Development of New Functional Food Traits in Peanuts

Kim-Yen Phan-Thien

Thesis submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy

School of Chemical Engineering The University of New South Wales Sydney, Australia August, 2012

25,*,1$/,7<67$7(0(17

µ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¶

6LJQHG«««««««««««««««««««««««««« «««

ϭϴͬϬϭͬϭϯ 'DWH«««««««««««««««««  &23<5,*+767$7(0(17

µ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

6LJQHG««««««««««««««««««««««««««««« ««

ϭϴͬϬϭͬϭϯ 'DWH«««««««««««««««««

 $87+(17,&,7<67$7(0(17

µ, FHUWLI\ WKDW WKH /LEUDU\ GHSRVLW GLJLWDO FRS\ LV D GLUHFW HTXLYDOHQW RI WKH ILQDO RIILFLDOO\DSSURYHGYHUVLRQRIP\WKHVLV1RHPHQGDWLRQRIFRQWHQWKDVRFFXUUHG DQG LI WKHUH DUH DQ\ PLQRU YDULDWLRQV LQ IRUPDWWLQJ WKH\ DUH WKH UHVXOW RI WKH FRQYHUVLRQWRGLJLWDOIRUPDW¶WDOIRRUPDW¶

6LJQHG««««««««««««««««««««««««««««« ««

'DWH«««««««««««««««««ϭϴͬϬϭͬϭϯ Acknowledgements

I am indebted to my supervisor Dr. N. Alice Lee and my co-supervisor Dr. Graeme C. Wright for their research direction, generous advice, and personal encouragement. They have been excellent mentors, from whom I have learnt a great deal during my PhD study. Teleconferences and trips to Kingaroy to visit Graeme were always stimulating. Alice, being rather more easily accessible, was no stranger to unscheduled knocks on the door and I thank her for her patience and kindness.

I received valuable advice from a number of other researchers and technicians. In particular, Dorothy Yu and Rabeya Akter (Elemental and Solid State Analysis Unit, Mark Wainwright Analytical Centre, UNSW) advised on the elemental analysis. Dr. Mirta Golic (Peanut Company of Australia) and Prof. Brynn Hibbert (School of Chemistry, UNSW) helped me with the NIRS analysis. Dr. Victor Wong and I spoke frequently about HPLC. Discussions with Dr. John Earl (Children’s Hospital at Westmead) clarified issues with the ECD and inspired some fun (if futile) sessions rigging up split-flow LC configurations. Lewis Adler and Dr. Leanne Stephenson advised on the HRMS. Dr. Martin Bucknall (Bioanalytical Mass Spectrometry Facility, Mark Wainwright Analytical Centre, UNSW) was a fount of knowledge on GC-MS, and I look forward to doing some data-mining with his advice. Camillo Taraborrelli has been an exceptionally organised lab manager; the lab wouldn’t run half as well without him.

I thank UNSW for enabling my PhD through provision of a University Postgraduate Award and PRSS travel award to attend Pacifichem 2010. The Grains Research and Development Corporation funded travel to the United States to attend the 2010 APRES meeting and generously provided me a one year scholarship during the extension of my PhD enrolment. The Peanut Company of Australia supplied all my peanuts and provided financial support for the research.

I thank my friends, old and new, lab mates and random unibar mates, for adding laughs, perspective, and a bit of life balance to the study.

Finally, I thank my family, Lai Kuen Wong, Kim-Chi Phan-Thien, and Minh Phan-Thien, for their unconditional love and belief in me.

| Acknowledgements 2

Abstract

Two categories of functional food traits were researched: dietary minerals and antioxidants. The primary objectives were to (1) characterise diverse peanut phenotypes using established methods or developing and validating analytical methods if necessary; (2) estimate genotypic variation in the functional food trait; and (3) investigate the stability of the functional food trait through studies of genotype-by-environment (G × E) interaction.

Essential mineral concentrations in kernels were analysed by inductively coupled plasma- optical emission spectroscopy (ICP-OES) and ICP-mass spectrometry (ICP-MS) with the use of a dynamic reaction cell (DRC) after preparation of samples by microwave-assisted closed acid digestion. Antioxidant capacity was assessed using ABTS•+, DPPH•, Folin-Ciocalteu total phenolics, and ORAC assays adapted to a 96-well microplate format for high-throughput analysis. The phytochemical profile was quantitatively analysed by high performance liquid chromatography (HPLC) with the use of a photodiode array (PDA) detector, after ultrasound- and enzyme-assisted extraction and solid phase extraction to purify and concentrate the extracts.

Genotypic variation for essential minerals and antioxidant capacity was estimated by analysis of 32 full-season maturity and 24 ultra-early maturity genotypes from the Australian Peanut Breeding Program (APBP). The studies established useful levels of variation of more than 10% relative standard deviation (RSD) among the genotypes in concentrations of most of the tested essential minerals, and of more than 20% RSD in antioxidant capacity, although only the ORAC assay distinguished statistically significant differences between genotypes.

Studies of G × E interaction affecting the essential mineral and antioxidant capacity traits revealed that genotype, environment, and G × E interaction all significantly affected trait expression. The results confirmed that there was substantial genetic control of essential mineral concentrations and antioxidant capacity in peanut kernels, but that it will be important to characterise environmental interaction to enable plant/seed selection in the APBP and potentially manipulate the interaction through agronomic or postharvest management.

The essential minerals data were used to develop approximately predictive calibrations for Ca, K, Mg, and P by near-infrared reflectance spectroscopy (NIRS) of sufficient accuracy to be useful as plant/seed selection tools in plant breeding. Techniques that enable high-throughput, non-destructive, time/cost-effective analysis of trait segregation are valuable due to the extremely large number of samples that are generated in breeding programs.

| Abstract 3

Five peanut genotypes with diverse antioxidant capacity phenotypes were quantitatively profiled for p-coumaric acid, salicylic acid, resveratrol, and daidzein. The co-eluting compounds, caffeic/vanillic acid and ferulic/sinapic acid, were quantified on caffeic acid equivalent and ferulic acid equivalent bases, respectively. The HPLC analysis established significant genotypic differences in phytochemical concentrations and also the importance of the bound (e.g., conjugated and matrix-embedded) fraction. Fractions of the HPLC eluate were evaluated by ORAC assay to evaluate relative contributions to antioxidant capacity, and allowed identification of a number of unknown compounds that made important contributions to antioxidant capacity. HPLC analysis of kernels subjected to various roasting treatments (150 °C, 0-70 min and 160 °C, 0-32.5 min) showed that ferulic/sinapic acid concentrations declined with roasting duration, but that most other tested analytes were relatively thermo-stable.

| Abstract 4

Table of contents

Acknowledgements ...... 2

Abstract ...... 3

Table of contents ...... 5

List of abbreviations ...... 8

List of tables ...... 11

List of figures ...... 15

1 Introduction ...... 18

2 Development and validation of analytical methodology ...... 21

2.1 Validation of elemental analysis by ICP-OES and ICP-MS ...... 21

2.1.1 Objectives ...... 21

2.1.2 Materials and methods ...... 21

2.1.3 Results and discussion ...... 24

2.1.4 Conclusion ...... 34

2.2 Discussion of antioxidant assays ...... 35

2.2.1 Objectives ...... 35

2.2.2 Comparison of ABTS•+, DPPH•, FC, and ORAC assays ...... 35

2.2.3 Optimisation of ORAC assay ...... 40

2.2.4 Conclusion ...... 45

2.3 Validation of HPLC analysis ...... 46

2.3.1 Objectives ...... 46

2.3.2 Materials and methods ...... 46

2.3.3 Results and discussion ...... 49

2.3.4 Conclusion ...... 78

3 Dietary minerals ...... 81

3.1 Literature review ...... 81

3.1.1 Importance of dietary minerals ...... 81

| Table of contents 5

3.1.2 Potential benefits of mineral biofortification ...... 88

3.1.3 Research and development of mineral-biofortified seed crops ...... 92

3.1.4 Research and development of mineral biofortified peanuts ...... 102

3.2 Genotypic variation in concentrations of essential minerals in peanut kernels ...... 107

3.2.1 Objectives ...... 107

3.2.2 Materials and methods ...... 107

3.2.3 Results and discussion ...... 109

3.2.4 Conclusion ...... 114

3.3 G × E interaction affecting concentrations of essential minerals in peanut kernels 117

3.3.1 Objectives ...... 117

3.3.2 Materials and methods ...... 117

3.3.3 Results and discussion ...... 119

3.3.4 Conclusion ...... 130

3.4 NIRS estimation of essential mineral concentrations in peanut kernels ...... 132

3.4.1 Objectives ...... 132

3.4.2 Materials and methods ...... 132

3.4.3 Results and discussion ...... 134

3.4.4 Conclusion ...... 143

4 Antioxidant capacity ...... 144

4.1 Literature review ...... 144

4.1.1 Importance of dietary antioxidants ...... 144

4.1.2 Assays of ‘total antioxidant capacity’ ...... 151

4.2 Genotypic variation in antioxidant capacity of peanut kernels ...... 158

4.2.1 Objectives ...... 158

4.2.2 Materials and methods ...... 158

4.2.3 Results and discussion ...... 163

4.2.4 Conclusion ...... 168

4.3 G × E interaction affecting antioxidant capacity of peanut kernels ...... 171

| Table of contents 6

4.3.1 Objectives ...... 171

4.3.2 Materials and methods ...... 171

4.3.3 Results and discussion ...... 172

4.3.4 Conclusion ...... 177

5 Antioxidant phytochemicals ...... 178

5.1 Literature review ...... 178

5.1.1 Phenolic acids in peanuts ...... 178

5.1.2 Flavonoids in peanuts ...... 183

5.1.3 Stilbenes in peanuts ...... 193

5.2 Quantification of antioxidant phytochemicals in peanut kernels...... 197

5.2.1 Objectives ...... 197

5.2.2 Materials and methods ...... 197

5.2.3 Raw kernel results and discussion ...... 200

5.2.4 Roasted kernel results and discussion ...... 241

5.2.5 Conclusion ...... 257

6 Conclusion ...... 259

7 Appendix ...... 264

8 Publications to date ...... 274

8.1 Peer-reviewed articles ...... 274

8.2 Oral presentations ...... 274

8.3 Poster presentations ...... 274

9 References ...... 275

| Table of contents 7

List of abbreviations

AAPH 2,2’-azobis(2-amidino-propane) dihydrochloride ABTS 2,2’-azino-bis-(3-ethylbenzothiazoline-6-sulfonic acid) AI Adequate intake AM Arbuscular mycorrhizae ANOVA Analysis of variance APBP Australian Peanut Breeding Program APCI Atmospheric pressure chemical ionization CAE Caffeic acid equivalent CGIAR Consultative Group for International Agricultural Research CI Confidence interval CRM Certified reference material CUPRAC Cupric ion reducing antioxidant power CV Coefficient of variation DALY Disability-adjusted life years DEEDI Queensland Government Department of Employment, Economic Development and Innovation DPPH 2,2-diphenyl-1-picrylhydrazyl DRC Dynamic reaction cell DT De-trending correction EAR Estimated average requirement ECD Electrochemical detector EI Electron ionisation ESI Electrospray ionisation ET Electron transfer mechanism FAE Ferulic acid equivalent FAO Food and Agriculture Organization of the United Nations FC Folin-Ciocalteu FRAP Ferric ion reducing antioxidant power FSANZ Food Standards Australia and New Zealand G × E Genotype-by-environment GAE Gallic acid equivalent GCGH Grand Challenges in Global Health GC-MS Gas chromatography-mass spectrometry GRDC Grains Research and Development Corporation GSH Reduced glutathione HAT Hydrogen atom transfer mechanism HPLC High performance liquid chromatography HRMS High resolution mass spectrometry ICP-MS Inductively coupled plasma-mass spectrometry ICP-OES Inductively coupled plasma-optical emission spectroscopy | List of abbreviations 8

LOD Limit of detection LOQ Limit of quantitation MSC Multiplicative scatter correction NHMRC National Health and Medical Research Council NIL Near-isogenic line NIRS Near-infrared reflectance spectroscopy NIST National Institute of Standards and Technology, USA NMI National Measurement Institute, Australia NUE Nutrient use efficiency OAE o-coumaric acid equivalent ORAC Oxygen radical absorbance capacity PBR Plant Breeders Rights PC Principal components PCA Peanut Company of Australia PCA Principal components analysis PDA Photodiode array detector PLS Partial least squares PTFE Polytetrafluoroethylene PUFA Polyunsaturated fatty acids QC Quality control QE Quercetin equivalent QTL Quantitative trait loci RDI Recommended dietary intake RF Fluorescence detector RIL Recombinant inbred line RMCD Randomly methylated β-cyclodextrin RMSECV Root mean square error for cross-validation RMSEP Root mean square error for prediction/performance RPD Residual predictive deviation RSA Root system architecture RSD Relative standard deviation SD Standard deviation SEM Standard error of the mean SIM Selective ion monitoring SNV Standard normal variate correction SPE Solid phase extraction SRL Specific root length TE Trolox equivalent TEAC Trolox equivalence antioxidant capacity TOSC Total oxyradical scavenging capacity TRAP Total radical-trapping antioxidant parameter | List of abbreviations 9

UAE Ultrasound-assisted extraction UFLC Ultra-fast liquid chromatography UL Upper limit UNSW The University of New South Wales USDA United States Department of Agriculture WHO World Health Organization YLD Years lived with disability YLL Years of life lost

| List of abbreviations 10

List of tables

Table 2.1 Operating conditions tested during ICP-OES method development ...... 22 Table 2.2 Operating conditions tested during ICP-(DRC)-MS method development ...... 22 Table 2.3 Final operating conditions for ICP-OES and ICP-(DRC)-MS analyses ...... 22 Table 2.4 Linearity of ICP-(DRC)-MS and ICP-OES calibrations ...... 30 Table 2.5 LOD, LOQ, and typical sample concentrations for ICP-(DRC)-MS and ICP-OES analyses ...... 31 Table 2.6 Accuracy of ICP-(DRC)-MS and ICP-OES analyses ...... 32 Table 2.7 Precision of ICP-(DRC)-MS and ICP-OES analyses ...... 33 Table 2.8 Variation in calibration linearity of ABTS•+, DPPH•, FC, and ORAC assays ...... 36 Table 2.9 Bivariate correlations among ABTS•+, DPPH•, FC, and ORAC measurements...... 39 Table 2.10 Key historical developments in the ORAC protocol ...... 41 Table 2.11 Review of optimisation/comparative studies of phytochemical extraction from peanuts ...... 51 Table 2.12 Review of phytochemical extraction methods applied to peanuts ...... 53 Table 2.13 Effect of temperature and duration on UAE efficiency ...... 55 Table 2.14 Effect of temperature on UAE efficiency ...... 55 Table 2.15 Correlations and differences between native and enzyme-treated peanut extracts...... 57 Table 2.16 Sorbents investigated during SPE method development ...... 58 Table 2.17 Effect of pH and organic solvent content of the loading solution on SPE recovery ...... 59 Table 2.18 Effect of storage conditions on solubility of phytochemical standards ...... 59 Table 2.19 Correlations and differences between 5% and 20% SPE peanut extracts ...... 60 Table 2.20 Review of methods for HPLC analysis of peanut extracts ...... 62 Table 2.21 Columns investigated during HPLC method development...... 67 Table 2.22 Experimentally-derived retention times, UV spectra, and fluorescence of tested phytochemical standards ...... 71 Table 2.23 Linearity of HPLC-PDA and HPLC-RF calibrations ...... 72 Table 2.24 LOD, LOQ, and typical sample concentrations for HPLC analysis ...... 75 Table 2.25 Accuracy (% spike recovery) of HPLC analysis ...... 76 Table 2.26 Precision of HPLC analysis ...... 77 Table 2.27 Summary of phytochemical extraction and HPLC method development ...... 79 Table 3.1 Major functions of essential minerals in the body ...... 82 Table 3.2 Functional food characteristics of dietary minerals ...... 83 Table 3.3 Definitions of nutrient reference values used in Australia ...... 84 Table 3.4 Nutrient reference values for adult men and women (19-30 years)...... 85 Table 3.5 Contribution of nutritional deficiencies to global morbidity and mortality ...... 86 Table 3.6 Comparison of intervention strategies for alleviation of micronutrient malnutrition ...... 90 Table 3.7 Examples of biofortified crop varieties ...... 94 Table 3.8 Review of root characteristics affecting NUE in plants ...... 96 Table 3.9 Revew of features of nutrient dynamics affecting NUE in plants...... 99 Table 3.10 Review of microbial factors affecting NUE in plants ...... 101 Table 3.11 Review of genotypic variation in the essential mineral concentrations of several legume and grain seed crops ...... 103 | List of tables 11

Table 3.12 Peanut genotypes analysed in genotype study of essential mineral concentrations ...... 108 Table 3.13 Variation in essential mineral concentrations (μg g-1 dw) of 32 full-season and 24 ultra-early peanut genotypes ...... 113 Table 3.14 Published variation in essential mineral concentrations of peanut genotypes ...... 113 Table 3.15 Significance of genotype on essential mineral concentrations ...... 114 Table 3.16 Full-season genotypes ranked by normalised essential mineral concentrations ...... 115 Table 3.17 Ultra-early genotypes ranked by normalised essential mineral concentrations ...... 116 Table 3.18 Peanut genotypes analysed in G × E study of essential mineral concentrations ...... 118 Table 3.19 Significance of genotype, environment, and G × E interaction on essential mineral concentrations ...... 119 Table 3.20 Essential mineral concentrations of peanut genotypes tested in the G × E study ...... 120 Table 3.21 Essential mineral concentrations in peanut kernels from each environment tested in the G × E study... 122 Table 3.22 Distribution of essential mineral concentrations (μg g-1 dw) in calibration and validation data sets ...... 134 Table 3.23 Statistics for the best Ca, K, Mg, and P NIRS calibration models ...... 134 Table 4.1 Review of human diseases associated with oxidative damage ...... 147 Table 4.2 Review of antioxidant treatments/therapies for diseases associated with oxidative stress ...... 150 Table 4.3 Review of in vitro chemical antioxidant assays ...... 152 Table 4.4 Review of genotypic variation in antioxidant capacity of several crop species ...... 155 Table 4.5 Peanut genotypes analysed in genotype study of antioxidant capacity ...... 160 Table 4.6 ORAC plate reader operating parameters ...... 162 Table 4.7 Variation in antioxidant capacity (μmol g-1 dw) of 32 full-season and 24 ultra-early maturity peanut genotypes according to different assays ...... 163 Table 4.8 Significance of genotype on antioxidant capacity in peanut kernels ...... 164 Table 4.9 ORAC antioxidant capacity (TE, μmol g-1 dw) of 32 full-season peanut genotypes ...... 169 Table 4.10 ORAC antioxidant capacity (TE, μmol g-1 dw) of 24 ultra-early peanut genotypes ...... 170 Table 4.11 Peanut genotypes analysed in G × E study of antioxidant capacity ...... 171 Table 4.12 Significance of genotype, environment, and G × E interaction on H-ORAC and FC antioxidant capacities ...... 173 Table 4.13 H-ORAC and FC antioxidant capacities of peanut genotypes tested in the G × E study ...... 174 Table 4.14 H-ORAC and FC antioxidant capacities of peanut kernels from each environment tested in the G × E study ...... 175 Table 5.1 Review of phenolic acids in peanut kernels ...... 180 Table 5.2 Review of flavonol concentrations in peanut kernels ...... 185 Table 5.3 Review of isoflavone concentrations in peanut kernels...... 186 Table 5.4 Review of lignan concentrations in peanut kernels ...... 190 Table 5.5 Review of stilbene concentrations in peanut kernels ...... 196 Table 5.6 Peanut genotypes analysed by HPLC ...... 198 Table 5.7 Correlations and differences in PDA and RF quantitation of analytes ...... 205 Table 5.8 HRMS identification of phytochemicals in peanut fractions ...... 207 Table 5.9 Retention times and spectral features of antioxidant standards analysed by GC-MS ...... 208 Table 5.10 Significance of genotype on phytochemical concentrations in peanut extracts ...... 213 Table 5.11 Caffeic/vanillic acid concentrations (μg g-1 dw CAE) in 5% SPE raw peanut extracts ...... 214 Table 5.12 p-Coumaric acid concentrations (μg g-1 dw) in 5% SPE raw peanut extracts ...... 215

| List of tables 12

Table 5.13 p-Coumaric acid concentrations (μg g-1 dw) in 20% SPE raw peanut extracts ...... 215 Table 5.14 Ferulic/sinapic acid concentrations (μg g-1 dw FAE) in 5% SPE raw peanut extracts ...... 216 Table 5.15 Ferulic/sinapic acid concentrations (μg g-1 dw FAE) in 20% SPE raw peanut extracts ...... 216 Table 5.16 OCU concentrations (μg g-1 dw OAE) in 5% SPE raw peanut extracts ...... 217 Table 5.17 OCU concentrations (μg g-1 dw OAE) in 20% SPE raw peanut extracts ...... 218 Table 5.18 Salicylic acid concentrations (μg g-1 dw) in 5% SPE raw peanut extracts ...... 219 Table 5.19 Salicylic acid concentrations (μg g-1 dw) in 20% SPE raw peanut extracts ...... 219 Table 5.20 Resveratrol concentrations (μg g-1 dw) in 20% SPE raw peanut extracts ...... 220 Table 5.21 Daidzein concentrations (μg g-1 dw) in 20% SPE raw peanut extracts ...... 221 Table 5.22 QCU concentrations (μg g-1 dw QE) in 20% SPE raw peanut extracts ...... 222 Table 5.23 Comparison of ORAC antioxidant capacities (TE, μmol g-1 dw) in crude, native, and enzyme-treated extracts of raw peanut kernels ...... 226 Table 5.24 Extract fraction contributions to total antioxidant capacity ...... 228 Table 5.25 ORAC antioxidant capacities of phytochemicals per g of peanut and per g of analyte ...... 230 Table 5.26 Published comparisons of the antioxidant capacity of phytochemical standards ...... 233 Table 5.27 Bivariate correlations between phytochemical concentrations and ORAC antioxidant capacity in raw peanut kernels ...... 234 Table 5.28 Bivariate (Pearson) correlations among phytochemical concentrations and ORAC antioxidant capacity in 20% SPE native extracts of raw peanut kernels ...... 235 Table 5.29 Bivariate (Pearson) correlations among phytochemical concentrations and ORAC antioxidant capacity in 20% SPE enzyme-treated extracts of raw peanut kernels ...... 236 Table 5.30 X-variable loadings, explained total variance, and X-variable variance in PCA of total phytochemical concentrations in native extracts of raw peanut kernels ...... 238 Table 5.31 X-variable loadings, explained total variance, and X-variable variance in PCA of total phytochemical concentrations in enzyme-treated extracts of raw peanut kernels ...... 238 Table 5.32 Literature regarding the effects of roasting on antioxidant capacity and concentrations in peanut kernels ...... 242 Table 5.33 Effect of roast duration on ORAC antioxidant capacity ...... 244 Table 5.34 ORAC antioxidant capacity (TE, μmol g-1 dw) of 5% SPE extracts of peanut kernels roasted at 150 °C .... 245 Table 5.35 ORAC antioxidant capacity (TE, μmol g-1 dw) of 20% SPE extracts of peanut kernels roasted at 150 °C .. 245 Table 5.36 ORAC antioxidant capacity (TE, μmol g-1 dw) of 5% SPE extracts of peanut kernels roasted at 160 °C .... 246 Table 5.37 ORAC antioxidant capacity (TE, μmol g-1 dw) of 20% SPE extracts of peanut kernels roasted at 160 °C .. 246 Table 5.38 Significance of roasting duration on phytochemical concentrations in peanut extracts ...... 247 Table 5.39 Caffeic/vanillic acid concentrations (μg g-1 dw CAE) in 5% SPE extracts of roasted peanut kernels ...... 248 Table 5.40 Ferulic/sinapic acid concentrations (μg g-1 dw FAE) in 5% SPE extracts of roasted peanut kernels ...... 249 Table 5.41 Ferulic/sinapic acid concentrations (μg g-1 dw FAE) in 20% SPE extracts of roasted peanut kernels ...... 249 Table 5.42 Resveratrol concentrations (μg g-1 dw) in 20% SPE extracts of roasted peanut kernels ...... 251 Table 5.43 Daidzein concentrations (μg g-1 dw) in 20% SPE extracts of roasted peanut kernels ...... 251 Table 5.44 QCU concentrations (μg g-1 dw QE) in 20% SPE extracts of roasted peanut kernels ...... 252 Table 5.45 OCU concentrations (μg g-1 dw OAE) in 5% SPE extracts of roasted peanut kernels ...... 253 Table 5.46 Salicylic acid concentrations (μg g-1 dw) in 5% SPE extracts of roasted peanut kernels ...... 253 Table 5.47 Salicylic acid concentrations (μg g-1 dw) in 20% SPE extracts of roasted peanut kernels ...... 254 Table 5.48 p-Coumaric acid concentrations (μg g-1 dw) in 5% SPE extracts of roasted peanut kernels ...... 255

| List of tables 13

Table 5.49 p-Coumaric acid concentrations (μg g-1 dw) in 20% SPE extracts of roasted peanut kernels ...... 255 Table 7.1 Chemical structures of phenolic acids reported in peanut kernels ...... 264 Table 7.2 Chemical structures of flavonoids reported in peanut kernels ...... 267 Table 7.3 Chemical structures of isoflavonoids reported in peanut kernels ...... 269 Table 7.4 Chemical structures of stilbenes reported in peanut kernels ...... 270

| List of tables 14

List of figures

Figure 2.1 Variation in slope coefficients of antioxidant assays...... 37 Figure 2.2 Variation in ABTS•+ measurements of standard solutions...... 38 Figure 2.3 Variation in DPPH• measurements of standard solutions...... 38 Figure 2.4 Variation in FC measurements of standard solutions...... 38 Figure 2.5 Variation in slope coefficients of ORAC analyses performed in 2009-2012...... 42 Figure 2.6 Spatial variation in ORAC fluorescence decay (AUC) measurements within rows (A-H) and between rows (within columns 1-12) of a 96-well microplate...... 43 Figure 2.7 ORAC plate layout designed to minimise well-to-well variability yet remain convenient for manual pipetting...... 43 Figure 2.8 Variation in ORAC measurements of standards performed in 2009-2012...... 44 Figure 2.9 Mean ORAC measurements of standards performed in 2009-2012...... 44 Figure 2.10 Thermal stability (change in HPLC peak area) after 30 min incubation of control and HCl-treated (a) p-coumaric acid, t-resveratrol, and polydatin; and (b) quercetin, genistein, and biochanin A standards...... 56 Figure 2.11 Change in HPLC chromatogram of crude peanut extract following acid hydrolysis (conc. HCl, 80 °C, 30 min)...... 56

Figure 2.12 Exploratory HPLC separations (305 nm) of representative phytochemicals on (a) Gemini C18 (1.5 mL -1 -1 -1 min ), (b) Synergi-polar (2 mL min ), and (c) Onyx C18 (2.5 mL min ) columns in equivalent elution volumes of 15- 95% phase B...... 68 -1 Figure 2.13 Exploratory HPLC separations (305 nm) of spiked peanut extract on (a) Gemini C18 (1.5 mL min ) and (b) Synergi-polar (2 mL min-1) columns in equivalent 15-45% aqueous acetonitrile gradients...... 68 Figure 2.14 Examples of tested HPLC system configurations using an adjustable split-valve or T-junction to regulate the mobile phase (a) immediately post-column; and (b) after passing through the PDA and/or RF detectors...... 70 Figure 2.15 Examples of tested HPLC system configurations using an external pump to introduce a make-up phase into the ECD split-flow (a) immediately post-column; and (b) after passing through the PDA and/or RF detectors. .. 70 Figure 2.16 Final HPLC-PDA-RF-(FRC/ECD) system configuration...... 70 Figure 2.17 Variation in slope coefficients of (a) caffeic acid; (b) ferulic acid; (c) salicylic acid; (d) p-coumaric acid; (e) o-coumaric acid; (f) resveratrol; (g) daidzein; (h) quercetin; and (i) kaempferol HPLC calibrations...... 73 Figure 3.1 Variation in essential mineral concentrations of (a) 32 full-season and (b) 24 ultra-early peanut genotypes...... 111 Figure 3.2 Variation in normalised essential mineral concentrations of 32 full-season and 24 ultra-early peanut genotypes...... 112 Figure 3.3 Genotype scores on PC-1 vs. PC-2 in PCA of auto-scaled essential mineral concentrations...... 123 Figure 3.4 Environment scores on PC-1 vs. PC-2 in PCA of auto-scaled essential mineral concentrations...... 124 Figure 3.5 Environment scores on PC-3 vs. PC-2 in PCA of auto-scaled essential mineral concentrations...... 124 Figure 3.6 Genotype scores on PC-1 vs. PC-2 in PCA of environment-scaled essential mineral concentrations...... 125 Figure 3.7 Genotype scores on PC-1 vs. PC-3 in PCA of environment-scaled essential mineral concentrations...... 126 Figure 3.8 Raw (log 1/R) NIR spectra of peanut kernels...... 135 Figure 3.9 Second derivative (d2A) NIR spectra of peanut kernels...... 135 Figure 3.10 Regression coefficients in NIRS models of (a), Ca (1130-2372 nm, d2A, SNV-DT) and (b) K (1246-1738 nm, d2A, SNV-DT) concentrations in peanut kernels...... 136 Figure 3.11 Regression coefficients in NIRS models of (a) Mg (1246-1738 nm, d2A, MSC) and (b) P (1400-1550 and 1850-2200 nm, d2A, SNV-DT) concentrations in peanut kernels...... 137 Figure 3.12 Accuracy of NIRS models of (a) Ca (1130-2372 nm, d2A, SNV-DT) and (b) K (1246-1738 nm, d2A, SNV-DT) concentrations in peanut kernels...... 138

| List of figures 15

Figure 3.13 Accuracy of NIRS models of (a) Mg (1246-1738 nm, d2A, MSC) and (b) P (1400-1550 and 1850-2200 nm, d2A, SNV-DT) concentrations in peanut kernels...... 139 Figure 3.14 Scores plot (PC1 vs. PC2) in NIRS models of (a) Ca (1130-2372 nm, d2A, SNV-DT) and (b) K (1246-1738 nm, d2A, SNV-DT) concentrations in peanut kernels...... 141 Figure 3.15 Scores plot (PC1 vs. PC2) in NIRS models of (a) Mg (1246-1738 nm, d2A, MSC) and (b) P (1400-1550 and 1850-2200 nm, d2A, SNV-DT) in peanut kernels...... 142 Figure 4.1 Kinetic fluorescence decay of Trolox standards in ORAC assay...... 162 Figure 4.2 Variation in antioxidant capacity of (a) 32 full-season and (b) 24 ultra-early maturity peanut genotypes...... 165 Figure 4.3 H-ORAC antioxidant capacity of 32 full-season maturity peanut genotypes...... 168 Figure 4.4 H-ORAC antioxidant capacity of 24 ultra-early maturity peanut genotypes...... 168 Figure 4.5 (a) H-ORAC and (b) FC antioxidant capacities of genotypes tested in the G × E study...... 176 Figure 4.6 (a) H-ORAC and (b) FC antioxidant capacities of peanut kernels from environments tested in the G × E study ...... 177 Figure 5.1 HPLC-PDA chromatograms for a mixed standard, native peanut extract, and enzyme-treated peanut extract...... 201 Figure 5.2 UV spectra at retention times where (a) caffeic, vanillic (b) ferulic, and sinapic acids were expected to co- elute in HPLC-PDA analysis of peanut extracts...... 202 Figure 5.3 3-D contour plots at retention times where (a) caffeic, vanillic, (b) ferulic, and sinapic acids were expected to co-elute in HPLC-PDA analysis of peanut extracts...... 202 Figure 5.4 UV spectra at retention times corresponding to (a) p-hydroxybenzoic acid, (b) syringic acid, (c) o-coumaric acid, (d) salicylic acid, (e) daidzein, and (f) quercetin in HPLC-PDA analysis of peanut extracts...... 203 Figure 5.5 UV spectra of peaks at retention times corresponding to (a) p-coumaric acid and (b) resveratrol in HPLC- PDA analysis of peanut extracts...... 204 Figure 5.6 HPLC-PDA chromatogram of peanut extract annotated with compounds identified by HRMS analysis. .. 206 Figure 5.7 Caffeic acid (20.95 min) GC-MS library match...... 209 Figure 5.8 Vanillic acid (15.87 min) GC-MS library match...... 209 Figure 5.9 p-Coumaric acid (18.39 min) GC-MS library match...... 210 Figure 5.10 Ferulic acid (20.45 min) GC-MS library match...... 210 Figure 5.11 Sinapic acid (22.38 min) GC-MS library match...... 211 Figure 5.12 Salicylic acid (9.23 min) GC-MS library match...... 211 Figure 5.13 Protocatechuic acid (16.60 min) GC-MS library match...... 212 Figure 5.14 Caffeic/vanillic acid concentrations (μg g-1 dw CAE) in raw peanut kernels...... 214 Figure 5.15 p-Coumaric acid concentrations (μg g-1 dw) in raw peanut kernels extracted by (a) 5% and (b) 20% SPE method...... 215 Figure 5.16 Ferulic/sinapic acid concentrations (μg g-1 dw FAE) in raw peanut kernels extracted by (a) 5% and (b) 20% SPE method...... 217 Figure 5.17 OCU concentrations (μg g-1 dw OAE) in raw peanut kernels extracted by (a) 5% and (b) 20% SPE method...... 218 Figure 5.18 Salicylic acid concentrations (μg g-1 dw) in raw peanut kernels extracted by (a) 5% and (b) 20% SPE method...... 219 Figure 5.19 Resveratrol concentrations (μg g-1 dw) in raw peanut kernels...... 220 Figure 5.20 Daidzein concentrations (μg g-1 dw) in raw peanut kernels...... 221 Figure 5.21 QCU concentrations (μg g-1 dw QE) in raw peanut kernels...... 222 Figure 5.22 Comparison of ORAC antioxidant capacities in crude, native, and enzyme-treated extracts of raw peanut kernels...... 226

| List of figures 16

Figure 5.23 Mean percentage contribution of time-collected fractions (2-22 min at 5 min intervals) to ORAC antioxidant capacity in native extracts of D147-p8-6F raw peanut kernels...... 229 Figure 5.24 Mean percentage contribution of time-collected fractions (2-22 min at 5 min intervals) to ORAC antioxidant capacity in enzyme-treated extracts of D147-p8-6F raw peanut kernels...... 229 Figure 5.25 Mean ORAC antioxidant capacities of eluted peaks collected during HPLC analysis of the enzyme-treated extract of D147-p8-6F raw peanut kernels...... 231 Figure 5.26 UV spectra of unidentified peaks at (a) 7.0 min, (b) 8.0 min, (c) 13.4 min, and (d) 19.8 min that were important contributors to ORAC antioxidant capacity in enzyme-treated extracts of D147-P8-6F raw peanut kernels...... 233 Figure 5.27 (a and b) X-variable loadings plots; (c and d) scores plots marked by antioxidant capacity; and (e and f) scores plots marked by genotype in PCA of native extracts of raw peanut kernels...... 239 Figure 5.28 (a and b) X-variable loadings plots; (c and d) scores plots marked by antioxidant capacity; and (e and f) scores plots marked by genotype in PCA of enzyme-treated extracts of raw peanut kernels...... 240 Figure 5.29 ORAC antioxidant capacity of (a) 5% and (b) 20% SPE extracts of peanut kernels roasted at 150 °C...... 245 Figure 5.30 ORAC antioxidant capacity of (a) 5% and (b) 20% SPE extracts of peanut kernels roasted at 160 °C...... 246 Figure 5.31 Caffeic/vanillic acid concentrations (μg g-1 dw CAE) in (a) 150 °C and (b) 160 °C roasted peanut kernels...... 248 Figure 5.32 Ferulic/sinapic acid concentrations (μg g-1 dw FAE) in 5% SPE extracts of (a) 150 °C and (b) 160 °C; and 20% SPE extracts of (c) 150 °C and (d) 160 °C roasted peanut kernels...... 250 Figure 5.33 Resveratrol concentrations (μg g-1 dw) in (a) 150 °c and (b) 160 °C roasted peanut kernels...... 251 Figure 5.34 Daidzein concentrations (μg g-1 dw) in (a) 150 °c and (b) 160 °C roasted peanut kernels...... 252 Figure 5.35 QCU concentrations (μg g-1 dw QE) in (a) 150 °c and (b) 160 °C roasted peanut kernels...... 252 Figure 5.36 OCU concentrations (μg g-1 dw OAE) in (a) 150 °c and (b) 160 °C roasted peanut kernels...... 253 Figure 5.37 Salicylic acid concentrations (μg g-1 dw) in 5% SPE extracts of (a) 150 °c and (b) 160 °C; and 20% SPE extracts of (c) 150 °C and (d) 160 °C roasted peanut kernels...... 254 Figure 5.38 p-Coumaric acid concentrations (μg g-1 dw) in 5% SPE extracts of (a) 150 °c and (b) 160 °C; and 20% SPE extracts of (c) 150 °C and (d) 160 °C roasted peanut kernels...... 256

| List of figures 17

1 Introduction

The understanding that food can provide benefits to human health that go beyond essential nutrition is not a new one. But it is only in relatively recent years that health-promotion has been conceived as a quality attribute of agricultural and food products that can be managed and marketed. The demand for ‘functional foods’ that contribute to optimisation of health or protection against chronic diseases is developing at several levels. Australian agri-food industries are challenged by increasingly competitive trade conditions and rely on product differentiation, value-adding, and niche-marketing of ‘quality rather than quantity’ to compete in the global market. There is a wide recognition of the link between diet and health and a growing consumer demographic interested in preventative self-care to supplement institutional health care, especially in industrialised countries with ageing populations. Meanwhile, food shortages associated with population growth and faltering food production are set to impact the world’s poorer communities, particularly in developing countries where starvation and micronutrient malnutrition already exact a devastating social toll. At international and national government levels, the alleviation of micronutrient deficiencies and malnutrition and optimisation of public health are high profile issues.

Genetic improvement has long been a key strategy by which primary industries have achieved production goals. Selection criteria have traditionally related to productivity; for example, maximising yield potential, reducing losses to pest and disease, and increasing adaptability to environmental conditions to allow expansion into new production zones. Breeding for nutritional quality is relatively recent, and the history of the Australian Peanut Breeding Program (APBP) is illustrative: since the program commenced in 1977, emphasis has progressed from drought adaptation and yield improvement in the 1990s, to a focus on kernel quality since 1995 that saw a major shift in kernel oil composition to a high oleic acid platform, resulting in improved shelf life and health benefits for consumers (Rachaputi and Chauhan, 2010).

Breeding for nutritional quality brings commercial opportunities, but can moreover benefit public health, particularly where malnutrition is a problem. For example, development of plant varieties with enhanced levels of essential vitamins or minerals presents a means to improve dietary micronutrient intake, alongside traditional interventions such as supplementation, fortification, and dietary diversification (White and Broadley, 2005). In seed or grain crops, such ‘bio-fortification’ has the added advantages of boosting productivity in nutrient-deficient

| Introduction 18 soils, and represents technology that is readily adopted by farmers because no change in agronomic practice or new capital investments are required (Welch and Graham, 2005).

Functional food traits are non-essential constituents that promote health by reducing the risks of chronic conditions and optimising immune function. To date, development of functional food products has often relied on novel processing (e.g., probiotics, phytosterol-enhanced margarine), or mere measurement and marketing of existing qualities (e.g., numerous antioxidant ‘super foods’). Yet plant breeding could increase the naturally-occurring levels of functional food traits in primary foods, for more effective delivery of health benefits to consumers. Increasing nutritional ‘bang for the buck’ may be particularly useful in communities afflicted by food shortages or food price crises.

Peanuts are bred conventionally in Australia, which means that new genetic combinations only arise from the natural processes of sexual recombination and segregation. At a fundamental level then, the breeding potential for specific traits depends largely on the genotypic variation among the population and the stability of trait expression.

In this PhD project, we have undertaken some of the initial exploratory research regarding two categories of functional food traits: dietary minerals and antioxidants. The primary objectives were to (1) characterise diverse peanut phenotypes using established methods or developing and validating analytical methods if necessary; (2) estimate genotypic variation in the functional food trait; and (3) investigate the stability of the functional food trait through studies of genotype-by-environment (G × E) interaction.

The thesis is structured so that the bulk of technical detail regarding methodology and chemical analysis are condensed in Chapter 2, while Chapters 3 to 5 present research findings for the three main bodies of work: studies of essential dietary minerals, studies of antioxidant capacity, and studies of the antioxidant phytochemical profile. The literature review is split accordingly, so that each chapter begins with an overview of relevant background material.

Scientific validation is critical to establish the limitations and usefulness of analytical methodology, and to provide assurance of the quality of data that they generate. Chapter 2 reports the development and validation of the principal analytical methods used during the PhD research. These included microwave-assisted closed acid digestion for mineralisation of peanut samples; inductively coupled plasma-optical emission spectroscopy (ICP-OES) and ICP- mass spectrometry (ICP-MS) in standard and dynamic reaction cell (DRC) modes for elemental analysis of dietary minerals; ultrasound-assisted solvent extraction (UAE) and solid phase extraction (SPE) for extraction of peanut antioxidants; and high performance liquid

| Introduction 19 chromatography (HPLC) for quantitative analysis of the peanut phytochemical profile. The chapter also discusses our experiences with in vitro antioxidant assays that have attained popularity as screens of ‘total antioxidant capacity’: the ABTS•+ decolourisation (ABTS•+), DPPH• scavenging (DPPH•), Folin-Ciocalteu total phenols (FC), and oxygen radical absorbance capacity (ORAC) assays.

Chapter 3 presents studies of genotypic variation in essential mineral concentrations among 56 breeding lines from the APBP, and G × E interaction in nine diverse genotypes grown in five distinct environments. The data generated from this study were used to develop approximately predictive calibrations of calcium, magnesium, potassium, and phosphorus by near-infrared reflectance spectroscopy (NIRS), which is the final study described in this chapter. The development of high-throughput, non-destructive, and time/cost-efficient tools for phenotyping, such as NIRS, has great practical relevance in plant breeding programs due to the extremely large numbers of samples that are generated.

Chapter 4 reports studies of antioxidant capacity based on similar experimental design to the essential mineral studies. Genotypic variation in the antioxidant capacity of 56 peanut breeding lines was analysed by ABTS•+, DPPH•, FC, and ORAC assays. The G × E interaction influencing antioxidant capacity was assessed in 18 diverse peanut genotypes grown in five distinct environments in two seasons using ORAC and FC assays.

Chapter 5 details experiments to quantitate the antioxidant phytochemicals in raw kernels of five peanut genotypes (D147-p8-6F, Fisher, Page, Sutherland, and UF32), and in kernels of one genotype (Middleton) that were subjected to various roasting treatments (150 °C, 0-70 min; and 160 °C, 0-32.5 min). The primary quantitative analysis was by HPLC using a photodiode array (PDA) detector. HPLC eluates were fractionated by time of elution and peak detection (i.e., signal thresholds), and further analysed by high resolution mass spectrometry (HRMS) and gas chromatography-mass spectrometry (GC-MS) for additional evidence of compound identification. The fractions were also analysed by ORAC assay to reveal the comparative contributions of analytes to kernel antioxidant capacity and to highlight unidentified antioxidants of potential value for investigation in future research.

Chapter 6 concludes the thesis with a succinct summary of the key findings and future research needs.

| Introduction 20

2 Development and validation of analytical methodology

2.1 Validation of elemental analysis by ICP-OES and ICP-MS

2.1.1 Objectives

The elemental analysis methods were developed for quantification of several essential minerals in peanut kernels. The immediate purpose was to screen a large set of samples from the APBP, in order to investigate the extent of genotypic variation (chapter 3.2) and G × E interaction (chapter 3.3) affecting dietary mineral composition among Australian peanut accessions, and thereby explore the breeding potential for phenotypes with high kernel concentrations of important elements. The development and validation of ICP-OES and ICP-MS methods for elemental analysis of closed acid digestates of peanut kernels are reported below.

Validation of the ICP-OES and ICP-MS methodology was published by Phan-Thien et al. (2012).

2.1.2 Materials and methods

2.1.2.1 Instruments Samples were digested in polytetrafluoroethylene (PTFE) vessels using Ethos Plus Labstation (10-vessel capacity) or Ethos 1 Advanced Microwave (24-vessel capacity) systems (Milestone Inc., Shelton, CT, USA). Elemental analyses were performed on Optima 7300 DV ICP-OES and Elan DRC II ICP-MS instruments (PerkinElmer SCIEX, Waltham, MA, USA), which used WinLab32 for ICP software version 4.0.0.0305 and Elan software version 3.4, respectively. The ICP-OES and ICP-MS operating conditions and detection settings that were investigated for analysis of each element are listed in Tables 2.1 and 2.2, while the parameters that were used in the final data analysis are shown in Table 2.3.

2.1.2.2 Reagents All reagents were analytical grade. Nitric acid was purified by sub-boiling distillation in a duoPUR acid purification system (Milestone Inc., USA) prior to use. Water was ultrapure grade (18.2 mΩ cm-1, Millipore Corp., USA). Analytical calibration standards diluted in 2% (v/v) nitric acid were prepared daily from stock standard solutions (Choice Analytical Pty Ltd). Rhodium and yttrium were used as internal standards for ICP-MS and ICP-OES. Wheat flour (1567a) and peanut butter (2387) Certified reference material (CRM) were obtained from the National Institute of Standards and Technology (NIST; Gaithersburg, MD, USA).

2.1.2.3 Samples The peanut samples were obtained from the APBP as described in chapters 3.2.2.3 and 3.3.2.3.

| Development and validation of analytical methodology 21

TABLE 2.1 OPERATING CONDITIONS TESTED DURING ICP-OES METHOD DEVELOPMENT

Parameters Plasma view: Radial Axial Sample flow rate (mL min-1): 1.50 1.50 Plasma gas flow rate (L min-1): 15 15 Auxiliary gas flow rate (L min-1): 0.5 0.5 Nebuliser gas flow rate (L min-1): 0.80 0.65 RF power (W): 1300 1400 View distance: 15.0 15.0 Genotype study analytes Ca (422.673), K (766.490), Mg (279.077), Na Cu (327.393), Fe (238.204), Mn (589.592), P (178.221) (257.610), Zn (206.200) (wavelength, nm): G × E study analytes Ca (422.673), Fe (238.204), K (766.490), Mg (279.077), Mn (257.610), Na (589.592), P (wavelength, nm): (178.221, 213.617)

TABLE 2.2 OPERATING CONDITIONS TESTED DURING ICP-(DRC)-MS METHOD DEVELOPMENT

Parameters Scan mode: Peak hopping Dwell time (ms): 25; 50 for 61Ni, 62Ni, 82Se Sweeps/reading: 20 Integration time (ms): 500; 1000 for 61Ni, 62Ni, 82Se Replicates: 3 Cell gas: 0.5 (DRC) RP q: 0.25 (standard); 0.75 (DRC) Standard mode isotopes: 11B, 59Co, 52Cr, 53Cr, 63Cu, 65Cu, 54Fe, 56Fe, 98Mo, 23Na, 58Ni, 60Ni, 61Ni, 62Ni, 82Se, 66Zn DRC mode isotopes: 59Co, 52Cr, 53Cr ,63Cu, 65Cu, 54Fe, 56Fe, 58Ni, 82Se, 66Zn

TABLE 2.3 FINAL OPERATING CONDITIONS FOR ICP-OES AND ICP-(DRC)-MS ANALYSES

ICP-OES parameters ICP-(DRC)-MS parameters Plasma view: Radial Scan mode: Peak hopping Sample flow rate (mL min-1): 1.50 Dwell time (ms): 25; 50 for 61Ni, 62Ni, 82Se Plasma gas flow rate (L min-1): 15 Sweeps/reading: 20 Auxiliary gas flow rate (L min-1): 0.5 Integration time (ms): 500; 1000 for 61Ni, 62Ni, 82Se Nebuliser gas flow rate (L min-1): 0.80 Replicates: 3 RF power (W): 1300 Cell gas: 0.5 (DRC) View distance: 15.0 RP q: 0.25 (standard); 0.75 (DRC) Analytes (λ, nm): Ca (422.673), Fe (238.204), K Analytes 11B, 59Co, 52Cr, 53Cr, 63Cu, 65Cu, 54Fe, 56Fe, (766.490), Mg (279.077), Mn (standard): 98Mo, 23Na, 58Ni, 60Ni, 61Ni, 62Ni, 82Se, 66Zn (257.610), Na (589.592), P (178.221, 213.617) Analytes 59Co, 52Cr, 53Cr ,63Cu, 65Cu, 54Fe, 56Fe, 58Ni, (DRC): 82Se, 66Zn

| Development and validation of analytical methodology 22

2.1.2.4 Preparation of closed acid digestates Peanut kernels were pulverised using a mortar and pestle after manual removal of the testa and addition of liquid nitrogen to encourage shattering. Only fine particles were used for digestion. Contact with potential metal contaminants was avoided as far as possible during sample preparation. For example, mortars and pestles were cleaned by abrasion with acid- washed sand, rinsed with acetone then ultrapure water, and oven-dried between samples. Each sample was ground in duplicate and stored in plastic sample containers until the time of digestion. A portion of each sample was set aside for gravimetric determination of kernel moisture content.

Samples were mineralised by closed acid microwave digestion. The reaction mixture comprised ground sample or CRM (0.4 g), nitric acid (5.0 mL), hydrogen peroxide (2.0 mL), and ultrapure water (3.0 mL) in each PTFE vessel. Digestion bombs were allowed to pre-digest for at least 15 min before loading into the microwave carousel according to the manufacturer’s instructions. A stepped microwave program of 100 °C and 190 °C for 10 min each was initially used. The program was adjusted after installation of a second microwave digestion system that had larger vessel capacity to: 0-20 min, ramp to 200 °C; 20-40 min, hold at 200 °C. Complete digestion was considered to have occurred when a clear, pale yellow-coloured solution was produced. Digestates were unloaded after internal temperatures had declined to less than 50 °C, diluted to 30 mL with ultrapure water, and stored at room temperature until the time of analysis.

2.1.2.5 Quality controls Each digestion batch contained a reagent blank containing all digestion reagents but no sample to allow background correction. Wheat flour and peanut butter CRM were measured regularly throughout the duration of the experiment. Calibration blank (2% v/v nitric acid) and quality check (QC) standard (i.e., a mid-concentration calibration standard) solutions were measured every 20 samples during ICP-OES analysis and every 15 samples during ICP-MS analysis. Matrix effects on sample uptake and nebulisation were monitored by internal standardisation with yttrium (ICP-OES) or rhodium (ICP-MS), and measurements automatically corrected by the ICP software. Genotype samples were grown as triplicate sub-samples in the field. Each of these sub-samples was digested and analysed in duplicate.

2.1.2.6 Moisture content The moisture content of the peanut samples was estimated gravimetrically (Young et al., 1982). Ground sample (2 g) was accurately weighed in an oven-dried aluminium pan and

| Development and validation of analytical methodology 23 heated in a convection oven (130 °C) for 2 h (i.e., until constant weight according to a preliminary experiment). The dish was again accurately weighed after cooling in a dessicator, and the weight loss considered to represent moisture content.

2.1.2.7 Data analysis Accuracy was evaluated by comparing the experimentally-derived concentrations of analytes in the CRM with the certified concentrations. Instrumental repeatability was estimated as the relative standard deviation (RSD) of consecutive QC measurements; within-run precision as the RSD of QC measurements taken at regular intervals within an analytical run; and between-run precision as the RSD of CRM measurements throughout the entire experiment, which comprised two ICP-(DRC)-MS and five ICP-OES analytical runs. The limit of detection (LOD) and limit of quantitation (LOQ) were calculated as three and ten times the standard deviation (SD) of ten blank measurements, respectively.

2.1.3 Results and discussion

2.1.3.1 Importance of method validation for elemental analysis of peanuts Essential minerals have axiomatic importance to human health and nutrition, and several may be considered to have ‘functional food’ qualities due to their expected preventative or therapeutic effects on chronic diseases. To evaluate the breeding potential of essential mineral traits in peanuts, we studied the genotypic variation (chapter 3.2) and G × E interaction (chapter 3.3) affecting concentrations of 15 essential minerals in representative peanut genotypes from the APBP.

Optical emission spectrometry, also known as atomic emission spectrometry, allows sensitive multi-element analysis because characteristic emission spectra may be generated for a range of elements under the same excitation conditions. ICP, usually argon, has become a popular atomiser because it produces a chemically inert environment, generates very high temperatures that ensure almost complete atomisation, and provides a calibration range of several orders of magnitude. ICP has also become popular as the ion source for MS, which permits multi-element analysis of even greater sensitivity, and achieves detection limits typically 2-3 orders of magnitude lower than ICP-OES (Ammann, 2007).

All spectrometric techniques for elemental analysis are subject to spectral interferences, depending on the sample matrix and elemental composition. Reaction/collision cells can be used to minimise spectral interferences, but at the expense of reduced signal intensity and additional complexity (e.g., potential new interferences). The optimal method for a given sample is often a compromise between minimising the respective interferences affecting

| Development and validation of analytical methodology 24 analytes and achieving acceptable detection limits in view of the sample matrix and expected elemental composition. In addition, there may be pragmatic concerns to minimise analytical run time and expense.

The literature describing validation of ICP methods for multi-element analysis of dietary samples, especially nuts and oilseeds, is relatively scant. Some of the few examples include ICP-OES analysis of essential minerals (including Ca, Cu, K, Mg, Mn, P, and Zn) in several nuts and seeds, albeit with reference to peach leaf CRM (Naozuka et al., 2011); and ICP-MS analysis of potential toxic contaminants (including Co, Cr, Cu, Mg, Mo, Ni, Se, and Zn) in a range of dietary CRM (Nardi et al., 2009). ICP-DRC-MS has rarely been used in the context of routine multi-element analysis, as DRC applications have primarily focused on honing the trace analysis of specific contaminants such as Cd in feed (Guo et al., 2011a), or for speciation of potential contaminants such as Cr (Ambushe et al., 2009) and Se (Cubadda et al., 2010).

The analytical methodology was validated in the course of optimisation to ensure and affirm data quality. The publication of this information will be useful to other researchers, in view of the scarce literature on validated ICP analysis of nuts and oilseeds. It may additionally be relevant for the purposes of product development, quality control, and food regulation: wherever there is a requirement for proven methods of accurate, reliable elemental analysis of food matrices.

2.1.3.2 Development of digestion method A closed acid digestion method was used to release cations, since these compose the majority of essential elements. The purpose of digestion was to decompose the sample matrix and solubilise the analytes of interest for nebulisation in the ICP instrument. The use of microwave radiation and high pressure for closed-vessel digestion has the advantages of accelerating sample decomposition and reducing reagent consumption, while minimising the risks of contamination and loss of volatile analytes (Borkowska-Burnecka, 2000). Throughput is limited by the capacity of the microwave system, but overall efficiency is improved because the run time is short (usually ~1 h or less) and digestion does not require supervision.

Various combinations of strong acids and oxidants have been used for digestion of plant materials, with nitric acid being the most commonly used acid. Previous research has validated the use of closed microwave digestion in nitric acid at 200 °C (600 psi) to prepare several CRM and foodstuffs for ICP-OES analysis of a range of dietary nutrients and contaminants, including the essential minerals of interest to our study (Dolan and Capar, 2002). We opted for a similar nitric acid-based method, with addition of hydrogen peroxide to boost the oxidising strength

| Development and validation of analytical methodology 25 of the bomb. Digestion was judged to be complete if a clear yellowish solution was formed, without visible particulate matter. The first microwave program was initially found to be adequate for peanut samples, but as the experiment proceeded, dissolution was occasionally incomplete, resulting in unacceptably high variability. This was attributed to gradual deterioration in the elasticity of the PTFE vessels due to frequent use, which intermittently affected pressure control. In view of this, the duration of the holding period (190-200 °C) was increased to ensure complete digestion.

2.1.3.3 Development of inductively coupled plasma methods It was desirable to maximise time- and cost-efficiency by simplifying the analytical process as much as practicable, but wide differences in the concentration ranges of analytes meant that the use of a single ICP technique to maximise throughput was not possible without unacceptable loss of accuracy for some elements. Specifically, the Co, Cr, Cu, Mo, and Se content of most samples were below the LOD of ICP-OES analysis; whereas very high concentrations of some macro-elements, especially K, caused signal saturation in ICP-MS and demanded analysis of at least one additional dilution level. Since more than one analytical run remained necessary, we used both ICP-OES and ICP-MS methods for elemental analysis of peanut acid digestates, in view of the validation statistics and the typical sample concentrations compared to the analytical LOD and LOQ.

Elements that were present in relatively high concentrations, including Ca, K, Mg, Mn, and P, were conveniently analysed by ICP-OES. Analytes were detected at the wavelength that maximised signal intensity and minimised spectral overlaps. P was measured at 178.221 nm for the genotype study, but 213.617 nm was later found to give more accurate and repeatable recovery of both peanut butter and wheat flour CRM. Sample concentration ranges were well above (>100 times greater) the LOD, even for less-sensitive elements that had relatively high LOD (i.e., K and P).

Elements that were present in samples at concentrations lower than the LOD of ICP-OES analysis, namely, Co, Cr, Mo, and Ni, were only measured by ICP-MS. The isotopes occurring in the greatest relative abundance, i.e., 59Co, 52Cr, 98Mo, and 58Ni, were found to be most suitable. ICP-OES quantitation of Na had a higher LOD than ICP-MS, but was preferable because it gave greater accuracy and precision.

We expected that Cu, Fe, and Zn would be adequately quantitated at the low end of the ICP- OES calibration range, or the high end of the ICP-MS range. In fact, the Cu content of some samples approached the ICP-OES LOQ, making it more reliable to quantitate on ICP-MS. Similar

| Development and validation of analytical methodology 26 accuracy and precision were achieved using 63Cu and 65Cu isotopes. Zn concentrations were determined just as well by ICP-OES as ICP-MS, making the former preferable due to the lower cost of operation. ICP-OES quantitation of Fe achieved a lower LOD, greater accuracy, and higher repeatability than ICP-MS. As discussed below, this was probably due to the spectral interferences affecting the most abundant Fe isotopes, 54Fe (38Ar16O and 40Ar14N) and 56Fe (40Ar16O and 40Ca16O). Ar, N, and Ca were expected to be major obstacles to Fe quantitation by ICP-MS for several reasons: Ar is not only ubiquitous in the environment but was also used as the nebuliser gas; nitric acid was the major digestion reagent and therefore present in the matrix; and the peanut samples contained high and, of course, variable levels of Ca. Compared to the standard ICP-MS configuration, ICP-OES was the superior method for Fe determination.

2.1.3.4 Use of a DRC in ICP-MS ICP-MS analysis is generally more susceptible to interferences than ICP-OES. Spectral overlaps may be caused by polyatomic and isobaric interferences whereby the analyte cannot be resolved from non-analyte elements of similar mass, species produced from the plasma, atmosphere, nebuliser gas, and matrix, or combinations of the above. In addition, doubly- charged ions cause interferences at half the mass-to-charge of the singly-charged ion, although this is minimised by instrument optimisation conducted prior to analysis. A number of argides are generated from the plasma gas, argon, which overlap with the major isotopes of several elements and thus degrade detection limits. Elemental analyses of food matrices are commonly subject to interferences caused by major food constituents such as C, Ca, and Cl. The major isotopes of some essential minerals expected to encounter interference problems in ICP-MS include 52Cr (40Ar12C, 35Cl16O1H, 36Ar16O, 38Ar14N, 37Cl15N, 36S16O), 53Cr (40Ar13C, 37Cl16O), 54Fe (40Ar14N, 38Ar16O, 37Cl16O1H), 56Fe (40Ar16O, 40Ca16O, 37Cl18O1H), 58Ni (23Na35Cl, 40Ar18O, 40Ca18O, 42Ca16O), 60Ni (44Ca16O, 23Na37Cl, 36Ar24Mg), 59Co (43Ca16O), 78Se (40Ar38Ar, 38Ar40Ca), and

82 12 35 40 1 Se ( C Cl2, Ar2 H2) (Cubadda et al., 2002, Pick et al., 2010).

Collision/reaction cells, such as the DRC designed by PerkinElmer SCIEX, are an important innovation in ICP-MS analysis for elimination of spectral interferences. In DRC mode, new species that are formed by the chemical reaction of plasma-generated ions with the DRC gas become the ions of interest. Unstable ions and new interferences formed within the DRC are ejected by a quadrupole mass filter before they enter the analyser quadrupole (PerkinElmer SCIEX, 2000). Problems with polyatomic interferences are reduced, but the resultant signal is also less intense, which degrades detection limits and may make DRC mode unsuitable when elements are present at very low levels.

| Development and validation of analytical methodology 27

The use of DRC mode was investigated for quantitation of some of the trace essential elements (Co, Cr, Cu, Fe, Ni, Se, and Zn) in peanut samples, using ammonia as the reaction gas. Without reference values for comparison, the usefulness of DRC mode for Co, Cr, and Ni quantitation was unclear. There were little differences between the standard and DRC mode measurements of 59Co and 58Ni. Cr measurements were lower and occasionally less than the LOD when using DRC compared to the standard mode, which suggests the elimination of some additive interferences. On the other hand, there were obvious benefits to the use of DRC mode for quantitation of Cu, Fe, and Zn. Cu measurement was accurate (95-105% recovery) in both standard and DRC modes, but the latter reduced the variability of measurement from 10-13% to just 5% for both 63Cu and 65Cu. DRC mode similarly improved the repeatability of Zn measurement, reducing the RSD from 8% to 4%, but there was no gain in accuracy. There were prominent improvements in both the accuracy and precision of Fe quantitation by 56Fe measurement, from 81-147% recovery and 19-25% RSD in standard mode, to 101-102% recovery and 6% RSD in DRC mode. Se measurements in standard and DRC modes of ICP-MS were similar. The validation statistics for Se, though not as good as the best configuration for other analytes, were acceptable considering that Se concentrations in the wheat flour CRM were very low (1.1 ± 0.2 μg g-1).

2.1.3.5 Validation statistics

2.1.3.5.1 Calibration linearity Correlation coefficients (r2) greater than 0.999 were obtained for all analyte calibrations, indicating good linearity over the specified concentration ranges (Table 2.4).

2.1.3.5.2 Limits of detection and quantitation LOD and LOQ for each analyte are shown in Table 2.5 alongside typical sample concentrations. LOD and LOQ values were of a similar magnitude to those previously reported (Cubadda et al., 2002, Nardi et al., 2009). The analytical methods were sufficiently sensitive to quantify all of the tested elements in peanut kernels, excluding Cr, Na, and Se, which were present at levels lower than the LOD in some samples.

2.1.3.5.3 Accuracy Experimentally-derived values of Ca, Cu, Fe, K, Mg, Mn, Mo, P, and Zn concentrations agreed well with certified values, achieving average recoveries of 95-100% in the peanut butter CRM, and 90-110% in the wheat flour CRM (Table 2.6). Measurements of Na and Se were less accurate (126 ± 11%; 88 ± 8%), as measured concentrations (6.1 ± 0.8 μg g-1 Na; 1.1 ± 0.2 μg g-1 Se) in the wheat flour CRM were low compared to the LOQ (41.9 μg g-1 Na; 0.07 μg g-1 Se). In

| Development and validation of analytical methodology 28 view of this, application to raw peanut kernels may be unreliable, since concentrations in samples were also low. Certified values of B, Co, Cr, and Ni concentrations were not available for the CRM used in this study so further testing is required for validation, preferably on a range of CRM unless an identical matrix becomes available.

2.1.3.5.4 Precision Instrumental repeatability for most elements was less than 2%; within-run precision less than 4%; and between-run precision less than 15% (Table 2.7). B had slightly lower instrumental repeatability (RSD 2.8%), probably due to ‘stickiness’ in the internal tubing; while K (2.9%), P (2.2%), and Se (2.3%) suffered from limited sensitivity at the very low concentrations measured. The between-run precision of B (32.4% in peanut butter CRM), Cr (46.6% in peanut butter CRM; 24.4% in wheat flour CRM), and Se (17.1% in peanut butter CRM) was also less than desirable.

Between-run, or intra-assay, precision is of greater practical importance than instrumental or within-run precision because it encompasses all stages of sample preparation and analysis. While 2% intra-assay precision is ideal (Hibbert, 1999), 15% or less is considered acceptable for complex matrices such as food samples (Dolan and Capar, 2002). It is interesting that other authors have reported problems of high variability in quantitation of Cr and Se (Dolan and Capar, 2002). The challenge of quantifying very low concentrations of Cr and Se may have been exacerbated by variability due to environmental contamination and spectral interferences, although the use of DRC should have minimised this. The use of the Elan DRC II ICP-MS reduced the tendency for B carryover compared to the ICP-OES, by virtue of a flow injection mechanism of sample introduction whereby samples are pushed into the carrier solution through a Teflon loop rather than introduction by peristaltic pumps, but high variability in B measurement was still apparent.

| Development and validation of analytical methodology 29

TABLE 2.4 LINEARITY OF ICP-(DRC)-MS AND ICP-OES CALIBRATIONS

Detection Calibration Correlation coefficient, r2 Analyte Method λ (nm) or isotope No. of pts Range (μg mL-1) B ICP-MS 11B 4 1 × 10-4 - 1 × 10-1 >0.9999 Ca ICP-OES 422.673 4-7 1 × 10-1 - 1 × 102 >0.9990 Co ICP-DRC-MS 59Co 4 1 × 10-4 - 1 × 10-1 >0.9999 Cr ICP-DRC-MS 52Cr 4 1 × 10-4 - 1 × 10-1 >0.9999 Cu ICP-DRC-MS 63Cu 4 1 × 10-4 - 1 × 10-1 >0.9999 Fe ICP-DRC-MS 56Fe 4 1 × 10-4 - 2 × 10-1 >0.9999 K ICP-OES 766.490 4-8 1 × 10-1 - 2 × 102 >0.9990 Mg ICP-OES 279.077 4-7 1 × 10-1 - 1 × 102 >0.9990 Mn ICP-OES 257.610 4-7 1 × 10-2 - 1 × 10 >0.9990 Mo ICP-MS 98Mo 4 1 × 10-4 - 1 × 10-1 >0.9999 Na ICP-OES 589.592 3-6 5 × 10-2 - 1 × 10 >0.9999 Ni ICP-DRC-MS 58Ni 4 1 × 10-4 - 1 × 10-1 >0.9999 P ICP-OES 213.617 6-7 1 × 10-1 - 1 × 102 >0.9990 Se ICP-DRC-MS 82Se 4 1 × 10-4 - 1 × 10-1 >0.9999 Zn ICP-DRC-MS 66Zn 4 1 × 10-4 - 1 × 10-1 >0.9999

| Development and validation of analytical methodology 30

TABLE 2.5 LOD, LOQ, AND TYPICAL SAMPLE CONCENTRATIONS FOR ICP-(DRC)-MS AND ICP-OES ANALYSES

Detection Limits (μg g-1) Sample distribution (μg g-1 dw) Analyte Method λ (nm) or isotope LOD LOQ Mean ± SD Min - Max

B ICP-MS 11B 2.14 × 10-1 7.12 × 10-1 1.91 × 10 ± 6.57 8.69 - 6.17 × 10 Ca ICP-OES 422.673 1.12 3.75 5.96 × 102 ± 1.10 × 102 3.63 × 102 - 9.21 × 102 Co ICP-DRC-MS 59Co 4.74 × 10-4 1.58 × 10-3 8.81 × 10-2 ± 2.59 × 10-2 4.18 × 10-2 - 1.82 × 10-1 52 -4 -3 -2 -2 -1

| Development and validation of analytical methodology Cr ICP-DRC-MS Cr 8.30 × 10 2.77 × 10 2.06 × 10 ± 1.89 × 10

TABLE 2.6 ACCURACY OF ICP-(DRC)-MS AND ICP-OES ANALYSES

Detection Peanut butter CRM (N=25) Wheat flour CRM (N=26) Analyte Method λ (nm) or isotope Certified conc. (μg g-1) Expt. conc. (μg g-1) Recovery (%) Certified conc. (μg g-1) Expt. conc. (μg g-1) Recovery (%)

Ca ICP-OES 422.673 411 ± 18 390 ± 22 95 ± 5 191 ± 4 177 ± 10 93 ± 6 Cu ICP-DRC-MS 63Cu 4.93 ± 0.15 4.94 ± 0.23 100 ± 5 2.1 ± 0.2 2.0 ± 0.1 95 ± 5 Fe ICP-DRC-MS 56Fe 16.4 ± 0.8 16.6 ± 1.1 101 ± 6 14.1 ± 0.5 14.4 ± 0.9 102 ± 6

| Development and validation of analytical methodology K ICP-OES 766.490 6070 ± 200 5859 ± 212 97 ± 3 1330 ± 3 1193 ± 36 90 ± 3 Mg ICP-OES 279.077 1680 ± 70 1620 ± 73 96 ± 4 400 ± 20 360 ± 15 90 ± 4 Mn ICP-OES 257.610 16 ± 0.6 16 ± 0.6 100 ± 4 9.4 ± 0.9 8.7 ± 0.3 93 ± 3 Mo ICP-MS 98Mo n/a 0.48 ± 0.03 0.53 ± 0.06 109 ± 12 Na ICP-OES 589.592 4890 ± 140 4816 ± 302 98 ± 6 6.1 ± 0.8 7.7 ± 0.8 126 ± 11 P ICP-OES 213.617 3378 ± 92 3395 ± 116 100 ± 3 1340 ± 60 1208 ± 44 90 ± 3 Se ICP-DRC-MS 82Se n/a 1.1 ± 0.2 1.0 ± 0.1 88 ± 8 Zn ICP-DRC-MS 66Zn 26.3 ± 1.1 25.1 ± 1.1 95 ± 4 11.6 ± 0.4 10.5 ± 0.5 91 ± 4 32

TABLE 2.7 PRECISION OF ICP-(DRC)-MS AND ICP-OES ANALYSES

Detection Instrumental repeatability (%) Within-run precision (%) Between-run precision (%) Analyte Method λ (nm) or isotope (N=7) (N=6-16) Peanut butter CRM (N=25) Wheat flour CRM (N=26)

B ICP-MS 11B 2.8 3.5 32.4A

| Development and validation of analytical methodology Cr ICP-DRC-MS Cr 1.0 2.2 46.6 24.4 Cu ICP-DRC-MS 63Cu 1.1 1.9 4.7 5.2 Fe ICP-DRC-MS 56Fe 0.7 2.5 6.3 6.3 K ICP-OES 766.490 2.9 3.1 3.6 2.8 Mg ICP-OES 279.077 1.5 2.2 4.5 4.2 Mn ICP-OES 257.610 1.1 2.3 3.7 3.7 Mo ICP-MS 98Mo 0.8 1.1 6.7A 10.6 Na ICP-OES 589.592 0.9 3.1 6.3 10.8 Ni ICP-DRC-MS 58NI 2.0 1.9 4.9A 6.3A P ICP-OES 213.617 2.2 2.5 3.4 3.6 Se ICP-DRC-MS 82Se 2.3 2.9 17.1A 9.6 Zn ICP-DRC-MS 66Zn 1.9 2.6 4.4 4.5 A Marked analytes were not certified in the CRM and imprecision may reflect absence in the CRM 33

2.1.4 Conclusion

Optimal configurations for ICP analysis of essential minerals in peanut kernels were selected after testing a range of ICP-OES, ICP-MS, and ICP-DRC-MS methods. The accuracy (Table 2.6) and precision (Table 2.7) of the selected methods for quantitation of Ca, Cu, Fe, K, Mg, Mn, Mo, P, and Zn concentrations were validated by analyses of peanut butter and wheat flour CRM. These methods were found to be appropriate for application to peanut kernels, in view of the distribution of concentrations among samples (Table 2.5). Na and Se suffered some problems with inaccuracy at the low concentrations tested, though precision was acceptable. Certified values of B, Co, Cr, and Ni concentrations were not available for the CRM used in this study, so further testing is required to validate the accuracy of the methods for these elements, preferably on a range of CRM unless an identical matrix becomes available. Having said this, we suggest that reliable quantitation of Co and Ni content will be possible, as these could be measured with acceptable levels of precision, and the ICP-MS analyses in standard and DRC modes yielded similar results. Cr and Se analyses, however, were insufficiently precise for reliable quantitation. B, Cr, Na, and Se analyses remain challenging because of the very low concentrations in peanuts, and inaccuracy or imprecision in analysis at these trace levels.

| Development and validation of analytical methodology 34

2.2 Discussion of antioxidant assays

2.2.1 Objectives

Chemical-based in vitro assays such as the methods employed in this research have become popular among antioxidant researchers as high-throughput screens for comparative study of antioxidant capacity among diverse dietary samples. The 2,2’-azino-bis-(3- ethylbenzothiazoline-6-sulfonic acid) (ABTS•+) decolourisation (Arnao et al., 2001), 2,2- diphenyl-1-picrylhydrazyl radical (DPPH•) scavenging (Brand-Williams et al., 1995), FC total phenolics (Waterhouse, 2002), and ORAC (Prior et al., 2003) assays were used to evaluate the antioxidant capacity of peanut kernels, with minor optimisation for the sample type and laboratory conditions. The principal goals of our experiments were to evaluate the genotypic variation in antioxidant capacity among 32 full-season and 24 ultra-early maturity representative genotypes from the APBP, to determine the breeding potential for antioxidant- related traits in kernels, and to identify specific breeding lines that may be useful for further study in this respect.

Several methods were employed as there has been considerable debate regarding the application and interpretation of ‘total antioxidant’ assays, no consensus on a standard or optimal protocol, and frequent reports of discrepant results generated by different methods or varying assay conditions. It was not our intention to undertake a thorough method validation, as the antioxidant assays have been extensively published and frequently reviewed. Data is thus limited to calibration statistics from assays conducted during the course of the study, which were compiled for diagnostic purposes. This section is thus not so much a validation of the methodology, but rather a critical discussion regarding practical aspects of the assays and the quality of the data that was generated from their use.

2.2.2 Comparison of ABTS•+, DPPH•, FC, and ORAC assays

The ABTS•+, DPPH•, and FC assays involve colorimetric reactions that are conveniently measured using a commonly-available UV-Visible spectrophotometer, whereas the ORAC assay involves the kinetic measurement of a fluorescence signal. Each of these methods entails apparently straightforward chemistry: in the ABTS•+ and DPPH• assays, absorbance decreases as the coloured radical species is scavenged by antioxidants; in the FC assay, the reduced form of the FC reagent is coloured, so that absorbance is proportional to the reducing power of antioxidants; whereas in the ORAC assay, the fluorescence of a probe molecule decreases due to oxidant activity, with the onset and rate of substrate oxidation influenced by the

| Development and validation of analytical methodology 35 concentration of antioxidants. Yet despite the apparent simplicity of the antioxidant assays, we experienced substantial difficulty with high experimental variability.

2.2.2.1 Assay variability

2.2.2.1.1 Calibration linearity and sensitivity Calibration equations generated from standard series measured on each microplate had adequate linearity in each assay (Table 2.8), but the sensitivity of calibrations varied excessively, especially in the ABTS•+ and lipophilic ORAC (L-ORAC) assays (Figure 2.1). The

•+ •+ ABTS /potassium persulfate (K2S2O8) version of the ABTS assay performed more consistently •+ than the ABTS /horseradish peroxidase (HRP)/hydrogen peroxide (H2O2) protocol, and was the •+ method adopted for sample analysis. Yet even after switching to the ABTS /K2S2O8 protocol, repeatability for analyses performed over the course of the study remained sub-optimal. External calibration standards that were measured in each microplate could compensate for minor shifts in reaction kinetics and experimental conditions, but large variations in sensitivity can render inter-plate comparisons less meaningful.

•+ • TABLE 2.8 VARIATION IN CALIBRATION LINEARITY OF ABTS , DPPH , FC, AND ORAC ASSAYS

Antioxidant assay Calibration Correlation coefficient, r2 Method Units No. of points Range (μM) No. of analyses Mean ± SD

ABTS•+ TE 5 100 - 1000 19 0.994 ± 0.014 DPPH• GAE 5 10 - 100 19 0.996 ± 0.004 FC GAE 5 100 - 500 19 0.991 ± 0.008 Hydrophilic ORAC TE 5 10 - 100 36 0.996 ± 0.005 Lipophilic ORAC TE 5 10 - 100 35 0.991 ± 0.014

2.2.2.1.2 Between-plate variability Absorbance measurements in the ABTS•+ assay (Figure 2.2), especially the higher concentration (1000 μM) Trolox standard, displayed inordinate variability over the course of 19 assays performed in 2009-2010. The DPPH• (Figure 2.3) and FC (Figure 2.4) methods produced more repeatable measurements, but had low sensitivity and were unable to statistically distinguish (P<0.05) the tested genotypes. It may be that the 96-well microplate platform was less robust than the original cuvette format, and more sensitive to variations in analytical conditions. Considerably more work was necessary to optimise the assays with identification of critical parameters that could be monitored for quality control. In view of this, we opted to focus on the hydrophilic ORAC (H-ORAC) assay, which was the only assay sufficiently sensitive and repeatable (Figures 2.8 and 2.9) to reveal genotypic differences.

| Development and validation of analytical methodology 36

20 15 10 FC analysisFC 5 0 1 SD 0 r -03 -03 -03 -04 10 10 10 10

u u u u

2 1 1 5 Slope coefficient Slope

1 CI (95%) 1 CI r 20 40 15 30 . 10 20 5 DPPH analysis DPPH 10 L-ORAC analysisL-ORAC 0 0 0 0 04 04 04 04 -01 -01 -01 -01 10 10 10 10 10 10 10 10 u u u u

u u u u

2 1 4 3

4 3 2 1

Slope coefficient Slope Slope coefficient Slope

40 20 30 15 20 10 5 ABTS analysisABTS 10 H-ORAC analysisH-ORAC ARIATION IN SLOPE COEFFICIENTS OF ANTIOXIDANT ASSAYS 0 0 V 0 0 04 04 04 04 03 -04 -04 -04 -04 10 10 10 10 10 2.1 10 10 10 10 u u u u u

u u u u

3 2 2 1 5

6 4 2 8

Slope coefficient Slope Slope coefficient Slope IGURE

F

| Development and validation of analytical methodology 37

FIGURE 2.2 0.6 100 PM Tr olox VARIATION IN •+ ABTS 500 PM Tr olox MEASUREMENTS 1000 PM Tr olox 0.4 OF STANDARD SOLUTIONS.

Absorbance 0.2 r 1 CI (95%) r 1 SD

0.0 0 5 10 15 20 25 Analysis

0.3 FIGURE 2.3 VARIATION IN 10 PM Gallic acid • DPPH 50 PM Gallic acid MEASUREMENTS 0.2 100 PM Gallic acid OF STANDARD SOLUTIONS.

r 1 CI (95%) r 1 SD Absorbance 0.1

0.0 0 5 10 15 20 25 Analysis 1.0 FIGURE 2.4 VARIATION IN 100 PM Gallic acid 0.8 FC 300 PM Gallic acid MEASUREMENTS 500 PM Gallic acid OF STANDARD 0.6 SOLUTIONS. 0.4 r 1 CI (95%) r 1 SD Absorbance

0.2

0.0 0 5 10 15 20 25 Analysis

2.2.2.2 Correlations between antioxidant assay methods Correlations between the antioxidant assays tested in this study are shown in Table 2.9. The total ORAC value was composed primarily (~90%) of the H-ORAC result so the strong correlation between the two variables was self-evident. There were also statistically significant positive correlations between the ABTS•+ and DPPH• (r=0.283, P=0.008), ABTS•+ and FC

| Development and validation of analytical methodology 38

(r=0.243, P=0.024), DPPH• and FC (r=0.444, P<0.001), and DPPH• and L-ORAC (r=0.207, P=0.043) results.

The antioxidant literature, mainly examining fruit and plant extracts, has typically reported positive correlations between ORAC value and phenolic content (e.g., Dudonné et al., 2009, Proteggente et al., 2002). Some authors have attempted to categorise assays according to their chemistry, e.g., whether they primarily involve electron transfer (ET) or proton donation (HAT) reactions (Huang et al., 2005, Karadag et al., 2009, Prior et al., 2005), but such implied correlations have been inconsistently borne out in experimental results. In some respects our results had general correspondence with the expected chemistry, i.e., strong positive correlation (P<0.01) between the ET-based DPPH• and FC assays. On the other hand, the ABTS•+ and ORAC assays are based on HAT mechanisms yet had no correlation at all.

•+ • TABLE 2.9 BIVARIATE CORRELATIONS AMONG ABTS , DPPH , FC, AND ORAC MEASUREMENTS

Method Statistic (N=96) ABTS•+ DPPH• FC H-ORAC L-ORAC Total ORAC ABTS•+ Pearson correlation 1 0.283** 0.243* 0.156 -0.204 0.112 Sig. (2-tailed) 0.008 0.024 0.153 0.060 0.304 DPPH• Pearson correlation 0.283** 1 0.444** 0.142 0.207* 0.178 Sig. (2-tailed) 0.008 <0.001 0.169 0.043 0.082 FC Pearson correlation 0.243* 0.444** 1 0.045 0.065 0.057 Sig. (2-tailed) 0.024 <0.001 0.662 0.527 0.583 H-ORAC Pearson correlation 0.156 0.142 0.045 1 0.007 0.981** Sig. (2-tailed) 0.153 0.169 0.662 0.950 <0.001 L-ORAC Pearson correlation -0.204 0.207* 0.065 0.007 1 0.199 Sig. (2-tailed) 0.060 0.043 0.527 0.950 0.052 Total ORAC Pearson correlation 0.112 0.178 0.057 0.981** 0.199 1 Sig. (2-tailed) 0.304 0.082 0.583 <0.001 0.052 *Significant (P<0.05) **Highly significant (P<0.01)

| Development and validation of analytical methodology 39

2.2.3 Optimisation of ORAC assay

The study of genotypic variation in the antioxidant capacity of peanut kernels revealed that the hydrophilic extract was the predominant contributor to total antioxidant capacity and that, of the assay methods employed, only the ORAC assay was sufficiently sensitive to statistically differentiate individual genotypes based on their antioxidant phenotype. In the course of subsequent studies concerning the G × E interaction affecting antioxidant capacity and the correlation of antioxidant capacity with phytochemical profile, we continued to employ the H- ORAC assay while undertaking to improve the repeatability and quality of results. As mentioned previously, a full method validation was not attempted as such work has already been published (esp. Dávalos et al., 2004, Prior, 2003). However the large number of analyses undertaken for this research provided an opportunity to perform quality diagnostics, which are reported and discussed here.

2.2.3.1 Modifications to the ORAC protocol There have been a number of improvements in the ORAC method since it was first published in 1993, most notably, the replacement of β-phycoerythrin with fluorescein as the fluorescent probe and conversion to high throughput platforms. We pursued a 96-well format because of the large number of samples planned for analysis, as this platform lends itself to manual (not robotic) yet high throughput screening. The assay protocol was based on that of Prior et al. (2003), but modifications beyond mere scaling were necessary in order to achieve acceptable repeatability and a practical assay runtime. Replicated experiments for the optimisation are not reported as changes to the initial protocol were minor; however a brief explanation of the modifications may be of interest to other researchers and is given here. The critical assay parameters are listed in Table 2.10 for convenient comparison.

First, the ratio of fluorescein and AAPH concentrations was adjusted so that the fluorescence decay of all Trolox calibration standards occurred within 2 h. The final fluorescein concentration per well was decreased to 52.5 nM and AAPH increased to 17.5 mM in order to achieve suitable fluorescence decay curves. It is unclear whether this was necessary because of differences in the activity of reagents or temperature control in different laboratories. The assay was monitored for 2 h to enable extension of the calibration range to 100 μM to ensure that diverse samples could be tested at a uniform dilution, in view of the large influence of matrix effects on results that has been reported (Pérez-Jiménez and Saura-Calixto, 2006). Injecting a smaller volume of AAPH at a higher concentration (25 μL, 140 mM) than is described in previous publications decreased the noisiness of the fluorescence signal in the

| Development and validation of analytical methodology 40 period immediately following injection, and improved the assay repeatability. This benefit was presumably due to a reduced impact on well temperature.

TABLE 2.10 KEY HISTORICAL DEVELOPMENTS IN THE ORAC PROTOCOL

Format Description Reference Cuvette Original publication of ORAC protocol. Cao et al. (1993) (2 mL) Reaction: β-phycoerythrin (16.7 nM); AAPH (3 mM); sample extract (20 μL) or Trolox standard (100 μM).

Measurement: λex/λem 540/565 nm; 5 min intervals for 60-90 min. COBAS FARA Fluorescein replaced β-phycoerythrin as fluorescent probe. Conversion to a (rarely Ou et al. II used) high throughput platform. (2001) (400 μL) Reaction: fluorescein (365 μL, 48 nM); AAPH (12.8 mM); sample extract (20 μL) or Trolox standard (0-100 μM).

Measurement: λex/λem 493/515 nm; 1 min intervals for 30 min COBAS FARA Lipophilic version of ORAC using randomly methylated β-cyclodextrin to enhance Huang et al. II solubility. (2002a) (400 μL) Reaction: fluorescein (63 nM); AAPH (12.8 mM). 96-well First protocol on a commonly used high-throughput platform. Huang et al. microplate (2002b) Reaction: fluorescein (150 μL, 61.2 nM final); AAPH (25 μL, 19 mM final); sample (200 μL) extract (25 μL) or Trolox standard (0-50 μM).

Measurement: λex/λem 485/530 nm; 1 min intervals for 35 min. 48-well Validation of extraction and 48-well format for food, plasma, and biological Prior et al. microplate samples. NB, the authors advised that 48-well yielded better repeatability than (2003) 96-well microplates. (515 μL or 590 μL) Reaction: fluorescein (400 μL, 83 nM final concentration in H-ORAC and 72 nM in L-ORAC); AAPH (75 μL in H-ORAC; 150 μL in L-ORAC, 9 mM final); sample extract (40 μL) or Trolox standard (0-50 μM).

Measurement: λex/λem 485/520 nm; ~1.5 min (unclear) intervals for 60 min. 96-well Further validation of 96-well microplate platform. Dávalos et al. microplate (2004) Reaction: fluorescein (120 μL, 70 nM final); AAPH (60 μL, 12 mM final); sample (200 μL) extract (20 μL) or Trolox standard (0-80 μM).

Measurement: λex/λem 485/520 nm; 1 min intervals for 80 min. Cuvette Pyrogallol red suggested as an alternative to fluorescein as the probe molecule. Alarcón et al. (2008) (? mL) Reaction: pyrogallol red (5 μM); AAPH (10 mM). Measurement: 540 nm; frequent readings until absorbance is 20% of initial.

| Development and validation of analytical methodology 41

2.2.3.2 Assay variability

2.2.3.2.1 Calibration linearity and sensitivity The assay had consistently good linearity within the calibration range, and correlation coefficients (r2) greater than 0.999 were obtained for each of 220 analyses performed in 2009- 2012. It is emphasised that calibration standards were run on each microplate in consideration of the assay sensitivity to fluctuations in test conditions. In particular, fluorescence intensity, thermal degradation of AAPH to generate peroxyl radicals, and the rate of assay reactions are all highly temperature-dependent.

Stability of the assay sensitivity, indicated by the calibration slope coefficient, is one of the most important parameters that circumscribe repeatability. The slope coefficients generated in 220 assay calibrations are plotted in Figure 2.5. Variability in the slope coefficient was nearly identical to that of 100 μM Trolox, with a RSD of 13%, perhaps not surprisingly given the influence of the highest concentration data point on the regression fit.

30000

20000 r 3 CI (95%) r 1 SD

10000 Slope coefficient

0 0 50 100 150 200 250 Analysis

FIGURE 2.5 VARIATION IN SLOPE COEFFICIENTS OF ORAC ANALYSES PERFORMED IN 2009-2012.

2.2.3.2.2 Within-plate variability Temperature control of the microplate contents is the paramount issue in ORAC assay variability. When comparing two models of microplate reader, Prior et al. (2003) observed that the Optima (BMG Labtech) model recorded consistently high readings for the first seven well positions. Our experience was that this was not so much an artefact of “erroneously high net areas” in specific wells, but reflected a general trend in the rate of fluorescence decay that extended through the entire plate. In other words, area-under-the-curve (AUC) measurements of replicate wells increased progressively from row A to row H with corresponding decrease in ORAC values. Thus, the variation between rows (10-16% RSD) was relatively high while that

| Development and validation of analytical methodology 42 within rows (1-6%) was more acceptable (Figure 2.6). It is likely that the reaction rates were influenced by a temperature gradient despite careful pre-incubation of reagents, possibly because instrument positioning of the microplate has row A innermost and row H closest to the drawer opening. Exclusion of edge wells, which were filled with water as a thermal buffer, and design of a plate layout that distributed well replicates along columns (Figure 2.7), were therefore important factors in minimising well-to-well variability among replicates.

RSD Row ± SD 1 2 3 4 5 6 7 8 9 10 11 12 (%) mean

A ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ 6 5.6 × 105 ± 3.6 × 104

B ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ 5 5.9 × 105 ± 3.0 × 104

C ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ 4 6.3 × 105 ± 2.8 × 104

D ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ 2 6.6 × 105 ± 1.4 × 104

E ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ 1 6.9 × 105 ± 8.5 × 103

F ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ 4 7.2 × 105 ± 2.8 × 104

G ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ 2 7.8 × 105 ± 1.3 × 104

H ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ 3 8.2 × 105 ± 2.8 × 104

RSD RSD (%) 10 10 13 14 15 16 15 15 15 15 13 12

4 4 4 4 4 5 5 4 5 4 4 4 10 × 10 × 10

SD 6.7 × 10 7.2 × 10 8.8 × 10 9.8 × 10 9.9 × 10 1.1 × 10 1.0 × 10 9.9 × 10 1.0 9.8 8.9 × 10 7.7 ×

± ± ± ± ± ± ± ± ± ± ± ± ±

5 5 5 5 5 5 5 5 5 5 5 5 Column mean 6.9 × 10 7.1 × 10 6.9 × 10 6.9 × 10 6.6 × 10 6.8 × 10 6.8 × 10 6.7 × 10 6.7 × 10 6.7 × 10 6.8 × 10 6.6 × 10

FIGURE 2.6 SPATIAL VARIATION IN ORAC FLUORESCENCE DECAY (AUC) MEASUREMENTS WITHIN ROWS (A-H) AND BETWEEN ROWS (WITHIN COLUMNS 1-12) OF A 96-WELL MICROPLATE.

1 2 3 4 5 6 7 8 9 10 11 12

A ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ B: phosphate buffer blank

B ⃝ B S5 X1 X3 X5 X7 X9 X11 B S5 ⃝ S1-S5: 10-100 μM Trolox

C ⃝ S1 S4 X11 X1 X3 X5 X7 X9 S1 S4 ⃝ X1-X12: extracts 1-12

D ⃝ S2 S3 X9 X11 X1 X3 X5 X7 S2 S3 ⃝ Edge wells: water

E ⃝ S3 S2 X8 X10 X12 X2 X4 X6 S3 S2 ⃝

F ⃝ S4 S1 X6 X8 X10 X12 X2 X4 S4 S1 ⃝

G ⃝ S5 B X4 X6 X8 X10 X12 X2 S5 B ⃝

H ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝ ⃝

FIGURE 2.7 ORAC PLATE LAYOUT DESIGNED TO MINIMISE WELL-TO-WELL VARIABILITY YET REMAIN CONVENIENT FOR MANUAL PIPETTING.

| Development and validation of analytical methodology 43

2.2.3.2.3 Between-plate variability Fluctuations in the fluorescence intensity and decay curve are to a large extent resolved by external calibration within each microplate. Previous authors have not suggested control parameters; however we monitored the stability of the assay in order to identify and if necessary control sources of variability. Figure 2.8 plots the blank-corrected AUC measurements for the lowest, highest, and mid-concentration Trolox standards, i.e. 10, 100, and 50 μM, analysed in 2009-2012 (N=220). A zone of three 95% confidence intervals is plotted as a visual reference to the high variability in fluorescence decay. The same data, but including all standard concentrations, is depicted as single points with error bars in Figure 2.9. This second representation emphasises the linearity of the calibrations, even averaged across 220 assays with highly variable decay kinetics.

06 3u10 100 PM Tr olox 50 PM Tr olox 10 PM Tr olox

2u10 06

r 3 CI (95%) r 1 SD 1u10 06 Area under curve

0 0 50 100 150 200 250 Analysis

FIGURE 2.8 VARIATION IN ORAC MEASUREMENTS OF STANDARDS PERFORMED IN 2009-2012.

3u10 06 Mean r SD

Mean r 3 CI (95%)

13% RSD 2u10 06 14% RSD

15% RSD 1u10 06 18% RSD Area under curve

26% RSD

0 0 50 100 150 Trolox concentration (PM)

FIGURE 2.9 MEAN ORAC MEASUREMENTS OF STANDARDS PERFORMED IN 2009-2012. In practice, observation of unusually fast (low AUC) or slow (high AUC) fluorescence decay in the laboratory could usually be attributed to inadequate temperature control. For example,

| Development and validation of analytical methodology 44 fluorescein that was inadequately warmed would generate less fluorescence and effectively performed like a lower concentration of substrate, resulting in low AUC. In contrast, a slower oxidation of the substrate and higher AUC would result if the AAPH or the plate reader had been given insufficient time to equilibrate at 37 °C.

2.2.3.3 Interpretation of ORAC results The proliferation of ‘total antioxidant’ assays since the 1990s has aroused ongoing debate regarding their robustness for screening purposes, the biological relevance of their chemistry, and the heterogeneity of sample responses when using different methods. The ORAC assay, one of several methods to gain popular application within research and industry, has naturally been accompanied by such critique. It is illustrative that the Nutrient Data Laboratory of the USDA maintained an online ORAC database for food products since 2007 (U. S. Department of Agriculture and Agricultural Research Service, 2010), only to remove it in 2012 citing that “values indicating antioxidant capacity have no relevance to the effects of specific bioactive compounds, including polyphenols on human health” and are “… routinely misused by food and dietary supplement manufacturing companies… and by consumers” (U. S. Department of Agriculture and Agricultural Research Service, 2012).

We suggest that the ORAC assay, along with other ‘total antioxidant’ methods, remains a relevant and useful tool for comparative analysis of food and biological samples, provided that the interpretation of results is restricted to the scope of the chemistry. Disputes regarding biological relevance and ‘misuse’ of results arise when practitioners focus excessively on extrapolated bioactivity, instead of considering the assay for what it is, an indicator of antioxidant function within a limited system of substrate, antioxidant, and pro-oxidant. In vitro assays are useful tools for convenient laboratory analysis; but they complement and do not replace in vivo research.

2.2.4 Conclusion

Comparison of ABTS•+, DPPH•, FC, and ORAC assays suggested that DPPH• and FC were less susceptible to experimental variation than the ABTS•+ assay, but that ORAC was the most sensitive of the tested methods in the 96-well microplate format. Temperature control was paramount to minimise well-to-well variability in the ORAC assay. Pre-incubation of reagents and microplates, thermal buffering of the edge wells, design of plate layout to spatially distribute standards and samples, and minimising injection volumes were several strategies used to optimise ORAC repeatability.

| Development and validation of analytical methodology 45

2.3 Validation of HPLC analysis

2.3.1 Objectives

HPLC was used to profile the constituents of hydrophilic extracts of peanut kernels, with intent to identify and quantify these, and correlate them with antioxidant capacity. The optimisation and validation of the extraction and analytical methodology are presented here.

2.3.2 Materials and methods

2.3.2.1 Instruments HPLC analysis was performed on a Shimadzu Prominence series UFLC system (Shimadzu Australasia, Rydalmere, NSW, Australia) comprising CBM-20A system controller, DGU-20A5 degasser, LC-20AD solvent delivery unit, SIL-20A autosampler, CTO-20A column oven, SPD- M20A photo diode array (PDA) detector, RF-10AXL fluorescence detector (RF), and FRC-10A fraction collector, which were controlled through Lab Solutions LCsolution version 1.25 software. The use of an Antec Decade II (Antec B.V., Zoeterwoude, The Netherlands) electrochemical (ECD) detector was also explored.

2.3.2.2 Reagents HPLC-grade standards including biochanin A (D2016), caffeic acid (C0625), daidzein (D7802), ferulic acid (128708), gallic acid (G7384), genistein (G6649), kaempferol (60010), m-coumaric acid (H23007), myricetin (70050), o-coumaric acid (H22809), p-coumaric acid (C9008), p- hydroxybenzoic acid (H9882), polydatin (15721), protocatechuic acid (P5630), quercetin dihydrate (Q0125), rutin (R5143), salicylic acid (84210), sinapic acid (D7927), syringic acid (S6881), t-cinnamic acid (C80857), t-resveratrol (R5010), and vanillic acid (94770) were purchased from Sigma-Aldrich Co. (St. Louis, MO, USA).

Alcalase 2.4 L FG, a protease (subtilisin) enzyme preparation, and Celluclast 1.5 L, a cellulase enzyme preparation, were gift samples from Novozymes Australia Pty. Ltd. (North Rocks, NSW, Australia). Analytical grade n-hexane and methanol were used in extraction. HPLC samples and mobile phase were made from ultrapure water (18.2 mΩ cm-1, Millipore Corp., Billerica, MA, USA) and HPLC-grade acetonitrile, methanol, and trifluoroacetic acid (TFA; 302031, Sigma- Aldrich Co.).

2.3.2.3 Samples The samples comprised six genotypes (D147-p8-6F, Fisher, Page, Sutherland, and UF32) of raw kernels obtained from the APBP and several roasted samples of ‘Middleton’ peanuts provided by DEEDI, as described in chapter 5.2.2.2.

| Development and validation of analytical methodology 46

2.3.2.4 Sample extraction Crude sample extracts were prepared by hexane-defatting and ultrasound-assisted extraction (UAE) in 90% aqueous methanol with several filtration, concentration, and reconstitution steps. Replicate samples were also treated with Alcalase (protease) and Celluclast (cellulase) enzyme preparations. Factors that were optimised included the temperature (ambient, 40, and 60 °C) and duration (5, 15, 30, and 60 min) of sonication. Several hydrolysis and enzyme treatments were also investigated. The final protocol was as follows.

Peanut sample (100 g) was embrittled with liquid nitrogen and pulverised in an analytical mill (A11 basic model, IKA Works (Asia), Rawang, Selangor, Malaysia). Ground sample (5.0 g) was transferred to a 50-mL screw-cap tube and vortexed with n-hexane (45 mL). The tube was sonicated (50 kHz, 40 °C, 5 min), centrifuged (9500 g, 10 min), and the supernatant discarded. The n-hexane extraction was repeated twice more and then the tube was flushed gently with nitrogen gas until the residual n-hexane had evaporated.

For ‘enzyme-treated extract’, the defatted sample was incubated (37 °C) for 24 h with Alcalase (1.0 mL), Celluclast (1.0 mL), and water (8.0 mL). Methanol (40 mL) was added and the mixture agitated overnight on a ten-roller mixer (Thermoline Scientific Pty. Ltd., Wetherill Park, NSW, Australia). The test tube was centrifuged (9500 g, 10 min) and the supernatant collected in a pear-shaped evaporation flask. The sample was extracted twice more in 90% aqueous methanol (25 mL), with sonication, centrifugation, and collection of the supernatant as described previously. ‘Native extract’ was prepared in the same way, but without the enzyme treatment. Instead, it was extracted thrice with 90% aqueous methanol (25 mL), firstly by overnight rolling and the latter times by sonication.

The pear-shaped evaporation flask containing methanolic extract was placed in a water bath on a hot plate (H30/30 model, CAT Ingenieurbüro M. Zipperer Gmbh, Staufen, Germany), which was adjusted to maintain the water temperature at 40-50 °C, and flushed with nitrogen gas until the volume had reduced to several millilitres. This was transferred to a screw-cap tube and centrifuged (9500 g, 10 min) before vacuum filtration through a 1.0 μm glass fibre filter paper (APFB04700, Millipore Corp., Billerica, MA, USA). The filtrate was diluted with water to produce either a 5% or 20% methanol solution and adjusted to pH 2 with hydrochloric acid (0.1 M) for use in SPE.

2.3.2.5 Solid phase extraction A Supelco Visiprep DL 24-port vacuum manifold with drying attachment (57265 and 57124, Sigma-Aldrich Co., St. Louis, MO, USA) and Strata-X (33 μm, 85 Å) polymeric reversed phase

| Development and validation of analytical methodology 47 sorbent (100 mg/3 mL) tubes (8B-S100-EBJ, Phenomenex, Inc., Lane Cove, NSW, Australia) were used to perform SPE. The sorbent was pre-conditioned with acidified methanol (5 mL, pH 2) followed by water (5 mL) prior to loading the extract solution under vacuum at a controlled flow rate (1 mL min-1). After loading, the sorbent was washed with water (10 mL) then flushed with nitrogen gas using the drying attachment (15 min). Retained compounds were eluted with HPLC-grade methanol (3 mL) in two aliquots (2 mL, 1 mL). The eluate was evaporated under nitrogen gas to 1 mL and diluted with ultrapure water to accurate volume in a 2-mL volumetric flask. The purified extract was passed through a Microscience Hydraflon 0.45 μm PTFE syringe filter (MS SF13HY045, MicroAnalytix Pty. Ltd., Taren Point, NSW, Australia) then transferred to an autosampler vial for HPLC analysis. Factors that were optimised included the choice of sorbent, pre-conditioning and loading solutions (i.e., pH and % organic solvent), and elution conditions (i.e., selection and volume of eluent).

2.3.2.6 HPLC analysis

HPLC separation was carried out on a Gemini reversed phase C18 (5 μm, 110 Å, 250 × 4.6 mm;

398123-3) column with a SecurityGuard C18 (4 × 3.0 mm) guard cartridge (00G-4435-E0 and AJ0-4287, Phenomenex Inc., Lane Cove, NSW, Australia). The mobile phase comprised solutions A, 0.1% aqueous TFA, and B, 0.1% TFA in acetonitrile, according to the following gradient elution: 0-30 min, 15-40% B; 30-31 min, 40-15% B; and 31-35 min, 15% B. The flow rate was 1.5 mL min-1 and the injection volume was 30 μL. The PDA, RF, and fraction collector were configured in series to allow simultaneous measurement. The delay volume from the PDA to the fraction collector (294 μL) was calculated so that the recorded retention times of these two modules matched; whereas retention times associated with the RF measurements were not corrected (the delay was ~0.1 min). The PDA was set to monitor 190-400 nm, while the RF measured at λex/λem 260/422 nm. Sample constituents were identified by comparing retention times and UV spectra with the respective standard compounds, and quantified using calibrations of standard concentration against peak area. HPLC parameters that were optimised included the selection of HPLC column, mobile phase, gradient, and configuration of detectors.

2.3.2.7 Quality controls Peanut kernels were sampled in triplicate and subsamples extracted in triplicate using each of the extraction methods, i.e. with and without enzyme treatment, and by 5% and 20% aqueous methanol-based SPE. Frequent inclusion of spiked replicates in which standard compounds (25 or 50 μg) were added prior to extraction allowed monitoring of extraction repeatability and provided an indication of accuracy. QC check standards were measured every ~20 injections to

| Development and validation of analytical methodology 48 monitor drift in retention time, signal intensity, and any potential degradation or change in peak spectra.

2.3.2.8 Moisture content The moisture content of the peanut samples was estimated gravimetrically (AOAC Official Method 925.40). Ground sample (2 g) was accurately weighed in an oven-dried aluminium pan and heated in a vacuum oven (<100 mm Hg, 95 °C) for 5 h. The dish was again accurately weighed after cooling in a dessicator, and the weight loss considered to represent moisture content.

2.3.2.9 Data analysis Accuracy was evaluated by the recovery of spiked phytochemical standards from the sample extracts. Instrumental repeatability was estimated as the RSD of consecutive QC measurements; within-run precision as the RSD of QC measurements taken at regular intervals within an analytical run; and between-run precision as the RSD of QC measurements throughout the entire experiment, which comprised 65 analytical runs over seven months. The LOQ was estimated as the minimum concentration within the peanut extract that could be quantified with a degree of certainty based on the linear calibration range for each tested analyte, and the LOD was taken as three-tenths of the LOQ.

2.3.3 Results and discussion

2.3.3.1 Development of extraction method

2.3.3.1.1 Ultrasound-assisted extraction The priority was to attain repeatable, convenient, and relatively high-throughput extraction, rather than to maximise extraction yield. Augmented extraction methods such as accelerated solvent, microwave-assisted, supercritical fluid, and UAE can have great practical benefits in decreasing extraction time and solvent consumption, as well as potentially increasing yields and extract quality (Wang and Weller, 2006). The basic mechanism of UAE is the generation of acoustic expansion/compression cycles (i.e., sound waves) in the extraction mixture, which create bubbles and points of negative pressure (i.e., cavitations). The collapse of bubbles at the solid boundaries of sample particles causes high-speed jets of liquid that impact the surface. Thus, ultrasound facilitates more efficient solvent penetration into the sample matrix and improves mass transfer. It can also act to disrupt cell walls and thereby release cellular contents (Wang and Weller, 2006).

The use of UAE in natural products and nutraceuticals research and industrial extractions has expanded during the past decade (Júnior et al., 2006, Vilkhu et al., 2008), but it has hitherto | Development and validation of analytical methodology 49 been rarely applied to peanuts (Table 2.11). In one study, sonication and microwave-assisted sonication were found to improve the extraction of resveratrol and isoflavones (i.e., daidzein, genistein, and biochanin A) from peanut kernels, compared to stirring or soxhlet reflux (Chukwumah et al., 2007b). The authors subsequently published an UAE optimisation study, in which different frequencies, durations, and sample/solvent ratios were tested (Chukwumah et al., 2008). UAE has been found to improve phytochemical extraction in other edible seed crops compared to conventional extraction, dependent on the solvent and UAE conditions. Examples include the UAE of isoflavones (Rostagno et al., 2003), phenolics, and antioxidants (Chung et al., 2010) from soybean; and phenolics, flavonoids, and antioxidants from grape seed/skin waste (Casazza et al., 2010).

The generic UAE method (Prior et al., 2003) that was used to generate crude peanut extracts in the genotype and G × E studies of antioxidant capacity was found to be convenient, repeatable, and adequate for our purposes. Therefore only minor optimisation experiments were performed to ascertain the optimal sonication conditions for peanut samples; besides these, methods from peanut-specific publications were surveyed to inform the development of our extraction protocol (Table 2.12). For example, n-hexane was used for defatting based on our previous experience with peanut extraction (Yusnawan et al., 2012), while 80-90% aqueous ethanol, methanol, and acetonitrile were the solvents most commonly employed for solid-liquid extractions of peanut in the literature. A sample/solvent ratio of 1:5 (w/v) has been sufficiently validated by previous research, being among the most commonly used, with 1:6 recommended in one optimisation study (Chukwumah et al., 2008).

Aside from sample characteristics and solvent selection, the critical operating parameters for UAE include the frequency, pressure, temperature, and duration of sonication. Ultrasound frequency can have a large effect on extraction yield and kinetics, whereas the effect of temperature in the range 40-66 °C is expected to be negligible (Wang and Weller, 2006). The ultrasound bath in this research was a simple model without capacities for adjusting frequency or pressure. It may be of interest to investigate optimisation of these parameters for industrial extraction, but it was not necessary for our purposes.

Experiments in which sonication temperature and duration were varied indicated that there were statistically-negligible (P>0.05) benefits in increasing the temperature above 40 °C and duration beyond 5 min (Table 2.13). Indeed, only temperature influenced (P<0.05) extraction efficiency (Table 2.14). Thus, the final UAE conditions were: 50 kHz, 40 °C, 5 min, 1:5 (w/v) sample/solvent (90% aqueous methanol). The UAE was repeated three times to ensure repeatable effective extraction. | Development and validation of analytical methodology 50

TABLE 2.11 REVIEW OF OPTIMISATION/COMPARATIVE STUDIES OF PHYTOCHEMICAL EXTRACTION FROM PEANUTS

Sample Method Solvent Duration; Sample/ Analytes Observations Reference temp. solvent

Raw kernels Stirring 80% aq. ethanol 2 h, ambient 1:10 Genistein, daidzein, Microwave-assisted (Chukwumah with testa; biochanin A, and sonication and soxhlet et al., 2007b) hexane- Sonication (25, 40, 80% aq. ethanol 2 h, 1:10 t-resveratrol by HPLC. reflux extracted maximum and 80 kHz) uncontrolled defatted and phytochemicals.

non-defatted Soxtec in 80% 80% aq. ethanol 30 min boiling; 1:10 Defatted kernels had more

| Development and validation of analytical methodology ethanol 1 h rinsing extractable phytochemicals Microwave-assisted 80% aq. ethanol 2 h; 70 °C 1:10 than non-defatted. (700 W) sonication (25, 40, and 80 kHz) Raw kernels Sonication (multi- 80% aq. ethanol 2 h 1:2, 1:4, 1:6, Genistein, daidzein, 1:6 was the best (Chukwumah with testa frequency 1:8, 1:10, biochanin A, and sample/solvent ratio et al., 2008) combinations of 25, 1:12, 1:14, t-resveratrol by HPLC. overall. 40, and 80 kHz) and 1:16 . Sonication efficiency at different frequencies varied with analyte. Multiple extractions better than once only. Legume seeds Shaking (300 rpm) 50% aq. acetone, 80% aq. 3 h; ambient 1:10 FC total phenols; total Different phenolic (Xu and Chang, acetone, 70:29.5:0.5 then left flavonoids; condensed components were 2007) acetone/water/ acetic acid, 70% overnight tannin content; DPPH• extracted more efficiently (Used by Dean aq. ethanol, 70% aq. methanol, scavenging; FRAP; and with different solvent et al., 2008) and absolute ethanol. ORAC. systems. 70% ethanol gave maximum ORAC in all legumes.

51

TABLE 2.11 CONTINUED

Sample Method Solvent Duration; Sample/ Analytes Observations Reference temp. solvent

Testa; hexane- Maceration Absolute acetone, ethanol, 24 h; ambient 1:10 Dry matter yield; FC Methanol and ethanol were (Nepote et al., defatted and methanol, ethanol, or water total phenolics; DPPH- the most effective 2002) non-defatted radical scavenging; and extraction solvents. peroxide value. Testa Maceration 0, 30, 50, 70, and 96% aq. 1, 10, 30, 60, 1:20, 1:30, Yield; FC total phenolics; 70% ethanol; non-crushed (Nepote et al., | Development and validation of analytical methodology ethanol and 120 min; 1:40, 1:50, and DPPH-radical testa; 1:20 sample/solvent; 2005) and 1:60 scavenging. 10 min shaking; 3 times 1-5 times (Used by gave maximum yield. Different Francisco and particle sizes Resurreccion, 2009a, Francisco and Resurreccion, 2009b) Testa Shaking 0, 30, 60, and 90% aq. methanol 10, 20, or 30 1:20 FC total phenolics; 30.8% ethanol; 12 min; 30.9 (Ballard et al., or ethanol min; 30, 45, or ORAC; and t-resveratrol °C maximised FC 2009) 60 °C by HPLC. 30% methanol; 30 min; Response surface 52.9°C maximised ORAC. analysis used to assess optimum conditions. Resveratrol detected in methanol but not ethanol extracts.

52

TABLE 2.12 REVIEW OF PHYTOCHEMICAL EXTRACTION METHODS APPLIED TO PEANUTS

Sample Method Solvent Duration; temp. Sample/ Analytes Reference solvent

Raw and roasted UAE 80% aq. methanol 2 h; ambient 1:6 Total flavonoids, FC total (Chukwumah et al., 2007a) kernels; defatted phenolics; HPLC; and MS. Roasted kernels Accelerated 1:1 hexane/dichloromethane 1500 psi; 70 °C (lipophilic) 1:22 H- and L-ORAC; FC total (Davis et al., 2010) without testa solvent extraction (lipophilic) and 70:29.5:0.5 and 80 °C (hydrophilic); 3 phenols; and tocopherols by acetone/water/acetic acid 5 min cycles HPLC. | Development and validation of analytical methodology (hydrophilic) Peanut butter Solid-liquid 80% aq. ethanol 30 min; ambient 1:10 t-resveratrol and t-piceid by (Ibern-Goméz et al., 2000) extraction HPLC. Roasted kernels Homogeniser 90% aq. acetonitrile 2 min; ambient 1:10 t-resveratrol by HPLC. (Lee et al., 2004) and peanut butter Imbibed kernels Homogeniser 80% aq. ethanol 2 min; 27000 rpm; on ice 1:3 t-resveratrol, t-piceid, and (Potrebko and Resurreccion, phenolphthalein by HPLC. 2009) (Used by Sales and Resurreccion, 2009) Imbibed kernels Wrist action Methanol 12 h; ambient 1:10 Thiocyanate antioxidant (Rudolf and Resurreccion, shaker capacity; FC total phenolics; 2005) and t-resveratrol by HPLC. Roasted kernels Homogeniser 80% aq. ethanol 2 min; on ice 1:3 t-resveratrol and (Sanders et al., 2000) without testa c-resveratrol by HPLC. (Used by Rudolf and Resurreccion, 2007, Rudolf et al., 2005, Tokuşoĝlu et al., 2005, Wang and Pittman, 2008) Imbibed kernels Blender 90% aq. acetonitrile 2-3 min; ambient 1:4 t-resveratrol and other (Sobolev et al., 1995) stilbenes by HPLC. 53

TABLE 2.12 CONTINUED

Sample Method Solvent Duration; temp. Sample/ Analytes Reference solvent Roasted and Blender with 90% aq. acetonitrile 1 min; ambient 1:4 t-resveratrol by HPLC. (Sobolev and Cole, 1999) boiled peanut; diatomaceous peanut butter earth followed by mini-column clean-up

| Development and validation of analytical methodology Raw kernels Blender 90% aq. acetonitrile 1 min; ambient 1:3 t-resveratrol and other (Sobolev et al., 2007) stilbenes by HPLC. Imbibed kernels Blender Methanol 1 min; ambient 1:5 t-resveratrol and other (Sobolev et al., 2009, Sobolev stilbenes by HPLC. et al., 2010) Roasted kernels Homogeniser 80% aq. methanol 1:4 FC total phenolics; ORAC; (Talcott et al., 2005a) without testa and phenolics by HPLC. Raw and roasted Homogeniser 80% aq. methanol 24h; ambient 1:2 FC total phenolics; ORAC; (Talcott et al., 2005b) kernels without and phenolics by HPLC. (Used by Sales and testa; imbibed Resurreccion, 2010a, Sales and kernels Resurreccion, 2010b) Roots Homogeniser 80% aq. methanol Homogenised (15000 1:10 t-resveratrol by HPLC. (Chen et al., 2002) rpm) 1 min then 30 min; (Used by Chang et al., 2006, 70°C Wang et al., 2005) Roots UAE Ethyl acetate 30s; ambient 1:20 t-resveratrol and other (Condori et al., 2010) stilbenes by HPLC and MS. Flowers Blender Methanol 2 min; ambient 1:5 Spermidine and flavonoid (Sobolev et al., 2008) conjugates by HPLC. 54

TABLE 2.13 EFFECT OF TEMPERATURE AND DURATION ON UAE EFFICIENCY

Temperature Duration N ORAC, TE (μmol g-1) FC, GAE (μmol g-1)

(°C) (min) Mean ± SEM Mean ± SEM

25 5 2 34.10 ± 1.00 9.68 ± 0.59 15 2 33.85 ± 1.95 9.89 ± 0.52 30 2 35.05 ± 0.95 9.19 ± 0.31 60 2 36.55 ± 1.25 10.16 ± 1.20 40 5 2 41.95 ± 0.35 11.27 ± 0.45 15 2 41.85 ± 2.05 11.64 ± 0.29 30 2 43.75 ± 0.05 11.67 ± 1.25 60 2 41.30 ± 1.80 12.94 ± 0.48 60 5 2 43.15 ± 2.55 12.56 ± 2.00 15 2 42.65 ± 3.55 12.80 ± 2.30 30 2 44.95 ± 4.85 13.24 ± 1.72 60 2 44.40 ± 2.40 11.21 ± 0.08

TABLE 2.14 EFFECT OF TEMPERATURE ON UAE EFFICIENCY

Temperature N ORAC, TE (μmol g-1) FC, GAE (μmol g-1) (°C) Mean ± SEM Mean ± SEM 25 8 34.89 ± 0.65 b 9.73 ± 0.31 b 40 8 42.21 ± 0.63 a 11.88 ± 0.36 ab 60 8 43.79 ± 1.36 a 12.45 ± 0.72 a

2.3.3.1.2 Hydrolysis and enzyme treatments Antioxidant phytochemicals occur in free and soluble conjugated forms. But insoluble bound forms are expected to be predominant in food and plant matrices, with implications for bioaccessibility and bioavailability. Hydrolysates of plant extracts contain many times higher concentrations of phenolic acids than native extracts (Kim et al., 2006b). The alkaline (10 M sodium hydroxide agitated overnight) and acid hydrolysis (concentrated hydrochloric acid, 85 °C, 1 h) methods described by Matilla and Kumpulainen (2002) were investigated, as well as the effects of different temperatures (i.e., 20, 40, 60, and 80 °C for 30 min) and durations (i.e., 15, 30, 60, and 120 min at 80 °C) of acid hydrolysis on several antioxidant standards (i.e., p-coumaric acid, polydatin, t-resveratrol, genistein, rutin, quercetin, and biochanin A).

The HPLC chromatograms of hydrolysates were vastly altered compared to the native (non- treated) extracts (Figure 2.11), but did not reveal expected hydrolysis reactions, e.g., polydatin → resveratrol, rutin → quercetin, and biochanin A → genistein. Yet, the tested antioxidant standards were apparently stable when incubated for 30 min at temperatures up to 80 °C (Figure 2.10). As it was clear that extensive experiments were required to characterise,

| Development and validation of analytical methodology 55 optimise, and validate the hydrolysis reactions, we decided to focus on the analysis of native extracts and leave analysis of hydrolysates for future work.

(A) 150000 Control: p-coumaric acid

HCl-treated: p-coumaric acid

Control: t-resveratrol 100000 HCl-treated: t-resveratrol

Control: polydatin

Peak area HCl-treated: polydatin 50000

0 20 40 60 80 Temperature (qC) B 150000 ( ) Control: quercetin

Control: genistein

HCl-treated: genistein 100000 Control: biochanin A

HCl-treated: biochanin A

Peak area 50000

0 20 40 60 80 Temperature (qC) FIGURE 2.10 THERMAL STABILITY (CHANGE IN HPLC PEAK AREA) AFTER 30 MIN INCUBATION OF CONTROL AND HCL-TREATED (A) P-COUMARIC ACID, T-RESVERATROL, AND POLYDATIN; AND (B) QUERCETIN, GENISTEIN, AND BIOCHANIN A STANDARDS.

100000 Native extract Hy droly s ate (HCl, 80qC, 30 min) 80000

60000

40000 Intensity

20000

0 0 5 10 15 20 25 Retention time (min)

FIGURE 2.11 CHANGE IN HPLC CHROMATOGRAM OF CRUDE PEANUT EXTRACT FOLLOWING ACID HYDROLYSIS (CONC. HCL, 80 °C, 30 MIN).

| Development and validation of analytical methodology 56

A number of researchers have reported that enzyme treatments improved the extraction of phenolic compounds from samples including peanut (Govindaraju and Srinivas, 2006, Wang et al., 2008b) and other oilseeds (Domínguez et al., 1994, Sengupta and Bhattacharyya, 1996, Vuorela et al., 2003, Zhang et al., 2007a). Enzymes have included various protease, esterase, glucanase, glucosidase, cellulase, pectinase, and multi-component preparations. The reactions may but do not necessarily involve hydrolysis. In contrast to the chemical hydrolysis treatments described earlier, the benefits of enzyme treatment lay not so much in its specific effects on analytes, but in the chemical and thermal degradation and resultant extractability of the sample matrix. Enzymes such as glucanases and pectinases may benefit extraction by degrading cell walls, thereby rendering intracellular contents more accessible to the solvent (Li et al., 2006).

TABLE 2.15 CORRELATIONS AND DIFFERENCES BETWEEN NATIVE AND ENZYME-TREATED PEANUT EXTRACTS

Analyte Method Paired samples T tests (enzyme-treated – native) SPE (%) Detector Setting (nm) N Correlation Sig. Difference Sig.

Caffeic/vanillic acidA 5 PDA 320 12 0.407 0.189 -45.5** 0.001 p-coumaric acid 5 PDA 320 12 0.319 0.312 97.3** <0.001 Ferulic/sinapic acidA 5 PDA 320 12 0.594* 0.042 43.7** <0.001 OCUB 5 PDA 320 12 0.541 0.069 -1.3 0.299 Salicylic acid 5 PDA 305 12 0.588* 0.044 6.3** <0.001 5 RF 260/422 9 0.109 0.781 11.3 0.053 20 PDA 305 14 0.670** 0.009 6.0** <0.001 20 RF 260/422 11 0.246 0.466 14.6** 0.002 Resveratrol 20 PDA 305 14 0.747** 0.002 2.5** <0.001 20 RF 260/422 14 0.384 0.175 3.0** <0.001 Daidzein 20 PDA 305 14 0.546* 0.043 9.5** <0.001 QCUC 20 PDA 370 14 -0.248 0.393 0.3 0.685 * Significant (P<0.05) ** Highly significant (P<0.01) A Co-eluting analytes B Unidentified compound initially thought to be o-coumaric acid and therefore termed o-coumaric acid-co-eluting unknown (OCU) C Unidentified compound initially thought to be quercetin and therefore termed quercetin-co-eluting unknown (QCU)

The application of protease and cellulase preparations increased (P<0.05) the concentrations of extractable p-coumaric acid, ferulic acid (and/or co-eluting peaks), salicylic acid, resveratrol, and daidzein (Table 2.15), presumably by enhancing the extractability of cell wall- and protein- bound compounds. However caffeic acid (and/or co-eluting peak) concentrations apparently

| Development and validation of analytical methodology 57 decreased with enzyme treatment. There were few correlations between concentrations in native and enzyme-treated extracts, aside for o-coumaric acid, salicylic acid, and resveratrol. This means that the native and enzyme-treated extracts provided distinct information regarding sample composition. Genotypic differences were therefore analysed separately for native and enzyme-treated extracts.

2.3.3.1.3 Solid phase extraction

TABLE 2.16 SORBENTS INVESTIGATED DURING SPE METHOD DEVELOPMENT

Sorbent Functional Selectivity Sorbent Surface area groupA type (m2 g-1)

Strata C18-U C18 sorbent with no end-capping, giving the Silica- 500 phase moderate hydrophobic selectivity based with slight polar selectivity due to the active silanol groups.

Strata-X A reversed phase functionalised polymeric Polymer- 800 polymeric sorbent that gives strong retention of based neutral, acidic, or basic compounds under aggressive, high organic wash conditions.

A Images and descriptions from www.phenomenex.com.au

C18 and polymeric reversed-phase sorbents (Table 2.16) have a similar moderately hydrophobic retention, and both achieved adequate recovery of tested standard compounds. The Phenomenex Strata-X polymeric sorbent was selected for use due to its apparent versatility and robustness over a wide range of pH and loading conditions, and higher surface area associated and hence loading capacity. Several aspects of the SPE loading and elution were optimised to ensure maximum and repeatable recovery of analytes.

Despite the manufacturer’s claims, the acidity and organic content of the SPE conditions was found to be critical to repeatable retention of the more polar phenolic acids. In particular, acidifying (pH 2) the loading solution was necessary to attain repeatable retention of p- hydroxybenzoic acid and protocatechuic acid (Table 2.17). The manufacturer recommended loading in 5-50% polar organic solvent depending on the analyte, but there was diminished recovery of these more weakly-retained phenolic acids even when loading in 10% methanol.

On the other hand, stilbenes and flavonoids have poor aqueous solubility. While troubleshooting problems with analyte recovery, daidzein and quercetin was found to precipitate out of solution when chilled overnight, even in 20% aqueous methanol (Table 2.18).

It was therefore expedient to use two SPE methods: 5% methanol loading solution to optimise recovery of phenolic acids and 20% methanol loading solution for resveratrol and flavonoids. | Development and validation of analytical methodology 58

Paired samples T tests of extracts prepared in 5% and 20% methanol loading solutions indicated that the two methods produced little difference (P>0.05) in the quantitation of p-coumaric, ferulic, o-coumaric, and salicylic acids in peanut samples (Table 2.19).

TABLE 2.17 EFFECT OF PH AND ORGANIC SOLVENT CONTENT OF THE LOADING SOLUTION ON SPE RECOVERY

Analyte Recovery (%) in different SPE loading conditions pH 2 pH 4 2% 10% 2% 10%

Protocatechuic acid 76 55 0 0 p-Hydroxybenzoic acid 85 0 0 0 Vanillic acid 80 62 51 54 Caffeic acid 82 64 55 61 Syringic acid 82 64 55 59 p-coumaric acid 81 64 58 72 Ferulic acid 81 64 57 67 m-coumaric acid 81 64 58 71 Sinapic acid 82 64 59 73 Polydatin 83 64 59 73 o-coumaric acid 81 64 58 71 Resveratrol 75 63 56 62 Cinnamic acid 80 64 57 68 Rutin 83 65 59 73 Daidzein 80 64 58 67 Quercetin 66 59 53 56 Genistein 75 63 56 61 Formononetin 75 63 56 61 Biochanin A 60 56 51 52

TABLE 2.18 EFFECT OF STORAGE CONDITIONS ON SOLUBILITY OF PHYTOCHEMICAL STANDARDS

Analyte Recovery (%) after overnight storage in: Freezer (0 °C) Fridge (4 °C) 20% methanol 80% methanol 20% methanol 80% methanol Caffeic acid 100 101 102 102 p-coumaric acid 100 101 102 102 Polydatin 107 100 109 101 Ferulic acid 103 101 104 102 Resveratrol 98 101 100 102 Daidzein 14 102 25 102 Quercetin 0 100 0 99

| Development and validation of analytical methodology 59

Salicylic acid was accurately measured by both methods (95 ± 18% cf. 99 ± 17%); however, validation statistics for accuracy (Table 2.25) suggested that the 5% SPE method produced less measurement variability than the 20% method for p-coumaric (90 ± 21% cf. 99 ± 38%), ferulic (92 ± 36% cf. 98 ± 49%), and o-coumaric acids (82 ± 23% cf. 76 ± 37%), although average recovery was acceptable by either method.

In contrast, the recovery of resveratrol (47 ± 16% cf. 91 ± 18%), daidzein (73 ± 20% cf. 103 ± 19%), and especially quercetin (25 ± 21% cf. 68 ± 28%) and kaempferol (28 ± 13% cf. 94 ± 61%) were dramatically diminished when the 5% SPE method was used, resulting in much lower quantitation (P<0.05) of these analytes in sample extracts compared to SPE based on 20% methanol (Table 2.19).

Various eluents including acetonitrile, methanol, acidified methanol (pH 2), and 50% (v/v) acetonitrile/methanol all gave adequate recovery, so methanol was used for simplicity. A single aliquot of 2 mL solvent was sufficient to elute analytes, but 3 mL methanol in two aliquots (2 mL followed by 1 mL) was used in order to confidently elute all retained analytes.

TABLE 2.19 CORRELATIONS AND DIFFERENCES BETWEEN 5% AND 20% SPE PEANUT EXTRACTS

Analyte Method Paired samples T tests (5% - 20% SPE) Detector Setting (nm) N Correlation Sig. Difference Sig.

Caffeic/vanillic acidA PDA 320 72 0.191 0.108 12.4* 0.010 p-coumaric acid PDA 320 72 0.946** <0.001 2.2 0.241 Ferulic/sinapic acidA PDA 320 72 0.868** <0.001 1.0 0.499 OCUB PDA 320 63 0.757** <0.001 -0.6 0.242 Salicylic acid PDA 305 73 0.834** <0.001 -0.2 0.573 RF 260/422 55 0.702** <0.001 0.4 0.683 Resveratrol PDA 305 73 0.750** <0.001 -1.5** <0.001 RF 260/422 73 0.663** <0.001 -1.1** 0.006 Daidzein PDA 305 73 0.565** <0.001 -1.2* 0.041 QCUC PDA 370 73 0.369** 0.001 -1.9** <0.001 Kaempferol PDA 370 73 0.379** 0.001 -2.1** 0.001 * Significant (P<0.05) ** Highly significant (P<0.01) A Co-eluting analytes B Unidentified compound initially thought to be o-coumaric acid and therefore termed o-coumaric acid-co-eluting unknown (OCU) C Unidentified compound initially thought to be quercetin and therefore termed quercetin-co-eluting unknown (QCU)

| Development and validation of analytical methodology 60

2.3.3.2 Development of HPLC method HPLC analysis of peanut kernels has most frequently targeted the quantification of resveratrol by UV detection predominantly using typical reversed phase conditions, i.e., C18 columns in aqueous-organic mobile phases (Table 2.20). Less commonly, normal phase conditions have been used for measurement of tocopherols in peanut oils or lipophilic extracts (Davis et al., 2010, Dean et al., 2009, Hashim et al., 1993a, Hashim et al., 1993b) and in one publication for resveratrol quantification (Sobolev and Cole, 1999). Few articles broach more than one class of phenolic compound let alone the simultaneous analysis of many analytes in peanut extracts. These include the identification of an extensive range of phenolic compounds in peanut plant tissues by HPLC-time-of-flight (TOF)-MS (Dean et al., 2008); the simultaneous quantitative analysis of 15 phenolic acids, flavonoids, and stilbenes in peanut testa (Francisco and Resurreccion, 2009a); and HPLC-electrospray ionisation (ESI)-MS analysis of various procyanidins in peanut testa (Yu et al., 2006).

In peanut kernel tissues specifically, one research group used HPLC-PDA to quantify phenolic acids and their derivatives, though without further characterisation these were only tentatively identified (Duncan et al., 2006, Talcott et al., 2005a, Talcott et al., 2005b). Chukwumah et al. (2007a) and Singleton et al. (2002) identified a range of flavonoids such as daidzein, genistein, biochanin A, rhamnetin, and their glycosides in peanut kernels by HPLC-UV with the aid of MS techniques; whereas Wang et al. (2007) quantified quercetin by HPLC-PDA, and identified daidzein, genistein, kaempferol, and myricetin somewhat more vaguely as occurring in ‘low’ or ‘trace’ amounts.

| Development and validation of analytical methodology 61

TABLE 2.20 REVIEW OF METHODS FOR HPLC ANALYSIS OF PEANUT EXTRACTS

Sample Column Mobile phase Gradient (min) Flow Inj. Detector: analytes Reference rate (mL vol. A B Time, %B Total min-1) (μL)

Aq. ethanol extracts of Hypersil ODS (250 50% aq. acetonitrile Isocratic 20 UV (335 nm): stilbenes (Aguamah et al., 1981) imbibed peanut kernels × 4.6 mm) Aq. acetone extracts of Spherisorb 10 ODS 1% aq. acetic Acetonitrile 0, 30; 3, 35; 6, 27 1.5 UV (338 nm): resveratrol (Arora and Strange, 1991) imbibed peanut (250 × 4.6 mm) acid 35; 9, 40; 12, and other stilbenes | Development and validation of analytical methodology cotyledons, pegs, pods, 40; 15, 50; 17, and kernels 50 Aq. methanol extracts of Thermo Hypersil 25% acetonitrile in 0.5% aq. Isocratic 30 1 PDA (307 nm): t- and (Burns et al., 2002) plant foods inc. roasted ODS (5 μm; 250 × formic acid c-resveratrol peanut and peanut butter 4.6 mm) Methanol extracts of Thermo Hypersil Water Methanol 0, 50; 22,100; 30 1 PDA (254 nm): (Chang et al., 2006) imbibed peanut kernels ODS (5 μm; 250 × 25, 100; 27, 50 t-resveratrol and other 4.6 mm) stilbenes Aq. methanol extracts of Agilent Zorbax SB 0.1% aq. 80% 0, 22; 10, 22; 35 1 50 PDA: daidzein (254 nm), (Chukwumah et al., 2007a) raw, boiled, and roasted C18 (250 × 4.6 mm) formic acid acetonitrile 30, 100 other isoflavones (260 peanut kernels in acetic acid nm), and resveratrol (310 nm) APCI-MS Aq. ethanol extracts of Agilent Zorbax SB Aq. acetic 80% 0, 22; 10, 22; 40 1.5 50 PDA: isoflavones (285 nm) (Chukwumah et al., 2008, raw peanut kernels C18 (250 × 4.6 acid (pH 2.4) acetonitrile 25, 100 and t-resveratrol (306 nm) Chukwumah et al., 2007b) mm) in A based on (Lamuela- Raventós et al., 1995) Aq. acetonitrile extracts of Spherisorb 10 ODS 1% aq. acetic Acetonitrile 0, 40; 7, 45; 20 1.5 10 UV (310 nm): resveratrol (Cooksey et al., 1988) imbibed peanut kernels (250 × 4.6 mm) acid 12, 45; 20, 65

Aq. methanol extracts of Bondpak C18 (10 1% aq. acetic Methanol 0, 10; 10, 35 10 2 UV (254 nm): flavonoids (Daigle et al., 1983) raw peanut kernels μm; 300 × 3.9 mm) acid 62

TABLE 2.20 CONTINUED

Sample Column Mobile phase Gradient (min) Flow Inj. Detector: analytes Reference rate (mL vol. A B Time, %B Total min-1) (μL)

Aq. methanol extracts of Bondpak C18 (10 1% aq. acetic Methanol 0, 10; 10, 35 10 2 UV (254 nm): flavonoids (Daigle et al., 1983) raw peanut kernels μm; 300 × 3.9 mm) acid 48% methanol in 1% aq. Isocratic 10 2 UV (254 nm) flavonoid acetic acid aglycones inc. quercetin, | Development and validation of analytical methodology rhamnetin, and isorhamnetin

Aq. methanol extracts of Bondpak C18 (10 42% methanol in 1% aq. Isocratic 1.9 UV (254 nm): 5,7- (Daigle et al., 1985) raw peanut kernels μm; 300 × 3.9 mm) acetic acid dihydroxyisoflavone Mechanically pressed oils Luna (5μm; 250 × 1% isopropanol in hexane Isocratic 1.2 20 UV (294 nm): tocopherols (Davis et al., 2010, Dean et and lipophilic extracts of 4.6 mm) al., 2009) raw and roasted peanut kernels Aq. and aq. methanol Waters Spherisorb 2% aq. acetic 30 % 0, 0; 20, 30; 70 0.8 PDA: polyphenolic (Duncan et al., 2006, extracts of roasted ODS-2 (250 × acid acetonitrile 30, 50; 50, 70; compounds inc. Talcott et al., 2005a, peanuts 4.6mm) in A 55, 100 p-coumaric acid Talcott et al., 2005b) based on (Talcott et al., 2000) Hexane extracts of raw LiChrosorb Si60 (5 1% 2-propanol in hexane Isocratic 1 100 RF (290/330 nm): (Hashim et al., 1993a, peanut kernels μm; 250 × 4 mm) tocopherols Hashim et al., 1993b) Aq. ethanol extracts of Nucleosil 120 C18 5.8% aq. 80% 0, 16.5; 13, 18; 30 1.5 100 PDA: t-resveratrol and (Ibern-Goméz et al., 2000) peanut butter (5 μm; 250 × 4.6 acetic acid acetonitrile 17, 23; 21, 25; piceid (306 nm); and based on (Romero-Pérez mm) ; 40 °C in A 27, 31.5 c-resveratrol and piceid et al., 1999) (285 nm)

Aq. acetonitrile extracts Nucleosil 100-5 C18 40% aq. acetonitrile Isocratic 20 0.3 20 UV: t-resveratrol (306 nm) (Lee et al., 2004) of raw and roasted (5 μm; 250 × 4.6 peanut kernels, and mm)

63 peanut butter

TABLE 2.20 CONTINUED

Sample Column Mobile phase Gradient (min) Flow Inj. Detector: analytes Reference rate (mL vol. A B Time, %B Total min-1) (μL)

Aq. ethanol extracts of Alltech 0.1% aq. Acetonitrile 0, 5; 23, 41.8; 34 1.5 15 PDA: t-resveratrol (Rudolf et al., 2005)A raw peanut kernels Econosphere C18 (5 acetic acid 28, 77; 29, 5 μm; 250 × 4.6 mm); 25 °C

| Development and validation of Aq. ethanol extracts of Agilent Eclipse Plus Water Acetonitrile 0, 5; 7, 22; 13, 34 1.5 40 PDA: t-resveratrol and (Sales and Resurreccion, imbibed peanut kernels C18 (5 μm; 250 × 23; 26, 63; 28, piceid (307 nm) 2009) 4.6 mm); 25 °C 80; 29, 5

Aq. ethanol extracts of Vydac C18 (150 × 0.1 % aq. Acetonitrile 0, 0; 1, 0; 3, 38 25 UV: t- and c-resveratrol (Sanders et al., 2000) roasted peanut kernels 4.5 mm) TFA 15; 23, 27; 28, (308 nm) 100

Methanol extracts of raw Supelco C18 (250 × 1 mM aq. Acetonitrile 0, 99; 50, 50 50 1 UV (262 nm) and ESI-MS: (Singleton et al., 2002) peanut kernels 4.6 mm) ammonium isoflavones acetate FABMS: flavonoid glycosides Aq. acetonitrile extracts of MAC-MOD Zorbax 1050:270:17:5:1 (v/v) n- Isocratic 20 1.5 20- PDA: t-resveratrol (307 or (Sobolev and Cole, 1999) analytical methodology boiled and roasted peanut Rx-SIL (5 μm; 250 × hexane/2- 120 320 nm) kernels, and peanut 4.6 mm) propanol/water/acetonitrile/ butter acetic acid Aq. ethanol extracts of Hypersil ODS (5 40% acetonitrile in 0.1% TFA Isocratic 15 1 10 PDA: t- and c-resveratrol (Tokuşoĝlu et al., 2005) raw peanut and pistachio μm; 250 × 4.6 mm) (308 nm) kernels Aq. methanol extracts of Thermo Hypersil Water Methanol 0, 30; 16, 90 18 1 20 UV: t-resveratrol (307 nm) (Wang et al., 2005) germinated peanut ODS (5 μm; 250 × kernels 4.6 mm)

Aq. methanol extracts of C18 (5 μm; 150 × Aq. formic Acetonitrile 2, 20; 20, 30; 33 2 PDA: flavonoids (Wang et al., 2007) 64 raw peanut kernels 4.6 mm); 40 °C acid (pH 2.5) 22, 100; 24, 100; 26, 20

TABLE 2.20 CONTINUED

Sample Column Mobile phase Gradient (min) Flow Inj. Detector: analytes Reference rate (mL vol. A B Time, %B Total min-1) (μL)

Aq. ethanol extracts of Agilent C18 (5 μm; Aq. formic Acetonitrile 0, 10; 2, 10; 26 1.5 30 PDA: t-resveratrol (310 (Wang and Pittman, 2008) raw peanut kernels 150 × 4.6 mm) acid (pH 2.5) 10, 30; 11, 30; nm) 17, 95; 26, 10 Aq. alcohol extracts of Phenomenex Luna 0.5% aq. 0.5% acetic 0, 0; 4, 4.7; 42, 65 0.8 PDA (280 nm): resveratrol (Ballard et al., 2009) | Development and validation of analytical methodology peanut testa C18 (5 μm; 250 × acetic acid acid in 74.7; 54.5, 95; 4.6 mm) methanol 55, 0; 65, 0 Aq. acetone extracts of Restex Ultra 0.1% aq. 50:50 (v/v) 0, 10; 30, 95; 40 0.4 5 TOF-MS: wide range of (Dean et al., 2008) peanut leaves, roots, and Aqueous C18 (100 × formic acid acetonitrile/ 35, 95;40, 5 phenolic compounds shells 2.1 mm) methanol Aq. ethanol extracts of Agilent Eclipse Plus 0.1% aq. 0.1% formic 0, 5; 7, 7; 75, 124 1.5 20 PDA: benzoic acids (250 (Francisco and A peanut testa C18 (5 μm; 250 × formic acid acid in 17; 110, 45; nm), flavanols (280 nm), Resurreccion, 2009a) 4.6 mm) acetonitrile 117, 100; 124, stilbenes (306 nm), 5 cinnamic acids (320 nm), and flavonols (370 nm) Hexane, ethyl acetate, Thermo Hypersil 75% acetonitrile in 0.1% aq. Isocratic 20 1 28 ECD (0.75 V): resveratrol (Ha et al., 2007) and methanol extracts of ODS (5 μm; 200 × phosphoric acid and 5 mM peanut testa 4.6 mm) NaCl 92% aq. methanol Isocratic 20 1 25 ECD: tocopherols

Aq. alcohol extracts of Nucleosil C18 (5 5% 80% 0, 10; 10, 20; 45 UV: phenolic acids, (Yu et al., 2005) peanut testa μm; 150 × 4.6 mm) acetonitrile acetonitrile 30, 50; 35, 50; flavonoids, and resveratrol in 0.035% in 0.025% 40, 10 (280 nm) TFA TFA

Aq. ethanol extracts of Nucleosil C18 (5 0.1% aq. 0.1% formic 0, 12; 15, 15; 50 0.8 UV (280 nm): catechins (Yu et al., 2006) peanut testa μm; 250 × 4.6 mm) formic acid acid in 25, 20; 30, 20; and procyanidins acetonitrile 40, 35; 45, 35; ESI-MS

65 50, 12

A Publication includes details of method optimisation and validation

HPLC method development was approached with the goal of simultaneously analysing the large range of phenolic acids (Dabrowski and Sosulski, 1984), stilbenes (Sobolev et al., 2011), and flavonoids (Chukwumah et al., 2007a, Singleton et al., 2002) that have been identified in peanut kernels. This sort of broad-spectrum simultaneous polyphenolic analysis has rarely been described for food matrices. Sakakibara et al. (2003) published an alluringly titled HPLC-

PDA determination of ‘all polyphenols in vegetables, fruits, and teas’ in a C18 column and phosphate buffer-methanol system, reporting the calibration curves for 100 phytochemical standards; however we note that their method validation was sparsely described and apparently conducted on only 23 of the analytes.

Tsao and Yang (2003) also used a C18 and binary solvent system, in this case acetate buffer (pH 2.55) and acetonitrile, to measure phenolic acids, flavan-3-ols, flavonols, anthocyanins, and dihydrochalcones, especially in apple and other fruit matrices. The method validation statistics for the six representative standards are good, but the selected analytes were chemically very distinct. One suspects that co-elution would render the statistics less attractive in a sample matrix more complex than the spiked juice that was used. Interestingly, the authors seemed to recognise this and suggested the method be employed for calculation of a total phenolic index, whereby peaks are classified into the five major phenolic classes based on their characteristic UV spectra, and reported in standard equivalent concentrations.

Nevertheless, we pursued a broad method for simultaneous analysis of peanut phytochemicals. The aim was to monitor as much of the extract as possible with a view to fractionating and analysing constituent contributions to total antioxidant capacity and potentially other bio-activity. This was an ambitious undertaking because the different chemical classes have somewhat conflicting analytical requirements. In particular, few systems will achieve retention of both highly hydrophilic phenolic acids and more hydrophobic flavonols with practical run times, yet remain able to resolve similar compounds within each chemical class. Our approach was to seek strong retention across analyte classes, optimise resolution as far as possible, and then seek selectivity via different detectors.

Several reversed phase columns with C18 or C18-like functionality were investigated, namely, the Phenomenex Kinetex C18, Gemini C18, Onyx monolithic C18, and Synergi-polar columns

(Table 2.21). The Gemini has a conventional C18 solid phase; the Kinetex is comprised of core- shell particles that permit higher peak capacity and resolution at back-pressures suitable for UHPLC and HPLC; whereas the Onyx has a monolithic solid phase support that generates low backpressure and can thus operate at very high flow rates for improved throughput. The chemistry of the Synergi-polar column also appeared promising as the relatively polar | Development and validation of analytical methodology 66 retention of the C6 chain combined with the aromatic selectivity of the phenyl group seemed appropriate for our analytes.

Exploratory gradients of water and methanol or acetonitrile (15-95%) to elute a mixture of standards confirmed that the columns provided similar retention (Figure 2.12), with the phenolic acids eluting much faster from the Kinetex and Onyx. When peanut extract was applied, the Kinetex did not resolve the most weakly retained phenolic acids, which co-eluted very early even at a low solvent strength (15% B). Several of the phenolic acid standards similarly co-eluted on the Onyx. The best Onyx results were achieved by running at low solvent strength (15% B) for an extended period, which led to long total run times (e.g., 65 min) and high solvent consumption, consequently defeating the major advantages of using a monolithic support.

TABLE 2.21 COLUMNS INVESTIGATED DURING HPLC METHOD DEVELOPMENT

Column Functional Selectivity Particle Pore Surface pH Dimensions groupA size (μm) size (Å) area range (mm) (m2 g-1)

Kinetex C18 with TMS end- 2.6 100 2.0-7.5 50 × 4.6 C18 capping for hydrophobic retention

and methylene selectivity

Gemini C18 C18 with TMS end- 5.0 110 375 1.0-12.0 250 × 4.6 capping for separation

of basic compounds under high pH conditions

Synergi- Polar end-capped, 4.0 80 475 1.5-7.0 150 × 4.6 polar ether-linked phenyl phase providing

separation of polar, aromatic compounds

Onyx C18 bound to a n/a 130 300 2.0-7.5 150 × 4.6 monolithic monolithic silica rod C18 for strong hydrophobic selectivity and fast

separation in dirty or

viscous samples

A images and descriptions from www.phenomenex.com.au

| Development and validation of analytical methodology 67

2 10 06 2 10 06 06 (A) u (B) u (C) 1u10 4 Phas e B: 4 5 Methanol 2 05 4 2 8u10 2 Acetonitrile 06 06 1 1u10 4 1u10 4 2 2 2 1 3 5u10 05 3 3 1 1 4 3 1

Intensity 3 Intensity 05 1 3 Intensity 5u10 5u10 05 5 5 05 5 5 5 3u10

6 6 6 6 | 6 6 0 0 0 Development and validation 0 5 10 15 20 0 5 10 15 0 2 4 6 8 Retention time (min) Retention time (min) Retention time (min) -1 FIGURE 2.12 EXPLORATORY HPLC SEPARATIONS (305 NM) OF REPRESENTATIVE PHYTOCHEMICALS ON (A) GEMINI C18 (1.5 ML MIN ), (B) SYNERGI-POLAR (2 ML -1 -1 MIN ), AND (C) ONYX C18 (2.5 ML MIN ) COLUMNS IN EQUIVALENT ELUTION VOLUMES OF 15-95% PHASE B. 1, caffeic acid; 2 p-coumaric acid; 3, ferulic acid; 4, resveratrol; 5, daidzein; and 6, quercetin

(A) 2u10 06 (B) 2u10 06

06 06 of analytical methodology 2u10 2u10

1u10 06 1u10 06 Intensity Intensity 5u10 05 5u10 05

0 0 0 5 10 15 20 0 5 10 15 Retention time (min) Retention time (min) -1 -1

68 FIGURE 2.13 EXPLORATORY HPLC SEPARATIONS (305 NM) OF SPIKED PEANUT EXTRACT ON (A) GEMINI C18 (1.5 ML MIN ) AND (B) SYNERGI-POLAR (2 ML MIN )

COLUMNS IN EQUIVALENT 15-45% AQUEOUS ACETONITRILE GRADIENTS.

When comparing the Gemini C18 and Synergi-polar columns on equivalent 15-45% acetonitrile gradients (Figure 2.13), the Gemini was able to resolve more of the peaks in the centre of the chromatogram, which corresponded to phenolic acids in experimental elutions of standard solutions, and at the end of the chromatogram, where flavonols were expected to elute. Running the Synergi-polar column with methanol instead of acetonitrile as the organic phase improved the resolution of these peaks, but the Gemini C18 remained superior.

A simple gradient elution using an aqueous and acetonitrile binary mobile phase was developed. There was co-elution at several peaks in the peanut chromatogram, but separation was not improved by altering or segmenting the mobile phase gradient, and peak resolution was optimal using C18 solid phase selectivity. Acidification of the mobile phase with 0.1% formic acid or 0.1% trifluoroacetic acid improved peak shape and repeatability of phenolic acid retention.

We attempted to design a configuration to simultaneously monitor analytes by detectors with distinct selectivity, namely, PDA, RF, and ECD detectors. PDA and RF can be operated similarly under rigorous conditions, but ECD of phytochemicals is susceptible to surface fouling of the electrode by phenolic oxidation products (Scampicchio et al., 2008, Wang and Martinez, 1991). However, it also has the capacity to operate at very sensitive signal thresholds (e.g., pA to μA signals with fmol-level LOD in some applications) such that it has typically been run in isocratic buffers with trace concentrations of analyte/sample. In view of this, various split-flow configurations were explored using several methods to regulate the divergent flow rates.

The simplest design utilised a post-column adjustable split-valve or T-junction with differing tubing dimensions (internal diameter and length) to moderate the flow of mobile phase (Figure 2.14). This sort of split-flow configuration is commonly used to interface HPLC-MS, with the majority of solution going to waste, since the MS mass analyser requires nebulisation of just a small quantity of sample. We also experimented with the use of an external (dual barrel or syringe) pump to introduce a make-up mobile phase to the post-column split (Figure 2.15). Unfortunately, post-column splitting introduced excessive noise into the ECD trace despite our best efforts to control pressure fluctuations through adjustments in the tubing, split valve, or flow rates of the make-up phase.

Finally, the PDA and RF were configured in series with options to run to the ECD or a fraction collector (Figure 2.16). The PDA provided the most sensitive calibrations (Table 2.23) and greatest precision (Table 2.26). This may be partially attributed to its position as the first detector in the serial configuration, where it benefited from the most stable system pressure.

| Development and validation of analytical methodology 69

Even when using theoretically zero-volume unions, passage through additional tubing and connections tends to introduce disturbances to the mobile phase that manifest as noise in the chromatogram. Indeed, the typically highly sensitive RF detector suffered diminished sensitivity (i.e., 10-50-fold higher LOQ than PDA) and precision in the serial configuration to the extent that measurements could not be considered quantitative. We suggest that RF and ECD are best operated as stand-alone detectors, and hope to optimise their application to phytochemical detection in future. For this research, the RF data was used only in support of the quantitative measurements by PDA.

(A) (B)

FIGURE 2.14 EXAMPLES OF TESTED HPLC SYSTEM CONFIGURATIONS USING AN ADJUSTABLE SPLIT- VALVE OR T-JUNCTION TO REGULATE THE MOBILE PHASE (A) IMMEDIATELY POST-COLUMN; AND (B) AFTER PASSING THROUGH THE PDA AND/OR RF DETECTORS.

(A) (B)

FIGURE 2.15 EXAMPLES OF TESTED HPLC SYSTEM CONFIGURATIONS USING AN EXTERNAL PUMP TO INTRODUCE A MAKE-UP PHASE INTO THE ECD SPLIT-FLOW (A) IMMEDIATELY POST-COLUMN; AND (B) AFTER PASSING THROUGH THE PDA AND/OR RF DETECTORS.

FIGURE 2.16 FINAL HPLC-PDA-RF-(FRC/ECD) SYSTEM CONFIGURATION. Calibrations were tested for as many of the phytochemicals that have been reported in peanut kernels as possible. Several standards were observed to co-elute under the HPLC test

| Development and validation of analytical methodology 70 conditions. These were caffeic and vanillic acids at 5.9 min, polydatin and rutin at 9.8 min, and sinapic and ferulic acids at 10.1 min (Table 2.22). It may be theoretically plausible to distinguish the compounds based on spectral characteristics, but it was not so in practice. Even when samples were spiked with standard compounds, characteristic UV spectra could not be distinguished for caffeic or vanillic acids among the peaks eluting at 5.9 min, let alone sinapic and ferulic acids, which were spectrally even more alike.

TABLE 2.22 EXPERIMENTALLY-DERIVED RETENTION TIMES, UV SPECTRA, AND FLUORESCENCE OF TESTED PHYTOCHEMICAL STANDARDS

Analyte Retention UV Absorbance (nm) Fluorescence (nm)

time (min) λmax λmin λex λem Protocatechuic acid 3.5 259.17/293.97 280.10/343.76 p-Hydroxybenzoic acid 5.3 254.79 374.41 260-270 360-380 Caffeic acidA 5.9 323.39 261.75 280-340 430-450 Vanillic acidA 6.0 260.32/291.98 280.69 260-280 360-390 Syringic acid 6.2 274.33 388.48/397.60 270-290 360-400 p-coumaric acid 8.9 309.26 390.8 280-330 400-440 PolydatinB 9.7 317.88/306.36 311.67/255.48/378.42 280-330 380-450 RutinB 9.9 255.59/353.37 281.64 Sinapic acidC 10.1 323.23 262.8 300-340 440-480 Ferulic acidC 10.1 322.57 261.02 280-340 430-480 m-coumaric acid 10.9 277.76 o-coumaric acid 13.1 276.14/324.60 303.99 270-290 520-560 Myricetin 15.1 371.61 284.94 Salicylic acid 15.2 303.01 260.58/370.70 280-300 390-450 t-resveratrol 16.8 305.54/317.00 314.10/255.05 270-340 380-440 Daidzein 18.1 301.15 288.33 280-320 440-540 t-Cinnamic acid 19.8 276.87 370.84 Quercetin 20.6 371.64 284.13 Genistein 24.5 259.67 Kaempferol 26.1 265.62/364.96 281.90

A, B, C Compounds marked with the same letter were co-eluting

For the purpose of sample analysis and HPLC QC, standards were monitored that were either detected in our peanut extracts or had a similar retention time to a peak in the chromatogram. Where there were co-eluting analytes, just one of the compounds was monitored. The selected standard compounds included caffeic acid, p-coumaric acid, ferulic acid, o-coumaric acid, salicylic acid, t-resveratrol, daidzein, quercetin, and kaempferol.

| Development and validation of analytical methodology 71

2.3.3.3 Validation statistics

2.3.3.3.1 Calibration linearity and sensitivity Correlation coefficients (r2) typically greater than 0.999 were obtained for all analyte calibrations by both PDA and RF, indicating consistently good linearity over the specified concentration ranges (Table 2.23). The HPLC method was least sensitive for quercetin and kaempferol, with variable signal response at low concentrations making it necessary to use relatively high calibration minima (5 and 1 μg mL-1, respectively, compared to 0.1 μg mL-1 for other analytes).

TABLE 2.23 LINEARITY OF HPLC-PDA AND HPLC-RF CALIBRATIONS

Detector λ (nm) Analyte No. of points Calibration range Correlation coefficientA, r2 (μg mL-1) Min - Max Mean PDA 320 Caffeic acid 7 0.1 - 100 >0.9900 - >0.9999 >0.9990 305 p-coumaric acid 6 0.1 - 50 >0.9990 - >0.9999 >0.9999 320 p-coumaric acid 6 0.1 - 50 >0.9990 - >0.9999 >0.9999 320 Ferulic acid 6 0.1 - 50 >0.9900 - >0.9999 >0.9990 320 o-coumaric acid 6 0.1 - 50 >0.9990 - >0.9999 >0.9999 305 Salicylic acid 6 0.1 - 50 >0.9995 - >0.9999 >0.9990 305 Resveratrol 6 0.1 - 50 >0.9990 - >0.9999 >0.9990 305 Daidzein 6 0.1 - 50 >0.9990 - >0.9999 >0.9999 370 Quercetin 3 5 - 50 >0.9900 - >0.9999 >0.9990 370 Kaempferol 4 1 - 50 >0.9900 - >0.9999 >0.9999 RF 260/422 Caffeic acid 4 5 - 100 >0.9900 - >0.9999 >0.9990 260/422 p-coumaric acid 4 5 - 100 >0.9900 - >0.9999 >0.9999 260/422 Ferulic acid 6 0.5 - 100 >0.9900 - >0.9999 >0.9990 260/422 o-coumaric acid 4 1 - 50 >0.9900 - >0.9999 >0.9990 260/422 Salicylic acid 3 5 - 50 >0.9990 - >0.9999 >0.9995 260/422 Resveratrol 4 1 - 50 >0.9990 - >0.9999 >0.9999

A Statistics are for 25 analytical runs Slope coefficients for calibrations developed over 25 analytical runs are plotted in Figure 2.17. Quercetin and kaempferol had the most variable sensitivity, which may in part explain the low precision of analysis for these compounds compared to other analytes. Calibration sensitivities for analytes excluding quercetin and kaempferol were relatively stable by both PDA and RF detection if the final two analyses are excluded. It is unclear why calibration sensitivities decreased during the final two HPLC runs, and this variability will need to be monitored in future analyses.

| Development and validation of analytical methodology 72

(A) (B) (C)

(D) (E) (F)

(G) (H) (I)

FIGURE 2.17 VARIATION IN SLOPE COEFFICIENTS OF (A) CAFFEIC ACID; (B) FERULIC ACID; (C) SALICYLIC ACID; (D) P-COUMARIC ACID; (E) O-COUMARIC ACID; (F) RESVERATROL; (G) DAIDZEIN; (H) QUERCETIN; AND (I) KAEMPFEROL HPLC CALIBRATIONS. Solid line is mean slope coefficient; dotted lines are ± 1 CI (95%); dashed lines are ± 1 SD

| Development and validation of analytical methodology 73

2.3.3.3.2 Limits of detection and quantitation The estimated LOD and LOQ for each analyte are reported in Table 2.24. It is emphasised that only the PDA measurements were considered quantitative, as the RF had reduced sensitivity and precision in the serial configuration used, as discussed in chapter 1.1.1.1. LOQ were well below sample concentrations for all tested analytes excluding kaempferol. Several samples had daidzein or quercetin levels below the LOQ, but on average the concentrations could be quantified. There are few publications to compare validation statistics with: LOQ for caffeic acid, p-coumaric acid, ferulic acid, and t-resveratrol were less, and the quercetin higher, than those (0.5, 0.2, 0.4, 0.2, and 0.5 ppm, respectively) determined by Francisco and Resurreccion (2009a); whereas t-resveratrol limits were similar to those reported by Rudolf et al. (2005).

2.3.3.3.3 Accuracy Spiked peanut samples were extracted and analysed in the same way as unadulterated samples throughout the study, and recovery of the spiked analytes used to estimate accuracy (Table 2.25). Acceptable mean recoveries by PDA analysis were achieved for p-coumaric, ferulic, and o-coumaric acids using the 5% methanol-based SPE method to purify and concentrate samples. On the other hand, good recoveries of resveratrol, daidzein, quercetin, and kaempferol were only attained using the 20% methanol-based SPE method. Not surprisingly, accuracy was problematic at points of co-elution. The high variability in caffeic acid, ferulic acid, and quercetin recoveries may have been due to slight run-to-run differences in the composition of co-eluting peaks and extent of overlap at those retention times. Kaempferol was well-resolved from other peaks, but it and quercetin had generally lower sensitivity and higher variability in signal response. There may be a system suitability issue yet to be unraveled.

The method of spike addition has a large impact on recovery statistics. In this research, peanut samples were spiked after hexane-defatting but prior to almost the entirety of the experimental process, including UAE, SPE, and HPLC analysis. The accuracy statistics in Table 2.25 are therefore a rigorous reflection of the capacity to measure analytes using the entire extraction and analysis protocol. The measurement of extracts spiked just prior to HPLC analysis achieves higher and more consistent recoveries (Francisco and Resurreccion, 2009a, Rudolf et al., 2005), but such results cannot be compared with those of this study.

| Development and validation of analytical methodology 74

TABLE 2.24 LOD, LOQ, AND TYPICAL SAMPLE CONCENTRATIONS FOR HPLC ANALYSIS

Detector λ (nm) SPE (%) Analyte Limits (μg g-1) Native extracts (μg g-1 dw)A Enzyme-treated extracts (μg g-1 dw)A LOD LOQ N Mean ± SD Min - Max N Mean ± SD Min - Max

PDA 320 5 Caffeic acid 0.01 0.04 30 103 ± 36 37 - 180 32 58 ± 26 11 - 96 305 5 p-coumaric acid 0.01 0.04 33 13 ± 8 3 - 29 35 111 ± 54 39 - 224 320 5 p-coumaric acid 0.01 0.04 33 13 ± 8 3 - 29 35 112 ± 55 38 - 226

| Development and validation of analytical methodology 320 5 Ferulic acid 0.01 0.04 33 39 ± 17 14 - 76 35 85 ± 20 46 - 114 320 5 o-coumaric acid 0.01 0.04 33 6.8 ± 3.8

TABLE 2.25 ACCURACY (% SPIKE RECOVERY) OF HPLC ANALYSIS

Detector λ (nm) Analyte Retention SPE: 5% methanol SPE: 20% methanol time (min) N Mean ± SD N Mean ± SD

PDA 320 Caffeic acid 5.9 13 78 ± 99 34 77 ± 143 305 p-coumaric acid 8.8 22 90 ± 21 36 99 ± 38 320 p-coumaric acid 8.8 22 90 ± 21 35 97 ± 36 320 Ferulic acid 10.0 22 92 ± 36 35 98 ± 49 320 o-coumaric acid 12.9 22 82 ± 23 18 76 ± 37 305 Salicylic acid 14.8 22 95 ± 18 35 99 ± 17 305 Resveratrol 16.6 22 47 ± 16 35 91 ± 18 305 Daidzein 17.9 12 73 ± 20 24 103 ± 19 370 Quercetin 20.5 13 25 ± 21 33 68 ± 28 370 Kaempferol 25.8 22 28 ± 13 35 94 ± 61 RF 260/422 Caffeic acid 6.0 10 44 ± 91 5 15 ± 22 260/422 p-coumaric acid 8.9 3 126 ± 14 13 146 ± 108 260/422 Ferulic acid 10.1 10 160 ± 168 5 95 ± 67 260/422 o-coumaric acid 13.0 19 51 ± 38 18 42 ± 48 260/422 Salicylic acid 14.9 21 91 ± 14 32 103 ± 20 260/422 Resveratrol 16.7 22 44 ± 15 33 92 ± 21

2.3.3.3.4 Instrumental, within-run, and between-run precision The repeatability of retention times, standard peak integration, and quantification of the peaks by the respective experimental calibrations, are reported in Table 2.26. Precision statistics for PDA quantification were the most important for our purposes; while between-run, or intra- assay, precision was of greater practical importance than instrumental or within-run precision because it encompasses all stages of sample preparation and analysis.

Instrumental repeatability was <0.6% for all analytes, while caffeic acid, p-coumaric acid, resveratrol, and daidzein were subject to almost negligible instrumental variation. Within-run precision, which reflects the variation throughout the analytical run (typically 35 h) on any given day, was <2% for most analytes; while between-run precision was <3% for most analytes. Analytes subject to high signal variation and hence greater variability included o-coumaric acid, quercetin, and kaempferol. Less than 3% between-run precision for most analytes, and 4-8% for less sensitive analytes, was acceptable given that 2% is considered to be ideal (Hibbert, 1999). Francisco and Resurreccion (2009a) and Rudolf et al. (2005) only reported instrumental precision of 1-2%.

| Development and validation of analytical methodology 76

TABLE 2.26 PRECISION OF HPLC ANALYSIS

Detector λ (nm) Analyte Instrumental repeatability, RSD (%)A Within-run precision, RSD (%)B Between-run precision, RSD (%)C Ret. time Peak area Quantification Ret. time Peak area Quantification Ret. time Peak area Quantification

PDA 320 Caffeic acid 0.1 <0.05 <0.05 0.2 0.3 0.6 0.8 5.5 2.8 305 p-coumaric acid 0.1 <0.05 <0.05 0.2 0.3 0.5 0.8 5.3 2.9 320 p-coumaric acid 0.1 <0.05 <0.05 0.2 0.3 0.5 0.8 5.4 2.6

| Development and validation of analytical methodology 320 Ferulic acid 0.1 0.1 0.1 0.2 0.2 0.5 0.8 5.5 2.6 320 o-coumaric acid 0.1 0.1 0.1 0.2 2.0 2.3 0.8 8.3 7.8 305 Salicylic acid <0.05 0.2 0.2 0.2 0.9 1.2 0.8 5.8 2.9 305 Resveratrol 0.1 <0.05 <0.05 0.2 0.1 0.3 0.7 5.4 2.8 305 Daidzein <0.05 <0.05 <0.05 0.2 0.1 0.3 0.7 4.3 2.3 370 Quercetin <0.05 0.5 0.5 0.2 4.6 3.3 0.6 10.6 5.6 370 Kaempferol <0.05 0.4 0.4 0.2 2.6 1.9 0.6 7.3 4.1 RF 260/422 Caffeic acid 0.1 0.6 0.6 0.2 1.1 1.3 0.8 7.8 4.3 260/422 p-coumaric acid 0.1 0.5 0.5 0.2 1.2 1.5 0.8 7.5 3.9 260/422 Ferulic acid <0.05 0.5 0.5 0.2 1.3 1.5 0.8 7.4 2.8 260/422 o-coumaric acid <0.05 0.3 0.3 0.2 1.1 1.3 0.8 6.9 2.4 260/422 Salicylic acid <0.05 0.5 0.5 0.2 1.8 1.8 0.8 8.2 2.8 260/422 Resveratrol <0.05 0.3 0.3 0.2 1.0 1.0 0.7 8.1 2.4 A Statistics are for 10 separate analytical runs B Statistics are for 16 separate analytical runs C Statistics are for 41 QC standard measurements made in the course of 12 analytical runs 77

2.3.4 Conclusion

Methods for extraction and HPLC analysis of a range of phenolic acids (i.e., caffeic acid, p-coumaric acid, ferulic acid, o-coumaric acid, and salicylic acid), stilbenes (resveratrol), and flavonoids (daidzein, quercetin, and kaempferol) in peanut kernels were developed and optimised (summarised in Table 2.27). Samples were defatted with n-hexane, extracted in aqueous methanol with the assistance of ultrasound and enzyme (cellulase and protease) preparations, concentrated and purified by reversed phase SPE, and then reconstituted in mobile phase for HPLC analysis. A HPLC-PDA method was optimised for separation of most of the detectable antioxidant phytochemicals in peanut extracts using a C18 column in an aqueous acetonitrile binary solvent system (acidified with 0.1% TFA).

Validation statistics indicated that the method had LOQ (Table 2.24) as low as 0.04 μg g-1 for all analytes, excluding quercetin (2.00 μg g-1) and kaempferol (0.40 μg g-1) for which the method was less sensitive. The analysis was accurate (Table 2.25), as indicated by spike recovery of 90- 103% for most analytes, although quercetin (68%) and o-coumaric acid (82%) were less accurately measured. Instrumental repeatability was excellent (<0.6%), within-run precision was <2% for most analytes, and between-run precision was <3% for most analytes (Table 2.26). Therefore, the method can be considered a useful validated procedure for extraction, concentration, and HPLC analysis of antioxidant phytochemicals in peanuts and similar matrices.

| Development and validation of analytical methodology 78

TABLE 2.27 SUMMARY OF PHYTOCHEMICAL EXTRACTION AND HPLC METHOD DEVELOPMENT

Stage Factors Explanation Result Extraction Type of Previous work indicated sonication and microwave-assisted sonication Repeatable, convenient, and relatively high-throughput extraction was method improved the extraction of t-resveratrol from peanut kernels (Chukwumah et prioritised over maximum extraction efficiency. UAE was used, based on al., 2007b). the literature. Solvent Peanut extracts have been most commonly made in 80-90% aq. ethanol, 90% aq. methanol was used, based on the literature and previous methanol, or acetonitrile. experiences. The aqueous component helps to wet particles and thereby aids the penetration of methanol into the matrix. | Development and validation of analytical methodology Sonication The temperature (ambient, 40, and 60 °C) and duration (5, 15, 30, and 60 min) There was marginal benefit in increasing the temperature or duration of conditions of extraction was optimised. A 1:5 (w/v) sample/solvent ratio was selected and extraction. The final conditions were: 50 kHz, 40 °C, 5 min, 1:5 (w/v) the extraction repeated three times, based on previous experience. An sample/solvent, repeated three times. ultrasound frequency of 50 kHz has been commonly used in the literature. Hydrolysis Alkaline (10 M NaOH agitated overnight) and acid hydrolysis (conc. HCl, 85 °C, 1 The standards (i.e., p-coumaric acid, polydatin, t-resveratrol, genistein, h) of peanut sample (native and spiked) followed by extraction in 1:1 (v/v) rutin, quercetin, and biochanin A) were apparently thermally stable up diethyl ether/ethyl acetate was investigated, as described by Matilla and to 80 °C. Hydrolysis reactions, e.g., polydatin → resveratrol and rutin → Kumpulainen (2002). Different temperatures (i.e., 20, 40, 60, and 80 °C) and quercetin, were not obvious, and we decided to focus on analysis of durations (i.e., 15, 30, 60, and 120 min) of acid hydrolysis were tested and native extracts, leaving analysis of conjugated phenolic acids and effects on the HPLC chromatogram observed. flavonoids for future work. Enzyme Commercially-available protease (Alcalase 2.4L FG, Novozymes Ltd.) and Treatment with Alcalase 2.4L FG and Celluclast 1.5 L increased the treatment cellulase (Celluclast 1.5 L, Novozymes Ltd.) preparations were tested. extraction of antioxidant phytochemicals from our samples. Samples were extracted with and without enzyme treatment in order to assess the importance of bound phytochemicals in the peanut.

Solid phase Sorbent Strata-X polymeric and Strata C18 reversed phase-type SPE sorbents from Polymeric reversed phase sorbent gave the best recovery of a range of extraction Phenomenex were tested. standards tested. Pre- Pre-conditioning with methanol and with acidified methanol (pH 2) was There was better retention of the most polar phenolic acids when conditioning compared. sorbent was pre-conditioned with acidified methanol. Acidified solutions (pH 2) were used for pre-conditioning and loading. Loading Breakthrough of standard compounds when loaded in differing concentrations Acidifying the loading solution to pH 2 was critical for retention of the of aq. methanol (i.e., 2, 10, and 30%) and different acidity (pH 2 and 4) was most polar phenolic acids, e.g., p-hydroxybenzoic acid. Extracts were tested. NB, an undergraduate student tested pH 2, 4, and 6, and breakthrough loaded in 5% methanol to optimise recovery of phenolic acids. 79 of t-cinnamic acid in 0, 10, 20, 30, and 40% aq. methanol solutions.

TABLE 2.27 CONTINUED

Stage Factors Explanation Result Elution Eluents tested were methanol, acidified methanol (pH 2), and 50:50 All eluents gave good recovery, so methanol was used for simplicity. The results (v/v) acetonitrile/methanol. The eluates from four sequential aliquots indicated that 2 mL solvent was sufficient to elute analytes. 3 mL methanol in two of methanol (1 mL) were analysed by HPLC to determine the optimal aliquots (2 mL then 1 mL) was used to ensure all retained analytes were eluted. eluent volume.

HPLC Column Several reversed phase columns with C18 or C18-like functionality were The Gemini C18 achieved the best peak resolution despite the theoretical benefits of tested including Kinetex C18, Gemini C18, Onyx monolithic C18, and the newer Kinetex and Synergi-polar columns. Separation using the Onyx column

|Development and validation Synergi-polar. was achievable, but solvent consumption was very high. Mobile phase A simple gradient elution using methanol or acetonitrile and aqueous Acetonitrile as the organic phase gave the best retention range. Acidification of the binary mobile phase was developed. mobile phase with 0.1% formic acid or 0.1% trifluoroacetic acid improved peak shape and repeatability of the phenolic acids. System Split-flow and series configurations for simultaneous detection of Post-column flow-splitting introduced excessive noise in the ECD trace despite our configuration analytes by PDA, RF, and ECD detectors were explored. attempts to control fluctuations in pressure by split valve and tubing adjustments. The best calibrations and repeatability were achieved using PDA. We suggest that RF and ECD are best operated as stand-alone detectors. of analytical methodology 80

3 Dietary minerals

3.1 Literature review

3.1.1 Importance of dietary minerals

3.1.1.1 Essentiality and functional food traits of dietary minerals An element is considered to be an ‘essential mineral’ if it is necessary to support normal growth, reproduction or health during the life cycle, when all other nutritional requirements are held at optimal levels. Essentiality has conventionally been proven either by the demonstration of biological function or by the presence/negation of deficiency symptoms in studies of controlled mineral intake (Webster-Gandy et al., 2006). The categorisation of essential minerals thus undergoes ongoing review as scientific knowledge of the roles of elements in the body expands.

Minerals are sometimes described as macro- or micro-elements depending on their relative concentrations in the body, which is related to the daily intake needed to maintain adequate levels. Macro-elements are required in quantities of 100 mg or more and include calcium, fluoride, iron, potassium, magnesium, sodium, phosphorus, and zinc. Micro- or trace elements are required in quantities of only several mg or less and include chromium, copper, iodine, manganese, molybdenum, and selenium (Webster-Gandy et al., 2006). The essentiality of this list is widely accepted, though the description of elements as ‘macro’ or ‘micro’ varies among authors. Minerals perform a range of physiological functions in the human body as components of body tissue, hormones, enzymes, and other regulatory compounds.

A number of elements, such as boron, cobalt, nickel, and vanadium are regarded as potentially essential. Arsenic, cadmium, lithium, lead, and strontium are also regarded to have potentially essential functions in the body, even though they are toxic environmental contaminants (Nabrzyski, 2007, Webster-Gandy et al., 2006). Such distinctions are somewhat semantic and date back to traditional definitions of essentiality, as these very trace elements are now known to play important roles in the body, particularly as cofactors and transcription factors (Harris, 2005, Kudrin, 2000).

The major functions of essential minerals, and some of the potentially essential minerals, are listed in Table 3.1.

| Dietary minerals 81

TABLE 3.1 MAJOR FUNCTIONS OF ESSENTIAL MINERALS IN THE BODY

Element Essential function B B status affects a range of metabolic processes and is suggested to have a regulatory role in the metabolism of other minerals. However specific functions and essentiality have not been proven. There is evidence that B deprivation has adverse effects on Ca metabolism, brain function, and energy metabolism. Ca The most obvious role of Ca is providing structural integrity and strength to bones and teeth. Ca is involved in energy regulation and fat deposition. Long-term deficiency leads to bone loss in the elderly and failure to achieve peak bone mass in the young. Cr Cr is involved in insulin and glucose metabolism. Deficiency is associated with impaired glucose tolerance and increased risk factors for type 2 diabetes mellitus and cardiovascular diseases.

Co Co is a cofactor in vitamin B12. Cu Cu is a cofactor for many enzymes involved in cellular oxidation, including cytochrome c-oxidase, tyrosinase, dopamine B-hydroxylase, ceruloplasmin, hephaestin, and peptidylglycine α-amidating monooxygenase. Deficiency most frequently manifests as refractory anaemia (i.e., anaemia that does not respond to iron treatment), neutropenia (abnormally low neutrophil levels in blood), and bone demineralisation presenting as fractures. F F is beneficial rather than essential. RDI protects against dental caries and mottling of tooth enamel, but there is little evidence that deprivation causes any nutritional deficiency. I I is a component of hormones involved in thyroid function, and has a regulatory role in the synthesis and secretion of other hormones related to growth and development. Deficiency inhibits normal brain development, which in severe cases can lead to cretinism. Insufficient I is the most common cause of preventable mental retardation worldwide. Other effects include goitre, hypothyroidism, decreased fertility rates, increased rates of stillbirth and spontaneous abortion, and increased perinatal and infant mortality. Fe The most obvious role of Fe is in haem protein transport of oxygen in the body. Fe functions in the oxygen and energy metabolism of mitochondrial cytochromes, in the antioxidant enzyme catalase, and enzymes involved in energy metabolism (e.g., aconitase, NADH dehydrogenase, and succinate dehydrogenase). Serious deficiency causes anaemia, which leads to physical weakness, impaired cognitive function, and adverse effects on pregnancy and child development. Mg Mg is involved in many cellular functions as a cofactor in more than 300 enzymatic reactions including oxidative phosphorylation, glycolysis, DNA transcription, and protein synthesis. Deficiency causes complex neuromuscular symptoms such as muscular fasciculation, tremor, tetany, nausea, and vomiting. Mg deficit disturbs Ca homeostasis, and is also associated with diabetes and insulin resistance, micro- and macro-angiopathy, and neuropathy, possibly due to effects on intracellular signaling and processing. Mn Mn is a constituent of enzymes including arginase, pyruvate carboxylase, and manganese-superoxide dismutase. It is an activator of many more enzymes including a range of hydrolases, kinases, decarboxylases, and transferases. Mn deficiency is uncommon under normal living conditions, but can cause impaired growth and reproduction, skeletal abnormalities, ataxia, and defects in carbohydrate, glucose, and lipid metabolism. Mo Mo is an essential cofactor in enzymes such as xanthine oxidase, sulfite oxidase, and aldehyde oxidase, which are involved in the metabolism of purines, pyrimidines, pterins, aldehydes, and sulfites. Ni Ni is an infrequent cofactor. P Phosphate is a component of many biomolecules including phospholipids, creatine phosphate, nucleotides, nucleic acids, and ATP, and has many cellular functions. P metabolism is closely linked with Ca, and it too has an important role in skeletal growth and maintenance. P is abundant in food, so deficiency is extremely rare. K K is the major intracellular cation and thereby affects nerve transmission, muscle contraction, and vascular tone. Severe deficiency causes cardiac arrhythmia, muscle weakness, and glucose intolerance. Moderate deficiency leads to increases in blood pressure, salt sensitivity, risk of kidney stones, and bone turnover, and may also increase the risks of stroke and other cardiovascular diseases.

| Dietary minerals 82

TABLE 3.1 CONTINUED

Element Essential function Se Se has numerous enzymatic, structural, and detoxifying roles in the body, mainly as a component of various selenoproteins (e.g., glutathione peroxidases, thioredoxin reductases, and iodothyronine deiodinases). The essentiality of Se was recognised relatively recently, and these functions are incompletely understood. Deficiency is implicated in Keshan and Kashin-Beck diseases, although neither are purely nutritional disorders. Na Na regulates the volume of extracellular (ECF) and intracellular fluid (ICR), and thus has an essential role in maintaining the osmotic concentration and integrity of cells. ECF/ICF abnormalities cause swelling or shrinkage of cells, with potentially severe impacts on the brain. Excess Na can lead to oedema, while deficits that significantly decrease plasma volume can compromise cardiovascular function. Nutritional concern regarding Na is primarily with excess dietary intake, age-related increase in arterial pressure, and hypertensive diseases such as stroke and coronary heart disease. Zn Zn is a component of more than 200 enzymes and even more transcription factors. It has functions throughout the body, for example, in DNA and protein synthesis, cell division and growth, and macronutrient metabolism. Deficiency has adverse effects on growth and wound healing, foetal (esp. brain) and pubertal development (e.g., spermatogenesis), immune system, and endogenous antioxidant defense.

Source: Caballero et al., (2005)

TABLE 3.2 FUNCTIONAL FOOD CHARACTERISTICS OF DIETARY MINERALS

Element Functional food trait Reference Mg Mg supplementation is suggested to relieve asthma symptoms, alleviate the frequency (Beckstrand and duration of migraine headaches, prevent or alleviate seizures associated with pre- and Pickens, eclampsia, decrease insulin resistance, and delay the onset of type 2 diabetes mellitus. It 2011, alleviates mild hypertension and has an additive antihypertensive effect with all classes of Houston, antihypertensive drugs. Oral or dietary Mg supplementation has very few adverse effects 2011) in subjects with normal renal function since excess Mg is excreted in the urine. Se Supplementation and case-control studies suggest increased Se status may have (Bates, protective effects on the risk of specific cancers, e.g., a US study of prostate cancer 2005) incidence among male health professionals (N=34000); and hepatocellular carcinoma in a Chinese controlled intervention. There are ongoing large-scale intervention studies including Prevention of Cancer by Intervention with Selenium (PRECISE) and Selenium and Vitamin E Cancer Prevention Trial (SELECT). Zn There is a strong relationship between Zn status, redox metabolism, and oxidative stress. (Freake, Zn may contribute to the body’s endogenous antioxidant defense in several ways: it can 2005, Maret bind the thiol groups in proteins and make them less susceptible to oxidation; it can and displace Cu and Fe from proteins and lipids and thereby decrease metal-catalysed Sandstead, formation of hydroxyl radicals; it induces metallothionein, which is involved in Zn 2006) homeostasis and also scavenges hydroxyl radicals; and it is a component of Cu/Zn superoxide dismutase, an important antioxidant enzyme that is impaired by Zn deficiency. On the other hand, Zn deficiency is associated with increased oxidative stress in animal studies, and supplementation offers protective effects against diabetic retinopathy. A potential anti-cancer link is suggested by the frequent mutation of the tumor-suppressor gene p53, a Zn-dependent transcription factor, during carcinogenesis. Most therapeutic applications of Zn supplementation target conditions in which Zn loss is (Haase et a secondary complication of the disease. However there is also evidence that enriched al., 2008) intake in Zn-sufficient subjects may diminish the incidence or progression of specific diseases. According to some studies, Zn supplementation may: diminish the severity of age-related immunosenescence, shorten the duration of the common cold, decrease the side effects of radiotherapy in cancer patients, and alleviate psychiatric illnesses including unipolar depression and attention deficit hyperactivity disorder. A large scale study (3640 participants over 5 years) on age-related macular degeneration found significant benefits from Zn treatment alone, and enhanced benefits when Zn was combined with the antioxidant vitamins A, C, and E.

| Dietary minerals 83

Several dietary minerals are of interest to consumers and food industry, not only for their contribution to essential nutrition, but because of their expected preventative or therapeutic effects on specific diseases. These relationships have not been conclusively proven and are the subject of ongoing research. The methodological issues that often confound comparison of clinical studies are partially to blame, such as heterogeneous sample populations with differing clinical profiles and lack of standardised dosing and analysis protocols. With this in mind, the potential ‘functional food’ qualities of some dietary elements are outlined in Table 3.2.

3.1.1.2 Dietary recommendations The governments of Australia and New Zealand provide nutrition advice to the public in the form of dietary guidelines for different ages and stages of the lifecycle through the auspices of the National Health and Medical Research Council (NHMRC). These are communicated by the use of nutrient reference values, defined in Table 3.3. Terminology and dietary recommendations generally conform to the international consensus as predicated by the World Health Organization (FAO/WHO, 2004), though policy is of course formulated at a national level. The NHMRC judged calcium, chromium, copper, fluoride, iodine, iron, potassium, magnesium, manganese, molybdenum, sodium, phosphorus, selenium, and zinc to be essential in 1991 (NHMRC, 2006). Reference values are provided for different age and gender groups of children and adults, with pregnant and lactating women listed separately. An example of nutrient reference values, for adult men and women, is shown in Table 3.4.

TABLE 3.3 DEFINITIONS OF NUTRIENT REFERENCE VALUES USED IN AUSTRALIA

Reference value Definition Calculation

Adequate intake (AI) Daily nutrient intake assumed to be Based on observed or experimentally- adequate; used when there is insufficient determined estimates of nutrient intake by a data to calculate EAR and RDI. group(s) of apparently healthy people. Estimated average Daily nutrient intake estimated to meet Based on average requirements of a group of requirement (EAR) the requirements of 50% of healthy subjects, to meet specific criteria of individuals in the life stage/gender group. adequacy, adjusted for inter-individual variability. Recommended daily Daily nutrient intake that is sufficient to EAR plus two standard deviations (SD); or intake (RDI) meet the requirements of 97-98% of twice an assumed coefficient of variation (CV) healthy individuals in the life stage/gender of 10% if the SD is unavailable. group. Upper level of intake Highest daily nutrient intake likely to pose Based on observed or experimentally- (UL) no adverse health effects to 97-98% of determined estimates of nutrient intake by a individuals in the general population. group(s) of apparently healthy people. Source: NHMRC {2006)

| Dietary minerals 84

TABLE 3.4 NUTRIENT REFERENCE VALUES FOR ADULT MEN AND WOMEN (19-30 YEARS)

Element Units Men, 19-30 years Women, 19-30 years AI EAR RDI AI EAR RDI Calcium mg day-1 840 1000 840 1000 Chromium μg day-1 35 25 Copper mg day-1 1.7 1.2 Fluoride mg day-1 4 3 Iodine μg day-1 100 150 100 150 Iron mg day-1 6 8 8 18 Magnesium mg day-1 330 400 255 310 Manganese mg day-1 5.5 5 Molybdenum μg day-1 34 45 34 45 Phosphorus mg day-1 580 1000 580 1000 Potassium mg day-1 3800 2800 Selenium μg day-1 60 70 50 60 Zinc mg day-1 12 14 6.5 8

Source: NHMRC (2006) 3.1.1.3 Global prevalence of micronutrient malnutrition The Food and Agriculture Organization of the United Nations (FAO, 2010) estimated that 1023 million people were malnourished in 2009, with 98% of these located in the developing countries of Asia and the Pacific (62%), sub-Saharan Africa (26%), Latin America and the Caribbean (6%), and the Near East and North Africa (4%). As a group, about 16% of the developing world suffers hunger, though this proportion of course varies from country to country. While protein and energy insufficiency may be the most dramatic manifestation of malnutrition, the ‘hidden hunger’ of vitamin and mineral deficiencies likewise takes a devastating toll on human health and productivity, through unnecessary child and maternal morbidity, impaired physical or mental ability, increased susceptibility to disease, and the consequent social and economic costs of these.

The concept of ‘disability-adjusted life years’ (DALY) was developed to quantify the magnitude of such health consequences (Stein, 2010, Stein et al., 2005). One DALY denotes the loss of one ‘healthy’ life year to cause-specific morbidity and mortality, measured as ‘years lived with disability’ (YLD, weighted) and ‘years of life lost’ (YLL), respectively. The World Health Organisation (WHO) uses DALY estimates to evaluate the relative importance of contributing factors to the global health burden (WHO, 2004). The latest statistics for major causal categories of morbidity and mortality are summarised in Table 3.5.

| Dietary minerals 85

Nutritional deficiencies directly account for about 2.5% of the world’s overall disease burden, notwithstanding their indirect contribution, for example, by reducing overall health and immunity and thereby increasing susceptibility to diseases and impeding recovery. While WHO no longer reports zinc deficiency individually, as it does iron and iodine, these three elements remain the most significant mineral deficiencies. Calcium and selenium are important to a lesser extent, while copper and magnesium deficiencies are common in specific sub- populations (Stein, 2010). Vitamin A and folate are the most important vitamin deficiencies (Micronutrient Initiative, 2009).

TABLE 3.5 CONTRIBUTION OF NUTRITIONAL DEFICIENCIES TO GLOBAL MORBIDITY AND MORTALITY

Cause YLD (‘000s) YLL (‘000s) Total DALY (‘000s) % of total DALY Non-communicable 413408 318243 731652 48.0 diseases Communicable diseases 53956 345974 399930 26.3 Injuries 58841 135157 187614 12.3 Perinatal conditions 20454 105970 126423 8.3 Maternal conditions 23165 15770 38936 2.6 Nutritional deficiencies 27097 11606 38703 2.5 x Protein- x 10438 x 7024 x 17462 x 1.1 energy x 13204 x 2948 x 16152 x 1.1 x Iron x 3359 x 170 x 3529 x 0.2 x Iodine x 36 x 593 x 629 x 0.0 x Vitamin A Source: WHO (2004) YLD, years lived with disability; YLL, years of life lost; and DALY, disability-adjusted life years

Micronutrient malnutrition is caused by insufficient dietary intake and diversity, exacerbated by disease, and underlaid by complex issues of limited access to food, healthcare, education, and sanitation (Micronutrient Initiative, 2009). These issues are frequently associated with poverty (Peña and Bacallao, 2002), conflict (Egal, 2006), and emergency situations (Weise Prinzo and de Benoist, 2002), and thus affect underprivileged populations worldwide. Having said this, micronutrient malnutrition may still occur in populations that have sufficient or excessive food consumption if dietary quality is poor. For example, 9% of UK and US adults are considered to be at risk of calcium and magnesium deficiency and 14% at risk of potassium deficiency, due to the displacement of fruit and vegetables from the diet by cereal products, which tend to be naturally low in these minerals (Broadley and White, 2010).

In Australia, mild micronutrient deficiencies at levels not necessarily causing clinical symptoms are surprisingly widespread. For example, a national survey in 1991 found that 50% of men

| Dietary minerals 86 and 39% of women had suboptimal magnesium intake, while 67% of men and 85% of women had suboptimal zinc intake (Baghurst et al., 1991). In a West Australian survey of 15 year old schoolchildren (n=555), about 12% of boys and 33% of girls were at risk of calcium deficiency, 21% of girls risked inadequate zinc intake, and many adolescents risked suboptimal carotene, fibre, riboflavin, and niacin intakes (Milligan et al., 1995). Iodine deficiency was recognised as a public health issue in the late 1940s, when state-wide surveys revealed about 50% prevalence of goitre among Tasmanian school-age children (Hynes et al., 2009). These were alleviated by supplementation and fortification programs, but mild to moderate iodine deficiency re- emerged in Tasmania and other parts of Australia in the 1990s (Eastman and Li, 2009) and led to the introduction of mandatory use of iodised salt in bread making since 2009 (FSANZ, 2009).

Specific sub-groups of the population have a greater likelihood of suffering micronutrient deficiencies. In Australia, Aboriginal communities are a special risk group. Like the indigenous populations of Canada, the US, and New Zealand, indigenous Australians have higher rates of malnutrition than non-indigenous, especially those living in rural and remote areas (Ruben, 2009). Indigenous children suffer a much higher prevalence of diet-related illness than non- indigenous children, particularly anaemia, iron depletion, eosinophilia (parasite-related deficiencies), and zinc deficiency linked to diarrhoeal disease (Brewster, 2006, Heath and Panaretto, 2005).

The elderly often experience physiological factors that lead to reduced food intake, such as decreased appetite, diminished smell and taste perception, difficulty chewing, and poor dentition; while the absorption and bioavailability of nutrients may decline because of oral or gastrointestinal illness and drug interactions (Wood, 1991). This is compounded by factors such as disability, psychological deterioration, financial challenges, and social isolation, especially in women whose greater life expectancy brings increased likelihood of experiencing these burdens (Chernoff, 2005). A seven-year cohort study of more than 900 community-living 70-85 year old women in Western Australia is illustrative: micronutrient intake declined with age; folate, vitamin E, and calcium intakes were suboptimal throughout the study; while magnesium and potassium intakes were frequently suboptimal (Zhu et al., 2010).

Vegetarian diets have a higher risk of inadequate iron and zinc than non-vegetarian diets, though deficiency disorders generally only occur in impoverished communities. High intake of phytic acid or other modifiers of mineral absorption, which are found in some commonly- consumed plant foods, can lower the bioavailability of iron, zinc, and possibly other trace elements (Sanders and Reddy, 1994, Hunt, 2003). Vegetarians have an increased risk of iodine

| Dietary minerals 87 deficiency in areas where soils are naturally iodine-poor, unless consuming supplements or iodised food (Remer et al., 1999).

3.1.2 Potential benefits of mineral biofortification

3.1.2.1 Improvement of public nutrition Genetic improvement has long been a key strategy by which primary industries have achieved production goals. Selection criteria have traditionally related to productivity, for example, maximising yield potential, reducing losses to pests and diseases, and increasing adaptability to environmental conditions to allow expansion into new production zones; whereas breeding for nutritional quality is relatively recent. Enrichment of mineral concentrations in staple crops has been identified as an efficient and sustainable approach to improving public nutrition, particularly in malnourished rural populations, whose capacity to diversify their diet or access commercial fortified foods is often limited (Bouis, 2003).

Genetic improvement of nutritional quality, or ‘biofortification’, complements the more traditional interventions of dietary diversification, supplementation, and fortification. Sufficient consumption of a varied diet is the simplest and most effective way to achieve optimal nutrition; however this may be impractical due to production constraints, limited market access, compliance issues, and so on. Supplementation is the best short-term method of alleviating acute mineral deficiencies, but is curative rather than preventative due to adherence and compliance problems (Nantel and Tontisiri, 2002). Food fortification is an effective long-term strategy that has been used successfully in industrialised countries (notably for iodine and iron), but is challenging to implement in developing nations as it requires strong processing infrastructure and a high level of technical sophistication and training to ensure appropriate safe dosage and monitoring. Moreover, commercial fortified products may lack penetration into rural communities (Nantel and Tontisiri, 2002).

The success of both supplementation and fortification is dependent on a high level of sustained funding, good governance, an extensive distribution network, and in the case of fortification, good processing infrastructure, technical competency and training (Gómez-Galera et al., 2010). In comparison, plant breeding requires a large initial investment, but costs after varietal release are low. Adoption tends to be widespread because farmers merely substitute seed varieties, without need for cultural change, technical upgrade, or extraordinary capital investment. Micronutrient-enriched seeds have the added advantage of increasing farm productivity by increasing plant resistance to disease and abiotic stresses, increasing seedling

| Dietary minerals 88 survivability and vigour, and boosting yields in nutrient-deficient soils, which are typically cultivated by the rural poor (Nestel et al., 2006).

The cost-effectiveness of biofortification for alleviating micronutrient malnutrition has been analysed in several ongoing breeding programs, based on estimations of the research and development investment, breeders’ estimation of feasible crop enrichment by conventional breeding, and ex ante calculations of health benefits (Meenakshi et al., 2010, Qaim et al., 2007). While these studies were conducted mainly to aid research priority-setting within the CGIAR HarvestPlus Challenge Program, the projections certainly suggest that biofortification is a cost-effective strategy. For example, iron-biofortified beans could reduce the burden of iron deficiency by 9-36% in northeast Brazil at a cost of US$ 134-20 DALY-1, under pessimistic and optimistic scenarios, compared to US$ 487 DALY-1 for supplementation and US$ 215 DALY-1 for fortification, assuming 50% coverage (Meenakshi et al., 2010).

Characteristics of the main types of dietary intervention and examples of each are summarised in Table 3.6. These should not be considered in opposition, but as complementary approaches best used in combination, to accommodate the varying features of target nutrients and populations (Mannar, 2006).

| Dietary minerals 89

TABLE 3.6 COMPARISON OF INTERVENTION STRATEGIES FOR ALLEVIATION OF MICRONUTRIENT MALNUTRITION

Type of Dietary diversification Supplementation Fortification Biofortification intervention

Description Strategies that enable and promote both Administration of supplements Artificial addition of micronutrients Enhancement of natural micronutrient content production and consumption of a diverse (e.g., pills, tablets, syrup) to to foodstuffs (esp. staples). and bioavailability by plant breeding or genetic diet with improved micronutrient content target populations. engineering. Definition sometimes includes and bioavailability. agronomic interventions. Examples of Diversification of food production (e.g., Medical-administration by Mandatory or commercially- Select for increased efficiency in micronutrient strategies livestock, aquaculture); technology to health personnel; self- motivated fortification by uptake, translocation, and storage; reduced facilitate more diverse consumption (e.g., administration of prescribed processors or distributors; inhibitor (e.g., phytate, tannins) and increased refrigeration); techniques to increase supplements; voluntary self- voluntary self-fortification, e.g., promoter (e.g., ascorbate) levels to improve bioavailability (e.g., soaking, fermenting, administration. consumers add micronutrient bioavailability (Welch and Graham, 2004). and germinating certain grains and pulses); ‘sprinkles’ before consumption. targeted nutritional education (Gibson and Hotz, 2001). Specific case Community-based intervention in rural High-dose vitamin A capsules Universal salt iodisation and flour CGIAR HarvestPlus Challenge Program aiming to examples southern Malawi that focused on administered to 75% of young fortification (often folic acid and increase the Fe, Zn, provitamin A, and (secondarily) increasing dietary diversity and quality, children in 90 developing iron) in many countries (Mannar, I and Se concentrations in a number of staple increasing animal-derived foods, and countries in 2002 WHO 2006), including Australia (FSANZ, crops; i.e., bean, cassava, maize, rice, sweet reducing phytic acid intake (Gibson et al., campaign (Mannar, 2006). 2009, FSANZ, 2010). potato, and wheat in Phase I; and 2003). bananas/plantain, barley, cowpea, groundnut, Voluntary consumption of See Darnton-Hill and Nalubola lentil, millet, pigeon pea, potato, sorghum, and Revival of traditional foods among over-the-counter (2002) for a list of mandatory yam in Phase II (Pfeiffer and McClafferty, 2007). indigenous peoples in Hawaii, Micronesia, vitamin/mineral supplements, fortification programs in | Dietary minerals Canada, and Australia (reviewed by e.g., in Australia (MacLennan developing countries. Kuhnlein and Receveur, 1996). et al., 2006), Canada (Guo et See Pérez-Expósito and Klein (2009) al., 2009), and the US (Balluz for a review of food aid delivered in et al., 2000). the form of fortified blended foods.

See Bishai and Nalubola (2002) for a historical perspective on fortification in the US, esp. Fe, I, vit- D, vit-B, and Ca. 90

TABLE 3.6 CONTINUED

Type of Dietary diversification Supplementation Fortification Biofortification intervention

Effectiveness Effective in long-term; behavioural changes Effective in short-term; Effective in long-term; good Effective in long-term; good coverage as new seed may require cultural adaptation. adherence (e.g., forgetfulness) coverage when centrally-processed is easily adopted; no compliance issues. and compliance (e.g., side staple foods are fortified, but may effects) issues. miss poor and remote communities. Sustainability Self-perpetuating after behavioural change Curative rather than Cost-effective preventative Cost-effective long-term strategy; additional achieved. preventative; unsustainable in strategy. agronomic benefits. long-term. Costs Initial investments mainly in education, but Expensive ongoing costs. Processing, regulatory, and Expensive, but little investment required after R&D low ongoing costs. monitoring costs absorbed by phase; there may be ongoing price premium for government (subsidies) or improved seed. consumers (increased product price). Labour Low; dependent on delivery medium (e.g., For mandatory Low; delivery through usual Low; delivery through usual seed distributors. requirements health, education, extension services). supplementation, intense distribution networks. need for health personnel. Infrastructure Some changes in production and Needs good health and Needs centralised processing Breeding program and usual seed distribution requirements processing may be necessary. distribution networks; facilities, good distribution, and network. adequate QC capacity to good governance. | Dietary minerals ensure safe dosing.

91

3.1.2.2 Improvement of food quality and marketability Health-promoting ‘functional foods’ represent one of if not the single most rapidly expanding segment of food markets in the developed world, particularly within the health-conscious populations of Japan, USA, EU, and Australia (Arai et al., 2002). It is difficult to estimate the size of the market because of differences in regulatory definitions and general understandings of ‘functional food’, but market research reporting global sales growth of USD 75 billion to USD 109 billion between 2007 and 2010 indicates the immense value of the market (Sloan, 2008).

Published research regarding consumer preferences and acceptance of functional foods is scarce, although pharmaceutical and biotechnology companies have made substantial investments in researching these (Childs and Poryzees, 1997). Even less information is available regarding consumer preferences for mineral-enriched food products specifically, whether developed by fortification or bio-fortification. Generally, the key drivers of demand for functional foods are held to be burgeoning interest in personalised and preventative health care, wide acceptance of the links between diet and health, and expansion of regulatory infrastructure and scientific research to support product development and verify health benefits (Milner, 2000).

Description of the consumer segment is varied, possibly due to diverse methods and divergent research foci, but is overall circumscribed by socioeconomic, cognitive, and attitudinal determinants. Most consensus is reached on socioeconomic factors, with many reports suggesting that females, parents, and people with ill family members have stronger purchase interest in functional foods; whereas findings regarding the impacts of age bracket and education are variable (Verbeke, 2005). Aging populations may be urged to pursue preventative health care in order to avoid invasive and costly procedures, while declining birth rates may be associated with greater scrutiny of children’s diet and nurture (Tapsell et al., 2005). Beliefs and knowledge about health, and trust in the industry/government sources of nutritional information, are also crucial (Verbeke, 2005).

3.1.3 Research and development of mineral-biofortified seed crops

3.1.3.1 Progress in plant breeding for mineral-biofortified seed crops Traits related to mineral uptake and utilisation have long been of interest to plant breeding programs, though more from the perspective of improving crop nutrition in order to optimise yields, reduce susceptibility to mineral deficiency disorders (such as iron chlorosis), and decrease reliance on soil amendments, rather than to benefit consumers (Ross, 1986). Heightened international focus on the alleviation of poverty and malnutrition, and commercial

| Dietary minerals 92 interest to increase the value-addedness of food products, have led to recent renewed interest in nutritional biofortification. Plant breeding programs that have explicitly published results have mostly been under the umbrella of the CGIAR HarvestPlus Initiative, an international multidisciplinary alliance of about 50 institutions that aims to develop biofortified varieties of key staple crops, with priority given to Fe, Zn, and provitamin A enrichment (Pfeiffer and McClafferty, 2007, Ortiz-Monasterio et al., 2007). The Grand Challenges in Global Health (GCGH) initiative is a second large biofortification consortium, which focuses on micronutrient biofortification of banana, cassava, and sorghum for Africa by genetic modification (Mayer et al., 2008). Some examples of biofortified crop varieties are listed in Table 3.7.

There has been significant pre-breeding research into mineral biofortification, but as yet there have been limited varietal releases. Where biofortification is aimed at alleviation of micronutrient deficiencies, nutrient targets must reflect the levels required to achieve physiological impact in view of typical dietary intake and composition. Collection, assessment, and selection of germplasm are crucial in conventional plant breeding since potential trait expression is dependent on the naturally-occurring variation inherent to the breeding population grown in typical target environments. Design of breeding schemes is influenced by factors such as the biology of the plant species (e.g., self-pollinating or outcrossing), source of the trait (e.g., back-crossing to a recurrent well-established parent can be used for interspecific gene transfer or the development of slightly different ‘isogenic’ lines; different methods have more or less opportunities for genetic recombination), and logistical considerations (e.g., single seed descent schemes save time and space compared to field trials, but have limited recombination and less scope for natural selection). Validated protocols for sampling and analysis underlie the objective scientific evaluation of seeds and plants and, in this regard, cost-effective methods for rapid, non-destructive screening of potentially extensive quantities of segregating breeding material are of great practical importance.

For some crops there have also been concerted efforts directed towards characterising the molecular genetic basis of traits and understanding their expression, with a view to developing marker-assisted selection (MAS) methods and possibly transgenic lines. For example, Brinch- Pedersen (2007) reviewed the current understanding of genetic and biochemical mechanisms to increase Fe, Zn and vitamin availability in cereal grains; Ali et al. (2002) reviewed genetic manipulation of mineral acquisition in chickpea, though less linkage data was available for this more minor crop; and a great deal of quantitative trait loci (QTL) analysis has been performed on rice (Ismail et al., 2007).

| Dietary minerals 93

TABLE 3.7 EXAMPLES OF BIOFORTIFIED CROP VARIETIES

Crop Mineral trait Program reference Transgenic banana for Uganda Increased Fe, provitamin A, or vitamin E Grand Challenges in Global Health (GCGH, 2011) Bean varieties for D. R. Congo and Increased Fe from 50 μg g-1 (baseline) to HarvestPlus (HarvestPlus, Rwanda (spillover to other African 94 μg g-1 2009) nations) – due for release 2012 Cassava varieties for D. R. Congo and Increased provitamin A from 0.5 μg g-1 HarvestPlus (HarvestPlus, Nigeria (spillover to other African (baseline) to 15.5 μg g-1 2009) nations) – due for release 2011 Transgenic cassava varieties – field Increased Fe, Zn, protein, vitamins A and BioCassava Plus, Grand trials in 2007 E; decreased cyanogenic glycosides Challenges in Global Health (GCGH, 2011) Transgenic lettuce – research lines Increased Ca from 14.7 mg g-1 (control) to (Park et al., 2009) 18.6-18.9 mg g-1 Maize varieties for Zambia (spillover to Increased provitamin A from 0.5 μg g-1 HarvestPlus (HarvestPlus, other African nations) – due for release (baseline) to 15.5 μg g-1 2009) 2012 Pearl millet varieties for India (spillover Increased Fe from 47 μg g-1 (baseline) to HarvestPlus (HarvestPlus, to Mali and Niger) – due for release 77 μg g-1 2009) 2012 Rice varieties for Bangladesh and India Increased Zn from 16 μg g-1 (baseline) to HarvestPlus (HarvestPlus, (spillover to other Asian nations) – due 24 μg g-1 2009) for release 2013 Transgenic ‘golden rice’ varieties – field Increased provitamin A (not synthesised (Al-Babili and Beyer, 2005, trials 2008-2009; releases expected in natural rice) of 1.6 μg g-1 in initial Virk and Barry, 2009) 2012-2013 prototypes (GR), to 6 μg g-1 (GR2), and 37 μg g-1 (GR3) in later releases Stacked transgenic ‘golden rice’ Golden rice with additional micronutrient ProVitaMin Rice varieties – field trials in 2007 (Fe, Zn, vitamin E, lysine) and Consortium, Grand bioavailability traits Challenges in Global Health (GCGH, 2011) Transgenic sorghum – in development Increased provitamin A, Fe and Zn African Bio-fortified stages bioavailability (by phytate-silencing), Sorghum Project, Grand lysine, threonine, tryptophan, starch and Challenges in Global Health protein digestibility, vitamin E (GCGH, 2011) Sweet potato varieties for Uganda and Increased provitamin A from 2 μg g-1 HarvestPlus (HarvestPlus, Mozambique (spillover to other African (baseline) to 32 μg g-1 2009) nations) – released in 2007 Wheat varieties for India and Pakistan Increased Zn from 25 μg g-1 (baseline) to HarvestPlus (HarvestPlus, (spillover to Afghanistan, Bangladesh, 33 μg g-1 2009) Egypt, Nepal, and northwest Mexico) – due for release in 2013 3.1.3.2 Factors affecting variation in essential mineral composition of seed crops Considerable research examines nutrient use efficiency (NUE) in plants generally, rather than in seed tissues specifically. NUE is circumscribed by (a) uptake efficiency, particularly root morphology, soil characteristics, and nutrient dynamics affecting influx kinetics into the plant; (b) incorporation efficiency, related to nutrient transport within the plant to shoots and leaves; and (c) utilisation efficiency, or the remobilisation/recirculation of nutrients, influenced mainly

| Dietary minerals 94 by root and shoot parameters (Baligar et al., 2001). Critical variables influencing NUE are reviewed in this section, with an attempt to highlight the interaction between factors under genetic control and environmental influences.

A notable research trend not directly concerned with plant breeding, but that is yielding valuable knowledge regarding plant mineral nutrition that could inform and aid breeding efforts, is the application of bioinformatics type approaches to examination of whole-plant changes in mineral composition. Such study, termed ‘ionomics’, has been defined as the “quantitative and simultaneous measurement of the elemental composition of living organisms and changes in this composition in response to physiological stimuli, developmental state, and genetic modifications” (Salt et al., 2008). Holistic profiling rather than single element measurement has become an important approach to examining nutrient dynamics, as it allows systemic examination of the myriad variables that can affect acquisition of any given mineral and the relationships among these, in view of the interconnectedness of the physiological/biochemical mechanisms and regulatory network governing them.

The extensive analytical screening intrinsic to ionomics became feasible with advances in technologies for high-throughput simultaneous multi-element analysis, especially ICP-OES and ICP-MS. Combination with other `omics – most obviously, with genomics, proteomics, and transcriptomics – presents powerful possibilities for developing a comprehensive understanding of plant nutrition, and has contributed a great deal to current knowledge thus far (for reviews of ionomics, see Baxter, 2009, Baxter, 2010, Salt et al., 2008).

3.1.3.2.1 Root characteristics Nutrient uptake by plant roots occurs via direct contact with nutrients in the soil, termed ‘root interception’; mass flow of soluble nutrients along the water potential gradient driven by root water absorption and transpiration; and passive diffusion of nutrient ions along ion concentration gradients from the soil to the roots (Wang et al., 2006). The efficiency of these mechanisms is greatly affected by root morphology and root system architecture (RSA). RSA describes the spatial composition and configuration of the root system, and thus encompasses root type, topology, and distribution in multiple axes. Important root parameters affecting nutrient uptake are listed in Table 3.8.

| Dietary minerals 95

TABLE 3.8 REVIEW OF ROOT CHARACTERISTICS AFFECTING NUE IN PLANTS

Characteristic Explanation Reference Size of root Root interception is particularly important for less mobile ions such as Cu, Fe, K, Mn, Zn, and especially P. A larger root system will (Gahoonia and Nielsen, 2004) system contribute to increased uptake of these diffusion-limited nutrients. Positive correlations between total root length and yield have

been reported in several crop species, though it should be noted that assumptions regarding cause and effect are confounded by complex variables, e.g., trade-offs in the diversion of assimilates and changes in root: shoot ratios. Selection for root length is considered to be feasible in some species. Root diameter Root diameter determines the volume of soil contacted per unit of root surface area. Some researchers have suggested a trade-off (Gahoonia and Nielsen, 2004) between increased NUE associated with finer roots, and additional cost in photosynthate to maintain the more easily-damaged roots. Root hairs As with root diameter, root hair morphology influences contact at the root-soil interface. Effective root surface area is particularly (Gahoonia et al., 2007, important for uptake of less mobile ions such as P. The genetic basis for root hair formation has been characterised using Gahoonia and Nielsen, 2004, Arabidopsis and barley mutants, and some researchers suggest potential to select for root hair morphology in plant breeding. For Jungk, 2001, Wissuwa and Ae, example, significant genotypic differences in root hair density, as well as ability to form root hairs, have been reported in chickpea 2001) (N=10 genotypes) and peanut (N=18 genotypes). Root system Adventitious root proliferation and basal root (growth angle) shallowness improve topsoil foraging. Surface root proliferation, (Hodge, 2009, Nibau et al., architecture linked to decreased gravitropic response, is correlated generally with P-efficiency in plants, since P tends to concentrate in topsoil. 2008, Ochoa et al., 2006, Shen However, overall benefit for the plant depends on the relative respiratory costs of the respective root types, soil stratification, and et al., 2011, Walk et al., 2006) P mobility because increased adventitious root formation is compensated by decreased basal root proliferation and total root system size, and leads to increased root competition since adventitious roots originate near the basal roots. Lateral root parameters, including lateral root number, density, and growth angle, are also important components of RSA. Lateral root formation, regulation, plasticity (response to environmental factors), and transcriptomics are reviewed in depth by Nibau et | Dietary minerals al. (2008). Hodge (2009) gives an excellent review of RSA regulation, coordination, and competition in response to complex nutrient environments. 96

Genotypic variation in root characteristics, or in nutrient uptake attributed at least partially to root characteristics, has been described in several crop species. For example, genetic differences in root hair density, root hair morphology (i.e., rosette or ‘normal’ long hairs), and ability to form root hairs at all, were observed in peanuts (N=18 genotypes) and considered to be the predominant cause of genotypic differences in P uptake (Wissuwa and Ae, 1999). Similarly, P-efficiency in lupin (Lupinus albus) was attributed to the development of cluster roots with high root hair density; while P-efficient common beans (Phaseolus vulgaris) had increased topsoil root proliferation (Shen et al., 2011) Significant differences in total root length (i.e., system size) and root hair density were reported in chickpea (N=10 genotypes; Gahoonia et al., 2007); while root hair length correlated to K-efficiency in pea (Pisum sativum), red clover (Trifolium pratense), lucerne (Medicago sativa), barley (Hordeum vulgare), rye (Secale cereale), perennial ryegrass (Lolium perenne), and oilseed rape (Brassica napus) (Høgh- Jensen and Pedersen, 2003). Quantitative variation in root traits including adventitious root formation, basal root shallowness, and root hair length and density have been demonstrated in common bean (Phaseolus vulgaris), soybean (Glycine max), Rumex spp., tomato (Lycopersicum esculentum), maize (Zea mays), and rice (Oryza sativa) (Ochoa et al., 2006).

The emergence of tomographic imaging techniques during the past decade for non-invasive analysis of root topology has greatly improved the accuracy and robustness of RSA phenotyping (de Dorlodot et al., 2007). Combining this with genomics/bioinformatics techniques, especially QTL mapping, has helped to elucidate the genetic basis for variation in RSA. To a lesser extent, transcriptome profiling of differential gene activity and proteomic analyses have provided insights on the regulation and hormonal modulation of RSA, and helped to explain response to environmental variables (Osmont et al., 2007). For example, recombinant inbred lines (N=84 F5:8 RIL) generated from two contrasting common bean parents were used for linkage mapping to examine the genetic control of root adaptations in response to P deficiency (Ochoa et al., 2006). Twenty QTL for adventitious root-related traits were identified under either high-P or low-P conditions; moreover, transgressive segregation (i.e., more or less expression than in the parents) in root traits including adventitious root proliferation, biomass, density and specific root length (SRL) were observed.

Manipulation of the root system through plant breeding is considered to be a potential strategy for improving nutrient- and water-use efficiency in crops, but actual implementation of this is constrained by practical difficulties. For example, in identification of RSA breeding targets appropriate to specific environments, and screening large breeding populations in the field.

| Dietary minerals 97

3.1.3.2.2 Nutrient dynamics The evolution in higher plants of complex adaptations to exploit fluctuating soil conditions has resulted in diverse mechanisms of nutrient uptake, translocation, accumulation, and remobilisation that vary among elements, plant species, and developmental states, and are responsive to numerous external factors. Our understanding of the molecular mechanisms underpinning these remains incomplete, particularly for low-flux micronutrients which can be difficult to measure and may not follow the same patterns of movement or regulation as the better-characterised macronutrients. Some of the major variables affecting nutrient dynamics are outlined in Table 3.9.

It is clear that significant inter-element interactions characterise nutrient accumulation in plants. Most obviously, uptake of nutrients through non-specific (low-affinity) processes will be highly influenced by their relative concentrations in the rhizosphere. Well-established empirical examples include competitive uptake among Ca, K, Mg, and Na cations; analogous

+ + - - Na /NH4 and Cl /NO3 antagonisms (Hu and Schmidhalter, 2005); and competition among - 2- 2- - 2- chloride (Cl ), molybdate (MoO4 ), phosphate (HPO4 or H2PO4 ), and sulfate (SO4 ) anions (Rebafka et al., 1993). In addition to coordinated uptake and translocation processes, other factors influencing inter-element interactions include the varying redox potential and pH- dependencies of nutrient bioavailability, diverse plant mechanisms for modulating these in response to soil concentrations, evidence of cross-talk in the regulatory network of certain elements, and nutrient dynamics within the plant itself.

3.1.3.2.3 Soil physical properties Physical constraints such as high bulk-density layers or pans, surface sealing, and crusting cause mechanical impedance to root penetration and adverse changes to root morphology, including stunting, thickening, decrease in the development of higher order lateral roots, loss of root caps, and changes in cellular morphology (Baligar et al., 2001). Such limitations restrict the ability of roots to explore larger and deeper volumes of soil.

3.1.3.2.4 Rhizosphere ecology The important role of rhizobial associations and mycorrhizal symbiosis in biological nitrogen fixation in legumes is well-known. Such plant-microbe interactions not only affect nitrogen nutrition, but also influence the availability and uptake of other elements, as outlined in Table 3.10.

| Dietary minerals 98

TABLE 3.9 REVEW OF FEATURES OF NUTRIENT DYNAMICS AFFECTING NUE IN PLANTS

Characteristic Explanation Reference Uptake Of the essential elements, most is known about the uptake and transport of the major transition metal cations, Ca, K, (Dakora and Phillips, 2002, Graham and mechanisms and Mg. These cations are delivered into the plant primarily by mass-flow of the soil solution; while influx is mediated Stangoulis, 2003, Karley and White, 2009, by a large number of transporter channels. The micronutrients Co, Cu, Fe, Mn, Ni, and Zn are thought generally to be Pilon et al., 2009, Puig and Peñarrubia, absorbed as divalent ions through possibly selective but probably non-specific channels, such as via Ca2+ channels. 2009, Tejada-Jiménez et al., 2009, Welch, Intake through non-selective channels will evidently be propelled by concentration-dependent competition. 1995) On the other hand, some micronutrients (e.g., Cu2+ and Fe3+) may be preferentially bound by negatively-charged ion exchange sites and reactive ligand groups (e.g., carbonyl, hydroxyl, thiol, carboxylic, amide-N, amino-N, and phenolic) in - 2- cell wall constituents of root cells. Cl and Mo are present in soil solutions as anions (Cl and MoO4 ) and require active (proton-coupled) transport into the plant; whereas B and Si are metalloids present as non-charged neutral molecules (boric and silicic acid) and their intake is thought to be diffusion-driven, although B may be actively transported under deficiency conditions. Plant regulation Plant strategies to increase bioavailability include upregulation of high-affinity transport channels; and activation of (Graham and Stangoulis, 2003, Puig and of nutrient intake proton ATPases, metallochelate reductases, and phytosiderophores (nonprotein amino acids that chelate poorly soluble Peñarrubia, 2009, Welch, 1995, White and minerals and thereby facilitate their diffusion to the roots). Reductase (Strategy I) systems are postulated to play a Broadley, 2009) particularly important role in dicotyledonous plants for the influx of micronutrients such as Cu, Fe, and Mn, which are often complexed or bound to metal chelators; whereas phytosiderophores are notable in graminaceous (Strategy II) plants. Root exudates also include a variety of phosphatases that act to hydrolyse and mobilise inorganic P in the rhizosphere, thereby increasing bioavailability to the plant. Some plants release phenolic compounds under conditions of Fe deficiency, which form chelates with the Fe and Al in insoluble Fe- and Al-phosphates and thereby increase the mobility and bioavailability of Fe and P.

| Dietary minerals

99

TABLE 3.9 CONTINUED

Characteristic Explanation Reference Plant Plants also influence nutrient bioavailability by modulating cation-anion balance and root cell proton efflux. (Dakora and Phillips, 2002, Gahoonia et al., modification of Acidification of the rhizosphere helps to solubilise precipitated forms of P, and improves the bioavailability of 2007, Gahoonia and Nielsen, 2004, rhizosphere pH micronutrients such as Fe, Mn, and Zn in calcareous (i.e., high pH) soils. On the other hand, excessive soil acidity (e.g., Hoffland et al., 2006, Kochian, 1995, below pH 5.5) is limiting to all macronutrients and can lead to phytotoxic concentrations of Al, Fe, and Mn. In simplified Kochian et al., 2005, Reid and Hayes, 2003, terms, cation intake in excess of anions stimulates proton efflux, which lowers pH; whereas anion intake in excess of Shen et al., 2011) cations is balanced by extrusion of inorganic ions (e.g., HCO -). Low molecular weight organic acid anions (e.g., oxalate, 3 citrate, and malate) in root exudates also lower rhizosphere pH, and play an important role in plant resistance to P deficiency, and tolerance to low Zn and P availability. Conversely, rhizosphere alkalinisation by root exudates is a mechanism of resistance to Al, Fe, and Mn toxicity.

Gaseous root exudates likewise influence mineral nutrition. CO2 released by growing roots during carbohydrate respiration promotes disassociation of carbonic acid and dissolution of CaCO3, thereby increasing Ca availability (as - direct exudation of HCO3 does). Release of H2 as a by-product of nitrogen fixation boosts yields, apparently by enhancing the energy efficiency of symbiotic rhizobia. Nutrient Long-distance nutrient translocation within plants is broadly described by root-to-shoot xylem transport driven by (Abbo et al., 2008, Karley and White, 2009, translocation negative hydrostatic pressure (water transpiration) and source-to-sink phloem transport driven by positive hydrostatic Waters and Sankaran, 2011, Welch, 1995, pressure (carbohydrate/macronutrient source-loading and sink-unloading). Sap composition and concentration is White and Broadley, 2009, White and influenced by such factors as pH, redox potential, ionic strength, and organic constituents (e.g., organic acids, amino Brown, 2010, Zhang et al., 2007b) acids, ligand groups). As with mineral uptake and homeostasis, the actual mechanisms by which plants regulate loading,

sap transport, and unloading in xylem and phloem vessels involve intricate systems for sensing, signaling, transport, and feedback regulation; our knowledge of these continues to grow.

| Dietary minerals Since delivery of nutrients to non-transpiring or xylem-deficient tissues (e.g., fruits, seeds, and tubers) depends on recirculation via the phloem, the phloem-mobility of minerals is pivotal to seed quality and propagation, and pertinent when discussing the mineral biofortification of seed crops. Ca and Mn are largely phloem-immobile; hence, seed concentrations tend to be low, while accumulation is greatest in highly-transpiring tissues such as leaves. B, Cu, Fe, I, Mo, and Zn are classified as having ‘variable’ or species-dependent phloem-mobility; whereas Cl, K, Mg, Na, N, P, S, and Se are readily translocated via the phloem. 100

TABLE 3.10 REVIEW OF MICROBIAL FACTORS AFFECTING NUE IN PLANTS

Characteristic Explanation Reference Mycorrhizal The hyphae of (vesicular-) arbuscular mycorrhizae (AM) serve to extend the root system, though with trade-offs in (Gahoonia and Nielsen, 2004, Schultz et al., symbiosis additional photosynthate cost. Cavagnaro and co-workers demonstrated variation in the growth and nutrient uptake of 2010, Asghari and Cavagnaro, 2011, AM symbioses in response to heterogeneous nutrient distribution (i.e., patchy soil), termed ‘functional diversity’. Such Cavagnaro et al., 2010, Cavagnaro et al., hyphal proliferation plays an important role in efficient acquisition of P in particular, but also N, Zn, and other nutrients. 2005, Zhu et al., 2001) The extent of AM colonisation has been shown to affect the Ca and Fe concentrations in shoots, Mg in roots and shots, and P in roots of the model legume, Medicago truncatula. The extent of AM colonisation and its influence on plant nutrient uptake is dependent on complex interactions between the AM and plant species composition, soil nutrient distribution, and growing conditions. The genetic basis for AM symbiosis is unclear, though there have been some gene expression studies, and the importance of plant genotypic variation has not been determined. Studies of wheat and barley germplasm did not find any significant genotypic variation in the degree of AM colonisation; whereas a comparison of old and new wheat lines led Zhu et al. (2001) to suggest that plant breeding may have inadvertently reduced responsiveness to AM. Rhizobial Rhizobacteria that produce Nod factors or lipo-chito-oligosaccharides (LCO) can stimulate growth in root mass and (Mehboob et al., 2009) associations length. Rhizobacteria may also produce enzymes that affect root morphology, e.g. ACC-deaminase lowers ethylene levels in the plant, and is associated with increased root length. Many Rhizobium, Pseudomonas, and Bacillus species promote solubilisation of P through production of organic acids or acid phosphatases. P is often immobilised by fixation/precipitation with oxides and hydroxides in acid soils, and Ca in alkaline soils. A review of rhizobial associations in non-legumes (Mehboob et al., 2009) has been published. Plant mediation The association of nitrogen-fixing legumes with rhizobia is well known. The ecology of such interactions is complex; root (Dakora and Phillips, 2002) of microbial exudates contain a wide range of chemicals (e.g., flavonoids, aromatic acids, amino acids, and dicarboxylic acids) that interactions have specific chemo-attractant and signaling properties, not only on rhizobia, but also on pathogens, pathogen-

| Dietary minerals antagonistic rhizobia, AM fungi, mutualistic fungi, and other plants. Thus, the quality of root exudate strongly influences not only the extent of rhizobial (and mycorrhizal) interactions and hence plant nutrient uptake, but also the rhizosphere microbiology and overall competition for soil nutrients. This is reviewed in Dakora and Phillips (2002).

101

3.1.4 Research and development of mineral biofortified peanuts

The development of peanut varieties with functional food traits is a potential avenue for improving the nutritional quality and value-addedness of Australian peanut products. However the baseline data regarding genotypic variation and the relative importance of genetic and environmental influences on the expression of mineral traits has not been established. In fact, no published surveys of mineral composition in domesticated peanut genotypes have been published since Branch and Gaines’ (1983) evaluation of B, Ca, Cu, Fe, K, Mg, Mn, N, P, S, and Zn concentrations by atomic absorption spectrometry and colorimetry (for B, N, P, and S) in 26 genotypes in 1979-81. Nevertheless, we hypothesise useful levels of variation in peanut kernels by extrapolating the available data on grains and other legumes, which are expected to be subject to similar nutrient dynamics. Variation reported in some of the larger scale studies are listed in Table 3.11.

| Dietary minerals 102

TABLE 3.11 REVIEW OF GENOTYPIC VARIATION IN THE ESSENTIAL MINERAL CONCENTRATIONS OF SEVERAL LEGUME AND GRAIN SEED CROPS

Element Genotypes Method Units Concentration Reference Min - Max Mean ± SD RSD (%)

B 27 cowpea (Vigna unguiculata) ICP-MS μg g-1 42 - 55 48 ± 4 9 (Belane and Dakora, 2011) 26 peanut (Arachis hypogaea) Colorimetry μg g-1 9 - 21 14 ± 2.4 17 (Branch and Gaines, 1983) Ca 155 bean (Phaseolus vulgaris) AAS mg g-1 0.7 - 2.9 1.3 ± 0.3 23 (Pinheiro et al., 2010) 27 cowpea (V. unguiculata) ICP-MS mg g-1 12.2 - 27.1 16.7 ± 3.6 21 (Belane and Dakora, 2011) 26 peanut (A. hypogaea) AAS mg g-1 0.4 - 0.8 0.6 ± 0.1 15 (Branch and Gaines, 1983) 274 rice (Oryza sativa)A ICP-MS μg g-1 49 - 283 118 ± 51 43 (Jiang et al., 2008) 45 soybean (Glycine max)B ICP-OES mg g-1 1.9 - 2.8 2.4 ± 0.3 10 (Spehar, 1994) 63 spring wheat (Triticum aestivum ssp. aestivum) ICP-OES μg g-1 288 - 568 418 ± 64 15 (Murphy et al., 2008) 43 spring wheat (T. aestivum ssp. aestivum) ICP-OES μg g-1 237 - 492 332 ± 57 17 (Tang et al., 2008) 57 durum wheat (Triticum turgidum ssp. durum) ICP-OES μg g-1 406 - 639 488 ± 49 10 (Ficco et al., 2009) Cu 155 bean (P. vulgaris) AAS μg g-1 4.7 - 13.5 8.7 ± 1.3 15 (Pinheiro et al., 2010) 27 cowpea (V. unguiculata) ICP-MS μg g-1 8.6 - 19.7 12.7 ± 2.5 20 (Belane and Dakora, 2011) 26 peanut (A. hypogaea) AAS μg g-1 8 - 17 13 ± 2.6 20 (Branch and Gaines, 1983) A -1

| Dietary minerals 274 rice (O. sativa) ICP-MS μg g 3.4 - 21.6 7.5 ± 2.7 36 (Jiang et al., 2008) 45 soybean (G. max)B ICP-OES μg g-1 36.7 - 49.0 43.0 ± 2.8 6 (Spehar, 1994) 63 spring wheat (T. aestivum ssp. aestivum) ICP-OES μg g-1 3.4 - 7.2 4.7 ± 0.7 16 (Murphy et al., 2008) 43 spring wheat (T. aestivum ssp. aestivum) ICP-OES μg g-1 5.2 - 8.1 6.9 ± 0.7 10 (Tang et al., 2008) 57 durum wheat (T. turgidum ssp. durum) ICP-OES μg g-1 5.8 - 14.0 7.5 ± 1.4 18 (Ficco et al., 2009) 103

TABLE 3.11 CONTINUED

Element Genotypes Method Units Concentration Reference Min - Max Mean ± SD RSD (%)

Fe 155 bean (P. vulgaris) genotypes AAS μg g-1 32.2 - 88.4 56.8 ± 11.1 20 (Pinheiro et al., 2010) 27 cowpea (V. unguiculata) genotypes ICP-MS μg g-1 142 - 813 307 ± 142 46 (Belane and Dakora, 2011) 26 peanut (A. hypogaea) genotypes AAS μg g-1 24 - 41 34 ± 5.9 17 (Branch and Gaines, 1983) 274 rice (O. sativa) genotypesA ICP-MS μg g-1 1.2 - 18.6 5.1 ± 3.0 59 (Jiang et al., 2008) 45 soybean (G. max) genotypesB ICP-OES μg g-1 54.8 - 104.2 86.9 ± 8.7 10 (Spehar, 1994) 63 spring wheat (T. aestivum ssp. aestivum) ICP-OES μg g-1 27.1 - 52.2 35.3 ± 4.9 14 (Murphy et al., 2008) 43 spring wheat (T. aestivum ssp. aestivum) ICP-OES μg g-1 21.6 - 33.8 28.2 ± 2.7 10 (Tang et al., 2008) 57 durum wheat (T. turgidum ssp. durum) ICP-OES μg g-1 33.6 - 65.6 43.3 ± 5.8 14 (Ficco et al., 2009) K 155 bean (P. vulgaris) AAS mg g-1 12.4 - 21.2 16.1 ± 1.7 10 (Pinheiro et al., 2010) 27 cowpea (V. unguiculata) ICP-MS mg g-1 26.3 - 36.7 32.1 ± 3.3 10 (Belane and Dakora, 2011) 26 peanut (A. hypogaea) AAS mg g-1 5.2 - 7.1 6.1 ± 0.4 6 (Branch and Gaines, 1983) 274 rice (O. sativa)A ICP-MS μg g-1 429 - 1526 865 ± 210 24 (Jiang et al., 2008) 45 soybean (G. max)B ICP-OES mg g-1 16.2 - 19.8 18.0 ± 1.0 5 (Spehar, 1994) -1

| Dietary minerals 43 spring wheat (T. aestivum ssp. aestivum) ICP-OES μg g 2438 - 4461 3534 ± 396 11 (Tang et al., 2008) 57 durum wheat (T. turgidum ssp. durum) ICP-OES μg g-1 4087 - 5274 4772 ± 223 5 (Ficco et al., 2009) 104

TABLE 3.11 CONTINUED

Element Genotypes Method Units Concentration Reference Min - Max Mean ± SD RSD (%)

Mg 155 bean (P. vulgaris) AAS mg g-1 1.3 - 2.5 1.9 ± 0.2 12 (Pinheiro et al., 2010) 27 cowpea (V. unguiculata) ICP-MS mg g-1 3.5 - 5.7 4.2 ± 0.5 13 (Belane and Dakora, 2011) 26 peanut (A. hypogaea) AAS mg g-1 1.7 - 2.3 2.0 ± 0.1 6 (Branch and Gaines, 1983) 274 rice (O. sativa)A ICP-MS μg g-1 66 - 445 211 ± 72 34 (Jiang et al., 2008) 45 soybean (G. max)B ICP-OES mg g-1 2.2 - 2.7 2.5 ± 0.1 4 (Spehar, 1994) 63 spring wheat (T. aestivum ssp. aestivum) ICP-OES μg g-1 1145 - 1723 1388 ± 120 9 (Murphy et al., 2008) 43 spring wheat (T. aestivum ssp. aestivum) ICP-OES μg g-1 950 - 1501 1289 ± 103 8 (Tang et al., 2008) 57 durum wheat (T. turgidum ssp. durum) ICP-OES μg g-1 1056 - 1417 1171 ± 67 6 (Ficco et al., 2009) Mn 155 bean (P. vulgaris) AAS μg g-1 8.0 - 20.1 12.6 ± 2.6 21 (Pinheiro et al., 2010) 27 cowpea (V. unguiculata) ICP-MS μg g-1 196 - 457 289 ± 58 20 (Belane and Dakora, 2011) 26 peanut (A. hypogaea) AAS μg g-1 13 - 20 16 ± 1.5 9 (Branch and Gaines, 1983) 274 rice (O. sativa)A ICP-MS μg g-1 6.1 - 16.3 9.0 ± 2.0 22 (Jiang et al., 2008) 45 soybean (G. max)B ICP-OES μg g-1 18.9 - 30.4 25.4 ± 2.6 10 (Spehar, 1994) -1

| Dietary minerals 63 spring wheat (T. aestivum ssp. aestivum) ICP-OES μg g 39.0 - 62.0 49.4 ± 5.0 10 (Murphy et al., 2008) 43 spring wheat (T. aestivum ssp. aestivum) ICP-OES μg g-1 19.7 - 36.8 29.4 ± 4.0 14 (Tang et al., 2008) 57 durum wheat (T. turgidum ssp. durum) ICP-OES μg g-1 41.3 - 57.4 47.8 ± 3.7 8 (Ficco et al., 2009) Mo 45 soybean (G. max)B ICP-OES μg g-1 0.4 - 1.6 0.9 ± 0.4 36 (Spehar, 1994) Na 274 rice (O. sativa)A ICP-MS μg g-1 7.3 - 79.9 26.6 ± 16.4 62 (Jiang et al., 2008) 45 soybean (G. max)B ICP-OES μg g-1 13.0 - 23.8 17.5 ± 3.5 19 (Spehar, 1994) -1 105 57 durum wheat (T. turgidum ssp. durum) ICP-OES μg g 19.2 - 37.7 26.1 ± 3.7 14 (Ficco et al., 2009)

TABLE 3.11 CONTINUED

Element Genotypes Method Units Concentration Reference

Min - Max Mean ± SD RSD (%)

P 155 bean (P. vulgaris) Colorimetry mg g-1 5.0 - 8.0 5.1 ± 0.6 12 (Pinheiro et al., 2010) 27 cowpea (V. unguiculata) ICP-MS mg g-1 2.8 - 7.4 4.8 ± 1.0 20 (Belane and Dakora, 2011) 26 peanut (A. hypogaea) Colorimetry mg g-1 3.8 - 5.5 4.4 ± 0.4 8 (Branch and Gaines, 1983) 45 soybean (G. max)B ICP-OES mg g-1 4.3 - 5.8 5.2 ± 0.3 6 (Spehar, 1994) 63 spring wheat (T. aestivum ssp. aestivum) ICP-OES μg g-1 3092 - 4738 3753 ± 311 8 (Murphy et al., 2008) 43 spring wheat (T. aestivum ssp. aestivum) ICP-OES μg g-1 2279 - 3535 3042 ± 253 8 (Tang et al., 2008) Se 63 spring wheat (T. aestivum ssp. aestivum) AAS μg g-1 2.0 - 30.0 15.3 ± 5.3 35 (Murphy et al., 2008) Zn 155 bean (P. vulgaris) AAS μg g-1 11.5 - 45.3 33.8 ± 5.2 15 (Pinheiro et al., 2010) 27 cowpea (V. unguiculata) ICP-MS μg g-1 49 - 125 84 ± 13 15 (Belane and Dakora, 2011) 26 peanut (A. hypogaea) AAS μg g-1 25 - 41 32 ± 4.1 13 (Branch and Gaines, 1983) 274 rice (O. sativa)A ICP-MS μg g-1 19.8 - 43.7 26.8 ± 3.9 15 (Jiang et al., 2008) 45 soybean (G. max)B ICP-OES μg g-1 6.8 - 11.0 8.5 ± 1.1 13 (Spehar, 1994) 63 spring wheat (T. aestivum ssp. aestivum) ICP-OES μg g-1 23.0 - 43.0 33.0 ± 4.3 13 (Murphy et al., 2008) -1

| Dietary minerals 43 spring wheat (T. aestivum ssp. aestivum) ICP-OES μg g 21.2 - 34.8 28.6 ± 3.1 11 (Tang et al., 2008) 57 durum wheat (T. turgidum ssp. durum) ICP-OES μg g-1 28.5 - 46.3 33.9 ± 2.8 8 (Ficco et al., 2009)

A Milled rice (N=69) results; see article for glutinous vs. non-glutinous, japonica vs. indica, and white vs. red vs. black rice B Full-limed trial results; see article for partially limed acid soils 106

3.2 Genotypic variation in concentrations of essential minerals in peanut kernels

3.2.1 Objectives

The primary objective of this study was to evaluate the extent of genotypic variation in essential mineral concentrations among a representative sample population from the APBP. We also sought to identify breeding lines that contained diverse concentrations of minerals for use in the subsequent G × E study and in further research.

The results of the genotype study were published by Phan-Thien et al. (2012).

3.2.2 Materials and methods

3.2.2.1 Instruments Samples were mineralised in an Ethos Plus Labstation microwave system (Milestone Inc., USA) in PTFE vessels. Elemental analyses were performed on an Optima 7300 DV ICP-OES (PerkinElmer SCIEX, USA) as described in chapter 2.1.2.1, using WinLab32 for ICP software version 4.0.0.0305.

3.2.2.2 Reagents The reagents were analytical grade as described in chapter 2.1.2.2.

3.2.2.3 Samples Peanut kernels were obtained from the APBP, collaboratively run by the Peanut Company of Australia (PCA), Grains Research and Development Corporation (GRDC), and the Queensland Department of Employment, Economic Development and Innovation (DEEDI). These comprised 32 full-season maturity (140 days; Arachis hypogaea ssp. hypogaea) and 24 ultra-early maturity (100-110 days; A. hypogaea ssp. fastigiata) genotypes from the APBP, which represented current commercially-grown varieties in addition to new breeding lines progressing for prospective release (Table 3.12).

The peanuts were grown in 2007/08 in triplicate randomly-allocated field plots at Kairi Research Station in the Atherton Tableland region of northern Queensland, Australia under non-limiting (nutrient- and water-replete) conditions. Sound mature jumbo and size one (i.e., >11.9 mm for Virginia market type; >9.9 mm for Runner market type; and >9.5 in kernel diameter for ultra-early) peanut kernels were sent to UNSW after harvesting and pre- processing in accordance with commercial practice. Specifically, peanuts were threshed, dried to approximately 7% moisture content, shelled (de-hulled), and graded on pilot-scale

| Dietary minerals 107 equipment at the DEEDI Kingaroy research station. The raw kernels were stored in vacuum- sealed polyethylene bags in a cool room (<10°C) prior to analysis.

TABLE 3.12 PEANUT GENOTYPES ANALYSED IN GENOTYPE STUDY OF ESSENTIAL MINERAL CONCENTRATIONS

Full-season genotypes Ultra-early genotypes Genotype Source program Genotype Source program

Ashton DEEDI D136-p7-5 DEEDI AT43 US private program D174-3-p8-7 DEEDI Bruce University of Florida D174-7-p19-4 DEEDI Cook University of Georgia D175-3-p17-3 DEEDI D147-p3-115 DEEDI D188-1-p371-1 DEEDI D147-p8-3H DEEDI D188-1-p371-2 DEEDI D147-p8-6F DEEDI D189-p8-2 DEEDI D164-p39-2 DEEDI D189-p8-3 DEEDI D164-p50-4 DEEDI D191-p1-1 DEEDI D197-9-p586-2 DEEDI D191-p1-3 DEEDI D204-p6-3 DEEDI D191-p13-1 DEEDI Farnsfield US private program D192-p386-1 DEEDI Fisher North Carolina State University D192-p394-1 DEEDI Holt University of Florida D192-p394-2 DEEDI Lyons University of Florida D192-p394-3 DEEDI Middleton DEEDI D192-p397-1 DEEDI Page University of Florida D193-p3-6 DEEDI PCA048 US private program D194-22-p209-2 DEEDI PCA051 US private program D194-24-p362-1 DEEDI PCA213 US private program D194-26-p136-2 DEEDI Reid University of Georgia D194-2-p65-1 DEEDI Scullin University of Georgia D194-2-p65-2 DEEDI Sutherland DEEDI Tingoora DEEDI UF11 University of Florida Walter DEEDI UF16 University of Florida UF31 University of Florida UF32 University of Florida UF36 University of Florida UF37 University of Florida UF38 University of Florida UF39 University of Florida Wheeler DEEDI

| Dietary minerals 108

3.2.2.4 Mineralisation The peanut samples were prepared and mineralised by closed acid digestion as described in chapter 1.1.1.1.

3.2.2.5 Quality controls Quality controls for the digestion and ICP-OES analysis were as described in chapter 2.1.2.5.

3.2.2.6 Moisture content Sample moisture content was determined gravimetrically as described in chapter 2.1.2.6 and used to convert the essential mineral concentrations to a dry weight basis.

3.2.2.7 Data analysis The validation of the analytical methodology is reported in depth in chapter 2.1. Genotypic variation was described as the RSD of the mean essential mineral concentrations within each set of peanut genotypes. The significance of genotype on essential mineral concentrations was examined by standard analysis of variance (ANOVA) in IBM SPSS Statistics software version 20.0.0. The covariance of analyte concentrations was assessed using Pearson’s Correlations coefficient. Genotypic differences were investigated in the raw and normalised (i.e., with mean=0 and SD=1) essential mineral concentrations. Summation of the normalised concentrations yielded a dummy variable that was considered indicative of overall mineral density.

3.2.3 Results and discussion

3.2.3.1 Genotypic variation in essential mineral concentrations Full-season and ultra-early maturity breeding lines are different subspecies of A. hypogaea and were analysed separately because they have distinct growing seasons, morphology (esp. seed size), and end uses. Combining the data generates wider ranges and higher coefficients of variation in mineral concentrations, but distinguishing hypogaea and fastigiata subspecies had more practical relevance for the breeding program.

The genotypic variation in essential mineral concentrations of the APBP peanut samples are presented in Table 3.13 and depicted graphically in Figure 3.1. The data were normalised to examine the relative distribution of mineral concentrations regardless of their magnitude (Figure 3.2). The normalised data were used to rank genotypes in order to select diverse samples for further research.

In general, greater genotypic variation indicates greater opportunity to influence phenotypic expression by conventional plant selection, although the success of breeding efforts is of course affected by many factors including the heritability of each mineral trait. The | Dietary minerals 109 experimental results revealed useful levels of variation of more than 10% RSD for most of the elements tested in either the full-season or ultra-early sample set, if not both (Wright, 2009). Data was generated from one trial location in one season and results may have been inflated by G × E effects. Nevertheless, the level of variation was sufficient to justify further investigation of breeding potential, particularly for Ca and Mn, which varied by 13% and 18% across the full-season lines, and 18% and 17% across the ultra-early lines, and were also present in relatively high concentrations.

The paucity of data concerning genotypic variation in the mineral composition of peanut kernels was mentioned in the literature review (chapter 3.1). The published data includes only one survey of 26 breeding lines (Branch and Gaines, 1983), and an evaluation of 20 Ghanaian landraces (Asibuo et al., 2008). Other studies were less concerned with genotypic trait variation and involved much smaller numbers of samples. For example, a range of traits including mineral composition were compared in eight genotypes grown under induced drought stress (Conkerton et al., 1989); six high-oleic cultivars (Jonnala et al., 2005b); and three transgenic lines (Jonnala et al., 2005a).

Results from Branch and Gaines (1983) and Asibuo et al. (2008) are reproduced in Table 3.14 for convenient comparison with those of our study, although it should be remembered that these other researchers employed AAS for elemental analysis (and colorimetry for P). Variation in Ca, Mg, and P concentrations were somewhat similar to the 1983 study, while there was lower variation in Fe, and higher variation in K and Mn. The Ghanaian peanuts had quite distinct mineral profiles to our APBP samples and those of the US study, with markedly elevated concentrations of Cu, K, Mg, Na, and Zn reported.

| Dietary minerals 110

(A) 1.5 15 25 10 2.5 50 8.0 40 20 8 2.0 40 6.0 1.0 10 30 15 6 1.5 30 dw dw dw dw dw dw dw dw -1 -1 -1 -1 -1 -1 -1 -1 4.0 Q3 + Mean 20 g g g g g g g g 10 4 1.0 20

P P P P Q  Median (Q2) mg g mg g mg g mg g mg 1 0.5 5 2.0 1.5 IQR 5 2 0.5 10 10

0.0 0 0 0 0.0 0 0.0 0 Ca Cu Fe K Mg Mn P Zn

(B) 1.5 15 25 10 2.5 50 8.0 40 20 8 2.0 40 6.0 1.0 10 30 15 6 1.5 30 dw dw dw dw dw dw dw dw -1 -1 -1 -1 -1 -1 -1 -1 4.0 Q3 + Mean 20 g g g g g g g g 10 4 1.0 20 P P P P Q  Median (Q2) mg g mg mg g mg g mg g mg 1 0.5 5

| Dietary minerals 2.0 1.5 IQR 5 2 0.5 10 10

0.0 0 0 0 0.0 0 0.0 0 Ca Cu Fe K Mg Mn P Zn

FIGURE 3.1 VARIATION IN ESSENTIAL MINERAL CONCENTRATIONS OF (A) 32 FULL-SEASON AND (B) 24 ULTRA-EARLY PEANUT GENOTYPES. 111

4 4

2 2

0 0 Mean r 1 SD

Q -2 -2 3 Concentration ~ N (0,1) N ~ Concentration Median (Q2)  Q1 1.5 IQR -4 -4 Ca Cu Fe K Mg Mn P Zn Ca Cu Fe K Mg Mn P Zn Full-season data Ultra-early data

FIGURE 3.2 VARIATION IN NORMALISED ESSENTIAL MINERAL CONCENTRATIONS OF 32 FULL-SEASON AND 24 ULTRA-EARLY PEANUT GENOTYPES. | Dietary minerals 112

-1 TABLE 3.13 VARIATION IN ESSENTIAL MINERAL CONCENTRATIONS (μG G DW) OF 32 FULL-SEASON AND 24 ULTRA-EARLY PEANUT GENOTYPES

Element λ (nm) Full-season genotypes Ultra-early genotypes

Min - Max Mean ± SD RSD Min - Max Mean ± SD RSD (%) (%)

Ca 422.673 475 - 739 619 ± 78 13 370 - 714 499 ± 88 18

Cu 327.393 4.5 - 8.3 6.6 ± 0.9 14 4.2 - 7.0 5.8 ± 0.7 12

Fe 238.204 9.7 - 15.1 12.4 ± 1.1 9 12.6 - 17.0 15.0 ± 1.2 8

K 766.490 5252 - 8151 6371 ± 708 11 6254 - 7920 6941 ± 417 6

Mg 279.077 1355 - 1818 1495 ± 126 8 1529 - 1834 1675 ± 81 5

Mn 257.610 17.3 - 33.5 24.0 ± 4.4 18 11.9 - 26.1 20.8 ± 3.4 17

Na 589.592 2.8 - 25.5 11.0 ± 4.5 41 4.4 - 10.9 7.0 ± 1.7 24

P 178.221 2879 - 4224 3518 ± 329 9 4300 - 6149 5207 ± 511 10

Zn 206.200 23.5 - 32.7 26.0 ± 2.1 8 24.0 - 35.1 29.4 ± 3.1 11

TABLE 3.14 PUBLISHED VARIATION IN ESSENTIAL MINERAL CONCENTRATIONS OF PEANUT GENOTYPES

Element Units Branch and Gaines (1983) Asibuo et al. (2008) Min - Max Mean ± SD RSD (%) Min - Max Mean ± SD RSD (%)

Ca mg g-1 0.4 - 0.8 0.6 ± 0.1 15 0.4 - 1.3 0.8 ± 0.3 33 Cu μg g-1 8 - 17 13 ± 2.6 20 17.0 - 27.0 20.3 ± 2.8 14 Fe μg g-1 24 - 41 34 ± 5.9 17 2.0 - 37.0 28.3 ± 8.2 29 K mg g-1 5.2 - 7.1 6.1 ± 0.4 6 11.8 - 16.9 13.6 ± 1.4 11 Mg mg g-1 1.7 - 2.3 2.0 ± 0.1 6 3.1 - 4.6 3.7 ± 0.4 10 Mn μg g-1 13 - 20 16 ± 1.5 9 15.0 - 29.0 20.9 ± 3.3 16 Na mg g-1 0.2 - 0.5 0.3 ± 0.1 23 P mg g-1 3.8 - 5.5 4.4 ± 0.4 8 Zn μg g-1 25 - 41 32 ± 4.1 13 44.0 - 65.0 54.8 ± 5.7 10 3.2.3.2 Comparison of genotypes and selection for further study Essential mineral concentrations were clearly influenced by genotype, with statistical significance (P<0.05) established for each element in either the full-season or ultra-early sample set, and for all elements when sample sets were combined (Table 3.15). In order to select genotypes with diverse mineral phenotypes for subsequent study, the concentration data was normalised (mean-centred and auto-scaled) in order to highlight genotypes that fell at the statistical extremes of the population distribution. Given the number of analytes, a dummy variable was also created through summation of the normalised data that was considered to represent overall mineral density. These rankings were used to select five full- season (Table 3.16) and four ultra-early (Table 3.17) peanut genotypes that varied both in

| Dietary minerals 113 concentration of individual analytes and in overall mineral density: D147-p3-115, Middleton, Page, PCA213, Sutherland, D175-p17-3, D192-p397-1, D193-p3-6, and Tingoora.

TABLE 3.15 SIGNIFICANCE OF GENOTYPE ON ESSENTIAL MINERAL CONCENTRATIONS

Element ANOVA P-values Full-season data Ultra-early data Combined data

Ca 0.007** 0.002** <0.001** Cu <0.001** <0.001** <0.001** Fe <0.001** 0.005** <0.001** K <0.001** <0.001** <0.001** Mg <0.001** 0.661 <0.001** Mn <0.001** 0.052 <0.001** P 0.006** 0.678 <0.001** Zn 0.133 <0.001** <0.001** * Significant (P<0.05) **Highly significant (P<0.01) 3.2.4 Conclusion

32 full-season maturity (A. hypoagea ssp. hypogaea) and 24 ultra-early maturity (A. hypogaea ssp. fastigiata) genotypes from the APBP were evaluated for the essential dietary minerals, Ca, Cu, Fe, K, Mg, Mn, P, and Zn, by closed acid digestion and ICP-OES analysis. Useful levels of genotypic variation of more than 10% RSD in the concentrations of most elements were found in either the full-season or ultra-early sample set. The results affirmed plant breeding potential for mineral phenotypes in the Australian peanut breeding population and suggests that further research will be worthwhile. The data generated in this study was a useful addition to our understanding of variability in the mineral accumulation of peanuts, given the paucity of data that has been published on the subject. A subset of five full-season and four ultra-early maturity genotypes displaying diverse mineral profiles were selected for subsequent study.

| Dietary minerals 114

TABLE 3.16 FULL-SEASON GENOTYPES RANKED BY NORMALISED ESSENTIAL MINERAL CONCENTRATIONS

Genotype Mean analyte concentrations (μg g-1 dw) Mineral

Ca Cu Fe K Mg Mn P Zn density

Ashton 647 8.6 13.8 6214 1732 25.5 3770 30.5 5.1

AT43 747 6.7 13.4 7325 1651 32.4 3988 28.3 4.2

Bruce 786 5.5 11.8 5585 1449 24.9 3062 25.4 -5.6

Cook 729 6.4 13.7 6920 1832 35.4 4325 29.0 7.8

D147-p3-115 577 6.2 12.0 8220 1441 24.7 4099 26.1 -3.1

D147-p8-3H 694 6.8 13.5 7738 1663 24.9 3735 27.4 1.2

D147-p8-6F 683 7.4 14.4 8020 1486 24.0 3465 26.2 1.1

D164-p39-2 714 7.2 12.6 6248 1772 26.3 3693 29.2 2.8

D164-p50-4 728 7.1 12.7 7374 1570 20.1 3893 25.4 -0.2

D197-9-p586-2 537 8.5 14.5 6847 1826 25.5 4505 31.2 7.5

D204-p6-3 638 6.1 11.5 6828 1486 23.5 3453 26.0 -5.2

Farnsfield 679 6.9 12.9 6788 1550 30.6 3393 27.9 -1.6

Fisher 509 6.1 12.8 7157 1572 20.8 3719 27.2 -3.1

Holt 759 7.1 14.8 6638 1462 20.2 3209 25.0 -3.4

Lyons 637 8.1 13.5 6314 1597 31.5 3443 27.8 0.6

PCA048 559 7.8 14.6 6239 1619 28.2 3814 31.0 2.8

PCA051 709 7.7 13.1 5970 1476 20.1 3570 26.9 -3.4

PCA213 640 8.7 16.1 6805 1847 33.9 4109 34.8 12.2

Middleton 733 8.9 14.4 7485 1941 25.3 4455 30.4 12.5

Page 761 7.7 13.5 5803 1460 19.7 3273 26.2 -4.5

Reid 739 6.5 13.8 7268 1616 35.6 3873 28.7 5.3

Scullin 731 6.2 14.3 5691 1489 21.1 3762 24.9 -1.6

Sutherland 517 7.0 13.7 8593 1484 25.0 4250 29.0 2.0

UF11 712 7.5 12.4 7257 1624 28.9 3600 27.0 0.3

UF16 666 6.2 12.0 6122 1441 25.2 3657 27.4 -5.0

UF31 708 8.2 11.9 7272 1547 20.6 3918 25.5 -0.6

UF32 620 6.7 12.6 6355 1569 26.1 3790 26.6 -1.3 Selected for G × E study

UF36 590 6.2 12.8 6536 1472 26.4 3350 26.4 -3.5

UF37 512 6.5 13.4 6415 1495 25.2 3472 27.2 -3.8 >95% of distribution

UF38 685 6.0 12.4 5862 1490 22.8 3742 25.5 -6.6 >68% of distribution

UF39 618 5.9 12.9 5938 1532 18.4 3489 25.9 -5.0 <68% of distribution

Wheeler 510 4.8 10.3 6881 1666 22.7 3627 27.3 -7.8 <95% of distribution

| Dietary minerals 115

TABLE 3.17 ULTRA-EARLY GENOTYPES RANKED BY NORMALISED ESSENTIAL MINERAL CONCENTRATIONS

Genotype Mean analyte concentrations (μg g-1 dw) Mineral

Ca Cu Fe K Mg Mn P Zn density

D136-p7-5 42 5.3 17.4 7865 1835 28.1 6357 30.1 3.4

D174-3-p8-7 46 5.8 15.8 7200 1645 12.8 4995 25.8 -9.1

D174-7-p19-4 64 6.8 16.5 6731 1759 17.3 5396 28.9 -1.9

D175-3-p17-3 76 7.5 18.2 7187 1782 18.9 4995 32.1 4.0

D188-1-p371-1 66 7.2 18.2 6923 1635 19.0 4598 26.5 -3.0

D188-1-p371-2 56 6.5 14.9 6823 1651 21.7 5034 28.5 -5.1

D189-p8-2 47 5.7 14.9 6763 1739 19.7 5462 29.1 -6.8

D189-p8-3 55 6.8 17.1 7332 1860 23.2 6280 32.1 2.9

D191-p1-1 59 5.9 17.2 7575 1854 24.7 6210 35.7 5.0

D191-p1-3 54 5.5 16.5 7565 1779 26.6 5285 35.7 1.1

D191-p13-1 52 4.5 13.4 6955 1753 21.1 5109 32.5 -8.4

D192-p386-1 60 6.0 15.9 7080 1826 23.1 5715 35.7 1.6

D192-p394-1 43 6.6 17.2 8484 1896 25.5 6171 32.9 5.7

D192-p394-2 47 5.9 14.2 7956 1836 23.5 5367 29.8 -2.4

D192-p394-3 48 7.4 16.4 8107 1795 20.9 5221 29.6 1.0

D192-p397-1 48 6.1 17.3 7665 1862 24.3 6401 37.5 7.3

D193-p3-6 55 6.0 14.7 7442 1668 20.5 5136 29.4 -5.4

D194-22-p209- 39 6.0 17.6 7347 1829 25.8 6597 36.0 4.4

D194-24-p362- 46 6.3 15.2 7554 1795 24.7 5561 30.5 -1.5 Selected for G × E study

D194-26-p136- 39 5.6 14.1 7416 1694 16.4 5452 27.8 -8.2

D194-2-p65-1 59 7.4 16.1 7664 1862 27.3 5522 32.2 5.1 >95% of distribution

D194-2-p65-2 50 6.6 16.9 7240 1899 24.4 5052 34.7 1.9 >68% of distribution

Tingoora 51 5.6 15.8 7460 1965 21.2 6256 34.1 4.5 <68% of distribution

Walter 70 6.4 14.9 7945 1807 22.8 5657 27.6 3.8 <95% of distribution

| Dietary minerals 116

3.3 G × E interaction affecting concentrations of essential minerals in peanut kernels

3.3.1 Objectives

The principal objective of the G × E study was to evaluate the stability of peanut mineral phenotypes by determining the significance of genotype, environment, and G × E interaction components of variation on essential mineral concentrations.

The results of this study were published by Phan-Thien et al. (2010).

3.3.2 Materials and methods

3.3.2.1 Instruments Samples were digested in an Ethos Plus Labstation (10-vessel capacity) or Ethos 1 Advanced (24-vessel capacity) microwave digestion system (Milestone Inc., USA) in PTFE vessels. Elemental analyses were performed on Optima 7300 DV ICP-OES and Elan DRC II ICP-MS (PerkinElmer SCIEX, USA) instruments as described in 2.1.2.1. Data were analysed using WinLab32 for ICP software version 4.0.0.0305 and Elan software version 3.4, respectively.

3.3.2.2 Reagents The reagents were analytical grade as described in chapter 2.1.2.2.

3.3.2.3 Samples Nine peanut genotypes (Table 3.18) were selected to represent the phenotypic diversity in essential mineral composition of the APBP breeding population as described in chapter 3.2.3.2. The peanuts were grown in five distinct growing environments as part of the PCA-GRDC-DEEDI variety evaluation trials. These were situated at DEEDI research stations in the primary peanut production zones of Australia, namely, Bundaberg in the coastal Burnett region of south Queensland, Taabinga in the south Burnett, and Kairi in the Atherton Tableland region of north Queensland.

At Kairi, peanuts were subjected to differing conditions of stress: controlled (low disease pressure) conditions, soil-borne disease pressure (mainly ‘Black Root Rot’ caused by Cylindrocladium parasiticum Crous, Wingfield & Alfenas), and foliar disease pressure (mainly ‘Leaf Spot’ and ‘Leaf Rust’ associated with Cercosporidium personatum (Berk. & Curt) Deighton, Cercospora arachidicola Hori, and Puccinia arachidis Speg.). These latter environments were designated as Kairi, Kairi CBR, and Kairi FDR, respectively. The peanuts were grown in 2007/08 in triplicate/quadruplicate randomly-allocated field plots, pre- processed, and stored as described in chapter 3.2.2.3. | Dietary minerals 117

TABLE 3.18 PEANUT GENOTYPES ANALYSED IN G × E STUDY OF ESSENTIAL MINERAL CONCENTRATIONS

Genotype Maturity Source Registration Genetic relatednessA program D147-p3-115 Full-season DEEDI Breeding line Sister-line to Sutherland

D175-3-p17-3 Ultra-early DEEDI Breeding line Unrelated

D192-p397-1 Ultra-early DEEDI Breeding line Unrelated

D193-p3-6 Ultra-early DEEDI Breeding line Sister-line to Tingoora

Middleton Full-season DEEDI PBR: Plant Varieties Journal, 2003, 16 (3) Unrelated

Page Full-season Uni Florida PBR: Plant Varieties Journal, 2009, 22 (3) Unrelated

PCA213 Full-season US private Breeding line Unrelated

Tingoora Ultra-early DEEDI PBR: Plant Varieties Journal, 2010, 23 (4) Sister-line to D193-p3-6

Sutherland Full-season DEEDI PBR: Plant Varieties Journal, 2007, 20 (2) Sister-line to D147-p3-115

A Sister-lines were derived from the same F2 plant 3.3.2.4 Mineralisation The peanut samples were prepared and mineralised by closed acid digestion as described in chapter 1.1.1.1.

3.3.2.5 Quality controls Quality controls for the digestion, ICP-OES, and ICP-MS analyses were as described in chapter 2.1.2.5.

3.3.2.6 Moisture content Sample moisture content was determined gravimetrically as described in chapter 2.1.2.6 and used to convert the essential mineral concentrations to a dry weight basis.

3.3.2.7 Data analysis Validation statistics for the analytical methodology are reported in chapter 2.1. Genotype, environment, and G × E interaction effects on the essential mineral composition were analysed as a complete factorial design by standard ANOVA in IBM SPSS Statistics software version 20.0.0. Tukey’s HSD test was used to make between-level comparisons of genotypes and environments. Multivariate relationships between peanut genotypes, environments, and essential mineral composition were explored by principal components analysis (PCA) of auto- scaled (i.e., grand mean-centred) and environment-scaled (i.e., environment mean-centred) data in The Unscrambler v9.8 (CAMO Software AS, 1986-2008).

| Dietary minerals 118

3.3.3 Results and discussion

3.3.3.1 Importance of genotype, environment, and G × E interaction Peanuts are bred conventionally in Australia, so new genetic combinations can arise only from the natural processes of sexual recombination and segregation that occur in a heterozygous breeding population. Variation is broadly considered to have three components, i.e. genetic/genotypic, environmental, and G × E interaction. G × E interaction introduces uncertainty into the selection process and therefore influences estimation of trait heritability and response to selection. There is no consensus on the best way to manage G × E interaction in plant breeding programs. This is a major research topic in agricultural genetics, and a range of predominantly statistical strategies of characterising G × E interaction and selecting genotypes are proposed in literature (Cooper and De Lacy, 1994).

TABLE 3.19 SIGNIFICANCE OF GENOTYPE, ENVIRONMENT, AND G × E INTERACTION ON ESSENTIAL MINERAL CONCENTRATIONS

Analyte ANOVA P-values Genotype main effect Environment main effect G × E interaction B 0.001** <0.001** 0.110 Ca <0.001** 0.003** <0.001** Co 0.341 0.113 0.088 Cr 0.852 0.694 0.740 Cu <0.001** <0.001** <0.001** Fe <0.001** <0.001** 0.092 K <0.001** <0.001** <0.001** Mg <0.001** <0.001** 0.057 Mn <0.001** <0.001** 0.001** Mo <0.001** <0.001** <0.001** Na <0.001** <0.001** <0.001** Ni <0.001** <0.001** 0.581 P <0.001** <0.001** <0.001** Se 0.004** <0.001** 0.385 Zn <0.001** <0.001** <0.001** * Significant (P<0.05) ** Highly significant (P<0.01)

| Dietary minerals 119

TABLE 3.20 ESSENTIAL MINERAL CONCENTRATIONS OF PEANUT GENOTYPES TESTED IN THE G × E STUDY

Genotype N Concentration (μg g-1 dw)A B Ca Co Cr Cu Fe K Mg

D147-p3-115 17 16.2 b 570 cde 0.114 a 0.021 a 6.37 de 14.3 e 8294 a 1595 cd D175-3-p17-3 17 23.7 a 768 a 0.085 a 0.024 a 7.73 bc 19.0 ab 6831 e 1867 b D192-p397-1 17 21.2 ab 502 e 0.089 a 0.021 a 6.59 de 20.1 a 7558 cd 1879 ab D193-p3-6 17 18.3 ab 565 cde 0.135 a 0.021 a 5.65 e 16.8 abcd 7317 d 1825 b Middleton 17 19.6 ab 667 b 0.085 a 0.017 a 9.24 a 16.4 bcd 7838 b 1813 b Page 17 16.2 b 630 bc 0.098 a 0.047 a 7.86 bc 16.8 bcd 6315 f 1690 c PCA213 17 18.0 ab 591 cd 0.105 a 0.015 a 8.35 ab 17.0 abcd 6608 e 1675 c Sutherland 20 16.8 b 553 de 0.102 a 0.017 a 7.17 cd 15.2 de 8314 a 1576 d Tingoora 17 21.7 ab 524 de 0.067 a 0.019 a 6.69 d 17.9 abc 7710 bc 1972 a

Genotype N Concentration (μg g-1 dw)A Mn Mo Na Ni P Se Zn D147-p3-115 17 24.1 a 0.225 c 36.2 b 7.71 ab 3669 c 0.088 ab 27.0 b D175-3-p17-3 17 20.0 cd 0.446 b 13.3 c 4.69 bc 3803 c 0.111 a 28.0 b

| Dietary minerals D192-p397-1 17 23.7 a 0.376 bc 12.2 c 4.98 abc 4425 a 0.110 a 31.8 a D193-p3-6 17 22.3 abc 0.383 bc 14.0 c 5.64 abc 4153 b 0.115 a 27.6 b Middleton 17 17.9 d 0.748 a 16.8 c 4.43 c 3726 c 0.096 ab 29.7 ab Page 17 21.4 abc 0.299 bc 15.3 c 7.41 abc 3679 c 0.079 ab 26.9 b PCA213 17 22.8 abc 0.260 c 16.1 c 7.94 a 3664 c 0.077 ab 26.8 b Sutherland 20 23.4 ab 0.772 a 45.3 a 6.24 abc 3790 c 0.063 b 29.6 ab

120 Tingoora 17 20.4 bcd 0.382 bc 17.4 c 6.07 abc 4501 a 0.115 a 28.8 ab

A Lower-case letters mark significantly different (P<0.05) genotype means according to Tukey’s HSD test

Partitioning of total variation into genotype, environment, interaction, and error components is widely practised using standard ANOVA or alternatives that emphasise the extent to which the interaction is due to heterogeneous environmental variance in different genotypes or heterogeneous genetic variance in different environments (Muir et al., 1992). The latter type of G × E interaction most impedes genotype selection because it can lead to re-ranking in different environments. In this study, standard ANOVA revealed that both genotype and environment main effects significantly influenced (P<0.05) the concentrations of all essential minerals tested, with the exception of Co and Cr. In addition, most elements were also subject (P<0.05) to G × E interaction (Table 3.19). This means that statistical distinctions could be made between genotypes and environments based on the overall trend in the mineral profiles of samples, but that the relative profiles of genotypes shifted when grown in different environments.

3.3.3.2 Genotype and environment main effects Genotypes (Table 3.20) and growing environments (Table 3.21) were compared using Tukey’s pairwise HSD test. Several genotypes had relatively high concentrations of specific elements. Most notably, D175-3-p17-3 contained outstanding concentrations of Ca (768 μg g-1 dw) compared to the other genotypes tested and relatively high levels of Fe (19.0 μg g-1 dw). Slightly higher Fe content was found in D192-p397-1 (20.1 μg g-1 dw) kernels. D147-p3-115 and Sutherland had very high levels of K (8294 and 8314 μg g-1 dw) and Na (36.2 and 45.3 μg g-1 dw); Middleton and Sutherland were high in Mo (0.748 and 0.772 μg g-1 dw); while high P (4425 and 4501 μg g-1 dw) was measured in D192-p397-1 and Tingoora kernels.

Environmental effects lacked prominence, despite the presence of statistically significant differences according to Tukey’s pairwise comparisons (Table 3.21). The only exceptional findings were that samples from Bundaberg contained very high levels of Mo (1.335 μg g-1 dw) and Na (66.5 μg g-1 dw). The former may be attributed to the practice of Mo application as soil and/or foliar treatment during crop growth. High Na was probably due to the use of irrigation water sourced from underground bores that tend to become increasingly saline towards the end of the season in the absence of rainfall.

| Dietary minerals 121

TABLE 3.21 ESSENTIAL MINERAL CONCENTRATIONS IN PEANUT KERNELS FROM EACH ENVIRONMENT TESTED IN THE G × E STUDY

Environment N Concentration (μg g-1 dw)A B Ca Co Cr Cu Fe K Mg

Bundaberg 30 20.7 ab 607 ab 0.084 a 0.014 a 8.19 a 15.2 b 7282 c 1890 a Kairi 27 16.0 c 619 a 0.134 a 0.022 a 7.05 b 16.6 b 7552 b 1731 b Kairi CBR 36 16.9 bc 560 b 0.088 a 0.030 a 6.34 c 15.5 b 7257 c 1606 c Kairi FDR 36 19.5 abc 595 ab 0.083 a 0.025 a 8.61 a 19.0 a 8106 a 1751 b Taabinga 27 22.5 a 608 a 0.109 a 0.018 a 6.06 c 18.8 a 6847 d 1875 a

TABLE 3.21 CONTINUED

Environment N Concentration (μg g-1 dw)A Mn Mo Na Ni P Se Zn Bundaberg 30 21.0 c 1.335 a 66.5 a 2.3 c 4037 b 0.062 c 31.8 a Kairi 27 24.6 a 0.198 bc 12.6 b 8.6 a 3892 bc 0.143 a 29.4 b Kairi CBR 36 21.3 bc 0.150 c 11.8 b 3.8 c 3657 d 0.104 b 24.1 c Kairi FDR 36 23.1 ab 0.399 b 3.8 c 6.4 b 4223 a 0.119 ab 29.7 ab

| Dietary Taabinga 27 18.9 d 0.122 d 15.2 b 10.7 a 3832 c 0.035 c 28.0 b A Lower-case letters mark significantly different (P<0.05) environment means according to Tukey’s HSD test minerals 122

3.3.3.3 Genotype and G × E relationships with essential mineral concentrations When G × E interaction has a stronger influence than the genetic correlations between environments, it becomes an indirect selection for adaption to distinct environments. In view of this, pattern analysis methods have been developed for the classification and ordination of environments, particularly in large multi-environment trials, so that these ‘mega- environments’ are analysed rather than individual and possibly highly disparate individual environments (Cooper and De Lacy, 1994). The foremost methods are variants of PCA known as the Additive Main Effects and Multiplicative Interaction model (AMMI) and Genotype Main Effects and G × E Interaction model (GGE), both of which have strong proponents (Gauch, 2006, Yan et al., 2007).

PCA of auto-scaled data revealed little underlying structure, and many principal components (PC) were required to explain the data variance (i.e., 10 PC to explain 90% of variance). There was little clustering of genotypes, with only D192-p397-1 and Tingoora grouping in the upper right quadrant of the scores plot (Figure 3.3), mainly due to positive correlations with Mg, P, and Zn content. Some clustering of growing environments could be observed along the first three PC, although these accounted for only 47% of variance. Kairi and Kairi FDR formed a group along the PC-2 axis, while Bundaberg samples grouped in the lower right quadrant of the scores plot (Figure 3.4). Kairi CBR and Taabinga scored similarly on PC-1, but could be distinguished along PC-3, which correlated mainly with Fe and K content (Figure 3.5).

FIGURE 3.3 GENOTYPE SCORES ON PC-1 VS. PC-2 IN PCA OF AUTO-SCALED ESSENTIAL MINERAL CONCENTRATIONS. Full-season genotypes: F1, D147-p3-115; F2, PCA213; F3, Middleton; F4, Page; and F5, Sutherland Ultra-early genotypes: U1, D175-3-p17-3; U2, D192-p397-1; U3, D193-p3-6; and U4, Tingoora

| Dietary minerals 123

FIGURE 3.4 ENVIRONMENT SCORES ON PC-1 VS. PC-2 IN PCA OF AUTO-SCALED ESSENTIAL MINERAL CONCENTRATIONS. BB and BB2, Bundaberg; K, Kairi; KCBR, Kairi CBR; KFDR, Kairi FDR; and TB, Taabinga

FIGURE 3.5 ENVIRONMENT SCORES ON PC-3 VS. PC-2 IN PCA OF AUTO-SCALED ESSENTIAL MINERAL CONCENTRATIONS. BB and BB2, Bundaberg; K, Kairi; KCBR, Kairi CBR; KFDR, Kairi FDR; and TB, Taabinga

The GGE approach of environment-scaling the data prior to PCA was followed in order to reveal patterns in the genetic and G × E components of variation. Block-scaling the data to remove the environmental main effects meant that PCA modelled only the genotype and G × E effects. The data remained complex, i.e., the number of PC required to explain data variance remained high, but genotype clusters that were not apparent in the auto-scaled PCA could now be observed.

| Dietary minerals 124

Most prominently, ultra-early genotypes were distinct from full-season genotypes along PC-1 mainly because of generally higher Fe, Mg, and P concentrations (Figure 3.6). Subsequent PC gave further separation of genotypes. In particular, D175-9-p17-3 and Middleton were distinguished by high Ca and relatively low Mn and Zn contents; D147-p3-115 and Sutherland had relatively high K and low Co and Cu levels; while Page and PCA213 had relatively low K and high Co and Cu contents (Figure 3.7). Thus, both high-concentration elements, such as Ca, K, Mg, and P, and trace elements, such as Co, Cu, Fe, Mn, and Zn, were important descriptors of genotypic difference.

The PCA results reflect genotypic trends only, since each PC explains but a small portion of data variance. Larger-scale data analysis covering a greater number of genotypes, environments and seasons is required to fully characterise the G × E interaction affecting mineral accumulation in peanut kernels, and to determine whether it can be partitioned into mega-environments. Nevertheless, the separation of ultra-early and full-season genotypes along PC-1 (Fe, Mg, and P), and clustering of the sister-lines D147-p3-115 and Sutherland (K, Co, and Cu), provide enough suggestion of genetic control to warrant further investigation of G × E and trait segregation.

FIGURE 3.6 GENOTYPE SCORES ON PC-1 VS. PC-2 IN PCA OF ENVIRONMENT-SCALED ESSENTIAL MINERAL CONCENTRATIONS. Full-season genotypes: F1, D147-p3-115; F2, PCA213; F3, Middleton; F4, Page; and F5, Sutherland Ultra-early genotypes: U1, D175-3-p17-3; U2, D192-p397-1; U3, D193-p3-6; and U4, Tingoora

| Dietary minerals 125

FIGURE 3.7 GENOTYPE SCORES ON PC-1 VS. PC-3 IN PCA OF ENVIRONMENT-SCALED ESSENTIAL MINERAL CONCENTRATIONS. Full-season genotypes: F1, D147-p3-115; F2, PCA213; F3, Middleton; F4, Page; and F5, Sutherland Ultra-early genotypes: U1, D175-3-p17-3; U2, D192-p397-1; U3, D193-p3-6; and U4, Tingoora 3.3.3.4 Breeding potential for essential mineral composition The stability of genetic differences in several essential minerals were confirmed by significant (P<0.05) genotype effects according to ANOVA (Table 3.20). In particular, D175-3-p17-3 kernels contained consistently high concentrations of Ca; Middleton and Sutherland had high Mo levels; D147-p3-115 and Sutherland had very high K and Na levels; while D192-p397-1 and Tingoora were high in P.

Ca is distinct from other major cations in that it enters the peanut seed mainly by direct diffusion from the soil and is largely phloem-immobile, meaning that there is little downwards re-distribution after transpiration-driven uptake through the xylem (Stalker, 1997). Genotypes that are less susceptible to ‘pops’ syndrome caused by Ca deficiency demonstrate improved translocation of Ca from the pod pericarp to the seed, which may be due to differing seed/shell contact, ‘sink’ signalling by phytohormones such as auxins, or shift in the balance between the assimilate demands of vegetative growth and pods (Beringer and Tekete, 1979).

A field trial of gypsum (CaSO4) applications to Virginia peanuts found that the smaller-seeded cultivar (73 g/100 seed) had a lower Ca requirement than the larger-seeded cultivar (101 g/100 seed) (Jordan et al., 2010), supporting the postulate that greater surface-to-mass ratio should favour Ca accumulation in seed in view of its phloem immobility.

However, studies of Ca concentrations in F1-F3 progenies of a chickpea cross indicated that Ca content was determined by genetic factors other than weight genes (Abbo et al., 2008), while

| Dietary minerals 126

Branch and Gaines (1983) found no association between Ca concentration and seed size in peanut. It is clear that there is a strong genetic component to Ca accumulation, and controlled studies to better characterise Ca physiology in peanuts could yield specific genetic or biochemical markers for breeding targets that could aid selection. Traits that lead to high Ca uptake and accumulation in seeds could reduce the need for soil amelioration, as well as improving peanut nutritional quality.

Mo is essential for biological nitrogen fixation (BNF) as a co-factor of the most common class of nitrogenase, the enzyme catalysing reduction of atmospheric nitrogen (N2) into ammonium + (NH4 ) in nitrogen-fixing (diazotrophic) bacteria (Hernandez et al., 2009). For crops that form symbiotic relationships with diazotrophic rhizobacteria, including legumes and several mainly graminaceous non-legumes, maximising BNF is a far more efficient and inexpensive method of delivering N requirements than artificial N fertilisation. Mo deficiency often limits BNF in the field, especially where intensive cultivation has led to depletion of soil micronutrients, and Mo supplementation through fertiliser mixtures, Mo frits, seed coatings, and foliar sprays have been used to sustain yields (Campo et al., 2009). Breeding for high Mo would be valuable from both nutritional and agronomic perspectives, and the relatively high Mo accumulation in Middleton and Sutherland lines is worthy of further investigation.

The sister lines, D147-p3-115 and Sutherland, had K and Na concentrations that were significantly (P<0.05) higher than other genotypes (Table 3.20). The genotypic differences in Na content were especially large, with D147-p3-115 and Sutherland kernels having on average two to three-fold more Na than the other lines, regardless of the growing environment. This result is interesting when we consider that saline soil is known to inhibit Ca and K uptake, while promoting Fe, Mn, P, and Na accumulation (Chavan and Karadge, 1980), and that Ca, Mg, and K have similar affinities for carrier sites across plant cellular membranes such that uptake of these cations is typically competitive (Fageria, 1974).

D147-p3-115 and Sutherland demonstrated the ability to sustain very high uptake of both K and Na, although their Ca and Mg concentrations were amongst the lowest. Although salinity is not a priority of the APBP, an approach for manipulating mineral composition may lie in strategies used for breeding salinity tolerance. It should be noted that the ‘high-Na’ D147-p3- 115 and Sutherland lines contained an average of 36.2 and 45.3 μg g-1 dw (i.e. 0.91 and 1.1 mg per 25 g serving), respectively, which is far below the adequate daily intake of 460-920 mg recommended for an adult (NHMRC, 2006), and would not be nutritionally detrimental.

| Dietary minerals 127

3.3.3.5 Agronomic manipulation of essential mineral concentrations The statistical differences between the Mo contents of Middleton and Sutherland and other genotypes suggest a strong genetic component for Mo concentration. However, the prominent response to Mo supplementation irrespective of genotype, which resulted in three to ten-fold more Mo in Bundaberg kernels compared to other locations (Table 3.21), suggests that agronomic manipulation may be of greater practical significance.

Research on several legumes indicate that the G × E interactions affecting seed Mo content and yield responses to Mo fertilisation are varied. Plants grown from soybean seed that were highly Mo-enriched by Mo fertilisation in the previous season sustained a high level of BNF and produced enhanced yields that were not improved by further fertilisation, regardless of increases in seed Mo content (Campo et al., 2009). On the other hand, Mo requirements to achieve optimal BNF and yields in bean plants grown from high- and low-Mo seed varied with cultivar, such that cultivar dominated the response to fertilisation regardless of the Mo concentrations of the seed (Brodrick et al., 1992).

Our data confirmed that the Mo content of peanut kernels is strongly affected by genotype, growing environment, and G × E interaction. Further experiments could determine the extent to which agronomic/cropping history contributes to Mo in seed, compared to Mo fertilisation during the current growing season. It is feasible that a combination of genetic and agronomic manipulation could be used to consistently achieve optimal levels of Mo in peanut seed, for maximum BNF, productivity, and postharvest nutritional quality.

The interaction of genotype and endogenous/exogenous Mo is further complicated by the soil

2- 2- chemical ecology. The structural similarity of molybdate (MoO4 ), sulphate (SO4 ), and 2- - phosphate (HPO4 or H2PO4 ) anions suggests that non-specific uptake will be competitive. This premise is supported by the finding that single superphosphate, which contains CaSO4, failed to improve yields of Nigerian peanuts despite overcoming the S and P deficiencies typical of the region. In contrast, (sulphate-free) triple superphosphate enhanced yields and Mo uptake, particularly in P- and Mo-deficient soils (Rebafka et al., 1993). S is thought to depress Mo uptake by lowering soil pH when it undergoes microbial oxidation, which leads to decreased Mo availability and an acidified rhizosphere that favours uptake of cations over anions, as well as competing directly with molybdate for uptake into the roots (Rebafka et al., 1993). Failure to take such considerations into account could be responsible for the apparently inconsistent responses reported by different studies of Mo fertilisation in the literature.

| Dietary minerals 128

The prospect of combining genetic improvement with agronomic management is worth exploring for other elements of interest. In general, response to fertilisation has been evaluated in terms of yield outcomes rather than effects on mineral concentrations. Plant nutrition is obviously relevant to productivity, but it may be desirable to boost the mineral content of peanut kernels beyond the maximum levels that deliver yield benefits.

In the peanut gypsum trial discussed earlier (Jordan et al., 2010), gypsum recommendations were assessed on yield, grade, and seed germination responses. Treatments, however, would also have affected Ca and Mo concentrations in seed and indirectly influenced the uptake of other minerals. The typical K rates were found to be suitable as they had no effect on pod yield or market grade. Yet K fertilisation is known to promote proton release, thereby reducing soil pH and enhancing cation uptake, although Fe acquisition may remain impaired by high soil

CaCO3 (Zaharieva and Römheld, 1991). Thus, rhizosphere acidification has been shown to enhance Ca, Mg, Mn, and Zn uptake in wheat and chickpea (Li et al., 2004).

Several articles have reported significant genotypic variability in the Se content of wheat, rye, and soybean seed, as well as the edible parts of some vegetables (Lyons et al., 2005), although differences may be small compared to the background variation. Scope for agronomic enrichment of Se content is evidenced by at least one study showing that Se in winter wheat grain increased by 0.016-0.029 μg g-1, or ten-fold, following Se fertilisation at ten trial sites over two years (Broadley et al., 2009). In our study, Se concentration in peanut kernels was significantly influenced by genotype, environment, and G × E interaction, albeit with small absolute differences, as in these previous studies. Again, it seems likely that genetic improvement could be coupled with agronomic management to achieve consistently high mineral content.

3.3.3.6 Further work There are several avenues of research that would support the incorporation of mineral composition into the breeding program. First, the reported sample set was limited to nine genotypes grown in five environments in a single season and results may have been inflated by seasonal/temporal interactions. Enlarging the dataset to include more genotypes and environments harvested over at least several seasons would greatly improve the characterisation of the G × E interaction and allow more explorative statistical analyses. The breeding lines that were tested, while genetically diverse, have already been subjected to selection pressures centring on program priorities of yield, disease and kernel quality criteria.

| Dietary minerals 129

The current APBP accessions display a promising level of genotypic variation, but breeding efforts would benefit from assessment of the genetic variability within the international gene bank, especially germplasm adapted to extremes in soil quality. Having said this, several tested lines accumulated high concentrations of Ca (D175-3-p17-3), Mo (Middleton and Sutherland), K, and Na (D147-p3-115 and Sutherland). It would be worth examining the heritability of these traits, as well as the effect of combining genetic and agronomic improvement. Some of the basic physiology regarding interactions of soil nutrients and factors affecting mineral accumulation remain poorly understood, and elucidation of these topics through controlled studies would illuminate efforts to manipulate mineral composition, both through breeding and soil management.

On a different note, breeding can be regarded as one of the first links in a peanut supply chain that ends with the consumers. The need to contend in an increasingly competitive global marketplace is the driving force for the APBP, pushing continual improvements in productivity and kernel quality, although outcomes are expected to eventually benefit agriculture and public health more generally. A key question for the peanut industry, then, is how much of which elements are desirable? It is straightforward to compare the mineral content of a typical serving of peanuts with recommended dietary mineral intakes, but rather more challenging to evaluate the health impacts and consumer acceptance of a mineral-rich snack food. The issue is further complicated by the varying bio-availability of different species of minerals (Sandberg, 2002), which may be adversely affected by antinutrients such as phytate-mineral complexes (Ravindran et al., 1994), and may itself be subject to G × E interaction (Šimić et al., 2009).

Staple food products may be subject to mandatory or voluntary fortification in the interests of public health; the enrichment of snack foods, however, is essentially market-driven. In this respect, research into consumer perceptions of enriched peanut products, awareness of health benefits, and willingness-to-pay are as important as the chemistry, physiology, and genetics, to ensure that breeding targets are in accord with the market opportunities for potential new peanut varieties. Snack foods make a large contribution to daily nutrition, especially of children, and healthier food options are attractive as a form of preventative healthcare, from both public health and food marketing perspectives.

3.3.4 Conclusion

A multi-environment study of essential mineral concentrations in nine diverse peanut genotypes was conducted. Genotype, environment, and G × E interaction were all found to have significant effects on mineral concentrations according to standard ANOVA. Several of

| Dietary minerals 130 the current APBP peanut accessions accumulated relatively high concentrations of specific minerals in their kernels, i.e., D175-3-p17-3 had high Ca; Middleton and Sutherland had high Mo; D147-p3-115 and Sutherland had high K and Na; while D192-p397-1 and Tingoora had high P content. Clustering of some genotypes in environment-centred PCA confirmed that there was substantial genetic control of both micro- and macro-element concentrations in peanut kernels, and expansion of the dataset is warranted to explore these relationships in greater depth. In addition to the genetic components of mineral composition, several elements were strongly influenced by the growing environment, especially Mo and Na. It is feasible that a strategy combining genetic improvement and agronomic management could be designed to ensure optimal concentrations of minerals are consistently achieved, bringing benefits during the growing season as well as for postharvest nutritional quality.

| Dietary minerals 131

3.4 NIRS estimation of essential mineral concentrations in peanut kernels

3.4.1 Objectives

One of the challenges in selecting plants for nutritional and chemical traits is that typical analytical methods are often destructive and time-consuming to perform. The extremely large sample numbers generated during the progression of breeding programs become prohibitive if varietal evaluation depends on capital- and labour-intensive analysis. For this reason, the development of rapid screening techniques will be pivotal to facilitating the development of peanut varieties with enhanced nutritional or functional food profiles.

The Australian peanut industry currently employs NIRS for the routine estimation of kernel moisture, oil content, and fatty acid composition. This study investigates the feasibility of using NIRS to estimate essential mineral concentrations in peanut kernels, which would have potential applications both in the APBP and in quality control as a tool for rapid, non- destructive selection of seeds and plants.

The results of this study were published by Phan-Thien et al. (2011).

3.4.2 Materials and methods

3.4.2.1 Samples

The data was generated from two sample sets: A, 56 breeding lines (N=163 samples) that were analysed in the genotype study (chapter 3.2); and B, nine breeding lines grown in five environments (N=156 samples) that were analysed in the G × E study described in chapter 3.3.

3.4.2.2 Spectral analysis

Peanut kernel samples were scanned by a monochromator in reflectance mode (6500 model, Foss NIRSystems, Silver Spring, MD, USA), in a rectangular sample holder carrying about 75 g of whole kernels. The spectral range was 400 to 2498 nm recorded at 2 nm intervals, but narrower spectral regions were used for calibration.

3.4.2.3 Elemental analysis Samples were mineralised by closed acid microwave digestion and analysed for essential minerals by ICP-OES and ICP-MS as described in chapters 3.2.2 and 3.3.2.

| Dietary minerals 132

3.4.2.4 Development of NIRS calibrations

Data was separated into calibration and validation subsets in a ratio of 3:1 after sorting in order of concentration so that both subsets contained representative numerical distributions for each analyte. Several mathematical pre-treatments of the spectral data were explored including first (dA) and second (d2A) derivatisation according to the Savitsky-Golay algorithm (second order polynomial, 9 smoothing points), de-trending (DT, second order polynomial), multiplicative scatter correction (MSC), and standard normal variate correction (SNV). Variables were normalised (mean-centred and standardised) for calibration against the spectral data. Outliers that reflected error in reference analysis or spectral aberrations were also removed, noting that whole kernels and variable particle size result in less perfect spectra than uniform samples. Models were developed by partial least squares (PLS) regression using The Unscrambler v9.8 (CAMO Software AS, Oslo, Norway), then tested by ‘leave-one-out’ cross-validation and application to the validation subsets.

3.4.2.5 Statistical analysis

The accuracy of calibrations were compared by reference to the root mean square error for cross-validation (RMSECV) and root mean square error for prediction/performance (RMSEP); the ratio of the standard deviation of the response variable to the RMSECV or RMSEP, termed the residual predictive deviation (RPD); the coefficient of determination of the calibration (R2) and validation (r2); and the bias. RMSECV and RMSEP are measures of the average modelling error and average prediction error, and are used to determine the number of latent variables (principal components, PC) in the model (Nicolaï et al., 2007). RPD allows comparison of model performances in populations with different distributions and is commonly used as a guide to the practical usefulness of calibrations. Opinions differ (Batten, 1998), but we have found that RPD above 2 can achieve discrimination between high and low values of the response variable, while RPD of more than 3 is required for reliable prediction (Nicolaï et al., 2007). Bias is the systematic component of prediction error and contributes to RMSEP (Golic and Walsh, 2006), and may reflect the robustness of the model (Nicolaï et al., 2007).

| Dietary minerals 133

3.4.3 Results and discussion

3.4.3.1 Development of calibration models

Meaningful regression models could not be obtained for most of the essential minerals measured in peanut kernels. This was not surprising given that essential minerals are present in low concentrations and lack NIR-absorbent molecular bonds, so correlations with NIR spectra are necessarily indirect. Nevertheless, potentially useful calibrations were achieved for several of the higher-concentration elements, namely, Ca, K, Mg, and P (Table 3.22).

-1 TABLE 3.22 DISTRIBUTION OF ESSENTIAL MINERAL CONCENTRATIONS (μG G DW) IN CALIBRATION AND VALIDATION DATA SETS

Analyte Reference Set A Set B

ICP-OES λ (nm) Calibration Validation Calibration Validation Ca 422.673 605 ± 136 599 ± 138 598 ± 110 590 ± 109 K 766.490 7056 ± 793 7025 ± 816 7452 ± 878 7400 ± 873 Mg 279.077 1678 ± 192 1669 ± 200 1766 ± 197 1752 ± 204 P 178.221 (A), 213.617 (B) 4556 ± 1200 4516 ± 1204 3940 ± 466 3909 ± 472

TABLE 3.23 STATISTICS FOR THE BEST CA, K, MG, AND P NIRS CALIBRATION MODELS

Analyte Pre-treat. Model Cross-validation Prediction

2 2 2 Math: Range Set PC RMSEC R C RMSECV r CV RPDCV RMSEP r P RPDP Bias Ca d2A-SNV-DT: A 3 108.26 0.222 116.66 0.107 1.17 123.79 0.172 1.11 -9.0 1130-2372 B 7 48.99 0.759 59.34 0.649 1.85 71.70 0.553 1.52 -19.5 K d2A-SNV-DT: A 10 369.28 0.768 493.30 0.602 1.61 578.81 0.484 1.41 -48.7 1246-1738 B 9 307.09 0.876 390.70 0.800 2.25 392.79 0.792 2.22 -10.9 Mg d2A-MSC: A 4 104.16 0.656 115.64 0.584 1.66 205.06 n/a 0.98 -35.3 1246-1738 B 7 73.03 0.846 88.01 0.786 2.24 117.34 0.660 1.74 1.9 P d2A-SNV-DT: A 5 541.05 0.728 629.42 0.636 1.91 843.05 0.497 1.43 -179.7 1400-1550, B 5 238.24 0.693 263.64 0.634 1.77 286.46 0.622 1.65 38.0 1850-2200

Calibration models for Ca, K, Mg, and P were developed after exploring the use of several mathematical pre-treatments to decrease the influence of uninformative data and optimise accuracy, and removal of outliers (4-14 data points) that reflected error in reference analysis or spectral aberrations. Statistics for the models achieving the lowest prediction errors and highest RPD are presented in Table 3.23.

| Dietary minerals 134

3.4.3.2 Spectra and spectral correlations

The NIR spectra (Figures 3.8 and 3.9) had a similar form to those previously reported for peanut samples. Major peaks at 1134-1200, 1400-1440, and 1900-1950 nm have been associated with water content; 1705-1795 and 2300-2350 nm with oil; and 2115-2165 nm with protein (Fox and Cruickshank, 2005, Misra et al., 2000, Sundaram et al., 2009, Tillman et al., 2006).

1.0

0.8

0.6 Absorbance

0.4

0.2 1250 1500 1750 2000 2250 Wavelength (nm)

FIGURE 3.8 RAW (LOG 1/R) NIR SPECTRA OF PEANUT KERNELS.

0.002

0.001

0.000

Absorbance 1414 2346 1162 1760 2306 -0.001 1728

1212 1920

-0.002 1250 1500 1750 2000 2250 Wavelength (nm)

2 FIGURE 3.9 SECOND DERIVATIVE (D A) NIR SPECTRA OF PEANUT KERNELS. NB, Characteristic peaks in peanut spectra are annotated (Fox and Cruickshank, 2005, Golic et al., 2003, Smith et al., 1991, Sundaram et al., 2009)

| Dietary minerals 135

Regression coefficients are typically examined to help explain the physical basis of the NIRS model, as the coefficients of greatest magnitude identify the most correlated wavelengths. While many of the largest spectral correlations in our optimised calibrations (Figures 3.10 and 3.11) lie in the regions usually associated with oil (e.g., Ca ~2300 nm and K ~1700 nm) or water (e.g., Ca ~1200 and ~1900 nm; K ~1450 nm; Mg ~1400 nm; and P ~ 1900 nm), the plots of regression coefficients are somewhat noisy, perhaps reflecting the indirect nature of the correlations. Interpretation of these is not possible without in-depth study of the individual element interactions in the peanut matrix.

(A) 100

50

0

Regression coefficient Regression -50

1250 1500 1750 2000 2250 -100 Wavelength (nm) (B) 2000

1000

0

Regression coefficient -1000

-2000 1300 1400 1500 1600 1700 Wavelength (nm) 2 FIGURE 3.10 REGRESSION COEFFICIENTS IN NIRS MODELS OF (A), CA (1130-2372 NM, D A, SNV- 2 DT) AND (B) K (1246-1738 NM, D A, SNV-DT) CONCENTRATIONS IN PEANUT KERNELS.

| Dietary minerals 136

(A) 1000000

500000

0

Regression coefficient Regression -500000

-1000000 1300 1400 1500 1600 1700 Wavelength (nm)

(B) 200

0

-200 Regression coefficient Regression

-400 1450 1550 1950 2050 2150 Wavelength (nm)

2 FIGURE 3.11 REGRESSION COEFFICIENTS IN NIRS MODELS OF (A) MG (1246-1738 NM, D A, MSC) 2 AND (B) P (1400-1550 AND 1850-2200 NM, D A, SNV-DT) CONCENTRATIONS IN PEANUT KERNELS.

| Dietary minerals 137

3.4.3.3 Accuracy of calibration models

Models developed for Set B data were more accurate than for Set A. Considering the Set B data only, the Ca, K, Mg, and P calibration models explained slightly more than 60% of the

2 2 variation in Ca (RPDCV 1.85, r CV 0.649, RPDP 1.52) and P (RPDCV 1.77, r CV 0.634, RPDP 1.65), and 2 2 about 80% of the variation in K (RPDCV 2.25, r CV 0.800, RPDP 2.22) and Mg (RPDCV 2.24, r CV

0.786, RPDP of 1.74) concentrations. Although the models were not ideal, 60-80% explained variance reflected substantial correlation between the observed and fitted data. This was apparent in plots of predicted versus reference concentrations, with linear regression lines lying close to the ‘one-to-one’ target lines (Figures 3.12 and 3.13).

(A) Regression line 800

Target line dw) -1 g P

600 Predicted conc. ( 400

400 600 800 Reference conc. (Pg g-1 dw) B 10000 ( ) Regression line

Target line 9000 dw) -1

g 8000 P

7000

Predicted conc. ( Predicted conc. 6000

5000 5000 6000 7000 8000 9000 10000 -1 Reference conc. (Pg g dw) 2 FIGURE 3.12 ACCURACY OF NIRS MODELS OF (A) CA (1130-2372 NM, D A, SNV-DT) AND (B) K 2 (1246-1738 NM, D A, SNV-DT) CONCENTRATIONS IN PEANUT KERNELS.

| Dietary minerals 138

(A) 2250 Regression line

Target line 2000 dw) -1 g P 1750

1500 Predicted conc. (

1250

1250 1500 1750 2000 2250 Reference conc. (Pg g-1 dw)

B 5000 ( ) Regression line

Target line 4500 dw) -1 g P 4000

3500 Predicted conc. ( Predicted conc.

3000

3000 3500 4000 4500 5000 -1 Reference conc. (Pg g dw) 2 FIGURE 3.13 ACCURACY OF NIRS MODELS OF (A) MG (1246-1738 NM, D A, MSC) AND (B) P 2 (1400-1550 AND 1850-2200 NM, D A, SNV-DT) CONCENTRATIONS IN PEANUT KERNELS. None of the calibrations were sufficiently accurate for quantitative application, which generally requires RPD>3 (Nicolaï et al., 2007). Nevertheless RPD>2 suggests that sound potential exists for approximate modelling of K content and to a lesser extent Mg. This would allow, for example, semi-quantification of peanut kernels into groups containing low, medium, or high concentrations. This type of approximate screening for Ca and P content may be feasible, but the statistics at this stage are less convincing, and further work to reduce prediction error and improve RPD is recommended for this purpose.

| Dietary minerals 139

3.4.3.4 Application for approximate prediction

From a practical viewpoint, NIRS models with r2>0.6 remain useful as rapid, non-destructive selection tools to progress breeding lines with superior nutritional traits in breeding populations. The approximate classification of plants for nutritional quality would add value to NIRS analysis, especially since this is already performed for routine monitoring of core quality criteria such as total oil content and fatty acid composition. At the least, it would reduce the number of reference analyses that must be carried out if accurate selection for the trait is desired. This application of NIRS models to plant breeding was previously suggested for improvement of Mg content in perennial ryegrass, when only approximate calibration models

2 (r C 0.68-0.82) could be achieved (Smith et al., 1991).

The values of RPD were too low for predictive accuracy, but RPD ~2.0 for K and Mg, in particular, may be useful for approximate quantification (Nicolaï et al., 2007). This can be observed in scores plots as clustering or stratification of samples having the highest and lowest concentrations. This is demonstrated in Figures 3.14 and 3.15 by marking data points in the highest and lowest quintiles of the sample concentration range. Clustering of the higher and lower values can still be observed, even though the models contained many latent variables (PC), and PC-1/PC-2 explained only 29%, 66%, 54%, and 38% of variance in Ca, K, Mg, and P, respectively.

3.4.3.5 Robustness of the calibration models

Robustness describes the sensitivity of calibration models to unknown changes in external factors (Nicolaï et al., 2007). The most practically relevant include changes in the instrumental response due, for example, to temperature fluctuations (Golic et al., 2003), shift in detector stability, and transfer to another instrument; and batch-to-batch changes in sample characteristics such as those due to field or seasonal variability, or changes in handling or storage conditions that lead to differences in the sample matrix.

In general, expansion of NIR data sets to increase inherent variation enables more robust calibrations (Nicolaï et al., 2007). In this study however, useful calibrations could be generated for Set B, while much poorer results were obtained for Set A alone and for Sets A and B combined. Two main distinctions between the sample sets may help to explain the ‘adverse’ influence of Set A on the calibrations.

First, Set B contained samples collected from multiple locations, while Set A was a varietal assessment of plants grown at one site. Mineral uptake and accumulation in plants is known to be strongly influenced by a host of environmental factors, particularly soil chemical

| Dietary minerals 140 interactions related, for example, to nutrient status/fertilisation (Rebafka et al., 1993), soil chemistry (Fageria, 1974, Li et al., 2004, Zaharieva and Römheld, 1991), salinity (Chavan and Karadge, 1980), and water activity (Conkerton et al., 1989, Moinuddin and Imas, 2007). While Set A was genetically more diverse, Set B contained greater phenotypic heterogeneity as it was subject to more varied climatic and growing conditions, as well as a more extended harvest period. This greater inherent variation in the sample matrix may have contributed to better calibrations for Set B.

(A) PC2 2 Low er Ca (<500 Pg g-1 dw )

Higher Ca (>750 Pg g-1 dw )

PC1 -10 10

Sample range: 396-853 Pg g-1 dw

-2 (B) PC2 1.5

Low er K (<6500 Pg g-1 dw )

Higher K (>8500 Pg g-1 dw )

PC1 -4 4

Sample range: 5903-9604 Pg g-1 dw

-1.5 2 FIGURE 3.14 SCORES PLOT (PC1 VS. PC2) IN NIRS MODELS OF (A) CA (1130-2372 NM, D A, SNV- 2 DT) AND (B) K (1246-1738 NM, D A, SNV-DT) CONCENTRATIONS IN PEANUT KERNELS.

| Dietary minerals 141

(A) PC2

0.001 Low er Mg (<1500 Pg g-1 dw )

Higher Mg (>2000 Pg g-1 dw )

PC1 -0.0005 0.0005

Sample range: 1300-2222 Pg g-1 dw

-0.001

(B) PC2 2 Low er P (<3200 Pg g-1 dw )

Higher P (>4400 Pg g-1 dw )

PC1 -4 4

Sample range: 2868-4761 Pg g-1 dw

-2 2 FIGURE 3.15 SCORES PLOT (PC1 VS. PC2) IN NIRS MODELS OF (A) MG (1246-1738 NM, D A, MSC) 2 AND (B) P (1400-1550 AND 1850-2200 NM, D A, SNV-DT) IN PEANUT KERNELS. Second, there were some slight differences in the reference analyses used that may have affected experimental error, and hence the quality of calibrations. This was particularly relevant for ICP-OES analysis of P content. Set A, analysed at 178.221 nm, had a much larger standard deviation (4556 ± 1200 μg g-1) than Set B (3940 ± 466 μg g-1), which was analysed at 213.617 nm. This difference in population distributions was more likely to reflect inaccuracy in the reference analysis than to represent truly greater variation in the P content of Set A samples.

3.4.3.6 Implications of NIRS study

Even though Ca, K, Mg, and P are considered to be dietary ‘macro-elements’, concentrations in the peanut samples were much lower than analytes to which predictive NIRS calibrations have previously been applied with success. Ca, K, Mg, and P were present in peanut kernels at the

| Dietary minerals 142 parts-per-million to low parts-per-thousand level, which is ten or more times lower than can typically be quantified by NIRS. In view of this, it is remarkable that even approximate quantification could be derived from the data.

Few publications have described NIRS quantification of essential minerals in food and plant materials, and only a small portion of these demonstrated predictive potential. For example, some of the calibrations for Ca, K, Mg, and P in botanical fractions taken from semi-arid grasslands provided approximate quantification, as shown for our results (RPDP approximately 2 or above), but none were accurately predictive (Ruano-Ramos et al., 1999). Calibrations of B, Cu, Fe, Mn, Na, S and Zn in white clover and lucerne yielded neither predictive nor approximate models (Cozzolino and Moron, 2004).

2 2 On the other hand, good calibrations were reported for Ca (r P 0.90, RPDP 3.1), Cl (r P 0.92, 2 2 RPDP 3.6), K (r P 0.92, RPDP 3.4), and Mg (r P 0.89, RPDP 3.0) in timothy forage grass (Tremblay 2 2 2 2 et al., 2009); Ca (r P 0.92, RPDP 4.6), Cu (r P 0.93, RPDP 3.7), Fe (r P 0.89, RPDP 4.4), K (r P 0.91, 2 2 RPDP 3.3), Mn (r P 0.85, RPDP 3.0), and P (r P 0.90, RPDP 3.3) in animal feed (González-Martín et 2 2 2 al., 2006); Ca (r P 0.95, RPDP 4.62) and Zn (r P 0.93, RPDP 3.75) in fresh cheeses; and Zn (r P 0.89,

RPDP 3.01) in freeze-dried cheeses (Lucas et al., 2008). The timothy and cheese samples contained about ten times more Ca (>5 mg g-1) than peanut and the timothy also had considerably more K (23.7 mg g-1). However, concentration did not entirely explain the greater accuracy that these other authors achieved, in view of the very good results for trace elements in animal feed (González-Martín et al., 2006) and Zn in cheese (Lucas et al., 2008).

3.4.4 Conclusion

Our results indicate that it should be feasible to develop useful calibration models for essential elements present at relatively high concentrations in peanut kernels, such as Ca, K, Mg, and P, with an accuracy that would be useful for application in plant breeding programs. Such NIRS models represent a cost-effective means of selecting segregating plants in breeding populations with superior nutritional composition for progression in the APBP, especially as samples are already subject to routine analysis by NIRS for determination of oil chemistry, moisture, and other constituents. Further analysis of more diverse peanut samples is warranted to confirm batch-to-batch accuracy and to improve the robustness of calibrations.

| Dietary minerals 143

4 Antioxidant capacity

4.1 Literature review

4.1.1 Importance of dietary antioxidants

4.1.1.1 Cellular oxidants and antioxidant defence The majority of reactive oxygen and nitrogen species (RONS) in the human body (and other eukaryotic organisms) are generated by regular cellular activity, particularly mitochondrial respiration, microsomal cytochrome P450 enzymes, flavoprotein oxidases, and peroxisomal fatty acid metabolism (Castro and Freeman, 2001). The enzymatic reduction of oxygen in mitochondrial respiration produces water, but a small percentage of molecular oxygen is

•- reduced by one electron to yield superoxide radical (O2 ). Subsequent reductions, which can be catalysed by transition metals such as copper and iron, generate hydrogen peroxide (H2O2) • •- •- and hydroxyl radical ( OH). Not all O2 production is accidental: the O2 generated by NADPH oxidases in phagocytes is a precursor for secondary oxidants such as H2O2, hypochlorous acid (HOCl), and singlet oxygen, which are employed in antimicrobial defence against pathogens. Lysosomes also generate RONS for oxygen-dependent killing of invasive microbes (Giustarini et al., 2009). The most important reactive nitrogen species is nitric oxide radical (•NO). Constitutive and inducible nitric oxide synthase convert L-arginine to citrulline and •NO during essential regulatory functions in neurotransmission, blood pressure, platelet aggregation, and immune response (Castro and Freeman, 2001). Peroxynitrite (ONOO-) is another reactive nitrogen species that is generated in the cellular respiratory burst through the reaction

• •- between NO and O2 .

There are of course salient environmental contributors to the body’s burden of oxidative stress. UVA and UVB radiation induce generation of RONS in skin cells, which is linked to inflammation, skin ageing, phototoxicity, and malignant tumours. The toxicity from pharmaceutical drug abuse (e.g., acetaminophen) and exposure to pesticides such as paraquat and organophosphates is primarily due to RONS production and alterations in redox metabolism. Air pollutants such as halogenated and polycyclic aromatic hydrocarbons cause carcinogenesis, teratogenesis, and immune dysfunction, associated with the generation of mutagenic metabolites and RONS, oxidative injury to DNA and proteins, and impaired function of the endogenous antioxidant system. The toxic effects of chronic exposure to fine particulate matter (<2.5 μm) and heavy metals and metalloids are also associated with RONS production and accumulation in tissues (Limón-Pacheco and Gonsebatt, 2009).

| Antioxidant capacity 144

RONS can attack all molecular targets: DNA, lipids, and proteins. The principal sites of lipid oxidation are polyunsaturated fatty acids (PUFA) associated with the plasma and organelle membranes and plasma lipoproteins. Lipid peroxidation is frequently initiated by a radical- mediated abstraction of hydrogen from a methylene (-CH2-) carbon, which yields reactive lipid, peroxyl (LOO•), and alkyl (L•) radicals in an autocatalytic chain of oxidation reactions (Castro and Freeman, 2001). The importance of lipid peroxidation in the pathophysiology of ageing and a variety of diseases was reviewed by Negre-Salvayre et al. (2010).

Oxidative damage to DNA can take the form of oxidised bases, single-strand, and double- strand breaks. RONS can also attack DNA indirectly through reactions with other cellular components, which generate secondary reactive intermediates that irreversibly couple to DNA bases to form DNA adducts. Quantitative data on DNA adducts formed in human tissues, and the types of mutations caused by oxidative DNA lesions, were extensively reviewed by De Bont and van Larebeke (2004). Enzymatic mechanisms to repair oxidative modifications of DNA, especially base excision repair enzymes, ensure that most damage is removed before cell replication; otherwise, the miscoded DNA becomes fixed as a mutation (De Bont and van Larebeke, 2004).

Oxidation of proteins is predominantly initiated by •OH and propagated by diverse reactions. Oxidation can cause modification of amino acid residue side chains, protein-protein cross- linkages, and protein fragmentation. In some cases protein unfolding or conformational changes may occur. Proposed reactions in protein oxidation and its products are reviewed by Dean et al. (1997) and Berlett and Stadtman (1997). Some oxidation is reversible by cellular repair mechanisms: modest modifications appear to signal degradation and trigger removal from the cell, as oxidised proteins become more susceptible to proteolytic attack. Heavily modified proteins, however, appear less sensitive, and it is unclear that large-scale unfolding or even limited conformational changes can be repaired.

Cells employ several mechanisms of endogenous defence against oxidative damage, which feature a wide variety of enzymes. Some, such as the superoxide dismutases (SOD), catalyse

•- the dismutation of O2 to H2O2; whereas others, such as catalase and glutathione peroxidase, convert H2O2 to water. There are enzymes that help to maintain the reduced forms of antioxidants (e.g., glutathione reductase) and protein thiol groups (e.g., thioredoxin reductase), and enzymes that act to maintain a reducing environment for NADH, NADPH, and FADH-dependent systems (Castro and Freeman, 2001). The enzymatic repair systems that remove oxidatively damaged cells can also be regarded as part of the defence system, as miscoded DNA become fixed mutations if they are not eliminated before cell replication. | Antioxidant capacity 145

Dietary antioxidants act as hydrophilic (e.g., ascorbate, urate, and glutathione) and lipophilic (e.g., tocopherols, flavonoids, carotenoids, and ubiquinol) radical scavengers, and thereby have important functions in counteracting the initiation and propagation of oxidative stress in cells.

4.1.1.2 Oxidative stress in human disease The ‘oxidative stress hypothesis’ implicates oxidative tissue injury in the sustained inflammation, cell proliferation, cytotoxicity, and proangiogenic environment (i.e., that promotes formation of blood vessels) that is common to the development of diverse diseases. Carcinoma and cardiovascular disease are classic oxidative stress-related diseases. There is also rapidly expanding data on pathological conditions as diverse as neurodegenerative, renal, and liver diseases, diabetes, atherosclerosis, foetal vascular syndrome in pre-eclampsia, metabolic syndrome, and ocular degeneration. The pathophysiological intricacies of these reactions are beyond the scope of this thesis; but the importance of redox reactions in human health cannot be overstated. To illustrate, some reviews of conditions associated with oxidative damage are outlined in Table 4.1. Roberts et al. (2009) and Scalbert et al. (2005) have published extensive overviews of the subject.

| Antioxidant capacity 146

TABLE 4.1 REVIEW OF HUMAN DISEASES ASSOCIATED WITH OXIDATIVE DAMAGE

Disease Links to oxidative stress Reference Neurodegenerative diseases, e.g., The central nervous system (CNS) is especially vulnerable to oxidative damage due to the high oxygen consumption (Emerit et al., 2004) Alzheimer’s disease, Parkinson’s disease, of the brain. Mitochondria are both the main source (through normal cellular respiration) and the main scavenger Huntington’s disease, Friedrich’s ataxia, of RONS. Cells modulate and counteract oxidative stress through coordinated, genetically-encoded defence amyotrophic lateral sclerosis, ageing brain, mechanisms. Excess levels of RONS beyond the cellular capacity to detoxify them activate apoptosis and AIDS dementia, and neurologic excitotoxicity (glutamate toxicity) processes, leading to neuronal cell death. Mitochondrial impairment is central to manifestations of mitochondrial dysfunction. all neurodegenerative diseases. While RONS do not directly cause neurodegenerative diseases or ageing, oxidative stress is clearly implicated in the pathology, and optimisation of antioxidant defences is expected to be therapeutic. The ‘neurotoxic trace element hypothesis’ for Alzheimer’s disease, which implicates elevated concentrations of Al, (Markesbery, 1997, Hg, and Fe in the neurodegenerative pathophysiology, relates to oxidative stress as forms of these elements can Praticò, 2000) act as catalysts for RONS production. The CNS has high lipid content, esp. PUFA that are relatively susceptible to oxidation, and lipid peroxidation is the predominant manifestation of oxidative stress in CNS tissues. Numerous markers of protein, DNA, and glyco-oxidation have also been identified. Environmental toxicity and carcinogenesis Environmental carcinogens such as cigarette smoking, alcohol consumption, radiation, xenobiotics, and chlorinated (De Bont and van compounds may cause cellular damage through RONS-mediated toxicity. Ultraviolet A and B radiation directly Larebeke, 2004, damage DNA and increase the accumulation of RONS in epidermal and tumour cells, while chronic exposure to air Ziech et al., 2010) pollutants is thought to cause irreversible damage to cellular macromolecules such as DNA, proteins, and lipids by inducing the production and accumulation of RONS in tissues. RONS can directly attack DNA to form oxidative DNA lesions that have been observed in controlled studies of human carcinogenesis. DNA adducts are prominent in carcinogenesis because they lead to miscoding and mutations if they escape cellular repair mechanisms and persist

| Antioxidant capacity in the body. Atherosclerosis and cardiovascular disease Macromolecule oxidation is implicated in the development of ‘foam cells’, which involves the deposition of cells (Dean et al., 1997) (CVD) and lipids on vessel walls and is associated with atherosclerosis. For example, about 30% of the cholesterol linoleate, which is found in atherosclerotic but not healthy vessel walls, is oxidised. Oxidised lipoproteins are taken up by macrophages and foam cells, forming sub-endothelial ‘fatty streaks’ that proliferate and form complex lesions and fibrous plaques that contribute to the thrombosis or persistent stenosis associated with CVD. Metabolic syndrome (MetS) MetS is described as accumulation of visceral fat with cardiovascular risk factors including impaired glucose (Otani, 2011) tolerance, dyslipidaemia, and hypertension. These factors feature oxidative stress-mediated endothelial dysfunction in common. 147

TABLE 4.1 CONTINUED

Disease Links to oxidative stress Reference Age-related ocular degeneration, e.g., The ocular lens is especially sensitive to oxidative damage due to the high exposure of eyes to UV radiation and low (Dean et al., 1997) cataract and age-related macular turnover of the crystallin proteins in the lens. Extensive oxidation leading to aggregation of denatured protein degeneration (AMD) products may restrict proteolytic access and thus inhibit cellular repair. This is clearly implicated in human cataract development. Respiratory disease, e.g., pollution or Air pollutants contribute to the development of chronic obstructive pulmonary diseases, CVD, asthma, and cancer. (Yang and Omaye, smoking-related conditions, chronic For example, ozone oxidises lipids and protein thiol groups, and decomposes to various by-products that also 2009) obstructive pulmonary disease, asthma, initiate oxidative damage. Nitrogen oxides have similar effects on lung tissue, esp. PUFA. Both ozone and neonatal lung disease, cystic fibrosis, and particulate matter (<2.5 μm) have been shown to initiate redox-sensitive transcription factors like nuclear factor-κB adult respiratory distress syndrome in airway epithelial cells, which activate pro-inflammatory cytokines that mediate inflammation and immune response. Sulfur oxides induce rapid bronchoconstriction, mucus production, coughing, and acute irritation, especially in the upper airway. Sulfur dioxide is even more toxic when reduced to a radical form in reaction with •- O2 , and may damage lung tissue through a radical mechanism. Renal disease Oxidative stress has been shown to mediate diverse renal impairments, including acute and chronic renal failure, (Singh et al., 2006) rhabdomyolosis, obstructive nephropathy, hyperlipidaemia, glomerular damage, and haemodialysis. The kidney is particularly vulnerable to lipid peroxidation due to its high PUFA content. Oxidative injury most frequently manifests in the glomerulus, tubular epithelia, and vascular epithelia of the renal system. | Antioxidant capacity 148

4.1.1.3 Dietary antioxidants and effects on human health and diseases Oxidative stress is certainly involved in the pathophysiology of many diseases, and dietary antioxidants in the body’s defence repertoire; however the extent to which disease outcomes can be influenced by dietary interventions is controversial. Studies are frequently inconsistent if not outright contradictory, and need to be viewed as a whole since each type of study is susceptible to different methodological problems.

Epidemiologic studies that track the health of large sample populations provide convincing observations of long-term dietary exposure to antioxidants, but suffer implicit problems with multicollinearity (e.g., effects could be due to any number of known and unknown bioactive compounds), confounding factors in populations that are subject to differing and indeterminate lifestyle factors (since even the best-designed prospective studies can only adjust for the most important factors such as cigarette smoking and alcohol consumption), and the publication bias of having relatively few researchers who conduct large cohort studies (Arts and Hollman, 2005). Controlled interventions can demonstrate causality, but often in short- term dosing regimens that do not reflect chronic dietary exposure and often rely on animal models or human biomarkers rather than disease outcomes.

Contrary to what one would expect, clinical trials of antioxidant interventions (e.g., vitamin supplements) have generally yielded disappointing results. The most convincing evidence has been produced by epidemiologic studies of dietary patterns and health. It seems that the interactions or synergies of the bioactive compounds with each other and with the food matrix are necessary to attain health benefits (Bhupathiraju and Tucker, 2011). Some researchers suggest that one of the essential physiological functions of dietary fibre is to transport polyphenolic antioxidants through the gastrointestinal tract, where non-extractable polyphenolic compounds (i.e., that are not extracted by aqueous solvent extraction) can be released by the activity of intestinal microflora (Saura-Calixto, 2011).

Table 4.2 compiles several reviews of epidemiologic studies of associations between dietary antioxidant intake and outcomes of oxidative stress-related disease. This is by no means exhaustive, but should reflect our current understanding of links between diet and chronic disease, and prospects for therapeutic antioxidant consumption.

| Antioxidant capacity 149

TABLE 4.2 REVIEW OF ANTIOXIDANT TREATMENTS/THERAPIES FOR DISEASES

Disease Interesting points Reference Cardiovascular disease (CVD), e.g., coronary Epidemiologic studies: Seven of twelve prospective cohort studies found significant correlations between flavonols, (Arts and Hollman, artery disease (CAD), coronary heart disease flavones, or catechins and protective effects on fatal or nonfatal CAD. Two of five studies found flavonoid had 2005) (CHD), stroke benefits for stroke risk. Cancer Epidemiologic studies: Two of seven prospective cohort studies found flavonoid intake conferred benefits for lung (Arts and Hollman, cancer risk and borderline benefits for colorectal cancer risk, but there were no significant results for other cancers 2005) such as epithelial, stomach, urinary tract, prostate, and breast. Respiratory diseases, e.g., chronic Epidemiologic studies: A Finnish prospective cohort study found that higher flavonoid intake reduced was (Kalantar-Zadeh et obstructive pulmonary disease (COPD), significantly correlated with lower asthma incidence. Kalantar-Zadeh et al. (2004) reviewed sixteen prospective or al., 2004, Knekt et asthma case-control studies that analysed the development and persistence of atopy and asthma in relation to dietary al., 2002) patterns. Some data suggests benefits on symptoms such as wheezing, bronchitis, lung function, and atopic symptoms (e.g., eczema), but the authors note that the studies have been in ‘prevalent’ asthma, i.e., in subjects over 6 years of age, when more than 80% children have already had their first manifestations of asthma or allergy, and there have been no studies on incidence of asthma. Type 2 diabetes mellitus Epidemiologic studies: A Finnish prospective cohort study found a trend between higher flavonoid intake and lower (Knekt et al., 2002) risk of type 2 diabetes. Alzheimer’s disease Epidemiologic studies: There have been three prospective dietary studies that investigated links to Alzheimer’s (Morris, 2009) disease. Two studies found that high vitamin E intake was associated with reduced risk of Alzheimer’s disease, but there were contradictory results for vitamin C. Several prospective studies explored the use of vitamin C and E

| Antioxidant capacity supplements in relation to risk of Alzheimer’s disease but did not have significant findings. So far, the most persuasive evidence is for high dietary vitamin E intake. Neurodegenerative diseases, e.g., Antioxidant therapies that have been tested in clinical or population studies include treatment with vitamin B, C, E, (Ienco et al., 2011) Alzheimer’s disease, Parkinson’s disease, and Se; lipoic acid and omega-3 fatty acids; flavonoids, e.g., epigallocatechin-gallate; carotenoids, e.g., vitamins A, amyotrophic lateral sclerosis, Huntington’s C, E, β-carotene, and lutein; cysteine donors, e.g., Gelsolin, which may enhance reduced glutathione (GSH) disease, and Friedrich’s ataxia. metabolism and thus improve cellular antioxidant defence; mitochondrial-targeted antioxidants, e.g., coenzyme Q10 (ubiquinone) and analogues (e.g., ubiquinol, Idebenone), MitoQ; mitochondrial membrane-targeted peptides, e.g., Szeto-Schiller peptide (SS-31); and Pioglitazone, an insulin sensitiser. Some studies are ongoing, but so far results have been inconsistent.

150

4.1.2 Assays of ‘total antioxidant capacity’

4.1.2.1 Types and chemistry of antioxidant assays There have been numerous reviews of in vitro antioxidant assays, corresponding with the exponential increase in their development and application in research over the past decade. Rather than reiterate these many articles, we have attempted to summarise diverse methods in Table 4.3, and recommend the recent publication by Gülçin (2012) for further reading, as it is the most comprehensive review thus far. In addition, Karadag et al. (2009) has done a good job of synthesising the chemical and practical information from previous reviews; while the articles by Huang et al. (2005), MacDonald-Wicks et al. (2006), and Prior et al. (2005) are also useful reading. Chemical-based assays dominate the literature, but there has been increased interest in the use of cell-based in vitro assays, which are expected to offer information more reflective of in vivo biological activity. This is likely to be a growth area in dietary antioxidant research. There is one extensive review by Cheli and Baldi (2011), and an overview of oxidative stress biomarkers by Wood et al. (2006).

| Antioxidant capacity 151

TABLE 4.3 REVIEW OF IN VITRO CHEMICAL ANTIOXIDANT ASSAYS

Assay type Method Description Additional reviews Chemiluminescence Luminol-based Oxidation of luminol or related compounds generates low-intensity chemiluminescent light emission. (Christodouleas et al., 2012, chemiluminescence The light emission is quenched by radical-scavenging antioxidants but returns when these are Vladimirov and Proskurnina, depleted. Peroxyl radical is commonly used, but the assay may be modified for use with different 2009) oxidants/initiators. DNA oxidation Comet aka single cell The comet assay is the most common method for assessing DNA damage. Cells are lysed and (Aruoma, 1999, Azqueta et gel electrophoresis electrophoresed. The nucleoids have a comet-like shape when stained with DNA-binding dye and al., 2009, Cemeli et al., 2009, (SCGE) assay viewed by fluorescence microscopy. The nucleoids can be digested with lesion-specific enzymes to Fairbairn et al., 1995, identify different types of base/strand damage. McKelvey-Martin et al., 1993) Lipid peroxidation Low density lipoprotein Conjugated diene peroxides formed during the autoxidation of linoleic acid or LDL are monitored (LDL) autoxidation spectrophotometrically (λmax 234 nm). Thiobarbituric acid Malondialdehyde is a marker of PUFA oxidation that reacts with thiobarbituric acid to form red reactive substances condensation products that can be monitored spectrophotometrically (λ 535 nm). (TBARS) Thiocyanate Peroxides formed during the oxidation of linoleic acid react with Fe2+ to form Fe3+. The latter forms a - complex with thiocyanate (SCN ) that can be monitored spectrophotometrically (λmax 500 nm). Radical scavenging Crocin bleaching The autoxidation of carotenoids following exposure to light, heat, or peroxyl radical causes bleaching (decolourisation) that is monitored by UV-Vis spectrometry (λ 443 nm). Crocin (a β-carotene derivative) has replaced β-carotene in this assay because it has more stable decolourisation. | Antioxidant capacity DMPD•+ scavenging N,N-dimethyl-p-phenylenediamine (DMPD) forms a stable coloured radical in the presence of Fe3+ at (Sánchez-Moreno, 2002) •+ (DMPD ) acidic pH. Radical scavenging causes decolourisation that is monitored spectrophotometrically (λmax 505 nm). The DMPD•+ assay is in aqueous phase and categorised as an electron transfer (ET) mechanism. DPPH• scavenging The DPPH· assay was first described by Brand-Williams (1995). 2,2-diphenyl-1-picrylhydrazyl (DPPH) (Bondet et al., 1997, Kedare (DPPH•) forms a stable nitrogen radical that is deep purple with in solution, which is reduced to yellow and Singh, 2011, Mishra et al., diphenyl-picrylhydrazine in methanol. Radical scavenging causes decolourisation that is monitored 2012, Villaño et al., 2007) spectrophotometrically (λmax 517 nm). The reaction occurs through two mechanisms: a hydrogen atom transfer (HAT) and a sequential proton loss electron transfer (SPLET). HAT: Ar-OH + DPPH• → Ar-O• + - + - • • - - +

152 DPPH-H. SPLET: Ar-OH → Ar-O + H ; Ar-O + DPPH → Ar-O + DPPH ; DPPH + H → DPPH-H.

TABLE 4.3 CONTINUED

Assay type Method Description Reviews Radical Hydroxyl radical (•OH) The research group that developed the ORAC assay described a similar assay that they named the hydroxyl radical averting (Ou et al., scavenging scavenging capacity (HORAC). •OH is generated by a Co(II)-mediated Fenton-like reaction. As in the ORAC assay, the oxidation of a 2002) fluorescent substrate (fluorescein) is monitored, and the presence of compounds with metal-chelating ability inhibits the radical-induced decay in fluorescence. Oxygen radical ORAC is a kinetic assay in which the peroxyl-induced oxidation of a fluorescent probe is monitored by decay in fluorescence. absorbance capacity Antioxidant capacity inhibits the decay in fluorescence, which is converted to standard equivalents by linear regression of the (ORAC) area under the curve and standard concentration. The assay has been improved since the original protocol by replacing β- phycoerythrin with fluorescein as the fluorescent probe, modifications for high throughput platforms, and for lipophilic antioxidants. ORAC has been characterised as a HAT mechanism. *- Superoxide radical anion O2 generated by hypoxanthine or xanthine/xanthine oxidase reduces cytochrome c or nitroblue tetrazolium, which can be •- (O2 ) scavenging monitored spectrophotometrically (λmax 550 or 560 nm) Trolox equivalence 2,2’-azinobis(3-ethylbenzothiazoline-6-sulfonic acid) radical cation (ABTS•+) is generated enzymatically (e.g., antioxidant capacity metmyoglobin/H2O2 and horseradish peroxidase/H2O2) or chemically (e.g., manganese dioxide and potassium persulfate) in •+ (TEAC or ABTS ) aqueous solution. Radical scavenging causes decolourisation that can be monitored spectrophotometrically (λmax 415, 734 nm). Total oxyradical Gas chromatography is used to measure the ethylene generated by the AAPH-induced oxidation of α-keto-γ- (Garrett et al., scavenging capacity (methylthio)butyric acid sodium salt (KMBA). Radical scavenging by antioxidants inhibits ethylene production. 2010) (TOSC) Total radical-trapping The TRAP assay was the first in vitro method to become popular (Wayner et al., 1985), and can be considered the precursor to | Antioxidant capacity antioxidant parameter many radical scavenging assays. The basic format entailed the competitive influence of peroxyl radical and radical-scavenging (TRAP) antioxidants on lipid peroxidation. TRAP was originally monitored by oxygen consumption using an oxygen electrode. Later improvements included the use of chemiluminescence to measure the reaction endpoint and alternative radical initiators. Reducing Cupric ion reducing The CUPRAC reagent is bis(neocuproine) copper(II) cation, which is reduced by antioxidants to form an orange-yellow Cu(I)- (Apak et al., power antioxidant power chelate that can be monitored spectrophotometrically (λ 450 nm). The CUPRAC assay is in aqueous ethanolic phase (pH 7) and 2005, Özyürek (CUPRAC) is suitable for hydrophilic and lipophilic antioxidants. et al., 2011) 3+ 3+ Ferric ion reducing Ferric 2,4,6-tripyridyl-s-triazine complex [Fe -(TPTZ)2] is reduced to a blue ferrous complex in acidic solution that can be antioxidant power monitored spectrophotometrically (λ 593 nm). The FRAP assay is based on a SET reaction. (FRAP)

153 •- Folin-Ciocalteu (FC) FC reagent is a mixture of phosphomolybdate and phosphotungstate that reacts with the O2 generated by phenolic oxidation

in alkaline solution to form blue products (e.g., molybdenum oxide), which are monitored spectrophotometrically (λmax 765 nm). The assay is in aqueous phase and based on SET at basic pH.

4.1.2.2 Applications of antioxidant assays Antioxidant assays have most commonly be used to compare contrasting samples. There have been several attempts, conspicuous for the large quantity of analyses that were required, to establish antioxidant capacity databases of commonly consumed food and beverages and, less often, plant extracts and supplements. For example, ORAC analyses of 326 foods composited from a nation-wide sampling scheme over multiple seasons were used to establish an ORAC database of commonly consumed foods in the USA, which was available at the USDA Nutrient Data Laboratory website until recently (U. S. Department of Agriculture and Agricultural Research Service, 2012, Wu et al., 2004a, Wu et al., 2004b). A Norwegian project conducted analyses of more than 3100 internationally-sourced food and beverage samples by a modified FRAP assay over an eight-year period, to establish an Antioxidant Food Database that will eventually be made available at the University of Oslo website (Carlsen et al., 2010). Similar efforts have been conducted in Italy, using a combination of TEAC (i.e., ABTS•+), TRAP, and FRAP assays (Pellegrini et al., 2003, Pellegrini et al., 2006); India, using FC and β- carotene/linoleate emulsion auto-oxidation assays (Saxena et al., 2007); and Russia, using an amperometric (i.e., electrochemical) method to measure antioxidant capacity (Yashin et al., 2010). There have also been numerous smaller scale studies of diverse sample types (e.g., Dudonné et al., 2009, Saxena et al., 2007, Velioglu et al., 1998). An online database (www.portalantioxidantes.com) of ORAC and FC measurements for more than 120 species/cultivars of fruit from the South Andes region of South America is currently available (Speisky et al., 2012).

Genotypic variation in antioxidant capacity has rarely been evaluated, though the number of studies has been increasing in recent years, probably associated with the development of in vitro screening assays that facilitated convenient and relatively high-throughput analyses. A summary of these publications, including the distribution and variation in antioxidant values according to common assay methods is collated in Table 4.4. There have been no publications on peanut and few publications on edible seed crops. These included small studies of apricot kernel, grape seed, pecan, pistachio, soybean, and wheat, and one large study of rice.

| Antioxidant capacity 154

TABLE 4.4 REVIEW OF GENOTYPIC VARIATION IN ANTIOXIDANT CAPACITY OF SEVERAL CROP SPECIES

Method Sample Antioxidant capacity Reference Genotypes Tissue Units Min - Max Mean ± SD RSD (%)

ABTS/TEAC 12 loquat (Eriobotrya japonica) Fruit TE (μmol g-1 fw) 1.3 - 3.3 2.2 ± 0.6 27 (Xu and Chen, 2011) 14 strawberry (Fragaria × ananassa) Fruit TE (μmol g-1 fw) 10 - 21 13 ± 2.8 22 (Capocasa et al., 2008) 481 rice (Oryza sativa) Grain TE (μmol g-1 dw) 0.1 - 55 4.1 ± 7.0 169 (Shen et al., 2009) 26 eggplant (Solanum melongena) Fruit TE (μmol g-1 fw) 2.7 - 8.2 4.4 ± 1.2 28 (Okmen et al., 2009) 12 grape (Vitis vinifera) Seed TE (mol g-1 dw) 2.5 - 4.1 3.2 ± 0.6 18 (Yemis et al., 2008) DPPH 6 pecan (Carya illinoinensis) Kernel TE (mg g-1 fw) 81 - 135 97 ± 18 18 (Villarreal-Lozoya et al., 2007) 12 loquat (E. japonica) Fruit TE (μmol g-1 fw) 1.5 - 4.2 2.5 ± 0.8 32 (Xu and Chen, 2011) 18 soybean (Glycine max) Seed % Abs 16 - 84 44 ± 24 54 (Kumar et al., 2010) 9 soybean (G. max) Seed % Abs 9 - 30 20 ± 8 42 (Kim et al., 2006a) 124 sweet potato (Ipomoea batatas) Root TE (μmol g-1 fw) 10 - 70 (Takahata et al., 2011) 8 pistachio (Pistachia vera) Kernel TE (μmol g-1 dw) 46 - 123 (Tsantili et al., 2011) -1 14 apricot (Prunus persica) Kernel EC50 (mg mL ) 44 - 123 64 ± 20 32 (Korekar et al., 2011)

| Antioxidant capacity 11 red wine grape (V. vinifera) Seed EC50 (%) 2.7 - 4.6 3.3 ± 0.6 18 (Bozan et al., 2008) 12 grape (V. vinifera) Seed TE (mol g-1 dw) 3.6 - 5.8 4.8 ± 0.7 14 (Yemis et al., 2008) 155

TABLE 4.4 CONTINUED

Method Sample Antioxidant capacity Reference Genotypes Tissue Units Min - Max Mean ± SD RSD (%)

FC 6 pecan (C. illinoinensis) Kernel CAE (mg g-1 fw) 62 - 106 76 ± 14 18 (Villarreal-Lozoya et al., 2007) 12 loquat (E. japonica) Fruit GAE (μg g-1 fw) 241 - 572 420 ± 104 25 (Xu and Chen, 2011) 12 strawberry (Fragaria × ananassa) Fruit CAE (mg g-1 fw) 1.6 - 5.9 3.3 ± 1.5 45 (Pincemail et al., 2012) 14 strawberry (Fragaria × ananassa) Fruit GAE (mg g-1 fw) 2.0 - 4.2 2.8 ± 0.7 24 (Capocasa et al., 2008) 18 soybean (G. max) Seed GAE (mg g-1 dw) 0.8 - 5.9 1.9 ± 1.4 74 (Kumar et al., 2010) 8 soybean (G. max) Seed GAE (mg g-1 fw) 1.2 - 5.4 (Whent et al., 2009) 12 tomato (Lycopersicon esculentum) Fruit CAE (μg g-1 fw) 92 - 270 172 ± 56 33 (George et al., 2004) 27 mulberry (Morus atropurpurea) Fruit GAE (mmol g-1 dw) 0. - 0.2 0.16 ± 0.06 35 (Isabelle et al., 2008) 481 rice (O. sativa) Grain CAE (mg g-1 dw) 1.1 - 12 2.0 ± 1.4 73 (Shen et al., 2009) 8 pistachio (P. vera) Kernel GAE (mg g-1 dw) 8.0 - 16 (Tsantili et al., 2011) 14 apricot (P. persica) Kernel GAE (mg g-1 dw) 0.9 - 1.6 1.3 ± 0.2 17 (Korekar et al., 2011) 26 eggplant (S. melongena) Fruit GAE (mg g-1 fw) 0.6 - 1.4 1.0 ± 0.2 24 (Okmen et al., 2009) -1

| Antioxidant capacity 13 potato (Solanum tuberosum) Tuber 5-CQAE (mg g dw) 1.4 - 27 6.9 (André et al., 2009) 6 wheat (Triticum aestivum) Grain GAE (μmol g-1 dw) 8.4 - 11 9.8 ± 1.2 12 (Okarter et al., 2010) 13 blueberry (Vaccinium spp.) Fruit GAE (mg g-1 fw) 2.0 - 5.9 3.4 (Howard et al., 2003) 11 red wine grape (V. vinifera) Seed GAE (mg g-1 dw) 79 - 155 113 ± 24 21 (Bozan et al., 2008) 12 grape (V. vinifera) Seed GAE (mg g-1 dw) 339 - 587 500 ± 71 14 (Yemis et al., 2008) FRAP 12 loquat (E. japonica) Fruit TE (μmol g-1 fw) 2.2 - 3.7 3.1 ± 0.5 17 (Xu and Chen, 2011) 18 soybean (G. max) Seed Fe2+ (μmol g-1 dw) 11 - 64 41 ± 15 36 (Kumar et al., 2010) 2+

156 12 tomato (L. esculentum) Fruit Fe (mM) 0.64 - 2.3 (George et al., 2004) -1

8 pistachio (P. vera) Kernel TE (μmol g dw) 51 - 133 (Tsantili et al., 2011) 14 apricot (P. persica) Kernel FeSO .7H O (μg mL-1) 154 - 244 193 ± 24 13 (Korekar et al., 2011) 4 2

TABLE 4.4 CONTINUED

Method Sample Antioxidant capacity Reference Genotypes Tissue Units Min - Max Mean ± SD RSD (%)

H-ORAC 22 broccoli (Brassica oleracea) Florets TE (μmol g-1 dw) 152 ± 43 28 (Eberhardt et al., 2005) 6 pecan (C. illinoinensis) Kernel TE (μmol g-1 fw) 373 - 817 583 ± 147 25 (Villarreal-Lozoya et al., 2007) 12 strawberry (Fragaria × ananassa) Fruit TE (μmol g-1 fw) 17 - 57 35 ± 11 30 (Pincemail et al., 2012) 8 soybean (G. max) Seed TE (μmol g-1 fw) 22 - 58 (Whent et al., 2009) 27 mulberry (M. atropurpurea) Fruit TE (mmol g-1 dw) 0.3 - 1.7 1.0 ± 0.4 38 (Isabelle et al., 2008) 30 potato (S. tuberosum) Tuber TE (μmol g-1 fw) 2.3 - 14 7.2 ± 3.3 47 (Brown et al., 2005) 13 potato (S. tuberosum) Tuber TE (μmol g-1 dw) 25 - 347 104 (André et al., 2009) 6 wheat (T. aestivum) Grain TE (μmol g-1 dw) 51 - 96 75 ± 15 21 (Okarter et al., 2010) 13 blueberry (Vaccinium spp.) Fruit TE (μmol g-1 fw) 21 - 73 37 (Howard et al., 2003) 11 red wine grape (V. vinifera) Seed TE (μmol g-1 dw) 1426 - 3009 2058 ± 450 22 (Bozan et al., 2008) L-ORAC 22 broccoli (B. oleracea) Florets TE (μmol g-1 dw) 28 ± 8 28 (Eberhardt et al., 2005) 30 potato (S. tuberosum) Tuber TE (nmol g-1 fw) 0.05 - 0.15 0.10 ± 0.03 28 (Brown et al., 2005) | Antioxidant capacity 157

4.2 Genotypic variation in antioxidant capacity of peanut kernels

4.2.1 Objectives

Plant breeding to improve the antioxidant profile of peanut kernels presents one approach to the eventual development of enhanced ‘functional food’ peanut products. Peanuts are bred conventionally in Australia, so it is crucial to establish the stability and levels of naturally- occurring genotypic variation among the germplasm collection in order to ascertain the potential to select for the trait. The primary objectives of this study were to evaluate the genotypic variation in kernel antioxidant capacity, identify several genotypes that displayed distinct antioxidant capacity phenotypes, and also to investigate some of the high-throughput methods for screening total antioxidant capacity (chapter 2.2).

4.2.2 Materials and methods

4.2.2.1 Instruments A PolarSTAR Optima multi-mode microplate reader (BMG Labtech GmbH, Ortenberg, Germany) installed with 485 nm and 520 nm filters and two onboard reagent injectors was used to measure fluorescence intensity in the ORAC assay. The assay was performed in black 96-well FLUOTRAC 200 (flat bottom, chimney well, medium-binding) microplates (655076, Greiner Bio-One GmbH, Germany).

A Molecular Devices SpectraMax M2 spectrophotometer (Molecular Devices, LLC, Sunnyvale, CA, USA) was used to measure absorbance in the ABTS•+, DPPH•, and FC assays. The assays were performed in clear 96-well (flat bottom, tissue culture suspension) microplates (83.1835.500, SARSTEDT AG and Co., Numbrecht, Germany).

4.2.2.2 Reagents Stock solutions of gallic acid (G7384), 6-hydroxy-2,5,7,8-tetramethylchroman-2-carboxylic acid (Trolox; 238813), and fluorescein (F2456) were purchased from Sigma Inc. (St. Louis, MO, USA) and frozen in aliquots until use. Radical-generating reagents (Sigma Inc.) including 2,2’- azobis(2-amidino-propane) dihydrochloride (AAPH; 11633), 2,2’-azino-bis-(3- ethylbenzothiazoline-6-sulfonic acid) (ABTS; A1888), and 2,2-diphenyl-1-picrylhydrazyl (DPPH; D9132) were made fresh at the time of analysis. Folin and Ciocalteu’s phenol reagent (FCR; F9252) was also from Sigma, Inc. (St. Louis, MO, USA), while sodium carbonate (AJA463) and potassium persulfate (AJA2504) were from Ajax FineChem Pty. Ltd. (Australia, part of Thermo Fisher Scientific Inc.). Randomly methylated β-cyclodextrin (RMCD) was obtained from Cyclodextrin Technologies Development Inc. (High Springs, FL, USA). Solutions for the ORAC

| Antioxidant capacity 158

assay were in phosphate buffer (75 mM, pH 7.4) made from potassium phosphate dibasic (Ajax Finechem Pty. Ltd., Australia) and sodium phosphate monobasic (Riedel-de Haën, Honeywell Specialty Chemicals International Inc., Seelze, Germany). The acetone, n-hexane, and methanol used in extraction were analytical grade.

4.2.2.3 Samples Samples comprised 32 full-season maturity (140 days; Arachis hypogaea ssp. hypogaea) and 24 ultra-early maturity (100-110 days; A. hypogaea ssp. fastigiata) genotypes from the APBP, which is collaboratively run by PCA, GRDC, and DEEDI. The genotypes represented varieties that are currently grown commercially in addition to new breeding lines progressing to prospective release (Table 4.5).

The peanuts were grown in triplicate randomly-allocated field plots under non-limiting (nutrient- and water-replete) conditions in 2008/09. Samples of sound, mature, jumbo and size 1 kernels (>11.9 mm for Virginia and >9.9 mm for Runner market types) were de-hulled, dried to approximately 7% moisture content in accordance with commercial practice, transported to UNSW, and stored in a cool room (3-5 °C) in vacuum-sealed polyethylene bags until the time of analysis. (NB, the ultra-early maturity genotypes were extracted and analysed by a UNSW undergraduate student, Hui Neng Wong, according to the methods described below but in duplicate rather than triplicate).

4.2.2.4 Preparation of crude extracts The extraction method was based on Prior et al. (2003). Peanut kernels (10 g) were embrittled in liquid nitrogen then pulverised in an electric coffee grinder after manual removal of the testa. Ground sample (1.0 g) was mixed with n-hexane (10 mL) in a 10-mL screw-cap tube, vortexed, sonicated (50 kHz, 37 °C, 5 min), and centrifuged (3500 rpm, 15 min). The supernatant was collected and the extraction repeated. The combined supernatant was evaporated by flushing with nitrogen gas, and re-solubilised in acetone (5.0 mL) and 1% (w/v) RMCD/methanol (5.0 mL) to produce the lipophilic extract. The peanut residue was flushed with nitrogen gas to evaporate any residual n-hexane and then extracted by sonication (50 kHz, 37 °C, 5 min) in 70:29.5:0.5 (v/v/v) acetone/water/acetic acid (10.0 mL). The tube was centrifuged (3500 rpm, 15 min) and the supernatant used as the hydrophilic extract.

4.2.2.5 Moisture content Sample moisture content was determined gravimetrically as described in chapter 2.1.2.6 and used to convert the values of antioxidant capacity to a dry weight basis.

| Antioxidant capacity 159

TABLE 4.5 PEANUT GENOTYPES ANALYSED IN GENOTYPE STUDY OF ANTIOXIDANT CAPACITY

Full-season genotypes Ultra-early genotypes

Genotype Source program Genotype Source program

Ashton DEEDI D136-p7-5 DEEDI

AT43 US private program D174-3-p8-7 DEEDI

Bruce University of Florida D174-7-p19-4 DEEDI

Cook University of Georgia D175-3-p17-3 DEEDI

D147-p3-115 DEEDI D188-1-p371-1 DEEDI

D147-p8-3H DEEDI D188-1-p371-2 DEEDI

D147-p8-6F DEEDI D189-p8-2 DEEDI

D164-p39-2 DEEDI D189-p8-3 DEEDI

D164-p50-4 DEEDI D191-p1-1 DEEDI

D197-9-p586-2 DEEDI D191-p1-3 DEEDI

D204-p6-3 DEEDI D191-p13-1 DEEDI

Farnsfield US private program D192-p386-1 DEEDI

Fisher North Carolina State University D192-p394-1 DEEDI

Holt University of Florida D192-p394-2 DEEDI

Lyons University of Florida D192-p394-3 DEEDI

Middleton DEEDI D192-p397-1 DEEDI

Page University of Florida D193-p3-6 DEEDI

PCA048 US private program D194-22-p209-2 DEEDI

PCA051 US private program D194-24-p362-1 DEEDI

PCA213 US private program D194-26-p136-2 DEEDI

Reid University of Georgia D194-2-p65-1 DEEDI

Scullin University of Georgia D194-2-p65-2 DEEDI

Sutherland DEEDI Tingoora DEEDI

UF11 University of Florida Walter DEEDI

UF16 University of Florida

UF31 University of Florida

UF32 University of Florida

UF36 University of Florida

UF37 University of Florida

UF38 University of Florida

UF39 University of Florida

Wheeler DEEDI

| Antioxidant capacity 160

4.2.2.6 ABTS•+ decolourisation assay The cuvette method described by Re et al. (1999) was scaled to a 96-well microplate. The ABTS•+ reagent was made 12-16 h in advance from ABTS (7 mM) and potassium persulfate (2.45 mM) in water. Each microplate well contained sample extract (10 μL) and diluted ABTS•+ reagent (190 μL, 1/40 in water). Absorbance at 734 nm was measured after 5 min. Trolox standards (100-1000 μM) were measured in quadruplicate in each microplate, while sample extracts were measured in triplicate. Calibration equations were calculated by linear regression of the absorbance and Trolox concentration, and antioxidant capacity of sample extracts estimated in terms of Trolox equivalents (TE).

4.2.2.7 DPPH•-scavenging assay The method described by Brand-Williams et al. (1995) was converted to a 96-well microplate format. DPPH• (0.2 mM) was prepared in methanol 30 min before use. The reaction mixture of sample extract (40 μL) and DPPH• reagent (160 μL) was sealed and incubated at room temperature for 30 min before measurement of absorbance at 517 nm. Gallic acid standards (10-100 μM) were measured in quadruplicate in each microplate, while sample extracts were measured in triplicate. Calibration equations were calculated by linear regression of the absorbance and gallic acid concentration, and antioxidant capacity estimated in terms of gallic acid equivalents (GAE).

4.2.2.8 FC ‘total phenols’ assay The version of the FC method (Singleton and Rossi, 1965) described by Waterhouse (2002) was adapted to a 96-well microplate format. Each well contained sample extract (30 μL), diluted FCR (140 μL, 1/14 in water), and the reaction begun with addition of 20% (w/v) aqueous sodium carbonate (30 μL). The microplate was incubated at room temperature for 2 h before measurement of absorbance at 765 nm. Gallic acid standards (100-500 μM) were measured in quadruplicate in each microplate, while sample extracts were measured in triplicate. Antioxidant capacity was estimated in terms of GAE based on calibration of the absorbance against standard concentrations.

4.2.2.9 ORAC assay The ORAC assay described by Prior et al. (2003) was optimised for our sample type and laboratory conditions. The reaction mixture comprised the standard or sample extract (25 μL), fluorescein (150 μL, 70 nM), and AAPH (25 μL, 140 mM). Standards and extracts were manually transferred to the microplate using a single-channel pipette, while the fluorescein and AAPH were delivered by the microplate reader’s onboard injectors. The fluorescein and AAPH were

| Antioxidant capacity 161

pre-incubated at 37 °C for 60 and 30 min prior to use. Efforts were made to minimise fluctuations in temperature. In particular, a 15 min delay in the initiation of the assay was incorporated into the plate reader program in order to allow equilibration of the plate temperature. More plate reader operating parameters are listed in Table 4.6. Trolox standards (10-100 μM) were measured in quadruplicate in each microplate, while sample extracts were measured in triplicate. Edge wells were excluded from use, and filled with water only to provide a thermal buffer. Calibration equations were calculated by linear regression of the area under the fluorescence decay curve and Trolox concentration (Figure 4.1), and antioxidant capacity estimated in terms of TE.

TABLE 4.6 ORAC PLATE READER OPERATING PARAMETERS

ORAC parameters

Temperature: 37 °C Excitation λ: 485 nm Emission λ: 520 nm Gain (signal amplification): 80% of fluorescein blank (~1500) Measurement: Top optic; 10 flashes Kinetic program: 1. Fluorescein (150 μL) injection 2. 15 min delay 3. AAPH (25 μL) injection 4. Intensity read at 1 min intervals for 2 h

8u10 04 100 PM Trolox 75 PM Tr olox 50 PM Tr olox 6 10 04 u 25 PM Tr olox 10 PM Tr olox Blank (phosphate buffer) 4u10 04

2u10 04 Relative fluorescence units Relative

0 0 30 60 90 120 Time (min)

FIGURE 4.1 KINETIC FLUORESCENCE DECAY OF TROLOX STANDARDS IN ORAC ASSAY. NB, AUC reflects change in both lag time to initiation of oxidation and rate of oxidation, and is thereby correlated to antioxidant capacity

| Antioxidant capacity 162

4.2.2.10 Statistical analysis Sample variances were analysed by standard ANOVA using IBM SPSS Statistics Version 20 (Release 18.0.0, IBM Corp., Armonk NY, USA). Pairwise comparisons of genotypes were made using Tukey’s HSD test.

4.2.3 Results and discussion

4.2.3.1 Variation in antioxidant capacity across genotypes The primary objective of this study was to establish the extent and significance of variation in antioxidant capacity among diverse genotypes that were representative of the APBP breeding population. The variation in antioxidant capacity according to each assay method is reported in Table 4.7 and depicted graphically in Figure 4.2. Data for full-season and ultra-early maturity genotypes are presented separately, as they represent distinct market types with differing morphology (e.g., number and size of kernels per pod), growing seasons, and end uses. Moreover, the ultra-early maturity genotypes were analysed by an undergraduate student with fewer replicates, and were probably subject to greater error than the full-season samples. In view of this, the ultra-early data is presented for comparative purposes, but the full-season data is emphasised in the interpretation and discussion of results.

-1 TABLE 4.7 VARIATION IN ANTIOXIDANT CAPACITY (μMOL G DW) OF 32 FULL-SEASON AND 24 ULTRA- EARLY MATURITY PEANUT GENOTYPES ACCORDING TO DIFFERENT ASSAYS

Antioxidant capacity Full-season maturity genotypes Ultra-early maturity genotypes Method Units Min - Max Mean ± SD RSD Min - Max Mean ± SD RSD (%) (%) ABTS•+ TE 0.3 - 18.9 7.1 ± 3.9 55 1.1 - 14.7 5.4 ± 2.9 53 DPPH• GAE 0.6 - 4.7 2.4 ± 0.8 31 1.1 - 6.0 2.6 ± 0.9 34 FC GAE 3.6 - 15.2 8.8 ± 2.1 24 4.6 - 23.3 10.7 ± 2.9 27 H-ORAC TE 11.5 - 43.5 23.8 ± 6.7 28 12.5 - 49.8 29.0 ± 7.6 26 L-ORAC TE 0.7 - 6.3 3.4 ± 1.3 37 0.8 - 6.0 3.0 ± 1.3 44 Total ORAC TE 13.1 - 46.1 27.3 ± 6.8 25 2.8 - 52.5 31.7 ± 8.2 26

ANOVA revealed that genotype significantly influenced (P<0.05) the hydrophilic and total ORAC values of the full-season breeding lines, but not the ABTS•+, DPPH•, and FC results in either data set (Table 4.8). The genotypic variation in ORAC results attested to a significant range in antioxidant capacity traits among the population, whereas the ABTS•+, DPPH•, and FC assays were subject to experimental variability that obscured genotypic differences. There was an approximately three-fold difference between the highest and lowest total ORAC values, and a RSD of more than 24%, depending on whether the full-season and ultra-early sample sets were viewed separately or combined (Table 4.7). This level of naturally-occurring genotypic

| Antioxidant capacity 163

variation provides a positive indication of potential to breed for high antioxidant capacity in peanut kernels.

TABLE 4.8 SIGNIFICANCE OF GENOTYPE ON ANTIOXIDANT CAPACITY IN PEANUT KERNELS

Method ANOVA P-values Full-season data Ultra-early data Combined data

ABTS•+ 0.804 0.796 0.741 DPPH• 0.105 0.788 0.247 FC 0.913 0.803 0.224 H-ORAC <0.001** 0.106 <0.001** L-ORAC 0.936 0.838 0.963 Total ORAC <0.001** 0.224 <0.001**

* Significant (P<0.05) ** Highly significant (P<0.01)

Antioxidant assays have frequently been used to confirm the fact of a given sample’s antioxidant capacity. There have also been several analysis-intensive attempts to create databases of antioxidant values for diverse samples, especially to facilitate comparisons of dietary antioxidant intake among consumer populations and the respective contribution of different food sources to this (Carlsen et al., 2010, Pellegrini et al., 2003, Pellegrini et al., 2006, Saxena et al., 2007, Wu et al., 2004a, Yashin et al., 2010).

In contrast, there has been relatively little research to establish the extent of genetic control or genotypic variation of antioxidant capacity among either breeding lines or commercial cultivars of the same species. Only a limited number of studies of edible seed crops are available in which the antioxidant capacity of different genotypes was assessed, including studies of apricot kernel, grape seed, pecan, pistachio, soybean and wheat, and one large study of rice (Table 4.4).

| Antioxidant capacity 164

. Median Median Mean Mean + +   1.5 IQR 1.5 IQR PEANUT GENOTYPES 3 1 3 1 Q Q Q Q Total ORAC Total ORAC 0 0

50 40 30 20 10 50 40 30 20 10

EARLY MATURITY MATURITY EARLY

P

dw) g mol ( TE P dw) g mol

TE ( TE - -1 -1 ULTRA 24

L-ORAC L-ORAC )

B

8 6 4 2 0 ( 8 6 4 2 0

P dw) g mol ( TE P dw) g mol ( TE

-1 -1 SEASON AND H-ORAC H-ORAC -

0 0

50 40 30 20 10

50 40 30 20 10

P dw) g mol ( TE P dw) g mol ( TE

-1 -1 FULL 32

) A ( FC FC

5 0 5 0

25 20 15 10

25 20 15 10

P dw) g mol ( GAE P dw) g mol ( GAE

-1 -1 DPPH DPPH

6 4 2 0 6 4 2 0

P dw) g mol ( GAE P dw) g mol ( GAE

-1 -1 ABTS ABTS

5 0 5 0

20 15 10 20 15 10

P P P P dw) g mol ( TE dw) g mol

TE ( TE OF CAPACITY ANTIOXIDANT IN ARIATION -1 -1 V

4.2 ) ) A B ( ( IGURE

F

| Antioxidant capacity 165

Large experimental variation constrained the usefulness of ABTS·+ and DPPH· assays in our study (chapter 2.2.2.1) and the ORAC data is therefore emphasised in our analysis. Of course, other researchers may have had different experiences. The number of genotypes tested and assay methods employed in the previous genotype studies varied, and the propriety of comparing results is slightly questionable. Nevertheless, it is worth noting that statistically significant differences were observed among H-ORAC values of seeds from six pecan cultivars (Villarreal-Lozoya et al., 2007), eight soybean genotypes (Whent et al., 2009), six wheat cultivars (Okarter et al., 2010), and eleven grape cultivars (Bozan et al., 2008); and that useful levels of variation of more than 20% and occasionally much higher were reported generally, by all assay methods in diverse crop species (Table 4.4).

4.2.3.2 Hydrophilic vs. lipophilic antioxidant capacity in peanut kernels Hydrophilic ORAC constituted 88 ± 5% of the total ORAC value in peanut kernels and was the predominant source of variation in the total ORAC measurement. The results highlighted the importance of hydrophilic antioxidants to the total antioxidant capacity in peanut kernels, and justified the sole use of the hydrophilic assay for screening peanut samples. This is not to dismiss the importance or breeding potential of the lipophilic fraction in peanuts, which after all composes about 50% of the peanut and has been shown to contain high and genetically variable concentrations of tocopherols, especially α- and γ-tocopherols (Dean et al., 2009). Moreover the importance of vitamin E, and α-tocopherol in particular, as an inhibitor of lipid peroxidation according to both in vitro and in vivo studies has long been widely accepted (Burton and Traber, 1990).

The constituents of the lipophilic extract were not quantitated, but several possibilities are suggested: the lipophilic antioxidant composition of our peanut samples contrasted with the American study (Dean et al., 2009) and simply did not vary substantially; the lipophilic antioxidants varied but the lipophilic ORAC assay was an insufficiently sensitive method to measure genotypic differences in general; or that the lipophilic ORAC assay is a sensitive method for some lipophilic antioxidants but not tocopherols, which happen to be the predominant constituent in lipophilic peanut fractions (Isanga and Zhang, 2007).

One of the latter two explanations seems likely, i.e., a feature of the method rather than the sample, as the ORAC assay is known to be highly sensitive to matrix effects (Pérez-Jiménez and Saura-Calixto, 2006). Indeed, marked difference in the hydrophilic and lipophilic reaction kinetics can be directly observed in the differing shapes of the respective fluorescence decay curves. For this reason, the hydrophilic and lipophilic ORAC assays are, strictly speaking, neither comparable nor additive. Yet summation to a total ORAC value is widespread and we | Antioxidant capacity 166

have followed the convention by presenting total ORAC alongside the hydrophilic and lipophilic values here.

Other researchers have also noted unexpectedly poor correlations between lipophilic ORAC results and concentrations of important lipophilic antioxidants. For example, the antioxidant capacity of lipophilic broccoli extracts correlated with relatively low concentrations of lutein and zeaxanthin, but had no correlation with α- and γ-tocopherols, which were the predominant lipophilic antioxidants by concentration (Kurilich et al., 2002). It is possible that the insignificant lipophilic antioxidant capacity results for peanut extracts reflected the inherent constraints of the ORAC assay, which, even with the use of RMCD to enhance the solubility of less-polar constituents, nevertheless remains an aqueous system more suited to the measurement of polar compounds.

Aguilar-Garcia et al. (2007) attempted to explain ABTS•+, FRAP, and ORAC results for rice bran and brown rice by PCA and multiple regression of the primary phytochemical antioxidants: γ- oryzanol, polyphenols, tocopherols, and tocotrienols. They contended that ORAC was sensitive to polyphenols and total tocopherols; FRAP was sensitive to polyphenols and total tocotrienols; while the ABTS•+ assay did not respond to any component particularly. This analysis is compelling because 96% of data variance was modelled by just three PC that diverged quite neatly into phytochemical groups. However, the results are surprising given that the extracts were hydrophilic (methanolic) and assays were in aqueous media, whereas the tested phytochemicals were quite hydrophobic (e.g., tocopherol has an intrinsic mass aqueous solubility of 5.2 μg L-1).

4.2.3.3 Comparison of genotypes and selection for further study Hydrophilic, lipophilic, and total ORAC values (mean and SEM) for each of the full-season genotypes are reported in Table 4.9 and depicted graphically in Figure 4.3; and the ultra-early genotypes in Table 4.10 and Figure 4.4. Full-season breeding lines that demonstrated relatively high antioxidant capacity included D164-p50-4, AT43, Sutherland, D147-p8-3H, and UF32; whereas Page, Lyons, Holt, and Reid had relatively low antioxidant capacity. The ORAC results were used to select peanut genotypes displaying diverse antioxidant phenotypes for use in the subsequent G × E study of antioxidant capacity. Ultra-early breeding lines could not be compared statistically due to insufficient replicates, and were simply ranked according to mean hydrophilic ORAC values for selection.

| Antioxidant capacity 167

50

40

dw) 30 -1

mol g mol 20 P P TE ( TE 10

0 Mean r SEM P Holt Reid Page AT43 UF11 UF16 UF31 UF32 UF36 UF37 UF38 UF39 Cook Bruce Lyons Fisher Scullin Ashton PCA048 PCA051 PCA213 Wheeler Middleton Farnsfield D204-p6-3 Sutherland D164-p39-2 D164-p50-4 D147-p8-6F D147-p8-3H D147-p3-115 D197-9-p586-2 Genotype FIGURE 4.3 H-ORAC ANTIOXIDANT CAPACITY OF 32 FULL-SEASON MATURITY PEANUT GENOTYPES. 50

40

dw) 30 -1 l g l

mo 20 P TE ( TE 10

0 Mean r SEM P Walter Tingoora D136-p7-5 D189-p8-2 D189-p8-3 D191-p1-1 D191-p1-3 D193-p3-6 D191-p13-1 D174-3-p8-7 D192-p386-1 D192-p394-1 D192-p394-2 D192-p394-3 D192-p397-1 D174-7-p19-4 D175-3-p17-3 D194-2-p65-1 D194-2-p65-2 D188-1-p371-1 D188-1-p371-2 Genotype D194-22-p209-2 D194-24-p362-1 D194-26-p136-2

FIGURE 4.4 H-ORAC ANTIOXIDANT CAPACITY OF 24 ULTRA-EARLY MATURITY PEANUT GENOTYPES. 4.2.4 Conclusion

A sample population of 32 full-season Australian peanut genotypes was evaluated to establish a ‘baseline’ level of variation in antioxidant capacity using the ABTS•+ decolourisation, DPPH• scavenging, FC ‘total phenols’, and both hydrophilic and lipophilic ORAC assays. There was a useful (>20%) level of phenotypic variation across the population, but only the hydrophilic ORAC assay revealed statistically significant differences between genotypes. The results affirmed that there is potential to breed for high antioxidant capacity traits in peanuts. The hydrophilic extract was found to be the predominant contributor to total antioxidant capacity in peanut kernels, according to the ORAC assay, which was the most useful screening method of the assays tested. Several genotypes that expressed relatively high antioxidant capacity were identified, including D164-p50-4, AT43, Sutherland, D147-p8-3H, and UF32. | Antioxidant capacity 168

-1 TABLE 4.9 ORAC ANTIOXIDANT CAPACITY (TE, μMOL G DW) OF 32 FULL-SEASON PEANUT GENOTYPES

GenotypeA H-ORACB L-ORAC Total ORACB Mean ± SD Rank Mean ± SD Mean ± SD

Ashton 25.43 ± 5.17 abcd 10 2.63 ± 1.53 28.06 ± 6.60 abcd AT43 34.12 ± 7.36 ab 2 2.61 ± 0.30 36.73 ± 7.14 ab Bruce 22.21 ± 8.20 bcd 19 3.18 ± 2.51 25.39 ± 6.28 bcd Cook 22.42 ± 2.27 bcd 17 3.16 ± 0.99 25.58 ± 2.57 bcd D147-p3-115 27.36 ± 8.82 abcd 7 3.43 ± 0.72 30.79 ± 8.55 abcd D147-p8-3H 29.97 ± 3.04 abc 4 3.31 ± 1.44 33.28 ± 1.93 abcd D147-p8-6F 28.08 ± 3.17 abcd 6 3.94 ± 0.99 32.02 ± 2.23 abcd D164-p39-2 22.38 ± 3.44 bcd 18 4.16 ± 1.95 26.54 ± 4.85 bcd D164-p50-4 39.71 ± 2.18 a 1 3.31 ± 0.53 43.02 ± 2.41 a D197-9-p586-2 22.08 ± 7.22 bcd 20 3.50 ± 1.52 25.58 ± 8.63 bcd D204-p6-3 24.71 ± 2.75 bcd 12 4.23 ± 1.01 28.94 ± 3.61 abcd Farnsfield 22.05 ± 5.58 bcd 21 3.51 ± 1.17 25.56 ± 4.81 bcd Fisher 20.34 ± 1.89 bcd 24 2.88 ± 0.24 23.22 ± 2.12 bcd Holt 18.68 ± 1.24 cd 30 3.53 ± 1.11 22.21 ± 2.29 bcd Lyons 17.76 ± 4.61 cd 31 2.68 ± 0.36 20.44 ± 4.96 cd Middleton 25.17 ± 2.06 abcd 11 4.07 ± 1.06 29.24 ± 3.08 abcd Page 14.23 ± 1.85 d 32 3.74 ± 0.82 17.97 ± 1.03 d PCA048 26.01 ± 1.75 abcd 9 2.74 ± 1.71 28.75 ± 2.27 abcd PCA051 23.57 ± 4.95 bcd 14 3.85 ± 1.20 27.42 ± 4.91 bcd PCA213 19.85 ± 1.97 bcd 28 3.71 ± 2.19 23.56 ± 4.12 bcd Reid 18.95 ± 1.05 cd 29 2.74 ± 1.12 21.69 ± 0.10 bcd Scullin 19.92 ± 1.21 bcd 27 3.71 ± 0.95 23.63 ± 0.82 bcd Sutherland 30.38 ± 8.26 abc 3 3.33 ± 1.58 33.71 ± 7.16 abc UF11 23.41 ± 5.37 bcd 16 2.79 ± 1.06 26.20 ± 5.68 bcd UF16 23.47 ± 0.85 bcd 15 3.12 ± 1.42 26.58 ± 1.78 bcd UF31 24.16 ± 5.21 bcd 13 2.36 ± 0.96 26.51 ± 6.17 bcd UF32 29.51 ± 5.64 abc 5 4.60 ± 1.32 34.11 ± 6.87 abc UF36 26.04 ± 5.08 abcd 8 4.57 ± 1.53 30.61 ± 4.59 abcd UF37 20.21 ± 3.24 bcd 25 3.95 ± 1.07 24.16 ± 3.34 bcd UF38 19.99 ± 4.11 bcd 26 3.17 ± 1.69 23.16 ± 5.58 bcd UF39 20.40 ± 4.87 bcd 23 3.06 ± 1.20 23.45 ± 4.94 bcd Wheeler 20.59 ± 3.62 bcd 22 4.08 ± 1.81 24.67 ± 4.31 bcd

A Boxed genotypes were used in the subsequent G × E study B Lower-case letters mark significantly different (P<0.05) means according to Tukey’s HSD test

| Antioxidant capacity 169

-1 TABLE 4.10 ORAC ANTIOXIDANT CAPACITY (TE, μMOL G DW) OF 24 ULTRA-EARLY PEANUT GENOTYPES

GenotypeA H-ORAC L-ORAC Total ORAC Mean ± SD Rank Mean ± SD Mean ± SD

D136-p7-5 36.17 ± 0.95 4 3.11 ± 2.45 39.28 ± 1.51 D174-3-p8-7 30.18 ± 10.13 10 2.47 ± 0.81 32.64 ± 9.32 D174-7-p19-4 32.47 ± 7.60 7 3.49 ± 0.81 35.95 ± 8.41 D175-3-p17-3 23.81 ± 18.29 19 3.14 ± 1.02 26.95 ± 17.27 D188-1-p371-1 23.42 20 4.47 27.89 D188-1-p371-2 38.45 1 2.25 40.70 D189-p8-2 32.90 ± 10.75 5 1.52 ± 0.29 34.42 ± 11.03 D189-p8-3 37.27 ± 4.28 3 2.95 ± 1.39 40.23 ± 3.01 D191-p1-1 25.36 ± 6.39 17 3.01 ± 2.21 28.36 ± 8.60 D191-p1-3 32.69 ± 2.35 6 3.12 ± 0.22 35.80 ± 2.57 D191-p13-1 20.71 ± 3.31 23 2.59 ± 2.15 23.30 ± 1.16 D192-p386-1 29.42 ± 5.73 12 2.09 ± 1.37 31.51 ± 4.37 D192-p394-1 37.93 ± 1.75 2 3.45 ± 0.47 41.38 ± 2.23 D192-p394-2 30.45 8 2.25 32.70 D192-p394-3 21.14 22 2.74 23.88 D192-p397-1 26.34 ± 4.52 15 2.79 ± 0.53 29.12 ± 5.05 D193-p3-6 29.72 ± 6.88 11 2.24 ± 0.62 31.96 ± 7.50 D194-22-p209-2 24.17 ± 7.65 18 2.78 ± 1.97 26.95 ± 9.62 D194-24-p362-1 27.28 ± 5.76 13 2.17 ± 0.11 29.45 ± 5.87 D194-26-p136-2 25.82 ± 1.95 16 3.84 ± 0.04 29.66 ± 1.99 D194-2-p65-1 18.10 ± 1.24 24 3.76 ± 2.28 21.86 ± 1.03 D194-2-p65-2 26.60 ± 6.21 14 3.50 ± 2.15 30.10 ± 8.37 Tingoora 30.39 ± 5.35 9 3.52 ± 0.32 33.91 ± 5.52 Walter 22.61 ± 1.90 21 3.31 ± 0.69 25.92 ± 1.21 A Boxed genotypes were used in the subsequent G × E study

| Antioxidant capacity 170

4.3 G × E interaction affecting antioxidant capacity of peanut kernels

4.3.1 Objectives

The principal objective of the G × E study was to evaluate the stability of peanut antioxidant phenotypes by determining the significance of genotype, environment, and G × E interaction components of variation on values of antioxidant capacity.

4.3.2 Materials and methods

4.3.2.1 Samples

TABLE 4.11 PEANUT GENOTYPES ANALYSED IN G × E STUDY OF ANTIOXIDANT CAPACITY

Maturity Genotype Market type Source program Registration Full- D147-p3-115 Runner DEEDI Breeding line season D147-p8-6F Runner DEEDI Breeding line D164-p50-4 Virginia DEEDI Breeding line Farnsfield Runner Private breeder, USA PBR: Plant Varieties Journal 2010, 23 (4) Fisher Virginia North Carolina State Uni PBR: Plant Varieties Journal 2009, 22 (3) Page Runner University of Florida PBR: Plant Varieties Journal 2009, 22 (3) PCA213 Runner Private breeder, USA Breeding line Sutherland Runner DEEDI PBR: Plant Varieties Journal 2002, 20 (2) UF32 Virginia University of Florida Breeding line UF37 Runner University of Florida Breeding line Ultra- D136-p7-5 Spanish DEEDI Breeding line early D175-3-p17-3 Spanish DEEDI Breeding line D188-1-p371-2 Spanish DEEDI Breeding line D189-p8-3 Spanish DEEDI Breeding line D193-p3-6 Spanish DEEDI Breeding line D194-22-p209-2 Spanish DEEDI Breeding line Walter Spanish DEEDI PBR: Plant Varieties Journal 2007, 20 (2) Tingoora Spanish DEEDI PBR: Plant Varieties Journal 2010, 23 (4)

Ten full-season and eight ultra-early maturity genotypes (Table 4.11) representing a large range of measured antioxidant capacity were harvested from five distinct growing environments in the principal peanut production regions of Australia in 2009/10 and 2010/11. The sites were Bundaberg Research Station (RS) and a commercial farm (Bundaberg Russo) in the coastal Burnett region of Queensland, Redvale RS and Taabinga RS near Kingaroy in the south Burnett, and Kairi RS in the Atherton Tablelands of northern Queensland. The peanuts were grown and handled as described in 4.2.2.3.

| Antioxidant capacity 171

4.3.2.2 Preparation of crude extracts Crude extracts were prepared as described in 4.2.2.4.

4.3.2.3 Moisture content Sample moisture content was determined gravimetrically as described in chapter 2.1.2.6 and used to convert the values of antioxidant capacity to a dry weight basis.

4.3.2.4 FC ‘total phenols’ assay The FC assay was performed as described in 4.2.2.8.

4.3.2.5 ORAC assay The hydrophilic ORAC assay was performed as described in 4.2.2.9.

4.3.2.6 Statistical analysis Sample variances were analysed by standard ANOVA using IBM SPSS Statistics Version 20 (Release 18.0.0, IBM Corp., Armonk NY, USA). Pairwise comparisons of genotypes and environments were made using Tukey’s HSD test.

4.3.3 Results and discussion

4.3.3.1 Importance of genotype, environment, and G × E interaction The principal objective of the G × E study was to investigate the stability of the antioxidant capacity phenotypes of peanut kernels by using standard ANOVA to partition variance into genotypic/genetic, environmental, and G × E interaction components (Table 4.12). Analysis of the 2008-2009 season data revealed that genotype and environment main effects on hydrophilic ORAC results were highly significant (P<0.001), but G × E interaction was not. However, when 2009-2010 season data were added to the analysis, all components of variation were found to be significant (P<0.05). The FC values were significantly influenced (P<0.05) by genotype and environment main effects, but not G × E interaction.

The results attest that the relative antioxidant rankings of genotypes (Table 4.13) were largely consistent despite a substantial environmental influence. This suggests strong genetic control of kernel antioxidant capacity, even though many of the known antioxidant phytochemicals in peanuts are highly influenced by common environmental phenomena such as pest or pathogen challenge (Sobolev, 2008, Sobolev et al., 2007, Sobolev et al., 2010), abiotic stresses such as UV irradiation and physical wounding (Rudolf and Resurreccion, 2005), and developmental regulation by plant hormones such as ethylene, salicylic acid, and jasmonic acid associated with such stresses (Chung et al., 2003).

| Antioxidant capacity 172

TABLE 4.12 SIGNIFICANCE OF GENOTYPE, ENVIRONMENT, AND G × E INTERACTION ON H-ORAC AND FC ANTIOXIDANT CAPACITIES

Method Source ANOVA P-values Full-season data Ultra-early data Combined data

ORAC Genotype <0.001** <0.001** <0.001** Environment <0.001** <0.001** <0.001** G × E interaction 0.002** 0.523 <0.001** FC Genotype 0.001** 0.496 0.007** Environment <0.001** 0.415 <0.001** G × E interaction 0.399 0.558 0.099 *Significant (P<0.05) **Highly significant (P<0.01)

As discussed in chapter 3.3.3.1, G × E interaction is important in plant breeding because it introduces uncertainty into the plant/seed selection process and therefore influences estimation of trait heritability and response to selection. In this study, G × E interaction reflected not only the stability of antioxidant capacity phenotypes, but also the suitability of ORAC (and potentially other in vitro) assays for screening what is not, in breeding or genetic terms, a clearly defined ‘trait’. In other words, ‘total antioxidant’ assays like ORAC measure the composite reactions of multitudinous phytochemicals within a complex matrix, whereby the synthesis and expression of each could potentially be influenced by distinct G × E interactions.

Few studies have investigated the G × E interactions affecting antioxidant capacity in crop species. Analysis of two tartary buckwheat cultivars grown in three locations revealed that significant G × E interaction affected DPPH• values as well as measurements of total flavonoids, phenolics, and phenolic acids, and the concentrations of the main antioxidant phytochemicals (i.e., catechin, ferulic acid, p-hydroxybenzoic acid, protocatechuic acid, quercetin, and rutin); however neither ABTS•+ nor DPPH• measures of antioxidant capacity correlated with any of these (Guo et al., 2011b). The antioxidant capacity of soybean and wheat genotypes were similarly subject to significant G × E interactions that did not appear to correlate with phytochemical constituents (Whent et al., 2009, Zhou and Yu, 2004). Our results confirmed the utility of the ORAC assay for screening peanut breeding populations, although it remains important to identify the chemical basis of contrasting antioxidant phenotypes.

| Antioxidant capacity 173

4.3.3.2 Genotype and environment main effects Genotype and environment both had highly significant (P<0.001) effects on ORAC-based antioxidant capacity. The main effects had statistically significant influence on the full-season FC results, but the distinctions between genotypes were not strong. Full-season genotypes with consistently high antioxidant capacity included D164-p50-4, UF32, and the sister-lines, D147-p3-115, D147-p8-6F, and Sutherland. Ultra-early genotypes that had outstanding antioxidant capacity included D136-p7-5 and D189-p8-3. Kairi RS stood out as a growing environment that consistently produced peanuts with elevated antioxidant capacity. These results may be viewed in Tables 4.13 and 4.14, and graphically in Figures 4.5 and 4.6.

TABLE 4.13 H-ORAC AND FC ANTIOXIDANT CAPACITIES OF PEANUT GENOTYPES TESTED IN THE G × E STUDY

Maturity Genotype ORAC, TE (μmol g-1 dw)A FC, GAE (μmol g-1 dw)A Mean ± SD Rank Mean ± SD Rank

Full-season D147-p3-115 23.2 ± 5.1 ab 4 9.2 ± 2.2 ab 3 D147-p8-6F 22.4 ± 4.8 abc 6 8.4 ± 2.9 ab 8 D164-p50-4 23.4 ± 5.0 ab 3 9.9 ± 1.8 a 1 Farnsfield 16.7 ± 5.3 def 13 6.6 ± 2.8 b 18 Fisher 20.4 ± 5.2 bcd 8 8.4 ± 3.0 ab 7 Page 12.8 ± 4.0 f 18 7.3 ± 2.7 ab 13 PCA213 16.4 ± 4.9 def 14 6.9 ± 1.6 ab 16 Sutherland 22.5 ± 4.6 abc 5 7.7 ± 3.7 ab 10 UF32 21.7 ± 4.3 abc 7 8.3 ± 3.5 ab 9 UF37 15.8 ± 4.3 ef 15 7.6 ± 1.7 ab 11 Ultra-early D136-p7-5 25.6 ± 4.1 a 1 8.5 ± 1.9 ab 6 D175-3-p17-3 17.5 ± 3.5 de 11 7.1 ± 2.5 ab 14 D188-1-p371-2 19.2 ± 4.1 cde 9 6.8 ± 2.5 ab 17 D189-p8-3 24.6 ± 4.9 a 2 9.3 ± 0.8 ab 2 D193-p3-6 17.6 ± 2.7 de 10 7.3 ± 3.1 ab 12 D194-22-p209-2 15.3 ± 2.2 ef 17 8.6 ± 1.7 ab 5 Tingoora 16.7 ± 2.4 def 12 7.0 ± 2.1 ab 15 Walter 15.6 ± 4.0 ef 16 8.7 ± 3.3 ab 4

A Lower-case letters mark significantly different (P<0.05) means according to Tukey’s HSD test.

| Antioxidant capacity 174

TABLE 4.14 H-ORAC AND FC ANTIOXIDANT CAPACITIES OF PEANUT KERNELS FROM EACH ENVIRONMENT TESTED IN THE G × E STUDY

Trial Region Season ORAC, TE (μmol g-1 dw)A FC, GAE (μmol g-1 dw)A Mean ± SD Rank Mean ± SD Rank

Bundaberg RS Coastal Burnett 2008/09 15.6 ± 4.7 e 10 8.2 ± 1.6 abc 5 Bundaberg RS Coastal Burnett 2009/10 18.3 ± 5.0 bcd 6 7.8 ± 2.8 abcd 6 Bundaberg Russo Coastal Burnett 2008/09 19.6 ± 5.9 bc 4 8.2 ± 2.1 abc 4 Bundaberg Russo Coastal Burnett 2009/10 17.5 ± 4.2 cde 8 5.8 ± 2.6 e 10 Kairi RS NQLD 2008/09 23.5 ± 5.5 a 2 7.6 ± 1.9 bcde 7 Kairi RS NQLD 2009/10 24.4 ± 4.4 a 1 9.4 ± 3.4 a 2 Redvale RS South Burnett 2008/09 20.3 ± 4.9 b 3 9.5 ± 2.5 a 1 Redvale RS South Burnett 2009/10 16.6 ± 5.6 de 9 6.8 ± 3.8 cde 8 Taabinga RS South Burnett 2008/09 19.5 ± 5.1 bc 5 9.2 ± 2.1 ab 3 Taabinga RS South Burnett 2009/10 18.0 ± 4.1 bcd 7 6.3 ± 2.3 de 9 A Lower-case letters mark significantly different (P<0.05) means according to Tukey’s HSD test. 4.3.3.3 Potential for genetic and agronomic manipulation of antioxidant capacity Significant (P<0.05) genotype effects according to ANOVA (Table 4.12) attested to strong genetic control of peanut antioxidant capacity despite substantial environmental influences. This is an exciting result that suggests it would be worthwhile to carry out further research towards the enhancement of antioxidant expression in peanut kernels by plant breeding.

Manipulation of antioxidant capacity through agronomic management should not be discounted, although G × E interactions are expected to be complex and call for a great deal more research. Poiroux-Gonord et al. (2010) published an interesting review of agronomic approaches to enhancing vitamins and phytochemicals in fruit and vegetables, including phenolics, flavonoids, and carotenoids that have relevance in antioxidant research. Observations cannot be extrapolated to peanuts since there were no edible seed crops reviewed. Yet it is noteworthy that factors such as (brief heat or cold) temperature shock, light intensity (exposure or type), elevated carbon dioxide, high salinity, and high nitrogen have had proven, if mixed, effects in increasing phenolic or flavonoid concentrations in diverse crop species.

| Antioxidant capacity 175

SEM SEM r r Mean Mean

. Walter

STUDY

Tingoora

G × E G ×

D194-22-p209-2

D193-p3-6 D189-p8-3

Ultra-early genotypes

D188-1-p371-2 D175-3-p17-3

5 0

0

15 10 D136-p7-5 30 20 10

P

dw) g mol ( GAE

P

dw) g mol ( TE -1

-1

UF37

UF32 Sutherland

ANTIOXIDANT CAPACITIES OF GENOTYPES TESTED IN THE PCA213

FC

Page ) B

( Fisher

AND AND Farnsfield genotypesFull-season H-ORAC

) D164-p50-4 A

(

D147-p8-6F 4.5 5 0

0

15 10

30 20 10

P P dw) g mol ( GAE D147-p3-115

P P dw) mol ( TE g -1 -1 IGURE F

) ) A B ( (

| Antioxidant capacity 176

SEM SEM r r Mean Mean

ES OF PEANUT PEANUT OF ES

Taabinga RS Taabinga

STUDY

Redvale RS Redvale Kairi RS Kairi G × E 2010 0 5 0

30 20 10

15 10

ANTIOXIDANT CAPACITI ANTIOXIDANT Bundaberg Russo Bundaberg

P P dw) g mol ( GAE dw) g mol ( TE

-1 -1 Bundaberg RS Bundaberg FC

) B (

AND AND

Taabinga RS Taabinga Redvale RS Redvale H-ORAC

) Kairi RS Kairi A 2009 (

4.6 0 5 0

30 20 10

15 10

Bundaberg Russo Bundaberg

P P dw) g mol ( GAE P P dw) g mol

TE ( TE IGURE -1 -1 Bundaberg RS Bundaberg F KERNELS FROM ENVIRONMENTS TESTED IN THE THE IN TESTED ENVIRONMENTS FROM KERNELS

) ) A B ( (

4.3.4 Conclusion

A multi-environment study of antioxidant capacity in eighteen diverse peanut genotypes was conducted. Genotype, environment, and G × E interaction were all found to have significant effects on ORAC-based antioxidant capacity according to standard ANOVA; whereas effects on FC total phenols value were less significant. Several current Australian peanut accessions demonstrated relatively high levels of antioxidant capacity, i.e., D136-p7-5, D189-p8-3, D164- p50-4, D147-p3-115, Sutherland, D147-p8-6F, and UF32. The ORAC assay was found to be a useful screening method for potential high-throughput evaluation of peanut breeding populations, even though it is a composite measure and it remains necessary to identify the specific chemical constituents of extracts.

| Antioxidant capacity 177

5 Antioxidant phytochemicals

5.1 Literature review

5.1.1 Phenolic acids in peanuts

Peanut kernels contain numerous phenolic compounds, many of which are expected to have potent in vivo antioxidant capacity. Phenolic acids, especially the derivatives of benzoic and cinnamic acids, are a major class of phenolic compounds present in peanut kernels. Several studies have attempted to characterise the phenolic acid profile of peanut kernels, mainly by liquid chromatographic separation of aqueous methanolic extracts (Table 5.1).

The concentration of total phenolic acids in peanut kernels is high compared to many other oilseeds and legumes. For example, in a study of ten defatted oilseed flours, peanuts (1755 μg g-1) contained high levels of phenolic acids compared to coconut (95 μg g-1), glanded (10.4 μg g-1) or glandless (12.4 μg g-1) cottonseed, sesame (227 μg g-1), soybean (736 μg g-1), flax (811 μg g-1), and safflower (1366 μg g-1). Only mustard (10217 μg g-1), rapeseed (11661 μg g-1), and sunflower (10037 μg g-1) had a higher total phenolic acid content (Dabrowski and Sosulski, 1984). Using the same methodology, peanuts were also found to be richer in phenolic acids than mungbean (18 μg g-1), field pea (26 μg g-1), lentil (30 μg g-1), fababean (31 μg g-1), pigeon pea (18 μg g-1), navy bean (55 μg g-1), lupin (71 μg g-1), lima bean (88 μg g-1), chickpea (95 μg g-1), and cowpea (163 μg g-1) (Sosulski and Dabrowski, 1984).

Several studies have found the predominant phenolic compound in peanut kernels by concentration to be p-coumaric acid. For example, Talcott et al. (2005b) determined by HPLC quantification that p-coumaric acid accounted for 31-73% (w/w) of total phenolics in raw kernels of 12 distinct genotypes. The peanut sample in a basket study of Finnish vegetables determined that p-coumaric acid represented approximately 73% of the total phenolic acid content (Mattila and Hellström, 2007). An earlier study of defatted oilseed flours (Dabrowski and Sosulski, 1984) found that p-coumaric acid constituted 83% of the total 80% ethanol- soluble phenolic content.

The dominance of p-coumaric acid in the phenolic profile of peanuts does not adequately explain their antioxidant capacity. Duncan et al. (2006) used solid phase fractionation combined with HPLC and assays of total phenols, free amino acids, and antioxidant capacity to deduce the contribution of various aqueous isolates to the antioxidant capacity of the crude extract. The majority of p-coumaric acid as well as 95% of ninhydrin-reactive amino acids were fractionated into the C18-non-retained isolate, which had the lowest antioxidant capacity. The

| Antioxidant phytochemicals 178

C18-retained isolate had the greatest antioxidant capacity, which therefore could not be explained by the presence of soluble proteins or p-coumaric acid. Instead, the fraction had numerous compounds present in low concentrations that could not be identified by HPLC alone. More thorough characterisation of the phenolic profile and a more exact method of matching compounds against their antioxidant capacity are required.

The majority of phenolic acids are present in bound forms that can be released by acid and/or alkaline hydrolysis (Dabrowski and Sosulski, 1984). In Mattila and Hellström’s (2007) basket study of Finnish vegetables, only a small portion of the total phenolic acids (1060 μg g-1) in peanut were in aglycon form (77 μg g-1), according to quantification based on the most similar external standard when the phenolic acid could not be identified. There is evidence that a substantial proportion of bound phenolic acids may be released during thermal processing. For example, the average concentration of total phenolic compounds across 12 cultivars, analysed without hydrolysis, rose from 44 μg g-1 (range 23.6-89.8 μg g-1) to 153 μg g-1 (range 35.8-227 μg g-1) after dry-roasting. The concentration of p-coumaric acid experienced a similar increase, from an average of 25 μg g-1 (range 8.3-65.9 μg g-1) to 69 μg g-1 (range 19.5-117 μg g-1), presumably because esterified and bound forms of these compounds were released under roasting conditions (Talcott et al., 2005b).

Other phenolic acids that have been identified in peanut kernels include ferulic, caffeic, sinapic, vanillic, gentisic, m-coumaric, o-coumaric, and other hydroxybenzoic acids (Dabrowski and Sosulski, 1984, Kozlowskra et al., 1983). Many phenolic acids have been observed in the HPLC chromatograms, but have so far only been tentatively identified without the use of more conclusive techniques (Mattila and Hellström, 2007, Talcott et al., 2005a, Talcott et al., 2005b).

| Antioxidant phytochemicals 179

TABLE 5.1 REVIEW OF PHENOLIC ACIDS IN PEANUT KERNELS

Analyte Sample Method Concentration (μg g-1) Reference Type N Extract Detection Mean ± SD Min - Max Basis

Caffeic acid Raw kernels 1 wholesale 80% aq. ethanol extract; Spectro- 28 dw (Dabrowski and sample alkaline hydrolysis photometry Sosulski, 1984)

Raw kernels 1 research 80% aq. methanol extract; GC Chlorogenic acid Raw kernels 1 research 80% aq. methanol extract; GC

Ferulic acid Raw kernels 1 wholesale 80% aq. ethanol extract; GC 162 dw (Dabrowski and sample alkaline hydrolysis Sosulski, 1984)

| Antioxidant Raw kernels 1 research 80% aq. methanol extract; GC Gentisic acid Raw kernels 1 research 80% aq. methanol extract; GC 28 dw (Kozlowskra et al., phytochemicals sample alkaline hydrolysis 1983) m-coumaric acid Raw kernels 1 research 80% aq. methanol extract; GC 41 dw (Kozlowskra et al., (defatted) sample alkaline hydrolysis 1983)

180

TABLE 5.1 CONTINUED

Analyte Sample Method Concentration (μg g-1) Reference Type N Extract Detection Mean ± SD Min - Max Basis

o-coumaric acid Raw kernels 1 research 80% aq. methanol extract; GC 36 dw (Kozlowskra et al., sample alkaline hydrolysis 1983) p-coumaric acid Raw kernels 12 genotypes 80% aq. methanol extract HPLC-PDA 28.3 ± 7.7 8.3 - 65.9 dw (Talcott et al., 2005b) Roasted kernels 12 genotypes 80% aq. methanol extract HPLC-PDA 78.5 ± 12 19.5 - 117 dw (Talcott et al., 2005b) Raw kernels 1 wholesale 80% aq. ethanol extract; GC 1464 dw (Dabrowski and sample alkaline hydrolysis Sosulski, 1984) Raw kernels 1 research 80% aq. methanol extract; GC 25 dw (Kozlowskra et al., sample alkaline hydrolysis 1983) (Raw?) kernels 12 retail Methanol extract; alkaline HPLC-PDA 770 ± 31 fw (Mattila and samples and acid hydrolysis Hellström, 2007) p-Hydroxybenzoic acid Raw kernels 1 wholesale 80% aq. ethanol extract; GC 20 dw (Dabrowski and | Antioxidant sample alkaline hydrolysis Sosulski, 1984) Raw kernels 1 research 80% aq. methanol extract; GC 259 dw (Kozlowskra et al., sample alkaline hydrolysis 1983) (Raw?) kernels 12 retail Methanol extract; alkaline HPLC-PDA 9.8 ± 0.10 fw (Mattila and phytochemicals samples and acid hydrolysis Hellström, 2007) Protocatechuic acid Raw kernels 1 research 80% aq. methanol extract; GC

181

TABLE 5.1 CONTINUED

Analyte Sample Method Concentration (μg g-1) Reference Type N Extract Detection Mean ± SD Min - Max Basis

Salicylic acid Raw kernels 1 research 80% aq. methanol extract; GC 32 dw (Kozlowskra et al., sample alkaline hydrolysis 1983) Sinapic acid Raw kernels 1 wholesale 80% aq. ethanol extract; GC 81 dw (Dabrowski and sample alkaline hydrolysis Sosulski, 1984) Raw kernels 1 research 80% aq. methanol extract; GC 115 dw (Kozlowskra et al., sample alkaline hydrolysis 1983) (Raw?) kernels 12 retail Methanol extract; alkaline HPLC-PDA 140 ± 3.5 fw (Mattila and samples and acid hydrolysis Hellström, 2007) Syringic acid (Raw?) kernels 12 retail Methanol extract; alkaline HPLC-PDA

phytochemicals 182

5.1.2 Flavonoids in peanuts

There are few publications regarding the flavonoid content of peanut kernels. This does not seem to be because of a lack of flavonoids in kernels, but perhaps because the levels in peanut testa and hulls are comparatively much higher on a dry weight basis and the potential opportunity to utilise commercial waste products for nutraceutical development is attractive. Flavonoids identified in peanuts so far include several flavonols (Table 5.2) and isoflavones (Table 5.3) in the kernels, a range of flavan-3-ols especially proanthocyanidins in the testa, and various flavonoid phytoalexins in the hulls. A summary of lignans observed in peanut kernels is also presented (Table 5.4), as these have often been reported in phytoestrogen research in conjunction with isoflavones.

A study of diverse legume germplasm (Wang et al., 2008a) indicated that the dominant flavonoid in peanut kernels by concentration was the flavonol, quercetin, which accounted for 86 and 96% of the total flavonoid content in the two seasons tested. The quercetin concentration in peanut (133-289 μg g-1) was significantly higher than in soybean (47-55 μg g-1), mungbean (17-31 μg g-1), and lablab (35-39 μg g-1), although soybean still had the highest total flavonoid content. There was large genetic variation for quercetin content, i.e., 13-360 μg g-1 in raw kernels of 24 peanut breeding lines, suggesting good breeding potential for this trait although seasonal variation was also high. Elevated quercetin concentration was apparently associated with white seedcoat colour.

Other flavonols present at low or trace levels included kaempferol, myricetin, rhamnetin (methylquercetin), and isorhamnetin, and the isoflavones, daidzein, genistein, genistin, and glycitin (Singleton et al., 2002, Wang et al., 2008a, Wang et al., 2007). Flavonols and isoflavones mainly occur as glycoside conjugates in peanut, although aglycones are also present. For example, several distinct di- and mono-glucosides of quercetin and an isorhamnetin glucoside have been characterised by mass spectrometry in methanolic peanut extract (Singleton et al., 2002).

Liggins et al. (2000) conducted an extensive survey of raw and cooked food products for daidzein and genistein content using GC-MS, including raw and roasted peanuts and peanut butter. Genistein was the dominant isoflavone in this study, i.e., 158 ng g-1 compared to 77 ng g-1 daidzein. The effects of processing were unclear (e.g., reduced genistein in peanut butter compared to raw kernels, but increased genistein in roasted peanut); however only a few unrelated samples were measured and the study was not intended as a controlled investigation of processing effects.

| Antioxidant phytochemicals 183

Mazur et al. (1998) determined that secoisolariciresinol, the immediate precursor of the mammalian lignan enterodiol, was the main phytoestrogen in peanut. Genistein and daidzein, as well as formononetin and biochanin A in lower concentrations, were important flavonoid contributors to total phytoestrogen content. It is interesting to note that in this study as well as in subsequent analyses (Mazur and Adlercreutz, 1998), the published isoflavone concentrations were very low: some one to two orders of magnitude lower than in the aforementioned study by Wang et al. (2008a). This is probably due to differences in methodology, as Mazur and co-workers used GC-MS whereas Wang’s group used HPLC for quantification, and the magnitude differences in concentration applied to food samples across the board and not just peanut. Further controlled studies would be worthwhile to confirm the levels of sample variation, and the effects of processing, handling, and other factors.

| Antioxidant phytochemicals 184

TABLE 5.2 REVIEW OF FLAVONOL CONCENTRATIONS IN PEANUT KERNELS

Analyte Sample Method Concentration (μg g-1) Reference Type N Extraction Analysis Mean ± SD Min - Max Basis

Kaempferol Raw kernels 24 genotypes 80% aq. methanol extract; HPLC LowA fw (Wang et al., 2007) acid hydrolysis Raw kernels 8 genotypes; 2 80% aq. methanol extract; HPLC 4.66 fw (Wang et al., 2008a) seasons acid hydrolysis 1.92 Myricetin Raw kernels 24 genotypes 80% aq. methanol extract; HPLC

| Antioxidant A Only described as a ‘low concentration’ and not quantified phytochemicals 185

TABLE 5.3 REVIEW OF ISOFLAVONE CONCENTRATIONS IN PEANUT KERNELS

Analyte Sample Method Concentration (μg g-1) Reference Type N Extraction Analysis Mean ± SD Min - Max Basis

Biochanin A Raw kernels 5 retail samples 10% aq. methanol extract; LC-MS/MS 0.08 fw (Kuhnle et al., 2008) enzymatic (β- Dry-roasted 0.22 fw glucuronidase, cellulase, kernels and β-glucosidase) Roasted kernels hydrolysis 0.21 fw Peanut butter 0.01 fw Raw kernels (3?) commercial Diethyl ether extract; acid ID-GC- 0.065 dw (Mazur et al., 1998) samples hydrolysis MS/SIM (Raw?) kernels ? Diethyl ether extract; acid ID-GC- 0.31 dw (Mazur and hydrolysis MS/SIM Adlercreutz, 1998) Coumestrol Raw kernels 5 retail samples 10% aq. methanol extract; LC-MS/MS 0.02 fw (Kuhnle et al., 2008) enzymatic (β- Dry-roasted 0.01 fw glucuronidase, cellulase, kernels and β-glucosidase) | Antioxidant Roasted kernels hydrolysis

TABLE 5.3 CONTINUED

Analyte Sample Method Concentration (μg g-1) Reference Type N Extraction Analysis Mean ± SD Min - Max Basis

Daidzein Raw kernels 5 retail samples 10% aq. methanol extract; LC-MS/MS <0.01 fw (Kuhnle et al., 2008) enzymatic (β- Dry-roasted <0.01 fw glucuronidase, cellulase, kernels and β-glucosidase) Roasted kernels hydrolysis 0.04 fw Peanut butter <0.01 fw Raw kernels 5 retail samples 80% aq. methanol extract; GC-MS 0.077 dw (Liggins et al., 2000) enzymatic (cellulase) Roasted kernels 0.037 dw hydrolysis Peanut butter

TABLE 5.3 CONTINUED

Analyte Sample Method Concentration (μg g-1) Reference Type N Extraction Analysis Mean ± SD Min - Max Basis

Formononetin Raw kernels 5 retail samples 10% aq. methanol extract; LC-MS/MS 0.04 fw (Kuhnle et al., 2008) enzymatic (β- glucuronidase, cellulase, and β-glucosidase) hydrolysis Dry-roasted 0.01 fw kernels Roasted kernels 0.03 fw Peanut butter 0.03 fw Raw kernels (3?) commercial Diethyl ether extract; acid ID-GC- 0.068 dw (Mazur et al., 1998) samples hydrolysis MS/SIM Genistein Raw kernels 5 retail samples 10% aq. methanol extract; LC-MS/MS 0.48 fw (Kuhnle et al., 2008) enzymatic (β- Dry-roasted 0.58 fw | Antioxidant glucuronidase, cellulase, kernels and β-glucosidase) Roasted kernels hydrolysis 0.70 fw Peanut butter 0.36 fw Raw kernels 5 retail samples 80% aq. methanol extract; GC-MS 1.58 dw (Liggins et al., 2000) phytochemicals enzymatic (cellulase) hydrolysis Roasted kernels 1.72 dw Peanut butter 0.098 dw 188

TABLE 5.3 CONTINUED

Analyte Sample Method Concentration (μg g-1) Reference Type N Extraction Analysis Mean ± SD Min - Max Basis

Genistein (Raw?) kernels ? Diethyl ether extract; acid ID-GC- 0.64 dw (Mazur and hydrolysis MS/SIM Adlercreutz, 1998) Raw kernels (3?) commercial Diethyl ether extract; acid ID-GC-MS- 0.826 dw (Mazur et al., 1998) samples hydrolysis SIM Roasted kernels 1 retail sample 70% aq. methanol extract; GC-MS 0.049 fw (Thompson et al., alkaline and enzymatic (β- 2006) glucuronidase) hydrolysis Peanut butter 1 retail sample 0.382 fw Raw kernels 24 genotypes 80% aq. methanol extract; HPLC

| Antioxidant Glycitein Raw kernels 5 retail samples 10% aq. methanol extract; LC-MS/MS 0.10 fw (Kuhnle et al., 2008) enzymatic (β- Dry-roasted 0.12 fw glucuronidase, cellulase, kernels and β-glucosidase) Roasted kernels hydrolysis 0.56 fw

phytochemicals Peanut butter 0.21 fw Roasted kernels 1 retail sample 70% aq. methanol extract; GC-MS 0.004 fw (Thompson et al., alkaline and enzymatic (β- 2006) glucuronidase) hydrolysis Peanut butter 1 retail sample 0.002 fw A Only described as a ‘low concentration’ and not quantified 189

TABLE 5.4 REVIEW OF LIGNAN CONCENTRATIONS IN PEANUT KERNELS

Analyte Sample Method Concentration (μg g-1) Reference Type N Extraction Analysis Mean ± SD Min - Max Basis

Cyclolariciresinol (Raw?) kernels 1 retail sample ASE; enzymatic (β- LC-MS/MS 0.43 dw (Smeds et glucuronidase/sulfatase) hydrolysis al., 2007) (-)-7- (Raw?) kernels 1 retail sample ASE; enzymatic (β- LC-MS/MS 0.20 dw (Smeds et Hydroxymatairesinol glucuronidase/sulfatase) hydrolysis al., 2007) (+)-Lariciresinol (Raw?) kernels Composite 70% aq. methanol extract; alkaline LC-MS/MS 0.41 fw (Milder et retail sample and enzymatic (β- al., 2005) glucuronidase/sulfatase) hydrolysis (Raw?) kernels 1 retail sample Alkaline then acid extraction; LC-MS/MS 0.98 dw (Smeds et enzymatic (β- al., 2007) glucuronidase/sulfatase) hydrolysis Roasted kernels 1 retail sample 70% aq. methanol; alkaline and GC-MS 0.009 fw (Thompson enzymatic (β-glucuronidase) et al., 2006) hydrolysis

| Antioxidant Peanut butter 1 retail sample 0.088 fw Lariciresinol- (Raw?) kernels 1 retail sample Alkaline extraction; enzymatic (β- LC-MS/MS 0.028 dw (Smeds et sesquilignan glucuronidase/sulfatase) hydrolysis al., 2007) (-)-Matairesinol Raw kernels 5 retail samples 10% aq. methanol extract; LC-MS/MS 0.05 fw (Kuhnle et enzymatic (β-glucuronidase, al., 2008) phytochemicals Dry-roasted 0.07 fw cellulase, and β-glucosidase) kernels hydrolysis Roasted kernels 0.17 fw Peanut butter 0.10 fw Raw kernels (3?) commercial Diethyl ether extract; acid ID-GC-

TABLE 5.4 CONTINUED

Analyte Sample Method Concentration (μg g-1) Reference Type N Extraction Analysis Mean ± SD Min - Max Basis

(-)-Matairesinol (Raw?) kernels ? Diethyl ether extract; acid ID-GC-

| Antioxidant glucuronidase/sulfatase) hydrolysis al., 2007) 7-Oxomatairesinol (Raw?) kernels 1 retail sample ASE; enzymatic (β- LC-MS/MS 0.019 dw (Smeds et glucuronidase/sulfatase) hydrolysis al., 2007) (+)-PInoresinol (Raw?) kernels Composite 70% aq. methanol extract; alkaline LC-MS/MS

TABLE 5.4 CONTINUED

Analyte Sample Method Concentration (μg g-1) Reference Type N Extraction Analysis Mean ± SD Min - Max Basis

(-)-Secoisolariciresinol Raw kernels 5 retail samples 10% aq. methanol extract; LC-MS/MS 0.97 fw (Kuhnle et enzymatic (β-glucuronidase, al., 2008) Dry-roasted 0.71 fw cellulase, and β-glucosidase) kernels hydrolysis Roasted kernels 2.56 fw Peanut butter 0.69 fw Raw kernels (3?) commercial Diethyl ether extract; acid ID-GC- 3.33 dw (Mazur et samples hydrolysis MS/SIM al., 1998) (Raw?) kernels ? Diethyl ether extract; acid ID-GC- 2.98 dw (Mazur and hydrolysis MS/SIM Adlercreutz, 1998) (Raw?) kernels Composite 70% aq. methanol extract; alkaline LC-MS/MS 0.53 fw (Milder et retail sample and enzymatic (β- al., 2005) glucuronidase/sulfatase) hydrolysis | Antioxidant (Raw?) kernels 1 retail sample ASE; enzymatic (β- LC-MS/MS 1.16 dw (Smeds et glucuronidase/sulfatase) hydrolysis al., 2007) Roasted kernels 1 retail sample 70% aq. methanol extract; alkaline GC-MS 0.253 fw (Thompson and enzymatic (β-glucuronidase) et al., 2006)

phytochemicals hydrolysis Peanut butter 1 retail sample 0.286 fw Secoisolariciresinol- (Raw?) kernels 1 retail sample Alkaline then acid extraction; LC-MS/MS 0.32 dw (Smeds et sesquilignan enzymatic (β- al., 2007) glucuronidase/sulfatase) hydrolysis (+)-Syringaresinol (Raw?) kernels 1 retail sample Alkaline then acid extraction; LC-MS/MS 0.81 dw (Smeds et enzymatic (β- al., 2007) 192 glucuronidase/sulfatase) hydrolysis A Only described as a ‘trace concentration’ and not quantified

5.1.3 Stilbenes in peanuts

Stilbenes occur in a wide range of plant families and often function as ‘phytoalexins’ that protect the plant against attack by pests or pathogens. So far, trans-3,5,4’-trihydroxystilbene (t-resveratrol) has received the most attention because of its outstanding bioactivity and medicinal properties, which are thought to partially explain the ‘French paradox’ and apparent health benefits of regular red wine consumption. Other relatively well-known stilbenes include the 3-β-glucoside of resveratrol known as piceid and the viniferins. Most identified stilbenes are produced in non-food plants or inedible tissue such as roots and bark, and the main dietary sources of stilbenes are grapes, peanuts, and their products including juice, wine and peanut butter (Cassidy et al., 2000). Resveratrol occurs in both trans- and cis-isomeric forms, but only the former is thought to occur naturally in peanut.

Resveratrol was first isolated from the seeds and hypocotyls of fungus-infected peanuts and shown to have antifungal activity in the mid-1970s (Ingham, 1976). There have subsequently been a number of studies surveying the resveratrol concentration of raw peanut kernels and processed products, generally by HPLC quantification of ethanolic, methanolic, or acetonitrilic extracts. These studies indicate that the resveratrol concentration in peanut kernels ranges from trace to low ppm levels (Table 5.5).

Other stilbenes that have been isolated from peanut kernels include trans-4-(3-methyl-but-1- enyl)-3,5,3’,4’-tetrahydroxystilbene (t-arachidin-1), trans-4-(3-methyl-but-2-enyl)-3,5,4’- trihydroxystilbene (t-arachidin-2), trans-4-(3-methyl-but-1-enyl)-3,5,4’-trihydroxystilbene (t- arachidin-3), trans-3-isopentadienyl-4,3’,5’-trihydroxystilbene (t-IPD or t-arachidin-4), and a resveratrol derivative with prenylated benzenoid and but-2-enolide moieties ascribed the common name SB-1 (Aguamah et al., 1981, Cooksey et al., 1988, Keen and Ingham, 1976, Sobolev et al., 2006). Very little quantitative data has been published on these stilbenes, although there is evidence that resveratrol concentrations in peanut kernels are on par with red wines (e.g., Buiarelli et al., 2007, Burns et al., 2002, Lamuela-Raventós et al., 1995) especially after applications of stress (Sobolev, 2008, Sobolev et al., 2007); albeit loosely so given the impropriety of comparing a food (per g basis) with a beverage (per mL basis).

There is a wealth of publications on the biological activity and clinical implications of resveratrol (e.g., reviews by Baur and Sinclair, 2006, de la Lastra and Villegas, 2005, Frémont, 2000, Holme and Pervaiz, 2007, King et al., 2006, Savouret and Quesne, 2002), but currently there is little information available regarding the in vivo bioactivity of other stilbenes. At least one study suggests that arachidin-1, arachidin-3, and IPD have potent antioxidant and anti-

| Antioxidant phytochemicals 193

inflammatory activities that are comparable to resveratrol according to a number of in vitro bioassays (Chang et al., 2006).

Surveys of resveratrol content have attempted to explore the variation among peanut cultivars and marketed retail products, but fail to take into account the environmental factors influencing resveratrol production in the plant. Resveratrol is an important phytoalexin that can be induced by wounding (Arora and Strange, 1991), abiotic stresses such as UV irradiation (Rudolf and Resurreccion, 2005), and insect damage (Sobolev et al., 2007). Resveratrol is associated with disease resistance (Sobolev et al., 2007) and found in elevated concentrations in poor quality kernels (Sobolev and Cole, 1999). These findings are corroborated by evidence that resveratrol synthase gene expression is mediated by stress hormones such as salicylic acid, jasmonic acid, and ethylene, which are known to be key signaling molecules in plant responses to abiotic and biotic stresses (Chung et al., 2003, Chung et al., 2001). The de novo synthesis and accumulation of resveratrol in plant tissue as a result of insults such as wounding, pathogen infection, and irradiation can thus be explained at a physiological level.

The link between stilbene synthesis and plant stress suggests that a large G × E interaction is likely to affect the concentrations of resveratrol and other stilbenes in peanuts. However, the extent of constitutive resveratrol gene expression relative to inducible expression is yet to be elucidated, and the respective importance of genetic and environmental factors remains unclear. Sanders et al. (2000) speculated that ‘leaky’ expression of stilbene synthase may explain the presence of resveratrol in sound uninfected kernels, but this has not been substantiated. The distinction may be important when evaluating the breeding potential of resveratrol as a genetic trait.

Many pathogenesis-related plant compounds are elicited by infection or other stresses and in some cases by non-pathogenic and pathogenic microbes. For example, resveratrol production in peanuts can be induced by infection with non-pathogenic Helminthosporum carbonum (Ingham, 1976) as well as pathogenic Aspergillus spp., including both producers and non- producers of aflatoxin and cyclopiazonic acid (Sobolev, 2008). The type and level of exposure to elicitors that may affect stilbene concentrations in commercial peanut products need to be properly characterised to ensure that the quantification of variation among genotypes or market samples is meaningful.

Stilbenes may occur in tissues other than kernels. High concentrations of resveratrol are reported in peanut testa (Sanders et al., 2000) and hulls, and the elevated resveratrol content of boiled peanuts may be explained by leaching from these tissues during processing (Sobolev

| Antioxidant phytochemicals 194

and Cole, 1999). Resveratrol has also been quantified in other peanut tissues such as leaves, roots, shells, and germinated seeds (Chang et al., 2006, Chen et al., 2002, Chung et al., 2003, Wang et al., 2005). Although resveratrol content in tissues that are not normally consumed may be of little use to mainstream food industry, some research has been undertaken with a view to the extraction and formulation of resveratrol in nutraceutical products. For example, aquaculture of peanuts for continual harvest of roots (Liu et al., 2003), hairy root culture of peanuts for production of resveratrol and pterostilbene (Medina-Bolivar et al., 2007), and use of peanut sprouts as a vegetable to utilise increased resveratrol production during germination (Wang et al., 2005).

| Antioxidant phytochemicals 195

TABLE 5.5 REVIEW OF STILBENE CONCENTRATIONS IN PEANUT KERNELS

Analyte Sample Method Concentration (μg g-1) Reference . Type N Extract Detection Mean ± SD Min - Max Basis

t-Piceid Peanut butter 14 retail samples 80% aq. ethanol extract HPLC-PDA 0.136 ± 0.052 0.067 - 0.225 fw (Ibern-Goméz et al., 2000) t-resveratrol Roasted salted 1 retail sample 50% aq. methanol extract HPLC-PDA

196

5.2 Quantification of antioxidant phytochemicals in peanut kernels

5.2.1 Objectives

The significance of genotype on the antioxidant capacity of peanut kernels was established in chapter 3. The objective of this study was to identify and quantify specific phytochemical contributors to total antioxidant capacity. HPLC was used to investigate genotype and roasting effects on the phytochemical profile, and to explore correlations among phytochemicals and their associations with measures of antioxidant capacity.

5.2.2 Materials and methods

5.2.2.1 Instruments HPLC analysis was performed on a Shimadzu Prominence series UFLC system (Shimadzu Australasia, Rydalmere, NSW, Australia) as described in chapter 2.3.2.

High resolution mass spectrometry (HRMS) was performed on an Orbitrap LTQ XL (Thermo Fisher Scientific, San Jose, CA, USA) ion trap mass spectrometer with an electrospray ionisation (ESI) source to generate ions from analytes in solution, and analysed using the Qual Browser feature in Xcaliber 2.1 (Thermo Fisher Scientific, San Jose, CA, USA).

GC-MS was performed on a Trace DSQ II mass spectrometer (Thermo Fisher Scientific) operated in positive electron ionisation (EI) mode with an ionisation energy of 70 eV, and also analysed using Xcaliber software.

A PolarSTAR Optima multi-mode microplate reader (BMG Labtech GmbH, Ortenberg, Germany), installed with 485 nm and 520 nm filters and two onboard reagent injectors, was used to measure fluorescence intensity in the ORAC assay. The assay was performed in black 96-well FLUOTRAC 200 (flat bottom, chimney well, medium-binding) microplates (655076, Greiner Bio-One GmbH, Germany).

5.2.2.2 Reagents HPLC-grade standards, solvents, and enzyme preparations were as described in chapter 2.3.2.2. ORAC stock solutions of 6-hydroxy-2,5,7,8-tetramethylchroman-2-carboxylic acid (Trolox; 238813) and fluorescein (F2456) purchased from Sigma Inc. (St. Louis, MO, USA) were made up in phosphate buffer (75 mM, pH 7.4) and frozen in aliquots until use. 2,2’-azobis(2- amidino-propane) dihydrochloride (AAPH; 11633) was made fresh at the time of analysis.

| Antioxidant phytochemicals 197

5.2.2.3 Samples Samples comprised six full-season maturity genotypes (Table 5.6) from the APBP, which is collaboratively run by PCA, GRDC, and DEEDI. The genotypes had diverse antioxidant capacity in the genotype (chapter 4.2) and G × E interaction (chapter 4.3) studies of variation in antioxidant capacity. The peanuts were grown at Taabinga Research Station in the southern Burnett region of Queensland in 2009/10, and handled as described in 4.2.2.3. Several Middleton samples that were roasted using PCA pilot equipment were also analysed: a raw control; 150 °C for 5, 20, 30, 50, and 70 min; and 160 °C for 5, 17.5, and 32.5 min.

TABLE 5.6 PEANUT GENOTYPES ANALYSED BY HPLC

Genotype Market type Source program Registration Protein (%)A

D147-p3-115 Runner DEEDI Breeding line 27 Fisher Virginia North Carolina State Uni. PBR: Plant Varieties Journal 2009, 22 (3) 27 Middleton Virginia DEEDI PBR: Plant Varieties Journal 2003, 16 (3) 30 Page Runner University of Florida PBR: Plant Varieties Journal 2009, 22 (3) 28 Sutherland Runner DEEDI PBR: Plant Varieties Journal 2002, 20 (2) 30 UF32 Virginia University of Florida Breeding line 28

A Estimated by Truspec LECO analysis 5.2.2.4 Sample extraction Native and enzyme-treated aqueous methanolic extracts of peanut kernels were prepared as described in chapter 2.3.2.4.

5.2.2.5 Solid phase extraction Extracts were concentrated and purified by SPE as described in chapter 2.3.2.5.

5.2.2.6 HPLC analysis HPLC analysis was performed as described in chapter 2.3.2.6.

5.2.2.7 Fraction collection Fractions were collected in two ways. For 26 samples, all eluate was collected at evenly spaced time intervals, i.e., 2-7, 7-12, 12-17, 17-22, and 22-27 min, freeze-dried, and the relative contributions of fractions to antioxidant capacity assessed by ORAC assay. For one sample, individual peaks were collected from sets of ten replicate injections, freeze-dried, and analysed by HRMS, LC-MS-MS, GC-MS, and ORAC assay.

5.2.2.8 HRMS analysis ESI-HRMS was performed by Dr. Leanne Stephenson (Bioanalytical Mass Spectrometry Facility, Mark Wainwright Analytical Centre, UNSW, Sydney, Australia). The instrument was calibrated daily using standard solutions and optimised for sensitivity using LC Tune software. Freeze-

| Antioxidant phytochemicals 198

dried sample extract was reconstituted in aqueous methanol (2 μL) and injected into a glass needle for introduction to the nanospray source. Ions generated were monitored over the mass range 100-2000 da in positive ion mode by a Fourier Transform MS analyser at a resolution of 60000. Data was acquired in full scan and selective ion monitoring (SIM) modes. Spectra were examined for [M+H]+ and [M+Na]+ adducts, although the latter only in full scans since SIM analysis was performed on limited mass ranges for expected [M+H]+ adducts. Correspondence between the theoretical exact mass of an analyte and the spectral centroid mass of a peak within an error margin of 2 ppm (reflecting instrumental mass accuracy) was regarded as strong evidence of compound identification.

5.2.2.9 GC-MS analysis GC-MS was performed by Dr. Martin Bucknall (Bioanalytical Mass Spectrometry Facility, Mark Wainwright Analytical Centre, UNSW, Sydney, Australia). Mixed phenolic acid standard solutions were dissolved in ethanol to a concentration of 1 mg mL-1. Aliquots (20 μL) were allowed to evaporate before addition of derivatising agent (50 μL), which comprised N,O-bis- trimethylsilyl-trifluoroacetamide (800 μL) and pyridine (200 μL). After the derivatisation treatment (90 °C, 1 h), the solution was allowed to cool before addition of acetonitrile (950 μL) and analysis by GC-MS. Small quantities of peanut extract were derivatised as above with slight alterations to the volume of derivatising agent (100 μL) and acetonitrile dilution (1950 μL).

GC-MS analysis was performed as described in Khojah (2010). Samples (1 μL injection volume) were separated on a TR-5MS SQC capillary nonpolar column (length 15m; ID 250μm; film thickness 2.5μm). The ion source and inlet temperature were set at 260 °C, while the oven temperature gradient was: 0-1.5 min, 50 °C; 1.5-6.5 min, 50-100 °C; 6.5-25.1 min, 100-230 ˚C; 25.1-26.7 min, 230-310 ˚C; and 26.7-28.7 min, hold at 310 ˚C. Detection was performed in SIM and total ion capture modes. Peaks were identified and quantified based on the mass spectra and retention time of target ions.

5.2.2.10 ORAC assay The hydrophilic ORAC assay was performed as described in 4.2.2.9 on crude extracts, purified extracts prepared by 5% and 20% methanol-based SPE, and fractions of extract collected from HPLC eluate.

5.2.2.11 Quality control Quality controls included sample replication, extract replication, and frequent analysis of spiked replicates and QC check standards to monitor data quality and allow statistical validation as described in chapter 2.3.2.7.

| Antioxidant phytochemicals 199

5.2.2.12 Moisture content Sample moisture content was determined gravimetrically as described in chapter 2.3.2.8 and used to convert the phytochemical concentrations to a dry weight basis.

5.2.2.13 Data analysis The significance of genotype and roast treatments on the phytochemical profile was analysed by standard ANOVA, and between-level differences examined by Tukey’s HSD test. Differences and correlations between native and enzyme-treated extracts were analysed using paired samples T tests. Bivariate correlations among phytochemical concentrations were evaluated using Pearson correlation coefficients, while multivariate correlations among phytochemicals were explored using PCA.

5.2.3 Raw kernel results and discussion

5.2.3.1 Explanation of HPLC data analysis The development and validation of the extraction and HPLC methods are discussed in chapter 2.3. In summary, it was preferable to use an SPE method based on 5% aqueous methanol loading solution in order to improve the repeatability of phenolic acid quantitation, whereas it was necessary to use a higher organic content (20% aqueous methanol) to achieve acceptable recovery of stilbenes and flavonoids.

Notwithstanding, actual peanut samples appeared to be less prone to phenolic acid losses during extraction and analysis than the spikes that were used to estimate recovery. According to pairwise T tests, the 20% SPE method gave statistically identical (P>0.05) results for p-coumaric, ferulic, o-coumaric, and salicylic acids to the 5% SPE method. Therefore both sets of data are presented, noting that 5% and 20% SPE extracts were prepared separately, often weeks apart, and such excellent correspondence in the quantitation provides strong evidence of the sample trends under investigation such as genotype differences, roast effects, and intervariate correlations.

Samples were extracted with and without the assistance of cellulase and protease preparations. Enzyme treatments can improve the extractability of analytes by degrading the cellular matrix and disrupting the adherence or conjugation of analytes to cell components, depending on the nature of the enzyme (Li et al., 2006, Wang et al., 2008b). For the purposes of this study, the phytochemical concentrations in enzyme-treated extracts were considered to represent total extractable analyte, whereas those in native extracts were regarded as freely extractable analyte. The difference between free and total analyte concentrations provided an estimate of the bound analyte that was covalently linked or trapped within the peanut matrix.

| Antioxidant phytochemicals 200

In the serial HPLC-PDA-RF configuration, RF detection was semi-quantitative for resveratrol and salicylic acid, but imprecise for other tested analytes. However, RF is highly selective as relatively few molecules exhibit fluorescent emission, whereas the majority of organic species will absorb UV-visible radiation especially in the 200-215 nm region (Snyder et al., 2010). RF data was therefore used as supportive evidence in peak identification alongside the quantitative PDA analysis.

This explanation of the methodological complexities should help to clarify references to ‘5% or 20% SPE’, ‘free, bound, and total analyte’, and discrepancies between PDA and RF quantitation in the results and discussion below.

5.2.3.2 HPLC identification of antioxidant phytochemicals in peanut extracts Compound identification in HPLC-PDA analysis is based on correspondence between the retention times and UV spectra of standard compounds and sample constituents (Figure 5.1). Thus, reliable compound identification depends on the availability of appropriate standards and sufficient resolution of analytes under the chromatographic test conditions to allow spectral matching. Experimentally-derived UV and fluorescence spectra are shown in Table 2.22.

2u10 06 Native extract Enzyme-treated extract daidzein

quercetin -1 ferulic acid resveratrol Standard (50 Pg mL ) caffeic acid salicylic acid -coumaric acid -coumaric -coumaric acid -coumaric o p 2u10 06

1u10 06 Intensity (320 nm)

5u10 05

0 0 5 10 15 20 25 30 Retention time (min)

FIGURE 5.1 HPLC-PDA CHROMATOGRAMS FOR A MIXED STANDARD, NATIVE PEANUT EXTRACT, AND ENZYME-TREATED PEANUT EXTRACT.

| Antioxidant phytochemicals 201

(A) 1u10 05 (B) 5u10 04 Caffeic acid standard Ferulic acid standard Vanillic acid standard Sinapic acid standard 8u10 04 4 10 04 Native extract u Native extract Enzyme-treated extract Enzyme-treated extract 6u10 04 3u10 04

04 04

Intensity 4 10 u Intensity 2u10

2u10 04 1u10 04

0 0 250 300 350 400 250 300 350 400 Wavelength (nm) Wavelength (nm)

FIGURE 5.2 UV SPECTRA AT RETENTION TIMES WHERE (A) CAFFEIC, VANILLIC (B) FERULIC, AND SINAPIC ACIDS WERE EXPECTED TO CO-ELUTE IN HPLC-PDA ANALYSIS OF PEANUT EXTRACTS. There were several areas in the peanut chromatogram where co-elution was evident, in particular at approximately 6 and 10 min (Figure 5.3). These retention times corresponded with at least several tested standard compounds that were co-eluting under our HPLC conditions (Table 5.7), as discussed in chapter 2.3.3.2. Specifically, caffeic and vanillic acids eluted at about 6 min; whereas polydatin, rutin, sinapic acid, and ferulic acid had retention times near 10 min. Co-elution constrained UV spectral matching and the prospective analytes had similar fluorescence spectra so it was not possible to achieve selective detection of compounds such as ferulic and sinapic acids, which had near-identical retention times, UV spectra, and fluorescence responses (Figure 5.2). In view of this, the peaks at 6 and 10 min were quantified on the basis of caffeic acid and ferulic acid equivalence due to their frequent mention in peanut literature, although identification was recognised to be speculative based on the HPLC data alone.

(A) (B)

FIGURE 5.3 3-D CONTOUR PLOTS AT RETENTION TIMES WHERE (A) CAFFEIC, VANILLIC, (B) FERULIC, AND SINAPIC ACIDS WERE EXPECTED TO CO-ELUTE IN HPLC-PDA ANALYSIS OF PEANUT EXTRACTS.

| Antioxidant phytochemicals 202

05 05 (A) 2u10 (B) 1u10 p-Hydroxybenzoic acid standard Syringic acid standard Native extract Native extract 2u10 05 Enzyme-treated extract 8u10 04

1u10 05 Intensity Intensity 4u10 04 5u10 04

0 0 250 300 350 400 250 300 350 400 Wavelength (nm) Wavelength (nm)

05 04 (C) 3u10 (D) 6u10 o-Coumaric acid standard Salicylic acid standard Native extract Native extract 2u10 05 Enzyme-treated extract Enzyme-treated extract 4u10 04 2u10 05

05

Intensity 1u10 Intensity 2u10 04

5u10 04

0 0 250 300 350 400 250 300 350 400 Wavelength (nm) Wavelength (nm)

06 (E) 6u10 04 (F) 2u10 Daidzein standard Quercetin standard Native extract Native extract Enzyme-treated extract Enzyme-treated extract 06 4u10 04 1u10 Intensity Intensity 05 2u10 04 5u10

0 0 250 300 350 400 250 300 350 400 Wavelength (nm) Wavelength (nm)

FIGURE 5.4 UV SPECTRA AT RETENTION TIMES CORRESPONDING TO (A) P-HYDROXYBENZOIC ACID, (B) SYRINGIC ACID, (C) O-COUMARIC ACID, (D) SALICYLIC ACID, (E) DAIDZEIN, AND (F) QUERCETIN IN HPLC- PDA ANALYSIS OF PEANUT EXTRACTS. In preliminary HPLC-PDA analyses of native peanut extracts, peaks eluting at the retention times for protocatechuic acid, myricetin, genistein, and kaempferol were not observed. Peaks with retention times that matched p-hydroxybenzoic acid, syringic acid, daidzein, and quercetin did not have correspondent UV spectra (Figure 5.4) and therefore could not be

| Antioxidant phytochemicals 203

confidently identified. The spectra of the prospective salicylic acid peak in samples resembled that of standards, but the absorbance range was broader than the pure compound.

On the other hand, the retention time and spectral matches for p-coumaric acid and resveratrol (Figure 5.5), and the qualitative identification of caffeic acid and ferulic acid as described earlier, were satisfactory. Some isomerisation of t-resveratrol (λmax 306 nm) to the cis form (λmax 280 nm) was unavoidable, and evident in the spectral development of a characteristic shoulder at ~270 nm, despite efforts to minimise UV exposure during sample preparation and extraction. The spectral shift is undesirable from an analytical aesthetic perspective, but would not have affected the molar extinction, quantitation of resveratrol concentrations, or comparison of samples.

05 04 A 2u10 B 8u10 ( ) p-Coumaric acid standard ( ) trans-Resveratrol standard Native extract Native extract Enzyme-treated extract 6u10 04 Enzyme-treated extract 1u10 05

4u10 04 Intensity Intensity 5u10 04 2u10 04

0 0 250 300 350 400 250 300 350 400 Wavelength (nm) Wavelength (nm)

FIGURE 5.5 UV SPECTRA OF PEAKS AT RETENTION TIMES CORRESPONDING TO (A) P-COUMARIC ACID AND (B) RESVERATROL IN HPLC-PDA ANALYSIS OF PEANUT EXTRACTS. RF quantitation of fluorescent analytes was numerically higher (P<0.05) than the concentrations derived by PDA analysis, especially the peaks that were qualitatively identified as caffeic and ferulic acids (Table 5.7). Yet there were also strong positive correlations (P<0.001) between the PDA and RF quantitation of analytes excluding o-coumaric acid. The correlations affirmed the identification of p-coumaric acid, salicylic acid, and resveratrol; and suggested the co-elution of peaks at 6 and 10 min that had similar fluorescence response to the hydroxycinnamic acids, but relatively low UV absorbance at the monitored wavelength, 320 nm.

In summary, p-coumaric acid and resveratrol were confidently identified and quantified; o-coumaric acid, salicylic acid, and quercetin were tentatively identified due to imperfectly correspondent UV spectra; co-eluting peaks at 6 min were likely to contain caffeic acid and similar compounds and were quantified on the basis of caffeic acid equivalence (CAE); co-

| Antioxidant phytochemicals 204

eluting peaks at 10 min were likely to contain ferulic acid and similar compounds and were quantified on the basis of ferulic acid equivalence (FAE); while other peaks were not identified with confidence in preliminary HPLC analyses of native peanut extracts.

TABLE 5.7 CORRELATIONS AND DIFFERENCES IN PDA AND RF QUANTITATION OF ANALYTES

Analyte Retention UV (nm)D Fluorescence (nm)D Paired samples T tests (RF – PDA)

time (min) λmax λex λem N Corr. Sig. Diff. Sig. Caffeic acidA 5.9 323.39 280-340 430-450 148 0.804** <0.001 135** <0.001 Vanillic acidA 6.0 260.32/291.98 260-280 360-390 p-coumaric acid 8.9 309.26 280-330 400-440 99 0.985** <0.001 19.8** <0.001 PolydatinB 9.7 317.88/306.36 280-330 380-450 RutinB 9.9 255.59/353.37 Ferulic acidC 10.1 322.57 280-340 430-480 166 0.520** <0.001 141** <0.001 Sinapic acidC 10.1 323.23 300-340 440-480 o-coumaric acid 13.1 276.14/324.60 270-290 520-560 286 0.041 0.485 1.9** <0.001 Salicylic acid 15.2 303.01 280-300 390-450 258 0.572** <0.001 1.9** <0.001 Resveratrol 16.8 305.54/317.00 270-340 380-440 339 0.911** <0.001 0.4** <0.001 * Significant (P<0.05) ** Highly significant (P<0.01) A, B, C Compounds marked with the same letter were co-eluting D Experimentally determined by spectrophotometric scans 5.2.3.3 HRMS identification of antioxidant phytochemicals in peanut extracts HPLC peaks were subjected to HRMS for additional confirmation of compound identification in the peanut extracts. The mass spectra were examined for hydrogen adducts as the primary confirmation ion. Later, full scans were also examined for sodium adducts as these are sometimes formed when glass needles are used in sample introduction. The HRMS results are presented in Table 5.8, with the expected instrumental error listed alongside the difference between the theoretical exact mass and centroid mass of the prospective analyte peak to allow convenient comparisons. Phytochemical identifications that were supported by this evidence are annotated on a representative HPLC chromatogram of an enzyme-treated peanut extract in Figure 5.6.

HRMS does not distinguish isomers, which will ionise with the same mass-to-charge ratios. However, coupling the HRMS results with HPLC retention time data supported the identification of p-coumaric acid, o-coumaric acid, salicylic acid, resveratrol, and daidzein, and helped to clarify the constituents of co-eluting peaks. Specifically, HRMS provided evidence of caffeic and vanillic acids co-eluting at 5.8 min, and strong evidence of ferulic and sinapic acids co-eluting at 9.8 min. It was not unexpected that the large ferulic/sinapic acid peak was

| Antioxidant phytochemicals 205

detectable at 10.1-10.5 min; however it was interesting to find a strong confirmation ion for coumaric acid hydrogen adduct. Preliminary experiments indicated that m-coumaric acid eluted at this retention time, but it had not been included in the calibration standards due to co-elution with ferulic acid and the expected predominance of the latter. Cinnamic acid was another compound analysed in preliminary experiments but not monitored in samples because it had a similar retention time to quercetin and, based on the peanut literature, the latter was expected to be more important. HRMS provided evidence that the HPLC peak eluting at 19.2 min contained cinnamic acid, with detection of a confirmation ion corresponding to its hydrogen adduct. On the other hand, HRMS provided contrary evidence of syringic acid and quercetin identification, as the closest confirmation ions corresponded insufficiently with the theoretical exact mass of the hydrogen adducts.

In summary, HRMS supported the HPLC identification of caffeic, p-coumaric acid, ferulic, o-coumaric acid, salicylic acid, resveratrol, and daidzein. It helped to clarify the identity of the co-eluting analytes quantitated on a CAE (i.e., caffeic and vanillic acids) and FAE (i.e., sinapic and coumaric acids) basis. HRMS suggested the presence of cinnamic acid in the peanut extracts and conversely that the peaks thought to contain syringic acid and quercetin were incorrectly identified. Further experiments to elucidate these peaks would be worthwhile.

2u10 06 daidzein resveratrol salicylic acid cinnamic acid cinnamic -coumaric acid -coumaric -coumaric acid -coumaric p o

2u10 06 ferulic acid, sinapic acid caffeic acid, vanillic acid

1u10 06 -coumaric acid, ferulic acid, sinapic acid m Intensity (320 nm)

5u10 05

0 0 5 10 15 20 25 30 Retention time (min)

FIGURE 5.6 HPLC-PDA CHROMATOGRAM OF PEANUT EXTRACT ANNOTATED WITH COMPOUNDS IDENTIFIED BY HRMS ANALYSIS.

| Antioxidant phytochemicals 206

TABLE 5.8 HRMS IDENTIFICATION OF PHYTOCHEMICALS IN PEANUT FRACTIONS

Fraction (min) Tentative identification Confirmation ions MS experimental data Mass correspondence Start End Common name Formula Mol. weight Exact mass [M+H]+ [M+Na]+ Mode [M+H]+ [M+Na]+ AccuracyA [M+H]+ [M+Na]+

5.5 6.0 Vanillic acid C8H8O4 168.1470 168.0423 169.0495 191.0315 Full 169.0491 0.0003 0.0004

Caffeic acid C9H8O4 180.1578 180.0423 181.0495 203.0315 Full 181.0490 0.0004 0.0005

6.3 6.7 Syringic acid C9H10O5 198.1731 198.0528 199.0601 SIM 199.0577 0.0004 0.0024

8.4 9.0 Coumaric acid C9H8O3 164.1584 164.0473 165.0546 187.0366 Full 165.0545 187.0360 0.0003 0.0001 0.0006

9.6 9.9 Ferulic acid C10H10O4 194.1844 194.0579 195.0652 217.0471 Full 195.0653 217.0472 0.0004 -0.0001 -0.0001

Sinapic acid C11H12O5 224.2104 224.0685 225.0757 247.0577 Full 225.0758 247.0578 0.0004 -0.0001 -0.0001

Rutin C27H30O16 610.5186 610.1534 611.1607 633.1426 Full 611.1621 633.1429 0.0012 -0.0014 -0.0003

10.1 10.5 Coumaric acid C9H8O3 164.1584 164.0473 165.0546 187.0366 Full 165.0542 0.0003 0.0004

Ferulic acid C10H10O4 194.1844 194.0579 195.0652 217.0471 Full 195.0648 0.0004 0.0004

Sinapic acid C11H12O5 224.2104 224.0685 225.0757 247.0577 Full 225.0752 247.0570 0.0004 0.0005 0.0007 12.2 12.4 Coumaric acid C H O 164.1584 164.0473 165.0546 SIM 165.0543 0.0003 0.0003 | Antioxidant 9 8 3

14.5 14.7 Salicylic acid C7H6O3 138.1210 138.0317 139.0390 SIM 139.0387 0.0003 0.0003

16.2 16.6 Resveratrol C14H12O3 228.2438 228.0786 229.0859 SIM 229.0851 0.0005 0.0008

17.5 17.9 Daidzein C15H10O4 254.2381 254.0579 255.0652 SIM 255.0645 0.0005 0.0007 phytochemicals 18.9 19.5 Cinnamic acid C9H8O2 148.1590 148.0524 149.0597 171.0416 Full 149.0597 0.0003 0.0000

19.7 19.9 Quercetin C15H10O7 302.2363 302.0427 303.0499 SIM 303.0468 0.0006 0.0031 A The Orbitrap mass spectrometer is expected to have a mass accuracy of 2 ppm 207

5.2.3.4 GC-MS identification of antioxidant phytochemicals in peanut extracts A GC-MS method developed for the analysis of phenolic acids was used to provide additional confirmation of compound identification for the hydroxybenzoic and hydroxycinnamic acids in peanut extracts. Many of the phenolic acids that have been reported in the peanut literature were analysed to establish retention times and characteristic mass spectra (Table 5.9). The correspondence between sample spectra and those of the standards were used to assign compound identities in conjunction with searches of the Wiley and NIST mass spectral libraries.

TABLE 5.9 RETENTION TIMES AND SPECTRAL FEATURES OF ANTIOXIDANT STANDARDS ANALYSED BY GC- MS

Analyte Retention time (min) Base peak Secondary peaks Caffeic acid 20.95 219 396 (87%), 381 (25%), 191 (20%) o-coumaric acid 16.50 147 308 (40%), 293 (50%) m-coumaric acid 17.52 293 203 (95%), 308 (87%), 249 (60%) p-coumaric acid 18.39 293 308 (78%), 219 (83%), 249 (65%) Ferulic acid 20.45 338 323 (74%), 308 (70%), 293 (47%) Gallic acid 18.51 281 458 (85%), 443 (32%) p-Hydroxybenzoic acid 13.45 267 223 (86%), 193 (67%), 282 (23%) Protocatechuic acid 16.60 193 370 (54%), 355 (38%), 311 (28%) Salicylic acid 9.24 195 120 (47%), 177 (33%), 210 (17%) Sinapic acid 22.39 338 368 (83%), 353 (50%), 323 (33%) Syringic acid 17.84 327 312 (80%), 342 (70%), 297 (70%) Vanillic acid 15.88 297 267 (73%), 312 (67%), 223 (53%)

GC-MS results for an enzyme-treated extract of D147-p8-6F raw kernels confirmed the presence of caffeic (Figure 5.7), vanillic (Figure 5.8), p-coumaric (Figure 5.9), ferulic (Figure 5.10), sinapic (Figure 5.11), and salicylic (Figure 5.12) acids. Neither p-hydroxybenzoic nor gallic acid was detected, but extracts were found to contain protocatechuic acid (Figure 5.13), which had not been identified in the HPLC analysis. Interestingly, the GC-MS analysis found no o-coumaric or m-coumaric acid: only p-coumaric acid. This seemed to conflict with the HRMS, which indicated that coumaric acid eluted during HPLC separation at several points that coincided with the structural isomers. The high specificity of GC-MS analysis makes it the more definitive result. The HPLC elution of p-coumaric at different retention times may point to unexpected retention properties on the HPLC column, or possible carryover in the fraction collector given its occurrence at high concentration. We suspect the former to be the case, where the column may be displaying unexpected hydrophilic interaction selectivity (HILIC) but emphasise that such effects were repeatable and that the HPLC quantitation has been proven

| Antioxidant phytochemicals 208

by our validation statistics (chapter 1.1.1.1). Further experiments are planned to improve the identification and quantitation of co-eluting phytochemicals in peanut kernels. For the purposes of this thesis, however, the GC-MS analysis provides strong evidence of caffeic, vanillic, p-coumaric, ferulic, sinapic, and salicylic acid content, and supports the quantitative analysis for these compounds.

FIGURE 5.7 CAFFEIC ACID (20.95 MIN) GC- MS LIBRARY MATCH. NB, the m/z 202 may belong to a co-eluting compound.

FIGURE 5.8 VANILLIC ACID (15.87 MIN) GC- MS LIBRARY MATCH.

| Antioxidant phytochemicals 209

FIGURE 5.9 P-COUMARIC ACID (18.39 MIN) GC-MS LIBRARY MATCH.

FIGURE 5.10 FERULIC ACID (20.45 MIN) GC- MS LIBRARY MATCH.

| Antioxidant phytochemicals 210

FIGURE 5.11 SINAPIC ACID (22.38 MIN) GC- MS LIBRARY MATCH.

FIGURE 5.12 SALICYLIC ACID (9.23 MIN) GC- MS LIBRARY MATCH.

| Antioxidant phytochemicals 211

FIGURE 5.13 PROTOCATECHUIC ACID (16.60 MIN) GC-MS LIBRARY MATCH.

5.2.3.5 Genotype effects on phytochemical concentrations Most information regarding phytochemical concentrations in peanuts refers to quantitation in ‘representative samples’ taken from retail or research sources and used to make comparative estimates of dietary intake; whereas phytochemical differences among peanut genotypes have rarely been explored. The few genotype studies are dominated by publications on resveratrol (Lee et al., 2004, Sanders et al., 2000, Tokuşoĝlu et al., 2005, Wang and Pittman, 2008), although differences in p-coumaric acid (Talcott et al., 2005b) and quercetin concentrations (Wang et al., 2007) have also been reported. Review of the literature (chapter 5.1) reveals wide disparities in the available quantitative data that make it difficult to draw conclusions other than the absence/presence and rough magnitudes of tested analytes. In view of this, the results of our study are a useful addition to the literature and understanding of peanut chemistry.

Substantial genetic control of antioxidant capacity was demonstrated in the screening of 32 diverse genotypes from the full-season maturity peanut breeding population (chapter 3.2) and subsequent study of G × E interaction over two seasons (chapter 3.3). To examine the nature of these genotypic differences, breeding lines with diverse phenotypes according to the antioxidant capacity screening were selected for phytochemical profiling. Therefore, even though only a small number of genotypes were analysed by HPLC, these were carefully selected and the results should help to characterise the chemical basis of genotypic distinctions and the breeding potential for antioxidant capacity in peanuts.

The free and total concentrations of most analytes differed (P<0.05) among raw kernels of the five tested genotypes (Table 5.10), with relative rankings revealed by Tukey’s HSD tests (Tables

| Antioxidant phytochemicals 212

5.11 to 5.22). For example, Fisher and Sutherland had high caffeic/vanillic acid and salicylic acid contents compared to the other genotypes; D147-p8-6F and Sutherland had high concentrations of p-coumaric acid, ferulic/sinapic acid, daidzein, and an unknown compound that co-eluted with o-coumaric acid (OCU); UF32 had relatively high concentrations of resveratrol and a quercetin-co-eluting unknown (QCU); while Page was low in daidzein and p-coumaric, ferulic/sinapic, OCU, and salicylic acids. Kaempferol levels were below the LOD in most samples and are not discussed further.

TABLE 5.10 SIGNIFICANCE OF GENOTYPE ON PHYTOCHEMICAL CONCENTRATIONS IN PEANUT EXTRACTS

Analyte Method ANOVA P-values SPE (%) Detector Setting (nm) Native extract Enzyme-treated extract Caffeic/vanillic acid 5 PDA 320 0.004** 0.482 p-coumaric acid 5 PDA 320 <0.001** <0.001** 20 PDA 320 <0.001** <0.001** Ferulic/sinapic acid 5 PDA 320 0.016* <0.001** 20 PDA 320 <0.001** <0.001** OCU 5 PDA 320 <0.001** <0.001** 20 PDA 320 <0.001** <0.001** Salicylic acid 5 PDA 305 <0.001** <0.001** 5 RF 260/422 <0.001** 0.253 20 PDA 305 <0.001** 0.003** 20 RF 260/422 <0.001** 0.060 Resveratrol 20 PDA 305 <0.001** <0.001** 20 RF 260/422 <0.001** 0.783 Daidzein 20 PDA 305 <0.001** <0.001** QCU 20 PDA 370 <0.001** <0.001**

* Significant (P<0.05) ** Highly significant (P<0.01)

5.2.3.5.1 Caffeic and vanillic acids Co-eluting peaks at about 6 min retention time were qualitatively estimated in CAE concentrations (Table 5.11 and Figure 5.14). Total concentrations ranged from 46.0-68.6 μg g-1 dw among genotypes, but did not differ statistically (P>0.05). Caffeic acid was the only analyte where enzyme-treated extracts had statistically lower (P<0.05) concentrations than native extracts (Table 2.15). Enzyme treatment may have altered some co-eluting constituent(s), causing a shift from this retention time, and leading to the decreased quantitation of analyte.

HRMS supported the identification of caffeic and vanillic acids (Table 5.8), but further experiments are required to confirm the identity of the co-eluting peaks and elucidate the

| Antioxidant phytochemicals 213

chromatogram changes induced by enzyme treatment. Caffeic acid has previously been measured in commercial samples of raw peanut kernels at much lower concentrations of 28 μg g-1 dw (Dabrowski and Sosulski, 1984) and 24 ± 0.14 μg g-1 fw (Mattila and Hellström, 2007); whereas vanillic acid has been reported at 22 μg g-1 dw (Kozlowskra et al., 1983) and 29 ± 2.1 μg g-1 fw (Mattila and Hellström, 2007)

-1 TABLE 5.11 CAFFEIC/VANILLIC ACID CONCENTRATIONS (μG G DW CAE) IN 5% SPE RAW PEANUT EXTRACTS

Genotype Free conc.A Total conc.A N Mean ± SEM N Mean ± SEM D147-p8-6F 7 91.2 ± 10.4 ab 9 46.0 ± 7.7 a Fisher 7 135.5 ± 15.7 a 9 52.7 ± 11.6 a Page 8 101.9 ± 9.3 ab 6 64.1 ± 14.8 a Sutherland 5 125.9 ± 16.2 a 6 65.6 ± 11.7 a UF32 6 63.4 ± 7.9 b 6 68.6 ± 6.6 a

A Lower-case letters mark significantly different (P<0.05) means according to Tukey’s HSD test 200 Free Total

dw) 150 -1 g P P 100

50 Concentration ( Concentration

Mean r SEM 0

Page UF32 Fisher

D147-p8-6F Sutherland Genotype

-1 FIGURE 5.14 CAFFEIC/VANILLIC ACID CONCENTRATIONS (μG G DW CAE) IN RAW PEANUT KERNELS.

5.2.3.5.2 p-Coumaric acid p-coumaric acid was the predominant phenolic acid in the peanut kernels by concentration, a finding in agreement with previous observations (Duncan et al., 2006). The enzyme-treated extracts contained up to fifteen times more analyte than the native extracts (Tables 5.12-5.13 and Figure 5.15). In other words, 77-93% of the p-coumaric acid in peanut kernels may be present in a bound form that is inefficiently extracted by conventional methods. The total p-coumaric acid concentrations of tested peanut genotypes ranged from 53.7-169.0 μg g-1 dw. Genotypic differences (P<0.05) in free and total analyte differed slightly; for example, D147- p8-6F and UF32 had the highest free concentrations of p-coumaric acid, whereas Sutherland | Antioxidant phytochemicals 214

and D147-p8-6F had the greatest total concentrations. Quite disparate concentrations of p-coumaric acid have been reported in raw peanut kernels ranging, for example, from 25 μg g-1 dw in Kozlowskra et al. (1983), to 770 μg g-1 fw in Mattila and Kumpulainen (2002), and 1464 μg g-1 dw in Dabrowski and Sosulski (1984). The magnitude and range of our results were similar to Talcott et al. (2005b).

-1 TABLE 5.12 P-COUMARIC ACID CONCENTRATIONS (μG G DW) IN 5% SPE RAW PEANUT EXTRACTS

Genotype Free conc.A Total conc.A N Mean ± SEM N Mean ± SEM

D147-p8-6F 7 24.2 ± 1.0 a 9 156.7 ± 23.2 a Fisher 7 7.2 ± 0.7 cd 9 96.3 ± 7.3 b Page 8 4.6 ± 0.5 d 6 53.7 ± 5.5 b Sutherland 5 11.1 ± 1.6 c 6 169.0 ± 9.9 a UF32 6 18.8 ± 2.3 b 6 80.2 ± 5.2 b A Lower-case letters mark significantly different (P<0.05) means according to Tukey’s HSD test -1 TABLE 5.13 P-COUMARIC ACID CONCENTRATIONS (μG G DW) IN 20% SPE RAW PEANUT EXTRACTS

Genotype Free conc.A Total conc.A N Mean ± SEM N Mean ± SEM D147-p8-6F 8 31.7 ± 0.9 a 8 154.0 ± 15.7 a Fisher 9 8.6 ± 0.9 c 9 77.3 ± 4.5 b Page 9 4.7 ± 0.3 d 6 43.7 ± 4.6 c Sutherland 9 11.1 ± 1.4 c 8 131.2 ± 7.8 a UF32 8 21.5 ± 0.8 b 8 66.9 ± 3.4 bc

A Lower-case letters mark significantly different (P<0.05) means according to Tukey’s HSD test

(A) (B) 200 200 Free Bound Total

dw) 150 dw) 150 -1 -1 g g P P P 100 100

50 50 Concentration ( Concentration (

Mean r SEM 0 0

Page UF32 Page UF32 Fisher Fisher

D147-p8-6F Sutherland D147-p8-6F Sutherland Genotype Genotype

-1 FIGURE 5.15 P-COUMARIC ACID CONCENTRATIONS (μG G DW) IN RAW PEANUT KERNELS EXTRACTED BY (A) 5% AND (B) 20% SPE METHOD.

| Antioxidant phytochemicals 215

5.2.3.5.3 Ferulic and sinapic acids Co-eluting peaks at 10 min that were expected to mainly comprise ferulic and sinapic acids were estimated in terms of FAE concentrations (Table 5.14-5.15 and Figure 5.16). The total analyte concentration in tested genotypes varied from 57.8-105.4 μg g-1 dw. About half (44- 53%) of the analyte was present in a freely extractable form, while the remainder (47-56%) was bound. There were genotype differences in concentrations of free analyte, but the differences in total concentrations were much more pronounced. D147-p8-6F and Sutherland genotypes contained prominently higher concentrations of ferulic acid equivalents than other genotypes.

Estimations of ferulic acid in our peanut samples were similar to the 87 μg g-1 dw (Mattila and Hellström, 2007) and lower than the 162 μg g-1 dw (Dabrowski and Sosulski, 1984) concentrations reported in previous studies of raw peanut kernels (hydrolysed extracts). On the other hand, other authors determined only ‘trace’ levels of ferulic acid that were below the LOQ (Kozlowskra et al., 1983). Sinapic acid has been reported in concentrations of 81 μg g-1 dw (Dabrowski and Sosulski, 1984), 115 μg g-1 dw (Kozlowskra et al., 1983), and 140 ± 3.5 μg g-1 fw (Mattila and Hellström, 2007).

-1 TABLE 5.14 FERULIC/SINAPIC ACID CONCENTRATIONS (μG G DW FAE) IN 5% SPE RAW PEANUT EXTRACTS

Genotype Free conc.A Total conc.A N Mean ± SEM N Mean ± SEM D147-p8-6F 7 51.2 ± 6.7 a 9 105.4 ± 2.0 a Fisher 7 40.8 ± 4.7 ab 9 87.7 ± 2.3 b Page 8 25.7 ± 3.4 b 6 57.8 ± 0.9 c Sutherland 5 47.4 ± 8.5 ab 6 100.8 ± 2.0 a UF32 6 33.8 ± 5.4 ab 6 63.6 ± 4.3 c A Lower-case letters mark significantly different (P<0.05) means according to Tukey’s HSD test -1 TABLE 5.15 FERULIC/SINAPIC ACID CONCENTRATIONS (μG G DW FAE) IN 20% SPE RAW PEANUT EXTRACTS

Genotype Free conc.A Total conc.A N Mean ± SEM N Mean ± SEM

D147-p8-6F 8 52.6 ± 3.1 a 8 111.2 ± 6.7 a Fisher 9 39.9 ± 1.5 b 9 70.8 ± 6.1 b Page 9 30.6 ± 1.1 c 6 50.3 ± 3.1 c Sutherland 9 41.0 ± 2.3 b 8 100.0 ± 3.0 a UF32 8 36.1 ± 0.7 bc 8 65.7 ± 3.6 bc A Lower-case letters mark significantly different (P<0.05) means according to Tukey’s HSD test

| Antioxidant phytochemicals 216

(A) (B) 150 150 Free Bound Total dw) dw) -1 -1 100 100 g g P P P

50 50 Concentration ( Concentration Concentration ( Concentration

Mean r SEM 0 0

Page UF32 Page UF32 Fisher Fisher

D147-p8-6F Sutherland D147-p8-6F Sutherland Genotype Genotype

-1 FIGURE 5.16 FERULIC/SINAPIC ACID CONCENTRATIONS (μG G DW FAE) IN RAW PEANUT KERNELS EXTRACTED BY (A) 5% AND (B) 20% SPE METHOD.

5.2.3.5.4 o-Coumaric acid-co-eluting unknown (OCU) There was a poor correspondence between the UV spectra of o-coumaric acid standard and the prospective peak in sample chromatograms (Figure 5.4C). While HRMS analysis seemed to support the identification of o-coumaric acid at this retention time (Table 5.8), GC-MS indicated that there was no detectable o-coumaric acid in peanut extract. However, quantification of the peak was highly repeatable and replicated in both 5% and 20% SPE extracts (Tables 5.16-5.17 and Figure 5.17), so the OCU results are nonetheless reported here in terms of o-coumaric acid equivalent (OAE) concentrations. There was no statistical difference (P>0.05) between free and total concentrations of OCU in samples excluding the UF32 genotype, which had exceptionally high concentrations of free analyte due to one of the sample replicates (i.e., from a certain field plot) containing markedly higher concentrations than other replicates, leading to high average and SEM values for the genotype. Further work to confirm the compound identification would be of interest.

-1 TABLE 5.16 OCU CONCENTRATIONS (μG G DW OAE) IN 5% SPE RAW PEANUT EXTRACTS

Genotype Free conc.A Total conc.A N Mean ± SEM N Mean ± SEM D147-p8-6F 7 9.4 ± 0.8 a 9 11 ± 1.5 a Fisher 7 4.3 ± 0.5 b 9 2.1 ± 0.6 b Page 8 3.1 ± 0.7 b 6 2.2 ± 0.8 b Sutherland 5 8.9 ± 0.5 a 6 9.2 ± 0.2 a UF32 6 10.0 ± 1.9 a 6 3.0 ± 0.3 b A Lower-case letters mark significantly different (P<0.05) means according to Tukey’s HSD test

| Antioxidant phytochemicals 217

-1 TABLE 5.17 OCU CONCENTRATIONS (μG G DW OAE) IN 20% SPE RAW PEANUT EXTRACTS

Genotype Free conc.A Total conc.A N Mean ± SEM N Mean ± SEM

D147-p8-6F 8 12.9 ± 0.4 a 8 12.7 ± 0.8 a Fisher 9 4.6 ± 0.4 bc 9 2.7 ± 0.4 c Page 9 3.1 ± 0.1 c 6 3.4 ± 0.1 c Sutherland 9 9.2 ± 0.3 ab 8 10.0 ± 0.4 b UF32 8 11.2 ± 3.2 a 8 3.2 ± 0.2 c A Lower-case letters mark significantly different (P<0.05) means according to Tukey’s HSD test

(A) (B) 15 15 Free Total dw) dw) -1 -1 10 10 g g P P P

5 5 Concentration ( Concentration Concentration ( Concentration

Mean r SEM 0 0

Page UF32 Page UF32 Fisher Fisher

D147-p8-6F Sutherland D147-p8-6F Sutherland Genotype Genotype -1 FIGURE 5.17 OCU CONCENTRATIONS (μG G DW OAE) IN RAW PEANUT KERNELS EXTRACTED BY (A) 5% AND (B) 20% SPE METHOD. NB, a ‘bound’ fraction is not presented as there was no statistical difference (P>0.05) between native and enzyme- treated extract concentrations

5.2.3.5.5 Salicylic acid Salicylic acid concentrations were approximately three to ninefold higher in the enzyme- treated extract than the native extract, suggesting that 71-89% of the compound occurs in a bound form in the peanut matrix (Tables 5.18-5.19 and Figure 5.18). Both 5% and 20% SPE methods were shown to be accurate for salicylic acid quantitation (Table 2.25), but there was some re-ranking of D147-p8-6F and Fisher genotypes when 5% SPE extracts (7.2 and 10.4 μg g-1 dw) were compared with 20% SPE extracts (9.4 and 7.6 μg g-1 dw). Given the small SEM and repeatability of the salicylic acid quantitation, the differences in the results for 5% and 20% SPE extracts were considered to reflect real variation among kernels, remembering that sample replicates were taken from different field plots and that the 5% and 20% SPE extracts were prepared separately. Salicylic acid in raw peanut kernels has been reported once previously, at an average concentration of 32 μg g-1 dw (Kozlowskra et al., 1983).

| Antioxidant phytochemicals 218

-1 TABLE 5.18 SALICYLIC ACID CONCENTRATIONS (μG G DW) IN 5% SPE RAW PEANUT EXTRACTS

Genotype Free conc.A Total conc.A N Mean ± SEM N Mean ± SEM

D147-p8-6F 7 2.07 ± 0.15 a 9 7.24 ± 1.00 b Fisher 7 2.45 ± 0.13 a 9 10.44 ± 0.41 a Page 8 1.01 ± 0.15 b 6 6.64 ± 0.49 b Sutherland 5 2.18 ± 0.26 a 6 8.99 ± 0.42 ab UF32 6 0.71 ± 0.09 b 6 6.43 ± 0.44 b A Lower-case letters mark significantly different (P<0.05) means according to Tukey’s HSD test -1 TABLE 5.19 SALICYLIC ACID CONCENTRATIONS (μG G DW) IN 20% SPE RAW PEANUT EXTRACTS

Genotype Free conc.A Total conc.A N Mean ± SEM N Mean ± SEM

D147-p8-6F 8 2.28 ± 0.04 a 8 9.42 ± 0.47 a Fisher 9 1.93 ± 0.20 a 9 7.64 ± 0.58 abc Page 9 0.96 ± 0.06 b 6 5.51 ± 1.22 c Sutherland 9 2.09 ± 0.12 a 8 8.32 ± 0.36 ab UF32 8 0.94 ± 0.11 b 8 6.51 ± 0.71 bc

A Lower-case letters mark significantly different (P<0.05) means according to Tukey’s HSD test

(A) (B) 15 15 Free Bound Total dw) dw) -1 10 -1 10 g g P P P

5 5 Concentration ( Concentration Concentration ( Concentration

Mean r SEM 0 0

Page UF32 Page UF32 Fisher Fisher

D147-p8-6F Sutherland D147-p8-6F Sutherland Genotype Genotype

-1 FIGURE 5.18 SALICYLIC ACID CONCENTRATIONS (μG G DW) IN RAW PEANUT KERNELS EXTRACTED BY (A) 5% AND (B) 20% SPE METHOD.

5.2.3.5.6 Resveratrol PDA quantitation of resveratrol attested to significant differences (P<0.05) among the tested genotypes, with similar trends observed in both native and enzyme-treated peanut extracts (Table 5.20 and Figure 5.19). Enzyme treatment increased the extractable concentration of resveratrol about threefold, from 0.97-1.81 μg g-1 dw in the native extracts to 2.75-5.59 μg g-1

| Antioxidant phytochemicals 219

dw in the enzyme-treated extracts. Thus, about 59-68% of the resveratrol in the peanut samples occurred in a bound form not efficiently extracted by conventional methods.

Measurements of resveratrol in native extracts of raw and processed peanuts have been published relatively frequently. But average concentrations reported by previous researchers have typically been much lower than in this study, and frequently less than 0.5 μg g-1 peanut (Table 5.5). Only boiled kernels were reported to have higher concentrations, from 3.42-7.09 μg g-1 fw (Sobolev and Cole, 1999). Three publications evaluated resveratrol concentrations among multiple peanut genotypes rather than in retail samples. Yet even these revealed limited concentration ranges: 0.09-0.30 μg g-1 in raw kernels of 15 genotypes (Lee et al., 2004); 0.02-1.79 μg g-1 in roasted kernels of 15 genotypes (Sanders et al., 2000); and 0.13-1.63 μg g-1 dw in raw kernels of 20 genotypes (Wang and Pittman, 2008). It may be that much of the literature has underestimated resveratrol concentration in peanuts because of failure to extract analyte bound in the cell walls and matrix.

-1 TABLE 5.20 RESVERATROL CONCENTRATIONS (μG G DW) IN 20% SPE RAW PEANUT EXTRACTS

Genotype Free conc.A Total conc.A N Mean ± SEM N Mean ± SEM D147-p8-6F 8 1.13 ± 0.07 bc 8 2.96 ± 0.24 bc Fisher 9 1.57 ± 0.05 a 9 4.36 ± 0.31 ab Page 9 1.47 ± 0.11 ab 6 3.60 ± 0.82 bc Sutherland 9 0.97 ± 0.09 c 8 2.75 ± 0.20 c UF32 8 1.81 ± 0.09 a 8 5.59 ± 0.31 a

A Lower-case letters mark significantly different (P<0.05) means according to Tukey’s HSD test 8 Free Bound Total

dw) 6 -1 g P P 4

2 Concentration (

Mean r SEM 0

Page UF32 Fisher

D147-p8-6F Sutherland Genotype

-1 FIGURE 5.19 RESVERATROL CONCENTRATIONS (μG G DW) IN RAW PEANUT KERNELS.

| Antioxidant phytochemicals 220

5.2.3.5.7 Daidzein Correspondence between the UV spectra of daidzein standard and the prospective peak in sample chromatograms was unsatisfactory (Figure 5.4E). Yet the HRMS evidence for daidzein was persuasive (Table 5.8), and the precision and accuracy of spike recovery at this retention time were good (Tables 2.25 and 2.26). Further work is necessary to confirm the compound identification, but we are reasonably confident that daidzein is present. Enzyme treatment improved extractability of the analyte by nine to thirtyfold, suggesting that 89-97% of the compound was present in a bound form (Table 5.21 and Figure 5.20). Analysis of the total analyte concentration revealed clear genotype differences, with D147-p8-6F and Sutherland containing the highest levels among the tested genotypes.

In the majority of publications, the daidzein content of peanuts has been quantified by MS and tandem MS instruments at concentrations typically less than 1 μg g-1 and sometimes less than 0.1 μg g-1 raw kernels (Table 5.3). Wang et al. (2008a) used HPLC to estimate quite disparate levels of daidzein (0.78 and 36.82 μg g-1 fw) in raw kernels harvested in two different seasons.

-1 TABLE 5.21 DAIDZEIN CONCENTRATIONS (μG G DW) IN 20% SPE RAW PEANUT EXTRACTS

Genotype Free conc.A Total conc.A N Mean ± SEM N Mean ± SEM D147-p8-6F 8 0.50 ± 0.07 b 8 15.22 ± 1.00 a Fisher 9 0.32 ± 0.04 b 9 6.35 ± 0.54 bc Page 9 0.38 ± 0.06 b 6 3.49 ± 1.61 c Sutherland 9 0.81 ± 0.09 a 8 13.23 ± 0.55 a UF32 8 0.31 ± 0.04 b 8 9.06 ± 1.02 b

A Lower-case letters mark significantly different (P<0.05) means according to Tukey’s HSD test 20 Free Bound Total

dw) 15 -1 g P 10

5 Concentration ( Concentration

Mean r SEM 0

Page UF32 Fisher

D147-p8-6F Sutherland Ge notype -1 FIGURE 5.20 DAIDZEIN CONCENTRATIONS (μG G DW) IN RAW PEANUT KERNELS.

| Antioxidant phytochemicals 221

5.2.3.5.8 Quercetin-co-eluting unknown (QCU)

-1 TABLE 5.22 QCU CONCENTRATIONS (μG G DW QE) IN 20% SPE RAW PEANUT EXTRACTS

Genotype Free conc.A Total conc.A N Mean ± SEM N Mean ± SEM

D147-p8-6F 8 4.30 ± 0.21 a 8 2.06 ± 0.26 b Fisher 9 2.72 ± 0.38 ab 9 3.08 ± 0.25 b Page 9 1.63B ± 0.33 b 6 2.75 ± 0.17 b Sutherland 9 3.87 ± 0.16 a 8 2.97 ± 0.26 b UF32 8 3.18 ± 0.72 ab 8 5.54 ± 0.27 a A Lower-case letters mark significantly different (P<0.05) means according to Tukey’s HSD test B Less than LOQ (2.00 μg g-1 dw) 8 Free Total

dw) 6 -1 g P P 4

2 Concentration ( Concentration

Mean r SEM 0

Page UF32 Fisher

D147-p8-6F Sutherland Genotype

-1 FIGURE 5.21 QCU CONCENTRATIONS (μG G DW QE) IN RAW PEANUT KERNELS. NB, a ‘bound’ fraction is not presented as there was no statistical difference (P>0.05) between native and enzyme- treated extract concentrations

The quantification of quercetin was challenging and subject to poorer accuracy and precision (Tables 2.25 and 2.26) than the other analytes in our study. Quercetin is reported to be one of the predominant flavonoids in peanut by concentration (Wang et al., 2008a). However, neither a UV spectral match (Figure 5.4F) nor a HRMS confirmation ion (Table 5.8) were observed at the retention time of the quercetin standard, which suggests that quercetin was absent or below the LOD in the peanut extract. It is possible that co-elution interfered with quercetin quantitation and the peak was therefore quantified in quercetin equivalent (QE) concentrations.

On the other hand, the spectrum of the prospective peak in sample chromatograms appears pure (Figure 5.4F), while quercetin has a distinctive λ max near 372 nm that should allow quite selective detection. Although more work is required to identify the QCU peak, it is worth | Antioxidant phytochemicals 222

noting some significant genotypic differences in native and enzyme-treated peanut extracts (Table 5.22 and Figure 5.21). In particular, UF32 had higher concentrations of the analyte in enzyme-treated extracts compared to the other tested genotypes.

5.2.3.6 Matrix-dependent bioaccessibility and bioavailability Treatment of peanut samples with cellulase and protease preparations significantly increased (P<0.05) the extractable concentrations of p-coumaric acid, ferulic/sinapic acid, resveratrol, and daidzein according to pairwise samples T tests (Table 2.15). The differences in analyte concentrations of native and enzyme-treated extracts suggested that 77-93% of the p-coumaric acid, 44-53% of the ferulic/sinapic acid, 71-89% of the salicylic acid, 59-68% of the resveratrol, and 89-97% of the daidzein in raw peanut kernels were present in bound forms. The estimation of phenolic acids (Table 5.1) and flavonoids (Tables 5.2 and 5.3) in peanut tissues has typically utilised extraction methods incorporating some form of hydrolysis, usually an acidic or alkaline chemical treatment. However, researchers measuring resveratrol in peanuts have used simple solvent extractions (Table 5.5) that our data imply are likely to result in underestimates of actual concentrations.

Most investigations of bound phytochemicals have concerned phenolic compounds, especially hydroxycinnamate derivatives. The significance of the bound phenolic fraction in terms of concentration and contribution to antioxidant capacity has been described in cowpea (Vigna unguiculata) seed tissues (Gutiérrez-Uribe et al., 2011), various millet (Elusine coracana, Panicum miliacium, P. sumatrense, Paspalum scrobiculatum, Pennisetum glaucum, and Setaria italica,) species (Chandrasekara and Shahidi, 2011), rice bran (Min et al., 2012), rice milling fractions (Tuncel and Yılmaz, 2011), and wheat bran (Moore et al., 2006). Assessments of free (aglycon and occasionally soluble conjugates) and total concentrations of phytochemicals in peanut tissues have been published, but rarely compared to estimate the bound fraction.

From an analyst’s perspective, the extent of the bound phytochemical fraction calls for scrutiny of analytical methods and their validity. In practical terms, the structure and form of phytochemicals is critical because it influences their bioaccessibility, bioavailability, and other pharmacokinetic characteristics, and ultimately impacts their metabolic fate. Our understanding of these issues in regard to functional food components such as antioxidant phytochemicals is as yet deficient, but continues to grow as the research progresses.

Bound forms of hydroxycinnamic acids include glycosylated derivatives or esters of quinic, shikimic, and tartaric acids. A prominent example is chlorogenic acid, the quinic ester of caffeic acid, which is one of the most prevalent hydroxycinnamic derivatives in fruit but is absorbed

| Antioxidant phytochemicals 223

only minimally by microflora in the colon because human tissues and digestive fluids lack esterases capable of hydrolysing it (Manach et al., 2004). Hydroxycinnamates may also be bound through ester linkages to cell wall components such as arabinoxylans and hemicelluloses. Such esterification has been most extensively described for cereals such as wheat and rice, wherein cell wall-bound hydroxycinnamates are concentrated in the bran and outer husk layers. It has been estimated that 80-95% of the polyphenolic compounds in cereal grain are diferulates that are covalently bound to cell wall polysaccharides, especially arabinoxylans, through ester linkages (Saura-Calixto, 2011).

Most esterified forms of hydroxycinnamates such as caffeic, ferulic, p-coumaric, and sinapic acids are hydrolysed by esterases in the intestinal lumen to produce free forms that can then be absorbed and undergo further metabolism. However, the nature of the bound species influences the site of hydrolysis and uptake, absorbability and rate of uptake, and even the nature of metabolites. For example, about 90% of the predominant phenolic in wheat grain, ferulic acid, is esterified to arabinoxylans and hemicelluloses in the aleurone and pericarp layers of the bran. Feruloyl mono-, di-, and oligosaccharides may be hydrolysed by mucosal esterases in the gastrointestinal tract, but most of the cell-wall bound feruloyl polymers remain intact until they reach the colon and are degraded by microbial enzymes. The bioaccessibility of dietary ferulic and p-coumaric acids is thus estimated to be highest for free forms, followed by feruloyl/coumaroyl mono- and di-saccharides, and lowest for feruloyl/coumaroyl polysaccharides. This was confirmed by studies of urinary excretion rate as a marker of uptake and absorption (Zhao and Moghadasian, 2010).

There is relatively little data on the bioavailability and metabolism of hydroxybenzoates because their comparatively restricted distribution in foods has led to less research than for hydroxycinnamate derivatives. The uptake of free and glucuronidated forms of gallic acid was shown to be highly efficient, but esterified forms have not yet been examined (Manach et al., 2005).

Flavonols such as quercetin and kaempferol are often glycosylated, most frequently with glucose or rhamnose moieties. Quercetin is less efficiently absorbed than its glucosides, but is more bioaccessible than rutin, its rhamnoglucoside (Manach et al., 2005). Studies suggest that quercetin and glucosides are absorbed in the upper gastrointestinal tract, whereas rutin remains largely intact until the distal tract, probably the colon (Erlund, 2004). Meanwhile, isoflavones such as daidzein, genistein, and glycitein occur in aglycon, 7-O-glucoside, 6”-O- acetyl-7-O-glycoside, and 6”-O-malonyl-7-O-glucoside forms (Manach et al., 2004). Most flavonoid glycosides are resistant to acid hydrolysis in the stomach, and probably only undergo | Antioxidant phytochemicals 224

limited hydrolysis by microflora in the colon. This has been demonstrated for polyphenols linked to rhamnose moieties and is likely to be the case for arabinose- and xylose-linked compounds (Manach et al., 2004).

Until recently, research has focused on free forms of phytochemicals that are bioaccessible in the upper gastrointestinal tract by direct solubilisation or enzymatic hydrolysis in digestive fluids, whereas the bound fraction has been largely ignored because of its low bioaccessibility and presumed inconsequence. However, it is hypothesised that the transport of antioxidant phytochemicals that are linked and/or trapped in the cellular matrix may in fact be one of the essential physiological functions of dietary fibre, which is comprised largely of the (natural and now also synthetic) carbohydrate polymers to which antioxidant phytochemicals are frequently bound. This fraction has been termed non-extractable polyphenols (NEPP) and includes flavanol monomers (proanthocyanidins or condensed tannins), polyesters of gallic or ellagic acid and a sugar moiety (hydrolysable tannins), as well as simple or oligomeric forms of the cell wall-bound or trapped polyphenolic compounds of interest in our study (Saura-Calixto, 2011).

It is suggested that NEPP assume many of the physiological properties of dietary fibre such as resistance to digestive degradation and partial colonic fermentability, but also confer antioxidant capacity on the dietary fibre. NEPP associated with dietary fibre are suggested to exert antioxidant/reducing activity in the upper gastrointestinal tract by surface and protein interactions. This implies a significant function in maintenance of gut health because the insoluble material remains in the gastrointestinal tract for prolonged periods. Meanwhile, non- absorbed or non-fermented antioxidants in the colon may also act as radical scavengers and counteract dietary pro-oxidants. There have been limited human and animal studies that suggest benefits for cardiovascular disease and gastrointestinal health including prevention of colorectal cancer risk, but such work is still preliminary (Saura-Calixto, 2011).

5.2.3.7 Antioxidant capacity of raw peanut extracts Genotype significantly (P<0.05) influenced antioxidant capacity in crude (P=0.002) and native (P<0.001) extracts of raw peanut kernels, but not in enzyme-treated (P=0.343) extracts (Table 5.23 and Figure 5.22). This was a surprising result given that there were significant genotypic differences in the concentrations of specific antioxidant phytochemicals that were generally more marked in the enzyme-treated extracts than the native extracts. Genotypic variation in antioxidant capacity appeared to be described by the proportions of free and bound antioxidants rather than the total concentration, but this makes little sense when we consider the trend for phytochemicals to co-vary (correlations are discussed later, in 5.2.3.9). | Antioxidant phytochemicals 225

The crude and SPE-purified extracts differed in extraction efficiency; for example, the crude extract comprised one acetonic extraction, whereas the SPE extract comprised three methanolic extractions. The crude extract would also have contained more peanut matrix components such as proteins and cellular material than the SPE extracts, which were purified, filtered through 0.45 μm syringe filters, and diluted 1/2500 in phosphate buffer for ORAC assay, and would therefore have contained few matrix components.

-1 TABLE 5.23 COMPARISON OF ORAC ANTIOXIDANT CAPACITIES (TE, μMOL G DW) IN CRUDE, NATIVE, AND ENZYME-TREATED EXTRACTS OF RAW PEANUT KERNELS

Genotype Crude extractA Native extractA Enzyme-treated extractA N Mean ± SEM N Mean ± SEM N Mean ± SEM

D147-p8-6F 6 23.2 ± 1.9 a 15 37.9 ± 2.3 a 17 49.1 ± 4.3 a Fisher 6 20.2 ± 2.3 a 16 37.8 ± 2.2 a 18 50.4 ± 3.3 a Page 6 12.7 ± 1.1 b 17 28.4 ± 1.7 b 12 51.4 ± 3.5 a Sutherland 6 17.9 ± 1.4 ab 13 33.1 ± 1.8 ab 14 41.9 ± 3.7 a UF32 9 18.0 ± 0.9 ab 14 40.9 ± 1.9 a 14 51.8 ± 3.1 a A Lower-case letters mark significantly different (P<0.05) means according to Tukey’s HSD test 60 Crude extract Native extract Enzyme-treated extract 40 dw) -1 mol g mol P P 20 TE ( TE

Mean r SEM 0

Page UF32 Fisher

D147-p8-6F Sutherland Genotype

FIGURE 5.22 COMPARISON OF ORAC ANTIOXIDANT CAPACITIES IN CRUDE, NATIVE, AND ENZYME- TREATED EXTRACTS OF RAW PEANUT KERNELS. However, it is difficult to explain the differences in antioxidant capacities of the native and enzyme-treated extracts in terms of extraction methodology or matrix effects. It may be that the additional sample preparation associated with enzyme treatment produced greater experimental variability, as indicated by higher standard deviations, which reduced the significance of sample differences. It may be helpful to perform further optimisation of the enzyme treatment to reduce experimental variability, and to monitor the matrix constituents in the extracts. Nonetheless, the main purpose of the ORAC analysis in this study was to

| Antioxidant phytochemicals 226

examine correlations with phytochemical concentrations using case-by-case statistical analysis that remains salient regardless of genotype results.

Phytopathology may provide one explanation for the genotypic variation in free and bound antioxidants. Many of the phytochemicals of interest to our study are known to function in the peanut defence response to pest and pathogen challenges (Strange and Rao, 1994) and have been termed phytoalexins (if synthesised de novo) or phytoanticipins (if constitutively expressed) in the crop protection/pathology literature. In particular, it has been demonstrated that stilbenes such as resveratrol and some of the isoflavonoids can be induced in peanuts by exposure to abiotic (Chung et al., 2003, Rudolf and Resurreccion, 2005, Sales and Resurreccion, 2009) or biotic challenges (Azpilicueta et al., 2004, Sobolev, 2008, Sobolev et al., 2009, Sobolev et al., 2010, Yang et al., 2010), and some researchers have associated higher ability to produce phytoalexins with genotype resistance to disease (Mohanty et al., 1991, Sobolev et al., 2007). Increase in cell wall-bound phenolic compounds including caffeic, ferulic, p-coumaric, and syringic acids is suggested to decrease the extensibility of the cell wall, making it more resistant to the cell wall-degrading enzymes produced by some pathogens, as well as directly inhibiting pathogen activity. Some cell wall-bound phenolic compounds, especially esterified ferulic acid, support intra- and/or inter-molecular ester-ether bridges between lignin fragments and thereby improve mechanical reinforcement of the cell walls and disease resistance (Mandal et al., 2009, Santiago et al., 2009).

Genotypic variation in the antioxidant capacity of edible plant parts has rarely been assessed (Table 4.4); genotypic variation in the concentrations of specific antioxidants even less so (chapter 5.1); and the relative contributions of free and bound fractions not at all. As such, our results are very interesting. But their implications for biological activity and plant breeding are unclear based on our current understanding of phytochemical bioaccessibility and bioavailability, especially the function, physiology, and metabolism of the bound fraction. There is plainly great scope for further research in this area.

5.2.3.8 Antioxidant capacity of peanut fractions The eluted peaks from ten repeated sample injections were combined, freeze-dried, reconstituted, and subjected to ORAC analysis of antioxidant capacity. Eluate was collected at 5 min time intervals following the solvent front (at 2 min) in order to assess all constituents of the extract, and at fixed retention times to capture individual (or co-eluting) peaks for more specific evaluation of contribution to antioxidant capacity. Rather than attempt to identify the plethora of compounds in peanut extracts, the purpose of this fractional analysis was to direct our research by distinguishing peaks or areas of the chromatogram with prominent antioxidant | Antioxidant phytochemicals 227

capacity that could then be subjected to further examination focusing on compound identification, characterisation, and bioactivity.

Standard ANOVA of ORAC values for the eluate time fractions (Table 5.24) confirmed differences (P<0.05) in respective contributions to total antioxidant capacity, as expected, and echoed the comparison of genotypes discussed in 5.2.3.7. Interestingly, there were no genotype × fraction interactions (P>0.05) in either native or enzyme-treated extracts, which suggests that genotypic differences in antioxidant capacity were described more by variation in overall concentration rather than changes in relative composition. Indeed, analysis of the percentage contribution of each fraction to total ORAC, rather than the ORAC value, completely removed statistical genotypic differences (Table 5.24).

TABLE 5.24 EXTRACT FRACTION CONTRIBUTIONS TO TOTAL ANTIOXIDANT CAPACITY

Native extract ANOVA P-values Enzyme-treated extract ANOVA P-values Contribution to ORAC Absolute Percentage Absolute Percentage

GenotypeA 0.048* 1.000 0.630 1.000 Fraction <0.001** <0.001** <0.001** <0.001** Genotype × fraction 0.252 0.481 0.383 0.133 * Significant (P<0.05) ** Highly significant (P<0.001) A Four genotypes were used: D147-p8-6F, Fisher, Page, and Sutherland

The first (2-7 min), second (7-12 min), and fourth (17-22 min) fractions contained the greatest contributors to antioxidant capacity in native extracts (Figure 5.23); whereas the second, third, and fourth made the most prominent contribution to antioxidant capacity in enzyme-treated extracts (Figure 5.24). There appeared to be significant changes in the antioxidant capacity of time fractions following enzyme treatment associated with increases or decreases in concentrations of specific antioxidants. The results point to the significance of the identified antioxidant phytochemicals as well as a number that are yet to be identified.

Some prominent changes that may be observed in Figures 5.23 and 5.24 include:

x Dramatic decline (from 28 to 9%) in the antioxidant contribution of the first fraction that correlated with decreases in concentrations of caffeic and vanillic acids and an unidentified peak; x Increase (from 31 to 43%) in the antioxidant contribution of the second fraction that correlated with elevated concentrations of p-coumaric, ferulic, and sinapic acids;

| Antioxidant phytochemicals 228

x Increase (from 12 to 26%) in the antioxidant contribution of the third fraction associated with minor increase in resveratrol and salicylic acid concentrations as well as the emergence of several small peaks; and x Decline (from 26 to 17%) in the antioxidant contribution of the fourth fraction associated with decreased concentrations of cinnamic acid and QCU.

2u10 06 50 QCU OCU daidzein resveratrol salicylic acid cinnamic acid cinnamic 40 -coumaric acid 2u10 06 p ferulic/sinapic acid caffeic/vanillic acid ) 30 nm dw) -1

(320 1u10 06 y it mol g mol P P

ens 20 t TE ( TE n I

5u10 05 10

0 0 0 5 10 15 20 25 30 Retention time (min)

FIGURE 5.23 MEAN PERCENTAGE CONTRIBUTION OF TIME-COLLECTED FRACTIONS (2-22 MIN AT 5 MIN INTERVALS) TO ORAC ANTIOXIDANT CAPACITY IN NATIVE EXTRACTS OF D147-P8-6F RAW PEANUT KERNELS.

2u10 06 50 OCU QCU daidzein resveratrol salicylic acid cinnamic acid cinnamic 40 -coumaric acid 2u10 06 p caffeic/vanillic acid ferulic acid, sinapic acid ferulic acid, sinapic acid 30 dw) -1 1u10 06 mol g mol P 20 TE ( TE Intensity (320 nm)

5u10 05 10

0 0 0 5 10 15 20 25 30 Retention time (min)

FIGURE 5.24 MEAN PERCENTAGE CONTRIBUTION OF TIME-COLLECTED FRACTIONS (2-22 MIN AT 5 MIN INTERVALS) TO ORAC ANTIOXIDANT CAPACITY IN ENZYME-TREATED EXTRACTS OF D147-P8-6F RAW PEANUT KERNELS.

| Antioxidant phytochemicals 229

ORAC assays of individual (or co-eluting) peaks collected from an enzyme-treated extract of D147-p8-6F suggested that virtually all of the extracted detectable phytochemicals contributed to peanut antioxidant capacity. Peaks that were quantified by HPLC-PDA analysis are compared in Table 5.25 according to their contribution to antioxidant capacity in the raw peanut kernel, and by their relative antioxidant capacity on an equal weight basis.

The ORAC values for peak fractions are presented overlaid in Figure 5.25 with a representative chromatogram to emphasise the HPLC-ORAC results. Of course, the standardised ORAC values for caffeic/vanillic acid, ferulic/sinapic, acid, o-coumaric acid, and quercetin must be viewed cautiously in light of their calculation from CAE, FAE, OAE, and QE concentrations. Yet again, the data highlights the presence of many unidentified phytochemicals in peanut extracts that may be worthy of further investigation in view of their contribution to peanut antioxidant capacity and prospective bioactivity.

TABLE 5.25 ORAC ANTIOXIDANT CAPACITIES OF PHYTOCHEMICALS PER G OF PEANUT AND PER G OF ANALYTE

Retention time (min) Analyte Conc. ORAC in peanut ORAC standardised by conc. Start End (μg g-1 dw) TE (μmol 100 g-1 dw) TE (mmol g-1 analyte)

5.5 6.0 Caffeic/vanillic acid 76.8 11.9 1.5 6.3 6.7 4.5 6.9 7.3 15.6 7.9 8.2 10.0 8.4 9.0 p-coumaric acid 198.4 43.9 2.2 9.6 9.9 Ferulic/sinapic acid 101.7 12.2 1.2 10.1 10.5 9.8 10.9 11.6 3.9 11.8 12.1 3.3 12.2 12.4 8.5 12.5 12.8 OCU 9.6 7.1 7.4 13.3 13.7 12.5 14.0 14.3 4.5 14.5 14.7 Salicylic acid 8.8 3.9 4.4 15.5 16.0 3.4 16.2 16.6 Resveratrol 2.6 6.0 22.9 17.5 17.9 Daidzein 13.8 1.4 1.0 18.9 19.5 8.5 19.7 19.9 10.5 20.1 20.4 QCU 1.1 6.0 56.1

| Antioxidant phytochemicals 230

2u10 06 0.5 OCU QCU daidzein resveratrol salicylic acid cinnamic acid cinnamic 0.4 -coumaric acid -coumaric 2u10 06 p ferulic/sinapic acid ferulic/sinapic acid caffeic/vanillic acid ) 0.3 nm dw) -1

(320 1u10 06 y it mol g mol P

ens 0.2 t TE ( TE n I

5u10 05 0.1

0 0.0 0 5 10 15 20 25 30 Retention time (min)

FIGURE 5.25 MEAN ORAC ANTIOXIDANT CAPACITIES OF ELUTED PEAKS COLLECTED DURING HPLC ANALYSIS OF THE ENZYME-TREATED EXTRACT OF D147-P8-6F RAW PEANUT KERNELS. It is interesting that while the phytochemicals present in the highest concentrations, namely, p-coumaric, ferulic/sinapic, and caffeic/vanillic acids, contributed the most to peanut antioxidant capacity, resveratrol and QCU had the highest antioxidant capacity relative to their low concentration in kernels (Table 5.25). These are rough approximations, however, given that our caffeic and ferulic acid estimates measured co-eluting peaks and that the collection of peak fractions cannot be presumed exact. Still, the ranking of antioxidants was unexpected and the prominence of resveratrol and QCU was not anticipated from the comparisons of pure standards published previously (Table 5.26).

The results allude to matrix effects, for example, apparent synergistic effects on resveratrol antioxidant capacity, but our study was not designed to investigate these and the data is insufficient to do more than speculate. Certainly, additive, synergistic, and antagonistic interactions on antioxidant capacity (measured by DPPH•, FC, FRAP, and ORAC assays) have been reported among extensive combinations of ‘high antioxidant’ foods (Wang et al., 2011), and it is reasonable to presume that varied interactions occur among combinations of phytochemicals as well.

The results have outstanding practical implications as they suggest a relatively small increase in the resveratrol or QCU content of peanut kernels, through plant breeding or some method of elicitation, would generate a disproportionate escalation in antioxidant capacity, even though the predominant phenolic acids may continue to make the major contribution to kernel

| Antioxidant phytochemicals 231

antioxidant capacity by virtue of their high concentrations. It would be valuable to confirm our results through further experiments and to characterise the nature of the antioxidant interaction, if this does indeed underlie the exceptionally high relative antioxidant capacities of resveratrol and QCU.

It goes without saying that it would be worthwhile to conduct further experiments to elucidate the identity of QCU and to confirm the identity of the peak tentatively assigned as cinnamic acid. In addition, the ORAC analysis of peak fractions revealed several other major but unidentified contributors to kernel antioxidant capacity, in particular, at 7.0, 8.0, 13.4, and 19.8 min. The peaks had UV spectra that seemed characteristic of the hydroxycinnamate derivatives, especially the 8.0 and 13.4 min peaks, but with slight differences in the λmax (Figure 5.26). The identity of these peaks will hopefully be characterised and confirmed in future experiments employing, for example, MS configurations and nuclear magnetic resonance (NMR) for structural elucidation.

| Antioxidant phytochemicals 232

TABLE 5.26 PUBLISHED COMPARISONS OF THE ANTIOXIDANT CAPACITY OF PHYTOCHEMICAL STANDARDS

Assay Unit Caffeic acid p-coumaric acid Ferulic acid Salicylic acid Resveratrol Quercetin Reference ORAC TE (μM) 4.37 7.28 (Ou et al., 2001) ORAC TE (mM) 15.28 10.16 8.48 8.62 13.35 (Villaño et al., 2005) ABTS TE (mM) 1.01 0 0.23 0.40 1.14 (Villaño et al., 2005) DPPH TE (mM) 1.11 0 0.60 0.49 2.55 (Villaño et al., 2005) TEAC TE (mM) 1.26 2.22 1.90 0.04 4.72 (Rice-Evans et al., 1996) FC GAE (M) 0.958 1.08 0.262 1.23 2.08 (Everette et al., 2010) AUC persulfate TE (mM) 0.99 1.56 1.75 2.88 (Re et al., 1999)

Modified FRAP EC1 (μM) 139 390 729 92 (Pulido et al., 2000)

LDL peroxidation (AAPH) Tlag (min) 34.5 22.2 25.2 (Cheng et al., 2007) 2+ LDL peroxidation (Cu ) Tlag (min) 139 56 53 (Cheng et al., 2007)

(A) O max 314 nm (B) O max 325 nm (C) O max 315 nm (D) O max 319 nm | Antioxidant Intensity phytochemicals

250 300 350 400 250 300 350 400 250 300 350 400 250 300 350 400 Wavelength (nm) Wavelength (nm) Wavelength (nm) Wavelength (nm)

FIGURE 5.26 UV SPECTRA OF UNIDENTIFIED PEAKS AT (A) 7.0 MIN, (B) 8.0 MIN, (C) 13.4 MIN, AND (D) 19.8 MIN THAT WERE IMPORTANT 233 CONTRIBUTORS TO ORAC ANTIOXIDANT CAPACITY IN ENZYME-TREATED EXTRACTS OF D147-P8-6F RAW PEANUT KERNELS.

5.2.3.9 Correlations between antioxidant phytochemicals and antioxidant capacity Attribution of ORAC values to fractions of HPLC eluate was a direct method of evaluating the relative antioxidant capacities of peanut phytochemicals. Correlations analysis of antioxidant concentrations and total ORAC was another approach to explore the relationship between phytochemical composition and kernel antioxidant capacity, though it is limited by our capacity to identify and quantify analytes.

Bivariate (pairwise) correlations impute direct relationships between two variables and are simple to interpret; however multivariate correlations are more pertinent for biological subjects, where the relationships between variables are subject to complex labile interactions. Therefore, the multivariate relationships described by PCA are emphasised, although the bivariate correlations between antioxidant capacity and the free and total concentrations of antioxidant phytochemicals are also reported.

TABLE 5.27 BIVARIATE CORRELATIONS BETWEEN PHYTOCHEMICAL CONCENTRATIONS AND ORAC ANTIOXIDANT CAPACITY IN RAW PEANUT KERNELS

Analyte Method Native extract Enzyme-treated extract (SPE; PDA) N Pearson corr. Sig. (2-tailed) N Pearson corr. Sig. (2-tailed)

Caffeic/vanillic acid 5%; 320 nm 30 -0.186 0.325 32 -0.007 0.971 p-coumaric acid 5%; 320 nm 33 0.310 0.079 35 -0.538** 0.001 Ferulic/sinapic acid 5%; 320 nm 33 0.082 0.651 35 -0.103 0.557 OCU 5%; 320 nm 33 -0.109 0.547 36 -0.328 0.051 Salicylic acid 20%; 305 nm 41 -0.014 0.930 38 -0.263 0.111 Resveratrol 20%; 305 nm 42 0.083 0.601 38 0.248 0.133 Daidzein 20%; 305 nm 42 -0.025 0.878 37 -0.097 0.568 QCU 20%; 370 nm 42 0.298 0.055 38 0.015 0.928

* Significant (P<0.05) ** Highly significant (P<0.01)

ORAC antioxidant capacity was not directly correlated with any one of the tested phytochemicals except for a counterintuitive negative correlation (r=-0.538, P=0.001) with p-coumaric acid, the predominant antioxidant in peanut, in enzyme-treated extract (Table 5.27). This may reflect the inverse relationships (Table 5.29) of p-coumaric acid with resveratrol (r=-0.427, P=0.009) and QCU (r=-0.485, P=0.003), which had the greatest relative antioxidant capacity of all the detected peaks in the peanut extract (Table 5.25). The main trends described by bivariate analysis were that total p-coumaric acid, ferulic/sinapic acid, OCU, salicylic acid, and daidzein concentrations were all positively correlated with each other; caffeic/vanillic acid was negatively correlated with all other analytes and significantly (P<0.05)

| Antioxidant phytochemicals 234

so with salicylic acid, resveratrol, and daidzein; while resveratrol was negatively correlated with caffeic/vanillic acid, p-coumaric acid, and OCU (Table 5.29).

Interestingly, the correlations in free analyte concentrations (Table 5.28) were distinct from those of the total concentrations. In contrast to the enzyme-treated extracts, ORAC value correlated positively with p-coumaric acid (r=0.530, P<0.001) and ferulic/sinapic acid (r=0.373, P=0.015) concentrations. Again, there were positive correlations among a number of the phytochemicals including p-coumaric acid, ferulic/sinapic acid, OCU, and QCU. The differences between correlations in native and enzyme-treated extracts reflect the varying composition of the bound fraction in peanut kernels.

TABLE 5.28 BIVARIATE (PEARSON) CORRELATIONS AMONG PHYTOCHEMICAL CONCENTRATIONS AND ORAC ANTIOXIDANT CAPACITY IN 20% SPE NATIVE EXTRACTS OF RAW PEANUT KERNELS

Statistic CAF pCO FER OCU SAL RES DAI QCU ORAC CAF Correlation 1 -0.288 -0.307 -0.058 0.282 -0.226 -0.026 0.085 0.086 (N=41) Sig. (2-tailed) 0.072 0.051 0.750 0.077 0.155 0.873 0.596 0.594 pCO Correlation -0.288 1 0.718** 0.748** 0.166 -0.110 0.041 0.513** 0.530** (N=41) Sig. (2-tailed) 0.072 <0.001 <0.001 0.305 0.495 0.799 0.001 <0.001 FER Correlation -0.307 0.718** 1 0.457** 0.514** -0.216 0.353* 0.506** 0.373* (N=42) Sig. (2-tailed) 0.051 <0.001 0.007 0.001 0.170 0.022 0.001 0.015 OCU Correlation -0.058 0.748** 0.457** 1 0.226 -0.306 0.172 0.751** 0.330 (N=33) Sig. (2-tailed) 0.750 <0.001 0.007 0.213 0.084 0.337 <0.001 0.061 SAL Correlation 0.282 0.166 0.514** 0.226 1 -0.534** 0.270 0.354* -0.014 (N=41) Sig. (2-tailed) 0.077 0.305 0.001 0.213 <0.001 0.087 0.023 0.930 RES Correlation -0.226 -0.110 -0.216 -0.306 -0.534** 1 -0.343* -0.297 0.083 (N=42) Sig. (2-tailed) 0.155 0.495 0.170 0.084 <0.001 0.026 0.056 0.601 DAI Correlation -0.026 0.041 0.353* 0.172 0.270 -0.343* 1 0.295 -0.025 (N=42) Sig. (2-tailed) 0.873 0.799 0.022 0.337 0.087 0.026 0.058 0.878 QCU Correlation 0.085 0.513** 0.506** 0.751** 0.354* -0.297 0.295 1 0.298 (N=42) Sig. (2-tailed) 0.596 0.001 0.001 <0.001 0.023 0.056 0.058 0.055 ORAC Correlation 0.086 0.530** 0.373* 0.330 -0.014 0.083 -0.025 0.298 1 (N=42) Sig. (2-tailed) 0.594 <0.001 0.015 0.061 0.930 0.601 0.878 0.055 CAF = caffeic/vanillic acid (320 nm); pCO = p-coumaric acid (320 nm); FER = ferulic/sinapic acid (320 nm); OCU = o-coumaric acid-co-eluting unknown (320 nm); SAL = salicylic acid (305 nm); RES = resveratrol (305 nm); QCU = quercetin-co-eluting unknown (370 nm) * Significant (P<0.05) ** Highly significant (P<0.01)

| Antioxidant phytochemicals 235

TABLE 5.29 BIVARIATE (PEARSON) CORRELATIONS AMONG PHYTOCHEMICAL CONCENTRATIONS AND ORAC ANTIOXIDANT CAPACITY IN 20% SPE ENZYME-TREATED EXTRACTS OF RAW PEANUT KERNELS

Statistic CAF pCO FER OCU SAL RES DAI QCU ORAC CAF Correlation 1 -0.190 -0.274 -0.124 -0.662** -0.391* -0.614** -0.333 0.349* (N=35) Sig. (2-tailed) 0.274 0.111 0.520 <0.001 0.020 <0.001 0.051 0.043 pCO Correlation -0.190 1 0.890** 0.838** 0.530** -0.427** 0.732** -0.485** -0.085 (N=36) Sig. (2-tailed) 0.274 <0.001 <0.001 0.001 0.009 <0.001 0.003 0.627 FER Correlation -0.274 0.890** 1 0.852** 0.668** -0.263 0.787** -0.310 0.036 (N=36) Sig. (2-tailed) 0.111 <0.001 <0.001 <0.001 0.121 <0.001 0.066 0.839 OCU Correlation -0.124 0.838** 0.852** 1 0.480** -0.419* 0.792** -0.425* 0.237 (N=33) Sig. (2-tailed) 0.520 <0.001 <0.001 0.005 0.015 <0.001 0.014 0.192 SAL Correlation -0.662** 0.530** 0.668** 0.480** 1 0.103 0.698** -0.084 -0.263 (N=39) Sig. (2-tailed) <0.001 0.001 <0.001 0.005 0.532 <0.001 0.612 0.111 RES Correlation -0.391* -0.427** -0.263 -0.419* 0.103 1 -0.184 0.715** 0.248 (N=39) Sig. (2-tailed) 0.020 0.009 0.121 0.015 0.532 0.268 <0.001 0.133 DAI Correlation -0.614** 0.732** 0.787** 0.792** 0.698** -0.184 1 -0.129 -0.097 (N=38) Sig. (2-tailed) <0.001 <0.001 <0.001 <0.001 <0.001 0.268 0.440 0.568 QCU Correlation -0.333 -0.485** -0.310 -0.425* -0.084 0.715** -0.129 1 0.015 (N=39) Sig. (2-tailed) 0.051 0.003 0.066 0.014 0.612 <0.001 0.440 0.928 ORAC Correlation 0.349* -0.085 0.036 0.237 -0.263 0.248 -0.097 0.015 1 (N=38) Sig. (2-tailed) 0.043 0.627 0.839 0.192 0.111 0.133 0.568 0.928

CAF = caffeic/vanillic acid (320 nm); pCO = p-coumaric acid (320 nm); FER = ferulic/sinapic acid (320 nm); OCU = o-coumaric acid-co-eluting unknown (320 nm); SAL = salicylic acid (305 nm); RES = resveratrol (305 nm); QCU = quercetin-co-eluting unknown (370 nm) * Significant (P<0.05) ** Highly significant (P<0.01)

PCA is a statistical approach to simplify and characterise the variance in complex multivariate data. High-dimensional data are reconstructed into PC, wherein correlated variables are condensed to generate fewer dimensions. Each PC is calculated from all of the tested variables by loading factors that weight their respective importance to maximise variance in the first PC; meanwhile sample data are denoted by scores for each PC. Therefore, samples are described by the reconstructed PC data in terms of the loadings, while interrelationships among samples and variables can be conveniently visualised by overlaying scores and loadings plots.

About 85% of variance in the free (native extract) phytochemical data was described in four PC (Table 5.30). PC-1 mainly comprised p-coumaric acid, ferulic/sinapic acid, OCU, and QCU; whereas PC-2 mostly described caffeic/vanillic acid, salicylic acid, and resveratrol. Total phytochemical concentrations were rather better described by PCA, with 82% of the total variance explained in just two PC (Table 5.31). PC-1 described p-coumaric acid, ferulic/sinapic

| Antioxidant phytochemicals 236

acid, OCU, and daidzein; whereas PC-2 predominantly described caffeic/vanillic acid, salicylic acid, resveratrol, and QCU concentrations. This simplification of the data accorded well with the bivariate correlations, which found strong positive correlations among p-coumaric acid, ferulic/sinapic acid, OCU, and daidzein as well as salicylic acid (Table 5.29).

Correlated variables can be observed as clusters in the loadings plots, whereas samples that are described similarly by PCA can be observed as clusters in the scores plots. The peanut genotypes evaluated in this study were well described by PC-1 and PC-2 in PCA of analyte concentrations in both native (Figure 5.27) and enzyme-treated (Figure 5.28) extracts.

In the PCA of native extract, for example, D147-p8-6F and Sutherland samples scored positively on PC-1 (Figure 5.27E), which was associated with relatively high concentrations of all tested analytes apart from caffeic acid/vanillic and resveratrol (Figure 5.27A). However, the two sister-lines could be distinguished along the PC-2 axis, indicating that Sutherland had relatively high caffeic/vanillic acid compared to D147-p8-6F, which had relatively high resveratrol content. Fisher and UF32 could be similarly distinguished on the PC-2 axis. Meanwhile, Page samples clustered in the centre left of the scores plot, indicating relatively low concentrations of all tested analytes.

In the PCA of enzyme-treated extract, D147-p8-6F and Sutherland samples clustered very closely with negative PC-1 scores (Figure 5.28E), associated with high concentrations of p-coumaric, ferulic/sinapic, and salicylic acids, as well as daidzein and OCU (Figure 5.28A); whereas UF32 tended to have higher PC-2 scores than the other samples, indicating relatively high resveratrol and QCU content in this genotype.

Antioxidant capacity was poorly described by the PCA of native extracts, with no obvious clustering evident in the first four PC (Figures 5.27C and D). In PCA of enzyme-treated extracts, ORAC value was poorly described by PC-1 and PC-2 (Figure 5.28C), but there was better separation of the highest and lowest quintiles along PC-3 and PC-4 (Figure 5.28D). Most of the highest quintile (>58.6 TE) tended to score in the upper left quadrant of the PC-3 vs. PC-4 scores plot, which corresponded with the combination of relatively high caffeic/vanillic acid, p-coumaric acid, ferulic/sinapic acid, and resveratrol concentrations.

| Antioxidant phytochemicals 237

TABLE 5.30 X-VARIABLE LOADINGS, EXPLAINED TOTAL VARIANCE, AND X-VARIABLE VARIANCE IN PCA OF TOTAL PHYTOCHEMICAL CONCENTRATIONS IN NATIVE EXTRACTS OF RAW PEANUT KERNELS

PC Explained X-variable variance (%) Total explained variance (%) X-variable loading CAF pCO FER OCU SAL RES DAI QCU Calibration Validation CAF pCO FER OCU SAL RES DAI QCU 1 1 52 67 64 39 28 23 63 42 27 -0.039 0.391 0.447 0.440 0.338 -0.286 0.259 0.434

2 64 81 75 67 63 59 34 63 63 43 0.627 -0.404 -0.204 -0.151 0.368 -0.423 0.254 0.005

3 83 86 79 80 63 59 83 71 75 51 -0.448 -0.208 0.222 -0.361 -0.025 -0.031 0.707 -0.273 4 84 87 88 85 89 61 96 86 85 61 0.136 -0.127 -0.349 0.293 -0.590 0.162 0.420 0.455 CAF = caffeic/vanillic acid (320 nm); pCO = p-coumaric acid (320 nm); FER = ferulic/sinapic acid (320 nm); OCU = o-coumaric acid-co-eluting unknown (320 nm); SAL = salicylic acid (305 nm); RES = resveratrol (305 nm); QCU = quercetin-co-eluting unknown (370 nm)

TABLE 5.31 X-VARIABLE LOADINGS, EXPLAINED TOTAL VARIANCE, AND X-VARIABLE VARIANCE IN PCA OF TOTAL PHYTOCHEMICAL CONCENTRATIONS IN ENZYME- TREATED EXTRACTS OF RAW PEANUT KERNELS

PC Explained X-variable variance (%) Total explained variance (%) X-variable loading | Antioxidant CAF pCO FER OCU SAL RES DAI QCU Calibration Validation CAF pCO FER OCU SAL RES DAI QCU

1 16 84 86 81 52 15 79 19 54 41 0.195 -0.448 -0.454 -0.424 -0.346 0.183 -0.423 0.206 2 80 87 86 85 75 82 88 76 82 73 -0.519 -0.110 0.018 -0.128 0.322 0.551 0.193 0.508

phytochemicals 3 86 89 90 91 91 82 89 93 89 78 -0.329 -0.200 -0.282 -0.328 0.545 -0.099 -0.173 -0.573 4 93 92 96 91 92 97 94 95 94 84 0.417 0.268 0.383 -0.105 0.200 0.603 -0.358 -0.257

CAF = caffeic/vanillic acid (320 nm); pCO = p-coumaric acid (320 nm); FER = ferulic/sinapic acid (320 nm); OCU = o-coumaric acid-co-eluting unknown (320 nm); SAL = salicylic acid (305 nm); RES = resveratrol (305 nm); QCU = quercetin-co-eluting unknown (370 nm) 238

(A) PC-2 (B) PC-4 0.8 0.8 CAF QCU SAL DAI DAI oCO CAF RES QCU PC-1 PC-3 -0.8 FER 0.8 -0.8 pCO 0.8 oCO FER RES pCO SAL -0.8 -0.8

-1 (C) PC-2 (D) PC-4 ORAC, TE (Pmol g dw) 3 3 Quintile 1: <31.0 TE Quintile 2-4 Quintile 5: >41.6 TE

PC-1 PC-3 -4 4 -3 3

-3 -3

(E) PC-2 (F) PC-4 D147-p8-6F 3 3 Fisher Page Sutherland UF32

PC-1 PC-3 -4 4 -3 3

-3 -3

FIGURE 5.27 (A AND B) X-VARIABLE LOADINGS PLOTS; (C AND D) SCORES PLOTS MARKED BY ANTIOXIDANT CAPACITY; AND (E AND F) SCORES PLOTS MARKED BY GENOTYPE IN PCA OF NATIVE EXTRACTS OF RAW PEANUT KERNELS.

| Antioxidant phytochemicals 239

(A) PC-2 (B) PC-4 0.8 0.8

RES RES QCU FER CAF SAL pCO DAI SAL FER PC-1 PC-3 -0.8 pCO oCO 0.8 -0.8 oCO 0.8 QCU DAI CAF

-0.8 -0.8

-1 (C) PC-2 (D) PC-4 ORAC, TE (Pmol g dw) 4 2 Quintile 1: <37.2 TE Quintile 2-4 Quintile 5: >58.6 TE

PC-1 PC-3 -6 6 -2 2

-4 -2

(E) PC-2 (F) PC-4 D147-p8-6F 6 2 Fisher Page Sutherland UF32

PC-1 PC-3 -6 6 -2 2

-6 -2

FIGURE 5.28 (A AND B) X-VARIABLE LOADINGS PLOTS; (C AND D) SCORES PLOTS MARKED BY ANTIOXIDANT CAPACITY; AND (E AND F) SCORES PLOTS MARKED BY GENOTYPE IN PCA OF ENZYME- TREATED EXTRACTS OF RAW PEANUT KERNELS.

| Antioxidant phytochemicals 240

The genotype results generally concurred with the ANOVA and Tukey HSD tests (chapter 5.2.3.5). The PCA description of antioxidant capacity in enzyme-treated extracts held additional interest because it linked high antioxidant capacity in kernels with a combination of phytochemicals that were not directly correlated in bivariate analysis (e.g., resveratrol was negatively correlated with caffeic/vanillic and p-coumaric acid) and that were not delineated by genotype (Figure 5.28F), remembering that there were no significant genotypic differences in antioxidant capacity of enzyme-treated extracts (Table 5.23 and Figure 5.22). At this stage of the research, PCA was limited by the incomplete identification and quantification of antioxidant phytochemicals in peanuts. Most obviously, there are a number of important contributors to antioxidant capacity in peanut kernels that are yet to be properly identified and characterised (Table 5.25 and Figure 5.25). Improving the phytochemical profiling will enable better modelling of the antioxidant expression and should be a priority in future research.

5.2.4 Roasted kernel results and discussion

5.2.4.1 Roast effects on antioxidant capacity of peanut kernels Roasting is the primary method of processing peanut kernels for domestic consumption. As such, evaluating the changes in functional food qualities such as antioxidant capacity and phytochemical concentrations is critical in establishing the value of the trait, and therefore pertinent to our research. Several surveys of antioxidants in roasted peanut kernels have been published (Table 5.32), including the measurement of selected phenolic acids (Talcott et al., 2005b), phytoestrogens (Kuhnle et al., 2008, Liggins et al., 2000, Thompson et al., 2006), and resveratrol (Burns et al., 2002, Lee et al., 2004, Sanders et al., 2000, Sobolev and Cole, 1999). Such studies establish that the major identified antioxidant phytochemicals in peanuts are relatively thermo-stable, at least sufficiently for measurable quantities to endure commercial roasting at high temperatures of 160-180 °C.

On the other hand, studies designed to characterise roasting-induced changes in the peanut antioxidant profile have yielded somewhat contradictory results. Several researchers reported that antioxidant capacity increased with duration of roasting as measured by ORAC (Davis et al., 2010, Talcott et al., 2005b), DPPH• scavenging, inhibition of linoleic acid peroxidation, thiobarbituric acid (Win et al., 2011), and ferric-thiocyanate (Hwang et al., 2001) assays. Whereas Davis et al. (2010) and Win et al. (2011) suggested that roasting was correlated with elevated levels of phenolic compounds according to the FC total phenols assay, Chukwumah et al. (2007a) and Talcott et al. (2005b) found no statistically significant change in total phenols content following roasting. | Antioxidant phytochemicals 241

TABLE 5.32 LITERATURE REGARDING THE EFFECTS OF ROASTING ON ANTIOXIDANT CAPACITY AND CONCENTRATIONS IN PEANUT KERNELS

Roast treatment Antioxidant assessment Method Effects Reference Dry-roasting 166 °C (0 Antioxidant capacity ORAC assay H-ORAC was 1.7 to 2.3-fold higher in light- and dark-roasted kernels (59.1-79.9 μmol g-1 fw (Davis et al., to 77 min) TE) cf. raw (30.4 μmol g-1 fw TE) kernels. Roast-blanched, light-, and dark-roasted kernels 2010) had higher L-ORAC than raw kernels. There was a linear correlation between H- and L- ORAC and decreasing L-value (i.e., darkness) in kernels roasted at 166 °C up to 77 min. Total phenols FC assay There was a linear correlation between total phenols and H-ORAC in kernels roasted at 166 °C up to 77 min. α-, β-, γ-, and δ- HPLC-UV Degradation of peanut oil α- and γ-tocopherols under accelerated oxidative conditions Tocopherols was promoted by light-medium roasting cf. raw kernels, but significantly less so in dark- roasted kernels. Dry- roasting 160 °C (0- Total phenols FC assay Total phenols increased with duration of roasting. (Win et al., 50 min) 2011) Antioxidant capacity DPPH• assay Antioxidant capacity increased with roasting. Inhibition of linoleic Antioxidant capacity increased with roasting. acid peroxidation Thiobarbituric acid Antioxidant capacity increased with roasting.

| Antioxidant assay p-Hydroxybenzoic, HPLC-UV There were statistical increases in quercetin and kaempferol concentrations after 20 and chlorogenic, and 10 min roasting, and apparent declines in phytochemical concentrations after 40-50 min. p-coumaric acids, quercetin, kaempferol, and

phytochemicals t-resveratrol Dry-roasting 180 °C (22 Total flavonoids Spectrophotometric No significant change except a slight increase (from 0.05 to 0.06 mg g-1 catechin eq.) when (Chukwumah min); oil-roasting 140 °C assay peanuts were boiled with testa. et al., 2007a)

(7 min); and boiling (4 h -1 in 25 g L-1 salt water) Total phenols FC assay No significant change except a slight increase (from 28.71 to 36.42 mg g GAE) when peanuts were boiled with testa. Daidzein, daidzin, genistein, HPLC-PDA-APCI-MS Biochanin A, genistein, daidzein, and resveratrol concentrations increased in boiled genistin, biochanin A, peanuts. Roasting had no effects except on daidzein (no daidzein was measured in the raw 242 sissotrin, and t-resveratrol controls; roasted kernels had about 0.3-0.4 μg g-1 dw).

TABLE 5.32 CONTINUED

Roast treatment Antioxidant assessment Method Effects Reference Dry-roasting 175 °C (10 Antioxidant capacity ORAC assay Antioxidant capacity increased (from 18.4-34.4 to 23.9-41.2 μM TE) with roasting. (Talcott et min) al., 2005b) Total phenols FC assay No significant change. Total phenols, tryptophan, HPLC-PDA Soluble phenolic and protein concentrations increased with roasting. and soluble proteins p-coumaric acid HPLC-PDA p-coumaric acid concentration increased (from 25 to 69 μg g-1) with roasting. Other unidentified peaks, assumed to be hydroxybenzoates, also increased. Dry-roasting 180 °C (0 Antioxidant capacity Ferric- Antioxidant capacity increased with increasing roasting time. (Hwang et to 60 min) thiocyanate/linoleic al., 2001) acid assay Commercial dry- α-, β-, γ-, and δ- HPLC-RF Decreased vitamin E stability was associated with decreased oxidative stability in roasted (Chun et al., roasting (135 °C then Tocopherols peanuts during storage. Tocopherol concentrations were the same after roasting. But 2005) 190 °C, approx. 10 min following 12 weeks storage, there was 50% loss of tocopherols in vacuum-packed kernels each) and 90% loss in atmospheric storage.

| Antioxidant phytochemicals 243

In our study, peanut kernels were subjected to oven-roasting at 150 and 160 °C for various durations to generate degrees of roast to a maximum level that was slightly darker that a commercial dark roast. Results of ORAC analysis were in agreement with the reports by Chukwumah et al. (2007a) and Talcott et al. (2005b), as no significant differences (Table 5.33) were observed among 150 °C or 160 °C roast treatments in either of the crude, 5% SPE, or 20% SPE extracts. Only one sample set, the enzyme-treated 5% SPE extracts roasted at 150 °C, displayed statistically significant differences, but these presented as sample variability rather than as a trend associated with roasting.

TABLE 5.33 EFFECT OF ROAST DURATION ON ORAC ANTIOXIDANT CAPACITY

ANOVA P-values Temperature (°C) Crude extract Native extract Enzyme-treated extract 5% SPE 20% SPE 5% SPE 20% SPE 150 0.693 0.615 0.239 0.001** 0.728 160 0.064 0.275 0.395 0.839 0.583 * Significant (P<0.05) ** Highly significant (P<0.01)

The results for the 150 °C roast treatments are presented in Tables 5.34-5.35 and Figure 5.29; while the 160 °C roast treatments are presented in Tables 5.36-5.37 and Figure 5.30.

| Antioxidant phytochemicals 244

-1 TABLE 5.34 ORAC ANTIOXIDANT CAPACITY (TE, μMOL G DW) OF 5% SPE EXTRACTS OF PEANUT KERNELS ROASTED AT 150 °C

Time NativeA Enzyme-treatedA (min) N Mean ± SEM N Mean ± SEM

0 6 37.5 ± 3.9 a 3 50.6 ± 4.2 c 5 3 30.7 ± 2.8 a 3 48.0 ± 3.5 c 20 3 38.0 ± 2.6 a 3 69.6 ± 4.3 ab 30 3 35.2 ± 2.5 a 3 47.2 ± 3.2 c 50 3 40.5 ± 4.0 a 3 71.9 2.8 a 70 3 34.8 ± 3.0 a 3 52.2 ± 3.8 bc

A Lower-case letters mark significantly different (P<0.05) means according to Tukey’s HSD test -1 TABLE 5.35 ORAC ANTIOXIDANT CAPACITY (TE, μMOL G DW) OF 20% SPE EXTRACTS OF PEANUT KERNELS ROASTED AT 150 °C

Time NativeA Enzyme-treatedA (min) N Mean ± SEM N Mean ± SEM

0 5 36.5 ± 4.2 a 2 72.8 6.1 a 5 3 46.9 ± 1.0 a 3 72.1 2.1 a 20 3 35.2 ± 4.5 a 3 67.5 4.1 a 30 3 42.1 ± 2.7 a 3 67.2 2.2 a 50 70 3 42.7 ± 3.1 a 3 67.5 4.5 a A Lower-case letters mark significantly different (P<0.05) means according to Tukey’s HSD test

(A) (B) 80 80 Crude extract Native extract 60 60 Enzyme-treated extract dw) dw) -1 -1 40 40 mol g mol mol g mol P P P P TE ( TE 20 ( TE 20

0 0 0 20 40 60 80 0 20 40 60 80 Time (min) Time (min)

FIGURE 5.29 ORAC ANTIOXIDANT CAPACITY OF (A) 5% AND (B) 20% SPE EXTRACTS OF PEANUT KERNELS ROASTED AT 150 °C.

| Antioxidant phytochemicals 245

-1 TABLE 5.36 ORAC ANTIOXIDANT CAPACITY (TE, μMOL G DW) OF 5% SPE EXTRACTS OF PEANUT KERNELS ROASTED AT 160 °C

Time NativeA Enzyme-treatedA (min) N Mean ± SEM N Mean ± SEM

0 6 37.5 ± 3.9 a 3 50.6 ± 4.2 a 5 3 27.5 ± 6.0 a 3 54.6 ± 6.3 a 32.5 3 29.7 ± 3.6 a 3 51.5 ± 4.0 a

A Lower-case letters mark significantly different (P<0.05) means according to Tukey’s HSD test -1 TABLE 5.37 ORAC ANTIOXIDANT CAPACITY (TE, μMOL G DW) OF 20% SPE EXTRACTS OF PEANUT KERNELS ROASTED AT 160 °C

Time NativeA Enzyme-treatedA (min) N Mean ± SEM N Mean ± SEM

0 5 36.5 ± 4.2 a 2 72.8 ± 6.1 a 5 2 27.4 ± 6.8 a 2 74.1 ± 1.4 a 32.5 2 28.5 ± 1.6 a 2 68.4 ± 1.4 a

A Lower-case letters mark significantly different (P<0.05) means according to Tukey’s HSD test

(A) (B) 80 80 Crude extract Native extract 60 60 Enzyme-treated extract dw) dw) -1 -1 l g l 40 g l 40 mo mo P P TE ( TE 20 ( TE 20

0 0 0 10 20 30 40 0 10 20 30 40 Time (min) Time (min)

FIGURE 5.30 ORAC ANTIOXIDANT CAPACITY OF (A) 5% AND (B) 20% SPE EXTRACTS OF PEANUT KERNELS ROASTED AT 160 °C.

| Antioxidant phytochemicals 246

5.2.4.2 Roast effects on phytochemical concentrations in peanut kernels Although significant differences in ORAC-based antioxidant capacity were not discerned, the concentrations of many of the antioxidant phytochemicals measured in this study altered (P<0.05) with roasting according to standard ANOVA (Table 5.38). The data did not always present clear explicable trends, which may be ascribed to practical difficulty in achieving uniform roast treatments at the laboratory scale. Larger scale processing, larger sample sizes (our samples were 200 g), and more experimental replication were necessary to precisely describe roasting kinetics. At this stage of the research however, the primary concern was to investigate overall trends in phytochemical concentrations rather than the statistical differences between specific roast times.

Phytochemical quantitation of the roasted samples is presented graphically in Figures 5.31 to 5.39. Trends in the data were explored through statistical comparisons of the numerical data (i.e., total increase/decrease and Tukey HSD tests) and single-phase decay curves that were fitted to the data to aid visualisation. It is emphasised that the fitted curves should be viewed as illustrative only and do not represent validated kinetic models.

TABLE 5.38 SIGNIFICANCE OF ROASTING DURATION ON PHYTOCHEMICAL CONCENTRATIONS IN PEANUT EXTRACTS

Analyte Method 150 °C ANOVA P-values 160 °C ANOVA P-values SPE Detector Setting Native Enzyme-treated Native Enzyme-treated (%) (nm) extract extract extract extract

Caffeic/vanillic acid 5 PDA 320 0.001** <0.001** 0.896 0.488 p-coumaric acid 5 PDA 320 <0.001** <0.001** <0.001** 0.001** 20 PDA 320 <0.001** <0.001** <0.001** 0.002** Ferulic/sinapic acid 5 PDA 320 <0.001** <0.001** 0.018* <0.001** 20 PDA 320 <0.001** <0.001** 0.043* 0.002** OCU 5 PDA 320 <0.001** <0.001** 0.036* 0.003** 20 PDA 320 <0.001** 0.033* Salicylic acid 5 PDA 305 <0.001** <0.001** 0.001** 0.016* 20 PDA 305 <0.001** 0.002** 0.001** 0.045* Resveratrol 20 PDA 305 <0.001** 0.001** 0.256 0.791 20 RF 260/422 <0.001** <0.001** 0.004** 0.019* Daidzein 20 PDA 305 <0.001** <0.001** <0.001** 0.207 QCU 20 PDA 370 <0.001** 0.006** 0.310 0.114

* Significant (P<0.05) ** Highly significant (P<0.01)

| Antioxidant phytochemicals 247

5.2.4.2.1 Caffeic and vanillic acids trend: free and total concentrations fairly stable Total caffeic/vanillic acid concentrations (Table 5.39) appeared stable during roasting (Figure 5.31), notwithstanding variability in the 150 °C data, and there was no difference between raw kernels and kernels that had been roasted for maximum durations at either temperature.

-1 TABLE 5.39 CAFFEIC/VANILLIC ACID CONCENTRATIONS (μG G DW CAE) IN 5% SPE EXTRACTS OF ROASTED PEANUT KERNELS

150 °C 160 °C Time Free conc.A Total conc.A Time Free conc.A Total conc.A (min) N Mean ± SEM N Mean ± SEM (min) N Mean ± SEM N Mean ± SEM N 0 6 94.7 ± 5.1 a 3 59.1 ± 8.2 c 0 6 94.7 ± 5.1 a 3 59.1 ± 8.2 a 5 3 75.1 ± 5.5 a 3 63.3 ± 2.4 c 5 3 94.3 ± 25.4 a 3 67.8 ± 0.6 a 20 3 104.3 ± 0.8 a 3 109.0 ± 1.1 ab 17.5 3 88.5 ± 8.7 a 30 3 98.1 ± 18.4 a 3 53.0 ± 2.4 c 32.5 3 77.3 ± 28.7 a 3 59.2 ± 4.9 a 50 3 100.8 ± 4.3 a 3 108.9 ± 1.8 a 70 3 30.9 ± 3.4 b 3 74.5 ± 0.9 bc A Lower-case letters mark significantly different (P<0.05) means according to Tukey’s HSD test

(A) 150 (B) 150 Free Total dw) dw) -1 -1 100 100 g g P P P P

50 50 Concentration ( Concentration Concentration ( Concentration

Mean r SEM 0 0 0 20 40 60 80 0 10 20 30 40 Time (min) Time (min) -1 FIGURE 5.31 CAFFEIC/VANILLIC ACID CONCENTRATIONS (μG G DW CAE) IN (A) 150 °C AND (B) 160 °C ROASTED PEANUT KERNELS.

| Antioxidant phytochemicals 248

5.2.4.2.2 Ferulic and sinapic acids trend: free and total concentrations decrease Free and total concentrations (Tables 5.40-5.41) of ferulic/sinapic acid tended to decline during roasting, decreasing about 40-60% in the maximum roast durations compared to the raw kernels (Figure 5.32). The extractable free analyte did not increase with roasting, as in the case of p-coumaric acid, but the total concentrations (i.e., in enzyme-treated extracts) decreased to about the level of the free fraction.

-1 TABLE 5.40 FERULIC/SINAPIC ACID CONCENTRATIONS (μG G DW FAE) IN 5% SPE EXTRACTS OF ROASTED PEANUT KERNELS

150 ° C 160 °C Time Free conc.A Total conc.A Time Free conc.A Total conc.A (min) N Mean (min) SEM N Mean ± SEM (min) N Mean ± SEM N Mean ± SEM 0 6 44.2 ± 5.8 b 3 87.6 ± 2.1 a 0 6 44.2 ± 5.8 a 3 87.6 ± 2.1 a 5 3 37.2 ± 2.6 b 3 59.7 ± 4.6 c 5 3 28.2 ± 6.2 ab 3 65.0 ± 2.1 b 20 3 68.3 ± 4.3 a 3 73.3 ± 1.6 b 17.5 3 42.7 ± 0.7 a 30 3 27.9 ± 0.7 b 3 39.5 ± 0.3 d 32.5 3 17.8 ± 0.6 b 3 37.3 ± 1.1 c 50 3 69.2 ± 1.9 a 3 49.1 ± 0.0 d 70 3 22.2 ± 0.1 b 3 42.4 ± 1.4 d

A Lower-case letters mark significantly different (P<0.05) means according to Tukey’s HSD test -1 TABLE 5.41 FERULIC/SINAPIC ACID CONCENTRATIONS (μG G DW FAE) IN 20% SPE EXTRACTS OF ROASTED PEANUT KERNELS

150 °C 160 °C Time Free conc.A Total conc.A Time Free conc.A Total conc.A (min) N Mean (min) SEM N Mean ± SEM (min) N Mean ± SEM N Mean ± SEM 0 5 50.8 ± 3.8 a 2 90.1 ± 1.1 a 0 5 50.8 ± 3.8 a 2 90.1 ± 1.1 a 5 3 49.7 ± 5.0 ab 3 67.6 ± 1.1 b 5 2 41.9 ± 7.9 ab 2 63.8 ± 2.7 b 20 3 63.6 ± 2.4 a 3 73.9 ± 1.2 b 17.5 3 43.7 ± 1.7 ab 30 3 33.7 ± 2.4 bc 3 43.0 ± 1.1 c 32.5 2 28.8 ± 0.8 b 2 35.3 ± 3.6 c 70 3 29.8 ± 0.9 c 3 36.7 ± 2.4 c

A Lower-case letters mark significantly different (P<0.05) means according to Tukey’s HSD test

| Antioxidant phytochemicals 249

(A) 100 (B) 100 Free Total 80 80 dw) dw) -1 -1 g 60 g P P 60 P

40 40

20 20 Concentration ( Concentration Concentration ( Concentration

Mean r SEM 0 0 0 20 40 60 80 0 10 20 30 40 Time (min) Time (min) (C) 100 (D) 100 Free Total 80 80 dw) dw) -1 -1 g 60 g 60 P P P

40 40

20 20 Concentration ( Concentration Concentration ( Concentration

Mean r SEM 0 0 0 20 40 60 80 0 10 20 30 40 Time (min) Time (min) -1 FIGURE 5.32 FERULIC/SINAPIC ACID CONCENTRATIONS (μG G DW FAE) IN 5% SPE EXTRACTS OF (A) 150 °C AND (B) 160 °C; AND 20% SPE EXTRACTS OF (C) 150 °C AND (D) 160 °C ROASTED PEANUT KERNELS.

5.2.4.2.3 Resveratrol, daidzein, QCU, and o-coumaric acid trend: free and total concentrations stable Resveratrol, daidzein, and QCU concentrations (Tables 5.42- 5.44) also appeared to be stable during roasting, if the outlying 150 °C/20 min data point is excluded. Again, there was no marked change in the relative contents of free and total analyte (Figures 5.33-5.35). The concentrations of o-coumaric acid (Table 5.45 and Figure 5.36) displayed a similar trend if the raw and final data points are emphasised, as there seemed to be a number of samples in the 150 °C set that did not react as usual with enzyme treatment.

| Antioxidant phytochemicals 250

-1 TABLE 5.42 RESVERATROL CONCENTRATIONS (μG G DW) IN 20% SPE EXTRACTS OF ROASTED PEANUT KERNELS

150 °C 160 °C Time Free conc.A Total conc.A Time Free conc.A Total conc.A (min) N Mean (min) SEM N Mean ± SEM (min) N Mean ± SEM N Mean ± SEM 0 5 2.07 ± 0.13 b 2 7.56 ± 0.03 a 0 5 2.07 ± 0.13 a 2 7.56 ± 0.03 a 5 3 2.06 ± 0.12 b 3 8.31 ± 0.43 a 5 2 2.10 ± 0.46 a 2 7.47 ± 0.31 a 20 3 4.81 ± 0.04 a 3 5.36 ± 0.17 b 17.5 3 2.17 ± 0.11 a 30 3 2.31 ± 0.21 b 3 7.66 ± 0.47 a 32.5 2 2.65 ± 0.04 a 2 7.96 ± 0.84 a 70 3 2.34 ± 0.09 b 3 7.16 ± 0.17 a

A Lower-case letters mark significantly different (P<0.05) means according to Tukey’s HSD test

(A) 10 (B) 10 Free Total 8 8 dw) dw) -1 -1 g g 6 6 P P P P

4 4

2 2 Concentration ( Concentration Concentration ( Concentration

Mean r SEM 0 0 0 20 40 60 80 0 10 20 30 40 Time (min) Time (min) -1 FIGURE 5.33 RESVERATROL CONCENTRATIONS (μG G DW) IN (A) 150 °C AND (B) 160 °C ROASTED PEANUT KERNELS. -1 TABLE 5.43 DAIDZEIN CONCENTRATIONS (μG G DW) IN 20% SPE EXTRACTS OF ROASTED PEANUT KERNELS

150 °C 160 °C Time Free conc.A Total conc.A Time Free conc.A Total conc.A (min) N Mean (min) SEM N Mean ± SEM (min) N Mean ± SEM N Mean ± SEM

0 5 0.30 ± 0.02 b 2 12.54 ± 0.00 a 0 5 0.30 ± 0.02 b 2 12.54 ± 0.00 a 5 3 0.22 ± 0.03 b 3 7.97 ± 0.27 b 5 2 0.32 ± 0.06 b 2 12.68 ± 0.02 a 20 3 10.62 ± 1.11 a 3 12.08 ± 0.30 a 17.5 3 0.12 ± 0.00 c 30 3 0.13 ± 0.01 b 3 6.46 ± 0.33 c 32.5 2 0.48 ± 0.01 a 2 11.88 ± 0.45 a 70 3 0.16 ± 0.01 b 3 6.97 ± 0.15 bc A Lower-case letters mark significantly different (P<0.05) means according to Tukey’s HSD test

| Antioxidant phytochemicals 251

(A) 15 (B) 15 Free Total dw) dw) -1 -1 10 10 g g P P P P

5 5 Concentration ( Concentration Concentration ( Concentration

Mean r SEM 0 0 0 20 40 60 80 0 10 20 30 40 Time (min) Time (min) -1 FIGURE 5.34 DAIDZEIN CONCENTRATIONS (μG G DW) IN (A) 150 °C AND (B) 160 °C ROASTED PEANUT KERNELS. -1 TABLE 5.44 QCU CONCENTRATIONS (μG G DW QE) IN 20% SPE EXTRACTS OF ROASTED PEANUT KERNELS

150 °C 160 °C Time Free conc.A Total conc.A Time Free conc.A Total conc.A (min) N Mean (min) SEM N Mean ± SEM (min) N Mean ± SEM N Mean ± SEM

0 5 3.11 ± 0.29 b 2 4.51 ± 0.17 c 0 5 3.11 ± 0.29 a 2 4.51 ± 0.17 a 5 3 3.01 ± 0.19 b 3 5.84 ± 0.31 ab 5 2 1.56 ± 1.56 a 2 5.54 ± 0.09 a 20 3 5.17 ± 0.08 a 3 6.23 ± 0.22 a 17.5 3 2.57 ± 0.08 a 30 3 2.57 ± 0.08 b 3 5.03 ± 0.22 bc 32.5 2 2.54 ± 0.08 a 2 5.80 ± 0.50 a 70 3 2.59 ± 0.03 b 3 5.10 ± 0.22 abc A Lower-case letters mark significantly different (P<0.05) means according to Tukey’s HSD test

(A) 8 (B) 8 Free Total

dw) 6 dw) 6 -1 -1 g g P P P 4 4

2 2 Concentration ( Concentration Concentration ( Concentration

Mean r SEM 0 0 0 20 40 60 80 0 10 20 30 40 Time (min) Time (min) -1 FIGURE 5.35 QCU CONCENTRATIONS (μG G DW QE) IN (A) 150 °C AND (B) 160 °C ROASTED PEANUT KERNELS.

| Antioxidant phytochemicals 252

-1 TABLE 5.45 OCU CONCENTRATIONS (μG G DW OAE) IN 5% SPE EXTRACTS OF ROASTED PEANUT KERNELS

150 °C 160 °C Time Free conc.A Total conc.A Time Free conc.A Total conc.A (min) N Mean (min) SEM N Mean ± SEM (min) N Mean ± SEM N Mean ± SEM 0 6 5.3 ± 0.5 abc 3 21.6 ± 0.4 a 0 6 5.3 ± 0.5 a 3 21.6 ± 0.4 a 5 3 3.4 ± 0.1 c 3 4.2 ± 0.2 d 5 3 3.6 ± 0.4 a 3 22.0 ± 0.6 a 20 3 5.7 ± 0.2 ab 3 7.0 ± 1.1 c 17.5 3 4.6 ± 0.2 a 30 3 3.6 ± 0.2 bc 3 4.5 ± 0.1 cd 32.5 3 3.8 ± 0.2 a 3 18.5 ± 0.3 b 50 3 7.6 ± 1.0 a 3 7.1 ± 0.1 c 70 3 3.7 ± 0.1 bc 3 18.9 ± 0.5 b

A Lower-case letters mark significantly different (P<0.05) means according to Tukey’s HSD test

(A) 25 (B) 25 Free Total 20 20 dw) dw) -1 -1 g 15 g 15 P P P P

10 10

5 5 Concentration ( Concentration Concentration ( Concentration

Mean r SEM 0 0 0 20 40 60 80 0 10 20 30 40 Time (min) Time (min) -1 FIGURE 5.36 OCU CONCENTRATIONS (μG G DW OAE) IN (A) 150 °C AND (B) 160 °C ROASTED PEANUT KERNELS.

5.2.4.2.4 Salicylic acid trend: free and total concentrations increase

-1 TABLE 5.46 SALICYLIC ACID CONCENTRATIONS (μG G DW) IN 5% SPE EXTRACTS OF ROASTED PEANUT KERNELS

150 °C 160 °C Time Free conc.A Total conc.A Time Free conc.A Total conc.A (min) N Mean (min) SEM N Mean ± SEM (min) N Mean ± SEM N Mean ± SEM

0 6 1.82 ± 0.05 e 3 7.19 ± 0.04 cd 0 6 1.82 ± 0.05 c 3 7.19 ± 0.04 ab 5 3 1.81 ± 0.12 e 3 6.55 ± 0.38 d 5 3 2.03 ± 0.18 c 3 6.54 ± 0.23 b 20 3 7.15 ± 0.35 b 3 8.14 ± 0.19 c 17.5 3 3.48 ± 0.04 b 30 3 3.21 ± 0.15 d 3 7.49 ± 0.18 cd 32.5 3 3.95 ± 0.11 a 3 8.04 ± 0.37 a 50 3 8.22 ± 0.15 a 3 13.83 ± 0.40 a 70 3 5.59 ± 0.41 c 3 9.85 ± 0.52 b

A Lower-case letters mark significantly different (P<0.05) means according to Tukey’s HSD test

| Antioxidant phytochemicals 253

-1 TABLE 5.47 SALICYLIC ACID CONCENTRATIONS (μG G DW) IN 20% SPE EXTRACTS OF ROASTED PEANUT KERNELS

150 °C 160 °C Time Free conc.A Total conc.A Time Free conc.A Total conc.A (min) N Mean (min) SEM N Mean ± SEM (min) N Mean ± SEM N Mean ± SEM 0 5 1.93 ± 0.06 d 2 9.79 ± 0.07 ab 0 5 1.93 ± 0.06 c 2 9.79 ± 0.07 ab 5 3 2.15 ± 0.33 d 3 8.62 ± 0.04 b 5 2 2.30 ± 0.35 c 2 9.40 ± 0.45 b 20 3 7.36 ± 0.52 a 3 8.20 ± 0.14 b 17.5 3 3.41 ± 0.14 b 30 3 3.71 ± 0.41 c 3 9.28 ± 0.51 b 32.5 2 6.38 ± 0.03 a 2 11.83 ± 0.54 a 70 3 5.37 ± 0.35 b 3 11.25 ± 0.54 a

A Lower-case letters mark significantly different (P<0.05) means according to Tukey’s HSD test Salicylic acid concentrations (Tables 5.46-5.47) displayed slight increase during roasting at 150 °C and 160 °C, but there was little change in the relative contents of free and total analyte (Figure 5.37).

15 15 (A) (B) Free Total dw) dw) -1 -1 10 10 g g P P P

5 5 Concentration ( Concentration Concentration ( Concentration Mean r SEM 0 0 0 20 40 60 80 0 10 20 30 40 Time (min) Time (min) (C) 15 (D) 15 Free Total dw) dw) -1 10 -1 10 g g P P P

5 5 Concentration ( Concentration Concentration ( Concentration

Mean r SEM 0 0 0 20 40 60 80 0 10 20 30 40 Time (min) Time (min) -1 FIGURE 5.37 SALICYLIC ACID CONCENTRATIONS (μG G DW) IN 5% SPE EXTRACTS OF (A) 150 °C AND (B) 160 °C; AND 20% SPE EXTRACTS OF (C) 150 °C AND (D) 160 °C ROASTED PEANUT KERNELS.

| Antioxidant phytochemicals 254

5.2.4.2.5 p-Coumaric acid trend: free and total concentrations converge The concentrations of p-coumaric acid (Tables 5.48-5.49) converged at a level partway between the free and total concentrations of raw kernels (Figure 5.38). In other words, roasting apparently caused reactions that reduced the total amount of extractable p-coumaric acid, but at the same time improved the extractability of the bound fraction (i.e., enhanced free analyte concentrations, in the native extract). Trendlines in both the native and enzyme- treated extracts plateau after ~ 20 min of roasting at either temperature, implying that p-coumaric acid is thermo-stable during the roasting process. The decline in total p-coumaric acid concentrations (i.e., the total analyte in the enzyme-treated extract of raw kernels) may be due to reposition in thermally-induced reaction products rather than degradation, as the decrease in concentration does not continue as roasting time is prolonged.

-1 TABLE 5.48 P-COUMARIC ACID CONCENTRATIONS (μG G DW) IN 5% SPE EXTRACTS OF ROASTED PEANUT KERNELS

150 °C 160 °C Time Free conc.A Total conc.A Time Free conc.A Total conc.A (min) N Mean (min) SEM N Mean ± SEM (min) N Mean ± SEM N Mean ± SEM

0 6 19.1 ± 2.0 d 3 84.3 ± 5.9 a 0 6 19.1 ± 2.0 b 3 84.3 ± 5.9 a 5 3 20.1 ± 0.8 d 3 52.0 ± 4.6 cd 5 3 16.1 ± 3.8 b 3 49.6 ± 1.8 b 20 3 73.8 ± 6.7 a 3 61.9 ± 2.8 bc 17.5 3 48.3 ± 1.0 a 30 3 45.4 ± 0.5 c 3 44.1 ± 0.9 d 32.5 3 49.7 ± 1.0 a 3 53.7 ± 1.7 b 50 3 64.0 ± 1.2 ab 3 69.9 ± 0.3 ab 70 3 53.7 ± 0.6 bc 3 55.7 ± 1.0 bcd

A Lower-case letters mark significantly different (P<0.05) means according to Tukey’s HSD test -1 TABLE 5.49 P-COUMARIC ACID CONCENTRATIONS (μG G DW) IN 20% SPE EXTRACTS OF ROASTED PEANUT KERNELS

150 °C 160 °C Time Free conc.A Total conc.A Time Free conc.A Total conc.A (min) N Mean (min) SEM N Mean ± SEM (min) N Mean ± SEM N Mean ± SEM 0 5 21.9 ± 0.5 d 2 67.4 ± 1.1 a 0 5 21.9 ± 0.5 c 2 67.4 ± 1.1 a 5 3 33.0 ± 1.6 c 3 49.0 ± 1.5 b 5 2 24.4 ± 3.7 c 2 35.5 ± 1.2 b 20 3 61.1 ± 2.6 a 3 60.2 ± 1.4 a 17.5 3 50.7 ± 1.6 b 30 3 51.8 ± 3.2 b 3 45.6 ± 0.8 b 32.5 2 58.1 ± 0.3 a 2 38.6 ± 2.7 b 70 3 61.7 ± 2.3 a 3 46.0 ± 4.1 b

A Lower-case letters mark significantly different (P<0.05) means according to Tukey’s HSD test

| Antioxidant phytochemicals 255

(A) 100 (B) 100 Free Total 80 80 dw) dw) -1 -1 g g 60 60 P P P

40 40

20 20 Concentration ( Concentration Concentration ( Concentration

Mean r SEM 0 0 0 20 40 60 80 0 10 20 30 40 Time (min) Time (min) (C) 100 (D) 100 Free Total 80 80 dw) dw) -1 -1 g g 60 60 P P P

40 40

20 20 Concentration ( Concentration Concentration ( Concentration

Mean r SEM 0 0 0 20 40 60 80 0 10 20 30 40 Time (min) Time (min) -1 FIGURE 5.38 P-COUMARIC ACID CONCENTRATIONS (μG G DW) IN 5% SPE EXTRACTS OF (A) 150 °C AND (B) 160 °C; AND 20% SPE EXTRACTS OF (C) 150 °C AND (D) 160 °C ROASTED PEANUT KERNELS. Publications regarding quantitative changes in phytochemical composition associated with roasting in peanuts are scant. In one study, increased p-coumaric acid concentrations were observed following roasting (Talcott et al., 2005b). Another study found no significant changes in isoflavone or resveratrol concentrations after oil- or dry-roasting (Chukwumah et al., 2007a). Possibly the most interesting observation in the latter work was the lack of daidzein detected in raw kernels. Such results may reflect a failure to extract bound phytochemicals from the sample matrix, as the researchers did not employ any enzyme or hydrolysis procedure in their extraction method. Roasting appears to have increased the extractability of the daidzein in the peanut kernels, thus producing the emergence of daidzein in the roasted kernels. In contrast, the roast treatments in our study did not increase extractable daidzein, which were measured at much higher concentrations in the enzyme-treated extracts than the native extracts (Table 5.43 and Figure 5.34).

Only one time course study of roasting effects on the peanut antioxidant phytochemical profile is available thus far (Win et al., 2011). The authors reported statistical differences between

| Antioxidant phytochemicals 256

samples of peanuts with and without testa that were roasted at 160 °C for various durations up to 50 min. The phytochemical concentrations were considered to represent total content as acid hydrolysates were analysed. However the HPLC quantitation was not entirely persuasive due to evident problems with peak fronting (likely because injections had excessive solvent strength compared to the mobile phase) and retention time shift that can be directly discerned in the published chromatograms. Nonetheless, the general trends are of interest. Extractable concentrations of p-hydroxybenzoic acid, chlorogenic acid, p-coumaric acid, and quercetin in blanched kernels were reported to increase with roasting but declined by 15-30% after 30-50 min. Kaempferol and resveratrol were measured at very low concentrations and declined proportionately greater amounts (38 and 73%). The results of this study contradicted Win and co-workers’ interpretation that antioxidant compounds would degrade on extension of roasting time past their suggested ‘optimal’ duration of 20 min. Our results suggested that ferulic/sinapic acid concentrations declined with roasting, but that other tested compounds were relatively thermo-stable after roasting at 150 °C for 70 min and 160 °C for 32.5 min.

5.2.5 Conclusion

The antioxidant phytochemicals in native and enzyme-treated aqueous methanolic extracts of peanuts were profiled by HPLC-PDA, with HPLC-RF, HRMS, and GC-MS data to support the compound identification and quantitation. Five genotypes with diverse antioxidant capacity according to ORAC screening (chapters 4.2 and 4.3) were analysed, as well as kernel samples subjected to a range of roast treatments (150 °C for 0-70 min and 160 °C for 0-32.5 min). Genotype differences, roasting effects, and the relative concentrations of free and bound fractions of the peanut extracts were investigated. Peanut extracts and fractions of HPLC eluate (time intervals and specific peaks) were subjected to ORAC assay to evaluate total and relative contribution to kernel antioxidant capacity.

Key findings included significant differences in phytochemical composition among genotypes. For example, Fisher and Sutherland had high caffeic/vanillic acid and salicylic acid contents compared to the other genotypes; D147-p8-6F and Sutherland had relatively high concentrations of p-coumaric acid, ferulic/sinapic acid, daidzein, and OCU; UF32 had relatively high concentrations of resveratrol and QCU; while Page had low concentrations of most tested analytes.

Comparison of native and enzyme-treated extracts highlighted the importance of the bound fraction in peanuts. Our study suggested that 77-93% of the p-coumaric acid, 44-53% of the ferulic/sinapic acid, 71-89% of the salicylic acid, 59-68% of the resveratrol, and 89-97% of the

| Antioxidant phytochemicals 257

daidzein in raw peanut kernels were present in bound forms. The implications for plant breeding and functional food utility require an improved understanding of how chemical structure and bonding affects bioaccessibility and bioavailability after consumption. Certainly, analytical results need to be interpreted in view of the varying extractability of phytochemicals. For example, the results suggest that resveratrol is underestimated in much of the peanut literature, as only simple solvent extractions have been used.

A small-scale study of roasting effects on phytochemical composition was conducted. Ferulic/sinapic acid concentrations declined with roasting, but other tested compounds were relatively thermo-stable after processing at 150 °C for 70 min and 160 °C for 32.5 min. These findings have great practical relevance as, outside of oil production, peanuts are most commonly consumed as roasted kernels. There is little published data available regarding the effects of roasting on peanut phytochemicals, so our study makes a valuable contribution to our understanding. It would be worthwhile to expand the study to allow quantitative modelling.

PCA of antioxidant capacity and phytochemical concentrations allowed exploration of the interrelationships among variables. Interestingly, PC-1 (correlated with p-coumaric acid, ferulic/sinapic acid, OCU, and daidzein) and PC-2 (correlated with caffeic/vanillic acid, salicylic acid, resveratrol, and QCU) of total analyte concentrations were able to distinguish the tested peanut genotypes. But antioxidant capacity was better described by PC-3 and PC-4, which were associated with a combination of phytochemicals (caffeic/vanillic acid, p-coumaric acid, ferulic/sinapic acid, and resveratrol) that were not directly correlated in bivariate analysis (e.g., resveratrol was negatively correlated with caffeic/vanillic and p-coumaric acid).

ORAC assay of peanut fractions indicated that while the phytochemicals present in the highest concentrations, namely, p-coumaric, ferulic/sinapic, and caffeic/vanillic acids, contributed the most to peanut antioxidant capacity, resveratrol and QCU had the highest antioxidant capacity relative to their low concentration in kernels. This data is speculative given that the concentrations of QCU and several other compounds were calculated in standard equivalents, but remains intriguing as it suggests a small increase in the concentration of these compounds could induce a disproportionately large increase in antioxidant capacity. The analysis revealed a number of unidentified peaks that made important contributions to peanut antioxidant capacity at 7.0, 8.0, and 13.4 min, as well as QCU and a peak tentatively identified as cinnamic acid. It will be a priority of future research to identify and characterise these compounds.

| Antioxidant phytochemicals 258

6 Conclusion

This PhD project undertook exploratory research regarding two categories of functional food traits: dietary minerals and antioxidants. The primary objectives were to (1) characterise diverse peanut phenotypes using established methods or developing and validating analytical methods if necessary; (2) estimate genotypic variation in the functional food trait; and (3) investigate the stability of the functional food trait through studies of G × E interaction.

Essential mineral concentrations in kernels were analysed by ICP-OES and ICP-(DRC)-MS after preparation of samples by microwave-assisted closed acid digestion. These methods are not new, but the published validation of protocols specific for peanuts or oilseeds is scant. In view of this, the scientific validation of protocols for elemental analysis of peanuts in Chapter 2.1 will be useful for analysts. The accuracy and precision of ICP-OES analysis of Ca, K, Mg, Mn, Na, and P; and ICP-(DRC)-MS analysis of Co, Cu, Fe, Mo, Ni, and Zn in peanut kernels was established. Some further testing is warranted to improve B, Cr, Na, and Se analyses, which remain challenging because of the very low concentrations in peanuts, and inaccuracy or imprecision in analysis at such trace levels.

Antioxidant capacity was assessed using popular in vitro antioxidant assays that were adapted to a 96-well microplate format for high-throughput analysis. Established protocols for the ABTS•+ decolourisation, DPPH• scavenging, FC total phenolics, and ORAC assays were used with slight optimisations for instrument and laboratory conditions, and adaptation to the microplate platform. Comparison of the assays’ performance in our study of genotypic variation is discussed in Chapter 2.2, and led us to conclude that the ORAC protocol was the most sensitive and repeatable assay, and hence the most useful for screening of peanut germplasm.

Antioxidant assays are composite measures and it remains necessary to quantitatively profile peanut phytochemicals in order to understand the underlying basis for variation in kernel antioxidant capacity. A number of phenolic acids, most notably p-coumaric acid, have been reported in peanuts at relatively high concentrations; however many peanut phytochemicals of interest occur at trace concentrations. At such levels, it is especially important to ensure that the detection limits, accuracy, and precision of analytical protocols are validated. The validation of protocols for ultrasound- and enzyme-assisted solvent extraction, SPE to purify and concentrate extracts, and HPLC-PDA analysis of several phenolic acids, stilbenes, and flavonoids (i.e., caffeic acid, p-coumaric acid, ferulic acid, o-coumaric acid, salicylic acid, resveratrol, daidzein, quercetin, and kaempferol) is described in Chapter 2.3. The work assures

| Conclusion 259

the quality of the experimental data and may be useful to analysts as few methods for broad phytochemical profiling in peanuts have been published.

Genotypic variation was evaluated by analysis of 32 full-season maturity (Arachis hypogaea ssp. hypogaea) and 24 ultra-early maturity (A. hypogaea ssp. fastigiata) genotypes grown during one season in APBP varietal trials at Kairi research station in north Queensland. The analysis of essential minerals reported in Chapter 3.2 revealed that there were useful levels of genotypic variation of more than 10% RSD in concentrations of most of the tested elements. Meanwhile, the assays of antioxidant capacity described in Chapter 4.2 measured useful levels of variation of more than 20% RSD across the tested breeding lines, although only the hydrophilic ORAC assay revealed statistically significant differences between genotypes.

The studies affirmed the breeding potential for both essential mineral and antioxidant-related traits among the current Australian peanut germplasm. The APBP has a diverse germplasm collection including lines sourced from the USA, ICRISAT, and other programs; the genotypes tested in our study were accordingly considered to be diverse. However it may in future be worthwhile to analyse more exotic genotypes such as wild-types and landraces that potentially express more extreme phytochemical traits, especially given the role of phenolic compounds in plant stress and adaptation responses.

The multi-environment study of essential mineral phenotypes reported in Chapter 3.3 and antioxidant capacity in Chapter 4.3 indicated that genotype, environment, and G × E interaction all had significant effects on both traits. The results confirmed that there was substantial genetic control of essential mineral concentrations and antioxidant capacity in peanut kernels, but the interaction with environmental factors needs to be considered. Expansion of the dataset is warranted to explore these relationships in greater depth, especially the G × E interaction affecting antioxidant capacity, of which our knowledge is very deficient.

G × E interaction may complicate plant/seed selection in plant breeding, but may also present opportunities for pre-harvest management or novel processing to enhance trait expression. Certainly many antioxidant phytochemicals may be induced in plants by abiotic and biotic stresses, and some research has already been conducted on the use of elicitation treatments such as imbibition and UV irradiation to increase antioxidant concentrations. Experiments to elucidate the underlying biochemistry and effects on other kernel qualities would be worthwhile. On the other hand, the influence of soil nutrition and rhizosphere interactions on nutrient acquisition is well known. It is feasible that a strategy combining genetic improvement

| Conclusion 260

and agronomic management could be designed to ensure optimal concentrations of minerals are consistently achieved, bringing benefits during the growing season as well as for postharvest nutritional quality.

The essential minerals data were used to develop approximately predictive NIRS calibrations of Ca, K, Mg, and P, reported in Chapter 3.4. The models had a level of accuracy that would be useful for application in plant breeding programs. Such NIRS models represent a cost-effective means of selecting segregating plants in breeding populations with superior nutritional composition for progression in the APBP, especially as samples are already subject to routine analysis by NIRS for determination of oil chemistry, moisture and other constituents. Further analysis of more diverse peanut samples to expand the data set is warranted to confirm batch- to-batch accuracy and improve the robustness of calibrations. We are also interested to examine the feasibility of developing NIRS calibrations for antioxidant-related traits.

Quantitative analysis of the antioxidant phytochemical profile, reported in Chapter 5.2, was performed on five peanut genotypes with diverse antioxidant capacity (D147-p8-6F, Fisher, Page, Sutherland, and UF32), as well as Middleton kernels subjected to varying roasting treatments (150 °C, 0-70 min and 160 °C, 0-32.5 min). There were significant differences in the phytochemical concentrations of genotypes, for example, Fisher and Sutherland had high caffeic/vanillic acid and salicylic acid contents compared to the other genotypes, while D147- p8-6F and Sutherland had relatively high concentrations of p-coumaric acid, ferulic/sinapic acid, and daidzein. Interestingly, PCA suggested that the genotypes could be distinguished by their phytochemical profile (i.e., clustering on scores plots), but the same PC poorly described antioxidant capacity.

The result was unexpected, but may relate to the relative antioxidant capacity of phytochemicals, which was not necessarily correlated with their concentration in kernels. Assessment of the ORAC antioxidant capacity of peanut extracts and fractions indicated that while the phytochemicals present in the highest concentrations, namely, p-coumaric, ferulic/sinapic, and caffeic/vanillic acids, contributed the most to peanut antioxidant capacity, resveratrol and QCU (an unknown peak co-eluting with quercetin) had the highest antioxidant capacity relative to their low concentration in kernels. The finding is intriguing as it suggests a small increase in the concentration of these compounds could induce a disproportionately large increase in antioxidant capacity.

The analysis revealed a number of unidentified peaks that made important contributions to peanut antioxidant capacity at 7.0, 8.0, and 13.4 min, as well as QCU and a peak tentatively

| Conclusion 261

identified as cinnamic acid. It will certainly be a priority of future research to identify and characterise these unknown peaks, and examine the potential additive/synergistic interactions that may explain the exceptional antioxidant capacities of peanut fractions, which do not seem to reflect the relative antioxidant capacities of pure standards that have been published (e.g., in structure-activity relationship studies).

Roasting effects on phytochemical composition were investigated in a small-scale study presented in Chapter 5.2.4. Ferulic/sinapic acid concentrations declined with roasting, but other tested compounds were relatively thermo-stable after processing at 150 °C for 70 min and 160 °C for 32.5 min. The findings have great practical relevance as, outside of oil production, peanuts are most commonly consumed as roasted kernels. There is little published data available regarding the effects of roasting on peanut phytochemicals, so our study makes a valuable contribution to our understanding. The measurements were more variable than desirable. It would be worthwhile to expand the study with larger sample size and more replicates given the difficulty of achieving uniform roasting at the laboratory scale. It would also be interesting to expand the number of treatments in order to allow quantitative modelling, including effects on the relative free and bound fractions.

Comparison of native and enzyme-treated extracts suggested that the majority of p-coumaric acid, ferulic/sinapic acid, salicylic acid, resveratrol, and daidzein in peanut kernels may be present bound or trapped within the peanut matrix. Our knowledge of the conjugated (e.g., glycosidic, esterified) forms of phytochemicals in peanuts – moreover in plants generally – is deficient. There have been a few publications that identified flavonoid glycosides in peanuts, but almost no other information is available.

The chemical structure and bonding of phytochemicals affects their bioaccessibility and bioavailability after consumption. For example, most esterified hydroxycinnamates are hydrolysed and taken up in the intestinal lumen, but cell wall-bound phenolics are thought to remain intact until the colon, where there is limited microflora-assisted fermentation and absorption. Until recently, the bound fraction of phytochemicals was thought to be unimportant due to this poor bioaccessibility in the upper gastrointestinal tract. However, recent research suggests that phenolic compounds associated with dietary fibre may exert antioxidant/reducing activity in the upper gastrointestinal tract by surface and protein interactions, and possess significant function in the maintenance of gut health due to their prolonged presence in the gastrointestinal tract

| Conclusion 262

It would be interesting to analyse the cell wall-bound and protein-bound phytochemicals in peanuts. Our enzyme treatment was developed to improve extraction efficiency and was not properly optimised or validated for effects on conjugation. It would be worthwhile to develop more specific extraction and analytical methodology to examine the bound phytochemicals in peanut kernels, given the importance they have according to our quantitative data. Analysis of the conjugated and bound composition of phytochemicals in peanut kernels, and characterisation of their bioaccessibility, bioavailability, and metabolism will be rich areas of further research.

| Conclusion 263

7 Appendix

TABLE 7.1 CHEMICAL STRUCTURES OF PHENOLIC ACIDS REPORTED IN PEANUT KERNELS

Analyte Structure

Common name Caffeic acid Synonyms 3,4-dihydroxycinnamic acid; (E)-3-(3,4-dihydroxyphenyl)acrylic acid

Formula C9H8O4

Molecular weight 180.16 Exact mass 180.04 Common name Chlorogenic acid Synonyms

Formula C16H18O9 Molecular weight 354.31 Exact mass 354.10

Common name Cinnamic acid Synonyms

Formula C9H8O2 Molecular weight 148.16

Exact mass 148.05 Common name m-Coumaric acid Synonyms 3-hydroxycinnamic acid; (E)-3-(3-hydroxyphenyl)acrylic acid Formula C H O 9 8 3 Molecular weight 164.16 Exact mass 164.05 Common name o-Coumaric acid Synonyms 2-hydroxycinnamic acid; (E)-3-(2-hydroxyphenyl)acrylic acid

Formula C9H8O3

Molecular weight 164.16 Exact mass 164.05 Common name p-Coumaric acid Synonyms 4-hydroxycinnamic acid; (E)-3-(4-hydroxyphenyl)acrylic acid

Formula C9H8O3

Molecular weight 164.16 Exact mass 164.05

| Appendix 264

TABLE 7.1 CONTINUED

Analyte Structure

Common name Ferulic acid Synonyms 4-hydroxy-3-methoxycinnamic acid; (E)-3-(4-hydroxy-3-methoxyphenyl)acrylic acid

Formula C10H10O4

Molecular weight 194.18 Exact mass 194.06 Common name Gentisic acid Synonyms 2,5-dihydroxybenzoic acid

Formula C7H6O4 Molecular weight 154.12

Exact mass 154.03 Common name p-Hydroxybenzoic acid Synonyms 4- hydroxybenzoic acid

Formula C7H6O3 Molecular weight 138.12

Exact mass 138.03 Common name Protocatechuic acid Synonyms 3,4-dihydroxybenzoic acid

Formula C7H6O4 Molecular weight 154.12

Exact mass 154.03 Common name Salicylic acid Synonyms 2-hydroxybenzoic acid

Formula C7H6O3 Molecular weight 138.12

Exact mass 138.03 Common name Sinapic acid Synonyms 4-hydroxy-3,5-dimethoxycinnamic acid; (E)-3-(4-hydroxy-3,5-dimethoxyphenyl)acrylic acid

Formula C11H12O5 Molecular weight 224.21

Exact mass 224.07 Common name Syringic acid Synonyms 4-hydroxy-3,5-dimethoxybenzoic acid

Formula C9H10O5 Molecular weight 198.17 Exact mass 198.05

| Appendix 265

TABLE 7.1 CONTINUED

Analyte Structure

Common name Vanillic acid Synonyms 4-hydroxy-3-methoxybenzoic acid

Formula C8H8O4 Molecular weight 168.15

Exact mass 168.04

| Appendix 266

TABLE 7.2 CHEMICAL STRUCTURES OF FLAVONOIDS REPORTED IN PEANUT KERNELS

Analyte Structure

Common name Chrysoeriol Synonyms 5,7-dihydroxy-2-(4-hydroxy-3- methoxyphenyl)-4H-chromen-4-one

Formula C16H12O6 Molecular weight 300.26 Exact mass 300.06

Common name Dihydroquercetin Synonyms 2-(3,4-dihydroxyphenyl)-3,5,7- trihydroxychroman-4-one

Formula C15H12O7 Molecular weight 304.25 Exact mass 304.06

Common name 5,7-dihydroxychromone Synonyms 5,7-dihydroxy-4H-chromen-4-one

Formula C9H6O4 Molecular weight 178.14

Exact mass 178.03 Common name Eriodictyol Synonyms 2-(3,4-dihydroxyphenyl)-5,7- dihydroxychroman-4-one

Formula C15H12O6 Molecular weight 288.25 Exact mass 288.06

Common name Kaempferol Synonyms 3,5,7-trihydroxy-2-(4-hydroxyphenyl)-4H- chromen-4-one

Formula C15H10O6 Molecular weight 286.24 Exact mass 286.05

| Appendix 267

TABLE 7.2 CONTINUED

Analyte Structure

Common name Luteolin Synonyms 3’,4’,5’,7-tetrahydroxyflavone; 2-(3,4-dihydroxyphenyl)-5,7-dihydroxy-4H- chromen-4-one

Formula C15H10O6 Molecular weight 286.24 Exact mass 286.05

Common name Myricetin Synonyms 3,5,7-trihydroxy-2-(3,4,5-trihydroxyphenyl)- 4H-chromen-4-one

Formula C15H10O8 Molecular weight 318.24 Exact mass 318.04

Common name Quercetin Synonyms 2-(3,4-dihydroxyphenyl)-3,5,7-trihydroxy- 4H-chromen-4-one

Formula C15H10O7 Molecular weight 302.24 Exact mass 302.04

| Appendix 268

TABLE 7.3 CHEMICAL STRUCTURES OF ISOFLAVONOIDS REPORTED IN PEANUT KERNELS

Analyte Structure

Common name Biochanin A Synonyms 5,7-dihydroxy-3-(4-methoxyphenyl)-4H- chromen-4-one

Formula C16H12O5 Molecular weight 284.26

Exact mass 284.07 Common name Daidzein Synonyms 7-hydroxy-3-(4-hydroxyphenyl)-4H- chromen-4-one

Formula C15H10O4 Molecular weight 254.24

Exact mass 254.06 Common name Formononetin Synonyms 7-hydroxy-3-(4-methoxyphenyl)-4H- chromen-4-one

Formula C16H12O4 Molecular weight 268.26

Exact mass 268.07 Common name Genistein Synonyms 5,7-dihydroxy-3-(4-hydroxyphenyl)-4H- chromen-4-one

Formula C15H10O5 Molecular weight 270.24

Exact mass 270.05

| Appendix 269

TABLE 7.4 CHEMICAL STRUCTURES OF STILBENES REPORTED IN PEANUT KERNELS

Analyte Structure

Common name Arachidin-1 Synonyms trans-4-(3-methyl-1- butenyl)piceatannol; 4-((E)-3,5-dihydroxy-4-((E)-3- methylbut-1-en-1- yl)styryl)benzene-1,2-diol

Formula C19H20O4 Molecular weight 312.36 Exact mass 312.14 Common name Arachidin-2 Synonyms trans-4-(3-methyl-2- butenyl)resveratrol; (E)-5-(4-hydroxystyryl)-2-(3- methylbut-2-en-1-yl)benzene- 1,3-diol

Formula C19H20O3 Molecular weight 296.36 Exact mass 296.14 Common name Arachidin-3 Synonyms trans-4-(3-methyl-1- butenyl)resveratrol; 5-((E)-4-hydroxystyryl)-2-((E)- 3-methylbut-1-en-1- yl)benzene-1,3-diol

Formula C19H20O3 Molecular weight 296.36 Exact mass 296.14 Common name Arahypin-1 Synonyms trans-4’-deoxyarachidin-3; 2-((E)-3-methylbut-1-en-1-yl)- 5-((E)-styryl)benzene-1,3-diol

Formula C19H20O2

Molecular weight 280.36 Exact mass 280.15

| Appendix 270

TABLE 7.4 CONTINUED

Analyte Structure

Common name Arahypin-2 trans-3’-(2’’,3’’,-dihydroxy- 3’’-methylbutyl)resveratrol; Synonyms (S,E)-5-(3-(2,3-dihydroxy-3- methylbutyl)-4- hydroxystyryl)benzene-1,3- diol

Formula C19H22O5

Molecular 330.37 weight Exact mass 330.15 Common name Arahypin-3 Synonyms trans-4-(2’’,3’’-dihydroxy- 3’’-methylbutyl)resveratrol; (R,E)-2-(2,3-dihydroxy-3- methylbutyl)-5-(4- hydroxystyryl)benzene-1,3- diol

Formula C19H22O5 Molecular 330.37 weight Exact mass 330.15 Common name Arahypin-4 Synonyms trans-4-(2’’,3’’-dihydroxy- 3’’-methylbutyl)-4’- deoxyresveratrol; (R,E)-2-(2,3-dihydroxy-3- methylbutyl)-5- styrylbenzene-1,3-diol

Formula C19H22O4 Molecular 314.38 weight Exact mass 314.15 Common name Arahypin-5 Synonyms (E)-7-(4-hydroxystyryl)-2,2- dimethyl-2H-chromen-5-ol

Formula C19H18O3 Molecular 294.34 weight Exact mass 294.13

| Appendix 271

TABLE 7.4 CONTINUED

Analyte Structure

Common name Chiricanine A Synonyms trans-4’-deoxyarachidin-2; (E)-2-(3-methylbut-2-en-1-yl)-5- styrylbenzene-1,3-diol

Formula C19H20O2

Molecular weight 280.36 Exact mass 280.15 Common name Isopentadienylresveratrol Synonyms trans-3’-isopentadienyl-3,5,4’- trihydroxystilbene; 5-((E)-4-hydroxy-3-((E)-3-methylbuta- 1,3-dien-1-yl)styryl)benzene-1,3-diol

Formula C19H18O3 Molecular weight 294.34

Exact mass 294.13 Common name Piceatannol Synonyms 3,5,3',4'-Tetrahydroxy-trans-stilbene; (E)-4-(3,5-dihydroxystyryl)benzene- 1,2-diol

Formula C14H12O4 Molecular weight 244.24

Exact mass 244.07 Common name Piceid Synonyms Reservatrol-3-β-mono-D-glucoside; (3R,4S,5S,6R)-2-(3-hydroxy-5-((E)-4- hydroxystyryl)phenoxy)-6- (hydroxymethyl)tetrahydro-2H-pyran- 3,4,5-triol

Formula C20H22O8 Molecular weight 390.38 Exact mass 390.13

| Appendix 272

TABLE 7.4 CONTINUED

Analyte Structure

Common name Pterostilbene Synonyms 4-hydroxy-3’,5’-dimethoxystilbene; (E)-4-(3,5-dimethoxystyryl)phenol

Formula C16H16O3 Molecular weight 256.30 Exact mass 256.11 Common name Resveratrol Synonyms 3,5,4’-trihydroxystilbene; (E)-5-(4-hydroxystyryl)benzene-1,3- diol

Formula C14H12O3

Molecular weight 228.24 Exact mass 228.08 Common name SB-1 Synonyms 2-(3-((E)-3,5-dihydroxy-4-((E)-3- methylbut-1-en-1-yl)styryl)-5-oxo-2,5- dihydrofuran-2-yl)acetic acid

Formula C19H20O6 Molecular weight 344.36 Exact mass 344.13

| Appendix 273

8 Publications to date

8.1 Peer-reviewed articles

Phan-Thien K-Y, Wright GC, Lee NA (2012). ‘Inductively coupled plasma-mass spectrometry (ICP-MS) and -optical emission spectroscopy (ICP–OES) for determination of essential minerals in closed acid digestates of peanuts (Arachis hypogaea L.)’. Food Chemistry 134, 453-460.

Phan-Thien K-Y, Golic M, Wright GC, Lee NA (2011). ‘Feasibility of estimating peanut essential minerals by near infrared reflectance spectroscopy’. Sensing and Instrumentation for Food Quality and Safety 5, 43-49.

Phan-Thien K-Y, Wright GC, Lee NA (2010). ‘Genotype-by-environment interaction affects the essential mineral composition of peanut (Arachis hypogaea L.) kernels’. Journal of Agricultural and Food Chemistry 58, 9204-9213.

8.2 Oral presentations

Phan-Thien K-Y (2011). ‘Breeding better quality peanuts: Benefits for industry, consumers, and food security’. Universitas 21 Graduate Research Conference on Food, 22-26 June 2011, Kuala Lumpur, Malaysia.

Phan-Thien K-Y, Fleischfresser D, Wright GC, Lee NA (2010). ‘Genotype variation and genotype- by-environment interaction affecting the antioxidant capacity of peanuts’. 42nd Annual General Meeting of the American Peanut Research and Education Society, 12- 15 July 2010, Clearwater Beach, Florida, USA.

8.3 Poster presentations

Phan-Thien K-Y, Wong HN, Fleischfresser D, Wright GC, Lee NA (2010). ‘HPLC-ORAC for profiling of antioxidants and antioxidant activity in raw and processed peanuts (Arachis hypogaea L.)’. The 2010 International Chemical Congress of Pacific Basin Societies (Pacifichem), 15-20 December 2010, Oahu, Hawaii, USA.

Phan-Thien K-Y, Arcot J, Lee NA (2009). ‘Comparing the antioxidant activity in kernels of Australian peanut accessions’. 11th Government Food Analysts’ Conference, 22-24.

| Publications to date 274

9 References

ABBO, S., GRUSAK, M., TZUK, T. & REIFEN, R. 2008. Genetic control of seed weight and calcium concentration in chickpea seed. Plant Breeding, 119, 427-431. AGUAMAH, G. E., LANGCAKE, P., LEWORTHY, D. P., PAGE, J. A., PRYCE, R. J. & STRANGE, R. N. 1981. Two novel stilbene phytoalexins from Arachis hypogaea. Phytochemistry, 20, 1381-1383. AGUILAR-GARCIA, C., GAVINO, G., BARAGAÑO-MOSQUEDA, M., HEVIA, P. & GAVINO, V. C. 2007. Correlation of tocopherol, tocotrienol, γ-oryzanol and total polyphenol content in rice bran with different antioxidant capacity assays. Food Chemistry, 102, 1228- 1232. AL-BABILI, S. & BEYER, P. 2005. Golden Rice - five years on the road - five years to go? Trends in Plant Science, 10, 565-573. ALARCÓN, E., CAMPOS, A. M., EDWARDS, A. M., LISSI, E. & C, L.-A. 2008. Antioxidant capacity of herbal infusions and tea extracts: a comparison of ORAC-fluorescein and ORAC- pyrogallol red methodologies. Food Chemistry, 107, 1114-1119. AMBUSHE, A. A., MCCRINDLE, R. I. & MCCRINDLE, C. M. E. 2009. Speciation of chromium in cow's milk by solid-phase extraction/dynamic reaction cell inductively coupled plasma mass spectrometry (DRC-ICP-MS). Journal of Analytical Atomic Spectrometry, 24, 502- 507. AMMANN, A. A. 2007. Inductively coupled plasma mass spectrometry (ICP MS): a versatile tool. Journal of Mass Spectrometry, 42, 419-427. ANDRÉ, C. M., OUFIR, M., HOFFMANN, L., HAUSMAN, J.-F., ROGEZ, H., LARONDELLE, Y. & EVERS, D. 2009. Influence of environment and genotype on polyphenol compounds and in vitro antioxidant capacity of native Andean potatoes (Solanum tuberosum L.). Journal of Food Composition and Analysis, 22, 517-524. APAK, R., GÜÇLÜ, K., ÖZYÜREK, M., KARADEMR, S. E. & ALTUN, M. 2005. Total antioxidant capacity assay of human serum using copper(II)-neocuproine as chromogenic agent: the CUPRAC method. Free Radical Research, 39, 949-961. ARAI, S., MORINAGA, Y., YOSHIKAWA, T., ICHIISHI, E., KISO, Y., YAMAZAKI, M., MOROTOMI, M., SHIMIZU, M., KUWATA, T. & KAMINOGAWA, S. 2002. Recent trends in functional food science and the industry in Japan. Bioscience, Biotechnology, and Biochemistry, 66, 2017-2029. ARNAO, M. B., CANO, A. & ACOSTA, M. 2001. The hydrophilic and lipophilic contribution to total antioxidant activity. Food Chemistry, 73, 239-244. ARORA, M. & STRANGE, R. 1991. Phytoalexin accumulation in groundnuts in response to wounding. Plant Science, 78, 157-163. ARTS, I. C. W. & HOLLMAN, P. C. H. 2005. Polyphenols and disease risk in epidemiologic studies. American Journal of Clinical Nutrition, 81. ARUOMA, O. I. 1999. Antioxidant actions of plant foods: use of oxidative DNA damage as a tool for studying antioxidant efficacy. Free Radical Research, 30, 419-427. ASIBUO, J. Y., AKROMAH, R., SAFO-KANTANKA, O., ADU-DAPAAH, H. K., OHEMENG-DAPAAH, S. & AGYEMAN, A. 2008. Chemical composition of groundnut, Arachis hypogaea (L) landraces. African Journal of Biotechnology, 7, 2203-2208. AZPILICUETA, C. E., ZAWOZNIK, M. S. & TOMARO, M. L. 2004. Phytoalexins synthesis is enhanced in groundnut plants inoculated with Bradyrhizobium sp. (Arachis). Crop Protection, 23, 1069-1074. AZQUETA, A., SHAPOSHNIKOV, S. & COLLINS, A. R. 2009. DNA oxidation: investigating its key role in environmental mutagenesis with the comet assay. Mutation Research, 674, 101-108.

| References 275

BAGHURST, K. I., DREOSTI, I. E., SYRETTE, J. A., RECORD, S. J., BAGHURST, P. A. & BUCKLEY, R. A. 1991. Zinc and magnesium status of Australian adults. Nutrition Research, 11, 23-32. BALIGAR, V. C., FAGERIA, N. K. & HE, Z. L. 2001. Nutrient use efficiency in plants. Communications in Soil Science and Plant Analysis, 32, 921-950. BALLARD, T. S., MALLIKARJUNAN, P., ZHOU, K. & O’KEEFE, S. F. 2009. Optimizing the extraction of phenolic antioxidants from peanut skins using response surface methodology. Journal of Agricultural and Food Chemistry, 57, 3064-3072. BALLUZ, L. S., KIESZAK, S. M., PHILEN, R. M. & MULINARE, J. 2000. Vitamin and mineral supplement use in the United States. Archives of Family Medicine, 9, 258-262. BATES, C. J. 2005. Selenium. In: CABALLERO, B., ALLEN, L. & PRENTICE, A. (eds.) Encyclopedia of Human Nutrition. Elsevier Ltd. BATTEN, G. D. 1998. Plant analysis using near infrared reflectance spectroscopy: the potential and the limitations. Australian Journal of Experimental Agriculture, 38, 697-706. BAUR, J. A. & SINCLAIR, D. A. 2006. The therapeutic potential of resveratrol: The in vivo evidence. Nature Reviews, 5, 493-506. BAXTER, I. 2009. Ionomics: studying the social network of mineral nutrients. Current Opinion in Plant Biology, 12, 381-386. BAXTER, I. 2010. Ionomics: the functional genomics of elements. Briefings in Functional Genomics, 9, 149-156. BECKSTRAND, R. L. & PICKENS, J. S. 2011. Beneficial effects of magnesium supplementation. Journal of Evidence-Based Complementary and Alternative Medicine, 16, 181-189. BELANE, A. K. & DAKORA, F. D. 2011. Levels of nutritionally-important trace elements and macronutrients in edible leaves and grain of 27 nodulated cowpea (Vigna unguiculata L. Walp.) genotypes grown in the Upper West Region of Ghana. Food Chemistry, 125, 99-105. BERINGER, H. & TEKETE, A. 1979. Influence of mineral nutrition on seed development of two groundnut cultivars. Communications in Soil Science and Plant Analysis, 10, 491-501. BERLETT, B. S. & STADTMAN, E. R. 1997. Protein oxidation in aging, disease, and oxidative stress. Journal of Biological Chemistry, 272, 20313-20316. BHUPATHIRAJU, S. N. & TUCKER, K. L. 2011. Coronary heart disease prevention: nutrients, foods, and dietary patterns. Clinica Chimica Acta, 412, 1493-1514. BISHAI, D. & NALUBOLA, R. 2002. The history of food fortification in the United States: its relevance for current fortifi cation efforts in developing countries. Economic Development and Cultural Change, 51, 37-53. BONDET, V., BRAND-WILLIAMS, W. & BERSET, C. 1997. Kinetics and mechanisms of antioxidant activity using the DPPH• free radical method. Food Science and Technology - Lebensmittel Wissenschaft und Technologie, 30, 609-615. BORKOWSKA-BURNECKA, J. 2000. Microwave assisted extraction for trace element analysis of plant materials by ICP-AES. Fresenius' Journal of Analytical Chemistry, 368, 633-637. BOUIS, H. E. 2003. Micronutrient fortification of plants through plant breeding: can it improve nutrition in man at low cost? Proceedings of the Nutrition Society, 62, 403-411. BOZAN, B., TOSUN, G. & ÖZCAN, D. 2008. Study of polyphenol content in the seeds of red grape (Vitis vinifera L.) varieties cultivated in Turkey and their antiradical activity. Food Chemistry, 109, 426-430. BRANCH, W. D. & GAINES, T. P. 1983. Seed mineral composition of diverse peanut germplasm. Peanut Science, 10, 5-8. BRAND-WILLIAMS, W., CUVELIER, M. E. & BERSET, C. 1995. Use of a free radical method to evaluate antioxidant activity. Food Science and Technology - Lebensmittel Wissenschaft und Technologie, 28, 25-30. BROADLEY, M. R., ALCOCK, J., ALFORD, J., CARTWRIGHT, P., FOOT, I., FAIRWEATHER-TAIT, S. J., HART, D. J., HURST, R., KNOTT, P., MCGRATH, S. P., MEACHAM, M. C., NORMAN, K.,

| References 276

MOWAT, H., SCOTT, P., STROUD, J. L., TOVEY, M., TUCKER, M., WHITE, P. J., YOUNG, S. D. & ZHAO, F.-J. 2009. Selenium biofortification of high-yielding winter wheat (Triticum aestivum L.) by liquid or granular Se fertilisation. Plant and Soil, DOI 10.1007/s11104- 009-0234-4. BRODRICK, S. J., SAKALA, M. K. & GILLER, K. E. 1992. Molybdenum reserves of seed, and growth and N2 fixation by Phaseolus vulgaris L. Biology and Fertility of Soils, 13, 39-44. BROWN, C. R., CULLEY, D., YANG, C.-P., DURST, R. & WROLSTAD, R. 2005. Variation of anthocyanin and carotenoid contents and associated antioxidant values in potato breeding lines. Journal of the American Society for Horticultural Science, 130, 174-180. BUIARELLI, F., COCCIOLI, F., JASIONOWSKA, R., MEROLLE, M. & TERRACCIANO, A. 2007. Analysis of some stilbenes in Italian wines by liquid chromatography/tandem mass spectrometry. Rapid Communications in Mass Spectrometry, 21, 2955-2964. BURNS, J., YOKOTA, T., ASHIHARA, H., LEAN, M. E. J. & CROZIER, A. 2002. Plant foods and herbal sources of resveratrol. Journal of Agricultural and Food Chemistry, 50. BURTON, G. W. & TRABER, M. G. 1990. Vitamin E: antioxidant activity, biokinetics, and bioavailability. Annual Review of Nutrition, 10, 357-382. CABALLERO, B., ALLEN, L. & PRENTICE, A. (eds.) 2005. Encyclopedia of human nutrition: Elsevier Ltd. CAMPO, R. J., ARAUJO, R. S. & HUNGRIA, M. 2009. Molybdenum-enriched soybean seeds enhance N accumulation, seed yield, and seed protein content in Brazil. Field Crops Research, 110, 219-224. CAO, G., ALESSIO, H. M. & CUTLER, R. G. 1993. Oxygen-radical absorbance capacity assay for antioxidants. Free Radical Biology and Medicine, 14, 303-311. CAPOCASA, F., DIAMANTI, J., TULIPANI, S., BATTINO, M. & MEZZETTI, B. 2008. Breeding strawberry (Fragaria X ananassa Duch) to increase fruit nutritional quality. BioFactors, 34, 67-72. CARLSEN, M. H., HALVORSEN, B. L., HOLTE, K., BØHN, S. K., DRAGLAND, S., SAMPSON, L., WILLEY, C., SENOO, H., UMEZONO, Y., SANADA, C., BARIKMO, I., BERHE, N., WILLETT, W. C., PHILLIPS, K. M., JACOBS, D. R., JR., & BLOMHOFF, R. 2010. The total antioxidant content of more than 3100 foods, beverages, spices, herbs and supplements used worldwide. Nutrition [Online], 9. CASAZZA, A. A., ALIAKBARIAN, B., MANTEGNA, S., CRAVOTTO, G. & PEREGO, P. 2010. Extraction of phenolics from Vitis vinifera wastes using non-conventional techniques. Journal of Food Engineering, 100, 50-55. CASSIDY, A., HANLEY, B. & LAMUELA-RAVENTOS, R. M. 2000. Isoflavones, lignans and stilbenes – origins, metabolism and potential importance to human health. Journal of the Science of Food and Agriculture, 80, 1044-1062. CASTRO, L. & FREEMAN, B. A. 2001. Reactive oxygen species in human health and disease. Nutrition, 17, 161-165 CEMELI, E., BAUMGARTNER, A. & ANDERSON, D. 2009. Antioxidants and the Comet assay. Mutation Research, 681, 51-67. CHANDRASEKARA, A. & SHAHIDI, F. 2011. Inhibitory activities of soluble and bound millet seed phenolics on free radicals and reactive oxygen species. Journal of Agricultural and Food Chemistry, 59, 428-436. CHANG, J.-C., LAI, Y.-H., DJOKO, B., WU, P.-L., LIU, C.-D., LIU, Y.-W. & CHIOU, R. Y.-Y. 2006. Biosynthesis enhancement and antioxidant and anti-inflammatory activities of peanut (Arachis hypogaea L.) arachidin-1, arachidin-3, and isopentadienylresveratrol. Journal of Agricultural and Food Chemistry, 54, 10281-10287. CHAVAN, P. D. & KARADGE, B. A. 1980. Influence of salinity on mineral nutrition of peanut (Arachis hypogea). Plant and Soil, 54, 5-13.

| References 277

CHELI, F. & BALDI, A. 2011. Nutrition-based health: cell-based bioassays for food antioxidant activity evaluation. Journal of Food Science, 76, R197-R205. CHEN, R.-S., WU, P.-L. & CHIOU, R. Y.-Y. 2002. Peanut roots as a source of resveratrol. Journal of Agricultural and Food Chemistry, 50, 1665-1667. CHENG, J.-C., DAI, F., ZHOU, B., YANG, L. & LIU, Z.-L. 2007. Antioxidant activity of hydroxycinnamic acid derivatives in human low density lipoprotein: mechanism and structure-activity relationship. Food Chemistry, 104, 132-139. CHILDS, N. M. & PORYZEES, G. H. 1997. Foods that help prevent disease: consumer attitudes and public policy implications. Journal of Consumer Marketing, 14, 433-447. CHRISTODOULEAS, D., FOTAKIS, C., PAPADOPOULOS, K., DIMOTIKALI, D. & CALOKERINOS, A. C. 2012. Luminescent methods in the analysis of untreated edible oils: a review. Analytical Letters, 45, 625-641. CHUKWUMAH, Y., WALKER, L., VOGLER, B. & VERGHESE, M. 2007a. Changes in the phytochemical composition and profile of raw, boiled, and roasted peanuts. Journal of Agricultural and Food Chemistry, 55, 9266-9273. CHUKWUMAH, Y., WALKER, L. T., VERGHESE, M. & OGUTU, S. 2008. Effect of frequency and duration of ultrasonication on the extraction efficiency of selected isoflavones and trans-resveratrol from peanuts (Arachis hypogaea). Ultrasonics Sonochemistry, 16, 293-299. CHUKWUMAH, Y. C., WALKER, L. T., VERGHESE, M., BOKANGA, M., OGUTU, S. & ALPHONSE, K. 2007b. Comparison of extraction methods for the quantification of selected phytochemicals in peanuts (Arachis hypogaea). Journal of Agricultural and Food Chemistry, 55, 285-290. CHUN, J., LEE, J. & EITENMILLER, R. R. 2005. Vitamin E and oxidative stability during storage of raw and dry roasted peanuts packaged under air and vacuum. Journal of Food Science, 70, C292-C297. CHUNG, H., JI, X., CANNING, C., SUN, S. & ZHOU, K. 2010. Comparison of different strategies for soybean antioxidant extraction. Journal of Agricultural and Food Chemistry, 58, 4508- 4512. CHUNG, I.-M., PARK, M. R., CHUN, J. C. & YUN, S. J. 2003. Resveratrol accumulation and resveratrol synthase gene expression in response to abiotic stresses and hormones in peanut plants. Plant Science, 164, 103-109. CHUNG, I.-M., PARK, M. R., REHMAN, S. & YUN, S. J. 2001. Tissue specific and inducible expression of resveratrol synthase gene in peanut plants. Molecules and Cells, 12, 353- 359. CONDORI, J., SIVAKUMAR, G., HUBSTENBERGER, J., DOLAN, M. C., SOBOLEV, V. S. & MEDINA- BOLIVAR, F. 2010. Induced biosynthesis of resveratrol and the prenylated stilbenoids arachidin-1 and arachidin-3 in hairy root cultures of peanut: effects of culture medium and growth stage. Plant Physiology and Biochemistry, 48, 310-318. CONKERTON, E. J., ROSS, L. F., DAIGLE, D. J., KVIEN, C. S. & MCCOMBS, C. 1989. The effect of drought stress on peanut seed composition. II. Oil, protein and minerals. Oléagineux, 44, 593-602. COOKSEY, C. J., GARRATT, P. J., RICHARDS, S. E. & STRANGE, R. N. 1988. A dienyl stilbene phytoalexin from Arachis hypogaea. Phytochemistry, 27, 1015-1016. COOPER, M. & DE LACY, I. H. 1994. Relationships among analytical methods used to study genotypic variation and genotype-by-environment interaction in plant breeding multi- environment experiments. Theoretical and Applied Genetics, 88, 561-572. COZZOLINO, D. & MORON, A. 2004. Exploring the use of near infrared reflectance spectroscopy (NIRS) to predict trace minerals in legumes. Animal Feed Science and Technology, 111, 161-173.

| References 278

CUBADDA, F., AURELI, F., CIARDULLO, S., D’AMATO, M., RAGGI, A., ACHARYA, R., REDDY, R. A. V. & PRAKASH, N. T. 2010. Changes in selenium speciation associated with increasing tissue concentrations of selenium in wheat grain. Journal of Agricultural and Food Chemistry, 58, 2295-2301. CUBADDA, F., RAGGI, A., TESTONI, A. & , Z., FABIO 2002. Multielemental analysis of food and agricultural matrixes by inductively coupled plasma-mass spectrometry. Journal of AOAC International, 85, 113-121. DABROWSKI, K. J. & SOSULSKI, F. W. 1984. Composition of free and hydrolyzable phenolic acids in defatted flours of ten oilseeds. Journal of Agricultural and Food Chemistry, 32, 128- 130. DAIGLE, D. J., CONKERTON, E. J., HAMMONS, R. O. & BRANCH, W. D. 1983. A preliminary classification of selected white testa peanuts (Arachis hypogaea L.) by flavonoid analysis. Peanut Science, 10, 40-43. DAIGLE, D. J., ORY, R. L. & BRANCH, W. D. 1985. The presence of 5,7-dihydroxyisoflavone in peanuts by high performance liquid chromatography analyses. Peanut Science, 12, 60- 62. DAKORA, F. D. & PHILLIPS, D. A. 2002. Root exudates as mediators of mineral acquisition in low-nutrient environments. Plant and Soil, 245, 35-47. DARNTON-HILL, I. & NALUBOLA, R. 2002. Fortification strategies to meet micronutrient needs: successes and failures. Proceedings of the Nutrition Society, 61, 231-241. DÁVALOS, A., GÓMEZ-CORDOVÉS, C. & BARTOLOMÉ, B. 2004. Extending applicability of the oxygen radical absorbance capacity (ORAC-fluorescein) assay. Journal of Agricultural and Food Chemistry, 52, 48-54. DAVIS, J. P., DEAN, L. L., PRICE, K. M. & SANDERS, T. H. 2010. Roast effects on the hydrophilic and lipophilic antioxidant capacities of peanut flours, blanched peanut seed and peanut skins. Food Chemistry, 119, 539-547. DE BONT, R. & VAN LAREBEKE, N. 2004. Endogenous DNA damage in humans: a review of quantitative data. Mutagenesis, 19, 169-185. DE DORLODOT, S., FORSTER, B., PAGÈS, L., PRICE, A., TUBEROSA, R. & DRAYE, X. 2007. Root system architecture: opportunities and constraints for genetic improvement of crops. Trends in Plant Science, 12, 474-481. DE LA LASTRA, C. A. & VILLEGAS, I. 2005. Resveratrol as an anti-inflammatory and anti-aging agent: Mechanisms and clinical implications. Molecular Nutrition and Food Research, 49, 405-430. DEAN, L. L., DAVIS, J. P., SHOFRAN, B. G. & SANDERS, T. H. 2008. Phenolic profiles and antioxidant activity of extracts from peanut plant parts. Open Natural Products Journal, 1, 1-6. DEAN, L. L., HENDRIX, K. W., HOLBROOK, C. C. & SANDERS, T. H. 2009. Content of some nutrients in the core of the core of the peanut germplasm collection. Peanut Science, 36, 104-120. DEAN, R. T., FU, S., STOCKER, R. & DAVIES, M. J. 1997. Biochemistry and pathology of radical- mediated protein oxidation. Biochemistry, 324, 1-18. DOLAN, S. P. & CAPAR, S. G. 2002. Multi-element analysis of food by microwave digestion and inductively coupled plasma-atomic emission spectrometry. Journal of Food Composition and Analysis, 15, 593-615. DOMÍNGUEZ, H., NÛÑEZ, M. J. & LEMA, J. M. 1994. Enzymatic pretreatment to enhance oil extraction from fruits and oilseeds: a review. Food Chemistry, 49, 271-286. DUDONNÉ, S., VITRAC, X., COUTIÈRE, P., WOILLEZ, M. & MÉRILLON, J.-M. 2009. Comparative study of antioxidant properties and total phenolic content of 30 plant extracts of industrial interest using DPPH, ABTS, FRAP, SOD, and ORAC assays. Journal of Agricultural and Food Chemistry, 57, 1768-1774.

| References 279

DUNCAN, C. E., GORBET, D. W. & TALCOTT, S. T. 2006. Phytochemical content and antioxidant capacity of water-soluble isolates from peanuts (Arachis hypogaea L.). Food Research International, 39, 898-904. EBERHARDT, M. V., KOBIRA, K., KECK, A.-S., JUVIK, J. A. & JEFFERY, E. H. 2005. Correlation analyses of phytochemical composition, chemical, and cellular measures of antioxidant activity of broccoli (Brassica oleracea L. var. italica). Journal of Agricultural and Food Chemistry, 53, 7421-7431. EMERIT, J., EDEAS, M. & BRICAIRE, F. 2004. Neurodegenerative diseases and oxidative stress. Biomedicine and Pharmacotherapy, 58, 39-46. ERLUND, I. 2004. Review of the flavonoids quercetin, hesperetin, and naringenin. Dietary sources, bioactivities, bioavailability, and epidemiology. Nutrition Research, 24, 851- 874. EVERETTE, J. D., BRYANT, Q. M., GREEN, A. M., ABBEY, Y. A., WANGILA, G. W. & WALKER, R. B. 2010. Thorough study of reactivity of various compound classes toward the Folin- Ciocalteu reagent. Journal of Agricultural and Food Chemistry, 58, 8139-8144. FAGERIA, N. K. 1974. Absorption of magnesium and its influence on the uptake of phosphorus, potassium, and calcium by intact groundnut plants. Plant and Soil, 40, 313-320. FAIRBAIRN, D. W., OLIVE, P. L. & O'NEILL, K. L. 1995. The comet assay: A comprehensive review. Mutation Research, 339, 37-59. FAO 2010. The state of food insecurity in the world: addressing food insecurity in protracted crises. Rome, Italy: Food and Agriculture Organization of the United Nations. FICCO, D. B. M., RIEFOLO, C., NICASTRO, G., DE SIMONE, V., DI GESU, A. M., BELEGGIA, R., PLATANI, C., CATTIVELLI, L. & DE VIT, P. 2009. Phytate and mineral elements concentration in a collection of Italian durum wheat cultivars. Field Crops Research, 111, 235-242. FOX, G. & CRUICKSHANK, A. 2005. Near infrared reflectance as a rapid and inexpensive surrogate measure for fatty acid composition and oil content of peanuts. Journal of Near Infrared Spectroscopy, 13, 287-291. FRANCISCO, M. L. D. L. & RESURRECCION, A. V. A. 2009a. Development of a reversed-phase high performance liquid chromatography (RP-HPLC) procedure for the simultaneous determination of phenolic compounds in peanut skin extracts. Food Chemistry, 117, 356-363. FRANCISCO, M. L. L. D. & RESURRECCION, A. V. A. 2009b. Total phenolics and antioxidant capacity of heat-treated peanut skins. Journal of Food Composition and Analysis, 22, 16-24. FREAKE, H. C. 2005. Zn - physiology. In: CABALLERO, B., ALLEN, L. & PRENTICE, A. (eds.) Encyclopedia of Human Nutrition. Elsevier Ltd. FRÉMONT, L. 2000. Biological effects of resveratrol. Life Sciences, 66, 663-673. FSANZ 2009. Standard 2.1.1 Cereals and cereal products. Canberra, ACT, Australia: Food Standards Australia and New Zealand. FSANZ. 2010. Mandatory fortification [Online]. Food Standards Australia New Zealand. Available: http://www.foodstandards.gov.au/consumerinformation/fortification/ [Accessed 3 May 2011]. GAHOONIA, T. S., ALI, R., MALHOTRA, R. S., JAHOOR, A. & RAHMAN, M. M. 2007. Variation in root morphological and physiological traits and nutrient uptake of chickpea genotypes. Journal of Plant Nutrition, 30, 829-841. GAHOONIA, T. S. & NIELSEN, N. E. 2004. Root traits as tools for creating phosphorus efficient crop varieties. Plant and Soil, 260, 47-57. GARRETT, A. R., MURRAY, B. K., ROBISON, R. A. & O’NEILL, K. L. 2010. Measuring antioxidant capacity using the ORAC and TOSC assays. In: ARMSTRONG, D. (ed.) Advanced Protocols in Oxidative Stress II. Humana Press.

| References 280

GAUCH, H. G., JR. 2006. Statistical analysis of yield trials by AMMI and GGE. Crop Science, 46, 1488-1500. GCGH. 2011. Challenge #9: create a full range of optimal, bioavailable nutrients in a single staple plant species [Online]. Grand Challenges in Global Health. Available: http://www.grandchallenges.org/ImproveNutrition/Challenges/NutrientRichPlants/Pa ges/default.aspx [Accessed June 2011]. GEORGE, B., KAUR, C., KHURDIYA, D. S. & KAPOOR, H. C. 2004. Antioxidants in tomato (Lycopersium esculentum) as a function of genotype. Food Chemistry, 84, 45-51. GIBSON, R. S. & HOTZ, C. 2001. Dietary diversification/modification strategies to enhance micronutrient content and bioavailability of diets in developing countries. British Journal of Nutrition, 85, S159-S166. GIBSON, R. S., YEUDALL, F., DROST, N., MTITIMUNI, B. M. & CULLINAN, T. R. 2003. Experiences of a community-based dietary intervention to enhance micronutrient adequacy of diets low in animal source foods and high in phytate: a case study in rural Malawian children. Journal of Nutrition, 133, 39925-39995. GIUSTARINI, D., DALLE-DONNE, I., TSIKAS, D. & ROSSI, R. 2009. Oxidative stress and human diseases: origin, link, measurement, mechanisms, and biomarkers. Critical Reviews in Clinical Laboratory Science, 46, 241-281. GOLIC, M., WALSH, K. & LAWSON, P. 2003. Short-wavelength near-infrared spectra of sucrose, glucose, and fructose with respect to sugar concentration and temperature. Applied Spectroscopy, 57, 139-145. GOLIC, M. & WALSH, K. B. 2006. Robustness of calibration models based on near infrared spectroscopy for the in-line grading of stonefruit for total soluble solids content. Analytica Chimica Acta, 555, 286-291. GÓMEZ-GALERA, S., ROJAS, E., SUDHAKAR, D., ZHU, C., PELACHO, A. M., CAPELL, T. & CHRISTOU, P. 2010. Critical evaluation of strategies for mineral fortification of staple food crops. Transgenic Research, 19, 165-180. GONZÁLEZ-MARTÍN, I., ÁLVAREZ-GARCÍA, N., GONZÁLEZ-PÉREZ, C. & VILLAESCUSA-GARCÍA, V. 2006. Determination of inorganic elements in animal feeds by NIRS technology and a fibre-optic probe. Talanta, 69, 711-715. GOVINDARAJU, K. & SRINIVAS, H. 2006. Studies on the effects of enzymatic hydrolysis on functional and physico-chemical properties of arachin. Lebensmittel-Wissenschaft und Technologie, 39, 54-62. GRAHAM, R. D. & STANGOULIS, J. C. R. 2003. Trace element uptake and distribution in plants. Journal of Nutrition, 133, 1502S-1505S. GÜLÇIN, İ. 2012. Antioxidant activity of food constituents: an overview. Archives of Toxicology, 86, 345-391. GUO, W., HU, S., ZHAO, J., JIN, S., LIU, W. & ZHANG, H. 2011a. Removal of spectral interferences and accuracy monitoring of trace cadmium in feeds by dynamic reaction cell inductively coupled plasma mass spectrometry. Microchemical Journal, 97, 154- 159. GUO, X.-D., MA, Y.-J., PARRY, J., GAO, J.-M., YU, L.-L. & WANG, M. 2011b. Phenolics content and antioxidant activity of tartary buckwheat from different locations. Molecules, 16, 9850-9867. GUO, X., WILLOWS, N., KUHLE, S., JHANGRI, G. & VEUGELERS, P. J. 2009. Use of vitamin and mineral supplements among Canadian adults. Canadian Journal of Public Health, 100, 357-360. GUTIÉRREZ-URIBE, J. A., ROMO-LOPEZ, I. & SERNA-SALDÍVAR, S. O. 2011. Phenolic composition and mammary cancer cell inhibition of extracts of whole cowpeas (Vigna unguiculata) and its anatomical parts. Journal of Functional Foods, 3, 290-297.

| References 281

HA, H. V., POKORNY, J. & SAKURAI, H. 2007. Peanut skin antioxidants. Journal of Food Lipids, 14, 298-314. HAASE, H., OVERBECK, S. & RINK, L. 2008. Zinc supplementation for the treatment or prevention of disease: current status and future perspectives. Experimental Gerontology, 43, 394-408. HARRIS, E. D. 2005. Cofactors - inorganic. In: CABALLERO, B., ALLEN, L. & PRENTICE, A. (eds.) Encyclopedia of Human Nutrition. 2nd edition ed.: Elsevier Ltd. HARVESTPLUS. 2009. HarvestPlus [Online]. CGIAR. Available: http://www.harvestplus.org/ [Accessed June 2011]. HASHIM, I. B., KOEHLER, P. E. & EITENMILLER, R. R. 1993a. Tocopherols in Runner and Virginia peanut cultivars at various maturity stages. Journal of the American Oil Chemists' Society, 70, 633-635. HASHIM, I. B., KOEHLER, P. E., EITENMILLER, R. R. & KVIEN, C. K. 1993b. Fatty acid composition and tocopherol content of drought stressed Florunner peanuts. Peanut Science, 20, 21- 24. HERNANDEZ, J. A., GEORGE, S. J. & RUBIO, L. M. 2009. Molybdenum trafficking for nitrogen fixation. Biochemistry, 48, 9711-9721. HIBBERT, D. B. 1999. Method validation of modern analytical techniques. Accreditation and Quality Assurance, 4, 352-356. HODGE, A. 2009. Root decisions. Plant, Cell and Environment, 32, 628-640. HOFFLAND, E., WEI, C. & WISSUWA, M. 2006. Organic anion exudation by lowland rice (Oryza sativa L.) at zinc and phosphorus deficiency. Plant and Soil, 283, 155-162. HØGH-JENSEN, H. & PEDERSEN, M. B. 2003. Morphological plasticity by crop plants and their potassium use efficiency. Journal of Plant Nutrition, 26, 969-984. HOLME, A. L. & PERVAIZ, S. 2007. Resveratrol in cell fate decisions. Journal of Bioenergetics and Biomembranes, 39, 59-36. HOUSTON, M. 2011. The role of magnesium in hypertension and cardiovascular disease. Journal of Clinical Hypertension, 13, 843-847. HOWARD, L. R., CLARK, J. R. & BROWNMILLER, C. 2003. Antioxidant capacity and phenolic content in blueberries as affected by genotype and growing season. Journal of the Science of Food and Agriculture, 83, 1238-1247. HUANG, D., OU, B., HAMPSCH-WOODILL, M., FLANAGAN, J. A. & DEEMER, E. K. 2002a. Development and validation of oxygen radical absorbance capacity assay for lipophilic antioxidants using randomly methylated β-cyclodextrin as the solubility enhancer. Journal of Agricultural and Food Chemistry, 50, 1815-1821. HUANG, D., OU, B., HAMPSCH-WOODILL, M., FLANAGAN, J. A. & PRIOR, R. L. 2002b. High- throughput assay of oxygen radical absorbance capacity (ORAC) using a multichannel liquid handling system coupled with a microplate fluorescence reader in 96-well format. Journal of Agricultural and Food Chemistry, 50, 4437-4444. HUANG, D., OU, B. & PRIOR, R. L. 2005. The chemistry behind antioxidant capacity assays. Journal of Agricultural and Food Chemistry, 53, 1841-1856. HWANG, J.-Y., SHUE, Y.-S. & CHANG, H.-M. 2001. Antioxidative activity of roasted and defatted peanut kernels. Food Research International, 34, 639-647. IBERN-GOMÉZ, M., ROIG-PÉREZ, S., LAMUELA-RAVENTÓS, R. M. & DE LA TORRE-BORONAT, M. C. 2000. Resveratrol and piceid levels in natural and blended peanut butters. Journal of Agricultural and Food Chemistry, 48, 6352-6354. IENCO, E. C., LOGERFO, A., CARLESI, C., ORSUCCI, D., RICCI, G., MANCUSO, M. & SICILIANO, G. 2011. Oxidative stress treatment for clinical trials in neurodegenerative diseases. Journal of Alzheimer's Disease, 24, 111-126. INGHAM, J. L. 1976. 3,5,4'-trihydroxystilbene as a phytoalexin from groundnuts (Arachis hypogaea). Phytochemistry, 15, 1791-1793.

| References 282

ISABELLE, M., LEE, B. L., ONG, C. N., LIU, X. & HUANG, D. 2008. Peroxyl radical scavenging capacity, polyphenolics, and lipophilic antioxidant profiles of mulberry fruits cultivated in southern China. Journal of Agricultural and Food Chemistry, 56, 9410-9416. ISANGA, J. & ZHANG, G.-N. 2007. Biologically active components and nutraceuticals in peanuts and related products: review. Food Reviews International, 23, 123-140. JIANG, S., WU, J., NGUYEN, B. T., FENG, Y., YANG, X. & SHI, C. 2008. Genotypic variation of mineral elements contents in rice (Oryza sativa L.). European Food Research and Technology, 228, 115-122. JONNALA, R. S., DUNFORD, N. T. & CHENAULT, K. 2005a. Nutritional composition of genetically modified peanut varieties. Journal of Food Science, 70, S254-S256. JONNALA, R. S., DUNFORD, N. T. & DASHIELL, K. E. 2005b. New high-oleic peanut cultivars grown in the Southwestern United States. Journal of the American Oil Chemists' Society, 82, 125-128. JORDAN, D. L., JOHNSON, P. D., SPEARS, J. F., PENNY, B. & HARDY, D. 2010. Response of Virginia market type peanut to interactions of cultivar, calcium, and potassium. Crop Management, February. JUNGK, A. 2001. Root hairs and the acquisition of plant nutrients from soil. Journal of Plant Nutrition and Soil Science, 164, 121-129. JÚNIOR, D. S., KRUG, F. J., DE GODOI PEREIRA, M. & KORN, M. 2006. Currents on ultrasound- assisted extraction for sample preparation and spectroscopic analytes determination. Applied Spectroscopy Reviews, 41, 305-321. KALANTAR-ZADEH, K., LEE, G. H. & BLOCK, G. 2004. Relationship between dietary antioxidants and childhood asthma: more epidemiological studies are needed. Medical Hypotheses, 62, 280-290. KARADAG, A., OZCELIK, B. & SANER, S. 2009. Review of methods to determine antioxidant capacities. Food Analytical Methods, 2, 41-60. KARLEY, A. J. & WHITE, P. J. 2009. Moving cationic minerals to edible tissues: potassium, magnesium, calcium. Current Opinion in Plant Biology, 12, 291-298. KEDARE, S. B. & SINGH, R. P. 2011. Genesis and development of DPPH method of antioxidant assay. Journal of Food Science and Technology, 48, 412-422. KEEN, N. T. & INGHAM, J. L. 1976. New stilbene phytoalexins from American cultivars of Arachis hypogaea. Phytochemistry, 15, 1794-1795. KHOJAH, E. Y. 2010. Development of a GC-MS method and analysis of phenolic acids in fruit and spice matrices and evaluation of their antioxidant capacity. Doctor of Philosophy, University of New South Wales. KIM, J. A., JUNG, W. S., CHUN, S. C., YU, C. Y., MA, K. H., GWAG, J. G. & CHUNG, I. M. 2006a. A correlation between the level of phenolic compounds and the antioxidant capacity in cooked-with-rice and vegetable soybean (Glycine max L.) varieties. European Food Research and Technology, 224, 259-270. KIM, K.-H., TSAO, R., YANG, R. & CUI, S. W. 2006b. Phenolic acid profiles and antioxidant activities of wheat bran extracts and the effect of hydrolysis conditions. Food Chemistry, 95, 466-473. KING, R. E., BOMSER, J. A. & MIN, D. B. 2006. Bioactivity of resveratrol. Comprehensive Reviews in Food Science and Food Safety, 5, 65-70. KNEKT, P., KUMPULAINEN, J., JÄRVINEN, R., RISSANEN, H., HELIÖVAARA, M., REUNANEN, A., HAKULINEN, T. & AROMAA, A. 2002. Flavonoid intake and risk of chronic diseases. American Journal of Clinical Nutrition, 76, 560-568. KOCHIAN, L. V. 1995. Cellular mechanisms of aluminium toxicity and resistance in plants. Annual Review of Plant Physiology and Plant Molecular Biology, 46, 237-260.

| References 283

KOCHIAN, L. V., PIÑEROS, M. A. & HOEKENGA, O. A. 2005. The physiology, genetics and molecular biology of plant aluminum resistance and toxicity. Plant and Soil, 274, 175- 195. KOREKAR, G., STOBDAN, T., ARORA, R., YADAV, A. & SINGH, S. B. 2011. Antioxidant capacity and phenolics content of apricot (Prunus armeniaca L.) kernel as a function of genotype. Plant Foods for Human Nutrition, 66, 376-383. KOZLOWSKRA, H., ZADERNOWSKI, R. & SOSULSKI, F. W. 1983. Phenolic acids in oilseed flours. Nahrung, 27, 449-453. KUDRIN, A. V. 2000. Trace elements in regulation of NF-κB activity. Journal of Trace Elements in Medicine and Biology, 14, 129-142. KUHNLE, G. G. C., DELL’AQUILA, C., ASPINALL, S. M., RUNSWICK, S. A., MULLIGAN, A. A. & BINGHAM, S. A. 2008. Phytoestrogen content of beverages, nuts, seeds, and oils. Journal of Agricultural and Food Chemistry, 56, 7311-7315. KUHNLEIN, H. V. & RECEVEUR, O. 1996. Dietary change and traditional food systems of indigenous peoples. Annual Review of Nutrition, 16, 417-442. KUMAR, V., RANI, A., DIXIT, A. K., PRATAP, D. & BHATNAGAR, D. 2010. A comparative assessment of total phenolic content, ferric reducing-anti-oxidative power, free radical-scavenging activity, vitamin C and isoflavones content in soybean with varying seed coat colour. Food Research International, 43, 323-328. KURILICH, A. C., JEFFERY, E. H., JUVIK, J. A., WALLIG, M. A. & KLEIN, B. P. 2002. Antioxidant capacity of different broccoli (Brassica oleracea) genotypes using the oxygen radical absorbance capacity (ORAC) Assay. Journal of Agricultural and Food Chemistry, 50, 5053-5057. LAMUELA-RAVENTÓS, R. M., ROMERO-PÉREZ, A. I., WATERHOUSE, A. L. & DE LA TORRE- BORONAT, M. C. 1995. Direct HPLC analysis of cis- and trans-resveratrol and piceid isomers in Spanish red Vitis vinifera wines. Journal of Agricultural and Food Chemistry, 43, 281-283. LEE, S. S., LEE, S. M., KIM, M., CHUN, J., CHEONG, Y. K. & LEE, J. 2004. Analysis of trans- resveratrol in peanuts and peanut butters consumed in Korea. Food Research International, 37, 247-251. LI, B. B., SMITH, B. & HOSSAIN, M. M. 2006. Extraction of phenolics from citrus peels II. Enzyme-assisted extraction method. Separation and Purification Technology, 48, 189- 196. LI, L., TANG, C., RENGEL, Z. & ZHANG, F. S. 2004. Calcium, magnesium and microelement uptake as affected by phosphorus sources and interspecific root interactions between wheat and chickpea. Plant and Soil, 261, 29-37. LIGGINS, J., BLUCK, L. J. C., RUNSWICK, S., ATKINSON, C., COWARD, W. A. & BINGHAM, S. A. 2000. Daidzein and genistein content of fruits and nuts. Journal of Nutritional Biochemistry, 11, 326-331. LIMÓN-PACHECO, J. & GONSEBATT, M. E. 2009. The role of antioxidants and antioxidant- related enzymes in protective responses to environmentally induced oxidative stress. Mutation Research, 674, 137-147. LIU, C. D., WEN, Y.-Y., CHIOU, J.-M., WANG, K.-H. & CHIOU, R. Y.-Y. 2003. Comparative characterization of peanuts grown by aquatic floating cultivation and field cultivation for seed and resveratrol production. Journal of Agricultural and Food Chemistry, 51, 1582-1585. LUCAS, A., ANDUEZA, D., ROCK, E. & MARTIN, B. 2008. Prediction of dry matter, fat, pH, vitamins, minerals, carotenoids, total antioxidant capacity, and color in fresh and freeze-dried cheeses by visible-near-infrared reflectance spectroscopy. Journal of Agricultural and Food Chemistry, 56, 6801-6808.

| References 284

LYONS, G., ORTIZ-MONASTERIO, I., STANGOULIS, J. & GRAHAM, R. 2005. Selenium concentration in wheat grain: Is there sufficient genotypic variation to use in breeding? Plant and Soil, 269, 369-380. MACDONALD-WICKS, L. K., WOOD, L. G. & GARG, M. L. 2006. Methodology for the determination of biological antioxidant capacity in vitro: a review. Journal of the Science of Food and Agriculture, 86, 2046-2056. MACLENNAN, A. H., MYERS, S. P. & TAYLOR, A. W. 2006. The continuing use of complementary and alternative medicine in South Australia: costs and beliefs in 2004. Medical Journal of Australia, 184, 27-31. MANACH, C., SCALBERT, A., MORAND, C., RÉMÉSY, C. & JIMÉNEZ, L. 2004. Polyphenols: food sources and bioavailability. American Journal of Clinical Nutrition, 79, 727-747. MANACH, C., WILLIAMSON, G., MORAND, C., SCALBERT, A. & RÉMÉSY, C. 2005. Bioavailability and bioefficacy of polyphenols in humans. I. Review of 97 bioavailability studies. American Journal of Clinical Nutrition, 81, 230S-242S. MANDAL, S., MITRA, A. & MALLICK, N. 2009. Time course study on accumulation of cell wall- bound phenolics and activities of defense enzymes in tomato roots in relation to Fusarium wilt. World Journal of Microbiology and Biotechnology, 25, 795-802. MANNAR, M. G. V. 2006. Successful food-based programmes, supplementation and fortification. Journal of Pediatric Gastroenterology and Nutrition, 43, S47-S53. MARET, W. & SANDSTEAD, H. H. 2006. Zinc requirements and the risks and benefits of zinc supplementation. Journal of Trace Elements in Medicine and Biology, 20, 3-18. MARKESBERY, W. R. 1997. Oxidative stress hypothesis in Alzheimer's disease. Free Radical Biology and Medicine, 23, 134-147. MATTILA, P. & HELLSTRÖM, J. 2007. Phenolic acids in potatoes, vegetables, and some of their products. Journal of Food Composition and Analysis, 20, 152-160. MATTILA, P. & KUMPULAINEN, J. 2002. Determination of free and total phenolic acids in plant- derived foods by HPLC with diode-array detection. Journal of Agricultural and Food Chemistry, 50, 3660-3667. MAYER, J. E., PFEIFFER, W. H. & BEYER, P. 2008. Biofortified crops to alleviate micronutrient malnutrition. Current Opinion in Plant Biology, 11, 166-170. MAZUR, W. & ADLERCREUTZ, H. 1998. Naturally occurring oestrogens in food. Pure and Applied Chemistry, 70, 1759-1776. MAZUR, W. M., DUKE, J. A., WÄHÄLÄ, K., RASKU, S. & ADLERCREUTZ, H. 1998. Isoflavonoids and lignans in legumes: nutritional and health aspects in humans Journal of Nutritional Biochemistry, 9, 193-200. MCKELVEY-MARTIN, V. J., GREEN, M. H. L., SCHMEZER, P., POOL-ZOBEL, B. L., DE MÉO, M. P. & COLLINS, A. 1993. The single cell gel electrophoresis assay (comet assay): a European review. Mutation Research: Fundamental and Molecular Mechanisms of Mutagenesis, 288, 47-63. MEDINA-BOLIVAR, F., CONDORI, J., RIMANDO, A. M., HUBSTENBERGER, J., SHELTON, K., O’KEEFE, S. F., BENNETT, S. & DOLAN, M. C. 2007. Production and secretion of resveratrol in hairy root cultures of peanut. Phytochemistry, 68, 1992-2003. MEENAKSHI, J. V., JOHNSON, N. L., MANYONG, V. M., DEGROOTE, H., JAVELOSA, J., YANGGEN, D. R., NAHER, F., GONZALEZ, C., GARCÍA, J. & MENG, E. 2010. How cost-effective is biofortification in combating micronutrient malnutrition? An ex ante assessment. World Development, 38, 64-75. MEHBOOB, I., NAVEED, M. & ZAHIR, Z. A. 2009. Rhizobial association with non-legumes: mechanisms and applications. Critical Reviews in Plant Sciences, 28, 432-456. MICRONUTRIENT INITIATIVE 2009. A united call to action on vitamin and mineral deficiencies. Ottawa, Ontario, Canada: Canadian International Development Agency.

| References 285

MILDER, I. E. J., ARTS, I. C. W., VAN DE PUTTE, B., VENEMA, D. P. & HOLLMAN, P. C. H. 2005. Lignan contents of Dutch plant foods: a database including lariciresinol, pinoresinol, secoisolariciresinol and matairesinol. British Journal of Nutrition, 93, 393-402. MILNER, J. A. 2000. Functional foods: the US perspective. American Journal of Clinical Nutrition, 71, 1654S-1659S. MIN, B., GU, L., MCCLUNG, A. M., BERGMAN, C. J. & CHEN, M.-H. 2012. Free and bound total phenolic concentrations, antioxidant capacities, and profiles of proanthocyanidins and anthocyanins in whole grain rice (Oryza sativa L.) of different bran colours. Food Chemistry, 133, 715-722. MISHRA, K., OJHA, H. & CHAUDHURY, N. K. 2012. Estimation of antiradical properties of antioxidants using DPPH· assay: a critical review and results. Food Chemistry, 130, 1036-1043. MISRA, J. B., MATHUR, R. S. & BHATT, D. M. 2000. Near-infrared transmittance spectroscopy: a potential tool for non-destructive determination of oil content in groundnuts. Journal of the Science of Food and Agriculture, 80, 237-240. MOHANTY, B., BASHA, S. M., GORBET, D. W., COLE, R. J. & DORNER, J. W. 1991. Variation in phytoalexin production by peanut seed from several genotypes. Peanut Science, 18, 19-22. MOINUDDIN, M. & IMAS, P. 2007. Evaluation of potassium compared to other osmolytes in relation to osmotic adjustment and drought tolerance of chickpea under water deficit environments. Journal of Plant Nutrition, 30, 517-535. MOORE, J., CHENG, Z., SU, L. & YU, L. L. 2006. Effects of solid-state enzymatic treatments on the antioxidant properties of wheat bran. Journal of Agricultural and Food Chemistry, 54, 9032-9045. MORRIS, M. C. 2009. The role of nutrition in Alzheimer's disease: epidemiological evidence. European Journal of Neurology, 16, 1-7. MUIR, W., NYQUIST, W. E. & XU, S. 1992. Alternative partitioning of the genotype-by- environment interaction. Theoretical and Applied Genetics, 84, 193-200. MURPHY, K. M., REEVES, P. G. & JONES, S. S. 2008. Relationship between yield and mineral nutrient concentrations in historical and modern spring wheat cultivars Euphytica, 163, 381-390. NABRZYSKI, M. 2007. Functional role of some minerals in foods. In: SZEFER, P. & NRIAGU, J. O. (eds.) Mineral components in foods. Boca Raton, Florida: Taylor & Francis Group. NANTEL, G. & TONTISIRI, K. 2002. Policy and sustainability issues. Journal of Nutrition, 132, 839S-844S. NAOZUKA, J., VIEIRA, E. C., NASCIMENTO, A. N. & OLIVEIRA, P. V. 2011. Elemental analysis of nuts and seeds by axially viewed ICP OES. Food Chemistry, 124, 1667-1672. NARDI, E. P., EVANGELISTA, F. S., TORMEN, L., SAINŤPIERRE, T. D., CURTIUS, A. J., DE SOUZA, S. S. & BARBOSA, F., JR. 2009. The use of inductively coupled plasma mass spectrometry (ICP-MS) for the determination of toxic and essential elements in different types of food samples. Food Chemistry, 112, 727-732. NEGRE-SALVAYRE, A., AUGE, N., AYALA, V., BASAGA, H., BOADA, J., BRENKE, R., CHAPPLE, S., COHEN, G., FEHER, J., GRUNE, T., LENGYEL, G., MANN, G. E., PAMPLONA, R., POLI, G., PORTERO-OTIN, M., RIAHI, Y., SALVAYRE, R., SASSON, S., SERRANO, J., SHAMNI, O., SIEMS, W., SIOW, R. C. M., WISWEDEL, I., ZARKOVIC, K. & ZARKOVIC, N. 2010. Pathological aspects of lipid peroxidation. Free Radical Research, 44, 1125-1171. NEPOTE, V., GROSSO, N. R. & GUZMAN, C. A. 2002. Extraction of antioxidant components from peanut skins. Grasas Y Aceites, 53, 391-395. NEPOTE, V., GROSSO, N. R. & GUZMÁN, C. A. 2005. Optimization of extraction of phenolic antioxidants from peanut skins. Journal of the Science of Food and Agriculture, 85, 33- 38.

| References 286

NESTEL, P., BOUIS, H. E., MEENAKSHI, J. V. & PFEIFFER, W. 2006. Biofortification of staple food crops. Journal of Nutrition, 136, 1064-1067. NHMRC 2006. Nutrient reference values for Australia and New Zealand. In: NATIONAL HEALTH AND MEDICAL RESEARCH COUNCIL-DEPARTMENT OF HEALTH AND AGING (ed.). Commonwealth of Australia. NIBAU, C., GIBBS, D. J. & COATES, J. C. 2008. Branching out in new directions: the control of root architecture by lateral root formation. New Phytologist, 179, 595-614. NICOLAÏ, B. M., BEULLENS, K., BOBELYN, E., PEIRS, A., SAEYS, W., THERON, K. I. & LAMMERTYN, J. 2007. Nondestructive measurement of fruit and vegetable quality by means of NIR spectroscopy: a review. Postharvest Biology and Technology, 46, 99-118. OCHOA, I. E., BLAIR, M. W. & LYNCH, J. P. 2006. QTL analysis of adventitious root formation in common bean under contrasting phosphorus availability. Crop Science, 46, 1609-1621. OKARTER, N., LIU, C.-S., SORRELLS, M. E. & LIU, R. H. 2010. Phytochemical content and antioxidant activity of six diverse varieties of whole wheat. Food Chemistry, 119, 249- 257. OKMEN, B., SIGVA, H. O., MUTLU, S., DOGANLAR, S., YEMENICIOGLU, A. & FRARY, A. 2009. Total antioxidant activity and total phenolic contents in different Turkish eggplant (Solanum melongena L.) cultivars. International Journal of Food Properties, 12, 616- 624. OSMONT, K. S., SIBOUT, R. & HARDTKE, C. S. 2007. Hidden branches: developments in root system architecture. Annual Review of Plant Biology, 58, 93-113. OTANI, H. 2011. Oxidative stress as pathogenesis of cardiovascular risk associated with metabolic syndrome. Antioxidants and Redox Signaling, 15, 1911-1926. OU, B., HAMPSCH-WOODILL, M., FLANAGAN, J. A., DEEMER, E. K., PRIOR, R. L. & HUANG, D. 2002. Novel fluorometric assay for hydroxyl radical prevention capacity using fluorescein as the probe. Journal of Agricultural and Food Chemistry, 50, 2772-2777. OU, B., HAMPSCH-WOODILL, M. & PRIOR, R. L. 2001. Development and validation of an improved oxygen radical absorbance capacity assay using fluorescein as the fluorescent probe. Journal of Agricultural and Food Chemistry, 49, 4619-4626. ÖZYÜREK, M., GÜÇLÜ, K., TÜTEM, E., BAŞKAN, K. S., ERÇAĞ, E., ÇELIK, S. E., BAKI, S., YıLDıZ, L., KARAMANC, Ş. & APAK, R. 2011. A comprehensive review of CUPRAC methodology. Analytical Methods, 3, 2439-2453. PARK, S., ELLESS, M. P., PARK, J., JENKINS, A., LIM, W., CHAMBERS, E., IV & HIRSCHI, K. D. 2009. Sensory analysis of calcium-biofortified lettuce. Plant Biotechnology Journal, 7, 106- 117. PELLEGRINI, N., SERAFINI, M., COLOMBI, B., DEL RIO, D., SALVATORE, S., BIANCHI, M. & BRIGHENTI, F. 2003. Total antioxidant capacity of plant foods, beverages and oils consumed in Italy assessed by three different in vitro assays. Journal of Nutrition, 133, 2812-2819. PELLEGRINI, N., SERAFINI, M., SALVATORE, S., DEL RIO, D., BIANCHI, M. & BRIGHENTI, F. 2006. Total antioxidant capacity of spices, dried fruits, nuts, pulses, cereals and sweets consumed in Italy assessed by three different in vitro assays. Molecular Nutrition and Food Research, 50, 1030-1038. PÉREZ-EXPÓSITO, A. B. & KLEIN, B. P. 2009. Impact of fortified blended food aid products on nutritional status of infants and young children in developing countries. Nutrition Reviews, 67, 706-718. PÉREZ-JIMÉNEZ, J. & SAURA-CALIXTO, F. 2006. Effect of solvent and certain food constituents on different antioxidant capacity assays. Food Research International, 39, 791-800. PERKINELMER SCIEX. 2000. Elan DRC ICP-MS system with dynamic bandpass tuning [Online]. Available: http://las.perkinelmer.com/Content/technicalinfo/tch_elandrcicpmssystem.pdf.

| References 287

PFEIFFER, W. H. & MCCLAFFERTY, B. 2007. HarvestPlus: breeding crops for better nutrition. Crop Science, 47, S88-S105. PHAN-THIEN, K.-Y., GOLIC, M., WRIGHT, G. C. & LEE, N. A. 2011. Feasibility of estimating peanut essential minerals by near infrared reflectance spectroscopy. Sensing and Instrumentation for Food Quality and Safety, 5, 43-49. PHAN-THIEN, K.-Y., WRIGHT, G. C. & LEE, N. A. 2010. Genotype-by-environment interaction affects the essential mineral composition of peanut (Arachis hypogaea L.) kernels. Journal of Agricultural and Food Chemistry, 58, 9204-9213. PHAN-THIEN, K.-Y., WRIGHT, G. C. & LEE, N. A. 2012. Inductively coupled plasma-mass spectrometry (ICP-MS) and -optical emission spectroscopy (ICP–OES) for determination of essential minerals in closed acid digestates of peanuts (Arachis hypogaea L.). Food Chemistry, 134, 453-460. PICK, D., LEITERER, M. & EINAX, J. W. 2010. Reduction of polyatomic interferences in biological material using dynamic reaction cell ICP-MS. Microchemical Journal, 95, 315-319. PILON, M., COHU, C. M., RAVET, K., ABDEL-GHANY, S. E. & GAYMARD, F. 2009. Essential transition metal homeostasis in plants. Current Opinion in Plant Biology, 12, 347-357. PINCEMAIL, J., KEVERS, C., TABART, J., DEFRAIGNE, J.-O. & DOMMES, J. 2012. Cultivars, culture conditions, and harvest time influence phenolic and ascorbic acid contents and antioxidant capacity of strawberry (Fragaria x ananassa). Journal of Food Science, 77, C205-C210. PINHEIRO, C., BAETA, J. P., PEREIRA, A. M., DOMINGUES, H. & RICARDO, C. P. 2010. Diversity of seed mineral composition of Phaseolus vulgaris L. germplasm. Journal of Food Composition and Analysis, 23, 319-325. POIROUX-GONORD, F., BIDEL, L. P. R., FANCIULLINO, A.-L., GAUTIER, H., LAURI-LOPEZ, F. & URBAN, L. 2010. Health benefits of vitamins and secondary metabolites of fruits and vegetables and prospects to increase their concentrations by agronomic approaches. Journal of Agricultural and Food Chemistry, 58, 12065-12082. POTREBKO, I. & RESURRECCION, A. V. A. 2009. Effect of ultraviolet doses in combined ultraviolet-ultrasound treatments on trans-resveratrol and trans-piceid contents in sliced peanut kernels. Journal of Agricultural and Food Chemistry, 57, 7750-7756. PRATICÒ, D. 2000. Oxidative injury in diseases of the central nervous system: focus on Alzheimer's Disease. American Journal of Medicine, 109, 577-585. PRIOR, R. L. 2003. Fruits and vegetables in the prevention of cellular oxidative damage. American Journal of Clinical Nutrition, 78, 570S-578S. PRIOR, R. L., HOANG, H., GU, L., WU, X., BACCHIOCCA, M., HOWARD, L., HAMPSCH-WOODILL, M., HUANG, D., OU, B. & JACOB, R. A. 2003. Assays for hydrophilic and lipophilic

antioxidant capacity (oxygen radical absorbance capacity (ORACFL)) of plasma and other biological and food samples. Journal of Agricultural and Food Chemistry, 51, 3273-3279. PRIOR, R. L., WU, X. & SCHAICH, K. 2005. Standardized methods for the determination of antioxidant capacity and phenolics in foods and dietary supplements. Journal of Agricultural and Food Chemistry, 53, 4290-4302. PROTEGGENTE, A. R., PANNALA, A. S., PAGANGA, G., VAN BUREN, L., WAGNER, E., WISEMAN, S., VAN DE PUT, F., DACOMBE, C. & RICE-EVANS, C. A. 2002. The antioxidant activity of regularly consumed fruit and vegetables reflects their phenolic and vitamin C composition. Free Radicals Research, 36, 217-233. PUIG, S. & PEÑARRUBIA, L. 2009. Placing metal micronutrients in context: transport and distribution in plants. Current Opinion in Plant Biology, 12, 299-306. PULIDO, R., BRAVO, L. & SAURA-CALIXTO, F. 2000. Antioxidant activity of dietary polyphenols as determined by a modified ferric reducing/antioxidant power assay. Journal of Agricultural and Food Chemistry, 48, 3396-3402.

| References 288

QAIM, M., STEIN, A. J. & MEENAKSHI, J. V. 2007. Economics of biofortification. Agricultural Economics, 37, 119-133. RACHAPUTI, R. & CHAUHAN, Y. 2010. Peanut - history of crop improvement and evaluation research [Online]. Queensland Government. Available: http://www.dpi.qld.gov.au/4791_4410.htm [Accessed 05/04/11 2011]. RAVINDRAN, V., RAVINDRAN, G. & SIVALOGAN, S. 1994. Total and phytate phosphorus contents of various foods and feedstuffs of plant origin. Food Chemistry, 50, 133-136. RE, R., PELLEGRINI, N., PROTEGGENTE, A., PANNALA, A., YANG, M. & RICE-EVANS, C. 1999. Antioxidant activity applying an improved ABTS radical cation decolorization assay. Free Radical Biology and Medicine 26, 1231-1237. REBAFKA, F.-P., NDUNGURU, B. J. & MARSCHNER, H. 1993. Single superphosphate depresses molybdenum uptake and limits yield response to phosphorus in groundnut (Arachis hypogaea L.) grown on an acid sandy soil in Niger, West Africa. Fertilizer Research, 34, 233-242. REID, R. & HAYES, J. 2003. Mechanisms and control of nutrient uptake in plants. International Review of Cytology, 229, 73-114. RICE-EVANS, C. A., MILLER, N. J. & PAGANGA, G. 1996. Structure-antioxidant activity relationships of flavonoids and phenolic acids. Free Radical Biology and Medicine, 20, 933-936. ROBERTS, R. A., LASKIN, D. L., SMITH, C. V., ROBERTSON, F. M., ALLEN, E. M. G., DOORN, J. A. & SLIKKER, W. 2009. Nitrative and oxidative stress in toxicology and disease. Toxicological Sciences, 112, 4-16. ROMERO-PÉREZ, A. I., IBERN-GOMÉZ, M., LAMUELA-RAVENTOS, R. M. & DE LA TORRE- BORONAT, M. C. 1999. Piceid, the major resveratrol derivative in grape juices. Journal of Agricultural and Food Chemistry, 47, 1533-1536. ROSTAGNO, M. A., PALMA, M. & BARROSO, C. G. 2003. Ultrasound-assisted extraction of soy isoflavones. Journal of Chromatography A, 1012, 119-128. RUANO-RAMOS, A., GARCÍA-CIUDAD, A. & GARCÍA-CRIADO, B. 1999. Near infrared spectroscopy prediction of mineral content in botanical fractions from semi-arid grasslands. Animal Feed Science and Technology, 77, 331-343. RUDOLF, J. R. & RESURRECCION, A. V. A. 2005. Elicitation of resveratrol in peanut kernels by application of abiotic stresses. Journal of Agricultural and Food Chemistry, 53, 10186- 10192. RUDOLF, J. R. & RESURRECCION, A. V. A. 2007. Optimization of trans-resveratrol concentration and sensory properties of peanut kernels by slicing and ultrasound treatment, using response surface methodology. Journal of Food Science, 72, S450-S462. RUDOLF, J. R., RESURRECCION, A. V. A., SAALIA, F. K. & PHILLIPS, R. D. 2005. Development of a reverse-phase high-performance liquid chromatography method for analyzing trans- resveratrol in peanut kernels. Food Chemistry, 89, 623-628. SAKAKIBARA, H., HONDA, Y., NAKAGAWA, S., ASHIDA, H. & KANAZAWA, K. 2003. Simultaneous determination of all polyphenols in vegetables, fruits, and teas. Journal of Agricultural and Food Chemistry, 51, 571-581. SALES, J. M. & RESURRECCION, A. V. A. 2009. Maximising resveratrol and piceid contents in UV and ultrasound treated peanuts. Food Chemistry, 117, 674-680. SALES, J. M. & RESURRECCION, A. V. A. 2010a. Maximizing phenolics, antioxidants and sensory acceptance of UV and ultrasound-treated peanuts. Food Science and Technology, 43, 1058-1066. SALES, J. M. & RESURRECCION, A. V. A. 2010b. Phenolic profile, antioxidants, and sensory acceptance of bioactive-enhanced peanuts using ultrasound and UV. Food Chemistry, 122, 795-803.

| References 289

SALT, D. E., BAXTER, I. & LAHNER, B. 2008. Ionomics and the study of the plant ionome. Annual Review of Plant Biology, 59, 709-733. SÁNCHEZ-MORENO, C. 2002. Review: methods used to evaluate the free radical scavenging activity in foods and biological systems. Food Science and Technology - Lebensmittel Wissenschaft und Technologie, 8, 121-137. SANDBERG, A.-S. 2002. Bioavailability of minerals in legumes. British Journal of Nutrition, 88, 281-285. SANDERS, T. H., MCMICHAEL, R. W., JR. & HENDRIX, K. W. 2000. Occurrence of resveratrol in edible peanuts. Journal of Agricultural and Food Chemistry, 48, 1243-1246. SANTIAGO, R., DE ARMAS, R., FONTANIELLA, B., VICENTE, C. & LEGAZ, M.-E. 2009. Changes in soluble and cell wall-bound hydroxycinnamic and hydroxybenzoic acids in sugarcane cultivars inoculated with Sporisorium scitamineum sporidia. European Journal of Plant Pathology, 124, 439-450. SAURA-CALIXTO, F. 2011. Dietary fiber as a carrier of dietary antioxidants: an essential physiological function. Journal of Agricultural and Food Chemistry, 59, 43-49. SAVOURET, J. F. & QUESNE, M. 2002. Resveratrol and cancer: a review. Biomedecine and Pharmacotherapy, 56, 84-87. SAXENA, R., VENKAIAH, K., ANITHA, P., VENU, L. & RAGHUNATH, M. 2007. Antioxidant activity of commonly consumed plant foods of India: contribution of their phenolic content. International Journal of Food Sciences and Nutrition, 58, 250-260. SCALBERT, A., MANACH, C., MORAND, C., RÉMÉSY, C. & JIMÉNEZ, L. 2005. Dietary polyphenols and the prevention of diseases. Critical Reviews in Food Science and Nutrition, 45, 287- 306. SCAMPICCHIO, M., BULBARELLO, A., ARECCHI, A. & MANNINO, S. 2008. Electrospun nanofibers as selective barrier to the electrochemical polyphenol oxidation. Electrochemistry Communications, 10, 991-994. SENGUPTA, R. & BHATTACHARYYA, D. K. 1996. Enzymatic extraction of mustard seed and rice bran. Journal of the American Oil Chemists' Society, 73, 687-692. SHEN, J., YUAN, L., ZHANG, J., LI, H., BAI, Z., CHEN, X., ZHANG, W. & ZHANG, F. 2011. Phosphorus dynamics: from soil to plant. Plant Physiology, 156, 997-1005. SHEN, Y., JIN, L., XIAO, P., LU, Y. & BAO, J. 2009. Total phenolics, flavonoids, antioxidant capacity in rice grain and their relations to grain color, size and weight. Journal of Cereal Science, 49, 106-111. ŠIMIĆ, D., SUDAR, R., LEDENČAN, T., JAMBROVIĆ, A., ZDUNIĆ, Z., BRKIĆ, I. & KOVAČEVIĆ, V. 2009. Genetic variation of bioavailable iron and zinc in grain of a maize population. Journal of Cereal Science, 50, 392-397. SINGH, D., KAUR, R., CHANDER, V. & CHOPRA, K. 2006. Antioxidants in the prevention of renal disease. Journal of Medicinal Food, 9, 443-450. SINGLETON, J. A., STIKELEATHER, L. F. & SANFORD, J. H. 2002. LC-electrospray ionization and LC-FABMS study of flavonoid glycosides extracted from peanut meal. Journal of the American Oil Chemists' Society, 79, 741-748. SINGLETON , V. L. & ROSSI, J. A., JR. 1965. Colorimetry of total phenolics with phosphomolybdic-phosphotungstic acid reagents. American Journal of Enology and Viticulture, 16, 144-158. SLOAN, A. E. 2008. The top 10 functional food trends. Food Technology, 62, 25-35. SMEDS, A. I., EKLUND, P. C., SJÖHOLM, R. E., WILLFÖR, S. M., NISHIBE, S., DEYAMA, T. & HOLMBOM, B. R. 2007. Quantification of a broad spectrum of lignans in cereals, oilseeds, and nuts. Journal of Agricultural and Food Chemistry, 55, 1337-1346. SMITH, K. F., WILLIS, S. E. & FLINN, P. C. 1991. Measurement of the magnesium concentration in perennial ryegrass (Lolium perenne) using near infrared reflectance spectroscopy. Australian Journal of Agricultural Research, 42, 1399-1404.

| References 290

SNYDER, L. R., KIRKLAND, J. J. & DOLAN, J. W. 2010. Introduction to Modern Liquid Chromatography, Hoboken, New Jersey, John Wiley & Sons, Inc. SOBOLEV, V. S. 2008. Localized production of phytoalexins by peanut (Arachis hypogaea) kernels in response to invasion by Aspergillus species. Journal of Agricultural and Food Chemistry, 56, 1949-1954. SOBOLEV, V. S. & COLE, R. J. 1999. trans-Resveratrol content in commercial peanuts and peanut products. Journal of Agricultural and Food Chemistry, 47, 1435-1439. SOBOLEV, V. S., COLE, R. J. & DORNER, J. W. 1995. Isolation, purification, and liquid chromatographic determination of stilbene phytoalexins in peanuts. Journal of AOAC International, 78, 1177-1182. SOBOLEV, V. S., DEYRUP, S. T. & GLOER, J. B. 2006. New peanut (Arachis hypogaea) phytoalexin with prenylated benzenoid and but-2-enolide moieties. Journal of Agricultural and Food Chemistry, 54, 2111-2115. SOBOLEV, V. S., GUO, B. Z., HOLBROOK, C. C. & LYNCH, R. E. 2007. Interrelationship of phytoalexin production and disease resistance in selected peanut genotypes. Journal of Agricultural and Food Chemistry, 55, 2195-2200. SOBOLEV, V. S., KHAN, S. I., TABANCA, N., WEDGE, D. E., MANLY, S. P., CUTLER, S. J., COY, M. R., BECNEL, J. J., NEFF, S. A. & GLOER, J. B. 2011. Biological activity of peanut (Arachis hypogaea) phytoalexins and selected natural and synthetic stilbenoids. Journal of Agricultural and Food Chemistry, 59, 1673-1682. SOBOLEV, V. S., NEFF, S. A. & GLOER, J. B. 2009. New stilbenoids from peanut (Arachis hypogaea) seeds challenged by an Aspergillus caelatus strain. Journal of Agricultural and Food Chemistry, 57, 62-68. SOBOLEV, V. S., NEFF, S. A. & GLOER, J. B. 2010. New dimeric stilbenoids from fungal- challenged peanut (Arachis hypogaea) seeds. Journal of Agricultural and Food Chemistry, 58, 875-881. SOBOLEV, V. S., SY, A. A. & GLOER, J. B. 2008. Spermidine and flavonoid conjugates from peanut (Arachis hypogaea) flowers. Journal of Agricultural and Food Chemistry, 56, 2960-2969. SOSULSKI, F. W. & DABROWSKI, K. J. 1984. Composition of free and hydrolyzable phenolic acids in the flours and hulls of ten legume species. Journal of Agricultural and Food Chemistry, 32, 131-133. SPEHAR, C. R. 1994. Seed quality of soya bean based on mineral composition of seeds of 45 varieties grown in a Brazilian Savanna acid soil. Euphytica, 76, 127-132. SPEISKY, H., LÓPEZ-ALARCÓN, C., GÓMEZ, M., FUENTES, J. & SANDOVAL-ACUÑA, C. 2012. First web-based database on total phenolics and oxygen radical absorbance capacity (ORAC) of fruits produced and consumed within the South Andes region of South America. Journal of Agricultural and Food Chemistry [Online]. Available: DOI 10.1021/jf205167k. STALKER, H. T. 1997. Peanut (Arachis hypogaea L.). Field Crops Research, 53, 205-217. STEIN, A. J. 2010. Global impacts of human mineral malnutrition. Plant and Soil, 335, 133-154. STEIN, A. J., MEENAKSHI, J. V., QAIM, M., NESTEL, P., SACHDEV, H. P. S. & BHUTTA, Z. A. 2005. Analyzing the health benefits of biofortified staple crops by means of the disability- adjusted life years approach. A handbook focusing on iron, zinc and vitamin A, Washington, DC, International Center for Tropical Agriculture (CIAT). STRANGE, R. N. & RAO, P. V. S. 1994. The phytoalexin response of groundnut and its role in disease resistance. Oléagineux, 49, 227-233. SUNDARAM, J., KANDALA, C. V. & BUTTS, C. L. 2009. Application of near infrared spectroscopy to peanut grading and quality analysis: overview. Sensing and Instrumentation for Food Quality and Safety, 3, 156-164.

| References 291

TAKAHATA, Y., KAI, Y., TANAKA, M., NAKAYAMA, H. & YOSHINAGA, M. 2011. Enlargement of the variances in amount and composition of anthocyanin pigments in sweetpotato storage roots and their effect on the differences in DPPH radical-scavenging activity. Scientia Horticulturae, 127, 469-474. TALCOTT, S. T., DUNCAN, C. E., DEL POZO-INSFRAN, D. & GORBET, D. W. 2005a. Polyphenolic and antioxidant changes during storage of normal, mid, and high oleic acid peanuts. Food Chemistry, 89, 77-84. TALCOTT, S. T., HOWARD, L. R. & BRENES, C. H. 2000. Contribution of periderm material and blanching time to the quality of pasteruized peach puree. Journal of Agricultural and Food Chemistry, 48, 4590-4596. TALCOTT, S. T., PASSERETTI, S., DUNCAN, C. E. & GORBET, D. W. 2005b. Polyphenolic content and sensory properties of normal and high oleic acid peanuts. Food Chemistry, 90, 379- 388. TANG, J., ZOU, C., HE, Z., SHI, R., ORTIZ-MONASTERIO, I., QU, Y. & ZHANG, Y. 2008. Mineral element distributions in milling fractions of Chinese wheats. Journal of Cereal Science, 48, 821-828. TAPSELL, L., WILLIAMS, P., DROULEZ, V., SOUTHEE, D., PATCH, C. & LETHBRIDGE, A. 2005. Functional Foods for the Australian industry: definitions and opportunities. National Centre of Excellence in Functional Foods. TEJADA-JIMÉNEZ, M., GALVÁN, A., FERNÁNDEZ, E. & LLAMAS, Á. 2009. Homeostasis of the micronutrients Ni, Mo and Cl with specific biochemical functions. Current Opinion in Plant Biology, 12, 358-363. THOMPSON, L. U., BOUCHER, B. A., LIU, Z., COTTERCHIO, M. & KREIGER, N. 2006. Phytoestrogen content of foods consumed in Canada, including isoflavones, lignans, and coumestan. Nutrition and Cancer, 54, 184-201. TILLMAN, B. L., GORBET, D. W. & PERSON, G. 2006. Predicting oleic and linoleic acid content of single peanut seeds using near-infrared reflectance spectroscopy. Crop Science, 46, 2121-2126. TOKUŞOĜLU, Ö., ÜNAL, M. K. & YEMIŞ, F. 2005. Determination of the phytoalexin resveratrol (3.5.4'-trihydroxystilbene) in peanuts and pistachios by high-performance liquid chromatographic diode array (HPLC-DAD) and gas chromatography-mass spectrometry (GC-MS). Journal of Agricultural and Food Chemistry, 53, 5003-5009. TREMBLAY, G. F., NIE, Z., BÉLANGER, G., PELLETIER, S. & ALLARD, G. 2009. Predicting timothy mineral concentrations, dietary cation-anion difference, and grass tetany index by near-infrared reflectance spectroscopy. Journal of Dairy Science, 92, 4499-4506. TSANTILI, E., KONSTANTINIDIS, K., CHRISTOPOULOS, M. V. & ROUSSOS, P. A. 2011. Total phenolics and flavonoids and total antioxidant capacity in pistachio (Pistachia vera L.) nuts in relation to cultivars and storage conditions. Scientia Horticulturae, 129, 694- 701. TSAO, R. & YANG, R. 2003. Optimization of a new mobile phase to know the complex and real polyphenolic composition: towards a total phenolic index using high-performance liquid chromatography. Journal of Chromatography A, 1018, 29-40. TUNCEL, N. B. & YıLMAZ, N. 2011. Gamma-oryzanol content, phenolic acid profiles and antioxidant activity of rice milling fractions. European Food Research and Technology, 233, 577-585. U. S. DEPARTMENT OF AGRICULTURE & AGRICULTURAL RESEARCH SERVICE 2010. Oxygen radical absorbance capacity (ORAC) of selected foods, release 2. U. S. DEPARTMENT OF AGRICULTURE & AGRICULTURAL RESEARCH SERVICE. 2012. Oxygen radical absorbance capacity (ORAC) of selected foods, release 2 (2010) [Online]. Available: http://www.ars.usda.gov/Services/docs.htm?docid=15866 [Accessed 22 June 2012.

| References 292

VELIOGLU, Y. S., MAZZA, G., GAO, L. & OOMAH, B. D. 1998. Antioxidant activity and total phenolics in selected fruits, vegetables, and grain products. Journal of Agricultural and Food Chemistry, 46, 4113-4117. VERBEKE, W. 2005. Consumer acceptance of functional foods: socio-demographic, cognitive and attitudinal determinants. Food Quality and Preference, 16, 45-57. VILKHU, K., MAWSON, R., SIMONS, L. & BATES, D. 2008. Applications and opportunities for ultrasound assisted extraction in the food industry - a review. Innovative Food Science and Emerging Technologies, 9, 161-169. VILLAÑO, D., FERNÁNDEZ-PACHÓN, M. S., MOYÁ, M. L., TRONCOSO A, A. M. & GARCÍA- PARRILLA, M. C. 2007. Radical scavenging ability of polyphenolic compounds towards DPPH free radical. Talanta, 71, 230-235. VILLAÑO, D., FERNÁNDEZ-PACHÓN, M. S., TRONCOSO, A. M. & GARCÍA-PARRILLA, M. C. 2005. Comparison of antioxidant activity of wine phenolic compounds and metabolites in vitro. Analytica Chimica Acta, 538, 391-398. VILLARREAL-LOZOYA, J. E., LOMBARDINI, L. & CISNEROS-ZEVALLOS, L. 2007. Phytochemical constituents and antioxidant capacity of different pecan [Carya illinoinensis (Wangenh.) K. Koch] cultivars. Food Chemistry, 102, 1241-1249. VIRK, P. & BARRY, G. 2009. Biofortified rice - towards combating human micronutrient deficiencies. 14th Australasian Plant Breeding and 11th SABRAO Conference, 10-14 August 2009. Cairns, Queensland, Australia. VLADIMIROV, Y. A. & PROSKURNINA, E. V. 2009. Free radicals and cell chemiluminescence. Biochemistry, 74, 1545-1566. VUORELA, S., MEYER, A. S. & HEINONEN, M. 2003. Quantitative analysis of the main phenolics in rapeseed meal and oils processed differently using enzymatic hydrolysis and HPLC. European Food Research and Technology, 217, 517-523. WALK, T. C., JARAMILLO, R. & LYNCH, J. P. 2006. Architectural tradeoffs between adventitious and basal roots for phosphorus acquisition. Plant and Soil, 279, 347-366. WANG, H., INUKAI, Y. & YAMAUCHI, A. 2006. Root development and nutrient uptake. Critical Reviews in Plant Sciences, 25, 279-301. WANG, J. & MARTINEZ, T. 1991. Scanning tunneling microscopic investigation of surface fouling of glassy carbon surfaces due to phenol oxidation. Journal of Electroanalytical Chemistry, 313, 129-140. WANG, K.-H., LAI, Y.-H., CHANG, J.-C., KO, T.-F., SHYU, S.-L. & CHIOU, R. Y.-Y. 2005. Germination of peanut kernels to enhance resveratrol biosynthesis and prepare sprouts as a functional vegetable. Journal of Agricultural and Food Chemistry, 53, 242-246. WANG, L. & WELLER, C. L. 2006. Recent advances in extraction of nutraceuticals from plants. Trends in Food Science and Technology, 17, 300-312. WANG, M. L., GILLASPIE, A. G., MORRIS, J. B., PITTMAN, R. N., DAVIS, J. & PEDERSON, G. A. 2008a. Flavonoid content in different legume germplasm seeds quantified by HPLC. Plant Genetic Resources: Characterization and Utilization, 6, 62-69. WANG, M. L., PINNOW, D. L. & PITTMAN, R. N. 2007. Preliminary screening of peanut germplasm in the US collection for TSWV resistance and high flavonoid content. Plant Pathology, 6, 219-226. WANG, M. L. & PITTMAN, R. N. 2008. Resveratrol content in seeds of peanut germplasm quantified by HPLC. Plant Genetic Resources: Characterization and Utilization, 7, 80-83. WANG, S., MECKLING, K. A., MARCONE, M. F., KAKUDA, Y. & TSAO, R. 2011. Synergistic, additive, and antagonistic effects of food mixtures on total antioxidant capacities. Journal of Agricultural and Food Chemistry, 59, 960-968. WANG, Y., WANG, Z., CHENG, S. & HAN, F. 2008b. Aqueous enzymatic extraction of oil and protein hydrolysates from peanut. Food Science and Technology Research, 14, 533- 540.

| References 293

WATERHOUSE, A. L. 2002. Determination of total phenolics. Current Protocols in Food Analytical Chemistry. John Wiley and Sons, Inc. WATERS, B. M. & SANKARAN, R. P. 2011. Moving micronutrients from the soil to the seeds: genes and physiological processes from a biofortification perspective. Plant Science, 180, 562-574. WAYNER, D. D. M., BURTON, G. W., INGOLD, K. U. & LOCKE, S. J. 1985. Quantitative measurement of the total peroxyl radical-trapping antioxidant capability of human plasma by controlled peroxidations - the important controibution made by plasma proteins. FEBS Letters, 187. WEBSTER-GANDY, J., MADDEN, A. & HOLDSWORTH, M. 2006. Oxford handbook of nutrition and dietetics, New York, Oxford University Press. WELCH, R. M. 1995. Micronutrient nutrition of plants. Critical Reviews in Plant Sciences, 14, 49- 82. WELCH, R. M. & GRAHAM, R. D. 2004. Breeding for micronutrients in staple food crops from a human nutrition perspective. Journal of Experimental Botany, 55, 353-364. WELCH, R. M. & GRAHAM, R. D. 2005. Agriculture: the real nexus for enhancing bioavailable micronutrients in food crops. Journal of Trace Elements in Medicine and Biology, 18, 299-307. WHENT, M., HAO, J., SLAVIN, M., ZHOU, M., SONG, J., KENWORTHY, W. & YU, L. L. 2009. Effect of genotype, environment, and their interaction on chemical composition and antioxidant properties of low-linolenic soybeans grown in Maryland. Journal of Agricultural and Food Chemistry, 57, 10163-10174. WHITE, P. J. & BROADLEY, M. R. 2005. Biofortifying crops with essential mineral elements. Trends in Plant Science, 10, 586-593. WHITE, P. J. & BROADLEY, M. R. 2009. Biofortification of crops with seven mineral elements often lacking in human diets – iron, zinc, copper, calcium, magnesium, selenium and iodine. New Phytologist, 182, 49-84. WHITE, P. J. & BROWN, P. H. 2010. Plant nutrition for sustainable development and global health. Annals of Botany, 105, 1073-1080. WHO. 2004. Global Health Observatory database [Online]. World Health Organization. Available: apps.who.int/ghodata/ [Accessed 2 May 2011]. WIN, M. M., ABDUL-HAMID, A., BAHARIN, B. S., ANWAR, F. & SAARI, N. 2011. Effects of roasting on phenolics composition and antioxidant activity of peanut (Arachis hypogaea L.) kernel flour. European Food Research and Technology, 233, 599-608. WISSUWA, M. & AE, N. 2001. Genotypic differences in the presence of hairs on roots and gynophores of peanuts (Arachis hypogaea L.) and their significance for phosphorus uptake. Journal of Experimental Botany, 52, 1703-1710. WOOD, L. G., GIBSON, P. G. & GARG, M. L. 2006. A review of the methodology for assessing in vivo antioxidant capacity. Journal of the Science of Food and Agriculture, 86, 2057- 2066. WRIGHT, G. C. 2009. RE: Personal communication that >10% trait variation was indicative of breeding potential, based on extensive (>20 years) experience in peanut breeding. WU, X., BEECHER, G. R., HOLDEN, J. M., HAYTOWITZ, D. B., GEBHARDT, S. E. & PRIOR, R. L. 2004a. Lipophilic and hydrophilic antioxidant capacities of common foods in the United States. Journal of Agricultural and Food Chemistry, 52, 4026-4037. WU, X., GU, L., HOLDEN, J., HAYTOWITZ, D. B., GEBHARDT, S. E., BEECHER, G. & PRIOR, R. L. 2004b. Development of a database for total antioxidant capacity in foods: a preliminary study. Journal of Food Composition and Analysis, 17, 407-422. XU, B. J. & CHANG, S. K. C. 2007. A comparative study on phenolic profiles and antioxidant activities of legumes as affected by extraction solvents. Journal of Food Science, 72, 5159-5166.

| References 294

XU, H.-X. & CHEN, J.-W. 2011. Commercial quality, major bioactive compound content and antioxidant capacity of 12 cultivars of loquat (Eriobotrya japonica Lindl.) fruits. Journal of the Science of Food and Agriculture, 91, 1057-1063. YAN, W., KANG, M. S., MA, B., WOODS, S. & CORNELIUS, P. L. 2007. GGE biplot vs. AMMI analysis of genotype-by-environment data. Crop Science, 47, 643-655. YANG, M.-H., KUO, C.-H., HSIEH, W.-C. & KU, K.-L. 2010. Investigation of microbial elicitation of trans-resveratrol and trans-piceatannol in peanut callus led to the application of chitin as a potential elicitor. Journal of Agricultural and Food Chemistry, 58, 9537-9541. YANG, W. & OMAYE, S. T. 2009. Air pollutants, oxidative stress and human health. Mutation Research, 674, 45-54. YASHIN, Y. I., NEMZER, B. V., RYZHNEV, V. Y., YASHIN, A. Y., CHERNOUSOVA, N. I. & FEDINA, P. A. 2010. Creation of a databank for content of antioxidants in food products by an amperometric method. Molecules, 15, 7450-7466. YEMIS, O., BAKKALBASI, E. & ARTIK, N. 2008. Antioxidative activities of grape (Vitis vinifera) seed extracts obtained from different varieties grown in Turkey. International Journal of Food Science and Technology, 43, 154-159. YOUNG, J. H., WHITAKER, T. B., BLANKENSHIP, P. D., BRUSEWITZ, G. H., TROEGER, J. M., STEELE, J. L. & PERSON, N. K., JR. 1982. Effect of oven drying time on peanut moisture determination. Transactions of the ASAE, 25, 491-496. YU, J., AHMEDNA, M. & GOKTEPE, I. 2005. Effects of processing methods and extraction solvents on concentration and antioxidant activity of peanut skin phenolics. Food Chemistry, 90, 199-206. YU, J., AHMEDNA, M., GOKTEPE, I. & DAI, J. 2006. Peanut skin procyanidins: composition and antioxidant activities as affected by processing. Journal of Food Composition and Analysis, 19, 364-371. YUSNAWAN, E., MARQUIS, C. P. & LEE, N. A. 2012. Purification and characterization of Ara h1 and Ara h3 from four peanut market types revealed higher order oligomeric structures. Journal of Agricultural and Food Chemistry, 60, 10352-10358. ZAHARIEVA, T. & RÖMHELD, V. 1991. Factors affecting cation-anion uptake balance and iron acquisition in peanut plants grown on calcareous soils. Plant and Soil, 130, 81-86. ZHANG, S. B., WANG, Z. & XU, S. Y. 2007a. Optimization of the aqueous enzymatic extraction of rapeseed oil and protein hydrolysates. Journal of the American Oil Chemists' Society, 84, 97-105. ZHANG, W.-H., ZHOU, Y., DIBLEY, K. E., TYERMAN, S. D., FURBANK, R. T. & PATRICK, J. W. 2007b. Nutrient loading of developing seeds. Functional Plant Biology, 31, 314-331. ZHAO, Z. & MOGHADASIAN, M. H. 2010. Bioavailability of hydroxycinnamates: a brief review of in vivo and in vitro studies. Phytochemistry Reviews, 9, 133-145. ZHOU, K. & YU, L. L. 2004. Antioxidant properties of bran extracts from Trego wheat grown at different locations. Journal of Agricultural and Food Chemistry, 52, 1112-1117. ZIECH, D., FRANCO, R., GEORGAKILAS, A. G., GEORGAKILA, S., MALAMOU-MITSI, V., SCHONEVELD, O., PAPPA, A. & PANAYIOTIDIS, M. I. 2010. The role of reactive oxygen species and oxidative stress in environmental carcinogenesis and biomarker development. Chemico-Biological Interactions, 188, 334-339.

| References 295