INVESTIGATING the ROLE of TOMATO PHYTOCHEMICALS THROUGH TARGETED and UNTARGETED METABOLOMICS DISSERTATION Presented in Partial F

INVESTIGATING THE ROLE OF TOMATO PHYTOCHEMICALS THROUGH TARGETED AND UNTARGETED METABOLOMICS

DISSERTATION

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University

Morgan Julienne Cichon

Graduate Program in Food Science and Technology

The Ohio State University

2015

Dissertation Committee:

Steven J. Schwartz, Ph.D., Advisor

Steven K. Clinton, M.D., Ph.D.

Earl H. Harrison, Ph.D.

Luis E. Rodriguez-Saona, Ph.D.

Copyrighted by

Morgan Julienne Cichon

2015

ABSTRACT

Epidemiological studies have demonstrated a link between an increased consumption of tomatoes and a decreased risk of chronic diseases, such as cancer and cardiovascular disease. As the carotenoid lycopene is the predominant pigment in tomatoes and is an efficient singlet oxygen quencher, many have focused on lycopene as the main bioactive compound in tomatoes responsible for these observed health benefits. However, tomatoes contain many potential bioactive components and research has also suggested that some of the protective effects of tomatoes might be due to the combination of phytochemicals and their metabolites. Determining the biochemical changes tomato phytochemicals undergo and elicit in vivo is important for understanding the biological function of this fruit.

The objective of this dissertation work was to utilize both targeted and untargeted metabolomics to investigate lycopene and other tomato phytochemicals in foods, preclinical models, and humans. One of the benefits of metabolomics is that it allows for continuity that is necessary in the area of functional food research. This technique can be applied first to chemically profile foods and then to profile biological samples collected from dietary interventions with those foods. The primary objective was accomplished

ii through the metabolomic investigation of 1) lycopene oxidative metabolism in humans using 13C-labeled lycopene, 2) phytochemical differences between red and tangerine tomato juices intended for human clinical trials, and 3) the impact of lycopene and tomatoes on the plasma metabolome of mice.

From this research, lycopene cis isomers and lycopene 1,2-epoxide were identified in plasma by mass spectrometry as potential oxidative metabolites, while nuclear magnetic resonance spectroscopy metabolomics experiments have suggested the presence of small, polar catabolites of lycopene in the urine. Red and tangerine tomato juices intended for use as functional foods in human clinical interventions with prostate cancer patients were found to differ significantly in a number of phytochemicals and metabolites. Many of these compounds have been shown to possess important antioxidant and biological activities and may contribute to the health promoting properties of tomato products in the diet. In mice, several tomato alkaloids were found to increase in plasma after the consumption of tomatoes. Results of this study also revealed that the red tomato fed mice had a unique metabolic profile compared to mice fed either tangerine tomatoes or low carotenoid tomatoes, suggesting differences in biological effect based on variety.

This work demonstrates the utility of metabolomics in food science and nutrition research and supports continued investigation of the synergistic effects of tomato bioactive phytochemicals.

iii

Dedicated to my mom, Debra J. Trantolo

ACKNOWLEDGMENTS

First and foremost, I would like to thank my advisor, Dr. Steven J. Schwartz, for his unwavering support and guidance during all stages of my graduate studies. I appreciate the investment in my growth as a researcher and the exciting opportunity to work in an emerging field. I could not have asked for a better mentor.

I would like to thank my committee members, Drs. Earl H. Harrison, Steven K. Clinton, and Luis E. Rodriguez-Saona, for their interest in my project and insightful comments along the way.

A special thanks to Dr. Ken Riedl for teaching me everything I know about mass spectrometry. His analytical knowledge and expertise has certainly made me a better

scientist.

I would like to thank Lisa and Dan Wampler for their generous support of my research

and commitment to the study of foods and health. Their philanthropy and kindness is

inspiring.

Thank you to all of my lab members, past and present, for providing invaluable scientific

advice and more importantly, much needed comic relief during late nights in the lab.

Finally, I would like to thank my family for their encouragement during this journey and

especially my mom for giving me my first dissertation to read and showing me that the

possibilities are endless for women in science.

VITA

May 2010 ...... B.S. Chemistry; Economics, Summa Cum

Laude, Emory University, Atlanta, GA

September 2010 to present ...... Graduate Research Associate, Department

of Food Science & Technology, The Ohio

State University, Columbus, OH

September 2010 to August 2011 ...... University Fellow, The Ohio State

University, Columbus, OH

January 2015 to present ...... Lisa and Dan Wampler Endowed Fellow for

Foods and Health Research, Department of

Food Science and Technology, The Ohio

State University, Columbus, OH

Publications

Schwartz, S.J., Cooperstone, J.L., Cichon, M.J., von Elbe, J.H., Giusti, M.M. "Colorants." In Fennema’s Food Chemistry 5th Edition. Eds. Srinivasan Damodaran, Kirk L. Parkin, and Owen R. Fennema. Chichester, West Sussex, UK: Wiley-Blackwell. Accepted.

vii

Moran, N.E., Novotny, J.A., Cichon, M.J., Riedl, K.M., Rogers, R.B., Grainger, E.M., Schwartz, S.J., Erdman Jr, J.W., Clinton, S.K. Absorption and distribution kinetics of the 13C-labeled tomato carotenoid phytoene in healthy adults. J Nutr. Accepted.

Moran, N.E., Cichon, M.J., Riedl, K.M., Grainger, E.M., Schwartz, S.J., Novotny, J.A., Erdman Jr, J.W., Clinton, S.K. Compartmental and non-compartmental modeling of 13C- lycopene absorption, isomerization, and distribution kinetics in healthy adults. Am J Clin Nutr. 2015. doi: 10.3945/ajcn.114.103143.

Pumilia, G., Cichon, M.J., Cooperstone, J.L., Giuffrida, D., Dugo, G., Schwartz, S.J. Changes in chlorophylls, chlorophyll degradation products and lutein in pistachio kernels (Pistachia vera L.) during roasting. Food Res Int. 2014. 65(Part B): 193-198.

Tan, S.H., Moran, N.E., Cichon, M.J., Riedl, K.M., Schwartz, S.J., Erdman Jr, J.W., Pearl, D.K., Thomas-Ahner, J.M., Clinton, S.K. Beta-carotene-9’,10’- oxygenase status modulates the impact of dietary tomato and lycopene on hepatic nuclear receptor-, stress-, and metabolism-related gene expression in mice. J Nutr. 2014. 144(4):431-439.

Moran, N.E., Cichon, M.J., Novotny, J.A., Grainger, E.M., Riedl, K.M., Rogers, R.B., Schwartz, S.J., Erdman Jr, J.W., Clinton, S.K. 13C-phytoene from tomato cell suspension cultures for pharmacokinetic studies in healthy adults. FASEB J. 2014. 28:645.15.

Moran, N.E., Cichon, M.J., Riedl, K.M., Grainger, E.M., Schwartz, S.J., Erdman Jr, J.W., Clinton, S.K. (2013). Pharmacokinetics of 13C-lycopene in healthy adults. FASEB J. 2013. 27:38.6.

Kopec, R.E., Cooperstone, J.L., Cichon, M.J., Schwartz, S.J. "Analysis Methods of Carotenoids." In Analysis of Antioxidant-rich Phytochemicals. Eds. Zhimin Xu and Luke R. Howard. Chichester, West Sussex, UK: Wiley-Blackwell, 2012. 105-148.

Hou Y., Xu, L., Cichon, M.J., Lense, S., Hardcastle, K.I., Hill, C.L. A New Family of Sandwich-Type Polytungstophosphates Containing Two Types of Metals in the Central Belt: M′2M2(PW9O34)212− (M′ = Na or Li, M = Mn2+, Co2+, Ni2+, and Zn2+). Inorg Chem. 2010. 49(9): 4125-4132.

Fields of Study

Major Field: Food Science and Technology

viii

TABLE OF CONTENTS

Abstract ...... ii

Acknowledgments...... v

Vita ...... vii

Table of Contents ...... ix

List of Tables ...... xvi

List of Figures ...... xvii

Chapter 1: Literature Review ...... 1

1.1 Tomatoes and Chronic Disease ...... 1

1.1.1.1 Prostate Cancer ...... 1

1.2 Bioactive Tomato Phytochemicals ...... 4

1.2.1 Vitamins ...... 4

1.2.2 Phenolic Acids ...... 5

1.2.3 Flavonoids ...... 6

1.2.4 Carotenoids ...... 9

1.3 Lycopene Metabolism ...... 13 ix

1.3.1 Oxidative Metabolism ...... 13

1.3.1.1 Isomerization ...... 13

1.3.1.2 Epoxidation ...... 15

1.3.2 Lycopene Metabolizing Enzymes ...... 15

1.3.2.1 Lycopene Cleavage Enzymes in Plants ...... 15

1.3.2.2 Lycopene Cleavage Enzymes in Mammals ...... 16

1.3.3 Lycopene Metabolites ...... 18

1.3.3.1 In Vitro ...... 18

1.3.3.2 Preclinical Models ...... 22

1.3.3.3 Humans ...... 22

1.3.4 Biological Activity of Metabolites ...... 24

1.3.4.1 Anti-cancer Activity Ex Vivo and In Vitro ...... 24

1.3.4.2 Retinoid Signaling ...... 26

1.3.4.3 Activity In Vivo ...... 27

1.4 Metabolomics ...... 28

1.4.1 Background ...... 28

1.4.2 Sample Preparation ...... 33

1.4.3 Instrumentation ...... 33

1.4.4 Metabolite Identification ...... 35

1.4.5 Stable Isotope Metabolomics ...... 36

1.4.6 Metabolite Profiling of Foods ...... 37

1.4.6.1 Tomatoes ...... 38

1.4.7 Metabolomics and Nutrition ...... 39

Chapter 2: Specific Aims ...... 43

2.1 Aim 1. Investigate the oxidative metabolism of lycopene in humans using 13C-

labeling...... 43

2.2 Aim 2. Utilize a metabolomics approach to identify phytochemical and

metabolite differences in red and tangerine tomato juices...... 44

2.3 Aim 3. Evaluate the influence of lycopene and different tomato diets on the

plasma metabolomes of mice...... 44

Chapter 3: Investigation of the Oxidative Metabolism of Lycopene in Humans Using 13C-

Labeling ...... 46

3.1 Abstract ...... 47

3.2 Introduction ...... 48

3.3 Materials and Methods ...... 50

3.3.1 Chemicals ...... 50

3.3.2 13C-Lycopene Dose ...... 50

3.3.3 Plasma and Urine Collection ...... 51

3.3.4 Plasma Extraction for Analysis of Lycopene Isomers and Oxidative

Metabolites ...... 52

3.3.5 HPLC-MS Plasma Analysis ...... 53

3.3.6 Metabolomics Analysis of Plasma ...... 54

3.3.7 Urine NMR Metabolomics Experiments ...... 56

3.4 Results and Discussion ...... 57

3.4.1 Lycopene Isomerization ...... 57

3.4.2 Apo-lycopenals Are Not Major Metabolic Products of Lycopene ...... 62

3.4.3 LC-MS Metabolomics Analysis of Lycopene Oxidative Metabolism ...... 62

3.4.4 Investigation of Urinary Lycopene Metabolites Using 13C NMR-Based

Metabolomics ...... 69

3.5 Conclusions ...... 75

Chapter 4: A Metabolomic Study of Carotenoids and Other Phytochemicals in Tomato

Juices Intended for Human Clinical Trials ...... 76

4.1 Abstract ...... 77

4.2 Introduction ...... 78

4.3 Materials and Methods ...... 80

4.3.1 Chemicals ...... 80

4.3.2 Tomato Juices ...... 81

xii

4.3.3 Preparation of Lipophilic Extract ...... 81

4.3.4 Preparation of Polar/Semi-Polar Extract ...... 82

4.3.5 Non-polar Phytochemical Analysis by LC-QTOF-MS (APCI+) ...... 82

4.3.6 Polar/Semi-Polar Phytochemical Analysis by LC-QTOF-MS (ESI-) ...... 83

4.3.7 Data Processing and Statistical Analysis ...... 84

4.3.8 Compound Identification ...... 85

4.4 Results and Discussion ...... 86

4.4.1 Lipophilic Phytochemicals Correlated with the Red and Tangerine Tomato

Juices…...... 86

4.4.2 The Tomato Juices Differed in Other Pigments and Lipids besides

Carotenoids ...... 94

4.4.3 Significant Differences in Polar Tomato Phytochemicals ...... 97

4.4.4 Identification of Differentiating Polar Phytochemicals ...... 99

4.4.5 Red and Tangerine Tomato Juices Differ in Levels of Potentially Bioactive

Phenolic Compounds ...... 101

4.5 Acknowledgments ...... 104

Chapter 5: Influence of Lycopene and Different Tomato Diets on the Plasma

Metabolomes of Mice ...... 105

5.1 Abstract ...... 106

5.2 Introduction ...... 107 xiii

5.3 Materials and Methods ...... 109

5.3.1 Materials and Chemicals ...... 109

5.3.2 Animals, Diets, and Study Design ...... 109

5.3.3 Preparation of Plasma for Metabolomics Analysis ...... 111

5.3.4 Plasma Analysis by UHPLC-QTOF-MS ...... 112

5.3.5 Data Processing and Statistical Analysis ...... 112

5.3.6 Compound Identification ...... 113

5.4 Results and Discussion ...... 114

5.4.1 Animal Ages and Weights ...... 114

5.4.2 Assessment of Metabolomics Data Quality ...... 117

5.4.3 The Whole Tomato Has a Greater Impact on the Metabolome than

Lycopene Alone ...... 119

5.4.4 Differentiating Metabolites Include Tomato Alkaloids ...... 125

5.4.5 Tomato Varieties Alter the Plasma Metabolome in Unique Ways ...... 132

5.5 Conclusions ...... 139

References ...... 140

Appendix A: Considerations Regarding the Stability, Formation, and Analysis of

Apocarotenoids ...... 168

A.1 Evaluation of Extraction Procedure ...... 168

xiv

A.2 Effect of Nitrogen Drying ...... 171

A.3 Stability in Extract ...... 173

A.4 Analysis of Foods and Dietary Supplements ...... 175

A.5 Analysis of Mouse Liver ...... 179

A.6 Conclusions ...... 182

LIST OF TABLES

Table 1. Average 13C-labeled isomers detected in the plasma of 8 subjects at various time

points (± SD)...... 59

Table 2. Identified carotenoids and other lipophilic phytochemicals differentiating red and tangerine tomato juices...... 89

Table 3. Identification of triglycerides based on APCI in-source ion fragmentation...... 96

Table 4. Identified polar and semi-polar phytochemicals differentiating red and tangerine

tomato juices...... 100

Table 5. List of plasma metabolites altered with the consumption of either the lycopene

or red tomato diet ...... 124

Table 6. Plasma extraction methods for the analysis of β-apo-13-carotenone...... 169

Table 7. β-apo-13-carotenone concentrations in baby foods (± SD)...... 178

Table 8. β-apo-13-carotenone concentrations in dietary supplements (± SD)...... 178

Table 9. β-apo-13-carotenone concentrations in the livers of mice and their diets...... 181

xvi

LIST OF FIGURES

Figure 1. Structures of tomato flavonoids...... 8

Figure 2. Structures of common tomato carotenoids...... 10

Figure 3. Structures of the lycopene oxidation products apo-lycopenals...... 20

Figure 4. Positioning of metabolomics at the end of the “omics cascade.” ...... 30

Figure 5. Number of publications found on metabolomics between 1999 and 2014 using

SciFinder...... 32

Figure 6. Mass spectrum showing the normal isotope distribution for unlabeled lycopene

and the reverse isotope distribution for the 13C-labeled lycopene...... 55

Figure 7. Overlaid extracted ion chromatograms of plasma 13C-lycopene at different time

points (normalized to the all-trans peak) showing isomerization...... 61

Figure 8. Mass spectra of the 13C-labeled lycopene (A) and lycopene epoxide (B)

detected with the IROA ClusterFinder software...... 64

Figure 9. UV-Vis spectra of all-trans-lycopene (top) and the lycopene epoxide (bottom).

...... 66

Figure 10. Average appearance of the 13C-labeled lycopene (left) and the 13C-labeled

lycopene epoxide (right) in the plasma of 6 subjects (± SEM)...... 68

xvii

Figure 11. 1D 13C NMR spectra comparing 10x concentrated urine at baseline and after

the 13C-labeled lycopene dose...... 72

Figure 12. 1D INADEQUATE NMR spectrum of 10x concentrated urine post 13C- lycopene dose...... 74

Figure 13. Heat map from hierarchal clustering analysis (Euclidean distance and Ward’s

linkage rule) performed on significantly different compounds in the red and tangerine

tomato juices with a fold change > 2 for the lipophilic fraction (A) and polar/semi-polar

fraction (B)...... 91

Figure 14. Retention time versus mass plot of significantly different non-polar

compounds detected in the tomato juices (P < 0.05; fold change > 2) showing the

compound clusters (“metabolite streaks”) in the data...... 93

Figure 15. Overlaid extracted ion chromatograms of compounds detected in the tomato

juices illustrating the complexity of the data...... 98

Figure 16. Average starting ages of mice used in the metabolomics analysis (± SEM). 115

Figure 17. Average starting weights (A) and weight changes (B) of mice used in the

metabolomics analysis (± SEM)...... 116

Figure 18. Principle component analysis (PCA) scores plots showing the clustering of the

QC samples (blue) in positive (A) and negative (B) ionization modes...... 118

Figure 19. Overlaid extracted ion chromatograms showing the thousands of metabolites

detected in the mouse plasma by LC-QTOF-MS (ESI+)...... 120

xviii

Figure 20. PCA scores plots comparing plasma from the control, lycopene, and red tomato fed mice analyzed in positive (A) and negative (B) ionization modes following a one-way ANOVA...... 122

Figure 21. Extracted ion chromatograms for the plasma metabolites tomatidine (A) and trigonelline (B) in the control (top), lycopene (middle), and red tomato (bottom) fed mice.

...... 126

Figure 22. Structures of the tomato alkaloids α-tomatine, tomatidine, pimpifolidine, and trigonelline...... 128

Figure 23. PCA scores plots comparing plasma from the control, lycopene, red tomato, tangerine tomato, and low carotenoid tomato fed mice analyzed in positive (A) and negative (B) ionization modes...... 133

Figure 24. Comparison of mean ion intensities (± SEM) for tomatidine (A), trigonelline

(B), and pimpifolidine (C) identified in the plasma of the mice on the 5 different diets.

...... 135

Figure 25. Venn diagram comparing ESI+ significantly different metabolites detected in each of the tomato groups...... 137

Figure 26. Mass versus retention time plot of ESI+ plasma metabolites altered by the red tomato only, the tangerine and low carotenoid tomatoes only, and all three tomatoes. . 138

Figure 27. EICs for apo-13-carotenone in plasma extracted using methods (1), (2), and

(3)...... 170

Figure 28. EICs for apo-13-carotenone in plasma showing the effect of an extended dry down step in the extraction method...... 172

xix

Figure 29. EICs of apo-13-carotenone in a newly redissolved plasma extract and the same extract after sitting in the autosampler for 7 hr...... 174

Figure 30. EICs for apo-13-carotenone in sweet potato, squash, and carrot baby foods.176

Figure 31. EICs for apo-13-carotenone in a β-carotene supplement, vitamin A supplement, and multi-vitamin...... 177

Figure 32. TIC showing presence of retinol in a representative mouse liver sample (top) and EIC of apo-13-carotenone in liver (bottom)...... 180

CHAPTER 1: LITERATURE REVIEW

1.1 Tomatoes and Chronic Disease

Research has suggested that diets rich in tomatoes are protective against cardiovascular disease (Sesso et al. 2003; Willcox et al. 2003) and certain cancers (Giovannucci 1999).

These protective effects are believed to be partially driven by the antioxidant and anti- inflammatory properties associated with tomatoes (Burton-Freeman and Reimers 2011).

Sesso et al. (2003) reported a decreased risk for cardiovascular disease in women consuming greater than seven servings of tomatoes and/or tomato products per week.

Epidemiological evidence has also demonstrated a correlation between the consumption of tomatoes and a decreased risk of certain cancers, including those of the prostate, breast, cervix, ovary, endometrium, lung, bladder, oral cavity, esophagus, stomach, colon, and pancreas (Giovannucci 1999). Arguably the most extensively researched of these relationships is that between tomatoes and prostate cancer.

1.1.1.1 Prostate Cancer

A number of epidemiological studies have demonstrated a protective effect of tomatoes or the tomato phytochemical lycopene against prostate cancer (Giovannucci et al. 1995;

Mills et al. 1989; Tzonou et al. 1999). The largest study to date on this relationship is a prospective cohort study with male health professionals in the United States. Data from 1

this study showed men consuming two to four servings of tomatoes or tomato sauce per

week had a significantly lower incidence of prostate cancer compared to those who rarely consumed these foods (Giovannucci et al. 1995). In a case control study conducted in

Greece, the consumption of cooked tomatoes, but not raw tomatoes, was associated with a decreased risk of prostate cancer (Tzonou et al. 1999). Authors predict that increasing consumption of cooked tomatoes from twice a week to 4 times per week can decrease prostate cancer risk by approximately 15%. While lycopene is believed to be responsible for the health benefits attributed to tomatoes, it was not measured in either of the aforementioned studies. In a meta-analysis of 11 case-control studies and 10 cohort/ nested case-control studies, a high consumption of tomatoes was correlated with a decreased risk of prostate cancer, with cooked tomatoes having a slightly larger effect

(20% decrease) than raw tomatoes (10% decrease) when considering all studies together

(Etminan et al. 2004). In a separate phase II randomized clinical trial conducted by

Kucuk et al. (2001), 26 men with localized prostate cancer were given either 15 mg of lycopene twice daily or no supplementation for 3 weeks before prostatectomy. Results from the study suggest that lycopene may inhibit the progression of prostate cancer, as demonstrated by a decrease in high-grade prostatic intraepithelial neoplasia and an increase in tumor-free margins in prostate cancer patients supplemented with lycopene.

Effects on prostate-specific antigen (PSA) levels and connexin 43 (gap junctional gene) expression were also reported, but were not significant.

At the same time, some studies have reported no significant correlation between prostate cancer and tomato- or lycopene-rich diets (Bosetti et al. 2004; Cohen et al. 2000; Hayes et al. 1999; Kirsh et al. 2006). In attempt to better understand the racial disparity in prostate cancer between black and white men, Hayes et al. (1999) conducted a population-based case-control study with recently diagnosed prostate cancer patients.

Food frequency questionnaires were used to measure dietary intake of lycopene sources, including raw tomatoes, tomato sauce, tomato juice, and watermelon. A moderate decrease in risk of advanced prostate cancer was observed with increased consumption of raw tomatoes, but no significant correlations were observed with any of the other lycopene-rich food sources. In another case-control study with men from the Seattle, WA area, researchers found no significant associations found between prostate cancer and raw tomatoes, cooked tomatoes, or total lycopene intake (Cohen et al. 2000). However, this study focused on a low risk age group (< 65 years old) and it is possible that diet is not a strong risk factor in this population where genetics likely play a larger role in the incidence of prostate cancer. Inconsistent findings from epidemiological studies demonstrate the need for controlled clinical trials to investigate the proposed relationship between lycopene and prostate cancer.

Zu et al. (2014) suggested that some of the inconsistencies may be due the increased number of asymptomatic and localized prostate cancer diagnoses in the PSA screening era. The investigators sought to address this issue by examining lycopene intake in relation to not only total cancer incidence, but also the incidence of lethal prostate cancer

using data from The Health Professionals Follow-up Study. Zu et al. (2014) report a

significant correlation between lycopene intake and a reduced risk of lethal prostate

cancer, as well as tumor angiogenesis. These results suggest that lycopene may be

involved in slowing the progression of prostate cancer and that the stage and type of

prostate cancer should be considered when analyzing data from epidemiological studies.

1.2 Bioactive Tomato Phytochemicals

Little is known about the mechanism through which tomatoes have a health promoting

and cancer protective effect. Tomatoes contain numerous potential bioactive components

including vitamins, phenolic acids, flavonoids, and carotenoids (Beecher 1998). Research

has suggested that many of these components and their metabolites act synergistically in

vivo (T. W.-M. Boileau et al. 2003; Canene-Adams et al. 2007; Fuhrman et al. 2000;

Stahl et al. 1998).

1.2.1 Vitamins

Tomatoes contain moderate levels of vitamins C and E, which are essential micronutrients and good antioxidants (Rock et al. 1996). Research has suggested that vitamin C (ascorbic acid) in tomatoes may work synergistically with other antioxidants in

the tomato to reduce oxidative stress and inflammation (Jacob et al. 2008). However,

vitamin C is heat labile and the content in tomatoes has been found to decrease with

thermal processing (Gahler et al. 2003).

Vitamin E is a lipid-soluble vitamin present in tomatoes in the form of α-tocopherol. It is an efficient scavenger of peroxyl radicals (Burton and Traber 1990) and has been shown to act in combination with lycopene to reduce low-density lipoprotein (LDL) oxidation in humans (Fuhrman et al. 2000). Additionally, α-tocopherol has been reported to act synergistically with ascorbic acid, which is hydrophilic, to scavenge free radicals within lipid bilayers (Niki 1991).

1.2.2 Phenolic Acids

Tomatoes contain a number of phenolic acids, including ferulic acid, caffeic acid, and chlorogenic acid (Slimestad and Verheulb 2009). These compounds are widely distributed in nature and are believed to be responsible for some of the cancer protective effects associated with a variety of fruits and vegetables (Weng and Yen 2012). Phenolic acids are efficient free radical scavengers, which is believed to be one of the predominant mechanisms through which they exert a biological effect in humans

(Kikuzaki et al. 2002; Rice-Evans et al. 1996; Srinivasan et al. 2007). Ferulic, caffeic, and chlorogenic acids have all been shown to inhibit tumor promotion in mice (M.-T.

Huang et al. 1988). Ferulic acid specifically has been reported to protect against a number of conditions including inflammation, cancer, neurodegeneration, and diabetes

(Srinivasan et al. 2007). Chlorogenic acid is an ester formed from caffeic acid and quinic acid and is metabolized by the gut microflora to smaller phenolic compounds (Gonthier et al. 2003). Research has suggested that chlorogenic acid may protect against carcinogenesis by modulating phase II detoxifying enzymes (Feng et al. 2005). It has also

been shown to significantly attenuate intestinal glucose absorption (Johnston et al. 2003;

Welsch et al. 1989).

1.2.3 Flavonoids

The predominant flavonoids in the tomato are naringenin, quercetin, and kaempferol

(Figure 1), which are generally found in various glycosylated forms in the fruit

(Slimestad and Verheulb 2009). Epidemiological studies have suggested an inverse

relationship between flavonoid consumption and lung cancer incidence (Knekt et al.

1997), dementia (Commenges et al. 2000), and mortality from coronary heart disease

(Hertog et al. 1993). Flavonoids have demonstrated antioxidant capabilities in vitro,

which have been hypothesized to translate into protective effects in vivo (Pietta 2000).

While a number of studies have reported cancer cell growth inhibition with flavonoids in

vitro, results in preclinical models have been inconsistent and human clinical studies have

been limited (Kuo 1997; C. S. Yang et al. 2001). Campbell, King, Harmston, Lila, &

Erdman (2006) found that the aglycones naringenin, quercetin, and kaempferol inhibited

the growth of mouse liver cancer cells and human prostate cancer cells, but that the

glycosylated forms had no effect.

Quercetin specifically has been reported to reduce blood pressure in hypertensive rats fed

10 mg/kg of the flavonoid for 5 wk (Duarte et al. 2001). Additionally, in humans,

quercetin supplementation for two wk inhibited LDL oxidation (Chopra et al. 2000).

However, studies have also shown no effect (Exon et al. 1998) or a negative effect

(Pereira et al. 1996) of quercetin on colon carcinogenesis in rats. More preclinical and human studies are needed to better understand the biological implications of tomato flavonoids at physiologically relevant doses.

Figure 1. Structures of tomato flavonoids.

1.2.4 Carotenoids

Carotenoids are a group of tetraterpene molecules found widely in nature. Over 600 carotenoids have been described, but lycopene, β-carotene, phytoene, and phytofluene

(Figure 2) are the primary forms found in the traditional red tomato. Phytoene and phytofluene are the first carotenoids formed in the biosynthetic pathway of these compounds in plants (Cunningham and Gantt 1998). Unlike most carotenoids, phytoene and phytofluene are colorless due to their shortened chromophores. While phytoene and phytofluene are minor carotenoids in tomatoes, research has suggested that they are more bioavailable than other carotenoids, such as lycopene (Cooperstone et al. 2015; Moran,

Clinton, et al. 2013). Additionally, these carotenoids have been shown to accumulate in tissues of rats following the consumption of a tomato supplemented diet (Campbell et al.

2007). While they are not as effective antiradicals as lycopene, phytoene and phytofluene have relatively high antioxidant capacities for the limited number of conjugated double bonds they possess (Martínez et al. 2014). They have also been reported to protect against

UV-light –induced skin damage (Aust et al. 2005), limit LDL oxidation (Shaish et al.

2008), and inhibit breast cancer cell proliferation (Hirsch et al. 2007).

Figure 2. Structures of common tomato carotenoids.

Like phytoene and phytofluene, β-carotene is a minor carotenoid in the red tomato. β-

carotene is a provitamin A carotenoid and is centrally cleaved in vivo by β-carotene

15,15ʹ dioxygenase 1 (BCO1) to produce 2 molecules of retinol via retinal (Paik et al.

2001). Enzymatic cleavage can also occur eccentrically to produce various β- apocarotenals and β-apocarotenones (G. Tang et al. 1991). Despite the necessary function of vitamin A in human growth and development, the Beta-Carotene and Retinol Efficacy

Trial (CARET) revealed an increase in lung cancer incidence in smokers supplemented with large doses of β-carotene (Omenn et al. 1996). The oxidation products β-apo-13- carotenone and β-apo-14ʹ-carotenal have been shown to act as retinoic acid receptor antagonists and may contribute to some of the negative biological effects associated with large doses of β-carotene in smokers (Eroglu et al. 2012). These cleavage products are present in foods (Fleshman et al. 2011) and can also form under autooxidative conditions

(Handelman et al. 1991) so the route of exposure/formation in humans remains unclear

(see Appendix A).

Lycopene is the most abundant carotenoid found in tomatoes and is responsible for the red color often associated with this fruit. It is a highly unsaturated C40 hydrocarbon, but

unlike β-carotene, lycopene is acyclic and its lack of a β-ionone ring means it cannot be cleaved to form vitamin A. The structure of lycopene allows for many different theoretical mono- and poly-cis isomers. However, the all-trans configuration is thermodynamically favored and is the form most commonly found in nature (Britton

1995).

With 11 conjugated double bonds, lycopene is inherently unstable and prone to degradation by heat, light, oxygen, and metal ions (Henry et al. 1998; Philip and Francis

1971; Scita 1992). Research has revealed, though, that lycopene is the most efficient biological carotenoid singlet oxygen quencher (Di Mascio et al. 1989) and it is hypothesized that these potent antioxidant capabilities may translate into important in vivo effects. As the oxidation of biomolecules is believed to play a role in the initiation and progression of many cancers (Ames and Gold 1997; Hussain et al. 2003), dietary antioxidants have the potential to modulate disease risk. In fact, protective effects of lycopene against cancer initiation and progression have been demonstrated in vitro

(Hantz et al. 2005; Hwang and Bowen 2004; Obermüller-Jevic et al. 2003) and in rodent models of prostate carcinogenesis (Astorg et al., 1997; Wargovich et al., 2000; Pannellini et al., 2010). It is not clear whether these effects are directly linked to the function of lycopene as an antioxidant. Lycopene has also been shown to induce apoptosis (L. Tang et al. 2005), inhibit cell cycle progression (Cheng et al. 2007), increase gap junction communication (Livny et al. 2002), and alter androgen status (Campbell, Stroud, et al.

2006). Such studies support lycopene as a health promoting compound, but the biochemical mechanisms responsible for these effects are not well-defined.

1.3 Lycopene Metabolism

1.3.1 Oxidative Metabolism

Studying lycopene metabolism is a critical step in understanding the biological function

of this compound. More research is suggesting that metabolites of lycopene may be

responsible for the health benefits associated with lycopene and tomatoes (Ford and

Erdman 2012; Mein et al. 2008). Currently little is known about the identity of these

metabolites and how they are formed in vivo. It is also not clear whether reactions

involved in this process are chemical, enzymatic, or a combination of the two. One

hypothesis is that lycopene degradation progresses first through isomerization and then

epoxidation, followed by oxidative cleavage (Caris-Veyrat et al. 2003).

1.3.1.1 Isomerization

Lycopene in tomatoes is found predominantly (> 90%) in the all-trans configuration, but

over 50% of lycopene found in human plasma and tissues is in various cis configurations

(Stahl et al. 1992). It has been hypothesized that cis isomers may be more protective than

all-trans due to differences in free radical scavenging abilities (Müller et al. 2011). Some

researchers suggest that biological samples have a high percentage of cis isomers,

because this form is more bioavailable than all-trans (Unlu et al. 2007). Due to their

spatial configuration, cis isomers are more soluble in oil and less likely to aggregate than

the all-trans form so it is possible that they are more readily incorporated into mixed

micelles and absorbed across the intestinal brush border membrane. Enhanced

bioaccessibility of cis isomers compared to all-trans has been demonstrated through in

vitro micellarization and Caco-2 intestinal absorption experiments (Failla et al. 2008). In

an in vivo study, A. C. Boileau, Merchen, Wasson, Atkinson, & Erdman (1999) fed

lymph-cannulated ferrets 40 mg of lycopene that was 9.5% cis and reported 8.3%, 31.1%,

58.8%, and 78.9% cis isomers in the stomach, small intestine, intestinal mucosa, and lymph secretion, respectively. Their results suggest that cis isomers are more readily transported to the intestinal brush border membrane and preferentially incorporated into chylomicrons. In another study, a tomato sauce rich in all-trans lycopene and a tomato sauce rich in lycopene cis isomers were fed to 12 healthy subjects in a cross-over design

(Unlu et al. 2007). Total, total cis-, and all-trans-lycopene responses were higher with the cis-lycopene rich sauce, suggesting a difference in bioavailability based on isomer

distribution.

Other researchers report that isomerization is actually occurring in vivo as an early step in lycopene metabolism (Richelle et al. 2010). Isomerization of lycopene under in vitro gastric conditions has also been reported (Re et al. 2001). However, in vivo the food matrix may enhance the stability of all-trans lycopene and limit isomerization during the gastric phase of digestion. If isomerization is occurring in vivo, little is known about the mechanism behind this reaction. It has been hypothesized that isomerization of lycopene may be the result of singlet oxygen quenching (Foote et al. 1970; Stahl and Sies 1993).

This would support the role of lycopene as an antioxidant and protector against oxidative stress.

1.3.1.2 Epoxidation

To date, the presence of a lycopene epoxide has not been reported in animals or humans.

Khachik, Goli, et al. (1992) detected and quantified the 5,6-epoxy lycopene in tomato-

based products. The epoxide was found at approximately 14% of the concentration of

lycopene in raw tomatoes and 4% of the concentration of lycopene in tomato paste.

Khachik et al. (1997) were unable to detect the 5,6-epoxy lycopene in human plasma, but reported the presence of lycopene diols and hypothesized that the epoxide is an intermediate in the formation of these products.

Epoxycarotenoids are found widely in nature, but humans do not appear to readily absorb xanthophyll epoxides from the diet (Barua and Olson 2001). On the other hand, Barua

(1999) found dietary and synthetic 5,6-epoxy-β-carotene to be bioavailable in humans.

The same epoxide of β-carotene was also shown to induce differentiation of NB4 leukemia cells in vitro (Duitsman et al. 1999). More research needs to be done to understand what role epoxidation may play in lycopene metabolism.

1.3.2 Lycopene Metabolizing Enzymes

1.3.2.1 Lycopene Cleavage Enzymes in Plants

Enzymes known as carotenoid cleavage dioxygenases (CCDs) have been identified in plants, including tomatoes. CCDs have been shown to cleave a variety of carotenoid species, including lycopene, to form apocarotenoids (Auldridge et al. 2006). While the

function of CCDs in the plant is still not well defined, the resulting apocarotenoid cleavage products are believed to be involved in the regulation of carotenoid biosynthesis and the synthesis of phytohormones, such as abscisic acid (McQuinn et al. 2015). CCD1 in the tomato has been shown to cleave lycopene at the 9,10 or 9ʹ,10ʹ position, yielding pseudoionone and a C14 aldehyde (Simkin et al. 2004). Tomato CCD1 also cleaves

lycopene at the 5,6 or 5ʹ,6ʹ position to generate the flavor volatile 6-methyl-5-heptene-2-

one (MHO) (Vogel et al. 2008). Similarly, CCD7 and CCD8 have been shown to act

sequentially on β-carotene to first generate β -apo-10ʹ-carotenal followed by β-apo-13-

carotenone (Alder et al. 2012), which has been reported in β-carotene containing fruits

(Fleshman et al. 2011). The generation of apocarotenoids by CCDs in plants may help

explain their presence in foods.

1.3.2.2 Lycopene Cleavage Enzymes in Mammals

Compounds derived from the oxidative cleavage of lycopene have been termed

lycopenoids. So far, two carotenoid cleavage enzymes have been identified in humans, β-

carotene 15,15ʹ-oxygenase (BCO1) and β-carotene 9ʹ,10ʹ-oxygenase (BCO2). BCO1 is involved in the central cleavage of carotenoids and is known to act on β-carotene to form two molecules of retinol via retinal (von Lintig et al. 2005). BCO2 is responsible for the asymmetric cleavage of carotenoids and may be a major enzyme involved in the formation of early lycopene metabolites (Ford and Erdman 2012). BCO2 acts on β- carotene in vivo to generate aldehyde cleavage products called β-apocarotenals. The in

vivo activities of BCO1 and BCO2 on other carotenoids have yet to be determined, but

current in vitro research suggests that these enzymes also act on lycopene.

Cleavage of lycopene by BCO1 would theoretically result in the formation of

acycloretinal, the lycopene equivalent of retinal. Results from Redmond et al. (2001)

suggest that murine BCO1 has activity toward lycopene, but to a much lesser extent than

β-carotene, while Lindqvist & Andersson (2002) reported no activity of purified

recombinant human BCO1 toward lycopene. However, more recently, dela Seña et al.

(2013) found BCO1 to centrally cleave lycopene to yield acycloretinal with a catalytic efficiency similar to that exhibited with β-carotene, which suggests that BCO1 may be involved in lycopene metabolism.

Less research has been reported on the interaction between lycopene and BCO2. If BCO2 acts on lycopene the same way it acts on β-carotene, apo-lycopenals would be the expected products of this reaction. BCO2 knock-out mice fed tomato and lycopene diets were found to have higher serum lycopene concentrations than wild-type mice (Tan et al.

2014). Kiefer et al. (2001) reported murine BCO2 activity toward lycopene, in addition to

β-carotene. In another study, Hu et al. (2006) demonstrated cleavage activity of recombinant ferret BCO2 toward 5-cis- and 13-cis-lycopene, but not all-trans. The primary cleavage product was identified as apo-10ʹ-lycopenal by high performance liquid chromatography with photodiode array detection (HPLC-PDA) and authentic standards, but was not mass confirmed. This research suggests isomer specificity in enzymatic

cleavage of lycopene and supports the hypothesis that isomerization is an early step in metabolism.

1.3.3 Lycopene Metabolites

The identification of in vivo metabolites of lycopene has been challenging. A number of oxidation products have been generated ex vivo, with some demonstrating biological

activity. However, to date, only a select number of metabolites have been reported in

animal and human samples. Additionally, many of these compounds are also found in

lycopene-containing foods so it is unclear as to whether they are absorbed from the diet

or formed endogenously. Identification of biologically relevant lycopene metabolites

represents a gap in what is known about lycopene metabolism.

1.3.3.1 In Vitro

Biological metabolites of lycopene are believed to be chemical and/or enzymatic

cleavage products. For this reason, the oxidation of lycopene has been studied in vitro to

gain insight on the possible in vivo cleavage products of lycopene. The highly conjugated

structure of lycopene makes it especially prone to oxidation and degradation by heat,

light, and metal catalysis. Kim, Nara, Kobayashi, Terao, & Nagao (2001) generated

autoxidation products of lycopene by incubating the solubilized compound at 37 °C for

72 hours. The predominant cleavage products were aldehydes and short-chain carbonyls,

eight of which were identified as 3,7,11-trimethyl-2,4,6,10-dodecatetraen-1-al, 6,10,14-

trimethyl-3,5,7,9,13-pentadecapentaen-2-one, acycloretinal, apo-14ʹ-lycopenal, apo-12ʹ-

18 lycopenal, apo-10ʹ-lycopenal, apo-8ʹ-lycopenal, and apo-6ʹ-lycopenal (Figure 3). The same group also identified (E,E,E)-4-methyl-8-oxo-2,4,6-nonatrienal as another autoxidation product of lycopene (H. Zhang et al. 2003).

Figure 3. Structures of the lycopene oxidation products apo-lycopenals.

Singlet oxygen is very reactive and a known cause of damage to macromolecules in vitro

and in vivo. Photosensitized oxygenation of lycopene by singlet oxygen was shown to

produce apo-6ʹ-lycopenal and 2-methyl-2-hepten-6-one as identified by combined high

performance liquid chromatography (HPLC), mass spectrometry (MS), and nuclear

magnetic resonance (NMR) spectroscopy data (Ukai et al. 1994). In another study, Caris-

Veyrat et al. (2003) produced cleavage products of lycopene through chemical oxidation

with potassium permanganate (KMnO4) and were able to tentatively characterize a series

of apo-lycopenals, apo-lycopenones, and apo-carotendials. In the same study lycopene

was also oxidized by atmospheric oxygen catalyzed by a metalloporphyrin.

Metalloporphyrin models are commonly used to mimic in vivo oxidation by the

xenobiotic metabolizing enzymes, cytochrome P450s (Mansuy 2007). With this method,

Caris-Veyrat et al. (2003) did not detect apo-carotendials, but did detect all of the apo-

lycopenals and apo-lycopenones that were formed with KMnO4, except for apo-7-

lycopenal and apo-5-lycopenone.

Ferreira, Yeum, Russell, Krinsky, & Tang (2004) studied the formation of lycopene

metabolites in rat intestinal mucosa incubated with and without soy lipoxygenase.

Enzymatic metabolites were tentatively identified as 3-keto-apo-13-lycopenone and 3,4-

dehydro-5,6-dihydro-15-apo-lycopenal and oxidation products were tentatively identified

as 2-ene-5,8-lycopenal-furanoxide, lycopene-5,6,5ʹ,6ʹ-diepoxide, lycopene-5,8-

furanoxide, and 3-keto-lycopene-5ʹ,8ʹ-furanoxide. Currently it is unclear how these in vitro experiments translate in vivo.

1.3.3.2 Preclinical Models

Apo-8ʹ-lycopenal and putative apo-12ʹ-lycopenal were reported in the livers of rats fed

14C-labeled lycopene and estimated to be in the concentration of around 250 ng/g (Gajic

et al. 2006). Other polar lycopene metabolites were detected in this study, but were not

characterized. Tan et al. (2014) also detected both apo-8ʹ-lycopenal and apo-12ʹ-

lycopenal, as well as apo-6ʹ-lycopenal, in the livers of mice fed lycopene and tomato supplemented diets, but at lower concentrations (approx. 2 ng/g). While apo-6ʹ-, 8ʹ-, 10ʹ-,

12ʹ-, and 14ʹ-lycopenal were all found in the tomato and lycopene diets, only apo-6′-, 8′-, and 12′-lycopenal were found in the livers of the mice, suggesting that they are preferentially absorbed or formed in vivo. On the other hand, apo-10′-lycopenol was the only lycopene metabolite detected in the lungs of ferrets supplemented with lycopene (Hu et al. 2006). This metabolite was present in similar concentrations at around 2-4 ng/g.

Sicilia et al. (2005) detected three lycopene metabolites in the plasma of preruminant calves after supplementation with lycopene for two weeks. The investigators hypothesized that these are hydrogenation products of lycopene, but their exact structures could not be determined.

1.3.3.3 Humans

The first lycopene metabolite reported in humans was 5,6-dihydroxy-5,6- dihydrolycopene by Khachik, Beecher, Goli, Lusby, & Smith (1992). This metabolite was identified in human plasma and is hypothesized to form from the oxidation of lycopene to the intermediate, lycopene 5,6-epoxide, which is then metabolically reduced

to 5,6-dihydroxy-5,6-dihydrolycopene. Khachik, Beecher, & Smith (1995) have also

found both lycopene 5,6-epoxide and 5,6-dihydroxy-5,6-dihydrolycopene in tomato- based foods, but speculate that the concentrations are too low to contribute to the levels

of 5,6-dihydroxy-5,6-dihydrolycopene in human plasma. This group has also found

epimeric 2,6-cyclolycopene-1,5-diols in human serum and milk (Khachik et al. 1997).

Like 5,6-dihydroxy-5,6-dihydrolycopene, it is believed that 2,6-cyclolycopene-1,5-diols

may be formed from the enzymatic or acidic hydrolysis of lycopene epoxides. However,

in a study where subjects were chronically fed tomato juice, tomato oleoresin, or

lycopene beadlets, plasma 2,6-cyclolycopene-1,5-diol was only significantly increased

with the tomato juice diet and not the tomato oleoresin and lycopene beadlets (Paetau et

al. 1998). While all diets provided approximately the same amount of lycopene, the

tomato juice had a higher concentration of 2,6-cyclolycopene-1,5-diol, suggesting that

this compound may be absorbed from the diet.

Lycopene aldehyde cleavage products, apo-6′-lycopenal, apo-8′-lycopenal, apo-10′-

lycopenal, and apo-12′-lycopenal (Figure 3), have been identified in human plasma of

subjects consuming tomato juice for 8 weeks (Kopec et al. 2010). However, these

metabolites are present in low concentrations and therefore, may be transient

intermediates in lycopene metabolism. Additionally, apo-lycopenals are oxidation products found in various lycopene containing foods, so more research needs to be done to determine if they are formed in vivo.

In an accelerator mass spectrometry (AMS) study by Ross et al. (2011), two subjects

were fed a single dose of 14C-labeled lycopene. Analysis by AMS showed relatively high

levels of radioactivity in the urine of subjects fed a single dose of 14C-labeled lycopene.

Analyte concentrations were too low in this study to characterize specific lycopene

metabolites, but the results suggest that these compounds are very polar, low molecular

weight cleavage products.

1.3.4 Biological Activity of Metabolites

It has been hypothesized that some of the health benefits associated with lycopene are

due to its metabolites (Lindshield et al. 2007). As β-carotene is cleaved to form vitamin A

(central cleavage), lycopene may be similarly cleaved to biologically active compounds

involved in various cell signaling pathways (Ford and Erdman 2012; Lindshield et al.

2007; Mein et al. 2008). Even the eccentric cleavage products of β-carotene, β-

apocarotenoids, have demonstrated biological activity as antagonists of retinoic acid

receptors (Eroglu et al. 2012), which further supports the hypothesis that similar eccentric cleavage products of lycopene, apolycopeneoids, are bioactive.

1.3.4.1 Anti-cancer Activity Ex Vivo and In Vitro

Identified and proposed lycopene metabolites have demonstrated protective effects in various in vitro cancer models. Ford et al. (2011), who reported apo-8′-lycopenal and putative apo-12′-lycopenal in rat livers, found apo-12′-lycopenal, but not apo-8′- lycopenal, to inhibit cell proliferation in DU145 androgen-independent prostate cancer

cells. However, this activity was only demonstrated at supraphysiological concentrations

(15 and 25 µM) and not at physiological concentrations (1 µM) tested. In a separate

study, apo-8′-lycopenal was shown to inhibit metastasis in an invasive human hepatocarcinoma SK-Hep-1 cell line (C.-M. Yang et al. 2012). This metabolite decreased cancer cell invasion and migration to a greater extent than lycopene at the same concentration (10 µM). In another study, Lian, Smith, Ernst, Russell, & Wang (2007) reported that apo-10′-lycopenoic acid (3 µM) was able to inhibit growth of normal, premalignant, and malignant human lung cells. Similar to apo-12′-lycopenal, the growth inhibitory effects of apo-10′-lycopenoic acid appear to be due to the inhibition of cell proliferation, rather than the induction of apoptosis or cytotoxicity. Apo-10′-lycopeneoic acid has yet to be identified in vivo, but has been generated from the in vitro incubation of ferret hepatic homogenates with apo-10′-lycopenal (Hu et al. 2006).

The autoxidation product of lycopene, (E,E,E)-4-methyl-8-oxo-2,4,6-nonatrienal, identified by Zhang et al. (2003), was reported by the group to induce apoptosis in HL-60 human promyelocytic leukemia cells. In a separate study, a lycopene oxidation mixture was shown to inhibit cell growth in HL-60 cells to a greater extent than pure lycopene

(Nara et al. 2001).

Acyclo-retinoic acid has been proposed as another possible metabolite of lycopene.

Acyclo-retinoic acid is the open-chain lycopene equivalent of retinoic acid, which is a known metabolite formed from the central cleavage of β-carotene. Ben-Dor et al. (2001)

evaluated the effect of synthetic acyclo-retinoic acid on human mammary cancer cells

(MCF-7). Acyclo-retinoic acid was found to significantly slow cell growth and inhibit cell cycle progression with a similar potency to both lycopene and retinoic acid. Results from this study suggest that cell cycle inhibition by acyclo-retinoic acid is partially due to a reduction in cyclin D1, which is involved in the regulation of the G1-S phase transition.

Deterioration of gap junctional communication is believed to play a role in the progression of cancer (Saunders et al. 2001). Stahl, von Laar, Martin, Emmerich, & Sies

(2000) compared the effects of lycopene and acyclo-retinoic acid on the induction of gap junctional communication in HFFF2 cells in vitro. Lycopene was found to significantly increase the number of communicating cells at 0.1 µM, while 1.0 µM of acyclo-retinoic acid was needed to achieve the same effect. However, other in vitro oxidation products of

lycopene were able to enhance intercellular communication by gap junctions in rat

epithelial cells to a similar degree as lycopene and retinoic acid (Aust et al. 2003). The

active oxidation product in this study was tentatively identified as the dialdehyde, 2,7,11- trimethyl-tetradecahexane-1,14-dial.

1.3.4.2 Retinoid Signaling

All-trans retinoic acid is responsible for activating hundreds of genes containing retinoic acid response elements (RARE) in their promoters. These genes are involved in many cellular functions, including differentiation, proliferation, and apoptosis. Retinoic acid receptor (RAR) β has even been shown to function as a tumor suppressor gene in human

26 epidermoid lung cancer cells and to induce apoptosis in human mammary epithelial cells

(Houle et al. 1993; Y. Liu et al. 1996). Recent research has suggested a role of lycopene metabolites in the retinoid signaling pathway. Stahl et al. (2000) found that acyclo- retinoic acid was able to transactivate the RAR-β2-promoter in vitro at supraphysiological concentrations (50 µM), compared to 0.1 µM of retinoic acid.

Similarly, Ben-Dor et al. (2001) demonstrated that acyclo-retinoic acid will bind to

RARα, but activates RARE-regulated reporter genes at a 100-fold lower potency than retinoic acid. Apo-10′-lycopenoic acid was also shown to induce RARβ expression in cancerous and noncancerous human lung cells, but again, to a lesser extent than retinoic acid (Lian et al. 2007).

1.3.4.3 Activity In Vivo

In addition to their in vitro experiments with human lung cancer cells, Lian et al. (2007) also evaluated the capacity of apo-10ʹ-lycopenoic acid to inhibit 4-(N-methyl-N- nitrosamino)-1-(3-pyridal)-1-butanone (NNK)-induced lung tumorigenesis in mice. In this study, supplementation with apo-10′-lycopenoic acid (10, 40 and 120 mg/kg diet) for

16 weeks did not affect tumor incidence, but did dose-dependently decrease tumor multiplicity.

Much attention is given to the potential cancer protective effects of lycopene and its metabolites, but other biological activities of lycopene metabolites have also been demonstrated. It has been suggested that apo-10′-lycopenoic acid can modulate lipid

metabolism and maybe protect against steatosis (Chung et al. 2012). In this study, apo-

10′-lycopenoic acid supplementation (40 µg/g diet) increased sirtuin 1 expression and decreased hepatic fat accumulation in obese (ob/ob) mice after 16 weeks.

1.4 Metabolomics

While there is evidence that lycopene and lycopene metabolites contribute to the protective effects of tomatoes, more research is suggesting that some of the health

benefits of this fruit are due to a combination of tomato phytochemicals and their

metabolites (T. W.-M. Boileau et al. 2003; Canene-Adams et al. 2007). For this reason,

studies focusing on the whole tomato and biological effects of tomato consumption are

needed. The following section will focus on metabolomics as a tool for studying foods

and health.

1.4.1 Background

Metabolomics is an emerging area of research that can be defined as the systematic study

of all small molecule metabolites within a biological system and changes in these

metabolites in response to environmental or physiological influences. It joins genomics,

transcriptomics, and proteomics as the newest of the “omics” techniques (Figure 4). In

the same way that genomics is the study of the genome and differences in genes,

metabolomics is the study of the metabolome and differences in metabolites. Metabolites

are the output of biological processes and generally correlate directly with biochemical

activities and phenotype (Patti et al. 2012). It has been estimated that the plant kingdom

alone is comprised of around 200,000 different primary and secondary metabolites (Fiehn

2002). The mammalian metabolome is even more complex as metabolites include both endogenous or physiological metabolites and exogenous metabolites derived from the

diet or surrounding environment (Wishart 2008) (Figure 4).

Figure 4. Positioning of metabolomics at the end of the “omics cascade.” General differences between endogenous and exogenous metabolites are also illustrated. 30

Metabolites can be used as powerful indicators of what is occurring at the genome level, as they often reflect and amplify small changes in gene expression. Additionally, more so than genes or proteins, metabolites are highly sensitive to physiological changes or environmental influences. For this reason, metabolomics has been applied to the study of disease biomarkers, horticulture, food science, and nutrition (Hall et al. 2008; Scalbert et al. 2009; Spratlin et al. 2009; Sumner et al. 2003; Wishart 2008; A. Zhang et al. 2012). In fact, the application of metabolomics as an investigational tool in a wide variety of

disciplines has led to an exponential increase in the number of publications utilizing this

technique over the past decade (Figure 5).

Metabolomics can be either a targeted or untargeted analysis (Patti et al. 2012). Targeted

metabolomics is concerned with the concentrations of specific metabolites in a sample and is a hypothesis driven approach. Untargeted metabolomics is concerned with the global metabolic profile of a sample and is a hypothesis generating approach. While both fall under the umbrella of metabolomics, true metabolic profiling involves the unbiased

detection of all small molecules within a biological sample. The workflow for untargeted

metabolomics starts with the extraction of metabolites from a sample followed by

instrumental analysis to generate raw data, which is processed and statistically analyzed.

Resulting metabolites of interest are then identified and, when possible, further evaluated

within the context of systems biology.

4000

3500

3000

2500

2000

1500

# of Publications 1000

500

Year

Figure 5. Number of publications found on metabolomics between 1999 and 2014 using SciFinder.

1.4.2 Sample Preparation

Metabolomics can be applied to analyze small molecules in cells, tissues, and whole

organisms ranging from plants to animals to microbes. Sample preparation procedures are

specific to the type of sample and the metabolites of interest (lipids versus polar

compounds). Protocols have been published for the metabolomic analysis of urine (Want

et al. 2010), blood (Dunn et al. 2011), mammalian tissues (Wu et al. 2008), and plant

tissues (De Vos et al. 2007), but it is often necessary to adapt such methods to suit the

study objectives and design. For the analysis of neutral lipids, a non-polar extraction

solvent should be used, whereas methanol and water are generally suitable for polar/semi-polar metabolites (Moco et al. 2007).

Regardless of the sample matrix, metabolomics requires careful consideration of sample collection, storage, and handling (Scalbert et al. 2009). As metabolomics is generally employed to investigate relative differences between groups, it is imperative that all samples be treated the same to minimize unwanted variability. When controlling the handling and preparation of samples, biological variability should significantly outweigh technical or analytical variability. Lengthy and complicated extraction procedures can lead to the formation of artifacts and inconsistent results (Moco et al. 2007).

1.4.3 Instrumentation

Advancements in analytical instrumentation are allowing for the detection of thousands of compounds in a single sample. The most common platforms used for metabolomics

33 are mass spectrometry (MS) and nuclear magnetic resonance (NMR) spectroscopy, which both have their respective benefits and limitations as reviewed by Dunn & Ellis (2005).

NMR is a quantitative and reproducible platform, which requires little to no sample preparation. However, NMR has limited sensitivity, which greatly hinders the metabolite coverage possible with this platform. On the other hand, MS is not inherently quantitative and requires some amount of sample preparation, but these limitations are offset by its high sensitivity and dynamic range. Whereas NNR is restricted to the most abundant metabolites in a sample, MS can detect metabolites present at low concentrations in complex biological samples. It can be argued, though, that NMR and MS are most powerful when used as complementary tools in metabolomics (Z. Pan and Raftery 2007).

For example, Atherton et al. (2006) used both MS and NMR platforms to investigate the effect of the nuclear receptor PPARα on the metabolomes of mice. Similarly, both approaches have been employed to study the effects of foods and food components on the metabolome (H. Liu et al. 2015; Ross et al. 2013). NMR can also be used for the structural elucidation of metabolites first detected in an MS-based metabolomics experiment (Moco et al. 2007).

MS-based metabolomics frequently employs chromatographic separation upfront of detection using gas chromatography (GC) or liquid chromatography (LC) methods. GC requires chemical derivatization to analyze non-volatile compounds, whereas LC does not require this additional element of sample preparation. Both chromatographic methods are widely used, but here we will focus on LC-MS-based metabolomics.

Coverage is an important consideration for metabolomics experiments. As mentioned,

MS is a highly sensitive platform, capable of detecting thousands of metabolites.

However, metabolites can vary widely in structure, solubility, and ionization efficiency.

Therefore, metabolites detected are limited by the chromatography upfront (HILIC,

reversed-phase), the ionization source (electrospray ionization (ESI), atmospheric

pressure chemical ionization (APCI)), and mode of ionization (positive, negative). For

this reason, coverage is best when multiple analytical approaches are utilized for a

metabolomics study (Nordström et al. 2008).

1.4.4 Metabolite Identification

Sensitive MS instrumentation allows for the detection of many different metabolites in a

sample, but the identification of these metabolites remains challenging and time-

consuming. Unlike the genome, the metabolome is not well characterized and much of

this is due to the chemical diversity of metabolites (Baker 2011). Metabolite

identification involves the use of a combination of strategies. With accurate mass

measurements, compounds can be searched against metabolite databases, such as the

Human Metabolome Database (HMDB) (Wishart et al. 2013) and METLIN (Smith et al.

2005). The resulting metabolite assignments are tentative, though, and must be confirmed with MS/MS data, chemical standards, and/or NMR.

Advancements in metabolite identification will likely come from the expansion of food, plant, and mammalian metabolite databases. Recently, efforts have been taken to better

understand the composition and complexity of the human urine (Bouatra et al. 2013) and

serum (Psychogios et al. 2011) metabolomes using multi-platform approaches. In 2009,

HMDB contained approximately 6,500 metabolites and now contains around 42,000

metabolites, including physiological metabolites, drug and drug metabolites, toxicants

and pollutants, and food components (Wishart et al. 2013). In addition to mammalian

metabolite databases, a number of databases focusing on specific plants or

phytochemicals have also been developed (Moco et al. 2006; Neveu et al. 2010).

Currently, significant gaps exist in the characterization of the exposome or metabolites arising from environmental exposures, such as diet, pollution, or drugs (Wild 2012). Of particular interest are those metabolites derived from the digestion of food components, which often undergo various phase I and phase II biotransformations. The identification of such metabolites will lead to a better understanding of the relationship between diet and health (Scalbert et al. 2014).

1.4.5 Stable Isotope Metabolomics

Stable isotope labeling is a powerful tool for both targeted and untargeted metabolomics.

In a targeted approach, a stable isotope tracer is used to track the absorption, metabolism, and excretion of a chemical and its known metabolic products. In untargeted metabolomics, a stable isotope is followed in a biological system in an unbiased manner.

A challenge in untargeted stable isotope metabolomics is the processing and interpretation of the resulting data. For 13C-labeling studies, MS-based Isotope Ratio

Outlier Analysis (IROA) and 13C NMR experiments can be useful for the detection of

labeled metabolites (Clendinen et al. 2015). The IROA protocol, which calls for 13C enrichment at 95 and 5%, has been used to identify metabolites in C. elegans altered in response to heat shock (Stupp et al. 2013). X. Huang et al. (2014) have built a specific feature finding platform (X13CMS) for stable isotope LC/MS data. This platform has

been successfully used to track 13C-labeled glucose in HeLa cancer cells (Chen et al.

2014). This same group has also developed the isoMETLIN database to search for stable

isotope labeled isotopologues of METLIN metabolites (Cho et al. 2014). Stable isotope

metabolomics will become more powerful and accessible with further advancements in

software platforms designed for stable isotope metabolomic data.

1.4.6 Metabolite Profiling of Foods

A quickly growing area of research is the application of metabolomics to food science.

Foods are comprised of many chemical components, including secondary plant

metabolites, animal metabolites, and food additives. Metabolomics has been used in the

area of food science to study changes in the chemical composition of foods as affected by

variables, such as geographic origin (J.-E. Lee et al. 2010; Son et al. 2009), processing

and storage (Johanningsmeier and McFeeters 2011; Rudell et al. 2009), and adulteration

(Jandrić et al. 2014; Vaclavik et al. 2012). For example, J.-E. Lee et al. (2010) used

NMR-based metabolomics to evaluate the effects of 3 different geographical areas in

South Korea on compounds in green tea. Their results revealed a significant effect of

growing region on catechins and amino acids in the tea. This technique has also been

37 used to investigate differences in the metabolite profiles of grapes and wines originating from unique regions (Son et al. 2009).

Metabolomics can be applied to assess potentially unwanted chemical changes occurring during the storage of foods. Using untargeted metabolomics, Rudell et al. (2009) were able to identify metabolite changes in apples that preceded symptoms of scald (a peel necrosis). Another important concern in food science is product adulteration. Often adulterated products are indistinguishable based on taste and visual appearance alone.

Jandrić et al. (2014) used an untargeted LC-MS- based metabolomics approach to identify markers of fruit juice adulteration, which were then incorporated into targeted screening methods. These studies demonstrate the potential of metabolomics to enhance current practices and research in the areas of food quality and food safety.

1.4.6.1 Tomatoes

A number of studies have been conducted to profile tomatoes using metabolomics strategies. Schauer, Zamir, & Fernie (2005) compared the metabolites present in the leaves and fruits of different wild tomato species. All six species analyzed had unique metabolite fingerprints, with the fruits being more metabolically diverse than the leaves.

The same group also metabolically profiled tomato introgression lines to identify metabolites associated with particular genomic regions (Schauer et al. 2006). Using GC-

MS the investigators were able to quantitate 74 metabolites including, amino acids, sugars, and organic acids. Metabolomics techniques can also be employed to investigate

changes in tomato-based foods during processing. For example, Capanoglu, Beekwilder,

Boyacioglu, Hall, & de Vos (2008) monitored the changes in the metabolite profiles of tomatoes during the different stages of tomato paste production using LC-QTOF-MS. Of the 3,177 metabolites detected, 40% changed significantly during processing, with the most noticeable changes in metabolite profile occurring during the breaking and pulping steps. Increased interest in metabolic profiling of crops has led to the development and curation of more plant-based metabolite databases. “Respect” is a LC/MS database of plant metabolites compiled and annotated by the RIKEN Plant Science Center in Japan

(Sawada et al. 2012). Tomato specific metabolite resources have also been developed, such as the Metabolome Tomato Database (MoTo DB) (Moco et al. 2006).

1.4.7 Metabolomics and Nutrition

The principles of metabolomics have been used in the pharmaceutical industry for many years to investigate the biological effects of drugs, but now metabolomics is being used in other disciplines such as nutrition and food science to understand the global effects of diet on the metabolomes of healthy and diseased populations (Gibney et al. 2005; Hall et al. 2008; Wishart 2008). Food and nutritional metabolomics have enormous implications for understanding how foods, nutrients, other food-derived compounds and their bioactive metabolites interact with important biochemical pathways in humans to impact health (McGhie and Rowan 2012). The area of nutritional metabolomics is complicated by the inter-individual variability in physiological metabolites, as well as those metabolites produced from the thousands of compounds present in foods. Research has

shown that age and sex (Kochhar et al. 2006; Slupsky et al. 2007) can have an impact on

the metabolome, but that dietary standardization during clinical interventions can help minimize these differences (Winnike et al. 2009).

Unlike pharmaceutical agents, which are designed to target a certain pathway or endpoint, foods modulate a number of different metabolic pathways. Therefore, while some foods have demonstrated a protective effect against disease, understanding the mechanisms underlying these relationships has been challenging. Metabolomics has proven to be an effective and efficient approach for characterizing global metabolic outcomes from nutritional interventions (Mensack et al. 2012; Moazzami et al. 2011;

Takahashi et al. 2014). Knowledge from these types of studies will lead to the discovery of new biomarkers in humans and the identification and development of optimal foods for individuals to prevent disease and promote health.

To date there are no known studies using metabolomics to investigate the biological effects of tomatoes. However, MS-based metabolomics has been successfully used to gain information on the biological activity of other foods and nutrients. For example,

Mensack, McGinley, & Thompson (2012) reported differences in the metabolomic

profiles of control-fed and bean-fed rats using MS metabolomics. Their data revealed

clustering of mammary gland tissue samples based on diet, which was determined to be

partially due to differences in eicosanoid, fatty acid, triacylglycerol, and steroid

metabolism between the groups. Similarly, researchers have used metabolomics to

determine global metabolic alterations in mice resulting from coffee intake (Takahashi et

al. 2014). Results from this study showed an increase in urea cycle metabolites in the livers of mice fed coffee containing diets. Guertin et al. (2015) also evaluated biomarkers

of coffee consumption, but in the serum of subjects with and without colorectal cancer.

This study revealed 3 coffee related metabolites, theophylline, caffeine, and paraxanthine, which were also inversely correlated with colorectal cancer. Another study investigated

the impact of black raspberries on the metabolic profiles of colorectral cancer patients

and identified a number of differences in benzoates and amino acid derivatives (P. Pan et al. 2015). In addition to diet and cancer, metabolomics has also been used to evaluate the relationship between diet and exercise (R. Lee et al. 2010).

Other metabolomics research has employed 1H-nuclear magnetic resonance (NMR)

spectroscopy to investigate the effect of a whole-grain rich diet on the metabolomes of

healthy adults (Ross et al. 2013). After only one week of the intervention, researchers

reported a decrease in urinary metabolites related to protein catabolism, lipid metabolism,

and gut microbial metabolism compared to a control refined grain diet. NMR

metabolomics has also been used to evaluate the biological impact of other potentially

health promoting foods, such as green tea (S. Zhang et al. 2013), soy (Solanky et al.

2003), and cranberries (H. Liu et al. 2015).

The aforementioned MS and NMR studies illustrate the utility of metabolomics in

elucidating the functionality of foods and food components in biological systems. With a

41 metabolomics approach, researchers are not restricted to studying a select number of compounds, but can instead globally analyze thousands of metabolites in tissues and biofluids, leading to the discovery of unanticipated relationships between foods and health.

CHAPTER 2: SPECIFIC AIMS

The primary objective of this work was to utilize both targeted and untargeted metabolomics to investigate lycopene and other tomato phytochemicals in foods,

preclinical models, and humans. This was accomplished through the metabolomic investigation of 1) lycopene oxidative metabolism in humans using 13C-labeled lycopene,

2) phytochemical differences between red and tangerine tomato juices intended for

human clinical trials, and 3) the impact of lycopene and tomatoes on the plasma

metabolome of mice. It was expected that a metabolomics analytical approach would

advance the understanding of the relationship between tomatoes and health.

2.1 Aim 1. Investigate the oxidative metabolism of lycopene in humans using 13C-

labeling.

Metabolites of lycopene may be responsible for some of the health benefits associated

lycopene and tomatoes, but currently little is known about the identity of these

metabolites. Here a combination of stable isotope targeted and untargeted metabolomics

experiments were conducted to investigate lycopene oxidative metabolism in humans.

Both mass spectrometry and nuclear magnetic resonance spectroscopy experiments were

performed to identify 13C-labeled metabolites in plasma and urine. It was expected that a

metabolomics approach would lead to the identification of new biologically relevant

lycopene metabolites in human plasma and urine samples.

2.2 Aim 2. Utilize a metabolomics approach to identify phytochemical and

metabolite differences in red and tangerine tomato juices.

While we know that red and tangerine tomatoes have different carotenoid profiles, it is

important to understand all major phytochemical differences that could contribute to any

in vivo effects observed during human and animal studies with these tomato varieties.

Untargeted metabolomics was used to profile phytochemical differences between red and

tangerine tomato juices intended for human clinical trials. It was expected that a

metabolomics approach would allow us to identify important phytochemical differences

with potential biological relevance in functional food-based clinical trials with the red

and tangerine tomatoes.

2.3 Aim 3. Evaluate the influence of lycopene and different tomato diets on the

plasma metabolomes of mice.

In order to better understand the biological impact of tomato phytochemicals, untargeted

metabolomics was used to 1) compare the effects of lycopene and red tomatoes on the

plasma metabolomes of mice and 2) evaluate whether red, tangerine, and low carotenoid

44 tomato varieties differentially impact the metabolome. It was expected that the tomato diets would have a greater impact on the plasma metabolome than lycopene alone.

CHAPTER 3: INVESTIGATION OF THE OXIDATIVE METABOLISM OF LYCOPENE IN HUMANS USING 13C-LABELING

Morgan J. Cichon1, Nancy E. Moran2, Ken M. Riedl1,2, Robert W. Curley, Jr.3, Steven K. Clinton2,4, Steven J. Schwartz1,2

1 Department of Food Science & Technology, The Ohio State University, Columbus, OH, USA 2 Comprehensive Cancer Center, The Ohio State University, Columbus, OH, USA 3 Division of Medicinal Chemistry and Pharmacognosy, The College of Pharmacy, The Ohio State University, Columbus, OH, USA 4 Division of Medical Oncology, Department of Internal Medicine, The Ohio State University Wexner Medical Center, Columbus, OH, USA

3.1 Abstract

Research has suggested that the carotenoid lycopene is one of the bioactive components of tomatoes. Lycopene metabolites have been hypothesized to be partially responsible for the protective properties associated with lycopene. Given the chemical structure of lycopene, it is likely that metabolism of this compound involves chemical and/or enzymatic oxidation. However, while numerous lycopene metabolites have been proposed, few have actually been identified in vivo. Identifying novel lycopene metabolites is an important step in understanding the function of lycopene and tomatoes in the progression of cancer and other diseases. The objective of this study was to use

13C-labeling to investigate the oxidative metabolism of lycopene in humans. Here a combination of targeted and untargeted mass spectrometry (MS) and nuclear magnetic resonance (NMR) spectroscopy experiments were performed for the detection and identification of lycopene metabolites in human urine and plasma. Lycopene cis isomers and lycopene 1,2-epoxide were identified in plasma by MS as potential oxidative metabolites, while NMR results suggest the presence of small, polar lycopene metabolites in urine.

3.2 Introduction

Epidemiological studies have demonstrated a correlation between the consumption of the tomato carotenoid lycopene and a decreased risk of certain cancers (Clinton 1998).

Lycopene is the pigment responsible for the red color of tomatoes and is one of the predominant phytochemicals in the tomato. It is a C40 hydrocarbon that exists

predominantly in the all-trans configuration in the diet. However, a high proportion of

cis-lycopene isomers are found in human plasma and biological tissues (Clinton et al.

1996; Stahl et al. 1992). In fact, over 50% of lycopene in blood and approximately 80%

of lycopene in prostate tissue has been reported to be in the cis configuration (Clinton et al. 1996), whereas only about 5% of lycopene in tomato products is in the cis form

(Nguyen et al. 2001). Research has been inconclusive as to whether this is the result of greater bioavailability of cis isomers from the diet (Stahl and Sies 1992; Unlu et al. 2007) or in vivo trans-to-cis isomerization as an early step in lycopene metabolism (Re et al.

2001; Richelle et al. 2010).

Lycopene metabolism is also believed to progress through chemical or enzymatic oxidative cleavage of the hydrocarbon chain. As β-carotene is centrally cleaved to generate vitamin A, it has been proposed that oxidative cleavage metabolites of lycopene may be responsible for some of the biological effects associated with this compound

(Lindshield et al. 2007). Potential oxidation products (apo-lycopenoids) have been identified in vitro and include aldehydes, ketones, and acids (Caris-Veyrat et al. 2003; S.

J. Kim et al. 2001). Currently, understanding of lycopene metabolism is limited and only a few potential metabolites have been identified in vivo. Our laboratory has previously identified several apo-lycopenals (apo-6ʹ-, apo-8ʹ-, apo-10ʹ-, and apo-12ʹ-lycopenal) in human plasma (Kopec et al. 2010). These compounds were also found in lycopene- containing foods, including raw tomatoes, tomato sauce, and tomato juice, so it is unclear whether they are absorbed from the diet or are products of in vivo metabolism.

Isotopic labeling has been used as a strategy for investigating carotenoid absorption, metabolism, and excretion in vivo. 14C-labeled lycopene has been used to study the biodistribution of this compound in rats, with the majority of 14C-lycopene administered

being transported in the liver (Zaripheh et al. 2003). Ross et al. (2011) also used 14C-

labeling to investigate the long-term bioavailability of lycopene in human plasma from a

single oral dose. From this study the investigators were able to monitor the metabolism

and excretion of lycopene through the appearance of 14C in the breath, skin, and urine.

14C-labeling is a highly sensitive technique, but as 14C is radioactive, the amount of 14C-

lycopene that can be safely administered is limited and well below the levels found in the

diet. For this reason, it is difficult to obtain concrete structural information for lycopene

metabolites at such low concentrations. Stable isotope labeling provides an alternative

approach, where doses can be administered that are comparable to levels found in the

diet. This strategy has been used to study β-carotene absorption and conversion to

vitamin A (Fleshman et al. 2012; Kurilich et al. 2003), but has not yet been used to study

lycopene.

The objective of this study was to investigate isomerization and oxidative metabolism of

lycopene in humans using a 13C-label. Plasma and urine samples for this analysis were

obtained from a recently conducted study by Moran et al. (2015). We expect labeling

with 13C stable isotope will allow better sensitivity to identify biological metabolites

using targeted and untargeted metabolomics approaches.

3.3 Materials and Methods

3.3.1 Chemicals

Solvents and butylatedhydroxytoluene (BHT) were purchased from Fisher Scientific

(Pittsburgh, PA, USA). Ethanol, acetone, and toluene were HPLC grade, hexane was

Optima grade, and methyl tert-butyl ether (MtBE), methanol, and water were Optima

LC-MS grade.

3.3.2 13C-Lycopene Dose

The 13C-labeled lycopene dose used in the study conducted by Moran et al. (2015) was

isolated from tomato cell suspension cultures as previously described (Moran et al.,

2013). This method allowed for rigorous environmental control and the production of a

highly labeled compound. The 13C-lycopene was extracted and purified from the cell

cultures for use in the human clinical study. The resulting lycopene was over 91% 13C atomic purity and approximately 30% uniformly labeled. The 13C-labeled lycopene was administered in olive oil to aid in micellarization and absorption of the dose. To prevent

the isomerization and degradation of lycopene, the doses were prepared immediately

before being administered to the subjects. The full preparation of the dose has been

described in detail by Moran et al. (2015). In summary, each 13C-labeled lycopene dose

(~10.2 mg) was solubilized in dichloromethane (7.5 mL) and added to 10.2 mL of light

olive oil. The dichloromethane was removed by evaporation under nitrogen. The

resulting 13C-lycopene in oil was transferred to an English muffin to be administered to

subjects. A small sample of the lycopene in olive oil (10 uL) was analyzed by HPLC-

PDA to determine the isomer profile of the dose directly prior to ingestion and a portion

of 13C-lycopene before mixing with oil was reserved for use in standard calibration

curves for plasma 13C-lycopene analyses.

3.3.3 Plasma and Urine Collection

Plasma and urine from a previously conducted clinical trial by Moran et al. (2015) were used for the targeted and untargeted metabolomics analyses. Relevant details of the study are summarized here. The study was conducted in compliance with the ethical standards of and was approved by The Ohio State University Institutional Review Board

(#2009C0104), and written informed consent was obtained from all subjects. Briefly, the

13C-labeled lycopene dose was administered to 8 healthy adults (4 males and 4 females).

The first two subjects were instructed to follow a moderate lycopene diet (10-20 mg/d) for the two weeks prior to consuming the dose, with the objective of activating enzymes involved in the metabolism of carotenoids. After preliminary analysis of these two subjects revealed high plasma levels of native lycopene and no detectable lycopene

51 metabolites, the remaining 6 subjects were instructed to consume a low lycopene diet (0-

5 mg/d) for the two weeks prior to consuming the dose in order to control the levels of background circulating lycopene and enhance the detection of 13C-labeled lycopene metabolites. The 13C-labeled lycopene in olive oil was administered on an English muffin along with a low carotenoid breakfast. Subjects were given a low-carotenoid snack and lunch 3 and 5 h after dosing, respectively. Blood was drawn hourly from 0-10 h after dosing and at 1, 3, and 28 d after dosing. Urine was collected at baseline, 0-10 h, 10-24 h, and 72 h. Following day 1, subjects were instructed to consume a moderate lycopene diet

(10-20 mg/day) for the remaining four weeks of the study. All plasma and urine samples collected were stored at -80 °C until analysis.

3.3.4 Plasma Extraction for Analysis of Lycopene Isomers and Oxidative Metabolites

Lycopene and lycopene metabolites were extracted from plasma (1 mL) by adding 1 mL ethanol containing 0.1% BHT (w/v) followed by 5 mL of HEAT solvent mix

(hexane/ethanol/acetone/toluene, 10:6:7:7, v/v/v/v). Mixtures were probe sonicated

(Fisher Scientific Model 150I, immersion probe 4C15) for 8 s and centrifuged for 2 min at 3,000 rpm. The upper organic layer was removed and saved. The HEAT extraction was repeated on the lower phase and the upper layers were pooled and dried under nitrogen.

Dried extracts were stored at -80 °C until analysis. All samples were analyzed by HPLC-

MS the same day they were extracted to minimize potential degradation and isomerization of lycopene and its metabolites.

3.3.5 HPLC-MS Plasma Analysis

Plasma extracts were reconstituted in 300 µL MtBE/methanol (1:1) and microcentrifuged

for 2 min at 15,000 rpm. The supernatant was then analyzed immediately by HPLC-MS.

Lycopene and its isomers were separated by a model 2695 HPLC (Waters Corp., Milford,

MA) with a 250 x 4.6 mm i.d., 3 µm, YMC C30 column (Waters Corp.). Solvent A was

methanol/methyl tert-butyl ether/H2O (65/30/5, v/v/v) and solvent B was methyl tert-

butyl ether/methanol/H2O (78/20/2, v/v/v). Compounds were eluted at a flow rate of 1.3

mL/min beginning with 0% B, then linearly increasing to 35% B over 9 min, then to

100% B over 6.5 min, 100% B was held for 2.5 min, and conditions were returned to 0%

B over 3.5 min.

The HPLC was coupled to a quadrupole time-of-flight mass spectrometer (Q-TOF)

Premier hybrid mass spectrometer (Micromass UK Ltd., Manchester, United Kingdom) via an atmospheric pressure chemical ionization probe operated in negative ion mode.

Additional MS parameters were as follows: corona current, 30 µA; collision energy, 8

eV; cone voltage, 40 V; source block temperature, 110 °C; probe temperature, 600 °C;

and desolvation gas flow, 400 L•h-1. All MS experiments were run in Enhanced Duty

Cycle mode for increased sensitivity with V-optics enabled (7500 mass resolution).

HPLC and MS data were acquired using MassLynx software (Waters Corp.)

Plasma 13C-lycopene isomer concentrations were calculated from the peak area of 13C-

13 lycopene in plasma at the monoisotopic mass of uniformly labeled lycopene [ C40H56 ;

m/z 576.57] using an external calibration curve of 13C-lycopene. Plasma 13C-lycopene

was distinguished from native circulating lycopene based on mass [C40H56; m/z 536.44].

Additionally, the labeled dose has a unique reverse isotope distribution, which further

differentiates it from other compounds (Figure 6). Lycopene isomer identities were

assigned based on retention time, accurate mass, and characteristic UV-Vis spectra

(Clinton et al. 1996; Emenhiser et al. 1996).

3.3.6 Metabolomics Analysis of Plasma

The HPLC-QTOF-MS plasma data acquired from this study was mined for 13C-labeled metabolites of lycopene using an untargeted approach. Waters format data files (.raw) were converted to an open data format (.mzXML) using ProteoWizard (Chambers et al.

2012). A subset of the data files were then analyzed using Isotope Ratio Outlier Analysis

(IROA) ClusterFinder software (IROA Technologies, Bolton, MA), which is a feature finding program designed to search for 13C isotope patterns within raw mass spectral

data. The resulting features were manually assessed and potential 13C-labeled metabolites

were evaluated in the full dataset.

Figure 6. Mass spectrum showing the normal isotope distribution for unlabeled lycopene and the reverse isotope distribution for the 13C-labeled lycopene.

3.3.7 Urine NMR Metabolomics Experiments

Urine from baseline and 10-24 h post 13C-lycopene dose, were thawed in cold water.

From each time point, 500 µL of unprocessed urine was sampled and 100 µL of D2O was added prior to the nuclear magnetic resonance (NMR) experiments. Due to the low concentration of analytes in the urine, 5 mL of urine from each time point was also freeze-dried and dissolved in 600 µL of D2O to concentrate approximately 10x.

All experiments were conducted using a Bruker DRX 400 MHz NMR spectrometer

(Bruker Biospin, Rheinstetten, Germany) with a 5 mm triple resonance broadband

observe (TBO) probe for 1H decoupling. 1D 1H decoupled 13C NMR experiments were

performed on both concentrations of baseline and post-dose urine with the following

acquisition parameters: 21.1 kHz spectral width, 10 s relaxation delay, 90 degree 13C excitation pulse, and a WALTZ16 1H composite pulse decoupling sequence.

Approximately 525 and 1,200 scans were acquired for the unprocessed and 10x

concentrated urine, respectively. The free induction decay (FID) data was Fourier

transformed after exponential multiplication using a 1.5 Hz line broadening to give the

1D NMR spectra. Potential 13C-labeled metabolites were identified by comparing peaks

present in the spectra of the baseline and post-dose urine.

To enhance the selectivity for 13C-labeled metabolites, a 1D 13C Incredible Natural

Abundance Double Quantum Transfer Experiment (INADEQUATE) was also performed on the 10x concentrated post-dose urine to investigate 13C-13C coupling. Approximately

28,000 scans were acquired using the following parameters: 90 degree excitation pulse,

21.1 kHz spectral width, 7.5 s relaxation delay, 40 Hz coupling constant, and WALTZ-16 proton decoupling.

3.4 Results and Discussion

3.4.1 Lycopene Isomerization

Pharmacokinetic studies have been done to monitor the absorption of lycopene from a test meal, but to date, none have been able to monitor isomerization of a single dose past

24 hours. Even if the subject undergoes a washout to reduce circulating levels of lycopene before dosing, it is still difficult to distinguish between baseline (native) lycopene and the lycopene administered as part of a dietary intervention. Some studies have investigated lycopene isomer bioavailability by analyzing the triglyceride-rich lipoprotein fraction of plasma (Cooperstone et al. 2015; Unlu et al. 2007), which reflects newly absorbed lycopene, but this approach cannot used to monitor post-absorption isomerization. We were able to circumvent this problem by using 13C-labeling to track isomerization of lycopene from a single dose through 28 days. With the stable isotope tracer and sensitivity of our MS platform, 13C-labeled lycopene could be detected and quantitated in the plasma of 6 of the 8 subjects after only one hour and in the plasma of all subjects after two hours.

The relative amounts of 13C-labeled all-trans and cis isomers of lycopene in the plasma at various time points are reported in Table 1. The isomer profile at 2 h resembles that of the 57

dose, which was determined to be approximately 82% all-trans at the time of

administration. This suggests that there is no difference in bioavailability between all-

trans and cis-lycopene isomers. However, this observation could be partially driven by

the low levels of cis isomers in the dose. Additionally, there are other unique geometric

configurations of lycopene, such as tetra-cis-lycopene, which may have enhanced

bioavailability compared to mono-cis forms (Cooperstone et al. 2015).

As shown in Table 1, lycopene isomerization appears to be a slow occurring process. The

ratio of cis to trans isomers increases modestly between 24 and 72 h and then triples

between 72 h and 28 d. After 28 d, the 13C-labeled lycopene was found to be

approximately 37% all-trans and 63% cis, which is similar to ratios that have previously

been reported in free-living individuals (Clinton et al. 1996; Schierle et al. 1997). The

equilibrium reached in our study after 28 d is supported by computational work done by

Chasse et al. (2001), which shows 5-cis as being the most thermodynamically stable configuration of lycopene, followed by all-trans and then other cis isomers. Time points

in between 72 h and 28 d should be investigated in order to better understand the rate at which isomerization is occurring in vivo.

Table 1. Average 13C-labeled isomers detected in the plasma of 8 subjects at various time points (± SD).

% All-trans % other cis Ratio cis to Time Point % 5-cis LYC LYC LYC trans

2 80.98 ± 8.54 16.75 ± 7.91 2.27 ± 2.14 0.23

5 77.69 ± 6.13 15.81 ± 5.01 6.50 ± 2.24 0.29

10 79.27 ± 6.37 15.59 ± 5.50 5.14 ± 1.89 0.26

24 73.64 ± 8.58 17.61 ± 6.67 8.75 ± 2.39 0.36

72 65.26 ± 8.61 20.71 ± 4.82 14.03 ± 4.14 0.53

d28 37.61 ± 11.11 43.69 ± 4.30 18.70 ± 9.88 1.66

The isomerization of all-trans lycopene to 5-cis over 28 days is well illustrated in Figure

7. Extracted ion chromatograms of plasma 13C-lycopene from a representative subject at

10, 24, 72, and 672 h were overlaid and normalized to the all-trans peak. A small

increase in 5-cis relative to all-trans was observed after 72 h, but drastic changes in this

ratio were observed after 28 d. Again, this suggests that isomerization is occurring in vivo

post-absorption and is not an immediate process. As reported by (Moran et al. 2015), the

plasma appearance curve for the 13C-cis isomers does not follow the same shape as the

13C-all-trans curve in this study. Additionally, the cis-lycopene isomers were found to

have a longer half-life and reach maximum plasma concentration later than all-trans,

which is suggestive of endogenous conversion of all-trans lycopene to cis isomers. It has been suggested that isomerization is part of the metabolism or catabolism of lycopene

(Richelle et al. 2010) and may be the result of singlet oxygen quenching (Foote et al.

1970; Stahl and Sies 1993). Therefore, we propose that cis isomers can be considered oxidative biological metabolites of lycopene and that understanding the isomerization of lycopene is an important step in elucidating the metabolism of this compound.

Figure 7. Overlaid extracted ion chromatograms of plasma 13C-lycopene at different time points (normalized to the all-trans peak) showing isomerization.

3.4.2 Apo-lycopenals Are Not Major Metabolic Products of Lycopene

Given the previous identification of apo-lycopenals in human plasma, it was expected that 13C-labeled aldehyde cleavage products of lycopene would be identified in the plasma of subjects from this study. The LC-MS method used for the plasma analysis was, therefore, designed and optimized to detect the various apo-lycopenals expected.

However, 13C-labeled apo-lycopenals were not detected in the plasma of the 8 subjects at

the any of the time points. Additionally, unlabeled apo-lycopenals were not detected in

the plasma over the course of the study, even when subjects resumed a moderate

lycopene diet (10-20 mg/d). These results suggest that apo-lycopenals are neither major,

nor immediate oxidative metabolites of lycopene. In the work by Kopec et al. (2010),

where apo-lycopenals were reported in human plasma, the levels detected were quite low

(0.12-0.73 nmol/L). Additionally, the plasma samples analyzed were from subjects who

had been on a high tomato juice dietary intervention for 8 weeks. As apo-lycopenals are

present in tomato juice, but were not present in our 13C-labeled lycopene dose, we

hypothesize that apo-lycopenals can accumulate at circulating levels after chronic intake

of tomato products.

3.4.3 LC-MS Metabolomics Analysis of Lycopene Oxidative Metabolism

An untargeted approach was taken to mine the full-scan LC-MS data collected and identify new plasma metabolites of lycopene. The IROA ClusterFinder software was used to scan the data and pull out mass spectral features with an isotopic distribution characteristic of 13C enrichment at approximately 91%. This approach was validated with

62 the detection of 13C-lycopene in our samples (Figure 8A). All features detected by the software were manually reviewed to eliminate false positives resulting from noise in the spectrum. True 13C labeled compounds should only be present post-dose and not at baseline.

Figure 8. Mass spectra of the 13C-labeled lycopene (A) and lycopene epoxide (B) detected with the IROA ClusterFinder software.

From manual review of the ClusterFinder data, we discovered another compound with the same characteristic reverse isotope distribution as the labeled lycopene (Figure 8B) that was present post-dose, but not at baseline. With an m/z of 592.58, this compound was determined to be a 13C-labeled lycopene epoxide as the mass corresponds to the addition of oxygen to the parent molecule. The observed epoxide was found to have a similar absorption spectrum to all-trans-lycopene (Figure 9), and therefore, we hypothesize that epoxidation is occurring at the 1,2 position (see structure in Figure 8B) where the chromophore would be unaltered (Mercadante and Egeland 2004). Epoxidation of lycopene at the 5,6 position has also been reported, but this alteration causes a shortening of the chromophore and a hypsochromic shift of approximately 20 nm

(Eugster 1995; Khachik, Goli, et al. 1992).

Figure 9. UV-Vis spectra of all-trans-lycopene (top) and the lycopene epoxide (bottom).

The 13C-labeled lycopene 1,2-epoxide was present in the plasma of all subjects and was

first detectable between 2-4 h. The relative MS response of the epoxide compared to

lycopene was calculated and used to quantitate the 13C-labeled lycopene 1,2-epoxide in

plasma based on the external calibration curve for 13C-labeled lycopene. The maximum

concentration of the 13C-labeled lycopene 1,2-epoxide detected in plasma was found to be between 0.82 and 4.65 nmol/L. The unlabeled lycopene 1,2-epoxide was also detected in the plasma of all subjects. It was low, but quantifiable, the day of dosing and increased significantly once subjects resumed a moderate lycopene diet. At 28 d, the unlabeled lycopene 1,2-epoxide was found to be between 1.63 and 11.18 nmol/L in plasma.

Khachik et al. (1997) report the presence of lycopene diols in human serum and

hypothesize that the epoxide is an intermediate in the formation of these products. While the investigators were unable to detect lycopene epoxides in serum, they propose that epoxidation is the first step in the oxidative metabolism of lycopene, followed by enzymatic or acid hydrolysis to the observed diol forms. Both the lycopene 1,2-epoxide and 5,6-epoxide are known to be present in tomatoes (Britton and Goodwin 1975;

Khachik, Goli, et al. 1992). Upon further investigation, it was discovered that the 13C- labeled lycopene 1,2-epoxide was present in the 13C-labeled lycopene dose at

approximately 1% the concentration of lycopene. Additionally, the plasma appearance

curve of the 13C-labeled lycopene epoxide mimics that of the 13C-labeled lycopene over

the first 24 h (Figure 10), suggesting that it was absorbed from the diet rather than

generated in vivo.

Figure 10. Average appearance of the 13C-labeled lycopene (left) and the 13C-labeled lycopene epoxide (right) in the plasma of 6 subjects (± SEM).

Despite the presence of the lycopene 1,2-epoxide in the dose, it is surprising that this compound would survive digestion as epoxides are chemically reactive species prone to nucleophilic substitution. Future studies with the pure compound are needed to determine whether the lycopene 1,2-epoxide is absorbed from the diet or an early oxidative metabolite of lycopene. Epoxycarotenoids are widely distributed in nature, but xanthophyll epoxides do not appear to be absorbed by humans (Barua and Olson 2001).

Alternatively, orally administered dietary and synthetic 5,6-epoxy-β-carotene were found to be bioavailable in humans (Barua 1999) and have been shown to be biologically active in inducing differentiation of NB4 leukemia cells in vitro (Duitsman et al. 1999). Other epoxy lipids have also been reported to have important biological activity. For example, fatty acid monoepoxides have been found to have anti-inflammatory properties (Node et al. 1999), antinociceptive effects (Wagner et al. 2014), protect cardiovascular function

(Gauthier et al. 2007), and inhibit angiogenesis and tumorigenesis (G. Zhang et al. 2013).

Therefore, it can be argued that regardless of whether the lycopene epoxide is absorbed or formed in vivo, it is present in human plasma at biologically relevant concentrations and may contribute to some of the bioactivity associated with lycopene and tomatoes.

3.4.4 Investigation of Urinary Lycopene Metabolites Using 13C NMR-Based

Metabolomics

Using 14C-counting accelerator mass spectrometry, Ross et al. (2011) report that 18% of a 14C-labeled lycopene dose was excreted in the urine and the time to maximum concentration (Tmax) was calculated to be 24 h. Based on this research, we chose to

analyze the urine collected between 10 and 24 h for 13C-labeled metabolites. Here 13C

NMR was used to gain information about the structure of these urinary metabolites of

lycopene. While 13C NMR is inherently insensitive due to the low natural abundance of

13C (1.1%), urine from our study should contain metabolites derived from the labeled lycopene that are highly enriched with 13C. This greatly enhances the signal from

lycopene derived metabolites that would not normally be concentrated enough in urine to

be observed by 13C NMR. Potential 13C-labeled metabolites of lycopene can be detected

by comparing the NMR spectra of urine collected at baseline and 10-24 h post-dose.

Results from the 1D 1H decoupled 13C NMR experiments on the baseline and post-dose

urine are presented in Figure 11. The unprocessed urine was not concentrated enough to

obtain strong 13C signals so only spectra from the 10x concentrated urine are shown. Urea

and creatinine were identified as endogenous metabolites based on known chemical

shifts. New peaks in the NMR spectrum after dosing with 13C-lycopene are circled. There

were not any peaks observed in the baseline urine that were not observed in the urine

post-dose. New peaks in the post-dose urine were primarily in the aliphatic region (~20-

60 ppm) and the vinyl region (~120-140 ppm). While there are no known urinary

metabolites of lycopene, these chemical shifts are consistent with structural features that

might be expected from catabolites of lycopene.

Our results suggest that urinary metabolites of lycopene are small, polar compounds that

are possibly conjugated. This supports conclusions made by Ross et al. (2011), who

detected the highest amount of 14C in those fractions of urine that eluted early on a C18 reversed-phase HPLC column. The authors propose that these highly polar metabolites are the products of β-oxidation of lycopene.

Figure 11. 1D 13C NMR spectra comparing 10x concentrated urine at baseline and after the 13C-labeled lycopene dose. Circled peaks were detected post-dose, but not at baseline. 72

Given the promising results from the 1D 1H decoupled 13C NMR experiment, a 1D 13C

Incredible Natural Abundance Double Quantum Transfer Experiment (INADEQUATE) was performed to look at 13C-13C coupling. With this type of experiment, the signal from singly labeled molecules is suppressed and only molecules with adjacent 13C atoms are

seen. Due to the low natural abundance of 13C, most endogenous metabolites should not

be present at high enough concentrations in urine to be observed with a 13C

INADEQUATE. For this reason, it can be expected that any coupled carbons observed

belong to lycopene metabolites derived from the labeled dose. A similar approach has

been taken to identify metabolites of [1,2,3-13C]1-bromopropane in rat urine (Garner et

al. 2006).

The INADEQUATE 13C NMR spectrum of the 10x concentrated urine is shown in Figure

12. Despite the low natural abundance of 13C, urea is still observed in the INADEQUATE

spectrum due to its high concentration in the urine. Several other weak signals were

observed which are suggestive of 13C-enriched metabolites. However, the abundance of

13C-13C coupled metabolites is too low in the urine to be able to detect and identify these

compounds with confidence using this platform. Future experiments should be conducted

at a higher magnetic field strength to better evaluate the presence of 13C-enriched

metabolites.

Figure 12. 1D INADEQUATE NMR spectrum of 10x concentrated urine post 13C- lycopene dose. Arrows are pointing to suggestive 13C-enriched metabolites.

3.5 Conclusions

We have demonstrated that stable isotope labeling is an effective strategy for studying phytochemical metabolism in humans in both a targeted and untargeted approach. Using

13C-labeling, in vivo isomerization of lycopene post-absorption has been confirmed and may be considered an early step in the oxidative metabolism of lycopene. While expected

13C-labeled apo-lycopenals were not detected during the post-prandial absorption of the labeled lycopene dose, a lycopene epoxide has been observed in human plasma. As other lipid epoxides have been reported to have potent biological activity, the lycopene epoxide may contribute to the protective properties associated with lycopene and tomatoes. Our

NMR-based metabolomics experiments have suggested the presence of small polar catabolites of lycopene in the urine. Follow-up experiments are needed to elucidate the structures of lycopene metabolites in the urine using both NMR- and MS-based strategies. Future studies with multiple doses of 13C-labeled lycopene may also assist in the detection of low levels of lycopene metabolites in plasma and urine.

CHAPTER 4: A METABOLOMIC STUDY OF CAROTENOIDS AND OTHER PHYTOCHEMICALS IN TOMATO JUICES INTENDED FOR HUMAN CLINICAL TRIALS

Morgan J. Cichon1, Ken M. Riedl1,2, Steven J. Schwartz1,2

1 Department of Food Science & Technology, The Ohio State University, Columbus, OH, USA 2 Comprehensive Cancer Center, The Ohio State University, Columbus, OH, USA

4.1 Abstract

The consumption of tomato products has been associated with a decreased risk of chronic diseases and carotenoids, such as lycopene, are believed to be partially responsible for this protective effect. The tangerine tomato has a unique carotenoid profile compared to the traditional red tomato, which translates into a difference in bioavailability and potentially in bioactivity. For this reason, these two tomato varieties are being investigated as functional foods in human clinical trials. However, it is unknown how the tangerine and red tomatoes differ in other biologically relevant phytochemicals beyond carotenoids. The objective of this study was to use a liquid-chromatography mass spectrometry (LC-MS)-based metabolomics approach to identify differences in carotenoids and other phytochemicals between red and tangerine tomato juices intended for clinical interventions. Red and tangerine tomatoes grown and processed into juices were analyzed for polar and non-polar phytochemicals using ESI and APCI ionization modes, respectively. The red and tangerine tomato juices were found to differ significantly in a number of phytochemicals and metabolites, including carotenoids, chlorophylls, neutral lipids, dihydrochalcones, hydroxycinnamic acids, and hydroxybenzoic acid derivatives. Many of these compounds have been shown to possess important antioxidant and biological activity and may contribute to the health promoting properties of tomato products in the diet.

4.2 Introduction

Epidemiological research has shown a correlation between increased consumption of

tomato products and a decreased risk of certain diseases, including prostate cancer

(Giovannucci 1999, 2002; Hadley et al. 2002). This relationship has also been observed

in studies with animals fed diets supplemented with tomatoes. However, we still know

little about the mechanism behind this observed protective effect. Much research has

focused on the tomato carotenoid lycopene as a potential bioactive compound due to its

abundance in the tomato and its antioxidant capacity. Lycopene is an efficient singlet oxygen quencher and free radical scavenger (Bohm et al. 2002; Di Mascio et al. 1989) and there is evidence suggesting that these properties translate into a protective effect in

vivo. Additionally, lycopene has been shown to accumulate in human tissues, such as the prostate, where it may have some biological consequence (Clinton et al. 1996).

Research has traditionally focused on the red tomato, but in a recent study, lycopene from a unique tangerine tomato variety was found to be 8.5 times more bioavailable than lycopene from the red tomato (Cooperstone et al. 2015). Tangerine tomatoes are orange in color, which is a result of lycopene being biosynthesized in a tetra-cis geometrical configuration rather the all-trans configuration found in red tomatoes. This conformational change causes a shift in the absorption spectrum of lycopene, resulting in the orange rather than red color of the tomato. The enhanced bioavailability of lycopene from the tangerine tomato has been attributed in part to the tetra-cis geometrical configuration of lycopene, as it has been hypothesized that the cis isomers are more 78

easily incorporated into micelles and absorbed in the small intestine than the all-trans

form found in the red tomato (A. C. Boileau et al. 1999).

Given the significant differences in lycopene bioavailability between the red tomato and

the tangerine tomato, there is interest in evaluating whether this translates into a

difference in biological activity. Red and tangerine tomatoes have been grown and

processed into juices at The Ohio State University for use as functional foods in human

clinical interventions with prostate cancer patients. While we know how these two

tomatoes differ in their carotenoid profiles, it is unknown how they differ in other

potentially bioactive phytochemicals. Tomatoes contain many phenolic compounds,

including flavonoids, such as naringenin and kaempferol, and hydroxycinnamic acids,

such as ferulic acid and coumaric acid (Moco et al. 2006). These compounds have been

shown to possess important bioactivity (Erlund 2004; Heim et al. 2002; Meyer et al.

1998) and therefore, it is reasonable to believe that they contribute to the health benefits

associated with tomatoes. In fact, some research has demonstrated an enhanced protective

effect when feeding whole tomatoes versus lycopene alone (T. W.-M. Boileau et al.

2003; Canene-Adams et al. 2007). These data suggest a synergistic effect between the

carotenoids and other phytochemicals in the tomato. In order to better understand any

biological effects observed in the clinical trials with the red and tangerine tomato juices, it is important to have a more comprehensive understanding of their phytochemical

differences.

The objective of this study is to use a liquid-chromatography mass spectrometry

(LC/MS)-based metabolomics approach to identify phytochemical and metabolite

differences in the polar and non-polar fractions of the red and tangerine tomato juices.

Untargeted metabolomic profiling allows for the unbiased detection and analysis of

thousands of phytochemical species belonging to a number of different compound classes

in a single analysis. This approach has been used to characterize phenolic compounds and

other secondary metabolites in tomatoes (Gómez-Romero et al. 2010; Moco et al. 2006)

and to evaluate the effects of thermal processing on tomato phytochemicals (Capanoglu

et al. 2008). We hypothesize that a metabolomics approach will allow us to identify

important phytochemical differences with potential biological relevance in functional

food-based clinical trials with the red and tangerine tomatoes.

4.3 Materials and Methods

4.3.1 Chemicals

All solvents were from Fisher Scientific (Pittsburgh, PA, USA). Methyl tert-butyl ether

(MtBE) and acetone were HPLC grade, hexanes was Optima grade, and acetonitrile

(ACN) was Optima LC/MS grade. Methanol (MeOH) and water were either HPLC grade

(extraction solvents) or Optima LC/MS grade (LC/MS analysis). Ammonium acetate was from J.T. Baker (Phillipsburg, NJ, USA) and formic acid was from Fisher Scientific.

Chlorogenic acid standard was purchased from Cayman Chemical (Ann Arbor, MI, USA) and 2-isopropylmalic acid standard was purchased from Santa Cruz Biotechnology

(Dallas, TX). Pheophytin a standard was generated from chlorophyll a standard using the 80 method described by Pumilia et al. (2014). All-trans-lycopene standard was isolated from tomato paste as previously reported (Kopec et al. 2010).

4.3.2 Tomato Juices

Tangerine tomatoes (Solanum lycopersicum L., hybrid FG10-314) and red tomatoes

(Solanum lycopersicum L., hybrid PS696) were grown in Fremont, OH at The Ohio State

University (OSU) North Central Agricultural Research Station. Both tomato varieties were processed into juice in the OSU Food Industries Center Pilot Plant in the

Department of Food Science & Technology (Columbus, OH, USA). Five cans of each juice were sampled for the metabolomics analyses. In order to expand compound coverage, two separate extractions were performed on the juices to extract both lipophilic and polar/semi-polar phytochemicals.

4.3.3 Preparation of Lipophilic Extract

To extract lipid soluble phytochemicals, 5 mL of MeOH was added to 2 g of tomato juice, probe sonicated for 8 s, and centrifuged for 5 min at 2,000 x g. The supernatant was removed and saved. The remaining pellet was then extracted with 5 mL hexane/acetone

(1:1, v/v), probe sonicated for 8 s, and centrifuged for 5 min at 2,000 x g. The supernatant was again removed and added to the previously saved supernatant. The hexane/acetone extraction was repeated 2 more times. To the pooled supernatants, 5 mL of water was added to drive phase separation and 1 mL of the upper organic layer was dried under

81 nitrogen. Dried extracts were stored at -80 °C for no more than 24 h before analysis by

LC-QTOF-MS.

4.3.4 Preparation of Polar/Semi-Polar Extract

Polar and semi-polar phytochemicals were extracted using a method adapted from Moco et al. (2006). In summary, 3 mL of MeOH was added to 1 g of tomato juice to yield an extract of approximately 75% MeOH and 25% water. The sample was then vortexed for

10 s, bath sonicated for 15 min, and centrifuged for 10 min at 2,000 x g. The supernatant was subsequently removed and immediately analyzed by LC-QTOF-MS.

4.3.5 Non-polar Phytochemical Analysis by LC-QTOF-MS (APCI+)

Both tomato juice extracts were analyzed using a 1290 Infinity UHPLC system with a diode array detector (DAD) coupled to an iFunnel 6550 QTOF-MS (Agilent, Santa Clara,

CA). Data were acquired in the range of 100-1700 m/z using Agilent MassHunter

Acquisition software.

Lipophilic tomato juice extracts were redissolved in 1 mL MtBE followed by 1 mL of

MeOH. Extracts were filtered through a 0.2 µm nylon filter and 5 µL were injected onto a

C30 reversed-phase column (250 mm x 4.6 mm, 3 µm particle size) (YMC America, Inc.,

Allentown, PA) maintained at 40 °C. Compounds were eluted using a mobile phase of A

= MeOH/MtBE/2% ammonium acetate (60:35:5, v/v/v) and B = MtBE/MeOH/2% ammonium acetate (78:20:2, v/v/v). A gradient was applied at 1.3 mL/min starting at 0%

B and increasing to 35% B over 9 min, then increasing to 100% B over 6.5 min, and

returning to 0% B over the final 6 min. For the analysis of lipophilic phytochemicals, the

UHPLC system was coupled to the QTOF-MS via atmospheric pressure chemical

ionization (APCI) operated in positive ion mode. The following MS parameters were

used: gas temperature, 290 °C; vaporizer, 500 °C; gas flow, 13 L/min; nebulizer, 20 psig;

VCap, 3500 V; corona current, 5 μA; nitrogen CID gas, 22 psig. Data were acquired in 2

GHz extended dynamic range (EDR) mode with 20K resolution in the range of 100-1700

m/z at a scan rate of 1 spectra/s. Prior to each experiment, the TOF mass axis was

calibrated with ESI-L solution (G1969-85000, Agilent). Reference solution (G1969-

85001, Agilent) containing HP-0921 was infused concurrently with the analytical

nebulizer through the dedicated reference sprayer providing mass calibration correction

for each scan.

4.3.6 Polar/Semi-Polar Phytochemical Analysis by LC-QTOF-MS (ESI-)

The methanol extracts were filtered and 2 µL were injected onto a Zorbax Eclipse Plus

RRHD C18 column (150 mm x 2.1 mm i.d., 1.8 µm particle size) (Agilent) maintained at

40 °C. Compounds were eluted with A = water (0.1% formic acid) and B = ACN (0.1%

formic acid). A 20 min gradient was applied at 0.6 mL/min starting at 5% B, increasing

to 35% B over 10 min, holding at 35% B for 2 min, increasing to 75% B over 3 min,

holding at 75% B for 2 min, and then returning immediately to 5% B and re-equilibrating

for 3 min. The UHPLC system was coupled without flow splitting to the QTOF-MS via electrospray ionization (ESI) operated in negative ion mode. The following MS

parameters were used: gas temperature, 290 °C; gas flow, 13 L/min; nebulizer, 30 psig;

sheath gas temperature, 400 °C; sheath gas flow, 12 L/min; VCap, 4000 V; nozzle voltage, 2000 V. Data were acquired in 2 GHz extended dynamic range (EDR) mode with 20K resolution in the range of 50-1700 m/z at a scan rate of 3 spectra/s. Mass

calibration and reference sprayer operation was performed as described under the

lipophilic analytical section above.

4.3.7 Data Processing and Statistical Analysis

Raw LC/MS data were processed using the Batch Recursive Feature Extraction option in

the Agilent MassHunter Profinder software (version B.06.00). This option extracts mass

spectral features and collapses related isotopes and common adducts ([M+Na]+, [M+K]+,

[M+HCOO]-) into one feature. This is followed by mass and retention time alignment of

all features. Features detected in only one of the tomato juice samples were disregarded.

A recursive feature extraction was then performed on the raw data where the mass and

retention time results from the untargeted feature extraction in the first step are used for a

targeted feature extraction. This improves the accuracy of the feature extraction by

reducing the number of false negatives and false positives in the dataset, thereby

increasing the quality of the data exported for differential analysis.

Extracted compounds, comprised of a mass, retention time and intensity, were exported

as compound exchange files (CEF) for further analysis using the Agilent Mass Profiler

Professional (MPP) chemometrics software (version 13.1). Only those compounds that

were present in at least 80% of the samples in the red and/or tangerine tomato juice

groups and had a coefficient of variation < 25% within a group were retained.

Statistically significant differences between the red and tangerine tomato juices were determined using an unpaired t-test (P < 0.05; Benjamini-Hochberg false discovery rate

(FDR) multiple testing correction). Compounds differing by a fold change greater than two between the red tomato juice and the tangerine tomato juice were considered for identification.

4.3.8 Compound Identification

Compounds were identified using a combination of UV/Vis spectra (carotenoids), accurate mass, isotope ratios, MS/MS fragmentation patterns, and authentic standards when available. This information was compared against published literature on the chemical composition of tomatoes and publically available online metabolite databases

(Metlin (Smith et al. 2005) and the Human Metabolome Database (Wishart et al. 2013)).

Targeted MS/MS experiments for structural identification were conducted using the same

LC-QTOF-MS systems and mobile phase gradients described previously. MS/MS data were collected by isolating precursor ions with a quadrupole resolution of 1.3 amu and using fixed collision energies of 10, 20, and 40 eV in the mass range of 50-1700 m/z at an acquisition rate of 2 spectra/s.

4.4 Results and Discussion

An untargeted metabolomics approach was taken to profile carotenoid and other

phytochemical differences between the red and tangerine tomato juices. Metabolomics

has been used as a tool for studying food composition in a variety of studies including

those focused on cultivar variation (Brown et al. 2012; Dobson et al. 2008; Gómez-

Romero et al. 2010), food quality (Johanningsmeier and McFeeters 2011; Le Gall et al.

2003), and product adulteration (Jandrić et al. 2014; Vaclavik et al. 2012). Here

metabolomics was used to enhance functional food research of tomatoes at a molecular

level. It is important to understand the phytochemical differences between the red and

tangerine tomato juices in order to better interpret clinical outcomes from dietary interventions with these food products.

4.4.1 Lipophilic Phytochemicals Correlated with the Red and Tangerine Tomato Juices

Carotenoids and other lipophilic phytochemicals were analyzed using APCI+. Of the

1,120 compounds detected in the lipophilic fraction, 501 (45%) were found to be significantly different with a corrected P < 0.05 (Benjamini-Hochberg false discovery rate (FDR) multiple testing correction) between the red tomato juice and the tangerine tomato juice. Focusing on those compounds differing by a fold change of at least two when comparing the red and the tangerine tomato juices, 430 compounds remained with

212 being significantly higher in the red tomato juice and 218 being significantly higher in the tangerine tomato juice.

A number of lipophilic compounds highly correlated with either the red or tangerine tomato juices were identified as carotenoids based on authentic standards, accurate mass, and characteristic UV-Vis spectra as reported previously (Cooperstone et al. 2015). As shown in Table 2, the red tomato juice and the tangerine tomato juice differed significantly in their carotenoid profiles. The all-trans configuration of lycopene was 50 times higher in the red tomato juice than the tangerine tomato juice, while the lycopene precursors phytoene and phytofluene were 4.5 and 3.6 times higher in the tangerine tomato juice compared to the red, respectively. The oxidative metabolite, lycopene 1,2- epoxide, was identified by UV-Vis/MS and was only detected in the red tomato juice. On the other hand, three carotenoids were found to be unique to the tangerine tomato juice-

ζ-carotene, neurosporene, and tetra-cis-lycopene. Several isomers of both ζ-carotene and neurosporene were detected in our metabolomics analysis. Tangerine tomatoes are rich in carotenoid cis isomers due to the lack of a functional carotenoid isomerase to enzymatically convert poly-cis-lycopene isomers to all-trans-lycopene. This results in the accumulation of cis-isomers of lycopene and its precursors ζ-carotene and neurosporene in the tangerine tomato. For this reason, chromatography was particularly important for our analysis and we chose a C30 analytical column to aid in the chromatographic separation of these carotenoid species. Additionally, many of these carotenoids only differ by one double bond and therefore have overlapping first and third isotopes that require high mass resolution or chromatographic separation to distinguish. As the carotenoids detected and identified in this analysis have been previously reported in red

87 and tangerine tomatoes (Clough and Pattenden 1979), they served as validation of the untargeted metabolomics approach.

Table 2. Identified carotenoids and other lipophilic phytochemicals differentiating red and tangerine tomato juices. Observed Retention Molecular Mass Difference from Compound IDa Fold Change [M+H]+ Time (min) Formula Theoretical (Δppm) (Red vs. Tangerine)

537.4461 15.27 C40H56 1.12 All-trans lycopene 49.6 c 537.4458 9.33 C40H56 0.56 Tetra-cis lycopene ------c 539.4608 9.62 C40H58 -0.56 Neurosporene isomer ------c 539.4603 12.18 C40H58 -1.48 Neurosporene isomer ------c 539.4609 12.89 C40H58 -0.37 Neurosporene isomer ------c 541.4762 8.92 C40H60 -1.11 ζ-carotene isomer ------c 541.4769 9.91 C40H60 0.18 ζ-carotene isomer ------c 541.4760 10.28 C40H60 -1.48 ζ-carotene isomer ------

543.4926 8.17 C40H62 0.37 Phytofluene -3.60 89 545.5085 7.92 C40H64 0.73 Phytoene -4.46 d 553.4404 13.03 C40H56O 0.00 Lycopene 1,2-epoxide ------

613.4816 7.48 C39H64O5 -1.79 DG (18:3/18:3/0:0) -2.67

871.5726 7.24 C55H74N4O5 -0.69 Pheophytin a -5.06

813.5681 10.04 C53H72N4O3 0.49 Pyropheophytin a -4.15 b 875.7122 8.01 C57H94O6 -0.11 TG (18:3/18:3/18:2) [iso3] -3.21 b 851.7121 8.94 C55H94O6 -0.23 TG (18:3/18:3/16:0) [iso3] -2.99 b 853.7278 9.50 C55H96O6 -0.23 TG (18:3/18:2/16:0) [iso6] -2.18

873.6969 7.487 C57H92O6 0.23 TG (18:3/18:3/18:3) -5.48 TG: Triglyceride; DG: Diglyceride a Bolded compounds have been confirmed using authentic standards. b Fatty acid positions on the glycerol backbone were not confirmed. Number of positional isomers possible is listed in brackets. c Compound was not detected in the red tomato juice. d Compound was not detected in the tangerine tomato juice.

Unsupervised hierarchical clustering analysis was performed on the normalized intensity values of the 430 significantly different compounds with a fold change greater than two between the tomato varieties using Euclidean distances and Ward’s linkage rule (Figure

13A). As is shown by the heat map coloring, many of these lipophilic compounds were present in high abundance (red) in one of the juices and present in much lower abundance or absent (blue) in the other juice. These results were unexpected as it was assumed that the aforementioned carotenoids were the major differentiating lipophilic phytochemicals.

B 91

Figure 13. Heat map from hierarchal clustering analysis (Euclidean distance and Ward’s linkage rule) performed on significantly different compounds in the red and tangerine tomato juices with a fold change > 2 for the lipophilic fraction (A) and polar/semi-polar fraction (B). The heat map is colored by normalized relative ion intensities, with red representing relatively high abundance compounds and blue representing relatively low abundance or absent compounds.

In attempt to understand and identify these other phytochemical differences, we plotted them by retention time versus mass and colored by fold change regulation (Figure 14). In doing so, we observed clusters of compounds in the data with the same retention times, but unique masses. Interestingly, these clusters lined up with the predominant carotenoids in the tomato juices. For example, the vertical streaks at 7.9, 8.2, 9.3, 9.6, and 9.9 min appear to correspond with phytoene, phytofluene, tetra-cis-lycopene, neurosporene, and

ζ-carotene, respectively, in the tangerine tomato juice, while the streak at 15.2 min appears to correspond with all-trans-lycopene in the red tomato juice. These are the most abundant carotenoids in the two juices. As carotenoids are highly conjugated structures, they are fairly labile and susceptible to degradation once extracted from the plant or food matrix (Kopec et al. 2012). Therefore, we hypothesize that these clusters are arising in- source from complex gas-phase chemistry of carotenoids and carotenoid fragments.

These “metabolite streaks” presents a unique challenge when analyzing carotenoids using an untargeted metabolomics approach and researchers attempting to do so should be aware of this issue when data mining. Here we collapsed each “metabolite streak” into one feature, which greatly reduced the complexity of the data.

Figure 14. Retention time versus mass plot of significantly different non-polar compounds detected in the tomato juices (P < 0.05; fold change > 2) showing the compound clusters (“metabolite streaks”) in the data. Red compounds were higher in the red tomato juice, while blue compounds were higher in the tangerine tomato juice.

4.4.2 The Tomato Juices Differed in Other Pigments and Lipids besides Carotenoids

In addition to carotenoids, a number of other lipophilic compounds were identified and

found to be significantly differentiated between the red and tangerine tomato juices. Plant pigments belonging to the chlorophyll class of compounds were identified as differentiating phytochemicals in the tomato juices. Pheophytin a, identified by accurate mass and confirmed by authentic standard, was found to be 5 times higher in the tangerine tomato juice compared to the red tomato juice. Pheophytin is a chlorophyll degradation product resulting from the chemical displacement of magnesium from the porphyrin ring of the molecule. This compound is formed with heat and is commonly found in thermally processed vegetables (Steven J. Schwartz et al. 1981; Weemaes et al.

1999). Another chlorophyll degradation product, pyropheophytin a, was also found to be higher in the tangerine tomato juice (4 fold difference). Pyropheophtyin is characterized by the loss of a carbomethoxy group from pheophytin and is also formed during thermal treatment of chlorophyll-containing foods (Pumilia et al. 2014; S. J. Schwartz and von

Elbe 1983). All immature tomatoes contain the green pigment chlorophyll, which is enzymatically degraded during ripening (Kozukue and Friedman 2003). As the tomatoes were thermally processed into juice for this study, the presence of both pheophytin a and pyropheophytin a can be rationalized. We were intrigued to learn that using an untargeted metabolomics approach we were able to identify other differentiating plant pigments

beyond the more obvious carotenoids.

Our analysis also revealed significant differences in neutral lipids, namely triglycerides,

between the red and tangerine tomato juices. The fatty acid compositions of the

triglycerides in Table 2 were identified based on their characteristic in-source

fragmentation patterns (Holcapek et al. 2003). The corresponding diglyceride fragment

ions [M – RCOO]+ observed for each triglyceride are reported in Table 3. All identified triglycerides were comprised of some combination of linoleic (C18:2), linolenic (C18:3), and palmitic (C16:0) acids, which corresponds with the fatty acids that have been previously reported in tomatoes (Lenucci et al. 2012; Sun et al. 2010). The positions of the fatty acids on the glycerol backbone (sn-1, 2, or 3) were not confirmed for this analysis and a number of positional isomers are possible for some of the identified triglycerides.

Carotenoids need to be incorporated into lipid micelles to be absorbed and research has demonstrated that dietary fat is important for lipid micelle formation (Erdman et al.

1993). While tomatoes alone are not a rich source of dietary lipids, the 2-3 fold increase in triglycerides in the tangerine tomato juice could contribute to the observed enhanced absorption of lycopene from tangerine tomato juice compared to red tomato juice

(Cooperstone et al. 2015).

Table 3. Identification of triglycerides based on APCI in-source ion fragmentation. Observed APCI in-source Corresponding fragment Proposed TG [M+H]+ fragments structure fatty acid [M+H]+ [M+H-RCOOH]+ composition 851.7121 573.4875 18:3/16:0 18:3/18:3/16:0 595.4730 18:3/18:3 853.7278 573.5880 18:3/16:0 18:3/18:2/16:0 597.4877 18:3/18:2 575.5030 18:2/16:0 873.6969 595.4731 18:3/18:3 18:3/18:3/18:3 875.7122 597.4874 18:3/18:2 18:3/18:3/18:2 595.4723 18:3/18:3 TG = Triglyceride

4.4.3 Significant Differences in Polar Tomato Phytochemicals

As with the non-polar fraction, many compounds were detected in the methanol extract

using an untargeted approach. The complexity of the data is illustrated by the extracted

ion chromatograms in Figure 15. With our LC-QTOF-MS method, we were able to detect

both high and low abundance compounds in our tomato juices. In analyzing the polar

fraction of the tomato juices using ESI-, 1649 (37%) of the 4452 phytochemicals detected were found to be significantly different with a corrected P < 0.05 (Benjamini-Hochberg

false discovery rate (FDR) multiple testing correction) between the red and tangerine

tomatoes. Of those 1649 phytochemicals, 506 had a fold change of at least two between

the two varieties with 371 being higher in the red tomato juice and 135 being higher in

the tangerine tomato juice. As with the lipophilic phytochemicals, we performed a

hierarchical clustering analysis on the 506 significantly different compounds with a fold

change greater than two between the tomato varieties (Figure 13b). When comparing the

hierarchical clustering results for the lipophilic tomato fraction to the polar fraction, it is

apparent from the heat map that the lipophilic phytochemicals vary between the juices to

a greater degree than do the polar phytochemicals. While there are a number of

carotenoids that are particular to the red or tangerine tomato, many of the polar

phytochemicals and metabolites appear to be common to both tomato groups and have

smaller fold change differences. However, these differentiating compounds still have a

fold change of at least two between the red and tangerine tomato juices and can be

considered important chemical differences with potential biological implications.

Figure 15. Overlaid extracted ion chromatograms of compounds detected in the tomato juices illustrating the complexity of the data. The inset is zoomed to the region between 6 and 9.3 min to show the many compounds detected within the dynamic range of the instrument.

4.4.4 Identification of Differentiating Polar Phytochemicals

A number of polar phytochemicals were identified as being significantly different

between the red and tangerine tomato juices (Table 4). This was particularly interesting,

because differences in polar phytochemicals between the red and tangerine tomatoes have

yet to be reported. Compound identification is arguably the most challenging and time-

consuming part of any metabolomics analysis. Here we used literature and accurate mass

database searching along with in-source and collision induced fragmentation data to

identify phytochemicals of interest. Phytochemicals of interest were those that were

found to be significantly different between the red and tangerine tomato juices (P < 0.05)

and differing in abundance by a fold change of two or greater. Priority was also given to high intensity ions that were well above background noise and for which reliable MS/MS fragmentation could be obtained. For example, the compound with an observed mass of

325.0908 at 3.04 min was identified as a hexose derivative of coumaric acid based on the fragment ions 163.0398 and 119.0500. Fragment 163.0398 is consistent with coumaric acid (loss of a hexose moiety), while the 119.0500 fragment matches MS/MS spectra for coumaric acid in the Metlin metabolite database. Additionally, a coumaric acid hexose derivative has previously been reported as a component of tomatoes (Gómez-Romero et al. 2010).

Table 4. Identified polar and semi-polar phytochemicals differentiating red and tangerine tomato juices. Observed Retention Chemical Mass MS/MS Fragments Compound IDa Fold [M-H]- Time Formula Difference Change (min) from (Red vs. Theoretical Tangerine) (Δppm) 175.0609 2.70 C7H12O5 -1.70 115.0339, 85.0654, 2-isopropylmalic acid 2.44 59.0136, 113.0606

209.0458 3.73 C10H10O5 1.21 137.0606, 165.0554 Hydroxyferulic acid 2.38

245.0929 5.27 C13H14N2O3 -1.08 74.0246, 58.0293, Acetyl tryptophan 2.10 203.0822, 116.0500, 98.0243

285.0609 2.30 C12H14O8 -2.42 152.0111, 108.0212 Dihydroxybenzoic acid pentose -2.86 derivative

100 325.0927 3.04 C15H18O8 -0.59 163.0398, 119.0500 Coumaric acid hexose derivative 2.50

329.2330 11.29 C18H34O5 -1.06 171.1023, 139.1124, Hydroxyoctadecanedioic acid 3.35 211.1339

343.1033 2.85 C15H20O9 -0.45 181.0505, 137.0605, Homovanillic acid hexose 2.28 59.0136 derivative

353.0876 2.99 C16H18O9 -0.58 191.0565, 173.0455, Chlorogenic acid -2.05 179.0352, 135.0450

355.1032 3.59 C16H20O9 -0.72 175.0398, 193.0506, Ferulic acid hexose derivative -4.63 160.0167

474.2621 14.96 C23H42NO7P -1.08 277.2161, 152.9952, LysoPE(18:3/0:0) -2.64 78.9587

597.1821 5.74 C27H34O15 -0.66 357.0963, 387.1066, Phloretin-di-C-glycoside 2.53 477.1381, 417.1171 aBolded compounds have been confirmed using authentic standards. 100

In some cases, Molecular Structure Correlator (MSC) software (version B.05.00, Agilent) was used to correlate collected MS/MS fragmentation data with candidate chemical structures. MS/MS experiments were performed using same QTOF-MS system and

mobile phase gradient as the untargeted analyses, yielding accurate mass information for the fragment ions as well as the parent compound. With accurate mass MS/MS spectra,

MSC was able to propose chemical formulas for the parent ions as well as their fragments. MSC was then used to search the ChemSpider database and to score matching structures based on how well the observed fragments could be chemically rationalized.

For example, the compound with an observed [M-H]- mass of 353.0876 at retention time

2.99 min with the fragments 191.0565, 173.0455, 179.0352, and 135.0450, was matched

with chlorogenic acid (compatibility score of 93 out of 100). Chlorogenic acid is a

reported phenolic compound in tomatoes (Gómez-Romero et al. 2010; Vallverdú-Queralt

et al. 2010) and following tentative identification by MSC, was confirmed in the tomato

juice samples using an authentic standard. This demonstrates the utility of the MSC

software in elucidating the structures of unknown phytochemicals and metabolites.

4.4.5 Red and Tangerine Tomato Juices Differ in Levels of Potentially Bioactive

Phenolic Compounds

Many of the phenolic compounds differentiating the red and tangerine tomato juices

(Table 4) have been reported to have some bioactivity, including dihyrochalcones and

hydroxycinnamic acids. From a functional food perspective, these differences are

particularly important as they have the potential to translate into differences in clinical

101

efficacy. Phloretin-di-C-glycoside was reported to be a significant flavonoid in tomatoes

as analyzed by NMR (Slimestad et al. 2008). In fact, it has been found to be present in tomatoes at 5-14% of the total flavonoid content. We have putatively identified this

dihyrochalcone in the tomato juices and detected it at levels greater than 2.5 times higher in the red tomato juice compared to the tangerine tomato juice. Gómez-Romero et al.

(2010) similarly found significant differences in phloretin-di-C-glycoside among

different Spanish tomato cultivars with levels being much higher in the red tomato

variety compared to the greener-colored varieties analyzed. Research has demonstrated a

chemoprotective effect of phloretins, including suppression of inflammation (Chang et al.

2012; W. C. Huang et al. 2013; J. H. Lee et al. 2011), inhibition of oxidative stress (Y. C.

Yang et al. 2011), inhibition of cell proliferation (Devi and Das 1993), and induction of apoptosis (M.-S. Kim et al. 2009; Kobori et al. 1997).

The hydroxycinnamic acid derivatives coumaric acid hexoside, chlorogenic acid,

hydroxyferulic acid, and a hexose derivative of ferulic acid were also found to be significantly different between the red and tangerine tomato juices. Both chlorogenic acid

and ferulic acid have known antioxidant activity in vitro and have been reported to inhibit

tumor promotion in mice (M.-T. Huang et al. 1988). Chlorogenic acid (3-O-

caffeoylquinic acid) is the conjugate of coumaric acid with quinic acid and was

approximately two times higher in the tangerine tomato juice compared to the red tomato

juice. Research has suggested that chlorogenic acid may protect against carcinogenesis by

modulating phase II detoxifying enzymes (Feng et al. 2005). It has also been shown to

102

significantly attenuate intestinal glucose absorption (Johnston et al. 2003; Welsch et al.

1989). Unlike chlorogenic acid, the ferulic acid derivative hydroxyferulic acid was found to be over two times higher in the red tomato juice. The position of the hydroxyl group could not be determined in this study as the compound had the same accurate mass and retention time as 5-hydroxyferulic acid, but different MS/MS fragmentation. A hexose

derivative of ferulic acid was detected at a 4.6 times higher abundance in the tangerine

tomato juice compared to the red. Ferulic acid is an effective free radical scavenger,

which is believed to be one of the predominant mechanisms through which it exerts a

biological effect in humans (Kikuzaki et al. 2002; Srinivasan et al. 2007). It has been

reported to protect against a number of conditions including inflammation, cancer,

neurodegeneration, and diabetes (Srinivasan et al. 2007).

The objective for using an unbiased or untargeted metabolomics approach was to capture

a greater number of phytochemical and metabolite differences between the two tomato

juices than could be accomplished with a targeted approach. Through our analysis we

were able to not only detect differences in carotenoid composition, but we were also able

to detect and identify other phytochemical differences that we would have likely missed

had we focused on only a select group of compounds. We believe that while carotenoids

may contribute to the protective effects of tomatoes, it is likely that there are other phytochemical differences between the red and tangerine tomato juices that could

translate into differences in biological activity. Here we demonstrate the effectiveness of

103 using untargeted metabolomics to understand chemical differences between foods and to enhance nutritional and dietary interventions being conducted with these products.

4.5 Acknowledgments

We would like to acknowledge David Francis, Ph.D. (Department of Horticulture and

Crop Science, The Ohio State University) for providing the tomatoes used in this study and Dan Cuthbertson, Ph.D. (Agilent Technologies) for his assistance with the metabolomics analysis.

104

CHAPTER 5: INFLUENCE OF LYCOPENE AND DIFFERENT TOMATO DIETS ON THE PLASMA METABOLOMES OF MICE

Morgan J. Cichon1, Ken M. Riedl1,2, Lei Wan3, Steven K. Clinton2,4, Steven J. Schwartz1,2

1 Department of Food Science & Technology, The Ohio State University, Columbus, OH, USA 2 Comprehensive Cancer Center, The Ohio State University, Columbus, OH, USA 3 Interdisciplinary Nutrition Program, The Ohio State University, Columbus, OH, USA 4 Division of Medical Oncology, Department of Internal Medicine, The Ohio State University Wexner Medical Center, Columbus, OH, USA

105

5.1 Abstract

Research has suggested a correlation between increased consumption of tomatoes and a

decreased risk of certain diseases, such as prostate cancer. As the carotenoid lycopene is

the predominant pigment in tomatoes and an efficient singlet oxygen quencher, many

have focused on lycopene as the main bioactive compound in tomatoes responsible for

these protective effects. In order to better understand the biological impact of tomato

phytochemicals, untargeted metabolomics was used to 1) compare the effects of lycopene

and red tomatoes on the plasma metabolomes of mice and 2) evaluate whether red, tangerine, and low carotenoid tomato varieties differentially impact the metabolome.

Untargeted metabolomic profiling of plasma was performed using UHPLC-QTOF-MS.

Of the thousands of metabolites detected using this approach, 169 were found to be significantly different (P < 0.05) between the red tomato, lycopene, and control fed mice.

The tomato alkaloids tomatidine, trigonelline, and pimpifolidine were identified as being

significantly upregulated in response to the tomato diet. Additionally, the plasma

metabolite profiles of mice on the tangerine tomato diet were very similar to those of

mice on the low carotenoid tomato diets, while the red tomato fed mice were quite

different, with a number of plasma metabolites detected only in the red tomato fed mice.

Further studies need to be performed to determine if these markers of tomato

consumption can be translated to humans.

106

5.2 Introduction

Epidemiological studies have suggested a correlation between diets rich in tomatoes and

a decreased risk for a number of chronic diseases, including certain cancers and cardiovascular disease (Giovannucci 1999; Willcox et al. 2003). The relationship between tomatoes and health is still not well understood, but much research has focused on lycopene as being one of the main bioactive components of the tomato. The C40 hydrocarbon lycopene belongs to the carotenoid family of compounds and is the lipophilic pigment responsible for the red color of tomatoes. Lycopene is known to accumulate in human tissues, such as the prostate, where it may have some biological effect (Clinton et al. 1996; Stahl et al. 1992). The mechanism behind its proposed bioactivity has yet to be fully elucidated, but lycopene is an efficient singlet oxygen quencher in vitro (Di Mascio et al. 1989) and it has been suggested that the protective properties of lycopene in vivo may be related to its antioxidant capacity. Lycopene has also been shown to induce apoptosis (L. Tang et al. 2005), inhibit cell cycle progression

(Cheng et al. 2007), increase gap junction communication (Livny et al. 2002), and alter androgen status (Campbell, Stroud, et al. 2006). Such studies support lycopene as a health promoting tomato phytochemical.

Approximately 95% of lycopene found in the traditional red tomato is present in the all- trans configuration, with the rest found as various mono-cis isomers (Nguyen et al.

2001). The tangerine tomato is a unique variety which biosynthesizes lycopene in a tetra- cis geometrical configuration compared to the all-trans form found in the red tomato. The 107

tetra-cis configuration causes a shift in the UV-Vis absorption spectrum of lycopene, giving the tangerine tomato an orange color. Interestingly, research from our group has found that lycopene from the tangerine tomato is much more bioavailable than lycopene from the red tomato (Cooperstone et al. 2015). If lycopene is one of the bioactive components of the tomato, the tangerine tomato may offer enhanced health benefits over the red tomato.

There is also research that has shown that whole tomatoes are more effective than lycopene alone in preventing prostate cancer in rodents (T. W.-M. Boileau et al. 2003;

Canene-Adams et al. 2007). This suggests that other tomato components beyond lycopene contribute to the protective effects associated with tomatoes. Tomatoes contain other potentially bioactive compounds besides lycopene, such as vitamins C, K, and E, flavonoids, phenolics, folate, potassium, and other carotenoids (Beecher 1998). Many of these compounds are also antioxidants and have their own reported bioactivities. For example, flavonoids present in tomatoes, such as naringenin, quercetin, and kaempferol, have been shown to inhibit cancer cell proliferation (Campbell, King, et al. 2006; Wang et al. 2003). Additionally, tomato phenolic acids ferulic acid and caffeic acid have been found to be efficient free radical scavengers (Kikuzaki et al. 2002) and to inhibit tumor promotion in mice (M.-T. Huang et al. 1988). These studies suggest that there is likely a synergistic effect of tomato phytochemicals in chronic disease prevention.

108

Metabolomics has been used to as an effective tool for studying biological response to foods and food components (Bondia-Pons et al. 2013; Mensack et al. 2012; P. Pan et al.

2015; Takahashi et al. 2014). Using LC-MS-based metabolomics, thousands of metabolites can be detected in a single biological sample, yielding a global picture of the influence of different dietary interventions or treatments on the metabolic profile.

The objectives of this study were to use an LC-MS-based metabolomics approach to 1) compare the effects of lycopene and red tomatoes on the plasma metabolomes of mice and 2) evaluate whether red, tangerine, and low carotenoid tomato varieties differentially

impact the metabolome. Red, tangerine, and low carotenoid tomatoes were chosen for

this study due to their differing phytochemical profiles, in particular, differing carotenoid

composition and content.

5.3 Materials and Methods

5.3.1 Materials and Chemicals

Methanol (Optima LC/MS grade), water (Optima LC/MS grade), and formic acid were

purchased from Fisher Scientific (Pittsburgh, PA). Tomatidine and trigonelline standards

were purchased from Sigma-Aldrich (St. Louis, MO).

5.3.2 Animals, Diets, and Study Design

The tangerine and low carotenoid tomato powders were prepared at The Ohio State

University (OSU) from freeze-dried tomatoes grown in Fremont, OH at the OSU North

109

Central Agricultural Research Station. The red tomato powder was purchased from

FutureCeuticals (Momence, IL). An AIN-93G semi-purified diet (Research Diets Inc.,

New Brunswick, NJ) was supplemented with either the red, tangerine or low carotenoid

tomato powder at 10% (w/w) or RediVivo (10% lycopene) beadlets (DSM, Heerlen,

Netherlands) at 0.25% (w/w). The AIN-93G diet alone was used for the control group.

The tomato and control diets contained placebo beadlets to mimic the lycopene beadlets

without the lycopene. The red tomato diet and lycopene diet were formulated to deliver

the same amount of lycopene. Diets were stored in the dark at -20 °C to minimize the degradation and oxidation of carotenoids and other phytochemicals over the course of the

study.

The animal study was approved through the Institutional Animal Care and Use

Committee at The Ohio State University. Male C57BL/6 mice were randomized to one of

the five experimental diets: control, lycopene, red tomato, tangerine tomato, or low

carotenoid tomato (54 mice total). Mice were fed every 2 days and allowed to eat ad

libitum. After 4 wk of feeding, mice were sacrificed by CO2 followed by cardiac

puncture. Plasma was collected and stored at -80 °C until analysis by UHPLC-QTOF-

MS.

Plasma samples from the mice in all 5 diet groups were selected for the metabolomics

analysis. Ten mice from the control, lycopene, red tomato, and tangerine tomato groups

and 9 mice from the low carotenoid group were analyzed. Ages, starting weights, and

110

weight changes of the mice analyzed were statistically compared across groups by one-

way analysis of variance (ANOVA) using SPSS 22.0 (IBM Corp., Armonk, NY).

5.3.3 Preparation of Plasma for Metabolomics Analysis

Plasma samples were thawed on ice. To 50 µL of plasma, 100 µL of cold MeOH was

added to precipitate proteins. The resulting solution was then vortexed briefly and placed

on ice for 5 min to aid in protein precipitation. Following this period, the sample was centrifuged at 21,130 x g for 5 min. Of the supernatant, 100 µL was removed and evaporated to dryness under vacuum at ambient temperature using a Speedvac

Concentrator (Thermo Fisher Scientific, Waltham, MA). The dried extract was stored at

-80 °C for no more than 24 h before analysis by UHPLC-QTOF-MS.

Prior to analysis, dried extracts were removed from -80 °C and reconstituted in 100 µL of

water, vortexed briefly, bath sonicated for 15 s, and centrifuged for 5 min at 21,130 x g to remove any insoluble material. The supernatant was then transferred to a 300 µL HPLC vial for analysis. Pooled aliquots of plasma samples were extracted as quality control

(QC) samples for the UHPLC-QTOF-MS analysis to monitor platform stability and reproducibility. A method blank was prepared by carrying out the full extraction procedure, leaving out the plasma.

111

5.3.4 Plasma Analysis by UHPLC-QTOF-MS

Plasma samples were analyzed with a 1290 Infinity UHPLC system coupled to an iFunnel 6550 QTOF-MS (Agilent, Santa Clara, CA). Plasma extracts were injected (5

µL) onto an Acquity UPLC BEH C18 column (100 mm x 2.1 mm, 1.7 µm particle size)

(Waters Corp., Milford, MA) maintained at 50 °C. The run order of all samples was randomized and QCs were injected every 7 samples to monitor instrument performance.

Compounds were eluted with A = water (0.1% formic acid) and B = MeOH (0.1% formic acid). A 30 min gradient was applied at 0.4 mL/min starting at 0% B and holding for 2 min, increasing linearly to 100% B over 19 min, holding at 100% B for 4 min, returning to 0% B over 2 min and re-equilibrating for 3 min. The UHPLC system was coupled without flow splitting to the QTOF-MS via Dual Agilent Jet Stream electrospray ionization (ESI). The following MS parameters were used: gas temperature, 150 °C; gas flow, 15 L/min; nebulizer, 30 psig; sheath gas temperature, 300 °C; sheath gas flow, 12

L/min; VCap, 4000 V; nozzle voltage, 2000 V. Data were acquired in 2 GHz extended dynamic range (EDR) mode with 20K mass resolution in the range of 50-1700 m/z at a scan rate of 3 spectra/s. Samples were analyzed in both positive and negative ionization modes in separate runs.

5.3.5 Data Processing and Statistical Analysis

Raw LC/MS data were processed using the Batch Recursive Feature Extraction option in the Agilent MassHunter Profinder software (version B.06.00). This option extracts mass spectral features and collapses related isotopes and common adducts ([M+Na]+, [M+K]+,

112

+ - [M+NH4] , [M+HCOO] ) into one feature. This is followed by mass and retention time

alignment of all features. A recursive feature extraction was then performed on the raw

data where the mass and retention time results from the untargeted feature extraction in

the first step are used for a targeted feature extraction. This serves to both reduce the

number of false negatives and false positives in the dataset, thereby increasing the quality

of the data exported for differential analysis.

Extracted metabolite features, comprised of a mass, retention time, and intensity, were

exported as compound exchange files (CEF) for further analysis using the Agilent Mass

Profiler Professional (MPP) chemometrics software (version 13.1). Data were baselined to the median of all QC samples. Metabolites that were detected in the method blank

were removed from the dataset. To limit inconsistent features, only those metabolites that

were present in at least 60% of the samples in one or more of the diet groups were

retained. Significant differences between the groups were determined using an ANOVA

with a corrected P < 0.05 (Benjamini-Hochberg false discovery rate (FDR) multiple

testing correction). Compounds differing by a fold change greater than 1.5 between any

group and the control were considered for identification.

5.3.6 Compound Identification

Compounds were identified using a combination of accurate mass, isotope ratios, MS/MS

fragmentation patterns, and authentic standards when available. This information was

compared against published literature and publically available online metabolite

113

databases (Metlin (Smith et al. 2005) and the Human Metabolome Database (Wishart et al. 2013)). Targeted MS/MS experiments for structural identification were conducted using the same UHPLC-QTOF-MS system and mobile phase gradient described

previously. MS/MS data were collected by isolating precursor ions with a quadrupole

resolution of 1.3 amu and using fixed collision energies of 10, 20, and 40 eV in the mass

range of 50-1700 m/z at an acquisition rate of 2 spectra/s.

5.4 Results and Discussion

5.4.1 Animal Ages and Weights

Age and weight can have an impact on metabolism so it was important to assess whether

there were significant differences in age, starting weight, and weight change between the

mice on the 5 different diets. Average ages of the 49 mice at the start of the 4 week study

are shown in Figure 16. While there were differences in age among the mice, the average

ages were not significantly different between the diet groups (P = 0.604). The average

starting weights and weight changes over the course of the study are shown in

Figure 17. Most mice gained weight on the study and there were no significant

differences in starting weight (P = 0.182) and weight change (P = 0.789) between the

different groups. These results suggest that age and weight were not confounding factors

in the metabolomics analysis.

114

14.0

12.0

10.0

8.0 (wk)

6.0 Age

4.0

2.0

0.0 Control Lycopene Red Tomato Tangerine Low Carotenoid Tomato Tomato Diet Group

Figure 16. Average starting ages of mice used in the metabolomics analysis (± SEM). There were no significant differences in age between the diet groups.

115

A 40.00

35.00

30.00 (g)

25.00

Weight 20.00

15.00

Starting 10.00

5.00

0.00 Control Lycopene Red Tomato Tangerine Low Carotenoid Tomato Tomato Diet Group

B 4.50 4.00 3.50 (g) 3.00 2.50 Change 2.00 1.50 Weight 1.00 0.50 0.00 Control Lycopene Red Tomato Tangerine Low Carotenoid Tomato Tomato Diet Group

Figure 17. Average starting weights (A) and weight changes (B) of mice used in the metabolomics analysis (± SEM). There were no significant differences in weight or weight change between the diet groups.

116

5.4.2 Assessment of Metabolomics Data Quality

Pooled plasma samples from all 49 mice were injected every 7 biological samples for

quality control. As the pooled aliquots should be homogeneous, they can be used to

monitor instrument stability and performance over the course of the experiment. The total

ion chromatograms (TICs) of the QC samples from each experiment were visually

compared to assess retention time shift. The chromatography was found to be highly

reproducible with very small differences in retention times across QC samples. The

retention time alignment window was set accordingly at 0.3 min for data processing.

QC samples were also evaluated using principle component analysis (PCA) (Figure 18).

PCA is an effective tool for reducing high dimensional data in order to visualize patterns within a dataset. As shown in Figure 18, the QC samples (blue) cluster together in the

PCA scores plots, illustrating little variation between these replicates and good

instrument stability. Based on assessment of the QC samples, the LC-MS platform was

found to be reproducible and the data of acceptable quality for differential analysis.

117

118

Figure 18. Principle component analysis (PCA) scores plots showing the clustering of the QC samples (blue) in positive (A) and negative (B) ionization modes.

118

5.4.3 The Whole Tomato Has a Greater Impact on the Metabolome than Lycopene

Alone

While lycopene is believed to be one of the bioactive components of the tomato, there is research suggesting that other phytochemicals in the tomato that may be absorbed and exert some biological effect (T. W.-M. Boileau et al. 2003; Canene-Adams et al. 2007).

For the first objective of this study, the plasma metabolite profiles from the lycopene and red tomato fed mice were compared to the plasma metabolite profiles of the mice on the control diet. Between these 3 groups, 14,670 and 12,639 metabolites were detected in positive and negative modes, respectively, with 12,605 (+) and 8,992 (-) metabolites present in at least 60% of any one group. The complexity of the data is illustrated in

Figure 19 with overlaid extracted ion chromatograms of all of the metabolites detected in one QC plasma sample in positive mode. As polar extracts of the plasma were analyzed, it should be noted that carotenoids are not part of the metabolites detected. However, small, polar cleavage products of lycopene and other carotenoids may be present.

119

120

Figure 19. Overlaid extracted ion chromatograms showing the thousands of metabolites detected in the mouse plasma by LC- QTOF-MS (ESI+). Inset is zoomed to the region between 9 and 17 minutes.

120

Following a one-way ANOVA, 90 metabolites in positive mode and 79 metabolites in

negative mode were found to be significantly different between the 3 groups with a corrected P < 0.05 (Benjamini-Hochberg false discovery rate (FDR) multiple testing

correction). PCA scores plots based on these significantly different metabolites are shown

in Figure 20. All 3 groups separate well based on these metabolites. In positive mode, the

red tomato group is separated from the lycopene and control groups on the first principle

component, which explains 59.02% of the variation in the data. In negative mode, the red

tomato group is well separated from the control on the first principle component, which

explains 44.17% of the variation in the data. In negative mode, the lycopene and red

tomato groups are clearly different from the control group, but are not as different from

each other. In both ionization modes, the mice within each group clustered relatively

tightly, indicating low biological variability in these metabolites.

Results of the Tukey’s post-hoc test are reported below the PCA scores plots in Figure

20. Of the 90 significantly different metabolites in positive mode, only 29 were

significantly different between the lycopene and control groups, while 84 were

significantly different between the red tomato and control groups. Similarly, in negative

mode 45 of the 79 metabolites were significantly different between the lycopene and

control groups, while 67 were significantly different between the red tomato and control

groups. In both positive and negative modes, more metabolites were altered in response

to the tomato diet than the lycopene diet. These results demonstrate that the tomato diet

had a greater impact on the plasma metabolome than lycopene alone.

121

122

Figure 20. PCA scores plots comparing plasma from the control, lycopene, and red tomato fed mice analyzed in positive (A) and negative (B) ionization modes following a one-way ANOVA. Results of the Tukey HSD Post Hoc test are reported below the PCA plots.

122

Significantly different metabolites that changed by at least 1.5 fold compared to the control group were considered for identification. This slightly reduced the data to 86 and

74 metabolites in positive and negative modes, respectively. Of these metabolites, 31

(36%) in positive mode and 10 (14%) in negative mode were specific to the red tomato group. These are likely exogenous metabolites derived from the tomato diet. Arguably the most challenging and time consuming part of untargeted metabolomics is the identification of metabolites, which involves a combination of accurate mass detection,

MS/MS fragmentation, database and literature searching, and chemical standard confirmation. Tentatively identified plasma metabolites altered with the consumption of either the lycopene or red tomato diet are reported in Table 5. Metabolites detected in only the red tomato group are reported as having a fold change >1000 and have been putatively identified as pyridoxal (vitamin B6), porphobilinogen (porphyrin metabolite), dihydroferuloylglycine (polyphenol metabolite), tomatidine (tomato alkaloid), and pimpifolidine (tomato alkaloid). There were also 10 metabolites detected in both the lycopene and red tomato fed mice, but not in the control group, which could be metabolites derived from lycopene. Further work is needed to structurally identify these compounds and confirm them as lycopene metabolites.

123

Table 5. List of plasma metabolites altered with the consumption of either the lycopene or red tomato diet Mass Retention Mode Molecular Mass Putative Metabolite Fold Change (Da) Time Formula Difference IDa Compared to Controlb (min) from Lycopene Red Theoretical (Δppm) 137.0477 1.00 ESI+ C7H7NO2 -3.6 Trigonelline -1.7 23.6

159.1257 1.23 ESI+ C8H17NO2 -5.0 2-aminooctanoic acid 1.3 2.2

167.0582 5.46 ESI+ C8H9NO3 -3.6 Pyridoxal 1.0 >1000

193.0743 7.90 ESI- C10H11NO3 -0.5 Phenylacetylglycine -1.1 1.9

226.0948 5.05 ESI+ C10H14N2O4 -4.9 Porphobilinogen 1.0 >1000

236.1156 5.94 ESI+ C12H16N2O3 -4.2 Phenylalanyl-Alanine 1.9 5.9

253.0955 7.74 ESI+ C12H15NO5 -0.4 Dihydroferuloylglycine 1.0 >1000 124 347.0625 6.21 ESI+ C10H14N5O7P -3.2 AMP/dGMP -1.1 2.6

352.2610 20.85 ESI- C21H36O4 -2.6 MG(18:3) -1.6 -1.4

354.2760 21.23 ESI- C21H38O4 -4.5 MG(18:2) -2.3 -1.4

415.3447 15.13 ESI+ C27H45NO2 -2.2 Tomatidine 1.0 >1000

431.3390 13.87 ESI+ C27H45NO3 -3.5 Pimpifolidine 1.0 >1000 MG: Monoglyceride a Bolded metabolites have been confirmed using authentic standards. b Highlighted metabolites were significantly different from the control group (P < 0.05).

124

5.4.4 Differentiating Metabolites Include Tomato Alkaloids

Two of the metabolites found to differentiate the red tomato fed mice from the control

and lycopene fed mice were positively identified as the tomato alkaloids tomatidine and trigonelline. Alkaloids are a diverse group of cyclic nitrogen containing compounds that can be found as secondary metabolites in many different plant species. The compound with a mass of 415.3447 at 15.13 min was identified as tomatidine based on comparison of mass, MS/MS fragment ions, and retention time with an authentic standard. Similarly, the metabolite with a mass of 138.0549 at 1.00 min was confirmed as trigonelline using an authentic standard. As shown in Figure 21, tomatidine was not detected in the control or lycopene fed mice and trigonelline was only detected at very low levels in some of the mice in those groups. Both metabolites were detected in the plasma of all mice in the red tomato fed group. Based on the positive identification of tomatidine and trigonelline in plasma, metabolite 431.3390 at 13.87 min has been tentatively identified as the tomato alkaloid pimpifolidine.

125

126

Figure 21. Extracted ion chromatograms for the plasma metabolites tomatidine (A) and trigonelline (B) in the control (top), lycopene (middle), and red tomato (bottom) fed mice.

126

Tomatidine is the aglycone of the tomato glycoalkaloid α-tomatine (Figure 22).

Tomatidine can be formed from the acid or enzymatic hydrolysis of the tetrasaccharide side chain (lycotetraose) of α-tomatine. In the plant, tomatine is believed to protect against fungi, bacteria, and pests, possibly through the disruption of cell membranes

(Friedman 2002). α-tomatine is highest in green tomatoes and is enzymatically degraded

during the ripening process. Friedman & Levin (1995) have measured α-tomatine in unripe green tomatoes at approximately 1.6 mg/100 g. Ripe red tomatoes were reported to have approximately 0.03 mg/100 g, while cherry tomatoes were found to have around

9 times more α-tomatine than the standard red tomato.

127

Figure 22. Structures of the tomato alkaloids α-tomatine, tomatidine, pimpifolidine, and trigonelline.

128

Research has suggested that tomatine may be a biologically active phytochemical in

animals and humans. Tomatine has been reported to possess anticancer,

immunopotentiating, and cholesterol lowering properties. For example, in vitro tomatine

has been shown to inhibit the growth of human colon and liver cancer cells (K.-R. Lee et al. 2004), prevent the metastasis of lung cancer cells (Shieh et al. 2011), and induce apoptosis in prostate cancer cells (S.-T. Lee et al. 2011). In mice, tomatine has been found to suppress androgen-independent prostate cancer tumors (S.-T. Lee, Wong, He, et al. 2013) and enhance prostate cancer cell death when administered in combination with the cancer drug paclitaxel (S.-T. Lee, Wong, Hooper, et al. 2013). While some research has indicated that tomatidine is a less potent bioactive than tomatine (Choi et al. 2012), studies have still supported tomatidine as a contributing bioactive compound. Tomatidine has been found to inhibit the invasion of human lung cancer cells (Yan et al. 2013) and reduce inflammation in LPS-induced macrophages (Chiu and Lin 2008). In vivo, dietary tomatidine was shown to lessen hyperlipidemia and atherosclerosis in mice (Fujiwara et al. 2012).

Tomatine has been reported to inhibit cholesterol absorption by forming insoluble complexes with cholesterol in the small intestine (Cayen 1971; Friedman et al. 2000). It is because of this property that tomatine is believed to have low bioavailability. In fact, while tomatine (m/z 1034.5539) was easily detectable in the red tomato diet, it was not detected in the plasma of those mice. A small amount of tomatidine was also detectable in the red tomato diet so it is unclear as to whether the aglycone is absorbed from the diet

129 or formed from the hydrolysis of tomatine in vivo and then absorbed. Further studies are needed to better understand the absorption and metabolism of tomatine and tomatidine.

Trigonelline (N-methylnicotinic acid) (Figure 22) is a pyridine alkaloid derived from niacin (vitamin B3 or nicotinic acid) and is believed to have a number of functions in plants, including cell cycle regulation and signal transduction (Ashihara et al. 2015).

Trigonelline has been reported in tomatoes (Le Gall et al. 2003; Mirnezhad et al. 2010;

Moco et al. 2008) and has been found to accumulate in response to salt stress

(Rajasekaran et al. 2001). It is present in urine and plasma (Lang et al. 2008) and increases with the consumption of trigonelline rich foods, such as coffee (Lang et al.

2010). As trigonelline was present in the red tomato diet, it appears that the mice in this group absorbed it intact from the diet.

In the plasma of the control and lycopene fed mice, trigonelline was detectable at low levels. The control and lycopene diets do not contain trigonelline, but do contain niacin

(nicotinic acid). While it is not a major niacin metabolite, trigonelline has been reported to be excreted in urine after the administration of nicotinic acid (Kodicek and Wang

1941; Kutscher and Ackermann 1933). Therefore, we hypothesize that trigonelline in the plasma of the control and lycopene fed mice is derived from the metabolism of niacin, while the elevated levels found in the plasma of the red tomato fed mice is the result of absorption of trigonelline from the diet.

130

Trigonelline has been shown to possess biological activity, including hypoglycemic,

neuroprotective, and anticancer properties. Zhou, Zhou, & Zeng (2013) report significantly decreased blood glucose and lipid levels in diabetic rats treated with trigonelline for 4 wk. Similarly, Yoshinari, Sato, & Igarashi (2009) observed improved glucose tolerance and lower serum and liver triglyceride levels in rats fed a trigonelline rich diet. Trigonelline has also been shown to regenerate dendrites and axons in rat cortical neurons and improve memory in mice (Tohda et al. 2005). Reported anticancer effects of trigonelline include inhibiting invasion of hepatoma cells (Hirakawa et al.

2005) and enhancing anticancer therapy through suppression of transcription factor Nrf2 activation in pancreatic cancer cells (Arlt et al. 2013).

Little research exists on the tomato alkaloid pimpifolidine (Figure 22). Pimpifolidine and its isomer, 22-isopimpifolidine, have been reported in tomato roots (Ripperger and Porzel

1994) and the tetrasaccharide lycotetraosyl-22-isopimpifolidine has been reported in tomato fruits (Yahara et al. 2004). Based on the confirmed presence of other tomato alkaloid aglycones in plasma, the identification of pimpifolidine is reasonable. This compound was only detected in the tomato fed mice and not in the control or lycopene fed mice. Additionally, a compound with a mass corresponding to lycotetraosyl- pimpifolidine (m/z 1150.5468) giving a pimpifolidine in-source fragment was detected in the red tomato diet. More work is needed to confirm the identification of pimpifolidine in plasma, but our data suggest that, like tomatine, the glycoalkaloid form is ingested and hydrolyzed to the aglycone, which is then absorbed.

131

5.4.5 Tomato Varieties Alter the Plasma Metabolome in Unique Ways

While the red tomato variety is the most commonly consumed, there is interest in other tomato varieties and their impacts on health. For the second objective of this study, we evaluated whether red, tangerine, and low carotenoid tomato varieties differentially impact the plasma metabolome of mice. Using a one-way ANOVA, 125 (+) and 96 (-) metabolites were found to be statistically different (P < 0.05; Benjamini-Hochberg false discovery rate multiple testing correction) between the 5 diet groups and have a fold change greater than 1.5 compared to the control group. PCA was used to visualize clustering differences between the groups based on these significantly different metabolites (Figure 23). The first two principle components explain 65% and 57% of the variation in the data in positive and negative modes, respectively. As expected, the tomato diets had an impact on the plasma metabolome and the tomato fed mice clustered away from the control mice. However, the red tomato fed mice clearly separated from the tangerine tomato and low carotenoid tomato fed mice on the first principle component.

Additionally, the tangerine tomato and low carotenoid tomato groups clustered together in the PCA scores plot and could not be distinguished based on the significantly different metabolites detected in positive or negative modes. So while all of the tomato diets altered the plasma metabolome, the different varieties had unique effects.

132

133

Figure 23. PCA scores plots comparing plasma from the control, lycopene, red tomato, tangerine tomato, and low carotenoid tomato fed mice analyzed in positive (A) and negative (B) ionization modes.

133

The tomato alkaloids identified in the first objective were compared across all 5 diet

groups (Figure 24). All 3 metabolites were higher in the tomato fed mice, with tomatidine and pimpifolidine being absent in the control and lycopene fed groups. Additionally, the

3 tomato alkaloids were most abundant in the red tomato fed group, while levels in the tangerine tomato and low carotenoid tomato groups were more similar. Tomatidine, trigonelline, and pimpifolidine were not significantly different between the tangerine and low carotenoid groups, which supports the clustering observed in the PCA scores plots in

Figure 23. Trends in these plasma metabolites reflect relative levels of tomatine, trigonelline, and lycotetraosyl-pimpifolidine detected in the different tomato diets. These results suggest that tomato alkaloids and their metabolites can be used as novel markers of tomato consumption.

134

Figure 24. Comparison of mean ion intensities (± SEM) for tomatidine (A), trigonelline (B), and pimpifolidine (C) identified in the plasma of the mice on the 5 different diets. Mean values with different letters were significantly different (Tukey HSD; P < 0.05). 135

As illustrated in Figure 23, the red tomato diet altered the plasma metabolome of the mice differently than the tangerine tomato and low carotenoid tomato diets. Upon investigation of the ANOVA results, it was discovered that certain metabolites were only altered by the red tomato diet, while other metabolites were only altered by the tangerine tomato and low carotenoid tomato diets. In fact, in positive mode, 50 metabolites were uniquely altered by the red tomato diet, while only one metabolite was uniquely altered by the low carotenoid diet and two metabolites were uniquely altered by the tangerine tomato diet

Figure 25. However, 34 metabolites were altered by both the low carotenoid and tangerine tomatoes, further demonstrating the similarities between the plasma metabolomes of these two groups. Metabolites significantly altered by the tomato diets are plotted by mass versus retention time in Figure 26 and are colored to show those altered by the red tomato diet only (red), both the tangerine and low carotenoid tomato diets (green), and all three tomatoes (blue). Interestingly, the metabolites altered by the red tomato diet only eluted earlier in the chromatographic run than those altered by the tangerine and low carotenoid tomato diets. These results suggest that metabolites uniquely altered by the red tomato diet are smaller, more polar compounds than those uniquely altered by the tangerine and low carotenoid tomato diets. Work is ongoing to identify these metabolites, but they appear to have considerable structural differences. As the tangerine and low carotenoids tomatoes used in this study were grown at OSU, while the red tomatoes were from FutureCeuticals, more research is needed to verify that differences in handling and storage are not contributing the observed results.

136

Figure 25. Venn diagram comparing ESI+ significantly different metabolites detected in each of the tomato groups.

137

Figure 26. Mass versus retention time plot of ESI+ plasma metabolites altered by the red tomato only, the tangerine and low carotenoid tomatoes only, and all three tomatoes.

138

5.5 Conclusions

We have demonstrated that an MS-based untargeted metabolomics approach can be used to identify changes to the global plasma metabolome in response to different dietary interventions. Our analysis has revealed that tomatoes more drastically alter the plasma metabolomes of mice than lycopene alone. This suggests that other phytochemicals may be contributing to the biological effects associated with tomato consumption.

Differentiating plasma metabolites included both endogenous as well as exogenous compounds from the diets. Most notably, several tomato alkaloids were found to increase in plasma after the consumption of tomatoes. To our knowledge, tomatidine and pimpifolidine have not been reported in plasma previously and may contribute to some of the health promoting properties associated with tomatoes. Results of this study also revealed that the red tomato fed mice had a unique metabolic profile compared to the tangerine and low carotenoid tomato fed mice, suggesting differences in biological effect based on variety. Untargeted metabolomics experiments such as this one will help to better understand the intricate relationship between tomatoes, tomato phytochemicals, and health.

139

REFERENCES

Alder, A., Jamil, M., Marzorati, M., Bruno, M., Vermathen, M., Bigler, P., et al. (2012). The path from β-carotene to carlactone, a strigolactone-like plant hormone. Science, 335(6074), 1348–1351. Ames, B. N., & Gold, L. S. (1997). The causes and prevention of cancer: Gaining perspective. Environmental Health Perspectives, 105(Suppl 4), 865–873. Arlt, A., Sebens, S., Krebs, S., Geismann, C., Grossmann, M., Kruse, M.-L., et al. (2013). Inhibition of the Nrf2 transcription factor by the alkaloid trigonelline renders pancreatic cancer cells more susceptible to apoptosis through decreased proteasomal gene expression and proteasome activity. Oncogene, 32(40), 4825–4835. Ashihara, H., Ludwig, I. A., Katahira, R., Yokota, T., Fujimura, T., & Crozier, A. (2015). Trigonelline and related nicotinic acid metabolites: occurrence, biosynthesis, taxonomic considerations, and their roles in planta and in human health. Phytochemistry Reviews, 14(5), 765–798. Astorg, P., Gradelet, S., Berges, R., & Suschetet, M. (1997). Dietary lycopene decreases the initiation of liver preneoplastic foci by diethylnitrosamine in the rat. Nutrition and Cancer, 29(1), 60–68. Atherton, H. J., Bailey, N. J., Zhang, W., Taylor, J., Major, H., Shockcor, J., et al. (2006). A combined 1H-NMR spectroscopy- and mass spectrometry-based metabolomic study of the PPAR-alpha null mutant mouse defines profound systemic changes in metabolism linked to the metabolic syndrome. Physiological Genomics, 27(2), 178– 186. Auldridge, M. E., McCarty, D. R., & Klee, H. J. (2006). Plant carotenoid cleavage oxygenases and their apocarotenoid products. Current Opinion in Plant Biology, 9(3), 315–321.

140

Aust, O., Ale-Agha, N., Zhang, L., Wollersen, H., Sies, H., & Stahl, W. (2003). Lycopene oxidation product enhances gap junctional communication. Food and Chemical Toxicology, 41(10), 1399–1407. Aust, O., Stahl, W., Sies, H., Tronnier, H., & Heinrich, U. (2005). Supplementation with tomato-based products increases lycopene, phytofluene, and phytoene levels in human serum and protects against UV-light-induced erythema. International Journal for Vitamin and Nutrition Research, 75(1), 54–60. Baker, M. (2011). Metabolomics: from small molecules to big ideas. Nature Methods, 8(2), 117–121. Barua, A. B. (1999). Intestinal absorption of epoxy-beta-carotenes by humans. The Biochemical Journal, 339(Pt 2), 359–362. Barua, A. B., & Olson, J. A. (2001). Xanthophyll epoxides, unlike beta-carotene monoepoxides, are not detectibly absorbed by humans. The Journal of Nutrition, 131(12), 3212–3215. Beecher, G. R. (1998). Nutrient content of tomatoes and tomato products. Experimental Biology and Medicine, 218(2), 98–100. Ben-Dor, A., Nahum, A., Danilenko, M., Giat, Y., Stahl, W., Martin, H.-D., et al. (2001). Effects of acyclo-retinoic acid and lycopene on activation of the retinoic acid receptor and proliferation of mammary cancer cells. Archives of Biochemistry and Biophysics, 391(2), 295–302. Bohm, V., Puspitasari-Nienaber, N. L., Ferruzzi, M. G., & Schwartz, S. J. (2002). Trolox equivalent antioxidant capacity of different geometrical isomers of α-carotene, β- carotene, lycopene, and zeaxanthin. Journal of Agricultural and Food Chemistry, 50(1), 221–226. Boileau, A. C., Merchen, N. R., Wasson, K., Atkinson, C. A., & Erdman, J. W. (1999). Cis-lycopene is more bioavailable than trans-lycopene in vitro and in vivo in lymph- cannulated ferrets. The Journal of Nutrition, 129(6), 1176–1181. Boileau, T. W.-M., Liao, Z., Kim, S., Lemeshow, S., Erdman, J. W., & Clinton, S. K. (2003). Prostate carcinogenesis in N-methyl-N-nitrosourea (NMU)-testosterone- treated rats fed tomato powder, lycopene, or energy-restricted diets. Journal of the National Cancer Institute, 95(21), 1578–1586.

141

Bondia-Pons, I., Barri, T., Hanhineva, K., Juntunen, K., Dragsted, L. O., Mykkänen, H., & Poutanen, K. (2013). UPLC-QTOF/MS metabolic profiling unveils urinary changes in humans after a whole grain rye versus refined wheat bread intervention. Molecular Nutrition and Food Research, 57(3), 412–422. Bosetti, C., Talamini, R., Montella, M., Negri, E., Conti, E., Franceschi, S., & La Vecchia, C. (2004). Retinol, carotenoids and the risk of prostate cancer: a case- control study from Italy. International Journal of Cancer, 112(4), 689–692. Bouatra, S., Aziat, F., Mandal, R., Guo, A. C., Wilson, M. R., Knox, C., et al. (2013). The human urine metabolome. PloS One, 8(9), e73076. Britton, G. (1995). Structure and properties of carotenoids in relation to function. The FASEB Journal, 9(15), 1551–1558. Britton, G., & Goodwin, T. W. (1975). Carotene epoxides from the Delta tomato mutant. Phytochemistry, 14(11), 2530–2532. Brown, P. N., Murch, S. J., & Shipley, P. (2012). Phytochemical diversity of cranberry (Vaccinium macrocarpon Aiton) cultivars by anthocyanin determination and metabolomic profiling with chemometric analysis. Journal of Agricultural and Food Chemistry, 60(1), 261–271. Burton, G. W., & Traber, M. G. (1990). Vitamin E: antioxidant activity, biokinetics, and bioavailability. Annual Review of Nutrition, 10, 357–382. Burton-Freeman, B., & Reimers, K. (2011). Tomato consumption and health: Emerging benefits. American Journal of Lifestyle Medicine, 5(2), 182–191. Campbell, J. K., Engelmann, N. J., Lila, M. A., & Erdman, J. W. (2007). Phytoene, phytofluene, and lycopene from tomato powder differentially accumulate in tissues of male Fisher 344 rats. Nutrition Research, 27(12), 794–801. Campbell, J. K., King, J. L., Harmston, M., Lila, M. A., & Erdman, J. W. (2006). Synergistic effects of flavonoids on cell proliferation in Hepa-1c1c7 and LNCaP cancer cell lines. Journal of Food Science, 71(4), S358–S363. Campbell, J. K., Stroud, C. K., Nakamura, M. T., Lila, M. A., & Erdman, J. W. (2006). Serum testosterone is reduced following short-term phytofluene, lycopene, or tomato powder consumption in F344 rats. The Journal of Nutrition, 136(11), 2813– 2819.

142

Canene-Adams, K., Lindshield, B. L., Wang, S., Jeffery, E. H., Clinton, S. K., & Erdman, J. W. (2007). Combinations of tomato and broccoli enhance antitumor activity in Dunning r3327-h prostate adenocarcinomas. Cancer Research, 67(2), 836–843. Capanoglu, E., Beekwilder, J., Boyacioglu, D., Hall, R., & de Vos, R. (2008). Changes in antioxidant and metabolite profiles during production of tomato paste. Journal of Agricultural and Food Chemistry, 56(3), 964–973. Caris-Veyrat, C., Schmid, A., Carail, M., & Böhm, V. (2003). Cleavage products of lycopene produced by in vitro oxidations: characterization and mechanisms of formation. Journal of Agricultural and Food Chemistry, 51(25), 7318–7325. Cayen, M. N. (1971). Effect of dietary tomatine on cholesterol metabolism in the rat. Journal of Lipid Research, 12(4), 482–490. Chambers, M. C., Maclean, B., Burke, R., Amodei, D., Ruderman, D. L., Neumann, S., et al. (2012). A cross-platform toolkit for mass spectrometry and proteomics. Nature Biotechnology, 30(10), 918–920. Chang, W. T., Huang, W. C., & Liou, C. J. (2012). Evaluation of the anti-inflammatory effects of phloretin and phlorizin in lipopolysaccharide-stimulated mouse macrophages. Food Chemistry, 134(2), 972–979. Chasse, G. A., Mak, M. L., Deretey, E., Farkas, I., Torday, L. L., Papp, J. G., et al. (2001). An ab initio computational study on selected lycopene isomers. Journal of Molecular Structure: THEOCHEM, 571(1-3), 27–37. Chen, Y. J., Huang, X., Mahieu, N. G., Cho, K., Schaefer, J., & Patti, G. J. (2014). Differential incorporation of glucose into biomass during Warburg metabolism. Biochemistry, 53(29), 4755–4757. Cheng, H.-C., Chien, H., Liao, C.-H., Yang, Y.-Y., & Huang, S.-Y. (2007). Carotenoids suppress proliferating cell nuclear antigen and cyclin D1 expression in oral carcinogenic models. The Journal of Nutritional Biochemistry, 18(10), 667–675. Chiu, F.-L., & Lin, J.-K. (2008). Tomatidine inhibits iNOS and COX-2 through suppression of NF-kappaB and JNK pathways in LPS-stimulated mouse macrophages. FEBS Letters, 582(16), 2407–2412. Cho, K., Mahieu, N., Ivanisevic, J., Uritboonthai, W., Chen, Y.-J., Siuzdak, G., & Patti, G. J. (2014). isoMETLIN: A database for isotope-based metabolomics. Analytical Chemistry, 86(19), 9358–9361.

143

Choi, S. H., Ahn, J.-B., Kozukue, N., Kim, H.-J., Nishitani, Y., Zhang, L., et al. (2012). Structure-activity relationships of α-, β1-, γ-, and δ-tomatine and tomatidine against human breast (MDA-MB-231), gastric (KATO-III), and prostate (PC3) cancer cells. Journal of Agricultural and Food Chemistry, 60(15), 3891–3899. Chopra, M., Fitzsimons, P. E., Strain, J. J., Thurnham, D. I., & Howard, A. N. (2000). Nonalcoholic red wine extract and quercetin inhibit LDL oxidation without affecting plasma antioxidant vitamin and carotenoid concentrations. Clinical Chemistry, 46(8), 1162–1170. Chung, J., Koo, K., Lian, F., Hu, K. Q., Ernst, H., & Wang, X.-D. (2012). Apo-10 ’- lycopenoic acid, a lycopene metabolite, increases sirtuin 1 mRNA and protein levels and decreases hepatic fat accumulation in ob/ob mice. The Journal of Nutrition, 142(3), 405–410. Clendinen, C. S., Stupp, G. S., Ajredini, R., Lee-McMullen, B., Beecher, C., & Edison, A. S. (2015). An overview of methods using 13C for improved compound identification in metabolomics and natural products. Frontiers in Plant Science, 6(611), 1–13. Clinton, S. K. (1998). Lycopene: chemistry, biology, and implications for human health and disease. Nutrition Reviews, 56(2), 35–51. Clinton, S. K., Emenhiser, C., Schwartz, S. J., Bostwick, D. G., Williams, A. W., Moore, B. J., & Erdman, J. W. (1996). cis-trans lycopene isomers, carotenoids, and retinol in the human prostate. Cancer Epidemiology, Biomarkers & Prevention, 5(10), 823– 833. Clough, J. M., & Pattenden, G. (1979). Naturally occurring poly-cis carotenoids. Stereochemistry of poly-cis lycopene and its congeners in “Tangerine” tomato fruits. Journal of the Chemical Society, Chemical Communications, (14), 616–619. Cohen, J. H., Kristal, A. R., & Stanford, J. L. (2000). Fruit and vegetable intakes and prostate cancer risk. Journal of the National Cancer Institute, 92(1), 61–68. Commenges, D., Scotet, V., Renaud, S., Jacqmin-Gadda, H., Barberger-Gateau, P., & Dartigues, J. F. (2000). Intake of flavonoids and risk of dementia. European Journal of Epidemiology, 16(4), 357–363.

144

Cooperstone, J. L., Ralston, R. A., Riedl, K. M., Haufe, T. C., Schweiggert, R. M., King, S. A., et al. (2015). Enhanced bioavailability of lycopene when consumed as cis- isomers from tangerine compared to red tomato juice, a randomized, cross-over clinical trial. Molecular Nutrition and Food Research, 59(4), 658–669. Cunningham, F. X., & Gantt, E. (1998). Genes and enzymes of carotenoid biosynthesis in plants. Annual Review of Plant Physiology and Plant Molecular Biology, 49, 557– 583. De Vos, R. C., Moco, S., Lommen, A., Keurentjes, J. J., Bino, R. J., & Hall, R. D. (2007). Untargeted large-scale plant metabolomics using liquid chromatography coupled to mass spectrometry. Nature Protocols, 2(4), 778–791. dela Seña, C., Narayanasamy, S., Riedl, K. M., Curley, R. W., Schwartz, S. J., & Harrison, E. H. (2013). Substrate specificity of purified recombinant human β- carotene 15,15’-oxygenase (BCO1). The Journal of Biological Chemistry, 288(52), 37094–37103. Devi, M. A., & Das, N. P. (1993). In vitro effects of natural plant polyphenols on the proliferation of normal and abnormal human lymphocytes and their secretions of interleukin-2. Cancer Letters, 69(3), 191–196. Di Mascio, P., Kaiser, S., & Sies, H. (1989). Lycopene as the most efficient biological carotenoid singlet oxygen quencher. Archives of Biochemistry and Biophysics, 274(2), 532–538. Dobson, G., Shepherd, T., Verrall, S. R., Conner, S., McNicol, J. W., Ramsay, G., et al. (2008). Phytochemical diversity in tubers of potato cultivars and landraces using a GC-MS metabolomics approach. Journal of Agricultural and Food Chemistry, 56(21), 10280–10291. Duarte, J., Pérez-Palencia, R., Vargas, F., Ocete, M. A., Pérez-Vizcaino, F., Zarzuelo, A., & Tamargo, J. (2001). Antihypertensive effects of the flavonoid quercetin in spontaneously hypertensive rats. British Journal of Pharmacology, 133(1), 117–124. Duitsman, P. K., Barua, A. B., Becker, B., & Olson, J. A. (1999). Effects of epoxycarotenoids, beta-carotene, and retinoic acid on the differentiation and viability of the leukemia cell line NB4 in vitro. International Journal for Vitamin and Nutrition Research, 69(5), 303–308.

145

Dunn, W. B., Broadhurst, D., Begley, P., Zelena, E., Francis-McIntyre, S., Anderson, N., et al. (2011). Procedures for large-scale metabolic profiling of serum and plasma using gas chromatography and liquid chromatography coupled to mass spectrometry. Nature Protocols, 6(7), 1060–1083. Dunn, W. B., & Ellis, D. I. (2005). Metabolomics: Current analytical platforms and methodologies. Trends in Analytical Chemistry, 24(4), 285–294. Emenhiser, C., Simunovic, N., Sander, L. C., & Schwartz, S. J. (1996). Separation of geometrical carotenoid isomers in biological extracts using a polymeric C30 column in reversed-phase liquid chromatography. Journal of Agricultural and Food Chemistry, 44(12), 3887–3893. Erdman, J. W., Bierer, T. L., & Gugger, E. T. (1993). Absorption and transport of carotenoids. Annals of the New York Academy of Sciences, 691, 76–85. Erlund, I. (2004). Review of the flavonoids quercetin, hesperetin, and naringenin. Dietary sources, bioactivities, bioavailability, and epidemiology. Nutrition Research, 24(10), 851–874. Eroglu, A., Hruszkewycz, D. P., dela Sena, C., Narayanasamy, S., Riedl, K. M., Kopec, R. E., et al. (2012). Naturally occurring eccentric cleavage products of provitamin A β-carotene function as antagonists of retinoic acid receptors. The Journal of Biological Chemistry, 287(19), 15886–15895. Etminan, M., Takkouche, B., & Caamaño-Isorna, F. (2004). The role of tomato products and lycopene in the prevention of prostate cancer: a meta-analysis of observational studies. Cancer Epidemiology, Biomarkers & Prevention, 13(3), 340–345. Eugster, C. H. (1995). Chemical derivatization: Microscale tests for the presence of common functional groups in carotenoids. In G. Britton, S. Liaaen-Jensen, & H. Pfander (Eds.), Carotenoids Volume 1A: Isolation and Analysis (pp. 71–80). Basel, Switzerland: Birkhauser Verlag. Exon, J. H., Magnuson, B. A., South, E. H., & Hendrix, K. (1998). Dietary quercetin, immune functions and colonic carcinogenesis in rats. Immunopharmacology and Immunotoxicology, 20(1), 173–190. Failla, M. L., Chitchumroonchokchai, C., & Ishida, B. K. (2008). In vitro micellarization and intestinal cell uptake of cis isomers of lycopene exceed those of all-trans lycopene. The Journal of Nutrition, 138(3), 482–486.

146

Feng, R., Lu, Y., Bowman, L. L., Qian, Y., Castranova, V., & Ding, M. (2005). Inhibition of activator protein-1, NF-κB, and MAPKs and induction of phase 2 detoxifying enzyme activity by chlorogenic acid. Journal of Biological Chemistry, 280(30), 27888–27895. Ferreira, A. L., Yeum, K.-J., Russell, R. M., Krinsky, N. I., & Tang, G. (2004). Enzymatic and oxidative metabolites of lycopene. The Journal of Nutritional Biochemistry, 15(8), 493–502. Fiehn, O. (2002). Metabolomics–the link between genotypes and phenotypes. Plant Molecular Biology, 48(1-2), 155–171. Fleshman, M. K., Lester, G. E., Riedl, K. M., Kopec, R. E., Narayanasamy, S., Curley, R. W., et al. (2011). Carotene and novel apocarotenoid concentrations in orange- fleshed Cucumis melo melons: determinations of β-carotene bioaccessibility and bioavailability. Journal of Agricultural and Food Chemistry, 59(9), 4448–4454. Fleshman, M. K., Riedl, K. M., Novotny, J. A., Schwartz, S. J., & Harrison, E. H. (2012). An LC/MS method for d8-β-carotene and d4-retinyl esters: β-carotene absorption and its conversion to vitamin A in humans. Journal of Lipid Research, 53(4), 820– 827. Foote, C. S., Chang, Y. C., & Denny, R. W. (1970). Chemistry of singlet oxygen. XI. Cis-trans isomerization of carotenoids by single oxygen and a probable quenching mechanism. Journal of the American Chemical Society, 92(17), 5218–5219. Ford, N. A., Elsen, A. C., Zuniga, K., Lindshield, B. L., & Erdman, J. W. (2011). Lycopene and apo-12’-lycopenal reduce cell proliferation and alter cell cycle progression in human prostate cancer cells. Nutrition and Cancer, 63(2), 256–263. Ford, N. A., & Erdman, J. W. (2012). Are lycopene metabolites metabolically active? Acta Biochimica Polonica, 59(1), 1–4. Friedman, M. (2002). Tomato glycoalkaloids: role in the plant and in the diet. Journal of Agricultural and Food Chemistry, 50(21), 5751–5780. Friedman, M., Fitch, T. E., & Yokoyama, W. E. (2000). Lowering of plasma LDL cholesterol in hamsters by the tomato glycoalkaloid tomatine. Food and Chemical Toxicology, 38(7), 549–553.

147

Friedman, M., & Levin, C. E. (1995). α-Tomatine content in tomato and tomato products determined by HPLC with pulsed amperometric detection. Journal of Agricultural and Food Chemistry, 43(6), 1507–1511. Fuhrman, B., Volkova, N., Rosenblat, M., & Aviram, M. (2000). Lycopene synergistically inhibits LDL oxidation in combination with vitamin E, glabridin, rosmarinic acid, carnosic acid, or garlic. Antioxidants & Redox Signaling, 2(3), 491– 506. Fujiwara, Y., Kiyota, N., Tsurushima, K., Yoshitomi, M., Horlad, H., Ikeda, T., et al. (2012). Tomatidine, a tomato sapogenol, ameliorates hyperlipidemia and atherosclerosis in apoE-deficient mice by inhibiting acyl-CoA:cholesterol acyl- transferase (ACAT). Journal of Agricultural and Food Chemistry, 60(10), 2472– 2479. Gahler, S., Otto, K., & Böhm, V. (2003). Alterations of vitamin C, total phenolics, and antioxidant capacity as affected by processing tomatoes to different products. Journal of Agricultural and Food Chemistry, 51(27), 7962–7968. Gajic, M., Zaripheh, S., Sun, F., & Erdman, J. W. (2006). Apo-8′-lycopenal and apo-12′- lycopenal are metabolic products of lycopene in rat liver. The Journal of Nutrition, 136(6), 1552–1557. Garner, C. E., Sumner, S. C. J., Davis, J. G., Burgess, J. P., Yueh, Y., Demeter, J., et al. (2006). Metabolism and disposition of 1-bromopropane in rats and mice following inhalation or intravenous administration. Toxicology and Applied Pharmacology, 215(1), 23–36. Gauthier, K. M., Yang, W., Gross, G. J., & Campbell, W. B. (2007). Roles of epoxyeicosatrienoic acids in vascular regulation and cardiac preconditioning. Journal of Cardiovascular Pharmacology, 50(6), 601–608. Gibney, M. J., Walsh, M., Brennan, L., Roche, H. M., German, B., & van Ommen, B. (2005). Metabolomics in human nutrition: opportunities and challenges. The American Journal of Clinical Nutrition, 82(3), 497–503. Giovannucci, E. (1999). Tomatoes, tomato-based products, lycopene, and cancer: review of the epidemiologic literature. Journal of the National Cancer Institute, 91(4), 317– 331. Giovannucci, E. (2002). A review of epidemiologic studies of tomatoes, lycopene, and prostate cancer. Experimental Biology and Medicine, 227(10), 852–859. 148

Giovannucci, E., Ascherio, A., Rimm, E. B., Stampfer, M. J., Colditz, G. A., & Willett, W. C. (1995). Intake of carotenoids and retinol in relation to risk of prostate cancer. Journal of the National Cancer Institute, 87(23), 1767–1776. Gómez-Romero, M., Segura-Carretero, A., & Fernández-Gutiérrez, A. (2010). Metabolite profiling and quantification of phenolic compounds in methanol extracts of tomato fruit. Phytochemistry, 71(16), 1848–1864. Gonthier, M.-P., Verny, M.-A., Besson, C., Rémésy, C., & Scalbert, A. (2003). Chlorogenic acid bioavailability largely depends on its metabolism by the gut microflora in rats. The Journal of Nutrition, 133(6), 1853–1859. Guertin, K. A., Loftfield, E., Boca, S. M., Sampson, J. N., Moore, S. C., Xiao, Q., et al. (2015). Serum biomarkers of habitual coffee consumption may provide insight into the mechanism underlying the association between coffee consumption and colorectal cancer. The American Journal of Clinical Nutrition, 101(5), 1000–1011. Hadley, C. W., Miller, E. C., Schwartz, S. J., & Clinton, S. K. (2002). Tomatoes, lycopene, and prostate cancer: progress and promise. Experimental Biology and Medicine, 227(10), 869–880. Hall, R. D., Brouwer, I. D., & Fitzgerald, M. A. (2008). Plant metabolomics and its potential application for human nutrition. Physiologia Plantarum, 132(2), 162–175. Handelman, G. J., Van Kuijk, F. J. G. M., Chatterjee, A., & Krinsky, N. I. (1991). Characterization of products formed during the autoxidation of β-carotene. Free Radical Biology and Medicine, 10(6), 427–437. Hantz, H. L., Young, L. F., & Martin, K. R. (2005). Physiologically attainable concentrations of lycopene induce mitochondrial apoptosis in LNCaP human prostate cancer cells. Experimental Biology and Medicine, 230(3), 171–179. Hayes, R. B., Ziegler, R. G., Gridley, G., Swanson, C., Greenberg, R. S., Swanson, G. M., et al. (1999). Dietary factors and risks for prostate cancer among blacks and whites in the United States. Cancer Epidemiology, Biomarkers & Prevention, 8(1), 25–34. Heim, K. E., Tagliaferro, A. R., & Bobilya, D. J. (2002). Flavonoid antioxidants: chemistry, metabolism and structure-activity relationships. Journal of Nutritional Biochemistry, 13(10), 572–584.

149

Henry, L. K., Catignani, G. L., & Schwartz, S. J. (1998). Oxidative degradation kinetics of lycopene , lutein , and 9-cis and all-trans β-carotene. Journal of the American Oil Chemists’ Society, 75(7), 823–829. Hertog, M. G., Feskens, E. J., Hollman, P. C., Katan, M. B., & Kromhout, D. (1993). Dietary antioxidant flavonoids and risk of coronary heart disease: the Zutphen Elderly Study. Lancet, 342(8878), 1007–1011. Hirakawa, N., Okauchi, R., Miura, Y., & Yagasaki, K. (2005). Anti-invasive activity of niacin and trigonelline against cancer cells. Bioscience, Biotechnology, and Biochemistry, 69(3), 653–658. Hirsch, K., Atzmon, A., Danilenko, M., Levy, J., & Sharoni, Y. (2007). Lycopene and other carotenoids inhibit estrogenic activity of 17 β-estradiol and genistein in cancer cells. Breast Cancer Research and Treatment, 104(2), 221–230. Holcapek, M., Jandera, P., Zderadicka, P., & Hrubá, L. (2003). Characterization of triacylglycerol and diacylglycerol composition of plant oils using high-performance liquid chromatography-atmospheric pressure chemical ionization mass spectrometry. Journal of Chromatography A, 1010(2), 195–215. Houle, B., Rochette-Egly, C., & Bradley, W. E. C. (1993). Tumor-suppressive effect of the retinoic acid receptor β in human epidermoid lung cancer cells. Proceedings of the National Academy of Sciences of the United States of America, 90(3), 985–989. Hu, K.-Q., Liu, C., Ernst, H., Krinsky, N. I., Russell, R. M., & Wang, X.-D. (2006). The biochemical characterization of ferret carotene-9’,10'-monooxygenase catalyzing cleavage of carotenoids in vitro and in vivo. The Journal of Biological Chemistry, 281(28), 19327–19338. Huang, M.-T., Smart, R. C., Wong, C.-Q., & Conney, A. H. (1988). Inhibitory effect of curcumin, chlorogenic acid, caffeic acid, and ferulic acid on tumor promotion in mouse skin by 12-O-tetradecanoylphorbol-13-acetate. Cancer Research, 48(21), 5941–5946. Huang, W. C., Chang, W. T., Wu, S. J., Xu, P. Y., Ting, N. C., & Liou, C. J. (2013). Phloretin and phlorizin promote lipolysis and inhibit inflammation in mouse 3T3-L1 cells and in macrophage-adipocyte co-cultures. Molecular Nutrition and Food Research, 57(10), 1803–1813.

150

Huang, X., Chen, Y.-J., Cho, K., Nikolskiy, I., Crawford, P. A., & Patti, G. J. (2014). X13CMS: global tracking of isotopic labels in untargeted metabolomics. Analytical Chemistry, 86(3), 1632–1639. Hussain, S. P., Hofseth, L. J., & Harris, C. C. (2003). Radical causes of cancer. Nature Reviews Cancer, 3(4), 276–285. Hwang, E.-S., & Bowen, P. E. (2004). Cell cycle arrest and induction of apoptosis by lycopene in LNCaP human prostate cancer cells. Journal of Medicinal Food, 7(3), 284–289. Jacob, K., Periago, M. J., Böhm, V., & Berruezo, G. R. (2008). Influence of lycopene and vitamin C from tomato juice on biomarkers of oxidative stress and inflammation. British Journal of Nutrition, 99(1), 137–146. Jandrić, Z., Roberts, D., Rathor, M. N., Abrahim, A., Islam, M., & Cannavan, A. (2014). Assessment of fruit juice authenticity using UPLC-QToF MS: a metabolomics approach. Food Chemistry, 148, 7–17. Johanningsmeier, S. D., & McFeeters, R. F. (2011). Detection of volatile spoilage metabolites in fermented cucumbers using nontargeted, comprehensive 2- dimensional gas chromatography-time-of-flight mass spectrometry (GCxGC- TOFMS). Journal of Food Science, 76(1), C168–C177. Johnston, K. L., Clifford, M. N., & Morgan, L. M. (2003). Coffee acutely modifies gastrointestinal hormone secretion and glucose tolerance in humans: glycemic effects of chlorogenic acid and caffeine. American Journal of Clinical Nutrition, 78(4), 728–733. Khachik, F., Beecher, G. R., Goli, M. B., Lusby, W. R., & Smith, J. C. (1992). Separation and identification of carotenoids and their oxidation products in the extracts of human plasma. Analytical Chemistry, 64(18), 2111–2122. Khachik, F., Beecher, G. R., & Smith, J. C. J. (1995). Lutein, lycopene, and their oxidative metabolites in chemoprevention of cancer. Journal of Cellular Biochemistry, 22, 236–246. Khachik, F., Goli, M. B., Beecher, G. R., Holden, J., Lusby, W. R., Tenorio, M. D., & Barrera, M. R. (1992). Effect of food preparation on qualitative and quantitative distribution of major carotenoid constituents of tomatoes and several green vegetables. Journal of Agricultural and Food Chemistry, 40(3), 390–398.

151

Khachik, F., Spangler, C. J., Smith, J. C., Canfield, L. M., Steck, A., & Pfander, H. (1997). Identification, quantification, and relative concentrations of carotenoids and their metabolites in human milk and serum. Analytical Chemistry, 69(10), 1873– 1881. Kiefer, C., Hessel, S., Lampert, J. M., Vogt, K., Lederer, M. O., Breithaupt, D. E., & von Lintig, J. (2001). Identification and characterization of a mammalian enzyme catalyzing the asymmetric oxidative cleavage of provitamin A. The Journal of Biological Chemistry, 276(17), 14110–14116. Kikuzaki, H., Hisamoto, M., Hirose, K., Akiyama, K., & Taniguchi, H. (2002). Antioxidant properties of ferulic acid and its related compounds. Journal of Agricultural and Food Chemistry, 50(7), 2161–2168. Kim, M.-S., Kwon, J. Y., Kang, N. J., Lee, K. W., & Lee, H. J. (2009). Phloretin induces apoptosis in H-Ras MCF10A human breast tumor cells through the activation of p53 via JNK and p38 mitogen-activated protein kinase signaling. Annals of the New York Academy of Sciences, 1171, 479–483. Kim, S. J., Nara, E., Kobayashi, H., Terao, J., & Nagao, A. (2001). Formation of cleavage products by autoxidation of lycopene. Lipids, 36(2), 191–199. Kirsh, V. A., Mayne, S. T., Peters, U., Chatterjee, N., Leitzmann, M. F., Dixon, L. B., et al. (2006). A prospective study of lycopene and tomato product intake and risk of prostate cancer. Cancer Epidemiology, Biomarkers & Prevention, 15(1), 92–98. Knekt, P., Järvinen, R., Seppänen, R., Hellövaara, M., Teppo, L., Pukkala, E., & Aromaa, A. (1997). Dietary flavonoids and the risk of lung cancer and other malignant neoplasms. American Journal of Epidemiology, 146(3), 223–230. Kobori, M., Shinmoto, H., Tsushida, T., & Shinohara, K. (1997). Phloretin-induced apoptosis in B16 melanoma 4A5 cells by inhibition of glucose transmembrane transport. Cancer Letters, 119(2), 207–212. Kochhar, S., Jacobs, D. M., Ramadan, Z., Berruex, F., Fuerholz, A., & Fay, L. B. (2006). Probing gender-specific metabolism differences in humans by nuclear magnetic resonance-based metabonomics. Analytical Biochemistry, 352(2), 274–281. Kodicek, E., & Wang, Y. L. (1941). A new method for the estimation of trigonelline in urine and foodstuffs. Nature, 148(3740), 23–24.

152

Kopec, R. E., Cooperstone, J. L., Cichon, M. J., & Schwartz, S. J. (2012). Analysis methods of carotenoids. In Z. Xu & L. R. Howard (Eds.), Analysis of Antioxidant Rich Phytochemicals (pp. 105–148). Blackwell, Oxford, UK: Wiley-Blackwell. Kopec, R. E., Riedl, K. M., Harrison, E. H., Curley, R. W., Hruszkewycz, D. P., Clinton, S. K., & Schwartz, S. J. (2010). Identification and quantification of apo-lycopenals in fruits, vegetables, and human plasma. Journal of Agricultural and Food Chemistry, 58(6), 3290–3296. Kozukue, N., & Friedman, M. (2003). Tomatine, chlorophyll, β-carotene and lycopene content in tomatoes during growth and maturation. Journal of the Science of Food and Agriculture, 83(3), 195–200. Kucuk, O., Sarkar, F. H., Sakr, W., Djuric, Z., Pollak, M. N., Khachik, F., et al. (2001). Phase II randomized clinical trial of lycopene supplementation before radical prostatectomy. Cancer Epidemiology, Biomarkers & Prevention, 10(8), 861–868. Kuo, S.-M. (1997). Dietary flavonoid and cancer prevention: evidence and potential mechanism. Critical Reviews in Oncogenesis, 8(1), 47–69. Kurilich, A. C., Britz, S. J., Clevidence, B. A., & Novotny, J. A. (2003). Isotopic labeling and LC-APCI-MS quantification for investigating absorption of carotenoids and phylloquinone from kale (Brassica oleracea). Journal of Agricultural and Food Chemistry, 51(17), 4877–4883. Kutscher, F., & Ackermann, D. (1933). The comparative biochemistry of vertebrates and invertebrates. Annual Review of Biochemistry, 2, 355–376. Lang, R., Wahl, A., Skurk, T., Yagar, E. F., Schmiech, L., Eggers, R., et al. (2010). Development of a hydrophilic liquid interaction chromatography-high-performance liquid chromatography-tandem mass spectrometry based stable isotope dilution analysis and pharmacokinetic studies on bioactive pyridines in human plasma and urine after coffee. Analytical Chemistry, 82(4), 1486–1497. Lang, R., Yagar, E. F., Eggers, R., & Hofmann, T. (2008). Quantitative investigation of trigonelline, nicotinic acid, and nicotinamide in foods, urine, and plasma by means of LC-MS/MS and stable isotope dilution analysis. Journal of Agricultural and Food Chemistry, 56(23), 11114–11121.

153

Le Gall, G., Colquhoun, I. J., Davis, A. L., Collins, G. J., & Verhoeyen, M. E. (2003). Metabolite profiling of tomato (Lycopersicon esculentum) using 1H NMR spectroscopy as a tool to detect potential unintended effects following a genetic modification. Journal of Agricultural and Food Chemistry, 51(9), 2447–2456. Lee, J. H., Regmi, S. C., Kim, J. A., Cho, M. H., Yun, H., Lee, C. S., & Lee, J. (2011). Apple flavonoid phloretin inhibits Escherichia coli O157:H7 biofilm formation and ameliorates colon inflammation in rats. Infection and Immunity, 79(12), 4819–4827. Lee, J.-E., Lee, B.-J., Chung, J.-O., Hwang, J.-A., Lee, S.-J., Lee, C.-H., & Hong, Y.-S. (2010). Geographical and climatic dependencies of green tea (Camellia sinensis) metabolites: a 1H NMR-based metabolomics study. Journal of Agricultural and Food Chemistry, 58(19), 10582–10589. Lee, K.-R., Kozukue, N., Han, J.-S., Park, J.-H., Chang, E.-Y., Baek, E.-J., et al. (2004). Glycoalkaloids and metabolites inhibit the growth of human colon (HT29) and liver (HepG2) cancer cells. Journal of Agricultural and Food Chemistry, 52(10), 2832– 2839. Lee, R., West, D., Phillips, S. M., & Britz-McKibbin, P. (2010). Differential metabolomics for quantitative assessment of oxidative stress with strenuous exercise and nutritional intervention: thiol-specific regulation of cellular metabolism with N- acetyl-L-cysteine pretreatment. Analytical Chemistry, 82(7), 2959–2968. Lee, S.-T., Wong, P.-F., Cheah, S.-C., & Mustafa, M. R. (2011). Alpha-tomatine induces apoptosis and inhibits nuclear factor-kappa B activation on human prostatic adenocarcinoma PC-3 cells. PloS One, 6(4), e18915. Lee, S.-T., Wong, P.-F., He, H., Hooper, J. D., & Mustafa, M. R. (2013). Alpha-tomatine attenuation of in vivo growth of subcutaneous and orthotopic xenograft tumors of human prostate carcinoma PC-3 cells is accompanied by inactivation of nuclear factor-kappa B signaling. PloS One, 8(2), e57708. Lee, S.-T., Wong, P.-F., Hooper, J. D., & Mustafa, M. R. (2013). Alpha-tomatine synergises with paclitaxel to enhance apoptosis of androgen-independent human prostate cancer PC-3 cells in vitro and in vivo. Phytomedicine, 20(14), 1297–1305.

154

Lenucci, M. S., Serrone, L., De Caroli, M., Fraser, P. D., Bramley, P. M., Piro, G., & Dalessandro, G. (2012). Isoprenoid, lipid, and protein contents in intact plastids isolated from mesocarp cells of traditional and high-pigment tomato cultivars at different ripening stages. Journal of Agricultural and Food Chemistry, 60(7), 1764– 1775. Lian, F., Smith, D. E., Ernst, H., Russell, R. M., & Wang, X.-D. (2007). Apo-10’- lycopenoic acid inhibits lung cancer cell growth in vitro, and suppresses lung tumorigenesis in the A/J mouse model in vivo. Carcinogenesis, 28(7), 1567–1574. Lindqvist, A., & Andersson, S. (2002). Biochemical properties of purified recombinant human beta-carotene 15,15’-monooxygenase. The Journal of Biological Chemistry, 277(26), 23942–23948. Lindshield, B. L., Canene-Adams, K., & Erdman, J. W. (2007). Lycopenoids: are lycopene metabolites bioactive? Archives of Biochemistry and Biophysics, 458(2), 136–140. Liu, H., Garrett, T. J., Tayyari, F., & Gu, L. (2015). Profiling the metabolome changes caused by cranberry procyanidins in plasma of female rats using (1) H NMR and UHPLC-Q-Orbitrap-HRMS global metabolomics approaches. Molecular Nutrition & Food Research, Epub ahead of print. Liu, Y., Lee, M.-O., Wang, H.-G., Li, Y., Hashimoto, Y., Klaus, M., et al. (1996). Retinoic acid receptor beta mediates the growth-inhibitory effect of retinoic acid by promoting apoptosis in human breast cancer cells. Molecular and Cellular Biology, 16(3), 1138–1149. Livny, O., Kaplan, I., Reifen, R., Polak-Charcon, S., Madar, Z., & Schwartz, B. (2002). Lycopene inhibits proliferation and enhances gap-junction communication of KB-1 human oral tumor cells. The Journal of Nutrition, 132(12), 3754–3759. Mansuy, D. (2007). A brief history of the contribution of metalloporphyrin models to cytochrome P450 chemistry and oxidation catalysis. Comptes Rendus Chimie, 10(4- 5), 392–413. Martínez, A., Stinco, C. M., & Meléndez-Martínez, A. J. (2014). Free radical scavenging properties of phytofluene and phytoene isomers as compared to lycopene: a combined experimental and theoretical study. The Journal of Physical Chemistry B, 118(33), 9819–9825.

155

McGhie, T. K., & Rowan, D. D. (2012). Metabolomics for measuring phytochemicals, and assessing human and animal responses to phytochemicals, in food science. Molecular Nutrition and Food Research, 56(1), 147–158. McQuinn, R. P., Giovannoni, J. J., & Pogson, B. J. (2015). More than meets the eye: from carotenoid biosynthesis, to new insights into apocarotenoid signaling. Current Opinion in Plant Biology, 27, 172–179. Mein, J. R., Lian, F., & Wang, X.-D. (2008). Biological activity of lycopene metabolites: implications for cancer prevention. Nutrition Reviews, 66(12), 667–683. Mensack, M. M., McGinley, J. N., & Thompson, H. J. (2012). Metabolomic analysis of the effects of edible dry beans (Phaseolus vulgaris L.) on tissue lipid metabolism and carcinogenesis in rats. The British Journal of Nutrition, 108(Suppl 1), 155–165. Mercadante, A. Z., & Egeland, E. S. (2004). Carotenoids Handbook. (H. Britton, George; Liaaen-Jensen, Synnove; Pfander, Ed.). Basel, Switzerland: Birkhauser Verlag. Meyer, A. S., Donovan, J. L., Pearson, D. A., Waterhouse, A. L., & Frankel, E. N. (1998). Fruit hydroxycinnamic acids inhibit human low-density lipoprotein oxidation in vitro. Journal of Agricultural and Food Chemistry, 46(5), 1783–1787. Mills, P. K., Beeson, W. L., Phillips, R. L., & Fraser, G. E. (1989). Cohort study of diet, lifestyle, and prostate cancer in Adventist men. Cancer, 64(3), 598–604. Mirnezhad, M., Romero-González, R. R., Leiss, K. A., Choi, Y. H., Verpoorte, R., & Klinkhamera, P. G. L. (2010). Metabolomic analysis of host plant resistance to thrips in wild and cultivated tomatoes. Phytochemical Analysis, 21(1), 110–117. Moazzami, A. A., Zhang, J.-X., Kamal-Eldin, A., Aman, P., Hallmans, G., Johansson, J.- E., & Andersson, S.-O. (2011). Nuclear magnetic resonance-based metabolomics enable detection of the effects of a whole grain rye and rye bran diet on the metabolic profile of plasma in prostate cancer patients. The Journal of Nutrition, 141(12), 2126–2132. Moco, S., Bino, R. J., Vorst, O., Verhoeven, H. A., de Groot, J., van Beek, T. A., et al. (2006). A liquid chromatography-mass spectrometry-based metabolome database for tomato. Plant Physiology, 141(4), 1205–1218.

156

Moco, S., Forshed, J., De Vos, R. C. H., Bino, R. J., & Vervoort, J. (2008). Intra- and inter-metabolite correlation spectroscopy of tomato metabolomics data obtained by liquid chromatography-mass spectrometry and nuclear magnetic resonance. Metabolomics, 4(3), 202–215. Moco, S., Vervoort, J., Moco, S., Bino, R. J., De Vos, R. C. H., & Bino, R. (2007). Metabolomics technologies and metabolite identification. Trends in Analytical Chemistry, 26(9), 855–866. Moran, N. E., Cichon, M. J., Riedl, K. M., Grainger, E. M., Schwartz, S. J., Novotny, J. A., et al. (2015). Compartmental and non-compartmental modeling of 13C-lycopene absorption, isomerization, and distribution kinetics in healthy adults. American Journal of Clinical Nutrition, Accepted. Moran, N. E., Clinton, S. K., & Erdman, J. W. (2013). Differential bioavailability, clearance, and tissue distribution of the acyclic tomato carotenoids lycopene and phytoene in mongolian gerbils. The Journal of Nutrition, 143(12), 1920–1926. Moran, N. E., Rogers, R. B., Lu, C.-H., Conlon, L. E., Lila, M. A., Clinton, S. K., & Erdman, J. W. (2013). Biosynthesis of highly enriched 13C-lycopene for human metabolic studies using repeated batch tomato cell culturing with 13C-glucose. Food Chemistry, 139(1-4), 631–639. Müller, L., Goupy, P., Fröhlich, K., Dangles, O., Caris-Veyrat, C., & Böhm, V. (2011). Comparative study on antioxidant activity of lycopene (Z)-isomers in different assays. Journal of Agricultural and Food Chemistry, 59(9), 4504–4511. Nara, E., Hayashi, H., Kotake, M., Miyashita, K., & Nagao, A. (2001). Acyclic carotenoids and their oxidation mixtures inhibit the growth of HL-60 human promyelocytic leukemia cells. Nutrition and Cancer, 39(2), 273–283. Neveu, V., Perez-Jimenez, J., Vos, F., Crespy, V., du Chaffaut, L., Mennen, L., et al. (2010). Phenol-Explorer: an online comprehensive database on polyphenol contents in foods. Database (Oxford), 2010, bap024. Nguyen, M., Francis, D., & Schwartz, S. (2001). Thermal isomerisation susceptibility of carotenoids in different tomato varieties. Journal of the Science of Food and Agriculture, 81(9), 910–917.

157

Niki, E. (1991). Action of ascorbic acid as a scavenger and stable oxygen radicals. The American Journal of Clinical Nutrition, 54(6 Suppl), 1119S–1124S. Node, K., Huo, Y., Ruan, X., Yang, B., Spiecker, M., Ley, K., et al. (1999). Anti- inflammatory Properties of Cytochrome P450 Epoxygenase-Derived Eicosanoids. Science, 285(5431), 1276–1279. Nordström, A., Want, E., Northen, T., Lehtiö, J., & Siuzdak, G. (2008). Multiple ionization mass spectrometry strategy used to reveal the complexity of metabolomics. Analytical Chemistry, 80(2), 421–429. Obermüller-Jevic, U. C., Olano-Martin, E., Corbacho, A. M., Eiserich, J. P., van der Vliet, A., Valacchi, G., et al. (2003). Lycopene inhibits the growth of normal human prostate epithelial cells in vitro. The Journal of Nutrition, 133(11), 3356–3360. Omenn, G. S., Goodman, G. E., Thornquist, M. D., Balmes, J., Cullen, M. R., Glass, A., et al. (1996). Effects of a combination of beta carotene and vitamin A on lung cancer and cardiovascular disease. The New England Journal of Medicine, 334(18), 1150– 1155. Paetau, I., Khachik, F., Brown, E. D., Beecher, G. R., Kramer, T. R., Chittams, J., & Clevidence, B. A. (1998). Chronic ingestion of lycopene-rich tomato juice or lycopene supplements significantly increases plasma concentrations of lycopene and related tomato carotenoids in humans. The American Journal of Clinical Nutrition, 68(6), 1187–1195. Paik, J., During, A., Harrison, E. H., Mendelsohn, C. L., Lai, K., & Blaner, W. S. (2001). Expression and characterization of a murine enzyme able to cleave β-carotene. The Journal of Biological Chemistry, 276(34), 32160–32168. Pan, P., Skaer, C. W., Stirdivant, S. M., Young, M. R., Stoner, G. D., Lechner, J. F., et al. (2015). Beneficial Regulation of Metabolic Profiles by Black Raspberries in Human Colorectal Cancer Patients. Cancer Prevention Research, 8(8), 743–750. Pan, Z., & Raftery, D. (2007). Comparing and combining NMR spectroscopy and mass spectrometry in metabolomics. Analytical and Bioanalytical Chemistry, 387(2), 525–527. Patti, G. J., Yanes, O., & Siuzdak, G. (2012). Innovation: Metabolomics: the apogee of the omics trilogy. Nature Reviews Molecular Cell Biology, 13(4), 263–269.

158

Pereira, M. A., Grubbs, C. J., Barnes, L. H., Li, H., Olson, G. R., Eto, I., et al. (1996). Effects of the phytochemicals, curcumin and quercetin, upon azoxymethane-induced colon cancer and 7,12-dimethylbenz[a]anthracene-induced mammary cancer in rats. Carcinogenesis, 17(6), 1305–1311. Philip, T., & Francis, F. J. (1971). Oxidation of capsanthin. Journal of Food Science, 36(1), 96–97. Pietta, P. G. (2000). Flavonoids as antioxidants. Journal of Natural Products, 63(7), 1035–1042. Psychogios, N., Hau, D. D., Peng, J., Guo, A. C., Mandal, R., Bouatra, S., et al. (2011). The human serum metabolome. PloS One, 6(2), e16957. Pumilia, G., Cichon, M. J., Cooperstone, J. L., Giuffrida, D., Dugo, G., & Schwartz, S. J. (2014). Changes in chlorophylls, chlorophyll degradation products and lutein in pistachio kernels (Pistacia vera L.) during roasting. Food Research International, 65(Part B), 193–198. Rajasekaran, L. R., Aspinall, D., Jones, G. P., & Paleg, L. G. (2001). Stress metabolism. IX. Effect of salt stress on trigonelline accumulation in tomato. Canadian Journal of Plant Science, 81(3), 487–498. Re, R., Fraser, P. D., Long, M., Bramley, P. M., & Rice-Evans, C. (2001). Isomerization of lycopene in the gastric milieu. Biochemical and Biophysical Research Communications, 281(2), 576–581. Redmond, T. M., Gentleman, S., Duncan, T., Yu, S., Wiggert, B., Gantt, E., & Cunningham, F. X. (2001). Identification, expression, and substrate specificity of a mammalian β-carotene 15,15’-dioxygenase. The Journal of Biological Chemistry, 276(9), 6560–6565. Rice-Evans, C. A., Miller, N. J., & Paganga, G. (1996). Structure-antioxidant activity relationships of flavonoids and phenolic acids. Free Radical Biology and Medicine, 20(7), 933–956. Richelle, M., Sanchez, B., Tavazzi, I., Lambelet, P., Bortlik, K., & Williamson, G. (2010). Lycopene isomerisation takes place within enterocytes during absorption in human subjects. British Journal of Nutrition, 103(12), 1800–1807.

159

Ripperger, H., & Porzel, A. (1994). Pimpifolidine and 22-isopimpifolidine, 22,26- epimino-16,23-epoxycholestane alkaloids from the wild tomato Lycopersicon pimpinellifolium. Phytochemistry, 35(3), 813–815. Rock, C. L., Jacob, R. A., & Bowen, P. E. (1996). Update on the biological characteristics of the antioxidant micronutrients: vitamin C, vitamin E, and the carotenoids. Journal of the American Dietetic Association. Ross, A. B., Pere-Trepat, E., Montoliu, I., Martin, F.-P. J., Collino, S., Moco, S., et al. (2013). A whole-grain–rich diet reduces urinary excretion of markers of protein catabolism and gut microbiota metabolism in healthy men after one week. The Journal of Nutrition, 143(6), 766–773. Ross, A. B., Vuong, L. T., Ruckle, J., Synal, H. A., Schulze-Konig, T., Wertz, K., et al. (2011). Lycopene bioavailability and metabolism in humans : an accelerator mass spectrometry study. The American Journal of Clinical Nutrition, 93(6), 1263–1273. Rudell, D. R., Mattheis, J. P., & Hertog, M. L. A. T. M. (2009). Metabolomic change precedes apple superficial scald symptoms. Journal of Agricultural and Food Chemistry, 57(18), 8459–8466. Saunders, M. M., Seraj, M. J., Li, Z., Zhou, Z., Winter, C. R., Welch, D. R., & Donahue, H. J. (2001). Breast cancer metastatic potential correlates with a breakdown in homospecific and heterospecific gap junctional intercellular communication. Cancer Research, 61(5), 1765–1767. Sawada, Y., Nakabayashi, R., Yamada, Y., Suzuki, M., Sato, M., Sakata, A., et al. (2012). RIKEN tandem mass spectral database (ReSpect) for phytochemicals: A plant-specific MS/MS-based data resource and database. Phytochemistry, 82, 38–45. Scalbert, A., Brennan, L., Fiehn, O., Hankemeier, T., Kristal, B. S., van Ommen, B., et al. (2009). Mass-spectrometry-based metabolomics: limitations and recommendations for future progress with particular focus on nutrition research. Metabolomics, 5(4), 435–458. Scalbert, A., Brennan, L., Manach, C., Andres-Lacueva, C., Dragsted, L. O., Draper, J., et al. (2014). The food metabolome: a window over dietary exposure. American Journal of Clinical Nutrition, 99(6), 1286–1308. Schauer, N., Semel, Y., Roessner, U., Gur, A., Balbo, I., Carrari, F., et al. (2006). Comprehensive metabolic profiling and phenotyping of interspecific introgression lines for tomato improvement. Nature Biotechnology, 24(4), 447–454. 160

Schauer, N., Zamir, D., & Fernie, A. R. (2005). Metabolic profiling of leaves and fruit of wild species tomato: a survey of the Solanum lycopersicum complex. Journal of Experimental Botany, 56(410), 297–307. Schierle, J., Bretzel, W., Bühler, I., Faccin, N., Hess, D., Steiner, K., & Schüep, W. (1997). Content and isomeric ratio of lycopene in food and human blood plasma. Food Chemistry, 59(3), 459–465. Schwartz, S. J., & von Elbe, J. (1983). Kinetics of chlorophyll degradation to pyropheophytin in vegetables. Journal of Food Science, 48(4), 1303–1306. Schwartz, S. J., Woo, S. L., & Von Elbe, J. H. (1981). High-performance liquid chromatography of chlorophylls and their derivatives in fresh and processed spinach. Journal of Agricultural and Food Chemistry, 29(3), 533–535. Scita, G. (1992). Stability of β-carotene under different laboratory conditions. The Journal of Nutritional Biochemistry, 3(3), 124–128. Sesso, H. D., Liu, S., Gaziano, J. M., & Buring, J. E. (2003). Dietary lycopene, tomato- based food products and cardiovascular disease in women. Journal of Nutrition, 133(7), 2336–2341. Shaish, A., Harari, A., Kamari, Y., Soudant, E., Harats, D., & Ben-Amotz, A. (2008). A carotenoid algal preparation containing phytoene and phytofluene inhibited LDL oxidation in vitro. Plant Foods for Human Nutrition, 63(2), 83–86. Shieh, J.-M., Cheng, T.-H., Shi, M.-D., Wu, P.-F., Chen, Y., Ko, S.-C., & Shih, Y.-W. (2011). α-Tomatine suppresses invasion and migration of human non-small cell lung cancer NCI-H460 cells through inactivating FAK/PI3K/Akt signaling pathway and reducing binding activity of NF-κB. Cell Biochemistry and Biophysics, 60(3), 297– 310. Sicilia, T., Bub, A., Rechkemmer, G., Kraemer, K., Hoppe, P. P., & Kulling, S. E. (2005). Novel lycopene metabolites are detectable in plasma of preruminant calves after lycopene supplementation. The Journal of Nutrition, 135(11), 2616–2621. Simkin, A. J., Schwartz, S. H., Auldridge, M., Taylor, M. G., & Klee, H. J. (2004). The tomato carotenoid cleavage dioxygenase 1 genes contribute to the formation of the flavor volatiles β-ionone, pseudoionone, and geranylacetone. The Plant Journal, 40(6), 882–892.

161

Slimestad, R., Fossen, T., & Verheul, M. J. (2008). The flavonoids of tomatoes. Journal of Agricultural and Food Chemistry, 56(7), 2436–2441. Slimestad, R., & Verheulb, M. (2009). Review of flavonoids and other phenolics from fruits of different tomato (lycopersicon esculentum mill.) cultivars. Journal of the Science of Food and Agriculture, 89(8), 1255–1270. Slupsky, C. M., Rankin, K. N., Wagner, J., Fu, H., Chang, D., Weljie, A. M., et al. (2007). Investigations of the effects of gender, diurnal variation, and age in human urinary metabolomic profiles. Analytical Chemistry, 79(18), 6995–7004. Smith, C. A., O’Maille, G., Want, E. J., Qin, C., Trauger, S. A., Brandon, T. R., et al. (2005). METLIN: a metabolite mass spectral database. Therapeutic Drug Monitoring, 27(6), 747–751. Solanky, K. S., Bailey, N. J. C., Beckwith-Hall, B. M., Davis, A., Bingham, S., Holmes, E., et al. (2003). Application of biofluid 1H nuclear magnetic resonance-based metabonomic techniques for the analysis of the biochemical effects of dietary isoflavones on human plasma profile. Analytical Biochemistry, 323(2), 197–204. Son, H. S., Hwang, G. S., Kim, K. M., Ahn, H. J., Park, W. M., Van Den Berg, F., et al. (2009). Metabolomic studies on geographical grapes and their wines using 1H NMR analysis coupled with multivariate statistics. Journal of Agricultural and Food Chemistry, 57(4), 1481–1490. Spratlin, J. L., Serkova, N. J., & Eckhardt, S. G. (2009). Clinical applications of metabolomics in oncology: a review. Clinical Cancer Research, 15(2), 431–440. Srinivasan, M., Sudheer, A. R., & Menon, V. P. (2007). Ferulic acid: therapeutic potential through its antioxidant property. Journal of Clinical Biochemistry and Nutrition, 40(2), 92–100. Stahl, W., Junghans, A., de Boer, B., Driomina, E. S., Briviba, K., & Sies, H. (1998). Carotenoid mixtures protect multilamellar liposomes against oxidative damage: synergistic effects of lycopene and lutein. FEBS Letters, 427(2), 305–308. Stahl, W., Schwarz, W., Sundquist, A. R., & Sies, H. (1992). cis-trans isomers of lycopene and β-carotene in human serum and tissues. Archives of Biochemistry and Biophysics, 294(1), 173–177.

162

Stahl, W., & Sies, H. (1992). Uptake of lycopene and its geometrical isomers is greater from heat-processed than from unprocessed tomato juice in humans. The Journal of Nutrition, 122(11), 2161–2166. Stahl, W., & Sies, H. (1993). Physical quenching of singlet oxygen and cis-trans isomerization of carotenoids. Annals of the New York Academy of Sciences, 691, 10– 19. Stahl, W., von Laar, J., Martin, H.-D., Emmerich, T., & Sies, H. (2000). Stimulation of gap junctional communication: comparison of acyclo-retinoic acid and lycopene. Archives of Biochemistry and Biophysics, 373(1), 271–274. Stupp, G. S., Clendinen, C. S., Ajredini, R., Szewc, M. A., Garrett, T., Menger, R. F., et al. (2013). Isotopic ratio outlier analysis global metabolomics of Caenorhabditis elegans. Analytical Chemistry, 85(24), 11858–11865. Sumner, L. W., Mendes, P., & Dixon, R. A. (2003). Plant metabolomics: large-scale phytochemistry in the functional genomics era. Phytochemistry, 62(6), 817–836. Sun, Y. L., Li, F., Su, N., Sun, X. L., Zhao, S. J., & Meng, Q. W. (2010). The increase in unsaturation of fatty acids of phosphatidylglycerol in thylakoid membrane enhanced salt tolerance in tomato. Photosynthetica, 48(3), 400–408. Takahashi, S., Saito, K., Jia, H., & Kato, H. (2014). An integrated multi-omics study revealed metabolic alterations underlying the effects of coffee consumption. PLoS ONE, 9(3), e91134. Tan, H.-L., Moran, N. E., Cichon, M. J., Riedl, K. M., Schwartz, S. J., Erdman, J. W. J., et al. (2014). β-carotene-9’,10'-oxygenase status modulates the impact of dietary tomato and lycopene on hepatic nuclear receptor- , stress- , and metabolism-related gene expression in mice. The Journal of Nutrition, 144(4), 431–439. Tang, G., Wang, X.-D., Russell, R. M., & Krinsky, N. I. (1991). Characterization of β- apo-13-carotenone and β-apo-14’-carotenal as enzymatic products of the excentric cleavage of β-carotene. Biochemistry, 30(41), 9829–9834. Tang, L., Jin, T., Zeng, X., & Wang, J.-S. (2005). Lycopene inhibits the growth of human androgen-independent prostate cancer cells in vitro and in BALB/c nude mice. The Journal of Nutrition, 135(2), 287–290. Tohda, C., Kuboyama, T., & Komatsu, K. (2005). Search for natural products related to regeneration of the neuronal network. Neurosignals, 14(1-2), 34–45.

163

Tzonou, A., Signorello, L. B., Lagiou, P., Wuu, J., Trichopoulos, D., & Trichopoulou, A. (1999). Diet and cancer of the prostate: a case-control study in Greece. International Journal of Cancer, 80(5), 704–708. Ukai, N., Lu, Y., Etoh, H., Yagi, A., Ina, K., Oshima, S., et al. (1994). Photosensitized oxygenation of lycopene. Bioscience, Biotechnology, and Biochemistry, 58(9), 1718–1719. Unlu, N. Z., Bohn, T., Francis, D. M., Nagaraja, H. N., Clinton, S. K., & Schwartz, S. J. (2007). Lycopene from heat-induced cis-isomer-rich tomato sauce is more bioavailable than from all-trans-rich tomato sauce in human subjects. The British Journal of Nutrition, 98(1), 140–146. Vaclavik, L., Schreiber, A., Lacina, O., Cajka, T., & Hajslova, J. (2012). Liquid chromatography-mass spectrometry-based metabolomics for authenticity assessment of fruit juices. Metabolomics, 8(5), 793–803. Vallverdú-Queralt, A., Jáuregui, O., Medina-Remón, A., Andrés-Lacueva, C., & Lamuela-Raventós, R. M. (2010). Improved characterization of tomato polyphenols using liquid chromatography/electrospray ionization linear ion trap quadrupole Orbitrap mass spectrometry and liquid chromatography/electrospray ionization tandem mass spectrometry. Rapid Communications in Mass Spectrometry, 24(20), 2986–2992. Vogel, J. T., Tan, B.-C., McCarty, D. R., & Klee, H. J. (2008). The carotenoid cleavage dioxygenase 1 enzyme has broad substrate specificity, cleaving multiple carotenoids at two different bond positions. The Journal of Biological Chemistry, 283(17), 11364–11373. von Lintig, J., Hessel, S., Isken, A., Kiefer, C., Lampert, J. M., Voolstra, O., & Vogt, K. (2005). Towards a better understanding of carotenoid metabolism in animals. Biochimica et Biophysica Acta, 1740(2), 122–131. Wagner, K., Vito, S., Inceoglu, B., & Hammock, B. D. (2014). The role of long chain fatty acids and their epoxide metabolites in nociceptive signaling. Prostaglandins & Other Lipid Mediators, 113-115, 2–12. Wang, S., DeGroff, V. L., & Clinton, S. K. (2003). Tomato and soy polyphenols reduce insulin-like growth factor-I-stimulated rat prostate cancer cell proliferation and apoptotic resistance in vitro via inhibition of intracellular signaling pathways involving tyrosine kinase. The Journal of Nutrition, 133(7), 2367–2376.

164

Want, E. J., Wilson, I. D., Gika, H., Theodoridis, G., Plumb, R. S., Shockcor, J., et al. (2010). Global metabolic profiling procedures for urine using UPLC-MS. Nature Protocols, 5(6), 1005–1018. Wargovich, M. J., Jimenez, A., McKee, K., Steele, V. E., Velasco, M., Woods, J., et al. (2000). Efficacy of potential chemopreventive agents on rat colon aberrant crypt formation and progression. Carcinogenesis, 21(6), 1149–1155. Weemaes, C. A., Ooms, V., van Loey, A. M., & Hendrickx, M. E. (1999). Kinetics of chlorophyll degradation and color loss in heated broccoli juice. Journal of Agricultural and Food Chemistry, 47(6), 2404–2409. Welsch, C. A., Lachance, P. A., & Wasserman, B. P. (1989). Dietary phenolic compounds: inhibition of Na+-dependent D-glucose uptake in rat intestinal brush border membrane vesicles. The Journal of Nutrition, 119(11), 1698–1704. Weng, C. J., & Yen, G. C. (2012). Chemopreventive effects of dietary phytochemicals against cancer invasion and metastasis: Phenolic acids, monophenol, polyphenol, and their derivatives. Cancer Treatment Reviews, 38(1), 76–87. Wild, C. P. (2012). The exposome: from concept to utility. International Journal of Epidemiology, 41(1), 24–32. Willcox, J. K., Catignani, G. L., & Lazarus, S. (2003). Tomatoes and cardiovascular health. Critical Reviews in Food Science and Nutrition, 43(1), 1–18. Winnike, J. H., Busby, M. G., Watkins, P. B., & O’Connell, T. M. (2009). Effects of a prolonged standardized diet on normalizing the human metabolome. American Journal of Clinical Nutrition, 90(6), 1496–1501. Wishart, D. S. (2008). Metabolomics: applications to food science and nutrition research. Trends in Food Science & Technology, 19(9), 482–493. Wishart, D. S., Jewison, T., Guo, A. C., Wilson, M., Knox, C., Liu, Y., et al. (2013). HMDB 3.0-The Human Metabolome Database in 2013. Nucleic Acids Research, 41(Database issue), D801–D807. Wu, H., Southam, A. D., Hines, A., & Viant, M. R. (2008). High-throughput tissue extraction protocol for NMR- and MS-based metabolomics. Analytical Biochemistry, 372(2), 204–212. Yahara, S., Uda, N., Yoshio, E., & Yae, E. (2004). Steroidal alkaloid glycosides from tomato (Lycopersicon esculentum). Journal of Natural Products, 67(3), 500–502. 165

Yan, K.-H., Lee, L.-M., Yan, S.-H., Huang, H.-C., Li, C.-C., Lin, H.-T., & Chen, P.-S. (2013). Tomatidine inhibits invasion of human lung adenocarcinoma cell A549 by reducing matrix metalloproteinases expression. Chemico-Biological Interactions, 203(3), 580–587. Yang, C. S., Landau, J. M., Huang, M. T., & Newmark, H. L. (2001). Inhibition of carcinogenesis by dietary polyphenolic compounds. Annual Review of Nutrition, 21, 381–406. Yang, C.-M., Hu, T.-Y., & Hu, M.-L. (2012). Antimetastatic effects and mechanisms of apo-8’-lycopenal, an enzymatic metabolite of lycopene, against human hepatocarcinoma SK-Hep-1 cells. Nutrition and Cancer, 64(2), 274–285. Yang, Y. C., Lii, C. K., Lin, A. H., Yeh, Y. W., Yao, H. T., Li, C. C., et al. (2011). Induction of glutathione synthesis and heme oxygenase 1 by the flavonoids butein and phloretin is mediated through the ERK/Nrf2 pathway and protects against oxidative stress. Free Radical Biology and Medicine, 51(11), 2073–2081. Yoshinari, O., Sato, H., & Igarashi, K. (2009). Anti-diabetic effects of pumpkin and its components, trigonelline and nicotinic acid, on Goto-Kakizaki rats. Bioscience, Biotechnology, and Biochemistry, 73(5), 1033–1041. Zaripheh, S., Boileau, T. W.-M., Lila, M. A., & Erdman, J. W. (2003). [14C]-lycopene and [14C]-labeled polar products are differentially distributed in tissues of F344 rats prefed lycopene. The Journal of Nutrition, 133(12), 4189–4195. Zhang, A., Sun, H., & Wang, X. (2012). Serum metabolomics as a novel diagnostic approach for disease: a systematic review. Analytical and Bioanalytical Chemistry, 404(4), 1239–1245. Zhang, G., Panigrahy, D., Mahakian, L. M., Yang, J., Liu, J.-Y., Stephen Lee, K. S., et al. (2013). Epoxy metabolites of docosahexaenoic acid (DHA) inhibit angiogenesis, tumor growth, and metastasis. Proceedings of the National Academy of Sciences of the United States of America, 110(16), 6530–6535. Zhang, H., Kotake-Nara, E., Ono, H., & Nagao, A. (2003). A novel cleavage product formed by autoxidation of lycopene induces apoptosis in HL-60 cells. Free Radical Biology and Medicine, 35(12), 1653–1663. Zhang, S., Myracle, A., Xiao, K., Yan, P., Ye, T., Ferruzzi, M., et al. (2013). Metabolic profiling of green tea treatments in Zucker diabetic rats using 1H NMR. Journal of Nutrition & Food Sciences, 3(6), 239. 166

Zhou, J., Zhou, S., & Zeng, S. (2013). Experimental diabetes treated with trigonelline: effect on β cell and pancreatic oxidative parameters. Fundamental & Clinical Pharmacology, 27(3), 279–287. Zu, K., Mucci, L., Rosner, B. A., Clinton, S. K., Loda, M., Stampfer, M. J., & Giovannucci, E. (2014). Dietary lycopene, angiogenesis, and prostate cancer: a prospective study in the prostate-specific antigen era. Journal of the National Cancer Institute, 106(2), djt430.

167

APPENDIX A: CONSIDERATIONS REGARDING THE STABILITY, FORMATION, AND ANALYSIS OF APOCAROTENOIDS

Additional experiments conducted to investigate the stability, formation, and analysis of

apocarotenoids from foods and biological samples will be summarized here, with

particular focus on apo-13-carotenone.

All analyses of β-apo-13-carotenone were conducted by HPLC-QTRAP-MS.

Identification and quantitation were performed using an authentic standard.

A.1 Evaluation of Extraction Procedure

We compared the three plasma extraction methods for the analysis of β-apo-13-

carotenone detailed in Table 6. β-apo-13-carotenone was not detected by LC-MS in the

small volume solvent extracts (1) and (2), but was detectable in extract (3) (Figure 27).

These results suggest that extraction method (3) is either more efficient or artificially

generating β-apo-13-carotenone.

168

Table 6. Plasma extraction methods for the analysis of β-apo-13-carotenone. Extraction Method (1) Small volume HEAT extraction (2) Small volume hexanes (3) Large volume hexanes extraction extraction  To 1 mL plasma, add 1 mL EtOH Same as (1) with hexanes instead of  1 mL of serum divided between 5 w/ 0.1% BHT and probe sonicate HEAT solvent mixture vials (200 uL) per vial  Add 5 mL HEAT  Add 1 mL EtOH to each vial and (10:6:7:7 Hexane/ Ethanol/ probe sonicate Acetone/ Toluene) and probe  Add 10 mL hexanes to each vial sonicate and probe sonicate  Centrifuge 2 min  Centrifuge 5 min  Remove upper phase  Remove upper phase from each 169  Repeat HEAT extraction vial and pool  Pool upper layers and dry under  Repeat hexanes extraction N2  Dry pooled extracts under N2  Reconstitute with 300 uL 1:1  Reconstitute in 200 uL 1:1 MTBE/MeOH MTBE/ MeOH

169

170

Figure 27. EICs for apo-13-carotenone in plasma extracted using methods (1), (2), and (3). 170

A.2 Effect of Nitrogen Drying

An experiment was conducted to determine whether β-apo-13-carotenone is more efficiently extracted with a larger volume of hexanes or formed during the dry down step in the extraction method. Plasma extracts (2) and (3) were prepared as described up to the dry down step. Prior to drying under nitrogen, additional hexanes was added to extract (2) to equal the solvent volume in extract (3). Both extracts were subsequently dried to completion under nitrogen. By adding additional hexanes following the extraction steps, the effect of prolonged drying under nitrogen could be isolated and evaluated.

As shown in Figure 28, β-apo-13-carotenone was detected in both extract (3) and extract

(2) with the extended dry down step at approximately the same concentrations. This result suggests that the detection of β-apo-13-carotenone in plasma following a larger volume solvent extraction is the result of the artifactual formation of this compound during prolonged nitrogen drying.

171

172

Figure 28. EICs for apo-13-carotenone in plasma showing the effect of an extended dry down step in the extraction method.

172

A.3 Stability in Extract

Given the stability concerns of β-apo-13-carotenone, we evaluated whether this compound would form spontaneously in a plasma extract under ambient conditions. A plasma extract prepared using the small volume solvent extraction (2) was analyzed after initial redissolution in MTBE/MeOH (1:1, v/v), recapped, and analyzed again after 7 hours in the HPLC autosampler.

β-apo-13-carotenone was not detected in plasma after initial redissolution, but was detectable at low levels after 7 hours in the autosampler (Figure 29). This provides further evidence of the instability β-apo-13-carotenone and raises additional concerns regarding the formation of this compound.

173

174

Figure 29. EICs of apo-13-carotenone in a newly redissolved plasma extract and the same extract after sitting in the autosampler for 7 hr.

174

A.4 Analysis of Foods and Dietary Supplements

β-apo-13-carotenone was quantitated in different baby foods (sweet potato, squash, and

carrot) and dietary supplements (β-carotene, vitamin A, multi-vitamin). Baby foods were analyzed in quadruplicate and supplements were analyzed in triplicate.

EICs for β-apo-13-carotenone in baby foods and supplements are shown in Figure 30 and

Figure 31. Quantitative results are reported in Table 7 and Table 8. These data demonstrate that β-apo-13-carotenone is ubiquitous in β-carotene and vitamin A containing foods and supplements. Our results suggest that β-apo-13-carotenone is an oxidation product of both β-carotene and vitamin A. The MS data for the squash and carrot baby foods reveals a second apo-13-carotenone peak with a different ratio of

MS/MS transitions. Both squash and carrots contain α-carotene in addition to β-carotene.

We hypothesize that this second peak is ε-apo-13-carotenone resulting from the oxidation

of α-carotene. This compound appears to exist in α-carotene containing foods, such as

squash and carrots, and can be distinguished from β-apo-13-carotenone based on

chromatographic separation and the unique ratio of MS/MS transitions.

175

176

Figure 30. EICs for apo-13-carotenone in sweet potato, squash, and carrot baby foods. 176

177

Figure 31. EICs for apo-13-carotenone in a β-carotene supplement, vitamin A supplement, and multi-vitamin.

177

Table 7. β-apo-13-carotenone concentrations in baby foods (± SD). Concentrations in Baby Foods

ug/100g pmol/g

Sweet potato 2.87 ± 0.15 111.30 ± 5.68

Squash 3.21 ± 0.39 124.36 ± 15.05

Carrot 2.77 ± 0.32 94.93 ± 11.15

Table 8. β-apo-13-carotenone concentrations in dietary supplements (± SD). Concentrations in Supplements

ug/ pill

β-carotene 1.66 ± 0.13

Vitamin A 1.63 ± 0.04

Multi-Vitamin 1.84 ± 0.02

178

A.5 Analysis of Mouse Liver

Based on the presence of β-apo-13-carotenone in vitamin A containing foods and

supplements, we investigated the presence of this compound in control mice (fed no β-

carotene) from a previously conducted study (Tan et al. 2014). Livers from two wild-type

(WT) mice and one β-carotene 9ʹ,10ʹ-oxygenase knock-out (KO) mouse fed a purified

AIN-93G diet were analyzed. This diet contains vitamin A, but does not contain β-

carotene.

Our results show that β-apo-13-carotenone is present in the livers of mice on a purified control diet containing vitamin A, but no β-carotene (Figure 32). We also detected β-apo-

13-carotenone, in addition to retinol, in the purified diets administered to these mice

(Table 9). From these results it is unclear whether β-apo-13-carotenone is absorbed from the diet, formed in vivo, or a combination of the two.

179

180

Figure 32. TIC showing presence of retinol in a representative mouse liver sample (top) and EIC of apo-13-carotenone in liver (bottom).

180

Table 9. β-apo-13-carotenone concentrations in the livers of mice and their diets. Concentrations in Liver

ng/g pmol/g

Control WT Liver (1) 3.41 13.23

Control WT Liver (2) 5.23 20.26

Control KO Liver 2.39 9.27

Concentrations in Diet

Control Diet 0.44 1.71

181

A.6 Conclusions

Results from experiments detailed in Appendix A demonstrate the spontaneous chemical

formation of β-apo-13-carotenone during extraction and analysis of human plasma.

Future studies on β-apo-13-carotenone should strive to reduce the artifactual formation of

this compound during sample handling and preparation.

Additionally, β-apo-13-carotenone is found in both β-carotene and vitamin A containing products. Therefore, it appears to form from the chemical degradation of both β-carotene and retinol. While β-apo-13-carotenone is low, if present at all, in plasma, it appears to accumulate in quantifiable levels in the liver of mice fed vitamin A but no β-carotene.

These results should be confirmed in humans and the mechanism of its appearance in biological tissues should be further evaluated. As β-apo-13-carotenone has reported biological activity, it may be possible to modulate tissue concentrations of this apocarotenoid through β-carotene or vitamin A containing diets.

182