with duration of flowering (0–50%, 50–100% and 0–100%) 

and days to heading. Rice genotypes with early heading had relatively 

shorter flowering durations, and genotypes with shorter 

flowering duration had higher head rice and total milled rice 

percentages. Faster duration to attain 100% flowering and more 

days from heading to maturity were favorable in increasing 

percentage head rice and total grain. Estimating the duration from 

start of flowering to heading or to 100% flowering can be potential 

selection criteria in the indirect evaluation and selection of 

breeding lines for high head rice and total milled rice percentages. 

The number of days from heading to maturity can be considered in 

indirectly selecting for high total milled rice percentage. 


The authors would like to acknowledge the financial support 

from Texas Rice Research Foundation and the field and laboratory 

assistance, especially the harvesting and milling of samples, 

provided by Chersty Harper, Patrick Frank, Joel Pace, Richard Boyd, 

and Pat Carre. 

References 

Banaszek, M.M., Siebenmorgen, T.J., 1990. Head rice yield reduction rates caused by 

moisture adsorption. Transactions of the ASAE 33, 1263–1269. 

Bautista, R.C., Siebenmorgen, T.J., 2005. Individual kernel rice moisture content 

variability trends. Applied Engineering in Agriculture 21 (4), 637–643.

Bautista, R.C., Siebenmorgen, T.J., 2002. Evaluation of laboratory mills for milling 

small samples of rice. Applied Engineering in Agriculture 18 (5), 577–583. 

Bautista, R.C., Siebenmorgen, T.J., Counce, P.A., 2007. Rice kernel dimensional variability 

trends. Applied Engineering in Agriculture 23 (2), 207–217. 

Chau, N.N., Kunze, O.R., 1982. Moisture content variation among harvested rice 

grains. Transactions of the ASAE 25, 1037–1040. 

Cheng, F.M., Zhang, Q.F., Zhu, H.J., Zhao, N.C., Wang, F., Chen, K.M., Zhang, G.P., 2007. 

The difference in amylose content within a panicle as affected by the panicle 

morphology of rice cultivars. Journal of Cereal Science 46, 49–57. 

Cnossen, A.G., Jiménez, M.J., Siebenmorgen, T.J., 2003. Rice fissuring response to high 

drying and tempering temperatures. Journal of Food Engineering 59, 61–69. 

Cooper, N.T.W., Siebenmorgen, T.J., Counce, P.A., Meullenet, J.-F., 2006. Explaining 

rice milling quality variation using a historical weather data analysis. Cereal 

Chemistry 83 (4), 447–450. 

Cooper, N.T.W., Siebenmorgen, T.J., Counce, P.A., 2008. Effects of night time 

temperature during kernel development on rice physicochemical properties. 

Cereal Chemistry 83 (4), 447–450. 

Counce, P.A., Siebenmorgen, T.J.,Vories, E.D.,Pitts, D.J.,1990. Time ofdraining and harvest 

effects on rice grain yield and quality. Journal of Production Agriculture 3, 436–445. 

Counce, P.A., Bryant, R.J., Bergman, C.J., Bautista, R.C., Wang, Y.-J., Siebenmorgen, T.J., 

Moldenhauer, K.A.K., Meullenet, J.-F.C., 2005. Rice milling quality, grain 

dimensions, and starch branching as affected by high night temperatures. 

Cereal Chemistry 82 (6), 645–648. 

Dilday, R.H., 1989. Milling quality of rice: cylinder speed vs. grain-moisture content 

at harvest. Crop Science 29, 1532–1535. 

Fan, J., Siebenmorgen, T.J., Yang, W., 2000. A study of head rice yield reduction of 

long and medium grain rice varieties in relation to various harvest and drying 

conditions. Transactions of the ASAE 43, 1709–1714. 

Gravois, K.A., Moldenhauer, K.A.K., Rohman, P.C., 1991. Genotypic and genotype 

environment effects for rough rice and head rice yields. Crop Science 31, 907–911. 

Gravois, K.A., Helms, R.S., 1996. Seeding rate effects on rough rice yield, head rice, 

and total milled rice. Agronomy Journal 88, 82–84. 

Holloway, G.E., Siebenmorgen, T.J., Counce, P.A., Lu, R., 1995. Causes of multi-modal 

moisture content frequency distribution among rice kernels. Applied Engineering 

in Agriculture 11 (4), 561–565. 

Jodari, F., Linscombe, S.D., 1996. Grain fissuring and milling yields of rice cultivars as 

influenced by environmental conditions. Crop Science 36, 1496–1502. 

Jongkaewwattana, S., Geng, S., 2001. Inter-relationships amongst grain characteristics, 

grain-filling parameters and rice (Oryza sativa L.) milling quality. Journal 

of Agronomy and Crop Science 187, 223–229. 

Jongkaewwattana, S., Geng, S., 2002. Non-uniformity of grain characteristics and 

milling quality of California rice (Oryza sativa L.) of different maturities. Journal 

of Agronomy and Crop Science 188, 161–167. 

Jongkaewwattana, S., Geng, S., Brandon, D.M., Hill, J.E., 1993. Effect of nitrogen and 

harvest grain moisture on head rice yield. Agronomy Journal 85, 1143–1146. 

Kocher, M.F., Siebenmorgen, T.J., Norman, R.J., Wells, B.R., 1990. Rice kernel moisture 

content variation at harvest. Transactions of the ASAE 33 (2), 541–548. 

Leesawatwong, M., Jamjod, S., Kuo, J., Dell, B., Rerkasem, B., 2005. Nitrogen fertilizer 

increases seed protein and milling quality of rice. Cereal Chemistry 82, 588–593. 

Lin, H.X., Liang, Z.W., Sasaki, T., Yano, M., 2003. Fine mapping and characterization 

of quantitative trait loci Hd4 and Hd5 controlling heading date in rice. Breeding 


Liu, Z.H., Cheng, F.M., Zhang, G.P., 2005. Positional variations in phytic acid and 

protein content within a panicle of japonica rice. Journal of Cereal Science 41, 

297–303. 

R.E. Tabien et al. / Journal of Cereal Science 49 (2009) 122–127 127 

Lloyd, B.B., Cnossen, A.G., Siebenmorgen, T.J., 2001. Evaluation of two methods for 

separating head rice from brokens for head rice yield determination. Applied 

Engineering in Agriculture 17 (5), 643–648. 

Lu, R., Siebenmorgen, T.J., 1995. Correlation of head rice yield to selected physical 

and mechanical properties of rice kernels. Transaction of the ASAE 38 (2), 

889–894. 

Luh, S., Luh, B., 1991. Properties of rice caryopsis. In: Rice Production, vol. 2. Van 

Nostrand Reinhold, New York. 394. 

McCauley, G.N., Way, M.O., 2002. Drain and harvest timing affects on rice grain 

drying and whole-milled grain. Field Crops Research 74, 163–172. 

Pan, Z., Thompson, J.F., Amarantunga, K.S.P., Anderson, T., Zheng, X., 2005. Effect of 

cooling methods and milling procedures on the appraisal of rice milling quality. 

Transactions of the ASAE 48 (5), 1865–1871. 

Perez, C.M., Juliano, B.O., Liboon, S.P., Alcantara, J.M., Cassman, K.G., 1996. Effects of 

late nitrogen fertilizer application on head rice yield, protein content, and grain 

quality of rice. Cereal Chemistry 73, 556–560. 

Reid, J.D., Siebenmorgen, T.J., Mauromoustakos, A., 1998. Factors affecting the slope 

of head rice yield vs. degree of milling. Cereal Chemistry 75, 738–741. 

Samonte, S.O.P.B., Wilson, L.T., McClung, A.M., Tabien, R.E., 2006. Selection for stable 

high income grossing rice genotypes. In: Proceedings, 31st Rice Technical 

Working Group, p. 60. The Woodlands, Texas. 

Schluterman, D.A., Siebenmorgen, T.J., 2007. Relating rough rice moisture content 

reduction and tempering duration to head rice yield reduction. Transactions of 

the ASABE 50 (1), 137–142. 

Seetanum, W., De Datta, S.K., 1973. Grain yield, milling quality and seed viability of 

rice as influenced by time of nitrogen application and time of harvest. 

Agronomy Journal 65, 390–394. 

Siebenmorgen, T.J., Jindal, V.K., 1986. Effects of moisture adsorption on the head rice 

yields of long-grain rough rice. Transactions of the ASAE 29, 1767–1771. 

Siebenmorgen, T.J., Counce, P.A., Lu, R., Kocher, M.R., 1992. Correlation of head rice 

yield with individual kernel moisture content distribution at harvest. Transactions 

of the ASAE 35, 1879–1884. 

Siebenmorgen, T.J., Bautista, R.C., Counce, P.A., 2007. Optimal harvest moisture 

contents for maximizing milling quality of long- and medium-grain rice cultivars. 

Applied Engineering in Agriculture 23 (4), 517–527. 

Thompson, J.F., Mutters, R.G., 2006. Effect of weather and rice moisture at harvest 

on milling quality of California medium grain rice. Transactions of the ASABE 

49 (2), 435–440. 

Wang, F., Chen, S., Cheng, F., Liu, Y., Zhang, G., 2007. Differences in grain weight and 

quality within a rice (Oryza sativa L.) panicle as affected by panicle type and 

source-sink relation. Journal of Agronomy and Crop Science 193, 63–73. 

Xiao, J., Li, L., Grandillo, S., Ahn, S.N., Yuan, L.P., Tanksley, S.D., McCouch, S.R., 1998. 

Identification of trait improving quantitative loci alleles from wild rice relatives, 

Oryza rufipogon. Genetics 150, 899–909. 

Xiong, L., Lui, K., Dai, X., Xu, C., Zhang, Q., 1999. Identification of genetics factors 

controlling domestication-related traits in rice using F2 population of 

a cross between Oryza sativa and O. rufipogon. Theoretical and Applied 

Genetics 98, 243–251. 

Zhang, Y., Luo, L., Xu, C., Zhang, Q., Xing, Y., 2006. Quantitative trait loci for panicle 

size, heading date and plant height co-segregating in trait performance 

derived near isogenic lines of rice (Oryza sativa). Theoretical and Applied 

Genetics 113, 361–368. 

Zhuang, J.Y., Lin, H.X., Lu, J., Qian, H.R., Hittalmani, S., Huang, N., Zheng, K.L., 1997. 

Analysis of QTL x environment interaction for yield components and plant 

height. Theoretical and Applied Genetics 95, 799–808.

Texture, processing and organoleptic properties of chickpea-fortified spaghetti 

with insights to the underlying mechanisms of traditional durum pasta quality 

Jennifer Ann Wood * 

NSW Department of Primary Industries, Tamworth Agricultural Institute, 4 Marsden Park Road, Calala, NSW 2340, Australia 

article info 


Received 10 March 2008 



Keywords: 

Cicer arietinum L 

Pasta quality 

Rheology 

Sensory 


abstract 

Nutritionists recommend pulses (grain legumes) such as 

chickpea (Cicer arietinum) in the diet as they have many nutritional 

benefits. Chickpea seed has a high protein digestibility, contains 

high levels of complex carbohydrates (low glycaemic index), is rich 

in vitamins and minerals and is relatively free from anti-nutritional 

factors (Muzquiz and Wood, 2007; Wood and Grusak, 2007). 

Consumers are becoming increasingly health conscious and 

while many admit to knowing pulses are good for them, they are 

not sure how to use them in their diet. There is also a perception 

that cooking pulses is difficult and/or time consuming. 

Australian production of desi chickpea is currently around 

200,000 tonnes per year (Knights et al., 2007), yet less than 0.5% of 

this is consumed by Australians. In comparison, pasta, produced 

from durum wheat (Triticum turgidum) flour, is consumed worldwide. 

It is relatively non-perishable, inexpensive, easy to prepare, 

and readily accepted by all age groups. 

The nutritional benefits, deep yellow–orange cotyledon colour 

and lack of offensive ‘beany’ flavours compared to other pulses 

make chickpea a suitable candidate for incorporation with durum 

* Tel.: þ61 2 67631157; fax: þ61 2 67631222. 

E-mail address: jenny.wood@dpi.nsw.gov.au 


doi:10.1016/j.jcs.2008.07.016 





Nutritionally enhanced spaghetti was prepared from durum semolina fortified with 0–30% desi chickpea 

‘besan’ flour. This study examined the dough rheology, processing ease and quality attributes of the 

fortified spaghetti including protein, starch, texture (firmness, resilience and stickiness), colour, cooking 

loss, and organoleptic acceptability. Chickpea-fortified spaghetti was acceptable to consumers, had 

reasonable pasta quality, including lower cooking loss and less stickiness than the control spaghetti and 

retained firmness better than durum after refrigeration. This study suggests that chickpea-fortified 

spaghetti may be suited to uses such as fresh pasta, in soups, canning, and microwave re-heating. In 

addition, this study has added to the understanding of the underlying mechanisms of pasta quality. The 

main findings were: (1) gluten content/composition appears to be more important than protein content 

for pasta firmness; (2) the protein–polysaccharide matrix appears to be more important than the starch 

composition for cooking loss; (3) increased protein and amylose contents were associated with 

decreased pasta stickiness; (4) cooking loss and stickiness were not necessarily as strongly related as 

commonly believed. Further research into these theories is necessary to fully understand the underlying 

mechanisms of pasta quality. 


into pasta. In this way consumers could increase their consumption 

of pulses with little extra effort or thought. 

The inclusion of pulses in cereal based foods is known to 

increase the nutritive value by improving protein content and 

lysine availability (Bahnassey et al., 1986; Hernandez and Sotelo, 

1984; Kurien et al., 1971; Reyes-Moreno et al., 2004; Siddique et al., 

1996; Wood and Grusak, 2007). Several studies have examined 

various aspects of chickpea incorporation into pasta (Goni and 

Valentin-Gamazo, 2003; Sabanis et al., 2006) however the endproduct 

qualities of pasta produced from ‘besan’ (dehulled desi 

chickpea flour) has not been thoroughly investigated. 

Goni and Valentin-Gamazo (2003) showed that spaghetti containing 

25% chickpea flour had a significantly lower glycaemic 

index (GI) than traditional durum spaghetti. Chickpea inclusion 

also increased the mineral and fat content without affecting the 

total starch content. Zhao et al. (2005) incorporated 5-30% of 

different pulse flours into spaghetti and found that firmness and 

colour intensity increased, while overall quality decreased. 

However, the chickpeas used in this study were kabuli types and 

the seed coats were not removed prior to grinding. Sabanis et al. 

(2006) investigated 5-50% inclusion of chickpea flour in durum 

lasagne and found that the physical properties of the dough were 

improved; however, processing, handling and cooking characteristics 

deteriorated with the higher substitution levels.

This study aimed to more comprehensively identify whether, 

besan, could be successfully incorporated into durum spaghetti to 

obtain a convenient health product with organoleptic properties 

acceptable to consumers. Furthermore, this research provides 

additional information to the existing literature on the effects of 

starch and protein on the quality of traditional durum pasta. 


2.1. Samples 

Unconditioned seed (50.0 g) of desi chickpea (cv. Amethyst) 

was dehulled in the ‘Sheller’ component (attrition-style) of an SK 

Engineering mill (SK Engineering and Allied Works, Bahraich, 

India) to produce dhal (split cotyledons) according to Wood et al. 

(2008). The resulting dhal was subsequently milled through a 

1.0 mm sieve in a Newport Scientific mill (Newport Scientific 

Pty Ltd, Narabeen, Australia) to obtain coarse, besan flour. 

Commercial durum semolina was obtained from Goodman Fielders 

(Tamworth, Australia) and blended with 0, 10, 15, 20, 25 and 30% of 

the chickpea flour by weight, in duplicate. No significant difference 

was found between the moisture contents of the blended flours 

as determined by the AACC approved method 44-15A (AACC, 

1995b). 

2.2. Dough quality and pasta colour evaluation 

Dough properties were determined with a farinograph and 

constant dough weight using the AACC approved method 54-21 

(AACC, 1995d). Pasta was produced under vacuum with standard 

33.3% water addition and dried in a pasta oven at low temperature 

according to Wood et al. (2001). Pasta colour (L*, a*, b*) of multiple 

layers of parallel spaghetti strands were measured (mean of six 

readings) using a Minolta chroma meter CR-310 (Minolta Camera 

Co., Osaka, Japan) on both dry and cooked spaghetti. The change in 

colour due to cooking was determined by calculating 

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 

DE* ¼ ðDL Þ 2 þðDa Þ 2 þðDb Þ 2 

q 

2.3. Cooked pasta quality evaluation 

The quality of the cooked spaghetti was evaluated according to 

Wood et al. (2001). Spaghetti strands (10 g, 7 cm lengths) were 

placed in 250 ml of vigorously boiling distilled water containing 

5 ml of stock solution (0.125 g NaHCO3, 175 g NaCl in 500 ml 

distilled H2O). Optimum cooking time of the spaghetti was recorded 

when the core in the middle of the spaghetti strand was no 

longer visible when squashed between two perspex sheets. Texture 

analysis (firmness, resilience and stickiness) and cooking loss were 

performed on spaghetti cooked to optimum cooking time according 

to Wood et al. (2001). Cooking loss was calculated as described 

by Matsuo et al. (1992). 

2.4. Rapid visco analyser (RVA) 

Spaghetti was milled into flour through a 0.5 mm sieve in 

a Newport Scientific mill (Newport Scientific Pty, Ltd, Narabeen, 

Australia). Moisture content of the flours and spaghetti blends were 

determined by the AACC approved method 44-15A (AACC, 1995b). 

Pasting properties were evaluated in duplicate using 29 g of sample 

(13.8% solids, moisture corrected) in the Rapid Visco Analyser 

(Newport Scientific Pty, Ltd, Narabeen, Australia). The temperature 

was held at 25 C for 2 min, increased to 95 C over 5 min, held 

at 95 C for 3 min and decreased to 25 C over 5 min. Peak 

J.A. Wood / Journal of Cereal Science 49 (2009) 128–133 129 

time, peak viscosity, breakdown, final viscosity and setback were 

recorded. 

2.5. Protein content and amino acid analysis 

Total protein content of the fortified products and controls were 

determined in duplicate with a LECO nitrogen analyser using AACC 

approved method 46-30 (AACC, 1995c). Protein determinations 

were based on nitrogen factors of 5.7 for wheat flour and 6.25 

for chickpea flour. Nitrogen factors for spaghetti blends were 

adjusted on the basis of the relative proportion of proteins in the 

ingredients. Following extraction by acid hydrolysis, amino acid 

analysis of the 0%, 15% and 30% fortified spaghetti were determined 

by HPLC using AACC approved method 07-01 (AACC, 1995a). 

Tryptophan content was not determined. 

2.6. Amylose content 

The amylose content of the flours and spaghetti were determined 

using the Megazyme amylose and amylopectin assay kit 

(Megazyme International, Ireland). 

2.7. Sensory evaluation 

Sensory evaluation was performed on cooked spaghetti cooled 

in distilled water (5 min), drained, covered with plastic, and 

refrigerated overnight. Spaghetti fortified with %0, %15 and 30% 

chickpea flour was randomly numbered and scored by 27 untrained 

volunteers the following day. Texture analysis was performed on 

this spaghetti as previously described. A hedonic scale was chosen 

for the sensory evaluation questionnaire. The volunteers were 

asked to give a ranking of the attributes (colour, firmness, flavour, 

overall acceptability) for the three samples on a linear scale. The 

distance from the undesirable end of the scale was measured for 

each sample and subjected to ANOVA. 


All the parameters were analysed in duplicate. The data was 

statistically analysed by analysis of variance (ANOVA). Each 

parameter was tested for significance (P < 0.05) between the 

fortified and control spaghetti samples. When significant differences 

were found, the least significant difference (LSD) test was 

used to determine the differences among means. A correlation 

coefficient between cooking loss and stickiness was determined 

using Microsoft Excel. 


3.1. Dough and processing properties 

Increasing chickpea fortification significantly (P < 0.05) decreased 

water absorption and resulted in longer development times and less 

stable doughs (Table 1). Similar farinograph trends have been reported 

previously for wheat incorporating legume flours or their protein 

concentrates (Lorenz et al., 1979; Rasmay et al., 2000; Yanez-Farias 

et al., 1999). Many of these effects can be attributed to weakening of 

the gluten matrix due to the incorporation of chickpea flour which 

contains no gluten. Chickpea proteins are comprised mainly of globulins 

(53–60%) with lesser concentrations of albumins, prolamins and 

glutelins (Dhawan et al., 1991). 

Fortification made the dough particles increasingly sticky, 

causing them to aggregate during mixing. This made extrusion to 

produce the spaghetti increasingly difficult. For this reason, 

spaghetti production above 30% fortification was not undertaken. 

Chickpea flour contains significant levels of soluble non-starch

130 

Table 1 

Effect of chickpea flour fortification on spaghetti quality parameters 

Quality parameters Fortification (chickpea %) P 

0% 10% 15% 20% 25% 30% 

Water absorption (%) 57.6 a 

57.6 a 

57.2 b 

56.3 c 

55.8 d 

54.2 e 

0.00 

DDT (min) 3.75 d 

4.50 c 

4.50 c 

4.50 c 

5.00 b 

5.75 a 

0.00 

B10 (BU) 50 d 

60 c 

70 b 

70 b 

75 ab 

80 a 

0.00 

Protein (%) 12.43 e 

14.33 d 

13.82 d 

15.37 c 

16.99 b 

17.42 a 

0.00 

Glutamine/glutamicacid (g/100 g) 4.81 a 

NA 4.48 b 

NA NA 4.22 c 

0.00 

Proline (g/100 g) 1.67 a 

NA 1.57 b 

NA NA 1.51 c 

0.00 

Lysine(g/100 g) 0.22 c 

NA 0.36 b 

NA NA 0.62 a 

0.00 

Cysteine (g/100 g) 0.33 a 

NA 0.34 a 

NA NA 0.33 a 

0.12 

Methionine (g/100 g) 0.21 a 

NA 0.21 a 

NA NA 0.20 a 

0.08 

Colour, L* (dry) 57.73 a 

57.25 ab 

55.98 abc 

55.20 bc 

54.61 c 

53.96 c 

0.00 

Colour, a* (dry) 3.56 c 

6.25 b 

7.13 ab 

7.61 ab 

8.00 a 

8.28 a 

0.00 

Colour, b* (dry) 40.74 a 

33.76 b 

36.33 ab 

36.72 ab 

37.86 ab 

38.74 ab 

0.04 

Colour, L* (cooked) 78.37 a 

75.85 b 

74.01 c 

73.10 cd 

72.33 d 

71.26 e 

0.00 

Colour, a* (cooked) 0.33 e 

3.42 d 

4.66 c 

5.21 bc 

6.10 ab 

6.52 a 

0.00 

Colour, b* (cooked) 32.52 a 

27.03 c 

28.65 bc 

29.95 abc 

30.27 ab 

30.61 ab 

0.01 

Colour change, DE 32.74 a 

28.16 b 

28.18 b 

27.07 b 

27.21 b 

27.19 b 

0.01 

Amylose (%) 23.01 b 

22.52 b 

23.63 b 

23.54 b 

23.45 b 

26.65 a 

0.01 

Peak time (min) 8.50 a 

8.30 a 

8.30 a 

7.73 b 

7.70 b 

7.70 b 

0.02 

Peak viscosity (cP) 2191 c 

2303 b 

2231 c 

2246 bc 

2518 a 

2529 a 

0.00 

Breakdown (cP) 922 c 

957 c 

959 c 

965 c 

1089 b 

1180 a 

0.00 

Final viscosity (cP) 5976 a 

5802 c 

5891 b 

4532 e 

5345 d 

5407 d 

0.00 

Setback (cP) 4707 a 

4456 b 

4619 a 

3250 d 

3916 c 

4058 c 

0.03 

Firmness (g) 624.6 a 

641.1 ab 

671.6 a 

574.8 bc 

550.9 cd 

510.1 d 

0.01 

Resilience (g) 62.28 a 

46.02 ab 

51.81 a 

33.07 bc 

15.45 c 

13.93 c 

0.00 

Stickiness (gs) 5.91 a 

4.72 bc 

4.74 bc 

4.67 bc 

4.81 b 

3.81 c 

0.00 

Cooking loss (%) 5.15 a 

4.84 ab 

4.64 ab 

4.66 ab 

4.53 b 

4.50 b 

0.03 

NA, not analysed. For each quality parameter, means within rows followed by different letters are significantly different (P < 0.05) by least significant difference (LSD) test. 

polysaccharides (NSP), about ten times higher than bread wheat 

flour (Naivikul and D’Appolonia, 1979), and this may have 

contributed to the increased stickiness. Based on these observations 

and the lower farinograph water absorptions of the chickpeafortified 

doughs, it is possible that reducing the amount of water 

added may improve mixing and extrudability to some extent. 

3.2. Dry pasta properties 

Protein content and amino acid composition generally increased 

(P < 0.05) with fortification. Protein content of control spaghetti was 

12.4% compared to 17.4% for the 30% fortified spaghetti (Table 1). A 

similar trend was observed by Zhao et al. (2005). Chickpea blends 

showed increased levels of all amino acids except for cysteine and 

methionine (no significant change) and glutamine/glutamic acid 

and proline which significantly (P < 0.05) decreased with fortification 

(Table 1). Chickpea flour is higher in protein but limiting in 

cysteine and methionine, hence no significant change in their 

contents were observed in the fortified spaghetti. Glutamine/glutamic 

acid and proline are indicative of gluten-type proteins which 

explains their lower levels since chickpea flour is gluten free. On the 

other hand, wheat is limiting in lysine. Lysine content significantly 

(P < 0.05) increased by 64 and 182% in the 15 and 30% fortified 

spaghetti respectively (Fig. 1). 

The colour of all the dried spaghetti generally became significantly 

(P < 0.05) less bright (lower L*), more red (higher a*) and less 

yellow (lower b*) as the percentage of chickpea flour increased 

(Fig. 2). The 10% and 15% chickpea-fortified spaghetti were rated by 

the sensory panel as visually similar to the control, however the 

spaghetti fortified with 20% or more tended to display an undesirable 

brownish tint (supported by L*a*b* measurements, Table 1). 

This is similar to the findings of Zhao et al. (2005). 

3.3. Rapid visco analysis 

The chickpea besan flour slurry had a much lower and flatter RVA 

viscosity profile than the durum semolina slurry (Fig. 3). Wood 

J.A. Wood / Journal of Cereal Science 49 (2009) 128–133 

(2004) found similar comparisons in RVA profiles between chickpea 

besan flour (cv. Howzat) and a hard bread wheat flour (cv. Drysdale). 

Chickpea has a lower starch content with a higher proportion of 

amylose and a greater non-starch polysaccharide (NSP) content and 

lipid content than durum and bread wheats (Fabriani and Lintas, 

1988; Wood and Grusak, 2007). This probably explains much of the 

differences in the RVA profiles of the besan and semolina slurries. 

The 10% and 15% fortified spaghetti displayed similar pasting 

profiles to the control spaghetti. They generally had higher peak 

viscosities and lower final viscosities than the control spaghetti 

(Table 1). The higher peak viscosities indicate larger water binding 

capacities of the starches in these blends. The 25% and 30% fortified 

spaghetti peaked earlier and higher than the control spaghetti with 

slightly lower final viscosities and much lower setbacks. The slurries 

of these blends swelled faster and bound more water; however, 

less retrogradation occurred, resulting in a less viscous gel after 

cooking and cooling. Most of these observations can be explained 

by the higher NSP content of the fortified spaghetti. NSP has a high 

Lysine Content (g / 100 g) 

0.7 

0.6 

0.5 

0.4 

0.3 

0.2 

0.1 

0.0 

Lysine Content 

Protein Content 

0 15 30 

% Chickpea in pasta 

Fig. 1. Protein and lysine content of spaghetti increases with desi chickpea flour 

fortification. 

17 

16 

15 

14 

13 

12 

11 

10 

Protein Content (%)

L* (brightness) 

60 

59 

58 

57 

56 

55 

54 

53 

52 

51 

50 

L 

b 

0 10 15 20 25 30 

% Chickpea in pasta (dry) 

Fig. 2. Colour changes in dry spaghetti with increasing desi chickpea flour fortification. 

water absorbing capacity which has been shown to increase the 

viscosity of bread wheat slurries (Sasaki et al., 2000). On cooling, 

the NSP can impede the formation of a three dimensional starch 

network, preventing hydrogen bonds between amylose and 

amylopectin from forming during retrogradation, resulting in a less 

viscous gel (Kim and D’Appolonia, 1977; Yoshimura et al., 1996). 

The 20% fortified spaghetti pasting curve did not track between 

the 10–15% and 25–30% spaghetti as expected (Fig. 3). Instead it 

peaked at a similar viscosity to the control and had a much lower 

final viscosity. This indicates that the starch granules in the blend 

swelled similarly to the control, but again with less retrogradation 

and reduced viscosity after cooking and cooling. The sample was reanalysed 

twice more but all generated the same pasting curve. This 

unexpected result is difficult to explain. There may be some sort of 

synergistic process between the besan and semolina constituents 

occurring at 20% fortification. 

Viscosity cP 

8000 

6000 

4000 

2000 

0 

0 4 8 12 16 20 

Time mins 

Semolina 

0% Pasta 

10% Pasta 

30% Pasta 

20% Pasta 

Chickpea 

Flour 

Fig. 3. RVA profile of raw flours (durum semolina and desi chickpea ‘course besan’) 

and spaghetti with different levels of chickpea fortification. 


120 

80 

40 

0 

41 

39 

37 

35 

33 

31 

29 

27 

25 

b* (yellowness) 

Temp 'C 

3.4. Cooked pasta properties 

The optimum cooking time was determined to be 10.5 minutes 

for all the spaghetti samples. After cooking, all the spaghetti 

samples were more bright (higher L*) and less yellow (lower b*) 

due to leaching and/or degradation of colour pigments such as 

carotenoids and xanthophyll. Bleaching was most pronounced in 

the control spaghetti, as calculated by DE* (Table 1). Despite this, 

the cooked chickpea-fortified spaghetti samples were generally less 

bright (L*), more red (a*) and less yellow (b*) than the control 

spaghetti (P > 0.05). However, there was no statistical difference 

between the 0%, 15% and 30% fortified spaghetti in panellist 

perceptions of cooked spaghetti colour. 

The 10% and 15% substituted spaghetti had similar firmness and 

resilience to the durum control spaghetti. Similarly, Zhao et al. 

(2005) found no significant difference in the firmness of most of 

their chickpea–spaghetti blends compared to the durum control. 

However, our results showed that firmness and resilience 

decreased with fortification of 20–30% (Table 1; Fig. 4). 

Increasing the protein content of durum spaghetti has been 

shown to increase firmness (Nobile et al., 2005; Sissons et al., 

2005). In addition, lowering the amylose content has been shown 

to decrease firmness (Gianibelli et al., 2005). This suggests that the 

fortified spaghetti firmness should have increased as protein 

content and amylose content increased due to fortification. This 

appears to be the case up to a threshold of 15% fortification. 

However, chickpea flour contains no gluten. As more chickpea flour 

was added, the semolina gluten was effectively diluted (observed as 

a decrease in glutamine/glutamic acid contents; P < 0.05) leading 

to weakening of the gluten matrix and a decrease in spaghetti 

firmness. This result suggests that it may not be the protein content 

per se (nor the high amylose content), rather the gluten content and 

possibly gluten composition that may be more important in 

determining spaghetti firmness. This finding is consistent with the 

work of Sissons et al. (2005) who showed that gluten content 

increased spaghetti firmness with no consistent trend relating to 

glutenin/gliadin composition. The larger NSP content of the 

chickpea flour may similarly contribute to weakening of the protein 

matrix. 

Legume starches generally contain more amylose than cereal 

starches. Zhao et al. (2005) found an increase in cooking loss 

(measured by solids loss) with fortification but this method is not 

comparable to amylose loss. The cooking loss test used in this study 

mainly measures the amount of amylose leached into the water 

during cooking (Matsuo et al., 1992). Sharma et al. (2002) and 

Gianibelli et al. (2005) found cooking loss to decrease in spaghetti 

with a low amylose content. Hence, it was initially presumed that 

the chickpea-fortified spaghetti would have increased cooking loss 

Firmness (g) 

700 

650 

600 

550 

500 

450 

400 

Firmness (g) 

Resilience (g) 

0 10 15 20 25 30 


Fig. 4. General decrease in firmness and resilience of spaghetti with increasing desi 

chickpea flour fortification. 

70 

60 

50 

40 

30 

20 

10 

0 

Resilience

132 

due to the higher amylose contents. However, cooking loss was 

shown to decrease with fortification (Table 1; Fig. 5). This result 

suggests that amylose content is not the major factor in cooking 

loss. 

Sissons et al. (2005) found no relationship of cooking loss with 

protein content in durum spaghetti. However, our results showed 

decreased cooking loss with increased protein and NSP contents of 

fortified spaghetti. This suggests that the protein–polysaccharide 

matrix (involving both starch and NSP) may be responsible for the 

retention of amylose during spaghetti cooking and not necessarily 

the starch composition per se, as suggested by Sissons et al. (2005). 

The chickpea-fortified spaghetti was also less sticky (P < 0.05) 

than the control (Fig. 5). Pasta surface stickiness is believed to be 

influenced by both the surface structure of the spaghetti strand and 

starch exuded onto the strand surface during cooking (Cunin et al., 

1995; Dexter et al., 1985; Perovic, 2000). Lowering the amylose 

content of spaghetti has been shown to increase stickiness (Gianibelli 

et al., 2005; Sharma et al., 2002). Furthermore, increasing 

the protein content of pasta has been associated with decreased 

stickiness (Nobile et al., 2005; Sissons et al., 2005). Hence, the 

reduced stickiness of the chickpea -fortified spaghetti is probably 

a result of both higher protein and higher amylose contents. 

However, whilst both the cooking loss and stickiness generally 

improved with chickpea fortification, they were only weakly 

correlated with one another (r ¼ 0.51). 

Since cooking loss, stickiness and firmness all decreased with 

the addition of besan chickpea flour, it appears that gluten, per se, 

may have little effect on the retention of amylose and other 

carbohydrates during cooking. The protein–polysaccharide matrix 

as a whole may be more likely to influence spaghetti quality in 

terms of firmness, stickiness and cooking loss. Additional components 

of chickpea flour that may aid the protein–polysaccharide 

matrix in carbohydrate retention include monomeric proteins and 

starch-bound phospholipids. This clearly requires further investigation 

and the knowledge may be useful for genotype selection in 

durum wheat breeding programs. 

3.5. Cooked and refrigerated pasta properties 

The sensory evaluation study gave an indication of consumer 

preference. Most panellists scored the 15 and 30% cooked fortified 

spaghetti to be equally acceptable as the durum control. No 

statistically significant (P < 0.05) difference could be found for 

colour, flavour or overall acceptability. However, the 30% blend was 

scored significantly (P < 0.05) firmer than the 0% and 15% blends. 

This was unexpected, as the texture analyser results of freshly 

cooked spaghetti showed a decreasing firmness with higher 

chickpea fortification. 

Stickiness (gs) 

6.0 

5.5 

5.0 

4.5 

4.0 

3.5 

Stickiness (gs) 

Cooking Loss (%) 

0 10 15 20 25 30 


Fig. 5. General decrease in cooking loss and stickiness of spaghetti with increasing 

desi chickpea flour fortification. 

J.A. Wood / Journal of Cereal Science 49 (2009) 128–133 

5.2 

5.1 

5.0 

4.9 

4.8 

4.7 

4.6 

4.5 

4.4 

% Cooking Loss 

Firmness (g) 

800.0 

700.0 

600.0 

500.0 

400.0 

300.0 

200.0 

100.0 

0.0 

When re-assessed by the texture analyser, the panel’s findings 

were confirmed. The fortified spaghetti were slightly firmer than 

the control after refrigeration (P < 0.05). Whilst all the spaghetti 

softened substantially after refrigeration, the effect was greater in 

the control durum spaghetti: a firmness decrease of 69% for the 

durum spaghetti compared to 66% for the fortified spaghetti 

(Fig. 6). Refrigeration differentially influenced the chickpea-fortified 

spaghetti and durum control spaghetti to alter the firmness 

trend. 

Riva et al. (2000) found that the rate and degree of starch 

retrogradation was exaggerated at cold temperatures. This too 

could cause firmer spaghetti, if the chickpea-fortified spaghetti 

were more predisposed to retrogradation under cold temperatures 

then the durum control. However, the RVA results showed the 

fortified spaghetti to have less retrogradation than the durum 

control. 

No difference in spaghetti diameter between the durum and 

fortified spaghetti was detected. However, water absorption may 

have occurred by diffusion through the gelatinised system to cause 

an increase in weight with negligible volume change (Riva et al., 

2000). If the chickpea blends were more susceptible to water 

absorption, spaghetti density would increase which may increase 

firmness. This is a possibility, as the RVA results showed the fortified 

spaghetti to have a larger water binding capacity during 

cooking and this may remain after cooling. This finding requires 

further investigation. 

4. Conclusion 

Normal Fridge 

100% Durum pasta 

20% Chickpea pasta 

Fig. 6. Effect of refrigeration on the firmness of 100% durum spaghetti and spaghetti 

fortified with 20% desi chickpea flour. 

Chickpea-fortified spaghetti was acceptable to consumers and 

provided an enhanced nutritional status via the amino acid profile. 

Lysine content increased by 64 and 182% in the 15 and 30% blends 

respectively whilst the total protein content and the content of 

most amino acids increased with fortification. Spaghetti processing 

and handling characteristics deteriorated as the level of fortification 

increased. Functional dough properties and spaghetti firmness 

were generally hindered by increasing amounts of chickpea flour. 

However, spaghetti stickiness improved with increasing fortification 

and cooking loss was reduced. Chickpea-fortified spaghetti 

retained firmness much better than durum after refrigeration. A 

marketing advantage may exist if this desirable firmness is retained 

when blended pasta is subjected to canning, microwave re-heating 

and inclusion in soups. This firmness retention property may also 

be of interest to fresh pasta manufacturers. Chickpea-fortified pasta 

would make a cheap, attractive and convenient health food which 

is acceptable to consumers in western society. In addition, this

study has added to the understanding of the underlying mechanisms 

of pasta quality. The findings were: (1) gluten content/ 

composition appears to be more important than protein content for 

pasta firmness; (2) the protein–polysaccharide matrix appears to 

be more important than the starch composition for cooking loss; 

(3) supportive of previous findings that increased protein and 

amylose contents are associated with decreased pasta stickiness; 

(4) cooking loss and stickiness are not necessarily as strongly 

related as commonly believed. Further research into these theories 

is necessary to fully understand the underlying mechanisms of 

pasta quality. 


The author wishes to thank the Grains Research and Development 

Corporation (GRDC) for funding (DAN402), Goodman 

Fielders, Tamworth for providing semolina, L. Ayre (NSW Department 

of Primary Industries, Tamworth) for statistical analyses, and 

M. Sissons (NSW Department of Primary Industries, Tamworth) for 

use of the pasta extruder/dryer. 

References 

American Association of Cereal Chemists (AACC), 1995a. Approved Methods of the 

AACC. Method 07-01, ninth ed. AACC, St Paul, MN. 

American Association of Cereal Chemists (AACC), 1995b. Approved Methods of the 

AACC. Method 44-15A, ninth ed. AACC, St Paul, MN. 

American Association of Cereal Chemists (AACC), 1995c. Approved Methods of 

the AACC. Method 46-30, ninth ed. AACC, St Paul, MN. 

American Association of Cereal Chemists (AACC), 1995d. Approved Methods of the 

AACC. Method 54-21, ninth ed. AACC, St Paul, MN. 

Bahnassey, Y., Khan, K., Harrold, R., 1986. Fortification of spaghetti with edible 

legumes. 1. Physicochemical, antinutritional, amino acid, and mineral composition. 

Cereal Chem 63, 210–215. 

Cunin, C., Handschin, S., Walther, P., Escher, F., 1995. Structural changes of starch 

during cooking of durum wheat pasta. Lebensm. Wiss. Technol 28, 323–328. 

Dexter, J.E., Matsuo, R.R., MacGregor, A.W., 1985. Relationship of instrumental 

assessment of spaghetti cooking quality to the type and the amount of material 

rinsed from cooked spaghetti. J. Cereal Sci 3, 39–53. 

Dhawan, K., Malhotra, S., Dahiya, B., Singh, D., 1991. Seed protein fractions and amino 

acid composition in gram (Cicer arietinum). Plant Food Hum. Nutr 41, 225–232. 

Fabriani, G., Lintas, C. (Eds.), 1988. Durum Wheat: Chemistry and Technology. AACC, 

St. Paul, MN. 

Gianibelli, M.C., Sissons, M.J., Batey, I.L., 2005. Effect of source and proportion of 

waxy starches on pasta cooking quality. Cereal Chem 82, 321–327. 

Goni, I., Valentin-Gamazo, C., 2003. Chickpea flour ingredient slows glycemic 

response to pasta in healthy volunteers. Food Chem 81, 511–515. 

Hernandez, M., Sotelo, A., 1984. Nutritional evaluation of wheat flour cookies 

supplemented with chickpea, cheese whey and amino acids. Nutr. Rep. Int 29, 

845–858. 

Kim, S.K., D’Appolonia, B.L., 1977. Effect of pentosans on the retrogradation of wheat 

starch gels. Cereal Chem 54, 150–160. 


Knights, E.J., Açikgöz, N., Warkentin, T., Bejiga, G., Yadav, S.S., Sandhu, J.S., 2007. 

Area, production and distribution. In: Yadav, S.S., Redden, B., Chen, W., 

Sharma, B. (Eds.), Chickpea Breeding and Management. CAB International, 

Wallingford, UK, pp. 167–178. 

Kurien, S., Narayanaswamy, D., Daniel, V.A., Swaminathan, M., Parpia, H.A., 1971. 

Supplementary value of pigeonpea (Cajanus cajan) and chickpea to poor diets 

based on kaffir corn and wheat. Nutr. Rep. Int 4, 229–236. 

Lorenz, K., Dilsaver, W., Wolt, M., 1979. Fababean flour and protein concentrate in 

baked goods and in pasta products. Bakers’ Dig 53, 39–45. 

Matsuo, R.R., Malcolmson, L.J., Edwards, N.M., Dexter, J.E., 1992. A colorimetric 

method for estimating spaghetti cooking losses. Cereal Chem 69, 27–29. 

Muzquiz, M., Wood, J.A., 2007. Antinutritional factors. In: Yadav, S.S., Redden, B., 

Chen, W., Sharma, B. (Eds.), Chickpea Breeding and Management. CAB International, 

Wallingford, UK, pp. 143–166. 

Naivikul, O., D’Appolonia, B.L., 1979. Carbohydrates of legume flours compared with 

wheat flour. 3. Nonstarchy polysaccharides. Cereal Chem 56, 45–49. 

Nobile, M.A., Baiano, A., Conte, A., Mocci, G., 2005. Influence of protein content on 

spaghetti cooking quality. J. Cereal Sci 41, 347–356. 

Perovic, B., 2000. Role of starches in pasta cooking. Tecnica Molitoria 51, 726–731. 

Rasmay, N.M.H., El-Shatanovi, G.A., Hassan, K.E.W., 2000. High-protein macaroni 

from legume flours and their protein concentrates. Ann. Agric. Sci 45, 555–570. 

Reyes-Moreno, C., Cuevas-Rodriguez, E.O., Milan-Carrillo, J., Cardenas- 

Valenzuela, O.G., Barron-Hoyos, J., 2004. Solid state fermentation process for 

producing chickpea (Cicer arietinum L.) tempeh flour: physicochemical and 

nutritional characteristics of the product. J. Sci. Food Agric 84, 271–278. 

Riva, M., Fessas, D., Schiraldi, A., 2000. Starch retrogradation in cooked pasta and 

rice. Cereal Chem 77, 433–438. 

Sabanis, D., Makri, E., Doxastakis, G., 2006. Effect of durum flour enrichment with 

chickpea flour on the characteristics of dough and lasagne. J. Sci. Food Agric 86, 

1938–1944. 

Sasaki, T., Yasui, T., Matsuki, J., 2000. Influence of non-starch polysaccharides isolated 

from wheat flour on the gelatinization and gelation of wheat starches. 

Food Hydrocolloids 14, 295–303. 

Sharma, R., Sissons, M.J., Rathjen, A.J., Jenner, C.F., 2002. The null-4A allele at the 

waxy locus in durum wheat affects pasta cooking quality. J. Cereal Sci 35, 

287–297. 

Siddique, I.M., Awan, J.A., Salim-ur-RehmanGilani, A.H., Ali, M., 1996. Evaluation of 

composite flour based on wheat (Triticum aestivum) and chickpea (Cicer arietinum) 

for chapati making. Sci. Int 8, 165–168. 

Sissons, M.J., Egan, N.E., Gianibelli, M.C., 2005. New insights into the role of gluten 

on durum pasta quality using reconstitution method. Cereal Chem 82, 601–608. 

Wood, J.A., 2004. RVAÔ Viscosity of Chickpea Flours. Tech. J. Newport Sci 1, 1. 

Wood, J.A., Batey, I.L., Hare, R.A., Sissons, M.J., 2001. A comparison of Australian and 

imported spaghetti. Food Aust 53, 349–354. 

Wood, J.A., Grusak, M.A., 2007. Nutritional Value of Chickpea. In: Yadav, S.S., 

Redden, B., Chen, W., Sharma, B. (Eds.), Chickpea Breeding and Management. 

CAB International, Wallingford, UK, pp. 101–142. 

Wood, J.A., Knights, E.J., Harden, S., 2008. Milling performance in desi type chickpea 

(Cicer arietinum L.): effects of genotype, environment and seed size. J. Sci. Food 

Agric 88, 108–115. 

Yanez-Farias, G.A., Bernal-Aguilar, V., Ramirez-Rodriguez, L., Barron-Hoyos, J.M., 

1999. Fortification of some cereal foods with a chickpea protein concentrate. 

Food Sci. Tech. Int 5, 89–93. 

Yoshimura, M., Takaya, T., Nishinari, K., 1996. Effects of konjac-glucomannan on the 

gelatinization and retrogradation of corn starch as determined by rheology and 

differential scanning calorimetry. J. Agric. Food Chem 44, 2970–2976. 

Zhao, Y.H., Manthey, F.A., Chang, S.K.C., Hou, H.J., Yuan, S.H., 2005. Quality characteristics 

of spaghetti as affected by green and yellow pea, lentil, and chickpea 

flours. J. Food Sci 70, S371–S376.

Impact of re-grinding on hydration properties and surface composition 

of wheat flour 

M. Mohamad Saad a , C. Gaiani a, *, J. Scher a , B. Cuq b , J.J. Ehrhardt c , S. Desobry a 

a LSGA, Laboratoire de Sciences et Génie Alimentaires, Nancy University, 2 avenue de la Forêt de Haye, B.P. 172, 54505 Vandoeuvre Les Nancy Cedex, France 

b Montpellier SupAgro, INRA, UMR 1208 Unit for Emerging Technology and Polymer Engineering, 2, place Viala, 34060 Montpellier Cedex 1, France 

c LCPME, Laboratoire de Chimie Physique et de Microbiologie pour l’Environnement, Nancy Université, CNRS, 405, rue de Vandoeuvre, 54600 Villers Lès Nancy, France 

article info 


Received 30 November 2007 


Accepted 5 August 2008 

Keywords: 

DVS 

Sorption isotherm 

Wheat flour 

XPS 


abstract 

One of the major control variables in food preservation technology 

is water activity (aw). This term indicating the ‘‘quality’’ of 

water content in food, describes the degree of ‘‘boundness’’ of 

water and thus its availability to participate in physical, chemical, 

and microbiological reactions (Brunauer et al., 1938; Van den Berg 

and Bruin, 1981). The moisture sorption isotherm is defined as the 

relationship between the total moisture content and water activity 

of the food at a constant temperature and under equilibrium 

conditions. Moisture sorption isotherms are useful, not only in 

showing at which moisture content certain desirable or undesirable 

levels of aw is achieved, but also in indicating what significance 

small changes in moisture content will have in terms of changes in 

aw. It is therefore a useful guide to the storage life of foods held at 

moderate temperatures and preserved only by reduced aw 

(Abdullah et al., 2000). 

The mathematical description of the sorption phenomena in 

foods is of interest. More than 200 equilibrium moisture content– 

equilibrium relative humidity relationships have been reported in 

Abbreviations: BET, Brunauer–Emmett–Teller; DVS, dynamic vapor sorption; 

GAB, Guggenheim–Andersen–de Boer; RH, relative humidity; TSS, third stage 

sorption; XPS, X-ray photoelectron spectroscopy. 

* Corresponding author. Tel.: þ33 3 83 59 58 78; fax: þ33 3 83 59 58 04. 

E-mail address: claire.gaiani@ensaia.inpl-nancy.fr (C. Gaiani). 


doi:10.1016/j.jcs.2008.08.001 





The combination of two analytical methodologies (water vapor sorption isotherm by using the DVS and 

chemical surface composition by using the XPS) has been used to enhance the understanding of the 

impact of re-grinding on the wheat flour hydration mechanism. A controlled atmosphere microbalance 

was used to construct water sorption isotherms at 25 C of different samples of wheat flours obtained by 

successive re-grinding of native wheat flour. 

Experimental water adsorption isotherms were modeled using different complementary models, based 

on two-parameter (BET), three-parameter (GAB), and four-parameter (TSS) models. A slight increase in 

water sorption capacity of wheat flour due to the re-grinding process was observed. The most affected 

parameters of the sorption isotherm models were C (the energy constant) and Xm (the monolayer water 

content capacity). The X-ray photoelectron spectroscopy (XPS) analysis showed changes in chemical 

bonds on wheat particle surfaces due to re-grinding process and particularly a significant increase in 

hydrophilic and decrease in hydrophobic bonds. 


the literature (McMinn et al., 2004). Both the International Union of 

Pure and Applied Chemistry (IUPAC) and the Commission on 

Colloid and Surface Chemistry (report of 1985) recommend the 

usefulness of the two-parameter model BET (Brunauer, Emmett, 

and Teller) because of its simplicity of application. This model is 

based on a multilayer sorption and is used in the low water activity 

range (0.05 < aw < 0.40). A three-parameter model (GAB model) 

was recommended by the European Project Group COST 90 (Wolf 

et al., 1984) and has been successfully applied to various foods (Van 

den Berg, 1985). The GAB (Guggenheim, Andersen, and de Boer) 

equation represents a model based on multi-layers and condensation 

and covers a wider range of water activity (0.05 < aw < 0.8) 

(Chirife et al., 1992; Timmermann and Chirife, 1991; Van den Berg 

and Bruin, 1981). Timmermann (1989) and Timmermann and 

Chirife (1991) developed a four-parameter equation, named the 

third stage sorption isotherm (TSS) to extend the GAB isotherm 

model to water activity ranges approaching unity. It is based on the 

premise that after a certain number of moisture layers exist, the 

moisture behaves as liquid water which has a dilution effect. The 

TSS model has thus the ability to predict infinite moisture 

adsorption (Bronlund and Paterson, 2004) and to give experimental 

data for many food systems a better fit when compared to the GAB 

equation. At the experimental level, the water adsorption capacity 

of food products has recently been evaluated using controlled 

atmosphere microbalances and dynamic automated sorption 

methods (Bell and Labuza, 2000; Johnson and Brennan, 2000). The

small size of samples and the dynamic airflow around the samples 

enable generation of a complete isotherm in less than a week (Bell 

and Labuza, 2000). Several studies using controlled atmosphere 

microbalances have already been reported (Ketal et al., 2004; 

Levoguer and Willians, 1997) 

The surface composition and particle properties can be considered 

as critical parameters of powders as they are supposed to play 

an important role within the powder’s end use processes and 

should certainly not be neglected (Fäldt, 1995; Kim et al., 2002). 

Understanding the mechanism of the powder surface formation in 

terms of the compositional aspect will be highly useful in powder 

quality improvement and new product development (Kim et al., 

2002). A classical way to characterise the powder’s surface 

composition is the use of X-ray photoelectron spectroscopy, also 

referred in the literature to electron spectroscopy for chemical 

analysis (ESCA) (Mistry et al., 1992). XPS is a well established 

technique for identifying elements and determining differences in 

surface chemistry of particle surfaces (Sionkowska et al., 2006). XPS 

has been commonly applied to spray-dried dairy powders (skim 

milk powder, whole milk powder, cream powder and whey protein 

concentrate) in order to investigate the relationship between milk 

powder processing, particle surface structure, and wetting properties 

to make advances in the understanding of the rehydration 

process or storage effects (Gaiani et al., 2006, 2007; Kim et al., 

2002). 

The main objective of the current work was to study the impact 

of the re-grinding process on hydration properties of wheat flour. 

This study is primarily concerned with determining the experimental 

sorption data of different re-ground wheat flour samples 

coming from the same batch of wheat flour and describing them by 

using the two-, three-, and four-parameter isotherm models. Water 

adsorption measurements by using the dynamic vapor sorption 

have been conducted in combination with measurement of changes 

in surface bonds on the surface of wheat flour particles using the Xray 

photoelectron spectroscopy. 


2.1. Wheat flour sample preparation and conservation 

The different samples were prepared from the selected 

commercial wheat flour (Grands Moulins de Paris, Ivry Sur Seine, 

France) intended for bread making (extraction rate 75%). Initial 

water content of the native wheat flour (FNative) was 13.7%. The 

different flour samples (F1, F2, and F3) were obtained by successive 

re-grinding (R1,R2, and R3) of the native flour to reduce particle size 

using a laboratory grinder, under ambient relative humidity 

conditions (ZM 200, Retsch, France). For the first re-grinding 

process R1, a 50 g sample of native wheat flour was placed inside 

the laboratory grinder. The resulting wheat flour was named F1. The 

flour F1 was twice re-ground under the same conditions R2 and R3, 

and then the re-ground flours F2 and F3 were produced. During our 

study, all samples were stored at 18 C in hermetically sealed 

cans, in order to preserve their physicochemical characteristics (Da 

Costa-Correia, 1997; Kusunose et al., 2002). Before use, samples 

were defrosted for 12 h at 20 C. 

2.2. Wheat flour characterization 

2.2.1. Chemical analyses 

Water and damaged starch contents of flour samples were 

estimated according to American Association of Cereal Chemists 

(AACC) methods 44-15A and 76-30A, respectively. The total 

nitrogen content (TN) was determined by the Kjeldahl method and 

protein content was calculated according to TN 5.7 AACC (2000) 

Method (No: 46-10). Fat content was determined by acid hydrolysis 

M.M. Saad et al. / Journal of Cereal Science 49 (2009) 134–140 135 

with HCl followed by extraction of hydrolysed lipids with ethers 

according to AACC Method (No: 30-10). All results represent the 

average of three tests. 

2.2.2. Physical properties 

The particle size distributions were measured by static light 

scattering (Mastersizer S, Malvern Instruments Ltd, Malvern, UK) 

with a 5 mW He–Ne laser operating at a wavelength of 632.8 nm 

with a 300F lens. Five grams of wheat sample were taken and 

introduced in 100 mL ethanol to reach a correct obscuration (Berton 

et al., 2002). The criterion selected was d50 which means that 

50% of the particles have a diameter lower than this criterion (i.e. 

midpoint of volume cumulative distribution). Results are the 

average of triplicate experiments carried out on different days. 

2.3. Sorption isotherms 

Water sorption isotherms were determined gravimetrically 

using DVS technique (Surface Measurement Systems, London, UK). 

The DVS apparatus monitors the moisture sorption capacities of 

wheat flour as function of relative humidity (RH). The changes in 

sample weight over time at 25 C and at any desired RH (between 

0% and 98%) were recorded. About 15–20 mg of sample were loaded 

onto the quartz sample pan. The program was initially set to control 

the humidity at 0% for 12 h (drying phase). This step allowed the 

sample water activity to decrease to zero and internally equilibrate. 

The sample was then subjected to successive steps of 10% RH 

increase, up to 98%. For each step, mass changes (m) and the rate of 

mass changes (dm/dt) were plotted against time. The equilibrium 

was considered to be reached when changes in mass with time 

(dm/dt) were lower than 0.002%/min (i.e. 2 g water/100 g db/day). 

All experiments were run at 25 C and 2–4 tests were carried out 

for each sample. The accuracy of the system was 1.0% RH 

and 0.2 C, respectively. 

2.4. XPS analysis 

The XPS analyses were carried out with a Kratos Axis Ultra 

(Kratos Analytical, Manchester, UK) spectrometer using a monochromatic 

Al Ka source. The delay-line detector allows a high count 

rate and the power applied to the X-ray anode was reduced to 90 W 

in order to avoid the X-ray induced degradation of the sample. The 

instrument work function was calibrated to give a binding energy 

(BE) of 83.96 eV for the Au 4f7/2 line for metallic gold and the 

spectrometer dispersion was adjusted to give a binding energy of 

932.62 eV for Cu 2p3/2 line for metallic copper. The wheat flour 

samples were attached to the sample holder using a double sided 

conductive tape and then evacuated overnight prior to analyses. All 

spectra were recorded at a 90 take-off angle, the analysed area 

being currently about 700 300 mm. Survey spectra were recorded 

with 1.0 eV steps and 160 eV analyser pass energy and the high 

resolution regions with 0.05 eV steps and 20 eV pass energy. In 

both cases the hybrid lens mode was used. During the data acquisition 

the Kratos charge neutralizer system was used on all specimens 

with the following settings: filament current 1.6 A, charge 

balance 2.4 V, filament bias 1.0 V and magnetic lens trim coil 

0.375 A. As overcompensation is always observed, the C1s line for 

adventitious carbon and C–C carbon was set to 284.60 eV and 

therefore used as an internal energy reference. With these 

parameters we could obtain a C1s signal with sharp, symmetric 

components with a FWHM of 1.2 eV. Spectra were analysed using 

the Vision software from Kratos (Vision 2.2.2). A Shirley baseline 

was selected to subtract the background and Gaussian–Lorentzian 

(70–30%) shapes were used for spectral decomposition. Quantification 

was performed using the photoemission cross-sections and 

the transmission coefficients given in the Vision package.

136 

Table 1 

Wheat flour samples chemical composition 

Wheat flour Composition (g/100 g flour) 

Water Protein Damaged starch Lipids 

FNative 13.7 0.1 12.5 0.1 4.7 0.4 1.5 0.2 

F1 13.6 0.2 12.4 0.1 8.5 0.0 1.4 0.1 

F2 13.4 0.2 12.4 0.2 9.1 0.4 1.6 0.1 

F3 13.4 0.3 12.4 0.3 12.6 0.8 1.2 0.3 

2.5. Data analysis 

The water vapor adsorption isotherms were described by using 

three models: the BET (Brunauer–Emmett–Teller) model (Eq. (1)), 

GAB (Guggenheim–Andersen–de Boer) model (Eq. (2)), and TSS 

(third stage sorption) model (Eq. (3a–c)). 

C BETaw 

X ¼ Xm 

ð1 awÞð1 aw þ CBETawÞ C GABK GABaw 

X ¼ Xm 

ð1 KGABawÞð1 KGABaw þ CGABKGABawÞ HðawÞH 0 ðawÞC TSSK TSSaw 

X ¼ Xm 

ð1 KTSSawÞ½1 þðCTSSHðawÞ 1ÞKTSSawŠ HðawÞh1 þ 1 K TSS 

K TSS 

ðK TSSawÞ hTSS 

ð1 awÞ 

M.M. Saad et al. / Journal of Cereal Science 49 (2009) 134–140 

(1) 

(2) 

(3a) 

(3b) 

Table 2 

Wheat flour samples particle size distribution 

Wheat flour d10 (mm) d50 (mm) d90 (mm) 

FNative 15.1 0.8 67.4 0.6 147.8 1.4 

F1 11.4 2.6 49.7 4.6 122.2 3.7 

F2 10.0 2.6 40.4 4.6 108.4 3.7 

F3 4.3 0.9 17.2 3.8 44.5 3.8 

H 0 ðawÞh1 þ HðawÞ 1 

HðawÞ 

ð1 KTSSawÞ ½hTSS þ ð1 h 

ð1 awÞ 

TSSÞawŠ (3c) 

where Xm is the monolayer moisture content (% db), X is the 

equilibrium water content (% db), CBET, CGAB, and CTSS are characteristic 

energy constants, KBET, KGAB, KTSS are the characteristic 

constants correcting the properties of the multilayer molecules 

with respect to the bulk liquid, and hTSS is the third stage sorption 

isotherm constant. 

The model parameters were directly determined from experimental 

water vapor sorption isotherms. The parameters of the BET 

model (Xm, C), GAB model (Xm,C, and K), and TSS model (Xm, C, K, 

and hTSS) were identified, respectively, with Eqs. (1)–(3). The 

parameters were calculated by an optimization procedure according 

to the Gauss–Newton algorithm using the software Excel 2007 

(Microsoft). The minimized objective function was the sum of the 

absolute difference between experimental and predicted points. 


All statistical analysis was carried out by using Microsoft Excel 

2007 (Microsoft Corporation, USA). 

Fig. 1. Sorption isotherm profile obtained for the FNative, F1, F2 and F3 wheat flour samples estimated at 25 C from 0% to 98% RH and modeled up with the TSS model.

Table 3 

Parameters obtained from the fitted curves with BET, GAB, and TSS models for FNative, F1, F2 and F3 wheat flour samples 

Model Flour Xm C K hTSS R 2 

Abs. difference 

BET FNative 6.41 0.69 18.9 10.1 – – 0.992 0.266 

F1 6.28 0.46 18.2 1.2 – – 0.990 0.296 

F2 6.36 0.03 17.4 1.0 – – 0.994 0.249 

F3 6.47 0.08 21.1 2.2 – – 0.994 0.263 


3.1. Wheat flour characterization 

The chemical composition of the selected samples of wheat 

flour is presented in Table 1. As reported by Berton et al. (2002) and 

Wang and Flores (2000), the re-grinding process produced 

a significant increase in damaged starch content (from 5% to 13%), 

while the protein and lipid contents remain stable (about 12.5% and 

1.5%, respectively). The water content slightly decreases from 13.7% 

to 13.2% with re-grinding and this could be due to the slight heat 

effects during the re-grinding process (þ10 C). As expected, the 

size distribution of flour samples decreases as re-grinding progresses 

(Table 2). The d50 of the samples is around 67 mm for native 

flour (FNative) and decreases to 50, 40, and 17 mm, respectively, for F1, 

F2 and F3 flours. The decrease in particle size with re-grinding is due 

to the successive breaking of flour particles inside the grinder. The 

breaks in particles occurring during re-grinding have been 

considered as responsible for starch granule rupture and thus 

induce increase in damaged starch content (Hoseney, 1994). 

According to Dubois (1949), starch granule exhibits elastic properties 

that lead to different types of damage such as cracks and 

breaks during grinding, which play an essential role in increasing 

the surface subjected to water vapor. 

3.2. Water vapor adsorption isotherms 

The adsorption isotherm profiles at 25 C, from 0% to 98% RH are 

presented in Fig. 1 for the selected wheat flour samples. As 

expected, these isotherms demonstrate an increase in equilibrium 

moisture content with increasing water activity. As has been shown 

by Roman-Gutierrez et al. (2002), this behavior is obviously 

a sigmoidal shaped curve reflecting a Type II isotherm according to 

the Brunauer classification (Brunauer et al., 1938). This form can be 

described in terms of a three-step moisture adsorption process. As 

previously observed by other researchers (Quirijns et al., 2005), 

sorption isotherms involve, during the first step, polar groups of 

high binding energy to hydrophilic components (starches, proteins, 

and pentosans) being saturated with water molecules (i.e. the 

monolayer coverage). During the second stage, additional water 

molecules are bound onto the monolayer (i.e. multilayer coverage) 

and water clusters begin to form. During the third step, accumulation 

of water in intermolecular free spaces occurs and results in 

partial swelling that in turn may expose additional hydrophilic 

binding sites. It can be noticed that very slight differences seem to 

be observed between the experimental adsorption isotherms for 

the native wheat flours and the three re-ground flour samples. 


GAB FNative 8.50 0.75 12.9 4.9 0.686 0.015 – 0.997 0.159 

F1 8.36 0.94 13.4 1.5 0.701 0.046 – 0.997 0.184 

F2 8.54 0.02 12.8 0.5 0.678 0.014 – 0.998 0.138 

F3 8.64 0.04 13.6 0.3 0.609 0.008 – 0.997 0.190 

TSS FNative 8.84 1.14 11.4 3.0 0.665 0.037 14.3 1.9 0.998 0.234 

F1 8.22 1.21 14.5 3.0 0.710 0.067 17.1 5.8 0.988 0.178 

F2 9.32 3.19 10.9 0.1 0.634 0.036 17.5 1.1 0.997 0.290 

F3 9.09 0.64 11.5 1.6 0.678 0.042 14.4 0.5 0.998 0.224 

Xm: monolayer moisture content g of water/100 g of dry base; C: constant; K: constant; hTSS: TSS model constant; Abs. difference: the average value of absolute difference 

between experimental data and calculated value of water content. 

3.3. Water vapor sorption mathematical modeling 

3.3.1. Brunauer–Emmett–Teller Model 

The experimental sorption data of wheat flour samples were 

first fitted with the two parameter BET model (Eq. (1)), between 0% 

and 40% RH (Table 3). There is a good agreement between experimental 

data and predicted values (average R 2 ¼ 0.992 and average 

absolute difference between experimental and calculated 

data ¼ 0.27%). The calculated Xm (monolayer moisture content) and 

CBET values are presented in Table 3 for native and re-ground flour 

samples. 

No differences have been found between the calculated values 

of BET model parameters for the four different products. We can 

report almost the same values of Xm (6.28–6.47 g/100 g dry bases) 

for the native wheat flour and the re-ground flour samples. Almost 

Fig. 2. XPS spectra obtained for the FNative wheat flour (survey scan).

138 

the same values for monolayer water contents (Xm) have already 

been reported in the literature for wheat flours (Roman-Gutierrez 

et al., 2002). 

We observe a slight increase in the CBET values with re-grinding, 

from CBET ¼ 18.9 for the native flour to CBET ¼ 21.1 for the sample F3. 

That is to say, when using the BET model, the energy constant (CBET) 

is found to be slightly increased as re-grinding rate increases. This 

behavior may be manifested by the liberation of hydrophilic sites 

during the re-grinding process, whereas different types of starch 

damage as cracks and breaks take place. 

3.3.2. Guggenheim–Andersen–de Boer Model 

The experimental sorption data of wheat flour samples have 

also been fitted with the three-parameter GAB model (Eq. (2)), 

between 0% and 80% RH (Table 3). As expected, there is good 

agreement between experimental data and predicted values 

(average R 2 ¼ 0.997 and average absolute difference between 

experimental and calculated data ¼ 0.17%). The calculated GAB 

model parameter Xm (monolayer moisture content), KGAB, and CGAB 

values are presented in Table 3 for the native wheat flour and the 

three re-ground flour samples. No significant differences have been 

found between the calculated values of GAB model parameters for 

the four products. We can report almost the same values of Xm 

(8.36–8.64 g/100 g dry bases) for the native wheat flour and the 

re-ground samples, nonetheless the highest values were observed 

for the most re-ground flour (sample F3). It can be noticed that the 


Xm values determined using the GAB model (8.36–8.64 g/100 g dry 

bases) are slightly higher than those calculated using the BET model 

(6.28–6.47 g/100 g dry bases). Timmermann (2003) also stated that 

the value Xm given by the BET model is always smaller than the 

monolayer value corresponding to the GAB model. 

No differences were observed among CGAB values (12.8–13.7 g/ 

100 g dry bases) for the native wheat flour and the re-ground flour 

samples. Still, the highest values were observed for the most 

re-ground flour (sample F3). 

The KGAB values remain constant (between 0.678 and 0.701) for 

all flour samples. The estimated value confirms observations, 

carried out on many starchy foods (Chirife et al., 1992), which 

reveals that KGAB values are about 0.74. Probably, this stability in 

KGAB values during the re-grinding process may be attributable to 

the steady chemical composition of all studied flour samples. 

Similar values of Xm, CGAB, and KGAB have already been reported in 

the literature for wheat flour (Chirife et al., 1992; Roman-Gutierrez 

et al., 2002). 

3.3.3. Third stage sorption model 

The experimental sorption data of wheat flour samples have 

then been fitted with the four-parameter TSS model (Eq. (3a–c)) 

between 0% and 98% RH as illustrated in Fig. 1 and Table 3. There is 

good agreement between experimental data and predicted values 

(average R 2 ¼ 0.995 and average absolute difference between 

experimental and calculated data ¼ 0.23%). The calculated TSS 

Fig. 3. Example of XPS narrow spectra of O1s obtained for FNative, F1, F2 and F3 wheat flour samples, respectively.

model parameters Xm (monolayer moisture content), CTSS, KTSS, and 

hTSS values are presented in Table 3 for the native wheat flour and 

the three re-ground flour samples. 

The calculated values of the TSS model (Xm, CTSS, KTSS, and hTSS) 

appear to stay constant for all the selected samples, with Xm values 

ranging between 8.84 and 9.32 g/100 g dry bases, CTSS values 

ranging between 10.9 and 14.5, KTSS ranging between 0.634 and 

0.710, and with hTSS values ranging between 14.3 and 17.5. Therefore, 

it may be stated that no significant change was observed 

between the native wheat flour and the three re-ground flour 

samples, when describing the whole adsorption isotherms 

between 0% and 98% relative humidity with the TSS model. The 

calculated values of the hTSS parameter appear to be low in 

comparison with ‘‘usual’’ values that have been found (Timmermann, 

1989). Bronlund and Paterson (2004) have found that hTSS 

value is around 30. Because of wheat flour insolubility in water 

(finite sorption at RH ¼ 100%), the TSS model takes into consideration 

the relative insolubility behavior of wheat components by 

lowering the hTSS value. This could explain why low values of h 

parameter have been obtained. 

3.4. Wheat flour chemical surface analyses using X-ray 

photoelectron spectroscopy 

The survey scan of the native wheat flour sample is represented 

as an example with the identification of the O1s,N1s,C1s,S2p and P2p 

peaks as illustrated in Fig. 2. The detection of S2p and P2p elements 

was possible thanks to the high sensitivity of the XPS equipment 

(Gaiani et al., 2006). Nevertheless, the concentration of these two 

elements was found below 1% for all the samples. Almost the same 

XPS curve shapes have been previously observed for milk powders 

(Gaiani et al., 2006). According to a model for biochemical 

compounds, the carbon (C), oxygen (O) and azotes (N) peaks were 

decomposed (Gerin et al., 1995). The C1s peak was decomposed into 

four peaks corresponding to the C–(C, H), C–(O, N), C]O and O– 

C]O functions. The O1s peak was decomposed in three peaks 

attributed to the O]C, O–C and H2O functions. The N1s peak was 

decomposed into C–NH and C–NH 3þ functions. 

As a typical example, decomposition of the oxygen peak for the 

native wheat flour and the three re-ground flour samples are 

presented in Fig. 3. Similar decomposition curves have been 

constructed for carbon and nitrogen (data not shown). From these 

decompositions, the data were analysed in terms of rate of surface 

bonds (Table 4). From the C1s peak, it appears that the re-grinding 

process (from FNative, F1, F2, toF3 flour) induces a slight decrease of 

Table 4 

Elemental surface composition (bold) and rate of surface bond for FNative,F1,F2 and F3 

wheat flour samples 

Element Bond Wheat flour sample 

FNative F1 F2 F3 

% C 77.0 77.4 76.7 74.7 

% C–H, C–C 61.0 60.1 59.1 57.2 

% C–O, C–N 28.2 27.9 29.3 31.0 

%C]O 6.2 8.6 8.0 8.7 

% O–C]O 4.6 3.4 3.6 3.1 

% O 18.6 18.0 18.9 19.3 

% O–C 82.0 77.5 76.9 75.2 

%O]C 13.7 14.1 15.2 18.9 

%H2O 4.3 8.4 7.9 5.9 

% N 3.9 4.1 4.0 5.5 

% C–NH 93.2 92.7 96.5 93.6 

% C–NH 3þ 

6.8 7.3 3.5 6.4 

% S 0.3 0.3 0.2 0.3 

% P 0.2 0.2 0.2 0.2 


Fig. 4. Changes in surface chemical bonds percentage as re-grinding rate increases for 

FNative, F1, F2 and F3 wheat flour samples, where white columns represent C–(H,C) and 

black columns represent C–(O–N), C]O, O–C]O. 

the C–(C, H) bond (from 61, 60, 59 to 57%, respectively, for FNative,F1, 

F2,toF3 samples). Concurrently the total of C–O functions (C–(O, N), 

C]O and O–C]O) increased. From the O1s peaks, the O]C functions 

increased significantly whereas the O–C functions decreased 

(from 82.0, 77.5, 76.9 to 75.2, respectively, for FNative, F1, F2, toF3 

samples). These results clearly reflect a decrease in the number of 

hydrophobic bonds in contrast to hydrophilic bonds which increase 

in its turn (Fig. 4). This behavior may be due to physical changes 

induced by re-grinding process resulting in starch granule rupture. 

4. Conclusion 

Whatever the model applied (BET, GAB, or TSS), wheat flour 

shows a slight trend to adsorb more water vapor after the regrinding 

process. This bonded water was adsorbed more strongly. 

One could assume that starch damages occurring during the regrinding 

process could affect the surface and modify interactions of 

particles with water vapor, whereas the hydrophilic bonds increase. 

Hence, the TSS (third stage sorption) model can be used coupled 

with the DVS technique to describe sorption isotherms of wheat 

flour throughout the entire range of water activity. Techniques such 

as DVS can be used coupled with XPS to obtain valuable information 

related to total and partial contribution of chemical components 

of wheat flours in their hydration properties. Therefore, 

further studies including surface composition evolution during 

milling coupled with sorption isotherms of pure wheat flour 

components (starch, protein,.) would improve understanding 

contributions of wheat flour components with water vapor 

adsorption. 


We are grateful to J. Lambert, LCPME engineer CNRS (Nancy), for 

providing XPS analysis and technical assistance. The authors would 

like also to acknowledge the support of Syrian Ministry of Higher 

Education for financial support. 

References 

American Association of Cereal Chemists, 2000. Official Methods of the AACC. 

Method 44-15A approved October 1975, revised October 1981; revised 

November 1999; Method 76-30A, approved May 1969, revised October 1982, 

and October 1984; revised November 1999; Method 46-10, approved April 

1961, revised November 1999; Method 30-10, approved April 1961, revised 

November 1999, 10th ed. The Association, St. Paul, MN, USA. 

Abdullah, N., Nawawi, A., Othman, I., 2000. Fungal spoilage of starch-based foods in 

relation to its water activity (aw). Journal of Stored Products Research 36, 47–54.

140 

Bell, L.N., Labuza, T.P., 2000. Moisture Sorption: Practical Aspects of Isotherm 

Measurement and Use. American Association of Cereal Chemists, Inc., St Paul, MN. 

Berton, B., Scher, J., Villieras, F., Hardy, J., 2002. Measurement of hydration capacity 

of wheat flour: influence of composition and physical characteristics. Powder 

Technology 128, 326–331. 

Bronlund, J., Paterson, T., 2004. Moisture sorption isotherms for crystalline, amorphous 

and predominantly crystalline lactose powders. International Dairy 


Brunauer, S., Emmett, P.H., Teller, E., 1938. Adsorption of gases in multimolecular 

layers. Journal of American Chemical Society 60, 309–319. 

Chirife, J., Timmermann, E.O., Iglesias, H.A., Boquet, R., 1992. Some features of the 

parameter k of the GAB equation as applied to sorption isotherms of selected 

food materials. Journal of Food Engineering 15, 75–82. 

Da Costa-Correia, J.M., 1997. Contribution à l’étude des propriétés rhéologiques des 

farines de blé: Influence de la composition, de l’hydratation de la distribution 

granulométrique des particules. Ph.D. thèsis, Institue Nationale Polytechnique 

de Lorraine. 

Dubois, M., 1949. Incidences du réglage de la mouture sur les propriétés des farines. 

Bulletin ENSMIC 113, 170–187. 

Fäldt, P., 1995. Surface Composition of Spray-dried Emulsions. Ph.D. thesis, Lund 

University. 

Gaiani, C., Ehrhardt, J.J., Scher, J., Hardy, J., Desobry, S., Banon, S., 2006. Surface 

composition of dairy powders observed by X-ray photoelectron spectroscopy 

and effects on their rehydration properties. Colloids and Surfaces B: Biointerfaces 

49, 71–78. 

Gaiani, C., Scher, J., Ehrhardt, J.J., Linder, M., Schuck, P., Desobry, S., Banon, S., 2007. 

Relationships between dairy powder surface composition and wetting properties 

during storage: importance of residual lipids. Journal of Agricultural and 


Gerin, P.A., Dengis, P.B., Rouxhet, P.G., 1995. Performance of XPS analysis of model 

biochemical compounds. Journal of Chemical Physics 92, 1043–1065. 

Hoseney, R.C., 1994. Dry milling of cereals. In: Principles of Cereal Science 

and Technology. American Association of Cereal Chemists, St. Paul, MN, USA, 

pp. 125–145. 

Johnson, P.N.T., Brennan, J.G., 2000. Kinetics of moisture absorption by plantain 

flour. Journal of Food Engineering 45, 33–36. 

Ketal, L.V., Graham, B., David, C., 2004. The use of dynamic vapour sorption and near 

infra-red spectroscopy (DVS-NIR) to study the crystal transitions of theophylline 

and the report of a new solid-state transition. European Journal of 

Pharmaceutical Sciences 22, 97–105. 

Kim, E.H.J., Chen, X.D., Pearce, D., 2002. Surface characterization of four industrial 

spray dried dairy powders in relation to chemical composition, 


structure and wetting properties. Colloids and Surfaces B: Biointerfaces 26 

(3), 197–212. 

Kusunose, C., Noguchi, S., Yamagishi, T., Seguichi, M., 2002. Binding starch to tailing 

fraction by proteins in stored wheat flour. Food Hydrocolloids 1, 73–77. 

Levoguer, C.L., Willians, D.R., 1997. Moisture sorption properties of food products 

and packaging materials studied by dynamic vapour sorption. Food Technology 

Europe 6, 28–30. 

McMinn, A.M., Al-Muhtaseb, A., Ronald, A., Magi, T., 2004. Assessment of two- and 

three-parameter Lewicki models for description of sorption phenomena of 

starch materials. Journal of the Science of Food and Agriculture 84, 1695–1700. 

Mistry, V.V., Hassan, H.N., Robinson, D.J., 1992. Effect of lactose and protein on the 

microstructure of dried milk. Food Structure 11, 73–82. 

Quirijns, E.J., Boxtel, A.J.B.V., Straten, W., 2005. Sorption isotherms, GAB parameters 

and isosteric heat of sorption. Journal of the Science of Food and Agriculture 85, 

1805–1814. 

Roman-Gutierrez, A., Guilbert, S., Cuq, B., 2002. Distribution of water between 

wheat flour components: a dynamic water vapor adsorption study. Journal of 


Sionkowska, A., Wisniewski, M., Kaczmarek, H., Skopinska, J., Chevallier, P., 

Mantovani, D., Lazare, S., Tokarev, V., 2006. The influence of UV irradiation on 

surface composition of collagen/PVP blended films. Applied Surface Science 

253, 1970–1977. 

Timmermann, E.O., 1989. A BET-like three sorption stage isotherm. Journal of 

Chemical Society 85, 1631–1645. 

Timmermann, E.O., Chirife, J., 1991. The physical state of water sorbed at high 

activities in starch in terms of the GAB sorption equation. Journal of Food 

Engineering 13, 171–179. 

Timmermann, E.O., 2003. Multilayer sorption parameters: BET or GAB values? 

Colloids and Surfaces A: Physicochemical and Engineering Aspects 220, 

235–260. 

Van den Berg, C., Bruin, S., 1981. Water activity and its estimation in food systems. 

In: Rockland, L.B., Stewart, F. (Eds.), Water Activity: Influence on Food Quality. 

Academic Press, New York, pp. 147–177. 

Van den Berg, C., 1985. Development of B.E.T.-like models for sorption of water on 

foods, theory and relevance. In: Simatos, D., Multon, J.L. (Eds.), Properties of 

Water in Foods. Martinus Nijhoff, Dordrecht, p. 119. 

Wang, L., Flores, R.A., 2000. Effects of flour particle size on the textural properties of 

flour tortillas. Journal of Cereal Science 31, 263–272. 

Wolf, W., Spiess, W.E.L., Jung, G., Weisser, H., Bizot, H., Duckworth, R.B., 1984. The 

water vapor sorption isotherms of microcrystalline cellulose (MCC) and of 

purified potato starch: results of a collaborative study. Journal of Food Engineering 

3, 51–72.

Identification of novel haze-active beer proteins by proteome analysis 

Takashi Iimure a, *, Nami Nankaku b , Megumi Watanabe-Sugimoto c , Naohiko Hirota a , Zhou Tiansu a , 

Makoto Kihara a , Katsuhiro Hayashi a , Kazutoshi Ito d , Kazuhiro Sato b 

a Bioresources Research and Development Department, Sapporo Breweries Ltd., 37-1, Nittakizaki, Ota, Gunma 370-0393, Japan 

b Barley Germplasm Center, Research Institute for Bioresources, Okayama University, 2-20-1, Chuo, Kurashiki, Okayama 710-0046, Japan 

c Graduate school of Natural Science and Technology, Okayama University, 1-1-1, Tsushimanaka, Okayama 700-8530, Japan 

d Frontier Laboratories of Value Creation, Sapporo Breweries Ltd., 10 Okatohme, Yaizu, Shizuoka 425-0013, Japan 

article info 


Received 30 June 2008 



Keywords: 

Beer colloidal haze 

Barley 

Proteome analysis 


abstract 

In clear beers, colloidal storage haze is one of the principal 

indicators of beer quality. Consumers rely greatly on visual 

impressions and will generally judge a beer as stale or not fit to 

drink if it displays haze when it is supposed to be clear. The cause of 

storage haze has been identified as due to interactions between 

haze-active proteins and certain polyphenols (Asano et al., 1982; 

Siebert, 1999; Siebert et al., 1996). The addition of silica gel during 

the beer filtration process is effective in removing haze-active 

proteins (Leiper et al., 2003; Siebert and Lynn, 1997). Haze-active 

polyphenols can be removed by the addition of polyvinylpolypyrolidone 

(PVPP) in beer filtration (McMurrough et al., 

Abbreviations: BDAI-1, barley dimeric alpha-amylase inhibitor; CMb, CMb 

component of tetrameric alpha-amylase inhibitor; CMe, trypsin inhibitor CMe 

precursor; 2DE, two-dimensional gel electrophoresis; LC-MS/MS, liquid chromatography 

mass spectrometry/mass spectrometry; LTP, lipid transfer protein; MALDI 

TOF-MS, matrix-assisted laser desorption/ionization time-of-flight mass spectrometry; 

NCBI-nr, the non-redundant amino acid database of the National Center 

for Biotechnology Information; PAS, proteins adsorbed onto the silica gel; TAI, 

trypsin/amylase inhibitor pUP13; Trx-2p, thioredoxin from yeast. 


E-mail address: takashi.iimure@sapporobeer.co.jp (T. Iimure). 


doi:10.1016/j.jcs.2008.08.004 





Colloidal haze reduces beer quality considerably. Four haze samples were analyzed by two-dimensional 

gel electrophoresis (2DE) in order to identify haze-active proteins. Several protein spots were observed in 

all of the four haze samples. Using mass spectrometry analysis followed by a database search identified 

these spots as barley dimeric alpha-amylase inhibitor (BDAI-1), CMb component of tetrameric alphaamylase 

inhibitor (CMb) and trypsin inhibitor CMe precursor (CMe). These proteins were considered to 

be haze-active. Since haze-active proteins are adsorbed by silica gel in the beer filtration process, we 

eluted proteins adsorbed onto silica gel (PAS) and identified their species. These major PAS were identified 

as protein Z4, protein Z7 and trypsin/amylase inhibitor pUP13 (TAI), rather than BDAI-1, CMb and 

CMe. Furthermore, we analyzed proline compositions in the beer proteins, PAS and the haze proteins. 

Consequently, we found that the proline compositions of PAS were higher (ca. 20 mol%) than those in the 

beer proteins (ca. 10 mol%), although those of the haze-active proteins such as BDAI-1, CMb and CMe 

were 6.6–8.7 mol%. Our results suggest that BDAI-1, CMb and CMe are not predominant haze-active 

proteins, but growth factors of beer colloidal haze. 


1995; Mikysˇka et al., 2002). Using barley cultivars that lack 

proanthocyanidins is also effective to produce beer without hazeactive 

polyphenols (Fukuda et al., 1999; von Wettstein et al., 1980, 

1985). Although research has been conducted on haze-active 

proteins and polyphenols (Apperson et al., 2002; Evans et al., 2003; 

Leiper et al., 2003; Robinson et al., 2004), the factors controlling 

colloidal haze formation are still not clear. 

A series of beer proteins such as hordeins (Asano et al., 1982), 

lipid transfer protein 1 (Evans and Hejgaard, 1999; van Nierop et al., 

2004) and protein Z (Evans and Hejgaard, 1999) are derived from 

barley malt. Hordeins are the predominant barley storage proteins 

and contain high levels of proline and glutamine. They are known 

to be involved in haze formation (Asano et al., 1982). Silica gel 

adsorbs proline rich proteins such as hordeins (Leiper et al., 2003). 

Robinson et al. (2007a,b) suggested that the barley trypsin inhibitor 

CMe precursor (CMe) was a haze-active protein based on the 

analysis of eluate proteins from silica gel. They distinguished the 

cultivars by the presence of CMe bands on immuno blot analysis 

with antibodies raised against silica eluate proteins, and revealed 

that beer brewed from cultivars without the CMe band formed less 

haze than the one with the CMe band (Evans et al., 2003; Robinson 

et al., 2004). Although several non-hordein proteins are regarded as 

haze-active, functions of these proteins in colloidal haze formation 

have not been well characterized. To reveal the mechanism of haze

142 

formation, beer proteins should be more comprehensively 

investigated. 

Proteome analysis, such as two-dimensional gel electrophoresis 

(2DE), followed by protein mass spectrometry, is a powerful tool to 

identify potential proteins of interest. Using this technique, barley 

grain, barley malt (Bak-Jensen et al., 2004; Østergaard et al., 2002) 

and beer (Perrocheau et al., 2005) have been analyzed. We also 

analyzed the beer proteins using 2DE and identified a possible 

foam-promoting protein (Okada et al., 2008). 

In this study, we prepared four haze-positive beer samples, 

analyzed proteins in haze by 2DE and identified haze-active 

proteins. We assumed that the protein spots observed on all four 

haze samples were from haze-active proteins, and that any proteins 

not appearing on the gel were haze-inactive proteins. In addition, in 

order to investigate the composition of proteins adsorbed onto the 

silica gel (PAS), we analyzed the PAS by 2DE and compared the 

proline compositions of the beer proteins, PAS and the haze 

proteins. 


2.1. Beer and haze samples 

To prepare haze-positive beer samples, commercial beers were 

stored for over one year at 4 C. These haze-positive beer samples 

were used for the analyses. Beer brewed from the malt of North 

American cultivar A, as described in our previous paper (Okada 

et al., 2008), was used to prepare a 25% salt-precipitated protein 

fraction. Beer samples filtered with and without silica gel 

(SiO 2 þbeer and SiO 2 beer) were brewed using only malt and hop, 

according to the standard method of Sapporo Breweries Ltd., as 

described by Okada et al. (2008). To prepare SiO 2 þ beer, the beer 

sample was filtered with 200 ppm of silica gel. 

2.2. Beer quality analysis 

Beer characteristics were analyzed according to the standard 

methods of the European Brewery Convention (EBC Analytica, 

1987). Colloidal stability was scored by a forcing test (FT-3). The 

bottled beer samples were stored at 60 C for 3 days, and then 0 C 

for 1 day. Subsequently, the beer clarity, FT-3, was assessed in terms 

of haze formation. 

2.3. Protein sample preparation 

To separate the haze and the beer fractions, the haze-positive 

beer samples were centrifuged at 2000 g, for 30 min at 4 C. The 

precipitate was washed twice in 5% ethanol solution, before being 

dissolved in 8 M urea (Wako, Japan) þ 2% 3-[(3-cholamidopropyl) 

dimethylammonio] propansulfonic acid (CHAPS) (Dojindo Laboratories, 

Japan) solution containing 0.28% dithiothreitol (Wako). 

Three milliliter of this solution and supernatant were applied to 

a PD-10 column (GE Healthcare Biosciences, Japan), and the 

desalted proteins were eluted by 4 ml of distilled water. The protein 

concentration was determined by the Bradford’s (1976) method 

using bovine serum albumin as a standard. These solutions were 

lyophilized and the lyophilized protein samples were subsequently 

used in 2DE and amino acid composition analyses. The 25% saltprecipitated 

protein fraction of the beer brewed from the malt of 

cultivar A was prepared according to Okada et al. (2008). The 

proteins adsorbed onto the silica gel (PAS) were prepared as 

described below by modification of the method described by Evans 

et al. (2003). Silica gel was added to the degassed beer sample upto 

200 ppm, whereupon the solution was stirred for 1 h at room 

temperature and then centrifuged at 2000 g for 30 min at 4 C. 2% 

(v/v) of ammonium solution was added to the precipitate. 

T. Iimure et al. / Journal of Cereal Science 49 (2009) 141–147 

Subsequently, to separate the silica gel and proteins, the solution 

was stirred for 1 h at room temperature and centrifuged at 

2000 g for 30 min at 4 C. By adding HCl, the pH of the solution 

was adjusted to 8.0. Subsequently, the solution was desalted using 

a PD-10 column (GE Healthcare Biosciences) and then lyophilized. 

The lyophilized protein was defined as PAS. 

2.4. Two-dimensional gel electrophoresis (2DE) 

The 2DE of both the first and second dimensions using the 

Multiphor II system (GE Healthcare Biosciences), and silver staining 

were carried out according to Okada et al. (2008). 

2.5. Mass spectrometry analysis and protein identification 

Selected protein spots were excised from the 2DE gel and 

digested with trypsin as described in a previous paper (Okada et al., 

2008). Eluted proteins were applied to matrix-assisted laser 

desorption/ionization time-of-flight mass spectrometry (MALDI 

TOF-MS) as described in Okada et al. (2008). Proteins were identified 

by peptide mass finger printing on the non-redundant amino 

acid database of the National Center for Biotechnology Information 

(NCBI-nr) using MASCOT software (Perkins et al., 1999). When 

a protein was not identified, it was digested using trypsin and/or 

lysil endopeptidase and the sample was analyzed using liquid 

chromatography mass spectrometry/mass spectrometry (LC-MS/ 

MS). A Tris–HCl buffer (pH 8.0) containing lysil endopeptidase was 

added to decolorized sample gel prepared according to Okada et al. 

(2008). Then the gel was incubated for 3 h at 35 C. Subsequently, 

trypsin was added to this sample and incubated for 20 h at 35 C. 

Each sample solution was applied to LC-MS/MS. The condition of 

the LC-MS/MS analysis is described below. Equipment: MAGIC 

2002 (Michrom BioResources, Inc., USA), column: Magic C18 

(0.1 150 mm, Michrom BioResources, Inc.), mobile phase: 2% 

acetonitrile þ 0.1% formic acid, and 90% acetonitrile þ 0.1% formic 

acid, flow rate: 250–300 nL min 1 , mass spectrometer: Q-Tof2 

(Micromass, U.K.), ionization method: Nanoflow-LC ESI, ionization 

mode: positive mode, electric potential of capillary: 1.8 kV, collision 

energy: 20–56 eV. To identify the protein species, NCBI-nr was 

searched for using the resultant values of product ion from all 

precursor ions using the MASCOT search engine (Perkins et al., 

1999). 

2.6. Analysis of the proline composition of protein fraction 

The proline compositions of beer proteins, haze proteins and 

PAS were analyzed using the Waters Pico$Tag system (Waters, 

Japan). A 20 mL aliquot of 6 M HCl (1% phenol) was added to the 

lyophilized protein sample, and then the proteins were completely 

hydrolyzed at 110 C for 20 h. Subsequently, the amino acid 

concentration was measured. Amino acid concentrations were 

determined as described below. The sample was ultra-filtered by 

Microcon Ultracel YM-10, MW 10,000 (Nihon Millipore Ltd., Japan). 

Once the filtrate had been diluted to a suitable concentration, 

fluorescence derivatization was carried out. A 20 mL aliquot of 

sample solution, 60 mL of AccQ$Fluorborate buffer (Waters) and 

20 mL of AccQ Fluor reagent (Waters) were mixed, and the mixture 

was subsequently incubated for 10 min at 55 C. The conditions for 

high performance liquid chromatography are described as follows: 

equipment: 2695 separation module, detector: 2475 multi l fluorescent 

detector (Waters) with Ex. 250 nm and Em. 395 nm, 

column: an AccQ Tag column (Waters), mobile phase: 100 mM 

sodium acetate þ 5.6 mM triethylamine (pH 5.7), 100 mM sodium 

acetate þ 5.6 mM triethylamine (pH 6.8), acetonitrile and distilled 

water. The chromatography data produced was analyzed using 

Empower personal software (Waters).

3. Results 

3.1. 2DE analysis of the proteins in haze 

After we prepared the haze-positive beer samples, the haze and 

beer protein fractions were separated by centrifugation. Then the 

protein fractions were analyzed by 2DE (Fig. 1). On the 2DE gel of 

the beer proteins, a large intensely staining spot (spot b0) at about 

pI 4.5–5.5, and a Mr of 35–45 kDa was observed. However, the 

intensity of spot b0 from the haze protein was lower than that of 

the beer (Fig. 1A, Table 1). In the haze proteins, intense spots were 

observed in region I-I, b8–10, b13, and b14 (Fig. 1). Spots in region 

I-I and b8 were intense both in the beer and the haze proteins. On 

the other hand, spots b9, b10, b13 and b14 were only intense in the 

haze proteins, and spots b11 and b12 were only intense in the beer 

proteins (Fig. 1, Table 1). The low molecular weight region of the 

2DE image of the 25% salt-precipitated protein fraction was similar 

to those of the haze proteins, but different from those of the beer 

proteins (Fig. 1, Table 1). The spots of the 25% salt-precipitated 

fraction were distinct in region I-I, b0 and b8–b10, but faint or 

invisible in b5 and b11–b14. Fig. 2 shows the 2DE images in region I 

of the four haze protein samples. The protein spots were classified 

into three groups i.e. (1) haze-active, because the protein spot was 

observed in all four haze samples, (2) moderately haze-active 

because the protein spot was observed in few samples as a relatively 

faint spot or observed in all samples but as a faint spot, and 

(3) haze-inactive because the protein spot was faint or invisible in 

all four samples. Protein spots in region I-I, b10, b13 and b14 were 

observed in all four samples. Conversely, protein spots of b11 and 

b12 were invisible, and b0 (data not shown) was faint in all four 

samples. In addition, spot b5 was only observed in sample i faintly, 

and spots b8 and b9 were observed in 2 and 3 samples respectively 

(Table 1, Fig. 2). Overall, the protein spots are: (1) haze-active in 

A 

Mr 

(kDa) 

97.4 

66.2 

45.0 

31.0 

21.5 

14.4 

97.4 

66.2 

45.0 

31.0 

21.5 

14.4 

97.4 

66.2 

45.0 

31.0 

21.5 

14.4 

T. Iimure et al. / Journal of Cereal Science 49 (2009) 141–147 143 

pI3 pI 10 

Beer (supernatant) 

Mw 

b0 

(kDa) 

14.4 

Haze (precipitate) 

25% salt precipitate 

14.4 

14.4 

B 

region I-I, b10, b13 and b14, (2) moderately haze-active in spots b0, 

b8 and b9, and (3) haze-inactive in spots b5, b11 and b12, respectively 

(see also Table 1). 

3.2. Mass spectrometry analysis and protein identification 

To identify protein species, we analyzed the protein spots in 

region I-I, b0, b5, and b8–b14 by MALDI TOF-MS or LC-MS/MS 

followed by a database search. Table 2 shows the identified protein 

spots by the analyses. Considering the classification of regions by 

haze-activity, barley dimeric alpha-amylase inhibitor (BDAI-1), 

CMb component of tetrameric alpha-amylase inhibitor (CMb), and 

trypsin inhibitor CMe precursor (CMe) were found in haze-active 

regions. Protein Z-type serpin (protein Z4), serpin (protein Z7), and 

chain A, non-specific lipid transfer protein 1 (LTP) were found in 

moderately haze-active regions. Thioredoxin from yeast (Trx-2p) 

and trypsin/amylase inhibitor pUP13 (TAI) were found in hazeinactive 

regions. 

3.3. 2DE analysis of the proteins adsorbed onto the silica gel (PAS) 

Silica gel is known to adsorb haze-active proteins (Leiper et al., 

2003; Siebert and Lynn, 1997) therefore, an addition of silica gel at 

beer filtration is effective to prevent beer colloidal haze. We 

prepared two beer samples with and without silica gel in filtration, 

described as SiO 2 þ beer and SiO 2 beer, respectively. The quality 

profiles of these beer samples were similar except for FT-3 i.e. the 

beer clarity after the forcing test (Table 3). The SiO 2 þ beer showed 

resistance to beer colloidal haze. On the other hand, the SiO 2 beer 

was shown to be prone to haze formation in response to force testing 

and contained haze-active protein, which was partly adsorbed onto 

the silica gel. To prepare the proteins adsorbed onto the silica gel 

(PAS), we added silica gel to the SiO 2 beer, and then eluted proteins 

b11 b12 

Region I Region I-I 

Fig. 1. The two-dimensional gel electrophoresis (2DE) images of the beer proteins of sample i, the haze proteins of sample i and the 25% salt-precipitated proteins of the beer 

brewed from the malt of cultivar A. A: The whole images of 2DE, pI 3–10, B: the enlarged images of region I in A. Arrows indicate the spot numbers. 

b5 

b9 

b2 

b10 

b8 

b8 

b13 

b14

144 

Table 1 

Summary of the spot intensities of region I-I, b0, b5 and b8–b14 in the beers, the hazes, the 25% salt-precipitated proteins and the proteins adsorbed onto the silica gel (PAS). 

The beer proteins The haze proteins The 25% 

Haze-activity PAS 

salt-precipitated 

proteins 

a 

Sample i Sample ii Sample iii Sample iv 

b0 B b 6 6 6 6 B 6c B 

Region I-I B B B B B B B 6 

b5 6 6 

b8 B B 6 B 6 6 

b9 B 6 B B 6 6 

b10 B 6 6 6 B B 6 

b11 B B 

b12 B B 

b13 B 6 6 B B 

b14 B 6 B B B 

a PAS, proteins adsorbed onto the silica gel. 

b ‘B’ indicates that the spot was distinct, ‘6’ indicates that the spot was faint, and ‘ ’ indicates that the spot was invisible. 

c ‘B’ indicates that the protein was haze-active because the protein spot was observed in all four haze samples, ‘6’ indicates that the protein was moderately haze-active 

because the protein spot was observed in few samples as relatively faint spot or observed in all samples but as faint spot, and ‘ ’ indicates that the protein was haze-inactive 

because the protein spot was faint or invisible in all four samples. 

from the silica gel. Subsequently, we analyzed the proteins of two 

beer samples and PAS. The 2DE images of the two beer proteins were 

quite similar (Fig. 3). The 2DE images in PAS and the haze proteins 

were different both at the intensity of the spot b0 (protein Z) (Figs.1A 

and 4A) and the region I in the low molecular region (Figs. 2 and 4). 

The intensity of spot b0 in PAS was higher than those in the haze 


sample (Figs.1A and 4A). In the PAS at region I (Fig. 4B), intensities of 

spots b8 (CMb), b9 (LTP), b10 (CMb), b13 (CMe) and b14 (CMe) were 

faint or invisible, those in region I-I (BDAI-1) were low except for the 

spot b2, and those in b11 and b12 (TAI) were distinct, respectively. 

These results suggest that the major PAS are protein Z4, protein Z7, 

and TAI, rather than BDAI-1, CMb and CMe. 

Fig. 2. The two-dimensional gel electrophoresis (2DE) images of the low molecular region, region I (Fig. 1A) of the haze samples i, ii, iii and iv. Arrows indicate the spot numbers.

Table 2 

Summary of protein identification by mass spectrometry analysis followed by 

a database search. 

Protein spot 

number 

Protein name Accession number Organism 

b0 protein Z-type serpin CAA66232 barley 

serpin CAA64599 barley 

Region I-I a 

barley dimeric 

alpha-amylase 

inhibitor; BDAI-1 

CAA08836 barley 

b5 thioredoxin; Trx-2p NP_011725 yeast 

b8 CMb component of 

tetrameric alpha-amylase 

inhibitor 


b9 chain A, non-specific 

lipid transfer protein 1 

1MID-A barley 

b10 CMb component of 

tetrameric alpha-amylase 

inhibitor 


b11 trypsin/amylase 

inhibitor pUP13 

1208404A barley 

b12 trypsin/amylase 

inhibitor pUP13 

1208404A barley 

b13 trypsin inhibitor 

CMe precursor 

P01086 barley 

b14 trypsin inhibitor 

CMe precursor 

P01086 barley 

a All spots in region I-I. 

Table 3 

The characteristics of the beer samples filtrated with silica gel (SiO 2 þ beer) and filtrated 

without silica gel (SiO 2 beer). 

SiO2 beer SiO2þ beer 

original gravity (%) 11.67 11.49 

final extract (%) 3.95 3.88 

apparent extract (%) 2.11 2.07 

alcohol (vol.%) 5.07 4.99 

pH 4.49 4.50 

color (EBC) 7.9 7.6 

bitter unit (mg/L) 28.2 28.1 

FT-3 a 

>10.0 1.68 

a FT-3 means the beer clarity after the forcing test (see Section 2). 

3.4. Proline compositions of the beer, the haze and the PAS 

It is well known that hordeins, one of the haze-active proteins, 

are proline rich (Asano et al., 1982). We examined the proline 

compositions of the beer proteins, the haze proteins and PAS (Table 

4). The proline compositions in the beer proteins were ca. 10 mol%, 

while those in PAS were ca. 20 mol%. These results suggest that 

proline rich proteins are adsorbed onto the silica gel. On the other 

Mr 

(kDa) 

97.4 

66.2 

45.0 

31.0 

21.5 

14.4 


hand, the proline compositions of the two haze proteins (samples i 

and iii) differed considerably (Table 4). 

4. Discussion 

To identify haze-active proteins, we prepared four haze-positive 

beer samples and analyzed the haze proteins using 2DE and mass 

spectrometry followed by a database search. Consequently, we 

assume that BDAI-1, CMb and CMe are haze-active, protein Z4, 

protein Z7 and LTP are moderately haze-active, and Trx-2p and TAI 

are haze-inactive proteins, respectively. Among the haze-active 

proteins the spots b10 (CMb), b13 (CMe) and b14 (CMe) were not 

observed in the beer protein sample i (Fig. 1) but spots in region I-I 

(BDAI-1) were intense (Fig. 1B). It is suggested, therefore, that CMb 

and CMe are concentrated in the haze fraction. In salting-out of the 

beer proteins, proteins did not precipitate at up to 20% saturation, 

but precipitated at 25% saturation of ammonium sulfate (data not 

shown). Therefore, the 25% salt-precipitated protein fraction may 

contain relatively hydrophobic proteins. The intense protein spots 

in region I-I (BDAI-1), b8 and b10 (CMb) were observed in a 25% 

salt-precipitated fraction (Fig. 1B), suggesting that BDAI-1 and CMb 

are potentially haze-active due to their high hydrophobicity. CMe is 

a possible haze-active protein although the mechanism is unknown 

at present. 

The spot intensities of regions I-I (BDAI-1), b10 (CMb), b13 and 

b14 (CMe) were different among the four haze samples (Fig. 2). In 

haze sample ii in Fig. 2 the protein spots b10 (CMb), b13 and b14 

(CMe) were less prominent despite the high intensity of the protein 

spots in region I-I (BDAI-1). This could have been due to the 

difference in the level of haze formability and suggested that BDAI- 

1 was more haze-active than both CMb and CMe. 

From the analyses of our four haze-positive beer samples, we 

consider that BDAI-1, CMb and CMe contribute to haze formation. 

The association of CMe with haze formation has already been 

suggested by Evans et al. (2003) and Robinson et al. (2004). They 

indicated that beer brewed from the malt of several barley varieties 

without the CMe band in western blot analysis using the antibody 

of silica eluate proteins showed improved haze stability compared 

to those with the CMe band. In order to further confirm the function 

of other haze-active proteins, i.e. BDAI-1 and CMb, we need to 

perform further experiments; for example force tests where these 

proteins are added to beer, and/or the development of novel 

cultivars without BDAI-1 or CMb which are then subject to the 

same trials. 

It has been shown that the addition of silica gel during beer 

filtration contributes to haze reduction due to adsorption of hazeactive 

proteins onto silica gel (Leiper et al., 2003; Siebert and Lynn, 

1997). We prepared two beer samples that were filtered with and 

without silica gel, i.e. SiO 2 þ beer and SiO 2 beer, respectively. The 

pI 3 pI 10 pI 3 pI 10 

SiO2 + beer SiO2-beer Fig. 3. The two-dimensional gel electrophoresis images of the beer filtered with silica gel (SiO 2 þ beer) and of that filtered without silica gel (SiO 2 beer).

146 

A 

Mr 

(kDa) 

B 

97.4 

66.2 

45.0 

31.0 

21.5 

14.4 

pI 3 pI 10 

Region I 

b11 

Region I-I 

b9 

b12 

PAS a 

PAS 

Fig. 4. The two-dimensional gel electrophoresis (2DE) images of the proteins adsorbed 

onto the silica gel (PAS). A: The whole image of the 2DE, pI 3–10, B: the enlarged image 

of region I. Arrows indicate the spot numbers. 

haze stability of SiO 2 þ beer was improved (Table 3). We then 

analyzed PAS using 2DE and identified protein species using mass 

spectrometry followed by a database search. As a result, the major 

proteins adsorbed onto the silica gel were identified as protein Z4, 

protein Z7 and TAI, rather than BDAI-1, CMb and CMe. Therefore, 

the majority of BDAI-1, CMb and CMe were transferred to beer after 

filtration. If BDAI-1, CMb and CMe were the predominant hazeactive 

proteins, the level of haze stability in SiO 2 þ beer might be 

comparable with that of SiO 2 beer. However, the haze stability in 

SiO 2 þ beer was improved compared to SiO 2 beer (Table 3). This 

evidence demonstrates that current silica treatment practices are 

only partially effective in improving haze stability because the 

majority of the BDAI-1, CMb and CMe remain in the treated beer. It 

follows that if these proteins could be removed, either by selection 

of novel barley varieties that do not contain these proteins, or 

improved haze stability treatments, the colloidal stability of beer 

could be further improved. 

Table 4 

The proline compositions of the beer proteins, the proteins adsorbed onto the silica 

gel (PAS) and the haze proteins. 

Sample name Proline (mol%) 

Beer SiO2þ beer 11.4 

SiO2 beer 10.7 

PAS a 

b2 

b10 

b8 

b13 

SiO 2 þ beer 16.2 

SiO 2 beer 23.4 

Haze sample i 5.6 

sample iii 27.9 

a PAS, proteins adsorbed onto the silica gel. 


b14 

It is conventionally accepted that haze proteins have elevated 

levels of proline (Siebert and Lynn, 1997). The proline composition 

of PAS was ca. 20 mol% (Table 4), while those of BDAI-1, CMb and 

CMe were as low as 6.6, 8.7 and 6.8 mol%, respectively. On the other 

hand, the proline composition of the haze proteins varied considerably 

between samples i (5.6 mol%) and iii (27.9 mol%), respectively 

(Table 4). If the beer colloidal haze consists of only proline 

rich proteins, such as hordein, the proline composition in all haze 

samples would be expected to be high (ca. 20 mol%). In addition, if 

the beer colloidal haze consists of only relatively proline poor 

proteins, such as BDAI-1, CMb and CMe, the proline compositions in 

all haze samples would be low (6–9 mol%). We estimate, therefore, 

that the proline composition of colloidal haze proteins might be 

determined by the degree of interaction between proline poor 

proteins, such as BDAI-1, CMb and CMe, and a core colloidal haze 

consisting of proline rich proteins such as hordeins. It is suggested 

that BDAI-1, CMb and CMe are not predominant haze-active 

proteins, but in fact initiation or growth factors in the formation of 

colloidal haze. 

Proline rich protein such as hordein was not detected in the 2DE 

analysis of haze proteins, because hordein may not have been 

detectable by the 2DE analysis applied in this study due to being 

poorly silver stained. Therefore, our results do not exclude the 

contribution of hordein-derived polypeptides to haze formation. To 

further improve the beer colloidal haze, we need to develop novel 

silica gel having adsorbing affinity to BDAI-1, CMb and CMe or 

develop cultivars with low haze-active proteins. Okada et al. (2008) 

suggested that BDAI-1 is a possible foam-positive protein. Therefore 

the amount of BDAI-1 may well need to be optimized to 

produce beer with high qualities such as foam retention and haze 

stability. 


We are grateful to T. Yazawa, K. Ito and H. Kato of the Bioresources 

Research and Development Department, Sapporo Breweries 

Ltd. for their technical assistance. This study was supported by 

the Program for Promotion of Basic Research Activities for Innovative 

Biosciences, Japan (PROBRAIN). 

References 

Apperson, K., Leiper, K.A., McKeown, I.P., Birch, D.J.S., 2002. Beer fluorescence and 

the isolation, characterization and silica adsorption of haze-active beer 

proteins. Journal of the Institute of Brewing 108, 193–199. 

Asano, K., Shinagawa, K., Hashimoto, N., 1982. Characterization of haze-forming 

proteins of beer and their roles in chill haze formation. Journal of the American 

Society of Brewing Chemists 40, 147–154. 

Bak-Jensen, K.S., Laugesen, S., Roepstorff, P., Svensson, B., 2004. Two-dimensional 

gel electrophoresis pattern (pH 6–11) and identification of water-soluble barley 

seed and malt proteins by mass spectrometry. Proteomics 4, 728–742. 

Bradford, M.M., 1976. A rapid and sensitive method for the quantization of microgram 

quantities of protein utilizing the principle of protein–dye binding. 

Analytical Biochemistry 72, 248–254. 

European Brewery Convention, Analytica, 1987. European Brewery Convention, 

fourth ed. Brauerei und Getraenke Rundschau, Zurich. 

Evans, D.E., Hejgaard, J., 1999. The impact of malt derived proteins on beer foam 

quality part I: the effect of germination and kilning on the level of proteins Z4, 

protein Z7 and LTP1. Journal of the Institute of Brewing 105, 159–169. 

Evans, D.E., Robinson, L.H., Sheehan, M.C., Tolhurst, R.L., Hill, A., Skerritt, J.S., 

Barr, A.R., 2003. Application of immunological methods to differentiate 

between foam-positive and haze-active proteins originating from malt. Journal 

of the American Society of Brewing Chemists 61, 55–62. 

Fukuda, K., Saito, W., Arai, S., Aida, Y., 1999. Production of a novel proanthocyanidin-free 

barley line with high quality. Journal of the Institute of Brewing 105, 

179–183. 

Leiper, K.A., Stewart, G.G., McKeown, I.P., 2003. Beer polypeptides and silica gel part 

I. Polypeptides involved in haze formation. Journal of the Institute of Brewing 

109, 57–72. 

McMurrough, I., Madigan, D., Smyth, M.R., 1995. Adsorption by polyvinylpolypyrrolidone 

of catechins and proanthocyanidins from barley. Journal 

of Agricultural and Food Chemistry 43, 2687–2691.

Mikysˇka, A., Hrabák, M., Hasˇková, D., Sˇrogl, J., 2002. The role of malt and hop 

polyphenols in beer quality, flavour and haze stability. Journal of the Institute of 

Brewing 108, 78–85. 

Okada, Y., Iimure, T., Takoi, K., Kaneko, T., Kihara, M., Hayashi, K., Ito, K., Sato, K., 

Takeda, K., 2008. The influence of barley malt protein modification on beer 

foam stability and their relationship to the barley dimeric a-amylase inhibitor-I 

(BDAI-I) as a possible foam-promoting protein. Journal of Agricultural and Food 

Chemistry 56, 1458–1464. 

Østergaard, O., Melchior, S., Roepstorff, P., Svensson, B., 2002. Initial proteome 

analysis of mature barley seeds and malt. Proteomics 2, 733–739. 

Perkins, D.N., Pappin, D.J., Creasy, D.M., Cottrell, J.S., 1999. Probability-based protein 

identification by searching sequence databases using mass spectrometry data. 

Electrophoresis 20, 3551–3567. 

Perrocheau, L., Rogniaux, H., Boivin, P., Marion, D., 2005. Probing heat-stable watersoluble 

proteins from barley to malt and beer. Proteomics 5, 2849–2858. 

Robinson, L.H., Evans, D.E., Kaukovirta-Norja, A., Vilpola, A., Aldred, P., Home, S., 

2004. The interaction between malt protein quality and brewing conditions and 

their impact on beer colloidal stability. Technical Quarterly – Master Brewers 

Association of the Americas 41, 353–362. 

Robinson, L.H., Healy, P., Stewart, D.C., Eglinton, J.K., Ford, C.M., Evans, D.E., 2007a. 

The identification of a barley haze active protein that influences beer haze 

stability: the genetic basis of a barley malt haze active protein. Journal of Cereal 



Robinson, L.H., Juttner, J., Milligan, A., Lahnstein, J., Eglinton, J.K., Evans, D.E., 2007b. 

The identification of a barley haze active protein that influences beer haze 

stability: cloning and characterization of the barley SE protein as a barley 

trypsin inhibitor of the chloroform/methanol type. Journal of Cereal Science 45, 

343–352. 

Siebert, K.J., Carrasco, A., Lynn, P.Y., 1996. Formation of protein–polyphenol haze in 

beverages. Journal of Agricultural and Food Chemistry 44, 1997–2005. 

Siebert, K.J., Lynn, P.Y., 1997. Mechanisms of beer colloidal stabilization. Journal of 

the American Society of Brewing Chemists 55, 73–78. 

Siebert, K.J., 1999. Effects of protein–polyphenol interactions on beverage haze, stabilization, 

and analysis. Journal of Agricultural and Food Chemistry 47, 353–362. 

van Nierop, S.N.E., Evans, D.E., Axcell, B.C., Cantrell, I.C., Rautenbach, M., 2004. 

Impact of different wort boiling temperatures on the beer foam stabilizing 

properties of lipid transfer protein 1. Journal of Agricultural and Food Chemistry 

52, 3120–3129. 

von Wettstein, D., Jende-Strid, B., Ahrenst-Larsen, B., Erdal, K., 1980. Proanthocyanidin-free 

barley prevents the formation of beer haze. Technical Quarterly – 

Master Brewers Association of the Americas 17, 16–23. 

von Wettstein, D., Nilan, R.A., Ahrenst-Larsen, B., Erdal, K., Ingversen, J., Jende- 

Strid, B., Kristiansen, K.N., Larsen, J., Outtrup, H., Ullrich, S.E., 1985. Proanthocyanidin-free 

barley for brewing: progress in breeding for high yield and 

research tool in polyphenol chemistry. Technical Quarterly – Master Brewers 

Association of the Americas 22, 41–52.

Mixing properties and dough functionality of transgenic lines of a commercial 

wheat cultivar expressing the 1Ax1, 1Dx5 and 1Dy10 HMW glutenin subunit 

genes 

Elena León a , Santiago Marín a , María J. Giménez a , Fernando Piston a , Marta Rodríguez-Quijano b , 

Peter R. Shewry c , Francisco Barro a, * 

a 

Instituto de Agricultura Sostenible, CSIC, Alameda del Obispo s/n, 14080 Córdoba, Spain 

b 

Universidad Politécnica de Madrid, 28040 Madrid, Spain 

c 

Rothamsted Research, Harpenden, Hertfordshire AL5 2JQ, UK 

article info 





Keywords: 

1Ax1 

1Dx5 

1Dy10 

Anza 

GM wheat 


abstract 

The viscoelastic properties of wheat flour dough allow wheat to 

be used for making bread and many other food products such as 

cake, biscuits, pasta and noodles. The unique properties of wheat 

result from the unusual biomechanical properties of the gluten 

proteins, which form a network conferring elasticity and 


E-mail address: ge1balof@uco.es (F. Barro). 


doi:10.1016/j.jcs.2008.08.002 





In this work we report the effects of the HMW-GS 1Ax1, 1Dx5 and 1Dy10 on the breadmaking quality of 

the bread wheat cultivar Anza that contains the HMW-GS pairs 1Dx2 þ 1Dy12 and 1Bx7* þ 1By8, and is 

null for the Glu-A1 locus. This allows the characterization of individual subunits 1Dx5 and 1Dy10 in the 

absence of subunit 1Dx5, and the interactions between these subunits and subunits 1Dx2 and 1Dy12 to 

be determined. Three transgenic lines termed T580, T581 and T590, containing, respectively, the HMW- 

GS 1Ax1, 1Dx5 and 1Dy10 were characterized over 3 years using a range of widely-used grain and dough 

testing methods. The transgenic subunits 1Ax1, 1Dx5 and 1Dy10 accounted for 25.2%, 20.3% and 17.9%, 

respectively, of the total HMW-GS in the three transgenic lines. Although lines T581 and T590 expressed 

similar levels of subunits 1Dx5 and 1Dy10 they had different effects on other aspects of protein 

composition, including changes in the ratios of glutenin/gliadin, of HMW/LMW-GS, the 1Dx2/1Dy12, the 

x-type/y-type HMW-GS and the proportions of high molecular mass glutenin polymers. In contrast, lines 

transformed to express subunits 1Ax1 and 1Dx5 showed similar changes in protein composition, with 

higher protein contents and decreased ratios of glutenin/gliadin and 1Dx2/1Dy12. In addition, both 

transgenic lines showed similar increases in the ratio of x-type/y-type subunits compared to the control 

line. The transgenic lines were analysed using Farinograph, Mixograph and Alveograph. This confirmed 

that the expression of all three subunits resulted in increased dough strength (and hence breadmaking 

quality) of the cultivar Anza. A beneficial effect of subunit 1Dx5 has not been reported previously, 

transgenic wheat lines expressing this subunit giving overstrong dough unsuitable for breadmaking. 

However, the expression of subunit 1Dy10 had a greater effect on breadmaking quality than subunits 

1Ax1 and 1Dx5. The Farinograph parameters such as dough stability and peak time were increased by 

9.2-fold and 2.4-fold, respectively, in line T590 (expressing 1Dy10) with respect to the control line. 

Similarly, the Mixograph mixing time was increased by four-fold and the resistance breakdown 

decreased by two-fold in line T590 compared with the control line. The Alveograph W value was also 

increased by 2.7-fold in line T590 compared to the control line. These transgenic lines are of value for 

studying the contribution of specific HMW-GS to wheat flour functional properties. 


extensibility to the dough. One group of gluten proteins, the high 

molecular weight glutenin subunits (HMW-GS), play an important 

role in determining the functional properties of wheat dough. The 

HMW-GS of wheat have been studied in detail, demonstrating that 

allelic differences in their composition result in effects on glutenin 

polymers and hence breadmaking quality (Payne, 1987; Shewry 

et al., 2003). Bread wheat contains six HMW-GS genes, with tightly 

linked pairs of genes encoding x- and y- type subunits being 

present at each of the Glu-A1, Glu-B1, and Glu-D1 loci on the long 

arms of chromosomes 1A, 1B, and 1D, respectively. Wheat cultivars

containing HMW-GS 1Dx5 and 1Dy10 have stronger doughs than 

those containing the allelic subunits 1Dx2 and 1Dy12 (Shewry 

et al., 2003). However, because the x- and y-type genes are so 

tightly linked in the D genome, it has not been possible to separate 

them by recombination and hence to determine their individual 

contributions to the breadmaking quality of wheat. 

Genetic engineering provides an opportunity to develop cultivars 

with new HMW-GS combinations in order to explore the 

molecular and biochemical bases for the breadmaking quality of 

wheat and to generate lines with improved or novel functional 

properties. This approach has also been used to investigate the 

effects of HMW-GS genes when expressed in other cereals. Thus, 

genes encoding HMW-GS have been expressed in wheat (Altpeter 

et al., 1996; Barro et al., 1997; Blechl and Anderson, 1996), tritordeum 

(Rooke et al., 1999a), maize (Sangtong et al., 2002), and rye 

(Altpeter et al., 2004). 

Expression of additional HMW-GS genes in wheat has resulted 

in increased amounts of HMW-GS protein and changes in the 

grain functional properties, the effects varying depending on the 

specific genes and genetic backgrounds. Thus, expression of 

subunit 1Ax1 in transgenic wheat resulted in lines with improved 

rheological properties while lines expressing subunit 1Dx5 

resulted in unsuitable properties for breadmaking (Barro et al., 

2003; Darlington et al., 2003). In particular, high levels of 

expression of subunit 1Dx5 resulted in greatly increased dough 

strength, being too strong for conventional breadmaking (Alvarez 

et al., 2001; Darlington et al., 2003; Rooke et al., 1999b). The 

differences between the results obtained with subunits 1Ax1 and 

1Dx5 may be due to the fact that subunit 1Dx5 has an additional 

cysteine residue which promotes a more highly cross-linked 

glutenin network (Darlington et al., 2003; Popineau et al., 2001). 

Also, because subunit 1Dx5 only occurs with 1Dy10 it has been 

suggested that the effects might be mitigated by increasing 

subunit 1Dy10 at the same time (Barro et al., 2003; Darlington 

et al., 2003). 

More recently, the effects of expression of transgenic subunits 

1Dx5 and/or 1Dy10 on flour properties in a wheat line which 

already expresses the endogenous forms of these subunits have 

been reported (Blechl et al., 2007). The results confirmed the 

effects of the increased levels of subunit 1Dx5 on dough strength 

but showed that increases in subunit 1Dy10 alone also resulted in 

curves typical of strong doughs. Furthermore, the expression of 

subunits 1Dx5 and 1Dy10 in the same lines failed to alleviate the 

overstrong characteristics associated with subunit 1Dx5 (Blechl 

et al., 2007). However, the effects of subunits 1Dx5 and 1Dy10 

have not been tested in wheat backgrounds that do not contain 

these subunits. We report here the transformation of wheat with 

subunits 1Ax1, 1Dx5 and 1Dy10 using a genetic background 

which does not express endogenous forms of subunits 1Dx5 and 

1Dy10 but the allelic subunits 1Dx2 and 1Dy12. This allows the 

effects of individual subunits 1Dx5 and 1Dy10, and the interactions 

between these and subunits 1Dx2 and 1Dy12 to be 

determined. 

2. Material and methods 

2.1. Plasmids 

Three plasmids, p1Dx5, p1Ax1 and pK-Dy10A containing, 

respectively, the genomic fragment for the HMW glutenin subunits 

1Dx5 (Anderson et al., 1989), 1Ax1 (Halford et al., 1992), and 1Dy10 

(Anderson et al., 1989), were used in combination with plasmid 

pACH25 (Christensen and Quail, 1996) which contains the bar gene 

that confers tolerance to phosphinothricin (PPT) under the control 

of maize ubiquitin promoter. 

E. León et al. / Journal of Cereal Science 49 (2009) 148–156 149 

2.2. Plant transformation and in vitro culture 

Immature scutella (0.5–1.5 mm in length) of wheat were used as 

target tissue for transformation by particle bombardment. Caryopses 

were harvested 16 days after anthesis and explants isolated 

as described by Barcelo and Lazzeri (1995). For each bombardment, 

30 scutella were excised and placed in the centre of a 9 cm Petri 

dish containing MP4 medium which consisted of MS (Murashige 

and Skoog, 1962) induction medium supplemented with 30 g l 1 

sucrose and 4 mg l 1 picloram. Explants were cultured in the dark 

at 25 C for 5 days prior to bombardment. Explants were subjected 

to a 4 h osmotic treatment, both before and after bombardment, in 

the same induction medium described above, but supplemented 

with 0.4 M mannitol. 

For bombardment, plasmids were precipitated onto 0.6 mm gold 

particles at 0.5 pmol mg 1 gold for pAHC25 and 0.75 pmol mg 1 

gold for the plasmids containing the HMW glutenin subunit genes. 

For each shot, 58 mg of coated particles were delivered at a pressure 

of 1100 psi using a PDS 1000/He particle gun (BioRad). Following 

bombardment, plates containing explants were cultured in MP4 

medium (see above) in the dark at 25 C for 3 weeks and then 

transferred to a RZ regeneration medium (Barro et al., 1998) supplemented 

with 2 mg l 1 of PPT and cultured for a further 3 weeks 

at 25 C with a photoperiod of 12 h. Surviving explants were 

transferred to R regeneration medium (Barro et al., 1998) supplemented 

with 2 mg l 1 PPT and cultured for a further two rounds of 

3 weeks each. Putative transgenic plants were then transferred to 

soil and grown to maturity in the greenhouse. Homozygous 

progeny of plants containing HMW-GS transgenes were identified 

by SDS–PAGE of endosperm proteins by single half-seed descent 

(see below). Homozygous lines were self-pollinated for three 

generations and assayed during 3 years, using a randomized 

complete block design with two replicates, as described (Barro 

et al., 2002). 

2.3. Genomic Southern blot analysis 

Total genomic DNA was extracted from leaf tissue using a CTAB 

method (Stacey and Isaac, 1994). Genomic DNA was digested with 

either EcoRV (which cuts once within the p1Ax1 and pK-Dy10A 

constructs) or ScaI (which cuts once within the p1Dx5 construct). 

Digested DNA was separated by electrophoresis using 0.8% (w/v) 

agarose gel, blotted onto Hybond-N þ membrane (Amersham 

Biosciences) and baked for 2 h at 80 C. Hybridization and signal 

detection were performed according to Roche Diagnostic protocols. 

Briefly, the membrane was prehybridized for 1 h with DIG Easy Hyb 

Granules (Roche Diagnostics) at 48 C. Southern filters were 

hybridized overnight at 48 C with PCR-generated digoxigeninlabelled 

probes produced using primers for the 1Ax1 (Rooke et al., 

1999a), 1Dx5 (D’Ovidio and Anderson, 1994) and 1Dy10 genes (5 0 - 

CCA CAA CAC CGA GCA CCA CAA-3 0 ,5 0 -GGG CGG CAC CAC AGT TTG 

CTC-3 0 ), with p1Ax1, p1Dx5 and pK-Dy10A as the DNA template, 

respectively. After hybridization the membrane was washed twice 

with a solution of 0.1% (w/v) SDS in 2 SSC for 15 min at 65 C. The 

DIG Luminescent Detection Kit (Roche Diagnostics) was used for 

signal detection. Finally, the membrane was exposed to AGFA 

medical X-ray (Agfa-Gevaert NV) film for 1–6 h. 

2.4. Protein and SDS–PAGE analysis 

The protein content of flour was calculated from the Kjeldahl 

nitrogen content (%N 5.7) and expressed on a dry matter basis. 

For protein extraction and quantification of the different protein 

fractions, all subplots in a year from the same line were bulked. 

Therefore, 30 individual seeds per line (ten seeds from each year) 

were crushed into a fine powder and used to extract the endosperm

150 

storage proteins. Gliadins were extracted in 60% (v/v) of aqueous 

ethanol using a rotary shaker for 40 min. Samples were centrifuged 

at 13,000 g for 5 min and the supernatant collected. Glutenins were 

extracted in 625 mM Tris–HCl pH 6.8, 5% (v/v) 2-mercaptoethanol, 

10% (v/v) glycerol, 0.02% (w/v) bromophenol blue and 2% (w/v) 

dithiothreitol in a 5:1 ratio (ml:mg) to wholemeal and separated 

using a Tris–borate buffer system and 10% (w/v) acrylamide gels 

(Shewry et al., 1995). Gliadin and glutenin contents were determined 

in the above extracts using a Bradford assay system (Bradford, 

1976). For densitometry analysis, glutenins were separated by 

SDS–PAGE gels and analysed using a Kodak Image Station 440CF 

and Kodak 1D Image Analysis Software using the SDS–PAGE 

Molecular Weight Standards from Bio-Rad as reference. 

Falling Number was determined according to the standard ICC 

method no. 107 (ICC, 1995). Sedimentation volume was determined 

according to Zeleny using the standard ICC method no. 116 (ICC, 

1994). 

SE-HPLC was carried out in duplicate at the Campden and 

Chorleywood Food Research Association (Chipping Campden, UK) 

as described by Morel et al (2000) and Millar (2003). 

2.5. Alveograph 

Approximately 0.8 kg samples of seeds per line and year were 

milled with a Chopin CD1 mill to obtain white flour. Alveograph 

tests were performed using an Alveograph MA 82 (Tripette et 

Renaud, France) and the dough properties determined in five 

replicates according to standard ICC method no. 121 (ICC, 1992a). 

The alveograms were evaluated to determine overpressure 

(tenacity) (P), the average length of the curve (extensibility) (L), the 

P/L ratio, the deformation work (W), the swelling index (G) and the 

elasticity index. 

2.6. Farinograph 

Approximately 0.2 kg samples of seeds per line and year were 

milled and mixed using a 50 g mixer for standard Farinograph 

analysis (Brabender GmbH & Co. KG, Germany) test. The Farinograph 

characteristics were determined in two different replicates 

per year according to standard ICC method no. 115/1 (ICC, 1992b). 

Farinograph parameters evaluated included water absorption, 

arrival time, departure time, stability, peak time, mixing tolerance 

time and degree of softening. 

2.7. Mixograph 

Dough mixing properties were determined with a 10 g Mixograph 

(National Manufacturing Co., Lincoln NE). Samples were 

mixed to optimum water absorption following 54-40A method 

(AACC, 2000). The mixing parameters determined were mixing 

time, peak resistance and resistance breakdown. 

2.8. Statistics 

Data were analysed using the SPSS version 11.0 statistical software 

package (SPSS Inc., Chicago, IL, USA). The general analysis of 

variance and the least significant difference pairwise comparisons 

of means were used to determine significant differences. 

3. Results 

Plasmids p1Ax1, p1DX5 and pKS-Dy10A, containing, respectively, 

the HMW-GS 1Ax1, 1Dx5 and 1Dy10, were used for transformation 

of the bread wheat cultivar Anza. Transformation 

efficiency (transformants per scutellum bombarded) for this 

cultivar was 0.4%. This cultivar contains four endogenous HMW-GS: 

E. León et al. / Journal of Cereal Science 49 (2009) 148–156 

1Bx7*, 1By8, 1Dx2 and 1Dy12 (Fig. 1A). Single seed descent was 

used to obtain homozygotes for lines containing HMW glutenin 

subunits transgenes by screening 30 half-seeds in each generation 

by SDS–PAGE. Homozygous transgenic lines expressing the HMW- 

GS 1Ax1, 1Dx5 and 1Dy10 transgenes (Fig. 1A) were termed T580, 

T581 and T590, respectively. These lines were multiplied by selfpollinating 

for three generations and assayed for 3 years as 

described (Barro et al., 2002). 

Southern blot analysis was complicated by the presence of 

cross-hybridizing bands in the control line, which was probably 

due to the sequence similarity between the endogenous and 

transgenic HMW-GS. Nevertheless, the banding pattern after 

hybridization showed that transgenic lines contained four, two and 

three insertion events for HMW-GS 1Ax1, 1Dx5 and 1Dy10, 

respectively (Fig. 1B). 

3.1. Protein characterization 

The total protein contents of the flours ranged from 11.7% to 

13.7% for the control line and line T581 transformed with HMW-GS 

1Dx5, respectively (Table 1). All three transgenic lines had higher 

protein contents than the control line but only the differences 

between lines T580 and T581, containing HMW-GS 1Ax1 and 1Dx5, 

respectively, were statistically significant in comparison to the 

control line (Table 1). As shown in Table 1, the glutenin fraction 

decreased from 4.21 mg/mg flour in the control line to 3.25 and 3.62 

in lines T580 and T581, but increased to 5.28 in line T590. In 

contrast, the gliadin contents increased significantly in all three 

transgenic lines with respect to the control line, with no significant 

differences among the three transgenic lines. These changes are 

also reflected in the glutenin/gliadin ratio. Thus, lines T580 and 

T581 had lower glutenin/gliadin ratios than the control line, but 

line T590 had a similar ratio to that of control line (Table 1). 

The expression of HMW-GS in the transgenic lines resulted in 

changes in amounts and proportions of HMW-GS in the endosperm. 

Thus, the transgenic subunits 1Ax1, 1Dx5 and 1Dy10 

accounted for 25.2%, 20.3% and 17.9%, respectively, of the total 

HMW in the three transgenic lines (Table 1). These increases in 

HMW subunit content were associated with decreases in LMW 

subunit content in all transgenic lines, but the differences were only 

significant for line T581 expressing the 1Ax1 subunit (Table 1). 

Expression of the transgenic HMW-GS was also associated with 

a significant decrease in the amounts of endogenous HMW-GS 

(Table 1). Furthermore, although all endogenous subunits were 

decreased, the relative effects varied. The expression of subunits 

1Ax1 and 1Dx5 in lines T580 and T581, respectively, led to greater 

decreases in the contents of the endogenous subunit 1Dx2 (Table 

1), with the Dx2/Dy12 ratio decreasing significantly from 1.11 in the 

control line to 0.83 and to 0.88 in lines T580 and T581, respectively. 

Furthermore, the ratio of x-type/y-type HMW-GS also increased in 

lines T580 and T581, respectively. In contrast, the contents of the 

endogenous subunits 1Dx2 and 1Dy12 were decreased in line T590 

expressing subunit 1Dy10 but to a similar extent (Table 1). Therefore, 

the ratio of Dx2/Dy12 in this line was comparable to that in the 

control line (Table 1), although the ratio of the x-type/y-type was 

decreased from 1.25 in the control line to 0.91 as a consequence of 

the addition of a new y-type subunit. 

There is a well-established correlation between the breadmaking 

performance of flour and the proportion of high molecular mass 

glutenin polymers (Gupta and MacRitchie, 1994; Popineau et al., 

1994). The proportions of monomeric, oligomeric and polymeric 

gluten proteins in the flours were therefore determined by SE-HPLC 

essentially as described by Morel et al. (2000). This method uses 

sonication in 1% (w/v) SDS in 0.1 M phosphate buffer, pH 6.9, to 

extract the total grain proteins, which are then separated into five 

fractions corresponding broadly to high molecular mass glutenin

A 

B 

1Ax1 

polymers (F1), lower molecular mass glutenin polymers (F2), ugliadins 

(F3), a- and g-type gliadins (F4) and albumins and globulins 

(F5). The sonication procedure results in shearing of glutenin 

polymers and hence the sizes of the polymers separated by SE- 

HPLC do not accurately reflect the sizes of those present in vivo. 

However, the relative amounts of the peaks do relate to dough 

strength with %F1/%F2 and (%F3 þ %F4)/%F1 showing particularly 

strong correlations (Millar, 2003). 

All three transgenic lines had higher values for %F1 and %F1/%F2, 

indicating that they had higher proportions of high molecular mass 

glutenin polymers. This effect was also greater in line T590 

expressing the 1Dy10 subunit than in lines T580 and T581. 

However, values for (%F3 þ %F4)/%F1 decreased in all the transgenic 

lines relative to the untransformed parent (Table 1). 

3.2. Technological properties of flour 

T580 

1Dx2 

1Bx7* 

1By8 

1Dy12 

The potential breadmaking quality of the flours was determined 

over 3 years using a range of widely-used grain and dough testing 

methods (Table 2). The Falling numbers of all of the samples were 

higher than the recommended limit of 180 s with no differences 

among all four lines. The grain test weight was also similar for all 

four lines. The thousand seed weight of line T580 was higher than 

that of the other lines. The Zeleny sedimentation was increased in 

all three transgenic lines and this was greatest in the line 

expressing subunit 1Dy10. 

The rheological properties and water absorption of the dough 

were determined using a Brabender Farinograph (Fig. 2) and10g 

Mixograph (Fig. 3) while the dough viscoelastic properties were 

determined using a Chopin Alveograph (Fig. 4). The Farinograph 


WT 

WT 

T581 

1Dx5 

T590 

T580 WT WT T581 WT T590 

Fig. 1. (A) SDS–PAGE of seed protein extracts from transgenic wheat lines and their non-transformed parent (WT). The location of the HMW-GS native to Anza (1Dx2, 1Bx7*, 1By9 

and 1Dy12) are indicated. (B) Southern blot analysis of the transgenic lines. Genomic DNA from lines T580, T581 and T590 was digested with EcoRV, ScaI and EcoRV, respectively, 

and hybridized with probes amplified from plasmids p1Ax1, p1Dx5 and pK-Dy10A. Four, three and two extra hybridized bands, compare to non-transformed parent (WT), are 

present in DNA of lines T580, T581 and T590, respectively. 

WT 

1Dy10 

characteristics of transgenic lines are shown in Table 2. No 

significant differences were detected for water absorption but the 

rheological properties differed with the arrival and peak times, 

increasing significantly in all three transgenic lines in comparison 

to the control line. The transgenic lines expressing subunits 1Ax1 

and 1Dx5 gave similar values for arrival time and peak time while 

the line expressing subunit 1Dy10 gave the highest values for 

these two parameters (Table 2). The dough stability and departure 

time were also significantly higher for the line transformed with 

subunit 1Dy10. Dough stability is an indicator of the overall 

quality of the protein in the flour and this increased from 1.3 in the 

control line to 11.9 in the line expressing subunit 1Dy10 (Table 2). 

The mixing tolerance index (MTI) is an indicator of how well the 

dough will perform during the critical final stages of mixing and is 

related to strength (Shuey, 1984). Thus, a high MTI means that the 

dough will tend to break down quickly while a low MTI may 

indicate that the dough will require longer mixing to fully 

develop. The MTI was decreased significantly in all three transgenic 

lines with respect to that of the control line. However, this 

decrease was greater for transgenic line T590 expressing subunit 

1Dy10, for which the MTI value fell from 132.8 (BU) in the control 

line to 35.2 (BU) (Table 2). The degree of softening varied from 

132.5 (BU) in the control line to 56.1 (BU) in the line transformed 

with subunit 1Dy10. Higher values of degree of softening, 

measured in Brabender units, are characteristic of weak flours and 

lower values of strong flours. 

The rheological properties of the transgenic wheats were also 

determined using a 10 g Mixograph (Fig. 3). The Mixograph is 

widely used in cereal research as it measures a range of rheological 

parameters that relate to the behaviour of the dough in bread

152 

Table 1 

Characterization of the storage proteins in transgenic and control lines of wheat cultivar Anza 

Parameter Line 

Anza T580 T581 T590 

Endogenous HMW 7* þ 8, 2 þ 12 7* þ 8, 2 þ 12 7* þ 8, 2 þ 12 7* þ 8, 2 þ 12 

Transgenic HMW NA 1Ax1 1Dx5 1Dy10 

Number of insertions NA 4 2 3 

Crude protein (%) a 

11.7 0.3 13.2 0.4 13.7 0.6 12.2 0.3 

Glutenin (mgmg 1 flour) b 

4.2 0.2 3.3 0.2 3.6 0.4 5.3 0.2 

Gliadin (mgmg 1 flour) b 

2.7 0.1 3.4 0.1 3.7 0.4 3.5 0.1 

Glutenin/gliadin 

HMW composition 

1.53 0.04 0.96 0.05 0.98 0.06 1.51 0.04 

c 

1Ax1 NA 25.2 0.6 NA NA 

1Dx2 27.8 0.6 16.8 0.5 19.1 0.4 21.5 0.7 

1Dx5 NA NA 20.3 0.8 NA 

1Bx7* 27.7 0.9 22.2 0.3 23.4 0.5 26.2 0.2 

1By8 19.4 0.9 15.5 0.3 15.6 0.9 15.9 0.9 

1Dy10 NA NA NA 17.9 0.6 

1Dy12 25.1 0.4 20.3 0.8 21.6 0.9 18.5 0.7 

making and other food processing systems. The machine measures 

the resistance as the dough is mixed, the most important parameters 

being the mixing time (time to maximum resistance), the 

peak resistance (maximum resistance obtained during mixing) and 

the resistance breakdown (the loss of resistance on over-mixing). In 


HMW (% glutenins) 31.7 0.79 32.4 1.21 36.8 0.52 32.9 0.45 

LMW (% glutenins) 68.3 0.61 67.6 1.13 63.2 0.49 67.1 0.51 

HMW/LMW ratio 0.47 0.01 0.49 0.02 0.58 0.01 0.49 0.01 

1Dx2/1Dy12 ratio 1.11 0.03 0.83 0.05 0.88 0.06 1.16 0.07 

1Bx7/1By8 ratio 1.43 0.11 1.43 0.04 1.50 0.12 1.64 0.09 

x-type/y-type ratio 1.25 0.06 1.80 0.07 1.69 0.06 0.91 0.03 

SE-HPLC d 

%F1 12.7 0.35 13.6 0.35 13.0 0.05 14.1 0.30 

%F1/%F2 0.60 0.02 0.61 0.01 0.62 0.00 0.68 0.01 

(%F3 þ %F4)/%F1 

NA, not applicable. 

4.08 0.12 3.63 0.12 3.88 0.03 3.54 0.10 

a 

Kjeldahl method. Average of six replications standard error. 

b 

Glutenin and gliadins were determined using a Bradford assay system. Average of 30 replications standard error. 

c 

HMW composition was determined by densitometry analysis. Average of 30 replications standard error. 

d 

SE-HPLC was carried out in duplicate by bulking samples from the third year. Average standard error. 

general, strong doughs have long mixing times, high peak resistances 

and low resistance breakdown. Transformation with the 

three HMW subunit genes led in all cases to increased dough 

strength, with quantitative differences depending on the subunit 

expressed (Fig. 3, Table 2). Thus, the mixing time increased from 

Table 2 

Technological properties of the transgenic and control lines of wheat cultivar Anza 

Parameter a 

Line 

Anza T580 T581 T590 

Falling number (s) 350 9.5 350 9.8 350 3.0 346 1.2 

Grain test weight (g l 1 ) 79.9 3.0 79.8 5.0 79.4 4.3 80.2 4.1 

Thousand seed weight (g) 38.4 0.9 42.1 1.6 37.5 0.7 39.7 0.9 

Zeleny sedimentation (ml) 20.2 0.4 28.2 2.0 31.3 3.8 41.1 3.0 

Farinograph 

Water absorption 72.2 1.5 72.8 0.9 73.8 2.8 73.7 1.2 

Arrival time (min) 2.1 0.1 3.4 0.6 3.7 0.2 4.6 0.3 

Peak time (min) 2.7 0.3 4.4 0.5 4.8 0.4 6.6 0.8 

Dough stability (min) 1.3 0.3 2.7 0.2 2.7 0.3 11.9 2.2 

Departure time (min) 3.5 0.4 6.0 0.6 6.3 0.6 16.5 2.1 

Mixing Tolerance Index (BU) 132.8 14.4 77.5 9.2 80.8 7.6 35.2 3.8 

Degree of softening (BU) 132.5 18.9 76.7 8.5 71.7 9.2 56.1 6.2 

Mixograph 

Mixing time (s) 30.0 72.0 87.0 120.0 

Peak resistance (AU) 58.0 75.0 60.0 61.5 

Resistance breakdown (%) 37.9 26.7 18.3 18.7 

Alveograph 

Tenacity, P (mm) 42.2 5.1 65.8 3.1 71.7 3.7 79.0 5.4 

Extensibility, L (mm) 94.2 4.5 91.8 3.0 61.0 0.6 82.3 5.9 

P/L ratio 0.4 0.03 0.7 0.04 1.2 0.07 1.0 0.12 

W-value ( 10 4 J) 97 8.2 170 13.2 180 12.1 258 4.9 

a 

Values are the average of six replications standard error, except for Mixograph parameters for which bulked samples from the third year were used.

30 s in the control line to 72 in that with subunit 1Ax1, to 87 in that 

with subunit 1Dx5, and to 120 in that with subunit 1Dy10 (Table 2). 

However, the peak resistance values were similar for all lines except 

for the line T580 expressing subunit 1Ax1 where it was higher 

(Table 2). The resistance breakdown decreased from 37.9% in the 

control line to 26.7, to 18.3, and to 18.7 in the lines with subunits 

1Ax1, 1Dx5 and 1Dy10, respectively (Table 2). 

The Chopin Alveograph is a dough-testing instrument that 

simulates the expansion of bubbles in fermenting dough and as 

such the parameters derived from this test should give some 

indication as to how the dough will act during fermentation. The 

Resistance (AU) 


100 

50 

0 

100 

50 

0 


Fig. 2. Farinograms of the doughs prepared from non-transformed control (Anza) and the flour from transgenic lines T580, T581 and T590, expressing the HMW-GS 1Ax1, 1Dx5 and 

1Dy10, respectively. 

most important parameters obtained from an alveogram are the 

tenacity (P) of the dough (expressing the resistance of the dough to 

deformation), the dough extensibility (L), the P/L ratio (which gives 

a general indication of the viscoeslastic properties of the dough) 

and the deformation energy (W), which is proportional to the 

energy required for deformation. This W-value is used to measure 

the dough strength by bakers. In general, low W values are related 

to weak flours and high values to strong flours. Transformation 

with the HMW subunit genes led to increased tenacity (P) in all 

three transgenic lines with respect to the control line (Table 2). 

However, the dough extensibility (L) decreased significantly from 

Anza T580 (1Ax1) 

0 

0 90 180 270 360 

0 90 180 270 360 

Time (s) Time (s) 

T581 (1Dx5) T590 (1Dy10) 

0 

0 90 180 270 360 

0 90 180 270 360 

Time (s) Time (s) 

Fig. 3. Mixograph curves of the doughs prepared from non-transformed control (Anza) and the flour from transgenic lines T580, T581 and T590, expressing the HMW-GS 1Ax1, 

1Dx5 and 1Dy10, respectively. Resistance in arbitrary units (AU) is plotted against time in minutes. 



100 

50 

100 

50

154 

Fig. 4. Alveograph curves of the doughs prepared from non-transformed control (Anza) and the flour from transgenic lines T580, T581 and T590, expressing the HMW-GS 1Ax1, 

1Dx5 and 1Dy10, respectively. 

94.2 in the control line to 61.0 in the line transformed with subunit 

1Dx5. The P/L ratio increased significantly from 0.4 in the control 

line to 0.7, 1.2, and 1.0 in that for lines transformed, respectively, 

with subunits 1Ax1, 1Dx5 and 1Dy10 (Table 2). Finally, the W value 

increased from 96.7 in the control line to 170.0, 180.3, and 258.3 in 

the lines with subunits 1Ax1, 1Dx5 and 1Dy10, respectively. 

4. Discussion 

Dough mixing and development are critical processes in bread 

production. Much research has therefore been conducted to 

determine the parameters that influence them and to facilitate the 

production of high quality bread that satisfies not only processing 

requirements but also customer expectations. The HMW-GS of 

wheat are major determinants of the breadmaking quality of wheat 

and, in particular, the subunit pair 1Dx5 þ 1Dy10 and subunit 1Ax1 

have high quality scores (Payne, 1987; Shewry et al., 2003). The 

production of transgenic lines of wheat differing in their HMW-GS 

composition makes it possible to determine the effects of individual 

subunits on the rheological and technological properties of 

wheat gluten. However, the behaviour of individual subunits may 

also be affected by the HMW-GS composition in which they are 

expressed as subunit interactions are crucial in forming the glutenin 

polymers. In this work we report the effects of the HMW-GS 

1Ax1, 1Dx5 and 1Dy10 on the breadmaking quality of the bread 

wheat cultivar Anza that contain the HMW-GS pairs 1Dx2 þ 1Dy12 

and 1Bx7* þ 1By8 and is null for the Glu-A1 locus. Wheat cultivars 

containing subunits 1Dx2 þ 1Dy12 generally have lower dough 

resistance than those containing subunits 1Dx5 þ 1Dy10 (Shewry 

et al., 2003). Therefore, this genotype is an excellent background to 

determine separately the effects of subunits 1Dx5 and 1Dy10 and 

subunit 1Ax1 on breadmaking quality. Although subunits 1Dx5 and 

1Dy10 have been separately introduced into the bread wheat 

Bobwhite (Blechl et al., 2007), this cultivar already contain both 

subunits which could influence the behaviour of the additional 

subunit protein encoded by the transgenes. 

This is therefore the first report on the characterization of wheat 

expressing subunit 1Dy10 in the absence of subunit 1Dx5. The 


results show that subunit 1Dy10 had a greater effect on breadmaking 

quality than subunits 1Ax1 and 1Dx5. However, the 

expression of subunits 1Ax1 and 1Dx5 also led to increased 

breadmaking quality. This is not surprising for subunit 1Ax1 as this 

protein has been reported to have positive effects in a range of 

wheat backgrounds (Barro et al., 2003; Darlington et al., 2003). 

However, the beneficial effect of subunit 1Dx5 was unexpected as 

previous transgenic wheat lines expressing this subunit had overstrong 

doughs unsuitable for breadmaking (Darlington et al., 2003; 

Rooke et al., 1999b). 

The expression levels of the transgenic 1Ax1, 1Dx5 and 1Dy10 

HMW-GS were comparable to those of the endogenous subunits. 

However, although lines T581 and T590 expressed similar levels of 

subunits 1Dx5 and 1Dy10 (Table 1) they had different effects on 

other aspects of protein composition, including changes in the 

ratios of glutenin/gliadin, the HMW/LMW, the 1Dx2/1Dy12, x-type/ 

y-type HMW-GS and the proportions of high molecular mass glutenin 

polymers (Table 1). In contrast, the lines transformed to 

express subunits 1Ax1 and 1Dx5 showed more similar changes in 

protein composition, with higher protein contents and decreased 

ratios of glutenin/gliadin and of 1Dx2/1Dy12. In addition, both 

transgenic lines showed similar increases in the ratio of x-type/ytype 

subunits compared to the control line (Table 1) while the line 

expressing subunit 1Dx5 also showed a significant increase in the 

HMW to LMW ratio. The increases in the HMW/LMW ratio and 

the changes in the glutenin/gliadin ratio as a consequence of the 

expression of transgenic HMW-GS have been reported previously 

(Alvarez et al., 2001; Blechl et al., 2007; Rakszegi et al., 2005). These 

changes also indicate that the new HMW-GS affect the structure of 

the glutenin network with the x-type subunits (i.e. 1Ax1 and 1Dx5) 

having similar effects which are different to those of subunit 1Dy10. 

In the case of T590 (expressing 1Dy10) the composition is more 

comparable to that of the control line. For example, both the control 

line and line T590 had similar protein contents and ratios of glutenin/gliadin, 

HMW/LMW and 1Dx2/1Dy12 (Table 1). However, the 

x-type/y-type ratio was lower in line T590 than in the control line 

while the proportion of high molecular mass polymers was 

significantly increased.

The transgenic lines were analysed over 3 years using standard 

dough testing methods (Farinograph, Mixograph and Alveograph). 

This confirmed that the expression of all three subunits increased 

the dough strength (and hence breadmaking quality) of the cultivar 

Anza. Furthermore, a stepwise increase was observed depending on 

the HMW-GS expressed. Subunits 1Ax1 and 1Dx5 resulted in 

similar changes in the protein composition and structure, as 

described above, and this is reflected in their similar effects on 

breadmaking quality. Thus, the Farinograph showed increased 

dough resistance and stability and increased dough tolerance to 

mixing, the values being characteristic of medium grade flours 

(Table 2). Improvements in the rheological properties of the 

transgenic lines expressing subunits 1Ax1 and 1Dx5 were 

confirmed using the 10 g Mixograph (Fig. 3). Both transgenic lines 

had longer mixing times and lower values for resistance breakdown 

than the control line (Table 2). Subunit 1Ax1 has previously 

been reported to either increase dough resistance (Barro et al., 

2003; Darlington et al., 2003) or to have little or no effect on dough 

resistance (Alvarez et al., 2001). In the former case, the genotype 

was derived from an Olympic Gabo cross and contained no 1Ax 

subunit and in the latter report the genotypes (BW and Federal) 

both expressed subunit 1Ax2*. However, the expression of subunit 

1Ax2* was suppressed in the transgenic line of Federal and the 

breadmaking quality was comparable to those of control lines 

expressing subunit 1Ax2*. It was therefore suggested that subunit 

1Ax1 had a similar effect to subunit 1Ax2*. 

Most previous studies of lines expressing the 1Dx5 transgene 

reported overstrong dough properties which were unsuitable for 

breadmaking. In contrast, in the present study the expression of 

subunit 1Dx5 in the breadmaking variety Anza resulted in 

improved dough properties for breadmaking. Rooke et al. (1999b) 

suggested that although the transgenic subunits are incorporated 

into glutenin polymers, their organization within the polymers may 

differ from that in the corresponding control lines. An imbalance in 

the ratio of x-type/y-type subunits may therefore be responsible for 

the effects of expression of the 1Dx5 transgene without expression 

of subunit 1Dy10 (Alvarez et al., 2001). Our results indicate that 

subunits 1Dx5 and 1 Ax1 are similar in this respect, having similar 

effects on protein composition and dough properties. 

In contrast, transformation with subunit 1Dy10 led to stronger 

flours than those made from the control line and lines transformed 

with subunits 1Ax1 and 1Dx5. Farinograph parameters such as 

dough stability and peak time were increased by 9.2-fold and 2.4fold, 

respectively, in line T590 (expressing 1Dy10) with respect to 

the control line, while the mixing tolerance index and the degree of 

softening decreased by 3.8-fold and 2.4-fold, respectively. The 

Mixograph and Alveograph parameters similarly showed that 

subunit 1Dy10 had a greater effect on breadmaking quality. Thus, 

the mixing time was increased by four-fold and resistance breakdown 

decreased by two-fold in line T590 compared with the 

control line. The W value was also increased by 2.7-fold in line T590 

compared to the control line. 

In this study, subunits 1Dx5 and 1Dy10 were compared in 

a cultivar of low breadmaking quality which provides an appropriate 

background to explore their effects on dough properties. 

Small scale mixing experiments incorporating purified 1Dx5 and 

1Dy10 subunits into dough (Uthayakumaran et al., 2000) showed 

that subunit 1Dx5 resulted in longer mixing times and lower 

resistance breakdown than subunit 1Dy10. In contrast, our results 

showed that expression of subunits 1Dx5 and 1Dy10 in transgenic 

wheat resulted in doughs with similar low resistance breakdown 

but that dough made from flour expressing the subunit 1Dy10 

transgene had a longer mixing time. This effect of the subunit 

1Dy10 transgene on dough strength is in agreement with the 

results reported by Blechl et al. (2007) for the expression of subunit 

1Dy10 in Bobwhite. This expression of subunit 1Dy10 was also 


associated with a greater effect on the proportion of high molecular 

mass glutenin polymers than those observed with subunits 1Dx5. 


The authors acknowledge funding by the Spanish CICYT (project 

AGL2007-65685-C02-01). Rothamsted Research receives grantaided 

support from the Biotechnology and Biological Sciences 

Research Council (BBSRC) of the UK. Plasmid pK-Dy10A was 

provided by Dr. Ann E. Blechl. The technical assistance of Azahara 

Vida is also acknowledged. 

References 

AACC, 2000. Physical Dough Tests. Mixograph Method tenth ed Method 54-40A. 

Altpeter, F., Vasil, V., Srivastava, V., Vasil, I.K., 1996. Integration and expression of the 

high molecular weight glutenin subunit 1Ax1 gene into wheat. Nature 


Altpeter, F., Popelka, J.C., Wieser, H., 2004. Stable expression of 1Dx5 and 1Dy10 

high-molecular-weight glutenin subunit genes in transgenic rye drastically 

increases the polymeric glutelin fraction in rye flour. Plant Molecular Biology 

54, 783–792. 

Alvarez, M.L., Gómez, M., Carrillo, J.M., Vallejos, R.H., 2001. Analysis of dough 

functionality of flours from transgenic wheat. Molecular Breeding 8, 103–108. 

Anderson, O.D., Greene, F.C., Yip, R.E., Halford, N.G., Shewry, P.R., Malpica- 

Romero, J.M., 1989. Nucleotide sequences of the two high-molecular-weight 

glutenin genes from the D-genome of a hexaploid bread wheat, Triticum aestivum 

L. cv Cheyenne. Nucleic Acids Research 17, 461–462. 

Barcelo, P., Lazzeri, P.A., 1995. Transformation of cereals by microprojectile 

bombardment of immature inflorescence and scutellum tissues. In: Jones, H. 

(Ed.), Methods in Molecular Biology: Plant Gene Transfer and Expression 

Protocols. Humana Press, Totowa, NJ, pp. 113–123. 

Barro, F., Rooke, L., Bekes, F., Gras, P., Tatham, A.S., Fido, R., Lazzeri, P.A., Shewry, P.R., 

Barcelo, P., 1997. Transformation of wheat with high molecular weight subunit 

genes results in improved functional properties. Nature Biotechnology 15, 

1295–1299. 

Barro, F., Cannell, M.E., Lazzeri, P.A., Barcelo, P., 1998. The influence of auxins on 

transformation of wheat and tritordeum and analysis of transgene integration 

patterns in transformants. Theoretical and Applied Genetics 97, 684–695. 

Barro, F., Barcelo, P., Lazzeri, P.A., Shewry, P.R., Martín, A., Ballesteros, J., 2002. Field 

evaluation and agronomic performance of transgenic wheat. Theoretical and 


Barro, F., Barcelo, P., Lazzeri, P.A., Shewry, P.R., Ballesteros, J., Martin, A., 2003. 

Functional properties of flours from field grown transgenic wheat lines 

expressing the HMW subunit 1Ax1 and 1Dx5 genes. Molecular Breeding 12, 

223–229. 

Blechl, A.E., Anderson, O.D., 1996. Expression of a novel high-molecular-weight 

glutenin subunit gene in transgenic wheat. Nature Biotechnology 14, 875–879. 

Blechl, A., Lin, J., Nguyen, S., Chan, R., Anderson, O.D., Dupont, F.M., 2007. Transgenic 

wheats with elevated levels of Dx5 and/or Dy10 high-molecular-weight glutenin 

subunits yield doughs with increased mixing strength and tolerance. 


Bradford, M.M., 1976. A rapid and sensitive method for quantitation of microgram 

quantities of protein utilizing the principle of protein-dye-binding. Analytical 


Christensen, A.H., Quail, P.H., 1996. Ubiquitin promoter-based vectors for high-level 

expression of selectable and/or screenable marker genes in monocotyledonous 

plants. Transgenic Research 5, 213–218. 

Darlington, H., Fido, R., Tatham, A.S., Jones, H., Salmon, S.E., Shewry, P.R., 2003. 

Milling and baking properties of field grown wheat expressing HMW subunit 

transgenes. Journal of Cereal Science 38, 301–306. 

D’Ovidio, R., Anderson, O.D., 1994. PCR analysis to distinguish between alleles of 

a member of a multigene family correlated with wheat bread-making quality. 


Gupta, R.B., MacRitchie, F., 1994. Allelic variation at glutenin subunit and gliadin 

loci, Glu-1, Glu-3 and Gli-1 of bread wheats: biochemical basis of the allelic 

effects on dough properties. Journal of Cereal Science 19, 19–28. 

Halford, N.G., Field, J.M., Blair, H., Urwin, P., Moore, K., Robert, L., Thompson, R., 

Flavell, R.B., Tatham, A.S., Shewry, P.R., 1992. Analysis of HMW glutenin subunits 

encoded by chromosome 1A of bread wheat (Triticum aestivum L.) indicates 

quantitative effects on grain quality. Theoretical and Applied Genetics 83, 

373–378. 

ICC, 1992a. Method for using of the Chopin-Alveograph. International Association 

for Cereal Science and Technology Method No. 121. 

ICC, 1992b. Method for using the Brabender Farinograph. International Association 

for Cereal Science and Technology Method No. 115/1. 

ICC, 1994. Determination of the sedimentation value (according to Zeleny) as an 

approximate measure of baking quality. International Association for Cereal 

Science and Technology Method No. 116/1. 

ICC, 1995. Determination of the ‘‘Falling Number’’ according to Hagberg-Perten as 

a mesure of the degree of Alpha-Amylase activity in grain and flour. International 

Association for Cereal Science and Technology Method No. 107/1.

156 

Millar, S.J., 2003. The development of near infrared (NIR) spectroscopy calibrations 

for the prediction of wheat and flour quality. Project Report No. 310. HGCA, UK. 

101. 

Morel, M.-H., Dehlon, P., Autran, J.C., Leygue, J.P., Bar-L’Helgouac’h, C., 2000. Effects 

of temperature, sonication time and power settings on size distribution and 

extractability of total wheat flour proteins as determined by size-exclusion 

high-performance liquid chromatography. Cereal Chemistry 77, 685–691. 

Murashige, T., Skoog, F., 1962. A revised medium for rapid growth and bio assays 

with tobacco tissue cultures. Physiologia Plantarum 15, 473–497. 

Payne, P.I., 1987. Genetics of wheat storage proteins and the effect of allelic variation 

on bread-making quality. Annual Review of Plant Physiology 38, 141–153. 

Popineau, Y., Cornec, M., Lefebvre, J., Marchylo, B., 1994. Influence of high Mr glutenin 

subunits on glutenin polymers and rheological properties of glutens and 

gluten subfractions of near-isogenic lines of wheat Sicco. Journal of Cereal 


Popineau, Y., Deshayes, G., Lefebvre, J., Fido, R., Tatham, A.S., Shewry, P.R., 2001. 

Prolamin aggregation, gluten viscoelasticity, and mixing properties of transgenic 

wheat lines expressing 1Ax and 1Dx high molecular weight glutenin 

subunit transgenes. Journal of Agricultural and Food Chemistry 49, 395–401. 

Rakszegi, M., Bekes, F., Lang, L., Tamas, L., Shewry, P.R., Bedo, Z., 2005. Technological 

quality of transgenic wheat expressing an increased amount of a HMW glutenin 

subunit. Journal of Cereal Science 42, 15–23. 

Rooke, L., Barro, F., Tatham, A.S., Fido, R., Steele, S., Bekes, F., Gras, P., Martin, A., 

Lazzeri, P.A., Shewry, P.R., Barcelo, P., 1999a. Altered functional properties of 


tritordeum by transformation with HMW glutenin subunit genes. Theoretical 

and Applied Genetics 99, 851–858. 

Rooke, L., Bekes, F., Fido, R., Barro, F., Gras, P., Tatham, A.S., Barcelo, P., Lazzeri, P., 

Shewry, P.R., 1999b. Overexpression of a gluten protein in transgenic wheat 

results in greatly increased dough strength. Journal of Cereal Science 30,115–120. 

Sangtong, V., Moran, D., Chikwamba, R., Wang, K., Woodman-Clikeman, W., 

Long, M., Lee, M., Scott, M., 2002. Expression and inheritance of the wheat Glu- 

1DX5 gene in transgenic maize. Theoretical and Applied Genetics 105, 937–945. 

Shewry, P.R., Tatham, A.S., Fido, R.J., 1995. Separation of plant proteins by electrophoresis. 

In: Jones, H. (Ed.), Methods in Molecular Biology: Plant Gene Transfer 

and Expression Protocols. Humana Press, Totowa, NJ, pp. 399–422. 

Shewry, P.R., Halford, N.G., Tatham, A.S., Popineau, Y., Lafiandra, D., Belton, P.S., 

2003. The high molecular weight subunits of wheat glutenin and their role in 

determining wheat processing properties. Advances in Food and Nutrition 

Research 45, 219–302. 

Shuey, W.C., 1984. Interpretation of the Farinogram. In: D’Appolonia, B.L., 

Kunerth, W.H. (Eds.), The Farinograph Handbook. American Association of 

Cereal Chemistry, St. Paul, MN, pp. 31–32. 

Stacey, J., Isaac, P.G., 1994. Isolation of DNA from plants. In: Isaac, P.G. (Ed.), Methods 

in Molecular Biology: Protocols for nucleic acid analysis by nonradiactive 

probes. Humana Press, Totowa, NJ, pp. 9–15. 

Uthayakumaran, S., Stoddard, F.L., Gras, P.W., Bekes, F., 2000. Effects of incorporated 

glutenins on functional properties of wheat dough. Cereal Chemistry 77, 

737–743.

Wheat glutenin proteins assemble into a nanostructure with unusual 

structural features 

Sarah H. Mackintosh a , Susie J. Meade a , Jackie P. Healy b , Kevin H. Sutton a , Nigel G. Larsen a , 

Adam M. Squires c,1 , Juliet A. Gerrard b, * 

a 

Food and Biomaterials Innovation, New Zealand Institute for Crop & Food Research Ltd., Private Bag 4704, Christchurch, New Zealand 

b 

School of Biological Sciences, University of Canterbury, 20 Kirkwood Ave, Private Bag 4800, Christchurch 8020, New Zealand 

c 

Cavendish Laboratory, Cambridge University, Cambridge, UK 

article info 





Keywords: 

Glutenins 

X-ray fibre diffraction 

Wheat protein 

Protein nanostructure 


abstract 

Proteins have an innate ability to adopt specific three dimensional 

forms with a plethora of functions. This affords the potential 

for design of nanoscale materials via the manufacture of selfassembled 

protein nanostructures, with properties that are not 

easily duplicated with traditional organic molecules and other 

polymers (Sanford and Kumar, 2005). Significantly, protein nanomaterials 

incorporate surface chemical functionality via the amino 

acid side chains, therefore providing additional possibilities for the 

construction of advanced nanomaterials, with a combination of 

structural and (bio)chemical function. The enormous promise for 

the use of peptides and protein as smart materials has attracted 

much recent attention (Shen et al., 2006; Waterhouse and Gerrard, 

2004). 

Amyloid fibrils are highly ordered nanofibres of insoluble 

protein, rich in b-strand content (Dobson, 1999). The mature fibril 

comprises a number of protofilaments (sub-structures) that typically 

twist together to form a nanostructure with a diameter of 


E-mail address: juliet.gerrard@canterbury.ac.nz (J.A. Gerrard). 

1 

Present address: School of Chemistry, University of Reading, Whiteknights, 

Reading RG6 6AD, UK. 


doi:10.1016/j.jcs.2008.08.003 





The proteins of wheat have a known propensity to aggregate into a variety of forms. We report here 

a novel nanostructure from wheat proteins, derived from a crude extract of high molecular weight 

glutenins. The structure was characterised by a significant thioflavin T (ThT) fluorescence and a fibrillar 

morphology by transmission electron microscopy (TEM). The ThT fluorescence and TEM data are 

suggestive of an amyloid structure, but the X-ray fibre diffraction data show a reflection pattern (4.02, 

4.2–4.3, 4.6, 12.9, 19.3 and 38.7 Å) inconsistent with both the classic amyloid form and the previously 

described b-helix structure. The 4.6 Å reflection is consistent with that predicted for the amyloid inter-bstrand, 

and the absence of the inter-b-sheet distance at z10–11 Å is not unprecedented in amyloid-like 

structures. However, our observed X-ray reflection pattern has not been previously reported and suggests 

a novel wheat glutenin nanostructure. 


approximately 5–15 nm, and a rope-like appearance when imaged 

at high magnifications (Serpell, 2000). Whilst there are a number of 

competing structural models (Makin and Serpell, 2005), amyloid 

fibrils are defined by observations of characteristic electron 

microscopy, specific chemical staining and X-ray fibre diffraction. 

The specific arrangement of the b-strands within the mature fibril 

is defined as a cross-b pattern and leads to characteristic reflections 

in the X-ray fibre diffraction pattern of these fibrils, which is 

increasingly accepted as a diagnostic indicator of the presence of 

amyloid fibrils (Makin and Serpell, 2004). As well as their well 

recognised importance in the diseases of protein misfolding 

(Dobson, 2001), the use of amyloid fibrils as nanomaterials is 

attracting increasing attention (Gras, 2007; Waterhouse and Gerrard, 

2004). Other amyloid-like nanostructure models have been 

proposed, including b-helix and spiral structures (DeMarco et al., 

2006; Makin and Serpell, 2005). 

Since polyglutamine proteins are associated with amyloid 

formation in the body (Chen et al., 2002), our attention was drawn 

to the proteins of wheat as a potential raw material for amyloid 

fibril manufacture. Wheat proteins are known to have a high 

glutamine content (Sugiyama et al., 1985), including the high 

molecular weight glutenin subunits (HMW-GS) associated with 

flour quality which have a (glutamic acid þ glutamine content) in 

the order of 30% (Anjum et al., 2008; Kipp et al., 1996). A BLAST 

search (Altschul et al., 1997) on the wheat (Triticum aestivum)

158 

genome returned matches for polyglutamine sequences up to 

eleven amino acids in length. Additionally, gluten proteins are well 

known to form aggregated structures; indeed it is this ability that 

facilitates their use in the production of a variety of baked goods 

and pasta products (Shewry et al., 2002). They have also been 

explored for biomaterial use (Guilbert et al., 2001). 

The gluten network itself is extremely complex, with the 

molecular weight of these HMW-GS components ranging from 

approx. 67–88 kDa (Shewry et al., 2002). The units themselves 

appear to be comprised of three distinct domains: a central 

repetitive domain of variable length (440–680 residues), and much 

shorter flanking non-repetitive terminal domains (N – 81–104 and 

C – 42 residues) (Tatham, 1995). Structural information to date has 

been based on predictive modeling, microscope imaging, indirect 

methods such as FTIR and CD, and limited SAXS analysis (Lindsay 

and Skerritt, 1999). Indications are that the secondary structure of 

the proteins in the matrix includes a high proportion of b-turn 

structure for the central domain, while the terminal domains 

comprise a-helices (Tatham, 1995 and references therein). The low 

molecular weight group of glutenin proteins (LMW-GS) is less well 

characterised. It is known they have a glutamine rich N-terminal 

domain (again predicted to form b-reverse turns), and are rich in 

cysteine residues located primarily in the C-terminal end, but also 

in the central domain (predicted to be a-helical). This group of 

proteins predominantly exists in monomeric form, and contains 

numerous internal disulfide bonds (Lew et al., 1992). Subgroups 

within the LMW-GS are also known to possess similar secondary 

structure to the gliadin protein group, which has been documented 

to aggregate under appropriate conditions (Kasarda et al., 1967). 

The gliadin fibril aggregates were observed to be 80 Å thick, up to 

several thousand Ångstroms long and dissociated to globular 

protein subunits at very low ionic strength and low pH. 

Despite considerable research into their aggregation properties, 

the tendency for gluten proteins to form amyloid-like structures 

has not been previously explored. We report herein an exploration 

of whether or not gluten proteins can form amyloid structures and 

the characterisation of a novel nanostructure derived from wheat 

glutenins, characterised by spectroscopic binding assays, transmission 

electron microscopy and X-ray fibre diffraction. 



Unless otherwise stated, all chemicals and reagents were 

purchased from Sigma Chemical Company Ltd. (St Louis, U.S.A.), 

Aldrich Chemicals (Milwaukee, U.S.A.) or BDH Laboratory Supplies 

(Poole. U.K.), and were of analytical grade. The wheat flour derived 

protein was extracted from a Domino cultivar, sourced by Crop and 

Food Research Ltd. and milled in-house. 

2.2. Extraction protocol 

Wheat proteins were crudely fractionated using an extraction 

method based on the methods of Gerrard et al. (2001), which were 

derived from the protocols of Osborne (1907). The albumins and 

globulins were extracted in dilute saline (2% w/v). The mixture was 

mixed by pulse vortex every 5 min for 30 min to allow extraction of 

the appropriate protein group, then centrifuged at 10 000 g for 

5 min. The resulting pellet was resuspended in a concentrated 

ethanol solution (70% v/v) and mixed by pulse vortex every 5 min 

for 30 min to resuspend the pellet and to separate the soluble and 

insoluble components. After centrifugation (10 000 g, 5 min) the 

gliadin extract was decanted. The pellet was resuspended in a SDSphosphate 

buffer (0.05% w/v SDS, 0.05 M phosphate, pH 6.9) with 

mixing by pulse vortex (every 5 min for 30 min), to extract the 

S.H. Mackintosh et al. / Journal of Cereal Science 49 (2009) 157–162 

LMW-GS and HMW-GS. After centrifugation (10 000 g, 5 min) the 

pellet was again resuspended in SDS-phosphate buffer (0.05% w/v 

SDS, 0.05 M phosphate, pH 6.9) with mixing by pulse vortex (every 

5 min for 30 min). This final solution of glutenins was sonicated at 

30 W for 15 s using a Branston model 250 sonic disruptor, to 

solubilise any protein components still held within the pellet. 

2.3. Protein concentration 

Protein concentrations were determined using a modified 

version of the Bradford’s (1976) method. A standard curve was 

determined using bovine serum albumin. 

2.4. PAGE 

Polyacrylamide gel electrophoresis was routinely run using an 

Invitrogen XCell SureLock Mini-Cell powered by a Bio-Rad 300 

power pack, and precast 4–12% BisTris gels from Invitrogen. The gel 

system was run in a 1 3-(N-morpholino) propanesulfonic acid 

(MOPS) SDS running buffer, over approx. 90 min at 150 V, and 

stained with Coomassie blue. The marker was wide molecular 

range SigmaMarker (product no. S8445 from Sigma). 

2.5. Screening protocols 

In preliminary screens, all four fractions of wheat protein were 

incubated in an array of solution conditions and periodically tested 

using thioflavin T (ThT, see below) for the presence of amyloid 

fibrils. In particular, additions of the following compounds to the 

extraction buffers were tested: acid (HCl or H2SO4); sodium chloride 

(0–0.2 M); denaturing compounds (2–6 M urea, 0–30% (v/v) 

TFE, 1 M mercaptoethanol or 0.5% (w/v) SDS); and insulin fibrils (2% 

v/v 5.8 mg/ml solution of preformed insulin fibrils/wheat protein 

solution). The extracted glutenin protein fractions proved to be the 

most interesting and were explored further. No further experiments 

were carried out on the albumin and globulin or gliadin 

fractions. Glutenin fractions were then extracted, and the proteins 

lyophilised and resuspended at a concentration of 10 mg/ml (w/v) 

in a variety of modified buffers. A screen of conditions was trialled 

in order to optimise conditions for fibril formation. 

The solution conditions of the glutenin extracts were varied in 

order to optimise the conditions for fibril formation, as judged by 

the ThT assay. A range of conditions and variables were investigated 

both individually and in combination and they are summarized in 

Table 1. The solutions were then incubated at temperatures of 

either 25 Cor37 C and monitored for periods of up to 105 days 

and compared to control samples kept frozen at 20 C. As 

required, trypsin solution was prepared with 1 mM HCl and mercaptoethanol 

(10 ml/ml) and the glutenin extracts were treated at 

a ratio of 1:20 (v/v) prior to incubation. Insulin fibrils were formed 

by incubation of insulin from bovine pancreas (Sigma product no. 

Table 1 

Range of conditions for glutenin incubations. 

Variable Range 

Glutenin concentration 2–10 mg/ml 

Temperature 20 to 37 C 

Time 0–105 days 

pH 2–7 

Urea 0–6 M 

TFE 0–30% 

Sodium chloride 0–0.2 M 

Sonication þ/ 

Trypsin pre-treatment þ/ 

Insulin seeding 2% v/v (5.8 mg/ml solution of preformed 

insulin fibrils/wheat protein solution)

I5500) at pH 2 (HCl in distilled water) at 50 C for 48 h according to 

the method of Nilsson and Dobson (2003). Fibril formation was 

confirmed by TEM and ThT assay. These seed fibrils were added to 

selected glutenin samples prior to incubation. In addition to 

changing the solution conditions, the impact of shaking and/or 

periodic sonication during the incubation was explored. Aliquots 

removed from the incubating reactions were stored at 20 C until 

required for analysis by ThT and TEM (below). 

2.6. ThT assay 

The ThT method used was based on the protocols of LeVine 

(1999). The protein samples were diluted to a concentration of 10– 

20 mg/ml in Tris buffer (50 mM Tris, 100 mM NaCl, pH 7.5). To this 

solution, the ThT dye was added to a final concentration of 5 mM. 

The solution was mixed and left to stand at room temperature for 

3 min to allow binding between the dye and protein to equilibrate, 

before the emission spectra were recorded. The spectra of buffer 

only, buffer with ThT and buffer with protein were used as the 

experimental controls. Fluorescence spectroscopy was performed 

using a Cary Eclipse Varian fluorescence spectrophotometer interfaced 

with Cary Eclipse operating software (V 2.0), using 1 cm path 

length quartz cuvettes and the excitation was set to 450 10 nm 

and emission set to 470–520 nm. The relative fluorescence is 

graphed. This reading is the incubated sample fluorescence with 

the control (non-incubated protein with ThT) fluorescence subtracted. 

The error bars are standard errors for a triplicate 

measurement. 

2.7. Transmission electron microscopy 

Samples were analysed by TEM, based on the protocols 

described by Brenner and Horne (1959), to confirm the presence of 

fibrils. The preparation involved placing small aliquots (5 ml) of 

protein fibril samples on formvar-coated copper grids. After 90– 

120 s, the excess solution was drained with filter paper, the grids 

washed with double distilled water and the samples negatively 

stained with 1% (w/v) uranyl acetate for 45–60 s, before the excess 

solution was again drained with filter paper and the grids left to 

dry. For each sample, four images were recorded and a representative 

image was selected from amongst these. 

2.8. X-ray diffraction 

The glutenin samples were prepared for X-ray fibre diffraction 

by placing a 10 ml droplet of the incubated protein solution between 

two wax-filled glass capillary ends (Serpell et al., 1999). The capillaries 

were manually separated as the sample dried, to promote 

fibril alignment. The resulting fibre was then analysed using an Xray 

beam produced by a Rigaku Cu-K rotating-anode source with 

the data collected using an R-AXIS IV image-plate X-ray detector. 

The data were analysed using Fit2D (V12.043) software. Images 

were recorded in the Biochemistry Department, University of 

Cambridge. 


3.1. Screening for fibril-forming conditions 

All four wheat protein fractions – the albumins and globulins, 

the gliadins, the SDS-soluble glutenins and the SDS-insoluble 

glutenins – were put through an initial screen including a range of 

solution conditions that have been shown to induce fibril 

formation in other systems (Kim et al., 2004). The initial screen 

showed promising results for the SDS-soluble glutenins and the 

SDS-insoluble glutenins only (data not shown) and no further 

S.H. Mackintosh et al. / Journal of Cereal Science 49 (2009) 157–162 159 

work was carried out on the albumin and globulin and gliadin 

fractions. 

The solution conditions of the two glutenin extracts were then 

varied in a systematic, combinatorial fashion, as described in 

Section 2. These included acidic conditions and addition of 

a denaturant. Also explored was the effect of shaking the solution 

during the incubation, sonicating the solution periodicially during 

the incubation, and temperature. At various time points throughout 

the incubation, aliquots were extracted from each of the incubated 

protein mixture samples and analysed by ThT assay. Evidence for 

increased b-sheet, consistent with fibril-like structures, was found 

under some conditions, particularly in acidic conditions in the 

presence of urea, consistent with the requirement for partial 

unfolding of the proteins prior to fibril formation (Chiti and Dobson, 

2006). The most successful conditions involved a pH of 6–7 in the 

presence of either 2 M urea or 30% (v/v) trifluoroethanol. The SDSinsoluble 

glutenins gave more encouraging results under these 

conditions than the SDS-soluble glutenins. The SDS-soluble glutenin 

conditions that gave significant increases in ThT fluorescence 

are shown in Table 2. 

An example of the results is shown in Fig. 1. A significant lag 

period was observed for all treatments in which positive results 

were obtained, characteristic of a period of equilibrium where 

fibril growth is at a minimum and interactions between the glutenin 

subunits are limited and reversible, prior to the growth 

phase (Nilsson and Dobson, 2003). The lag phase observed is 

significantly longer than any lag phase reported for amyloid 

formation in vitro in the literature to date. This increase in ThT 

fluorescence is consistent with a structure with increased b-sheet 

content. 

3.2. Characterising the aggregated structures: SDS-PAGE 

In an attempt to establish which of the many proteins in the 

glutenin fractions were forming the aggregated structure, the 

incubated solutions were analysed by SDS-PAGE. The solution was 

spun down to remove any aggregated protein and the supernatant 

solution analysed as a function of time, for many of the conditions 

tested. The electrophoretic profile of this fraction is well studied 

(Gerrard et al., 2001) but typically, no specific proteins were found 

to be selectively lost from solution; rather, all proteins were lost at 

a similar rate, although larger proteins were removed from the 

supernatant somewhat faster than smaller ones. A typical result is 

found in Fig. 2. These results suggest that a mixture of proteins is 

incorporated into the aggregate, or that different proteins form 

different structures that are incorporated into the assembly, and/or 

that those proteins not incorporated were subject to hydrolysis 

and/or proteolysis during the long timeframe of the incubation. 

Extensive further work would be required to distinguish between 

these possibilities, and due to the long timeframe and low yield of 

the process, these experiments were not continued further. 

Table 2 

Summary of SDS-insoluble glutenin incubation conditions that resulted in significant 

increase ThT fluorescence after 105 days. Frozen controls showed no increase in 

ThT fluorescence. 

Treatment SDS-insoluble 

Temp. ( C) Conc. (mg/ml) Conditions 

glutenin 

25 10 pH 6–7 þThT 

25 10 pH 6–7, 0.5 M H2SO4 þThT 

37 10 pH 6–7 þThT 

37 10 pH 6–7, 0.5 M HCl þThT 

37 10 pH 6–7, 2 M urea þThT 

37 10 pH 6–7, 30% (v/v) TFE þThT

160 

Relative Fluorescence 

12 

10 

8 

6 

4 

2 

0 

0 10 20 30 40 50 60 70 

Time (Days) 

Fig. 1. Formation of amyloid-like structures in the SDS-soluble glutenins of wheat 

incubated at 25 C in the presence of 2 M urea. 

3.3. Characterising the aggregated structures: TEM and X-ray fibre 

diffraction 

ThT is commonly used to monitor amyloid fibril formation, as it 

forms a fluorescent species when bound to protein material with 

a high content of b-structure (LeVine, 1999). A positive ThT result is 

thus strongly suggestive of the presence of b-sheet structure such 

as an amyloid fibril, but other methods are required to corroborate 

this observation. 

Samples which showed a positive ThT test were observed under 

TEM. A variety of morphologies were apparent in the sample, as 

shown in Fig. 3. Fibril structures were typically unbranched and of 

indeterminate length (at least several microns), with widths of 

between 5 and 15 nm, of similar dimensions to amyloid fibrils 

described in the literature (Serpell, 2000). Some displayed the 

Fig. 2. SDS-PAGE gel of glutenins, in 2 M urea, incubated for various times up to 70 

days. 1. Marker (205, 116, 97.4, 84, 66, 55, 45, 36, 29, 24, 20.1, 14.2, and 6.5 kDa); 2. 7 

days; 3. 14 days; 4. 21 days; 5. 28 days; 6. 42 days; 7. 56 days; and 8. 70 days. 


classical twisted rope appearance (Fig. 3A, B). Consistent with the 

low yield of ThT positive material (Fig. 1), the nanofibrillar structures 

were found to be contaminated with amorphously aggregated 

protein. The sticky nature of the amorphous aggregates made it 

impracticable to separate and purify the nanostructures. We estimate 

that the yield of these nanostructures was less than 1%. 

X-ray fibre diffraction is increasingly used as a diagnostic indicator 

for amyloid fibril formation (Makin et al., 2005). The diffraction 

pattern produced when an aligned fibril sample is exposed to 

an X-ray beam can be used to detect the cross-b pattern indicative 

of b-sheet structure, of indeterminate length, with the b-strands 

aligned perpendicular to the fibril axis and the b-sheets running 

parallel to it. For amyloid fibrils, the interstrand separation classically 

appears as a 4.7 Å meridonal reflection (i.e. in a parallel 

direction to the mounted fibril fibre) and the intersheet separation 

as an equatorial reflection at approx. 10–11 Å. However, polyglutamine 

peptide fibre patterns have been reported with a reflection 

at 4.7 Å, but the 10 Å reflection is often faint or not observed (Perutz 

et al., 2002). 

The diffraction pattern obtained from the wheat glutenin 

protein fibrillar structure is shown in Fig. 4. It is clear that the 

sample is anisotropic, but the reflections show distinct differences 

from the classic amyloid form. Equatorially aligned bands at 12.9, 

19.3 and 38.7 Å, and meridionally aligned bands at 4.6, 4.2–4.3, and 

4.02 Å were observed. A number of fainter bands are also visible. 

The faint band at 4.6 Å is suggestive of an aligned structure with an 

intra-strand distance corresponding to that of a b-sheet. The 

absence of a diffraction ring at approximately 10–11 Å is also 

consistent with amyloid fibrils formed from high glutamine 

protein. In classic amyloid fibrils (i.e. those linked with disease) this 

distance corresponds to the interstrand spacing, and forms an 

integral part of definition for cross-b-structure. 

Thus, whilst it is tempting to speculate that the observed 

nanostructures are amyloid-like, this model does not fit all the data. 

The other wide-angle meridional reflections and the inner equatorial 

reflections (12.9, 19.3, and 38.7 Å) have not, to our knowledge, 

been reported in any fibrillar system; the equatorial reflections are 

more consistent with a lateral assembly (i.e. perpendicular to the 

fibril axis), with a spacing of 38.7 Å. In light of this, the 4.6 Å 

reflection may be due to another feature within the fibrils of a nonamyloid 

structure, or perhaps indicates that the sample may 

contain a small amount of amyloid-like fibrils mixed with nonamyloid 

fibrils. The glutenin nanostructure thus appears to contain 

novel features distinct from both the previously reported classical 

amyloid fibril structure and the hypothesised b-sheet helix structure 

(DeMarco et al., 2006). 

3.4. Attempts to improve the rate and yield of fibril formation 

The novel wheat protein structure was formed at a low rate and 

the overall yield was too small to be considered as a viable process 

for commercial manufacturing. All treatments consistently resulted 

in extremely low yields. In a subsequent screen, therefore, two 

further methods were explored in order to increase the rate of 

formation of fibrillar structures, and boost the yield: seeding the 

solution with preformed fibrils, and pre-treatment of the protein 

with trypsin. 

An analysis of the sequences of glutenin proteins using the 

aggregating predicting programme Zygregattor (Pawar et al., 2005) 

revealed that the regions of local sequence most likely to hinder 

aggregation of the glutenin proteins were those containing lysine 

or arginine residues. Since trypsin cuts proteins adjacent to both 

these residues in a selective manner (Wang and Carpenter, 1967), it 

was hoped that pre-digestion with trypsin might improve the rate 

and yield of formation of fibrils from the wheat protein incubations. 

Unfortunately, despite confirmation by SDS-PAGE that the trypsin

Fig. 3. Wheat glutenin protein structure images derived from transmission electron microscopy analysis. Images were taken from SDS-insoluble glutenin proteins incubated at 

25 C (2 mg/ml, pH 7.4) for between 54 and 105 days. Scale bars represent 200 nm. 

had indeed cut the glutenins into the desired smaller fragments, no 

subsequent increase in rate of formation of ThT positive species, or 

yield was observed. 

Since amyloid formation is a seeded process (Ban et al., 2006), 

attempts were made to enhance the rate of production by 

Fig. 4. Diffraction patterns obtained from dried SDS-insoluble glutenin proteins at 

2 mg/ml, pH 7.4 for 105 days. 

S.H. Mackintosh et al. / Journal of Cereal Science 49 (2009) 157–162 161 

sonication and addition of preformed fibril seeds. These processes 

have been reported to accelerate fibril formation (Stathopulos et al., 

2004). The sonication provides strong agitation that can fragment 

the fibrils, thus potentially providing more ‘sticky ends’ for fibril 

elongation. Samples were sonicated periodically during the incubation 

of wheat protein in a range of different buffer conditions. 

Insulin seeds were formed by sonication of insulin fibrils and these 

insulin seeds were added to the glutenin protein extracts prior to 

incubation. Neither method was successful. The lack of sequence 

identity between the insulin seeds and the glutenin structure may 

account for the failure of the latter method. 

3.5. Concluding comments 

Although our original aim of manufacturing large quantities of 

protein nanostructures from a readily available heterogeneous 

protein substrate was not met, a novel aggregated structure has 

been assembled and characterised from the wheat glutenin 

proteins, without prior purification. The structure displays some 

characteristic features of amyloid fibrils, including a detectable 

increase in ThT fluorescence assays and a fibrillar morphology 

when viewed under an electron microscope. While the aggregated 

material does not display the classic cross-b pattern commonly 

observed for amyloid material, bands do exist within the pattern 

at z4.8 Å (corresponding to inter-b-strand distance). The absence 

of an inter-b-sheet distance at z10–11 Å is not unprecedented 

and a number of recent articles describing amyloid-like 

structures, that may represent a distinct but related form of 

amyloid-like nanostructures (Kwan et al., 2006). However, our 

observed X-ray reflection pattern has not been previously reported 

and appears to represent a new nanostructured form of wheat 

proteins. 

Rauscher et al. (2006) have discussed the role of proline and 

glycine in control of protein self-organization into elastomeric or

162 

amyloid-like fibrils. Given that HMW-GS contain repeating 

sequences of proline, glycine and glutamine (Tatham and Shewry, 

2002), it may well be that this novel structure is dependent on the 

interaction of these three amino acids and could shed light on the 

relationship between sequence, structure and elastomeric qualities 

of these intriguing proteins. 


The authors thank Harald Dobberstein and Manfred Ingerfeld 

for assistance with TEM, Grant Pearce and Amol Pawar for assistance 

with the Zygreggator algorithm, and Drs Cait MacPhee, Kate 

McGrath, Sarah Meehan, and Margie Sunde for useful discussions. 

Funding was provided by the New Zealand Foundation for 

Research, Science and Technology via the New Economy Research 

Fund contracts C02X0204 (Advanced Biological Materials) and 

C02X0404 (Amyloid Fibrils). 

References 

Altschul, S.F., Madden, T.L., Schäffer, A.A., Zhang, J., Zhang, Z., Miller, W., Lipman, D.J., 

1997. Gapped BLAST and PSI-BLAST: a new generation of protein database 

search programs. Nucleic Acids Research 25, 3389–3402. 

Anjum, F.M., Khan, M.R., Din, A., Saaed, M., Pasha, I., Arshad, M.U., 2008. Wheat 

gluten: high molecular weight glutenin subunits – structure, genetics, and 

relation to dough elasticity. Journal of Food Science 72, R56–R63. 

Ban, T., Yamaguchi, K., Goto, Y., 2006. Direct observation of amyloid fibril growth, 

propagation, and adaptation. Accounts of Chemical Research 39, 663–670. 

Bradford, M.A., 1976. A rapid and sensitive method for the quantitation of microgram 

quantities of protein utilising the principle of protein–dye binding. 

Analytical Biochemistry 72, 248–254. 

Brenner, S., Horne, R.W., 1959. A negative staining method for high resolution 

electron microscopy of viruses. Biochimica et Biophysica Acta 34, 103–110. 

Chen, S., Berthelier, V., Hamilton, J.B., O’Nuallain, B., Wetzel, R., 2002. Amyloid-like 

features of polyglutamine aggregates and their assembly kinetics. Biochemistry 

41, 7391–7399. 

Chiti, F., Dobson, C.M., 2006. Protein misfolding, functional amyloid, and human 

disease. Annual Review of Biochemistry 75, 333–366. 

DeMarco, M.L., Silveira, J., Caughey, B., Daggett, V., 2006. Structural properties of 

prion protein protofibrils and fibrils: an experimental assessment of atomic 

models. Biochemistry 45, 15573–15582. 

Dobson, C.M., 1999. Protein misfolding, evolution and disease. Trends in 

Biochemical Sciences 24, 329–332. 

Dobson, C.M., 2001. The structural basis of protein folding and its links with human 

disease. Philosophical Transactions of the Royal Society of London. Series B, 

Biological Sciences 356, 133–145. 

Gerrard, J.A., Fayle, S.E., Brown, P.K., Sutton, K.H., Simmons, L., Rasiah, I., 2001. 

Effects of microbial transglutaminase on the wheat proteins of bread and 

croissant dough. Journal of Food Science 66, 782–786. 

Guilbert, S., Gontard, N., Morel, M.H., Chalier, P., Micard, V., Redl, A., 2001. Formation 

and properties of wheat gluten films and coatings. In: Gennadios, A. (Ed.), 

Protein-based Films and Coatings. CRC Press, Boca Raton, pp. 69–122. 

Gras, S., 2007. Amyloid fibrils: from disease to design. New biomaterial applications 

for self-assembling cross-b fibrils. Australian Journal of Chemistry 60, 333–342. 

Kasarda, D.D., Bernardin, J.E., Thomas, R.S., 1967. Reversible aggregation of a-gliadin 

to fibrils. Science 155, 203–205. 

Kim, J.R., Muresan, A., Lee, K.Y., Murphy, R.M., 2004. Urea modulation of b-amyloid 

fibril growth: experimental studies and kinetic models. Protein Science 13, 

2888–2898. 


Kipp, B., Belitz, H.-D., Seilmeier, W., Wieser, H., 1996. Comparative studies of high Mr 

subunits of rye and wheat. I. Isolation and biochemical characterisation and 

effects on gluten extensibility. Journal of Cereal Science 23, 227–234. 

Kwan, A.H., Winefield, R.D., Sunde, M., Matthews, J.M., Haverkamp, R.G., 

Templeton, M.D., Mackay, J.P., 2006. Structural basis for rodlet assembly in 

fungal hydrophobins. Proceedings of the National Academy of Sciences 103, 

3621–3626. 

LeVine III, H., 1999. Quantification of beta-sheet amyloid fibril structures with thioflavin 

T. In: Wetzel, R. (Ed.), Methods in Enzymology 309: Amyloid, Prions, and 

Other Protein Aggregates. Academic Press, US, pp. 274–284. 

Lew, E.J.-L., Kuzmicky, D.D., Kasarda, D.D., 1992. Characterization of low molecular 

weight glutenin subunits by reversed-phase high-performance liquid chromatography, 

sodium dodecyl sulfate-polyacrylamide gel electrophoresis, and Nterminal 

amino acid sequencing. Cereal Chemistry 69, 505–515. 

Lindsay, M.P., Skerritt, J.H., 1999. The glutenin macropolymer of wheat flour doughs 

structure–function perspectives. Trends in Food Science & Technology 10, 

247–253. 

Makin, O.S., Serpell, L.C., 2004. Structural characterisation of islet amyloid polypeptide 

fibrils. Journal of Molecular Biology 335, 1279–1288. 

Makin, O.S., Serpell, L.C., 2005. Structures for amyloid fibrils. FEBS Journal 272, 

5950–5961. 

Makin, O.S., Atkins, E., Sikorski, P., Johansson, J., Serpell, L.C., 2005. Molecular basis 

for amyloid fibril formation and stability. Proceedings of the National Academy 

of Sciences of the United States of America 102, 315–320. 

Nilsson, M.R., Dobson, C.M., 2003. Chemical modification of insulin in amyloid 

fibrils. Protein Science 12, 2637–2641. 

Osborne, T.B., 1907. The proteins of the wheat kernel. In: Judd, Detweiller (Eds.), 

Publications of the Carnegie Institution Washington No. 84, pp. 1–119. Washington, 

DC. 

Pawar, A.P., DuBay, K.F., Zurdo, J., Chiti, F., Vendruscolo, M., Dobson, C.M., 2005. 

Prediction of ‘‘aggregation-prone’’ and ‘‘aggregation-susceptible’’ regions in 

proteins associated with neurodegenerative diseases. Journal of Molecular 

Biology 350, 379–392. 

Perutz, M.F., Finch, J.T., Berriman, J., Lesk, A., 2002. Amyloid fibers are water-filled 

nanotubes. Proceedings of the National Academy of Sciences 99, 5591–5595. 

Rauscher, S., Baud, S., Miao, M., Keeley, F.W., Pomes, R., 2006. Proline and glycine 

control protein self-organization into elastomeric or amyloid-like fibrils. 

Structure 14, 1667–1676. 

Sanford, K., Kumar, M., 2005. New proteins in a materials world. Current Opinion in 


Serpell, L.C., Fraser, P.E., Sunde, M., 1999. In: Wetzel, R. (Ed.), Methods in Enzymology 

309: Amyloid, Prions, and Other Protein Aggregates. Academic Press, 

US, pp. 526–536. 

Serpell, L.C., 2000. Alzheimer’s amyloid fibrils: structure and assembly. Biochimica 

et Biophysica Acta 1502, 16–30. 

Shen, X., Mo, X., Moore, R., Frazier, S.J., Iwamoto, T., Tomich, J.M., Sun, X.S., 2006. 

Adhesion and structural properties of protein nanomaterials containing 

hydrophobic and charged amino acids. Journal of Nanoscience and Nanotechnology 

6, 837–844. 

Shewry, P.R., Halford, N.G., Belton, P.S., Tatham, A.S., 2002. The structure and 

properties of gluten: an elastic protein from wheat grain. Philosophical Transactions 

B 357, 133–142. 

Stathopulos, P.B., Scholz, G.A., Hwang, Y.M., Rumfeldt, J.A., Lepock, J.R., 

Meiering, E.M., 2004. Sonication of proteins causes formation of aggregates that 

resemble amyloid. Protein Science 13, 3017–3027. 

Sugiyama, T., Rafalski, A., Peterson, D., Soll, D., 1985. A wheat HMW glutenin subunit 

gene reveals a highly repeated structure. Nucleic Acids Research 13, 8729–8737. 

Tatham, A.S., 1995. The structures of wheat proteins. In: Schofield, J.D. (Ed.), Wheat 

Structure, Biochemistry and Functionality. The Royal Society of Chemistry, 

Reading, U.K., pp. 53–62. 

Tatham, A.S., Shewry, P.R., 2002. Comparative structures and properties of elastic 

proteins. Philosophical Transactions B 357, 229–234. 

Wang, S.S., Carpenter, F.H., 1967. Kinetics of the tryptic hydrolysis of the oxidized B 

chain of bovine insulin. Biochemistry 6, 215–224. 

Waterhouse, S.H., Gerrard, J.A., 2004. Amyloid fibrils in bionanotechnology. 

Australian Journal of Chemistry 57, 519–523.

Journal of Cereal Science

Create successful ePaper yourself

Delete template?

Save as template?