Predictive Modelling of Upper Intestinal Absorption of Dietary Phytochemicals to Optimise for Health Benefits in Humans

PREDICTIVE MODELLING OF UPPER INTESTINAL ABSORPTION OF DIETARY PHYTOCHEMICALS TO OPTIMISE FOR HEALTH BENEFITS IN HUMANS

Sophie Ngoc Bich Selby-Pham

orcid.org/0000-0001-8514-3294

A thesis submitted in total fulfilment of the requirements

of the degree of Doctor of Philosophy

November 2017

Department of Agricultural and Food Systems

Faculty of Veterinary and Agricultural Sciences

The University of Melbourne PREDICTIVE MODELLING OF UPPER INTESTINAL ABSORPTION OF DIETARY PHYTOCHEMICALS TO OPTIMISE FOR HEALTH BENEFITS IN HUMANS

Abstract

Phytochemical-rich diets protect against chronic diseases such as cardiovascular diseases, diabetes and cancer via regulation of oxidative stress and inflammation (OSI). Transient

OSI also occurs in healthy bodies during exercise and meal digestion whilst persistent

OSI is associated with chronic diseases. Whilst dietary phytochemicals have been demonstrated to protect against both transient and chronic OSI, current ad hoc uses of dietary phytochemicals fail to maximise their health potential, considering their low bioavailability and transient presence in the body. This PhD project explored the absorption kinetics of phytochemicals regarding the time required to reach peak plasma concentration (Tmax) during human absorption. The main hypothesis of this project was that matching Tmax of dietary phytochemicals to the onset of OSI can achieve optimal nutritional efficacy of phytochemicals.

An in silico phytochemical absorption prediction (PCAP) model was successfully developed for prediction of Tmax of phytochemicals based on their physicochemical properties and dietary intake forms. The PCAP model comprised of two mathematical relationships including either molecular mass and lipophilicity descriptor log P, or molecular mass and polar surface area, for three dietary intake forms of liquid, semi-solid and solid. The accurate prediction of Tmax was reliable for a broad range of dietary phytochemical classes. Subsequently, research methods were developed for measuring molecular mass and log P of complex phytochemical mixtures using liquid chromatography mass spectrometry (LC-MS). The PCAP model coupled with LC-MS analysis allowed prediction of Tmax profiles of complex phytochemical mixtures, referred

PREDICTIVE MODELLING OF UPPER INTESTINAL ABSORPTION OF DIETARY PHYTOCHEMICALS TO OPTIMISE FOR HEALTH BENEFITS IN HUMANS

to as ‘functional fingerprints’. The usefulness of the functional fingerprinting was demonstrated for blueberry and green tea using published clinical studies reporting that their ability to regulate OSI in humans was only observed at post-ingestion times matching the predicted Tmax from their associated functional fingerprints.

The absorption kinetics of phytochemicals were studied in vitro using cell monolayers as a model of the human epithelium. Whilst lipophilic compounds were transported across the monolayers at faster rates in vitro than hydrophilic compounds, they were predicted by the PCAP model to require longer Tmax to accumulate peak plasma concentrations in vivo. Validation of the PCAP model was subsequently undertaken in vivo using healthy pigs as an animal model for human nutrition. Following oral intake, absorption kinetics of two selected extracts (red cabbage and grape skin) were tested via measurements of plasma Trolox equivalent antioxidant capacity (TEAC) and plasma glutathione peroxidase (GPx) activity. Both extracts increased plasma TEAC and plasma GPx activity with no clear Tmax peak detected, indicating that phytochemicals utilise both direct

(increased plasma TEAC) and indirect (increased plasma GPx activity) antioxidant mechanisms. The latter could be associated with their capacity to produce hydrogen peroxide which in turn induces the onset of cellular antioxidant defence.

This project substantiated new concepts on how understanding the absorption kinetics of dietary phytochemicals can be utilised to optimise the health benefits of plant foods. The functional fingerprinting methodology developed in this project can be applied to allow matching of absorption kinetic properties to the timing of biological need thus achieving optimal health protection of plant foods against OSI.

iii

PREDICTIVE MODELLING OF UPPER INTESTINAL ABSORPTION OF DIETARY PHYTOCHEMICALS TO OPTIMISE FOR HEALTH BENEFITS IN HUMANS

Declaration

I hereby declare that:

i. The thesis comprises only my original work towards the PhD except where

indicated in the Preface;

ii. Due acknowledgment has been made in the text to all other material used; and

iii. The thesis is fewer than 100,000 words in length, exclusive of tables, maps,

bibliographies and appendices.

SOPHIE NGOC BICH SELBY-PHAM

Date: 14/02/2018

PREDICTIVE MODELLING OF UPPER INTESTINAL ABSORPTION OF DIETARY PHYTOCHEMICALS TO OPTIMISE FOR HEALTH BENEFITS IN HUMANS

Preface

Professor Louise Bennett (CSIRO Agriculture and Food), Dr Kate Howell (The

University of Melbourne), Prof Frank Dunshea (The University of Melbourne) and Dr

Ken Ng (The University of Melbourne) were supervisors of this PhD project, and were involved in the development of overall experimental design, data interpretation and final thesis editing.

Publications included in this thesis were with collaboration with the following co-authors.

Ms Rosalind Miller (CSIRO Data61) was involved in the statistical modelling for Chapter

2. Dr Adrian Lutz (Metabolomics Australia) was involved in the LC-MS analysis for

Chapter 3. Mr Joel Ludbey (CSIRO Information Management and Technology) was involved in data visualisation for Chapter 3. Dr Simone Osborne (CSIRO Agriculture and

Food) was involved in preparation of cell culture for experiments in Chapter 4. Dr Jeremy

Cottrell (The University of Melbourne) was involved in ethic application and animal trial in Chapter 5. All co-authors read and approved the final manuscripts.

The PhD project was funded by Horticulture Innovation Australia Limited using the

Vegetable levy and funds from the Australian Government.

Part of this work has been presented in the following publications:

Chapter 2

Selby-Pham, SNB, Miller, RB, Howell, K, Dunshea, F & Bennett, LE 2017,

‘Physicochemical properties of dietary phytochemicals can predict their passive absorption in the human small intestine’, Scientific Reports, vol. 7, no. 1, p. 1931.

PREDICTIVE MODELLING OF UPPER INTESTINAL ABSORPTION OF DIETARY PHYTOCHEMICALS TO OPTIMISE FOR HEALTH BENEFITS IN HUMANS

Chapter 3

Selby-Pham, SNB, Howell, K, Dunshea, F, Ludbey, J, Lutz, A & Bennett, LE 2018,

‘Statistical modelling coupled with LC-MS analysis to predict human upper intestinal absorption of phytochemical mixtures’, Food Chemistry, vol. 245, p. 353-363.

Chapter 4

Selby-Pham, SNB, Osborne, SA, Howell, KS, Dunshea, F & Bennett, LE 2017,

‘Transport rates of dietary phytochemicals in cell monolayers is inversely correlated with absorption kinetics in humans’, Journal of Functional Foods, vol. 39, p. 206-214.

Chapter 5

Selby-Pham, SNB, Cottrell, JJ, Dunshea, F, Ng, K, Bennett, LE & Howell, K 2017,

‘Dietary phytochemicals promote health by enhancing antioxidant defence in a pig model’, Nutrients, vol. 9, no. 7, p. 758.

Part of this work has been patented:

Bennett, LE, Miller RB & Selby-Pham, SNB 2016, ‘Methods for formulating dietary foodstuffs’, Australian Provisional Patent Application no. 2014903013 and International

Patent Application no. PCT/AU2015/000461.

PREDICTIVE MODELLING OF UPPER INTESTINAL ABSORPTION OF DIETARY PHYTOCHEMICALS TO OPTIMISE FOR HEALTH BENEFITS IN HUMANS

Acknowledgements

I have had the great fortune to work with a lot of fantastic friends and colleagues throughout my PhD and I feel privileged to have worked with such supportive and knowledgeable teams. I would like to thank my supervisor and mentor, Prof. Louise

Bennett, for her advice, guidance and patience throughout my PhD. I started working with

Louise during my Masters and she has taken me under her wing and given me the opportunity to complete my PhD study with her at CSIRO. I have learned many valuable skills and lessons that will remain with me throughout my career. Thank you Louise for teaching me how to become a scientist and for encouraging me through the ‘HPLC heartbreak’.

I would like to thank my supervisor Dr Kate Howell for her support and positivity. I have very much appreciated her guidance in the development of my science communications skills. I would like to thank my supervisor Prof. Frank Dunshea for his guidance and support especially during the animal trial. Thanks you Frank for making time in your busy schedule to wear overalls over your suit and helping me with surgery of the pigs. I would also like to thank my supervisor Dr Ken Ng for his support and guidance during data interpretation.

I would like to thank the following people who had provided expertise and services during the project. I would like to thank Dr Hema Jagasothy and Dr Cheryl Taylor from CSIRO,

Katja Bitter from Berlin Institute of Technology and Jaime Ley from Thermo Fisher

Scientific for their help in setting up the HPCL analysis. I would like to thank Rosalind

Miller (CSIRO) for her expertise in statistical modelling, Dr Simone Osborne, Wei Chen

vii

PREDICTIVE MODELLING OF UPPER INTESTINAL ABSORPTION OF DIETARY PHYTOCHEMICALS TO OPTIMISE FOR HEALTH BENEFITS IN HUMANS

and Rama Addepalli (CSIRO) for their help and warm hospitality during the cell studies in Brisbane. I would like to thank Richard Phillips (CSIRO) for his excellent service with mineral analysis, Dr Thomas Leung (CSIRO) for his advice on ultrasonication techniques which was used to produce plant extracts from fresh materials. I would also like to thank

Dr Jeremy Cottrell, Maree Cox, Shannon Holbrook, Dr Peter Cakebread, Dr Ruslan

Pustovit, Udanni Wijesiriwardana, Paula Parra and Caroline Stoner for their help and warm hospitality during the animal trials at The University of Melbourne Animal House.

I would also like to thank Dr Adrian Lutz and Dr Berin Boughton at Metabolomics

Australia for their excellent services and expertise in LC-MS analysis.

Many thanks to Horticulture Innovation Australia Limited, CSIRO and The University of

Melbourne for funding this research. I’d like to acknowledge the donation of the grape skin and grape seed extracts from Tarac Technologies and the analytical services obtained from Waite Institute, South Australia.

I have spent the majority of my PhD at CSIRO in Werribee and I have very much enjoyed my time there. I would like to thank my lab mates and office mates, past and present, for their company and friendship. Thank you Claudia Phillips, Tamsyn Stanborough, Jessica

Gray, Anastasia Devi, Hema Jegasothy, Cheryl Taylor, Thu Vu, Amanda Bergamin,

Ranjit Peddapuli, Mala Gamage, Amirtha Puvanenthiran, Phil Muller, Marie Collier, Sie

Ng, Sean Moore, Peter Watkins, Diana Castree and Jared Raynes for their friendship, caring and warmth.

Last but not least, thank you to my family and friends for their continuous support and always believing in me through both the highs and lows during the candidature. I would like to thank my parents Tung Pham and Ha Do and my sister Cun Meo for their love,

viii

PREDICTIVE MODELLING OF UPPER INTESTINAL ABSORPTION OF DIETARY PHYTOCHEMICALS TO OPTIMISE FOR HEALTH BENEFITS IN HUMANS

care and ongoing support. I would also like to thank my in laws Michele, Bob and Brad

Selby for their love and encouragement. I would like to especially thank my wonderful husband Jamie Selby-Pham for his love and continuous support. Thank you for making me laugh every day and being the shoulder for me to lean on when I needed the most. I would also like to thank my friends at Absolute MMA St Kilda for keeping me active, reducing the PhD stress and celebrating small milestones of the PhD with me.

Thank you to everyone who I have had the pleasure of working with and/or getting to know during my PhD study. It has been an incredible journey which I have very much enjoyed.

PREDICTIVE MODELLING OF UPPER INTESTINAL ABSORPTION OF DIETARY PHYTOCHEMICALS TO OPTIMISE FOR HEALTH BENEFITS IN HUMANS

Table of Contents

Abstract ...... ii

Declaration ...... iv

Preface ...... v

Acknowledgements ...... vii

Table of contents ...... x

List of Tables ...... xii

List of Figures ...... xiii

Chapter 1: Introduction, Project Hypotheses and Aims...... 1 1.1 Introduction ...... 1 1.2 Regulation of oxidative stress and inflammation by dietary phytochemicals ...... 2 1.3. Bioavailability and absorption kinetics of phytochemicals ...... 5 1.4. Project hypotheses ...... 11 1.5. Research objectives ...... 11

Chapter 2: Physicochemical Properties of Dietary Phytochemicals Can Predict Their Passive Absorption in The Human Small Intestine ...... 12 2.1 Introduction ...... 12 2.2 Accepted manuscript ...... 13

Chapter 3: Statistical Modelling Coupled with LC-MS Analysis to Predict Human Upper Intestinal Absorption of Phytochemical Mixtures ...... 41 3.1 Introduction ...... 41 3.2 Accepted manuscript ...... 42

Chapter 4: Transport Rates of Dietary Phytochemicals in Cell Monolayers Is Inversely Correlated with Absorption Kinetics in Humans ...... 71 4.1 Introduction ...... 71 4.2 Accepted manuscript ...... 72

PREDICTIVE MODELLING OF UPPER INTESTINAL ABSORPTION OF DIETARY PHYTOCHEMICALS TO OPTIMISE FOR HEALTH BENEFITS IN HUMANS

Chapter 5: Dietary phytochemicals promote health by antioxidant defence in a pig model ...... 107 5.1 Introduction ...... 107 5.2 Accepted manuscript ...... 108

Chapter 6: General Discussion and Conclusion ...... 128 6.1 In silico phytochemical absorption prediction (PCAP) model ...... 129 6.2 LC-MS method for the characterisation of phytochemical mixtures ...... 130 6.3 Absorption kinetics of phytochemicals in vitro ...... 132 6.4 Absorption kinetics and modes of action of phytochemicals in vivo...... 134 6.5 Conclusion ...... 136 6.6 Future research ...... 137

References...... 141

Appendices ...... 155 Appendix I. Epidemiological studies from 2000 to 2016 reporting effects of plant-rich diets on biomarkers of oxidative stress and inflammation ...... 155 Appendix II. Clinical studies from 2000 to 2016 reporting effects of plant-rich diets on biomarkers of oxidative stress and inflammation ...... 160 Appendix III. Common biomarkers of oxidative stress and inflammation ...... 169 Appendix IV. Supporting Information for Chapter 2 ...... 171 Appendix V. Supporting Information for Chapter 3 ...... 192 Appendix VI. Supporting Information for Chapter 4 ...... 195

PREDICTIVE MODELLING OF UPPER INTESTINAL ABSORPTION OF DIETARY PHYTOCHEMICALS TO OPTIMISE FOR HEALTH BENEFITS IN HUMANS

List of Tables

Table 1.1. Frequency of observation of effects of plant-rich diets on biomarkers of oxidative stress and inflammation reported in epidemiological studies from 2000 to 2016.

...... 4

Table 1.2. Frequency of observation of effects of plant-rich diets on biomarkers of oxidative stress and inflammation reported in clinical studies from 2000 to 2016...... 5

xii

PREDICTIVE MODELLING OF UPPER INTESTINAL ABSORPTION OF DIETARY PHYTOCHEMICALS TO OPTIMISE FOR HEALTH BENEFITS IN HUMANS

List of Figures

Figure 1.1. Chemical classification of dietary phytochemicals...... 2

Figure 1.2. The time course of plasma cytokine concentration after an oxidative stress and inflammation (OSI) challenge...... 3

Figure 1.3. The pharmacokinetics of phytochemicals...... 10

xiii

Chapter 1: Introduction, Project Hypotheses and Aims

1.1 Introduction

Plant foods represent an important component of the human diet, responsible for the supply of macronutrients in the form of carbohydrates (including fibre) and micronutrients (including minerals and phytochemicals). Research has consistently shown that plant-rich diets confer human health and associated with reducing the risk of chronic diseases including diabetes (Stravodimos et al. 2016), cancer (Key 2011), cardiovascular (Dauchet, Amouyel & Dallongeville 2009) and neurodegenerative diseases (D'Onofrio et al. 2016). Daily intakes of two serves of fruits and five serves of vegetables are recommended for maintenance of good health and overall longevity

(NHMRC 2005).

The nutritional value of plant foods is related to the complex carbohydrate macronutrient fraction and the functional value is related to the micronutrient phytochemical fraction which exhibits capacity to regulate of oxidative stress and inflammation (OSI). Within the plant, the bioactive phytochemicals are non-nutritional secondary metabolites that play protective roles such as anti-microbial agents, attractants for seed-dispersing animals, UV protectants and signalling molecules (Crozier, Jaganath & Clifford 2007).

These bioactive, non-nutritional phytochemicals represent an important component of the human diet with significant potential for the development as “health-functional” ingredients, new natural drugs and antibiotics.

Phytochemicals are chemically diverse and can be divided into seven main groups including carotenoids, vitamins, phenolics, selenium compounds, nitrogen compounds, organosulfur compounds and miscellaneous group (Figure 1.1). This PhD research

Chapter 1: Introduction, Project Hypotheses and Aims

focused on carotenoids, phenolics and vitamins, due to their relative high contents in dietary plants and significant demonstration of bioactivity.

Figure 1.1. Chemical classification of dietary phytochemicals. Adapted from Liu (2004) and Mehta et al. (2010)

1.2 Regulation of oxidative stress and inflammation by dietary phytochemicals

Dietary phytochemicals have been associated with regulation of both transient and persistent OSI. Transient OSI is associated with daily activities such as meal digestion

(Burton-Freeman 2010) and exercise (van der Merwe & Bloomer 2016) in healthy individuals whilst persistent OSI is associated with chronic diseases (Calder et al. 2009).

Post-prandial lipidemia and glycaemia generate excess free radicals that can trigger a biochemical cascade (Figure 1.2), resulting in OSI and endothelial dysfunction (Esposito et al. 2002; Van Oostrom, Sijmonsma, Rabelink, et al. 2003; van Oostrom, Sijmonsma,

Verseyden, et al. 2003). Dietary phytochemicals have been demonstrated to diminish the

OSI induced by consumption of a high-fat meal (Ghanim et al. 2011; Miglio et al. 2013;

Peluso et al. 2013) or by exercise (McAnulty et al. 2013; Nieman et al. 2009) in healthy human subjects. These results indicate that regulation of OSI after meals or exercise via phytochemical intake may be a promising strategy to reduce disease risk.

Chapter 1: Introduction, Project Hypotheses and Aims

Pro-inflammatoryPro-OSI Anti-inflammatoryAnti-OSI

TNF IL-1 IL-1ra IL-10

Time after initiation (h)

Figure 1.2. The time course of plasma cytokine concentration after an oxidative stress and inflammation (OSI) challenge. Pro-OSI cytokines (TNF, tumour factor necrosis and IL-1, interleukin 1) are initially accumulated, followed by anti-OSI cytokines (IL-10, interleukin 10 and IL-1a, IL-1 receptor antagonist); adapted from Andreasen et al. (2008).

While there are considerable data suggesting the regulation of OSI by phytochemicals in vitro (Krishnaiah, Sarbatly & Nithyanandam 2011; Lee, Koo & Min 2004; Seifried et al.

2007), evidence of OSI regulation by phytochemicals in epidemiological and clinical studies is reportedly inconclusive yet promising (Barbaresko et al. 2013; D'Onofrio et al.

2016; Dell'Agli et al. 2013; Kay et al. 2012). For example, from the year 2000 to 2016, a total of 39 epidemiological (Appendix I) and 42 clinical studies (Appendix II) reported the effects of plant-rich diets on common biomarkers of OSI (Appendix III).

Approximately half of these studies reported a positive effect whilst the other half observed a neutral effect (Table 1.1 and 1.2). The inconsistency in findings of the bioefficacy of phytochemicals in human studies might be attributed to their low bioavailability and their elimination by the body, as they are recognised as xenobiotics

(Holst & Williamson 2008).

Chapter 1: Introduction, Project Hypotheses and Aims

Table 1.1. Frequency of observation of effects of plant-rich diets on biomarkers of oxidative stress and inflammation reported in epidemiological studies from 2000 to 2016. (details of studies are provided in Appendix I) Biomarkers Increase Reduce Neutral Adiponectin 3 0 4 37 51 (1 in women only (3 in women only 3 in men only CRP 1 2 in men only 1 in women; 1 in men ≥ 45 y and men < 45 y only) only) E-selectin 0 6 6 9 5 (1 in women; Fibrinogen 0 (1 in men ≥ 45 y and men < 45 y only) only) Homocysteine 0 2 1 ICAM-1 0 4 12 IL-10 1 0 3 IL-1β 0 1 0 IL-6 1 16 16 MCP-1 0 0 3 SAA 1 2 3 TNF-α 2 9 14 VCAM-1 0 3 5 WBC 1 3 4 CRP, C-reactive protein; ICAM-1, intercellular adhesion molecule 1; IL, interleukin; MCP-1, monocyte chemoattractant protein 1; SAA, serum amyloid A; TNF-α, tumour necrosis factor α; VCAM-1, vascular cell adhesion protein 1; WBC, white blood cell.

Chapter 1: Introduction, Project Hypotheses and Aims

Table 1.2. Frequency of observation of effects of plant-rich diets on biomarkers of oxidative stress and inflammation reported in clinical studies from 2000 to 2016. (details of studies are provided in Appendix II) Biomarkers Increase Reduce Neutral Adiponectin 1 0 0 CRP 0 14 15 E-selectin 0 1 5 Fibrinogen 0 1 2 GCSF 0 1 2 Homocysteine 0 4 1 ICAM-1 0 8 5 IFN-α 0 1 0 IFN-γ 0 1 1 IL-10 0 2 2 IL-13 0 1 0 IL-1α 0 1 0 IL-1β 0 2 3 IL-4 0 1 0 IL-6 1 9 11 IL-8 0 1 1 LT4 0 1 0 MCP-1 0 5 10 MIP-1β 0 2 0 NF-κB activation 0 1 0 P-selectin 0 2 2 RANTES 0 3 0 SAA 0 1 0 1 7 TNF-α (only for the first 8 (1 only after 8 wk 13 wk of intervention) of intervention) TXB2 0 1 0 VCAM-1 0 7 5 CRP, C-reactive protein; ICAM-1, intercellular adhesion molecule 1; IL, interleukin; MCP-1, monocyte chemoattractant protein 1; SAA, serum amyloid A; TNF-α, tumour necrosis factor α; VCAM-1, vascular cell adhesion protein 1; WBC, white blood cell.

1.3. Bioavailability and absorption kinetics of phytochemicals

Bioavailability of phytochemicals is defined as the fraction of compounds absorbed into the plasma after ingestion in comparison to the intake dose (Cheng, Li & Uss 2008).

Following consumption, up to 53% of ingested phytochemicals are absorbed into the blood stream via the small intestine depending on their chemical structures (Andlauer &

Fürst 2003; D’Archivio et al. 2010). These phytochemicals may be metabolised by the

Chapter 1: Introduction, Project Hypotheses and Aims

liver to release their hepatic metabolites back into circulation (Holst & Williamson 2008).

The main hepatic metabolites of phytochemicals include methylation, glucuronide and sulfate derivatives (Mullen, Edwards & Crozier 2006). The unabsorbed phytochemicals reach the large intestine and are substantially transformed by the colonic microbiota to produce microbial metabolites that can re-enter circulation (D’Archivio et al. 2010).

Chemical modifications of phytochemicals by the colonic microbiota involve dihydroxylation, hydrolysation, ring-cleavage, reduction and demethylation of both parent phytochemicals and their hepatic metabolites (Gonzalez-Barrio, Edwards &

Crozier 2011; Jaganath et al. 2006). The health benefits of phytochemicals can be attributed to the parent compounds and also their hepatic and microbial metabolites

(Donnini et al. 2006; Forester & Waterhouse 2010; Koga & Meydani 2001). However, this PhD study only focused on the parent compounds, i.e., phytochemicals that are passively absorbed in the small intestine without any chemical modification.

The absorption kinetics of phytochemicals are investigated using pharmacokinetic studies

(Khojasteh, Wong & Hop 2011). In pharmacokinetic studies, phytochemicals are administered to healthy human volunteers and blood samples are collected and analysed to quantify concentrations of phytochemicals and their metabolites in the plasma

(Khojasteh, Wong & Hop 2011). Important parameters of a pharmacokinetic profile include: (i) Cmax, the maximal concentration of a metabolite in plasma; (ii) Tmax, the time required for a metabolite to reach its Cmax in plasma and (iii) half-life T1/2, the time required for the amount of a metabolite in plasma to decrease to one half of the original amount (Figure 1.3a). Bioavailability of a phytochemical is determined by comparing the area under the curve of the pharmacokinetics of the compound by oral intake to by intravenous injection (Borgstrom et al. 1989). Considering the transient presence of

Chapter 1: Introduction, Project Hypotheses and Aims

phytochemicals and their metabolites in the human body (Holst & Williamson 2008),

Tmax is an important factor to understand and optimise the health benefits of phytochemical-rich foods.

In a recent study, the significance of phytochemical’s Tmax has been demonstrated in healthy individuals who ingested a strawberry drink at three different time points: two hours before, during or two hours after a high-fat meal (Huang et al. 2016). The effect of the strawberry drink in reducing OSI associated with the high-fat meal was only observed when the drink was ingested at two hours before the meal (Huang et al. 2016). Further, the same group reported that Tmax of the phytochemicals and their metabolites in the strawberry drink was approximately 1–2 h (Sandhu et al. 2016). Considering these observations together, it can be concluded that ingestion of the strawberry drink two hours before the high-fat meal guaranteed their presence at maximal concentrations in circulation concurrently with the OSI associated with the high-fat meal to minimise the

OSI damage induced by the meal (Burton-Freeman 2010).

Factors affecting bioavailability and absorption kinetics of phytochemicals include environmental factors, food matrix and physicochemical properties (D'Archivio et al.

2010). Environmental factors such as sun exposure, rainfall and growing temperature influence the biosynthesis of phytochemicals and subsequently the composition of the phytochemicals in the plants and their relative abundances (Gomez-Rico et al. 2006;

Manach et al. 2004). Food matrix, referring to the structural organelles containing phytochemicals in plants or the formulation in processed foods, has a pronounced effect on bioavailability of phytochemicals due to its role in the release, mass transfer and accessibility of the phytochemicals (Parada & Aguilera 2007). For example, dietary fibre may slow down digestion and absorption of phenolic acids and anthocyanidins due to the

Chapter 1: Introduction, Project Hypotheses and Aims

affinity between fibre and the phytochemicals (Padayachee et al. 2013). Similarly, milk protein has been observed to reduce bioavailability indicated by decreased area under the curve of the pharmacokinetics of chocolate phytochemicals (Serafini et al. 2003).

However in contrast, olive oil increases plasma concentrations of the tomato phytochemical lycopene by improving solubilisation (Fielding et al. 2005).

Physicochemical properties of phytochemicals have substantial impact on their bioavailability which results in a broad range of Tmax in the human body (Figure 1.3b)

(Crozier, Jaganath & Clifford 2009). For example, green tea flavan-3-ols reach maximal concentrations in human plasma 1–2 hour (h) after consumption and are eliminated from the human plasma after the next few hours (Stalmach et al. 2009). However, tomato lycopene peaks in human plasma between 15 and 33 h after consumption and is completely eliminated over the next few days (Gustin et al. 2004).

Considering the diverse Tmax ranges of phytochemicals (Figure 1.3b) and their associated presence in the human body, it is possible that the inconsistency in findings of the OSI- mediating bioefficacy of phytochemicals (Table 1 and 2) is due to blood sampling outside the timespan of Tmax. For example, neither supplementation of vitamin C for 1 day or 2 weeks had any effects on plasma biomarkers of OSI in healthy human subjects (Alessio,

Goldfarb & Cao 1997). However, administration of a single dose of vitamin C 2 h before exercise was reported to prevent exercise-associated OSI (Ashton et al. 1999).

Furthermore, the inconsistency in reported bioefficacy of vitamin C can be attributed to the time of blood sampling wherein sampling distant from the short Tmax of ~ 3 h is expected to miss detection of bioefficacy (Zetler et al. 1976). Therefore, the timing of phytochemical consumption relative to the OSI cycles associated with daily activities

Chapter 1: Introduction, Project Hypotheses and Aims

(meal or exercise) could be important to understand and optimise the health potential of dietary phytochemicals.

Chapter 1: Introduction, Project Hypotheses and Aims

Cmax Cmax Absorption phase Absorption phase Elimination phase Elimination phase (T1/2)

(T1/2)

Plasma concentration Plasma Plasma concentration Plasma

Tmax TTime1/2 after ingestion T max Time after ingestion

Figure 1.3. The pharmacokinetics of phytochemicals. (a) Pharmacokinetic parameters describing a typical plasma concentration-time profile after an oral administration of phytochemicals; Tmax, time required for a phytochemical to reach its maximum plasma concentration (Cmax); adapted from Apredica (2013). (b) Pharmacokinetics of selected phytochemicals in human subjects; adapted from Donovan et al. (2007).

Chapter 1: Introduction, Project Hypotheses and Aims

1.4. Project hypotheses

The main hypothesis of the project is that the absorption kinetics and health benefits of dietary plant foods can be understood and predicted, based on their physicochemical properties. Additionally, it was hypothesised that the bioefficacy of phytochemicals is dependent on the presence of phytochemicals in circulation and ‘bio-matching’ to onset of cycles of OSI. This PhD project involved the study of the absorption kinetics of phytochemicals using in silico modelling, in vitro and in vivo research methods, to study the time required for phytochemicals to reach their maximal plasma concentration (Tmax).

1.5. Research objectives

This study ultilised in silico, in vitro and in vivo approaches to explore the absorption kinetics of phytochemicals with an overal aim of establising a method to predict Tmax of phytochemicals during human consumption. This aim was accomplished through completion of four major research objectives:

1. To develop an in silico statistical model that predicts the Tmax of phytochemicals

in humans.

2. To develop analytical methods to characterise the Tmax profiles of phytochemical

mixtures from dietary plants.

3. To characterise the absorption kinetics of selected phytochemicals in vitro using

a cell model of epithelial transport and compare to the in silico predictions.

4. To characterise the absorption kinetics of selected phytochemicals in vivo using

a pig model and compare to the in silico predictions.

Chapter 2: Physicochemical properties of dietary phytochemicals can predict their passive absorption in the human small intestine

Chapter 2: Physicochemical Properties of Dietary Phytochemicals Can Predict Their Passive Absorption in The Human Small Intestine

2.1 Introduction

This chapter described an in silico approach to evaluating the absorption kinetics of dietary phytochemicals in healthy humans and led to identifying factors affecting the time required for a phytochemical to reach the peak plasma concentration (Tmax). In this chapter, Tmax data obtained from the literature was used to develop a predictive model that could compute Tmax from a specific selection of physicochemical properties. The predictive model, referred to as the phytochemical absorption prediction (PCAP) model, was developed using clinical studies in which Tmax was measured for 67 dietary phytochemicals spanning three dietary intake forms of liquid, semi-solid and solid.

In this chapter, the research hypothesis was that Tmax of phytochemicals can be predicted by measurable or calculable physicochemical properties and depends on dietary intake forms.

The aim of this chapter was to develop an in silico statistical model that predicted Tmax of dietary phytochemicals in humans based on specific physicochemical properties that regulate absorption, and on dietary intake forms. The aims were fulfilled via completion of the following objectives:

 Development of an in silico statistical model that predicted Tmax of dietary

phytochemicals based on key physicochemical properties of phytochemicals.

 Validation of the model using independent datasets of dietary phytochemicals

and pharmaceutical compounds, following oral intake.

Chapter 2: Physicochemical properties of dietary phytochemicals can predict their passive absorption in the human small intestine

2.2 Accepted manuscript

Physicochemical properties of dietary phytochemicals can predict their passive absorption in the human small intestine Sophie N.B. Selby-Pham1,2, Rosalind B. Miller3, Kate Howell1, Frank Dunshea1 and Louise E. Bennett2,*

1Faculty of Veterinary and Agricultural Science, The University of Melbourne, Parkville, 3010, Australia 2CSIRO Agriculture and Food, 671 Sneydes Road, Werribee, 3030, Australia 3CSIRO Data61, North Ryde, 2113, Australia *[email protected]

ABSTRACT

A diet high in phytochemical-rich plant foods is associated with reducing the risk of chronic diseases such as cardiovascular and neurodegenerative diseases, obesity, diabetes and cancer. Oxidative stress and inflammation (OSI) is the common component underlying these chronic diseases. Whilst the positive health effects of phytochemicals and their metabolites have been demonstrated to regulate OSI, the timing and absorption for best effect is not well understood. We developed a model to predict the time to achieve maximal plasma concentration (Tmax) of phytochemicals in fruits and vegetables. We used a training dataset containing 67 dietary phytochemicals from 31 clinical studies to develop the model and validated the model using three independent datasets of 108 dietary phytochemicals and 98 pharmaceutical compounds. The developed model based on dietary intake forms and the physicochemical properties lipophilicity and molecular mass accurately predicts Tmax of dietary phytochemicals and pharmaceutical compounds over a broad range of chemical classes. This is the first direct model to predict Tmax of dietary phytochemicals in the human body. The model informs the clinical dosing frequency for optimising uptake and sustained presence of dietary phytochemicals in circulation, to maximise their bio-efficacy for positively affect human health and managing OSI in chronic diseases.

Introduction

Chronic diseases are the leading causes of mortality in the world, responsible for 68% of all deaths1. Current evidence strongly supports that diets rich in plant foods are associated with reduced risk of chronic diseases such as cardiovascular2 and neurodegenerative diseases3, obesity4, diabetes5 and cancer6. Oxidative stress and inflammation (OSI) are consistently high in people suffering from chronic diseases7. These transient elevated states of OSI can also be associated with daily cycles of activity including meal digestion8 and exercise9 in healthy individuals. Ingestion of a phytochemical-rich fruit juice or grape extracts can prevent post-prandial OSI induced by a high-fat meal challenge in healthy volunteers10-12. Similarly, positive health effects of phytochemicals have been demonstrated to attenuate the OSI associated with exercise

in athletes13, 14. Accordingly, managing OSI after every meal or after exercise is likely an important strategy for reducing disease risk.

Uptake of dietary phytochemicals in the human body and their bioavailability to target cells facilitate their bio-efficacy to protect our health15. However, phytochemicals have relatively low bioavailability as they are handled by the body as xenobiotics therefore the presence in the body is transient16. Following the ingestion of phytochemicals, some but not all components are absorbed into the circulatory system via the small intestine15. These phytochemicals may be subjected to metabolism in the liver and their hepatic metabolites are released back into the circulatory system15. The phytochemicals that are not absorbed in the small intestine reach the colon whereby substantial structural modification by the colonic microbiota occurs and their microbial metabolites are released back into the circulatory system16. The main factors affecting the bioavailability of phytochemicals include chemical structures and dietary intake forms15. The chemical heterogeneity of key bioactive phytochemicals within dietary plants results in a broad range of associated time required to reach maximal plasma 17 concentration (Tmax) in the body . For example, green tea flavan-3-ols peak in human plasma within 1–2 hour (h) post ingestion and cleared over the next few hours18 whilst maximal levels of tomato lycopene was observed between 15 and 33 h post-ingestion and completely cleared over the next few days19. Additionally, dietary intake forms of 20 phytochemicals may also have an impact on their Tmax in the body . Ellagic acid from a pomegranate extract was reported to have a Tmax of 0.5–1 h when ingested as liquid form, but 2–3 h when ingested in a solid form21. It is possible that previous studies have underestimated the OSI-reducing effects of dietary phytochemicals if blood sampling was performed outside the timespan of Tmax in the body. For example, no effects of vitamin C supplementation (1 g/d) on plasma biomarkers of OSI were reported after either 1 day or 2 week treatment durations22. However, bolus dose of vitamin C given 2 h before exercise prevented exercise-induced OSI23. The inconsistency in findings of bio-efficacy of vitamin C could be due to the time of blood sampling that mismatched 24 the short Tmax of vitamin C (~ 3 h ). The timing of dietary phytochemical consumption relative to OSI challenges (e.g., meal or exercise) could be an important factor in understanding and optimising the health benefits of phytochemicals.

Oral bioavailability of phytochemicals can be informed by the application of in silico modelling widely used in pharmaceutical sciences25 and drug discovery26. These models correlate in vitro and/or in vivo passive absorption of drugs with their chemical structures described by physicochemical properties to predict the absorption of similar compounds27. Physicochemical properties of importance in drug absorption includes molecular mass (Mr), lipophilicity (expressed as the logarithm of the partition coefficient between water and 1-octanol, log P), number of hydrogen (H) donors and acceptors28, polar surface area (PSA), number of freely-rotatable bonds29 and molecular volume30. Multiple models have been developed to predict absorption kinetics and bioavailability of pharmaceutical compounds27. However, there is currently no such model for predicting Tmax of dietary phytochemicals from physicochemical properties.

The aim of this study was to determine if Tmax of dietary phytochemicals in healthy individuals could be predicted from standard physicochemical properties and dietary intake forms. To develop the predictive model, we used a training dataset that modelled the Tmax of 67 dietary phytochemicals collected from 31 clinical studies of healthy

volunteers18, 19, 21, 24, 31-57 to their calculated physicochemical properties. To validate the predictive model for dietary phytochemicals, we used an independent phytochemical validation dataset (PCv) containing 108 dietary phytochemicals collected from a further 34 clinical studies58-91. We validated the predictive model using pharmaceutical compounds and evaluated the effects of food on the prediction accuracy of the model by using two datasets containing 60 pharmaceutical compounds ingested without food (PHv-fasted)92-148 and 38 pharmaceutical compounds ingested with food (PHv-fed)92-95, 97, 98, 102-104, 106-111, 113, 116, 117, 121, 122, 126, 128, 130-133, 136, 138, 140, 143-146, 148-151. This study demonstrates that physicochemical properties and dietary intake forms can be used to predict Tmax of dietary phytochemicals and pharmaceutical compounds when ingested without food.

Results

Correlation analysis of the training dataset

The model training dataset contained 11 variables including Tmax, 8 physicochemical properties and 3 categories of dietary intake forms (Supplementary Table S1 online). The included physicochemical properties were Mr, log P, PSA, number of freely rotatable bonds, number of H donors, number of H acceptors and molecular volume. As there is a high correlation between variables, multi-collinearity affects the estimation of the coefficients and inflates the standard errors (SE). Therefore, to investigate the relationships between the physicochemical properties in the training dataset, Pearson correlation analyses were performed. Table 1 provides these Pearson’s correlation coefficients (r) with their associated P-values. Significantly high correlations (│r│> 0.75, P < 0.05) were observed between Mr and number of freely rotatable bonds (r = 0.772, P < 0.001), Mr and molecular volume (r = 0.949, P < 0.001), log P and number of H acceptors (r = -0.755, P < 0.001), number of freely rotatable bonds and molecular volume (r = 0.901, P < 0.001), number of H acceptors and H donors (r = 0.949, P < 0.001), number of H acceptors and PSA (r = 0.998, P < 0.001), number of H donors and PSA (r = 0.955, P < 0.001). For correlated variables, only one of the baseline variables was chosen to be included in the predictive model and were Mr, PSA and log P.

To test the effects of dietary intake forms, Pearson correlation analyses between Tmax, Mr, PSA and log P were performed with the inclusion of dietary intake forms (liquid, semi-solid and solid). Table 2 shows significantly high correlations between PSA and log P in the liquid intake form (r = -0.82, P < 0.001) and in the semi-solid intake form (r = -0.93, P < 0.001). Therefore, the predictive model of Tmax was developed including 2 separate models: the ‘log P model’ containing log P and Mr and the ‘PSA model’ containing PSA and Mr.

Development of the predictive model

To develop the predictive model of Tmax for phytochemicals, we used regression modelling with a natural logarithm transformation of Tmax (ln (Tmax)) and standard error (SE) of Tmax as weights to account for the uncertainty of each data points. We used the training dataset containing 67 phytochemicals collected from 31 clinical studies with a total number of 384 healthy participants (Table 3). The predictive model included 2 mathematical models: the log P model and the PSA model that appeared to

approximately equally well fit the data with coefficients depending on dietary intake forms (Fig. 1). All models had statistical power of > 0.999.

The log P model estimated Tmax based on log P and Mr (Fig. 1a–c). When phytochemicals were administered in liquid form, ln (Tmax) was positively associated with log P and Mr (Fig. 1a). When phytochemicals were administered in semi-solid (Fig. 1b) or solid (Fig. 1c) forms, ln (Tmax) was independent of Mr and followed a quadratic relationship with log P. The PSA model estimated Tmax based on PSA and Mr (Fig. 1d–f). In the PSA model, ln (Tmax) was positively associated with Mr and negatively associated with PSA. Overall, the predictive model covered a Mr range of 2 122–1270, a log P range of -4.7–9.8 and a PSA range of 0–465 Å corresponding a Tmax range of 0.3–32.6 h (Table 3). Distribution patterns of log P, Mr and PSA in the training dataset were demonstrated in Fig. 2. Log P was relatively evenly distributed across the range from -4.7–3 and 8.7–10 (Fig. 2a). Therefore, the log P model had to interpolate values between 3 and 8.5 because they were not represented in the training dataset. Mr and PSA of the training dataset were evenly distributed (Fig. 2b and 2c).

The prediction accuracy of the log P model and the PSA model in the training dataset was assessed by the root mean weighted square error normalized by the weights (RNMSWE) and the percentage relative error (%RE) of predictions (Table 4). Comparison of the measured versus predicted values of ln (Tmax) was plotted in Fig. 3a– c. The RNMSWE of prediction is an estimate of the standard deviation of the prediction normalized by the weights. As Tmax required a natural logarithm transformation, the RNMSWE in ln (hours) was transformed to %RE of prediction which is approximately average % error of Tmax (in hours) over the mean of Tmax (in hours). The %RE of prediction of the log P model was 18.27%, 19.13% and 47.08% for the liquid, semi- solid and solid intake, respectively. The %RE of prediction of the PSA model was 37.46%, 25.43% and 45.8% for the liquid, semi-solid and solid intake, respectively (Table 4). Overall, for the training dataset, despite the similar R2, the log P model had lower %RE of prediction across all three intakes and thus higher prediction accuracy.

Validation of the predictive model

To validate the predictive model, we used three independent datasets: the PCv, PHv- fasted and PHv-fed datasets. In comparison with the training dataset, all three validation datasets covered smaller ranges of log P, Mr and PSA (Table 3, Fig. 2). The PCv dataset contained phytochemicals of similar chemistry classes to the training dataset whilst the PHv-fasted and the PHv-fed datasets contains pharmaceutical compounds. The PCv dataset contained 108 phytochemicals including anthocyanins, flavanols, flavonols, hydrobenzoic acids, hydroxycinnamic acids, stilbenes, carotenoids and vitamins (Supplementary Table S2 online). Comparing to the training dataset, the PCv dataset covered a similar range of log P of -4.7–10 and measured Tmax of 0.5–37 h (Table 3) with sparsely distributed data of log P (Fig. 2a). Log P values of the PCv dataset were more concentrated in the range of -2.8–-2.5 and 1.2–2.3. Similar to the training dataset, the PCv dataset lacked log P values from 5.6–8.4 (Fig. 2a). The PCv dataset covered a 2 Mr range of 138–758 and a PSA range of 0–330 Å (Table 3, Fig. 2b and 2c). In comparison the training dataset, Mr and PSA of the PCv dataset were less evenly distributed (Fig. 2b and 2c).

To evaluate the prediction accuracy of the predictive model on the PCv dataset, we compared the measured versus predicted values of ln (Tmax) in Fig. 3d–f and calculated the %RE in Table 4. The %RE of prediction of the log P model was 55.84%, 57.07% and 76.7% for the liquid, semi-solid and solid intakes, respectively. The %RE of prediction of the PSA model was 66.07%, 92.95% and 89.4% for the liquid, semi-solid and solid intakes, respectively (Table 4). Overall, for the PCv dataset and in comparison with the PSA model, the log P model had lower %RE of prediction across three intakes and thus higher prediction accuracy. Comparing to the training dataset, the PCv dataset had higher %RE of prediction and thus lower prediction accuracy across all intake forms.

To validate the predictive model on pharmaceutical compounds, we used two pharmaceuticals validation datasets: PHv-fasted and PHv-fed. All pharmaceutical compounds in the two datasets were administered in the solid form (Table 3). The PHv- fasted dataset contains 60 compounds collected from 59 clinical studies and the PHv- fed dataset contains 38 compounds collected from 37 clinical studies (Table 3). The entire list of pharmaceutical compounds in the PHv-fasted dataset can be found as Supplementary Table S3 online and the PHv-fed dataset as Supplementary Table S4 online. The two PHv datasets covered a similar range of log P -1.7–5.4 (Table 3) with a similar distribution pattern (Fig. 2a). Comparing to the PHv-fasted dataset, the PHv-fed 2 dataset covered a slightly broader range of Mr of 123–823 and PSA of 3–221 Å while 2 the PHv-fasted dataset covered Mr range of 123–552 and PSA of 3–146 Å (Table 3). Similar distribution patterns of Mr and PSA were observed in the two PHv datasets (Fig. 2b and 2c).

To evaluate the effects of food on the prediction accuracy of the model, we compared the measured versus predicted values of ln (Tmax) in Fig. 4 and calculated the %RE in Table 4. The %RE of prediction for the log P model was 45.18% for the PHv-fasted dataset and 93.37% for the PHv-fed dataset. The %RE of prediction for the PSA model was 162.69% for the PHv-fasted dataset and 92.01% for the PHv-fed dataset (Table 4). For the log P model, food increased the %RE of prediction and therefore reduced the prediction accuracy. By contrast, for the PSA model, food reduced the %RE of prediction and thus increased the prediction accuracy. Overall, the log P model and PSA model had similar %RE for the PHv-fed dataset. However, the log P model had substantially lower %RE for the PHv-fasted dataset and thus had higher prediction accuracy.

Discussion

This is the first direct model to predict the time of maximal plasma concentration (Tmax) of dietary phytochemicals in the human body based on their physicochemical properties and dietary intake forms. The model was developed based on Tmax data from clinical studies of healthy individuals and therefore accurately describes the absorption of phytochemicals in the human body. To select the most important variables for the predictive model, we analysed the correlation between several physicochemical properties that are well known in pharmaceutical science to have significant impacts on oral bioavailability of drugs such as molecular mass, lipophilicity, polar surface area, molecular volume, number of freely rotatable bonds, number of hydrogen donors and acceptors28-30. We found significantly high correlation between some of the

physicochemical properties and selected three independent physicochemical properties to use in the model including molecular mass, lipophilicity and polar surface area. These phytochemical properties were selected due to their well-known impacts on drug bioavailability as they are related to intestinal membrane permeability of a compound28, 29. In order for a drug to cross the membrane, the compound needs to break hydrogen bonds with its aqueous environment and partition through the membrane152. Polar surface area is related to the hydrogen-bonding potential of a compound whilst molecular mass and lipophilicity are related to the membrane permeability. Consistently with the literature28, 29, 152, we found that these physicochemical properties have significant impacts on the Tmax of dietary phytochemicals in the human body. Further, dietary intake forms were identified to have an essential impact on absorption of dietary phytochemicals and were included in the model development. Similarly to drug compounds, the effects of dietary intake forms on bioavailability of phytochemicals are related to the dissolution and release of phytochemicals into the gastrointestinal tract making them available for absorption153. Therefore, comparing to the liquid form, dietary phytochemicals consumed in the semi-solid or solid forms would require longer time to dissolve into the gastrointestinal environment before they are available for absorption.

The predictive model based on lipophilicity and molecular mass provides a quantitative and high-throughput tool for prediction of Tmax of dietary phytochemicals and also pharmaceutical compounds ingested without food. Tmax of a phytochemical or pharmaceutical compound that has not been studied in vivo can simply be calculated from its molecular mass and log P for three different intake forms of liquid, semi-solid or solids using the equations reported in this predictive model (Fig. 1a–c). For example, phytochemical phloretin (Mr = 274.27, log P =2.66) found in apple would be predicted to have Tmax of 1.05, 0.62 and 1.6 h when consumed in liquid, semi-solid and solid forms, respectively. The model covers a broad range of chemical classes from phenolic compounds to carotenoids, from very hydrophilic (log P ~ -4.7) to very lipophilic (log P ~ 10) with a wide molecular mass range of Mr ~ 122–1270. The prediction accuracy of the model was indicated by relative error of prediction from 18–77% for total 175 dietary phytochemicals tested and 45% for 60 pharmaceutical compounds ingested without food (Table 4). The relative error of prediction is an indication of the total error of prediction compared to the mean. Our literature searches show that published Tmax have a SE between 0 and 200% of the mean (Supplementary Tables S1–S4 online). Therefore, the prediction accuracy of our model was deemed more accurate. Additionally, considering that a statistical power of 0.8 is the standard for adequacy154, our model with power of > 0.999 had high statistical power for accurate prediction.

The predictive model was of course limited by the literature reports of the experimental data. The Tmax variable was logarithmically transformed to alleviate the non-normality of the errors. However, there were gaps in the independent variables of log P from 3–8.5 and Mr from 750–1270 that the model had to overcome (Fig. 2). Therefore, further data covering a complete range of the parameter space would increase the rigour of the model. Additionally, we observed an increase of relative error of prediction for pharmaceutical compounds when ingested with food (Table 4). Mechanisms whereby food affects the bioavailability of drug absorption have been well studied. Food promotes absorption of lipophilic drugs due to improved drug solubilisation whilst

reducing absorption of hydrophilic drugs due to delayed drug permeation155. Similar effects of food on absorption of dietary phytochemicals have been observed20. Increased absorption of the lipophilic compound lycopene in tomato was reported when consumed with olive oil156. Hydrophilic compounds such as phenolic acids and anthocyanins were observed to bind to fibre and compromised their absorption during stimulated gastric and small intestinal digestion157. Further, protein in food has been reported to reduce absorption of dietary phytochemicals in chocolate158. Our predictive model was developed based on dietary phytochemicals administered as single-source phytochemicals or phytochemical extracts and also phytochemicals consumed in their natural matrices of whole fruits and vegetables (Supplementary Table S1 online). Apart from the models for phytochemicals consumed in liquid (Fig. 1a) or solid (Fig. 1c) forms, mostly in isolation or extracts, a statistically valid model was also developed from consumption of phytochemicals mostly (75%) in whole fruits and vegetables and accounted for the effects of these matrices on phytochemical absorption in semi-solid form (Fig. 1b). Therefore, the effects of interactions of phytochemicals with macronutrients such as fibre and protein from the natural matrices were accounted for to a small extent. Accordingly, the impact of macronutrients from food sources other than natural plant food matrices on Tmax of phytochemicals are not accounted for. Considering that macronutrients are known to interact with phytochemicals and thereby 20 alter their Tmax , the developed model may less accurately predict the Tmax of phytochemicals when consumed in conjunction with other foods. Accordingly, the predictive model reported herein is most applicable for prediction of Tmax of dietary phytochemicals and pharmaceuticals ingested without foods.

In this study, the time of maximal plasma concentration (Tmax) was chosen as the most relevant molecular data for the predictive model due to its importance in understanding and optimising the health benefits of dietary phytochemicals. Phytochemicals are treated as xenobiotic species and therefore display transient presence in circulation16. Under this circumstance, the Tmax is of prime importance in predicting the presence of any phytochemicals with the expectation that it will be substantially eliminated after a few hours or a few days depending on the phytochemicals18, 19. The protective efficacy of dietary phytochemicals can mitigate oxidative stress and inflammation (OSI) associated with daily activity and found consistently elevated in chronic diseases7-9. Managing OSI associated with daily activity is likely an important strategy for reducing disease risk in both healthy and unhealthy people. The time of maximal plasma concentration of dietary phytochemicals has recently been reported to have an important impact on their ability to regulate OSI159. Consumption of a strawberry drink 2 h before a high fat meal maximises protection against OSI compared with having the drink with 159 or 2 h after the meal , demonstrating that the Tmax of dietary phytochemicals must be 159 matched to the OSI challenge for optimal health protection . The Tmax of strawberry phytochemicals were reported to be about 1–2 h therefore consumption of the strawberry drink 2 h before the meal allowed their presence at maximal plasma concentration to reduce the OSI burden stimulated by the high fat meal160. Here, we chose Tmax instead of maximal plasma concentration (Cmax) in the predictive model as Tmax seems to be less affected by dose. For example, Tmax of lycopene was reported to 65 be about 5 h irrespective of the dose whilst Cmax increased with dose escalation . Furthermore, the anti-OSI response of phytochemicals does not necessarily continue to increase with dose and higher concentrations of phytochemicals may become pro-

161-163 oxidants and promote OSI . Without good understanding of the target Cmax for maximising phytochemical efficacy, Cmax is less useful than Tmax.

Although the study is not concerned with post-primary absorption of phytochemicals formed during hepatic and microbial metabolism, it is acknowledged that these metabolites may also contribute to the regulation of OSI similarly to their parent 164-166 compounds . Therefore, it is important to consider the reported Tmax of these derived metabolites (not predicted by the model) together with Tmax of the parent compounds predicted by this model. The main hepatic metabolites of phytochemicals are glucuronide, sulphate and methylation derivatives with short Tmax values that range from 0.5 h to up to 2.5 h42, indicative of rapid clearance by the hepatic portal system. Colonic microbiota chemical transformations of phytochemicals include hydrolysation, reduction, ring-cleavage, demethylation and dihydroxylation of both parent compounds 167, 168 and their hepatic derivatives , Accordingly, metabolites with Tmax > 5 h are likely to be absorbed or transformed with the involvement of the colon169.

The ability to predict Tmax of dietary phytochemicals offers a valuable tool for designing clinical studies to capture the time of maximal phytochemicals in the human body and to avoid underestimation of their impacts on regulation of OSI. We propose that by matching Tmax to the biological cycle of OSI, suppression of OSI is maximised and the associated tissue damage would be minimised. Therefore, the strategy for optimising the protective efficacy of dietary phytochemicals involves selection of phytochemical sources to achieve desirable Tmax that target different needs for OSI regulation. Using the unique approach of combining phytochemical-rich foods based on computable physicochemical properties, we can understand the absorption characteristics of dietary phytochemicals to achieve their full potential for protective health benefits.

Methods

Clinical data collection

Clinical measures of Tmax were obtained from the literature using the PubMed database. Information collected included compound name and family, sources, dose, intake forms and Tmax in hours (as mean ± SE, hours). When Tmax was given as median and range, conversion to mean and SE was performed as described in Hozo et al.170. The inclusion selection criteria for publications included: 1) randomised controlled clinical trials in healthy volunteers; 2) inclusion of a wash-out period when the study followed a cross over design; 3) PCs analysed were passively absorbed, i.e., compounds found in the plasma or serum were unchanged from those ingested; and 4) plasma analysed without enzymatic deconjugation.

The data collected here were included in the training dataset.

Physicochemical property data collection

Physicochemical properties of phytochemicals were calculated from the molecular structures using the Molinspiration Chemoinformatics calculator (www.molinspiration.com). The physicochemical properties calculated included Mr, log

P, PSA, number of freely rotatable bonds, number of H acceptors, number of H donors and molecular volume.

Pearson correlation analysis between variables in the training dataset

Pearson correlation analyses of all variables included in the training dataset were performed using the statistical package R version 3.3.2 171. Results were reported as Pearson’s correlation coefficient (r) and P-values.

Development of the predictive model

The predictive model was developed by a linear model framework using the statistical package R. The dependent variable Tmax required a natural logarithm transformation (ln(Tmax)) to capture the non-normality of errors in the variance across all observations of Tmax. The SE of each sample was used as weights during the regression modelling of Tmax. Because Tmax required a log normal distribution, and since:

푆퐸2(푌) 푉푎푟(푙푛(푌)) ≈ , where E(Y) = expected value of y=mean(y) (Equation 1) 퐸2(푌) the calculated weights for the regression modelling were:

1 w = ⁄ 2 (Equation 2) (푆퐸(푇푚푎푥)⁄푇푚푎푥)

When SE was missing, the weight was set to 4 and when SE was zero the weight was set to 400. Significance testing between Tmax and the physicochemical properties of PCs was carried out using multivariate regression.

Power analysis of the predictive model

Post hoc power analysis of the predictive model was performed using the power calculation program G*Power 3.1.9.2172, 173.

Validation of the predictive model

The prediction accuracy of the predictive model was validated using three independent datasets of measured Tmax obtained from clinical studies using the same selection criteria, including the PCv, PHv-fasted and PHv-fed datasets. Measured Tmax was collected as mean ± SE (hours). The prediction accuracy of the predictive model was evaluated by the normalised mean square weighted error (NMSWE) and % relative error of prediction for each dataset. The NMSWE of prediction was calculated:

푁 ̂ 2 ̂ ∑1 푤푖(푌푖−푌푖) 푁푀푆푊퐸(푌) = 푁 (Equation 3) ∑1 푤푖

Where wi is the weights calculated as in Equation 1, Yi is ln(Tmax_measured), Ŷi is ln(Tmax_predicted) and N is the number of data points.

Root NMSWE (RNMSWE) was calculated:

푅푁푀푆푊퐸 = √푁푀푆푊퐸 (Equation 4)

Let Δ=RNMSWE of prediction. If ɛ is the error in predicted values of Tmax and ln(Tmax+ɛ) is predicted from the predictive model, then:

푇푚푎푥+휀 휀 ∆≈ ln(푇푚푎푥 + 휀) − ln⁡(푇푚푎푥) ≈ ln⁡( ) ≈ 푙푛 (1 + ) (Equation 5) 푇푚푎푥 푇푚푎푥

Converting Δ (ln hours) to hours:

휀 푒∆ = 1 + (Equation 6) 푇푚푎푥

The % relative error (RE) of prediction is an approximately averaged error over all data points in the dataset:

휀 %⁡푅퐸 = ⁡× 100 = (푒∆ − 1) × 100 (Equation 7) 푇푚푎푥 References

1. Global status report on noncommunicable diseases. (World Health Organisation, Geneva; 2014). 2. Dauchet, L., Amouyel, P. & Dallongeville, J. Fruits, vegetables and coronary heart disease. Nat. Rev. Cardiol. 6, 599-608 (2009). 3. D'Onofrio, G. et al. Phytochemicals in the treatment of Alzheimer's disease: a systematic review. Curr. Drug Targets (2016). 4. Jarzab, A. & Kukula-Koch, W. Recent advances in obesity: the role of turmeric tuber and its metabolites in the prophylaxis and therapeutical strategies. Curr. Med. Chem. (2016). 5. Stravodimos, G.A. et al. Phytogenic polyphenols as glycogen phosphorylase inhibitors: the potential of triterpenes and flavonoids for glycaemic control in type 2 diabetes. Curr. Med. Chem. (2016). 6. Key, T.J. Fruit and vegetables and cancer risk. Br. J. Cancer 104, 6-11 (2011). 7. Calder, P.C. et al. Inflammatory disease processes and interactions with nutrition. Br. J. Nutr. 101 Suppl 1, S1-S45 (2009). 8. Burton-Freeman, B. Postprandial metabolic events and fruit-derived phenolics: a review of the science. Br. J. Nutr. 104 Suppl 3, S1-S14 (2010). 9. van der Merwe, M. & Bloomer, R.J. The influence of methylsulfonylmethane on inflammation-associated cytokine release before and following strenuous exercise. J. Sports Med. (Hindawi Publ. Corp.) 2016, 7498359 (2016). 10. Miglio, C. et al. Fruit juice drinks prevent endogenous antioxidant response to high-fat meal ingestion. Br. J. Nutr., 1-7 (2013). 11. Peluso, I. et al. Consumption of mixed fruit-juice drink and vitamin C reduces postprandial stress induced by a high fat meal in healthy overweight subjects. Curr. Pharm. Des., 1020-1024 (2013). 12. Ghanim, H. et al. A resveratrol and polyphenol preparation suppresses oxidative and inflammatory stress response to a high-fat, high-carbohydrate meal. J. Clin. Endocrinol. Metab. 96, 1409-1414 (2011).

13. Nieman, D.C. et al. Effects of quercetin and EGCG on mitochondrial biogenesis and immunity. Med. Sci. Sports Exerc. 41, 1467-1475 (2009). 14. McAnulty, L.S. et al. Effect of resveratrol and quercetin supplementation on redox status and inflammation after exercise. Appl. Physiol. Nutr. Metab. 38, 760-765 (2013). 15. D’Archivio, M., Filesi, C., Varì, R., Scazzocchio, B. & Masella, R. Bioavailability of the polyphenols: status and controversies. Int. J. Mol. Sci. 11, 1321-1342 (2010). 16. Holst, B. & Williamson, G. Nutrients and phytochemicals: from bioavailability to bioefficacy beyond antioxidants. Curr. Opin. Biotechnol. 19, 73-82 (2008). 17. Crozier, A., Jaganath, I.B. & Clifford, M.N. Dietary phenolics: chemistry, bioavailability and effects on health. Nat. Prod. Rep. 26, 1001-1043 (2009). 18. Stalmach, A., Troufflard, S., Serafini, M. & Crozier, A. Absorption, metabolism and excretion of Choladi green tea flavan-3-ols by humans. Mol. Nutr. Food Res. 53, S44-S53 (2009). 19. Gustin, D.M. et al. Single-dose pharmacokinetic study of lycopene delivered in a well-defined food-based lycopene delivery system (tomato paste-oil mixture) in healthy adult male subjects. Cancer Epidemiol. Biomarkers Prev. 13, 850-860 (2004). 20. Parada, J. & Aguilera, J.M. Food microstructure affects the bioavailability of several nutrients. J. Food Sci. 72, R21-32 (2007). 21. Seeram, N.P. et al. Pomegranate juice and extracts provide similar levels of plasma and urinary ellagitannin metabolites in human subjects. J. Med. Food 11, 390-394 (2008). 22. Alessio, H.M., Goldfarb, A.H. & Cao, G. Exercise-induced oxidative stress before and after vitamin C supplementation. Int. J. Sport Nutr. 7, 1-9 (1997). 23. Ashton, T. et al. Electron spin resonance spectroscopy, exercise, and oxidative stress: an ascorbic acid intervention study. J. Appl. Physiol. 87, 2032-2036 (1999). 24. Zetler, G., Seidel, G., Siegers, C.P. & Iven, H. Pharmacokinetics of ascorbic acid in man. Eur. J. Clin. Pharmacol. 10, 273-282 (1976). 25. van de Waterbeemd, H. & Gifford, E. ADMET in silico modelling: towards prediction paradise? Nat. Rev. Drug. Discov. 2, 192-204 (2003). 26. Camp, D., Campitelli, M., Carroll, A.R., Davis, R.A. & Quinn, R.J. Front- loading natural-product-screening libraries for log P: background, development, and implementation. Chem. Biodivers. 10, 524-537 (2013). 27. Van de Waterbeemd, H. From in vivo to in vitro/in silico ADME: progress and challenges. Expert Opin. Drug Metab. Toxicol. 1, 1-4 (2005). 28. Lipinski, C.A., Lombardo, F., Dominy, B.W. & Feeney, P.J. Experimental and computational approaches to estimate solubility and permeability in drug discovery and development settings. Adv. Drug Deliv. Rev. 23, 3-26 (1997). 29. Veber, D.F. et al. Molecular properties that influence the oral bioavailability of drug candidates. J. Med. Chem. 45, 2615-2623 (2002). 30. Abraham, M.H., Chadha, H.S. & Mitchell, R.C. Hydrogen bonding. 33. Factors that influence the distribution of solutes between blood and brain. J. Pharm. Sci. 83, 1257-1268 (1994). 31. Kay, C.D., Mazza, G. & Holub, B.J. Anthocyanins exist in the circulation primarily as metabolites in adult men. J. Nutr. 135, 2582-2588 (2005).

32. Boocock, D.J. et al. Phase I dose escalation pharmacokinetic study in healthy volunteers of resveratrol, a potential cancer chemopreventive agent. Cancer Epidemiol. Biomarkers Prev. 16, 1246-1252 (2007). 33. Castelli, M.C., Wong, D.F., Friedman, K. & Riley, M.G. Pharmacokinetics of oral cyanocobalamin formulated with sodium N-[8-(2- hydroxybenzoyl)amino]caprylate (SNAC): an open-label, randomized, single- dose, parallel-group study in healthy male subjects. Clin. Ther. 33, 934-945 (2011). 34. Del Rio, D. et al. Bioavailability and catabolism of green tea flavan-3-ols in humans. Nutrition 26, 1110-1116 (2010). 35. Ferslew, K.E., Acuff, R.V., Daigneault, E.A., Woolley, T.W. & Stanton, P.E. Pharmacokinetics and bioavailability of the RRR and all racemic stereoisomers of alpha-tocopherol in humans after single oral administration. J. Clin. Pharmacol. 33, 84-88 (1993). 36. Garcia-Alonso, M., Minihane, A.M., Rimbach, G., Rivas-Gonzalo, J.C. & de Pascual-Teresa, S. Red wine anthocyanins are rapidly absorbed in humans and affect monocyte chemoattractant protein 1 levels and antioxidant capacity of plasma. J. Nutr. Biochem. 20, 521-529 (2009). 37. Hartmann, D. et al. Plasma kinetics of zeaxanthin and 3′-dehydro-lutein after multiple oral doses of synthetic zeaxanthin. Am. J. Clin. Nutr. 79, 410-417 (2004). 38. Jia, Y. et al. Pharmacokinetics of single-dose and multi-dose of lovastatin/niacin ER tablet in health volunteers. Chromatogr. Res. Int., 1-11 (2012). 39. Mascher, H. & Kikuta, C. High-performance liquid chromatographic determination of total thiamine in human plasma for oral bioavailability studies. J. Pharm. Sci. 82, 56-59 (1993). 40. Mertens-Talcott, S.U., Jilma-Stohlawetz, P., Rios, J., Hingorani, L. & Derendorf, H. Absorption, metabolism, and antioxidant effects of pomegranate (Punica granatum l.) polyphenols after ingestion of a standardized extract in healthy human volunteers. J. Agric. Food Chem. 54, 8956-8961 (2006). 41. Moon, Y.J., Wang, L., DiCenzo, R. & Morris, M.E. Quercetin pharmacokinetics in humans. Biopharm. Drug Dispos. 29, 205-217 (2008). 42. Mullen, W., Edwards, C.A. & Crozier, A. Absorption, excretion and metabolite profiling of methyl-, glucuronyl-, glucosyl-and sulpho-conjugates of quercetin in human plasma and urine after ingestion of onions. Br. J. Nutr. 96, 107-116 (2006). 43. Novotny, J.A., Kurilich, A.C., Britz, S.J., Baer, D.J. & Clevidence, B.A. Vitamin K absorption and kinetics in human subjects after consumption of 13C- labelled phylloquinone from kale. Br. J. Nutr. 104, 858-862 (2010). 44. Odeberg, J.M., Lignell, A., Pettersson, A. & Hoglund, P. Oral bioavailability of the antioxidant astaxanthin in humans is enhanced by incorporation of lipid based formulations. Eur. J. Pharm. Sci. 19, 299-304 (2003). 45. Østerlie, M., Bjerkeng, B. & Liaaen-Jensen, S. Plasma appearance and distribution of astaxanthin E/Z and R/S isomers in plasma lipoproteins of men after single dose administration of astaxanthin. J. Nutr. Biochem. 11, 482-490 (2000).

46. Paetau, I., Chen, H., Goh, N.M. & White, W.S. Interactions in the postprandial appearance of beta-carotene and canthaxanthin in plasma triacylglycerol-rich lipoproteins in humans. Am. J. Clin. Nutr. 66, 1133-1143 (1997). 47. Parker, R.S., J., T.B., Swanson, J.E., Goodman, K.J. & Marmor, B. in Methods Enzymol., Vol. Volume 282. (ed. J.W.S.C.W. Donald B. McCormick) 130-140 (Academic Press, 1997). 48. Petley, A., Macklin, B., Renwick, A.G. & Wilkin, T.J. The pharmacokinetics of nicotinamide in humans and rodents. Diabetes 44, 152-155 (1995). 49. Rechner, A.R. et al. The metabolism of dietary polyphenols and the relevance to circulating levels of conjugated metabolites. Free Radic. Res. 36, 1229-1241 (2002). 50. Seeram, N.P. et al. Pomegranate juice ellagitannin metabolites are present in human plasma and some persist in urine for up to 48 hours. J. Nutr. 136, 2481- 2485 (2006). 51. Stalmach, A. et al. Metabolite profiling of hydroxycinnamate derivatives in plasma and urine after the ingestion of coffee by humans: identification of biomarkers of coffee consumption. Drug Metab. Dispos. 37, 1749-1758 (2009). 52. Stoner, G.D. et al. Pharmacokinetics of anthocyanins and ellagic acid in healthy volunteers fed freeze-dried black raspberries daily for 7 days. J. Clin. Pharmacol. 45, 1153-1164 (2005). 53. Tallaksen, C.M.E., Sande, A., Bøhmer, T., Bell, H. & Karlsen, J. Kinetics of thiamin and thiamin phosphate esters in human blood, plasma and urine after 50 mg intravenously or orally. Eur. J. Clin. Pharmacol. 44, 73-78 (1993). 54. Vitaglione, P. et al. Protocatechuic acid is the major human metabolite of cyanidin-glucosides. J. Nutr. 137, 2043-2048 (2007). 55. Yao, L., Liang, Y., Trahanovsky, W.S., Serfass, R.E. & White, W.S. Use of a 13C tracer to quantify the plasma appearance of a physiological dose of lutein in humans. Lipids 35, 339-348 (2000). 56. Yap, S.P., Yuen, K.H. & Wong, J.W. Pharmacokinetics and bioavailability of alpha-, gamma- and delta-tocotrienols under different food status. J. Pharm. Pharmacol. 53, 67-71 (2001). 57. Zempleni, J., Galloway, J.R. & McCormick, D.B. Pharmacokinetics of orally and intravenously administered riboflavin in healthy humans. Am. J. Clin. Nutr. 63, 54-66 (1996). 58. Azachi, M., Yatuv, R., Katz, A., Hagay, Y. & Danon, A. A novel red grape cells complex: health effects and bioavailability of natural resveratrol. Int. J. Food Sci. Nutr. 65, 848-855 (2014). 59. Barnett, C.F. et al. Pharmacokinetic, partial pharmacodynamic and initial safety analysis of (-)-epicatechin in healthy volunteers. Food Funct. 6, 824-833 (2015). 60. Bianchi, M.L. et al. No difference in intestinal strontium absorption after an oral or an intravenous 1,25(OH)2D3 bolus in normal subjects. For the European Study Group on Vitamin D in children with renal failure. J. Bone Miner. Res. 14, 1789-1795 (1999). 61. Bowen, P.E., Herbst-Espinosa, S.M., Hussain, E.A. & Stacewicz-Sapuntzakis, M. Esterification does not impair lutein bioavailability in humans. J. Nutr. 132, 3668-3673 (2002).

62. Cao, G., Muccitelli, H.U., Sanchez-Moreno, C. & Prior, R.L. Anthocyanins are absorbed in glycated forms in elderly women: a pharmacokinetic study. Am. J. Clin. Nutr. 73, 920-926 (2001). 63. Carr, A.C., Bozonet, S.M. & Vissers, M.C. A randomised cross-over pharmacokinetic bioavailability study of synthetic versus kiwifruit-derived vitamin C. Nutrients 5, 4451-4461 (2013). 64. Czank, C. et al. Human metabolism and elimination of the anthocyanin, cyanidin-3-glucoside: a (13)C-tracer study. Am. J. Clin. Nutr. 97, 995-1003 (2013). 65. Diwadkar-Navsariwala, V. et al. A physiological pharmacokinetic model describing the disposition of lycopene in healthy men. J. Lipid Res. 44, 1927- 1939 (2003). 66. Frank, T., Netzel, M., Strass, G., Bitsch, R. & Bitsch, I. Bioavailability of anthocyanidin-3-glucosides following consumption of red wine and red grape juice. Can. J. Physiol. Pharmacol. 81, 423-435 (2003). 67. Frank, T. et al. Pharmacokinetics of anthocyanidin-3-glycosides following consumption of Hibiscus sabdariffa L. extract. J. Clin. Pharmacol. 45, 203-210 (2005). 68. González-Sarrías, A. et al. Identifying the limits for ellagic acid bioavailability: a crossover pharmacokinetic study in healthy volunteers after consumption of pomegranate extracts. J. Funct. Foods 19, Part A, 225-235 (2015). 69. Hashimoto, T. et al. Pharmacokinetics of fucoxanthinol in human plasma after the oral administration of kombu extract. Br. J. Nutr. 107, 1566-1569 (2012). 70. Haustein, K.O. & Huller, G. Pharmacokinetics of phenprocoumon. Int. J. Clin. Pharmacol. Ther. 32, 192-197 (1994). 71. Keane, K.M. et al. Phytochemical uptake following human consumption of Montmorency tart cherry (L. Prunus cerasus) and influence of phenolic acids on vascular smooth muscle cells in vitro. Eur. J. Nutr. 55, 1695-1705 (2016). 72. Kostic, D., White, W.S. & Olson, J.A. Intestinal absorption, serum clearance, and interactions between lutein and beta-carotene when administered to human adults in separate or combined oral doses. Am. J. Clin. Nutr. 62, 604-610 (1995). 73. Kuntz, S. et al. Uptake and bioavailability of anthocyanins and phenolic acids from grape/blueberry juice and smoothie in vitro and in vivo. Br. J. Nutr. 113, 1044-1055 (2015). 74. Ludwig, I.A. et al. New insights into the bioavailability of red raspberry anthocyanins and ellagitannins. Free Radic. Biol. Med. 89, 758-769 (2015). 75. Matsumoto, H. et al. Orally administered delphinidin 3-rutinoside and cyanidin 3-rutinoside are directly absorbed in rats and humans and appear in the blood as the intact forms. J. Agric. Food. Chem. 49, 1546-1551 (2001). 76. McKay, D.L., Chen, C.Y., Zampariello, C.A. & Blumberg, J.B. Flavonoids and phenolic acids from cranberry juice are bioavailable and bioactive in healthy older adults. Food Chem. 168, 233-240 (2015). 77. Mertens-Talcott, S.U. et al. Pharmacokinetics of anthocyanins and antioxidant effects after the consumption of anthocyanin-rich acai juice and pulp (Euterpe oleracea Mart.) in human healthy volunteers. J. Agric. Food Chem. 56, 7796- 7802 (2008).

78. Milbury, P.E., Vita, J.A. & Blumberg, J.B. Anthocyanins are bioavailable in humans following an acute dose of cranberry juice. J. Nutr. 140, 1099-1104 (2010). 79. Miller, R.J. et al. A preliminary investigation of the impact of catechol-O- methyltransferase genotype on the absorption and metabolism of green tea catechins. Eur. J. Nutr. 51, 47-55 (2012). 80. Narumi, K. et al. Simultaneous detection of green tea catechins and gallic acid in human serum after ingestion of green tea tablets using ion-pair high- performance liquid chromatography with electrochemical detection. J. Chromatogr. B Analyt. Technol. Biomed. Life Sci. 945–946, 147-153 (2014). 81. Nunes, T. et al. Pharmacokinetics of trans-resveratrol following repeated administration in healthy elderly and young subjects. J. Clin. Pharmacol. 49, 1477-1482 (2009). 82. Renouf, M. et al. Plasma pharmacokinetics of catechin metabolite 4'-O-Me-EGC in healthy humans. Eur. J. Nutr. 50, 575-580 (2011). 83. Rotches-Ribalta, M., Andres-Lacueva, C., Estruch, R., Escribano, E. & Urpi- Sarda, M. Pharmacokinetics of resveratrol metabolic profile in healthy humans after moderate consumption of red wine and grape extract tablets. Pharmacol. Res. 66, 375-382 (2012). 84. Schramm, D.D. et al. Food effects on the absorption and pharmacokinetics of cocoa flavanols. Life Sci. 73, 857-869 (2003). 85. Seymour, E.M. et al. Anthocyanin pharmacokinetics and dose-dependent plasma antioxidant pharmacodynamics following whole tart cherry intake in healthy humans. J. Funct. Foods 11, 509-516 (2014). 86. Song, Q. et al. HPLC–APCI–MS for the determination of vitamin K1 in human plasma: method and clinical application. J. Chromatogr. B Analyt. Technol. Biomed. Life Sci. 875, 541-545 (2008). 87. Stalmach, A. et al. Absorption, metabolism, and excretion of green tea flavan-3- ols in humans with an ileostomy. Mol. Nutr. Food Res. 54, 323-334 (2010). 88. Stalmach, A., Edwards, C.A., Wightman, J.D. & Crozier, A. Gastrointestinal stability and bioavailability of (poly)phenolic compounds following ingestion of Concord grape juice by humans. Mol. Nutr. Food Res. 56, 497-509 (2012). 89. Steinberg, F.M., Holt, R.R., Schmitz, H.H. & Keen, C.L. Cocoa procyanidin chain length does not determine ability to protect LDL from oxidation when monomer units are controlled. J. Nutr. Biochem. 13, 645-652 (2002). 90. Umigai, N. et al. The pharmacokinetic profile of crocetin in healthy adult human volunteers after a single oral administration. Phytomedicine 18, 575-578 (2011). 91. Zhi, J., Melia, A.T., Koss-Twardy, S.G., Arora, S. & Patel, I.H. The effect of orlistat, an inhibitor of dietary fat absorption, on the pharmacokinetics of β- carotene in healthy volunteers. J. Clin. Pharmacol. 36, 152-159 (1996). 92. Auclair, B., Nix, D.E., Adam, R.D., James, G.T. & Peloquin, C.A. Pharmacokinetics of ethionamide administered under fasting conditions or with orange juice, food, or antacids. Antimicrob. Agents Chemother. 45, 810-814 (2001). 93. Audet, P.R., Chiang, S., Maroli, A., Locniskar, A. & Morrison, G. The effect of food on ciramadol bioavailability in normal subjects. Biopharm. Drug Dispos. 8, 299-304 (1987).

94. Barbhaiya, R.H., Shukla, U.A., Gleason, C.R., Shyu, W.C. & Pittman, K.A. Comparison of the effects of food on the pharmacokinetics of cefprozil and cefaclor. Antimicrob. Agents Chemother. 34, 1210-1213 (1990). 95. Barone, J.A. et al. Food interaction and steady-state pharmacokinetics of itraconazole capsules in healthy male volunteers. Antimicrob. Agents Chemother. 37, 778-784 (1993). 96. Bastami, S. et al. Pharmacogenetic aspects of tramadol pharmacokinetics and pharmacodynamics after a single oral dose. Forensic Sci. Int. 238, 125-132 (2014). 97. Bekersky, I., Dressler, D. & Mekki, Q.A. Effect of low- and high-fat meals on tacrolimus absorption following 5 mg single oral doses to healthy human subjects. J. Clin. Pharmacol. 41, 176-182 (2001). 98. Bramer, S.L. & Forbes, W.P. Relative bioavailability and effects of a high fat meal on single dose cilostazol pharmacokinetics. Clin. Pharmacokinet. 37, 13- 23 (1999). 99. Bu, H.Z. et al. Metabolism and excretion of capravirine, a new non-nucleoside reverse transcriptase inhibitor, alone and in combination with ritonavir in healthy volunteers. Drug Metab. Dispos. 32, 689-698 (2004). 100. Chang, X.Y., Guo, T. & Xia, D.Y. Pharmacokinetics of tinidazole in Chinese subjects: comparison of Mongolian, Korean, Hui, Uighur and Han nationalities. J. Pharm. Pharm. Sci. 12, 175-180 (2009). 101. Chen, R. et al. A more rapid, sensitive, and specific HPLC-MS/MS method for nifedipine analysis in human plasma and application to a pharmacokinetic study. Drug Res. 63, 38-45 (2013). 102. Cheng, H. et al. Pharmacokinetics and food interaction of MK-462 in healthy males. Biopharm. Drug Dispos. 17, 17-24 (1996). 103. de Campos, D.R. et al. Bioequivalence of two enteric coated formulations of pantoprazole in healthy volunteers under fasting and fed conditions. Arzneimittel-Forschung 57, 309-314 (2007). 104. Dingemanse, J., Bodin, F., Weidekamm, E., Kutz, K. & van Giersbergen, P. Influence of food intake and formulation on the pharmacokinetics and metabolism of bosentan, a dual endothelin receptor antagonist. J. Clin. Pharmacol. 42, 283-289 (2002). 105. Doose, D.R., Walker, S.A., Gisclon, L.G. & Nayak, R.K. Single-dose pharmacokinetics and effect of food on the bioavailability of topiramate, a novel antiepileptic drug. J. Clin. Pharmacol. 36, 884-891 (1996). 106. Eshelman, F.N. & Spyker, D.A. Pharmacokinetics of amoxicillin and ampicillin: crossover study of the effect of food. Antimicrob. Agents Chemother. 14, 539- 543 (1978). 107. Francis, R.J., Dixon, J.S., Lowe, J.R. & Harris, P.A. The effects of food and of antacid on the single oral dose pharmacokinetics of tenoxicam. Eur. J. Drug Metab. Pharmacokinet. 10, 309-314 (1985). 108. Frost, C. et al. Apixaban, an oral, direct factor Xa inhibitor: single dose safety, pharmacokinetics, pharmacodynamics and food effect in healthy subjects. Br. J. Clin. Pharmacol. 75, 476-487 (2013). 109. Hamelin, B.A. et al. The effect of timing of a standard meal on the pharmacokinetics and pharmacodynamics of the novel atypical antipsychotic agent ziprasidone. Pharmacotherapy 18, 9-15 (1998).

110. Herrstedt, J., Jorgensen, M. & Angelo, H.R. The effect of food on serum concentrations of metopimazine. Br. J. Clin. Pharmacol. 30, 237-243 (1990). 111. Jang, Y.J. et al. Prandial effect on the systemic exposure of amisulpride. Arch. Pharm. Res., 1325-1328 (2014). 112. Kassahun, K. et al. Disposition and biotransformation of the antipsychotic agent olanzapine in humans. Drug Metab. Dispos. 25, 81-93 (1997). 113. Khosravan, R., Grabowski, B., Wu, J.T., Joseph-Ridge, N. & Vernillet, L. Effect of food or antacid on pharmacokinetics and pharmacodynamics of febuxostat in healthy subjects. Br. J. Clin. Pharmacol. 65, 355-363 (2008). 114. Lee, L. et al. The safety, tolerability, pharmacokinetics, and pharmacodynamics of single oral doses of RO5068760, an MEK inhibitor, in healthy volunteers: assessment of target suppression. J. Clin. Pharmacol. 50, 1397-1405 (2010). 115. Lee, L. et al. The safety, tolerability, pharmacokinetics, and pharmacodynamics of single oral doses of CH4987655 in healthy volunteers: target suppression using a biomarker. Clin. Cancer Res. 15, 7368-7374 (2009). 116. Lettieri, J., Vargas, R., Agarwal, V. & Liu, P. Effect of food on the pharmacokinetics of a single oral dose of moxifloxacin 400 mg in healthy male volunteers. Clin. Pharmacokinet. 40 Suppl 1, 19-25 (2001). 117. Li, Y. et al. Pharmacokinetic properties of rosuvastatin after single-dose, oral administration in Chinese volunteers: a randomized, open-label, three-way crossover study. Clin. Ther. 29, 2194-2203 (2007). 118. Lindholm, A., Henricsson, S. & Dahlqvist, R. The effect of food and bile acid administration on the relative bioavailability of cyclosporin. Br. J. Clin. Pharmacol. 29, 541-548 (1990). 119. Liu, M.Y. et al. Pharmacokinetic properties and bioequivalence of two formulations of arbidol: an open-label, single-dose, randomized-sequence, two- period crossover study in healthy Chinese male volunteers. Clin. Ther. 31, 784- 792 (2009). 120. Mahatthanatrakul, W. et al. Effect of cytochrome P450 3A4 inhibitor ketoconazole on risperidone pharmacokinetics in healthy volunteers. J. Clin. Pharm. Ther. 37, 221-225 (2012). 121. Mannisto, P.T., Mantyla, R., Nykanen, S., Lamminsivu, U. & Ottoila, P. Impairing effect of food on ketoconazole absorption. Antimicrob. Agents Chemother. 21, 730-733 (1982). 122. Melander, A., Wahlin, E., Danielson, K. & Rerup, C. On the influence of concomitant food intake on sulfonamide bioavailability. Acta Med. Scand. 200, 497-500 (1976). 123. Misaka, S. et al. Green tea ingestion greatly reduces plasma concentrations of nadolol in healthy subjects. Clin. Pharmacol. Ther. 95, 432-438 (2014). 124. Noh, Y.H. et al. Assessment of the influence of severe renal impairment on the pharmacokinetics of mirodenafil in Korean male volunteers. Int. J. Clin. Pharmacol. Ther. 50, 880-888 (2012). 125. Novalbos, J. et al. Effects of CYP2D6 genotype on the pharmacokinetics, pharmacodynamics, and safety of risperidone in healthy volunteers. J. Clin. Psychopharmacol. 30, 504-511 (2010). 126. Ochs, H.R., Ramsch, K.D., Verburg-Ochs, B., Greenblatt, D.J. & Gerloff, J. Nifedipine: kinetics and dynamics after single oral doses. Klin. Wochenschr. 62, 427-429 (1984).

127. Ohmori, S., Miura, M., Toriumi, C., Satoh, Y. & Ooie, T. Absorption, metabolism, and excretion of [14C]imidafenacin, a new compound for treatment of overactive bladder, after oral administration to healthy male subjects. Drug Metab. Dispos. 35, 1624-1633 (2007). 128. Overdiek, H.W. & Merkus, F.W. Influence of food on the bioavailability of spironolactone. Clin. Pharmacol. Ther. 40, 531-536 (1986). 129. Parrillo-Campiglia, S. et al. Bioequivalence of two film-coated tablets of imatinib mesylate 400 mg: a randomized, open-label, single-dose, fasting, two- period, two-sequence crossover comparison in healthy male South American volunteers. Clin. Ther. 31, 2224-2232 (2009). 130. Paulson, S.K. et al. Pharmacokinetics of celecoxib after oral administration in dogs and humans: effect of food and site of absorption. J. Pharmacol. Exp. Ther. 297, 638-645 (2001). 131. Peloquin, C.A. et al. Pharmacokinetics of pyrazinamide under fasting conditions, with food, and with antacids. Pharmacotherapy 18, 1205-1211 (1998). 132. Peloquin, C.A., Namdar, R., Singleton, M.D. & Nix, D.E. Pharmacokinetics of rifampin under fasting conditions, with food, and with antacids. Chest 115, 12- 18 (1999). 133. Peloquin, C.A. et al. Pharmacokinetics of ethambutol under fasting conditions, with food, and with antacids. Antimicrob. Agents Chemother. 43, 568-572 (1999). 134. Shi, J.G. et al. The effect of CYP3A4 inhibition or induction on the pharmacokinetics and pharmacodynamics of orally administered ruxolitinib (INCB018424 phosphate) in healthy volunteers. J. Clin. Pharmacol. 52, 809- 818 (2012). 135. Soyinka, J.O. & Onyeji, C.O. Alteration of pharmacokinetics of proguanil in healthy volunteers following concurrent administration of efavirenz. Eur. J. Pharm. Sci. 39, 213-218 (2010). 136. Stoch, S.A. et al. Odanacatib, a selective cathepsin K inhibitor to treat osteoporosis: safety, tolerability, pharmacokinetics and pharmacodynamics-- results from single oral dose studies in healthy volunteers. Br. J. Clin. Pharmacol. 75, 1240-1254 (2013). 137. Sturm, S. et al. Metabolism, excretion, and pharmacokinetics of [14c]- radiolabeled aleglitazar: a phase I, nonrandomized, open-label, single-center, single-dose study in healthy male volunteers. Clin. Ther. 34, 420-429 (2012). 138. Tenero, D. et al. Pharmacokinetics of intravenously and orally administered eprosartan in healthy males: absolute bioavailability and effect of food. Biopharm. Drug Dispos. 19, 351-356 (1998). 139. Usman, M. et al. Comparative pharmacokinetics of levofloxacin in healthy volunteers and in patients suffering from typhoid fever. Iran J. Pharm. Res. 12, 147-154 (2013). 140. van Marle, S., van Vliet, A., Sollie, F., Kambayashi, Y. & Yamada-Sawada, T. Safety, tolerability and pharmacokinetics of oral S-3304, a novel matrix metalloproteinase inhibitor, in single and multiple dose escalation studies in healthy volunteers. Int. J. Clin. Pharmacol. Ther. 43, 282-293 (2005). 141. O'Farrell, A.M. et al. Pharmacokinetic and pharmacodynamic assessments of the dipeptidyl peptidase-4 inhibitor PHX1149: double-blind, placebo-controlled,

single- and multiple-dose studies in healthy subjects. Clin. Ther. 29, 1692-1705 (2007). 142. Yan, M. et al. Lack of effect of continuous glycyrrhizin administration on the pharmacokinetics of the P-glycoprotein substrate talinolol in healthy volunteers. Eur. J. Clin. Pharmacol. 69, 515-521 (2013). 143. Yeh, K.C. et al. Single-dose pharmacokinetics of indinavir and the effect of food. Antimicrob. Agents Chemother. 42, 332-338 (1998). 144. Zhang, J. et al. The effects of food on the pharmacokinetics of mitiglinide tablets in healthy volunteers and a novel mass-spectrometric (UPLC-MS/MS) method for such studies. J. Clin. Pharm. Ther. 37, 95-99 (2012). 145. Zhou, X., Lloyd, D.M., Chao, G.C. & Brown, N.A. Absence of food effect on the pharmacokinetics of telbivudine following oral administration in healthy subjects. J. Clin. Pharmacol. 46, 275-281 (2006). 146. Zimmermann, T., Yeates, R.A., Laufen, H., Pfaff, G. & Wildfeuer, A. Influence of concomitant food intake on the oral absorption of two triazole antifungal agents, itraconazole and fluconazole. Eur. J. Clin. Pharmacol. 46, 147-150 (1994). 147. Zollinger, M. et al. Absorption, distribution, metabolism, and excretion (ADME) of C-sonidegib (LDE225) in healthy volunteers. Cancer Chemother. Pharmacol., 63-75 (2014). 148. Peloquin, C.A., Namdar, R., Dodge, A.A. & Nix, D.E. Pharmacokinetics of isoniazid under fasting conditions, with food, and with antacids. Int. J. Tuberc. Lung Dis. 3, 703-710 (1999). 149. Courtney, R., Wexler, D., Radwanski, E., Lim, J. & Laughlin, M. Effect of food on the relative bioavailability of two oral formulations of posaconazole in healthy adults. Br. J. Clin. Pharmacol. 57, 218-222 (2004). 150. Dien, T.K. et al. Effect of food intake on pharmacokinetics of oral artemisinin in healthy Vietnamese subjects. Antimicrob. Agents Chemother. 41, 1069-1072 (1997). 151. Williams, D.D. et al. Effects of food and antacids on the pharmacokinetics of eltrombopag in healthy adult subjects: two single-dose, open-label, randomized- sequence, crossover studies. Clin. Ther. 31, 764-776 (2009). 152. van de Waterbeemd, H., Smith, D.A. & Jones, B.C. Lipophilicity in PK design: methyl, ethyl, futile. J. Comput. Aided Mol. Des. 15, 273-286 (2001). 153. Dressman, J.B., Amidon, G.L., Reppas, C. & Shah, V.P. Dissolution testing as a prognostic tool for oral drug absorption: immediate release dosage forms. Pharm. Res. 15, 11-22 (1998). 154. MacCallum, R.C., Browne, M.W. & Sugawara, H.M. Power analysis and determination of sample size for covariance structure modeling. Psychol. Methods 1, 130-149 (1996). 155. Gu, C.H. et al. Predicting effect of food on extent of drug absorption based on physicochemical properties. Pharm. Res. 24, 1118-1130 (2007). 156. Fielding, J.M., Rowley, K.G., Cooper, P. & O' Dea, K. Increases in plasma lycopene concentration after consumption of tomatoes cooked with olive oil. Asia Pac. J. Clin. Nutr. 14, 131-136 (2005). 157. Padayachee, A. et al. Lack of release of bound anthocyanins and phenolic acids from carrot plant cell walls and model composites during simulated gastric and small intestinal digestion. Food Funct. 4, 906-916 (2013).

158. Serafini, M. et al. Plasma antioxidants from chocolate. Nature 424, 1013-1013 (2003). 159. Huang, Y., Park, E., Edirisinghe, I. & Burton-Freeman, B.M. Maximizing the health effects of strawberry anthocyanins: understanding the influence of the consumption timing variable. Food Funct. (2016). 160. Sandhu, A.K. et al. Pharmacokinetic characterization and bioavailability of strawberry anthocyanins relative to meal intake. J. Agric. Food Chem. 64, 4891- 4899 (2016). 161. Azam, S., Hadi, N., Khan, N.U. & Hadi, S.M. Prooxidant property of green tea polyphenols epicatechin and epigallocatechin-3-gallate: implications for anticancer properties. Toxicol. In Vitro 18, 555-561 (2004). 162. Raza, H. & John, A. Green tea polyphenol epigallocatechin-3-gallate differentially modulates oxidative stress in PC12 cell compartments. Toxicol. Appl. Pharmacol. 207, 212-220 (2005). 163. Watjen, W. et al. Low concentrations of flavonoids are protective in rat H4IIE cells whereas high concentrations cause DNA damage and apoptosis. J. Nutr. 135, 525-531 (2005). 164. Donnini, S. et al. Divergent effects of quercetin conjugates on angiogenesis. Br. J. Nutr. 95, 1016-1023 (2006). 165. Koga, T. & Meydani, M. Effect of plasma metabolites of (+)-catechin and quercetin on monocyte adhesion to human aortic endothelial cells. Am. J. Clin. Nutr. 73, 941-948 (2001). 166. Forester, S.C. & Waterhouse, A.L. Gut metabolites of anthocyanins, gallic acid, 3-O-methylgallic acid, and 2,4,6-trihydroxybenzaldehyde, inhibit cell proliferation of Caco-2 cells. J. Agric. Food Chem. 58, 5320-5327 (2010). 167. Jaganath, I.B., Mullen, W., Edwards, C.A. & Crozier, A. The relative contribution of the small and large intestine to the absorption and metabolism of rutin in man. Free Radic. Res. 40, 1035-1046 (2006). 168. Gonzalez-Barrio, R., Edwards, C.A. & Crozier, A. Colonic catabolism of ellagitannins, ellagic acid, and raspberry anthocyanins: in vivo and in vitro studies. Drug Metab. Dispos. 39, 1680-1688 (2011). 169. Del Rio, D. et al. Dietary (poly)phenolics in human health: structures, bioavailability, and evidence of protective effects against chronic diseases. Antioxid. Redox Signal. 18, 1818-1892 (2013). 170. Hozo, S.P., Djulbegovic, B. & Hozo, I. Estimating the mean and variance from the median, range, and the size of a sample. BMC Med. Res. Methodol. 5, 13 (2005). 171. R Core Team (R Foundation for Statistical Computing, Vienna, Austria; 2016). 172. Faul, F., Erdfelder, E., Lang, A.G. & Buchner, A. G*Power 3: a flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behav. Res. Methods 39, 175-191 (2007). 173. Cohen, J. in Statistical power analysis for the behavioral sciences (revised edition) 407-453 (Academic Press, 1977).

Acknowledgements

This project has been funded by Horticulture Innovation Australia Limited using the Vegetable levy and funds from the Australian Government.

Author contributions statements

Initial idea: L.B. and S.SP. Data collection: S.SP. Statistical analysis: R.M, S.SP. Project coordination: L.B, K.H, F.D. Manuscript writing and review: S.SP, L.B, R.M, K.H and F.D.

Competing financial interest

The authors declared no competing financial interest with respect to the research, authorship, and/or publication of this article.

Figure legends

Figure 1. Prediction of Tmax by the predictive model. (a) The log P model in liquid, (b) semi-solid and (c) solid intakes. (d) The PSA model in liquid, (e) semi-solid and (f) solid intakes.

Figure 2. Summary of variables included in datasets for the development and validation of the predictive model. Dot plots demonstrate distributions of (a) log P, (b) Mr and (c) PSA of four datasets: training (N = 67), PCv (N = 108), PHv-fasted (N= 60) and PHv-fed (N = 38) datasets.

Figure 3. Comparison of measured versus predicted values of Tmax of the training dataset and the PCv dataset. Natural logarithm of Tmax measured from the training dataset (N = 67) were plotted against natural logarithm of predicted Tmax based on the log P model (black circle), the PSA model (clear circle) and compared to the regression of measured Tmax = predicted Tmax (dotted line) when intake as (a) liquid, (b) semi-solid and (c) solid forms. Similar comparison was plotted for the PCv dataset (N = 108) when intake as (d) liquid, (e) semi-solid and (f) solid forms.

Figure 4. Comparison of measured versus predicted values of Tmax of the PHv datasets. Natural logarithm of Tmax measured from (a) the PHv-fasted (N = 60) dataset and (b) the PHv-fed (N = 38) dataset were plotted against natural logarithm of predicted Tmax based on the log P model (black circle), the PSA model (clear circle) and compared to the regression of measured Tmax = predicted Tmax (dotted line) when intake solid forms.

Figure 1.

a 2 d Ln (Tmax) = 0.008 (log P) + 0.155 log P Ln (Tmax) = -0.009 PSA + 0.007 Mr − 0.959 2 + 0.002 Mr − 0.966; R = 0.93 R2 = 0.76

10 2 Mr 10 PSA (Å ) 100 0 300 8 8 150 500 (h) 300 700 (h) 6 450 max 6

max

4 4

Predicted T Predicted 2 Predicted T Predicted 2

0 -4 -2 0 2 4 6 8 10 0 0 200 400 600 800 1000 Log P M r 2 b Ln (Tmax) = 0.047 (log P) − 0.107 log P − 0.524 e Ln (Tmax) = -0.011 PSA + 0.009 Mr − 1.831 2 R = 0.95 R2 = 0.91

10 10 PSA (Å2)

8 0 8 150 (h) 300

(h) max 6 450 6

max 4 4

Predicted T 2

Predicted T Predicted 2 0 -4 -2 0 2 4 6 8 10 0 Log P 0 200 400 600 800 1000 M r 2 c Ln (Tmax) = 0.038 (log P) − 0.161 log P + 0.629 f Ln (Tmax) = -0.007 PSA + 0.004 Mr + 0.59 R2 = 0.64 R2 = 0.65

2 18 PSA (Å ) 10 16 0 14 150 8 300 (h) 12

(h) 450 max 10 6 8 max ieferLn (Tmax) = -0.009 PSA + 0.007 Mr − 6 0.959 4 2 Predicted T Predicted R = 0.76 4

2 T Predicted 2 0 -4 -2 0 2 4 6 8 10 Log P 0 0 200 400 600 800 1000 M r

Figure 2. a

Training

a D PCv

PHv-fasted PHv-fed -5.0 -2.5 0.0 2.5 5.0 7.5 10.0 LogP b

Training

D PCv

PHv-fasted PHv-fed 160 320 480 640 800 960 1120 1280 Mr c

Training

D PCv

PHv-fasted

PHv-fed 0 70 140 210 280 350 420 PSA

Figure 3. a d

LogP model (%RE = 18.27%) 4 PSA model (%RE = 37.46%) 4 LogP model (%RE = 55.84%) Measured = Predicted PSA model (%RE = 66.07%) Measured = Predicted 3 3

) ) 2 2

1 1

max_measured max_measured

Ln (T Ln 0 (T Ln 0

-1 -1

-1 0 1 2 3 4 -1 0 1 2 3 4 Ln (T ) Ln (T ) max_predicted max_predicted b e

4 LogP model (%RE = 19.13%) 4 LogP model (%RE = 57.07%) PSA model (%RE = 25.43%) PSA model (%RE = 92.95%) Measured = Predicted Measured = Predicted 3 3

) ) 2 2

1 1

max_measured max_measured

Ln (T Ln 0 (T Ln 0

-1 -1

-1 0 1 2 3 4 -1 0 1 2 3 4 Ln (T ) Ln (T ) max_predicted max_predicted c f

LogP model (%RE = 76.70%) 4 LogP model (%RE = 47.08%) 4 PSA model (%RE = 89.40%) PSA model (%RE = 45.80%) Measured = Predicted Measured = Predicted 3 3

)

) 2 2

1 1

max_measured

0 Ln(T 0

Ln (T

-1 -1

-1 0 1 2 3 4 -1 0 1 2 3 4 Ln (T ) Ln(T ) max_predicted max_predicted

Figure 4. a b

4 LogP model (%RE = 45.18%) 4 LogP model (%RE = 93.37%) PSA model (%RE = 162.69%) PSA model (%RE = 92.01%) Measured = Predicted Measured = Predicted 3 3

) ) 2 2

1 1

max_measured max_measured

Ln (T Ln 0 (T Ln 0

-1 -1

-1 0 1 2 3 4 -1 0 1 2 3 4 Ln (T ) Ln (T ) max_predicted max_predicted

Table 1. Pearson correlations between physicochemical properties of phytochemicals in the training dataset (N = 67) Physico- Freely H H chemical Mr Log P rotatable PSA acceptors donors properties bonds r = 0.174 Log P P = 0.159 Freely r = 0.772 r = 0.554 rotatable P < 0.001 P < 0.001 bonds H r = 0.442 -0.755 r = -0.094

acceptors P < 0.001 P < 0.001 P = 0.449 r = 0.424 r = -0.712 r = -0.127 r = 0.949 H donors P < 0.001 P < 0.001 P = 0.306 P < 0.001 0.435 r = -0.748 r = -0.110 r = 0.998 r = 0.955 PSA P < 0.001 P < 0.001 P = 0.377 P < 0.001 P < 0.001 Molecular r = 0.949 r = 0.445 r = 0.901 r = 0.413 r = 0.141 r = 0.135 volume P < 0.001 P < 0.001 P < 0.001 P = 247 P = 0.254 P = 0.277 Data reported as Pearson’s correlation coefficient (r) with p-values. Significantly high correlations (│r│ > 0.75, P < 0.05) are highlighted with bold.

Table 2. Pearson correlations between selected physicochemical properties and Tmax in the training dataset (N = 67)

Dietary intake form Variable Tmax Mr Log P Liquid Mr r = 0.48 P = 0.01 Log P r = 0.80 r = 0.45 P < 0.001 P = 0.015 PSA r = -0.72 r = -0.03 r = -0.82 P < 0.001 P = 0.88 P < 0.001 Semi-solid Mr r = 0.28 P = 0.388 Log P r = 0.75 r = -0.07 P = 0.005 P = 0.832 PSA r = -0.64 r = 0.30 r = -0.93 P = 0.026 P = 0.342 P < 0.001 Solid Mr r = 0.52 P < 0.001 Log P r = 0.65 r = 0.19 P < 0.001 P = 0.332 PSA r = -0.13 r = 0.66 r = -0.56 P = 0.517 P < 0.001 P = 0.002 Data reported as Pearson’s correlation coefficient (r) with p-values. Significantly high correlations (│r│ > 0.75, P < 0.05) are highlighted with bold.

Table 3. Summary of datasets for development and validation of the predictive model. No. No. No. Tmax PSA Mr Log P Dataset Intake of of of rangea rangeb rangeb rangeb studies people PCs (h) (Å2) Training 0.3– Liquid 12 112 28 122–612 -4.4–10 0–271 32.6 Semi- 0.6– 5 29 12 302–728 -4.7–9.8 0–309 solid 14.8 0.7– Solid 16 257 27 122–1270 -3.4–9.4 29–465 15.2 0.3– Total 31 384 67 122–1270 -4.7–9.8 0–465 32.6 PCv Liquid 19 667 70 0.5–19 138–659 -4.3–10 0–271 Semi- 5 129 12 1.0–4.0 176–758 -4.7–-1.4 107–330 solid Solid 14 354 26 0.8–37 176–569 -2.8–9.8 0–197 Total 34 1150 108 0.5–37 138–758 -4.7–10 0–330 PHv- Solid 59 963 60 0.8–3.6 123–552 -1.7–5.2 3–146 fasted PHv-fed Solid 37 617 38 1.4–6.5 123–823 -1.7–5.4 3–221 a Tmax of phytochemicals were collected from clinical studies in the literature. b Physicochemical properties of phytochemicals including Mr, Log P and PSA were calculated using the Molinspiration Chemoinformatics calculator.

Table 4. Comparison of prediction accuracy of the predictive model for each datasets Log P model PSA model Parameter Training PCv PHv- PHv- Training PCv PHv- PHv- dataset dataset fasted fed dataset dataset fasted fed Liquid intake NMSWE 0.0282 0.1968 NA NA 0.1012 0.2573 NA NA RNMSWE 0.1678 0.4436 NA NA 0.3181 0.5072 NA NA %RE 18.27 55.84 NA NA 37.46 66.07 NA NA N 28 70 NA NA 28 70 NA NA Semi-solid intake NMSWE 0.0306 0.2039 NA NA 0.0513 0.4320 NA NA RNMSWE 0.1750 0.4516 NA NA 0.2266 0.6573 NA NA %RE 19.13 57.07 NA NA 25.43 92.95 NA NA N 12 12 NA NA 12 12 NA NA Solid intake NMSWE 0.1488 0.3241 0.1390 0.4349 0.1422 0.4079 0.9327 0.4256 RNMSWE 0.3858 0.5693 0.3728 0.6594 0.3771 0.6387 0.9658 0.6524 %RE 47.08 76.70 45.18 93.37 45.80 89.40 162.69 92.01 N 27 26 60 38 27 26 60 38

Chapter 3: Statistical modelling coupled with LC-MS analysis to predict human upper intestinal absorption of phytochemical mixtures

Chapter 3: Statistical Modelling Coupled with LC-MS Analysis to Predict Human Upper Intestinal Absorption of Phytochemical Mixtures

3.1 Introduction

The in silico statistical PCAP model developed in Chapter 2 allows direct calculation of

Tmax of known individual phytochemicals using molecular mass, lipophilicity descriptor log P, for different dietary intake forms. However, dietary phytochemicals are typically consumed as complex mixtures of unidentified phytochemicals. Therefore, identification of molecular mass and log P of complex mixtures of phytochemicals is required to expand the practical usefulness of the in silico statistical model.

In this chapter, the research hypothesis was that characterisation of molecular mass and lipophilicity descriptor log P of phytochemicals present in a complex phytochemical mixture would allow prediction of Tmax profiles of the mixture.

Chapter 3: Statistical modelling coupled with LC-MS analysis to predict human upper intestinal absorption of phytochemical mixtures

The aim of this chapter was to develop analytical methods coupled with the application of the in silico statistical model to characterise Tmax profiles of phytochemicals mixtures from 17 selected dietary plants. The aims were fulfilled via completion of the following objectives:

 Application of liquid chromatography mass spectrometry (LC-MS) methodology

to determine values of molecular mass and lipophilicity descriptor log P of

complex phytochemical mixtures.

 Development of a data processing workflow to convert the LC-MS data output to

predicted Tmax profiles of phytochemical mixtures.

 Validation of the predicted Tmax profiles by comparing with published literature

of clinical plasma Tmax and regulation of OSI for similar extracts.

3.2 Accepted manuscript

Statistical modelling coupled with LC-MS analysis to predict human

upper intestinal absorption of phytochemical mixtures

Sophie N.B. Selby-Phama,b, Kate S. Howella, Frank R. Dunsheaa, Joel Ludbeyc, Adrian

Lutzd and Louise Bennettb,*1 aFaculty of Veterinary and Agricultural Sciences, The University of Melbourne,

Parkville, Victoria, 3010, Australia bCSIRO Agriculture and Food, Werribee, Victoria, 3030, Australia cCSIRO Information Management and Technology, Clayton South, Victoria, 3169,

Australia dMetabolomics Australia, School of Biosciences, The University of Melbourne, Parkville,

Victoria, 3010, Australia

Abstract

A diet rich in phytochemicals confers benefits for health by reducing the risk of chronic diseases via regulation of oxidative stress and inflammation (OSI). For optimal protective bio-efficacy, the time required for phytochemicals and their metabolites to reach maximal plasma concentrations (Tmax) should be synchronised with the time of increased OSI. A statistical model has been reported to predict Tmax of individual phytochemicals based on molecular mass and lipophilicity. We report the application of the model for predicting the absorption profile of an uncharacterised phytochemical mixture, herein referred to as the ‘functional fingerprint’. First, chemical profiles of phytochemical extracts were acquired using liquid chromatography mass spectrometry (LC-MS), then the molecular

*Corresponding author Email address: [email protected] Phone number: +613 9905 4593 43

features for respective components were used to predict their plasma absorption maximum, based on molecular mass and lipophilicity. This method of ‘functional fingerprinting’ of plant extracts represents a novel tool for understanding and optimising the health efficacy of plant extracts.

Keywords: untargeted profiling, LC-MS, secondary metabolites, log P, molecular mass,

Tmax, Phytochemical Absorption Prediction model

1. Introduction

Phytochemicals, also referred to as secondary metabolites, are the non-nutrient compounds in fruits, vegetables and other dietary plants which have been associated with reductions in the risk of major chronic diseases including cancer (Key, 2011), cardiovascular (Dauchet, Amouyel, & Dallongeville, 2009) and neurodegenerative diseases (D'Onofrio, Sancarlo, Ruan, Yu, Panza, Daniele, et al., 2016). More than 200,000 phytochemical structures have been identified but only a small percentage have been investigated with regard to their application in medicine, making them interesting candidates as pharmaceutically active agents (Hartmann, 2007). The health benefits of phytochemicals have been linked with their capacity to regulate oxidative stress and inflammation (OSI), which occurs as part of normal metabolism, but is also involved in the aetiology of most chronic diseases (Calder, Albers, Antoine, Blum, Bourdet-Sicard,

Ferns, et al., 2009). The regulation of OSI by phytochemicals may occur by direct antioxidant activity or by an indirect mechanism via induction of antioxidant stress defence (Selby-Pham, Cottrell, Dunshea, Ng, Bennett, & Howell, 2017). Cells in the human body are continuously exposed to oxidising agents from the environment, foods or those produced by metabolic activities within cells. Maintaining the balance between oxidants and anti-oxidants is crucial for optimal physiological conditions in the body

(Calder, et al., 2009). Over-production of oxidants can cause OSI and unregulated OSI can damage large biomolecules such as proteins, DNA and lipids, which in turn results in an increased risk of chronic diseases (Kryston, Georgiev, Pissis, & Georgakilas, 2011).

Therefore, regulating transient and cumulative OSI associated with daily activities and chronic diseases is important to lower OSI-related mortality.

The bio-efficacy of phytochemicals to protect human health is dependent on their absorption into circulation and delivery to the target cells (D’Archivio, Filesi, Varì,

Scazzocchio, & Masella, 2010). However, phytochemicals are only transiently present in circulation after consumption because they are recognised as xenobiotics by the human body (Holst & Williamson, 2008). After consumption of dietary plants, some phytochemicals are absorbed in the small intestine and enter the circulatory system

(D’Archivio, Filesi, Varì, Scazzocchio, & Masella, 2010). These phytochemicals may be modified by the liver and their hepatic metabolites re-enter the circulatory system

(D’Archivio, Filesi, Varì, Scazzocchio, & Masella, 2010). The unabsorbed phytochemicals reach the large intestine and are subjected to structural transformation by the colonic microbiota. These microbial metabolites can also be absorbed via the colon

(Holst & Williamson, 2008). The time required for phytochemicals or their metabolites to reach their maximal concentrations in the circulatory system (Tmax) can be an important factor to understand and optimise the health benefits of plant foods.

The importance of Tmax was demonstrated in a recent study where healthy volunteers consumed a strawberry drink two hours before, during, or two hours after a high fat meal

(Huang, Park, Edirisinghe, & Burton-Freeman, 2016). The strawberry drink was observed to reduce the OSI associated with the high fat meal only when being consumed at two hours before the meal (Huang, Park, Edirisinghe, & Burton-Freeman, 2016). Considering

that Tmax of the measured phytochemicals in strawberry was approximately 1–2 h

(Sandhu, Huang, Xiao, Park, Edirisinghe, & Burton-Freeman, 2016), consumption of the drink two hours before the high fat meal ensured that maximal concentration in the circulatory system coincided with the post-prandial OSI to minimise the OSI damage

(measured by plasma concentration of the biomarker interleukin-6) triggered by the meal

(Burton-Freeman, 2010).

The bioavailability of phytochemicals is dependent on their chemical structures and dietary intake forms (D’Archivio, Filesi, Varì, Scazzocchio, & Masella, 2010). We have developed a statistical model, the Phytochemical Absorption Prediction (PCAP) model, to predict Tmax of dietary phytochemicals absorbed in human upper intestine based on their molecular mass, lipophilicity (expressed as log P, the logarithm of the partition coefficient between water and 1-octanol) and dietary intake forms (Selby-Pham, Miller,

Howell, Dunshea, & Bennett, 2017). Application of this model allows direct calculation of values of Tmax of phytochemicals using molecular mass and log P. However, the PCAP model can only be applied to individual (known) phytochemicals. In order to expand the practical usefulness of the model, it is necessary that molecular mass and log P of complex mixtures of phytochemicals, reflecting their typical mode of consumption, can be identified. Accordingly, further development of methods to predict the Tmax range arising from uncharacterised phytochemicals mixtures is required. Chemical identification, which presents additional challenges such as the need for a previously purified, synthesised or characterised chemical standard of specific phytochemical, is not required for this purpose.

The retention of compounds on C18 reverse phase columns during liquid chromatography

(LC) is controlled by lipophilicity and therefore correlated with log P (Valko, 2004). This

feature of reverse phase LC allows for the development of multiple methods to estimate the log P of drug compounds (Valko, 2004) and natural products (Camp, Campitelli,

Carroll, Davis, & Quinn, 2013). Further, LC may be coupled with mass spectrometry

(MS) so that, in addition to allowing the determination of log P from retention time, MS identifies the accurate molecular mass of the compound, referred to as a ‘molecular feature’, until the chemical identity of the compound is confirmed (Flamini, De Rosso,

De Marchi, Dalla Vedova, Panighel, Gardiman, et al., 2013; Tsao & Li, 2013).

The aim of this research was to apply LC-MS methodology to simultaneously determine values of log P and molecular mass of individual phytochemicals present in extracts of selected dietary plants. A further aim was to develop a data processing workflow to convert the LC-MS data output to individual Tmax values of phytochemicals absorbed in the human upper intestine, using the PCAP model. The Tmax values were then used to generate a characteristic ‘functional fingerprint’ which represents the human upper intestinal absorption kinetic profile of the tested plant extract. Finally, validation of predicted functional fingerprints was investigated by comparison with published clinical evidence of plasma Tmax and regulation of OSI, for similar extracts.

2. Materials and methods

2.1. Chemicals and reagents

Chemicals and reagents including chloroform, methanol, Na2CO3, gallic acid, formic acid, acetonitrile, L-histidine, (S)-dihydroorotate, shikimate, 4-pyridoxate, 3- hydroxybenzyl alcohol, 2,5-dihydroxybenzoate, 3-hydroxybenzaldehyde, trans- cinnamate, estradiol-17α, deoxycholate, retinoate, oleic acid and heptadecanoate were from Sigma-Aldrich (St Louis, MO, USA). Folin-Ciocalteu reagent was from Merck

(Darmstadt, Germany).

2.2. Preparation of plant extracts

Fresh forms of dietary plant materials were purchased from local retailers (Woolworth,

Werribee, VIC, Australia) and included broccoli (Brassica oleracea var. italica), carrot

(Daucus carota ssp. sativus), red sweet potato (Impomoea batatas), rhubarb (Rheum rhabarbarum), squash (Cucurbita pepo var. ovifera), eggplant (Solanum melongena), kale (Brassica oleracea var. acephala) and Vietnamese coriander (Persicaria odorata).

All samples were subjected to Stage 1 of a three-stage generic processing as described previously (Bennett & Muench, 2011). Briefly, plant material was homogenised in a food processor (Breville, Sydney, NSW. Australia) with water (1:2 ratio w/v) before cooking by microwave at 800 W (Sharp Carousel, Huntingwood, NSW, Australia) for 10 min to a final temperature of ~ 70 ºC. After cooling to room temperature, the mixture was ultrasonicated at 300 W for 11 min (Hielscher 400UPS, Hielscher, Germany) before bag filtration (1 µm pore size, Sefar Filtration Inc., Depew, NY, USA). The filtrate was freeze- dried, ground to a fine powder and stored with a desiccant at -18 oC. Processed products were referred as “project extracts”.

Commercial plant products were used as “reference extracts” and were obtained from the following suppliers: blueberry (Super Sprout, Campbellfield, VIC, Australia), green tea powder (Absolute Green, DeDu Pty Ltd., Ermington, NSW, Australia ), olive leaf powder

(Austral herbs, Uralla, NSW, Australia) and tomato powder (Herbies’s Spices, Rozelle,

NSW, Australia).

2.3. Proximate composition analysis of plant extracts

Proximate composition of the plant extracts was determined using standard analytical methods. Moisture content was determined using a halogen moisture analyser (Model

HR73, Mettler Toledo, Columbus, OH, USA). Nitrogen analysis was determined by

LECO Trumac® N analyser (LECO Corporation, Michigan, USA). Protein content was calculated by multiplying nitrogen content by the recommended conversion factor of 6.25 for plant foods (Sosulski & Imafidon, 1990). Total lipid content was quantified by Waite

Analytical Services (The University of Adelaide, SA, Australia) using gas chromatography (GC) after extraction with chloroform: methanol (9:1 v/v) and subsequent methylation, as described previously (Makrides, Neumann, & Gibson, 1996).

Mineral analysis quantified Al, B, Ca, Cd, Co, Cr, Cu, Fe, K, Mg, Mn, Mo, Na, P, Pb, S,

Se, Si, Ti and Zn using inductively coupled plasma atomic emission spectroscopy (ICP-

AES). The ICP-AES analyses were conducted using a Varian Vista Pro (Varian Australia,

Melbourne, Australia) as described previously (Bennett, Singh, & Clingeleffer, 2011).

Total ash content was estimated as the sum of all mineral contents of plant extracts

(James, 1995). The total carbohydrate content was determined by subtracting the sum of protein content, fat content, ash content and moisture from 100 (Njinkoue, Gouado,

Tchoumbougnang, Ngueguim, Ndinteh, Fomogne-Fodjo, et al., 2016). All analyses were performed in duplicate.

2.4. Total phenolic content of plant extracts

Total phenolic content of plant extracts was quantified using a modified Folin-Ciocalteu spectrophotometric methodology (Singleton & Rossi, 1965). In brief, 20 μL plant extract

(2 mg/mL in 20% methanol) was mixed with 1 mL of 0.2 N Folin-Ciocalteu reagent and

180 µL of Milli-Q water for 15 s. The mixture stood for 3 min before the addition of 800

µL of 7.5% Na2CO3. The mixture was further shaken for 15 s and then incubated in the dark at 37 °C for 1 h. The absorbance at 765 nm was measured using a Varioskan Flash microplate reader (Thermo Fisher Scientific, Waltham, MA, USA). The total phenolic

content of plant extracts was expressed as gallic acid equivalent (GAE) by plotting the gallic acid calibration curve (from 0 to 500 µg/mL in 20% methanol).

2.5. Characterisation of physicochemical properties of plant extracts using liquid chromatography mass spectrometry (LC-MS) analysis

Characterisation of log P and molecular mass of plant extracts was performed using untargeted LC-MS profiling analysis by Metabolomics Australia (School of BioSciences,

University of Melbourne). Prior to LC-MC analysis, plant extracts were reconstituted at a concentration of 20 mg/mL in 20% methanol and injection volume was 1 µL. Instrument and LC-MS setup were as follows. Agilent 6520 quadrupole time-of-flight (QTOF) MS system (Santa Clara, CA, USA) with a dual sprayer electrospray ionisation (ESI) source and attached to Agilent 1200 series high performance liquid chromatography (HPLC) system (Santa Clara, CA, USA) comprised of a vacuum degasser, binary pump, with a thermostated auto-sampler and column oven. The MS was operated in positive or negative mode using the following conditions (positive/negative, respectively): nebulizer pressure 30/45 psi, gas flow-rate 10 L/min, gas temperature 300°C, capillary voltage

4000/-3500 V, fragmentor 150 and skimmer 65 V. Instrument was operated in the extended dynamic range mode with data collected in mass to charge ratio (m/z) range of

70–1700. Chromatography was carried out using an Agilent Zorbax Eclipse XDB-C18,

2.1 x 100 mm, 1.8 µm column maintained at 40 °C (± 1 °C) at a flow rate of 400 µL/min with a 20 minute run time. A gradient LC method was used with mobile phases comprised of (A) 0.1% formic acid in deionized water and (B) 0.1% formic acid in acetonitrile.

Gradient: A 5 min linear gradient from 5% to 30% mobile phase B, followed by 5 minute gradient to 100% mobile phase B and then a 5 min hold, followed by a 5 minute re- equilibration at 5% mobile phase B. Molecular feature extraction (MFE) was conducted

using Agilent MassHunter Qualitative analysis (version B.07.00) and MassHunter

Profinder (version B.06.00). Binning and alignment tolerances were set to: retention time

= ±0.1% + 0.15 min; mass window = ±20 ppm + 2 mDa. Allowed ion species: +H, +Na,

+K, +NH4 (positive mode); -H, +Cl, +HCOO (negative mode) and neutral losses: H2O,

H3PO4, CO2, C6H12O6. MFE was restricted to the 1000 largest features and 1-2 charge states.

For determination of log P values of plant extracts based on retention times, a calibration curve of retention time and log P of twelve standards was established. Log P values of standards were calculated from their molecular structures using the Molinspiration

Chemoinformatics calculator (www.molinspiration.com). Standards included L- histidine, (S)-dihydroorotate, shikimate, 4-pyridoxate, 3-hydroxybenzyl alcohol, 2,5- dihydroxybenzoate, 3-hydroxybenzaldehyde, trans-cinnamate, estradiol-17α, deoxycholate, retinoate, oleic acid and heptadecanoate. After single injections, retention times were recorded in both positive and negative modes. Averages of the retention times from the positive and negative modes were then used to develop the calibration curve.

2.6. Functional fingerprint profiling of plant extracts by application of LC-MS analysis and the PCAP model

The human upper intestinal absorption kinetics of plant extract characterised by application of LC-MS and the PCAP model was referred to as the ‘functional fingerprint’, which displays the predicted Tmax ranges of phytochemicals and their relative abundance within the plant extract. A molecular mass range of 122–1270 amu was chosen according to the range in the PCAP model which was used to calculate Tmax from log P and molecular mass (Selby-Pham, Miller, Howell, Dunshea, & Bennett, 2017). Molecular features present in all plant samples were categorised as ‘primary metabolites’ and the

molecular features only present in some plants were categorized as ‘secondary metabolites’, as primary metabolites are common to all plants whilst secondary metabolites are specific to selected plant groups (Pichersky & Gang, 2000). For example, sugars and some amino acids are primary metabolites as they are present in all plants whilst phenolics, carotenoids and terpenoids are secondary metabolites as they are present in certain restricted groups of plants (Kutchan, Gershenzon, Moller, & Gang, 2015). As phenolics are one of three major groups of secondary metabolites in plant extracts

(Crozier, Jaganath, & Clifford, 2006), Pearson’s correlation analysis between total phenolics and LC-MS relative quantification of secondary metabolites as total peak area was performed. The functional fingerprints of plant extracts were generated by plotting peak area (relative ion count) of the molecular features detected versus their associated

Tmax.

2.7. Literature evaluation of acute bio-efficacy of dietary phytochemicals in humans

A literature search on acute bio-efficacy of dietary phytochemicals in managing OSI in human was performed for blueberry and green tea consumed as either whole foods or extracts. The inclusion selection criteria included: 1) randomised controlled clinical trials in healthy volunteers; 2) inclusion of a wash-out period when the study followed a cross over design; 3) time course of blood sampling was performed after a single dose of blueberry or green tea and the effects of phytochemicals consumption on biomarkers of

OSI were acute. Schematic diagrams of the selected studies were generated in combination with the predicted Tmax of dietary phytochemicals obtained from their functional fingerprint profiles.

2.8. Data analysis

All curve-fitting and Pearson’s correlation analysis were performed using SigmaPlot for

Windows Version 12.5 (Systat Software Inc., Chicago, IL, USA).

3. Results

3.1. Proximate analyses of plant extracts

Proximate analyses of eight project extracts and four reference extracts included total solids, total nitrogen, total lipids and total ash (Supplementary Table S1) and mineral profiling (Supplementary Table S2). The total solids of the project extracts varied between 83.35 and 99.65% total weight whilst the total solids of the reference extracts ranged from 98.78 to 99.43% total weight (Supplementary Table S1). The nitrogen contents ranged from 0.9 to 5.49% total weight for the project extracts and from 0.77 to

4.59% total weight for the reference extracts (Supplementary Table S1). The total lipids were 0.48–7.02% total weight for the project extracts and 1.7–44.64% total weight for the reference extracts with the highest total lipids of 44.64% total weight for olive leaf

(Supplementary Table S1). The total ash, estimated as the sum of all minerals detected by ICP analysis, varied from 1.86 to 16.34% total weight for the project extracts and 0.06–

2.21% total weight for the reference extract (Supplementary Table S1). The total carbohydrates were 37.41–88.18% total weight for the project extracts and 43.13–92.38% total weight for the reference extracts with the highest total carbohydrates of 92.38% total weight for blueberry (Supplementary Table S1).

3.2. Validation of the processing method of the LC-MS output data

The LC-MS outputs included molecular mass, retention time and relative quantification as peak area (relative ion count) of the molecular features detected in the plant extracts.

Data processing workflow showing the translation of molecular mass and log P obtained

from LC-MS analysis into predicted functional fingerprints of plant extracts is shown in

Fig. 1A. The molecular mass range of 122–1270 amu (Fig. 1A) was chosen according to the range in the PCAP model, which was used to calculate Tmax from log P and molecular mass (Selby-Pham, Miller, Howell, Dunshea, & Bennett, 2017). Retention times of the molecular features within a plant extract were converted into values of log P using the calibration curve of retention time and log P of twelve standards. This calibration curve described a fourth-degree polynomial relationship between log P and retention time (Fig.

1B).

Molecular features detected in plant extracts were categorised into two groups: primary and secondary metabolites (referred to as phytochemicals in the present study). This method of categorisation was validated by a significant correlation (Pearson’s correlation coefficient r = 0.77, p = 0.003) between the total phenolic content and LC-MS relative quantification of secondary metabolites as total peak area (Fig. 1C).

3.3. Modelling functional fingerprints of plant extracts

The functional fingerprints of ten selected plant extracts were generated by plotting peak area (relative ion count) of the molecular features detected versus their associated Tmax.

The calculation of Tmax was dependent on plant extract intake in either drink (Fig. 2) or solid form (Fig. 3). For plant extracts consumed in drink form, the primary metabolites displayed relatively short Tmax of 0.4–1 h with a median of 0.5 h whilst Tmax of the secondary metabolites ranged from 0.4 to11 h with a median of 1 h (Fig. 2, Supplementary

Fig. 1A). For plant extracts consumed in solid form, the primary metabolites and the secondary metabolites displayed similar Tmax range of 1.6–3.7 h (Fig. 3, Supplementary

Fig. 1B).

For consumption in drink form, all project extracts displayed similar functional fingerprints with Tmax of 0.4–11 h for broccoli, carrot, kale and Vietnamese coriander

(Fig. 2A–D), Tmax of 0.4–8 h for red sweet potato, rhubarb and eggplant (Fig. 2E–G) and

Tmax of 0.4–6 h for squash (Fig. 2H). However, distinct functional fingerprints were observed for the olive leaf and tomato reference extracts. Tmax of olive leaf were divided into three segments: 0.4–2.4 h, 2.9–3 h and 4.2–5 h (Fig. 2I) whilst Tmax of tomato included four segments: 0.4–2.8 h, 3.6 h, 4.3 h and 7.3–8.5 h (Fig. 2H).

3.4. Comparison of LC-MS functional fingerprints and chemical analysis of plant

extracts

The measured LC-MS functional fingerprints of selected phytochemical-rich plants were compared with published analytical data of phytochemical composition (Del Rio,

Stewart, Mullen, Burns, Lean, Brighenti, et al., 2004; Gao & Mazza, 1994). Blueberry in drink (Fig. 4A) and paste forms (Fig. 5A) and green tea in drink form (Fig 6A) were compared to their published phytochemical compositional data (Fig. 4B, 5B and 6B). In the present study, the molecular features detected in blueberry and green tea extracts had molecular mass ranges of 122–1258 and log P ranges of -2 to 6.9 (Fig. 4A, 5A and 6A).

However, the published phytochemical compositional data (Del Rio, et al., 2004; Gao &

Mazza, 1994) showed a substantially narrower range of molecular mass and log P for these extracts. Blueberry was reported to contain phytochemicals with a molecular mass range of 354–535 and a log P range of -3 to -0.5 (Fig. 4B and 5B). Additionally, green tea was reported to contain phytochemicals with a molecular mass range of 170–611 and a log P range of -1.2 to 2.5 (Fig. 6A).

The measured LC-MS functional fingerprint of blueberry when consumed in drink form were generated based on all the molecular features detected (Fig. 4A, grey) and the subset

reflecting the narrower ranges of molecular mass (354–535) and log P (-2 to -0.5) of the published compositional data was highlighted (Fig. 4A, black). Accordingly, blueberry phytochemicals when consumed in drink form had three segments of Tmax: 0.4–1.8 h, 4.8–

5.4 h and 7.6–8.4 h (Fig 4A, grey). The functional fingerprint of blueberry in drink form generated from the published compositional data (Fig. 4B) shared a similar peak profile to the portion of the measured LC-MS functional fingerprint (Fig. 4A, black) occurring within the highlighted Tmax range of 0.6–0.9 h.

The measured LC-MS functional fingerprint of blueberry when consumed in paste form were generated based on all the molecular features detected (Fig. 5A, grey) and the subset reflecting the narrower ranges of molecular mass (354–535) and log P (-2 to -0.5) of the published compositional data was highlighted (Fig. 5A, black). Accordingly, blueberry phytochemicals when consumed in paste form displayed three segments of Tmax: 0.6–0.9 h, 1.5 h and 2.3–2.6 h (Fig 5A, grey). The functional fingerprint of blueberry in paste form generated from the published compositional data (Fig. 5B) showed a slightly longer

Tmax of 0.6–1.3 h compared to the portion of the measured LC-MS functional fingerprint

(Fig. 5A, black) occurring within the highlighted Tmax range of 0.7–0.9 h.

The measured LC-MS functional fingerprint of green tea when consumed in drink form were generated based on all the molecular features detected (Fig. 6A, grey) the subset reflecting the narrower ranges of molecular mass (170–611) and log P (-1.2 to 2.5) of the published compositional data was highlighted (Fig. 6A, black). Accordingly, green tea phytochemicals when consumed in drink form had two segments of Tmax: 0.4–8 h and

10–11 h (Fig 6A, grey). The functional fingerprint of green tea in drink form generated from the published compositional data (Fig. 6B) shared a similar peak profile to the

portion of the measured LC-MS functional fingerprint (Fig. 6A, black) occurring within the highlighted Tmax range of 0.5–1.5 h.

3.5. The relationship between functional fingerprints of blueberry and green tea

phytochemicals to their bio-efficacy of OSI regulation

The association between the functional fingerprint of blueberry in drink form (Fig. 4A) and its previously reported bio-efficacy of OSI regulation in human subjects (Kay &

Holub, 2002) was investigated (Fig. 4C) by comparing OSI biomarkers after the consumption of a high fat meal with/without concomitant ingestion of blueberry in drink form. In the study (Kay & Holub, 2002), blood sampling was performed hourly for 4 h and analysed for serum anti-oxidant status using the oxygen radical absorbance capacity

(ORAC) assay as a biomarker of anti-OSI (aOSI) capacity. Subjects who consumed the high fat meal without the blueberry drink showed minimal changes to the aOSI biomarker. However, in the subjects who consumed the high fat meal with the blueberry drink, a significant increase of the aOSI biomarker was detected at 1 h but not at other blood sampling times. This observed time of increased aOSI biomarker levels at 1 h occurred within the 0.4–1.8 h predicted Tmax of the blueberry in drink form (Fig. 4C).

The association between the functional fingerprint of blueberry in paste form (Fig. 5A) and its previously reported bio-efficacy of OSI regulation in human subjects (Del Bo,

Riso, Campolo, Moller, Loft, Klimis-Zacas, et al., 2013) was investigated (Fig. 5C) by measuring OSI biomarkers after the consumption of blueberry in paste form. In the study

(Del Bo, et al., 2013), blood sampling was performed hourly for 3 h and at 24 h and analysed for DNA damage as a biomarker of OSI. A significant decrease of the OSI biomarker was detected at 1 h but not at other blood sampling times. This observed time

of reduce OSI biomarker levels at 1 h occurred within the 0.6–0.9 h predicted Tmax of the blueberry in paste form (Fig. 5C).

The association between the functional fingerprint of green tea in drink form (Fig. 6A) and its previously reported bio-efficacy of OSI regulation in human subjects (Ho, Choi,

Siu, & Benzie, 2014; Leenen, Roodenburg, Tijburg, & Wiseman, 2000) was also investigated (Fig. 6C and D) by measuring OSI and aOSI biomarkers after the consumption of green tea in drink form. In the first study (Ho, Choi, Siu, & Benzie, 2014), blood sampling was performed hourly for 2 h and analysed for DNA damage as a biomarker of OSI and DNA repair enzyme as a biomarker of aOSI (Fig. 6C). A significant decrease of the OSI biomarker and a significant increase of the aOSI biomarker was detected at 1 h and 2 h after the consumption of green tea in drink form (Fig. 6C). In the second study (Leenen, Roodenburg, Tijburg, & Wiseman, 2000), blood sampling was performed every 0.5 h for 2 h and analysed for plasma phytochemicals and plasma antioxidant status using the ferric antioxidant power (FRAP) assay as a biomarker of aOSI

(Fig. 6D). Significant increases of plasma phytochemicals were observed at all time points and significant increases of the aOSI biomarker were observed at 1 h, 1.5 h and 2 h (Fig. 6D). In both studies (Ho, Choi, Siu, & Benzie, 2014; Leenen, Roodenburg,

Tijburg, & Wiseman, 2000), the observed time of increased aOSI and reduced OSI biomarkers levels at 1 h and 2 h for the first study (Fig. 6C) and at 1 h, 1.5 h and 2 h for the second study (Fig. 6D) occurred within the 0.4–8 h predicted Tmax of the green tea in drink form.

4. Discussion

4.1. Functional fingerprint profiling by LC-MS using log P and molecular mass

This study reported a novel and high-throughput method to obtain a prediction of human upper intestinal absorption of plant extracts, referred to as a ‘functional fingerprint’.

Firstly, plant extracts were analysed by LC-MS to identify the log P, molecular mass and relative abundance of the phytochemicals present in the plant mixture. Each combination of log P and molecular mass represents a phytochemical within the plant mixture. The

PCAP model (Selby-Pham, Miller, Howell, Dunshea, & Bennett, 2017) was then applied to calculate the predicted time required for each phytochemical to reach plasma maximal concentration (Tmax) based on its log P and molecular mass. Finally, the Tmax of phytochemicals were plotted against their relative abundances to generate the functional fingerprint of the tested plant extract.

In the present study, LC-MS analysis was used for the determination of log P and molecular mass of the phytochemical composition of plant extracts. LC-MS analysis has been commonly used for log P profiling, identification and quantitation of drug compounds (Valko, 2004) and plant extracts (Camp, Campitelli, Carroll, Davis, & Quinn,

2013). However, the direct prediction of Tmax of compounds from log P and molecular mass was not feasible until the recently reported PCAP model, which directly predict Tmax of phytochemicals from their log P and molecular mass (Selby-Pham, Miller, Howell,

Dunshea, & Bennett, 2017). This model allows the information obtained from LC-MS analysis to be converted into a prediction of human upper intestine absorption.

When blueberry was ingested as paste form, the functional fingerprint generated from data reported in the literature covered a Tmax range of 0.6–1.3 h (Fig. 5B). However, when limited to the log P and molecular mass ranges presented in the literature, the measured

LC-MS functional fingerprint predicted a slightly narrower Tmax range of 0.6–0.9 h (Fig.

5A, black). Accordingly, within this range of log P and molecular mass, Gao and Mazza

(1994) detected slightly higher number of compounds. This difference may be explained by the use of weak acid in combination with methanol and water for the extraction of phytochemicals in their study (Gao & Mazza, 1994) that facilitate better partitioning of phytochemicals into the extract (Tsao & Li, 2013) compared to methanol and water used in the present study. As the functional fingerprint reflects actual composition, the profile from a given type of plant is expected to depend on the selectivity and efficiency of the extraction method.

The data processing method reported herein can be applied to any plant extract and any other analytical technique that characterises molecular mass and log P, for example other types of LC-MS featuring different ionisation methods and mass analysers (Wu, Guo,

Chen, Liu, Zhou, Zhang, et al., 2013) or nuclear magnetic resonance spectroscopy

(Wolfender, Ndjoko, & Hostettmann, 2001). Additionally, as this method does not rely on identification of compounds present in plant extracts, the results are not limited by the availability of standards. This feature is particularly appealing as only a very small percent out of the estimated 250,000 phytochemicals have been identified

(Waksmundzka-Hajnos & Sherma, 2010). Whilst previous attempts to profile blueberry

(Fig. 4B) and green tea (Fig. 6B) were limited by the availability of pure chemical standards, the functional fingerprints presented herein not only identified these previously reported profiles but also identified more comprehensive fingerprints over a broader range of compounds (Fig. 4A and 6A, grey). Further, within the coinciding ranges of log

P and molecular mass between the previously reported profiles (Fig. 4B and 6B) and the generated functional fingerprints (Fig. 4A and 6A, black), a strong resemblance was

observed highlighting that the untargeted approach of this method does not inhibit profiling of identifiable compounds.

4.2. Significance of phytochemical functional fingerprints for clinical research and

human health

In the present study, the functional fingerprints of primary and secondary metabolites of selected plant extracts provided insights into their absorption in drink and solid forms.

The ability to distinguish primary and secondary metabolites is of particular interest in the context of impacts on human health. Sugars are one of the main primary metabolites in plants (Patrick, Botha, & Birch, 2013) and postprandial plasma glucose spike is a risk factor of cardiovascular diseases especially in diabetic individuals (O’Keefe & Bell,

2007). When plant extracts were consumed in drink form, the primary metabolites displayed very short Tmax (median of 0.5 h) and appeared in the plasma faster than the secondary metabolites with median Tmax of 1 h (Fig. 2, Supplementary Fig. S1A). In contrast, when plant extracts were consumed in solid form, the primary metabolites had longer Tmax (median of 2.6 h) compared to the secondary metabolites with median Tmax of 1.7 h (Fig. 3, Supplementary Fig. S1B). These results may indicate that consumption of plant products in the solid form (e.g., whole foods) may impart better glycaemic control and avoidance of a plasma glucose spike. These results are consistent with findings from large prospective cohort studies indicating that consumption of vegetables and whole fruits was associated with lower risks of diabetes, whilst consumption of fruit juice was associated with increased risk (Bazzano, Li, Joshipura, & Hu, 2008; Muraki, Imamura,

Manson, Hu, Willett, van Dam, et al., 2013).

The functional fingerprints of plant extracts provide prediction of the Tmax range of their phytochemicals in human plasma after consumption. Considering that phytochemicals

may stimulate OSI defence mechanism (Selby-Pham, Cottrell, Dunshea, Ng, Bennett, &

Howell, 2017) whilst being treated by the human body as xenobiotics (Holst &

Williamson, 2008), it is reasonable to propose that the transient anti-OSI benefits of phytochemicals are stimulated by the presence of the phytochemicals and their metabolites and are resolved once they are eliminated (Fig. 4C, 5C and 6C–D). The functional fingerprints of blueberry and green tea analysed in the present study were used to explain results of clinical trials reporting their effects on regulation of OSI in human

(Fig. 4C, 5C and 6C–D). In all four clinical trials, the bio-efficacy of blueberry (Del Bo, et al., 2013; Kay & Holub, 2002) and green tea (Ho, Choi, Siu, & Benzie, 2014; Leenen,

Roodenburg, Tijburg, & Wiseman, 2000) on regulation of OSI was only observed at the time matching the Tmax ranges determined by their functional fingerprints. These findings provide convincing evidence that the bio-efficacy of phytochemicals in the human body is dependent on their transient presence and specific Tmax. Therefore, the functional fingerprint of a plant extract is an important feature that should be taken in account when designing clinical studies to ensure that the phytochemical intake is aligned to the time that anti-OSI activity is required, and also to monitoring effects, including blood sampling. These tools are expected to significantly impact and improve capacity for phytochemicals to optimise anti-OSI activity, with broad significance for human health.

5. Conclusion

Human upper intestinal absorption of plant extracts can be predicted using LC-MS analysis in combination with the statistical PCAP model. The data processing method utilises an untargeted approach that does not rely on compound identification and therefore is not limited by the availability of standards. The predicted absorption profile obtained by this method provides a unique functional fingerprint for each individual plant extract showing the time required for the phytochemical mixtures to reach their maximal plasma concentration. Considering that phytochemicals are eliminated by the human body as xenobiotics, the functional fingerprint of a plant extract is an important feature that should be taken in account when designing clinical studies to ensure that the phytochemicals are present in circulation at the time of biological need so as to exert maximal protection. We propose that functional fingerprinting of plant extracts provides a novel and useful tool for understanding and optimising the health benefits of plant extracts.

Acknowledgments

This project has been funded by Horticulture Innovation Australia Limited using the

Vegetable levy and funds from the Australian Government. Metabolite analysis was conducted at Metabolomics Australia (The University of Melbourne, Australia), a NCRIS initiative under Bioplatforms Australia Pty Ltd.

Conflict of interest

The authors declare no conflict of interests.

References

Bazzano, L. A., Li, T. Y., Joshipura, K. J., & Hu, F. B. (2008). Intake of fruit, vegetables, and fruit juices and risk of diabetes in women. Diabetes Care, 31(7), 1311-1317. Bennett, L. E., & Muench, G. (2011). Process for plant bioactive enrichment. In, vol. PCT/AU2010/0015102010). Bennett, L. E., Singh, D. P., & Clingeleffer, P. R. (2011). Micronutrient mineral and folate content of Australian and imported dried fruit products. Critical Reviews in Food Science and Nutrition, 51(1), 38-49. Burton-Freeman, B. (2010). Postprandial metabolic events and fruit-derived phenolics: a review of the science. British Journal of Nutrition, 104 Suppl 3, S1-14. Calder, P. C., Albers, R., Antoine, J. M., Blum, S., Bourdet-Sicard, R., Ferns, G. A., Folkerts, G., Friedmann, P. S., Frost, G. S., Guarner, F., Lovik, M., Macfarlane, S., Meyer, P. D., M'Rabet, L., Serafini, M., van Eden, W., van Loo, J., Vas Dias, W., Vidry, S., Winklhofer-Roob, B. M., & Zhao, J. (2009). Inflammatory disease processes and interactions with nutrition. British Journal of Nutrition, 101 Suppl 1, S1-45. Camp, D., Campitelli, M., Carroll, A. R., Davis, R. A., & Quinn, R. J. (2013). Front- loading natural-product-screening libraries for log P: background, development, and implementation. Chemistry & Biodiversity, 10(4), 524-537. Crozier, A., Jaganath, I. B., & Clifford, M. N. (2006). Phenols, polyphenols and tannins: an overview. In A. Crozier, M. N. Clifford & H. Ashihara (Eds.), Plant Secondary Metabolites: Occurrence, Structure and Role in the Human Diet). Oxford: Blackwell Publishing Ltd. D'Onofrio, G., Sancarlo, D., Ruan, Q., Yu, Z., Panza, F., Daniele, A., Greco, A., & Seripa, D. (2016). Phytochemicals in the treatment of Alzheimer's disease: a systematic review. Current Drug Targets. D’Archivio, M., Filesi, C., Varì, R., Scazzocchio, B., & Masella, R. (2010). Bioavailability of the polyphenols: status and controversies. International Journal of Molecular Sciences, 11(4), 1321-1342. Dauchet, L., Amouyel, P., & Dallongeville, J. (2009). Fruits, vegetables and coronary heart disease. Nature Reviews Cardiology, 6(9), 599-608. Del Bo, C., Riso, P., Campolo, J., Moller, P., Loft, S., Klimis-Zacas, D., Brambilla, A., Rizzolo, A., & Porrini, M. (2013). A single portion of blueberry (Vaccinium corymbosum L) improves protection against DNA damage but not vascular function in healthy male volunteers. Nutrition Research, 33(3), 220-227. Del Rio, D., Stewart, A. J., Mullen, W., Burns, J., Lean, M. E., Brighenti, F., & Crozier, A. (2004). HPLC-MSn analysis of phenolic compounds and purine alkaloids in green and black tea. Journal of Agricultural and Food Chemistry, 52(10), 2807- 2815. Flamini, R., De Rosso, M., De Marchi, F., Dalla Vedova, A., Panighel, A., Gardiman, M., Maoz, I., & Bavaresco, L. (2013). An innovative approach to grape metabolomics: stilbene profiling by suspect screening analysis. Metabolomics, 9(6), 1243-1253. Gao, L., & Mazza, G. (1994). Quantitation and distribution of simple and acylated anthocyanins and other phenolics in blueberries. Journal of Food Science, 59(5), 1057-1059. Hartmann, T. (2007). From waste products to ecochemicals: Fifty years research of plant secondary metabolism. Phytochemistry, 68(22–24), 2831-2846.

Ho, C. K., Choi, S. W., Siu, P. M., & Benzie, I. F. (2014). Effects of single dose and regular intake of green tea (Camellia sinensis) on DNA damage, DNA repair, and heme oxygenase-1 expression in a randomized controlled human supplementation study. Molecular Nutrition and Food Research, 58(6), 1379-1383. Holst, B., & Williamson, G. (2008). Nutrients and phytochemicals: from bioavailability to bioefficacy beyond antioxidants. Current Opinion in Biotechnology, 19(2), 73- 82. Huang, Y., Park, E., Edirisinghe, I., & Burton-Freeman, B. M. (2016). Maximizing the health effects of strawberry anthocyanins: understanding the influence of the consumption timing variable. Food and Function, 7, 4745-4752. James, C. S. (1995). Theory of analytical methods for specific food constituents. In C. S. James (Ed.), Analytical Chemistry of Foods, (pp. 37-67). New York: Springer. Kay, C. D., & Holub, B. J. (2002). The effect of wild blueberry (Vaccinium angustifolium) consumption on postprandial serum antioxidant status in human subjects. British Journal of Nutrition, 88(04), 389-397. Key, T. J. (2011). Fruit and vegetables and cancer risk. British Journal of Cancer, 104(1), 6-11. Kryston, T. B., Georgiev, A. B., Pissis, P., & Georgakilas, A. G. (2011). Role of oxidative stress and DNA damage in human carcinogenesis. Mutation Research/Fundamental and Molecular Mechanisms of Mutagenesis, 711(1–2), 193-201. Kutchan, T. M., Gershenzon, J., Moller, B. L., & Gang, D. R. (2015). Natural products. In B. B. Buchanan, W. Gruissem & R. L. Jones (Eds.), Biochemistry & Molecular Biology of Plants, (pp. 1132-1206). Chichester: John Wiley & Sons. Leenen, R., Roodenburg, A. J., Tijburg, L. B., & Wiseman, S. A. (2000). A single dose of tea with or without milk increases plasma antioxidant activity in humans. European Journal of Clinical Nutrition, 54(1), 87-92. Makrides, M., Neumann, M. A., & Gibson, R. A. (1996). Effect of maternal docosahexaenoic acid (DHA) supplementation on breast milk composition. European Journal of Clinical Nutrition, 50(6), 352-357. Muraki, I., Imamura, F., Manson, J. E., Hu, F. B., Willett, W. C., van Dam, R. M., & Sun, Q. (2013). Fruit consumption and risk of type 2 diabetes: results from three prospective longitudinal cohort studies. BMJ, 347, f5001. Njinkoue, J. M., Gouado, I., Tchoumbougnang, F., Ngueguim, J. H. Y., Ndinteh, D. T., Fomogne-Fodjo, C. Y., & Schweigert, F. J. (2016). Proximate composition, mineral content and fatty acid profile of two marine fishes from Cameroonian coast: Pseudotolithus typus (Bleeker, 1863) and Pseudotolithus elongatus (Bowdich, 1825). NFS Journal, 4, 27-31. O’Keefe, J. H., & Bell, D. S. H. (2007). Postprandial hyperglycemia/hyperlipidemia (postprandial dysmetabolism) is a cardiovascular risk factor. The American Journal of Cardiology, 100(5), 899-904. Patrick, J. W., Botha, F. C., & Birch, R. G. (2013). Metabolic engineering of sugars and simple sugar derivatives in plants. Plant Biotechnology Journal, 11, 142-156. Pichersky, E., & Gang, D. R. (2000). Genetics and biochemistry of secondary metabolites in plants: an evolutionary perspective. Trends in Plant Science, 5(10), 439-445. Sandhu, A. K., Huang, Y., Xiao, D., Park, E., Edirisinghe, I., & Burton-Freeman, B. (2016). Pharmacokinetic characterization and bioavailability of strawberry anthocyanins relative to meal intake. Journal of Agricultural and Food Chemistry, 64(24), 4891-4899.

Selby-Pham, S. N. B., Cottrell, J. J., Dunshea, F., Ng, K., Bennett, L. E., & Howell, K. (2017). Dietary phytochemicals promote health by enhancing antioxidant defence in a pig model. Nutrients, 9(7), 758. Selby-Pham, S. N. B., Miller, R. B., Howell, K., Dunshea, F., & Bennett, L. E. (2017). Physicochemical properties of dietary phytochemicals can predict their passive absorption in the human small intestine. Scientific Reports, 7, 1931. Singleton, V. L., & Rossi, J. A. (1965). Colorimetry of total phenolics with phosphomolybdic-phosphotungstic acid reagents. American Journal of Enology and Viticulture, 16(3), 144-158. Sosulski, F. W., & Imafidon, G. L. (1990). Amino acid composition and nitrogen-to- protein conversion factos for animal and plant foods. Journal of Agricultural and Food Chemistry, 38, 1351-1356. Tsao, R., & Li, H. (2013). Analytical techniques for phytochemicals. In B. K. Tiwari, N. P. Brunton & C. S. Brennan (Eds.), Handbook of Plant Food Phytochemicals, (pp. 434-451). Hoboken: John Wiley & Sons Ltd. Valko, K. (2004). Application of high-performance liquid chromatography based measurements of lipophilicity to model biological distribution. Journal of Chromatography A, 1037(1-2), 299-310. Waksmundzka-Hajnos, M., & Sherma, J. (2010). Overview of the field of high performance liquid chromatography in phytochemical analysis and the structure of the book. In M. Waksmundzka-Hajnos & J. Sherma (Eds.), High Performance Liquid Chromatography in Phytochemical Analysis, (pp. 3-12). Boca Raton: CRC Press. Wolfender, J. L., Ndjoko, K., & Hostettmann, K. (2001). The potential of LC-NMR in phytochemical analysis. Phytochemical Analysis, 12(1), 2-22. Wu, H., Guo, J., Chen, S., Liu, X., Zhou, Y., Zhang, X., & Xu, X. (2013). Recent developments in qualitative and quantitative analysis of phytochemical constituents and their metabolites using liquid chromatography-mass spectrometry. J. Pharm. Biomed. Anal., 72, 267-291.

Figure Captions

Figure 1. Translation of physicochemical data obtained from LC-MS analysis into predicted phytochemical absorption ‘functional fingerprints’ of plant extracts. (A)

Flowchart showing data processing workflow of LC-MS data. (B) Calibration of retention time of standards analysed by LC-MS for determination of log P of individual phytochemicals. Data represents mean of positive and negative mode retention times after single injections. (C) Relationship between total phenolics and LC-MS relative quantification of secondary metabolites as total peak area (relative ion count) of selected plant extracts.

Figure 2. Predicted ‘functional fingerprint’ of selected plant extracts if consumed in drink form. Values of molecular mass and log P of primary and secondary metabolites obtained from LC-MS were used to calculate their respective predicted plasma Tmax (h) which were plotted against their respective relative abundance for (A) broccoli, (B) carrot,

(C) kale (D), Vietnamese coriander, (E) red sweet potato, (F) rhubarb, (G) eggplant, (H) squash, (I) olive leaf and (J) tomato.

Figure 3. Predicted ‘functional fingerprint’ of selected plant extracts if consumed in solid form. Values of molecular mass and log P of primary and secondary metabolites obtained from LC-MS were used to calculate their respective predicted plasma Tmax (h) which were plotted against their respective relative abundance for (A) broccoli, (B) carrot,

(C) kale (D), Vietnamese coriander, (E) red sweet potato, (F) rhubarb, (G) eggplant, (H) squash, (I) olive leaf and (J) tomato.

Figure 4. Relationship of predicted Tmax range of blueberry consumed in drink form with measured bio-efficacy for regulating measures of oxidative stress and inflammation (OSI) in human subjects. Functional fingerprints showing range of Tmax

(h) predicted from molecular mass and log P data obtained from either (A) LC-MS method or (B) analysis of phytochemicals in blueberry reported by Gao and Mazza

(1994). (C) Bio-efficacy of blueberry in regulation of OSI associated with a high fat (HF) meal is indicated by significant change in biomarkers of anti-OSI (aOSI) compared to baseline, as reported by Kay and Holub (2002). The aOSI biomarker measured in this study was serum anti-oxidant status using the oxygen radical absorbance capacity

(ORAC) assay.

Figure 5. Relationship of predicted Tmax range of blueberry consumed in paste form with measured bio-efficacy for regulating measures of oxidative stress and inflammation (OSI) in human subjects. Functional fingerprints showing range of Tmax

(h) predicted from molecular mass and log P data obtained from either (A) LC-MS method or (B) analysis of phytochemicals in blueberry reported by Gao and Mazza

(1994). (C) Bio-efficacy of OSI regulation by blueberry is indicated by significant change in biomarkers of OSI compared to baseline, as reported by Del Bo, et al. (2013). The OSI biomarker measured in this study was DNA damage using the comet assay.

Figure 6. Relationship of predicted Tmax range of green tea consumed in drink form with measured bio-efficacy for regulating measures of oxidative stress and inflammation (OSI) in human subjects. Functional fingerprints showing range of Tmax

(h) predicted from molecular mass and log P data obtained from either (A) LC-MS method or (B) analysis of phytochemicals in green tea reported by Del Rio, et al. (2004).

Bio-efficacy of OSI regulation by green tea is indicated by significant change in biomarkers of OSI and aOSI compared to baseline, as reported by (C) Ho, Choi, Siu, and

Benzie (2014), the OSI measured in this study was DNA damage using the comet assay, the aOSI was measured by DNA repair enzyme activity; and (D) Leenen, Roodenburg,

Tijburg, and Wiseman (2000), phytochemical (PC) in this study was measured by plasma catechin, the aOSI was measured by plasma anti-oxidant status using the ferric antioxidant power (FRAP) assay.

Figure 1.

B C 8 R2 = 0.98 4.5e+8 6 4.0e+8

4 3.5e+8 3.0e+8 Pearson’s correlation 2 2.5e+8 coefficient r = 0.77,

Log P Log 0 2.0e+8 P = 0.003 1.5e+8 -2 1.0e+8

-4 count) ion area (relative peak Total 5.0e+7 0 2 4 6 8 10 12 14 0 50 100 150 200 250 Retention time (min) Total phenolic content (mg GAE/g)

Figure 2. A B Primary metabolites Primary metabolites Secondary metabolites Secondary metabolites

Peak area (relative ion count) (relative area Peak

Peak area (relative ion count) (relative area Peak 0 2 4 6 8 10 12 0 2 4 6 8 10 12 Plasma T (h) Plasma T (h) max max

C Primary metabolites D Primary metabolites Secondary metabolites Secondary metabolites

Peak area (relative ion count) (relative area Peak

Peak area (relative ion count) (relative area Peak 0 2 4 6 8 10 12 0 2 4 6 8 10 12 Plasma T (h) Plasma T (h) max max

E Primary metabolites F Primary metabolites Secondary metabolites Secondary metabolites

Peak area (arbitrary units) (arbitrary area Peak

Peak area (relative ion count) (relative area Peak

0 2 4 6 8 10 12 0 2 4 6 8 10 12 Plasma T (h) Plasma T (h) max max

G Primary metabolites H Primary metabolites Secondary metabolites Secondary metabolites

Peak area (relative ion count) (relative area Peak

Peak area (relative ion count) (relative area Peak 0 2 4 6 8 10 12 0 2 4 6 8 10 12 Plasma T (h) Plasma T (h) max max

I Primary metabolites J Primary metabolites Secondary metabolites Secondary metabolites

Peak area (relative ion count) (relative area Peak

Peak area (relative ion count) (relative area Peak 0 2 4 6 8 10 12 0 2 4 6 8 10 12 Plasma T (h) Plasma T (h) max max

Figure 3.

A B Primary metabolites Primary metabolites Secondary metabolites Secondary metabolites

Peak area (relative ion count) (relative area Peak

Peak area (relative ion count) (relative area Peak 0 2 4 6 8 10 12 0 2 4 6 8 10 12 Plasma T (h) Plasma T (h) max max

C Primary metabolites D Primary metabolites Secondary metabolites Secondary metabolites

Peak area (relative ion count) (relative area Peak

Peak area (relative ion count) (relative area Peak 0 2 4 6 8 10 12 0 2 4 6 8 10 12 Plasma T (h) Plasma T (h) max max

E Primary metabolites F Secondary metabolites Primary metabolites Secondary metabolites

Peak area (relative ion count) (relative area Peak

0 2 4 6 8 10 12 0 2 4 6 8 10 12 Plasma T (h) Plasma T (h) max max

G Primary metabolites H Primary metabolites Secondary metabolites Secondary metabolites

Peak area (relative ion count) (relative area Peak

0 2 4 6 8 10 12 0 2 4 6 8 10 12 Plasma T (h) Plasma T (h) max max

I Primary metabolites J Primary metabolites Secondary metabolites Secondary metabolites

Peak area (relative ion count) (relative area Peak

0 2 4 6 8 10 12 0 2 4 6 8 10 12 Plasma T (h) Plasma T (h) max max

Figure 4.

A B 8 Extraction: MeOH/H2O Extraction: MeOH/acid/H2O Mass 122 to 1258, Mass 354 to 535, log P -2 to 6.9 log P -3 to -0.5 Mass 354 to 535, 6 log P -2 to -0.5

Peak area (relative ion count) ion (relative area Peak

Abundance (mg/g dry solids) dry (mg/g Abundance 0 0 2 4 6 8 10 12 0 2 4 6 8 10 12 Plasma T (h) Plasma T (h) max max C

INTAKE: INTAKE HF meal + Blueberry (100g ↑aOSI ↔aOSI ↔ aOSI ↔ aOSI powder in 500 mL water) 0 h 1 h 2 h 3 h 4 h 5 h 6 h 7h 8 h 9 h Tmax Tmax Tmax

INTAKE: Blood HF meal Sampling

↔aOSI ↔aOSI ↔ aOSI ↔ aOSI INTAKE

Figure 5.

A B 8 Extraction: MeOH/H2O Extraction: MeOH/acid/H2O Mass 122 to 1258, Mass 354 to 535, log P -2 to 6.9 log P -3 to -0.5 Mass 354 to 535, 6 log P -2 to -0.5

Abundance (mg/g dry solids) dry (mg/g Abundance

Peak area (relative ion count) ion (relative area Peak 0 0 2 4 6 8 10 12 0 2 4 6 8 10 12 Plasma T (h) Plasma T (h) max max

INTAKE: INTAKE Blueberry paste (300g) ↓ OSI ↔ OSI ↔ OSI

0 h 1 h 2 h 3 h …. 24 h Tmax Tmax Tmax

Blood Sampling

Figure 6.

A B Extraction: H O Extraction: MeOH/H O 2 2 30 Mass: 170 to 611, Mass 122 to 1258, log P -1.2 to 2.5 log P -2 to 6.9 Mass 170 to 611, log P -1.2 to 2.5 20

Abundance (mg/g dry leaves) dry (mg/g Abundance

Peak area (relative ion count) ion (relative area Peak 0 0 2 4 6 8 10 12 0 2 4 6 8 10 12 Plasma T (h) Plasma T (h) max max C

INTAKE: INTAKE Green tea ↓OSI ↓OSI (200 mL of ↑aOSI ↑aOSI 1.5% w/v)

0 h 0.5 h 1 h 1.5 h 2 h … 8 h … 10 h 11 h Tmax Tmax

Blood Sampling

INTAKE: INTAKE Green tea ↑PC ↑PC ↑PC ↑PC (2g in 300 mL water) ↔aOSI ↑aOSI ↑aOSI ↑aOSI

0 h 0.5 h 1 h 1.5 h 2 h … 8 h … 10 h 11 h Tmax Tmax

Blood Sampling

Chapter 4: Transport rates of dietary phytochemicals in cell monolayers is inversely correlated with absorption kinetics in humans

Chapter 4: Transport Rates of Dietary Phytochemicals in Cell Monolayers Is Inversely Correlated with Absorption Kinetics in Humans

4.1 Introduction

The intestinal epithelium represents a physical barrier that separates the contents of the gut lumen and the blood stream, and therefore plays an important role in uptake of nutrients. Accordingly, in vitro cell-based models of the small intestinal epithelium have been used to study absorption of pharmaceuticals and phytochemicals. It is understood that the apparent permeability (Papp) of a compound across the in vitro cell monolayer is positively correlated with its in vivo bioavailability i.e. the % absorbed after oral consumption in humans. However, Papp has not previously been related to kinetics of human absorption (Tmax), as predicted by the in silico statistical PCAP model.

In this chapter, the research hypothesis was that in vitro Papp of phytochemicals can be correlated with their in silico human Tmax, according to predictions of the PCAP model.

The aim of this chapter was to characterise the absorption kinetics of selected phytochemical mixtures in vitro using a cell model of epithelial transport. A further aim was to compare the in vitro cell model Papp of pure standards to their in silico predictions of human Tmax calculated from the in silico statistical PCAP model described in Chapter

Chapter 4: Transport rates of dietary phytochemicals in cell monolayers is inversely correlated with absorption kinetics in humans

The aims were fulfilled via completion of the following objectives:

 Application of the co-culture monolayers model using Caco-2/HT29-MTX-E12

cells to investigate the transport kinetics of pure standards and selected

phytochemical mixtures across the monolayers.

 Determination of a relationship between the in vitro Papp of pure standards and the

in silico prediction of human Tmax, thereby enabling the cell model to predict Tmax

of phytochemical mixtures from their in vitro Papp.

4.2 Accepted manuscript

Title: Transport rates of dietary phytochemicals in cell

monolayers is inversely correlated with absorption kinetics

in humans

Authors: Sophie N. B. Selby-Phama,b, Simone A. Osbornec, Kate S.

Howella, Frank R. Dunsheaa and Louise E. Bennettb,*

Affiliations: aFaculty of Veterinary and Agricultural Science,

The University of Melbourne,

Parkville, Victoria, 3010, Australia

bCSIRO Agriculture and Food,

671 Sneydes Road, Werribee,

Victoria, 3030, Australia

cCSIRO Agricultural and Food,

306 Carmody Road, St Lucia,

Queensland, 4067, Australia

*Corresponding Professor Louise Bennett

Author: School of Chemistry, Monash University

Clayton, Victoria, 3800, Australia

E-mail: [email protected]

Phone: +613 9905 4593

Abstract

Dietary phytochemicals promote health and reduce the risk of chronic diseases. The

Phytochemical Absorption Prediction Model (PCAP) predicts the time required for phytochemicals to reach maximal plasma concentrations (Tmax) in humans based on their lipophilicity and molecular mass. Cell-based transport models have been used to quantify transport rate and efficiency of pharmaceuticals and phytochemicals, however these parameters have not previously been related to the human absorption Tmax. Caco-2/HT29-

MTX-E12 monolayers were used to characterise transport of phytochemical standards and extracts and to establish a relationship between the in vitro permeability (Papp) of standards and their in vivo Tmax predicted from the PCAP model. Lipophilic compounds transported through the cell monolayer at relatively faster rates (higher Papp) than hydrophilic compounds, whilst having slower predicted in vivo absorption rates (longer

Tmax). The results infer differences between in vitro (cell monolayer) and in vivo (human gastrointestinal tract) absorption kinetics of phytochemicals.

Keywords: molecular mass, log P, small intestine, Caco-2, HT29-MTX-E12, phytochemical absorption prediction

1. Introduction

Diets rich in fruits and vegetables promote health and are associated with a reduced risk of chronic diseases including obesity (Jarzab & Kukula-Koch, 2017), diabetes

(Stravodimos et al., 2017), cancer (Key, 2011), cardiovascular (Dauchet, Amouyel, &

Dallongeville, 2009) and neurodegenerative diseases (D'Onofrio et al., 2017). The protection provided by these plant-based foods has been attributed to their high levels of phytochemicals that help to regulate oxidative stress and inflammation (OSI) (Del Rio et al., 2013; Schinella, Tournier, Prieto, de Buschiazzo, & Rı́os, 2002). Elevated OSI is persistent in individuals suffering from chronic diseases (Calder et al., 2009) and also associated with normal activities including exercise (van der Merwe & Bloomer, 2016) and meal digestion (Burton-Freeman, 2010) in healthy individuals.

Uptake of phytochemicals into circulation and their absorption by target cells is necessary for providing biological benefits (Lee, 2013). However, phytochemicals are recognised by the body as xenobiotics resulting in their low bioavailability and transient presence

(Holst & Williamson, 2008). Depending on their chemical structures and dietary intake forms, the time required for phytochemicals to reach maximal plasma concentrations

(Tmax) can be 1–2 hour (h) or 15–33 h post consumption and are completely cleared over the next few hours or days, respectively (Gustin et al., 2004; Stalmach, Troufflard,

Serafini, & Crozier, 2009). Therefore, Tmax is important to define the optimal temporal window to observe the bioefficacy of phytochemicals. For example, a strawberry drink with Tmax of the associated phytochemicals of 1–2 h was reported to significantly attenuate the OSI induced by a high-fat meal when the drink was consumed 2 h before the meal and not with the meal or after the meal (Huang, Park, Edirisinghe, & Burton-

Freeman, 2016; Sandhu et al., 2016).

Following ingestion, absorption of some phytochemicals into the blood stream occurs in the small intestine via passive diffusion or carrier mediated routes (Donovan, Manach,

Faulks, & Kroon, 2006). Uptake routes of passive diffusion are dependent on physicochemical properties of phytochemicals such as lipophilicity and molecular mass.

Lipophilic compounds diffuse through the lipid core of the membrane (transcellular diffusion) whereas hydrophilic compounds pass through the water-filled junctions between adjacent cells (paracellular diffusion) (Artursson, Palm, & Luthman, 2001). The molecular mass of phytochemicals also affects passive absorption as large molecules are limited to diffusing through the membrane by their size (Lipinski, Lombardo, Dominy,

& Feeney, 1997). Alternatively, phytochemicals can be transported across the intestinal epithelium by active carrier-mediated pathway, however the mechanisms and extent of transport via this pathway is largely unknown (Day, Gee, DuPont, Johnson, &

Williamson, 2003; Kottra & Daniel, 2007).

A model to predict passive absorption of phytochemicals in humans was recently reported

(Selby-Pham, Miller, Howell, Dunshea, & Bennett, 2017). This model, referred to as the phytochemical absorption prediction (PCAP) model, permits direct calculation of phytochemicals’ Tmax based on their lipophilicity descriptor log P and molecular mass.

The PCAP model is applicable for phytochemicals that are passively absorbed in the small intestine and does not account for absorption pathways involving either chemical modification by brush border enzymes, nor fermentation by the gut microflora. Further, the PCAP model can be applied to individual or mixtures of phytochemicals as isolated compounds or as present in native matrices of vegetables and whole fruits (Selby-Pham,

Miller, et al., 2017). The PCAP model is unique in its ability to account for the interaction

of phytochemicals with macronutrients such as protein and fibre, and to predict phytochemical absorption kinetics following oral intake in humans.

The human colorectal adenocarcinoma cell line, Caco-2 has been widely used to study intestinal passive absorption of drugs (Artursson, 1990; Hubatsch, Ragnarsson, &

Artursson, 2007) and phytochemicals (Boyer, Brown, & Liu, 2004; Liu, Glahn, & Liu,

2004). The apparent permeability (Papp) of compounds across the Caco-2 monolayer has been positively correlated with their bioavailability in human in vivo, as indicated by % absorption i.e., the fraction of compounds absorbed after oral administration in comparison to the intake dose (Artursson & Karlsson, 1991; Cheng, Li, & Uss, 2008).

Additionally, Papp of several drug compounds and phenolic acids across the Caco-2 monolayer via the transcellular diffusion pathway were dependent on their lipophilicity descriptor log D and molecular mass (Farrell, Poquet, Dew, Barber, & Williamson, 2012).

This relationship was successfully developed into a predictive model showing that the transcellular Papp of compounds decreased with increasing molecular mass and reducing lipophilicity. However, the relationship between in vitro Papp and in vivo Tmax has not been previously investigated.

The aim of this study was to use the co-cultured monolayers of the Caco-2 cells and the mucus-producing human colorectal adenocarcinoma HT29-MTX-E12 cells as an improved model of the human small intestinal epithelium (Walter, Janich, Roessler,

Hilfinger, & Amidon, 1996), to determine apparent permeability (Papp) of pure standards and phytochemical extracts across the co-cultured cell monolayer. Further, this study then investigated the relationship of respective measures of in vitro Papp and in vivo Tmax predicted by the PCAP model.

2. Materials and Methods

2.1. Cell culture and materials

Human colorectal adenocarcinoma Caco-2 (ATCC no. HTB-37) and HT29-MTX-E12

(ATCC no. HTB-38) cell lines were purchased from the American Type Culture

Collection (ATCC, Rockville, MD, USA) and Sigma-Aldrich (St Louise, MO, USA), respectively. Dulbecco’s modified eagle’s medium (DMEM) with high glucose (4500 mg/mL), heat-inactivated fetal bovine serum (FBS), nonessential amino acids (NEAA),

Hank’s balanced salt solutions (HBSS) and penicillin/streptomycin were from Gibco-Life

Technology (Rockville, MD, USA). Dimethyl sulfoxide (DMSO) was from Sigma-

Aldrich (St Louise, MO, USA) and gradient grade methanol was from Merck (Darmstadt,

Germany). Dimethylthiazol carboxymethoxyphenyl sulfophenyl tetrazolium (MTS) and phenazine methosulfate (PMS) reagents were from Promega (Madision, WI, USA).

Transwell® 24-well plates including permeable supports with polycarbonate membranes

(0.33 cm2 growth area) were purchased from Corning (Sydney, Australia) and 96-well plates were from Nunc (Roskilde, Denmark).

Caco-2 and HT29-MTX-E12 cells were cultured separately in DMEM supplemented with

10% FBS, 1% w/v penicillin/streptomycin and 1% w/v NEAA. All cells were grown at

o ® 37 C/5% CO2 in a humidified atmosphere. Cells grown in Transwell at passage 5–20 were used for the experiment.

Standards for cell transport experiment covering a dynamic range of log P from -3.5 to

6.7 included ascorbic acid, gallic acid, thiamine HCl, curcumin (Sigma-Aldrich, St

Louise, MO, USA) and bromophenol blue (Bio-Rad Laboratories, Richmond, CA, USA).

Two groups of plant extracts were used in this study. Group one, referred to as ‘project extracts’ included ten vegetable extracts made from fresh materials purchased from local

retailers (Woolworths, Werribee, VIC, Australia): broccoli (Brassica oleracea var. italica), carrot (Daucus carota ssp. sativus), red cabbage (Brassica oleracea var. italica), red sweet potato (Impomoea batatas), rhubarb (Rheum rhabarbarum), squash (Cucurbita pepo var. ovifera), eggplant (Solanum melongena), white zucchini (Cucurbita pepo var. melopepo), kale (Brassica oleracea var. acephala) and Vietnamese coriander (Persicaria odorata). Vegetables were subjected to Stage 1 of a three-stage generic processing as described previously (Bennett & Muench, 2011). Group two, referred to as ‘reference extracts’ included seven commercial extracts: blueberry powder (Super Sprout,

Campbellfield, Victoria, Australia), cacao powder (TRU-RA Cacao, Big Tree Farm,

Ashland, OR, USA), grape seed and grape skin extracts (Grapex, Tarac Technologies,

Nuriootpa, South Australia, Australia), green tea powder (Absolute Green, DeDu Pty

Ltd., Ermington, NSW, Australia ), olive leaf powder (Austral herbs, Uralla, NSW,

Australia) and tomato powder (Herbies’s Spices, Rozelle, NSW, Australia).

Characterisation of the project and reference extracts was performed including total phenolics content and proximate analyses with results reported elsewhere (Selby-Pham,

Howell, et al., 2017).

2.2. Determination of standards and plant extracts toxicity

Prior to the cell monolayer transport experiments, cell viability in response to standards and plant extracts was analysed to ensure that the concentrations of standards and plant extracts to be used in the cell monolayer transport experiment were nontoxic for the cells.

Cell viability was evaluated by MTS assay using a CellTiter 96 Aqueous One Solution

Cell Proliferation Assay (Promega, Madision, WI, USA) according to manufacturer’s directions. Briefly, 1x104 Caco-2/HT29-MTX-E12 (ratio 9:1) cells per well were grown in a flat bottom 96 well plate for 7 days. On day 7, standards and plant extracts were

prepared in 1% DMSO/ HBSS at concentrations of 0.05–1 mg/mL and centrifuged at

15000 × g for 5 min (Model 5417R, Eppendorf AG, Hamburg, Germany) to obtain the supernatants. Growth media was removed and replaced with the supernatants of standards

o and plant extracts prior to incubation at 37 C/ 5% CO2 for 60 min. The cells were then washed once in HBSS and assayed using 100 μL HBSS with 20 μL MTS/PMS reagents.

o Absorbance at 492 nm was monitored for 1–4 h of 37 C/ 5% CO2 incubation as an indicator of cell viability. Confluency of the cell monolayers was also monitored prior to,

24 h and 48 h following the transport experiments by measuring the transepthelial electrical resistance (TEER) using a Millicell-ERS Volt-ohm meter (Millipore, Bedford,

MA, USA). Concentrations of standards and plant extracts resulting in cell viability >

90% were chosen for the cell monolayer transport experiment. Accordingly, for the cell monolayer transport experiment, the chosen concentrations of standards were 0.09 mg/mL ascorbic acid, 0.09 mg/mL gallic acid, 0.13 mg/mL thiamine hydrochloride, 0.07 mg/mL curcumin and 0.07 mg/mL bromophenol blue. The chosen concentrations of project extracts and reference extracts were 1 mg/mL, except for grape seed extract at 0.5 mg/mL. All standards and extracts were prepared in 1% DMSO/ HBSS and centrifuged at 15000 × g for 5 min to obtain the supernatants to be used in the cell monolayer experiment.

2.3. Cell monolayer transport of standards and plant extracts

Prior to seeding, Caco-2 and HT29-MTX-E12 cells were mixed to yield cell ratio of 9:1, respectively. The co-cultures from Caco-2/HT29-MTX-E12 cells (1.2 × 105 cells/cm2) were then seeded into Transwell® apparatus and cultured in growth media for 21 days to facilitate cell differentiation and formation of an intact monolayer. During the 21-day differentiation period, growth media was removed and replaced with fresh media every

2-3 days. Cell differentiation and formation of an intact monolayer was monitored by measuring the TEER in an apical to basolateral direction.

On the day of the monolayer transport experiments (day 21), cells were depleted for 2 h by replacing the growth media with transport medium HBSS to enhance uptake.

Schematic diagram of the experiment is shown in Fig. 1a. A 200 μL of standards or plant extracts were applied to the apical chamber and 600 μL of HBSS was placed in the basolateral chamber. Subsamples of 110 μL were collected from the basolateral chamber and replaced with pre-warmed HBSS at 30 min intervals for up to 5 h for standards and

4 h for plant extracts. HBSS was used as the control to determine the baseline activity of cells in the absence of sample. The monolayer transport experiment was performed in 4 independent biological replicates.

2.4. Analysis of standards and plant extracts transported through the cell monolayer

The transport of compounds present in the plant extracts and standards through the cell monolayer, herein referred to as ‘solids transported’, were analysed by quantifying the basolateral and apical samples by high performance liquid chromatography (HPLC).

Analyses were performed on an automated Waters Alliance series 2690 HPLC system

(Watford, Hertfordshire, UK) connected to a Phenomenex guard column, 35 mm x 7.80 mm (Phenomenex, Torrance, CA, USA). The samples were injected at 100 μL and 20 μL for the basolateral and apical samples, respectively. The mobile phase was 50% methanol at a flow rate of 0.5 ml/min for 5 min. Absorbance at 254 nm was monitored by a Waters

2487 dual wavelength absorbance detector (Watford, Torrance, CA, USA). Total peak area (TPA) was determined as an indicator of the total amount of standards or plant extracts in the samples after correction for reagent control.

Ultraviolet absorptivity characteristics of standards and plant extracts were quantified to standardise and convert TPA to relative mass. For each sample, absorbance at 254 nm was determined as the slope of the linear regression of absorbance and concentrations of samples, based on the Beer–Lambert’s law: A = ε × l × C, where A is the absorbance at

254 nm quantified by TPA, ε is the absorptivity coefficient, l is the light path length through the sample and C is the concentration of the sample (Swinehart, 1962).

2.5. Calculation of apparent permeability coefficient (Papp)

The rate of standards and plant extracts transported across the Caco-2/HT29-MTX-E12 monolayer was assessed by calculating the apparent permeability coefficient (Papp, cm/h) as previously described (Hubatsch et al., 2007): Papp = (dQ/dt) x (1/A × 1/Co). Where dQ/dt is the change of accumulative mass in the basolateral chamber over time (µg/h); A

® 2 is the surface area of the Transwell membrane (cm ) and Co is the initial concentration of standards or plant extracts in the apical chamber (µg/mL). Accumulative TPAs of the transported samples in the basolateral chamber were plotted over time and fitted to linear regressions to determine dQ/dt. The slope from the linear regression of the blank was subtracted from the slopes of sample linear regressions. These blank-corrected slopes were then standardised to account for variations in average absorptivity at 254 nm. The conversion from TPA to Papp was demonstrated as a schematic diagram in Fig. 1b.

2.6. Relationship between cell monolayer transport and human absorption

The relationship of the cell monolayer Papp and the in vivo Tmax in healthy humans calculated from the PCAP model (Selby-Pham, Miller, et al., 2017) was established. The predicted in vivo Tmax of five standards was calculated from their molecular mass and log

P according to the PCAP model (Selby-Pham, Miller, et al., 2017). The relationship of the cell monolayer Papp and the predicted Tmax of five standards was fitted to a linear

regression. This relationship was then applied to calculate predicted average Tmax of plant extracts from respective values of Papp.

2.7. Data analysis and curve fitting

All curve-fitting including linear regression for computing rates of solids transported across cell monolayers in an apical-basolateral direction, were calculated using SigmaPlot for Windows Version 12.5 (Systat Software Inc., Chicago, IL, USA). Statistical analysis was performed using Minitab 16 statistical software (Minitab Inc., State College,

Pennsylvania, USA).

3. Results

3.1. Effects of standards and plant extracts on cell viability

Prior to the cell monolayer transport experiments, effects of exposure to standards and plant extracts on cell viability at treatment concentrations up to 0.13 mg/mL and 1 mg/mL for standards and plant extracts, respectively. Cell viability was > 90% (Fig. 2).

Additionally, the integrity of the monolayers was evaluated by the TEER values prior to,

24 h and 48 h after the monolayer transport experiments. Prior to the monolayer transport,

TEER values were > 270 Ω cm2 and TEER variation 24 h and 48 h post transport was <

25% (Fig. 2).

3.2. Transport of standards across the cell monolayer in an apical to basolateral

direction and correlation to human absorption

Five standards with log P range from -3.45 to 6.75 and molecular mass from 170.12 to

669.96 were analysed for their transport across the cell monolayer in an apical-basolateral direction (Table 1). The rate of cumulative quantities of solids transported (in peak area units) across the cell monolayer in an apical-basolateral direction were calculated from

regression analysis of the slope (Fig. 3a). After correction for reagent control and standardisation of absorptivity (Table A1), the transport rates of standards were found to range from 0.0526 µg/h for ascorbic acid to 0.3218 µg/h for bromophenol blue, corresponding to values of apparent permeability Papp from 0.0018 cm/h for ascorbic acid to 0.0714 cm/h for bromophenol blue (Table 1). A positive correlation between Papp and log P of standards was observed and fitted to a quadratic relationship (Fig. 3b).

Using the PCAP model, values of the predicted time of plasma peak Tmax in healthy humans was calculated from substitution of values of log P and molecular mass of standards (Table 1). Accordingly, the in vivo Tmax predicted for uptake of standards from the natural digestive system of healthy humans ranged from 0.42 h for thiamine HCl to

5.95 h for bromophenol blue (Table 1). The in vivo Tmax predicted of standards were positively correlated to their cell monolayer Papp and fitted to a linear regression (Fig. 3c).

3.3. Transport of plant extracts across the cell monolayer in an apical-basolateral

direction and prediction of their human absorption

A selection of ten project plant extracts and seven reference plant extracts were analysed for their transport behaviours across the cell monolayer from an apical-basolateral direction (Table 2). The rate of cumulative quantities of solids transported (in peak area units) across the cell monolayers in an apical-basolateral direction were calculated from regression analysis of the slope (Fig. 4). After correction for reagent control and standardisation of absorptivity (Table A1), the transport rates of plant extracts were found to be 1.5–9.78 µg/h for the project extracts and were 0.23–5.68 µg/h for the reference extracts (Table 2). The apparent permeability Papp of the plant extracts were 0.0059–

0.0337 cm/h for the project extracts and were 0.0038–0.0235 cm/h for the reference extracts. After 4 h of the experiment, the % solids transported across the cell monolayers

in an apical-basolateral direction ranged from 3.64 to 31.52% for the project extracts and from 2.14 to 18.88% for the reference extracts (Table 2).

Correlations between parameters of the cell monolayer transport in an apical-basolateral direction of plant extracts were reported as Pearson’s correlation coefficient (r) with p- values (Table 3). Significant correlations were observed between % solids transported and Papp (r = 0.868, P < 0.001), % solids transported and transport rate (r = 0.847, P <

0.001), % solids transported and final cumulative mass in the basolateral chamber (r =

0.976, P < 0.001), Papp and transport rate (r = 0.983, P < 0.001), Papp and final cumulative mass in the basolateral chamber (r = 0.881, P < 0.001), transport rate and final cumulative mass in the basolateral chamber (r = 0.894, P < 0.001) (Table 3).

To estimate the predicted average in vivo Tmax of the project and reference plant extracts, the relationship between Papp and Tmax of standards (Fig. 3c) was used representing an averaged transport behaviour of the complex mixture of phytochemicals in the cell monolayer model. Calculated values of predicted average Tmax of the project extracts ranged from 0.68 h for kale to 2.9 h for white zucchini for the project plant extracts and from 0.51 h for olive leaf to 2.09 h for blueberry for the reference plant extracts (Table

2).

4. Discussion

4.1. Transport behaviour through cell monolayers in an apical-basolateral direction of

standards and plant extracts

In the current study, co-cultures of Caco-2 and HT29-MTX-E12 cells were used to allow incorporation of the two major cell types of the small intestinal epithelium (enterocytes and goblet cells, respectively) and therefore representing the human intestinal epithelium

more accurately (Hilgendorf et al., 2000). The monolayers of co-cultured Caco-2/HT29-

MTX-E12 cells have been widely used to investigate absorption of drugs (Hilgendorf et al., 2000; Roine et al., 2015), peptides (Stuknytė et al., 2015; Wikman-Larhed &

Artursson, 1995) and phytochemicals (Kuntz et al., 2015; Pacheco-Palencia, Talcott,

Safe, & Mertens-Talcott, 2008). Concentrations of standards and plant extracts used for apical loading were chosen so as to avoid any cell toxicity, with cell viabilities > 90%

(Fig. 2). The sampling intervals (30 min) and the total transport time (5 h for standards and 4 h for plant extracts) were chosen to accommodate both slow and rapid transport characteristics and thus to produce valid apparent permeability (Papp) values (Hubatsch et al., 2007).

The cell monolayer apparent permeability (Papp) was measured for five standards and ten plant extracts. The standards including ascorbic acid, gallic acid, thiamine HCl and curcumin are compounds commonly found in plants therefore they are good representatives the tested plant extracts. Considering that the linear fits R2 of cumulative solids (peak area units) of plant extracts in the basolateral chamber are all > 0.9, the plant extracts appeared to be transported across the monolayers from the apical chamber in a manner similar to that of the pure standards. This behaviour represented an approximation of the average transport rate as a ‘single’ species knowing that the plant extracts were chemically heterogeneous. Compounds with high Papp transported across the monolayer faster, indicated by increased transported rate and therefore increased % solids transported (Table 2). Considering that the % of solids transported is an indicator of bioavailability, this result is consistent with previous studies reporting a positive correlation of Papp and bioavailability in human (Artursson & Karlsson, 1991; Cheng et al., 2008) with bioavailability measured by % the area under the curve of the

pharmacokinetics of the drug by oral intake compared to by intravenous administration

(Borgstrom, Nyberg, Jonsson, Lindberg, & Paulson, 1989).

The relationship of the cell monolayer Papp versus lipophilicity log P of standards (Fig.

3b) was in agreement with the relationship previously reported for a wide range of drugs and phytochemicals (Artursson & Karlsson, 1991; Farrell et al., 2012). This relationship in Caco-2 monolayer absorption was modelled by Farrell et al. (2012) for 30 drugs, indicating that Papp was correlated with the lipophilicity descriptor log D (similar to log

P) (van de Waterbeemd & Gifford, 2003) and molecular mass (R2 = 0.93). According to the model of Farrell et al. (2012), Papp increased with increased log D and reduced molecular mass. Similarly, in this study, more lipophilic standards (high log P, e.g., curcumin and bromophenol blue) were associated with higher values of Papp and therefore were transported at faster rates than the hydrophilic standards (Table 1, Fig. 3b). The relationship between Papp and log P was expected as lipophilic compounds can easily diffuse through the lipid core of the membrane via the transcellular diffusion pathway

(Artursson et al., 2001).

4.2. Relationship of cell monolayer transport and human absorption

Tmax values of standards were calculated from their log P and molecular mass using the recently reported PCAP model (Selby-Pham, Miller, et al., 2017). Whilst the relationship between the cell monolayer permeability (Papp) and in vivo bioavailability as indicated by

% absorption has been well established, this is the first study to investigate the relationship between Papp and the predicted in vivo plasma Tmax. The Papp of standards were linearly correlated to their Tmax (Fig. 3c) with more hydrophilic compounds (low log P) transported at a slower rate in the cell monolayer model (lower Papp). However, whilst these more hydrophilic compounds were absorbed relatively slower in the cell

model in vitro, they were absorbed into the human plasma in vivo at a relatively faster rate (shorter Tmax). This result is consistent with previous studies which observed that, compared with more lipophilic compounds, the permeability of hydrophilic compounds was relatively slower in the Caco-2 monolayers than in the human intestine (Artursson,

1990; Hidalgo, Raub, & Borchardt, 1989; Lennernäs, Palm, Fagerholm, & Artursson,

1996).

A possible explanation for the difference between absorption in vitro and in vivo may be that hydrophilic compounds were likely absorbed through the water-filled junctions via the paracellular pathway. In comparison to human small intestine in vivo, the tighter junctions of the cell monolayers have smaller pore diameters which limit flux of hydrophilic compounds in vitro (Artursson et al., 2001). Alternatively, the difference between absorption in vitro and in vivo may be attributed to the greater number of openings in the tight junction of the human small intestine compared to the cell monolayers in vitro which promote flux of hydrophilic compounds in vivo (Artursson,

Ungell, & Löfroth, 1993; Tanaka et al., 1995).

2 The TEER values of the cell monolayers used in this study (~ 300 Ω cm , Fig. 2) were higher than that of the human small intestine (~40 Ω cm2, Sjöberg et al., 2013), indicating that the cell monolayers are tighter than the human small intestine. Therefore, the inverse relationship between the cell monolayer transport rates and the in vivo absorption kinetics

Tmax is attributed to the differences in routes of passive absorption as a result of differences in the tight junctions. The cell monolayers with tighter water-filled junctions facilitate transport of lipophilic compounds through the lipid core of the monolayers via transcellular diffusion route whilst the human intestine with larger water-filled junctions

facilitate transport of hydrophilic compounds through the water-filled junctions via the paracellular diffusion route.

Differences between absorption of phytochemicals in the cell model and in human studies have been previously reported. For example, in human subjects, Hollman, De Vries, Van

Leeuwen, Mengelers, and Katan (1995) reported 52% absorption of quercetin glucosides

(low log P, more hydrophilic) compared to 20% absorption of quercetin aglycone (higher

Log P, more lipophilic). However, in the Caco-2 cell model, quercetin glucosides were

- observed to pass through the epithelial monolayer at a much slower rate (Papp < 0.02 x 10

6 -6 cm/s) than quercetin aglycone (Papp 5.8 x 10 cm/s) (Murota, Shimizu, Chujo, Moon, &

Terao, 2000; Walgren, Walle, & Walle, 1998). These differences in absorption rates may possibly be explained by transport in humans being assisted by digestive secretions that compete with adsorption by the membrane, a mechanism not present in the cell model

(Ingels & Augustijns, 2003). Whilst the results presented herein were consistent with previous findings supporting the proposed interpretations, the methodology utilised did not include simulated gastric digestion of the standards and plant extracts prior to absorption in vitro. Accordingly, the presented results describe the absorption kinetics of phytochemicals only in their native forms. Previous studies have shown that 57% of anthocyanins were retained in their native forms during gastrointestinal digestion

(Bermúdez-Soto, Tomás-Barberán, & García-Conesa, 2007). These compounds were identified as stable during gastric digestion with the changes occurring almost exclusively during pancreatic digestion (McDougall, Fyffe, Dobson, & Stewart, 2005). Accordingly, future studies should investigate if the changes to forms of the subset of phytochemicals altered during gastrointestinal digestion impact overall phytochemical absorption kinetics

in vitro. Further, testing of a broader range of standards and phytochemical extract concentrations would be required to identify if these phenomena are dose dependant.

The Caco-2/HT29-MTX-E12 cell monolayers are a widely used model for human absorption at the epithelium with a well-established relationship between Papp in vitro and bioavailability in vivo (Artursson & Karlsson, 1991; Cheng et al., 2008). It is worth highlighting that whilst Papp is a reliable indicator of in vivo bioavailability (Artursson &

Karlsson, 1991; Cheng et al., 2008), the relationship between bioavailability and Tmax in vivo is less clear. For example, the pharmaceutical compounds nadolol and fluconazole have similar Tmax of 3 h in vivo (Misaka et al., 2014; Zimmermann, Yeates, Laufen, Pfaff,

& Wildfeuer, 1994), however their bioavailabilities have been identified as 34% and

100%, respectively (Cheng et al., 2008). Therefore, compounds that are absorbed at the same rate (same Tmax) will not necessarily have the same bioavailabilities in vivo. The

Caco-2/HT29-MTX-E12 cell model and PCAP statistical model may be used in conjunction to provide complimentary predictions regarding phytochemical absorption kinetics wherein Papp (in vitro) and Tmax (in silico) may be used to infer the bioavailability and time of maximal plasma concentration in vivo, respectively.

5. Conclusion

In this study, trans-epithelial absorption of a selection of phytochemical standards and extracts across the Caco-2/HT29-MTX-E12 co-cultured monolayers were determined.

We have compared these results with the kinetics from the phytochemical absorption prediction (PCAP) model from human clinical data which predicts time of peak absorption (Tmax) of phytochemicals based on their physicochemical properties. The apparent permeability (Papp) determined here as an indicator of trans-epithelial absorption observed for the pure compounds were consistent with those described in previous cell

culture studies. Whilst Papp is a good indicator of in vivo bioavailability (% absorption), this measure of in vitro absorption rate is not necessarily an accurate predictor of absorption kinetics (the rate of absorption) in vivo. We propose that in vitro cell culture analyses be complimented with the PCAP model to ensure maximal accuracy when predicting human absorption kinetics of phytochemicals.

Acknowledgements

This project has been funded by Horticulture Innovation Australia Limited using the

Vegetable levy and funds from the Australian Government. Donation of the grape skin and grape seed extracts from Tarac Technologies and technical assistance by Hema

Jegasothy, Wei Chen and Rama Addepalli are gratefully acknowledged.

Conflict of interest

The authors declare no conflict of interests.

References

Artursson, P. (1990). Epithelial transport of drugs in cell culture. I: A model for studying the passive diffusion of drugs over intestinal absorbtive (Caco-2) cells. Journal of Pharmaceutical Sciences, 79(6), 476-482. doi: http://dx.doi.org/10.1002/jps.2600790604 Artursson, P., & Karlsson, J. (1991). Correlation between oral drug absorption in humans and apparent drug permeability coefficients in human intestinal epithelial (Caco- 2) cells. Biochemical and biophysical research communications, 175(3), 880-885. Artursson, P., Palm, K., & Luthman, K. (2001). Caco-2 monolayers in experimental and theoretical predictions of drug transport. Advanced drug delivery reviews 46(1-3), 27-43. Artursson, P., Ungell, A. L., & Löfroth, J. E. (1993). Selective paracellular permeability in two models of intestinal absorption: cultured monolayers of human intestinal epithelial cells and rat intestinal segments. [journal article]. Pharmaceutical Research, 10(8), 1123-1129. doi: 10.1023/a:1018903931777 Bennett, L. E., & Muench, G. (2011). Process for plant bioactive enrichment. PCT/AU2010/0015102010. Bermúdez-Soto, M. J., Tomás-Barberán, F. A., & García-Conesa, M. T. (2007). Stability of polyphenols in chokeberry (Aronia melanocarpa) subjected to in vitro gastric and pancreatic digestion. Food Chemistry, 102(3), 865-874. doi: http://dx.doi.org/10.1016/j.foodchem.2006.06.025 Borgstrom, L., Nyberg, L., Jonsson, S., Lindberg, C., & Paulson, J. (1989). Pharmacokinetic evaluation in man of terbutaline given as separate enantiomers and as the racemate. British Journal of Clinical Pharmacology, 27(1), 49-56. Boyer, J., Brown, D., & Liu, R. H. (2004). Uptake of quercetin and quercetin 3-glucoside from whole onion and apple peel extracts by Caco-2 cell monolayers. Journal of Agricultural and Food Chemistry, 52(23), 7172-7179. doi: 10.1021/jf030733d Burton-Freeman, B. (2010). Postprandial metabolic events and fruit-derived phenolics: a review of the science. British Journal of Nutrition, 104 Suppl 3, S1-S14. doi: 10.1017/s0007114510003909 Calder, P. C., Albers, R., Antoine, J. M., Blum, S., Bourdet-Sicard, R., Ferns, G. A., . . . Zhao, J. (2009). Inflammatory disease processes and interactions with nutrition. British Journal of Nutrition, 101 Suppl 1, S1-S45. doi: 10.1017/s0007114509377867 Cheng, K. C., Li, C., & Uss, A. S. (2008). Prediction of oral drug absorption in humans – from cultured cell lines and experimental animals. Expert Opinion on Drug Metabolism & Toxicology, 4(5), 581-590. doi: 10.1517/17425255.4.5.581 D'Onofrio, G., Sancarlo, D., Ruan, Q., Yu, Z., Panza, F., Daniele, A., . . . Seripa, D. (2017). Phytochemicals in the treatment of Alzheimer's disease: a systematic review. Current Drug Targets, 18(13), 1487-1498. Dauchet, L., Amouyel, P., & Dallongeville, J. (2009). Fruits, vegetables and coronary heart disease. Nature Reviews. Cardiology, 6(9), 599-608. doi: 10.1038/nrcardio.2009.131 Day, A. J., Gee, J. M., DuPont, M. S., Johnson, I. T., & Williamson, G. (2003). Absorption of quercetin-3-glucoside and quercetin-4'-glucoside in the rat small

intestine: the role of lactase phlorizin hydrolase and the sodium-dependent glucose transporter. Biochemical Pharmacology, 65(7), 1199-1206. Del Rio, D., Rodriguez-Mateos, A., Spencer, J. P., Tognolini, M., Borges, G., & Crozier, A. (2013). Dietary (poly)phenolics in human health: structures, bioavailability, and evidence of protective effects against chronic diseases. Antioxidants and Redox Signaling, 18(14), 1818-1892. doi: 10.1089/ars.2012.4581 Donovan, J. L., Manach, C., Faulks, R. M., & Kroon, P. A. (2006). Absorption and metabolism of dietary plant secondary metabolites. In A. Crozier, M. N. Clifford & H. Ashihara (Eds.), Plant secondary metabolites: Occurrence, structure and role in the human diet (pp. 303-351). Oxford, UK: Blackwell Publishing. Farrell, T. L., Poquet, L., Dew, T. P., Barber, S., & Williamson, G. (2012). Predicting phenolic acid absorption in Caco-2 cells: a theoretical permeability model and mechanistic study. Drug Metabolism and Disposition, 40(2), 397-406. doi: 10.1124/dmd.111.041665 Gustin, D. M., Rodvold, K. A., Sosman, J. A., Diwadkar-Navsariwala, V., Stacewicz- Sapuntzakis, M., Viana, M., . . . Bowen, P. E. (2004). Single-dose pharmacokinetic study of lycopene delivered in a well-defined food-based lycopene delivery system (tomato paste-oil mixture) in healthy adult male subjects. Cancer Epidemiology, Biomarkers and Prevention, 13(5), 850-860. Hidalgo, I. J., Raub, T. J., & Borchardt, R. T. (1989). Characterization of the human colon carcinoma cell line (Caco-2) as a model system for intestinal epithelial permeability. Gastroenterology, 96(3), 736-749. doi: http://dx.doi.org/10.1016/0016-5085(89)90897-4 Hilgendorf, C., Spahn-Langguth, H., Regardh, C. G., Lipka, E., Amidon, G. L., & Langguth, P. (2000). Caco-2 versus Caco-2/HT29-MTX co-cultured cell lines: permeabilities via diffusion, inside- and outside-directed carrier-mediated transport. Journal of Pharmaceutical Sciences, 89(1), 63-75. doi: 10.1002/(sici)1520-6017(200001)89:1<63::aid-jps7>3.0.co;2-6 Hollman, P. C. H., De Vries, J. H. M., Van Leeuwen, S. D., Mengelers, M. J. B., & Katan, M. B. (1995). Absorption of dietary quercetin glycosides and quercetin in healthy ileostomy volunteers. American Journal of Clinical Nutrition, 62(6), 1276-1282. Holst, B., & Williamson, G. (2008). Nutrients and phytochemicals: from bioavailability to bioefficacy beyond antioxidants. Current Opinion in Biotechnology, 19(2), 73- 82. doi: http://dx.doi.org/10.1016/j.copbio.2008.03.003 Huang, Y., Park, E., Edirisinghe, I., & Burton-Freeman, B. M. (2016). Maximizing the health effects of strawberry anthocyanins: understanding the influence of the consumption timing variable. Food and Function, 7(12), 4745-4752. doi: 10.1039/c6fo00995f Hubatsch, I., Ragnarsson, E. G., & Artursson, P. (2007). Determination of drug permeability and prediction of drug absorption in Caco-2 monolayers. Nature Protocols, 2(9), 2111-2119. doi: 10.1038/nprot.2007.303 Ingels, F. M., & Augustijns, P. F. (2003). Biological, pharmaceutical, and analytical considerations with respect to the transport media used in the absorption screening system, Caco-2. Journal of Pharmaceutical Sciences, 92(8), 1545-1558. doi: 10.1002/jps.10408 Jarzab, A., & Kukula-Koch, W. (2017). Recent advances in obesity: the role of turmeric tuber and its metabolites in the prophylaxis and therapeutical strategies. Current

Medicinal Chemistry, 24, 1-17. doi: https://doi.org/10.2174/0929867324666161118095443 Key, T. J. (2011). Fruit and vegetables and cancer risk. British Journal of Cancer, 104(1), 6-11. Kottra, G., & Daniel, H. (2007). Flavonoid glycosides are not transported by the human Na+/glucose transporter when expressed in Xenopus laevis oocytes, but effectively inhibit electrogenic glucose uptake. Journal of Pharmacology and Experimental Therapeutics, 322(2), 829-835. doi: 10.1124/jpet.107.124040 Kuntz, S., Asseburg, H., Dold, S., Rompp, A., Frohling, B., Kunz, C., & Rudloff, S. (2015). Inhibition of low-grade inflammation by anthocyanins from grape extract in an in vitro epithelial-endothelial co-culture model. Food and Function, 6(4), 1136-1149. doi: 10.1039/c4fo00755g Lee, C. Y. (2013). Challenges in providing credible scientific evidence of health benefits of dietary polyphenols. Journal of Functional Foods, 5(1), 524-526. doi: http://dx.doi.org/10.1016/j.jff.2012.10.018 Lennernäs, H., Palm, K., Fagerholm, U., & Artursson, P. (1996). Comparison between active and passive drug transport in human intestinal epithelial (Caco-2) cells in vitro and human jejunum in vivo. International Journal of Pharmaceutics, 127(1), 103-107. doi: http://dx.doi.org/10.1016/0378-5173(95)04204-0 Lipinski, C. A., Lombardo, F., Dominy, B. W., & Feeney, P. J. (1997). Experimental and computational approaches to estimate solubility and permeability in drug discovery and development settings. Advanced Drug Delivery Reviews, 23(1–3), 3-25. doi: http://dx.doi.org/10.1016/S0169-409X(96)00423-1 Liu, C. S., Glahn, R. P., & Liu, R. H. (2004). Assessment of carotenoid bioavailability of whole foods using a Caco-2 cell culture model coupled with an in vitro digestion. Journal of Agricultural and Food Chemistry, 52(13), 4330-4337. doi: 10.1021/jf040028k McDougall, G. J., Fyffe, S., Dobson, P., & Stewart, D. (2005). Anthocyanins from red wine – their stability under simulated gastrointestinal digestion. Phytochemistry, 66(21), 2540-2548. doi: http://dx.doi.org/10.1016/j.phytochem.2005.09.003 Misaka, S., Yatabe, J., Muller, F., Takano, K., Kawabe, K., Glaeser, H., . . . Kimura, J. (2014). Green tea ingestion greatly reduces plasma concentrations of nadolol in healthy subjects. [Articles]. Clinical Pharmacology and Therapeutics, 95(4), 432- 438. doi: 10.1038/clpt.2013.241 Murota, K., Shimizu, S., Chujo, H., Moon, J. H., & Terao, J. (2000). Efficiency of absorption and metabolic conversion of quercetin and its glucosides in human intestinal cell line Caco-2. Archives of Biochemistry and Biophysics, 384(2), 391- 397. doi: 10.1006/abbi.2000.2123 Pacheco-Palencia, L. A., Talcott, S. T., Safe, S., & Mertens-Talcott, S. (2008). Absorption and biological activity of phytochemical-rich extracts from acai (Euterpe oleracea Mart.) pulp and oil in vitro. Journal of Agricultural and Food Chemistry, 56(10), 3593-3600. doi: 10.1021/jf8001608 Roine, J., Kaasalainen, M., Peurla, M., Correia, A., Araujo, F., Santos, H. A., . . . Salonen, J. (2015). Controlled dissolution of griseofulvin solid dispersions from electrosprayed enteric polymer micromatrix particles: physicochemical characterization and in vitro evaluation. Molecular Pharmaceutics, 12(7), 2254- 2264. doi: 10.1021/mp500787b

Sandhu, A. K., Huang, Y., Xiao, D., Park, E., Edirisinghe, I., & Burton-Freeman, B. (2016). Pharmacokinetic characterization and bioavailability of strawberry anthocyanins relative to meal intake. Journal of Agricultural and Food Chemistry, 64(24), 4891-4899. doi: 10.1021/acs.jafc.6b00805 Schinella, G. R., Tournier, H. A., Prieto, J. M., de Buschiazzo, P. M., & Rı́os, J. L. (2002). Antioxidant activity of anti-inflammatory plant extracts. Life Sciences, 70(9), 1023-1033. doi: http://dx.doi.org/10.1016/S0024-3205(01)01482-5 Selby-Pham, S. N. B., Howell, K., Dunshea, F., Ludbey, J., Lutz, A., & Bennett, L. E. (2017). Statistical modelling coupled with LC-MS analysis to predict human upper intestinal absorption of phytochemical mixtures. Food Chemistry, manuscript submitted. Selby-Pham, S. N. B., Miller, R. B., Howell, K., Dunshea, F., & Bennett, L. E. (2017). Physicochemical properties of dietary phytochemicals can predict their passive absorption in the human small intestine. Scientific Reports, 7(1), 1931. Sjöberg, Å., Lutz, M., Tannergren, C., Wingolf, C., Borde, A., & Ungell, A. L. (2013). Comprehensive study on regional human intestinal permeability and prediction of fraction absorbed of drugs using the Ussing chamber technique. European Journal of Pharmaceutical Sciences, 48, 166-180. Stalmach, A., Troufflard, S., Serafini, M., & Crozier, A. (2009). Absorption, metabolism and excretion of Choladi green tea flavan-3-ols by humans. [Article]. Molecular Nutrition and Food Research, 53, S44-S53. doi: 10.1002/mnfr.200800169 Stravodimos, G. A., Chetter, B. A., Kyriakis, E., Kantsadi, A. L., Chatzileontiadou, D. S., Skamnaki, V. T., . . . Leonidas, D. D. (2017). Phytogenic polyphenols as glycogen phosphorylase inhibitors: the potential of triterpenes and flavonoids for glycaemic control in type 2 diabetes. Current Medicinal Chemistry, 24(4), 384-403. doi: 10.2174/0929867324666161118122534. Stuknytė, M., Maggioni, M., Cattaneo, S., De Luca, P., Fiorilli, A., Ferraretto, A., & De Noni, I. (2015). Release of wheat gluten exorphins A5 and C5 during in vitro gastrointestinal digestion of bread and pasta and their absorption through an in vitro model of intestinal epithelium. Food Research International, 72, 208-214. doi: http://dx.doi.org/10.1016/j.foodres.2015.04.002 Swinehart, D. F. (1962). The Beer-Lambert Law. Journal of Chemical Education, 39(7), 333. doi: 10.1021/ed039p333 Tanaka, Y., Taki, Y., Sakane, T., Nadai, T., Sezaki, H., & Yamashita, S. (1995). Characterization of drug transport through tight-junctional pathway in Caco-2 monolayer: Comparison with isolated rat jejunum and colon. [journal article]. Pharmaceutical Research, 12(4), 523-528. doi: 10.1023/a:1016245711557 van de Waterbeemd, H., & Gifford, E. (2003). ADMET in silico modelling: towards prediction paradise? [10.1038/nrd1032]. Nature Reviews Drug Discovery, 2(3), 192-204. van der Merwe, M., & Bloomer, R. J. (2016). The influence of methylsulfonylmethane on inflammation-associated cytokine release before and following strenuous exercise. Journal of Sports Medicine (Hindawi Publishing Corporation), 2016, 7498359. doi: 10.1155/2016/7498359 Walgren, R. A., Walle, U. K., & Walle, T. (1998). Transport of quercetin and its glucosides across human intestinal epithelial Caco-2 cells. Biochemical Pharmacology, 55(10), 1721-1727.

Walter, E., Janich, S., Roessler, B. J., Hilfinger, J. M., & Amidon, G. L. (1996). HT29- MTX/Caco-2 cocultures as an in vitro model for the intestinal epithelium: In vitro–in vivo correlation with permeability data from rats and humans. Journal of Pharmaceutical Sciences, 85(10), 1070-1076. doi: 10.1021/js960110x Wikman-Larhed, A., & Artursson, P. (1995). Co-cultures of human intestinal goblet (HT29-H) and absorptive (Caco-2) cells for studies of drug and peptide absorption. European Journal of Pharmaceutical Sciences, 3(3), 171-183. doi: http://dx.doi.org/10.1016/0928-0987(95)00007-Z Zimmermann, T., Yeates, R. A., Laufen, H., Pfaff, G., & Wildfeuer, A. (1994). Influence of concomitant food intake on the oral absorption of two triazole antifungal agents, itraconazole and fluconazole. European Journal of Clinical Pharmacology, 46(2), 147-150. doi: 10.1007/BF00199879

Table 1. Transport of standards across the Caco-2/HT29-MTX-E12 monolayer in an apical-basolateral direction Predicted Apparent Molecular Lipophilicity Transport Linear Standard human permeability massa log Pa ratec (µg/h) fitd R2 b Tmax (h) Papp (cm/h)

Ascorbic acid 176.12 -1.402 0.44 0.0018 0.0526 0.96

Gallic acid 170.12 0.59 0.59 0.0022 0.0615 0.92

Thiamine 337.27 -3.449 0.42 0.0043 0.1869 0.89 HCl

Curcumin 368.38 2.303 1.19 0.0148 0.2222 0.95

Bromophenol 669.96 6.745 5.95 0.0714 0.3218 0.94 blue aCalculated using the online software http://www.molinspiration.com/cgi-bin/properties bCalculated using the Phytochemical Absorption Prediction model (Selby-Pham, Miller, et al., 2017) cSlope of linear fits of cumulative solids in peak area units transported across the monolayer over time (Fig. 2a), corrected for blanks and for ultraviolet absorptivity of each standard dLinear fits R2 of cumulative solids in peak area units transported across the monolayer over time (Fig. 2a)

Table 2. Transport of plant extracts across the Caco-2/HT29-MTX-E12 monolayer

Apparent Predicted Initial apical Final cumulative Solids permeability average Transport rated concentratione mass in the Linear fitf Sample Transporteda b 2 Papp Tmax (µg/h) (µg/mL) basolateral R (%) (cm/h) (h) chamber (µg) Project extracts

Broccoli 5.43 ± 0.22 0.0153 1.43 4.21 834.58 ± 11.57 9.06 ± 0.61 0.995 Carrot 8.49 ± 0.23 0.0070 0.76 2.15 937.45 ± 9.52 15.92 ± 0.67 0.995 Red cabbage 11.56 ± 1.31 0.0143 1.35 4.49 948.73 ± 4.29 21.93 ± 2.04 0.988 Red sweet potato 12.66 ± 1.36 0.0129 1.24 4.04 949.58 ± 0.42 24.04 ± 3.78 0.956 Rhubarb 12.23 ± 0.48 0.0144 1.36 3.94 826.86 ± 5.53 20.23 ± 1.11 0.996 Squash 18.96 ± 0.37 0.0161 1.50 4.77 894.28 ± 26.28 33.91 ± 0.43 0.998 White zucchini 31.52 ± 1.13 0.0337 2.90 9.78 879.12 ± 9.12 55.42 ± 4.46 0.996 Eggplant 7.18 ± 0.29 0.0131 1.25 3.56 823.40 ± 8.18 11.82 ± 0.70 0.984 Kale 3.64 ± 0.14 0.0059 0.68 1.50 775.51 ± 20.41 5.65 ± 0.19 0.998 Vietnamese coriander 7.58 ± 0.12 0.0135 1.29 3.59 806.57 ± 14.57 12.23 ± 0.32 0.990 Reference extracts

Blueberry 18.88 ± 0.75 0.0235 2.09 5.68 731.96 ± 10.31 27.64 ± 1.99 0.996 Cacao 6.03 ± 0.82 0.0102 1.03 2.52 745.28 ± 0.18 8.99 ± 0.28 0.983 Grape seed 5.99 ± 0.80 0.0043 0.55 0.71 500.00 ± 0.23 5.99 ± 0.59 0.946

Grape skin 2.14 ± 0.33 0.0103 1.03 3.40 1000 ± 0.67 4.27 ± 0.69 0.981 Green tea 6.01 ± 1.29 0.0063 0.71 1.67 800.03 ± 1.86 9.62 ± 0.25 0.980 Olive leaf 7.27 ± 0.89 0.0038 0.51 0.23 188.03 ± 8.55 2.73 ± 0.17 0.985 Tomato 6.15 ± 0.52 0.0057 0.67 1.29 683.64 ± 23.26 8.41 ± 0.40 0.986 aThe percentage of the final cumulative mass in the basolateral chamber compared to the initial mass in the apical chamber b Predicted average Tmax was calculated from the fitted relationship between Papp and Tmax of standards calculated from the PCAP model dSlope of linear fits of cumulative solids in peak area units transported across the monolayer over time, corrected for blanks and for ultraviolet absorptivity of each plant extract eInitial concentration of soluble solids in the apical chamber fLinear fits R2 of cumulative solids in peak area units absorbed across the monolayer in an apical-basolateral direction over time (Fig. 4a and b)

Table 3. Pearson correlations between the Caco-2/HT-27 monolayer transport parameters of plant extracts

Parameters % Solids Apparent Transport Initial apical Transported permeability rate (µg/h) concentration

Papp (cm/h) (µg/mL)

Apparent r = 0.868 permeability P < 0.001 Papp (cm/h)

Transport rate r = 0.847 r = 0.983 (µg/h) P < 0.001 P < 0.001

Initial apical r = 0.192 r = 0.401 r = 0.528 concentration P = 0.461 P = 0.111 P = 0.029 (µg/mL)

Final cumulative r = 0.976 r = 0.881 r = 0.894 r = 0.390 mass in the P < 0.001 P < 0.001 P < 0.001 P = 0.122 basolateral chamber (µg)

Data reported as Pearson’s correlation coefficient (r) with p-values. Significantly high correlations (│r│ > 0.75, P < 0.05) are highlighted with bold.

100

Figure captions

Figure 1. Schematic diagram of the monolayer transport experiment and data processing method. (a) The monolayer transport experiment of plant extracts. (b) The data processing method to transform HPLC output to apparent permeability Papp of the plant extracts though the monolayer.

Figure 2. Effects of plant extracts on viability of Caco-2/HT29-MTX-E12 cells.MTS assay and TEER measurement were used to determine cell viability of (a) standards at concentrations of 0.09 mg/mL ascorbic acid, 0.09 mg/mL gallic acid, 0.13 mg/mL thiamine HCl, 0.07 mg/mL curcumin and 0.07 mg/mL bromophenol blue, (b) project extracts at concentration of 1 mg/mL and (c) reference extracts at concentration of 1 mg/mL and 0.5 mg/mL grape seed.

Figure 3. Calibration of the apparent permeability (Papp) of standards transported across the Caco-2/HT29-MTX-E12 monolayer in an apical-basolateral direction and the predicted time required for standards to reach maximal plasma concentration

(Tmax) calculated using the Phytochemical Absorption Prediction (PCAP) model. (a)

Transport rates represented by cumulative solids (peak area units) transported across the

Caco-2/HT29-MTX-E12 monolayer, showing linear fits for standards and control used to

2 compute Papp. (b) Relationship of Papp versus lipophilicity (log P): Papp = 0.0011 (log P)

2 + 0.0029 (log P) + 0.0018 with R of 0.998 and (c) relationship of Papp versus Tmax in human by the PCAP model reported previously (Selby-Pham, Miller, et al., 2017). Papp–

2 Tmax was fitted using a linear regression Tmax = 79.8349Papp + 0.20836 with R of 0.995.

101

Figure 4. Transport of plant extracts represented by cumulative solids (in peak area units) transported across the Caco-2/HT29-MTX-E12 monolayer in an apical- basolateral direction over time. Linear fits for (a) project extracts and (b) reference extracts. Coefficients and R2 of fits are given in Table 2.

102

Figure 1. a

Total peak area obtained • Absorption at 254 nm from HPLC

Cumulative absorbed • Peak area units solids vs time

Slope of linear fits, corrected for • Peak area units/time dPA/dt blanks

Corrected for • Mass units/time dQ/dt (µg/h) absorptivity

Calculation of apparent • P = dQ/dt / (C A) permeability app o

103

Figure 2.

a TEER before_mean TEER 24h_mean TEER 48h-mean Cell viability 90% cell viability

600 120

500 100

)

2 400 80

.cm 300 60

200 40

TEER (

Cell viability (%) Cell viability

100 20

0 0

Gallic acid Curcumin Ascorbic acid Thiamine HCl Bromophenol blue

b TEER before transport TEER 24 h post transport TEER 48 h post transport Cell viability 90% cell viability 400 100

80 300

)

2 60

.cm 200 40

TEER ( 100 (%) Cell viability 20

0 0

Kale Carrot Squash Broccoli Rhubarb Eggplant Red cabbage White zucchini Red sweet potato Vietnamese corriander

c TEER before transport TEER 24 h post transport TEER 48 h post transport Cell viability 90% cell viability 500

100 400

) 80

2 300 .cm 60 200 40

TEER (

Cell viability (%) Cell viability 100 20

0 0

Cacao Olive leaf Tomato Blueberry Grape seedGrape skinGreen tea

104

Figure 3.

a Ascorbic acid 5e+7 Gallic acid Thiamine HCl Curcumin 4e+7 Bromophenol blue HBSS

3e+7

(peak area units) area (peak 2e+7

Cumulative Transported Solids Transported Cumulative 1e+7 0 1 2 3 4 5 Time (h)

b 0.08 BPB

0.06

0.04

(cm/h) Cur

app 0.02 P Thia AA GA 0.00

-0.02 -4 -2 0 2 4 6 8 LogP

c 7 BPB 6

2 Cur

1 GA

predicted from PCAP model model (h) from PCAP predicted AA Thia

max

T 0 0.00 0.02 0.04 0.06 0.08

Papp (cm/h)

105

Figure 4.

a Broccoli Carrot Red cabbage Red sweet potato Rhubarb Squash White zucchini 8e+6 Eggplant Kale Vietnamese coriander 6e+6

4e+6

(peak area units) (peak 2e+6

Cumulative Transported Solids Solids Transported Cumulative 0 0 1 2 3 4 5 Time (h)

b Blueberry Cacao 1.6e+7 Grape seed Grape skin 1.4e+7 Green tea 1.2e+7 Olive leaf Tomato 1.0e+7 8.0e+6 6.0e+6 4.0e+6

(peak area units) (peak 2.0e+6 0.0

Cumulative Transported Solids Solids Transported Cumulative

0 1 2 3 4 5 Time (h)

106

Chapter 5: Dietary phytochemicals promote health by antioxidant defence in a pig model

5.1 Introduction

In this chapter, validation of the in silico statistical PCAP model was undertaken using an oral intervention study in pigs, which was chosen as an animal model with physiological and anatomical similarities to the digestive conditions of humans.

In this chapter, the hypothesis was that the absorption kinetics of phytochemicals in pigs in vivo can be informed by prediction of Tmax profiles using the in silico statistical model.

The aim of this chapter was to characterise the in vivo absorption kinetics of selected phytochemical extracts following oral intake using the pig model, and to compare with the in silico predictions of Tmax profiles obtained from the PCAP model described in

Chapter 3. Further, this chapter aimed to investigate the mechanism of actions by which phytochemicals regulate oxidative stress in pigs. The aims were fulfilled via completion of the following objectives:

 Determination of the absorption kinetics of selected phytochemical extracts in

pigs via measurement of plasma antioxidant status including total antioxidant

capacity and activity of the antioxidant enzyme glutathione peroxidase, as

biomarkers of phytochemical absorption.

 Investigation of the modes of action of phytochemicals for regulating oxidative

stress using both in vitro and ex vivo experimental measures.

107

Chapter 5: Dietary phytochemicals promote health by antioxidant defence in a pig model

5.2 Accepted manuscript

108

Article Dietary Phytochemicals Promote Health by Enhancing Antioxidant Defence in a Pig Model

Sophie N. B. Selby-Pham 1,2, Jeremy J. Cottrell 1, Frank R. Dunshea 1, Ken Ng 1, Louise E. Bennett 2 and Kate S. Howell 1,*

1 Faculty of Veterinary and Agricultural, The University of Melbourne, Parkville, VIC 3010, Australia; [email protected] (S.N.B.S.-P.); [email protected] (J.J.C.); [email protected] (F.R.D.); [email protected] (K.N.) 2 CSIRO Agriculture and Food, 671 Sneydes Road, Werribee, VIC 3010, Australia; [email protected] * Correspondence: [email protected]; Tel.: +61-3-9035-3119

Received: 8 June 2017; Accepted: 12 July 2017; Published: date

Abstract: Phytochemical-rich diets are protective against chronic diseases and mediate their protective effect by regulation of oxidative stress (OS). However, it is proposed that under some circumstances, phytochemicals can promote production of reactive oxygen species (ROS) in vitro, which might drive OS-mediated signalling. Here, we investigated the effects of administering single doses of extracts of red cabbage and grape skin to pigs. Blood samples taken at baseline and 30 min intervals for 4 hours following intake were analyzed by measures of antioxidant status in plasma, including Trolox equivalent antioxidant capacity (TEAC) and glutathione peroxidase (GPx) activity. In addition, dose-dependent production of hydrogen peroxide (H2O2) by the same extracts was measured in untreated commercial pig plasma in vitro. Plasma from treated pigs showed extract dose-dependent increases in non-enzymatic (plasma TEAC) and enzymatic (GPx) antioxidant capacities. Similarly, extract dose-dependent increases in H2O2 were observed in commercial pig plasma in vitro. The antioxidant responses to extracts by treated pigs were highly correlated with their respective yields of H2O2 production in vitro. These results support that dietary phytochemicals regulate OS via direct and indirect antioxidant mechanisms. The latter may be attributed to the ability to produce H2O2 and to thereby stimulate cellular antioxidant defence systems.

Keywords: hydrogen peroxide; reactive oxygen species; plant extracts; red cabbage; grape; glutathione peroxide; total antioxidant capacity; porcine; piglet; Landrace

1. Introduction

A phytochemical-rich diet is strongly associated with reducing the risk of chronic diseases including cancer [1], cardiovascular [2], and neurodegenerative diseases [3]. The health benefits of dietary phytochemicals have been attributed to their ability to mitigate oxidative stress and inflammation (OSI), which is associated with normal metabolism [4,5] but is also involved in the onset of chronic diseases [6]. Production of reactive oxygen species (ROS) occurs under normal conditions in cells, the main source from by-products of the electron transport chains [7]. Uncontrolled ROS can lead to OSI and unregulated OSI can result in molecular and cellular damage which in turn leads to an increased risk of chronic diseases [8]. However, OSI is an important defence mechanism of the body against infections and injuries [9]. Therefore, transient

109

peaks or optimal steady state levels of ROS in the body are likely involved in maintaining good health and reducing the risk of disease [10]. It was believed that dietary phytochemicals exert protection via direct scavenging of ROS, as observed in many in vitro studies [11–13]. However, this concept has been challenged as the concentrations of phytochemicals in human plasma in vivo after consumption of phytochemicals are much lower (in the nM to low µM range) compared to concentrations used in the in vitro studies (in the low µM to mM range) [14,15]. There are clearly discrepancies between the studies of these mechanisms in the whole organism. The health benefits of dietary phytochemicals are thought to be attributed to their ability to generate electrophilic or chemical stress signals, which trigger the cellular defence system to protect against molecular damage and subsequent chronic diseases [16–19]. The cellular antioxidant defence is made up of a non-enzymatic, inducible enzymatic defence and the DNA repair systems [20]. The non-enzymatic defence includes antioxidant molecules such as vitamin C, vitamin E, uric acid, glutathione, and thioredoxin that directly scavenge ROS and metal- chelating proteins such as transferrin, coeruloplasmin, and metallothionein that prevent ROS formation via controlling the level of pro-oxidative free metal ions [20]. The enzymatic antioxidant defence includes several pathways that remove ROS through enzymatic reactions. For example, superoxide dismutase converts superoxide anions into hydrogen peroxide (H2O2), which is subsequently transformed by catalase into oxygen and water or by glutathione peroxidase (GPx) into water [20]. The reduction of H2O2 by GPx consumes the reduced form of glutathione and generates the oxidised form, which can be recycled by glutathione reductase to restore the glutathione pool [20]. Dietary phytochemicals have been associated with increasing the levels of both non- enzymatic and enzymatic antioxidant defence in animal [21–26] and human dietary intervention studies [27–30]. Consumption of phytochemical-rich diets increased the expression of genes associated with DNA repair, immune, and inflammatory responses in humans [10,31–33]. The varied roles that dietary phytochemicals may play in the whole organism are complex, perhaps overlapping and have not been fully elucidated. The ability of dietary phytochemicals to generate stress signals can be related to their ability to produce ROS, in particular H2O2 [34]. Phytochemicals have been reported to produce H2O2 in cell culture media, which was potentially responsible for their cytotoxic effects in cell culture studies [35–38]. However, no research has been done on the ability of phytochemicals to produce H2O2 in plasma. Understanding this pro- oxidant action will provide information about how the phytochemicals can stimulate ROS- induced cellular antioxidant defence to provide protective effects against OSI. Absorption of phytochemicals into circulation and uptake by target cells are essential for phytochemicals to exert biological effects [39]. As phytochemicals are recognised by the human body as xenobiotics, their presence in the human body is transient [40] and influenced by their physicochemical properties. Recently, we have developed the phytochemical absorption prediction (PCAP) model, allowing direct calculation of the time required for phytochemicals to reach their maximal plasma concentrations (Tmax) after oral consumption, based on their molecular mass and lipophilicity descriptor log P [41]. Further, a liquid chromatography mass spectrometry (LC-MS) method has been developed to characterise Tmax ranges of phytochemical mixtures based on molecular mass and log P [42]. Here, we extend this modelling to dietary intervention in pigs, an animal model with physiological and anatomical similarities to the digestive tract of humans [43]. Phytochemicals across a broad range of chemical classes have been shown to impart positive health benefits [3,40]. Grape products and Brassica vegetables are among the most widely studied for their antioxidant capacity and protection against chronic diseases [44,45]. Grape skin contains predominately polyphenols including anthocyanidins, phenolic acids, and stilbenes [44], whilst

110

red cabbage (a member of the Brassica vegetables) contains a wider variety of phytochemicals including polyphenols (anthocyanidins, phenolic acids), glucosinolates, and vitamins [46]. The aim of this study was to use a pig model to establish the absorption kinetics of phytochemical extracts from red cabbage and grape skin and to examine their effects on two measures of antioxidant status (plasma total antioxidant capacity and plasma GPx activity). Direct induction of the pro-oxidant effects of the plant extracts in pig plasma was measured by H2O2 production in pig plasma when exposed to the plant extracts in vitro. This study provides both in vitro and ex vivo evidence to support that one of the likely modes of action by phytochemicals is to induce H2O2 in plasma and to thereby initiate protective action by enzymatic and non-enzymatic cellular defences.

2. Materials and Methods

2.1. Materials

All chemicals including gallic acid, Folin-Ciocalteu reagent, sodium carbonate (Na2CO3), hydrogen peroxide (H2O2), sulfuric acid (H2SO4), xylenol orange, Iron (II) sulphate (FeSO4), butylated hydroxytoluene (BHT), tris(hydroxymethyl)aminomethane (Tris), glycine, citrate, urea, hydrochloric acid (HCl), Trolox, bathocuproinedisulfonic acid sodium salt (BCS), copper (II) chloride (CuCl2), methanol, formic acid, acetonitrile, L-histidine, (S)-dihydroorotate, shikimate, 4-pyridoxate, 3-hydroxybenzyl alcohol, 2,5-dihydroxybenzoate, 3- hydroxybenzaldehyde, trans-cinnamate, estradiol-17α, deoxycholate, retinoate, oleic acid, and heptadecanoate were of analytical grade and from Sigma-Aldrich (St. Louis, MO, USA). 96 well plates were from Greiner UV-Star (Greiner Bio-One, Frickenhausen, Germany). Tris-glycine-urea buffer pH 7 contained 0.086 M Tris, 0.09 M glycine, 4 mM citrate, and 8 M urea, adjusted to pH 7 using 2 M HCl. Ferrous ion oxidation-xylenol orange (FOX) reagent contained 25 mM H2SO4 containing 0.1 mM xylenol orange, 0.25 mM FeSO4, and 4 mM BHT in 90% methanol.

2.2. Preparation of Plant Extracts

Grape skin extract was obtained from Tarac Technologies (Nuriootpa, South Australia, Australia). The extract was freeze-dried (Virtis Genesis 35EL, SP Scientific, Warmister, PA, USA) and stored with a small head space with desiccant at −18 °C. Red cabbage extract was produced by the following process. Fresh red cabbage was purchased from a local retailer (Coles supermarket, Werribee, Victoria, Australia). Edible parts of the red cabbage were washed and blended in a food processor with water (1:2 ratio, w/v) before boiling by microwave heating at 800 W for 10 min. After cooling to ambient temperature, the mixture was ultrasonicated at 300 W for 11 min (Hielscher, Germany) before bag filtration (1 µm pore size, Sefar Filtration Inc., Depew, NY, USA). The filtrate was freeze-dried (Virtis Genesis 35EL, SP Scientific, Warmister, PA, USA) and stored with desiccant and low head space at −18 °C.

2.3. Total Phenolic Content of Plant Extracts

Total phenolic content of the plant extracts was quantified using the Folin-Ciocalteu assay [47]. In brief, 20 µL samples (blank, standard, or 2 mg/mL plant extract in 20% methanol) was added to 1 mL of 0.2 N Folin-Ciocalteu reagent and 180 µL of Milli-Q water and mixed for 15 s, and allowed to stand for 3 min before 800 µL of 7.5% Na2CO3 was added to the mixture. The mixture was shaken for 15 s followed by incubation at 37 °C for 1 h in the dark. The absorbance at 765 nm was measured using a Varioskan Flash microplate reader (Thermo Fisher Scientific,

111

Waltham, MA, USA). The total phenolic content of plant extracts was reported as gallic acid equivalent (GAE) using a 7-point calibration curve of gallic acid standard with concentrations of 0–500 µg/mL in 20% methanol after blank subtraction. Total phenolic content of the plant extracts was 26.6 ± 1.5 mg GAE/g for red cabbage extract and 327.1 ± 13.9 mg GAE/g for grape skin extract. Analysis was performed in duplicate.

2.4. Prediction of Human Absorption Kinetics of Plant Extracts

Predicted human absorption kinetics, the ”functional fingerprints” of plant extracts, were determined using untargeted liquid chromatography mass spectrometry (LC-MS) profiling method [42] in combination with the PCAP model [41]. Untargeted LC-MS profiling analysis was performed using an Agilent 6520 quadrupole time-of-flight (QTOF) MS system (Agilent, Santa Clara, CA, USA) with a dual sprayer electrospray ionisation (ESI) source attached to the Agilent 1200 series high performance liquid chromatography (HPLC) system (Santa Clara, CA, USA) comprised of a vacuum degasser and binary pump with a thermostated auto-sampler and column oven. The MS was operated in positive or negative mode using the following conditions (positive/negative, respectively): nebulizer pressure 30/45 psi, gas flow-rate 10 L/min, gas temperature 300 °C, capillary voltage 4000/−3500 V, fragmentor 150, and skimmer 65 V. The instrument was operated in the extended dynamic range mode with data collected in the mass to change ratio (m/z) range of 70–1700. Chromatography was carried out using an Agilent Zorbax Eclipse XDB-C18, 2.1 × 100 mm, 1.8 µm column maintained at 40 °C (±1 °C) at a flow rate of 400 µL/min with a 20-min run time. A gradient LC method was used with mobile phases comprised of (A) 0.1% formic acid in deionized water and (B) 0.1% formic acid in acetonitrile. Gradient: A 5- min linear gradient from 5% to 30% mobile phase B, followed by 5-min gradient to 100% mobile phase B and then a 5-min hold, followed by a 5-min re-equilibration at 5% mobile phase B. Molecular feature extraction (MFE) was conducted using Agilent MassHunter Qualitative analysis (version B.07.00, Agilent) and MassHunter Profinder (version B.06.00, Agilent). Binning and alignment tolerances were set to: retention time: ±0.1% + 0.15 min; mass window: ±20 ppm + 2 mDa. Allowed ion species: H+, Na+, K+, NH4+, and neutral losses: H2O, H3PO4, CO2, C6H12O6. MFE was restricted to the 1000 largest features and 1–2 charge states. After elimination of the molecular features which were common in the two plant extracts (i.e., primary metabolites), the remaining molecular features represented the phytochemicals (i.e., secondary metabolites) of the plant extracts. The lipophilicity descriptor log P was determined using a calibration curve of retention time and log P of twelve standards including L-histidine, (S)-dihydroorotate, shikimate, 4-pyridoxate, 3-hydroxybenzyl alcohol, 2,5-dihydroxybenzoate, 3-hydroxybenzaldehyde, trans-cinnamate, estradiol-17α, deoxycholate, retinoate, oleic acid, and heptadecanoate. Log P values of standards were calculated using the Molinspiration Chemoinformatics calculator. The combination of log P and molecular mass were used to calculate predicted time of maximal plasma absorption (Tmax) in humans using the PCAP model [41]. The functional fingerprints of plant extracts were generated by plotting predicted human Tmax and peak area (relative ion count) of the phytochemicals detected by LC-MS [42].

2.5. Dietary Intervention Using an Animal Model

2.5.1. Animals and Background Diet

The study used six female pigs (Large White × Landrace, 2.5 months old, weight ~30 kg). The pigs weighed 23.2–25.4 kg (mean 24.4 kg, standard error (SE) 3 kg) at the start of the study and 42.8–45.4 kg (mean 44.5 kg, SE 0.3 kg) on study completion five week later. The pigs were housed in individual pens for the duration of the study (12 h light/dark cycle, temperature 18–24 °C). The

112

animals were fed a commercial background diet (Ridley AgriProducts, Melbourne, VIC, Australia) at an energy intake of 0.5 MJ digestible energy/kg body weight (BW)/day representing about 80% of usual energy intake and consumed water ad libitum. The composition of the feed includes 18% protein, 40.37% starch, 2.73% sugar, 4.9% fat, 19.35% fibre, 4.95% ash, 0.9% calcium, and 0.65% phosphorus. The study was approved by the Animal Ethics Committee of the Faculty of Veterinary and Agricultural Sciences, The University of Melbourne, Australia (approval number 1513762.1).

2.5.2. Cephalic Vein Catheterisation Procedure

The cephalic veins of the animals were catheterised under general anaesthesia allowing 7- day post-surgery recovery. Pigs were injected intramuscularly with ketamine hydrochloride (10 mg/kg BW; Ketalar, Pfizer, NY, USA) mixed with xylazine (1 mg/kg BW; Rompun, Bayer, Leverkusen, Germany) to induce sedation and anesthesia. Pigs were then intubated and maintained on 1–4% isoflurane inhalation anaesthesia (Rhone Merieux, Footscray West, VIC, Australia). A silastic catheter was inserted into the external cephalic vein and advanced to the anterior vena cava via the cephalic vein; exteriorisation of the catheter in the interscapular space and storage of the catheter in a cloth pouch glued to the back of the animals was performed as described previously [48]. After catheterisation, the neck incision and exit site were irrigated with benzyl penicillin (BenPen, CSL, Parkville, VIC, Australia) and the animals were given 2 mL of 150 mg/mL of antibiotic amoxicillin (Moxylan, Jurox, Rutherford, NSW, Australia) and 2 mL of 100 mg/mL analgesic/anti-inflammatory ketoprofen (Troy Labs Pty. Ltd., Smithfield, NSW, Australia). After surgery, the animals were monitored for feeding behaviour, general disposition, and rectal temperature. Any animals with elevated temperatures (>39 °C) were given 2 mL of 150 mg/mL amoxicillin. Catheters were flushed daily with physiological saline containing 100 units/mL (U/mL) heparin.

2.5.3. Experimental Design and Procedure

The study was performed in a crossover 4 × 2 factorial design with the factors being two plant extracts at four doses (including placebo control) in triplicate. The wash-out period between treatments was for a minimum of two days. To account for differences in the total phenolic contents of the plant extracts, doses of red cabbage and grape skin extracts were standardised for their total phenolic content as gallic acid equivalents (GAE). On each experiment day, the pigs received a single dose of one of two treatments: red cabbage or grape skin extracts at one of four doses: 0, 2.22, 4.44, and 11.11 mg GAE/kg BW. Considering that the grape skin extract had a higher total phenolic content compared to the red cabbage extract, the doses were selected based on previous studies of grape skin extract administered safely to mice [49–51]. The maximal dose of 11.11 mg GAE/kg BW corresponding to 30 mg grape skin extract/kg BW was selected for our pig study, as this dose is equivalent to the proven safe dose of 200 mg grape skin extract/kg BW in mice [52]. At 8 am on each experiment day, pigs were weighed after an overnight fast. After a baseline (0 h) blood sample, pigs were gavaged with a single dose of plant extract solids reconstituted in water to 50 mL and blood samples collected every 0.5 h for 4 h. The catheter was washed before collecting each blood sample by withdrawing 10 mL of fresh blood. A 10-mL blood sample was then collected using a syringe and immediately placed into a heparinised collection tube (BD Vacutainer®, BD Australia, North Ryde, NSW, Australia) and immediately placed on ice. Lastly, the cannulas were refilled with 100 U/mL heparin in saline and secured in the interscapular pouch. Plasma was obtained by withdrawing supernatants of blood centrifuged at 2000 g for 10

113

min at 4 °C, and aliquots were frozen at −20 °C until analysis. During 4 h of blood sampling period, no foods were given to pigs. After the last blood sampling, pigs were fed the background diet.

2.5.4. Plasma Total Antioxidant Capacity Assay

Plasma total antioxidant capacity ex vivo and in vitro was measured using the cupric reducing antioxidant capacity (CUPRAC) assay [53] and reported as Trolox equivalent antioxidant capacity (TEAC). Plasma TEAC ex vivo was performed on plasma samples collected from the pigs after oral intake of the plant extracts. Plasma TEAC in vitro was performed on reconstituted commercial pig plasma (3.8% trisodium citrate as anticoagulant, Sigma-Aldrich, St. Louis, MO, USA). Freeze-dried commercial pig plasma was reconstituted in Milli Q water to the indicated volume by the manufacture, and aliquots were frozen until analysis. On the day of plasma TEAC in vitro analysis, commercial pig plasma aliquots were thawed and spiked with either gallic acid standard, red cabbage, or grape skin extracts to final concentrations of 0.05, 0.1, 0.2, 0.4, and 0.5 mg GAE/mL. Plasma samples (collected from the pigs or spiked commercial plasma) were diluted 1:5 with Tris-glycine-urea buffer pH 7 before the CUPRAC assay. The CUPRAC assay is based on the capacity of a sample to reduce a Cu(II) complex to a Cu(I) complex, which can be measured at 485 nm wavelength. Equal volumes (50 µL) of 7.5 mM BCS, 10 mM CuCl2 and Tris-glycine-urea buffer were added to each well of a 96-well plate, followed by addition of 100 µL of samples (blank, standard, or diluted plasma). The plate was incubated at 22 °C for 1 h and absorbance at 485 nm was measured. Results were reported as TEAC based on a 6-point calibration curve using Trolox as the standard (0–100 µM) after blank subtraction. Analysis was performed in duplicate. Yields of increased plasma TEAC in vitro (nmol/µmol GAE) by the spiked phytochemicals were reported as the slope of linear regression of plasma TEAC as a function of phytochemical concentrations.

2.5.5. Plasma Glutathione Peroxidase Activity

Plasma GPx activity ex vivo was performed on plasma samples collected from the pigs after oral intake of the plant extracts using a commercial kit (Trevigen, Gaithersburg, MD, USA). Briefly, plasma samples (20 µL) were added to a reaction mixture containing premixed glutathione, reduced form of nicotiamide adenine dinucleotide phosphate (NADPH), glutathione reductase, and cumene hydroperoxide. Absorbance at 340 nm were monitored at 1 min intervals for 15 min, at 25 °C. The GPx activity was calculated from the rate of change in absorbance using GPx standard as a positive control. Results were reported as units/mL, where 1 unit of GPx activity was defined as the amount of enzyme that caused the oxidation of 1 nmol of NADPH to NADP+ per minute at 25 °C. Analysis was performed in triplicate.

2.6. Hydrogen Peroxide Production of Plant Extracts in Pig Plasma In Vitro

The dose response production of H2O2 by phytochemicals in reconstituted commercial pig plasma (Sigma-Aldrich, St. Louis, MO, USA) was measured using the FOX assay [54]. Reconstituted commercial pig plasma was spiked with either gallic acid standard, red cabbage, or grape skin extracts to final concentrations of 0.05, 0.1, 0.2, 0.4, and 0.5 mg GAE/mL and was incubated at 37 °C for 1 h before the FOX assay of H2O2. The concentrations were selected to be in the equivalent range of the doses used in the animal study with pigs having 70 mL circulating blood/kg BW and plasma making up 55% of blood volume [55]. After incubation, the plasma sample was diluted 1:5 with Tris-glycine-urea buffer pH 7 and assays were conducted as follows. 90 µL of samples (blank, standard, or diluted plasma) were mixed with 10 µL of methanol and 900 µL of FOX reagent. The mixture was vortexed for 5 s

114

followed by incubation at 22 °C for 30 min. After incubation, the mixture was centrifuged at 15,000 rpm for 10 min at 22 °C and absorbance of the supernatant was measured at 560 nm. Concentrations of plasma H2O2 were calculated based on a 6-point calibration curve using H2O2 as the standard (0–90 µM) after blank subtraction. Analysis was performed in duplicate. Yields of H2O2 production in vitro (nmol/µmol GAE) by the spiked phytochemicals were reported as the slope of the linear regression of H2O2 concentration as a function of phytochemical concentrations.

2.7. Data Analysis

All curve-fitting was performed using SigmaPlot for Windows Version 12.5 (Systat Software Inc., Chicago, IL, USA). The general linear model (GLM), analysis of covariance (ANCOVA), and Tukey’s test 95% confidence grouping analyses were performed in Minitab 16 statistical software (Minitab Inc., State College, PA, USA). Pearson’s correlation analysis was performed in Minitab 16 statistical software (Minitab Inc.).

3. Results

3.1. Predicted Human Absorption as Functional Fingerprints of Plant Extracts

Predicted absorption as ‘functional fingerprints’ of red cabbage and grape skin extracts were analysed by our LC-MS method with the application of the PCAP model. These functional fingerprints show the predicted ranges of time required for phytochemicals in the extracts to reach their maximal plasma concentrations in human (Tmax) after oral consumption. Accordingly, red cabbage was predicted to have a long Tmax range of 0.4–11 h (Figure 1a) whilst grape skin was predicted to have shorter Tmax ranges of 0.4–3.7 h and 8.2–8.3 h (Figure 1b). The functional fingerprints of the plant extracts informed blood sampling time between 0–4 h at 0.5 h intervals in the current animal study.

106 4x106

8x105 3x106

6x105 2x106 4x105

106 2x105

Peak area (relative ion counts) ion (relative area Peak

Peak area (relative ion counts) area (relative Peak 0 0 0 2 4 6 8 10 12 0 2 4 6 8 10 12 Plasma T (h) Plasma T (h) max max (a) (b)

Figure 1. “Functional fingerprints” of plant extracts predicting absorption in humans based on the PCAP model [41] and the LC-MS method [42]. Functional fingerprints of (a) red cabbage; and (b) grape skin extracts. Tmax, the time required for phytochemicals to reach their maximal plasma concentration.

115

3.2. Time Course Effects of Oral Consumption of Plant Extracts on Plasma Antioxidant Status Ex Vivo of Pigs

The animals consumed four doses of either red cabbage or grape skin extracts (0–11.11 mg GAE/kg BW) and plasma samples were taken every 0.5 h for 4 h. After oral consumption of red cabbage extract, in comparison to baseline at time 0, a significant increase in plasma TEAC was observed at 0.5 h in pigs consuming the maximal dose of 11.11 mg GAE/kg BW (Figure 2a) and a significant increase in plasma GPx activity was observed at 1.5 h in pigs consuming 2.22 mg GAE/kg BW (Figure 2b).

60 Red cabbage 0 140 Red cabbage 0 Red cabbage 2.22 Red cabbage 2.22 55 120 * Red cabbage 4.44 Red cabbage 4.44 Red cabbage 11.11 100 Red cabbage 11.11 M 50 80 45 60 40 40 * 35

Plasma TEAC Plasma TEAC 20 30 * (U/mL) Plasma GPx activity 0 25 0 1 2 3 4 0 1 2 3 4 Time (h) Time (h) (a) (b)

Figure 2. Effects of oral consumption of red cabbage extract on the plasma antioxidant status of pigs. Pigs consumed red cabbage extract at four doses in mg gallic acid equivalent/kg body weight: 0 (black circle), 2.22 (white circle), 4.44 (black triangle), and 11.11 (white triangle). Plasma antioxidant status was measured as: (a) plasma Trolox equivalent antioxidant capacity (TEAC); and (b) plasma glutathione peroxidase (GPx) activiy. Data points labelled “*” are significantly different from baseline at time 0 (p ≤ 0.05, Tukey’s test). Results represent the mean and error bars represent standard error of the mean (N = 3).

After consumption of grape skin extract, in comparison to baseline at time 0, a significant increase in plasma TEAC was observed after 1 h in pigs consuming 2.22 mg GAE/kg BW (Figure 3a) and significant increases of plasma GPx activity were observed at 2.5, 3.5, and 4 h in in pigs consuming 4.44 mg GAE/kg BW (Figure 3b). In contrast, a significant reduction in plasma GPx activity was observed at 1 h in pigs consuming 4.44 mg GAE/kg BW (Figure 3b).

116

60 Grape skin 0 140 Grape skin 0 Grape skin 2.22 Grape skin 2.22 55 Grape skin 4.44 120 Grape skin 4.44 Grape skin 11.11 Grape skin 11.11

M 50 100

45 * 80 * *

40 60 * * 35 40

Plasma TEAC Plasma TEAC 30 20

Plasma GPx activity (U/mL) Plasma GPx activity

25 0 * 0 1 2 3 4 0 1 2 3 4 Time (h) Time (h) (a) (b)

Figure 3. Effects of oral consumption of grape skin extract on the plasma antioxidant status of pigs. Pigs consumed grape skin extract at four doses in mg gallic acid equivalent/kg body weight: 0 (black circle), 2.22 (white circle), 4.44 (black triangle), and 11.11 (white triangle). Plasma antioxidant status was measured as: (a) plasma Trolox equivalent antioxidant capacity (TEAC); and (b) plasma glutathione peroxidase (GPx) activiy. Data points labelled “*” are significantly different from baseline at time 0 (p ≤ 0.05, Tukey’s test). Results represent the mean and error bars represent standard error of the mean (N = 3).

3.3. Effects of Plant Extracts on Plasma Total Antioxidant Capacity and Plasma Hydrogen Peroxide Concentration In Vitro

The dose response effects of the plant extracts on plasma TEAC and plasma H2O2 concentration in vitro were analysed after spiking plasma with either gallic acid standard, red cabbage, or grape skin extracts to final concentrations of 0.05, 0.1, 0.2, 0.4, and 0.5 mg GAE/mL. Proportional increase in plasma TEAC was observed with increased concentrations of all three phytochemical sources and followed linear regression relationships (Table 1). Yields of increased plasma TEAC in vitro by the phytochemicals were 1606.3 ± 98.1, 633.2 ± 74.7, and 1077.8 ± 120.4 nmol/µmol GAE for gallic acid standard, red cabbage, and grape skin extracts, respectively (Table 1).

Table 1. Effects of plant extracts on plasma Trolox equivalent antioxidant capacity (TEAC) and plasma levels of hydrogen peroxide (H2O2) in vitro.

Plasma TEAC Plasma H2O2 Phytochemical Sources Yield (nmol/µmol GAE) * Linear Fit R2 Yield (nmol/µmol GAE) * Linear Fit R2 Gallic acid standard 1606.3 ± 98.1 0.99 68.7 ± 4.5 0.97 Red cabbage extract 633.2 ± 74.7 0.96 22.4 ± 1.1 0.99 Grape skin extract 1077.8 ± 120.4 0.96 44.2 ± 2.1 0.99 * Gallic acid standard and plant extracts were directly spiked into commercial pig plasma at concentrations of 0.05–0.5 mg gallic acid equivalent (GAE)/mL. Increased plasma TEAC and plasma H2O2 levels followed linear regressions with slopes representing yields of increase. Comparing three phytochemical sources, significant differences in yields of plasma TEAC and

plasma H2O2 were observed (p ≤ 0.05, analysis of covariance (ANCOVA)). Significantly high correlation between plasma TEAC and plasma H2O2 was observed (r = 1, p ≤ 0.05, Pearson’s correlation analysis). Results represent the mean ± standard error of the mean (N = 2).

Similar to plasma TEAC in vitro, proportional increase in plasma H2O2 concentrations was observed with increased concentrations of all three phytochemical sources and followed linear regression relationships (Table 1). Yields of H2O2 production in vitro by the phytochemicals in

117

plasma were 68.7 ± 4.5, 22.4 ± 1.1, and 44.2 ± 2.1 nmol/µmol GAE for gallic acid standard, red cabbage, and grape skin extracts, respectively (Table 1). Comparing the three phytochemical sources, significant differences in yields of plasma TEAC and plasma H2O2 were observed (p ≤ 0.05, ANCOVA). Further, significantly high correlation between yields of plasma TEAC and plasma H2O2 were observed (r = 1, p ≤ 0.05, Pearson’s correlation analysis), with gallic acid having the strongest effect (highest yields) followed by grape skin and red cabbage extracts (Table 1).

3.4. Effects of Phytochemical Dose and Their H2O2 Production Capacity In Vitro on Plasma Antioxidant Status of Pigs Ex Vivo

Means across all pig plasma sampling points (0.5 h interval for 4 h, Figures 2 and 3) were combined to investigate the overall dose effects of the plant extracts on pig plasma antioxidant status (Figure 4). For both plant extracts, plasma TEAC ex vivo significantly increased at all three doses 2.22, 4.44, and 11.11 mg GAE/kg BW compared to dose 0 (Figure 4a). There was no significant difference in plasma TEAC among the three doses of red cabbage extract whilst plasma TEAC at grape skin extract dose of 4.44 and 11.11 mg GAE/kg BW was significantly reduced compared to the 2.22 mg GAE/kg BW dose (Figure 4a). The phytochemical dose (mg GAE/kg BW) of the two plant extracts was standardised to their in vitro H2O2 production yields (nmol/µmol GAE, Table 1) to estimate the H2O2 production (nmol/kg BW) by the plant extract dose used in the animal study. The in vitro H2O2 production yields of the two plant extracts had similar effects on the mean plasma TEAC of pigs compared to their phytochemical dose (Figure 4b).

118

46 Red cabbage 46 Red cabbage Grape skin Grape skin 44 44 * * *

M) 42 * M) 42 * * 40 40

38 * ,# 38 * * * * ,# * 36 36

Plasma TEAC ( TEAC Plasma

Plasma TEAC ( TEAC Plasma 34 34

32 32 0 2 4 6 8 10 12 0 500 1000 1500 2000 2500 3000 H O production (nmol/kg body weight) Dose (mg GAE/kg body weight) 2 2 (a) (b) 80 Red cabbage 80 Red cabbage Grape skin Grape skin 70 70 * *,# * 60 * 60 * * * 50 * 50 *,# 40 * * 40 * 30 30 *

20 (U/mL) Plasma GPx activity

Plasma GPx activity (U/mL) Plasma GPx activity

10 10 0 2 4 6 8 10 12 0 500 1000 1500 2000 2500 3000 H O production (nmol/kg body weight) Dose (mg GAE/kg body weight) 2 2 (c) (d)

Figure 4. Total plasma antioxidant capacity and glutathione peroxidase activity of pig plasma as a function of phytochemical dose and H2O2 production efficacy. Means across all pig plasma sampling time points (0.5 h interval for 4 h) of plasma TEAC versus (a) phytochemical doses and (b) H2O2 production efficacy. Means across all pig plasma sampling time points of plasma GPx activity versus (c) phytochemical doses and (d) H2O2 production efficacy. The H2O2 production

(nmol/kg body weight) was calculated based on the yield of H2O2 production (nmol/µmol GAE) of the plant extracts in vitro (Table 1). Data points labelled “*” are significantly different from dose 0 (p ≤ 0.05, Tukey’s test). Data points labelled “#” are significantly different from the previous dose (p ≤ 0.05, Tukey’s test). Results represent the mean and error bars represent standard error of the mean (N = 27).

Similarly, for both plant extracts, significant increases in pig plasma GPx activity were observed at all three doses (Figure 4c). There was no significant difference in GPx activity among the three doses of grape skin extract whilst GPx activity at a red cabbage extract dose of 4.44 mg GAE/kg BW was significantly increased compared to the 2.22 mg GAE/kg BW (Figure 4c). After standardisation of the phytochemical dose to their in vitro H2O2 production yields (Table 1), the plasma GPx activity in response to the two plant extracts was remarkably similar (Figure 4d).

119

4. Discussion

This study examines the consumption of dietary phytochemicals by pigs and shows that non-enzymatic and enzymatic antioxidant defences were increased. Absorption kinetics of red cabbage and grape skin extracts were characterised in pigs after oral consumption using plasma TEAC as a measure of the non-enzymatic antioxidant response [56] and plasma GPx (an antioxidant enzyme) activity [57]. The blood sampling time of the study (0.5 h interval for 4 hours) was chosen to capture the range of time expected for the phytochemicals to achieve their maximal plasma concentrations (Tmax), predicted from their functional fingerprints (0.4–4 h). Consistent with the predicted functional fingerprints, a significant increase in plasma TEAC was observed at 0.5 h after consumption of red cabbage (11.11 mg GAE/kg BW) and at 1 h after consumption of grape skin (2.22 mg GAE/kg BW). Peaks of plasma TEAC have been observed to coincide with peaks of plasma phytochemicals in humans after consumption of tea [58] and chocolate [59]. Therefore, the identification of increased plasma TEAC within this selected time frame after plant extract ingestion validates the utility of the phytochemical absorption prediction (PCAP) model [41] and its application to the production of the functional fingerprints. These results highlight the ability of the PCAP model to guide experimental design to ensure that the functional impact of the phytochemicals is captured during the sampling regime. For example, a previous study investigating the pharmacokinetics of three phytochemicals carvacrol, thymol, and eugenol in pigs reported the time of maximal absorption (Tmax) at 1.39, 1.35, and 0.83 h, respectively [60]. Using our PCAP model [41], the Tmax of these phytochemicals for humans was predicted to be 1.76, 1.67, and 1.58 h, respectively. Comparing to the reported Tmax in pigs [60], the predicted Tmax of these phytochemicals in humans was very similar and followed the same sequence with Tmax of carvacrol > thymol > eugenol. This similarity of observed Tmax compared to predicted Tmax suggests that the PCAP model can be useful for predicting absorption of phytochemicals in pigs as well as in humans. In the present study, the plant extract doses were standardised for their respective total phenolic contents as GAE analysed by the Folin-Ciocateu assay. Whilst ascorbic acid is known to interfere with this assay [61], based on analyses conducted by others [61–67], the contributions of ascorbic acid to the GAE results are estimated to be 0.3% and 4% for grape skin and red cabbage extracts, respectively. These minimal contributions of ascorbic acid to the GAE results reflected the naturally low ascorbic acid content of grape skin [62], and the effects of microwave cooking [66] and ultrasonication [67] which reduced the ascorbic acid content of red cabbage during plant processing. Accordingly, the GAE results presented herein are considered accurate indicators of the total phenolic content of the two plant extracts. In comparison to plasma TEAC, a delayed increase in plasma GPx activity was observed at 1.5 h after consumption of red cabbage (2.22 mg GAE/kg BW) and at 2.5, 3.5, and 4 h after consumption of grape skin (4.44 mg GAE/kg BW). The observed time delay of plasma GPx activity after plasma TEAC is consistent with a previous study [68]. This delay may be explained by the induction of GPx activity occurring in response to the presence of phytochemicals in the plasma, as indicated by increased plasma TEAC [58,59]. Accordingly, increased plasma TEAC and increased plasma GPx activity after consumption of the plant extracts indicate that phytochemicals provide health benefits via both direct antioxidant activity and indirectly via the induction of enzymatic antioxidant defence mechanisms. The dose response effects of red cabbage and grape skin extracts increased plasma TEAC in vitro after direct addition of the extracts to the pig plasma in the present study. As the phytochemical doses increased, there was a proportional increase in plasma TEAC in vitro (633.2– 1606.3 nmol/µmol GAE), supporting the direct antioxidant activity of phytochemicals in vitro as observed in many studies [69–71]. In comparison to the in vitro experiments, same doses of red cabbage and grape skin extracts consumed by the pigs did not result in a proportional increase

120

in plasma TEAC and plasma GPx activity ex vivo. When plant extracts were orally administered to pigs, increased plasma TEAC was observed at all doses compared to 0 but an increase in dose did not result in significant further increase of TEAC above the lowest dose. Further, an increase in dose of grape skin extract resulted in decreased plasma TEAC at doses of 4.44 and 11.11 mg GAE/kg BW compared to the dose at 2.22 mg/kg BW. The differences in plasma TEAC responses to the plant extracts may be attributed to their distinct phytochemical compositions [44,46]. Similarly, plasma GPx activity significantly increased for all doses compared to dose 0 but further increase of doses did not show a clear response relationship. The observed differences between in vitro and ex vivo have also been observed in other studies [69,70]. Direct addition of tea [69] or apple phytochemicals [70] to human plasma in vitro increased plasma TEAC. However, consumption of the same or higher concentrations of tea [69] and apple phytochemicals [70] by humans did not reproduce the same effects as observed in vitro. The differences between in vitro and ex vivo results can be explained by the low bioavailability of phytochemicals in vivo as they are handled by the body as xenobiotics [40]. Further, these differences may be attributed to the increased complexity of the in vivo system wherein both direct and indirect antioxidant mechanisms may arise, as indicated by increased plasma GPx activity ex vivo. Hypothetical pro-oxidant effects of phytochemicals in vitro via measurement of H2O2 levels in plasma were studied after direct addition of plant extracts. Similar to the results measuring plasma TEAC in vitro, incubation of red cabbage and grape skin extracts in pig plasma resulted in a proportional increase in plasma H2O2 levels (22.4–68.7 nmol/µmol GAE). Pro-oxidant effects of phytochemicals in vitro have been observed in the presence of oxygen and metal ions such as copper and iron [35–37,72–75]. Concentrations of iron and copper ions in human plasma are 2.13 and 0.81 µg/g, respectively [76], and iron levels of 0.1 µg/g [35] and copper levels of 3 µg/g [73] have been reported to initiate H2O2 production in vitro. The formation of H2O2 by phytochemicals in plasma observed here may be attributed to the electron transfer process between phytochemicals, oxygen, and metal ions present in plasma [73]. The ability of phytochemicals to produce H2O2 has been proposed to be responsible for the cytotoxic effects of phytochemicals in cell culture studies [35–37]. H2O2 has been widely used as an oxidative stress inducer in many studies investigating the protective effects of phytochemicals in response to oxidative stress [77–79]. However, the ability of phytochemicals to produce H2O2 may explain their indirect antioxidant protection mechanism. High concentrations of H2O2 (≥100 µM) are harmful for cells but low concentrations of H2O2 (≤50 µM) can be beneficial to initiate the antioxidant cellular defence [80,81]. Low concentrations of H2O2 have been observed to stimulate wound healing in keratinocytes [82] and in mice [83]. Similarly, the health benefits of regular exercise have been proposed to be associated with their production of low levels of ROS (such as H2O2) that induce adaptive responses to protect against molecular damage and, subsequently, aging [84,85]. Supporting this mechanism, H2O2 has been reported to activate the nuclear factor- erythroid-2-related factor 2 (Nrf2) de novo [86] which is a transcription factor involved in inducing the antioxidant response by regulating coordinated induction of stress response genes encoding antioxidant enzymes such as superoxide dismutase, catalase, and GPx [87]. Activation of Nrf2 has been proposed as a therapeutic potential for protection against chronic diseases [87,88]. Many phytochemicals are known as Nrf2 activators including curcumin (in turmeric) [89], epigallocatechin gallate (in green tea) [90], lycopene (in tomato) [91], resveratrol (in grape) [92], and sulforaphane (in broccoli) [93]. Accordingly, H2O2-mediated induction of Nrf2 in response to phytochemical supply may explain the correlation between H2O2 production and increased plasma GPx activity observed in our study.

121

5. Conclusions

The findings of the current study provide new insights in mechanisms by which dietary phytochemicals impact health, apart from direct ROS-scavenging pathways. An additional role is proposed whereby protection against oxidative tissue damage results from the promotion of cellular oxidative stress defence by dietary phytochemicals. This research demonstrates for the first time that H2O2 production analysis represents a useful predictive indicator of the in vivo efficacy of dietary phytochemicals.

Acknowledgments: This project has been funded by Horticulture Innovation Australia Limited using the Vegetable levy and funds from the Australian Government. We gratefully acknowledge the donation of the grape skin extract from Tarac Technologies and assistance during the animal trial by Maree Cox, Shannon Holbrook, Peter Cakebread, Ruslan Pustovit, Udanni Wijesiriwardana, Paula Andrea Giraldo Parra, and Caroline Storer.

Author Contributions: Conception and experimental design: S.N.B.S.-P., L.E.B., F.R.D. and J.J.C. Study execution: S.N.B.S.-P., J.J.C. and F.R.D. Data collection, analysis and interpretation: S.N.B.S.-P., L.E.B., F.R.D., J.J.C., K.N. and K.S.H. Manuscript writing and review: S. N.B.S.-P., L.E.B., K.S.H., J.J.C., K.N. and F.R.D. All authors read and approved the final manuscripts.

Conflicts of Interest: The authors declare no conflicts of interest.

Reference

1. Key, T.J. Fruit and vegetables and cancer risk. Br. J. Cancer 2011, 104, 6–11. 2. Dauchet, L.; Amouyel, P.; Dallongeville, J. Fruits, vegetables and coronary heart disease. Nat. Rev. Cardiol. 2009, 6, 599–608. 3. D’Onofrio, G.; Sancarlo, D.; Ruan, Q.; Yu, Z.; Panza, F.; Daniele, A.; Greco, A.; Seripa, D. Phytochemicals in the treatment of alzheimer’s disease: A systematic review. Curr. Drug Targets 2016, 17, doi:10.2174/1389450117666161102121553. 4. Burton-Freeman, B. Postprandial metabolic events and fruit-derived phenolics: A review of the science. Br. J. Nutr. 2010, 104, S1–S14. 5. van der Merwe, M.; Bloomer, R.J. The influence of methylsulfonylmethane on inflammation-associated cytokine release before and following strenuous exercise. J. Sports Med. 2016, 2016, 7498359. 6. Calder, P.C.; Albers, R.; Antoine, J.M.; Blum, S.; Bourdet-Sicard, R.; Ferns, G.A.; Folkerts, G.; Friedmann, P.S.; Frost, G.S.; Guarner, F.; et al. Inflammatory disease processes and interactions with nutrition. Br. J. Nutr. 2009, 101, S1–S45. 7. Halliwell, B. Reactive species and antioxidants. Redox biology is a fundamental theme of aerobic life. Plant Physiol. 2006, 141, 312–322. 8. Kryston, T.B.; Georgiev, A.B.; Pissis, P.; Georgakilas, A.G. Role of oxidative stress and DNA damage in human carcinogenesis. Mutat. Res. Fund. Mol. Mech. Mut. 2011, 711, 193–201. 9. Fang, F.C. Antimicrobial reactive oxygen and nitrogen species: Concepts and controversies. Nat. Rev. Microbiol. 2004, 2, 820–832. 10. Bøhn, S.K.; Myhrstad, M.C.; Thoresen, M.; Holden, M.; Karlsen, A.; Tunheim, S.H.; Erlund, I.; Svendsen, M.; Seljeflot, I.; Moskaug, J.Ø.; et al. Blood cell gene expression associated with cellular stress defense is modulated by antioxidant-rich food in a randomised controlled clinical trial of male smokers. BMC Med. 2010, 8, 54. 11. Krishnaiah, D.; Sarbatly, R.; Nithyanandam, R. A review of the antioxidant potential of medicinal plant species. Food Bioprod. Process. 2011, 89, 217–233. 12. Lee, J.; Koo, N.; Min, D.B. Reactive oxygen species, aging, and antioxidative nutraceuticals. Compr. Rev. Food Sci. Food Saf. 2004, 3, 21–33. 13. Seifried, H.E.; Anderson, D.E.; Fisher, E.I.; Milner, J.A. A review of the interaction among dietary antioxidants and reactive oxygen species. J. Nutr. Biochem. 2007, 18, 567–579.

122

14. Del Rio, D.; Rodriguez-Mateos, A.; Spencer, J.P.; Tognolini, M.; Borges, G.; Crozier, A. Dietary (poly)phenolics in human health: Structures, bioavailability, and evidence of protective effects against chronic diseases. Antioxid. Redox. Signal. 2013, 18, 1818–1892. 15. Lotito, S.B.; Frei, B. Consumption of flavonoid-rich foods and increased plasma antioxidant capacity in humans: Cause, consequence, or epiphenomenon? Free Radic. Biol. Med. 2006, 41, 1727–1746. 16. Kong, A.N.; Yu, R.; Lei, W.; Mandlekar, S.; Tan, T.H.; Ucker, D.S. Differential activation of MAPK and ICE/CED-3 protease in chemical-induced apoptosis. The role of oxidative stress in the regulation of mitogen-activated protein kinases (MAPKs) leading to gene expression and survival or activation of caspases leading to apoptosis. Restor. Neurol. Neurosci. 1998, 12, 63–70. 17. Kong, A.N.; Owuor, E.; Yu, R.; Hebbar, V.; Chen, C.; Hu, R.; Mandlekar, S. Induction of xenobiotic enzymes by the MAP kinase pathway and the antioxidant or electrophile response element (ARE/EpRE). Drug Metab. Rev. 2001, 33, 255–271. 18. Finley, J.W.; Kong, A.N.; Hintze, K.J.; Jeffery, E.H.; Ji, L.L.; Lei, X.G. Antioxidants in foods: State of the science important to the food industry. J. Agric. Food Chem. 2011, 59, 6837–6846. 19. Drew, J.E. Cellular defense system gene expression profiling of human whole blood: Opportunities to predict health benefits in response to diet. Adv. Nutr. 2012, 3, 499–505. 20. Lee, W.L.; Huang, J.Y.; Shyur, L.F. Phytoagents for cancer management: Regulation of nucleic acid oxidation, ROS, and related mechanisms. Oxid. Med. Cell. Longev. 2013, 2013, 925804. 21. Manjunatha, H.; Srinivasan, K. Protective effect of dietary curcumin and capsaicin on induced oxidation of low-density lipoprotein, iron-induced hepatotoxicity and carrageenan-induced inflammation in experimental rats. FEBS J. 2006, 273, 4528–4537. 22. Srihari, T.; Sengottuvelan, M.; Nalini, N. Dose-dependent effect of oregano (origanum vulgare L.) on lipid peroxidation and antioxidant status in 1,2-dimethylhydrazine-induced rat colon carcinogenesis. J. Pharm. Pharmacol. 2008, 60, 787–794. 23. Kim, B.; Ku, C.S.; Pham, T.X.; Park, Y.; Martin, D.A.; Xie, L.; Taheri, R.; Lee, J.; Bolling, B.W. Aronia melanocarpa (chokeberry) polyphenol-rich extract improves antioxidant function and reduces total plasma cholesterol in apolipoprotein E knockout mice. Nutr. Res. 2013, 33, 406–413. 24. Ali, H.A.; Afifi, M.; Abdelazim, A.M.; Mosleh, Y.Y. Quercetin and omega 3 ameliorate oxidative stress induced by aluminium chloride in the brain. J. Mol. Neurosci. 2014, 53, 654–660. 25. Gourineni, V.; Shay, N.F.; Chung, S.; Sandhu, A.K.; Gu, L. Muscadine grape (vitis rotundifolia) and wine phytochemicals prevented obesity-associated metabolic complications in C57BL/6J mice. J. Agric. Food Chem. 2012, 60, 7674–7681. 26. Belviranli, M.; Gokbel, H.; Okudan, N.; Basarali, K. Effects of grape seed extract supplementation on exercise-induced oxidative stress in rats. Br. J. Nutr. 2012, 108, 249–256. 27. Natella, F.; Belelli, F.; Gentili, V.; Ursini, F.; Scaccini, C. Grape seed proanthocyanidins prevent plasma postprandial oxidative stress in humans. J. Agric. Food Chem. 2002, 50, 7720–7725. 28. Serafini, M.; Bugianesi, R.; Maiani, G.; Valtuena, S.; De Santis, S.; Crozier, A. Plasma antioxidants from chocolate. Nature 2003, 424, 1013–1013. 29. Pedersen, C.B.; Kyle, J.; Jenkinson, A.M.; Gardner, P.T.; McPhail, D.B.; Duthie, G.G. Effects of blueberry and cranberry juice consumption on the plasma antioxidant capacity of healthy female volunteers. Eur. J. Clin. Nutr. 2000, 54, 405–408. 30. Sung, H.; Nah, J.; Chun, S.; Park, H.; Yang, S.E.; Min, W.K. In vivo antioxidant effect of green tea. Eur. J. Clin. Nutr. 2000, 54, 527–529. 31. Guarrera, S.; Sacerdote, C.; Fiorini, L.; Marsala, R.; Polidoro, S.; Gamberini, S.; Saletta, F.; Malaveille, C.; Talaska, G.; Vineis, P.; et al. Expression of DNA repair and metabolic genes in response to a flavonoid-rich diet. Br. J. Nutr. 2007, 98, 525–533. 32. Wang, J.; Siegmund, K.; Tseng, C.C.; Lee, A.S.; Wu, A.H. Soy food supplementation, dietary fat reduction and peripheral blood gene expression in postmenopausal women—A randomized, controlled trial. Mol. Nutr. Food Res. 2011, 55, S264–S277. 33. Bakker, G.C.; van Erk, M.J.; Pellis, L.; Wopereis, S.; Rubingh, C.M.; Cnubben, N.H.; Kooistra, T.; van Ommen, B.; Hendriks, H.F. An antiinflammatory dietary mix modulates inflammation and oxidative

123

and metabolic stress in overweight men: A nutrigenomics approach. Am. J. Clin. Nutr. 2010, 91, 1044– 1059. 34. Halliwell, B.; Whiteman, M. Measuring reactive species and oxidative damage in vivo and in cell culture: How should you do it and what do the results mean? Br. J. Pharmacol. 2004, 142, 231–255. 35. Long, L.H.; Clement, M.V.; Halliwell, B. Artifacts in cell culture: Rapid generation of hydrogen peroxide on addition of (−)-epigallocatechin, (−)-epigallocatechin gallate, (+)-catechin, and quercetin to commonly used cell culture media. Biochem. Biophys. Res. Commun. 2000, 273, 50–53. 36. Chai, P.C.; Long, L.H.; Halliwell, B. Contribution of hydrogen peroxide to the cytotoxicity of green tea and red wines. Biochem. Biophys. Res. Commun. 2003, 304, 650–654. 37. Long, L.H.; Hoi, A.; Halliwell, B. Instability of, and generation of hydrogen peroxide by, phenolic compounds in cell culture media. Arch. Biochem. Biophys. 2010, 501, 162–169. 38. Rodd, A.L.; Ververis, K.; Sayakkarage, D.; Khan, A.W.; Rafehi, H.; Ziemann, M.; Loveridge, S.J.; Lazarus, R.; Kerr, C.; Lockett, T.; et al. RNA sequencing supports distinct reactive oxygen species- mediated pathways of apoptosis by high and low size mass fractions of bay leaf (Lauris nobilis) in HT- 29 cells. Food Funct. 2015, 6, 2507–2524. 39. Lee, C.Y. Challenges in providing credible scientific evidence of health benefits of dietary polyphenols. J. Funct. Foods 2013, 5, 524–526. 40. Holst, B.; Williamson, G. Nutrients and phytochemicals: From bioavailability to bioefficacy beyond antioxidants. Curr. Opin. Biotechnol. 2008, 19, 73–82. 41. Selby-Pham, S.N.B.; Miller, R.B.; Howell, K.; Dunshea, F.; Bennett, L.E. Physicochemical properties of dietary phytochemicals can predict their passive absorption in the human small intestine. Sci. Rep. 2017, 7, 1931. 42. Selby-Pham, S.N.B.; Howell, K.S.; Dunshea, F.R.; Ludbey, J.; Lutz, A.; Bennett, L.E. High throughput prediction of human absorption kinetics of plant extracts using LC-MS and statistical modelling. Food Chem. 2017, in press. 43. Roura, E.; Koopmans, S.J.; Lalles, J.P.; Le Huerou-Luron, I.; de Jager, N.; Schuurman, T.; Val-Laillet, D. Critical review evaluating the pig as a model for human nutritional physiology. Nutr. Res. Rev. 2016, 29, 60–90. 44. Flamini, R.; Mattivi, F.; De Rosso, M.; Arapitsas, P.; Bavaresco, L. Advanced knowledge of three important classes of grape phenolics: Anthocyanins, stilbenes and flavonols. Int. J. Mol. Sci. 2013, 14, 19651–19669. 45. Wagner, A.E.; Terschluesen, A.M.; Rimbach, G. Health promoting effects of brassica-derived phytochemicals: From chemopreventive and anti-inflammatory activities to epigenetic regulation. Oxid. Med. Cell. Longev. 2013, 2013, 964539. 46. Huang, H.; Jiang, X.; Xiao, Z.; Yu, L.; Pham, Q.; Sun, J.; Chen, P.; Yokoyama, W.; Yu, L.L.; Luo, Y.S.; et al. Red cabbage microgreens lower circulating low-density lipoprotein (LDL), liver cholesterol, and inflammatory cytokines in mice fed a high-fat diet. J. Agric. Food Chem. 2016, 64, 9161–9171. 47. Ainsworth, E.A.; Gillespie, K.M. Estimation of total phenolic content and other oxidation substrates in plant tissues using folin-ciocalteu reagent. Nat. Protoc. 2007, 2, 875–877. 48. Ostrowska, E.; Cross, R.F.; Muralitharan, M.; Bauman, D.E.; Dunshea, F.R. Effects of dietary fat and conjugated linoleic acid on plasma metabolite concentrations and metabolic responses to homeostatic signals in pigs. Br. J. Nutr. 2002, 88, 625–634. 49. Asseburg, H.; Schafer, C.; Muller, M.; Hagl, S.; Pohland, M.; Berressem, D.; Borchiellini, M.; Plank, C.; Eckert, G.P. Effects of grape skin extract on age-related mitochondrial dysfunction, memory and life span in C57BL/6J mice. Neuromol. Med. 2016, 18, 378–395. 50. Costa, M.R.; Pires, K.M.; Nalbones-Barbosa, M.N.; Dos Santos Valenca, S.; Resende, A.C.; de Moura, R.S. Grape skin extract-derived polyphenols modify programming-induced renal endowment in prenatal protein-restricted male mouse offspring. Eur. J. Nutr. 2016, 55, 1455–1464. 51. Resende, A.C.; Emiliano, A.F.; Cordeiro, V.S.; de Bem, G.F.; de Cavalho, L.C.; de Oliveira, P.R.; Neto, M.L.; Costa, C.A.; Boaventura, G.T.; de Moura, R.S. Grape skin extract protects against programmed

124

changes in the adult rat offspring caused by maternal high-fat diet during lactation. J. Nutr. Biochem. 2013, 24, 2119–2126. 52. Nair, A.B.; Jacob, S. A simple practice guide for dose conversion between animals and human. J. Basic Clin. Pharm. 2016, 7, 27–31. 53. Marques, S.; Magalhães, L.; Tóth, I.; Segundo, M. Insights on antioxidant assays for biological samples based on the reduction of copper complexes—The importance of analytical conditions. Int. J. Mol. Sci. 2014, 15, 11387–11402. 54. Wolff, S.P. Ferrous ion oxidation in presence of ferric ion indicator xylenol orange for measurement of hydroperoxides. Methods Enzymol. 1994, 233, 182–189. 55. Wolfensohn, S.; Lloyd, M. The larger domestic species. In Handbook of Laboratory Animal Management and Welfare, 3rd ed.; Wolfensohn, S., Lloyd, M., Eds.; Blackwell Publishing: Ames, IA, USA, 2003; pp. 326–364. 56. Apak, R.; Guclu, K.; Ozyurek, M.; Karademir, S.E.; Altun, M. Total antioxidant capacity assay of human serum using copper(II)-neocuproine as chromogenic oxidant: The CUPRAC method. Free Radic. Res. 2005, 39, 949–961. 57. Hayes, J.D.; McLellan, L.I. Glutathione and glutathione-dependent enzymes represent a co-ordinately regulated defence against oxidative stress. Free Radic. Res. 1999, 31, 273–300. 58. Leenen, R.; Roodenburg, A.J.; Tijburg, L.B.; Wiseman, S.A. A single dose of tea with or without milk increases plasma antioxidant activity in humans. Eur. J. Clin. Nutr. 2000, 54, 87–92. 59. Rein, D.; Lotito, S.; Holt, R.R.; Keen, C.L.; Schmitz, H.H.; Fraga, C.G. Epicatechin in human plasma: In vivo determination and effect of chocolate consumption on plasma oxidation status. J. Nutr. 2000, 130, 2109s–2114s. 60. Michiels, J.; Missotten, J.; Dierick, N.; Fremaut, D.; Maene, P.; De Smet, S. In vitro degradation and in vivo passage kinetics of carvacrol, thymol, eugenol and trans-cinnamaldehyde along the gastrointestinal tract of piglets. J. Sci. Food Agric. 2008, 88, 2371–2381. 61. Everette, J.D.; Bryant, Q.M.; Green, A.M.; Abbey, Y.A.; Wangila, G.W.; Walker, R.B. Thorough study of reactivity of various compound classes toward the folin−ciocalteu reagent. J. Agric. Food Chem. 2010, 58, 8139–8144. 62. Singha, I.; Das, S.K. Free radical scavenging properties of skin and pulp extracts of different grape cultivars in vitro and attentuation of H2O2-induced oxidative stress in liver tissue ex vivo. Ind. J. Clin. Biochem. 2015, 30, 305–312. 63. Gabas, A.L.; Telis-Romero, J.; Menegalli, F.C. Thermodynamic models for water sorption by grape skin and pulp. Drying Technol. 1999, 17, 962–974. 64. Singh, J.; Upadhyay, A.K.; Bahadur, A.; Singh, K.P.; Rai, M. Antioxidant phytochemicals in cabbage (Brassica oleracea L. Var. Capitata). Sci. Hort. 2006, 108, 233–237. 65. Sivakumaran, S.; Huffman, L.; Sivakumaran, S. The Concise New Zealand Food Composition Tables; 12th ed.; The New Zealand Institute for Plant & Food Research Limited and Ministry of Health: Palmerston North, New Zealand, 2017. 66. Zeng, C. Effects of different cooking methods on the vitamin c content of selected vegetables. Nutr. Food Sci. 2013, 43, 438–443. 67. Tiwari, B.K.; O’Donnell, C.P.O.; Patras, A.; Cullen, P.J. Anthocyanin and ascorbic acid degradation in sonicated strawberry juice. J. Agric. Food Chem. 2008, 56, 10071–10077. 68. Alleva, R.; Di Donato, F.; Strafella, E.; Staffolani, S.; Nocchi, L.; Borghi, B.; Pignotti, E.; Santarelli, L.; Tomasetti, M. Effect of ascorbic acid-rich diet on in vivo-induced oxidative stress. Br. J. Nutr. 2012, 107, 1645–1654. 69. Cherubini, A.; Beal, M.F.; Frei, B. Black tea increases the resistance of human plasma to lipid peroxidation in vitro, but not ex vivo. Free Radic. Biol. Med. 1999, 27, 381–387. 70. Lotito, S.B.; Frei, B. Relevance of apple polyphenols as antioxidants in human plasma: Contrasting in vitro and in vivo effects. Free Radic. Biol. Med. 2004, 36, 201–211.

125

71. Kolodziejczyk-Czepas, J.; Nowak, P.; Moniuszko-Szajwaj, B.; Kowalska, I.; Stochmal, A. Free radical scavenging actions of three trifolium species in the protection of blood plasma antioxidant capacity in vitro. Pharm. Biol. 2015, 53, 1277–1284. 72. Sahu, S.C.; Gray, G.C. Interactions of flavonoids, trace metals, and oxygen: Nuclear DNA damage and lipid peroxidation induced by myricetin. Cancer Lett. 1993, 70, 73–79. 73. Zheng, L.F.; Wei, Q.Y.; Cai, Y.J.; Fang, J.G.; Zhou, B.; Yang, L.; Liu, Z.L. DNA damage induced by resveratrol and its synthetic analogues in the presence of Cu (II) ions: Mechanism and structure-activity relationship. Free Radic. Biol. Med. 2006, 41, 1807–1816. 74. Fukumoto, L.R.; Mazza, G. Assessing antioxidant and prooxidant activities of phenolic compounds. J. Agric. Food Chem. 2000, 48, 3597–3604. 75. Ahmad, M.S.; Fazal, F.; Rahman, A.; Hadi, S.M.; Parish, J.H. Activities of flavonoids for the cleavage of DNA in the presence of Cu(II): Correlation with generation of active oxygen species. Carcinogenesis 1992, 13, 605–608. 76. Tautkus, S.; Irnius, A.; Speiciene, D.; Barkauskas, J.; Kareiva, A. Investigation of distribution of heavy metals between blood plasma and blood cells. Ann. Chim. 2007, 97, 1139–1142. 77. Sandstrom, B.E.; Marklund, S.L. Effects of variation in glutathione peroxidase activity on DNA damage and cell survival in human cells exposed to hydrogen peroxide and t-butyl hydroperoxide. Biochem. J. 1990, 271, 17–23. 78. Ajila, C.M.; Prasada Rao, U.J. Protection against hydrogen peroxide induced oxidative damage in rat erythrocytes by Mangifera indica L. Peel extract. Food Chem. Toxicol. 2008, 46, 303–309. 79. Shui, G.; Bao, Y.M.; Bo, J.; An, L.J. Protective effect of protocatechuic acid from alpinia oxyphylla on hydrogen peroxide-induced oxidative PC12 cell death. Eur. J. Pharmacol. 2006, 538, 73–79. 80. Halliwell, B.; Clement, M.V.; Long, L.H. Hydrogen peroxide in the human body. FEBS Lett. 2000, 486, 10–13. 81. Ohguro, N.; Fukuda, M.; Sasabe, T.; Tano, Y. Concentration dependent effects of hydrogen peroxide on lens epithelial cells. Br. J. Ophthalmol. 1999, 83, 1064. 82. Loo, A.E.; Halliwell, B. Effects of hydrogen peroxide in a keratinocyte-fibroblast co-culture model of wound healing. Biochem. Biophys. Res. Commun. 2012, 423, 253–258. 83. Loo, A.E.; Wong, Y.T.; Ho, R.; Wasser, M.; Du, T.; Ng, W.T.; Halliwell, B. Effects of hydrogen peroxide on wound healing in mice in relation to oxidative damage. PLoS ONE 2012, 7, e49215. 84. Radak, Z.; Chung, H.Y.; Goto, S. Exercise and hormesis: Oxidative stress-related adaptation for successful aging. Biogerontology 2005, 6, 71–75. 85. Pallauf, K.; Giller, K.; Huebbe, P.; Rimbach, G. Nutrition and healthy ageing: Calorie restriction or polyphenol-rich “mediterrasian” diet? Oxid. Med. Cell Longev. 2013, 2013, 707421. 86. Covas, G.; Marinho, H.S.; Cyrne, L.; Antunes, F. Activation of Nrf2 by H2O2: De novo synthesis versus nuclear translocation. Methods Enzymol. 2013, 528, 157–171. 87. Surh, Y.J.; Kundu, J.K.; Na, H.K. Nrf2 as a master redox switch in turning on the cellular signaling involved in the induction of cytoprotective genes by some chemopreventive phytochemicals. Planta Med. 2008, 74, 1526–1539. 88. Hybertson, B.M.; Gao, B.; Bose, S.K.; McCord, J.M. Oxidative stress in health and disease: The therapeutic potential of Nrf2 activation. Mol. Aspects Med. 2011, 32, 234–246. 89. Farombi, E.O.; Shrotriya, S.; Na, H.K.; Kim, S.H.; Surh, Y.J. Curcumin attenuates dimethylnitrosamine- induced liver injury in rats through Nrf2-mediated induction of heme oxygenase-1. Food Chem. Toxicol. 2008, 46, 1279–1287. 90. Yuan, J.H.; Li, Y.Q.; Yang, X.Y. Inhibition of epigallocatechin gallate on orthotopic colon cancer by upregulating the Nrf2-ugt1a signal pathway in nude mice. Pharmacology 2007, 80, 269–278. 91. Ben-Dor, A.; Steiner, M.; Gheber, L.; Danilenko, M.; Dubi, N.; Linnewiel, K.; Zick, A.; Sharoni, Y.; Levy, J. Carotenoids activate the antioxidant response element transcription system. Mol. Cancer Ther. 2005, 4, 177– 186.

126

92. Kode, A.; Rajendrasozhan, S.; Caito, S.; Yang, S.R.; Megson, I.L.; Rahman, I. Resveratrol induces glutathione synthesis by activation of Nrf2 and protects against cigarette smoke-mediated oxidative stress in human lung epithelial cells. Am. J. Physiol. Lung Cell. Mol. Physiol. 2008, 294, L478–L488. 93. Hong, F.; Freeman, M.L.; Liebler, D.C. Identification of sensor cysteines in human keap1 modified by the cancer chemopreventive agent sulforaphane. Chem. Res. Toxicol. 2005, 18, 1917–1926.

© 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons by Attribution (CC-BY) license (http://creativecommons.org/licenses/by/4.0/).

127

Chapter 6: General discussion and conclusion

Chapter 6: General Discussion and Conclusion

Dietary phytochemicals represent an important component of the human diet with significant potential to protect against oxidative stress and inflammation (OSI) and ultimately chronic diseases. However, dietary phytochemicals are handled by the human body as xenobiotics therefore their presence in the body is transient, with low potential bioavailability. The overall aims of this project were to explore the absorption kinetics of phytochemicals and to establish a method to predict the time required for phytochemicals to reach their maximal plasma concentration (Tmax) following human consumption. The main hypothesis of this project was that Tmax of phytochemicals can be predicted based on their physicochemical properties and that matching Tmax of dietary phytochemicals to the onset of OSI can achieve optimal bioefficacy of phytochemicals.

Whilst pharmacokinetics of phytochemicals has been previously studied, these approaches involve costly labour and time intensive clinical studies to provide accurate absorption kinetic information. Further, these clinical studies are only informative to the specific phytochemicals tested. Reported evidence suggests that an absence of reliable absorption kinetic information has compromised the conclusiveness of bioefficacy studies as blood sampling time may not be optimised to Tmax of phytochemicals. In this project, an in silico phytochemical absorption prediction (PCAP) model was developed to allow for direct calculation of Tmax of phytochemicals for human consumption (Chapter

2), allowing for the production of “functional fingerprints” (predicted Tmax profiles) of complex mixtures of phytochemicals in plant extracts (Chapter 3). In Chapter 4 and 5, the absorption kinetics of phytochemicals were investigated in cell culture (in vitro) and in

128

Chapter 6: General discussion and conclusion pigs (in vivo), respectively. Additionally, Chapter 5 explored the modes of action contributing to health benefits of phytochemicals.

6.1 In silico phytochemical absorption prediction (PCAP) model

The in silico PCAP model was successfully developed using the Tmax data of 67 dietary phytochemicals gathered from 31 clinical studies, and validated against three independent datasets containing 108 dietary phytochemicals and 98 pharmaceutical compounds gathered from a further 129 clinical studies. The development of the PCAP model utilised regression modelling with a natural logarithm transformation of Tmax and standard error of Tmax as weights to account for the experimental variation of each data point. Correlation analysis between several physicochemical properties of phytochemicals identified that molecular mass, lipophilicity, polar surface area and dietary intake forms had significant impacts on Tmax of phytochemicals. Accordingly, the PCAP model comprised two predictive mathematical relationships with high statistical power for accurate prediction of Tmax, including either molecular mass and lipophilicity descriptor log P (the log P model), or molecular mass and polar surface area (the PSA model), for three dietary intake forms of liquid, semi-solid and solid.

In the log P model, Tmax was dependent on both log P and molecular mass when phytochemicals were administered in liquid form. However, when administration as semi-solid and solid forms, Tmax was found to be dependent on log P and independent of molecular mass. Absorption of phytochemicals consists of two steps including (1) dissolution and release of phytochemicals into the gastrointestinal tract, and (2) transport of phytochemicals across the intestinal wall. In comparison to the liquid form, phytochemicals in semi-solid and solid forms require longer time for dissolution and

129

Chapter 6: General discussion and conclusion release into the gastrointestinal tract before they are ready to cross the intestinal wall.

Lipophilicity log P may have a larger effect on the dissolution of phytochemicals compared to molecular mass. This may explain why Tmax of phytochemicals was independent of molecular mass when phytochemicals were ingested as semi-solid and solid forms.

Overall, the PCAP model covered a diverse range of phytochemical classes from phenolic compounds to carotenoids, from very hydrophilic (log P of -4.7, polar surface area of

465) to very lipophilic (log P of 9.8, polar surface area of 0) with a broad molecular mass range of 122–1270. The in silico PCAP model is the first model that can directly predict

Tmax following oral intake of dietary phytochemicals and pharmaceutical compounds, in the human body.

6.2 Liquid chromatography mass spectrometry method for the characterisation of phytochemical mixtures

The in silico PCAP model allows calculation of Tmax of known compounds. However, phytochemicals are consumed as plant extracts containing complex mixtures of unknown phytochemicals. Therefore, a liquid chromatography mass spectrometry (LC-MS) method and data processing workflow were developed to simultaneously characterise values of molecular mass and lipophilicity descriptor log P of complex phytochemical mixtures, and thereby calculate values of Tmax using the PCAP model. Values of log P of complex phytochemical mixtures were determined using a calibration curve of retention time and log P of twelve standards. The Tmax values of the phytochemical components were then plotted against the corresponding relative abundance (relative ion count) obtained from the LC-MS outputs to produce the predicted Tmax profile ‘functional

130

Chapter 6: General discussion and conclusion fingerprint’ of the phytochemical mixture. The methods were applied to characterise the functional fingerprints of 12 plant extracts including eight vegetable extracts and four commercial extracts, for 2 dietary intakes of liquid and solid.

The data processing workflow developed in this chapter also permitted the differentiation between primary and secondary metabolites without the need of compound identification.

Primary metabolites are compounds present in all plant species for examples sugars and some amino acids, whilst bioactive phytochemicals are secondary metabolites occurring in limited plant groups. Examples of secondary metabolites include carotenoids, terpenoids and phenolics which are usually produced in plants for environmental defence.

According to the LC-MS analysis of 12 plant extracts tested, primary metabolites tend to be smaller in molecular mass (298 ± 16 amu) and more hydrophilic (log P -0.6 ± 0.2) compared to secondary metabolites (molecular mass 431 ± 5 amu, log P 1.1 ± 0.1).

Consequently, application of the PCAP model revealed that when taken as liquid form, primary metabolites have shorter Tmax (0.8 ± 0.1 h) compared to secondary metabolites

(1.5 ± 0.03 h). By contrast, when taken as solid form, primary metabolites have longer

Tmax (2.5 ± 0.04 h) compared to secondary metabolites (Tmax 2.0 ± 0.02 h). These results highlight that in comparison to fruit and vegetable juice (liquid form), whole fruits and vegetables (solid form) may slow down the absorption of sugar (one of the primary metabolites) and provide better glycaemic control. In support and consistent with previous prospective cohort studies, diets high in whole fruits and vegetables are associated with reducing risks of diabetes, whilst fruit juice is associated with increased risks.

Usefulness of the functional fingerprints was demonstrated for blueberry and green tea, using four published clinical studies reporting their ability to regulate OSI in humans.

131

Chapter 6: General discussion and conclusion

Considering the transient presence of phytochemicals in the human body, it was hypothesised that the timespan of Tmax of phytochemicals may be an important factor affecting their bioefficacy. In all four clinical studies, blueberry and green tea were observed to regulate OSI in humans exclusively at the time matching the associated Tmax obtained from their functional fingerprints. These findings strongly support that bioefficacy of phytochemicals is contingent upon their timespan of Tmax. Accordingly, the functional fingerprint predicting Tmax profiles of plant extract is an important feature to be considered by researchers during experimental design of clinical studies so that time of blood sampling is aligned with timespan of Tmax to maximise effect monitoring.

Functional fingerprinting methods of plant extracts reported in this chapter represent novel tools allowing high-throughput prediction of the absorption kinetics of phytochemical mixtures in humans. These tools provide a research basis for characterisation of plant extracts so as to understand and optimise their health efficacy potential.

6.3 Absorption kinetics of phytochemicals in vitro

Trans-epithelial transport across the Caco-2/HT29-MTX-E12 co-cultured monolayers (in an apical-basolateral direction) was used to characterise passive absorption kinetics of five standards and 17 plant extracts including ten vegetable extracts and seven commercial extracts. The in vitro cell permeability Papp of standards and plant extracts, representing a measure of trans-epithelial transport rate, was determined. The plant extracts were transported across the monolayers in a manner of single species similarly to the pure standards. Considering the chemical heterogeneity of the plant extracts, this behaviour was interpreted as an estimation of the average transport rate of phytochemicals present in the extracts.

132

Chapter 6: General discussion and conclusion

Values of in vivo Tmax of five standards were calculated from the PCAP model and relationships between the in vitro Papp, lipophilicity descriptor log P and predicted in vivo

Tmax were investigated. Consistent with previous cell transport studies, lipophilic compounds (high log P) are transported through the monolayers at a faster rate (high Papp) compared to hydrophilic compounds (low log P). This result indicates that high lipophilicity facilitate diffusion of compounds through the lipid core of the cell monolayers via transcellular diffusion pathway. A linear correlation between the in vitro

Papp and the predicted in vivo Tmax of five standards was observed. Hydrophilic compounds (low log P) were transported at slower rates in vitro (low Papp), however were predicted by the PCAP model to accumulate peak plasma concentration within a faster time frame (short Tmax). The difference between absorption kinetics in vitro and in vivo indicates that hydrophilic compounds were likely transported via water-filled gap junctions between cells (i.e., paracellular diffusion pathway) and these junctions were tighter in the in vitro cell monolayers compared to the in vivo human small intestine.

The in vitro Papp has previously been correlated with in vivo bioavailability measured as the percentage of a compound absorbed after oral intake compared to the intake dose in humans. Accordingly, low in vitro Papp indicates low in vivo bioavailability. This chapter is the first study to report the relationship of absorption kinetics between the in vitro Papp and the predicted in vivo Tmax. The results observed in this chapter indicate that a compound with high in vivo bioavailability will not necessarily accumulate plasma peak concentration within a short time frame and vice versa. Accordingly, whilst in vitro Papp is a reliable indicator of in vivo bioavailability, the PCAP model is required to predict in vivo phytochemical pharmacokinetic Tmax. Therefore, the in silico PCAP model can be

133

Chapter 6: General discussion and conclusion used to compliment the in vitro cell culture to ensure maximal accuracy when predicting human absorption kinetics of phytochemicals.

6.4 Absorption kinetics and modes of action of phytochemicals in vivo

Following from studies in the cell monolayer model, validation of the PCAP model was undertaken in vivo using healthy pigs as an animal model with physiological and anatomical similarities to the digestive conditions of humans. This in vivo study tested absorption kinetics of two extracts (red cabbage and grape skin) at four doses, including placebo control. To account for the differences in phytochemical contents of the selected plant extracts, the doses were standardised for their total phenolic content. Blood sampling times (0.5 h interval for 4 h) were informed by functional fingerprints of the two extracts, predicting the expected time range of 0.4–4 h required for the associated phytochemicals to reach peak plasma concentrations. Absorption biomarkers of phytochemicals were measured using plasma Trolox equivalent antioxidant capacity

(TEAC) and plasma glutathione peroxidase (GPx) activity. Plasma TEAC and plasma

GPx activity represent two important lines of cellular antioxidant host defence against

OSI. Plasma TEAC represents the non-enzymatic antioxidant cellular defence whilst plasma GPx activity is an indicator of the enzymatic antioxidant cellular defence.

Overall, consumption of red cabbage and grape skin extracts increased plasma TEAC and plasma GPx activity in pigs, however no clear Tmax peak was detected. This result indicates that the phytochemicals present in the extracts provide health benefits by exerting direct antioxidant effects as indicated by increased plasma TEAC, and by indirect mechanisms via the induction of the enzymatic antioxidant defence. Plasma TEAC is the measure of the total activities of antioxidant molecules in plasma including vitamin C,

134

Chapter 6: General discussion and conclusion vitamin E, uric acid, glutathione and phytochemicals. Increased plasma TEAC after phytochemical consumption could be due to the additional effects of the low levels of phytochemicals in the plasma. Analysis of plasma concentrations of phytochemicals would be needed to confirm this interpretation.

Increased plasma GPx was observed after increased plasma TEAC suggesting that GPx activity was induced in response to the appearance of the phytochemicals in the plasma.

However, a reduction of plasma GPx activity was observed at 1 h in pigs consuming grape skin extract at 4.44 mg gallic acid equivalent/ kg body weight. This could be due to the enzyme activity in scavenging H2O2. Further analyses of H2O2 concentrations in plasma and activities of other antioxidant enzymes are required to understand this phenomenon.

Considering that GPx is one of the enzymes responsible for the removal of hydrogen peroxide (H2O2) in vivo, induction of plasma GPx by the phytochemicals could be due to the pro-oxidant effects of phytochemicals to produce H2O2. Therefore, the antioxidant and pro-oxidant effects of the selected plant extracts were further investigated in vitro in comparison to gallic acid standard. Direct addition of gallic acid standard, red cabbage and grape skin extracts into pig plasma was performed before analysis of plasma TEAC

(antioxidant effects) and plasma concentration of H2O2 (pro-oxidant effects). Linear responses in both plasma TEAC and plasma concentrations of H2O2 were observed depending on the doses and sources of the phytochemicals. Accordingly, gallic acid exerts the strongest antioxidant and pro-oxidant effects in vitro, followed by grape skin and red cabbage extracts. The ability of the tested plant extracts to produce H2O2 in vitro was highly correlated with the response activity of the antioxidant enzyme GPx in plasma.

135

Chapter 6: General discussion and conclusion

Further, the increased plasma GPx activity in response to phytochemical intakes can be due to activation of the nuclear factor-erythroid-2-related factor 2 (Nrf2), which is a transcription factor responsible for regulation of antioxidant defence. The Nrf2 has been reported to be activated by both H2O2 and phytochemicals including resveratrol (in grape) and sulforaphane (in red cabbage). In combination with results from this PhD study, activation of the Nrf2 by phytochemicals could be related to their ability to produce H2O2.

These results support that dietary phytochemicals utilise both direct and indirect antioxidant mechanisms for regulation of OSI and thereby provide protective health benefits. The indirect antioxidant mechanisms of dietary phytochemicals may be associated to their capacity to produce H2O2 which in turn stimulate cellular antioxidant defence. The research in this chapter demonstrates for the first time, that in vitro H2O2 production capacity could be used as an indicator of in vivo bioefficacy of phytochemicals.

6.5 Conclusion

This research studied the absorption kinetics of dietary phytochemicals using in silico, in vitro and in vivo approaches. The in silico PCAP model was developed and validated in vitro and in vivo, for the prediction of times of peak absorption Tmax of phytochemicals during human consumption.

The in silico PCAP model and in vitro research tools developed in this PhD project represent valuable resources for understanding and optimising the health benefits of plant foods. Whilst dietary phytochemicals have been demonstrated to protect health against both transient and chronic OSI, current ad hoc uses of dietary plants fail to maximise their health potential. This is in part due to the low bioavailability and transient presence of

136

Chapter 6: General discussion and conclusion phytochemicals in the body. Whilst absorption kinetics of phytochemicals have been previously studied, these approaches require labour intensive, time consuming and costly clinical studies which are only informative for the specific phytochemicals tested. This project is the first to present a systematic approach to predict absorption kinetics of these bioactive phytochemicals. The PCAP model is a world-first model of its kind, and was judged as such by the International Examination report for the filed patent. The research has also expanded the applicability of the PCAP model from known phytochemicals to uncharacterised mixture of phytochemicals, the common intake form of phytochemicals, via application of an LC-MS analytical method.

The research has generated unprecedented new knowledge demonstrating the different absorption timespans of phytochemicals. The research lays down the ground work for further understanding the immediate and long-term health benefits of plant foods associated with phytochemicals. The functional fingerprinting of phytochemical extracts can be applied to achieve “bio-matching” of absorption kinetic properties to the onset of

OSI for maximal anti-OSI bioefficacy. The results enable the development of personalised dietary tools based on dietary plants for a broad range of health and well- being endpoints.

6.6 Future research

Research in this project has advanced scientific understanding and substantiated new concepts of how absorption kinetics of dietary phytochemicals can be used to optimise the health benefits of plant foods. In this project, the PCAP model predicting the time required for phytochemicals to reach peak plasma concentration was validated in a pig model, with the presence of phytochemicals in plasma indicated indirectly by plasma

137

Chapter 6: General discussion and conclusion antioxidant capacity. Future research should include a human clinical trial in healthy subjects with analysis of plasma using LC-MS to directly quantify the low levels of phytochemicals in plasma and to further validate predictions of the PCAP model.

The capacity of phytochemicals to produce H2O2 in vitro was highly correlated with the response activity of the antioxidant enzyme GPx in pig plasma ex vivo, indicating the indirect antioxidant mechanisms of phytochemicals via stimulation of cellular antioxidant defence. Therefore, production of H2O2 in vitro by phytochemicals can be used as a simple assay to develop a ‘library’ of dietary phytochemicals with H2O2 production potential as an indicator of their OSI regulating bioefficacy. Further investigation of this mode of action of phytochemicals could be carried on in clinical trials or animal studies with direct measurement of plasma H2O2 concentration post phytochemical consumption.

Additionally, plasma samples obtained from these studies could be analysed for other enzymes of the cellular antioxidant defence such as catalase, glutathione reductase and superoxide dismutase to investigate the effects of dietary phytochemicals on activity of these cellular antioxidant enzymes.

In this study, the Folin-Ciocalteu assay was used to determine the total phenolic contents of plant extracts. Whilst this is a well-established assay, it is interfered by vitamins

(mainly ascorbic acid) and some amino acids. The total protein contents of plant extracts used in this study were low and therefore are less likely to interfere with the Folin-

Ciocalteu assay. However, ascorbic acid content was not tested. Future studies should include analyses of vitamins (especially ascorbic acid) to overcome the limitations of the

Folin-Ciocalteu assay.

138

Chapter 6: General discussion and conclusion

The concept of bio-matching emerging from this project suggests that matching of absorption kinetics Tmax of phytochemicals to the timing of body’s need appears to be essential for optimal protection of dietary phytochemicals against OSI. The timing of body’s need can be associated with onset of the OSI cycles that continuously occur in healthy body and associated with daily activities such as exercise and meal digestion.

Further, advances in genetic profiling has allowed identification of genetic polymorphisms that may predispose individuals to elevated OSI and consequently high disease risks. The research tools presented in this study could be complemented with genetic profiling to allow development of phytochemical-rich foods that meet personalised nutrition in both healthy and high-disease risk individuals.

In this project, the concept of bio-matching was substantiated using published clinical studies demonstrating that the bioefficacy of dietary phytochemicals was observed exclusively within the time frame Tmax of their predicted plasma peak. Future research could include animal studies or human clinical trials in healthy subjects to further explore this concept of matching the Tmax of phytochemicals with the time of OSI induced by exercise or a high-fat meal, to determine optimal time of phytochemical consumption for maximal protection against OSI associated with daily activities. Considering that OSI is associated with most chronic diseases, future research could test the bioefficacy of phytochemicals with a wide range of Tmax on regulating OSI in chronic diseases and ultimately on disease management.

The ability of phytochemicals to provide health benefits as both reactive oxygen species

(ROS) scavengers and ROS producers highlights the dual roles of ROS in aging and longevity. Whilst high levels of ROS are generally accepted to cause cellular damage and to promote aging, low levels of ROS can stimulate adaptive response and improve

139

Chapter 6: General discussion and conclusion systemic defence system. Similar to the effects of phytochemical intakes found in this study, the health benefits of moderate regular exercise and calorie-restriction diets have been associated with low levels of ROS formation. In contrast, maximal high-intensity exercise was reported to induce oxidative damage due to high levels of ROS formation.

Future research could investigate the effects of phytochemical intakes on regular moderate exercise versus maximal high-intensity exercise.

The research tools developed in this project have significant applications for enabling unprecedented value-addition to the phytochemical fraction of plants. The functional fingerprint tools can be applied to develop a ‘library’ of predicted Tmax for a broad range of dietary phytochemicals, providing a foundation for strategically combined plant extracts into blends of targeted absorption behaviours. For example, blends of short Tmax phytochemicals might be useful to minimise OSI associated with exercise, medium Tmax for OSI associated with meal digestion, and combination of short and long Tmax for chronic disease management.

140

Reference References Albertini, J-P, Valensi, P, Lormeau, B, Aurousseau, M-H, Ferrière, F, Attali, J-R & Gattegno, L 1998, 'Elevated Concentrations of Soluble E-Selectin and Vascular Cell Adhesion Molecule-1 in NIDDM: Effect of intensive insulin treatment', Diabetes Care, vol. 21, no. 6, pp. 1008-1013. Alessio, HM, Goldfarb, AH & Cao, G 1997, 'Exercise-induced oxidative stress before and after vitamin C supplementation', International Journal of Sport Nutrition, vol. 7, no. 1, pp. 1-9. Andlauer, W & Fürst, P 2003, 'Absorption and metabolism of phenolic phytochemicals in the isolated rat small intestine with special reference to isoflavones and anthocyanins - a review', Polish Journal of Food and Nutrition Science, vol. 12, pp. 117-127. Andreasen, AS, Krabbe, KS, Krogh-Madsen, R, Taudorf, S, Pedersen, BK & Moller, K 2008, 'Human endotoxemia as a model of systemic inflammation', Current Medicinal Chemistry, vol. 15, no. 17, pp. 1697-1705. Appay, V & Rowland-Jones, SL 2001, 'RANTES: a versatile and controversial chemokine', Trends in Immunology, vol. 22, no. 2, pp. 83-87. Apredica 2013, Pharmacokinetics services, viewed 12/12/2013, . Ashton, T, Young, IS, Peters, JR, Jones, E, Jackson, SK, Davies, B & Rowlands, CC 1999, 'Electron spin resonance spectroscopy, exercise, and oxidative stress: an ascorbic acid intervention study', Journal of Applied Physiology vol. 87, no. 6, pp. 2032-2036. Atteritano, M, Marini, H, Minutoli, L, Polito, F, Bitto, A, Altavilla, D, Mazzaferro, S, D'Anna, R, Cannata, ML, Gaudio, A, Frisina, A, Frisina, N, Corrado, F, Cancellieri, F, Lubrano, C, Bonaiuto, M, Adamo, EB & Squadrito, F 2007, 'Effects of the phytoestrogen genistein on some predictors of cardiovascular risk in osteopenic, postmenopausal women: A two-year randomized, double-blind, placebo-controlled study', The Journal of Clinical Endocrinology Metabolism, vol. 92, no. 8, pp. 3068-3075. Azzini, E, Polito, A, Fumagalli, A, Intorre, F, Venneria, E, Durazzo, A, Zaccaria, M, Ciarapica, D, Foddai, M, Mauro, B, Raguzzini, A, Palomba, L & Maiani, G 2011, 'Mediterranean Diet Effect: an Italian picture', Nutrition Journal, vol. 10, no. 1, p. 125. Banecka-Majkutewicz, Z, Sawula, W, Kadzinski, L, Wegrzyn, A & Banecki, B 2012, 'Homocysteine, heat shock proteins, genistein and vitamins in ischemic stroke-- pathogenic and therapeutic implications', Acta Biochimica Polonica, vol. 59, no. 4, pp. 495-499. Barbaresko, J, Koch, M, Schulze, MB & Nöthlings, U 2013, 'Dietary pattern analysis and biomarkers of low-grade inflammation: a systematic literature review', Nutrition Reviews, vol. 71, no. 8, pp. 511-527. Beavers, KM, Serra, MC, Beavers, DP, Cooke, MB & Willoughby, DS 2009, 'Soymilk supplementation does not alter plasma markers of inflammation and oxidative stress in postmenopausal women', Nutrition Research, vol. 29, no. 9, pp. 616-622. Bennett, LE, Miller, RB & Selby-Pham, SNB 2016, Methods for formulating dietary foodstuffs, Australian Provisional Patent Application no. 2014903013 and International Patent Application no. PCT/AU2015/000461.

141

Reference

Bhupathiraju, SN & Tucker, KL 2011, 'Greater variety in fruit and vegetable intake is associated with lower inflammation in Puerto Rican adults', The American Journal of Clinical Nutrition, vol. 93, no. 1, pp. 37-46. Blanco-Colio, LM, Munoz-Garcia, B, Martin-Ventura, JL, Alvarez-Sala, LA, Castilla, M, Bustamante, A, Lamuela-Raventos, RM, Gomez-Gerique, J, Fernandez-Cruz, A, Millan, J & Egido, J 2007, 'Ethanol beverages containing polyphenols decrease nuclear factor kappa-B activation in mononuclear cells and circulating MCP-1 concentrations in healthy volunteers during a fat-enriched diet', Atherosclerosis, vol. 192, no. 2, pp. 335-341. Block, G, Jensen, C, Dietrich, M, Norkus, EP, Hudes, M & Packer, L 2004, 'Plasma C- reactive protein concentrations in active and passive smokers: influence of antioxidant supplementation', Journal of The American College of Nutrition, vol. 23, no. 2, pp. 141- 147. Bogani, P, Galli, C, Villa, M & Visioli, F 2007, 'Postprandial anti-inflammatory and antioxidant effects of extra virgin olive oil', Atherosclerosis, vol. 190, no. 1, pp. 181-186. Borgstrom, L, Nyberg, L, Jonsson, S, Lindberg, C & Paulson, J 1989, 'Pharmacokinetic evaluation in man of terbutaline given as separate enantiomers and as the racemate', British Journal of Clinical Pharmacology, vol. 27, no. 1, pp. 49-56. Bruunsgaard, H, Andersen-Ranberg, K, Jeune, B, Pedersen, AN, Skinhoj, P & Pedersen, BK 1999, 'A high plasma concentration of TNF-alpha is associated with dementia in centenarians', The Journals of Gerontology. Series A, Biological Sciences and Medical Sciences, vol. 54, no. 7, pp. M357-364. Bruunsgaard, H, Poulsen, HE, Pedersen, BK, Nyyssonen, K, Kaikkonen, J & Salonen, JT 2003, 'Long-term combined supplementations with alpha-tocopherol and vitamin C have no detectable anti-inflammatory effects in healthy men', The Journal of Nutrition, vol. 133, no. 4, pp. 1170-1173. Burton-Freeman, B 2010, 'Postprandial metabolic events and fruit-derived phenolics: a review of the science', British Journal of Nutrition, vol. 104 Suppl 3, pp. S1-S14. Burton-Freeman, B, Talbot, J, Park, E, Krishnankutty, S & Edirisinghe, I 2012, 'Protective activity of processed tomato products on postprandial oxidation and inflammation: a clinical trial in healthy weight men and women', Molecular Nutrition & Food Research, vol. 56, no. 4, pp. 622-631. Calder, PC, Albers, R, Antoine, JM, Blum, S, Bourdet-Sicard, R, Ferns, GA, Folkerts, G, Friedmann, PS, Frost, GS, Guarner, F, Lovik, M, Macfarlane, S, Meyer, PD, M'Rabet, L, Serafini, M, van Eden, W, van Loo, J, Vas Dias, W, Vidry, S, Winklhofer-Roob, BM & Zhao, J 2009, 'Inflammatory disease processes and interactions with nutrition', British Journal of Nutrition, vol. 101 Suppl 1, pp. S1-S45. Carter, SJ, Roberts, MB, Salter, J & Eaton, CB 2010, 'Relationship between Mediterranean Diet Score and atherothrombotic risk: findings from the Third National Health and Nutrition Examination Survey (NHANES III), 1988-1994', Atherosclerosis, vol. 210, no. 2, pp. 630-636. Cassidy, A, Skidmore, P, Rimm, EB, Welch, A, Fairweather-Tait, S, Skinner, J, Burling, K, Richards, JB, Spector, TD & MacGregor, AJ 2009, 'Plasma adiponectin concentrations are associated with body composition and plant-based dietary factors in female twins', The Journal of Nutrition, vol. 139, no. 2, pp. 353-358.

142

Reference

Centritto, F, Iacoviello, L, di Giuseppe, R, De Curtis, A, Costanzo, S, Zito, F, Grioni, S, Sieri, S, Donati, MB, de Gaetano, G & Di Castelnuovo, A 2009, 'Dietary patterns, cardiovascular risk factors and C-reactive protein in a healthy Italian population', Nutrition, Metabolism, and Cardiovascular Diseases, vol. 19, no. 10, pp. 697-706. Charles, C, Yuskavage, J, Carlson, O, John, M, Tagalicud, AS, Maggio, M, Muller, DC, Egan, J & Basaria, S 2009, 'Effects of high-dose isoflavones on metabolic and inflammatory markers in healthy postmenopausal women', Menopause, vol. 16, no. 2, pp. 395-400. Cheng, KC, Li, C & Uss, AS 2008, 'Prediction of oral drug absorption in humans – from cultured cell lines and experimental animals', Expert Opinion on Drug Metabolism & Toxicology, vol. 4, no. 5, pp. 581-590. Chrysohoou, C, Panagiotakos, DB, Pitsavos, C, Das, UN & Stefanadis, C 2004, 'Adherence to the Mediterranean diet attenuates inflammation and coagulation process in healthy adults: The ATTICA Study', Journal of the American College of Cardiology, vol. 44, no. 1, pp. 152-158. Chun, OK, Chung, SJ, Claycombe, KJ & Song, WO 2008, 'Serum C-reactive protein concentrations are inversely associated with dietary flavonoid intake in U.S. adults', The Journal of Nutrition, vol. 138, no. 4, pp. 753-760. Crozier, A, Jaganath, IB & Clifford, MN 2007, 'Phenols, polyphenols and tannins: an overview', in A Crozier, MN Clifford & H Ashihara (eds), Plant secondary metabolites: Occurrence, structure and role in the human diet, Blackwell Publishing Ltd, Oxford, pp. 1-24. Crozier, A, Jaganath, IB & Clifford, MN 2009, 'Dietary phenolics: chemistry, bioavailability and effects on health', Natural Product Reports, vol. 26, no. 8, pp. 1001- 1043. D'Archivio, M, Filesi, C, Vari, R, Scazzocchio, B & Masella, R 2010, 'Bioavailability of the polyphenols: status and controversies', International Journal of Molecular Sciences, vol. 11, no. 4, pp. 1321-1342. D'Onofrio, G, Sancarlo, D, Ruan, Q, Yu, Z, Panza, F, Daniele, A, Greco, A & Seripa, D 2016, 'Phytochemicals in the treatment of Alzheimer's Disease: a systematic review', Current Drug Targets, vol. 17, no. 16. D’Archivio, M, Filesi, C, Varì, R, Scazzocchio, B & Masella, R 2010, 'Bioavailability of the polyphenols: status and controversies', International Journal of Molecular Sciences, vol. 11, no. 4, pp. 1321-1342. Dai, J, Miller, AH, Bremner, JD, Goldberg, J, Jones, L, Shallenberger, L, Buckham, R, Murrah, NV, Veledar, E, Wilson, PW & Vaccarino, V 2008, 'Adherence to the mediterranean diet is inversely associated with circulating interleukin-6 among middle- aged men: a twin study', Circulation, vol. 117, no. 2, pp. 169-175. Dauchet, L, Amouyel, P & Dallongeville, J 2009, 'Fruits, vegetables and coronary heart disease', Nature Reviews Cardiology, vol. 6, no. 9, pp. 599-608. De Bacquer, D, Clays, E, Delanghe, J & De Backer, G 2006, 'Epidemiological evidence for an association between habitual tea consumption and markers of chronic inflammation', Atherosclerosis, vol. 189, no. 2, pp. 428-435.

143

Reference

Deibert, P, Solleder, F, Konig, D, Vitolins, MZ, Dickhuth, HH, Gollhofer, A & Berg, A 2011, 'Soy protein based supplementation supports metabolic effects of resistance training in previously untrained middle aged males', Aging Male, vol. 14, no. 4, pp. 273- 279. Dell'Agli, M, Di Lorenzo, C, Badea, M, Sangiovanni, E, Dima, L, Bosisio, E & Restani, P 2013, 'Plant food supplements with anti-inflammatory properties: a systematic review (I)', Critical Reviews in Food Science and Nutrition, vol. 53, no. 4, pp. 403-413. Deshmane, SL, Kremlev, S, Amini, S & Sawaya, BE 2009, 'Monocyte chemoattractant protein-1 (MCP-1): an overview', Journal of Interferon & Cytokine Research, vol. 29, no. 6, pp. 313-326. Di Bona, D, Rizzo, C, Bonaventura, G, Candore, G & Caruso, C 2012, 'Association between interleukin-10 polymorphisms and Alzheimer's disease: a systematic review and meta-analysis', Journal of Alzheimer's Disease, vol. 29, no. 4, pp. 751-759. Diez, JJ & Iglesias, P 2003, 'The role of the novel adipocyte-derived hormone adiponectin in human disease', European Journal of Endocrinology, vol. 148, no. 3, pp. 293-300. Donnini, S, Finetti, F, Lusini, L, Morbidelli, L, Cheynier, V, Barron, D, Williamson, G, Waltenberger, J & Ziche, M 2006, 'Divergent effects of quercetin conjugates on angiogenesis', British Journal of Nutrition vol. 95, no. 5, pp. 1016-1023. Donovan, JL, Manach, C, Faulks, RM & Kroon, PA 2007, 'Absorption and metabolism of dietary plant secondary metabolites', in A Crozier, MN Clifford & H Ashihara (eds), Plant secondary metabolites: Occurrence, structure and role in the human diet, Blackwell Publishing, Oxford, pp. 303-351. Egert, S, Wolffram, S, Bosy-Westphal, A, Boesch-Saadatmandi, C, Wagner, AE, Frank, J, Rimbach, G & Mueller, MJ 2008, 'Daily quercetin supplementation dose-dependently increases plasma quercetin concentrations in healthy humans', Journal of Nutrition, vol. 138, no. 9, pp. 1615-1621. eMaggio, M, Guralnik, JM, Longo, DL & Ferrucci, L 2006, 'Interleukin-6 in aging and chronic disease: a magnificent pathway', The Journals of Gerontology. Series A, Biological Sciences and Medical Sciences, vol. 61, no. 6, pp. 575-584. Esmaillzadeh, A & Azadbakht, L 2008, 'Home use of vegetable oils, markers of systemic inflammation, and endothelial dysfunction among women', The American Journal of Clinical Nutrition, vol. 88, no. 4, pp. 913-921. Esmaillzadeh, A, Kimiagar, M, Mehrabi, Y, Azadbakht, L, Hu, FB & Willett, WC 2006, 'Fruit and vegetable intakes, C-reactive protein, and the metabolic syndrome', The American Journal of Clinical Nutrition, vol. 84, no. 6, pp. 1489-1497. Esmaillzadeh, A, Kimiagar, M, Mehrabi, Y, Azadbakht, L, Hu, FB & Willett, WC 2007, 'Dietary patterns and markers of systemic inflammation among Iranian women', The Journal of Nutrition, vol. 137, no. 4, pp. 992-998. Esposito, K, Nappo, F, Marfella, R, Giugliano, G, Giugliano, F, Ciotola, M, Quagliaro, L, Ceriello, A & Giugliano, D 2002, 'Inflammatory cytokine concentrations are acutely increased by hyperglycemia in humans: role of oxidative stress', Circulation, vol. 106, no. 16, pp. 2067-2072.

144

Reference

Estruch, R, Sacanella, E, Badia, E, Antunez, E, Nicolas, JM, Fernandez-Sola, J, Rotilio, D, de Gaetano, G, Rubin, E & Urbano-Marquez, A 2004, 'Different effects of red wine and gin consumption on inflammatory biomarkers of atherosclerosis: a prospective randomized crossover trial - Effects of wine on inflammatory markers', Atherosclerosis, vol. 175, no. 1, pp. 117-123. Eyles, JL, Roberts, AW, Metcalf, D & Wicks, IP 2006, 'Granulocyte colony-stimulating factor and neutrophils[mdash]forgotten mediators of inflammatory disease', Nature Clinical Practice Rheumatology, vol. 2, no. 9, pp. 500-510. Fargnoli, JL, Fung, TT, Olenczuk, DM, Chamberland, JP, Hu, FB & Mantzoros, CS 2008, 'Adherence to healthy eating patterns is associated with higher circulating total and high- molecular-weight adiponectin and lower resistin concentrations in women from the Nurses' Health Study', The American Journal of Clinical Nutrition, vol. 88, no. 5, pp. 1213-1224. Fielding, JM, Rowley, KG, Cooper, P & O' Dea, K 2005, 'Increases in plasma lycopene concentration after consumption of tomatoes cooked with olive oil', Asia Pacific Journal of Clinical Nutrition, vol. 14, no. 2, pp. 131-136. Filiberto, AC, Mumford, SL, Pollack, AZ, Zhang, CL, Yeung, EH, Perkins, NJ, Wactawski-Wende, J & Schisterman, EF 2013, 'Habitual Dietary Isoflavone Intake Is Associated with Decreased C-Reactive Protein Concentrations among Healthy Premenopausal Women', The Journal of Nutrition, vol. 143, no. 6, pp. 900-906. Ford, ES, Mokdad, AH & Liu, S 2005, 'Healthy Eating Index and C-reactive protein concentration: findings from the National Health and Nutrition Examination Survey III, 1988-1994', European Journal of Clinical Nutrition, vol. 59, no. 2, pp. 278-283. Forester, SC & Waterhouse, AL 2010, 'Gut metabolites of anthocyanins, gallic acid, 3- O-methylgallic acid, and 2,4,6-trihydroxybenzaldehyde, inhibit cell proliferation of Caco-2 cells', Journal of Agricultural and Food Chemistry, vol. 58, no. 9, pp. 5320-5327. Fragopoulou, E, Panagiotakos, DB, Pitsavos, C, Tampourlou, M, Chrysohoou, C, Nomikos, T, Antonopoulou, S & Stefanadis, C 2010, 'The association between adherence to the Mediterranean diet and adiponectin levels among healthy adults: the ATTICA study', The Journal of Nutritional Biochemistry, vol. 21, no. 4, pp. 285-289. Fuentes, F, Lopez-Miranda, J, Perez-Martinez, P, Jimenez, Y, Marin, C, Gomez, P, Fernandez, JM, Caballero, J, Delgado-Lista, J & Perez-Jimenez, F 2008, 'Chronic effects of a high-fat diet enriched with virgin olive oil and a low-fat diet enriched with alpha- linolenic acid on postprandial endothelial function in healthy men', The British Journal of Nutrition, vol. 100, no. 1, pp. 159-165. Fung, TT, McCullough, ML, Newby, P, Manson, JE, Meigs, JB, Rifai, N, Willett, WC & Hu, FB 2005, 'Diet-quality scores and plasma concentrations of markers of inflammation and endothelial dysfunction', The American Journal of Clinical Nutrition, vol. 82, no. 1, pp. 163-173. Gao, X, Bermudez, OI & Tucker, KL 2004, 'Plasma C-Reactive Protein and Homocysteine Concentrations Are Related to Frequent Fruit and Vegetable Intake in Hispanic and Non-Hispanic White Elders', The Journal of Nutrition, vol. 134, no. 4, pp. 913-918.

145

Reference

Ghanim, H, Sia, CL, Abuaysheh, S, Korzeniewski, K, Patnaik, P, Marumganti, A, Chaudhuri, A & Dandona, P 2010, 'An Antiinflammatory and Reactive Oxygen Species Suppressive Effects of an Extract of Polygonum Cuspidatum Containing Resveratrol', Journal of Clinical Endocrinology and Metabolism, vol. 95, no. 9, pp. E1-E8. Ghanim, H, Sia, CL, Korzeniewski, K, Lohano, T, Abuaysheh, S, Marumganti, A, Chaudhuri, A & Dandona, P 2011, 'A resveratrol and polyphenol preparation suppresses oxidative and inflammatory stress response to a high-fat, high-carbohydrate meal', Journal of Clinical Endocrinology and Metabolism, vol. 96, no. 5, pp. 1409-1414. Gomez-Rico, A, Salvador, MD, La Greca, M & Fregapane, G 2006, 'Phenolic and volatile compounds of extra virgin olive oil (Olea europaea L. Cv. Cornicabra) with regard to fruit ripening and irrigation management', Journal of Agricultural and Food Chemistry, vol. 54, no. 19, pp. 7130-7136. Gonzalez-Barrio, R, Edwards, CA & Crozier, A 2011, 'Colonic catabolism of ellagitannins, ellagic acid, and raspberry anthocyanins: in vivo and in vitro studies', Drug metabolism and disposition: the biological fate of chemicals, vol. 39, no. 9, pp. 1680- 1688. Greany, KA, Nettleton, JA, Wangen, KE, Thomas, W & Kurzer, MS 2008, 'Consumption of isoflavone-rich soy protein does not alter homocysteine or markers of inflammation in postmenopausal women', European Journal of Clinical Nutrition, vol. 62, no. 12, pp. 1419-1425. Gustin, DM, Rodvold, KA, Sosman, JA, Diwadkar-Navsariwala, V, Stacewicz- Sapuntzakis, M, Viana, M, Crowell, JA, Murray, J, Tiller, P & Bowen, PE 2004, 'Single- dose pharmacokinetic study of lycopene delivered in a well-defined food-based lycopene delivery system (tomato paste-oil mixture) in healthy adult male subjects', Cancer Epidemiology, Biomarkers and Prevention, vol. 13, no. 5, pp. 850-860. Hall, WL, Vafeiadou, K, Hallund, J, Bugel, S, Koebnick, C, Reimann, M, Ferrari, M, Branca, F, Talbot, D, Dadd, T, Nilsson, M, Dahlman-Wright, K, Gustafsson, JA, Minihane, AM & Williams, CM 2005, 'Soy-isoflavone-enriched foods and inflammatory biomarkers of cardiovascular disease risk in postmenopausal women: interactions with genotype and equol production', American Journal of Clinical Nutrition, vol. 82, no. 6, pp. 1260-1268. Heinz, SA, Henson, DA, Nieman, DC, Austin, MD & Jin, FX 2010, 'A 12-week supplementation with quercetin does not affect natural killer cell activity, granulocyte oxidative burst activity or granulocyte phagocytosis in female human subjects', British Journal of Nutrition, vol. 104, no. 6, pp. 849-857. Herder, C, Haastert, B, Muller-Scholze, S, Koenig, W, Thorand, B, Holle, R, Wichmann, HE, Scherbaum, WA, Martin, S & Kolb, H 2005, 'Association of systemic chemokine concentrations with impaired glucose tolerance and type 2 diabetes: results from the Cooperative Health Research in the Region of Augsburg Survey S4 (KORA S4)', Diabetes, vol. 54 Suppl 2, pp. S11-17. Hodgson, JM, Puddey, IB, Mori, TA, Burke, V, Baker, RI & Beilin, LJ 2001, 'Effects of regular ingestion of black tea on haemostasis and cell adhesion molecules in humans', European Journal of Clinical Nutrition, vol. 55, no. 10, pp. 881-886.

146

Reference

Holst, B & Williamson, G 2008, 'Nutrients and phytochemicals: from bioavailability to bioefficacy beyond antioxidants', Current Opinion in Biotechnology, vol. 19, no. 2, pp. 73-82. Holt, EM, Steffen, LM, Moran, A, Basu, S, Steinberger, J, Ross, JA, Hong, C-P & Sinaiko, AR 2009, 'Fruit and Vegetable Consumption and Its Relation to Markers of Inflammation and Oxidative Stress in Adolescents', Journal of The American Dietetic Association, vol. 109, no. 3, pp. 414-421. Housa, D, Housova, J, Vernerova, Z & Haluzik, M 2006, 'Adipocytokines and cancer', Physiological Research, vol. 55, no. 3, pp. 233-244. Huang, Y, Park, E, Edirisinghe, I & Burton-Freeman, BM 2016, 'Maximizing the health effects of strawberry anthocyanins: understanding the influence of the consumption timing variable', Food and Function, vol. 7, pp. 4745-4752. Jacob, K, Periago, MJ, Böhm, V & Berruezo, GR 2008, 'Influence of lycopene and vitamin C from tomato juice on biomarkers of oxidative stress and inflammation', British Journal of Nutrition, vol. 99, no. 01, pp. 137-146. Jaganath, IB, Mullen, W, Edwards, CA & Crozier, A 2006, 'The relative contribution of the small and large intestine to the absorption and metabolism of rutin in man', Free Radical Research, vol. 40, no. 10, pp. 1035-1046. Jimenez-Gomez, Y, Lopez-Miranda, J, Blanco-Colio, LM, Marin, C, Perez-Martinez, P, Ruano, J, Paniagua, JA, Rodriguez, F, Egido, J & Perez-Jimenez, F 2009, 'Olive oil and walnut breakfasts reduce the postprandial inflammatory response in mononuclear cells compared with a butter breakfast in healthy men', Atherosclerosis, vol. 204, no. 2, pp. e70-76. Jin, Y, Cui, X, Singh, UP, Chumanevich, AA, Harmon, B, Cavicchia, P, Hofseth, AB, Kotakadi, V, Stroud, B, Volate, SR, Hurley, TG, Hebert, JR & Hofseth, LJ 2010, 'Systemic inflammatory load in humans is suppressed by consumption of two formulations of dried, encapsulated juice concentrate', Molecular Nutritrion & Food Research, vol. 54, no. 10, pp. 1506-1514. Kant, AK & Graubard, BI 2005, 'A comparison of three dietary pattern indexes for predicting biomarkers of diet and disease', Journal of The American College of Nutrition, vol. 24, no. 4, pp. 294-303. Kantor, ED, Lampe, JW, Vaughan, TL, Peters, U, Rehm, CD & White, E 2012, 'Association between use of specialty dietary supplements and C-reactive protein concentrations', American Journal of Epidemiology, vol. 176, no. 11, pp. 1002-1013. Karlsen, A, Retterstol, L, Laake, P, Paur, I, Bohn, SK, Sandvik, L & Blomhoff, R 2007, 'Anthocyanins inhibit nuclear factor-kappaB activation in monocytes and reduce plasma concentrations of pro-inflammatory mediators in healthy adults', The Journal of Nutrition, vol. 137, no. 8, pp. 1951-1954. Kaspar, KL, Park, JS, Brown, CR, Mathison, BD, Navarre, DA & Chew, BP 2011, 'Pigmented Potato Consumption Alters Oxidative Stress and Inflammatory Damage in Men', The Journal of Nutrition, vol. 141, no. 1, pp. 108-111. Kawashima, A, Madarame, T, Koike, H, Komatsu, Y & Wise, JA 2007, 'Four week supplementation with mixed fruit and vegetable juice concentrates increased protective

147

Reference serum antioxidants and folate and decreased plasma homocysteine in Japanese subjects', Asia Pacific Journal of Clinical Nutrition, vol. 16, no. 3, pp. 411-421. Kay, CD, Hooper, L, Kroon, PA, Rimm, EB & Cassidy, A 2012, 'Relative impact of flavonoid composition, dose and structure on vascular function: a systematic review of randomised controlled trials of flavonoid-rich food products', Molecular Nutrition and Food Research, vol. 56, no. 11, pp. 1605-1616. Key, TJ 2011, 'Fruit and vegetables and cancer risk', British Journal of Cancer vol. 104, no. 1, pp. 6-11. Khojasteh, S, Wong, H & Hop, CCA 2011, 'Pharmacokinetics', in Drug Metabolism and Pharmacokinetics Quick Guide, Springer New York, pp. 1-15, DOI 10.1007/978-1-4419- 5629-3_1, . Knab, AM, Shanely, RA, Henson, DA, Jin, FX, Heinz, SA, Austin, MD & Nieman, DC 2011, 'Influence of Quercetin Supplementation on Disease Risk Factors in Community- Dwelling Adults', Journal of The American Dietetic Association, vol. 111, no. 4, pp. 542- 549. Koga, T & Meydani, M 2001, 'Effect of plasma metabolites of (+)-catechin and quercetin on monocyte adhesion to human aortic endothelial cells', American Journal of Clinical Nutrition, vol. 73, no. 5, pp. 941-948. Konstantinidou, V, Covas, MI, Munoz-Aguayo, D, Khymenets, O, de la Torre, R, Saez, G, Tormos Mdel, C, Toledo, E, Marti, A, Ruiz-Gutierrez, V, Ruiz Mendez, MV & Fito, M 2010, 'In vivo nutrigenomic effects of virgin olive oil polyphenols within the frame of the Mediterranean diet: a randomized controlled trial', FASEB Journal vol. 24, no. 7, pp. 2546-2557. Kornman, KS 2006, 'Interleukin 1 genetics, inflammatory mechanisms, and nutrigenetic opportunities to modulate diseases of aging', The American Journal of Clinical Nutrition, vol. 83, no. 2, pp. 475s-483s. Krishnaiah, D, Sarbatly, R & Nithyanandam, R 2011, 'A review of the antioxidant potential of medicinal plant species', Food and Bioproducts Processing, vol. 89, no. 3, pp. 217-233. Lamprecht, M, Oettl, K, Schwaberger, G, Hofmann, P & Greilberger, JF 2007, 'Several indicators of oxidative stress, immunity, and illness improved in trained men consuming an encapsulated juice powder concentrate for 28 weeks', The Journal of Nutrition, vol. 137, no. 12, pp. 2737-2741. Lee, J, Koo, N & Min, DB 2004, 'Reactive oxygen species, aging, and antioxidative nutraceuticals', Comprehensive Reviews in Food Science and Food Safety, vol. 3, no. 1, pp. 21-33. Liu, RH 2004, 'Potential Synergy of Phytochemicals in Cancer Prevention: Mechanism of Action', Journal of Nutrition vol. 134, no. 12, pp. 3479S-3485S. Lopez-Garcia, E, Schulze, MB, Fung, TT, Meigs, JB, Rifai, N, Manson, JE & Hu, FB 2004, 'Major dietary patterns are related to plasma concentrations of markers of inflammation and endothelial dysfunction', The American Journal of Clinical Nutrition, vol. 80, no. 4, pp. 1029-1035.

148

Reference

Lopez-Garcia, E, van Dam, RM, Qi, L & Hu, FB 2006, 'Coffee consumption and markers of inflammation and endothelial dysfunction in healthy and diabetic women', The American Journal of Clinical Nutrition, vol. 84, no. 4, pp. 888-893. Lowe, GD, Rumley, A & Mackie, IJ 2004, 'Plasma fibrinogen', Annals of Clinical Biochemistry, vol. 41, no. Pt 6, pp. 430-440. Maki, T, Pham, NM, Yoshida, D, Yin, G, Ohnaka, K, Takayanagi, R & Kono, S 2010, 'The relationship of coffee and green tea consumption with high-sensitivity C-reactive protein in Japanese men and women', Clinical Chemistry and Laboratory Medicine, vol. 48, no. 6, pp. 849-854. Manach, C, Scalbert, A, Morand, C, Remesy, C & Jimenez, L 2004, 'Polyphenols: food sources and bioavailability', The American Journal of Clinical Nutrition, vol. 79, no. 5, pp. 727-747. Maurer, M & von Stebut, E 2004, 'Macrophage inflammatory protein-1', The International Journal of Biochemistry & Cell Biology, vol. 36, no. 10, pp. 1882-1886. McAnulty, LS, Miller, LE, Hosick, PA, Utter, AC, Quindry, JC & McAnulty, SR 2013, 'Effect of resveratrol and quercetin supplementation on redox status and inflammation after exercise', Applied Physiology, Nutrition and Metabolism vol. 38, no. 7, pp. 760-765. Mehta, R, Murillo, G, Naithani, R & Peng, X 2010, 'Cancer Chemoprevention by Natural Products: How Far Have We Come?', Pharmaceutical Research, vol. 27, no. 6, pp. 950- 961. Miglio, C, Peluso, I, Raguzzini, A, Villano, DV, Cesqui, E, Catasta, G, Toti, E & Serafini, M 2013, 'Fruit juice drinks prevent endogenous antioxidant response to high-fat meal ingestion', British Journal of Nutrition, pp. 1-7. Mikkila, V, Rasanen, L, Raitakari, OT, Marniemi, J, Pietinen, P, Ronnemaa, T & Viikari, J 2007, 'Major dietary patterns and cardiovascular risk factors from childhood to adulthood. The Cardiovascular Risk in Young Finns Study', The British Journal of Nutrition, vol. 98, no. 1, pp. 218-225. Mooradian, A, Reed, R & Scuderi, P 1991, 'Serum levels of tumor necrosis factor alpha, interleukin-1 alpha and beta in healthy elderly subjects', Age, vol. 14, no. 3, pp. 61-64. Mullen, W, Edwards, CA & Crozier, A 2006, 'Absorption, excretion and metabolite profiling of methyl-, glucuronyl-, glucosyl-and sulpho-conjugates of quercetin in human plasma and urine after ingestion of onions', British Journal of Nutrition, vol. 96, no. 1, pp. 107-116. Nanri, A, Yoshida, D, Yamaji, T, Mizoue, T, Takayanagi, R & Kono, S 2008, 'Dietary patterns and C-reactive protein in Japanese men and women', The American Journal of Clinical Nutrition, vol. 87, no. 5, pp. 1488-1496. Nantz, MP, Rowe, CA, Bukowski, JF & Percival, SS 2009, 'Standardized capsule of Camellia sinensis lowers cardiovascular risk factors in a randomized, double-blind, placebo-controlled study', Nutrition, vol. 25, no. 2, pp. 147-154. Nettleton, JA, Matijevic, N, Follis, JL, Folsom, AR & Boerwinkle, E 2010, 'Associations between dietary patterns and flow cytometry-measured biomarkers of inflammation and cellular activation in the Atherosclerosis Risk in Communities (ARIC) Carotid Artery MRI Study', Atherosclerosis, vol. 212, no. 1, pp. 260-267.

149

Reference

Nettleton, JA, Schulze, MB, Jiang, R, Jenny, NS, Burke, GL & Jacobs, DR, Jr. 2008, 'A priori-defined dietary patterns and markers of cardiovascular disease risk in the Multi- Ethnic Study of Atherosclerosis (MESA)', The American Journal of Clinical Nutrition, vol. 88, no. 1, pp. 185-194. Nettleton, JA, Steffen, LM, Mayer-Davis, EJ, Jenny, NS, Jiang, R, Herrington, DM & Jacobs, DR 2006, 'Dietary patterns are associated with biochemical markers of inflammation and endothelial activation in the Multi-Ethnic Study of Atherosclerosis (MESA)', The American Journal of Clinical Nutrition, vol. 83, no. 6, pp. 1369-1379. NHMRC 2005, Food for health, dietary guidelines for Australians: A guide to healthy eating, National Health and Medical Research Council. Nicastro, HL, Mondul, AM, Rohrmann, S & Platz, EA 2013, 'Associations between urinary soy isoflavonoids and two inflammatory markers in adults in the United States in 2005-2008', Cancer Causes Control, vol. 24, no. 6, pp. 1185-1196. Nieman, DC, Henson, DA, Maxwell, KR, Williams, AS, McAnulty, SR, Jin, F, Shanely, RA & Lines, TC 2009, 'Effects of quercetin and EGCG on mitochondrial biogenesis and immunity', Medicine and Science in Sports and Exercise, vol. 41, no. 7, pp. 1467-1475. Opal, SM, Wherry, JC & Grint, P 1998, 'Interleukin-10: potential benefits and possible risks in clinical infectious diseases', Clinincal Infectious Diseases, vol. 27, no. 6, pp. 1497-1507. Pacheco, YM, Bemudez, B, Lopez, S, Abia, R, Villar, J & Muriana, FJ 2007, 'Minor compounds of olive oil have postprandial anti-inflammatory effects', The British Journal of Nutrtion, vol. 98, no. 2, pp. 260-263. Pacheco, YM, López, S, Bermúdez, B, Abia, R, Villar, J & Muriana, FJG 2008, 'A meal rich in oleic acid beneficially modulates postprandial sICAM-1 and sVCAM-1 in normotensive and hypertensive hypertriglyceridemic subjects', The Journal of Nutritional Biochemistry, vol. 19, no. 3, pp. 200-205. Padayachee, A, Netzel, G, Netzel, M, Day, L, Mikkelsen, D & Gidley, MJ 2013, 'Lack of release of bound anthocyanins and phenolic acids from carrot plant cell walls and model composites during simulated gastric and small intestinal digestion', Food and Function, vol. 4, no. 6, pp. 906-916. Panagiotakos, DB, Pitsavos, C & Stefanadis, C 2006, 'Dietary patterns: a Mediterranean diet score and its relation to clinical and biological markers of cardiovascular disease risk', Nutrition, Metabolism and Cardiovascular Diseases, vol. 16, no. 8, pp. 559-568. Panunzio, MF, Pisano, A, Antoniciello, A, Di Martino, V, Frisoli, L, Cipriani, V, Mongelli, MA & Bronzetti, G 2003, 'Supplementation with fruit and vegetable concentrate decreases plasma homocysteine levels in a dietary controlled trial', Nutrition Research, vol. 23, no. 9, pp. 1221-1228. Papayianni, A, Alexopoulos, E, Giamalis, P, Gionanlis, L, Belechri, AM, Koukoudis, P & Memmos, D 2002, 'Circulating levels of ICAM-1, VCAM-1, and MCP-1 are increased in haemodialysis patients: association with inflammation, dyslipidaemia, and vascular events', Nephrology, Dialysis, Transplantation, vol. 17, no. 3, pp. 435-441. Parada, J & Aguilera, JM 2007, 'Food microstructure affects the bioavailability of several nutrients', Journal of Food Science, vol. 72, no. 2, pp. R21-R32.

150

Reference

Paterson, E, Gordon, MH, Niwat, C, George, TW, Parr, L, Waroonphan, S & Lovegrove, JA 2006, 'Supplementation with fruit and vegetable soups and beverages increases plasma carotenoid concentrations but does not alter markers of oxidative stress or cardiovascular risk factors', The Journal of Nutrition, vol. 136, no. 11, pp. 2849-2855. Pellegrini, N, Valtuena, S, Ardigo, D, Brighenti, F, Franzini, L, Del Rio, D, Scazzina, F, Piatti, PM & Zavaroni, I 2010, 'Intake of the plant lignans matairesinol, secoisolariciresinol, pinoresinol, and lariciresinol in relation to vascular inflammation and endothelial dysfunction in middle age-elderly men and post-menopausal women living in Northern Italy', Nutrition, Metabolism and Cardiovascular Diseases, vol. 20, no. 1, pp. 64-71. Peluso, I, Villano, DV, Roberts, SA, Cesqui, E, Raguzzini, A, Borges, G, Crozier, A, Catasta, G, Toti, E & Serafini, M 2013, 'Consumption of mixed fruit-juice drink and vitamin C reduces postprandial stress induced by a high fat meal in healthy overweight subjects', Current Pharmaceutical Design, vol. 20, no. 6, pp. 1020-1024. Pepys, MB & Hirschfield, GM 2003, 'C-reactive protein: a critical update', The Journal of Clinical Investigation, vol. 111, no. 12, pp. 1805-1812. Popa, C, Netea, MG, van Riel, PL, van der Meer, JW & Stalenhoef, AF 2007, 'The role of TNF-alpha in chronic inflammatory conditions, intermediary metabolism, and cardiovascular risk', Journal of Lipid Research, vol. 48, no. 4, pp. 751-762. Puglisi, MJ, Vaishnav, U, Shrestha, S, Torres-Gonzalez, M, Wood, RJ, Volek, JS & Fernandez, ML 2008, 'Raisins and additional walking have distinct effects on plasma lipids and inflammatory cytokines', Lipids in Health and Disease, vol. 7, p. 14. Pusparini, Dharma, R, Suyatna, FD, Mansyur, M & Hidajat, A 2013, 'Effect of soy isoflavone supplementation on vascular endothelial function and oxidative stress in postmenopausal women: a community randomized controlled trial', Asia Pacific Journal of Clinical Nutrition, vol. 22, no. 3, pp. 357-364. Rebello, SA, Chen, CH, Naidoo, N, Xu, W, Lee, J, Chia, KS, Tai, ES & van Dam, RM 2011, 'Coffee and tea consumption in relation to inflammation and basal glucose metabolism in a multi-ethnic Asian population: a cross-sectional study', Nutrition Journal, vol. 10, p. 61. Robak, T, Gladalska, A, Stepien, H & Robak, E 1998, 'Serum levels of interleukin-6 type cytokines and soluble interleukin-6 receptor in patients with rheumatoid arthritis', Mediators of Inflammation, vol. 7, no. 5, pp. 347-353. Root, MM, McGinn, MC, Nieman, DC, Henson, DA, Heinz, SA, Shanely, RA, Knab, AM & Jin, F 2012, 'Combined fruit and vegetable intake is correlated with improved inflammatory and oxidant status from a cross-sectional study in a community setting', Nutrients, vol. 4, no. 1, pp. 29-41. Rutkowski, P, Kamińska, J, Kowalska, M, Ruka, W & Steffen, J 2003, 'Cytokine and cytokine receptor serum levels in adult bone sarcoma patients: Correlations with local tumor extent and prognosis', Journal of Surgical Oncology, vol. 84, no. 3, pp. 151-159. Sacanella, E, Vazquez-Agell, M, Mena, MP, Antunez, E, Fernandez-Sola, J, Nicolas, JM, Lamuela-Raventos, RM, Ros, E & Estruch, R 2007, 'Down-regulation of adhesion molecules and other inflammatory biomarkers after moderate wine consumption in

151

Reference healthy women: a randomized trial', The American Journal of Clinical Nutrition, vol. 86, no. 5, pp. 1463-1469. Samman, S, Sivarajah, G, Man, JC, Ahmad, ZI, Petocz, P & Caterson, ID 2003, 'A mixed fruit and vegetable concentrate increases plasma antioxidant vitamins and folate and lowers plasma homocysteine in men', The Journal of Nutrition, vol. 133, no. 7, pp. 2188- 2193. Sanchez-Moreno, C, Cano, MP, de Ancos, B, Plaza, L, Olmedilla, B, Granado, F & Martin, A 2006, 'Mediterranean vegetable soup consumption increases plasma vitamin C and decreases F2-isoprostanes, prostaglandin E2 and monocyte chemotactic protein-1 in healthy humans', The Journal of Nutritional Biochemistry, vol. 17, no. 3, pp. 183-189. Sandhu, AK, Huang, Y, Xiao, D, Park, E, Edirisinghe, I & Burton-Freeman, B 2016, 'Pharmacokinetic characterization and bioavailability of strawberry anthocyanins relative to meal intake', Journal of Agricultural and Food Chemistry, vol. 64, no. 24, pp. 4891- 4899. Seifried, HE, Anderson, DE, Fisher, EI & Milner, JA 2007, 'A review of the interaction among dietary antioxidants and reactive oxygen species', Journal of Nutritional Biochemistry, vol. 18, no. 9, pp. 567-579. Serafini, M, Bugianesi, R, Maiani, G, Valtuena, S, De Santis, S & Crozier, A 2003, 'Plasma antioxidants from chocolate', Nature, vol. 424, no. 6952, pp. 1013-1013. Shine, B, de Beer, FC & Pepys, MB 1981, 'Solid phase radioimmunoassays for human C-reactive protein', Clinica Chimica Acta, vol. 117, no. 1, pp. 13-23. Shinohara, K, Oeda, E, Nomiyama, J, Inoue, H, Kamei, S, Tajiri, M, Ichikawa, T, Kuwaki, T & Tachibana, K 1995, 'The levels of granulocyte colony-stimulating factor in the plasma of the bone marrow aspirate in various hematological disorders', Stem Cells, vol. 13, no. 4, pp. 421-427. Skov, L, Beurskens, FJ, Zachariae, COC, Reitamo, S, Teeling, J, Satijn, D, Knudsen, KM, Boot, EPJ, Hudson, D, Baadsgaard, O, Parren, PWHI & van de Winkel, JGJ 2008, 'IL-8 as Antibody Therapeutic Target in Inflammatory Diseases: Reduction of Clinical Activity in Palmoplantar Pustulosis', The Journal of Immunology, vol. 181, no. 1, pp. 669-679. Song, Y, Manson, JE, Buring, JE, Sesso, HD & Liu, S 2005, 'Associations of dietary flavonoids with risk of type 2 diabetes, and markers of insulin resistance and systemic inflammation in women: a prospective study and cross-sectional analysis', Journal of The American College of Nutrition, vol. 24, no. 5, pp. 376-384. Stalmach, A, Troufflard, S, Serafini, M & Crozier, A 2009, 'Absorption, metabolism and excretion of Choladi green tea flavan-3-ols by humans', Molecular Nutrition and Food Research, vol. 53, pp. S44-S53. Steinberg, FM, Guthrie, NL, Villablanca, AC, Kumar, K & Murray, MJ 2003, 'Soy protein with isoflavones has favorable effects on endothelial function that are independent of lipid and antioxidant effects in healthy postmenopausal women', The American Journal of Clinical Nutrition, vol. 78, no. 1, pp. 123-130. Steptoe, A, Gibson, EL, Vuononvirta, R, Hamer, M, Wardle, J, Rycroft, JA, Martin, JF & Erusalimsky, JD 2007, 'The effects of chronic tea intake on platelet activation and inflammation: a double-blind placebo controlled trial', Atherosclerosis, vol. 193, no. 2, pp. 277-282.

152

Reference

Stravodimos, GA, Chetter, BA, Kyriakis, E, Kantsadi, AL, Chatzileontiadou, DS, Skamnaki, VT, Kato, A, Hayes, JM & Leonidas, DD 2016, 'Phytogenic polyphenols as glycogen phosphorylase inhibitors: the potential of triterpenes and flavonoids for glycaemic control in type 2 diabetes', Current Medicinal Chemistry. Sung, HJ, Ahn, JM, Yoon, YH, Rhim, TY, Park, CS, Park, JY, Lee, SY, Kim, JW & Cho, JY 2011, 'Identification and validation of SAA as a potential lung cancer biomarker and its involvement in metastatic pathogenesis of lung cancer', Journal of Proteome Research, vol. 10, no. 3, pp. 1383-1395. Teede, HJ, Dalais, FS & McGrath, BP 2004, 'Dietary soy containing phytoestrogens does not have detectable estrogenic effects on hepatic protein synthesis in postmenopausal women', The American Journal of Clinical Nutrition, vol. 79, no. 3, pp. 396-401. Ueland, PM, Refsum, H, Stabler, SP, Malinow, MR, Andersson, A & Allen, RH 1993, 'Total homocysteine in plasma or serum: methods and clinical applications', Clinical Chemistry, vol. 39, no. 9, pp. 1764-1779. Urieli-Shoval, S, Linke, RP & Matzner, Y 2000, 'Expression and function of serum amyloid A, a major acute-phase protein, in normal and disease states', Current Opinion in Hematology, vol. 7, no. 1, pp. 64-69. van Breemen, MJ, de Fost, M, Voerman, JS, Laman, JD, Boot, RG, Maas, M, Hollak, CE, Aerts, JM & Rezaee, F 2007, 'Increased plasma macrophage inflammatory protein (MIP)-1alpha and MIP-1beta levels in type 1 Gaucher disease', Biochimica et Biophysica Acta, vol. 1772, no. 7, pp. 788-796. van der Merwe, M & Bloomer, RJ 2016, 'The influence of methylsulfonylmethane on inflammation-associated cytokine release before and following strenuous exercise', Journal of Sports Medicine (Hindawi Publishing Corporation), vol. 2016, p. 7498359. Van Oostrom, AJ, Sijmonsma, TP, Rabelink, TJ, Van Asbeck, BS & Cabezas, MC 2003, 'Postprandial leukocyte increase in healthy subjects', Metabolism, vol. 52, no. 2, pp. 199- 202. van Oostrom, AJ, Sijmonsma, TP, Verseyden, C, Jansen, EH, de Koning, EJ, Rabelink, TJ & Castro Cabezas, M 2003, 'Postprandial recruitment of neutrophils may contribute to endothelial dysfunction', Journal of Lipid Research, vol. 44, no. 3, pp. 576-583. Villegas, R, Xiang, YB, Cai, H, Elasy, T, Cai, Q, Zhang, X, Fazio, S, Linton, MF, Li, H, Xu, WH, Yang, G, Zheng, W & Shu, XO 2012, 'Lifestyle determinants of C-reactive protein in middle-aged, urban Chinese men', Nutrition, Metabolism and Cardiovascular Diseases, vol. 22, no. 3, pp. 223-230. Wannamethee, SG, Lowe, GD, Rumley, A, Bruckdorfer, KR & Whincup, PH 2006, 'Associations of vitamin C status, fruit and vegetable intakes, and markers of inflammation and hemostasis', The American Journal of Clinical Nutrition, vol. 83, no. 3, pp. 567-574. Watzl, B, Kulling, SE, Moseneder, J, Barth, SW & Bub, A 2005, 'A 4-wk intervention with high intake of carotenoid-rich vegetables and fruit reduces plasma C-reactive protein in healthy, nonsmoking men', The American Journal of Clinical Nutrition, vol. 82, no. 5, pp. 1052-1058. Williams, CJ, Fargnoli, JL, Hwang, JJ, van Dam, RM, Blackburn, GL, Hu, FB & Mantzoros, CS 2008, 'Coffee Consumption Is Associated With Higher Plasma

153

Reference

Adiponectin Concentrations in Women With or Without Type 2 Diabetes: A prospective cohort study', Diabetes Care, vol. 31, no. 3, pp. 504-507. Wu, SH, Shu, XO, Chow, W-H, Xiang, Y-B, Zhang, X, Li, H-L, Cai, Q, Ji, B-T, Cai, H, Rothman, N, Gao, Y-T, Zheng, W & Yang, G 2012, 'Soy Food Intake and Circulating Levels of Inflammatory Markers in Chinese Women', Journal of The Academy of Nutrition and Dietetics, vol. 112, no. 7, pp. 996-1004. Zampelas, A, Panagiotakos, DB, Pitsavos, C, Chrysohoou, C & Stefanadis, C 2004, 'Associations between coffee consumption and inflammatory markers in healthy persons: the ATTICA study', The American Journal of Clinical Nutrition, vol. 80, no. 4, pp. 862- 867. Zern, TL, Wood, RJ, Greene, C, West, KL, Liu, Y, Aggarwal, D, Shachter, NS & Fernandez, ML 2005, 'Grape polyphenols exert a cardioprotective effect in pre- and postmenopausal women by lowering plasma lipids and reducing oxidative stress', The Journal of Nutrition, vol. 135, no. 8, pp. 1911-1917. Zetler, G, Seidel, G, Siegers, CP & Iven, H 1976, 'Pharmacokinetics of ascorbic acid in man', European Journal of Clinical Pharmacology, vol. 10, no. 3-4, pp. 273-282.

154

Appendices Appendices

Appendix I. Epidemiological studies from 2000 to 2016 reporting effects of plant-rich diets on biomarkers of oxidative stress and inflammation Country and Dietary assessment Positive (↑), inverse (↓) and no (↔) Reference Dietary score/index population method statistically significant associations USA (Filiberto et al. 259 women 24-h recall Isoflavone intake CRP↓ 2013) 18 – 44 y Urinary genistein CRP↓, WBC↔ USA Urinary daidzen CRP↓, WBC↔ (Nicastro et al. 1,683 participants Urinary test Urinary equol (bacterial metabolite of daidzen) CRP↔, WBC↓ 2013) 44.3 ± 0.5 y Urinary O-desmethylangolensin (bacterial CRP↓, WBC↓ metabolite of daidzen) Garlic supplementation CRP↔ USA (Kantor et al. Ginkgo biloba suplementation CRP↔ 9,947 participants FFQ, 30-d recall 2012) Saw palmetto CRP↔ ≥ 25 y Pycnogenol (grape seed extract, pine-bark) CRP↔ Fruit intake CRP↓, IL-6↓, TNF-α↓, WBC↓, MCP- 1↔, IL-10↔ USA Vegetable intake CRP↓, IL-6↓, TNF-α↓, WBC↔, MCP- (Root et al. 2012) 1,000 participants FFQ 1↔, IL-10↔ 18 – 85 y CRP↔, IL-6↓, TNF-α↓, WBC↔, MCP- Combined fruit and vegetable intake 1↔, IL-10↔ China (Villegas et al. Fruit-based diet CRP↓ 3,646 men FFQ 2012) 40 – 74 y Vegetable-based diet CRP↔ China (Wu et al. 2012) 1,005 women FFQ, 24-h recall Soy food intake IL-6↓, IL-1β↓, TNF-α↓, CRP↔ 40 – 70 y Italy Dietary diary (4 days (Azzini et al. 2011) 131 participants Mediterranean diet score IL-10↑, TNF-α↓ including the weekend) 20 – 40 y USA (Bhupathiraju & 908 participants FFQ Fruit and vegetable intake CRP↓ Tucker 2011) 45 – 75 y (Rebello et al. Singapore Coffee intake CRP↔, Adiponectin↔ FFQ 2011) 4,139 participants Black tea intake CRP↔, Adiponectin↔ 155

Appendices

48 y (mean) Oolong tea intake CRP↔, Adiponectin↔ Green tea intake CRP↓, Adiponectin↔ CRP↓ (in men age ≥ 45 y only), CRP↔ USA (women and men < 45 y), Fibrinogen↓ (Carter et al. 2010) 13,197 participants FFQ, 24-h dietary recall Mediterranean diet score (in men age ≥ 45 y only), Fibrinogen↔ 18 – 90 y (women and men < 45 y) Greece (Fragopoulou et al. 3,042 participants FFQ Mediterranean diet score Adiponectin↑ 2010) 18 – 89 y Japan (Maki et al. 2010) Closed-ended questions, Green tea intake CRP↔ 10,325 participants 28-d record 49 – 76 y Coffee intake CRP↓ (men only), CRP↔ (in women) USA (Nettleton, J. A. et 1,101 participants FFQ Healthy diet CRP↓ al. 2010) men: 71.8 y (mean) women: 70.7 y (mean) Total lignan intake ICAM-1↔ Italy Matairesinol intake ICAM-1↓ (Pellegrini et al. 242 participants 3-d weighed food record Secoisolariciresinol intake ICAM-1↓ 2010) 60 y (mean) Pinoresinol intake ICAM-1↔ Lariciresinol intake ICAM-1↔ UK (Cassidy et al. 1,754 women FFQ Fruit and vegetable intake Adiponecin↓ 2009) 18 – 80 y Italy (Centritto et al. 7,464 participants FFQ Olive oil and vegetable intake CRP↓ 2009) ≥ 35 y Vitamin C intake CRP↓, TNF-α↔, IL-6↓ β-Carotene intake CRP↔, TNF-α↓, IL-6↓ Folate intake CRP↓, TNF-α↔, IL-6↔ Fruit intake (without juice) CRP↓, TNF-α↔, IL-6↓ Fruit juice CRP↔, TNF-α↔, IL-6↔ USA Vegetable intake CRP↔, TNF-α↑, IL-6↓ (Holt et al. 2009) 285 participants FFQ Fruit and vegetable intake CRP↓, TNF-α↓, IL-6↓ 13 – 17 y Total flavonoid intake CRP↔, TNF-α↔, IL-6↔ Myricetin intake CRP↔, TNF-α↔, IL-6↔ Kaempferol intake CRP↔, TNF-α↔, IL-6↔ Quercetin intake CRP↔, TNF-α↔, IL-6↔ Luteolin intake CRP↔, TNF-α↓, IL-6↔ 156

Appendices

Apigenin intake CRP↔, TNF-α↔, IL-6↔ USA Total flavonoid intake CRP↓

8,335 participants Flavonol intake CRP↓ (Chun et al. 2008) 24-h recall ≥ 19 y Anthocyanidin intake CRP↓ Isoflavone intake CRP↓ Flavone intake CRP↔ Flavanone intake CRP↔ Flavan-3-ol intake CRP↔ Vitamin C intake CRP↓ Vitamin E intake CRP↓

Carotene intake CRP↓ Apple intake CRP↓ Vegetable intake CRP↓ Citrus fruit and juice intake CRP↔ Tea intake CRP↔ Wine intake CRP↔ Vietnam (Dai et al. 2008) 345 men FFQ Mediterranean diet score IL-6↓, CRP↔ about 54 y (mean) Comparison of the consumption of partially (Esmaillzadeh, A. Iran hydrogenated vegetable oils (PHVOs, commonly Consumption of non-HVOs: CRP↓, & Azadbakht 486 women FFQ used for cooking in Iran) and non-HVOs (sunflower TNF-α↓, ICAM-1↓, SAA↓ 2008) 40 – 60 y oil, corn oil, canola oil, soybean oil and olive oil) Alternate healthy eating index (fruit, vegetables, the ratio of white meat to red USA (Fargnoli et al. meat, trans fat, the ratio of polyunsaturated fat to Adiponectin↑, CRP↓, E-selectin↓, IL- 1,922 women FFQ 2008) saturated fat, cereal fibre, nuts, soy, moderate 6↔, ICAM-1↔, VCAM-1↔, TNF-α↔ 30 – 55 y alcohol consumption and long-term multivitamin use) Japan (Nanri et al. 2008) 7,802 participants FFQ Healthy diet CRP↓ 50 – 74 y Comprehensive healthy dietary pattern USA CPR↓, IL-6↓, Fibrinogen↓ (Nettleton, J. A. et (47 food groups) 5,042 participants FFQ al. 2008) Simplified healthy dietary pattern 45 – 84 y CRP↓, IL-6↔, Fibrinogen↔ (6 food groups) (Williams et al. 1,058 women Semi FFQ, four 1-week Coffee intake Adiponectin↑ 2008) 55 y (mean) dietary records

157

Appendices

Iran 196 women CRP↓, VCAM-1↓, IL-6↔, (Esmaillzadeh, A. 18 – 84 y FFQ Healthy diet E-selectin↔, ICAM-1↔, TNF-α↔, et al. 2007) 486 women SAA↔ 40 – 60 y Finland (Mikkila et al. Health-conscious diet CRP↓ (in women), CRP↔ (in men), 1,037 participants 48-h recall 2007) Homocysteine↓ 24 – 39 y Belgium (De Bacquer et al. Tea intake CRP↓, SAA↓, Fibrinogen↔ 1,031 men Questionnaire 2006) 35 – 59 y Coffee intake CRP↔, SAA↔, Fibrinogen↔ Iran (Esmaillzadeh, Fruit intake CRP↓ 486 women FFQ Ahmad et al. 2006) 40 – 60 y Vegetable intake CRP↓ CRP↔, ICAM-1↔, E-selectin↔, TNF- USA Coffee intake (Lopez-Garcia, α↔ 730 women FFQ Esther et al. 2006) ICAM-1↔, E-selectin↓, CRP↓, TNF- 43 – 70 y Decaffeinated coffee intake α↔ CRP↔, IL-6↓, Homocysteine↔, ICAM- USA Vegetable and fish diet (Nettleton, Jennifer 1↔, E-selectin↔ 5, 089 participants FFQ A et al. 2006) CRP↓, IL-6↓, Homocysteine↓, ICAM- 45 – 84 y Whole grains and fruit diet 1↓, E-selectin↔ (Panagiotakos, Greece Pitsavos & 3,042 participants FFQ Mediterranean diet score CRP↓, Fibrinogen↓ Stefanadis 2006) > 18 y UK Vitamin C intake CRP↓, Fibrinogen↔ (Wannamethee et 3,258 men FFQ Fruit intake CRP↓, Fibrinogen↔ al. 2006) 60 – 79 y Vegetable intake CRP↔, Fibrinogen↔ Healthy eating index CRP↓ (women only), CRP↔ (men only) USA (Ford, Mokdad & Grain intake CRP↓ (women only), CRP↔ (men only) 13,811 participants Single 24-h recall, FFQ Liu 2005) Fruit intake CRP↓ (women only), CRP↔ (men only) ≥ 20 y Vegetable intake CRP↔ CRP↔, IL-6↔, E-selectin↔, ICAM- Healthy eating index 1↔, VCAM-1↔ USA CRP↓, IL-6↓, E-selectin↓, (Fung et al. 2005) 660 women FFQ Alternate healthy eating index ICAM-1↓, VCAM-1↔ 43 – 69 y CRP↔, IL-6↔, E-selectin↔, ICAM- Diet quality index revised 1↔, VCAM-1↔

158

Appendices

CRP↔, IL-6↔, E-selectin↓, ICAM-1↔, Recommended food score VCAM-1↔ CRP↓, IL-6↓, E-selectin↓, ICAM-1↔, Alternate Mediterranean diet score VCAM-1↓ USA Healthy eating index CRP↓, Fibrinogen↔ (Kant & Graubard 8,719 participants 24-h recall Recommended food score CRP↓, Fibrinogen↓ 2005) ≥ 20 y Dietary diversity score for recommended foods CRP↓, Fibrinogen↔ UK (Song et al. 2005) 344 women FFQ, 8.8 y of follow-up Total flavonoid intake CRP↔, IL-6↔ ≥ 45 y Greece (Chrysohoou et al. CRP↓, IL-6↓, Fibrinogen↓, 3,042 participants FFQ Mediterranean diet score 2004) TNF-α↔, SAA↔ 18 – 89 y USA (Gao, Bermudez & 599 participants FFQ Fruit and vegetable intake CRP↓ Tucker 2004) 70 y (mean) USA (Lopez-Garcia, E. CRP↓, IL-6↔, E-selectin↓, 732 women Prudent diet (high in vegetables, fruits and tomatoes) et al. 2004) ICAM-1↔, VCAM-1↓ 43 – 69 y Greece 1,514 men (Zampelas et al. 18 – 87 y FFQ Coffee intake CRP↑, IL-6↑, TNF-α↑, SAA↑, WBC↑ 2004) 1,528 women 18 – 89 y CRP, C-reactive protein; FFQ, food frequency questionnaire; ICAM-1, intercellular adhesion molecule 1; IL, interleukin; MCP-1, monocyte chemoattractant protein 1; SAA, serum amyloid A; TNF-α, tumour necrosis factor-alpha; VCAM-1, vascular cell adhesion 1; WBC, white blood cell count.

159

Appendices

Appendix II. Clinical studies from 2000 to 2016 reporting effects of plant-rich diets on biomarkers of oxidative stress and inflammation Country and Study Frequency of Reference Intervention Duration Comparison Biomarkers Hard end points population design measurement Consumption of supplement Indonesia Change in plasma Randomis containing 100 mg 182 inflammatory Reduction of oxidative (Pusparini ed double isoflavone and 500 mg At baseline, 6 postmenopau 12 mo biomarkers VCAM-1↔ stress in the isoflavone et al. 2013) blind calcium carbonate or just mo and 12 mo sal women compared to group controlled 500 mg calcium carbonate 47 – 60 y control group (control group) Consumption of high-fat USA Tomato (Burton- Crossover meal (45% fat) containing 30 min interval Change in plasma Reduction in oxidative 25 Postprand product: IL-6↓, Freeman et placebo- either processed tomato until 2 h after inflammatory stress in the tomato participants ial CRP↔, TNF- al. 2012) controlled product or non-tomato meal biomarkers group 27 ± 8 y α↔ product 3 groups: lifestyle education control group (LE), Change in plasma USA Randomis resistance training group IL-6↔, (Deibert et At baseline and inflammatory 40 men ed (RT) and resistance training 12 wk CRP↔, al. 2011) after 12 wk biomarkers after 50 – 65 y controlled with supplementation of 50 fibrinogen↔ intervention g of a soy-yogurt-honey per day (RTS) Consumption of supplement containing 100 mg USA resveratrol and 75 mg total 10 Crossover 2 h interval Change in plasma (Ghanim et polyphenols from a Postprand Reduction of oxidative participants placebo- until 5 h after inflammatory IL-1β↓ al. 2011) muscadine grape extract in ial stress 37 ± 4 y controlled meal biomarkers combination with a 900-kcal

high-fat high-carbohydrate meal 3 groups: consumption of 150 g of cooked white- Change in plasma (WP), yellow- (YP) (58 mg Randomis inflammatory YP group: IL- USA of carotenoids and 0.3 g of Reduction of oxidative (Kaspar et ed At baseline and biomarkers 6↓, CRP↔ 36 men anthocyanin/kg) or purple- 6 wk stress in the YP and PP al. 2011) placebo- after 6 wk compared with the PP group: IL- 18 – 40 y flesh (PP) (1.3 mg of groups controlled WP group (as 6↓, CRP↓ carotenoids and 6.2 placebo) anthocyanin/kg) potatoes daily

160

Appendices

Q-500 group: IL-6↔, IL- Supplementation of 10↔, GCSF↔, quercetin, vitamin C and MCP-1↔, USA niacin: Change in plasma Randomis TNF-α↔, 1,002 Group 1: 500 mg quercetin, inflammatory (Knab et al. ed 12 wk At baseline and CRP↔ participants 125 mg vitamin C and 5 mg biomarkers 2011) placebo- after 12 wk Q-1000 group: 18 – 85 y niacin (Q-500) compared to controlled IL-6↓, IL-10↔, Group 2: 1000 mg placebo GCSF↔, quercetin, 250 mg vitamin C MCP-1↔, and 10 mg niacin (Q-1000) TNF-α↔, CRP↔ Change in plasma USA Randomis Supplementation of 200 At baseline and inflammatory (Ghanim et 20 ed mg/d of P. cuspidatum TNF-α↓, IL-6↓, 6 wk at 1, 3 and 6 biomarkers al. 2010) participants placebo- extract, containing 40 mg/d CRP↓ wk compared to 36 ± 5 y controlled trans-resveratrol placebo Randomis Change in plasma ed, USA Supplementation of inflammatory (Heinz et double- At baseline and IL-6↔, TNF- 120 women encapsulated quercetin: 500 12 wk biomarkers al. 2010) blind after 12 wk α↔, CRP↔ 30 – 79 y mg/d or 1000 mg/d compared to placebo- placebo controlled 1. Fruit, vegetable and berry: MCP- Change in plasma Randomis Supplementation of 3 1↓, CRP↔, inflammatory USA ed capsules per day, containing MIP-1β↓, biomarkers in (Jin et al. 117 double- Group 1: Fruit, vegetable At baseline and RANTES↓ 60 d groups taking the 2010) participants blind, and berry after 60 d 2. Fruit and supplements 22 – 55 y placebo- Group 2: Fruit and vegetable: compared to controlled vegetable MCP-1↓, placebo group CRP↔, MIP- 1β↓, RANTES↓ Group 1: Mediterranean diet Change in plasma Iran Randomis MD+VO: IFN- (Konstantin with virgin olive oil inflammatory Reduction of oxidative 90 ed, At baseline and γ↓, MCP-1↔, idou et al. (MD+VO) 3 mo biomarkers from stress in both groups. participants parallel after 3 mo P-selectin↓, 2010) Group 2: Mediterranean diet baseline to 3 20 – 50 y controlled CRP↓ with washed virgin olive oil months 161

Appendices

(lower polyphenol content MD+WO: : than virgin olive oil) IFN-γ↔, (MD+WO) MCP-1↔, P- selectin↔, CRP↓ Randomis USA ed, Supplementation of 2 Change in plasma 31 (Beavers et single- serving of soy milk (~90 mg At baseline and inflammatory TNF-α↔, IL- postmenopau 4 wk al. 2009) blind of isolfavone) or diary milk after 4 wk biomarkers 1β↔, IL-6↔ sal women parallel per day between 2 groups 40 – 60 y controlled USA Change in plasma Randomis 75 Consumption of 20 g soy inflammatory Adiponectin↑, (Charles et ed At baseline and postmenopau protein with 160 mg of total 12 wk biomarkers IL-6↔, TNF- al. 2009) placebo- after 12 wk sal women isoflavones per day compared to α↔ controlled ~ 57 y placebo In plasma: 3 diet interventions for 4 wk Extra olive oil: each: Western diet rich in TNF-α↔, IL- saturated fatty acids, 6↔, MCP-1↔ Mediterranean diet rich in Change in Walnut: TNF- virgin olive oil or high- inflammatory α↔, IL-6↔, Spain (Jimenez- Crossover carbohydrate rich in vegetal 4 wk, 3 h interval biomarkers in MCP-1↔ 20 Gomez et placebo- n-3 fatty acids for 4 wk, postprand until 9 h after plasma and in participants al. 2009) controlled followed by a breakfast with ial meal peripheral blood In PBMCs: ~ 20 y a fat composition similar to mononuclear cells Extra olive oil: that consumed in each of the (PBMCs) TNF-α↓, IL-6↑, diets: butter breakfast, olive MCP-1↔ oil breakfast or walnut Walnut: TNF- breakfast α↓, IL-6↓, MCP-1↔ Randomis ed, USA parallel, Supplementation of Change in plasma (Nantz et 111 At baseline and Reduction of oxidative double- decaffeinated green tea 3 wk inflammatory SAA↓ al. 2009) participants after 3 wk stress blind, capsules. Dose: 2 capsules/d biomarkers 21 – 70 y placebo- controlled

162

Appendices

24 d (subject Supplementation of Change in plasma Randomis cycled for Q: CRP↔, IL- quercetin (Q) with or inflammatory USA ed, a 3-d Before and 6↔, IL-10↓, without epigallocatechin 3- biomarkers after (Nieman et 39 double- period after a 3-d GCSG↓ gallate, isoquercetin and exercise and al. 2009) participants blind, after the period of Q-EGCG: eicosapentaenoic acid and supplementation 26.3 ± 1.7 placebo- first 2 wk exercise CRP↓, IL-6↓, docosahexaenoic acid (Q- compared to controlled of IL-10↓, GCSF↓ EGCG) placebo suppleme ntation) Change in plasma Germany Supplementation of inflammatory (Egert et al. 35 Randomis encapsulated quercetin at 3 At baseline and 2 wk biomarkers TNF-α↔ 2008) participants ed doses: 50, 100 or 150 mg/d after 2 wk compared to 19 – 40 y (group Q 50 – Q150) baseline 3 diets: Extra virgin Group 1: Western diet, rich olive oil: Plasma Spain Randomis in saturated fat 4 wk, 2 h interval VCAM-1↓, (Fuentes et inflammatory 20 men ed Group 2: Mediterranean diet postprand after meal until ICAM-1↔ al. 2008) biomarkers 18 – 30 y crossover + extra virgin olive oil ial 8 h Walnut: between groups Group 3: a low-fat diet (with VCAM-1↓, walnut) ICAM-1↔ Consumption of soy protein USA Change in plasma Homocysteine concentrate (26 ± 5 g, 34 Randomis inflammatory ↔, CRP↔, E- (Greany et containing 44 ± 8 mg At baseline and postmenopau ed 6 wk biomarkers selectin↔, al. 2008) isoflavones per day) or milk after 6 wk sal women crossover compared to VCAM-1↔, protein isolate as placebo 47 - 69 placebo ICAM-1↔ group 1. Tomato juice Group 1: Tomato juice: group: CRP↓, Reduction of oxidative daily dose 20.6 mg lycopene Change in plasma IL-1β↔, TNF- Germany stress in the tomato juice and 45.5 mg vitamin C inflammatory α↓ (Jacob et al. 24 Randomis At baseline and fortified with vitamin C Group 2. Tomato juice 2 wk biomarkers 2. Tomato juice 2008) participants ed after 2 wk group fortified with vitamin C: compared to fortified with 19 - 27 Reduction of cholesterol daily dose 20.6 mg lycopene baseline vitamin C: levels in both groups and 435 mg vitamin C CRP↓, IL-1β↓, TNF-α↔ (Pacheco, Spain Randomis 1 wk, 1 h interval Postprandial Olive oil group Diet rich in refined olive oil Yolanda M. 14 ed postprand intil 8 h after plasma vs sunflower or high-palmitic sunflower et al. 2008) participants crossover ial meal inflammatory oil group: 163

Appendices

21 – 38 y oil for 1 wk, followed by a biomarkers VCAM-1↓, high-fat meal between groups ICAM-1↓, RAISINS: TNF-α↓, ICAM-1↓, IL- Group 1: Consumption of 1 8↔, MCP-1↔ Change in plasma USA cup raisins/d (RAISINS) WALK: TNF- in plasma 34 Group 2: Increase the α↔, ICAM-1↓, (Puglisi et Randomis At baseline and inflammatory Reduction of cholesterol participants amount of steps walked/d 6 wk IL-8↔, MCP- al. 2008) ed after 6 wk biomarkers levels in all groups 55 ± 3.8 y (WALK) 1↔ compared to Group 3: Combination of RAISINS + baseline RAISINS + WALK WALK: TNF- α↔, ICAM-1↓, IL-8↔, MCP- 1↔ Italy Change in plasma 389 Randomis Supplementation of At baseline, inflammatory (Atteritano osteopenic ed VCAM-1↓, genistein 54 mg/d in 2 24 mo after 12 mo and biomarkers et al. 2007) postmenopau placebo- ICAM-1↓ tablets after 24 mo compared to sal women controlled placebo 49 – 67 y Red wine, Randomis Consumption of red wine or Change in plasma Spain brandy, rum: (Blanco- ed vodka or brandy or rum or inflammatory 16 At baseline, 1 d NF-κB Colio et al. crossover, control without alcohol. All 5 d biomarkers participants and 5 d activation↓ 2007) placebo- groups accompanied by a compared to 22 – 29 y Red wine: controlled fat-enriched diet (44%) control MCP-1↓ High carbohydrate meal Change in plasma Italy Only in extra Crossover (150 g of potatoes) + At baseline and inflammatory Reduction of oxidative (Bogani et 12 men Postprand virgin olive oil placebo- different kind of oil (50 1, 2 and 6 h biomarkers and stress in the extra virgin al. 2007) 25 ± 3 y ial group: TXB ↓, controlled mL): extra virgin olive oil, after meal oxidative stress 2 olive oil group LT ↓ olive oil or corn oil between groups 4 Change in plasma IL-4↓, IL-8↓, Norway Parallel, Supplementation of inflammatory IL-13↓, (Karlsen et 120 At baseline and placebo- anthocyanin Medox 300 mg 3 wk biomarkers RANTES↓, al. 2007) participants after 3 wk controlled per day compared to IFN-α↓, 40 – 74 y placebo CRP↔ (Kawashim Japan Randomis Supplementation of Juice Change in plasma At baseline, 14 Reduction of oxidative a et al. 60 ed, Plus (commercial fruit and 28 d homocystein Homocysteine↓ d and 28 d stress 2007) participants double- vegetable extract capsules) concentration 164

Appendices

18 – 50 y blind, compared to placebo- placebo controlled compared to placebo group: TNF-α↓ at 16 Randomis Change in plasma Supplementation of capsules and 28 wk Austria ed, TNF-α during the containing fruit, vegetable At baseline, 4 In both group: (Lamprecht 41 trained double- intervention Reduction of duty days and berry powder 28 wk wk, 8 wk, 16 TNF-α↑ during et al. 2007) men blind, compared to lost due to illness concentrate. Dose: 6 wk and 28 wk the first 8 wk 34 ± 5 y placebo- baseline and capsules daily followed by controlled placebo group TNF-α↓ for the following 20 wk. 1 wk of supplementation of with extra-virgin olive oil containing 1125 mg Change in plasma Randomis Spain polyphenols/kg and 350 inflammatory (Pacheco, ed 1 wk, 1 h interval 14 tocopherols/kg, or refined biomarkers VCAM-1↓, Y. M. et al. crossover postprand until 8 h after participants olive oil with no compared to ICAM-1↓ 2007) placebo- ial meal 21 – 38 y polyphenols or tocopherols; placebo (refined controlled followed by a high-fat meal olive oil) enriched in extra virgin olive oil or refined olive oil Red wine: CRP↓, ICAM- 1↓, IL-6↓, Change in plasma VCAM-1↓, E- Spain Randomis Consumption of 30 g inflammatory (Sacanella At baseline and selectin 35 women ed ethanol/d as white or red 4 wk biomarkers et al. 2007) after 4 wk White wine: 20 – 50 y crossover wine compared to CRP↓, ICAM- baseline 1↓, IL-6↓, VCAM-1↔, E- selectin↔ Randomis Change in plasma ed, Consumption of black tea UK inflammatory (Steptoe et double- daily: 1050 mg tea extract At baseline and CRP↓, 75 men 6 wk biomarkers al. 2007) blind, dissolved in 250 mL hot after 6 wk P-selectin↔ 18 – 55 y compared to placebo- water placebo controlled 165

Appendices

2-wk run-in period with fish Randomis oil supplementation (4 g/d) UK ed single- and throughout 4 wk of the Change in plasma (Paterson et 36 blind, study: At baseline and homocysteine 4 wk Homocysteine↓ al. 2006) participants controlled 1. Consumption of after 6 wk compared to 20 – 67 y , carotenoid-rich foods control crossover 2. Control: consumption of carotenoid-poor foods Spain Consumption of 500 mL of Change in plasma 12 commercial gazpacho (a (Sanchez- inflammatory MCP-1↓, TNF- participants Randomis Mediterranean vegetable At baseline, 7 d Reduction of oxidative Moreno et 14 d biomarkers α↔, IL-1β↔, (6 men and 6 ed soup constituted mainly and 14 d stress al. 2006) compared to IL-6↔ women) tomato, pepper and baseline 22 ± 0.5 y cucumber) per day Randomis ed, UK Change in plasma double- Consumption of isoflavone- 117 inflammatory CRP↓, VCAM- (Hall et al. blind, enriched (genistein-t- At baseline and postmenopau 8 wk biomarkers 1↔, E- 2005) placebo- daidzein ratio of 2:1; 50 8 wk sal women compared to selectin↔ controlled mg/d) 45 – 70 y placebo . crossover Consumption of fruit and Germany Change in plasma Randomis vegetables: (Watzl et 63 At baseline and inflammatory Group 3 vs ed Group 1: 2 servings/d 4 wk al. 2005) participants 4 wk biomarkers group 1: CRP↓ controlled Group 2: 5 servings/d 31 ± 9 y between groups Group 3: 8 servings/d USA 24 Randomis premenopaus Change in plasma ed single- Consumption of 36 g/d of a al women inflammatory (Zern et al. blind, lyophilised grape powder At baseline and TNF-α↓, IL- Reduction of oxidative (39.7 ± 8.5 y) 4 wk biomarkers 2005) crossover, (mixed with water before 4 wk 6↔, CRP↔ stress 20 compared to placebo- intake) postmenopau placebo controlled sal women (58.5 ± 7.5 y) USA Parallel, Group 1: supplementation of Change in plasma Group 1: CRP↓ (Block et 160 randomis vitamin C (515 mg/d) At baseline and CRP concentration Group 2 and 2 mo al. 2004) participants ed, Group 2: supplementation of 2 mo compared to placebo: 46 ± 14 y double- anti-oxidant mixture (per placebo CRP↔ 166

Appendices

blind, day: 515 mg vitamin C, 371 placebo- mg α-tocopherol, 171 mg γ- controlled tocopherol, 252 mg mixed tocotrienols and 95 mg α- lipoic acid) Wine: Fibrinogen↓, Open, IL-1α↓, MCP- prospecti 1↓, CRP↓, ve, VCAM-1↓, Spain randomis Consumption of 30 g of Change in plasma (Estruch et At baseline and ICAM-1↓ 40 men ed, ethanol per day as red wine 2 wk inflammatory al. 2004) 2 wk Gin: 30 – 50 y crossover (320 mL) or gin (100 mL) biomarkers Fibrinogen↓, and IL-1α↓, MCP- single- 1↔, CRP↔, blind VCAM-1↔, ICAM-1↔ Group 1: consumption of Australia (Teede, Double- soy protein isolate (per day: Change in plasma 50 Dalais & blind, 40 g soy protein, 118 mg At baseline and CRP concentration postmenopau 3 mo CRP↔ McGrath placebo- isoflavones) 3 mo compared to sal women 2004) controlled Group 2: consumption of placebo 50 – 75 y casein placebo Change in plasma Denmark Randomis Combined supplementation (Bruunsgaa inflammatory 520 ed of vitamin E (91 mg) and At baseline and TNF-α↔, IL- rd et al. 3 y biomarkers participants placebo- vitamin C (250 mg) twice after 3 y 6↔, CRP↔ 2003) compared to 45 – 69 y controlled daily placebo Italy Change in plasma Supplementation of fruit and (Panunzio 24 Crossover At baseline, 13 homocysteine vegetable capsules: 2 4 wk Homocysteine↓ et al. 2003) participants controlled d, 27 d compared to capsules of each/d 20 – 56 y control Randomis ed Change in plasma Australia double- Supplementation of Juice (Samman et At baseline and homocysteine Reduction of heart 32 men blind, Plus (commercial fruit and 6 wk Homocysteine↓ al. 2003) after 6 wk compared to disease risk 18 – 50 y crossover vegetable extract capsules) placebo placebo- controlled

167

Appendices

Consumption of 25 g of 3 protein products/d: USA Group 1: isolated soy Randomis 28 protein with isoflavones ed, Change in plasma E-selectin↔, (Steinberg postmenopau (107 units/d) At baseline and double- 6 wk inflammatory VCAM-1↔, et al. 2003) sal women Group 2: ethanol-washed after 6 wk blind, biomarkers ICAM-1↔ 54.9 ± 1 y isolated soy protein with crossover trace isoflavones (2 units/d) Group 3: total milk protein (0 units/d) Plasma P-selectin↓, E- Australia Randomis inflammatory selectin↔, (Hodgson 22 ed- Consumption of 5 cups/d At baseline and 4 wk biomarkers ICAM-1↔, et al. 2001) participants controlled (250 mL) of black tea after 4 wk compared to VCAM-1↔, 59.1 ± 1.6 y crossover control (water) Fibrinogen↔ CRP, C-reactive protein; GCSF, granulocyte colony stimulating factor; ICAM-1, intercellular adhesion molecule 1; IL, interleukin; LTB4, leukotriene B4; MCP-1, monocyte chemoattractant protein 1; MIP-1β, macrophage inflammatory protein 1β; NF-κB, nuclear factor κB; RANTES, regulated upon activation normal T cell and expressed and secreted; SAA, serum amyloid A; TNF-α, tumour necrosis factor- alpha; TXB2, thromboxane B2; VCAM-1, vascular cell adhesion 1.

168

Appendices

Appendix III. Common biomarkers of oxidative stress and inflammation Normal plasma Marker Type Role Disease Reference concentration Pro-OSI (Pepys & Median: 0.8 mg/L; the 90th Diabetes, cardiovascular diseases, Hirschfield C-reactive protein Acute phase Pattern recognition receptor centiles is 3.0 mg/L and the systemic sclerosis and rheumatoid 2003; Shine, (CRP) protein 99th centiles is 10 mg/L arthritis de Beer & Pepys 1981) (Lowe, Acute phase Major plasma protein coagulation Rumley & Fibrinogen 1.5 – 4.5 g/L Thrombosis, cardiovascular diseases protein factor Mackie 2004) (Sung et al. Cardiovascular disease, obesity, 2011; Urieli- Serum amyloid A Acute phase Transport of cholesterol, amyloid rheumatoid arthritis, cancer, Alzheimer’s Shoval, 13.89 ± 37.18 μg/mL (SAA) protein fibril formation and cell adhesion disease Linke & Matzner 2000) Cell adhesion Recruiting leukocytes to the site Median: 58 ng/mL, range: (Albertini et E-selectin Cardiovascular diseases molecule of injury 43 – 80 ng/mL al. 1998) Intercellular adhesion Cell adhesion Median: 226 ng/mL, range: (Albertini et Transmigrating leukocytes Cardiovascular diseases molecule (ICAM-1) molecule 193 – 289 ng/mL al. 1998) Recruiting leukocytes to vascular Vascular cell adhesion Cell adhesion Median: 443 ng/mL, range: Cardiovascular disease (Albertini et endothelium molecule 1 (VCAM-1) molecule 395 – 573 ng/mL al. 1998)

(Eyles et al. Granulocyte colony- Regulating neutrophil production Rheumatoid arthritis, collagen-induced 2006; stimulating factor Cytokine 16.2 ± 2.4 pg/mL and survival arthritis Shinohara et (GCSF) al. 1995) (Kornman 2006; Interleukin 1α (IL-1α) Amplifying the inflammatory IL-1α: 156.5 ± 539.5 pg/mL Cardiovascular diseases, Alzheimer’s Mooradian, and interleukin 1β (IL- Cytokine cascades IL-1β: ~50 pg/mL disease, gastric cancer and periodontitis Reed & 1β) Scuderi 1991) (eMaggio et Diabetes, depression, cardiovascular Stimulating immune responses Median: 5.3 pg/mL, range: al. 2006; Interleukin 6 (IL-6) Cytokine diseases, Alzheimer’s disease, cancer, and acute-phase reactions 0.5 – 16.6 pg/mL Robak et al. rheumatoid arthritis, multiple sclerosis 1998) 169

Appendices

(Bruunsgaard promoting the expression of et al. 1999; Tumour necrosis factor adhesion molecules on endothelial Median: 1.4 pg/mL, range: Cardiovascular disease, diabetes , Cytokine Popa et al. α (TNF-α) cells, recruitment and activation of 0.4 – 5.5 pg/mL rheumatoid arthritis 2007) inflammatory cells

(Rutkowski Rheumatoid arthritis, inflammatory Immune cell activation and et al. 2003; Interleukin 8 (IL-8) Chemokine 10 – 12.5 pg/mL bowel disease, psoriasis and promotion of angiogenesis Skov et al. palmoplantar pustulosis 2008) Recruiting monocytes, neutrophils Cardiovascular diseases, multiple (Deshmane Monocyte and lymphocytes, regulating sclerosis, cancer, rheumatoid arthritis, et al. 2009; chemoattractant protein Chemokine 125 ± 42 pg/mL migration and infiltration of diabetes, obesity, brain ischemia, Papayianni et 1 (MCP-1) monocytes/macrophages Alzheimer’s disease al. 2002) (Maurer & Macrophage Attracting other pro-inflammatory Cardiovascular diseases, rheumatoid von Stebut median: 17 pg/mL, range: 1 inflammatory protein Chemokine cells, recruiting themselves to the arthritis, multiple sclerosis, sepsis, 2004; van – 41 pg/mL 1β (MIP-1β) sites of inflammation asthma, granuloma formation Breemen et al. 2007) (Appay & Regulated upon Median: 19.93 ng/mL, Rowland- activation normal T cell Inducing leukocyte migration to Cardiovascular diseases, arthritis, apotic Chemokine range: 14.22 – 29.28 ng/mL Jones 2001; expressed and secreted sites of inflammation dermatitis, asthma Herder et al. (RANTES) 2005) (Banecka-

Majkutewicz Non-protein Involved in transfer and metabolic Homocysteine 5 – 15 µmol/L et al. 2012; amino acid cycle of the methyl group Cardiovascular diseases Ueland et al.

1993) Anti-OSI (Diez & Reduced levels in plasma correlated with Antihyperglycemic, anti- Iglesias Adiponectin Cytokine 5 – 30 µg/ml obesity, diabetes mellitus type 2, atherogenic and anti-inflammatory 2003; Housa cardiovascular diseases and cancer et al. 2006) (Di Bona et al. 2012; Suppressing macrophage and the Protective roles in rheumatoid arthritis, Opal, Interleukin 10 (IL-10) Cytokines production of pro-inflammatory 2 – 7.3 pg/mL multiple sclerosis, asthma, bowel Wherry & cytokines, preventing host damage disease, cancer and Alzheimer’s disease Grint 1998; Rutkowski et al. 2003) 170

Appendices

Appendix IV. Supporting Information for Chapter 2

S1. Table. The training dataset used for the development of the predictive model. T # No. freely max PSA* No. H No. H Molecular (h) Phytochemical M * LogP* rotatable Intake# Reference# r (Å2) acceptor* donor* volume* (Å3) (mean ± bonds* SE) Anthocyanins Cyanidin 3-O- solid 3.5 ± 419.36 -1.627 171.234 4 10 7 342.085 31 arabinoside (extract) 0.5 Cyanidin 3-O- solid 2.5 ± 449.38 -2.793 191.462 4 11 8 366.931 31 galactoside (extract) 0.5 Cyanidin 3-O- liquid 449.38 -2.793 191.462 4 11 8 366.931 0.5 54 glucoside (juice) semi-solid (freeze-dried Cyanidin 3-O- 1.09 ± 449.38 -2.793 191.462 4 11 8 366.931 whole fruit 52 glucoside 0.3 made into a paste) Cyanidin 3-O- liquid 449.38 -2.793 191.462 4 11 8 366.931 1 49 glucoside (juice) semi-solid (freeze-dried Cyanidin 3-O- 1.64 ± 595.53 -3.492 250.386 6 15 10 490.793 whole fruit 52 rutinoside 0.5 made into a paste) Cyanidin 3-O- liquid 595.53 -3.492 250.386 6 15 10 490.793 1 49 rutinoside (juice) semi-solid (freeze-dried Cyanidin 3-O- 2.18 ± 581.50 -4.071 250.386 6 15 10 474.206 whole fruit 52 sambubioside 1.54 made into a paste)

171

Appendices

semi-solid (freeze-dried Cyanidin 3-O- 2.55 ± 727.65 -4.572 309.31 8 19 12 598.069 whole fruit 52 xylosyl-rutinoside 0.169 made into a paste) Delphinidin 3-O- liquid 465.39 -3.084 211.69 4 12 9 374.948 1 49 glucoside (juice) Delphinidin 3-O- liquid 611.53 -3.783 270.614 6 16 11 498.811 1 49 rutinoside (juice) semi-solid Malvidin 3-O- 1.8 ± 493.44 -2.469 189.702 6 12 7 410.004 (extract with 36 glucoside 0.6 yoghurt) semi-solid Peonidin 3-O- 1.4 ± 463.42 -2.486 180.468 5 11 7 384.458 (extract with 36 glucoside 0.8 yoghurt) Flavanols (−)-Epicatechin 3- liquid 1.6 ± 442.37 2.537 177.135 4 10 7 359.554 18 O-gallate (drink) 0.2 (−)- liquid 1.9 ± Epigallocatechin 3- 458.37 2.245 197.363 4 11 8 367.571 18 (drink) 0.1 O-gallate (−)- liquid 1.4 ± Epigallocatechin 3- 458.37 2.245 197.363 4 11 8 367.571 34 (drink) 0.6 O-gallate Flavonols semi-solid Isorhamnetin 3-O- 0.6 ± 492.39 -0.179 216.58 5 13 7 391.915 (fried red 42 glucuronide 0.1 onions) solid Quercetin aglycone 302.24 1.683 131.351 1 7 5 240.084 (pure 3 ± 2 41 compound) Quercetin 3-O- 0.6 ± 478.36 -0.487 227.574 4 13 8 374.387 semi-solid 42 glucuronide 0.1 172

Appendices

(fried red onions) semi-solid Quercetin 3'-O- 0.75 ± 381.29 -1.809 177.556 3 10 4 277.758 (fried red 42 sulfate 0.12 onions) Hydroxybenzoic acids liquid 0.65 ± Ellagic acid 302.19 0.943 141.334 0 8 4 221.776 21 (juice) 0.23 liquid 0.94 ± Ellagic acid 302.19 0.943 141.334 0 8 4 221.776 (extract 21 0.06 drink) solid 2.58 ± Ellagic acid 302.19 0.943 141.334 0 8 4 221.776 21 (extract) 0.42 semi-solid (freeze-dried 1.98 ± Ellagic acid 302.19 0.943 141.334 0 8 4 221.776 whole fruit 52 2.87 made into a paste) solid Ellagic acid 302.19 0.943 141.334 0 8 4 221.776 1 40 (extract) liquid 0.98 ± Ellagic acid 302.19 0.943 141.334 0 8 4 221.776 50 (juice) 0.06 Hydroxycinnamic acids 5-O-Caffeoylquinic liquid 354.31 -0.453 164.744 5 9 6 289.267 1 ± 0.2 51 acid (drink) 3-O-caffeoylquinic liquid 0.6 ± acid lactone-O- 415.35 -4.212 179.727 6 11 3 316.048 51 (drink) 0.1 sulfate 4-O-caffeoylquinic liquid 0.7 ± acid lactone-O- 415.35 -4.375 179.727 6 11 3 316.048 51 (drink) 0.1 sulfate 3-O-Feruloylquinic liquid 0.7 ± 368.34 -0.145 153.75 6 9 5 313.795 51 acid (drink) 0.1 173

Appendices

4-O-Feruloylquinic liquid 0.8 ± 368.34 -0.363 153.75 6 9 5 313.795 51 acid (drink) 0.1 5-O-Feruloylquinic liquid 0.9 ± 368.34 -0.145 153.75 6 9 5 313.795 51 acid (drink) 0.1 Stilbenes solid trans-Resveratrol 228.24 2.986 60.684 2 3 3 206.922 (pure 0.833 32 compound) solid trans-Resveratrol 228.24 2.986 60.684 2 3 3 206.922 (pure 0.759 32 compound) solid trans-Resveratrol 228.24 2.986 60.684 2 3 3 206.922 (pure 1.375 32 compound) solid trans-Resveratrol 228.24 2.986 60.684 2 3 3 206.922 (pure 1.5 32 compound) Carotenoids liquid 16.6 ± Lycopene 536.89 9.977 0 16 0 0 601.871 19 (drink) 10.1 liquid 19.8 ± Lycopene 536.89 9.977 0 16 0 0 601.871 19 (drink) 12.4 liquid 15.6 ± Lycopene 536.89 9.977 0 16 0 0 601.871 19 (drink) 13.8 liquid 26.1 ± Lycopene 536.89 9.977 0 16 0 0 601.871 19 (drink) 5.5 liquid 32.6 ± Lycopene 536.89 9.977 0 16 0 0 601.871 19 (drink) 18.6 semi-solid (pure β-Carotene 536.89 9.843 0 10 0 0 591.964 5 47 compound in emulsion)

174

Appendices

liquid β-Carotene 536.89 9.843 0 10 0 0 591.964 5 46 (drink) solid Astaxanthin 596.85 8.596 74.598 10 4 2 612.415 (pure 8 ± 0 44 compound) liquid 6.7 ± Astaxanthin 596.85 8.596 74.598 10 4 2 612.415 45 (drink) 1.2 liquid Canthaxanthin 564.85 9.293 34.142 10 2 0 596.328 12 46 (drink) solid 15.2 ± all-E-Zeaxanthin 568.89 9.39 40.456 10 2 2 608.052 (pure 37 7.6 compound) solid 14.7 ± all-E-Zeaxanthin 568.89 9.39 40.456 10 2 2 608.052 (pure 37 0.9 compound) semi-solid 14.8 ± Lutein 568.89 9.307 40.456 10 2 2 608.077 (pure 55 2.5 compound) Vitamins solid Vitamin C 2.76 ± 176.12 -1.402 107.22 2 6 4 139.71 (pure 24 (ascorbic acid) 0.62 compound) liquid Vitamin B1 265.36 -3.449 75.92 4 5 3 239.76 (pure 0.88 53 (thiamine) compound) solid Vitamin B1 2.34 ± 265.36 -3.449 75.92 4 5 3 239.76 (pure 39 (thiamine) 0.89 compound) solid Vitamin B2 376.40 -0.759 161.563 5 10 5 321.314 (pure 1.5 57 (riboflavin) compound) solid Vitamin B2 376.40 -0.759 161.563 5 10 5 321.314 (pure 1.4 57 (riboflavin) compound) 175

Appendices

solid Vitamin B2 376.40 -0.759 161.563 5 10 5 321.314 (pure 2 57 (riboflavin) compound) liquid 0.3 ± Nicotiamide 122.13 -0.48 55.986 1 3 2 110.159 (pure 48 0.1 compound) liquid 0.5 ± Nicotiamide 122.13 -0.48 55.986 1 3 2 110.159 (pure 48 0.3 compound) solid Nicotiamide 122.13 -0.48 55.986 1 3 2 110.159 (pure 1 ± 0.8 48 compound) solid 1.9 ± Nicotiamide 122.13 -0.48 55.986 1 3 2 110.159 (pure 48 1.2 compound) solid 2.1 ± Nicotinic acid 123.11 0.273 50.191 1 3 1 106.888 (pure 38 1.6 compound) solid 2.8 ± Nicotinuric acid 180.16 -1.773 79.289 3 5 2 155.076 (pure 38 1.4 compound) solid Cobalamin (vitamin 6.83 ± 1270.44 -2.137 464.966 26 27 15 1146.798 (pure 33 B12) 3.19 compound) solid R,R,R-α- 13.5 ± 430.72 9.043 29.462 12 2 1 474.499 (pure 35 Tocopherol 1.4 compound) solid 4.3 ± α-Tocotrienol 424.67 9.089 29.462 9 2 1 455.861 (pure 56 0.7 compound) solid 4.3 ± γ-Tocotrienol 410.64 9.03 29.462 9 2 1 439.3 (pure 56 0.7 compound) δ-Tocotrienol 396.62 8.671 29.462 9 2 1 422.739 solid 3 ± 0.4 56 176

Appendices

(pure compound) solid Phylloquinone 450.71 8.803 34.142 14 2 0 483.869 (whole 8 43 (vitamin K) vegetable) * calculated using the Molinspiration Chemoinformatics calculator # collected from the literature

177

Appendices

S2. Table. The PCv dataset for validation of the predictive model.

Predicted Predicted * Measured Phyto- PSA Tmax_ Tmax_ # Family Source# LogP* M * Tmax (h) Reference# chemical r (Å2) LogP** PSA** (h) (h) Mean SE N Liquid intake Cyanidin 3- Cranberry 1.17 O- (juice) -1.63 419.36 171.23 0.70 1.55 3.30 10 76 arabinoside Cyanidin 3- Cranberry O- -1.63 419.36 171.23 0.70 1.55 1.47 0.17 15 78 (juice) arabinoside Cyanidin 3- Cranberry O- -2.79 449.39 191.46 0.65 1.59 1.27 0.15 15 78 (juice) galactoside Cyanidin 3- Cranberry O- -2.79 449.39 191.46 0.65 1.59 2.30 1.01 10 76 (juice) galactoside Cyanidin 3- Acai berry Anthocyanins -2.79 449.39 191.46 0.65 1.59 2.00 0.22 12 77 O-glucoside (juice) Black Cyanidin 3- currant -2.79 449.39 191.46 0.65 1.59 1.25 0.16 8 75 O-glucoside (extract) Cyanidin 3- Elderberry -2.79 449.39 191.46 0.65 1.59 1.08 0.34 4 62 O-glucoside (extract) Cyanidin 3- Cranberry -2.79 449.39 191.46 0.65 1.59 1.13 0.21 15 78 O-glucoside (juice) Cyanidin 3- Red grape -2.79 449.39 191.46 0.65 1.59 0.50 0.07 9 66 O-glucoside (juice) Red grape Cyanidin 3- and -2.79 449.39 191.46 0.65 1.59 0.92 0.12 10 73 O-glucoside blueberry 178

Appendices

Cyanidin 3- Cranberry -2.79 449.39 191.46 0.65 1.59 1.70 0.19 10 76 O-glucoside (juice) Black Cyanidin 3- currant -3.49 595.53 250.39 0.80 2.60 1.50 0.19 8 75 O-rutinoside (extract) Cyanidin 3- O- Elderberry -4.07 581.5 250.39 0.74 2.36 1.19 0.28 4 62 sambubiosid (extract) e Cyanidin 3- O- Hibiscus -4.07 581.5 250.39 0.74 2.36 1.25 0.06 6 67 sambubiosid (extract) e Delphinidin Red grape 3-O- -3.08 465.39 211.69 0.65 1.48 1.40 0.40 8 88 (juice) glucoside Delphinidin Red grape 3-O- -3.08 465.39 211.69 0.65 1.48 0.63 0.08 9 66 (juice) glucoside Red grape Delphinidin and 3-O- -3.08 465.39 211.69 0.65 1.48 0.97 0.08 10 73 blueberry glucoside (juice) Delphinidin Black 3-O- currant -3.08 465.39 211.69 0.65 1.48 1.50 0.19 8 75 glucoside (extract) Delphinidin Black 3-O- currant -3.08 465.39 211.69 0.81 2.43 1.75 0.37 8 75 rutinoside (extract) Delphinidin 3-O- Hibiscus -4.31 597.5 270.61 0.75 2.20 1.38 0.20 6 67 sambubiosid (extract) e

179

Appendices

Malvidin 3- Cranberry -2.47 493.44 189.7 0.73 2.20 0.93 0.28 15 78 O-glucoside (juice) Malvidin 3- Red grape -2.47 493.44 189.7 0.73 2.20 0.63 0.08 9 66 O-glucoside (juice) Malvidin 3- Red grape -2.47 493.44 189.7 0.73 2.20 1.38 0.08 9 66 O-glucoside (juice) Red grape Malvidin 3- and -2.47 493.44 189.7 0.73 2.20 1.13 0.13 10 73 O-glucoside blueberry (juice) Peonidin 3- Cranberry O- -1.32 433.39 160.24 0.75 1.88 0.90 0.16 10 76 (juice) arabinoside Peonidin 3- Cranberry O- -1.32 433.39 160.24 0.75 1.88 1.27 0.15 15 78 (juice) arabinoside Peonidin 3- Cranberry O- -2.49 463.42 180.47 0.69 1.94 1.47 0.17 15 78 (juice) galactoside Peonidin 3- Cranberry O- -2.49 463.42 180.47 0.69 1.94 1.60 0.82 10 76 (juice) galactoside Peonidin 3- Red grape -2.49 463.42 180.47 0.69 1.94 0.50 0.07 9 66 O-glucoside (juice) Peonidin 3- Cranberry -2.49 463.42 180.47 0.69 1.94 1.40 0.21 15 78 O-glucoside (juice) Peonidin 3- Red wine -2.49 463.42 180.47 0.69 1.94 1.38 0.08 9 66 O-glucoside (drink) Red grape Peonidin 3- and -2.49 463.42 180.47 0.69 1.94 0.92 0.05 10 73 O-glucoside blueberry (juice) Peonidin 3- Cranberry -2.49 463.42 180.47 0.69 1.94 4.70 3.00 10 76 O-glucoside (juice) 180

Appendices

Petunidin 3- Red wine -2.78 479.41 200.7 0.69 1.94 1.25 0.05 9 66 O-glucoside (drink) Petunidin 3- Red grape -2.78 479.41 200.7 0.69 1.81 0.50 0.07 9 66 O-glucoside (juice) Red grape Petunidin 3- and -2.78 479.41 200.7 0.69 1.81 1.15 0.08 10 73 O-glucoside blueberry (juice) Petunidin 3- Red grape -2.78 479.41 200.7 0.69 1.81 1.30 0.5 8 88 O-glucoside (juice) Pure Epicatechin 1.37 290.27 110.37 0.85 1.08 1.00 0.50 9 59 compound Cranberry Epicatechin 1.37 290.27 110.37 0.85 1.08 2.60 0.76 10 76 (juice) Cocoa Epicatechin 1.37 290.27 110.37 0.85 1.08 2.00 NA 6 89 (drink) Green tea Epicatechin 1.37 290.27 110.37 0.85 1.08 0.78 0.19 12 82 (drink) Epicatechin Cocoa 1.37 290.27 110.37 0.85 1.08 1.50 NA 6 84 and catechin (drink) Epicatechin Green tea Flavanols 2.54 442.38 177.13 1.44 1.72 1.00 0.30 5 87 gallate (drink) Epigallocatec Green tea 1.08 306.27 130.6 0.84 1.01 0.72 0.12 12 82 hin (drink) Epigallocatec Green tea 1.08 306.27 130.6 0.84 1.01 0.50 0.00 5 87 hin (drink) Epigallocatec Green tea 2.25 458.38 197.36 1.40 1.61 1.69 0.50 12 82 hin gallate (drink) Epigallocatec Green tea 2.25 458.38 197.36 1.40 1.61 0.60 0.10 5 87 hin gallate (drink) Procyanidin Cocoa 2.58 578.53 220.75 1.90 3.02 2.00 NA 6 89 B2 dimer (drink) Cranberry Flavonols Myricetin 1.39 318.24 151.58 0.91 0.91 1.70 0.41 10 76 (juice) 181

Appendices

Cranberry Quercetin 1.68 302.24 131.35 0.92 0.97 1.40 0.41 10 76 (juice) 4- Cranberry Hydroxybenz 1.37 138.12 57.53 0.63 0.60 0.80 0.22 10 76 (juice) oicacid Montmoren Protocatechu cy tart 0.88 154.12 77.75 0.60 0.56 1.00 NA 12 71 ic acid cherry (juice) Montmoren Protocatechu cy tart 0.88 154.12 77.75 0.60 0.56 1.00 NA 12 71 ic acid cherry Hydroxy- (juice) benzoic acids Cranberry Vanillic acid 1.19 168.15 66.76 0.65 0.68 0.70 0.09 10 76 (juice) Montmoren cy tart Vanillic acid 1.19 168.15 66.76 0.65 0.68 2.00 NA 12 71 cherry (juice) Montmoren cy tart Vanillic acid 1.19 168.15 66.76 0.65 0.68 1.00 NA 12 71 cherry (juice) Cranberry Caffeic acid 0.94 180.16 77.75 0.64 0.67 0.80 0.16 10 76 (juice) Cranberry Ferulic acid 1.25 194.19 66.76 0.69 0.82 0.60 0.09 10 76 Hydroxy- (juice) cinnamic acids p-Coumaric Cranberry 1.43 164.16 57.53 0.67 0.72 0.60 0.09 10 76 acid (juice) Cranberry Sinapic acid 1.26 224.21 76 0.73 0.93 0.60 0.16 10 76 (juice) Hydroxy- 3,4- Cranberry phenylacetic 0.39 168.15 77.75 0.57 0.62 0.80 0.22 10 76 Dihydroxy- (juice) acids

182

Appendices

phenylacetic acid cis-Piceid (cis- Red wine Resveratrol 1.2 390.39 139.84 1.01 1.67 1.13 0.15 10 83 (drink) 3-O- glucoside) Stilbenes trans-Piceid (trans- Red wine Resveratrol 1.2 390.39 139.84 1.01 1.67 0.86 0.19 10 83 (drink) 3-O- glucoside) Pure β-Carotene 9.843 536.89 0 11.12 16.43 7.00 0.83 8 72 compound Fucoxanthin Kombu 8.49 658.92 96.36 9.44 16.22 4.00 NA 18 69 ol (extract) Marigold Lutein flower 9.307 568.89 40.46 10.05 14.28 19.00 2.05 8 72 (extract) Lycopene Tomato 9.977 536.89 0 11.59 16.43 5.00 NA 5 65 (drink) Carotenoids Lycopene Tomato 9.977 536.89 0 11.59 16.43 5.00 NA 5 65 (drink)

Tomato Lycopene 9.977 536.89 0 11.59 16.43 5.00 NA 5 65 (drink)

Lycopene Tomato 9.977 536.89 0 11.59 16.43 5.00 NA 5 65 (drink)

183

Appendices

Semi-solid intake Cyanidin 3- Acai berry -2.79 449.39 191.46 1.15 1.11 2.17 0.11 12 77 O-glucoside (pulp) Red grape and Cyanidin 3- blueberry -2.79 449.39 191.46 1.15 1.11 1.13 0.13 10 73 O-glucoside (juice and puree) Raspberry Cyanidin 3- (homogenis -2.79 449.39 191.46 1.15 1.11 1.00 NA 10 74 O-glucoside ed whole fruit) Montmoren Cyanidin 3- cy tart O- cherry -4.79 757.67 329.54 2.91 3.91 2.00 NA 12 85 glucosylrutin (whole oside Anthocyanins fruit) Montmoren Cyanidin 3- cy tart O- cherry -4.79 757.67 329.54 2.91 3.91 4.00 NA 12 85 glucosylrutin (whole oside fruit) Montmoren cy tart Cyanidin 3- cherry -3.49 595.53 250.39 1.52 2.17 1.50 NA 12 85 O-rutinoside (whole fruit) Montmoren cy tart Cyanidin 3- cherry -3.49 595.53 250.39 1.52 2.17 2.00 NA 12 85 O-rutinoside (whole fruit)

184

Appendices

Red grape Delphinidin and 3-O- blueberry -3.08 465.39 211.69 1.29 1.03 1.12 0.13 10 73 glucoside (juice and puree) Red grape and Malvidin 3- blueberry -2.47 493.44 189.7 1.03 1.69 1.25 0.13 10 73 O-glucoside (juice and puree) Red grape and Peonidin 3- blueberry -2.49 463.42 180.47 1.03 1.43 1.15 0.10 10 73 O-glucoside (juice and puree) Red grape and Petunidin 3- blueberry -2.49 463.42 180.47 1.15 1.32 1.18 0.17 10 73 O-glucoside (juice and puree) Kiwi fruit Ascorbic Vitamins (whole -1.4 176.12 107.22 0.75 0.24 2.00 NA 9 63 acid fruit) Solid intake Cyanidin 3- Pure Anthocyanins -2.79 449.39 191.46 3.95 2.85 1.81 0.16 8 64 O-glucoside compound Green tea Epicatechin 1.37 290.27 110.37 1.62 2.66 1.40 0.11 8 80 (extract) Green tea Epicatechin 1.37 290.27 110.37 1.62 2.66 1.65 NA 20 79 (extract) Flavanols Epicatechin Green tea 2.54 442.38 177.13 1.59 3.06 1.20 0.14 8 80 gallate (extract) Epicatechin Green tea 2.54 442.38 177.13 1.59 3.06 1.53 NA 20 79 gallate (extract) 185

Appendices

Epigallocatec Green tea 1.08 306.27 130.6 1.65 2.46 1.30 0.11 8 80 hin (extract) Epigallocatec Green tea 1.08 306.27 130.6 1.65 2.46 1.64 NA 20 79 hin (extract) Epigallocatec Green tea 2.25 458.38 197.36 1.58 2.83 0.90 0.07 8 80 hin gallate (extract) Epigallocatec Green tea 2.25 458.38 197.36 1.58 2.83 1.36 NA 20 79 hin gallate (extract) Pomegranat Ellagic acid e 0.94 302.19 141.33 1.67 2.25 1.97 0.41 20 68 Hydroxy- (extract) benzoic acids Pomegranat Ellagic acid e 0.94 302.19 141.33 1.67 2.25 1.42 0.24 20 68 (extract) Pure Resveratrol 2.99 228.25 60.68 1.63 2.94 1.58 0.27 15 58 compound trans- Pure 2.99 228.25 60.68 1.63 2.94 1.12 0.26 12 81 Resveratrol compound cis-Piceid (cis- Pure Resveratrol 1.2 390.39 139.84 1.63 3.23 1.00 0.06 10 83 Stilbenes compound 3-O- glucoside) trans-Piceid (trans- Pure Resveratrol 1.2 390.39 139.84 1.63 3.23 0.88 0.08 10 83 compound 3-O- glucoside) Pure β-Carotene 9.843 536.89 0 15.27 15.45 36.00 NA 12 91 compound Pure Carotenoids β-Carotene 9.843 536.89 0 15.27 15.45 36.00 NA 12 91 compound Pure β-Carotene 9.843 536.89 0 15.27 15.45 35.00 NA 12 91 compound 186

Appendices

Pure Crocetin 4.63 328.41 74.6 2.01 3.98 4.8 0.32 10 90 compound Pure Crocetin 4.63 328.41 74.6 2.01 3.98 4 0.28 10 90 compound Pure Crocetin 4.63 328.41 74.6 2.01 3.98 4.6 0.32 10 90 compound Pure Lutein 9.307 568.89 40.46 11.27 13.23 36.8 4.15 18 61 compound Ascorbic Pure -1.4 176.12 107.22 2.53 1.72 3.00 NA 9 63 acid compound Pure Calcitriol 5.56 416.65 60.68 2.48 6.25 4.00 NA 10 60 compound Vitamins Phenprocou Pure 4.09 280.32 50.44 1.83 3.89 2.25 NA 24 70 mon compound Phylloquinon Pure 8.803 450.71 34.14 8.64 8.62 8.8 0.38 20 86 e compound * calculated using the Molinspiration Chemoinformatics calculator ** calculated from the predictive model # collected from the literature

187

Appendices

S3. Table. The PHv_fasted dataset for validation of the predictive model. All compounds were taken as solid form. Predicted Predicted Measured Pharma- PSA* T _ T _ T #(h) ceutical LogP* M * max max max Reference# r (Å2) LogP** PSA** compound Mean SE N (h) (h) Nadolol 1.15 309.41 81.95 1.64 3.50 3.00 1 10 123 Nifedipine 3.07 346.34 110.46 1.64 3.33 2.80 0.5 20 101 Odanacatib 3.96 525.57 99.06 1.80 7.38 3.00 1 44 136 Tramadol 3.18 263.38 32.70 1.65 4.11 2.20 0.8 20 96 Amisulpride 1.56 369.49 101.74 1.60 3.88 2.30 1.9 8 111 Levofloxacin -0.26 361.37 75.01 1.96 4.53 1.84 0 12 139 Mirodenafil 3.23 531.68 120.77 1.66 6.50 3.25 2.75 6 124 Talinolol 3.36 363.50 86.11 1.68 4.23 2.10 0.81 14 142 Aleglitazar 4.73 437.52 81.80 2.05 5.86 3.00 0 6 137 Mitiglinide 3.37 315.41 57.61 1.68 4.26 0.36 0.16 8 144 Risperidone 2.96 410.49 64.17 1.62 5.95 1.20 0.7 25 125 Sonidegib 5.89 485.51 63.70 2.72 8.05 3.00 1 6 147 Imidafenacin 2.34 319.01 60.92 1.58 4.22 1.50 0.45 6 127 Olanzapine 3.47 312.44 35.16 1.70 4.92 4.92 1.8 6 112 Imatinib 3.89 493.62 86.28 1.78 7.10 3.63 1.2 30 129 Arbidol 4.86 477.42 54.71 2.10 8.30 0.75 0.75 20 119 Ruxolitinib 1.83 306.37 83.19 1.59 3.43 1.00 0.5 16 134 Risperidone 2.96 410.49 64.17 1.62 5.95 1.00 0.31 10 120 Ritonavir 6.93 720.96 145.78 3.81 11.63 4.20 2.2 10 135 Rosuvastatin 2.11 481.55 140.92 1.58 4.62 2.46 0.65 12 117 Capravirine 4.45 451.38 83.05 1.95 6.14 3.00 1.5 5 99 Proguanil 2.05 253.74 83.79 1.58 2.77 2.80 1.2 15 135 CH 4987655 3.22 565.29 100.12 1.66 8.59 1.00 1 40 115 RO 5068760 4.11 647.44 128.20 1.84 9.80 2.00 2 6 114 Tinidazole -0.06 247.28 97.79 1.89 2.45 2.15 0.47 50 100 Pantoprazole 1.95 383.38 86.35 1.58 4.57 2.27 0.52 28 103 Moxifloxacin 0.39 401.44 83.80 1.77 5.00 1.00 1.5 16 116 Tacrolimus 4.26 804.03 178.38 1.88 12.90 1.37 0.64 15 97

188

Appendices

Ethionamide 1.46 166.25 38.92 1.61 2.67 1.70 0.87 16 92 Metopimazine 2.73 445.61 85.41 1.60 5.90 0.87 0.5 6 110 Sulfonamide -0.29 172.21 86.19 1.97 1.97 1.98 0.51 8 122 Ketoconazole 3.77 531.44 69.08 1.75 9.32 1.40 0.1 10 121 Miconazole 5.72 416.14 27.06 2.59 7.89 2.60 0.3 12 121 Nifedipine 3.07 346.34 110.46 1.64 3.33 1.00 0 10 126 Spironolactone 3.03 416.58 3.03 1.63 9.35 1.70 1 9 128 Ciramadol 2.56 249.35 43.69 1.59 3.60 0.90 1 14 93 Cyclosporin 3.61 1202.64 278.78 1.72 31.47 2.00 2 11 118 S 3304 2.80 464.57 99.26 1.61 5.77 2.00 1 6 140 MK 462 1.39 269.35 49.75 1.61 3.74 1.60 1.2 12 102 Itraconazole 5.32 705.65 104.73 2.34 14.58 3.90 1 28 95 Indinavir 2.51 613.80 118.02 1.59 9.20 0.90 0.5 10 143 Celecoxib 3.61 381.38 77.99 1.72 4.80 2.44 0.83 24 130 Rifampin 2.62 822.95 220.16 1.60 10.39 2.18 1.44 14 132 Topiramate 0.16 339.37 115.57 1.83 3.12 1.75 1.17 28 105 Ethambutol 0.35 204.31 64.51 1.78 2.60 2.46 0.86 14 133 Isoniazid -0.97 137.14 68.01 2.27 1.94 1.02 1.1 14 148 Cilostazol 3.40 369.47 3.40 1.68 7.72 3.00 3 23 98 Febuxostat 3.68 316.38 83.22 1.74 3.57 0.80 0.6 24 113 Apixaban 1.78 459.51 110.77 1.59 5.22 3.00 3 43 108 Ziprasidone 4.05 412.95 48.47 1.82 6.70 3.60 0.7 8 109 Cefprozil -1.68 389.43 132.96 2.74 3.38 1.20 0.8 12 94 Eprosartan 4.89 424.52 92.42 2.12 5.16 1.50 1.5 6 138 Amoxicillin -1.35 365.41 132.86 2.50 3.07 1.86 0.3 16 106 Ampicillin -0.87 349.41 112.73 2.22 3.32 1.49 0.5 16 106 Pyrazinamide -0.71 123.12 68.88 2.14 1.82 1.71 1.19 14 131 Bosentan 4.16 551.63 145.67 1.85 5.91 3.50 4.5 16 104 Telbivudine -1.43 242.23 104.56 2.55 2.29 3.00 3 24 145 Tenoxicam 0.76 337.38 99.60 1.70 3.46 1.70 1.2 6 107 Eltrombopag 5.12 442.48 116.82 2.23 4.67 3.50 2 25 151 Fluconazole -0.12 306.28 81.66 1.91 3.47 3.08 0.79 12 146 * calculated using the Molinspiration Chemoinformatics calculator; ** calculated from the predictive model; # collected from the literature

189

Appendices

S4. Table. The PHv_fed dataset for validation of the predictive model. All compounds were taken as solid form. Predicted Predicted Measured Pharma- PSA* T _ T _ T #(h) ceutical LogP* M * max max max Reference# r (Å2) LogP** PSA** compound Mean SE N (h) (h) Odanacatib 3.96 525.57 99.06 1.80 7.38 4 1 44 136 Amisulpride 1.56 369.49 101.74 1.60 3.88 1.7 0.6 8 111 Mitiglinide 3.37 315.41 57.61 1.68 4.26 1.97 0.81 8 144 Rosuvastatin 2.11 481.55 140.92 1.58 4.62 4.28 1.35 8 117 Pantoprazole 1.95 383.38 86.35 1.58 4.57 6.287 4.41 28 103 Moxifloxacin 0.39 401.44 83.80 1.77 5.00 2.5 1.5 16 116 Tacrolimus 4.26 804.03 178.38 1.88 12.90 6.47 3.04 15 97 Ethionamide 1.46 166.25 38.92 1.61 2.67 2.6 0.94 16 92 Metopimazine 2.73 445.61 85.41 1.60 5.90 1.5 0.5 6 110 Sulfonamide -0.29 172.21 86.19 1.97 1.97 2.22 1.07 8 122 Ketoconazole 3.77 531.44 69.08 1.75 9.32 2.3 0.3 12 121 Nifedipine 3.07 346.34 110.46 1.64 3.33 3.5 0.5 10 126 Spironolactone 3.03 416.58 3.03 1.63 9.35 1.4 1.1 9 128 Ciramadol 2.56 249.35 43.69 1.59 3.60 2.1 2.1 14 93 S 3304 2.80 464.57 99.26 1.61 5.77 5 1 6 140 MK 462 1.39 269.35 49.75 1.61 3.74 2.9 1.4 12 102 Itraconazole 5.32 705.65 104.73 2.34 14.58 4.5 1.1 28 95 Indinavir 2.51 613.80 118.02 1.59 9.20 2.8 1.9 16 143 Celecoxib 3.61 381.38 77.99 1.72 4.80 3.42 1.28 24 130 Rifampin 2.62 822.95 220.16 1.60 10.39 4.43 1.12 14 132 Ethambutol 0.35 204.31 64.51 1.78 2.60 3.21 1.34 14 133 Isoniazid -0.97 137.14 68.01 2.27 1.94 1.93 1.62 14 148 Cilostazol 3.40 369.47 3.40 1.68 7.72 3.5 1.5 23 98 Febuxostat 3.68 316.38 83.22 1.74 3.57 1.9 0.9 24 113 Apixaban 1.78 459.51 110.77 1.59 5.22 4 5 43 108 Ziprasidone 4.05 412.95 48.47 1.82 6.70 4.5 1.4 8 109 Cefprozil -1.68 389.43 132.96 2.74 3.38 2 1.5 12 94 Eprosartan 4.89 424.52 92.42 2.12 5.16 3 1.5 4 138

190

Appendices

Artemisinin 3.32 282.34 54.01 1.67 3.82 1.78 1.23 16 150 Amoxicillin -1.35 365.41 132.86 2.50 3.07 2.4 0.41 16 106 Ampicillin -0.87 349.41 112.73 2.22 3.32 2.48 0.74 14 106 Pyrazinamide -0.71 123.12 68.88 2.14 1.82 3.09 1.75 16 131 Bosentan 4.16 551.63 145.67 1.85 5.91 4 4 24 104 Telbivudine -1.43 242.23 104.56 2.55 2.29 3 1 24 145 Posaconazole 4.33 700.79 115.72 1.91 13.24 5.5 1.76 20 149 Tenoxicam 0.76 337.38 99.60 1.70 3.46 3.8 2 6 107 Eltrombopag 5.12 442.48 116.82 2.23 4.67 4 8 25 151 Fluconazole -0.12 306.28 81.66 1.91 3.47 3.5 1 12 146 * calculated using the Molinspiration Chemoinformatics calculator ** calculated from the predictive model # collected from the literature

191

Appendices

Appendix V. Supporting Information for Chapter 3

Table S1. Proximate composition of plant extracts. Data represent the mean of analysis in duplicate ± standard error. Total Crude Total Total solids Total lipid Total ash Plant nitrogen protein carbohydrate Total phenolic Group (% total (% total (% total extract (% total (% total (% total (mg GAE/g) weight) weight) weight)* weight) # weight)## weight)** Broccoli 83.35 ± 0.28 5.49 ± 0.02 34.31 ± 0.11 4.6 ± 0.03 7.04 ± 0.09 37.41 ± 0.10 20.27 ± 0.97

Carrot 96.46 ± 0.22 1.20 ± 0.03 7.53 ± 0.17 1.27 ± 0.02 3.33 ± 0.04 84.33 ± 4.40 7.55 ± 0.25 Red sweet 96.78 ± 0.04 0.90 ± 0.02 5.65 ± 0.13 1.09 ± 0.01 1.86 ± 0.02 88.18 ± 3.82 5.43 ± 0.15 potato Rhubarb 88.00 ± 0.18 2.45 ± 0.01 15.31 ± 0.01 0.48 ± 0.01 7.10 ± 0.04 65.12 ± 1.88 22.68 ± 0.79 Project extracts Squash 99.65 ± 0.36 2.38 ± 0.01 14.85 ± 0.05 1.65 ± 0.04 5.70 ± 0.05 77.50 ± 3.11 13.58 ± 0.42

Eggplant 94.25 ± 0.35 2.11 ± 0.01 13.20 ± 0.03 0.79 ± 0.01 3.94 ± 0.03 76.32 ± 2.02 14.29 ± 0.29

Kale 84.91 ± 1.73 4.36 ± 0.03 27.26 ± 0.17 7.02 ± 0.2 7.53 ± 0.05 43.06 ± 2.66 47.31 ± 1.21 Vietnamese 88.94 ± 0.20 3.77 ± 0.04 23.58 ± 0.25 2.68 ± 0.02 16.34 ± 0.08 46.32 ± 1.17 102.17 ± 5.54 coriander Blueberry 99.34 ± 0.12 0.77 ± 0.04 5.20± 0.05 1.7 ± 0.05 0.06 ± 0.01 92.38 ± 18.23 29.81 ± 0.04

Green tea 98.78 ± 0.27 4.59 ± 0.07 31.00 ± 0.83 2.50 ± 0.06 0.86 ± 0.04 64.42 ± 4.72 218.74 ± 3.68 Reference extracts Olive leaf 99.43 ± 0.17 1.51 ± 0.03 9.45 ± 0.67 44.64 ± 0.9 2.21 ± 0.02 43.13 ± 1.42 53.55 ± 0.75

Tomato 99.18 ± 0.47 2.09 ± 0.10 14.10 ± 0.07 3.30 ± 0.01 0.30 ± 0.01 81.48 ± 3.35 12.30 ± 0.03 #sum of organic plus inorganic nitrogen ##based on total nitrogen × 6.25 *calculated from sum of minerals analysis (Table S2) **calculated by difference

192

Appendices

Table S2. Mineral analysis of plant extracts. Data represent the mean of analysis in duplicate ± standard error. Plant Al B Ca Cd Co Cr Cu Fe K Mg Mn Mo Na P Pb S Se Si Ti Zn extract (µg/g) (µg/g) (%) (µg/g) (µg/g) (µg/g) (µg/g) (µg/g) (%) (%) (µg/g) (µg/g) (%) (%) (µg/g) (%) (µg/g) (µg/g) (µg/g) (µg/g) 184.5 26.4 0.4 0.8 0.7 3.2 7 54.3 4.3 0.2 14.8 1.7 0.2 0.8 1.1 23.2 9.9 57.7 < 10 < 10 Broccoli ± 44.5 ± 3.2 ± 0 ± 0.4 ± 0.1 ± 2.5 ± 1.7 ± 14.2 ± 0 ± 0 ± 0.3 ± 0.4 ± 0 ± 0 ± 0 ± 3.5 ± 1.8 ± 1.4

43.2 8.8 0.1 0.2 0 1 3.3 11.8 2.3 0.1 5.6 0.5 0.4 0.3 0.1 7.2 5.4 11.1 < 10 < 10 Carrot ± 3.4 ± 0.8 ± 0 ± 0 ± 0 ± 0.4 ± 0.1 ± 2.2 ± 0 ± 0 ± 0.1 ± 0.1 ± 0 ± 0 ± 0 ± 0.2 ± 0.4 ± 0.6

Red sweet 71.5 4.7 0.1 0.1 0.2 0.2 7.3 53.4 1.3 0.1 21.4 0.3 0.2 0.1 0.1 52.4 9.6 9.9 < 10 < 10 potato ± 3.1 ± 0.3 ± 0 ± 0 ± 0.1 ± 0.1 ± 0.3 ± 2.6 ± 0 ± 0 ± 0.4 ± 0 ± 0 ± 0 ± 0 ± 2.1 ± 0.4 ± 0.2

29.6 37.8 0.1 0.4 0.2 0.3 11.1 17.6 6.1 0.3 58.8 0.3 0.2 0.3 0.1 144.5 5.5 58.1 < 10 < 10 Rhubarb ± 6 ± 0.2 ± 0 ± 0.2 ± 0 ± 0.1 ± 0.2 ± 1 ± 0 ± 0 ± 0.1 ± 0 ± 0 ± 0 ± 0 ± 0.5 ± 0.1 ± 1.2

6.3 23.2 0.2 0.1 0.4 0.4 9.9 32.8 4.2 0.3 22.9 2.8 0 0.7 0.2 364.5 3.5 44.9 < 10 < 10 Squash ± 1 ± 0.3 ± 0 ± 0 ± 0 ± 0.1 ± 0.1 ± 3.1 ± 0 ± 0 ± 0.1 ± 0.2 ± 0 ± 0 ± 0 ± 6.5 ± 0.1 ± 0.4

20.6 16.5 0.1 0.4 0 3.4 28.8 38.9 3.2 0.1 13.6 0.5 0 0.4 0.1 93.4 13 17 < 10 < 10 Eggplant ± 7.6 ± 0.1 ± 0 ± 0 ± 0 ± 1.1 ± 0 ± 8.1 ± 0 ± 0 ± 0.2 ± 0.1 ± 0 ± 0 ± 0 ± 2.4 ± 0.6 ± 0.1

163 13.8 0.7 0.5 0.8 9.2 9.6 117.8 4.1 0.3 374 1.1 0.5 0.6 1.3 230 13.3 54.4 < 10 < 10 Kale ± 49 ± 2.5 ± 0 ± 0 ± 0.1 ± 8.8 ± 0.2 ± 54.2 ± 0 ± 0 ± 4 ± 0.1 ± 0 ± 0 ± 0 ± 6 ± 2 ± 2.6

Vietnamese 118 35.7 0.2 0.4 3.1 10 50.8 139.5 10.9 1.8 473.5 0.8 0.1 1.5 1.8 311.5 21.6 216 < 10 < 10 coriander ± 7 ± 0.1 ± 0 ± 0.1 ± 0.3 ± 0.4 ± 0.2 ± 6.5 ± 0.1 ± 0 ± 0.5 ± 0.2 ± 0 ± 0 ± 0 ± 1.5 ± 1 ± 0

41.8 15.8 0.04 0.2 0.1 0.9 2.8 22.4 0 0 14.3 0 0 0 0 22.7 2.7 5.9 < 10 < 10 Blueberry ± 4.9 ± 8.4 ± 0 ± 0.2 ± 0 ± 0.4 ± 0.2 ± 6 ± 0 ± 0 ± 0.8 ± 0 ± 0 ± 0 ± 0 ± 36.2 ± 2.1 ± 0.7

1600 11.9 0.43 0.1 1.1 4.4 12.6 435.5 0 0 1945 0 0 0 0 276 13.8 35.1 < 10 < 10 Green tea ± 70 ± 0.4 ± 0 ± 0 ± 0.1 ± 0.2 ± 0.4 ± 7.5 ± 0 ± 0 ± 75 ± 0 ± 0 ± 0 ± 0 ± 21 ± 1.6 ± 1

184 21.1 2.1 0.04 1.4 1 12.3 171 0 0 39.1 0 0 0 0 452 9 20.5 < 10 < 10 Olive leaf ± 8 ± 0.7 ± 0 ± 0.1 ± 0.1 ± 0 ± 0 ± 8 ± 0 ± 0 ± 3.9 ± 0 ± 0 ± 0 ± 0 ± 2 ± 0.7 ± 0.4

38.8 18.1 0.1 0.2 0 1.3 9.1 49.7 0 0 11.1 0 0 0 0 1470 0.9 20.3 < 10 < 10 Tomato ± 0.9 ± 0.2 ± 0 ± 0.1 ± 0.2 ± 0 ± 0.1 ± 0.6 ± 0 ± 0 ± 0.1 ± 0 ± 0 ± 0 ± 0 ± 100 ± 0 ± 0

193

Appendices

Figure S1. Comparison of primary metabolites and secondary metabolites of plant extracts. Box plots shows (A) Tmax if consumed in drink form and (B) Tmax if consumed in solid form, (C) molecular mass and (D) log P.

A B

12 12

10 10

) )

h h

( ( 8 8

x x

a a

m m

T T

6 6

a a

m m

s 4 s 4

a a

l l

P P 2 2

0 0 Primary metabolites Secondary metabolites Primary metabolites Secondary metabolites

C D

1400 7 1200 6 5 s 1000

a 4

r 800

a g

u 2

L c 600

e l 1

M 400 0 200 -1 -2 0 Primary metabolites Secondary metabolites Primary metabolites Secondary metabolites

194

Appendices

Appendix VI. Supporting Information for Chapter 4

Figure A1. Calibration curve to determine absorptivity at 254 nm. Absorptivity was determined as the slope of the linear regression of peak are at 254 nm and mass of (a) and (b) standards, (c) project extracts and (d) reference extracts. Absorptivity and linear fit R2 are given in Table S1. a b 2.5e+6 2.5e+7 Ascorbic acid Gallic acid Bromophenol blue 2.0e+6 Thiamine HCl 2.0e+7 Curcumin 1.5e+6 1.5e+7 1.0e+6 1.0e+7 5.0e+5

Peak area at 253 nm 253 at area Peak 0.0 nm 254 at area Peak 5.0e+6

0.0 0 1 2 3 4 5 0 5 10 15 20 25 Mass ( g) Mass ( g)

c Broccoli d Blueberry Carrot Cacao Red cabbage Grape seed Red sweet potato Grape skin 2.5e+6 Rhubarb Green tea Eggplant 8e+6 Olive leaf Kale 2.0e+6 Tomato Vietnamese coriander Squash 6e+6 1.5e+6 White zucchini

1.0e+6 4e+6

Peak area at 254 nm 254 at area Peak 5.0e+5 2e+6

Peak area at 254 nm 254 at area Peak 0.0 0 5 10 15 20 0 Mass ( g) 0 2 4 6 8 10 12 14 16 Mass ( g)

195

Appendices

Table A1. Absorptivity of standards and plant extracts at 254 nm. Absorptivity was determined as the slope of the linear regression of peak area at 254 nm and mass of standards and plant extracts as shown in Figure S2. Sample Absorptivity at 254 nm Linear fit R2 (peak area/µg) Standards Ascorbic acid 3669735.39 0.98 Gallic acid 3887152.53 0.99 Thiamine HCl 2114699.02 0.99 Curcumin 52869.52 0.98 Bromophenol blue 995540.96 0.99 Project extracts Broccoli 78882.77 0.99 Carrot 27203.25 0.99 Red cabbage 68730.95 0.99 Red sweet potato 23020.08 0.99 Rhubarb 33995.98 0.99 Squash 101979.32 0.99 White zucchini 57674.41 0.99 Eggplant 45762.15 0.99 Kale 136549.75 0.99 Vietnamese coriander 11787.17 0.99 Reference extracts Blueberry 38690.74 0.99 Cacao 622719.29 0.99 Grape seed 287488.16 0.99 Grape skin 449295.60 0.99 Green tea 1428046.72 0.99 Olive leaf 1052769.03 0.99 Tomato 86289.08 0.99

196

Minerva Access is the Institutional Repository of The University of Melbourne

Author/s: Selby-Pham, Sophie Ngoc Bich

Title: Predictive modelling of upper intestinal absorption of dietary phytochemicals to optimise for health benefits in humans

Date: 2017

Persistent Link: http://hdl.handle.net/11343/198273

File Description: Predictive modelling of upper intestinal absorption of dietary phytochemicals to optimise for health benefits in humans

Terms and Conditions: Terms and Conditions: Copyright in works deposited in Minerva Access is retained by the copyright owner. The work may not be altered without permission from the copyright owner. Readers may only download, print and save electronic copies of whole works for their own personal non-commercial use. Any use that exceeds these limits requires permission from the copyright owner. Attribution is essential when quoting or paraphrasing from these works.