PROFILING THE METABOLOME CHANGES CAUSED BY...
Transcript of PROFILING THE METABOLOME CHANGES CAUSED BY...
PROFILING THE METABOLOME CHANGES CAUSED BY CRANBERRY JUICES OR CRANBERRY PROCYANIDINS USING 1H NMR AND UHPLC-Q-ORBITRAP-HRMS
GLOBAL METABOLOMICS APPROACHES
By
HAIYAN LIU
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
2015
© 2015 Haiyan Liu
To my parents for their unconditional love; my partner for encouraging me at every step
4
ACKNOWLEDGMENTS
I would like to express my gratitude to my major advisor, Dr. Liwei Gu, for his
patience, continuous encouragement and mentorship. Without his guidance and
support, this research could not be accomplished. I am grateful for my committee
members, Dr. Zhihua Su, Dr. Peggy R. Borum and Dr. Maurice R. Marshall for their
valuable time and suggestions.
I acknowledge all the assistance provided by Dr. Timothy J. Garrett, Dr. Arthur S.
Edison, Dr. Fariba Tayyari, Ramadan Ajredini and Sandi Batson Sternberg in the
Southeast Center for Integrated Metabolomics (SECIM). They’ve provided tremendous
help for this research project.
I cherished the friendship with my lab group members, Dr. Keqin Ou, Wei Wang,
Dr. Hanwei Liu, Dr. Amandeep K. Sandhu, Bo Zhao, Kaijie Song, Sara Marshall, Dr.
Zheng Li, Weixin Wang and Yajing Qi. They were always willing to offer helping hands.
The laughter we shared brought abundant joy and made our lives memorable.
Most of all, I would like to express my deepest gratitude to my parents for their
patience, constant love and unconditional support. I also would like give my heartful
thanks to my partner Qiuzhong Wu, who provided me the strength to succeed,
encouraged and guided me. Without their love and support, I would not be able to
successfully accomplish my graduate studies.
5
TABLE OF CONTENTS page
ACKNOWLEDGMENTS .................................................................................................. 4
LIST OF TABLES ............................................................................................................ 8
LIST OF FIGURES ........................................................................................................ 10
LIST OF ABBREVIATIONS ........................................................................................... 14
ABSTRACT ................................................................................................................... 16
CHAPTER
1 A REVIEW: BIOACTIVITY AND BIOAVAILABILITY OF PROCYANIDINS IN CRANBERRIES ...................................................................................................... 18
Procyanidins in Cranberries .................................................................................... 18 Cranberries & Urinary Tract Infections .................................................................... 18
Intervention Studies and Clinical trials .............................................................. 18
Mechanisms ..................................................................................................... 22 Bioavailability of Procyanidins ................................................................................. 26
Absorption and Metabolism in Stomach and Small Intestine ............................ 26 Microbial Catabolism of Procyanidins in Colon ................................................. 30
Metabolomics Approach to Assess Food Specific Molecular Profiles and Biomarkers after Intake........................................................................................ 31
Assessment of Food Intake .............................................................................. 31
Metabolomics ................................................................................................... 33 Applications of Metabolomics for Discovery of Biomarkers of Dietary Intake ... 35
Research Objectives ............................................................................................... 37
2 PROFILING THE METABOLOME CHANGES CAUSED BY CRANBERRY PROCYANIDINS IN PLASMA OF FEMALE RATS USING 1H NMR AND UHPLC-Q-ORBITRAP-HRMS GLOBAL METABOLOMICS APPROACHES.......... 39
Background ............................................................................................................. 39
Materials and Methods............................................................................................ 41 Chemicals and Materials .................................................................................. 41
Extraction, Purification and Characterization of Partially Purified Cranberry Procyanidins and Partially Purified Cranberry Procyanidins ......................... 42
Animals and Experiment Design ...................................................................... 44 1H NMR Analyses ............................................................................................. 45 UHPLC-Q-Orbitrap-HRMS Analyses ................................................................ 45
Multivariate Data Processing and Statistical Analyses ..................................... 47 Results and Discussion........................................................................................... 49
Procyanidin Composition and Content in PPCP and PPAP ............................. 49
6
Quality Control of Multivariate Analyses ........................................................... 49
NMR Metabolomics Analysis of Rat Plasma .................................................... 50 LC-HRMS Metabolomics Analysis of Rat Plasma ............................................ 51
Discriminant Metabolites Identification ............................................................. 53 Summary ................................................................................................................ 59
3 1H NMR-BASED METABOLOMICS REVEALS URINARY METABOLOME MODIFICATIONS IN FEMALE RATS BY CRANBERRY PROCYANIDINS ............ 74
Background ............................................................................................................. 74
Materials and Methods............................................................................................ 75 Chemicals and Materials .................................................................................. 75 Animal Experiment ........................................................................................... 76
1D 1H and 2D 1H-13C NMR analyses ................................................................ 76 Multivariate Statistical Analyses ....................................................................... 77
Results and Discussion........................................................................................... 78
Urinary Metabolome Modification after PPCP or PPAP ................................... 78 Discriminant Metabolites Identification ............................................................. 81
Summary ................................................................................................................ 84
4 A 1H NMR BASED APPROACH TO INVESTIGATE METABOLOMIC DIFFERENCES IN THE PLASMA AND URINE OF YOUNG WOMEN AFTER CRANBERRY JUICE OR APPLE JUICE CONSUMPTION .................................... 96
Background ............................................................................................................. 96
Materials and Methods............................................................................................ 97 Chemicals and Materials .................................................................................. 97
Total Phenolics, Total Anthocyanins, Procyanidin Composition and Content... 97 Sugar Analyses in Cranberry Juice and Apple Juice ........................................ 99 Subjects and Study Design .............................................................................. 99 1H NMR Metabolomics Analyses .................................................................... 100 Multivariate Data Processing .......................................................................... 101
Results and Discussion......................................................................................... 102 Juice Analyses ............................................................................................... 102 Quality Control Data ....................................................................................... 102
Multivariate Analyses of Plasma after Drinking Cranberry Juice vs. Drinking Apple Juice .................................................................................................. 103
Multivariate Analyses of Urine after Drinking Cranberry Juice vs. Drinking Apple Juice .................................................................................................. 104
Discriminant Metabolite Identification ............................................................. 105 Summary .............................................................................................................. 108
5 UHPLC-Q-ORBITRAP-HRMS-BASED GLOBAL METABOLOMICS REVEAL METABOLOME MODIFICATIONS IN PLASMA OF YOUNG WOMEN AFTER CRANBERRY JUICE OR APPLE JUICE CONSUMPTION .................................. 127
Background ........................................................................................................... 127
7
Materials and Methods.......................................................................................... 127
Chemicals and Materials ................................................................................ 127 Subjects and Study Design ............................................................................ 127
UHPLC-Q-Orbitrap-HRMS Analyses .............................................................. 127 Multivariate Data Processing and Statistical Analyses ................................... 129
Results and Discussion......................................................................................... 130 Quality Control of Multivariate Analyses ......................................................... 130 Baseline Plasma vs. Plasma after Drinking Cranberry Juice .......................... 131
Plasma after Drinking Apple Juice vs. Plasma after Drinking Cranberry Juice ............................................................................................................ 134
Discriminant Metabolites Identification ........................................................... 135 Summary .............................................................................................................. 141
6 MODIFICATION OF URINARY METABOLOME IN YOUNG WOMEN AFTER CRANBERRY JUICE CONSUMPTION WERE REVEALED USING UHPLC-Q-ORBITRAP-HRMS-BASED GLOBAL METABOLOMICS APPROACH ................ 165
Background ........................................................................................................... 165
Materials and Methods.......................................................................................... 165 Chemicals and Materials ................................................................................ 165 Subjects and Study Design ............................................................................ 165
UHPLC-Q-Orbitrap-HRMS Analyses .............................................................. 165 Multivariate Data Processing and Statistical Analyses ................................... 167
Results and Discussion......................................................................................... 168 Quality Control of Multivariate Analyses ......................................................... 168
Baseline Urine vs. Urine after Drinking Cranberry Juice................................. 168 Urine after Drinking Apple Juice vs. Urine after Drinking Cranberry Juice ..... 170 Discriminant Metabolites Identification ........................................................... 171
Summary .............................................................................................................. 175
7 CONCLUSIONS ................................................................................................... 198
LIST OF REFERENCES ............................................................................................. 200
BIOGRAPHICAL SKETCH .......................................................................................... 213
8
LIST OF TABLES
Table page 2-1 Content of procyanidins in PPCP and PPAP. ..................................................... 60
2-2 Summary of parameters for PCA, PLS-DA, and OPLS-DA models for rat plasma after administering PPCP or PPAP by oral gavage. ............................... 61
2-3 Identification of discriminant metabolites in rat plasma after administering PPCP or PPAP by oral gavage........................................................................... 62
2-4 Unidentified discriminant metabolic features for rat plasma after administering PPCP or PPAP by oral gavage. ................................................... 63
3-1 Summary of parameters for PLS-DA and OPLS-DA models for rat baseline urine and urine after administering PPCP or PPCP by oral gavage. .................. 86
3-2 Summary of the metabolite profile changes in rat baseline urine and urine after administering PPCP or PPCP by oral gavage. ........................................... 87
4-1 Timeline of intervention study on women. ........................................................ 109
4-2 Total phenolics, total anthocyanins, procyanidin composition and content of cranberry juice and apple juice. ........................................................................ 110
4-3 Summary of parameters for PCA, PLS-DA, and OPLS-DA models for human baseline plasma and plasma after drinking cranberry juice or apple juice. ....... 111
4-4 Summary of parameters for PCA, PLS-DA, and OPLS-DA models for human plasma after drinking cranberry juice or apple juice. ......................................... 112
4-5 Summary of parameters for PCA, PLS-DA, and OPLS-DA models for human urine after drinking cranberry juice or apple juice. ............................................ 113
4-6 Summary of metabolite profile changes in plasma and urine of young women after drinking cranberry juice and apple juice. .................................................. 114
5-1 Summary of parameters for PLS-DA and OPLS-DA models for human baseline plasma and plasma after drinking cranberry juice or apple juice. ....... 142
5-2 Identification of discriminant metabolites in human plasma after drinking cranberry juice or apple juice by negative ionization analysis. ......................... 143
5-3 Identification of discriminant metabolites in human plasma after drinking cranberry juice or apple juice by positive ionization analysis. ........................... 145
5-4 Unidentified discriminant metabolic features in human plasma after cranberry juice or apple juice by negative ionization analysis. .......................................... 147
9
5-5 Unidentified discriminant metabolic features in human plasma after cranberry juice or apple juice by positive ionization analysis. ........................................... 150
6-1 Summary of parameters for PLS-DA or OPLS-DA model for human baseline urine and urine after drinking cranberry juice or apple juice. ............................ 177
6-2 Identification of discriminant metabolites in human urine after drinking cranberry juice or apple juice by negative ionization analysis. ......................... 178
6-3 Identification of discriminant metabolites in human urine after drinking cranberry juice or apple juice by positive ionization analysis. ........................... 179
6-4 Unidentified discriminant metabolic features in human urine after cranberry juice or apple juice by negative ionization analysis. .......................................... 180
6-5 Unidentified discriminant metabolic features in human urine after cranberry juice or apple juice by positive ionization analysis. ........................................... 182
6-6 Summary of identified discriminant metabolites in rats and human. ................. 186
10
LIST OF FIGURES
Figure page 1-1 Structures of epicatechin and procyanidin oligomers isolated from
cranberries.......................................................................................................... 38
2-1 HPLC chromatogram of procyanidins in PPCP and PPAP using fluorescence detection.. ........................................................................................................... 64
2-2 The PCA score plot of rat plasma and quality control samples from 1H NMR metabolomics. .................................................................................................... 65
2-3 The PCA score plot of rat plasma from 1H NMR metabolomics after administering PPCP or PPAP. Red squares: rat plasma after administering PPCP. ................................................................................................................. 66
2-4 The PLS-DA and OPLS-DA score plots and cross-validated score plots of rat plasma derived from 1H NMR metabolomics. ..................................................... 67
2-5 The PCA score plot of rat plasma and quality control samples from LC-HRMS metabolomics.. ........................................................................................ 68
2-6 The PCA score plot of rat plasma from LC-HRMS metabolomics after administering PPCP or PPAP. ............................................................................ 69
2-7 The PLS-DA and OPLS-DA score plots and cross-validated score plots of rat plasma derived from LC-HRMS metabolomics.. ................................................. 70
2-8 Validation plot obtained from 200 permutation tests for the OPLS-DA model of rat plasma after administering PPCP or PPAP from LC-HRMS metabolomics. .................................................................................................... 71
2-9 S-plots associated with the OPLS-DA score plot of data derived from LC-HRMS of rat plasma after administering PPCP or PPAP. .................................. 72
2-10 VIP plot of variables with VIP score higher than 1. ............................................. 73
3-1 The PCA score plot of rat baseline urine and urine after administering PPCP or PPAP.. ............................................................................................................ 88
3-2 The PLS-DA and OPLS-DA score plots and cross-validated score plots of rat baseline urine and urine after administering PPCP. ........................................... 89
3-3 The PLS-DA and OPLS-DA score plots and cross-validated score plots of rat baseline urine and urine after administering PPAP. ........................................... 90
11
3-4 The PLS-DA and OPLS-DA score plots and cross-validated score plots of rat urine after administering PPCP or PPAP. ........................................................... 91
3-5 Validation plot obtained from 200 permutation tests for the OPLS-DA models of rat baseline urine and urine after administering PPCP or PPAP from 1H NMR metabolomics.. .......................................................................................... 92
3-6 S-line associated with the OPLS score plots of data derived from rat baseline urine and urine after PPCP or PPAP.. ................................................................ 93
4-1 Chromatograms of procyanidins extracted from cranberry juice and apple juice using fluorescence detection. ................................................................... 115
4-2 Chromatograms of sugar standards and juices using refractive index detector. ........................................................................................................... 116
4-3 The PCA score plot of human plasma and plasma quality control from 1H NMR metabolomics.. ........................................................................................ 117
4-4 The PCA and OPLS-DA score plots of human plasma after drinking cranberry juice or apple juice from 1H NMR metabolomics. .............................. 118
4-5 Model score plot and cross-validated score plot of OPLS-DA model for human plasma after drinking cranberry juice or apple juice from 1H NMR metabolomics. .................................................................................................. 119
4-6 Validation plot of 200 permutation tests for OPLS-DA model built for human plasma after drinking cranberry juice or apple juice from 1H NMR metabolomics. .................................................................................................. 120
4-7 The PCA and OPLS-DA score plot of human urine after drinking cranberry juice or apple juice from 1H NMR metabolomics. .............................................. 121
4-8 Cross-validated score plot of OPLS-DA model derived from human urine after drinking cranberry juice or apple juice from 1H NMR metabolomics. ........ 122
4-9 Validation plot of 200 permutation tests for OPLS-DA model built for human urine after drinking cranberry juice or apple juice from 1H NMR metabolomics. .................................................................................................. 123
4-10 S-line associated with the OPLS score plots of data derived from human plasma after cranberry juice or apple juice consumption. ................................. 124
4-11 S-line associated with the OPLS score plots of data derived from human urine after cranberry juice or apple juice consumption. .................................... 125
12
4-12 Box-and-whisker plot of the NMR signal intensities of eight significant metabolites detected in human plasma or human urine of young women after drinking cranberry juice and apple juice. .......................................................... 126
5-1 The PCA score plot of human plasma and quality control samples from LC-HRMS metabolomics.. ...................................................................................... 152
5-2 The PCA score plot of human baseline plasma and human plasma after drinking cranberry juice from LC-HRMS metabolomics. ................................... 153
5-3 The PLS-DA and OPLS-DA score plots of human baseline plasma and human plasma after drinking cranberry juice from LC-HRMS metabolomics. .. 154
5-4 The PLS-DA and OPLS-DA cross-validated score plots of human baseline plasma and human plasma after drinking cranberry juice from LC-HRMS metabolomics.. ................................................................................................. 155
5-5 Validation plot obtained from 200 permutation tests for the PLS-DA and OPLS-DA models of human baseline plasma vs. human plasma after cranberry juice by negative ionization analysis. ................................................ 156
5-6 Validation plot obtained from 200 permutation tests for the PLS-DA and OPLS-DA models of human baseline plasma vs. human plasma after cranberry juice by positive ionization analysis. ................................................. 157
5-7 The PCA score plot of human plasma after drinking apple juice or cranberry juice from LC-HRMS metabolomics. ................................................................. 158
5-8 The PLS-DA and OPLS-DA score plots of human plasma after drinking apple juice or cranberry juice from LC-HRMS metabolomics. .................................... 159
5-9 The PLS-DA and OPLS-DA cross validated score plots of human plasma after drinking apple juice or cranberry juice from LC-HRMS metabolomics. ..... 160
5-10 Validation plot obtained from 200 permutation tests for the PLS-DA and OPLS-DA models of human plasma after apple juice vs. plasma after cranberry juice by negative ionization analysis. ................................................ 161
5-11 Validation plot obtained from 200 permutation tests for the PLS-DA and OPLS-DA models of human plasma after apple juice vs. after cranberry juice by positive ionization. ....................................................................................... 162
5-12 S-plots associated with the OPLS-DA score plot of data derived from LC-HRMS of human baseline plasma and plasma after cranberry juice or apple juice by negative ionization. .............................................................................. 163
13
5-13 S-plots associated with the OPLS-DA score plot of data derived from LC-HRMS of human baseline plasma and plasma after cranberry juice or apple juice by positive ionization. ............................................................................... 164
6-1 The PCA score plot of human urine and quality control samples from LC-HRMS metabolomics. ....................................................................................... 189
6-2 The PCA score plot of human baseline urine and human urine after cranberry juice from LC-HRMS metabolomics.. ................................................................ 190
6-3 The PLS-DA, OPLS-DA score plots and cross-validated score plots of human baseline urine and urine after cranberry juice. .................................................. 191
6-4 Validation plot obtained from 200 permutation tests for the PLS-DA and OPLS-DA models of human baseline urine vs. human urine after cranberry juice. ................................................................................................................. 192
6-5 The PCA score plot of human urine after drinking apple juice or cranberry juice from LC-HRMS metabolomics. ................................................................. 193
6-6 The OPLS-DA score plots and cross-validated score plots of human urine after drinking apple juice or cranberry juice from LC-HRMS metabolomics.. .... 194
6-7 Validation plot obtained from 200 permutation tests for the OPLS-DA models of human urine after apple juice vs. human urine after cranberry juice.. .......... 195
6-8 S-plots associated with the OPLS-DA score plot of data derived from LC-HRMS of human baseline urine and urine after cranberry juice or apple juice by negative ionization.. ..................................................................................... 196
6-9 S-plots associated with the OPLS-DA score plot of data derived from LC-HRMS of human baseline urine and urine after cranberry juice or apple juice by positive ionization. ....................................................................................... 197
14
LIST OF ABBREVIATIONS
µg Microgram
µL Microliter
μm Micrometer
µmol Micromole
AJ Apple juice
arb Arbitrary unit
BS Baseline
CJ Cranberry juice
CPMG Carr-Purcell-Meiboom-Gill
DP Degree of polymerization
DSS 2,2-dimethyl-2-silapentane-5-sulfonate
FLD
FT
Fluorescent detector
Fourier transformed
g Gram
g Relative centrifugal force
h Hour(s)
HESI Heated electrospray ionization
HPLC High performance liquid chromatography
HPHPA 3-(3’-hydroxyphenyl)-3-hydroxypropanoic acid
HRMS High resolution mass spectrometer
HSQC Heteronuclear single quantum coherence
L Liter
min Minute (s)
mL Milliliter
15
mM Millimolar
mm Millimeter
MS Mass spectrometer
m/z Mass to charge ratio
nM Nanomolar
nm Nanometer
NMR Nuclear magnetic resonance
NOESY Nuclear overhauser effect spectroscopy
OPLS-DA Orthogonal projection on latent structure-discriminant analysis
OSC Orthogonal signal correction
PAFFT Peak alignment by fast fourier transform
PCA Principal component analysis
PHPAA p-Hydroxyphenylacetic acid
PLS-DA Projection on latent structure-discriminant analysis
PPAP Partially purified apple procyanidins
PPCP Partially purified cranberry procyanidins
PQN Probabilistic quotient normalization
psi Pounds per square inch
QC Quality control
SECIM Southeast Center for Integrated Metabolomics
TOF Time of flight
UHPLC Ultra high performance liquid chromatography
UTI Urinary tract infection
v Volume
VIP Variable Importance Projection
16
Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy
PROFILING THE METABOLOME CHANGES CAUSED BY CRANBERRY JUICES OR CRANBERRY PROCYANIDINS USING 1H NMR AND UHPLC-Q-ORBITRAP-HRMS
GLOBAL METABOLOMICS APPROACHES
By
Haiyan Liu
December 2015
Chair: Liwei Gu Major: Food Science
Cranberries are known to prevent urinary tract infections and other chronic
conditions. Procyanidins are thought to be the bioactive components. The objective of
this study was to identify specific molecular profiles and biomarkers of cranberry
procyanidin or cranberry juice intake in female rats or young women using 1H NMR and
UHPLC-Q-Orbitrap-MS global metabolomics approaches.
Twenty four female Sprague-Dawley rats were administered partially purified
cranberry (PPCP) or apple procyanidins (PPAP) by oral gavage for 3 times at 0, 12 and
24 hours using a 250 mg extracts/kg body weight dose each. A 24-h baseline urine
were collected before the 1st gavage. Second 24-hour urine were collected after the 1st
oral gavage. Six hours after the 3rd gavage, plasma samples of each rat were collected.
Urine and plasma were analyzed using 1H NMR and UHPLC-Q-Orbitrap-HRMS.
Multivariate analyses revealed that plasma and urinary metabolome in rats were
modified after administering PPCP or PPAP. A total of 36 metabolic features in rat
plasma were detected to be discriminant metabolites using UHPLC-Q-Orbitrap-HRMS
metabolomics. Among them, 11 exogenous metabolites originated from procyanidins
17
catabolism by gut microbiota were identified. Furthermore, urinary excretion of six
endogenous organic acids and three exogenous metabolites were changed after PPCP
or PPAP using 1H NMR metabolomics.
Seventeen young women were given either cranberry or apple juice for three
days using a randomized cross-over design. The metabolome in human plasma and
urine were modified following cranberry juice compared to baseline or apple juice. A
total of 26 and 18 metabolites were identified in human plasma and urine, respectively,
to differentiate cranberry juice consumption from baseline or apple juice consumption.
In conclusion, the plasma and urinary metabolome in female rats or young
women were changed after intake of cranberry procyanidins or cranberry juices. Food
specific metabolite profiles and biomarkers were identified in plasma and urine. These
biomarkers may be used to estimate cranberry juice or cranberry procyanidin intake.
The metabolomics differences between cranberry and apple procyanidins as well as the
differences between cranberry juices and apples juices may help to explain the unique
bioactivities of cranberry juice in mitigating urinary tract infections.
18
CHAPTER 1 A REVIEW: BIOACTIVITY AND BIOAVAILABILITY OF PROCYANIDINS IN
CRANBERRIES
Procyanidins in Cranberries
Cranberries (Vaccinium macrocarpon) are a native crop in North America grown
commercially in Wisconsin, Massachusetts, New Jersey, Washington, and part of
Canada. Cranberries are a rich source of flavan-3-ols and procyanidins (Center, 2004).
Procyanidins are oligomeric or polymeric of flavan-3-ols linked through interflavan
bonds. Procyanidins are classified as B-type and A-type based on type of interflavan
bonds (Ou & Gu, 2014). B-type interflavan linkages are C4→ C8 and/or C4→ C6. A-
type procyanidins contain an additional ether bond C2→O→C7(Ou & Gu, 2014). Most
foods including apple juice, grapes, and cocoa contain exclusively B-type procyanidins.
Cranberries are one of a few foods that contain predominantly A-type procyanidins (L.
Gu, Kelm, Hammerstone, Beecher, Holden, Haytowitz, et al., 2003). Procyanidins have
various degree of polymerization (DP). Procyanidins with DP 1, 2-5, and >10 are is
monomer, oligomers, polymers and high polymers, respectively. At about 448 mg/100g
fresh fruits, cranberries contained highest amount of procyanidins compared to all other
fruits (Center, 2004). Over 75% procyanidins in cranberries are polymers and high
polymers (L. Gu, Kelm, Hammerstone, Beecher, Holden, Haytowitz, et al., 2004). The
structures of epicatechin and procyanidin oligomers isolated from cranberries are shown
in Figure 1-1.
Cranberries & Urinary Tract Infections
Intervention Studies and Clinical trials
Urinary tract infection (UTI) is diagnosed by the presence of bacteria in the urine.
It affects over 11 million women in the U.S., and costs over $ 1.6 billion each year (Fihn,
19
2003). Most patients have recurrent infection during their lifetime (Hooton, 2001). The
standard treatment of UTI is to use antibiotics. However, antibiotic-resistant bacteria
cause frequent treatment failure and relapse. Cranberries have been used for UTI
prevention and treatment for over 100 years (Blatherwick, 1914). An epidemiological
study found that drinking cranberry juice on a regular basis decreased the chance of
UTI (Foxman, Geiger, Palin, Gillespie, & Koopman, 1995). The study was retrospective
and examined the relationship between first time UTI and health/sex behavior. The
authors concluded that young and sexually active women may benefit from cranberries
consumption.
The first clinical study to investigate the effects of cranberry on UTI was
conducted in 1966. Sixty patients with bacteriuria were recruited and given 480 mL
cranberry juice daily. After 3 weeks, 53% of the patients had a positive response, but
most patients had a recurrence 6 weeks after stopping drinking juice (Sobota, 1984). In
an open, randomized, controlled trial, three groups of women were given 3 different
treatments: group 1 drank 50 mL of cranberry-lingonberry juice (contained 7.5 g
cranberry concentrate and 1.7 g lingonberry concentrate) daily; group 2 had 100 mL of
a lactobacillus drink; group 3 received no treatment (Kontiokari, Sundqvist, Nuutinen,
Pokka, Koskela, & Uhari, 2001). After 6 months, 16% of the subjects in cranberry group
had ≥ 1 recurrence of UTI, compared to 39% in the lactobacillus group and 36% in the
control group. After stopping drinking, the recurrence of UTI at 12 months in cranberry
group was significantly lower than that in control group and lactobacillus drink group.
Similar results were obtained by Stothers et al. in a placebo-controlled, double-blind
clinical trial, in which a total of 150 women with previous UTI were divided into 3 groups
20
(Stothers, 2002). One group took placebo juice and placebo tablets; the second group
received cranberry juice and placebo tablets; the last group had placebo juice and
cranberry tables. The study lasted for 1 year and the results showed that 18% and 20%
of subjects in cranberry tablets and cranberry juice group experienced ≥ 1 recurrence of
UTI, lower than 32% recurrence rate in the control group. A double-blind crossover
study was done on 15 women with recurrent UTI (Walker, Barney, Mickelsen,
WALYON, & Mickelsen, 1997). All the subjects received either cranberry capsule or a
placebo capsule for 3 months. Then patients were switched to an alternative therapy for
another 3 months. It was found that cranberry treatment led to a lower recurrence rate.
Elderly women are more susceptible to UTI. In a randomized, double-blind study
(Avorn, Monane, Gurwitz, Glynn, Choodnovskiy, & Lipsitz, 1994), 153 asymptomatic
elderly women were provided with 300 mL of either cranberry juice cocktail or placebo
for 6 months. Urine sample were tested at baseline and at a 1-month interval for 6
months. There was no difference in the percentage of urine samples with bacteriuria at
the baseline and after 1 month of treatment (~20% at baseline and ~25% after 1
month). But from the 2-month, the percentage of urine sample with bacteriuria in
cranberry group was lower than that in placebo group. At the end of the study, urine
samples with bacteriuria in cranberry group was 42% less frequent than those in the
placebo group, suggesting that cranberry juice reduced the frequency of bacteriuria
(Avorn, Monane, Gurwitz, Glynn, Choodnovskiy, & Lipsitz, 1994). In another non-
blinded crossover study (Haverkorn & Mandigers, 1994), elderly patients received either
15 mL cranberry juice or the same amount of water twice daily. After 4 weeks, patients
switched to the other treatment for the next 4 weeks. At the end of the study, urine
21
samples with bacteriuria in 3 patients were observed during the entire study, and 7
patients had no bacteriuria in either cranberry or water treatment. In the rest of 7
patients’ urine samples, the chance of bacteriuria in cranberry treatment period was less
than that when receiving water.
However, several clinical trials showed that cranberry juices had no effect in
preventing UTI. In a double-blind, randomized, placebo-controlled trial, 319 college
women with an acute UTI were recruited (Barbosa-Cesnik, Brown, Buxton, Zhang,
DeBusscher, & Foxman, 2011). Participants received 240 mL of 27% low-calorie
cranberry juice cocktail or placebo juice twice a day for 6 months. At baseline, 3- and 6-
month time point, urine samples were collected from participants for uropathogens
assessment. The recurrence rate of UIT was 20% in cranberry juice group and 14% in
placebo group. The presence of urinary symptoms at 3 days, 1-2 weeks, and at>1
month was similar between study groups. In another double-blind, randomized,
placebo-controlled study, a total of 255 children (1-16 years old) treated for UTI were
randomized to receive 300 mL of either cranberry juice or placebo juice twice a day for
6 months (Salo, Uhari, Helminen, Korppi, Nieminen, Pokka, et al., 2012). The primary
end point was the occurrence of the first UTI episode during the 12-months follow-up.
Regular cranberry juice drinking reduced the number of UTI recurrence but did not
decrease the number of children experiencing at least 1 recurrence after initial UTI
episode. In a randomized, placebo-controlled, double-blind trial, 376 elderly patients in
hospital received 300 mL of either cranberry juice or placebo juice once daily (McMurdo,
Bissett, Price, Phillips, & Crombie, 2005). The primary outcome was time to the onset of
first UTI. A total of 5.6% of participants developed a symptomatic UTI, with 14/189 in the
22
placebo group and 7/187 in the cranberry juice group. These between-group differences
were not significant. The authors confirmed the acceptability of cranberry juice to elderly
people but not observed the effectiveness of cranberry juice in reducing UTI in elderly
hospital patients. When using cranberry products in clinical trials, participant adherence
is a challenge because of the bitterness and astringency of cranberry juices. Without
specific biomarkers, there is no effective way to evaluate participant compliance in a
clinical trial. Poor participant adherence, high withdrawal rate, and lack of sufficient
active ingredient have been attributed to the ineffectiveness of cranberry products in
preventing UTIs in some clinical trials (Jepson, Craig, & Williams, 2013).
Mechanisms
Although the exact mechanism remains unknown, several theories have been
proposed to explain the effects of cranberries in preventing UTI. Nearly 95% of UTIs are
caused by uropathogenic strain of Escherichia coli bacteria (Blatherwick & Long, 1923).
An earliest theory suggested that the acidity of cranberries played a key role in inhibiting
the activity of uropathogenic E.coli (Blatherwick & Long, 1923). However, latter research
dispelled this theory because a bacteriostatic pH in urine can hardly be achieved
following normal consumption of cranberry juices (Amy B Howell, 2007). A prevalent
theory nowadays is that the preventive effects of cranberries on UTI are due to ability of
cranberries to inhibit the adhesion of E.coli (Amy B Howell, 2007). Uropathogenic E.coli
adhere to the uroepithelium first, and then multiply and colonize the urinary tract,
resulting in UTI (Beachey, 1981). P-fimbriae and type-1 fimbriae of uropathogenic E.coli
adhere to the carbohydrate receptor on the surface of uroepithelium cells. These E.coli
are the virulence factors in the pathogenesis of UTI (Beachey, 1981). P-fimbriated E.coli
adhere to the oligosaccharide receptor sequences (Bond, Favero, Petersen, Gravelle,
23
Ebert, & Maynard, 1981) and typ-1 fimbriated E.coli adheres to mannose-like receptors
(Beachey, 1981).
Ex vivo and in vitro studies showed that cranberries juices and cranberry
capsules inhibited the adhesion of uropathogenic E.coli on urinary tract epithelium cells.
In a multicenter, randomized, double-blind, crossover study, 32 sexually active females
over 18 years old from 4 different countries were recruited (Amy B Howell, Botto,
Combescure, Blanc-Potard, Gausa, Matsumoto, et al., 2010). Eight subjects from
France and 8 subjects from Spain received treatment of 2 placebo capsules, or 1
placebo and 1 cranberry capsule, or 2 cranberry capsules during 3 treatment periods.
The other 8 volunteers from Hungary and 8 from Japan received the same regimen but
with double dosage of cranberry powder. Urine samples before and after
placebo/cranberry capsules were collected and tested for the anti-adhesion activity.
Anti-adhesion activity was detected in urine samples collected from volunteers who
consumed cranberry powder, but not observed in urines collected from placebo group.
The inhibition of bacteria adhesion was dose-dependent, prolonged and increased with
the amount of procyanidins consumed. The authors concluded that consuming
cranberry powder at dosages of 72 mg of procyanidins offered protection against
bacterial adhesion (Amy B Howell, et al., 2010). In a different randomized, double-blind,
placebo-controlled and crossover study, 20 volunteers including 10 men and 10 women
were recruited (Di Martino, Agniel, David, Templer, Gaillard, Denys, et al., 2006).
Volunteers received 750 mL of drinks composed of (A) 250 mL placebo + 500 mL
mineral water; or (B) 750 mL of the placebo; or (C) 250 mL of the cranberry juice and
500 mL of mineral water; or (D) 750 mL of the cranberry juice. Each volunteer took the
24
four regimens successively in a random order, with a washout period of at least 6 days
between regimens. The first morning urine following cranberry or placebo consumption
was collected and used to test bacterial growth. Cranberry consumption caused a dose-
dependent decrease in bacterial adherence. Cranberry juice consumption provided anti-
adherence activity against different E. coli uropathogenic strains in the urine compared
with placebo (Di Martino, et al., 2006).
Procyanidins in cranberries were suggested to be the active compounds in
preventing the adhesion of E. coli. Cranberry procyanidins inhibited adhesion of only P-
fimbriated E.coli to urinary tract epithelial cells, but not type-1 fimbriae (Gupta, Chou,
Howell, Wobbe, Grady, & Stapleton, 2007). It was found that A-type procyanidins but
not B-type procyanidins in cranberries prevented the adhesion of P-fimbriated E.coli to
urinary tract epithelial cells (Foo, Lu, Howell, & Vorsa, 2000; Gupta, Chou, Howell,
Wobbe, Grady, & Stapleton, 2007). A-type procyanidins extracted from cranberry juice
and B-type procyanidins isolated from grape juice, apple juice, chocolate and green tea
were tested for their anti-adhesion activities towards P-fimbriated uropathogenic E.coli
in vitro. A-type procyanidins from cranberry had anti-adhesion activity in vitro at 60 µg
procyanidins/mL (Amy B. Howell, Reed, Krueger, Winterbottom, Cunningham, & Leahy,
2005). The threshold is 1200 µg/mL for grape juice. B-type procyanidins from other
dietary sources did not show anti-adhesion activity (Amy B. Howell, Reed, Krueger,
Winterbottom, Cunningham, & Leahy, 2005). In the same study, human subjects were
enrolled and provided with a single serving of each food containing same amount of
procyanidins. Urine samples were collected and tested for the anti-adhesion activity.
Bacterial anti-adhesion activity was detectable only in the urine of volunteers who
25
consumed cranberry juices (Amy B. Howell, Reed, Krueger, Winterbottom,
Cunningham, & Leahy, 2005).
The anti-adhesion mechanisms of cranberry procyanidins are not fully
understood. Studies suggested that A-type procyanidins may work as receptor analogs
to competitively inhibit adhesion of E.coli by binding to the fimbrial tips (Amy B Howell,
2007). A-type procyanidins in cranberries may alter cell surface properties of the
bacteria to reduce its adhesion capabilities (Foo, Lu, Howell, & Vorsa, 2000). It was also
observed that cranberry juice changed the conformation of surface macromolecules of
P-fimbriated E.coli and specifically reduced fimbrial length and density (Y. Liu, Black,
Caron, & Camesano, 2006). Other studies suggested that A-type procyanidins in
cranberries reduced fimbrial expression at the genetic level and changed the shape of
bacteria (Ahuja, Kaack, & Roberts, 1998; Y. Liu, Black, Caron, & Camesano, 2006).
A-type oligomers with DP 3-5 were the most effective procyanidins in preventing
adhesion of E. coli in ex vivo assays. A-type dimers were slightly active (Foo, Lu,
Howell, & Vorsa, 2000; Gupta, Chou, Howell, Wobbe, Grady, & Stapleton, 2007).
However, such activity may not explain the anti-adhesion activity of urine after cranberry
intake because the bioavailability of procyanidins was extremely low. Only trace amount
of A-type dimers were detected in human urine after cranberry juices intake (McKay,
Chen, Zampariello, & Blumberg, 2015; Zampariello, McKay, Dolnikowski, Blumberg, &
Chen, 2012). No A-type trimers or tetramers were detected in human urine or blood. It is
likely that anti-adhesion activities in urine is due to unknown metabolites from A-type
procyanidins. One objective of this research is to identify those metabolites.
26
Bioavailability of Procyanidins
Absorption and Metabolism in Stomach and Small Intestine
Flavan-3-ols were stable in simulated stomach juice (pH=1.8) in vitro (Spencer,
Chaudry, Pannala, Srai, Debnam, & Rice-Evans, 2000). Procyanidin oligomers from
chocolate were degraded completely to flavan-3-ol monomers under simulated gastric
juice (37oC, pH 2.0, 1–4 h, no digestive enzymes) in vitro (Spencer, Chaudry, Pannala,
Srai, Debnam, & Rice-Evans, 2000). However, in vivo study showed that procyanidins
were stable in human stomachs by analyzing gastric juice after ingestion of procyanidin-
rich cocoa beverages. No significant depolymerization or degradation of procyanidins
was observed in the gastric juice after cocoa ingestion (Rios, Bennett, Lazarus,
Rémésy, Scalbert, & Williamson, 2002). The authors suggested that the acid in the
stomach was buffered by the foods so that procyanidin exposed to much lower acidic
conditions. Additional studies supported that procyanidins from sorghum or grape seed
extracts were not depolymerized to monomeric flavan-3-ols in the gastrointestinal tract
in rodents (L. Gu, House, Rooney, & Prior, 2007; Tsang, Auger, Mullen, Bornet,
Rouanet, Crozier, et al., 2005).
Procyanidins are absorbed in the small intestine. Because no active transporters
were identified for procyanidins, passive diffusion appears to be the major transportation
route (Ou & Gu, 2013). Paracellular absorption is probably the predominate pathway for
procyanidins since passing the lipid bilayer via the transcellular pathway is not very
likely due to the large number of hydrophilic hydroxyl groups (Ou & Gu, 2013; Ou,
Percival, Zou, Khoo, & Gu, 2012). (-)-Epicatechin is a procyanidin monomer and a
constituent unit of procyanidin oligomers. Epicatechin is absorbed on the epithelium of
the upper portion of small intestine. Absorbed epicatechin undergoes extensive phase II
27
metabolism in the small intestine and liver (Crozier, 2013; Ou & Gu, 2013). The main
metabolites of (–)-epicatechin are (–)-epicatechin-3’-glucuronide, (–)-epicatechin-3’-
sulfate, and 3’-methyl-(–)-epicatechin-5-sulfate. Peak plasma concentration of the
metabolites was achieved 1-3 h after intake (Ottaviani, Momma, Kuhnle, Keen, &
Schroeter, 2012). Urinary excretion of (–)-epicatechin was 21% to 50% of (–)-
epicatechin intake in human (Crozier, 2013).
The absorption of oligomeric procyanidins differ from (-)-epicatechin and is
primarily affected by molecular sizes. A fraction of procyanidin oligomers with DP< 5 are
absorbed in the small intestine. Higher polymers were not absorbed at all. They bypass
small intestine and underwent microbial catabolism in the large intestine. Ou et al. (Ou,
Percival, Zou, Khoo, & Gu, 2012) investigated the transportation of cranberry A-type
procyanidin dimers, trimers, and tetramers on monolayers of Caco-2 cells. They found
that the transportation of A-type dimers, trimers and tetramers was rather low, at 0.6%,
0.4%, and 0.2%, respectively. No conjugated forms of A-type procyanidins were
detected (Ou, Percival, Zou, Khoo, & Gu, 2012). In a different study, A-type procyanidin
dimer A1 and A2 from peanut skin were detected in rat plasma after in situ perfusion of
small intestinal. The plasma concentration of A1 was 0.12 µmol/L after 30 min of
perfusion. A2 was not quantified. The absorption of A-type dimers was only 5-10% of
that of epicatechin (Appeldoorn, Vincken, Gruppen, & Hollman, 2009). A-type dimer A2
and a trimer was detected in rat urine after administering A-type oligomeric procyanidins
which were purified from pericarp of litchi. About 0.85% and 0.21% of ingested A2 and
trimers were excreted into the urine (Li, Sui, Xiao, Wu, Hu, Xie, et al., 2013).
28
More studies have been conducted to investigate the absorption rate of B-type
procyanidins. Deprez et al.(Deprez, Mila, Huneau, Tome, & Scalbert, 2001) found that
(+)-catechin, B-type procyanidin dimers and trimers had similar permeability coefficients
to that of mannitol on the human intestinal epithelia Caco-2 monolayers. In contrast, the
permeability of oligomeric procyanidins was 10-fold lower than dimers. Higher polymers
were not permeable at all (Deprez, Mila, Huneau, Tome, & Scalbert, 2001). Ou et al.
also found that procyanidin B2 was able to transport across the Caco-2 monolayers with
a transport rate of 3.0% (Ou, Percival, Zou, Khoo, & Gu, 2012). B-type procyanidin
dimers in grape seed extracts were absorbed in the small intestine using in site
perfusion, but the absorption rate was only 5-10% of that of epicatechin (Appeldoorn,
Vincken, Gruppen, & Hollman, 2009). Procyanidin dimer B2 and B5 from cocoa
transported from the lumen of isolated rat small intestines to the serosal side of
enterocytes. But the transportation rate was <1% of the total transferred flavonols-like
compounds (Spencer, Schroeter, Shenoy, S Srai, Debnam, & Rice-Evans, 2001).
Epicatechin, catechin, and procyanidin dimer B2 reached the highest plasma
concentration 2 h after cocoa consumption (Holt, Lazarus, Sullards, Zhu, Schramm,
Hammerstone, et al., 2002). Peak plasma concentration of epicatechin after chocolate
intake was observed at Tmax 2-3 h (Richelle, Tavazzi, Enslen, & Offord, 1999).
Two early in vivo studies concluded that procyanidin oligomers were not
bioavailable. In the first study (Donovan, Lee, Manach, Rios, Morand, Scalbert, et al.,
2002), catechin, procyanidin dimer B3 and a grapeseed extracts containing monomers
and a mixture of procyanidins were fed to rats in a single meal. Only catechin and
epicatechin conjugates were found in both plasma and urine after the catechin meal. No
29
procyanidins or conjugates were detected in plasma or urine after the procyanidin B3
meal. Procyanidins in the grapeseed extracts were not cleaved into monomers, and
therefore did not increase the monomers concentration in plasma or urine (Donovan, et
al., 2002). In the second study (Gonthier, Donovan, Texier, Felgines, Remesy, &
Scalbert, 2003), catechin, procyanidin dimer B2, timer C2 and polymers were fed to the
rats in a single meal, respectively. After the meal, methylated catechin and phenolic
acids were detected in both plasma and urine. No procyanidin dimers, trimers or any
conjugates were detected in plasma or urine. Only a very small amount of phenolic
acids (0.5%-0.7%) were found in the plasma and urine (Gonthier, Donovan, Texier,
Felgines, Remesy, & Scalbert, 2003). However, most other in vivo studies did detect the
procyanidin dimers or trimers in plasma or urine after procyanidins consumption in
rodent model. Trace amount of procyanidin dimer B1, B2, B3, B4, C2 and trimers were
detected in the urine of rats after consuming grapeseed extracts (Tsang, et al., 2005).
Dimer B2 and trimers were also found in the plasma of rats and urine of pigs after
feeding grapeseed extracts (Rzeppa, Bittner, Döll, Dänicke, & Humpf, 2012). Dimer B2
was detected in the plasma and urine of rats after intake of pure dimer B2 (Baba,
Osakabe, Natsume, & Terao, 2002). In these animal studies, only intact procyanidin
dimers or trimers were detected at a low concentration. No procyanidin conjugates were
detected at any level. Similar results were also found in human studies. Procyanidin
dimers were detected in plasma after one single cocoa drink which contained only B-
type procyanidins (Holt, et al., 2002). Two grams of grapeseed extracts in capsules
were ingested by 4 healthy volunteers, and procyanidin dimers were found in plasma 2
hours after ingestion (Sano, Yamakoshi, Tokutake, Tobe, Kubota, & Kikuchi, 2003).
30
Although these studies found procyanidin dimers or trimers in plasma or urine after
procyanidin-rich foods ingestion, the amount of excreted procyanidin oligomers in
plasma or urine was very low.
Microbial Catabolism of Procyanidins in Colon
A small portion of flavan-3-ols and procyanidin oligomers are absorbed in the
small intestine. Majority of ingested procyanidins reach the colon intact and are
metabolized by gut microflora. Griffith et al. (Griffiths, 1964) first discovered that
procyanidins were fermented by gut microflora to generate microbial metabolites. They
found 3-(3’-hydroxyphenyl)propionic acid in the urine of rats fed a diet containing (+)-
catechin (Griffiths, 1964). Deprez et al. (Déprez, Brezillon, Rabot, Philippe, Mila,
Lapierre, et al., 2000) found that polymeric procyanidins were completely degraded after
48 h incubation with freshly human fecal bacteria and the major metabolites were 3-(3’-
hydroxyphenyl)propionic acid, 4-hydroxyphenylacetic acid, 3-(4’-
hydryoxphenyl)propionic acid, and 3-phenylpropionic acid (Déprez, et al., 2000).
Procyanidin B2 was degraded by human microflora faster than epicatechin. Degradation
produced metabolites unique to procyanidin B2 including 5-(2’,4’-dihydroxy) phenyl-2-
ene valeric acid and 5-(3’,4’-dihydroxyphenyl) valeric acid (Stoupi, Williamson, Drynan,
Barron, & Clifford, 2010). The total yield of phenolic acids in colon decreases drastically
with the increase of degree of polymerization. About 10% and 7% of microbial
metabolites in rat guts were from monomers and dimers, whereas 0.7% and 0.5% were
generated from trimers and polymers, respectively (Gonthier, Donovan, Texier,
Felgines, Remesy, & Scalbert, 2003). A-type linkage in procyanidins is more rigid and
stable than B-type linkages due to an additional covalent bond (L. Gu, Kelm,
Hammerstone, Beecher, et al., 2003). About 80% of procyanidin A-type dimers and
31
40% of A-type trimers were degraded within 8 h of incubation with pig cecum microflora
(Engemann, Hubner, Rzeppa, & Humpf, 2012). A-type trimers produced hydroxylated
catabolites with more complicated patterns than A-type dimers. The metabolites for both
A-type dimer and trimer procyanidins included hydroxy- or dihydroxy-benzoic acids,
phenylacetic acids, phenylpropionic acids, and phloroglucinol (Engemann, Hubner,
Rzeppa, & Humpf, 2012). Gu et al. found that 50-80% of ingested procyanidins from
sorghum were degraded in the gastrointestinal tract of rats, and 11% of them were
excreted in 24-hour urine as phenolic acids. The major microbial metabolites found in
the serum of rats were 3,4-dihydroxybenzoic acid, vanillic acid, and 4-
hydroxyphenylacetic acid (L. Gu, House, Rooney, & Prior, 2007).
Metabolomics Approach to Assess Food Specific Molecular Profiles and Biomarkers after Intake
Assessment of Food Intake
Accurate assessment of food intake is critical in epidemiological studies to
associate the intake of certain foods with health outcomes. Current measurement of the
dietary intake uses food intake records and food composition database. Food intake
data are commonly obtained from food frequency questionnaires, diet diaries, diet
histories, multiple 24 h recalls, etc. These methods rely on self-reporting by the study
subjects. Consequently, the accuracy of this measurement remains uncertain (Manach,
Hubert, Llorach, & Scalbert, 2009; Spencer, Abd El Mohsen, Minihane, & Mathers,
2008). The limitation of food composition databases is the lack of systematic approach
to comprehensively identify and quantify nutrients in foods. This is particularly true for
phytochemicals, because phytochemical contents in foods are affected by genetics,
environmental conditions, cultivar differences, horticultural practices, and food
32
processing methods (Spencer, Abd El Mohsen, Minihane, & Mathers, 2008). The impact
of these factors on the phytochemical level in foods was not distinguished in the food
composition table, which makes the estimation of dietary intake of phytochemicals
inaccurate. Estimation of food or phytochemical intake using specific biomarkers in
urine or blood is expected to overcome these problem.
The positive correlation between doses of quercetin in fruit juice and urinary
quercetin level suggested that urinary quercetin may serve as a quantitative biomarker
of dietary quercetin intake (Young, Nielsen, Haraldsdóttir, Daneshvar, Lauridsen,
Knuthsen, et al., 1999). Plasma and urinary isoflavone levels are semi-quantitative
indicators of isoflavone intake (Setchell, Brown, Desai, Zimmer-Nechimias, Wolfe,
Jakate, et al., 2003). Flavonoids metabolites in plasma or urine could also be used as
potential biomarkers of fruits and vegetables intake. A positive correlation (r=0.86)
between total polyphenol metabolites in 24-h urine and fruits and vegetables intake was
observed in a dietary intervention study (Krogholm, Haraldsdóttir, Knuthsen, &
Rasmussen, 2004). Urinary quercetin and flavones were found to be higher after high-
vegetable diets. Fruit and vegetable intake positively correlated with changes in urinary
excretion of flavonoids (Krogholm, Haraldsdóttir, Knuthsen, & Rasmussen, 2004).
Positive correlation between the intake of polyphenols-rich foods and urinary excretion
of the corresponding metabolites from spot urine samples was found in several foods
(Mennen, Sapinho, Ito, Bertrais, Galan, Hercberg, et al., 2006). For example, apple
intake positively correlated with phloretin, grapefruits intake with naringenin, orange
consumption with hesperetin, citrus consumption with urinary excretion of hesperetin
and naringenin. Wine consumption positively associated with caffeic acid in human
33
plasma (Simonetti, Gardana, & Pietta, 2001). Black tea and coffee consumption
correlated with 24-h urinary excretion of 4-O-methylgallic and isoferulic acids in human,
respectively (Hodgson, Yee Chan, Puddey, Devine, Wattanapenpaiboon, Wahlqvist, et
al., 2004). These findings supported that polyphenol metabolites can be used as
specific biomarkers for the intake of polyphenol-rich foods (Ito, Gonthier, Manach,
Morand, Mennen, Rémésy, et al., 2005; Spencer, Abd El Mohsen, Minihane, & Mathers,
2008).
These traditional methods to identify and validate biomarkers for food intake was
based on the “one metabolite-one food” approach (Manach, Hubert, Llorach, & Scalbert,
2009). Candidate biomarker metabolites are often the predominant ones for particular
phytochemicals. In the case of procyanidins, no specific or predominant metabolite was
discovered in the urine or blood to serve as possible intake biomarkers. This issue can
be addressed by untargeted metabolomics approach because of its ability to
simultaneously analyze hundreds or thousands phytochemicals and their metabolites in
biofluids. Untargeted Metabolomics approach is useful to identify new and unexpected
biomarkers of phytochemicals intake.
Metabolomics
Metabolomics refers to the comprehensive analysis of low molecular-weight
metabolites in biological samples (Nicholson, Lindon, & Holmes, 1999). It is a system
biology approach to monitor systematic physiological effects following genetic
modification, pathophysiological changes, or exogenous challenges (Griffin, 2006;
Nicholson, Lindon, & Holmes, 1999). Diet plays a pivotal role to shape human
metabolome. Part of the ingested phytochemicals from foods are absorbed through gut
barrier and metabolized. The resultant exogenous phytochemical metabolites are part of
34
food metabolome and may alter the endogenous metabolites (Manach, Hubert, Llorach,
& Scalbert, 2009). Alteration of endogenous metabolites is considered the amplified
‘end-point’ output of changes down the biochemical pathways (S. Lin, Chan, Li, & Cai,
2010). Cranberries are known to affect gene expression, protein activity, and signaling
transduction (Deziel, Patel, Neto, Gottschall‐Pass, & Hurta, 2010; Kresty, Howell, &
Baird, 2011). Cranberry juice or procyanidin consumption may alter the profile of
endogenous metabolites. The microbial catabolites of A-type procyanidins and altered
profile of endogenous metabolites may contribute to the unique bioactivities of cranberry
juices or procyanidins.
Multiple analytical platforms are employed to efficiently generate the metabolic
profiles of biological samples. Nuclear magnetic resonance (NMR) spectroscopy and
mass spectrometry (MS) are the commonly used analytic techniques. NMR
spectroscopy has the advantage of being quantitative, highly reproducible, non-
selective and minimal sample preparation (Dunn, Broadhurst, Atherton, Goodacre, &
Griffin, 2011). MS coupled with chromatographic separation techniques are able to
detect a wider range of metabolites and identify compounds based on their unique
spectrums of mass fragments (H. M. Lin, Helsby, Rowan, & Ferguson, 2011).
Metabolomics strategy produces high-dimensional and complex data set. Multivariate
statistic techniques including principal component analysis (PCA), projection on latent
structure-discriminant analysis (PLS-DA), and orthogonal projection on latent structure-
discriminant analysis (OPLS-DA) are often used to reduce the dimensionality of the data
(Bylesjö, Rantalainen, Cloarec, Nicholson, Holmes, & Trygg, 2006). They are useful to
reveal patterns related to the physiological or pathological perturbation and to aid
35
biological interpretation (Marchesi, Holmes, Khan, Kochhar, Scanlan, Shanahan, et al.,
2007).
Applications of Metabolomics for Discovery of Biomarkers of Dietary Intake
Urinary metabolome modifications in human after cocoa consumption were
explored using a HPLC-Q-TOF-MS-based metabolomics approach (Llorach, Urpi-
Sarda, Jauregui, Monagas, & Andres-Lacueva, 2009). In a randomized, crossover
study, 5 women and 5 men were recruited and consumed either a single dose of cocoa
powder with milk or cocoa powder with water, or milk without cocoa. Urine samples
were collected at baseline, 0-6 h, 6-12h, and 12-24 h after cocoa consumption.
Multivariate statistic models including PCA, PLS-DA, and orthogonal signal correction
(OSC)-PLS-DA were built to reveal differences in urinary metabolome between three
diets. It was found that milk in the cocoa drinks had little influence on the urinary
metabolome. A segregation between cocoa with milk and only milk was observed on the
valid supervised models. A total of 27 compounds including alkaloid derivatives,
metabolites of flavan-3-ols and procyanidins were identified as the main discriminant
biomarkers. In another open, blind, randomized and placebo-controlled trial, 24
volunteers ingested either 10 capsules containing almond skin extract or 10 capsules
containing placebo (Llorach, Garrido, Monagas, Urpi-Sarda, Tulipani, Bartolome, et al.,
2010). Urine samples were collected at baseline, 0-6h, 6-12h, and 12-24 h. Samples
were analyzed using HPLC-Q-TOF-MS followed by multivariate statistics including PCA
and OPLS-DA models. Urinary metabolome of 4 different urine sampling times after
intake of almond skin extract were different from those after intake of placebo. A total of
34 microbial metabolites of procyanidins including flavonoid conjugates,
hydroxylphenylvalerolactone conjugates, 4-hydroxy-5-(phenyl)-valeric acid conjugates,
36
hydroxyphenyl-propionic acid conjugates, hydroxyphenylacetic acid conjugates, and
other phenolic acid conjugates were identified as the potential biomarkers of almond
polyphenol intake. Biomarkers of citrus intake was discovered using three study designs
(Pujos-Guillot, Hubert, Martin, Lyan, Quintana, Claude, et al., 2013). In the first design,
volunteers consumed an acute dose of orange of grapefruit juice. In the second design,
volunteers consumed orange juice regularly for one month. The third design used
volunteers from a large cohort study who reported high or low consumption of citrus
products. PCA and PLS-DA were used to reveal the urinary metabolome modifications
after intake of citrus products. Different discriminant markers were found in these three
studies. Many signals that increased after citrus intake in the acute study were not
found to be the contributing markers in the cohort study. Proline betaine, hydroxyproline
betaine, hesperetin and naringenin glucuronides were identified as sensitive
biomarkers. Additionally, two terpene metabolites were identified as candidate
biomarkers. The authors proposed that data-driven metabolomics profiling of urinary
metabolome in cohort subjects is a powerful approach to discover sensitive biomarkers
for a wide range of foods. The biomarkers of citrus intake were also investigated in
volunteers who consumed a standardized diet supplemented with mix-fruits
(Heinzmann, Brown, Chan, Bictash, Dumas, Kochhar, et al., 2010). Urinary metabolome
of study subjects were profiled using 1H NMR-based metabolomics approach. The
authors identified proline betaine as a putative biomarker of citrus consumption. This
biomarker was validated in an epidemiological study and it showed a sensitivity of
86.3% and a specificity of 90.6% using a receiver operating characteristic curve.
Metabolome modifications in male subjects after green tea or black tea consumption
37
were revealed using a 1H NMR-based metabolomics approach (S. Lin, Chan, Li, & Cai,
2010). Seventeen healthy male volunteers consumed black tea, green tea, or caffeine in
a randomized crossover study. It was found that urinary excretion of hippuric acid, 1, 3-
Dihydroxyphenyl-2-O-sulfate was increased after green tea or black tea consumption
compared to the control of caffeine. The intake of green teal and black tea had different
impact on the endogenous metabolites in urine and plasma. Green tea consumption
caused a greater increase in urinary excretion of several citric acid cycle intermediates.
Research Objectives
Cranberries are known to prevent urinary tract infections and other chronic
conditions. However, there is no effective way to assess cranberry intake in
epidemiological studies or clinical trials. The mechanism by which cranberries mitigate
UTI remains unknown in part because the systematic physiological effects of cranberry
intake in not clear.
The overall goal of this research is to identify specific molecular profiles and
biomarkers of cranberry intake and to help identify the mechanisms of cranberry juices
or procyanidins in mitigating urinary tract infections or other chronic diseases. We
hypothesized that the plasma and urinary metabolome of female rats and young women
are modified after cranberry procyanidins or cranberry juices. These research goals
were reached and hypotheses were tested by pursuing the following two specific aims:
1. To perform metabolomics profiling and fingerprinting (1H NMR & UHPLC-Q-Orbitrap-HRMS) of plasma and urinary metabolome in female Sprague-Dawley rats after administering partially purified procyanidins from cranberry powder or apples.
2. To perform metabolomics profiling and fingerprinting (1H NMR & UHPLC-Q-Orbitrap-HRMS) of plasma and urinary metabolome in young women following cranberry juice and to differentiate metabolites from those formed after apple juice consumption.
38
Figure 1-1. Structures of epicatechin and procyanidin oligomers isolated from
cranberries. 1, 2, 3, 4, 5 and 6 are epicatechin, epicatechin-(4β→8)-epicatechin (procyanidin B2), epicatechin-(4β→8, 2β→O→7)-epicatechin (procyanidin A2), epicatechin-(4β→6)-epicatechin-(4β→8, 2β→O→7)-epicatechin, epicatechin-(4β→8, 2β→O→7)-epicatechin-(4β→8)-epicatechin, and epicatechin-(4β→8)-epicatechin-(4β→8, 2β→O→7)-epicatechin, respectively (Foo, Lu, Howell, & Vorsa, 2000).
39
CHAPTER 2 PROFILING THE METABOLOME CHANGES CAUSED BY CRANBERRY
PROCYANIDINS IN PLASMA OF FEMALE RATS USING 1H NMR AND UHPLC-Q-ORBITRAP-HRMS GLOBAL METABOLOMICS APPROACHES
Background
Procyanidins are oligomers and polymers of (−)-epicatechin or (+)-catechin (L.
Gu, Kelm, Hammerstone, Beecher, et al., 2003). The molecular weight of procyanidins
is described by degree of polymerization (DP). Monomeric procyanidins are (−)-
epicatechin and (+)-catechin. Procyanidins with DP 2, 3, and 4 are dimers, trimers, and
tetramers, respectively. The most widely distributed procyanidins in foods are the B-
type, which are linked through C4→ C8 and/or C4→ C6 interflavan bonds (Ou & Gu,
2014). Examples of foods that contain exclusively B-type procyanidins are apples,
pears, blueberries, and cocoa. A-type procyanidins are rare in foods and they have an
additional ether interflavan bond C2→O→C7. Cranberries are among a few foods that
contain A-type procyanidins. A previous study showed that cranberry press cake with at
least one A-type bond accounted for more than 90% of the oligomers between trimers
and decamers (Feliciano, Krueger, Shanmuganayagam, Vestling, & Reed, 2012).
Studies suggested that A-type procyanidins may have greater or unique bioactivity
compared with B-type (Amy B. Howell, Reed, Krueger, Winterbottom, Cunningham, &
Leahy, 2005). Such activity was attributed to A-type procyanidins but not B-type ones
(Amy B. Howell, Reed, Krueger, Winterbottom, Cunningham, & Leahy, 2005).
Metabolomics have been widely applied in clinical, pharmaceutical and
toxicological studies for identification of biomarkers (Lindon, Holmes, & Nicholson,
2006). It assesses the metabolic changes in a global manner in order to monitor
biological function alteration due to genetic modification, pathophysiological changes, or
40
exogenous challenges (Griffin, 2006; Nicholson, Lindon, & Holmes, 1999).
Phytochemicals originating from foods are ingested, metabolized and absorbed in the
gastrointestinal tract generating a characteristic metabolome profile, which may further
alter endogenous metabolites. Metabolomics is an effective approach to distinguish the
metabolome differences caused by different diets (Llorach, Urpi-Sarda, Jauregui,
Monagas, & Andres-Lacueva, 2009). NMR and UHPLC-HRMS are the two most widely
used metabolomic platforms (Dunn, Broadhurst, Atherton, Goodacre, & Griffin, 2011).
Both techniques are able to detect hundreds of wide ranging metabolites in biological
samples. NMR spectroscopy has the advantage of being quantitative, highly
reproducible, non-selective (Dunn, Broadhurst, Atherton, Goodacre, & Griffin, 2011) and
minimal sample preparation (Beckonert, Keun, Ebbels, Bundy, Holmes, Lindon, et al.,
2007), while UHPLC-HRMS is highly sensitive and able to identify the chemical
structures of metabolites (Dunn, Broadhurst, Atherton, Goodacre, & Griffin, 2011). The
high-dimensional data produced by a metabolomics study is often processed using
multivariate statistical techniques such as PLS-DA and OPLS-DA to reduce the
dimensionality of the data (Bylesjö, Rantalainen, Cloarec, Nicholson, Holmes, & Trygg,
2006).
The mechanism by which cranberry procyanidins mitigate urinary tract infection
remains elusive. A-type procyanidins from cranberry juice inhibited the adhesion of
uropathogenic E. coli, whereas those from apple juice showed no activity. Anti-adhesion
activity in human urine was detected following cranberry juice cocktail consumption, but
not after consumption of apple juice (Amy B. Howell, Reed, Krueger, Winterbottom,
Cunningham, & Leahy, 2005). We hypothesized that the metabolome changes caused
41
by cranberry procyanidins in female rats may be different from those caused by apple
procyanidins. The objective of this study is to identify molecular profile and putative
biomarkers in plasma of female rats after intake of partially purified cranberry
procyanidins (PPCP) using both 1H NMR and UHPLC-Q-Orbitrap-HRMS based global
metabolomics approaches.
Materials and Methods
Chemicals and Materials
Freeze-dried cranberry powder was provided by Ocean Spray Cranberries, Inc.
(Lakeville-Middleboro, MA, USA). Fresh granny smith apples were purchased from a
local grocery store. LC-MS grade acetonitrile, methylene chloride, methanol, acetic acid,
formic acid, and acetone were purchased from Fischer Scientific Co. (Pittsburgh, PA,
USA). (-)-Epicatechin was purchased from Sigma Chemical Co. (St. Louis, MO, USA).
A mixture of partially pure procyanidin oligomers (monomers through nonamers) was
provided by Mars Inc. (McLean, VA, USA). D2O (99.9% D) was provided from
Cambridge Isotope Laboratories, Inc. (Tewksbury, MA, USA). Creatine-D3, L-leucine-
D10, L-tryptophan-2, 3, 3-D3, caffeine-D3 were obtained from CDN Isotopes Inc. (Pointe-
Claire, Quebec, Canada). Sephadex LH-20 resin was purchased from Sigma-Aldrich
(St. Louis, MO, USA). Amberlite FPX 66 resin was a product of Rohm and Haas Co.
(Philadelphia, PA, USA). Pooled quality control plasma samples used in NMR
metabolomics were purchased from the American Red Cross and were collected over a
period of about 2 weeks.
42
Extraction, Purification and Characterization of Partially Purified Cranberry Procyanidins and Partially Purified Cranberry Procyanidins
One hundred and twenty grams of freeze-dried cranberry powder was extracted
with 1 L of methanol. The cranberry-methanol mixture was put into a beaker sealed with
Parafilm M and sonicated for 30 min. After sonication the cranberry-methanol mixture
was placed in darkness at room temperature for 48 h. Extracts obtained after vacuum
filtration were combined and concentrated under a partial vacuum using a rotary
evaporator which was performed at 45 oC. The concentrated extract was re-suspended
in 20 mL of water and loaded onto a column packed with Amberlite FPX 66 resin.
Column was eluted with 3 L of de-ionized water to remove free sugars and organic
acids. Column was then eluted with 500 mL of methanol to recover cranberry
phytochemicals absorbed in the resin. Methanol was then evaporated using a
SpeedVac Concentrator (Thermo scientific ISS110, Waltham, MA) under a reduced
pressure to yield dry cranberry sugar-free extract (5.40 g). The sugar-free powder (5.40
g) was suspended in 100 mL of 30% methanol and loaded onto a column (5.8×28 cm)
packed with Sephadex LH-20, which was soaked in 30% methanol for over 4 hours
before use. The column was eluted with 30% methanol to remove anthocyanins and
phenolic acids, and then eluted with 70% acetone to yield partially purified cranberry
procyanidins (3.95 g). To extract procyanidins from fresh apples, 5000 g fresh granny
smith apples were used. Fresh apples were stored at -20 oC and divided into two
batches before the extraction. Each 2500 g frozen apples were cut into small pieces
and pulverized to apple puree using a blender. Apple puree was mixed with 2 L of
methanol and sonicated for 40 min on an ice bath. Then the apple puree-methanol
suspension was placed in darkness at -10 oC for 48 h. Two batches of extracts obtained
43
after vacuum filtration were combined and concentrated under a partial vacuum using a
rotary evaporator which was performed at 45 oC The concentrated extract was re-
suspended in 50 mL of water and loaded onto a column packed with Amberlite FPX 66
resins. Column was eluted with 3 L of de-ionized water to remove free sugars and
organic acids. Column was then eluted with 500 mL of methanol to recover apple
procyanidins absorbed in the resin. Methanol was then evaporated using a SpeedVac
Concentrator (Thermo scientific ISS110, Waltham, MA) under a reduced pressure to
yield partially purified apple procyanidins (5.30 g).
Quantitative and qualitative analyses of procyanidins followed a previous
publication (Hanwei Liu, Zou, Gao, & Gu, 2013). The HPLC-MSn system had an HCT
ion trap mass spectrometer (Bruker Daltonics, Billerica, MA, USA) coupled with an
Agilent 1200 HPLC (Palo Alto, CA, USA) equipped with a binary pump, an autosampler,
and a fluorescence detector. Separation of procyanidins was carried out on a Luna
Silica (2) column (250 × 4.6 mm, 5 μm particle size, Phenomenex, Torrance, CA, USA)
at a column temperature of 37 oC. The binary mobile phase consisted of (A) methylene
chloride/methanol/acetic acid/water (82:14:2:2, v: v: v: v) and (B) methanol/acetic acid/
water (96:2:2, v: v: v: v). The 70 min gradient was as follows: 0−20 min, 0.0−11.7% B
linear; 20−50 min, 11.7−25.6% B linear; 50−55 min, 25.6−87.8% B linear; 55−65 min,
87.8% B isocratic; 65−70 min, 87.8−0.0% B linear; followed by 5 min of column re-
equilibration before the next injection. Excitation and emission of the fluorescent
detector were set at 231 and 320 nm, respectively. Electrospray ionization at negative
mode was performed using nebulizer 50 psi, drying gas 10 L/min, drying temperature
350 °C, and capillary 4000 V. Mass spectra were recorded from m/z 150 to 2000. The
44
most abundant ion in full scan was isolated, and its product ion spectra were recorded.
Identification of A- and B-type procyanidins using tandem mass spectrometry followed a
previously published method (L. Gu, Kelm, Hammerstone, Zhang, Beecher, Holden, et
al., 2003).
Procyanidins in PPCP and PPAP were quantified based on a method
standardized by Mars Inc. (Robbins, Leonczak, Li, Johnson, Collins, Kwik-Uribe, et al.,
2012). This method uses (−)-epicatechin as a calibrant and relative response factors for
procyanidin dimers through nonamers, because at the same concentration the ratio of
fluorescent responses between an oligomer and (−)-epicatechin stay constant under the
same HPLC condition. The relative response factor of nonamers was used as the
response factor to quantify high polymers.
Animals and Experiment Design
Approval for animal study was sought through the Institutional Animal Care and
Use Committee at the University of Florida (IACUC Study #201307837). Female
Sprague Dawley (n=24, 220-280 g) were housed in the animal facility and acclimated
for 5 days using a purified diet free of flavonoid compounds (D10012G, Research diet
Inc., New Brunswick, NJ, USA). Two female rats were housed in a cage. After the
acclimation period female rats were randomly divided into two groups with 12 female
rats per group, and fasted for six hours before the metabolomics study. PPCP or PPAP
were dispersed in water and administered by oral gavage at 0 and 12 hours using a
dose of 250 mg extracts/kg body weight. Female rats had free access to food and water
after dosing. At 24 hours, female rats were gavaged for a third time. Six hours after the
3rd gavage, female rats were anesthetized and blood samples were collected by cardiac
puncture into vials containing sodium heparin using heparinized syringes. Blood
45
collection time point was selected based on a previous study which showed that [14C]
procyanidin B2 in rats reached a peak plasma concentration at Tmax 5-6 hours after oral
administration (Stoupi, Williamson, Viton, Barron, King, Brown, et al., 2010). Blood
samples were centrifuged at 2,000 g for 10 min at 4 oC to obtain plasma. All plasma
samples were aliquoted and kept in a -80 oC freezer until analyses.
1H NMR Analyses
Plasma samples were thawed at 4oC in a cold room. Four hundred µL of saline
solution (NaCl 0.9% in 10% D2O) was added to 200 µL of plasma. The mixtures were
vortexed for 1 minute and centrifuged at 16, 000 g for 15 min at 4oC and 550 µL of
supernatant was transferred into 5 mm Bruker NMR tubes (Z105684 Bruker 96 well
rack) using Gilson 215 Liquid Handler (Trilution software version 2.0). All 1H-NMR
spectra were collected on a 600 MHz Avance II NMR spectrometer (Bruker Biospin,
Rheinstetten, Germany) equipped with a 5 mm CryoProbe. A Bruker sampleJet
operated by IconNMR in Topspin was used to record spectra automatically. 1D CPMG-
presaturated spectra for plasma were recorded. Optimal probe tuning and matching, 90°
pulse length, water offset, and receiver gain were adjusted on the representative
sample. The probe was automatically locked to H2O+D2O (90%+10%) and shimmed for
each sample. All NMR data were acquired at 300 K.
UHPLC-Q-Orbitrap-HRMS Analyses
Frozen plasma samples (-80 oC) were thawed at room temperature. One plasma
sample (50 µL) was mixed with 400 µL acetonitrile: acetone: methanol (8:1:1, v: v: v) to
precipitate the proteins. Ten µL isotopically-labeled standard solution (40 µg/mL L-
tryptophan-D3, 4 µg/mL L-leucine-D10, 4 µg/mL creatine-D3, and 4 µg/mL caffeine-D3)
was added to the above extraction mixture as internal standards. The sample was then
46
vortexed and placed in a 4 oC refrigerator for 30 min to assist protein precipitation. Then
the sample was centrifuged at 20,000 g for 10 min at <10 oC to pellet the protein. One
hundred and twenty five µl of supernatant was transferred to a new 1 ml Eppendorf tube
and dried under a gentle stream of Nitrogen (Organomation Associates, Inc., Berlin,
MA, USA). Dried sample was reconstituted in 50 µL 0.1% formic acid in water and
vortexted. The sample solution was put on an ice bath for 10-15 min and centrifuged at
20,000 g for 5 min at <10 oC to remove debris. The supernatant was transferred into a
LC glass vial with fused glass insert for analyses. All 24 rat plasma samples were
prepared in the same manner. Three pooled quality control (QC) samples were
prepared by mixing an equal volume of the supernatant from 24 rat plasma extracts. In
addition, three neat QC samples were prepared by adding 20 µL of isotopically-labeled
standard solution directly to three LC glass vials, respectively. To monitor the
performance of data acquisition, run sequence was started with 3 blanks (0.1% formic
acid in water), one neat QC, and one pooled QC followed by every 10 plasma samples
to ensure instrument drift was minimal.
Chromatographic separation was performed on a Thermo Scientific-Dionex
Ultimate 3000 UHPLC using an ACE Excel 2 C18-PFP column, 100 mm x 2.1 mm i.d., 2
µm (Advanced Chromatography Technologies, Aberdeen, UK). The mobile phase
consisted of (A) water with 0.1% formic acid and (B) acetonitrile. The gradient was as
follows: 0−3 min, 100% A isocratic; 3−13 min, 0−80% B linear; 13−16 min, 80% B
isocratic; 16−16.5 min, 80-0% B linear; followed by 3 min of re-equilibration of the
column before the next run. The flow rate was 350 μL/min and the injection volume was
4 μL. Before starting the sequence, UHPLC column was rinsed using 100% acetonitrile
47
and then equilibrated using 100% 0.1% formic acid for 10 min. The UHPLC system was
coupled to a Q Exactive™ Hybrid Quadrupole-Orbitrap High Resolution Mass
Spectrometer (Thermo Fisher Scientific, San Jose, CA, USA). The MS acquisition was
performed in negative ionization with a mass resolution of 70,000 at m/z 200 and
separate injections were performed in a data-dependent (top 5) MS/MS mode with the
full scan mass resolution reduced to 35,000 at m/z 200. The m/z range for all full scan
analyses was 70–1000. Heated electrospray ionization (HESI) parameters were as
follows: sheath gas flow 45 arb, auxiliary gas flow 10 arb, sweep gas flow 1 arb, spray
voltage 3.5 kV, capillary temperature 320 °C, and probe temperature 350°C. In source
CID (Collision-Induced Dissociation) was 2 eV. The mass spectrometer was calibrated
using Pierce™ negative ion calibration solution (Thermo Fisher Scientific, San Jose CA,
USA). To avoid possible bias, the sequence of injections for plasma samples was
randomized.
Multivariate Data Processing and Statistical Analyses
All NMR spectra were phased and baseline corrected using NMRPipe (Delaglio,
Grzesiek, Vuister, Zhu, Pfeifer, & Bax, 1995) and then converted to FT (Fourier
transformed) files. The FT files were imported into MATLAB (R2013B, the Mathworks,
Inc., Natick, MA, USA). Spectra were referenced to the alanine peak at δ1.469 ppm and
water resonance region (4.66-4.95 ppm) was excluded. Then the spectra were aligned
and normalized in MATLAB. The resultant data set was imported into SIMCA (Version
13.0.3, Umetrics, Umea, Sweden) for multivariate statistical analysis. Data were mean-
centered and Pareto scaled before PCA, PLS-DA and OPLS-DA analyses in SIMCA.
LC-HRMS data were converted to mzXML using MSConvert from ProteoWizard
(Chambers, Maclean, Burke, Amodei, Ruderman, Neumann, et al., 2012) and then
48
processed using MZmine 2.12 (Pluskal, Castillo, Villar-Briones, & Orešič, 2010). Peaks
in each sample were extracted, deconvoluted, and deisotoped. Alignment using join
aligner algorithm was conducted with a 10 ppm tolerance for m/z values and 0.2 min
tolerance for retention time. Gap filling using peak finder algorithm was performed to fill
in missing peaks. The resultant data set was imported into SIMCA (Version 13.0.3,
Umetrics, Umea, Sweden) for multivariate statistical analysis. Data were mean-
centered, Pareto scaled and log-transformed before PCA analysis. Data were mean-
centered and log-transformed before PLS-DA and OPLS-DA analyses in SIMCA.
Unsupervised PCA model was performed to initially examine intrinsic variation in the
data set. Then supervised pattern recognition methods include PLS-DA and OPLS-DA
(Bylesjö, Rantalainen, Cloarec, Nicholson, Holmes, & Trygg, 2006) were used to extract
maximum information on discriminant compounds from the data. Validation of the model
was tested using 7-fold internal cross-validation and permutation tests for 200 times. To
further evaluate the predictive ability of the PLS-DA and OPLS-DA models, an external
validation procedure was performed (Brindle, Antti, Holmes, Tranter, Nicholson, Bethell,
et al., 2002; Llorach, et al., 2010). The LC-HRMS metabolomics data set was split into a
training set and a test set. Approximately 80% of the samples were randomly selected
as the training set and the remaining 20% were treated as the test set. PLS-DA and
OPLS-DA models were built based on the training set and obtained models were used
to blindly predict the classification of the samples in the test set. This procedure was
repeated 30 times and correct classification rate was calculated. For univariate
analyses, mass spectral intensity data of selected metabolites which have been mean-
centered and log-transformed were subjected to Welch’s t test. Benjamini–Hochberg
49
procedure at α=0.01 (Benjamini & Hochberg, 1995) was conducted to control false
discoveries. The univariate analyses were done using Microsoft Excel (Version 2010,
Microsoft Corporation, Seattle, WA, USA).
Results and Discussion
Procyanidin Composition and Content in PPCP and PPAP
Procyanidins were extracted and partially purified from cranberry powder or fresh
apples. Figure 2-1 shows the HPLC fluorescence chromatograms of procyanidins in
PPCP and PPAP. The content of total procyanidins in PPCP was 511 mg/g extracts,
lower than that in PPAP (690 mg/g extracts). Over 90% of procyanidin oligomers (dimer
to tetramers) in PPCP were A-type (Table 2-1). Our results were consistent with a
previous study which showed that procyanidins with at least one A-type bond accounted
for more than 90% of trimers through undecamers in cranberry press cake (Feliciano,
Krueger, Shanmuganayagam, Vestling, & Reed, 2012). Monomers through tetramers
accounted for 45% of total procyanidin, with the rest being high polymers. A- and B-type
pentamers and hexamers were identified but not quantified in PPCP due to peak
overlapping (Figure 2-1). PPAP contained exclusively B-type procyanidins. Content of
dimers through tetramers in PPAP were higher than those in PPCP (Table 2-1).
Monomer through tetramers accounted for 65% of total procyanidins in PPAP, with
about 13% being high polymers.
Quality Control of Multivariate Analyses
In this study, the concept of biology QC which uses biological samples including
plasma, urine or tissue as quality controls was adopted (Gika, Theodoridis, Wingate, &
Wilson, 2007). Biological QCs consisting of 4 replicates of pooled Red Cross plasma
were analyzed together with rat plasma to validate NMR acquisition method. The PCA
50
model was built to investigate the metabolome differences between QCs and rat
plasma. The mechanism was based on the ability of the PCA model to cluster samples
in an unsupervised approach. The PCA score plot (Figure 2-2) showed that the 4
replicates were segregated from experimental samples, indicating that the NMR data
acquisition method was valid. Since variations between LC-HRMS injections and
artifacts due to the order of acquisition and carry-over, sensitivity changes or ion
suppression could occur during the experimental period (Burton, Ivosev, Tate, Impey,
Wingate, & Bonner, 2008). Sample acquisition was randomized, and QC samples were
used to monitor the instrument performance. Pooled QCs were further examined using
multivariate statistic techniques. A PCA model was constructed to visualize any
separation between three QCs. PCA score plot (Figure 2-5) demonstrates that the three
QCs across the entire sequence were tightly clustered, suggesting a high quality of data
acquisition.
NMR Metabolomics Analysis of Rat Plasma
PCA model was built on NMR metabolomics data before supervised multivariate
analyses. PCA score plot showed a separation between PPCP and PPAP, with one
sample from PPAP group mixed with the group of PPCP (Figure 2-3). To further confirm
and validate the metabolome differences between PPCP and PPAP, PLS-DA and
OPLS-DA models were constructed. Two principal components were selected to build
PLS-DA model. One principal component and one orthogonal component were used to
construct OPLS-DA model. The R2X and R2Y of both models was 0.433 and 0.676,
respectively (Table 2-2). R2 represents the goodness of fit, and the results indicated that
about 43% of variance in X data matrix and 68% of variance in Y was explained by
PLS-DA and OPLS-DA models. The high R2 values indicated the robustness of the
51
supervised models (Llorach, Urpi-Sarda, Jauregui, Monagas, & Andres-Lacueva, 2009).
Overfitting arises in PLS and OPLS models when the number of variables is much
larger than that of observations. It could be a problem with any high-dimensional data.
Accidental correlation between one or more variables becomes common for
metabolomics data (Kemsley, Le Gall, Dainty, Watson, Harvey, Tapp, et al., 2007).
Internal cross-validation is thus the first step to test the predictability of the supervised
models. If Q2 calculated from the cross-validation has a low value, then conclusion
could be drawn that the supervised model does not have predictability. In the present
study, 7-fold internal cross validation was performed on both PLS-DA and OPLS-DA
models derived from NMR metabolomics data. Q2 obtained from cross-validation for
PLS-DA and OPLS-DA was 0.254 and 0.291, respectively (Table 2-2). They were much
lower than 0.5, a threshold value for a good multivariate model of metabolomics data
(Hawkins, Basak, & Mills, 2003). Although a segregation between PPCP and PPAP was
observed on the PLS-DA and OPLS-DA score plots (Figure 2-4A, 2-4B), the low Q2
values suggested that both models had poor predictability and the segregation was
most likely due to overfitting. Misclassification that occurred during cross-validation
(Figure 2-4C, 2-4D) also confirmed that NMR metabolomics data did not reveal a
metabolome difference in rat plasma between PPCP and PPAP.
LC-HRMS Metabolomics Analysis of Rat Plasma
Similarly, LC-HRMS metabolomics data was analyzed using supervised models
to reveal the metabolic changes of rat plasma after administering PPCP or PPAP.
Figure 2-7A and 2-7B showed a clear segregation between two groups on the score plot
of both PLS-DA and OPLS-DA models. The advantage of OPLS-DA over PLS-DA is
that the “structure noise” of data matrix which is unrelated to the variation of interest is
52
filtered and described only by the orthogonal component. The variation of scientific
interest is described in the predictive component. Therefore the interpretability of the
resulting model is increased (Fonville, Richards, Barton, Boulange, Ebbels, Nicholson,
et al., 2010). In the present study, PLS-DA models derived from LC-HRMS metabolomic
data had high quality parameters which was not improved by OPLS-DA, suggesting
reduced “structure noise” in the data set. PLS-DA had two principal components with an
overall value of R2X and R2Y of 0.428 and 0.995, respectively (Table 2-2). Similarly,
OPLS-DA generated one principal component and one orthogonal component. The R2X
and R2Y of OPLS-DA model was 0.428 and 0.995 (Table 2-2). It showed that about
42% of variance in X data matrix and 99% of variance in Y data matrix was explained by
both supervised models.
To test for overfitting and the validity of PLS and OPLS models derived from LC-
HRMS metabolomic data, three validation methods were used. Seven-fold internal
cross validation was initially performed on both PLS-DA and OPLS-DA models.
Predictability Q2 obtained from cross-validation was 0.982 and 0.974 for PLS-DA and
OPLS-DA model, respectively. The high Q2 indicated both supervised models had
excellent predictability. The cross-validated score plots (Figure 2-7C, 2-7D) showed that
no rat plasma from two groups was misclassified which was consistent with the internal
validation result. In order to further confirm the predictability of PLS-DA and OPLS-DA
models, permutation test was conducted. The class labels of PPCP and PPAP group
were permuted and randomly assigned to different observations. Then a classification
model was calculated with the permutated class labels. The procedure was repeated
200 times. R2 and Q2 within each model were calculated and a regression line was
53
drawn. Ideally, all R2 and Q2 calculated from the permutation data should be lower than
those from the actual data and the Q2-intercept value obtained from the regression line
should be lower than 0.05 (Kang, Choi, Kang, Kwon, Wen, Lee, et al., 2008). The
rationale behind the permutation test is that the newly constructed classification models
should not be able to predict the classes well with a wrong class label (Westerhuis,
Hoefsloot, Smit, Vis, Smilde, van Velzen, et al., 2008). Figure 2-8 showed that the
goodness of fit (R2) and predictive powder (Q2) of newly constructed models with
permuted class labels were decreased compared to the actual model, indicating the
supervised model was statistically valid and the achieved segregation between PPCP
and PPAP was not due to overfitting. Cross-validation and permutation test provide a
reasonable estimate of the predictability of a PLS or OPLS model (Eriksson, 2006).
However, external validation that uses an independent set of test data to evaluate
predictability of a supervised model that is built on the training set is a more scrupulous
and demanding method (Eriksson, 2006). The correct classification rates of 100% for
both PLS-DA and OPLS-DA models (Table 2-2) were obtained, indicating that the
supervised models based on LC-HRMS metabolomics data had excellent predictability
and were able to correctly predict the unknown samples. The validation tests suggested
that the UHPLC-HRMS metabolomics approach was able to reveal the metabolome
changes in female rats after administering PPCP compared with PPAP.
Discriminant Metabolites Identification
No modification of rat plasma metabolome was detected using 1H NMR-based
metabolomics approach although it was proven to be an effective tool for metabolomics
profiling in other studies(Graham, Holscher, & Green, 2014; Kang, et al., 2008). This
was likely due to the inherent low sensitivity of NMR technique that failed to detect
54
procyanidin metabolites in plasma. Untargeted UHPLC-HRMS metabolomics was a
more sensitive method to reveal the metabolome differences and contributing markers.
S-plot (Figure 2-9) is a statistical tool that visualizes the variable influence in a
projection-based model and discover the responsible metabolites. It is a scatter plot that
combines the covariance (magnitude) and correlation loading (reliability) for the model
variables (Wiklund, Johansson, Sjöström, Mellerowicz, Edlund, Shockcor, et al., 2008).
S-plot can be applied to projection-based models including OPLS, PLS or PCA. The x-
axis in the S-plot describes the magnitude of each variable. The y-axis represents the
reliability of each variable. The y-axis has a theoretical minimum of -1 and maximum of
+1. Unless the variable variance is uniform, otherwise the scatter plot will look like an S-
shape. At a significance level p=0.05, a p(corr) of 0.5 was used as an arbitrary cutoff
value to select the potential markers (Llorach, Urpi-Sarda, Jauregui, Monagas, &
Andres-Lacueva, 2009). The markers with higher absolute p[1] and p(corr) values which
are located on the upper right or lower left corner of the S-plot were the statistical
relevant variables for explaining the separation between PPCP and PPAP. The
variables in the middle of the S-plot did not show any relevance in the model. Variable
importance for projection (VIP) is another statistical tool used to summarize the
importance of X-variable both for X- and Y- models (Eriksson, 2006). It is used to
determine the relevance and importance of a variable in a projection-based model. The
influence on the response of each variable is summed over all components and
categorical responses, relative to the total sum of squares of the model. For a given
model, there will be only one VIP-vector summarizing all components and Y-variables.
This makes the VIP an appealing measure of the global effect of diet intervention. A
55
threshold of VIP value ≥ 1 is usually considered appropriate for a metabolomics study
(Eriksson, 2006). In the present study, S-plot is used as the primary statistical tool for
determining the significant metabolites. We compared the markers selected from S-plot
to those selected using a VIP plot. All variables with VIP score >1 were plotted in Figure
2-10. Variables selected as significant ones from S-plot were colored in red. It was
found that variables selected from S-plot had a higher VIP score (VIP>1.7). This result
supported the reliability and effectiveness of S-plot, and indicated that S-plot is a more
scrupulous method. These selected significant metabolites were then subjected to
Welch’s t test, and the p-value obtained for each marker was smaller than 0.01 (Table
2-3). Benjamini–Hochberg procedure (α=0.01) was conducted to control false
discoveries.
A total of 1186 metabolic features were detected in rat plasma, among which 36
features were found to be discriminant metabolites on the basis of multivariate analysis
(Figure 2-9). Eleven metabolites were identified based on their accurate masses and/or
product ion spectra (Table 2-3). The other 25 unidentified metabolites were listed in
Table 2-4. HMDB (Wishart, Tzur, Knox, Eisner, Guo, Young, et al., 2007) and/or
Phenol-Explorer (Neveu, Perez-Jimenez, Vos, Crespy, Du Chaffaut, Mennen, et al.,
2010) were searched to assist metabolite identification. One metabolite that was higher
in rat plasma after PPCP was the ion at m/z 137.0246 [M-H]- producing a product ion at
m/z 93.0339 [M-H-COO]- after MS/MS. It was tentatively identified as p-hydroxybenzoic
acid as it matched the same compound in HMDB (Δ=0.0002 Da) and a previous
publication (Chen, Bozzo, Freixas-Coutin, Marcone, Pauls, Tang, et al., 2014). The
compound producing a [M-H]- ion at m/z 93.0337 [M-H]- was tentatively identified as
56
phenol according to HMDB (Δ=0.0009 Da). The ion at m/z 172.9915 [M-H]- was
tentatively identified as phenol sulfate which agreed with HMDB (Δ=0.0001 Da). The
plasma level of catechol sulfate was elevated in female rats after administering PPCP.
This metabolite was previously identified in human urine after drinking blackcurrant juice
(Törrönen, McDougall, Dobson, Stewart, Hellström, Mattila, et al., 2012). Identification
of catechol sulfate was based on the accurate m/z 188.9863[M-H]-, product ion at m/z
109.0296 [M-H-sulphate]-and HMDB match (Δ=0 Da). 3, 4-Dihydroxyphenylvaleric acid
and 4'-O-methyl-(-)-epicatechin-3'-O-beta-glucuronide were also detected and
tentatively identified in rat plasma after PPCP intake. It should be noted that the
difference between detected mass and theoretical mass of 4'-O-methyl-(-)-epicatechin-
3'-O-beta-glucuronide was 0.0707 Da, which was relatively higher than other mass
error. However, 4'-O-methyl-(-)-epicatechin-3'-O-beta-glucuronide was tentatively
assigned because it was the only database match that was biologically relevant and no
standard was available for further identification. Furthermore, consumption of PPCP
also decreased the plasma level of five metabolites. The metabolite producing a [M-H]-
ion at m/z 479.1190 and a product ion at m/z 303.0885 [M-H-glucuronide]- was
assigned as O-methyl-(-)-epicatechin-O-glucuronide by comparing with HMDB
(Δ=0.0001 Da). A previous publication revealed that methylation of (-)-epicatechin
occurred at 3'- position in rats (Natsume, Osakabe, Oyama, Sasaki, Baba, Nakamura,
et al., 2003). Another study showed that glucuronidation of daidzein occurred at the 7
position after daidzein was incubated with Sprague-Dawley rat liver microsome (Zhang,
Song, Cunnick, Murphy, & Hendrich, 1999). The positions of glucuronidation and
methylation in rats were markedly different from humans, mice, pigs, etc. In the present
57
study, the substitution positions were not able to be further confirmed without NMR
spectra of purified compounds. The metabolite producing a [M-H]- ion at m/z 289.0384
was putatively identified as 4-hydroxy-5-(hydroxyphenyl)-valeric acid-O-sulphate. The
identification agreed with HMDB (Δ=0.0003 Da) and was described in a previous study
(Garcia‐Aloy, Llorach, Urpi‐Sarda, Jáuregui, Corella, Ruiz‐Canela, et al., 2014). 5-
(hydroxyphenyl)-Ƴ-valerolactone-O-sulphate was putatively identified based on its m/z
at 271.0287 [M-H]- and HMDB match (Δ=0.0005 Da). The metabolite having a [M-H]- ion
at m/z 184.0757 was putatively identified as 4-hydroxydiphenylamine according to the
HMDB (Δ=0.0011 Da). 4-hydroxydiphenylamine is a metabolite of diphenylamine and
found in stored apples (Rudell, Mattheis, & Fellman, 2005). The metabolite giving a [M-
H]- ion at m/z 461.9787 and product ion at m/z 264.0330 [M-H-hexose-H2O]- was
tentatively identified as peonidin-3-O-hexose. The exact type of hexose could not be
determined due to lack of standard comparison. Previous studies showed that both
peonidin-3-O-glactoside and peonidin-3-O-glucoside were found in rat plasma after they
were administered with anthocyanin-rich extracts (Ichiyanagi, Shida, Rahman, Hatano,
& Konishi, 2006). In the present study, the detection of peonidin-3-O-hexose in rat
plasma after administering PPAP was likely due to the residual anthocyanins in PPAP.
Procyanidins purified from cranberry powder were predominantly A-type while
exclusively B-type procyanidins were found in PPAP. Procyanidins had a very low
absorption rate in vivo and only a small portion of epicatechin and oligomeric
procyanidins (DP<5) were able to be absorbed in the small intestine (Ou & Gu, 2014).
The majority of A- and B-type procyanidin oligomers and polymers were degraded by
gut microbiota in the colon to produce microbial metabolites. More than half of identified
58
discriminant metabolites in the present study corresponded to the phase II and microbial
metabolites of procyanidins. A previous study demonstrated that B-type procyanidin
dimers were catabolized by microbial cleavage of C-ring and/or oxidation of A-ring, and
further degraded into hydroxyphenyl-Ƴ-valerolactone (Stoupi, Williamson, Drynan,
Barron, & Clifford, 2010). Phenylvalerolactones were then slowly dehydroxylated by
bacteria to form phenylvaleric acids (Sánchez-Patán, Cueva, Monagas, Walton, Gibson,
Martín-Álvarez, et al., 2012). 5-(hydroxyphenyl)-Ƴ-valerolactone-O-sulphate which was
found to be decreased after rat receiving PPCP was formed after phase II metabolism
of 5-(hydroxyphenyl)-Ƴ–valerolactone. 4-hydroxy-5-(hydroxyphenyl)-valeric acid-O-
sulphate, which was also decreased after ingesting PPCP was probably generated by
dehydroxylation of dihydroxyphenyl-Ƴ-valerolactone after further sulfation. 3, 4-
Dihydroxyphenylvaleric acid was probably a dehydroxylation product from
dihydroxyphenyl-Ƴ-valerolactone. p-hydroxybenzoic acid was likely to be formed by
progressive shortening the aliphatic chain by α-and β-oxidation of phenylvaleric acids
(Sánchez-Patán, et al., 2012). Compared to extensive investigation on B-type dimers
catabolism, limited data is available on the microbial catabolism of A-type procyanidins.
A former study employed a pig cecum model and showed that, similar as B-type dimers
catabolism, A-type procyanidins degradation was initiated by cleavage of C-ring
followed by generation of various phenolic acids (Engemann, Hubner, Rzeppa, &
Humpf, 2012). A-type procyanidins oligomers exhibited a more complicated pattern of
hydroxylated catabolites probably due to their more rigid and complex interflavan ether
bonds (Engemann, Hubner, Rzeppa, & Humpf, 2012). However, in the present study we
failed to detect any metabolites that retain this unique ether linkage.
59
Summary
Female Sprague-Dawley rat plasma metabolome differences between PPCP and
PPAP were detected using an untargeted UHPLC-Q-Orbitrap-HRMS metabolomics
approach but not a 1H NMR metabolomics approach. This study is one of few
publications that use two metabolomics tools. Compared to 1H NMR metabolomics,
UHPLC-Q-Orbitrap-HRMS metabolomics is more effective to reveal the overall rat
plasma metabolome modifications caused by PPCP or PPAP and identify the
contributing makers. Discriminant metabolites including p-hydroxybenzoic acid, phenol,
phenol-sulfate, catechol sulphate, 3, 4-dihydroxyphenylvaleric acid, 4'-O-methyl-(-)-
epicatechin-3'-O-beta-glucuronide were significantly higher in rat plasma after PPCP
intake. On the contrary, plasma level of several metabolites including O-methyl-(-)-
epicatechin-O-glucuronide, 4-hydroxy-5-(hydroxyphenyl)-valeric acid-O-sulphate, 5-
(hydroxyphenyl)-Ƴ-valerolactone-O-sulphate, peonidin-3-O-hexose and 4-
hydroxydiphenylamine were increased after rats were gavaged with PPAP.
60
Table 2-1. Content of procyanidins in PPCP and PPAP. Procyanidins Partially Purified Cranberry
procyanidins (mg/g extracts) Partially Purified Apple procyanidins (mg/g extracts)
Monomer 8.64±0.36 61.2±1.59 Dimers 71.7±3.48(60.9±0.98)* 161±5.41
Trimers 71.6±0.40(60.9±0.66)* 101±4.28
Tetramers 75.9±0.22(75.9±0.22)* 125±6.31
Pentamers UQ 95.4±5.37
Hexamers UQ 58.1±3.30
High polymer 283±13.1 88.1±8.34
Total 511±17.6 690±34.6
Data are expressed as mean ± standard deviation. UQ: detected as mixture of A- and B-type oligomers, but not quantified due to peak overlapping. *Numbers in the parentheses represent the content of A-type procyanidins. Numbers out of parenthesis are total procyanidins.
61
Table 2-2.Summary of parameters for PCA, PLS-DA, and OPLS-DA models for rat plasma after administering PPCP or PPAP by oral gavage.
1H NMR metabolomics LC-HRMS metabolomics
PCA PLS-DA OPLS-DA PCA PLS-DA OPLS-DA
Na
5 2 1Pc+1Od 4 2 1Pc+1Od
R2
X(cum)b
0.783 0.433 0.433 0.516 0.428 0.428
R2
Y(cum)b
--- 0.676 0.676 --- 0.995 0.995
Q2
(cum)b
0.513 0.254 0.291 0.521 0.982 0.974
*Correct Classification Rate
--- --- --- --- 100%±0 100%±0
a N: number of components. b R2X (cum)and R2Y (cum) are the cumulative modeled variations in the X and Y matrix, respectively. Q2Y (cum) is the cumulative predicted variation in the Y matrix.
c Predictive component. d Orthogonal component.
62
Table 2-3. Identification of discriminant metabolites in rat plasma after administering PPCP or PPAP by oral gavage. NO. Detected
Mass [M-H]-
(MSMS)
Retention Tim (min)
p[1] (contribution)
p(corr)[1] (confidence)
VIP Welch t test a
Metabolites Putative Identification
Theoretical Mass [M-H]-
Mass Difference (Da)
Database ID
PPCPvs. PPAP b
1 137.0246 (93.0339 [M-H-COO]-)
9.231 0.063 0.931 2.16 <0.01 p-hydroxybenzoic acid c 137.0244 0.0002 HMDB 00500
2 172.9915 6.855 0.051 0.609 1.74 <0.01 phenyl sulfate 172.9914 0.0001 HMDB 60015
3 188.9863 (109.0296) [M-H-sulphate]-)
6.566 0.139 0.933 4.73 <0.01 catechol sulphate c 188.9863 0.0000 HMDB 61713
4 209.0904 9.525 0.057 0.587 1.99 <0.01 3, 4-dihydroxyphenylvaleric acid
209.0819 0.0085 HMDB 29233
5 493.2059 7.572 0.098 0.584 3.22 <0.01 4'-O-methyl-(-)-epicatechin-3'-O-beta-glucuronide
493.1352 0.0707 HMDB 29180
6 93.0337 6.854 0.096 0.653 3.25 <0.01 phenol 93.0346 0.0009 HMDB 00228
7 479.1194(303.0885[M-H-glucuronide]-)
7.463 -0.065 -0.608 2.24 <0.01 3'-O-methyl-(-)-epicatechin-7-O-glucuronide c
479.1195 0.0001 HMDB 41659
8 289.0384
7.095 -0.088 -0.725 3.06 <0.01 4-hydroxy-5-(hydroxyphenyl)-valeric acid-O-sulphate c
289.0387 0.0003 HMDB 59976
9 271.0287 7.793 -0.134 -0.812 4.68 <0.01 5-(hydroxyphenyl)-gamma-valerolactone-O-sulphate
271.0282 0.0005 HMDB 599993
10 184.0757 9.448 -0.236 -0.991 8.16 <0.01 4-hydroxydiphenylamine 184.0768 0.0011 HMDB 32597
11 461.9787(264.0330[M-H-hexose-H2O]-)
9.481 -0.249 -0.994 8.60 <0.01 peonidin-3-O-hexose 462.1168 0.1380 Phenol-Explorer
a Benjamini–Hochberg procedure was conducted to control false discoveries at α=0.01 b Arrows indicated a decrease or increase in metabolite level in rats plasma after administering PPCP compared to PPAP. c Identification agrees with those in Chen et al. (Chen, et al., 2014), Törrönen et al. (Törrönen, et al., 2012), Natsume et al. (Natsume, et al.,
2003), and Garcia-Aloy et al. (Garcia‐Aloy, et al., 2014).
63
Table 2-4. Unidentified discriminant metabolic features for rat plasma after administering PPCP or PPAP by oral gavage.
Detected Mass [M-H]-
Retention Tim (min)
p[1] (contribution)
p(corr)[1] (confidence)
Welch t test a PPCP vs. PPAP b
174.9867 6.854 0.057 0.616 <0.01
190.9822 6.584 0.163 0.920 <0.01
203.0021 7.173 0.173 0.943 <0.01
205.0112 9.232 0.101 0.960 <0.01
240.9789 6.853 0.057 0.590 <0.01
257.9695 6.853 0.069 0.614 <0.01
304.0135 6.606 0.083 0.876 <0.01
308.9664 6.851 0.088 0.600 <0.01
419.1698 7.616 0.097 0.576 <0.01
157.0871 8.324 -0.083 -0.835 <0.01
213.0193 9.293 -0.100 -0.956 <0.01
213.0196 9.194 -0.122 -0.944 <0.01
215.0383 9.457 -0.219 -0.984 <0.01
235.0823 5.094 -0.110 -0.573 <0.01
264.0338 9.451 -0.370 -0.991 <0.01
266.0291 9.547 -0.287 -0.985 <0.01
280.0286 7.824 -0.116 -0.944 <0.01
332.0207 9.463 -0.056 -0.982 <0.01
332.9666 7.560 -0.049 -0.682 <0.01
349.0116 9.448 -0.253 -0.995 <0.01
393.9909 9.464 -0.057 -0.981 <0.01
400.008 9.459 -0.282 -0.995 <0.01
467.9958 9.473 -0.244 -0.991 <0.01
529.9661 9.465 -0.069 -0.968 <0.01
535.9839 9.464 -0.071 -0.985 <0.01
a Benjamini–Hochberg procedure was conducted to control false discoveries at α=0.01. b Arrows indicated a decrease or increase in metabolite level in rats plasma after administering PPCP compared to PPAP.
64
Figure 2-1. HPLC chromatogram of procyanidins in PPCP and PPAP using
fluorescence detection. A) Partially purified cranberry procyanidins and B) partially purified apple procyanidins. Peak identification was performed using MSn. The numbers beside the peaks indicate the degree of polymerization. 2b-6b designate the peaks of B-type procyanidin dimers through hexamers. 2a-6a designate the peaks of A-type procyanidin dimers through hexamers with one A-type linkage.
65
Figure 2-2. The PCA score plot of rat plasma and quality control samples from 1H NMR
metabolomics. Green squares: 4 replicates of Red Cross pooled plasma. Red squares: rat plasma after administering PPCP. Blue squares: rat plasma after administering PPAP. Each square represents an individual rat.
66
Figure 2-3. The PCA score plot of rat plasma from 1H NMR metabolomics after
administering PPCP or PPAP. Red squares: rat plasma after administering PPCP. Blue squares: rat plasma after administering PPAP. Each square represents an individual rat.
-5
-4
-3
-2
-1
0
1
2
3
4
-8 -6 -4 -2 0 2 4 6
t[1]
t[2
]Rat plasma after administering PPAP
Rat plasma after administering PPCP
67
Figure 2-4. The PLS-DA and OPLS-DA score plots and cross-validated score plots of rat plasma derived from 1H NMR metabolomics. A) PLS-DA score plot, B) OPLS-DA score plot, C) PLS-DA cross-validated score plot and D) OPLS-DA cross-validated score plot. Red squares: rat plasma after administering PPCP. Blue squares: rat plasma after administering PPAP. Each square represents an individual rat.
-4
-3
-2
-1
0
1
2
3
-8 -6 -4 -2 0 2 4 6
t[1]
t[2
]
-4
-3
-2
-1
0
1
2
3
-6 -4 -2 0 2 4
t[1]
to [
1]
A B
Rat plasma after administering PPAP
Rat plasma after administering PPCP
-2
-1.5
-1
-0.5
0
0.5
1
1.5
-4 -3 -2 -1 0 1 2 3
tcv[1]
tcv
[2]
-2.5
-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
-3 -2 -1 0 1 2
tcv[1]
tocv
[1]
C D
68
Figure 2-5. The PCA score plot of rat plasma and quality control samples from LC-
HRMS metabolomics. Green squares: pooled plasma samples as quality control. Red squares: rat plasma after administering PPCP. Blue squares: rat plasma after administering PPAP. Each square represents an individual rat.
t[2]
-20
-15
-10
-5
0
5
10
15
-25 -20 -15 -10 -5 0 5 10 15 20
t[1]
QC
Rat plasma after administering PPAP
Rat plasma after administering PPCP
69
Figure 2-6. The PCA score plot of rat plasma from LC-HRMS metabolomics after
administering PPCP or PPAP. Red squares: rat plasma after administering PPCP. Blue squares: rat plasma after administering PPAP. Each square represents an individual rat.
t[2]
-20
-15
-10
-5
0
5
10
15
-25 -20 -15 -10 -5 0 5 10 15 20
t[1]
Rat plasma after administering PPAP
Rat plasma after administering PPCP
70
Figure 2-7. The PLS-DA and OPLS-DA score plots and cross-validated score plots of rat plasma derived from LC-HRMS
metabolomics. A) PLS-DA score plot, B) OPLS-DA score plot, C) PLS-DA cross-validated score plot and D) OPLS-DA cross-validated score plot. Red squares: rat plasma after administering PPCP. Blue squares: rat plasma after administering PPAP. Each square represents an individual rat.
to[1
]
-15
-10
-5
0
5
10
-20 -15 -10 -5 0 5 10 15
t[1]
t[2]
-15
-10
-5
0
5
10
-20 -15 -10 -5 0 5 10 15
t[1]
A B
Rat plasma after administering PPAP
Rat plasma after administering PPCP
tocv
[1]
-10
-8
-6
-4
-2
0
2
4
6
8
-8 -6 -4 -2 0 2 4 6
tcv[1]
-10
-8
-6
-4
-2
0
2
4
6
8
-8 -6 -4 -2 0 2 4 6tcv[1]
tcv[2
]
C D
71
Figure 2-8. Validation plot obtained from 200 permutation tests for the OPLS-DA model
of rat plasma after administering PPCP or PPAP from LC-HRMS metabolomics.
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
-0.2 0 0.2 0.4 0.6 0.8 1
R2Q2
R2
, Q2
r(y, permuted y)
72
Figure 2-9. S-plots associated with the OPLS-DA score plot of data derived from LC-
HRMS of rat plasma after administering PPCP or PPAP. p[1] is the loading vector of covariance in the first principal component. p(corr)[1] is loading vector of correlation in the first principal component. Variables with |p| ≥ 0.05 and |p(corr)| ≥ 0.5 are considered statistically significant. Significant variables in blue color were identified and numbered according to Table 3-3. Unidentified significant variables in red color were listed in Table 3-4. Non-significant variables were in green color.
1
2, 46
3
5
7
8
9
1011
73
Figure 2-10. VIP plot of variables with VIP score higher than 1. Variables selected as significant ones from S-plot were marked in red with a VIP score > 1.7.
Variables
74
CHAPTER 3 1H NMR-BASED METABOLOMICS REVEALS URINARY METABOLOME MODIFICATIONS IN FEMALE RATS BY CRANBERRY PROCYANIDINS
Background
Cranberries (Vaccinium macrocarpon) are known to have various health benefits
including preventing urinary tract infection (Amy B. Howell, Reed, Krueger,
Winterbottom, Cunningham, & Leahy, 2005), delaying aging process (Wilson, Singh,
Vorsa, Goettl, Kittleson, Roe, et al., 2008), decreasing the risk of cardiovascular
diseases (Caton, Pothecary, Lees, Khan, Wood, Shoji, et al., 2010), inhibiting the
glycation of human hemoglobin and serum albumin (Haiyan Liu, Liu, Wang, Khoo,
Taylor, & Gu, 2011). Many of these health-promoting properties of cranberries were
attributed to their procyanidins content. Procyanidins are oligomers and polymers of (−)-
epicatechin or (+)-catechin with various degree of polymerization (L. Gu, Kelm,
Hammerstone, Beecher, Cunningham, Vannozzi, et al., 2002). Procyanidins are
classified as A-type and B-type according to their interflavan bonds. Apples contain
exclusively B-type procyanidins while over 65% procyanidins in cranberries are A-type
(L. Gu, et al., 2004).
Untargeted metabolomics employ high-throughput analytical platforms to
investigate the metabolic changes in a global manner. NMR spectroscopy is able to
detect hundreds of metabolites in biological samples. This technique has the advantages
of being quantitative, highly reproducible, non-selective and minimal sample preparation
(Dunn, Broadhurst, Atherton, Goodacre, & Griffin, 2011). Multivariate statistic techniques
are very helpful to reduce the dimensionality of the high-dimensional data produced by
metabolomics study (Bylesjö, Rantalainen, Cloarec, Nicholson, Holmes, & Trygg, 2006).
75
The urinary metabolome modifications in female Sprague-Dawley rats after
administering partially purified cranberry procyanidins (PPCP) or partially purified apple
procyanidins (PPAP) were investigated. We hypothesized that cranberry or apple
procyanidins will modify urinary metabolome and metabolome changes caused by
cranberry procyanidins will be different from those caused by apple procyanidins. The
objective of this chapter is to test this hypothesis using a 1H NMR-based metabolomics
approach.
Materials and Methods
Chemicals and Materials
Freeze-dried cranberry powder was provided by Ocean Spray Cranberries, Inc.
(Lakeville-Middleboro, MA, USA). Fresh granny smith apples were purchased from a
local grocery store. HPLC-grade methanol, acetone, sodium phosphate dibasic
anhydrous, sodium phosphate monobasic anhydrous, sodium hydroxide and sodium
chloride were purchased from Fischer Scientific Co. (Pittsburgh, PA, USA). D2O (99.9%
D), 2, 2-dimethyl-2-silapentane-5-sulfonate (DSS, 98%) was a product from Cambridge
Isotope Laboratories, Inc (Tewksbury, MA, USA). Sephadex LH-20 resin was purchased
from Sigma-Aldrich (St. Louis, MO, USA). Amberlite FPX 66 resin was obtained from
Rohm and Haas Co. (Philadelphia, PA, USA).
Partially purified cranberry procyanidins (3.95 g) was extracted from freeze-dried
cranberry power and purified using column chromatography on Amberlite FPX 66 resins
and Sephadex LH-20. Partially purified apple procyanidins (5.30 g) were extracted and
purified from fresh granny smith apples using a similar method. PPCP contained a
mixture of A- type and B- procyanidin oligomers and polymers. The total procyanidin
content was 51.1% (w/w) with high polymer (DP>10) content being 28.3% (w/w). PPAP
76
contained exclusively B-type procyanidin oligomers and polymers. The total procyanidin
content was 69.1% (w/w) with 8.8% (w/w) being high polymers with DP>10. Detailed
extraction procedure and compositional data was described in Chapter 2.
Animal Experiment
Approval for animal study was sought through the Institutional Animal Care and
Use Committee at the University of Florida. Female Sprague Dawley (n=24, 220-280 g)
were acclimated in the animal facility for 5 days using a purified diet free of flavonoid
compounds (D10012G, Research diet Inc., New Brunswick, NJ, USA). After the
acclimation period, rats were housed individually in a metabolic cage for 24 hours to
collect 24-hour baseline urine. Afterwards, rats were randomly divided into two groups
with 12 rats per group, and fasted for six hours before metabolomics study. PPCP or
PPAP were dispersed in water and administered by oral gavage at 0 and 12 hour using a
dose of 250 mg extracts/kg body weight. Rats had free access to food and water after
the gavage. The 24-hour urine of each rat was collected starting from 0 hour after the 1st
gavage. All urine samples were aliquoted and kept in a -80 oC freezer until analysis. One
rat did not produce any urine sample during 24-hour urine collection period after the
gavage with PPAP. Therefore this rat was excluded from this study.
1D 1H and 2D 1H-13C NMR analyses
Urine samples were thawed at 4oC in a cold room and then were centrifuged at
16, 000 g for 15 min. Gilson 215 Liquid Handler (Trilution software version 2.0) was used
to transfer urine (540 µL) into a 5 mm Bruker NMR tube (Z105684 Bruker 96 well rack)
having 60 µL of 1.5 M phosphate buffer (pH 7.4) containing 1 mM DSS in 10% D2O. The
mixture in each NMR tube was vortexed for 30 seconds. All 1H-NMR spectra were
collected on a 600 MHzAvance II NMR spectrometer (Bruker Biospin, Rheinstetten,
77
Germany) equipped with a 5 mm cryo probe. A sample changer (SampleJet, Bruker
BioSpin, Rheinstetten, Germany) operated under IconNMR software (Bruker BioSpin,
Rheinstetten, Germany) was used for automation. All NMR data were acquired at 300K.
Probe tuning and matching were optimized for the representative sample. A 90° pulse
length, the offset of the water signal, water suppression and receiver gain for a data set
were also determined on the representative sample in each run. The probe was
automatically locked to H2O+D2O (90%+10%) and shimmed for each sample. 1D
NOESY-presaturated spectra for all urine samples were recorded. 2D 1H-13C
heteronuclear single quantum coherence (HSQC) were obtained on selected samples to
aid metabolite identification.
Multivariate Statistical Analyses
Raw data from the spectrometer were zero filled, Fourier transformed (FT) and
phase corrected using NMRPipe (Delaglio, Grzesiek, Vuister, Zhu, Pfeifer, & Bax, 1995).
1D FT files were imported into MATLAB (R2013B, the Mathworks, Inc., Natick, MA,
USA) for referencing, removing residual water, baseline correction, alignment, and
normalization. Chemical shifts were referenced to the left peak of lactate at 1.47 ppm.
Peaks were aligned using peak alignment by fast Fourier transform (PAFFT) method and
spectra were normalized using probabilistic quotient normalization (PQN) method. The
resultant data set was imported into SIMCA (Version 13.0.3, Umetrics, Umea, Sweden)
for multivariate statistical analysis. Data were mean-centered and Pareto scaled before
PCA, PLS-DA and OPLS-DA analyses in SIMCA. Unsupervised PCA model was
performed to initially examine intrinsic variation in the data set. Then supervised pattern
recognition methods include PLS-DA and OPLS-DA (Bylesjö, Rantalainen, Cloarec,
Nicholson, Holmes, & Trygg, 2006) were used to extract maximum information on
78
discriminant compounds from the data. Validation of the model was tested using 7-fold
internal cross-validation and permutation tests for 200 times. To further evaluate the
predictive ability of the PLS-DA and OPLS-DA models, an external validation procedure
was performed (Brindle, et al., 2002; Llorach, et al., 2010). The NMR metabolomics data
set was split into a training set and a test set. Approximately 70% of the samples were
randomly selected as training set and the remaining 30% were treated as test set. PLS-
DA and OPLS-DA models were built based on the training set and obtained models were
used to blindly predict the classes of the samples in the test set. This procedure was
repeated 30 times and correct classification rates were calculated.
2D 1H-13C HSQC FT files were imported to MestReNova software (Version 9.0,
Mestrelab Research S.L., A Coruña, Spain) for peak picking. 1H-13C HSQC peak lists
were transferred to the COLMAR 13C-1H HSQC query web server (Bingol, Li,
Bruschweiler-Li, Cabrera, Megraw, Zhang, et al., 2014) for metabolite identification.
Results and Discussion
Urinary Metabolome Modification after PPCP or PPAP
The score plot of PCA in Figure 3-1 showed that rat baseline urine clustered on
upper left of the graph. They were partially separated from urine after administering
PPCP or PPAP. It suggested that urinary metabolome was modified after administering
procyanidins from cranberries or apples. However, no segregation of rat urine between
PPCP and PPAP was observed on PCA score plot. To further examine the metabolic
changes, supervised multivariate statistic techniques were used. PLS-DA and OPLS-DA
models were constructed on the urine samples in three comparisons: baseline vs. PPCP,
baseline vs. PPAP, PPCP vs. PPAP. For all three comparisons, two principal
components were generated to build PLS-DA model. One principal component and one
79
orthogonal component were constructed to build OPLS-DA model. R2 was calculated to
evaluate the performance of the supervised models and the goodness of fit. For baseline
vs. PPCP comparison, the R2Y of both PLS-DA and OPLS-DA models was 0.966 (Table
3-1), indicating that about 97% of variance in Y was explained by these supervised
models. For baseline vs. PPAP comparison, both models had a R2Y of 0.969 (Table 3-1)
suggesting that approximate 97% of variance of Y was explained by the models.
Similarly, for PPCP vs. PPAP comparison, supervised models with R2Y of 0.889 were
obtained (Table 3-1), suggesting that both models had a valid goodness of fit. A clear
segregation between baseline urine and urine after administering PPCP (Figure 3-2A, 3-
2B), baseline urine and urine after administering PPAP (Figure 3-3A, 3-3B), urine after
administering PPCP or after PPAP (Figure 3-4A, 3-4B) was observed on the score plot
of PLS-DA and OPLS-DA models for all three comparisons. Furthermore, validation
methods were conducted to validate these models. When processing multivariate data
with hundreds or thousands variables, one should take extreme caution about the
possibility of overfitting. Accidental correlations between one or more variables may
result in the unreliable segregation between groups (Kemsley, et al., 2007). To avoid this
caveat, three methods including internal cross-validation, permutation test and external
validation were conducted to confirm the validity and predictability of the supervised
models. For baseline vs. PPCP comparison, Q2 obtained from cross-validation for PLS-
DA and OPLS-DA was 0.853 and 0.852, respectively (Table 3-1). They were much
higher than 0.5, a thresh hold value for a good multivariate model of metabolomics data.
Furthermore, the score plots of PLS-DA and OPLS-DA from the cross-validation also
showed a clear discrimination between two groups and no misclassification was
80
observed during cross-validation (Figure 3-2C, 3-2D), indicating the segregation
observed on the score plot was not due to overfitting. Similarly, for the baseline vs.
PPAP comparison, cross-validation showed a Q2 of 0.757 and 0.777 for PLS-DA and
OPLS-DA models, respectively (Table 3-1). The cross-validation score plots of PLS-DA
and OPLS-DA models showed a clear separation between two groups and no
misclassification was observed (Figure 3-3C, 3-3D). As for the PPCP vs. PPAP
comparison, the urinary metabolic profiles of rats were modified by both PPCP and
PPAP. Therefore the magnitude of differences in urinary metabolome between PPCP
and PPAP was lower than that between baseline and PPCP or baseline and PPAP
group. This was demonstrated by the relatively lower Q2 of 0.656 and 0.629 obtained
from PLS-DA and OPLS-DA models in the cross-validation (Table 3-1). However, the Q2
of 0.635 and 0.613 were still higher than 0.5, indicating the supervised models were
valid. The cross-validated score plots of PLS-DA and OPLS-DA showed a separation
between two groups, although three samples had cross validation score of 0 on the
OPLS-DA cross-validate score plot (Figure 3-4D).
The second validation method used was the permutation test. The class labels of
tested groups were permuted and randomly assigned to different observations. With the
permutated class labels, 200 new supervised models were built, respectively. R2 and Q2
within each model was calculated and a regression line was drawn. The Q2-intercept
value obtained from the regression line should be lower than 0.05 for a valid model.
Permutation regression line was obtained from the OPLS-DA model derived from
baseline vs. PPAP comparison (Figure 3-5A), baseline vs. PPCP comparison (Figure 3-
5B), and PPCP vs. PPAP comparison (Figure 3-5C). The negative Q2 intercept
81
suggested a good predictability of the OPLS-DA models. However, the relatively high R2
intercept indicated somewhat overfitting of the supervised models. To further confirm the
validity and predictability of PLS-DA and OPLS-DA models, external validation which is a
more scrupulous and demanding method was used (Eriksson, 2006). The results
showed that the correct classification rate of 99.4% and 95.3% was obtained for baseline
vs. PPAP comparison and baseline vs. PPAP comparison, respectively (Table 3-1). As
for the PPCP vs. PPAP comparison, correct classification rate of 95.8% and 96.4% was
calculated for PLS-DA and OPLS-DA model, respectively (Table 3-1). The external
validation results demonstrated that supervised models derived from rat urine NMR data
had excellent predictability and were able to correctly predict the unknown urine samples
with a correct classification rate of above 95%. All three validation tests suggested that
1H NMR global metabolomics approach was effective to reveal the urinary metabolome
modification in female rats after administering PPCP or PPAP.
Discriminant Metabolites Identification
The statistical S-lineTM is the tailored S-plot plot for NMR spectroscopy data and
was used to identify the potential metabolites that contribute to the urinary metabolome
modification by PPCP or PPAP. S-line combines the covariance (magnitude) and
correlation (reliability) for the model variables (Wiklund, et al., 2008) and visualizes both
in one graph. The p(ctr) is the centered loading vector of the first principal component. It
was colored according to the absolute value of the correlation loading p(corr). A
p(corr)>0.5 was selected as a significance level. The advantage of the S-line plot over S-
plot is that it displays the predictive loading in a form resembling the original NMR
spectra. The discriminant metabolites were identified by comparing their NMR spectra
with reported spectra (Bouatra, Aziat, Mandal, Guo, Wilson, Knox, et al., 2013), Human
82
Metabolome Database (Wishart, et al., 2007) and the COLMAR 13C-1H HSQC query
(Bingol, et al., 2014). These discriminant metabolites were summarized in Table 3-2.
The HPLC chromatograms of PPCP and PPAP depicted in Chapter 2 showed that
PPCP contained both A-type and B-type procyanidins, while PPAP had exclusively B-
type procyanidins. Majority of ingested procyanidins are not absorbed in small intestine.
They reach the colon intact and are degraded by gut microbiota (Ou & Gu, 2014). The
resultant exogenous procyanidins metabolites are part of food metabolome and may also
change the endogenous metabolome. A total of 17 metabolites were modified in the
urine of female rats after PPCP or PPAP. The urinary level of hippuric acid, succinic
acid, lactic acid, unknown metabolite 1 at 7.30-7.35 ppm, and unknown metabolite 2 at
7.37-7.42 ppm were increased after rats were administered with PPCP compared to
baseline urine. Endogenous metabolites including α-ketoglutaric acid and citric acid were
decreased after administering PPCP compared to baseline samples. Similarly, rats after
PPAP had a lower urinary level of α-ketoglutaric acid, citric acid and creatinine compared
to baseline urine. PPAP caused a stronger increase of several metabolites including D-
maltose, α-D-glucose, formic acid, 3-(3’-hydroxyphenyl)-3-hydroxypropanoic acid, p-
hydroxyphenylacetic acid, phenol, unknown metabolite 1 at 7.30-7.35 ppm, and unknown
metabolite 2 at 7.37-7.42 ppm, unknown metabolite 3 at 6.77 (s) ppm, unknown
metabolite 4 at 6.73 (dd) ppm, and unknown metabolite 5 at 7.04 (s) ppm. By comparing
the urinary metabolite profile of rats after PPCP and after PPAP, it was found that
hippuric acid, unknown metabolite 1 at 7.30-7.35 ppm, and unknown metabolite 2 at
7.37-7.42 ppm increased after PPCP. Metabolites including D-maltose, 3-(3’-
hydroxyphenyl)-3-hydroxypropanoic acid, p-hydroxyphenylacetic acid, phenol, unknown
83
metabolite 3 at 6.78 (s) ppm, and unknown metabolite 4 at 6.74 (dd) ppm and unknown
metabolite 5 at 7.04 (s) ppm decreased after PPCP. The most important metabolites that
were responsible for the separation between baseline vs. PPCP were hippuric acid,
unknown metabolite 1 at 7.30-7.35 ppm, and unknown metabolite 2 at 7.37-7.42 ppm.
The correlation loadings p(corr) of these three metabolites were 0.94, 0.91, and 0.73,
which were much higher than the statistically significant level of 0.5. For baseline vs.
PPAP, metabolites including 3-(3’-hydroxyphenyl)-3-hydroxypropanoic acid, p-
hydroxyphenylacetic acid, phenol, formic acid, unknown metabolite 1 at 7.30-7.35 ppm,
and unknown metabolite 2 at 7.37-7.42 ppm, unknown metabolite 3 at 6.78 (s) ppm,
unknown metabolite 4 at 6.74 (dd) and unknown metabolite 5 at 7.04 (s) ppm had the
highest correlation loadings p(corr) of 0.79, 0.84, 0.90, 0.78, 0.82, 0.73, 0.86, 0.83 and
0.92, respectively. By comparing PPCP vs. PPAP, the most important metabolite for the
separation was an increased hippuric acid after PPCP with a correlation loading p(corr)
of 0.94. This result was consistent with a previous rat study which showed that
consumption of cranberry powder caused an increase in urinary excretion of hippuric
acid. Its quantity in urine was higher than any other urinary phenolic acids (Prior, Rogers,
Khanal, Wilkes, Wu, & Howard, 2010). In addition, several exogenous metabolites
including 3-(3’-hydroxyphenyl)-3-hydroxypropanoic acid, p-hydroxyphenylacetic acid,
phenol, unknown metabolite 3 at 6.78 (s) ppm, unknown metabolite 4 at 6.74 (dd) ppm
and unknown metabolite 5 at 7.04 (s) ppm also contributed to the segregation of
metabolite profiles between PPCP and PPAP.
PPCP increased the urinary excretion of lactic acid, succinic acid and hippuric
acid. Both hippuric acid and succinic acid are the intermediates of phenylalanine
84
metabolism, suggesting that intake of PPCP altered phenylalanine metabolism pathway
at gene or protein levels. Both lactic acid and succinic acid participate in propanoate
metabolism, indicating an upregulation of propanoate metabolism by PPCP. Citric acid
and α-ketoglutaric acid are key intermediates in citrate cycle. Both compounds also
participate in the metabolism of glyoxylic acid and dicarboxylic acid. A reduction of citric
acid and α-ketoglutaric in rat urine indicated a downregulation of citrate cycle and
glyoxylic acid metabolism by both PPCP and PPAP. Compared to baseline, PPAP
increased the urinary excretion of D-glucose and D-maltose that are intermediates in
starch and sucrose metabolism, suggesting that PPAP had an impact on the metabolism
of carbohydrate in rats. Previous research suggested that procyanidins prevented or
alleviated type 2 diabetes in part by inhibiting enzymes in starch digestion (Y. Gu, Hurst,
Stuart, & Lambert, 2011; Lee, Cho, Tanaka, & Yokozawa, 2007). Phenol, p-
hydroxyphenylacetic acid and 3-(3’-hydroxyphenyl)-3-hydroxypropanoic acid also
increased after PPAP. They are microbial metabolites of procyanidins by gut microbiota
(Ou & Gu, 2014). Phenol and p-hydroxyphenylacetic acid may also originate from
tyrosine metabolism. Phenol is a metabolite degraded directly from tyrosine. p-
Hydroxyphenylacetic acid is an intermediate converted from 4-
hydroxyphenylacetaldehyde which is oxidized from p-tyramine in the pathway of tyrosine
metabolism. An increase of formic acid after PPAP suggested an alternation of pyruvate
metabolism pathway.
Summary
Female Sprague-Dawley rat urinary metabolome modifications after administering
PPCP or PPAP were detected using a global 1H NMR metabolomics approach. PPCP
caused an increase of hippuric acid, lactic acid, succinic acid, but a decrease of citric
85
acid and α-ketoglutaric acid in rat urine after administering PPCP compared to baseline
urine. The urinary level of α-D-glucose, D-maltose, 3-(3’-hydroxyphenyl)-3-
hydroxypropanoic acid, p-hydroxyphenylacetic acid and phenol were increased but citric
acid, α-ketoglutaric acid and creatinine were decreased after administering PPAP
compared to baseline urine. The metabolite profile differences between PPCP and PPAP
were observed. Discriminate metabolite included hippuric acid which was higher in rat
urine after PPCP. D-maltose, 3-(3’-hydroxyphenyl)-3-hydroxypropanoic acid, p-
hydroxyphenylacetic acid and phenol were lower after PPCP.
86
Table 3-1. Summary of parameters for PLS-DA and OPLS-DA models for rat baseline urine and urine after administering PPCP or PPCP by oral gavage.
Baseline vs. PPCP Baseline vs. PPAP PPCP vs. PPAP
PLS-DA OPLS-DA PLS-DA OPLS-DA PLS-DA OPLS-DA
Na
2 1Pc+1Od 2 1Pc+1Od 2 1Pc+1Od
R2
X(cum)b
0.248 0.248 0.326 0.326 0.278 0.278
R2
Y(cum)b
0.966 0.966 0.969 0.969 0.889 0.889
Q2
(cum)b
0.853 0.852 0.757 0.777 0.656 0.629
*Correct Classification Rate
0.950±0.089 0.956±0.075 0.967±0.068 0.967±0.068 0.900±0.143 0.922±0.105
a N: number of components. b R2X (cum)and R2Y (cum) are the cumulative modeled variations in the X and Y matrix, respectively. Q2Y (cum) is the cumulative predicted variation in the Y matrix.
c Predictive component. d Orthogonal component. * Correct classification rate was obtained from external validation procedure repeated for 30 times.
87
Table 3-2. Summary of the metabolite profile changes in rat baseline urine and urine
after administering PPCP or PPCP by oral gavage.
a Arrows indicated a decrease or increase in metabolites detected in rat urine after PPCP compared to baseline.
b Arrows indicated a decrease or increase in metabolites detected in rat urine after PPAP compared to baseline.
c Arrows indicated a decrease or increase in metabolites detected in rat urine after PPCP compared to PPAP.
*Compound was identification by COLMAR 13C-1H HSQC query (Bingol, et al., 2014).
Metabolites Chemical shift (multiplicity)
PPCP vs. Baseline a
PPAP vs. Baseline b
PPCP vs. PPAP c
lactic acid* 1.32 (d) ---- ----
succinic acid* 2.39 (s) ---- ----
citric acid* 2.52 (d), 2.67 (d) ----
α-ketoglutaric acid* 2.43 (t), 2.99 (t) ----
creatinine* 3.03 (s), 4.04 (s) ---- ----
α-D-glucose* 5.23 (d), 4.65 (d) ---- ----
D-maltose* 3.26 (dd), 3.42 (t), 3.58 (m), 3.62 (m), 3.71 (m), 3.76 (m), 3.83 (m), 3.89 (m), 3.96 (m), 5.40 (d), 5.25 (d)
----
phenol 6.94 (d) ----
p-hydroxyphenylacetic acid (PHPAA)*
7.15 (d), 6.85 (d) ----
3-(3’-hydroxyphenyl)-3-hydroxypropanoic acid (HPHPA)*
5.05 (dd), 6.84 (dd), 6.97 (d), 7.22 (t)
----
hippuric acid* 7.53 (t), 7.62 (t), 7.82 (d)
----
formic acid* 8.45 (s) ---- ----
unknown metabolite 1 7.30-7.35
unknown metabolite 2 7.37-7.42
unknown metabolite 3 6.77 (s) ----
unknown metabolite 4 6.73 (dd) ----
Unknown metabolite 5 7.04 (s) ----
88
Figure 3-1. The PCA score plot of rat baseline urine and urine after administering PPCP
or PPAP. Green squares: rat baseline urine. Red squares: rat urine after administering PPAP. Blue squares: rat urine after administering PPCP. Each square represents an individual rat.
Rat baseline urine
Rat urine after PPCP
Rat urine after PPAP
-8
-6
-4
-2
0
2
4
6
-8 -6 -4 -2 0 2 4 6t[1]
t [2
]
89
Figure 3-2. The PLS-DA and OPLS-DA score plots and cross-validated score plots of rat baseline urine and urine after
administering PPCP. A) PLS-DA score plot, B) OPLS-DA score plot, C) PLS-DA cross-validated score plot and D) OPLS-DA cross-validated score plot. Green squares: rat baseline urine before administering PPCP. Blue squares: rat urine after administering PPCP. Each square represents an individual rat.
Rat baseline urine before administering PPCPRat urine after administering PPCP
B
C
A
D
-6
-4
-2
0
2
4
-10 -8 -6 -4 -2 0 2 4 6 8t[1]
t [2
]
to [
1]
-8
-6
-4
-2
0
2
4
6
-8 -6 -4 -2 0 2 4 6t[1]
-2.5
-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
-4 -3 -2 -1 0 1 2 3tcv[1]
tcv
[2]
-3
-2
-1
0
1
2
3
-3 -2 -1 0 1 2tcv[1]
tocv
[1]
90
Figure 3-3. The PLS-DA and OPLS-DA score plots and cross-validated score plots of rat baseline urine and urine after
administering PPAP. A) PLS-DA score plot, B) OPLS-DA score plot, C) PLS-DA cross-validated score plot and D) OPLS-DA cross-validated score plot. Green squares: rat baseline urine before administering PPAP. Red squares: rat urine after administering PPAP. Each square represents an individual rat.
Rat baseline urine before administering PPAP
Rat urine after administering PPAP
A B
C D
t [2
]
-6
-4
-2
0
2
4
-10 -8 -6 -4 -2 0 2 4 6 8t[1]
to [
1]
-6
-4
-2
0
2
4
-10 -8 -6 -4 -2 0 2 4 6 8t[1]
-1.5
-1
-0.5
0
0.5
1
-5 -4 -3 -2 -1 0 1 2 3 4tcv[1]
-2
-1
0
1
2
-4 -3 -2 -1 0 1 2 3tcv[1]
tcv
[2]
tocv
[1]
91
Figure 3-4. The PLS-DA and OPLS-DA score plots and cross-validated score plots of rat urine after administering PPCP
or PPAP. A) PLS-DA score plot, B) OPLS-DA score plot, C) PLS-DA cross-validated score plot and D) OPLS-DA cross-validated score plot. Blue squares: rat urine after administering PPCP. Red squares: rat urine after administering PPAP. Each square represents an individual rat.
Rat urine after administering PPCPRat urine after administering PPAPA B
C D
t [2
]
-8
-6
-4
-2
0
2
4
6
-8 -6 -4 -2 0 2 4 6t[1]
-10
-8
-6
-4
-2
0
2
4
6
8
-8 -6 -4 -2 0 2 4 6t[1]
to [
1]
tocv
[1]
-2
-1
0
1
2
-2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2tcv[1]
tocv
[1]
-4
-3
-2
-1
0
1
2
3
-3 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5tcv[1]
92
Figure 3-5. Validation plot obtained from 200 permutation tests for the OPLS-DA models
of rat baseline urine and urine after administering PPCP or PPAP from 1H NMR metabolomics. A) Rat baseline urine vs. urine after administering PPAP, B) rat baseline urine vs. urine after administering PPCP and C) urine after administering PPCP vs. after PPAP.
R2,
Q2
r(y, permuted y)R
2, Q
2r(y, permuted y)
R2,
Q2
r(y, permuted y)
BA
CR
2, Q
2R
2, Q
2
93
Figure 3-6. S-line associated with the OPLS score plots of data derived from rat baseline urine and urine after PPCP or
PPAP. A) Baseline urine vs. urine after PPCP, B) baseline urine vs. urine after PPAP and C) urine after PPCP vs. urine after PPAP. The x-axis is chemical shift derived from NMR spectra. The y-axis p(ctr)[1] is the centered loading vector of the first principal component. p(ctrl)[1] is colored according to the absolute value of the correlation loading p(corr). p(corr)>0.5 is selected as a significance level.
94
Figure 3-6. Continued.
95
Figure 3-6. Continued.
96
CHAPTER 4 A 1H NMR BASED APPROACH TO INVESTIGATE METABOLOMIC DIFFERENCES IN
THE PLASMA AND URINE OF YOUNG WOMEN AFTER CRANBERRY JUICE OR APPLE JUICE CONSUMPTION
Background
Cranberries (Vaccinium macrocarpon) are a native crop in North America. Fresh
cranberries have a tart taste, therefore majority of them are processed into juice for
consumption. Cranberry procyanidins are oligomeric or polymeric of flavan-3-ols linked
through interflavan bonds. B-type interflavan linkage is C4→ C8 and/or C4→ C6. A-type
procyanidins contain an additional ether bond C2→O→C7 (Ou & Gu, 2014). Most foods
including apples or apple juice contain exclusively B-type procyanidins, while
cranberries or cranberry juice contains both A and B-type procyanidins (L. Gu, Kelm,
Hammerstone, Beecher, et al., 2003). Ingestion of cranberry juice has long been
associated with prevention of urinary tract infection (UTI) (Blatherwick, 1914). Studies
showed that A-type procyanidins from cranberry juice inhibited the adhesion of
uropathogenic E. coli, whereas B-type procyanidins from apple juice showed no activity
(Amy B. Howell, Reed, Krueger, Winterbottom, Cunningham, & Leahy, 2005). Anti-
adhesion activity in human urine was detected following cranberry juice cocktail
consumption, but not after consumption of the apple juice (Amy B. Howell, Reed,
Krueger, Winterbottom, Cunningham, & Leahy, 2005). A-type trimers demonstrated
anti-adhesion activity, whereas epicatechin and a B-type dimer showed no such effect
(Foo, Lu, Howell, & Vorsa, 2000).
We hypothesized that cranberry juice consumption may have a different impact
on human metabolome compared to apple juice. A 1H NMR-based metabolomics
97
approach with multivariate statistic techniques was applied to analyze the overall
metabolic impact by cranberry juice and differentiate that from apple juice consumption.
Materials and Methods
Chemicals and Materials
Cranberry juice cocktail (double strength, 54% juice) and 100% apple juice was
provided by Ocean Spray Cranberries, Inc. (Lakeville-Middleboro, MA). Gallic acid,
HPLC-grade acetonitrile, methylene chloride, methanol, acetic acid, Folin−Ciocalteau
reagent, sodium carbonate, sodium phosphate dibasic anhydrous, sodium phosphate
monobasic anhydrous, sodium hydroxide, sodium chloride, sodium azide, sucrose,
glucose, and fructose were purchased from Fischer Scientific Co. (Pittsburgh, PA,
USA). D2O (99.9% D), 2, 2-dimethyl-2-silapentane-5-sulfonate (DSS, 98%) were from
Cambridge Isotope Laboratories, Inc. (Tewksbury, MA, USA). Sephadex LH-20 resin
was purchased from Sigma-Aldrich (St. Louis, MO, USA). Amberlite FPX 66 resin was a
product of Rohm and Haas Co. (Philadelphia, PA, USA). Pooled plasma used as quality
control samples were purchased from American Red Cross and they were collected
over a period of about 2 weeks.
Total Phenolics, Total Anthocyanins, Procyanidin Composition and Content
Two hundred and fifty mL of cranberry juice or apple juice was loaded onto a
column packed with Amberlite FPX 66 resins. Column was eluted with 3 L of
deionization water to remove sugars and ascorbic acids. Column was eluted with 300-
400 mL of methanol to yield cranberry juice sugar-free extract (890 mg) or apple juice
sugar-free extract (393 mg). The total phenolic content of juice sugar-free extracts were
determined by Folin-Ciocalteu assay (Singleton & Rossi Jr, 1965). A pH differential
assay was used to determine the total anthocyanin content (Giusti & Wrolstad, 2001).
98
Another 250 mL of cranberry juice or apple juice was loaded onto a column (5.8×28 cm)
packed with Sephadex LH-20. The column was eluted with 30% methanol to remove
anthocyanins and phenolic acids, and then eluted with 70% acetone to yield cranberry
juice procyanidin extract (273 mg) or apple juice procyanidin extract (18.5 mg). The
procyanidin composition and content were analyzed on an Agilent 1200 HPLC system
(Palo Alto, CA) equipped with a binary pump, an autosampler, a fluorescence detector,
and a HCT ion trap mass spectrometer (Bruker Daltonics, Billerica, MA). Separation of
procyanidins was carried out on a Luna Silica (2) column (250 × 4.6 mm, 5 μm particle
size, Phenomenex, Torrance, CA) at a column temperature of 37 oC. The binary mobile
phase consisted of (A) methylene chloride/methanol/acetic acid/water (82:14:2:2, v: v: v:
v) and (B) methanol/acetic acid/ water (96:2:2, v: v: v: v). The 70 min gradient was as
follows: 0−20 min, 0.0−11.7% B linear; 20−50 min, 11.7−25.6% B linear; 50−55 min,
25.6−87.8% B linear; 55−65 min, 87.8% B isocratic; 65−70 min, 87.8−0.0% B linear;
followed by 5 min of column re-equilibration before the next injection. Excitation and
emission of the fluorescent detector were set at 231 and 320 nm, respectively.
Electrospray ionization at negative mode was performed using nebulizer 50 psi, drying
gas 10 L/min, drying temperature 350 °C, and capillary 4000 V. Mass spectra were
recorded from m/z 150 to 2000. The most abundant ion in full scan was isolated, and its
product ion spectra were recorded.
Procyanidins in PPCP and PPAP were quantified based on a method
standardized by Mars Inc.(Robbins, et al., 2012). This method uses (-)-epicatechin as a
calibrant and relative response factors for procyanidin dimers through nonamers,
because at the same concentration fluorescent signal response ratio between an
99
oligomer and (−)-epicatechin stays constant under the same HPLC condition. The
relative response factor of nonamers was used as the response factor of high polymers
Sugar Analyses in Cranberry Juice and Apple Juice
Sugar analysis was conducted on an Agilent 1200 HPLC system consisting of an
autosampler, a binary pump, and a refractive index detector (Agilent Technologies, Palo
Alto, CA). Separation was carried out on a Restek ultra amino column (5 μm, 250 × 4.6
mm). The column temperature was maintained at 30 °C and a 5 μL of sample was
injected. Acetonitrile/water (80:20, v: v) was used as the mobile phase at a constant
flow rate of 1.0 mL/min. The optical unit temperature was set at 35 °C and the refractive
index detector signal was monitored in positive polarity. The run time for each sample
was 15 min followed by 5 min post time before the next run. Calibration curves were
constructed by use of pure standards of glucose, fructose, and sucrose.
Subjects and Study Design
Human study was approved by Institutional Review Boards at University of
Florida. Eighteen healthy female college students between 21-29 years old with a
normal BMI of 18.5-25 were recruited. Each subject was provided with a list of foods
that contained significant amount of procyanidins, such as cranberries, apples, grapes,
blueberries, chocolate and plums. They were advised to avoid these foods during the 1-
6th day and the rest of the study. On the morning of the 7th day, a first-morning baseline
urine sample and blood sample were collected from all human subjects after overnight
fasting. Participants were then randomly allocated into two groups (n=9 for each group)
to consume either cranberry juice or apple juice. Six bottles (250 mL/bottle) of juice
were given to participants to drink in the morning and evening of the 7th, 8th, and 9th day.
On the morning of 10th day, all subjects returned to clinical unit to provide a first-morning
100
urine sample after overnight fasting. The blood sample was also collected from
participants 30-60 min later after they drank another bottle of juice in the morning. After
two-weeks of wash out period, participants switched to the alternative regimen and
repeat the protocol. The timeline of the trial was summarized in Table 4-1. Blood
samples were centrifuged at 2,000 g for 10 min at 4 oC to obtain plasma. All urine and
plasma samples were aliquot and kept in a -80 oC freezer until analysis. One human
subject was dropped off this study because she missed part of her appointments.
Another two human subjects were removed from urine metabolomics analyses because
they failed to provide required urine samples.
1H NMR Metabolomics Analyses
Plasma or urine samples were taken out of -80 oC freezer to thaw at 4 oC in a
cold room, and then centrifuged at 5,220 g for 5 min. Plasma (400 µL) was mixed with
200 µL of saline solution (0.9% NaCl in D2O). Urine (400 µL) was mixed with 200 µL of
phosphate buffer (pH 7.4, DSS added). Both plasma and urine samples were
transferred into 5 mm Bruker NMR tubes (Z105684 Bruker 96 well rack) using Gilson
215 Liquid Handler (Trilution software version 2.0). All 1H-NMR spectra were collected
on a 600 MHz Avance II NMR spectrometer (Bruker Biospin, Germany) equipped with a
5mm cryo probe. Instrument had a samples changer (Sample Xpress Lite Autosampler)
under Icon-NMR. 1D NOESY-presaturated spectra for urine and 1D CPMG-
presaturated spectra for plasma were recorded. All NMR data were acquired at 25°C.
Probe tuning and matching were optimized for the first sample in each run. A 90° pulse
length, the offset of the water signal, water suppression and receiver gain for a data set
were also determined on the first sample in each run. The probe was automatically
locked to H2O+D2O (90%+10%) and shimmed for each sample.
101
Multivariate Data Processing
All NMR spectra were phased and baseline corrected using NMRPipe and then
converted to FT files. The FT files were imported into MATLAB (R2013B, the
Mathworks, Inc., Natick, MA). Water regions (4.6-5.1 ppm) and DSS region (-0.2-0.5
ppm) were removed. Then the spectra were aligned and normalized in MATLAB. The
resultant data set was imported into SIMCA (Version 13.0.3, Umetrics, Umea, Sweden)
for multivariate statistical analysis. Data were mean-centered, Pareto scaled before
PCA, PLS-DA and OPLS-DA analysis in SIMCA. Unsupervised PCA model was
performed to initially examine intrinsic variation in the data set. Then supervised pattern
recognition methods PLS-DA and OPLS-DA (Bylesjö, Rantalainen, Cloarec, Nicholson,
Holmes, & Trygg, 2006) were used to extract maximum information on discriminant
compounds from the data. Validation of the model was tested using 7-fold internal
cross-validation and permutation tests for 200 times. To further evaluate the predictive
ability of the PLS-DA and OPLS-DA models, an external validation procedure was
performed (Brindle, et al., 2002; Llorach, et al., 2010). The whole data set was split into
a training set and a test set. Approximately 70% of the samples were randomly selected
as training set and the remaining 30% were treated as test set. PLS-DA and OPLS-DA
models were built based on the training set and obtained models were used to blindly
predict the classification of the samples in the test set. This procedure was repeated 30
times and the correct classification rate was calculated. For univariate analyses,
Welch’s t test was carried out on the NMR signal intensity of selected metabolites which
were considered to be responsible for the separation between treatments from
multivariate analyses. Benjamini–Hochberg (1995) procedure (α=0.01) was conducted
to control false discoveries. Box-and-whisker plot was drawn to display variations in
102
samples within each treatment. The univariate analyses were done using Excel
Microsoft (Version 2007, Microsoft Corporation., Seattle, WA, USA).
Results and Discussion
Juice Analyses
Figure 4-1 showed HPLC chromatograms of procyanidins extracted from
cranberry juice and apple juice using fluorescence detection. Both A- and B-type
procyanidins were detected in cranberry juice, while only B-type procyanidins and their
oxidized forms were found in apple juice. Oligomeric procyanidins with DP 1-5 and high
polymeric procyanidins in juice were identified and quantified by HPLC-MSn. The total
quantifiable procyanidin content was 566 µg/mL in cranberry juice and 9.68 µg/mL in
apple juice. The total phenolics and total anthocyanins of apple juice (124 µg gallic acid
/mL, 0.12 µg cyanidin 3, 5-diglucoside/mL) were lower than those in cranberry juice
(913 µg gallic acid/mL, 59.2 µg cyanidin 3,5-diglucoside/mL) (Table 4-2). It should be
noted that ascorbic acid was removed using a chromatographic method so it was not
counted as part of total phenolics.
Sugar composition and content were analyzed on HPLC using a refractive index
detector. Glucose, fructose and sucrose were found in apple juice, while only glucose
and fructose appeared in cranberry juice (Figure 4-2). The results were consistent with
previously reported findings (Fuleki, Pelayo, & Palabay, 1994). Table 4-2 showed total
sugar content in apple juice was 10 times higher than that in cranberry juice, with
fructose as the dominant type of sugar.
Quality Control Data
To validate NMR acquisition method, quality control samples consisting of 17
replicates of pooled plasma collected from American Red Cross were analyzed along
103
with experimental samples. PCA model was built to analyze the metabolic differences
between quality control and experimental samples. The mechanism was based on the
ability of PCA model to cluster samples in an unsupervised approach. PCA score plot
(Figure 4-3) showed that the 17 replicates were tightly clustered, suggesting that our
data acquisition method was valid.
Multivariate Analyses of Plasma after Drinking Cranberry Juice vs. Drinking Apple Juice
PCA, PLS-DA and OPLS-DA models were built to analyze the metabolic patterns
of plasma and urine. Neither PCA, PLS-DA or OPLS-DA models was able to detect
metabolic differences in baseline plasma and plasma after cranberry or apple juice
consumption (Table 4-3). Q2 calculated from cross-validation was all below 0.5
indicating the poor predictability of these supervised models. However, overall
metabolic profiles of plasma and urine after cranberry juice consumption were found to
be different from those after apple juice consumption. Figure 4-4 showed that two
groups of plasma samples were segregated on the score plot of OPLS-DA model but
not on a PCA model. Compared with PLS-DA model, OPLS-DA performed the analysis
with orthogonal filtration of matrix X on a vector Y. The variance in the X matrix was split
by OPLS model into predictive and orthogonal variance. “Structure noise” of data matrix
which was unrelated to the variation of interest such as genetic background, age,
physical activity, stress, etc., was filtered and described only by the orthogonal
component. The variation of scientific interest was only observed in the predictive
component, which is the first component. Therefore the interpretability of the resulting
model was increased (Fonville, et al., 2010). Plasma after drinking cranberry juice and
apple juice were clearly segregated on the score plot of OPLS-DA. One predictive
104
component and seven orthogonal components were generated in the OPLS-DA. This
model obtained high-quality parameters with an overall value of R2X, R2Y, and Q2 of
0.856, 0.979, and 0.652, respectively (Table 4-4). It indicated that about 85% in X data
set and 98% of variance in Y data set was explained by this model. Model validation
was performed by permutation test (n=200) and 7-fold internal validation. Q2 of 0.652
calculated from cross-validation was higher than 0.5, which was considered as good for
metabolomics data. Q2 indicated the OPLS-DA model had a good predictability. The Q2-
intercept values from permutation test (Figure 4-6) were lower than 0.05, indicating that
the achieved segregation was not due to overfitting. OPLS-DA score plot and cross-
validated score plot were visualized in Figure 4-5. Although one plasma sample after
cranberry juice consumption and two plasma samples after apple juice consumption
were misclassified during internal validation, the rest of samples were correctly
classified into two groups.
Multivariate Analyses of Urine after Drinking Cranberry Juice vs. Drinking Apple Juice
Similar results were obtained from urine metabolomics data. As an unsupervised
technique, PCA reveals the main structure in the data without considering a special
direction or type of information. The score plot of PCA in Figure 4-7A showed some
segregation between two groups of urine samples. By using a supervised pattern
recognition technique, a much more clear segregation was observed on the score plot
of OPLS-DA (Figure 4-7B). One predictive component and two orthogonal components
were generated from the OPLS-DA model. An overall value of R2X, R2Y, and Q2 of
0.548, 0.853, and 0.503 indicated good quality of the OPLS-DA model (Table 4-5).
Model validation was performed by permutation test (n=200) and 7-fold internal
105
validation. Q2 of 0.503 suggested that the model had a good predictive ability. The
regression line generated from permutation test (Figure 4-9) suggested the OPLS-DA
model was valid. The cross-validated score plot (Figure 4-8) was another way to
visualize the 7-fold internal validation. Both plasma and urine metabolomics data
suggested a differentiation between cranberry juice and apple juice consumption was
obtained.
Discriminant Metabolite Identification
To identify the contributing metabolites that are responsible for the separation of
cranberry juice and apple juice consumption, a S-lineTM technology, which is the tailored
S-plot (Llorach, Urpi-Sarda, Jauregui, Monagas, & Andres-Lacueva, 2009; Wiklund, et
al., 2008) for NMR spectroscopy data was used. It visualizes both the covariance and
the correlation structure between the X-variables and the predictive score. The p(ctr) is
the centered loading vector of the first principal component. It was colored according to
the absolute value of the correlation loading p(corr). A p(corr)>0.5 was selected as
significance level. The advantage of the S-line plot is that it displays the predictive
loading in a form resembling the original NMR spectra. A list of makers detected in the
S-line was then subjected to Welch’s t test, and the p-value obtained for each marker
was smaller than 0.01 (Table 4-6). Both multivariate and univariate analyses concluded
eight significant metabolites were responsible for the separation between drinking
cranberry juice and apple juice consumption. The contributing metabolites were
identified by comparing their NMR spectra with published papers (Bouatra, et al., 2013;
Psychogios, Hau, Peng, Guo, Mandal, Bouatra, et al., 2011) and those registered in the
Human Metabolome Database. These markers found from plasma and urine
metabolomic data were summarized in Table 4-6.
106
A-type procyanidins in cranberry juice or B-type procyanidins in apple juice have
very low absorption rate in vivo. Only a small portion of epicatechin and procyanidin
oligomers with DP<5 were absorbed in small intestine (Ou & Gu, 2014). The absorption
rate was below 5% (Gonthier, Donovan, Texier, Felgines, Remesy, & Scalbert, 2003;
Rzeppa, Bittner, Döll, Dänicke, & Humpf, 2012; Tsang, et al., 2005). The majority of
procyanidin oligomers and polymers reach colon, where both A- and B-type
procyanidins are degraded by gut microflora to form various microbial metabolites. Part
of the microbial metabolites were low molecular weight phenolic acids and
phenylvalerolactones (Ou & Gu, 2014). In the present study, the detection of these
microbial metabolites by NMR spectroscopy was limited due to their low levels in urine
or plasma. However, by using a global metabolomics approach we were able to find
several endogenous metabolites that are responsible for the separation of cranberry
juice and apple juice consumption. These metabolites were marked on the S-line in
Figure 4-10 and Figure 4-11. Cranberry juice and apple juice consumption had different
impact on endogenous metabolites in urine and plasma. The plasma level of citric acid
was considered to be increased after consumption of cranberry juice according to its
loading profile. Although its relatively low magnitude makes it not an ideal case, its
correlation loading p(corr) >0.5 is accepted for being statistically significant. Lactate, D-
glucose and two unidentified metabolites in plasma were higher after consumption of
apple juice. One unidentified metabolite with chemical shift at 3.56 (m) ppm and 4.01
(m) ppm was first identified as quinic acid by matching its NMR spectrum with published
data in Human Metabolome Database. We then spiked the plasma samples with pure
quinic acid to disprove the identification. Cranberry juice consumption caused a stronger
107
increase in urinary excretion of hippuric acid and one unidentified metabolites. The
result was consistent with our previous finding that urinary level of hippuric acid in
female rats was greatly increased after intake of cranberry procyanidins. Hippuric acid is
formed by the conjugation of benzoic acid with glycine in the liver, and then excreted in
urine. Production of hippuric acid is mainly from two routes. One is from the
consumption of foods containing benzoic acid. The other one is from the metabolism of
polyphenols into benzoic acid by the gut microflora (Walsh, Brennan, Pujos-Guillot,
Sébédio, Scalbert, Fagan, et al., 2007). Procyanidins were degraded by the gut
microflora into benzoic acid in colon and benzoic acid was converted to hippuric acid in
the liver (Rechner, Kuhnle, Bremner, Hubbard, Moore, & Rice-Evans, 2002). A previous
animal study showed that consumption of cranberry powder caused a strong increase in
urinary excretion of hippuric acid. Its quantity in urine were higher than any other urinary
phenolic acids (Prior, Rogers, Khanal, Wilkes, Wu, & Howard, 2010). Citric acid is an
intermediate in the citric acid cycle intermediates. Plasma level of citric acid was
elevated after cranberry juice consumption, suggesting an increased oxidative energy
metabolism. Reduction in plasma level of lactate suggested that cranberry juice
consumption may be associated with anaerobic glycolysis reduction (S. Lin, Chan, Li, &
Cai, 2010).
Box-and-whisker plots of signal intensity of these eight metabolites were used to
display their differences in plasma or urine level following juice consumption (Figure 4-
12). The median intensity of lactate, glucose, unknown 1 (singlet at 2.36 ppm), and
unknown 2 (multiplet at 4.01 ppm) in plasma following cranberry juice consumption
were about two times lower than those after drinking apple juice. It was consistent with
108
Welch’s t test, confirming their significantly low level in plasma after drinking cranberry
juice (Table 4-6). It is interesting to notice that the whiskers on the box plots of hippuric
acid and unknown 4 (singlet 2.11 ppm) following apple juice were considerably smaller
compared to those following cranberry juice, indicating that contents of hippuric acid
and unknown 4 (singlet 2.11 ppm) were consistently low throughout all urine samples
following apple juice consumption. The results were consistent with both univariate and
multivariate analyses that these two metabolites had significantly higher quantities in
subjects’ urine after drinking cranberry juice.
Summary
This study showed that global 1H NMR metabolomics was a very effective
approach to differentiate metabolic impact of cranberry juice from those of apple juice.
The metabolic differences observed in the present study were consistent with our
previous findings in female rats. Cranberry juice consumption caused a higher urinary
excretion of hippuric acid, while apple juice intake increased the plasma concentration
of lactate and D-glucose. Several health benefits were associated with consumption of
cranberry juices; however, the mechanisms remain unclear. The metabolic differences
observed in this study may help to explain the physiological activities of procyanidin-rich
cranberry juices.
109
Table 4-1. Timeline of intervention study on women. Volunteers 18 healthy females college students
Treatment A: Cranberry juice B: Apple juice
1st -6th day and the rest of the study
Avoid procyanidins-rich foods
7th day morning (8-10 am)
Collect first-morning baseline urine samples Collect baseline blood samples Consume 1 bottle of cranberry or apple juice
7th day evening
Consume 1 bottle of cranberry or apple juice
8th-9th day Consume 1 bottle of cranberry or apple juice in the morning and evening
10th day morning (8-10 am)
Collect first-morning urine samples Consume 1 bottle of cranberry or apple juice Collect blood samples
Wash out period for 2 weeks
25th day morning (8-10 am)
Collect first-morning baseline urine samples Collect baseline blood samples. Consume 1 bottle of cranberry or apple juice
25th day evening
Consume 1 bottle of cranberry or apple juice
26th-27th day Consume 1 bottle of cranberry or apple juice in the morning and evening
28th day morning (8-10 am)
Collect first-morning urine samples Consume 1 bottle of cranberry or apple juice Collect blood samples.
End
110
Table 4-2. Total phenolics, total anthocyanins, procyanidin composition and content of cranberry juice and apple juice.
Cranberry Juice Apple Juice
Procyanidins content (µg/mL juice)
Monomer 6.39±0.19 Not Detected
Dimers 53.8±0.1 0.225±0.170
Trimers 49.2±0.7 0.445±0.012
Tetramers 58.5±0.5 1.26±0.07
Pentamers 34.4±1.8 1.28±0.01
High Polymers 364±14 6.46±0.41
Total 566±17 9.68±0.52
Total phenolics (µg gallic acid equivalents/mL juice)*
Total phenolics 913±7 124±1
Total anthocyanins (µg cyanidin 3,5-diglucoside equivalents/mL juice)
Total anthocyanins 59.2±2.4 0.12±0.00
Sugar Composition and Content (mg/mL juice)
Fructose 3.46±0.12 157±8
Glucose 22.3±0.3 76.1±2.3
Sucrose Not Detected 41.6±1.5
Total 25.8±0.4 275±12
Data are expressed as mean ± standard deviation. *Ascorbic acid was not counted as total phenolics
111
Table 4-3. Summary of parameters for PCA, PLS-DA, and OPLS-DA models for human baseline plasma and plasma after drinking cranberry juice or apple juice.
Baseline vs. Cranberry Juice Baseline vs. Apple Juice
PCA PLS-DA OPLS-DA PCA PLS-DA OPLS-DA
Na
4 2 1Pc+1Od 4 2 1Pc+1Od
R2
X(cum)b
0.553 0.278 0.278 0.751 0.274 0.274
R2
Y(cum)b
--- 0.598 0.598 --- 0.656 0.656
Q2
(cum)b
0.346 -0.207 -0.478 0.633 0.105 0.403
a N: number of components. b R2X (cum)and R2Y (cum) are the cumulative modeled variations in the X and Y matrix, respectively. Q2Y (cum) is the cumulative predicted variation in the Y matrix. c Predictive component. d Orthogonal component.
112
Table 4-4. Summary of parameters for PCA, PLS-DA, and OPLS-DA models for human plasma after drinking cranberry juice or apple juice.
Model Na
R2
X(cum)b
R2
Y(cum)b
Q2
(cum)b
Correct classification Rate*
PCA 3 0.683 ---- 0.521 ----
PLS-DA 3 0.571 0.716 0.414 0.803±0.098
OPLS-DA 1Pc
+7Od
0.856 0.979 0.652 0.803±0.091
a N: number of components. b R2X (cum)and R2Y (cum) are the cumulative modeled variations in the X and Y matrix, respectively. Q2Y (cum) is the cumulative predicted variation in the Y matrix. c Predictive component. d Orthogonal component. *Correct classification rate was obtained from external validation procedure repeated for 30 times.
113
Table 4-5. Summary of parameters for PCA, PLS-DA, and OPLS-DA models for human urine after drinking cranberry juice or apple juice.
Model Na
R2
X(cum)b
R2
Y(cum)b
Q2
(cum)b
Correct classification Rate*
PCA 2 0.505 ---- 0.414 ----
PLS-DA 3 0.548 0.853 0.547 0.802±0.108
OPLS-DA 1Pc
+2Od
0.548 0.853 0.503 0.802±0.101
a N: number of components. b R2X (cum)and R2Y (cum) are the cumulative modeled variations in the X and Y matrix, respectively. Q2Y (cum) is the cumulative predicted variation in the Y matrix.
c Predictive component. d Orthogonal component. *Correct classification rate was obtained from external validation procedure repeated for 30 times.
114
Table 4-6. Summary of metabolite profile changes in plasma and urine of young women after drinking cranberry juice and apple juice.
Metabolites Chemical shift (multiplicity) p-value a Cranberry juice vs. apple juice b
Plasma lactate 1.32 (d), 4.11 (q) <0.01
D-glucose 3.22 (dd), 3.40 (q), 3.46 (m), 3.70 (m), 3.83 (m), 3.90 (m), 5.23 (d)
<0.01
citric acid 2.52 (d), 2.62 (d) <0.01
unknown 1 2.36 (s) <0.01
unknown 2 4.01 (m) <0.01
unknown 3 3.56 (m) <0.01
Urine hippuric acid 3.96 (d), 7.54 (t),7.63 (t), 7.82 (d), 8.53 (br,s)
<0.01
unknown 4 2.11 (s) <0.01
a p-value obtained from Welch’s t test. Benjamini–Hochberg procedure was conducted to control false discoveries and conclude that all these variables are significant different at α=0.01.
b Arrows indicated a decrease or increase in metabolite level in plasma or urine after cranberry juice consumption compared to apple juice.
115
Figure 4-1. Chromatograms of procyanidins extracted from cranberry juice and apple
juice using fluorescence detection. A) Cranberry juice and B) apple juice. Identification was performed using HPLC-FLD-MSn .The numbers beside the peaks indicate the degree of polymerization of B-type procyanidins. 2a-4a designates the peaks of procyanidins dimers through pentamers with one A-type linkage. O3-O5 designates the peaks of oxidized B-type procyanidins trimer, tetramers, and pentamers found in apple juice.
min0 10 20 30 40 50 60
LU
10
20
30
40
50
60
1
2a
2b3a
3a3 4a
4a4a 5a
3
High polymer
A
min0 10 20 30 40 50 60
LU
5
10
15
20
High polymer
2
O3
O3
O3O4
O4 O4
O5B
116
Figure 4-2. Chromatograms of sugar standards and juices using refractive index
detector. A) Sugar standards, B) apple juice and C) cranberry juice.
min0 2 4 6 8 10 12 14
nRIU
0
50000
100000
150000
A
FructoseGlucose Sucrose
min0 2 4 6 8 10 12 14
nRIU
0
20000
40000
60000
B Fructose
GlucoseSucrose
min0 2 4 6 8 10 12 14
nRIU
0
20000
40000
60000C
Fructose Glucose
117
Figure 4-3. The PCA score plot of human plasma and plasma quality control from 1H
NMR metabolomics. Green squares: plasma after drinking cranberry juice. Blue squares: plasma after drinking apple juice. Red squares: 17 replicates of pooled plasma samples.
-8
-6
-4
-2
0
2
4
6
-15 -10 -5 0 5 10
t[1]
Plasma after cranberry juice
Plasma after apple juice
17 replicates of pooled plasma
t [2]
118
Figure 4-4. The PCA and OPLS-DA score plots of human plasma after drinking
cranberry juice or apple juice from 1H NMR metabolomics. A) PCA score plot and B) OPLS-DA score plot. Green squares: human plasma after cranberry juice. Blue squares: human plasma after apple juice.
-6
-4
-2
0
2
4
-10 -8 -6 -4 -2 0 2 4 6 8
t[1]
t [2]
A
-8
-6
-4
-2
0
2
4
6
-3 -2 -1 0 1 2 3
t[1]
to [1
]
Plasma after cranberry juicePlasma after apple juice
B
119
Figure 4-5. Model score plot and cross-validated score plot of OPLS-DA model for
human plasma after drinking cranberry juice or apple juice from 1H NMR metabolomics. Circles: model scores of plasma. Squares: cross-validated scores of plasma. Green color: plasma after cranberry juice. Blue color: plasma after apple juice.
-2.5
-2
-1.5
-1
-0.5
0
0.5
1
1.5
B15
'
B5
'
B17'
D4
'
D17
'
D3'
B13
'
D9
'
D11
'
B16'
D7
'
B11
'
D5'
D12
'
D16
'
D8'
D2
'
B12
'
B4
'
B8'
D1
'
D13
'
B14'
B10
'
D10
'
B9'
B2
'
B6
'
D14
'
B1'
B7
'
D15
'
B3'
D6
'
Sample ID
Cross-validated scores of plasma after cranberry juice
Cross-validated scores of plasma after apple juice
Model scores of plasma after cranberry juice
Model scores of plasma after apple juice
t[1]
, tcv
[1]
120
Figure 4-6. Validation plot of 200 permutation tests for OPLS-DA model built for human plasma after drinking cranberry juice or apple juice from 1H NMR metabolomics.
r(y, permuted y)
R2
, Q2
121
Figure 4-7. The PCA and OPLS-DA score plot of human urine after drinking cranberry
juice or apple juice from 1H NMR metabolomics. A) PCA score plot and B) OPLS-DA score plot. Green squares: human urine after cranberry juice. Blue squares: human urine after apple juice.
t [2]
Urine after cranberry juiceUrine after apple juice
-0.06
-0.04
-0.02
0
0.02
0.04
-0.08 -0.06 -0.04 -0.02 0 0.02 0.04 0.06t[1]
A
-0.06
-0.04
-0.02
0
0.02
0.04
-0.05 -0.04 -0.03 -0.02 -0.01 0 0.01 0.02 0.03 0.04t[1]
t o [
1]
B
122
Figure 4-8. Cross-validated score plot of OPLS-DA model derived from human urine
after drinking cranberry juice or apple juice from 1H NMR metabolomics. Green squares: urine after cranberry juice. Blue squares: urine after apple juice.
Urine after cranberry juice
Urine after apple juice
tocv
[1]
-0.05
-0.04
-0.03
-0.02
-0.01
0
0.01
0.02
0.03
-0.03 -0.025 -0.02 -0.015 -0.01 -0.005 0 0.005 0.01 0.015 0.02
tcv[1]
123
Figure 4-9. Validation plot of 200 permutation tests for OPLS-DA model built for human
urine after drinking cranberry juice or apple juice from 1H NMR metabolomics.
r(y, permuted y)
R2
, Q2
124
Figure 4-10. S-line associated with the OPLS score plots of data derived from human plasma after cranberry juice or
apple juice consumption. The x-axis is chemical shift derived from NMR spectra. The y-axis p(ctr)[1] is the centered loading vector of the first principal component. p(ctr)[1] is colored according to the absolute value of the correlation loading p(corr). p(corr)>0.5 is selected as significance level.
125
Figure 4-11. S-line associated with the OPLS score plots of data derived from human urine after cranberry juice or apple
juice consumption. The x-axis is chemical shift derived from NMR spectra. The y-axis p(ctr)[1] is the centered loading vector of the first principal component. p(ctr)[1] is colored according to the absolute value of the correlation loading p(corr). p(corr)>0.5 is selected as significance level.
126
Figure 4-12. Box-and-whisker plot of the NMR signal intensities of eight significant
metabolites detected in human plasma or human urine of young women after drinking cranberry juice and apple juice.
127
CHAPTER 5 UHPLC-Q-ORBITRAP-HRMS-BASED GLOBAL METABOLOMICS REVEAL METABOLOME MODIFICATIONS IN PLASMA OF YOUNG WOMEN AFTER
CRANBERRY JUICE OR APPLE JUICE CONSUMPTION
Background
The objective of this study is to investigate the plasma metabolome modifications
of young women after drinking cranberry juice or apple juice and to identify putative
biomarkers using an UHPLC-Q-Orbitrap-HRMS-based metabolomics profiling method.
Materials and Methods
Chemicals and Materials
Cranberry juice cocktail (double strength, 54% juice) and 100% apple juice were
provided by Ocean Spray Cranberries, Inc. (Lakeville-Middleboro, MA, USA). LC-MS
grade acetonitrile, methylene chloride, methanol, acetic acid, formic acid and acetone
were purchased from Fischer Scientific Co.(Pittsburgh, PA, USA). Creatine-D3, L-
leucine-D10, L-tryptophan-2, 3, 3-D3, caffeine-D3 were from CDN Isotopes Inc. (Pointe-
Claire, Quebec, Canada). Pooled plasma from American Red Cross were collected over
a period of about 2 weeks.
Subjects and Study Design
Human study was approved by Institutional Review Boards at University of
Florida. Detailed protocol of the human study was described in Chapter 4. The timeline
was summarized in Table 4-1.
UHPLC-Q-Orbitrap-HRMS Analyses
Frozen plasma samples (-80 oC) were thawed at room temperature. One plasma
(100 µL) was mixed with 800 µL acetonitrile: acetone: methanol (8:1:1, v: v: v) to
precipitate the proteins. Twenty µL isotopically-labeled standard solution (40 µg/mL L-
128
tryptophan-D3, 4 µg/mL L-leucine-D10, 4 µg/mL creatine-D3, and 4 µg/mL caffeine-D3)
was added to the above extraction mixture as internal standards. The sample was
vortexed and placed in a 4 oC refrigerator for 30 min to assist protein precipitation. This
sample was then centrifuged at 20,000 g for 10 min at <10 oC to pellet the protein. Two
hundred and fifty µL of supernatant was transferred to a new 1 mL Eppendorf tube and
dried under a gentle stream of Nitrogen (Organomation Associates, Inc., Berlin, MA,
USA). Dried sample was reconstituted in 100 µL 0.1% formic acid in water and
vortexted. The sample solution was put on an ice bath for 10-15 min and centrifuged at
20,000 g for 5 min at <10 oC to remove debris. The reconstituted sample was
transferred into a glass vial with fused glass inserts for analyses. All 34 human plasma
samples were prepared in the same manner. Four groups of quality control (QC)
samples including pooled plasma from baseline group, cranberry juice group, apple
juice group and Red Cross group were prepared and analyzed with experimental
plasma samples to monitor the stability and validity of instrumental acquisition. Running
sequence started with 3 blanks (0.1% formic acid in water), one Red Cross QC, one
pooled QC from baseline group, one pooled QC from cranberry juice group and one
pooled QC from apple juice group, followed by 17 plasma samples to ensure instrument
drift was minimal.
Chromatographic separation was performed on a Thermo Scientific-Dionex
Ultimate 3000 UHPLC using an ACE Excel 2 C18-PFP column, 100 mm x 2.1 mm i.d., 2
µm (Advanced Chromatography Technologies, Aberdeen, UK). The mobile phase
consisted of (A) water with 0.1% formic acid and (B) acetonitrile. The gradient was as
follows: 0−3 min, 100% A isocratic; 3−13 min, 0−80% B linear; 13−16 min, 80% B
129
isocratic; 16−16.5 min, 80-0% B linear; followed by 3 min of re-equilibration of the
column before the next run. The flow rate was 350 μL/min. The UHPLC system was
coupled to a Q Exactive™ Hybrid Quadrupole-Orbitrap High Resolution Mass
Spectrometer (Thermo Fisher Scientific, San Jose, CA, USA). The MS acquisition was
performed using both negative and positive ionization with a mass resolution of 70,000
at m/z 200. Separate injections were performed in a data-dependent (top 5) MS/MS
mode with full scan mass resolution reduced to 35,000 at m/z 200. The injection volume
was 4 μL for negative ionization and 2 μL for positive ionization acquisition. The m/z
range for all full scan analyses was 70–1000. Heated electrospray ionization (HESI)
parameters were as follows: sheath gas flow 45 arb (arbitrary units) auxiliary gas flow
10 arb, sweep gas flow 1 arb, spray voltage 3.5 kV, probe temperature 350°C, capillary
temperature 320 °C for negative ionization and 325 °C for positive ionization. In source
CID (Collision-Induced Dissociation) was 2 eV. The mass spectrometer was calibrated
using Pierce™ negative and positive ion calibration solution (Thermo Fisher Scientific,
San Jose CA, USA). To avoid possible bias, the sequence of injections for plasma
samples was randomized.
Multivariate Data Processing and Statistical Analyses
LC-HRMS data were converted to mzXML using MSConvert from ProteoWizard
(Chambers, et al., 2012) and then processed using MZmine 2.12 (Pluskal, Castillo,
Villar-Briones, & Orešič, 2010). Peaks in each sample were extracted, deconvoluted,
and deisotoped. Alignment using join aligner algorithm was conducted with a 10 ppm
tolerance for m/z values and 0.2 min tolerance for retention time. Gap filling using peak
finder algorithm was performed to fill in missing peaks. The resultant data set was
imported into SIMCA (Version 14.0, Umetrics, Umea, Sweden) for multivariate statistical
130
analysis. Data acquired using negative ionization were mean-centered, Pareto scaled
and log-transformed before multivariate statistical analyses. Data obtained using
positive ionization were mean-centered, Pareto scaled and log-transformed before
building PCA model; mean-centered and log-transformed before PLS-DA and OPLS-DA
analyses. Unsupervised PCA model was performed to initially examine intrinsic
variation in the data set. Then supervised pattern recognition methods include PLS-DA
and OPLS-DA (Bylesjö, Rantalainen, Cloarec, Nicholson, Holmes, & Trygg, 2006) were
used to extract maximum information on discriminant compounds from the data.
Validation of the model was tested using 7-fold internal cross-validation and permutation
tests for 200 times. To further evaluate the predictive ability of the PLS-DA or OPLS-DA
models, an external validation procedure was performed (Brindle, et al., 2002; Llorach,
et al., 2010). The LC-HRMS metabolomics data set was split into a training set and a
test set. Approximately 70% of the samples were randomly selected as the training set
and the remaining 30% were treated as the test set. PLS-DA and OPLS-DA models
were built based on the training set and then blindly predicted the classes of the
samples in the test set. This procedure was repeated 30 times and a correct
classification rate was calculated.
Results and Discussion
Quality Control of Multivariate Analyses
Due to the variations between LC-HRMS injections and artifacts caused by the
order of acquisition and carry-over, sensitivity changes or ion suppression could occur
during the experimental period (Burton, Ivosev, Tate, Impey, Wingate, & Bonner, 2008).
Sample acquisition was randomized. QCs created from baseline group, cranberry juice
group, apple juice group, and Red Cross group were analyzed along with experimental
131
plasma samples to monitor the instrument performance. The PCA model was built to
investigate the metabolome difference between QCs and experimental samples. The
mechanism was based on the ability of the PCA model to cluster samples in an
unsupervised approach. QCs from Red Cross plasma clustered together and separated
from experimental plasma on the PCA score plots (Figure 5-1A, 5-1B). It confirmed the
stability of instrumental analysis. PCA score plots based on experimental samples and
QCs from cranberry group, apple group and baseline group (Figure 5-1C, 5-1D) showed
that replicates of pooled QCs from each group tended to cluster together across the
entire sequence indicating a good quality of data acquisition.
Baseline Plasma vs. Plasma after Drinking Cranberry Juice
Two sets of data acquired by LC-HRMS negative ionization and positive
ionization were subjected to multivariate analyses, respectively. No segregation
between baseline plasma and plasma after cranberry juice was observed on the PCA
score plots for either negative ionization or positive ionization acquisition (Figure 5-2A,
5-2B). Compared to unsupervised PCA model, supervised multivariate statistic
techniques including PLS-DA and OPLS-DA successfully segregated two groups of
plasma samples in both negative and positive ionization data acquisition. Score plots of
PLS-DA (Figure 5-3A, 5-3C) and OPLS-DA (Figure 5-3B, 5-3D) demonstrated a clear
separation of baseline plasma vs. plasma after drinking cranberry juice. Two principal
components were selected to build PLS-DA model. A principal component and an
orthogonal component were used to construct OPLS-DA model. R2 represents the
goodness of fit. The R2Y of supervised models based on negative ionization and
positive ionization was 0.901 and 0.951, respectively (Table 5-1). The results indicated
that above 90% of variance in Y data matrix was explained by both PLS-DA and OPLS-
132
DA models. The advantage of OPLS-DA over PLS-DA is that the “structure noise” of
data matrix which is unrelated to the variation of interest is filtered and described only
by the orthogonal component. The variation of scientific interest is described in the
predictive component. Therefore the interpretability of the resulting model is increased
(Fonville, et al., 2010). In this study the performance of PLS-DA was similar to OPLS-
DA, indicating little “structure noise” existed in the LC-HRMS data. Overfitting is
possible when analyzing high-dimensional data with thousands of variables. This is due
to accidental correlations between one or more variables. Therefore, validation of
supervised model was applied to detect overfitting. Three validation methods were used
to test the validity and predictability of PLS-DA and OPLS-DA models. Internal cross-
validation was the first step to test the predictability of the supervised models. Q2
calculated from the cross-validation higher than 0.5 indicates a good multivariate model.
Q2 higher than 0.9 suggests excellent metabolomics data. Seven-fold internal cross
validation was performed on both PLS-DA and OPLS-DA models. Q2 obtained from
PLS-DA and OPLS-DA model derived from negative ionization mode was 0.627 and
0.641, respectively. Q2 of 0.764 and 0.679 was calculated from PLS-DA and OPLS-DA
model derived from positive ionization mode, respectively (Table 5-1). These Q2 values
indicated that supervised models derived from both negative and positive ionization had
a good predictability and the segregation was not due to overfitting. Cross-validated
PLS-DA and OPLS-DA score plots (Figure 5-4) also showed a separation between two
groups of plasma. Three or four plasma samples were misclassified during cross
validation. Misclassification would not occur if a supervised model has an excellent
predictability with Q2 higher than 0.9. To further test the predictability of PLS-DA and
133
OPLS-DA models, permutation tests were conducted. The class labels of baseline and
cranberry juice drinking group were permuted and randomly assigned to different
observations. Then a new supervised model was calculated with the permutated class
labels. The procedure was repeated 200 times. R2 and Q2 of each newly constructed
model were calculated and a regression line was drawn. In an ideal case, all R2 and Q2
calculated from the permutation data should be lower than those from the actual data,
and the Q2-intercept value obtained from the regression line should be lower than 0.05.
The rationale behind the permutation test is that the newly constructed classification
models that are built based on permutated class labels should not be able to correctly
predict the class (Westerhuis, et al., 2008). Although the R2 calculated from permeated
models were higher than 0.7, the corresponding Q2 were smaller than 0.4 (Figure 5-5,
Figure 5-6), suggesting that the classification models based on permuted class labels
had poor predictability compared to actual model. Therefore, the achieved segregation
between baseline plasma and plasma after drinking cranberry juice was not likely due to
overfitting. Cross-validation and permutation test provided a reasonable estimation of
the predictability of a PLS or OPLS model (Eriksson, 2006). External validation uses an
independent set of test data to evaluate predictability of a supervised model and
therefore is a more scrupulous and demanding method (Eriksson, 2006). Correct
classification rate of 85% and 90% for supervised models derived from negative and
positive ionization analyses were obtained from external validation procedure (Table 5-
1). These results indicated that both PLS-DA and OPLS-DA models had excellent
predictabilities and were able to correctly predict unknown plasma samples with an error
rate smaller than 15%. These three validation tests confirmed that there were true
134
changes of plasma metabolome in young women after drinking cranberry juice. UHPLC-
HRMS metabolomics was an effective approach to reveal the metabolome modification.
Plasma after Drinking Apple Juice vs. Plasma after Drinking Cranberry Juice
Similarly, projection pattern techniques were applied to investigate the
metabolome differences in human plasma after drinking apple juice or cranberry juice.
Compared to unsupervised PCA model which did not segregate the two groups of
plasma (Figure 5-7), PLS-DA and OPLS-DA were able to separate human plasma after
drinking apple juice from those after cranberry juice consumption (Figure 5-8). Both
PLS-DA and OPLS-DA models derived from negative ionization and positive ionization
had high quality parameters. The R2Y of four supervised models were higher than
0.950, suggesting that more than 95% of the variance of Y data matrix was explained by
these models. Q2 calculated from 7-fold cross-validation for PLS-DA and OPLS-DA
models derived from negative ionization were 0.846 and 0.809, respectively, (Table 5-1)
indicating a very good predictability of supervised models. Q2 were higher than those
obtained from positive ionization (Table 5-1), suggesting that a better performance of
supervised models was achieved based on LC-HRMS data acquired using negative
ionization. The cross-validated score plots (Figure 5-9) showed a clear segregation of
two groups of plasma. No misclassification occurred during the cross validation.
Compared to supervised models derived from baseline plasma vs. plasma after
cranberry juice, the same models derived from plasma after apple juice vs. plasma after
cranberry juice had higher Q2 and a zero misclassification. It suggested that plasma
metabolome differences between apple juice and cranberry juice consumption were
more robust than those between baseline and cranberry juice consumption.
135
Permutation test and external validation test were conducted to further
investigate the validity and predictability of supervised models. Figure 5-10 and Figure
5-11 showed that the Q2 of each model constructed based on permutated class labels
was smaller than 0.4, suggesting a good predictability. Therefore, the achieved
segregation between plasma after apple juice or cranberry juice was not likely due to
overfitting. External validation test generated correct classification rates of 97.7% and
98% for supervised models derived from negative and positive ionization (Table 5-1),
indicating that both PLS-DA and OPLS-DA models had excellent predictability and were
able to correctly predict the unknown human plasma samples with an error rate smaller
than 3%. These results confirmed that the plasma metabolome of young women after
drinking cranberry juice were different than those after drinking apple juice.
Discriminant Metabolites Identification
S-plot is a statistical tool that visualizes the variable influence in a projection-
based model. It was used to discover the discriminant metabolites. At a significance
level p= 0.05, a p(corr) of 0.5 was used as an arbitrary cutoff value to select the
potential biomarkers (Llorach, Urpi-Sarda, Jauregui, Monagas, & Andres-Lacueva,
2009). Metabolites with higher absolute p[1] and p(corr) values were located on the
upper right or lower left corner of the S-plot. They were the statistically significant
variables contributing to the separation between apple and cranberry juice consumption.
Figure 5-12A and 5-13A show that a total of 57 and 28 metabolic features in negative
mode and positive mode, respectively, were discriminant metabolites that separated
baseline and cranberry juice consumption. Among them, 14 features in negative mode
and 8 features in positive mode were identified based on their accurate masses and/or
product ion spectra. Similarly, Figure 5-12B and 5-13B demonstrate that 39 and 42
136
metabolic features in negative mode and positive mode, respectively, were discriminant
metabolites that separated apple juice and cranberry juice consumption. A total of 12
metabolic features in both negative mode and positive mode were identified using
accurate masses and/or product ion spectra. These identified metabolites were
numbered in the Figure 5-12, 5-13 and summarized in Table 5-2 and Table 5-3.
Unidentified metabolic features were listed in Table 5-4 and Table 5-5. HMDB (Wishart,
et al., 2007), mzCloud, Metlin and Mass Bank (Horai, Arita, Kanaya, Nihei, Ikeda, Suwa,
et al., 2010) were searched to assist metabolite identification.
Seven exogenous metabolites were higher in human plasma after drinking
cranberry juice compared to baseline plasma. One metabolite producing a [M-H]- ion at
m/z 188.9854 and a product ion at m/z 109.0294 [M-H-sulfate]-. It was tentatively
identified as catechol sulfate as it matched the same metabolite in HMDB (Δ=0.0009
Da). This metabolite was also higher in rat plasma after intake of cranberry procyanidins
in Chapter 2. The compound producing a [M-H]- ion at m/z 151.0389 was tentatively
identified as hydroxyphenyl acetic acid (Δ=0.0012Da). It generated product ions at m/z
107.0499 [M-H-COO]-, m/z 64.8080, and m/z 59.0130. The fragmentation pattern
matched hydroxyphenylacetic acid in mzCloud. However, the position of hydroxyl group
(3- or 4- ) could not be determined based on product ion spectra. A metabolite
producing a [M-H]- ion at m/z 242.9967 was putatively identified as coumaric acid
sulfate (Δ=0.0002 Da). The product ion at m/z 163.0400 [M-H-sulfate]- was observed in
its MS2 spectra. Other product ions including m/z 146.9611, m/z 119.0503 and m/z
174.9560 matched the fragmentation pattern of coumaric acid (Liang, Xu, Zhang,
Huang, Zang, Zhao, et al., 2013). The compound with [M-H]- ion at m/z 273.0077 and a
137
major product ion at m/z 193.0504 [M-H-sulfate]- was tentatively identified as ferulic acid
sulfate (Δ=0.0003 Da). Characteristic fragment ions at m/z 178.0270, m/z 149.0606 and
m/z 134.0372 were in high accordance with ferulic acid (Liang, et al., 2013). 3, 4-
Dihydroxyphenyl ethanol sulfate (Δ=0.0004 Da) was detected and tentatively identified
in human plasma after drinking cranberry juice. It had a [M-H]- ion at m/z 233.0121 and
a product ion at m/z 153.0149 [M-H-sulfate]-. Two isomers of trihydroxybenzoic acid
were detected and tentatively identified based on their accurate m/z 171.0264 [M+H]+
and m/z 171.0265 [M+H]+. The product ions of one isomer included m/z 152.0704, m/z
148.9768, m/z 125.9610, m/z 88.0713 and m/z 84.9602. The fragment ions of the other
isomer were m/z 148.9768, m/z 125.9608, m/z 84.0812, m/z 89.0713 and m/z
109.0761. The fragmentation patterns and accurate masses of these two isomers
matched those of trihydroxybenzoic acid in mzCloud (Δ=0.0024 Da).
Furthermore, 11 endogenous metabolites were higher in human plasma after
drinking cranberry juice compared to baseline plasma. The metabolite producing a [M-
H]- ion at m/z 178.0501 and a fragment ion at m/z 134.0612 was tentatively identified as
hippuric acid (Δ=0.0008 Da) by comparing its spectra with those in mzCloud. This
metabolite also generated a [M+H]+ ion at m/z 180.0656 and a fragment ion at m/z
105.0337 in positive mode, which matched the same compound in HMDB and mzCloud.
2-Hydroxyhippuric acid (Δ=0.0008 Da) was detected and tentatively identified based on
the [M-H]- ion at m/z 194.0451 and product ions at m/z 150.0506, m/z 194.0456, and
m/z 93.0342. The fragmentation pattern matched those in mzCloud and HMDB. The
compound producing a [M+H]+ ion at m/z 196.0395 was tentatively identified as
hydroxyhippuric acid (Δ=0.0209 Da) by searching HMDB. However, the position of
138
hydroxyl group could not be determined for this compound because no MS2 spectra
was collected from the positive ionization analysis. The metabolite generating a [M-H]-
ion at m/z 168.0292 and a [M+H]+ ion at m/z 170.0449 was tentatively identified as 2-
furoyglycine. The product ions at m/z 124.0404 and m/z 67.0183 in negative mode
matched those in mzCloud (Δ=0.0010 Da). The MS2 spectra in positive mode showed
ions at m/z 95.0131, m/z 123.9658, m/z 146.9814, m/z 88.0398 matching the
fragmentation pattern of 2-furoyglycine in Metlin database (Δ=0.0001 Da).
Vanilloylglycine was detected and tentatively identified based on its accurate m/z
224.0561[M-H]-. Searching m/z 224.0561[M-H]- in HMDB yielded only vanilloylglycine
with Δ<0.001 Da. Hippuric acid, hydroxyhippuric acid, 2-furoyglycine, vanilloylglycine
belong to acyl glycine and they are formed by conjugation of benzoic acid,
hydroxybenzoic acid, vanillic acid or furan derivatives with glycine. These phenolic acids
were likely generated from procyanidin catabolism in colon by the gut microflora (Ou &
Gu, 2014). A previous animal study showed that consumption of cranberry powder
caused a strong increase in urinary excretion of hippuric acid. Its quantity were higher
than any other phenolic acids (Prior, Rogers, Khanal, Wilkes, Wu, & Howard, 2010). A
compound produced a [M-H]- ion at m/z 191.0554 and a [M+H]+ ion at m/z 193.0708. It
was identified as quinic acid (Δ=0.0007 Da, Δ=0.0001 Da) because its accurate mass
and MS2 spectra matched those of quinic acid standard. Retention time and mass
spectra of quinic acid were curated in the in-house database at SECIM. The metabolite
producing a [M-H]- ion at m/z 147.0287 was tentatively identified as citramalic acid by
comparing its fragment ions at m/z 129.0195, m/z 102.9458, m/z 873.9248 and m/z
58.9582 with those in HMDB (Δ=0.0012 Da). Citramalic acid is an analog of malic acid
139
and an inhibitor of malic acid production. The compound generating a [M-H]- ion at m/z
173.0092 and product ions at m/z 128.8782, m/z 99.9255, m/z 116.9285, m/z 118.9659
and m/z 85.0290 was tentatively identified as aconitic acid by searching HMDB and
mzCloud (Δ=0.0009 Da). However, the trans- or cis-form of aconitic acid could not be
determined based on only MS2 spectra. Cis- and trans- aconitic acids belong to
tricarboxylic acids. Cis-aconitic acid is an intermediate produced by the dehydration of
citric acid in citrate cycle. The compound producing a [M-H]- ion at m/z 180.0657 was
tentatively identified as tyrosine. Its product ions were m/z 163.0100, m/z 135.0452, and
m/z 119.0502, matching those in HMDB and mzCloud (Δ=0.0039 Da).
Hydroxyoctadecanoic acid, a hydroxyl fatty acid, was putatively identified based on its
accurate [M-H]- ion at m/z 299.2592 (Δ=0.00003 Da). Glycerol 3-phosphate generating
an accurate m/z 173.0236 [M+H]+ was tentatively identified according to HMDB
(Δ=0.0026 Da). Glycerol 3-phosphate is a metabolite in both glycerolipid and
glycerophospholipid metabolism pathway. The compound producing a [M+H]+ ion at m/z
162.0551 was tentatively assigned as dihydroxyquinoline. The positions of two hydroxyl
groups could not be determined. Among various isomers of dihydroxyquinoline, 4, 6-
and 4, 8-dihydroxyquinoline are products after conversion of 5-hydroxykynurenamine or
3-hydroxykynurenamine by monoamine oxidase. They are catabolites of tryptophan
through the kynurenine metabolic pathway. This pathway is employed by immune
system to modulate the balance between responsiveness to pathogens and tolerance to
non-harmful antigens (Moffett & Namboodiri, 2003). It was proposed that tryptophan
catabolism facilitates immune tolerance by suppressing T cell proliferation due to a
reduction of this critical amino acid. Another theory proposed was that catabolites of
140
tryptophan can suppress certain immune cells (Moffett & Namboodiri, 2003). Indole-3-
acetaldehyde is also an intermediate in tryptophan metabolism and its level increased
after drinking cranberry juice compared to apple juice consumption. Decarboxylation of
tryptophan generates tryptamine which is further catabolized by monoamine oxidase to
indole-3-acetaldehyde.
Most of these metabolites higher in human plasma following cranberry juice
compared to baseline also increased compared to apple juice consumption (Table 5-2,
5-3). A few exceptions were aconitic acid, tyrosine, and hydroxyphenyl acetic acid. 2-
Furoyglycine was found to be decreased in human plasma following drinking cranberry
juice compared to apple juice. Moreover, metabolites including vanilloloside, 5-
(trihydroxyphenyl)-ϒ-valerolactone, 3-(hydroxyphenyl) propionic acid, 4-acetamido-2-
aminobutanoic acid and indole-3-acetaldehyde increased following cranberry juice
compared to apple juice. Vanilloloside is a phenolic glucoside and was tentatively
assigned based on its accurate m/z 315.1088 [M-H]- matching HMDB (Δ= 0.0003 Da). A
compound generating a [M+H]+ ion at m/z 225.0733 was tentatively identified as 5-
(trihydroxyphenyl)-ϒ-valerolactone as it matched the same metabolite in HMBD (Δ=
0.0024 Da). 3-(Hydroxyphenyl) propionic acid was tentatively assigned because its
accurate m/z 167.0705 [M+H]+ and fragment ion at m/z 120.0808 matched those in
HMDB and mzCloud (Δ=0.0023 Da). Vanilloloside, 5-(trihydroxyphenyl)-ϒ-valerolactone
and 3-(hydroxyphenyl) propionic acid were exogenous metabolites and derived from
microbial degradation of procyanidins. 4-Acetamido-2-aminobutanoic acid belongs to
the family of alpha amino acids and derivatives. It was tentatively assigned according to
its accurate m/z 161.0921 [M+H]+ (Δ= 0.0004 Da).
141
More than two thirds of discriminant metabolites identified in the present study
were endogenous metabolite, with the rest being phenolic acids derived from
catabolism of procyanidins by gut microbiota. These identified endogenous metabolites
corresponded to quinic acids and derivatives, hydroxyl fatty acids, acyl glycine,
glycerophosphates, indoles and derivatives. They are intermediates or products in a
range of metabolic pathways.
Summary
This study showed that metabolite profiles of young women after drinking
cranberry juice were different from those before drinking cranberry juice or after apple
juice consumption. Compared to baseline condition, cranberry juice consumption
caused a greater increase of metabolites including catechol sulfate, 3, 4,-
dihydroxyphenyl ethanol sulfate, hydroxyphenyl acetic acid, coumaric acid sulfate,
ferulic acid sulfate, quinic acid, citramalic acid, aconitic acid, hippuric acid,
hydroxyhippuric acid, 2-furoylgycine, vanilloylglycine, tyrosine, hydroxyoctadecanoic
acid, trihydroxybenzoic acid, dihydroxyquinoline and glycerol 3-phosphate. Moreover,
compared to apple juice consumption, drinking cranberry juice increased the plasma
level of catechol sulfate, 3, 4,-dihydroxyphenyl ethanol sulfate, coumaric acid sulfate,
ferulic acid sulfate, 5-(trihydroxyphenyl)-ϒ-valerolactone, 3-(hydroxyphenyl) propionic
acid, vanilloloside, quinic acid, citramalic acid, hippuric acid, hydroxyhippuric acid,
vanilloylglycine, 4-acetamido-2-aminobutanoic acid, hydroxyoctadecanoic acid,
trihydroxybenzoic acid, glycerol 3-phosphate, dihydroxyquinoline and indole-3-
acetaldehyde. The plasma level of 2-furoylgycine was decreased following cranberry
juice compared to apple juice consumption.
142
Table 5-1. Summary of parameters for PLS-DA and OPLS-DA models for human baseline plasma and plasma after drinking cranberry juice or apple juice.
Negative Ionization Analyses Positive Ionization Analyses
Baseline vs. Cranberry Juice Cranberry Juice vs. Apple Juice
Baseline vs. Cranberry Juice
Cranberry Juice vs. Apple Juice
PLS-DA OPLS-DA PLS-DA OPLS-DA PLS-DA OPLS-DA PLS-DA OPLS-DA
Na
2 1Pc+1Od 2 1Pc+1Od 2 1Pc+1Od 2 1Pc+1Od
R2
X(cum)b
0.235 0.235 0.253 0.253 0.211 0.211 0.232 0.253
R2
Y(cum)b
0.901 0.901 0.951 0.951 0.951 0.951 0.959 0.951
Q2
(cum)b
0.627 0.641 0.846 0.809 0.764 0.679 0.796 0.776
*Correct Classification Rate
0.853±0.094 0.857±0.097 0.977±0.043 0.977±0.043 0.900±0.095 0.900±0.095 0.980±0.048 0.980±0.048
a N: number of components. b R2X (cum)and R2Y (cum) are the cumulative modeled variations in the X and Y matrix, respectively. Q2Y (cum) is the cumulative predicted variation in the Y matrix. c Predictive component. d Orthogonal component. *Correct classification rate was obtained from external validation procedure repeated for 30 times.
143
Table 5-2. Identification of discriminant metabolites in human plasma after drinking cranberry juice or apple juice by negative ionization analysis.
NO. Retention Time (min)
Detected Mass [M-H]-
p[1] (contribution)a
p(corr)[1] (confidence) b
MSMS Putative Identification
Theoretical Mass [M-H]-
Mass Difference (Da)
Reference
CJ vs AJ c
CJ vs BS d
1 0.898 191.0554 0.085 (0.115)
0.781 (0.860)
---- Quinic acid 191.0561 0.0007 HMDB, in-house DB
2 2.805 147.0287 0.089 (0.080)
0.827 (0.848)
129.0195, 102.9485, 87.9248, 58.9582
Citramalic acid 147.0299 0.0012 HMDB
3 2.990 173.0083 0.092 0.718 128.8782, 99.9255, 116.9285, 118.9659
Aconitic acid 173.0092 0.0009 HMDB mzCloud
----
4 6.722 168.0292 -0.080 (0.093)
-0.730 (0.741)
124.0404,67.0183 2-Furoylglycine 168.0302 0.0010 mzCloud
5 7.206 188.9854 0.070 (0.085)
0.754 (0.816)
109.0294 Catechol sulfate
188.9863 0.0009 HMDB
6 7.312 233.0121 0.053 (0.060)
0.567 (0.560)
153.0149 3,4-Dihydroxyphenyl ethanol sulfate
233.0125 0.0004 Liang et al. e
7 7.419 194.0451 0.072 (0.085)
0.767 (0.868)
150.0506, 194.0456, 93.0342
2-Hydroxyhippuric acid
194.0458 0.0008 HMDB mzCloud
8 7.739 224.0561 0.094 (0.099)
0.879 (0.884)
---- Vanilloylglycine 224.0564 0.0035 HMDB
9 7.887 315.1088 0.133 0.892 ---- Vanilloloside 315.1085 0.0003 HMDB ----
10 8.069 178.0501 0.086 (0.100)
0.831 (0.836)
134.0612 Hippuric acid 178.0510 0.0008 HMDB mzCloud
11 8.165 180.0657 0.064 0.525 163.0400, 135.0452, 119.0502
Tyrosine 180.0666 0.0039 HMDB mzCloud
----
12 8.279 242.9967 0.104 (0.118)
0.850 (0.871)
163.0400, 146.9611, 119.0503, 174.9560
Coumaric acid sulfate
242.9969 0.0002 Liang et al. e
144
Table 5-2. Continued. NO. Retention
Time (min)
Detected Mass [M-H]-
p[1] (contribution)a
p(corr)[1] (confidence) b
MSMS Putative Identification
Theoretical Mass [M-H]-
Mass Difference (Da)
Reference
CJ vs AJ c
CJ vs BS d
13 8.313 273.0077 0.104 (0.124)
0.831 (0.907)
193.0504, 178.0270, 149.0606, 134.0372
Ferulic acid sulfate
273.0074 0.0003 Liang et al. e
14 10.660 151.0389 0.052 0.599 107.0499, 64.8080, 59.0130
Hydroxyphenyl acetic acid
151.0401 0.0012 mzCloud ----
15 14.588 299.2592 0.059 0.500 ---- Hydroxyoctadecanoic acid
299.2592 0.00003 HMDB
a Number inside the parentheses is the p[1] value obtained from OPLS-DA model based on cranberry juice vs. baseline. b Number inside the parentheses is the p (corr) [1] value obtained from OPLS-DA model based on cranberry juice vs. baseline c Arrows indicated a decrease or increase in metabolite level in human plasma after drinking cranberry juice compared to apple juice. d Arrows indicated a decrease or increase in metabolite level in human plasma after drinking cranberry juice compared to baseline. e Identification of compounds were referred to publication by Liang et al. (Liang, et al., 2013).
145
Table 5-3. Identification of discriminant metabolites in human plasma after drinking cranberry juice or apple juice by positive ionization analysis.
NO. Retention Time (min)
Detected Mass [M+H]+
p[1] (contribution) a
p(corr)[1] (confidence) b
MSMS Putative Identification Theoretical Mass [M+H]+
Mass Difference (Da)
Reference
CJ vs AJ c
CJ vs BS d
1 0.846 193.0708 -0.108 (0.113)
-0.839 (0.890)
111.0442, 129.0544, 95.0459, 83.0496
Quinic acid 193.0707 0.0001 mzCloud in-house DB
2 2.911 171.0264 -0.254 (0.252)
-0.954 (0.958)
152.0704, 148.9768, 125.9610,89.0713, 84.9602
Trihydroxybenzoic acid
171.0288 0.0024 HMDB mzCloud
3 3.337 171.0265 -0.269 (0.284)
-0.930 (0.958)
148.9768, 125.9608, 84.0812,89.0714,109.0761
Trihydroxybenzoic acid
171.0288 0.0024 HMDB mzCloud
4 3.678 161.0921 -0.085 -0.504 ---- 4-Acetamido-2-aminobutanoic acid
161.0921 0.00004 HMDB ----
5 5.318 225.0733 -0.075 -0.540 ---- 5-(Trihydroxyphenyl)-gamma-valerolactone
225.0757 0.0024 HMDB ----
6 5.325 167.0705 -0.075 -0.531 120.0808 3-(Hydroxyphenyl)propionic acid
167.0723 0.0023 HMDB mzCloud
----
7 6.728 170.0449 0.083 (0.129)
0.661 (0.795)
95.0131, 123.9658, 146.9814, 88.0398
2-Furoyglycine 169.0375 0.0001 HMDB Metlin
8 8.059 180.0656 -0.076 (0.086)
-0.847 (0.853)
105.0337 Hippuric acid 179.0582 0.0008 mzCloud
9 8.066 196.0395 -0.073 (0.092)
-0.838 (0.842)
---- Hydroxyhippuric acid 196.0604 0.0209 HMDB
10 8.067 162.0551 -0.083 (0.114)
-0.832 (0.828)
139.9819, 116.9662
Dihydroxyquinoline 162.055 0.0002 MMDB KEGG
146
Table 5-3. Continued. NO. Retentio
n Time (min)
Detected Mass [M+H]+
p[1] (contribution) a
p(corr)[1] (confidence) b
MSMS Putative Identification
Theoretical Mass [M+H]+
Mass Difference (Da)
Reference
CJ vs AJ c
CJ vs BS d
11 8.067 173.0236 -0.076 (0.108)
-0.811 (0.806)
---- Glycerol 3-phosphate
173.0209 0.0026 HMDB
12 10.664 160.0758 -0.063 -0.531 118.0651, 132.0812, 146.9600
Indole-3-acetaldehyde
160.0757 0.0001 Mass Bank
----
a Number inside the parentheses is the p[1] value obtained from OPLS-DA model based on cranberry juice vs. baseline. b Number inside the parentheses is the p (corr) [1] value obtained from OPLS-DA model based on cranberry juice vs. baseline c Arrows indicated a decrease or increase in metabolite level in human plasma after drinking cranberry juice compared to apple juice. d Arrows indicated a decrease or increase in metabolite level in human plasma after drinking cranberry juice compared to baseline.
147
Table 5-4. Unidentified discriminant metabolic features in human plasma after cranberry juice or apple juice by negative ionization analysis.
Retention Time (min)
Detected Mass [M-H]-
p[1] (contribution) a
p(corr)[1] (confidence) b
CJ vs. AJ c CJ vs. BS d
0.762 183.0866 0.054 0.608 ----
0.869 189.0397 0.064 0.626 ----
0.887 249.0148 0.093 (0.111) 0.644 (0.715)
0.895 377.0860 0.080 0.500 ----
0.898 191.0554 0.115 0.860 ----
0.918 495.0007 0.088 0.524 ----
0.919 493.0039 0.088 0.527 ----
1.751 231.0619 0.051 0.500 ----
1.955 177.9805 0.119 (0.137) 0.783 (0.887)
2.182 159.0652 0.093 (0.099) 0.718 (0.764)
2.280 181.0125 0.052 0.534 ----
2.799 129.0181 0.083 (0.111) 0.766 (0.889)
2.805 147.0287 0.089 (0.080) 0.827 (0.848)
2.977 159.0652 0.098 (0.115) 0.769 (0.782)
3.425 147.0651 0.068 0.599 ----
3.487 129.0181 0.060 0.568 ----
6.058 291.9445 0.188 (0.187) 0.935 (0.936)
6.064 229.9737 0.192 (0.173) 0.925 (0.935)
6.067 161.9855 0.196 (0.217) 0.851 (0.884)
6.088 177.9806 0.190 (0.195) 0.930 (0.927)
6.579 143.0337 0.076 0.782 ----
7.205 400.9614 0.125 (0.132) 0.744 (0.761)
7.205 190.9813 0.070 (0.086) 0.746 (0.812)
148
Table 5-4. Continued. Retention Time (min)
Detected Mass [M-H]-
p[1] (contribution) a
p(corr)[1] (confidence) b
CJ vs. AJ c CJ vs. BS d
7.206 188.9854 0.085 0.816 ----
7.209 272.9386 0.082 (0.099) 0.720 (0.776)
7.209 256.9735 0.078 (0.099) 0.721 (0.784)
7.284 201.0761 0.051 0.513 ----
7.312 233.0121 0.060 0.560 ----
7.405 232.9759 0.110 (0.125) 0.931 (0.853)
7.419 194.0451 0.085 0.868 ----
7.557 264.9849 0.072 (0.077) 0.602 (0.600)
7. 739 224.0561 0.099 0.884 ----
7.887 315.1088 0.136 0.906 ----
8.068 246.0382 0.070 (0.087) 0.807 (0.823)
8.069 179.0470 0.087 (0.103) 0.800 (0.846)
8.069 134.0599 0.086 (0.102) 0.826 (0.841)
8.069 178.0501 0.100 0.837 ----
8.069 308.0092 0.072 (0.088) 0.811 (0.828)
8.069 276.0275 0.083 (0.095) 0.829 (0.841)
8.072 379.0911 0.108 (0.134) 0.805 (0.817)
8.073 263.0287 0.069 0.709 ----
8.195 143.0703 0.052 0.504 ----
8.279 242.9967 0.118 0.871 ----
8.313 273.0077 0.124 0.907 ----
8.426 182.0814 0.070 0.583 ----
149
Table 5-4. Continued. Retention Time (min)
Detected Mass [M-H]-
p[1] (contribution) a
p(corr)[1] (confidence) b
CJ vs. AJ c CJ vs. BS d
9.189 257.0125 0.076 0.644 ----
9.499 266.8968 0.088 0.625 ----
9.501 268.8948 0.090 0.635 ----
a Number inside the parentheses is the p[1] value obtained from OPLS-DA model based on cranberry juice vs. baseline.
b Number inside the parentheses is the p (corr) [1] value obtained from OPLS-DA model based on cranberry juice vs. baseline
c Arrows indicated a decrease or increase in metabolite level in human plasma after drinking cranberry juice compared to apple juice.
d Arrows indicated a decrease or increase in metabolite level in human plasma after drinking cranberry juice compared to baseline.
150
Table 5-5. Unidentified discriminant metabolic features in human plasma after cranberry juice or apple juice by positive ionization analysis.
Retention Time (min)
Detected Mass [M+H]+
p[1] (contribution) a
p(corr)[1] (confidence) b
CJ vs. AJ c CJ vs. BS d
0.792 183.0866 0.054 0.608
0.890 215.0526 -0.139 (0.185) -0.785 (0.884)
0.897 365.1054 0.102 0.539 ----
0.929 211.5523 0.121 0.515 ----
0.931 443.0483 0.118 0.509 ----
1.370 143.0316 0.054 0.645 ----
1.590 220.9743 0.126 0.831 ----
1.620 109.5209 0.109 0.813 ----
1.634 177.0072 0.093 0.665 ----
1.638 205.0021 0.110 0.816 ----
1.785 118.5261 0.080 0.816 ----
4.159 127.0756 -0.054 (0.078) -0.524 (0.675)
5.324 150.5415 -0.062 -0.513 ----
6.046 151.5097 -0.338 (0.302) -0.924 (0.942)
6.389 199.1077 0.054 0.601 ----
6.426 158.1176 -0.114 (0.112) -0.794 (0.846)
6.728 170.0449 0.083 0.661 ----
6.983 127.0756 0.054 0.633 -----
7.192 204.9820 -0.141 (0.138) -0.805 (0.801)
7.352 121.0287 -0.068 (0.077) -0.824 (0.896)
7.497 155.0780 -0.086 -0.535 ----
8.062 118.0655 -0.082 (0.098) -0.823 (0.842)
151
Table 5-5. Continued. Retention Tim (min)
Detected Mass [M+H]+
p[1] (contribution) a
p(corr)[1] (confidence) b
CJ vs. AJ c
CJ vs. BS d
8.066 155.0128 -0.072 (0.094) -0.827 (0.840)
8.067 150.5366 -0.068 (0.089) -0.833 (0.837)
8.067 219.5525 -0.155 (0.170) -0.807 (0.850)
8.067 105.0339 -0.078 (0.100) -0.825 (0.840)
8.067 171.0499 -0.077 (0.094) -0.846 (0.834)
8.067 159.5418 -0.103 (0.120) -0.824 (0.813)
8.067 292.0141 -0.084 (0.111) -0.789 (0.796)
8.826 153.0258 0.082 0.555 ----
12.820 467.2621 0.060 0.488 ----
12.833 445.2799 0.068 0.515 ----
14.409 666.4347 0.078 0.520 ----
15.587 333.1514 -0.069 -0.516 ----
16.066 350.1780 -0.063 -0.499 ----
a Number inside the parentheses is the p[1] value obtained from OPLS-DA model based on cranberry juice vs. baseline.
b Number inside the parentheses is the p (corr) [1] value obtained from OPLS-DA model based on cranberry juice vs. baseline
c Arrows indicated a decrease or increase in metabolite level in human plasma after drinking cranberry juice compared to apple juice.
d Arrows indicated a decrease or increase in metabolite level in human plasma after drinking cranberry juice compared to baseline.
152
Figure 5-1. The PCA score plot of human plasma and quality control samples from LC-HRMS metabolomics. A) PLS-DA
score plot of negative ionization data, B) OPLS-DA score plot of positive ionization data, C) PLS-DA score plot of negative ionization without QC from Red Cross and D) OPLS-DA score plot of positive ionization data without QC from Red Cross.
-30
-20
-10
0
10
20
30
-25 -20 -15 -10 -5 0 5 10 15 20t[1]
t [2
]A
B
-15
-10
-5
0
5
10
-25 -20 -15 -10 -5 0 5 10 15 20t[1]
t [2
]
C
-20
-15
-10
-5
0
5
10
15
-50 -40 -30 -20 -10 0 10 20 30t[1]
t [2
]
-15
-10
-5
0
5
10
-20 -15 -10 -5 0 5 10 15t[1]
t [2
]
D
PoolQC from apple juice group
PoolQC from Baseline group
PoolQC from cranberry juice group
Plasma of baseline, cranberry and apple group
PoolQC from Red Cross plasma
153
Figure 5-2. The PCA score plot of human baseline plasma and human plasma after drinking cranberry juice from LC-
HRMS metabolomics. A) Data were acquired by negative ionization and B) data were acquired by positive ionization. Blue squares: baseline plasma before drinking cranberry juice. Green squares: plasma after drinking cranberry juice.
Baseline plasma before cranberry juice
Plasma after cranberry juice
-25
-20
-15
-10
-5
0
5
10
15
20
-30 -20 -10 0 10 20t[1]
t [2
]
-25
-20
-15
-10
-5
0
5
10
15
20
-30 -20 -10 0 10 20t[1]
t [2
]A B
154
Figure 5-3. The PLS-DA and OPLS-DA score plots of human baseline plasma and human plasma after drinking cranberry
juice from LC-HRMS metabolomics. A) PLS-DA score plot by negative ionization, B) OPLS-DA score plot by negative ionization, C) PLS-DA score plot by positive ionization and D) OPLS-DA score plot by positive ionization. Blue squares: baseline plasma before drinking cranberry juice. Green squares: plasma after drinking cranberry juice.
-25
-20
-15
-10
-5
0
5
10
15
20
-20 -15 -10 -5 0 5 10 15t[1]
-25
-20
-15
-10
-5
0
5
10
15
20
-20 -15 -10 -5 0 5 10 15t[1]
t [2
]
to [
1]
Baseline plasma before cranberry juice
Plasma after cranberry juice
A B
-15
-10
-5
0
5
10
-20 -15 -10 -5 0 5 10 15t[1]
-20
-15
-10
-5
0
5
10
15
-15 -10 -5 0 5 10t[1]
to [
1]
CD
t [2
]
155
Figure 5-4. The PLS-DA and OPLS-DA cross-validated score plots of human baseline plasma and human plasma after
drinking cranberry juice from LC-HRMS metabolomics. A) PLS-DA cross-validated score plot by negative ionization, B) OPLS-DA cross-validated score plot by negative ionization, C) PLS-DA cross-validated score plot by positive ionization and D) OPLS-DA cross-validated score plot by positive ionization. Blue squares: baseline plasma before drinking cranberry juice. Green squares: plasma after drinking cranberry juice.
-10
-5
0
5
10
-10 -5 0 5 10tcv[1]
-15
-10
-5
0
5
10
-10 -8 -6 -4 -2 0 2 4 6 8 10tcv[1]
tcv[
2]
tocv
[1]
A B
Baseline plasma before cranberry juice
Plasma after cranberry juice
C D
-10
-8
-6
-4
-2
0
2
4
6
-10 -8 -6 -4 -2 0 2 4 6 8 10
tcv[1]
tcv[
2]
-12
-10
-8
-6
-4
-2
0
2
4
6
8
-10 -8 -6 -4 -2 0 2 4 6 8tcv[1]
tocv
[1]
156
Figure 5-5. Validation plot obtained from 200 permutation tests for the PLS-DA and
OPLS-DA models of human baseline plasma vs. human plasma after cranberry juice by negative ionization analysis. A) PLS-DA model and B) OPLS-DA model.
R2
, Q2
r(y, permuted y)
R2
, Q2
r(y, permuted y)
A B
157
Figure 5-6. Validation plot obtained from 200 permutation tests for the PLS-DA and
OPLS-DA models of human baseline plasma vs. human plasma after cranberry juice by positive ionization analysis. A) PLS-DA model and B) OPLS-DA model.
R2
, Q2
r(y, permuted y)
R2
, Q2
r(y, permuted y)
A B
158
Figure 5-7. The PCA score plot of human plasma after drinking apple juice or cranberry juice from LC-HRMS
metabolomics. A) Data were acquired by negative ionization and B) data were acquired by positive ionization. Purple squares: plasma after drinking apple juice. Green squares: plasma after drinking cranberry juice.
-20
-15
-10
-5
0
5
10
15
-30 -20 -10 0 10 20t[1]
t [2
]
A
Plasma after apple juice
Plasma after cranberry juice
t [2
]
-25
-20
-15
-10
-5
0
5
10
15
20
-25 -20 -15 -10 -5 0 5 10 15 20t[1]
B
159
Figure 5-8. The PLS-DA and OPLS-DA score plots of human plasma after drinking apple juice or cranberry juice from LC-
HRMS metabolomics. A) PLS-DA score plot by negative ionization, B) OPLS-DA score plot by negative ionization, C) PLS-DA score plot by positive ionization and D) OPLS-DA score plot by positive ionization. Purple squares: plasma after drinking apple juice. Green squares: plasma after drinking cranberry juice.
t [2
]
-25
-20
-15
-10
-5
0
5
10
15
20
-20 -15 -10 -5 0 5 10 15
t[1]
-25
-20
-15
-10
-5
0
5
10
15
20
-20 -15 -10 -5 0 5 10 15t[1]
to [
1]
A B
Plasma after apple juice
Plasma after cranberry juice
-20
-15
-10
-5
0
5
10
15
-15 -10 -5 0 5 10t[1]
t [2
]
-20
-15
-10
-5
0
5
10
15
-15 -10 -5 0 5 10t[1]
to [
1]
C D
to [
1]
160
Figure 5-9. The PLS-DA and OPLS-DA cross validated score plots of human plasma after drinking apple juice or
cranberry juice from LC-HRMS metabolomics. A) PLS-DA cross-validated score plot by negative ionization, B) OPLS-DA cross-validated score plot by negative ionization, C) PLS-DA cross-validated score plot by positive ionization and D) OPLS-DA cross-validated score plot by positive ionization. Purple squares: plasma after drinking apple juice. Green squares: plasma after drinking cranberry juice.
-15
-10
-5
0
5
10
-10 -8 -6 -4 -2 0 2 4 6 8tcv[1]
tcv
[2]
-20
-15
-10
-5
0
5
10
-8 -6 -4 -2 0 2 4 6tcv[1]
tocv
[1]
A B
Plasma after apple juice
Plasma after cranberry juice
C D
-8
-6
-4
-2
0
2
4
6
8
10
-8 -6 -4 -2 0 2 4 6tcv[1]
tcv
[2]
-10
-5
0
5
10
-8 -6 -4 -2 0 2 4 6tcv[1]
tocv
[1]
161
Figure 5-10. Validation plot obtained from 200 permutation tests for the PLS-DA and
OPLS-DA models of human plasma after apple juice vs. plasma after cranberry juice by negative ionization analysis. A) PLS-DA model and B) OPLS-DA model.
R2
, Q2
r(y, permuted y)
R2
, Q2
r(y, permuted y)
A B
162
Figure 5-11. Validation plot obtained from 200 permutation tests for the PLS-DA and
OPLS-DA models of human plasma after apple juice vs. after cranberry juice by positive ionization. A) PLS-DA model and B) OPLS-DA model.
R2
, Q2
r(y, permuted y)
R2
, Q2
r(y, permuted y)
A B
163
Figure 5-12. S-plots associated with the OPLS-DA score plot of data derived from LC-
HRMS of human baseline plasma and plasma after cranberry juice or apple juice by negative ionization. A) Human baseline plasma vs. plasma after cranberry juice and B) human plasma after cranberry juice vs. plasma after apple juice. p[1] is the loading vector of covariance in the first principal component. p(corr)[1] is loading vector of correlation in the first principal component. Variables with |p| ≥ 0.05 and |p(corr)| ≥ 0.5 are considered statistically significant. Significant variables in blue color were identified and numbered according to Table 5-2. Unidentified significant variables in red color were listed in Table 5-4. Non-significant variables were in green color.
1, 122
3
4
7 5
6
8
10
13
14
A
1
2, 10
4
5,7
6, 15
8
912, 13
B
164
Figure 5-13. S-plots associated with the OPLS-DA score plot of data derived from LC-HRMS of human baseline plasma and plasma after cranberry juice or apple juice by positive ionization. A) Human baseline plasma vs. plasma after cranberry juice and B) human plasma after cranberry juice vs. plasma after apple juice. p[1] is the loading vector of covariance in the first principal component. p(corr)[1] is loading vector of correlation in the first principal component. Variables with |p| ≥ 0.05 and |p(corr)| ≥ 0.5 are considered statistically significant. Significant variables in blue color were identified and numbered according to Table 5-3. Unidentified significant variables in red color were listed in Table 5-5. Non-significant variables were in green color.
13 2
7
8, 9
10, 11
A
1
23
4
5, 6 12
B
7
8, 9, 10, 11
165
CHAPTER 6 MODIFICATION OF URINARY METABOLOME IN YOUNG WOMEN AFTER
CRANBERRY JUICE CONSUMPTION WERE REVEALED USING UHPLC-Q-ORBITRAP-HRMS-BASED GLOBAL METABOLOMICS APPROACH
Background
The objective of this study is to investigate the urinary metabolome modifications
and identify putative biomarkers in young women after drinking cranberry juice or apple
juice using an UHPLC-Q-Orbitrap-HRMS-based metabolomics approach.
Materials and Methods
Chemicals and Materials
Cranberry juice cocktail (double strength, 54% juice) and 100% apple juice were
provided by Ocean Spray Cranberries, Inc. (Lakeville-Middleboro, MA, USA). LC-MS
grade acetonitrile, methylene chloride, methanol, acetic acid, formic acid, sodium azide
and acetone were purchased from Fischer Scientific Co.(Pittsburgh, PA, USA).
Creatine-D3, L-leucine-D10, L-tryptophan-2, 3, 3-D3, caffeine-D3 were from CDN
Isotopes Inc. (Pointe-Claire, Quebec, Canada).
Subjects and Study Design
Human study was approved by Institutional Review Boards at University of
Florida. Detailed protocol of the human study was described in Chapter 4. The timeline
was summarized in Table 4-1.
UHPLC-Q-Orbitrap-HRMS Analyses
Frozen urine samples (-80 oC) were thawed at room temperature. One urine (50
µL) was transferred to a clean, labeled microcentrifuge-filter tube. Twenty µL
isotopically-labeled standard solution (40 µg/mL L-tryptophan-D3, 4 µg/mL L-leucine-
D10, 4 µg/mL creatine-D3, and 4 µg/mL caffeine-D3) was added to the above tube as
166
internal standards. The urine sample was then diluted with 400 µL water: acetonitrile
(98:2, v: v) with 0.1% sodium azide. The diluted urine sample was vortexed and
centrifuged at 20,000 g for 10 min <10 oC to pellet debris. After centrifuge, the filter was
discarded and supernatant was transferred into a glass vial with a fused glass insert for
analyses. Three groups of triplicate quality control (QC) samples including pooled urine
from baseline group, pooled urine from cranberry juice group, and pooled urine from
apple juice group were prepared and analyzed concurrently with experimental urine
samples to monitor the stability and validity of instrumental acquisition. In addition, three
neat QC samples were prepared by adding 20 µL of isotopically-labeled standard
solution to three LC glass vials, respectively. Running sequence started with 3 blanks
(0.1% formic acid in water), one neat QC, one pooled QC from baseline group, one
pooled QC from cranberry juice group and one pooled QC from apple juice group,
followed by every 10 urine samples.
Chromatographic separation was performed on a Thermo Scientific-Dionex
Ultimate 3000 UHPLC using an ACE Excel 2 C18-PFP column, 100 mm x 2.1 mm i.d., 2
µm (Advanced Chromatography Technologies, Aberdeen, UK). The mobile phase
consisted of (A) water with 0.1% formic acid and (B) acetonitrile. The gradient was as
follows: 0−3 min, 100% A isocratic; 3−13 min, 0−80% B linear; 13−16 min, 80% B
isocratic; 16−16.5 min, 80-0% B linear; followed by 3 min of re-equilibration of the
column before the next run. The flow rate was 350 μL/min. The UHPLC system was
coupled to a Q Exactive™ Hybrid Quadrupole-Orbitrap High Resolution Mass
Spectrometer (Thermo Fisher Scientific, San Jose, CA, USA). The MS acquisition was
performed in both negative and positive ionization with a mass resolution of 70,000 at
167
m/z 200. Separate injections were performed in a data-dependent (top 5) MS/MS mode
with the full scan mass resolution reduced to 35,000 at m/z 200. The injection volume
was 4 μL for negative ionization and 2 μL for positive ionization acquisition. The m/z
range for full scan analyses was 70–1000. Heated electrospray ionization (HESI)
parameters were as follows: sheath gas flow 45 arb, auxiliary gas flow 10 arb, sweep
gas flow 1 arb, spray voltage 3.5 kV, probe temperature 350°C, capillary temperature
320 °C for negative ionization and 325 °C for positive ionization. In source CID was 2
eV. The mass spectrometer was calibrated using Pierce™ negative and positive ion
calibration solution (Thermo Fisher Scientific, San Jose CA, USA). To avoid possible
bias, the sequence of injections for urine samples was randomized.
Multivariate Data Processing and Statistical Analyses
LC-HRMS data were converted to mzXML using MSConvert from ProteoWizard
(Chambers, et al., 2012) and then processed using MZmine 2.12 (Pluskal, Castillo,
Villar-Briones, & Orešič, 2010). Peaks in each sample were extracted, deconvoluted,
and deisotoped. Alignment using join aligner algorithm was conducted with a 10 ppm
tolerance for m/z values and 0.2 min tolerance for retention time. Gap filling using peak
finder algorithm was performed to fill in missing peaks. This dataset was imported into
MetaboAnalyst. Normalization was conducted on each observation using its specific
gravity as the normalization factor. Specific gravity of each urine sample was measured
using a LED refractometer (Fisher Scientific, Pittsburgh, PA, USA). The resultant
dataset was exported from MetaboAnalyst and then imported into SIMCA (Version 14.0,
Umetrics, Umea, Sweden) for multivariate statistical analysis. Data acquired using both
negative and positive ionization were mean-centered, Pareto scaled and log-
transformed before PCA modelling; mean-centered and log-transformed before PLS-DA
168
or OPLS-DA analyses. Unsupervised PCA model was performed to initially examine
intrinsic variation in the data set. Then supervised pattern recognition methods include
PLS-DA and OPLS-DA were used to extract maximum information on discriminant
compounds from the data (Bylesjö, Rantalainen, Cloarec, Nicholson, Holmes, & Trygg,
2006). Validation of the models was tested using 7-fold internal cross-validation and
permutation tests for 200 times. To further evaluate the predictive ability of the PLS-DA
or OPLS-DA models, an external validation procedure was performed (Brindle, et al.,
2002; Llorach, et al., 2010). The LC-HRMS metabolomics data set was split into a
training set and a test set. Approximately 70% of the samples were randomly selected
as the training set and the remaining 30% were treated as the test set. PLS-DA and
OPLS-DA models were built based on the training set and then blindly predicted the
classes of the samples in the test set. This procedure was repeated for 30 times and
correct classification rates were calculated.
Results and Discussion
Quality Control of Multivariate Analyses
QCs created from internal standard group clustered tight and completely
separated from the experimental samples on the PCA score plots (Figure 6-1A, 6-1B). It
confirmed the stability of instrumental analyses and quality of sample preparation
method. PCA score plots also revealed that triplicates of pooled QCs from cranberry,
apple and baseline group clustered tight across the entire sequence (Figure 6-1C, 6-1D)
suggesting a good quality of data acquisition.
Baseline Urine vs. Urine after Drinking Cranberry Juice
LC-HRMS data of urine samples acquired using both negative and positive
ionization were subjected to multivariate analyses. A partial segregation of urine sample
169
between two groups was obtained on PCA score plot (Figure 6-2A) with two samples
from cranberry juice group clustered with baseline group. Using a PLS-DA model, a
complete separation between two groups was achieved (Figure 6-3A). The cross-
validated score plot also demonstrated a clear separation except that one sample from
cranberry juice group was misclassified into baseline group (Figure 6-3C). The R2Y of
the PLS-DA model was 0.939 indicating that about 94% of variance in the Y data matrix
was explained by the model. Q2 calculated from the 7-fold cross-validation was 0.873
suggesting a good predictability of the PLS-DA model (Table 6-1). All R2 and Q2
calculated from 200 times permutation tests (Figure 6-4A) were smaller than those from
actual model, supporting the results from cross-validation that the segregation between
two groups of urine samples was not due to overfitting. The external validation test
generated correct classification rate of 97.1% for the PLS-DA model derived from
negative ionization analyses. This model was able to correctly classify an unknown
human urine sample with an error rate of 2.9%. Similarly, a valid segregation between
baseline urine and human urine after drinking cranberry juice was obtained based on
positive ionization data. Both score plot and cross-validated score plot of OPLS-DA
model showed a clear separation between two groups (Figure 6-3B, 6-3D). Similar to
the model derived from negative ionization data, one urine sample from cranberry juice
group was misclassified into baseline group during the cross-validation. Validation plot
from 200 permutation tests showed that Q2 of each model built based on permutated
class labels were smaller than 0.4. Both cross-validation and permutation test showed a
good predictability of the OPLS-DA model and confirmed the segregation of two groups
of urine samples was not due to overfitting. Unknown urine samples collected from
170
similar studies could be correctly classified using this OPLS-DA model with an error rate
of 5% (Table 6-1). The urinary metabolome modifications of young women caused by
drinking cranberry juice were revealed using a UHPLC-HRMS metabolomics approach.
Urine after Drinking Apple Juice vs. Urine after Drinking Cranberry Juice
Projection pattern techniques were used to analyze the overall urinary
metabolome differences in young women after drinking apple juice or cranberry juice.
OPLS-DA models were built on the LC-HRMS data collected from both negative and
positive ionization. Approximately 92% and 94% of variance in Y data matrix were
explained by two OPLS-DA models, respectively, indicating an excellent goodness of fit
(Table 6-1). Figure 6-6 showed a clear separation between two groups of urine samples
on the score plots and cross-validated score plots of OPLS-DA models. One sample
from the cranberry juice group was misclassified into baseline group during cross-
validation. All other observations were correctly classified during the 7-fold internal
validation. Q2 calculated from the cross-validation of two OPLS-DA models was 0.825
and 0.762, respectively, indicating good predictabilities. To detect the possibility of
overfitting in the OPLS-DA models, 200 permutation tests were conducted and
validation plots were drawn in Figure 6-7. The results showed that all Q2 calculated from
the models with permutated class labels were smaller than 0.4. Therefore the achieved
segregation between two groups of urine samples were not likely due to overfitting.
Furthermore, correct classification rate of 95% was obtained for both OPLS-DA models
after external validation. Therefore, an unknown urine sample collected from a similar
study would be correctly classified with an error rate of 5% (Table 6-1).
171
Discriminant Metabolites Identification
S-plots were used to visualize the variable influence in OPLS-DA models derived
from LC-HRMS data of baseline urine vs. cranberry juice (Figure 6-8A, 6-9A) and
cranberry juice vs. apple juice (Figure 6-8B, 6-9 B). At a significance level of p = 0.05, a
p(corr) of 0.5 was used as an arbitrary cutoff value to select the potential biomarkers.
Discriminant metabolites located on the upper right or lower left corner of the S-plot had
higher absolute p[1] and p(corr) values.
Figure 6-8A and Figure 6-9A showed a total of 45 and 53 metabolic features in
negative mode and positive mode, respectively. They were discriminant metabolites that
separated baseline urine and urine after cranberry juice consumption. Among them, 7
features in negative mode and 6 in positive mode were identified based on their
accurate masses and/or product ion spectra. Similarly, Figure 6-8B and 6-9B showed
that 48 and 75 metabolic features in negative mode and positive mode, respectively,
were discriminant metabolites that separated urine after apple juice and urine after
cranberry juice consumption. Up to 8 metabolic features in negative mode and 9
features in positive mode were identified using accurate masses and/or product ion
spectra. These identified metabolites were numbered in the Figure 6-8, 6-9 and
summarized in Table 6-2 and Table 6-3. Unidentified metabolic features were listed in
Table 6-4 and Table 6-5. HMDB (Wishart, et al., 2007), KEGG (Liebich & Först, 1990),
mzCloud and Metlin were searched to assist metabolite identification.
Compared to apple juice consumption, 9 exogenous metabolites increased in
human urine after drinking cranberry juice. A metabolite producing a [M-H]- ion at m/z
289.0371 was tentatively identified as 4-hydroxy-5-(hydroxyphenyl)-valeric acid-O-
sulfate (Δ=0.0016 Da). Its product ions included m/z 209.0461 [M-H-sulfate]-, m/z
172
191.0561 [M-H-sulfate-H2O]-, and m/z 96.9598. The compound producing a [M-H]- ion
at m/z 287.0409 was tentatively identified as 5-(dihydroxyphenyl)-ϒ-valerolactone-O-
sulfate. Its product ions in negative mode included m/z 207.0209 [M-H-sulfate]-, m/z
171.0776 [M-H-sulfate-2H2O]-, m/z 97.0290, m/z 141.0550 and m/z 74.0241. HMDB
was searched for this identification (Δ=0.0178 Da). 5-(Dihydroxyphenyl)-ϒ-valerolactone
was detected and tentatively assigned according to HMDB (Δ=0.0021 Da). One
metabolite generating a [M-H]- ion at m/z 285.0620 and a major fragment ion at m/z
109.0294 [M-H-glucuronide]- was tentatively identified as diphenol glucuronide. It
matched the same compound in HMDB (Δ=0.0004 Da). The compound producing a
[M+H]+ ion at m/z 167.0303 and a fragment ion at m/z 121.0808 was tentatively
assigned as 3-(hydroxyphenyl)propionic acid according to HMDB (Δ=0.0003 Da).This
metabolite was found to be higher in human plasma following cranberry juice in Chapter
5. Coumaric acid (Δ=0.0002 Da) was detected and tentatively identified based on its
accurate m/z 165.0548 [M+H]+ and product ions at m/z 147.0439 and 119.0492. The
sulfated coumaric acid was previously detected in human plasma after drinking
cranberry juice in Chapter 5. Trihydroxybenzoic acid and O-methylgallic acid were
detected and tentatively assigned. Trihydroxybenzoic acid produced a [M+H]+ ion at m/z
171.0360 and fragment ions at m/z 148.9768, m/z 95.0131, m/z 125.9610, and m/z
107.9506. The product ions match those of trihydroxybenzoic acid in HMDB (Δ=0.0072
Da). However, the positons of three hydroxyl groups could not be determined unless the
retention time was compared to those of all possible isomers. Gallic acid could be one
possible isomer. O-Methylgallic acid was tentatively assigned because its accurate m/z
185.0447 [M+H]+ and product ions at m/z 125.9610, m/z 95.0131, m/z 148.9768 and
173
m/z 107.9506 matched those in Metlin and HMDB (Δ=0.0002 Da). The metabolite
producing a [M+H]+ ion at m/z 169.0862 were assigned as 1, 3, 5-trimethoxybenzene
according to HMDB (Δ=0.0003 Da). A previous study suggested that gallic acid, 4-O-
methylgallic acid and 1, 3, 5-trimethoxybenzene were candidate urinary biomarkers
after epigallocatechin gallate intake (Loke, Jenner, Proudfoot, McKinley, Hodgson,
Halliwell, et al., 2009).
In addition to these exogenous metabolites, a total of 6 endogenous metabolites
were detected and tentatively identified. A metabolite produced a [M-H]- ion at m/z
219.0512 and product ions at m/z 174.9699 [M-H-COO]-, m/z 115.9207, m/z 111.0087,
m/z 128.9645 and m/z 100.9333. It was tentatively assigned as 3-hydroxy-3-
carboxymethyl adipic acid according to HMDB (Δ=0.0002 Da). This metabolite belongs
to the family of tricarboxylic acids and derivatives (Liebich & Först, 1990). Pimelic acid
(Δ=0.0012 Da, Δ=0.00006 Da) that belongs to the family of dicarboxylic acids was
tentatively identified based on its accurate m/z 159.0651 [M-H]- and m/z 161.0810
[M+H]+. Its product ions in negative mode included m/z 115.0402, m/z 97.0290, m/z
141.0550 and m/z 74.0241. Its fragmentation pattern matched that in Metlin. A
metabolite generated a [M-H]- ion at m/z 175.0600 and product ions at m/z 73.0289, m/z
85.0290, m/z 87.0083 and m/z 132.0666. Its accurate mass and fragmentation pattern
matched those of 2- or 3-isopropylmalate in HMDB, mzCloud and Metlin (Δ=0.0012 Da).
The exact positions of hydroxyl group could not be determined based on MS2 spectra.
2- and 3-Isopropylmalate belong to hydroxyl fatty acids. 2-Isopropylmalate is an
intermediate in pyruvate metabolism. The compound generating a [M-H]- ion at m/z
205.0344 and a [M+H]+ ion at m/z 207.0499 was tentatively assigned as homocitric acid
174
according to HMDB and KEGG (Δ=0.0010 Da, Δ=0.0009 Da). Homocitric acid is a citric
acid analogue in human urine. It is an intermediate in pyruvate metabolism. Hippuric
acid was detected and tentatively identified based on its accurate m/z 178.0435 [M-H]-
and HMDB search (Δ=0.00747 Da). Elevated level of hippuric acid in human urine
following cranberry juice consumption was found using 1H NMR metabolomics
approach in Chapter 4. Quinic acid was increased in human urine after cranberry juice
consumption. Similar observation was made in human plasma in Chapter 5. The
accurate mass, retention time and fragmentation pattern of quinic acid standard was
curated as part of in-house database in SECIM. The identification of quinic acid was
based on the in-house database.
Compared to baseline urine, urinary excretion of two additional metabolites were
increased following cranberry juice consumption. One exogenous metabolite was
tentatively identified as 3, 4-dihydroxyphenyl propionic acid. It produced a [M-H]- ion at
m/z 181.0589 and products ions at m/z 137.0609, m/z 59.0131, m/z 109.0294 and m/z
121.0296, which matched those of 3, 4-dihydroxyphenyl propionic acid in mzCloud and
HMDB (Δ=0.0083 Da). One endogenous metabolite was tentatively assigned as N-
acetyl-L-glutamate 5-semialdehyde. This compound gave rise to a [M+H]+ ion at m/z
174.0733 which matched the same compound in HMDB and KEGG (Δ=0.0028 Da). N-
Acetyl-L-glutamate 5-semialdehyde belongs to the group of N-acyl-aliphatic-alpha
amino acids. It is an intermediate in the pathway of oxaloacetate metabolism.
Cranberry juice consumption caused a higher urinary level of metabolites that
belong to the family of di- or tri-carboxylic acids. Metabolites that are intermediates in
the pathway of 2-oxocarboxylic acids (oxaloacetate, pyruvate, etc.) metabolism also
175
increased after cranberry juice consumption. Carboxylic acids and 2-oxocarboxylic
acids supply energy to living cells. The results suggested that energy metabolism in
young women was changed by cranberry juice consumption.
All 53 identified discriminant metabolites in plasma and urine of human or rats
were summarized in Table 6-6. Among them, 35 metabolites contributed to the group
separation of plasma or urine among cranberry, apple and baseline. 3-(Hydroxyphenyl)
propionic acid, trihydroxybenzoic acid, quinic acid, hippuric acid were biomarkers of
cranberry intake in both human plasma and human urine. Catechol sulphate, 4-hydroxy-
5-(hydroxyphenyl)-valeric acid-O-sulphate and hippuric acid were detected in plasma or
urine of both human and rats after cranberry intake. Hippuric acid was the most
prominent biomarker detected in human plasma, human urine and rat urine.
Summary
This study demonstrated that the overall urinary metabolome in young women
were altered following cranberry juice consumption. Compared to baseline condition,
cranberry juice consumption caused a greater urinary excretion of metabolites including
5-(dihydroxyphenyl)-ϒ-valerolactone and its sulfated form, 3,4-dihydoxyphenyl propionic
acid, 3-(hydroxyphenyl) propionic acid, trihydroxybenzoic acid, 3-hydroxy-3-
carboxymethyl adipic acid, 2 or 3-isopropylmalate, pimelic acid, homocitric acid, hippuric
acid, quinic acid, and N-acetyl-L-glutamate 5-semialdehyde. Furthermore, urinary
metabolites discriminating cranberry juice and apple juice consumption included 4-
hydroxy-5-(hydroxyphenyl)-valeric acid-O-sulphate, 5-(dihydroxyphenyl)-ϒ-
valerolactone and its sulfated form, diphenol glucuronide, 3-(hydroxyphenyl) propanic
acid, O-methylgallic acid, trihydroxybenzoic acid, 1,3,5-trimethoxybenzene, coumaric
acid, 2-or 3-isopropylmalate, pimelic acid, homocitric acid, hippuric acid and quinic acid.
176
Elevated level of carboxylic acids and intermediates in 2-oxocarboxylic acids
metabolism suggested that cranberry juice consumption changed energy metabolism in
young women.
177
Table 6-1. Summary of parameters for PLS-DA or OPLS-DA model for human baseline urine and urine after drinking cranberry juice or apple juice.
Negative Ionization Analyses Positive Ionization Analyses
Baseline vs. Cranberry Juice
Cranberry Juice vs. Apple Juice
Baseline vs. Cranberry Juice
Cranberry Juice vs. Apple Juice
PLS-DA OPLS-DA OPLS-DA OPLS-DA
Na
2 1Pc+1Od 1Pc+1Od 1Pc+1Od
R2
X(cum)b
0.219 0.392 0.331 0.363
R2
Y(cum)b
0.939 0.920 0.951 0.941
Q2
(cum)b
0.873 0.825 0.847 0.762
*Correct Classification Rate
0.971±0.054 0.938±0.064 0.954±0.061 0.958±0.060
a N: number of components. b R2X (cum)and R2Y (cum) are the cumulative modeled variations in the X and Y matrix, respectively. Q2Y (cum) is the cumulative predicted variation in the Y matrix. c Predictive component. d Orthogonal component. *Correct classification rate was obtained from external validation procedure repeated for 30 times.
178
Table 6-2. Identification of discriminant metabolites in human urine after drinking cranberry juice or apple juice by negative ionization analysis.
No. Retention Time (min)
Detected Mass [M-H]-
p[1] (contribution) a
p(corr)[1] (confidence) b
MSMS Putative Identification
Theoretical Mass [M-H]-
Mass Difference (Da)
Reference
CJ vs. AJ c
CJ vs. BS d
1 0.948 289.0371 0.050 0.513 191.0561, 209.0461, 96.9598
4-Hydroxy-5-(hydroxyphenyl)-valeric acid-O-sulphate
289.0387 0.0016 HMDB ----
2 1.047 219.0512 0.057 (0.072)
0.729 (0.789)
174.9699, 115.9207, 111.0087, 128.9645, 100.9333
3-Hydroxy-3-carboxymethyl-adipic acid
219.0510 0.0002 HMDB
3 1.108 175.0600 0.063 (0.067)
0.750 (0.836)
73.0289, 85.0290, 87.0083, 132.0666
(2)3-Isopropylmalate
175.0610 0.0012 HMDB mzCloud Metlin
4 2.087 159.0651 0.065 (0.065)
0.693 (0.753)
115.0402, 97.0290, 141.0550, 74.0241
Pimelic acid 159.0663 0.0012 HMDB Metlin
5 3.490 287.0409 0.061 (0.067)
0.664 (0.749)
207.0209, 171.0776, 142.0511
5-(Dihydroxyphenyl)-ϒ-valerolactone sulfate
287.0231 0.0178 HMDB
6 5.970 205.0344 0.086 (0.076)
0.902 (0.926)
---- Homocitric acid 205.0354 0.0010 HMDB
7 7.771 285.0620 0.053 0.748 109.0294, 113.0243, 85.0290, 59.0131
Diphenol glucuronide
285.0616 0.0004 HMDB ----
8 8.017 181.0589 0.052 0.799 137.0609, 59.0131, 109.0294, 121.0296
3,4-Dihydroxyphenyl propionic acid
181.0506 0.0083 HMDB mzCloud
----
9 8.092 178.0435 0.069 (0.058)
0.660 (0.640)
---- Hippuric acid 178.0510 0.0075 HMDB
a Number inside the parentheses is the p[1] value obtained from OPLS-DA model based on cranberry juice vs. baseline. b Number inside the parentheses is the p (corr) [1] value obtained from OPLS-DA model based on cranberry juice vs. baseline c Arrows indicated a decrease or increase in metabolite level in human plasma after drinking cranberry juice compared to apple juice. d Arrows indicated a decrease or increase in metabolite level in human plasma after drinking cranberry juice compared to baseline.
179
Table 6-3. Identification of discriminant metabolites in human urine after drinking cranberry juice or apple juice by positive ionization analysis.
NO. Retention Time (min)
Detected Mass [M+H]+
p[1] (contribution)a
p(corr)[1] (confidence) b
MSMS Putative Identification
Theoretical Mass [M+H]+
Mass Difference (Da)
Reference
CJ vs AJ c
CJ vs BS d
1 0.890 193.0709 0.067 (0.050)
0.801 (0.733)
111.0442 129.0546, 95.0495, 83.0496, 69.0341
Quinic acid 193.0707 0.0002 HMDB in-house DB
2 2.025 161.0810 0.065 0.734 ---- Pimelic acid 161.0808 0.00006
HMDB
3 6.044 207.0490 0.093 (0.093)
0.814 (0.922)
139.0024, 143.0339, 157.0128, 129.0180, 119.0340
Homocitric acid 207.0499 0.0009 HMDB KEGG
4 6.107 167.0703 0.093 (0.087)
0.933 (0.937)
121.0808 3-(Hydroxyphenyl) propionic acid
167.0703 0.0003 HMDB
5 7.024 209.0787 0.116 (0.111)
0.936 (0.957)
149.0597, 166.0863, 84.9601
5-(Dihydroxyphenyl)-ϒ-valerolactone
209.0808 0.0021 HMDB
6 7.198 185.0447 0.070 0.691 126.0913, 125.1073, 143.1177
4-O-Methylgallic acid
185.0444 0.0002 HMDB Metlin
----
7 7.214 171.0360 0.071 (0.057)
0.741 (0.584)
148.9768, 95.0131, 125.9610, 107.9506
Trihydroxybenzoic acid
171.0288 0.0072 HMDB
8 7.865 169.0862 0.067 0.816 ---- 1,3,5-Trimethoxybenzene
169.0859 0.0003 HMDB ----
9 8.420 174.0733 0.059 (0.055)
0.739 (0.800)
146.0598 N-Acetyl-L-glutamate 5-semialdehyde
174.0761 0.0028 HMDB KEGG
----
10 9.103 165.0548 0.050 0.659 147.0439, 119.0492 Coumaric acid 165.0550 0.0002 HMDB mzCloud
----
a Number inside the parentheses is the p[1] value obtained from OPLS-DA model based on cranberry juice vs. baseline. b Number inside the parentheses is the p (corr) [1] value obtained from OPLS-DA model based on cranberry juice vs. baseline c Arrows indicated a decrease or increase in metabolite level in human plasma after drinking cranberry juice compared to apple juice. d Arrows indicated a decrease or increase in metabolite level in human plasma after drinking cranberry juice compared to baseline.
180
Table 6-4. Unidentified discriminant metabolic features in human urine after cranberry juice or apple juice by negative ionization analysis.
Retention Tim (min)
Detected Mass [M-H]-
p[1] (contribution) a
p(corr)[1] (confidence) b
CJ vs. AJ c CJ vs. BS d
0.912 192.0586 0.070 0.732 ----
0.997 179.9963 0.183 (0.174) 0.931 (0.953)
2.129 205.0711 0.076(0.078) 0.665 (0.721)
2.969 190.9646 0.064 (0.068) 0.610 (0.722)
3.732 200.0921 0.080 0.691 ----
4.167 141.0545 0.073 0.677 ----
5.485 162.9879 0.156 (0.160) 0.835 (0.905)
5.586 163.9890 0.163 (0.156) 0.936 (0.973)
5.620 82.0283 0.171 (0.164) 0.924 (0.965)
5.623 291.9446 0.177 (0.176) 0.909 (0.956)
5.658 346.9621 0.199 (0.185) 0.926 (0.953)
5.668 163.9816 0.189 (0.191) 0.908 (0.960)
5.692 273.9641 0.160 (0.156) 0.935(0.973)
5.736 177.9806 0.181 (0.185) 0.891 (0.958)
5.744 161.9856 0.188 (0.207) 0.842 (0.941)
5.746 162.9882 0.163 (0.169) 0.905 (0.958)
5.858 77.9635 0.155 (0.162) 0.922(0.952)
5.898 346.9621 0.212 (0.200) 0.947 (0.970)
5.905 163.9889 0.162 (0.159) 0.930 (0.967)
5.905 163.9815 0.175 (0.190) 0.899 (0.965)
5.929 161.9856 0.176 (0.190) 0.833 (0.931)
5.954 77.9635 0.156 (0.157) 0.917 (0.946)
181
Table 6-4. Continued. Retention Tim (min)
Detected Mass [M-H]-
p[1] (contribution) a
p(corr)[1] (confidence) b
CJ vs. AJ c CJ vs. BS d
5.995 177.9808 0.161 (0.147) 0.902 (0.892)
6.012 291.9443 0.154 (0.148) 0.915 (0.919)
6.054 82.0283 0.145 (0.140) 0.926 (0.953)
6.466 344.0305 0.084 (0.075) 0.813 (0.862)
6.562 432.0720 0.106 (0.108) 0.874 (0.928)
6.626 360.0609 0.088 (0.087) 0.699 (0.726)
7.105 229.9741 0.099 (0.074) 0.700 (0.688)
7.270 220.0535 0.089 0.649 ----
7.449 185.0811 0.082 (0.057) 0.826 (0.829)
7.568 199.0605 0.076 (0.066) 0.908 (0.886)
7.625 229.1076 0.082 (0.071) 0.826(0.783)
8.025 204.9968 0.062 0.628 -----
8.040 204.0045 0.056 (0.053) 0.769 (0.821)
8.099 181.0698 0.064 0.738 ----
8.117 274.0755 0.066 (0.059) 0.797 (0.847)
8.231 443.0101 0.109 (0.126) 0.799 (0.909)
8.239 430.9833 0.099 (0.126) 0.781 (0.884)
8.261 432.9802 0.101 (0.132) 0.750 (0.891)
8.269 441.0156 0.111 (0.117) 0.833 (0.914)
8.285 434.9778 0.155 (0.135) 0.796 (0.912)
a Number inside the parentheses is the p[1] value obtained from OPLS-DA model based on cranberry juice vs. baseline.
b Number inside the parentheses is the p (corr) [1] value obtained from OPLS-DA model based on cranberry juice vs. baseline
c Arrows indicated a decrease or increase in metabolite level in human plasma after drinking cranberry juice compared to apple juice.
d Arrows indicated a decrease or increase in metabolite level in human plasma after drinking cranberry juice compared to baseline.
182
Table 6-5. Unidentified discriminant metabolic features in human urine after cranberry juice or apple juice by positive ionization analysis.
Detected Mass [M+H]+
p[1] (contribution) a p(corr)[1] (confidence) b
CJ vs. AJ c CJ vs. BS d
197.1020 -0.071 -0.738 ----
179.0915 -0.093 -0.761 ----
196.0293 0.067 (0.072) 0.736 (0.816)
143.0704 0.057 (0.063) 0.744 (0.763)
264.0900 0.086 (0.080) 0.763 (0.750)
125.0601 0.055 0.826 ----
129.5364 0.073 (0.061) 0.744 (0.645)
120.5311 0.105 (0.108) 0.845 (0.836)
143.0704 0.064 (0.077) 0.816 (0.801)
202.108 0.061 0.690 ----
129.5364 0.074 (0.069) 0.800 (0.812)
285.0811 0.103 (0.103) 0.714 (0.632)
127.0757 0.059 (0.051) 0.692 (0.602)
149.0072 0.131 (0.133) 0.897 (0.927)
237.9691 0.197 (0.201) 0.922 (0.943) ----
255.9796 0.182 (0.188) 0.925 (0.944)
253.9415 0.158 (0.159) 0.929 (0.923)
265.9643 0.167 (0.181) 0.909 (0.937)
276.9577 0.175 (0.172) 0.935 (0.926)
140.0017 0.167 (0.176) 0.917 (0.930)
151.5098 0.202 (0.213) 0.900 (0.914)
260.9852 0.161 (0.161) 0.905 (0.922)
183
Table 6-5. Continued. Retention Tim (min)
Detected Mass [M+H]+
p[1] (contribution) a
p(corr)[1] (confidence) b
CJ vs. AJ c CJ vs. BS d
5.855 130.9966 0.121 (0.120) 0.926 (0.921)
5.881 142.5044 0.132 (0.146) 0.910 (0.924)
5.895 229.0314 0.105 (0.103) 0.933 (0.954)
5.947 152.5208 0.111 (0.109) 0.917 (0.920)
6.034 207.0620 0.124 (0.112) 0.904 (0.950)
6.080 141.5363 0.116 (0.109) 0.921 (0.945)
6.222 164.0288 0.089 (0.089) 0.902 (0.942)
6.573 116.5105 0.071 (0.091) 0.522 (0.581)
6.752 247.0131 0.072 (0.080) 0.618 (0.653)
6.895 133.5387 0.121 (0.114) 0.942 (0.952)
6.907 201.0752 0.051 0.838 ----
7.043 311.0009 0.076 (0.058) 0.722 (0.598)
7.068 288.1806 0.062 0.675 ----
7.144 149.5338 0.055 0.826 ----
7.245 129.0184 0.065 0.716 ----
7.257 155.5507 0.051 0.740 ----
7.296 159.5291 0.070 0.741 ----
7.320 243.0476 0.092 (0.067) 0.728 (0.601)
7.348 187.0966 0.058 0.832 ----
7.364 150.5233 0.060 0.698 ----
7.414 143.5701 0.063 0.605 ----
7.455 142.5424 0.064 (0.053) 0.861 (0.806)
7.682 213.1122 0.062 0.792 ----
7.717 204.0420 0.059 0.756 ----
184
Table 6-5. Continued. Retention Tim (min)
Detected Mass [M+H]+
p[1] (contribution) a
p(corr)[1] (confidence) b
CJ vs. AJ c CJ vs. BS d
7.777 183.5290 0.074 (0.070) 0.767 (0.798)
7.888 203.544 0.068 (0.073) 0.788 (0.824)
8.032 404.5529 0.066 (0.077) 0.755 (0.831)
8.033 591.0945 0.071 (0.068) 0.689 (0.790)
8.038 576.1292 0.056 (0.069) 0.690 (0.797)
8.038 413.0433 0.063 (0.057) 0.672 (0.776)
8.039 421.0505 0.051 0.636 ----
8.041 180.0144 0.050 (0.060) 0.663 (0.796)
8.045 415.0440 0.071 0.782 ----
8.051 181.2448 0.056 (0.060) 0.584 (0.583)
8.052 178.8979 0.065 (0.057) 0.561 (0.559)
8.060 204.1052 0.084 (0.108) 0.748 (0.810)
8.064 236.1807 -0.059 -0.545 ----
8.079 243.5619 -0.062 -0.699 ---
8.094 364.0349 -0.075 -0.762 ----
8.094 181.5549 -0.050 -0.755 ----
8.097 172.5497 -0.050 -0.774 ----
8.102 180.5359 -0.069 -0.753 ----
8.104 193.5648 -0.056 -0.692 ----
8.108 193.0628 -0.053 -0.737 ----
8.142 201.0492 -0.051 -0.728 ----
8.274 238.5014 0.126 (0.146) 0.913 (0.949)
8.408 153.5600 0.052 0.649 ----
185
Table 6-5. Continued. Retention Tim (min)
Detected Mass [M+H]+
p[1] (contribution) a
p(corr)[1] (confidence) b
CJ vs. AJ c CJ vs. BS d
8.472 83.0863 0.050 (0.050) 0.691 (0.768)
8.786 213.0655 0.051 (0.066) 0.620 (0.758)
a Number inside the parentheses is the p[1] value obtained from OPLS-DA model based on cranberry juice vs. baseline.
b Number inside the parentheses is the p (corr) [1] value obtained from OPLS-DA model based on cranberry juice vs. baseline
c Arrows indicated a decrease or increase in metabolite level in human plasma after drinking cranberry juice compared to apple juice.
d Arrows indicated a decrease or increase in metabolite level in human plasma after drinking cranberry juice compared to baseline.
186
Table 6-6. Summary of identified discriminant metabolites in rats and human. Discriminant Metabolites Cranberry vs. baseline Cranberry vs. apple Apple vs.
baseline Rat
urine Human plasma
Human urine
Rat urine
Rat plasma
Human plasma
Human urine
Rat urine
p-Hydroxybenzoic acid X
Coumaric acid X X
Coumaric acid sulfate X X
Ferulic acid sulfate X X
Phenol X X X
Phenyl sulfate X
Diphenol glucuronide X
Catechol sulphate X X X
3, 4-Dihydroxyphenylvaleric acid
X
3,4-Dihydroxyphenyl propionic acid
X
Hydroxyphenyl acetic acid
X X
p-Hydroxyphenylacetic acid
X X
3-(3’-Hydroxyphenyl)-3-hydroxypropanoic acid
X X
3-(Hydroxyphenyl) propionic acid
X X X
4-Hydroxy-5-(hydroxyphenyl)-valeric acid-O-sulphate
X X
5-(Hydroxyphenyl)-ϒ-valerolactone-O-sulphate
X
187
Table 6-6. Continued. Discriminant Metabolites Cranberry vs. baseline
Cranberry vs. apple
Apple vs. baseline
Rat urine
Human plasma
Human urine
Rat urine
Rat plasma
Human plasma
Human urine
Rat urine
5-(Dihydroxyphenyl)-ϒ-valerolactone
X X
5-(Dihydroxyphenyl)-ϒ-valerolactone sulfate
X X
5-(Trihydroxyphenyl)-ϒ-valerolactone
X
3,4-Dihydroxyphenyl ethanol sulfate
X X
4'-O-Methyl-(-)-epicatechin-3'-O-beta-glucuronide
X
3'-O-Methyl-(-)-epicatechin-7-O-glucuronide
X
4-O-Methylgallic acid X
1,3,5-Trimethoxybenzene X
Trihydroxybenzoic acid X X X X
4-Hydroxydiphenylamine X
Peonidin-3-O-hexose X
Quinic acid X X X X
Lactic acid X X
Succinic acid X
Citric acid X X X
α-Ketoglutaric acid X
Aconitic acid X
Citramalic acid X X
188
Table 6-6. Continued. Discriminant Metabolites Cranberry vs. baseline
Cranberry vs. apple
Apple vs. baseline
Rat urine
Human plasma
Human urine
Rat urine Human plasma
Human urine
α-D-glucose X X
D-maltose X X
Creatinine X
2-Furoylglycine X X
Hippuric acid X X X X X X X
Hydroxyhippuric acid X X
Vanilloylglycine X X
Vanilloloside X
Tyrosine X
Hydroxyoctadecanoic acid
X
4-Acetamido-2-aminobutanoic acid
X
Glycerol 3-phosphate X X
Indole-3-acetaldehyde X
Dihydroxyquinoline X X
3-Hydroxy-3-carboxymethyl-adipic acid
X X
Pimelic acid X X
Homocitric acid X X
(2)3-Isopropylmalate X X
N-Acetyl-L-glutamate 5-semialdehyde
X
189
Figure 6-1. The PCA score plot of human urine and quality control samples from LC-HRMS metabolomics. A) PLS-DA
score plot of negative ionization data, B) OPLS-DA score plot of positive ionization data, C) PLS-DA score plot of negative ionization without QC of internal standards and D) OPLS-DA score plot of positive ionization data without QC of internal standards.
-25
-20
-15
-10
-5
0
5
10
15
20
-150 -100 -50 0 50 100 150t[1]
t [2
]
PoolQC from apple juice group
PoolQC from Baseline group
PoolQC from cranberry juice group
Urine of baseline, cranberry and apple group
QC from internal standards
-30
-20
-10
0
10
20
30
-150 -100 -50 0 50 100t[1]
t [2
]
A B
-30
-20
-10
0
10
20
-50 -40 -30 -20 -10 0 10 20 30 40t[1]
t [2
]
-40
-30
-20
-10
0
10
20
-80 -60 -40 -20 0 20 40 60t[1]
t [2
]
C D
190
Figure 6-2. The PCA score plot of human baseline urine and human urine after cranberry juice from LC-HRMS
metabolomics. A) Data were acquired by negative ionization and B) data were acquired by positive ionization. Blue squares: baseline urine before drinking cranberry juice. Green squares: urine after drinking cranberry juice.
t [2
]
-30
-20
-10
0
10
20
-50 -40 -30 -20 -10 0 10 20 30 40t[1]
Baseline urine before cranberry juice
Urine after cranberry juice
-40
-30
-20
-10
0
10
20
30
-80 -60 -40 -20 0 20 40 60t[1]
t [2
]
A B
191
Figure 6-3. The PLS-DA, OPLS-DA score plots and cross-validated score plots of human baseline urine and urine after
cranberry juice. A) PLS-DA score plot by negative ionization, B) OPLS-DA score plot by positive ionization, C) PLS-DA cross-validated score plot by negative ionization and D) OPLS-DA cross-validated score plot by positive ionization. Blue squares: baseline urine before drinking cranberry juice. Green squares: urine after drinking cranberry juice.
-20
-15
-10
-5
0
5
10
15
B 8
B 1
0
B1
B 1
1
B 1
2
B 1
3
B 1
4
B 1
5
B 1
6
B 1
7
B 2
B 4
B 6
B 7 C1
C11
C12
C13
C14
C15
C16
C17
C2
C4
C6
C7
C8
Sample ID
t [1
]
-50
-40
-30
-20
-10
0
10
20
30
40
-40 -30 -20 -10 0 10 20 30
t[1]
to [
1]
A BBaseline Urine before cranberry juice
Urine after cranberry juice
-15
-10
-5
0
5
10
B 8
B 1
0
B1
B 1
1
B 1
2
B 1
3
B 1
4
B 1
5
B 1
6
B 1
7
B 2
B 4
B 6
B 7 C1
C11
C12
C13
C14
C15
C16
C17
C2
C4
C6
C7
C8
Sample ID
t [1
]
-25
-20
-15
-10
-5
0
5
10
15
20
25
-15 -10 -5 0 5 10
tcv[1]
tocv
[1]
C D
192
Figure 6-4. Validation plot obtained from 200 permutation tests for the PLS-DA and
OPLS-DA models of human baseline urine vs. human urine after cranberry juice. A) PLS-DA model by negative ionization and B) OPLS-DA model by positive ionization.
-0.2
0
0.2
0.4
0.6
0.8
-0.2 0 0.2 0.4 0.6 0.8 1
R2
, Q2
r(y, permuted y)
R2
, Q2
r(y, permuted y)
A BR2
Q2
193
Figure 6-5. The PCA score plot of human urine after drinking apple juice or cranberry juice from LC-HRMS metabolomics.
A) Data were acquired by negative ionization and B) data were acquired by positive ionization. Purple squares: urine after drinking apple juice. Green squares: urine after drinking cranberry juice.
t [2
]
-30
-20
-10
0
10
20
-50 -40 -30 -20 -10 0 10 20 30 40t[1]
-40
-30
-20
-10
0
10
20
30
-80 -60 -40 -20 0 20 40 60t[1]
t [2
]
A BUrine after apple juice
Urine after cranberry juice
194
Figure 6-6. The OPLS-DA score plots and cross-validated score plots of human urine after drinking apple juice or
cranberry juice from LC-HRMS metabolomics. A) OPLS-DA score plot by negative ionization, B) OPLS-DA score plot by positive ionization, C) OPLS-DA cross-validated score plot by negative ionization and D) OPLS-DA cross-validated score plot by positive ionization. Purple squares: urine after drinking apple juice. Green squares: urine after drinking cranberry juice.
-40
-30
-20
-10
0
10
20
30
-40 -30 -20 -10 0 10 20 30t[1]
to[1
]
-50
-40
-30
-20
-10
0
10
20
30
40
-30 -20 -10 0 10 20t[1]
to [
1]
tocv
[1]
-20
-15
-10
-5
0
5
10
15
-15 -10 -5 0 5 10tcv[1]
-30
-20
-10
0
10
20
-15 -10 -5 0 5 10tcv[1]
tocv
[1]
A B
C D
Urine after apple juice
Urine after cranberry juice
195
Figure 6-7. Validation plot obtained from 200 permutation tests for the OPLS-DA models
of human urine after apple juice vs. human urine after cranberry juice. A) Data were acquired by negative ionization and B) data were acquired by positive ionization.
R2
, Q2
r(y, permuted y)
R2
, Q2
r(y, permuted y)
A B
196
Figure 6-8. S-plots associated with the OPLS-DA score plot of data derived from LC-
HRMS of human baseline urine and urine after cranberry juice or apple juice by negative ionization. A) Human baseline urine vs. urine after cranberry juice and B) human urine after cranberry juice vs. urine after apple juice. p[1] is the loading vector of covariance in the first principal component. p(corr)[1] is loading vector of correlation in the first principal component. Variables with |p| ≥ 0.05 and |p(corr)| ≥ 0.5 are considered statistically significant. Significant variables in blue color were identified and numbered according to Table 6-2. Unidentified significant variables in red color were listed in Table 6-4. Non-significant variables were in green color.
2,5
3
4
68
9
A
B
1 4 5
2, 36
7
9
197
Figure 6-9. S-plots associated with the OPLS-DA score plot of data derived from LC-
HRMS of human baseline urine and urine after cranberry juice or apple juice by positive ionization. A) Human baseline urine vs. urine after drinking cranberry juice and B) human urine after cranberry juice vs. urine after apple juice. p[1] is the loading vector of covariance in the first principal component. p(corr)[1] is loading vector of correlation in the first principal component. Variables with |p| ≥ 0.05 and |p(corr)| ≥ 0.5 are considered statistically significant. Significant variables in blue color were identified and numbered according to Table 6-3. Unidentified significant variables in red color were listed in Table 6-5. Non-significant variables were in green color.
1
34
7
9A
5
2
B
10
2, 7
3
4 591 8
6
198
CHAPTER 7 CONCLUSIONS
1H NMR and UHPLC-Q-Orbitrap-HRMS based metabolomics methods were
developed and employed to discover that plasma and urinary metabolome of both
female rats and young women were changed after intake of cranberry procyanidins or
cranberry juices. Our project is among few metabolomics studies that combined both 1H
NMR and UHPLC-Q-Orbitrap-HRMS analytical techniques. Although 1H NMR was
successfully applied to metabolomics studies on rat urine, human urine and human
plasma, the study in Chapter 2 demonstrated that UHPLC-Q-Orbitrap-HRMS
metabolomics was more effective to reveal the plasma metabolome modifications in rats
caused by cranberry procyanidins. A list of exogenous compounds corresponding to
microbial metabolites of procyanidins were the major contributing markers in the rodent
model. Similarly, the plasma and urinary metabolite profiles of young women were
changed after drinking cranberry juice compared to their baseline profiles. The
metabolome in young women after cranberry juice consumption were different from
those after apple juice consumption. Both endogenous and exogenous metabolites
were discovered and putatively identified as discriminant biomarkers.
Pattern projection techniques were successfully applied in this metabolomics
study. Supervised PLS-DA and OPLS-DA models were developed to segregate rats or
human subjects that received different types of procyanidins or juice. These supervised
PLS-DA and OPLS-DA models had good predictability and could be used to predict the
class of unknown samples from similar studies with low error rates.
The incompleteness of in-house database prevented accurate identification of
new or unknown metabolites in this and other metabolomics studies. All putatively
199
identified metabolites need to be confirmed in the future when the in-house metabolome
database is complete.
This metabolomics research resulted in the identification of specific molecular
profiles and biomarkers of cranberry procyanidin intake in rats and cranberry juice
intake in human for the first time. The discriminant metabolites suggested that many
metabolic pathways were affected by cranberry juice or cranberry procyanidin intake.
The changes in metabolite profiles were likely caused by the ability of cranberries to
impact gene transcription and protein expression. This is also the first time that the
systematic physiological effects of cranberry juice intake was depicted at metabolite
levels. Findings made in this research will help to provide an effective way to assess
cranberry juice or procyanidin intake in epidemiological studies or clinical trials. This
knowledge will help to elucidate the mechanisms of cranberry juices or procyanidins in
mitigating urinary tract infections or other chronic diseases.
200
LIST OF REFERENCES
Ahuja, S., Kaack, B., & Roberts, J. (1998). Loss of fimbrial adhesion with the addition of Vaccinum macrocarpon to the growth medium of P-fimbriated Escherchia coli. The Journal of Urology, 159(2), 559-562.
Appeldoorn, M. M., Vincken, J.-P., Gruppen, H., & Hollman, P. C. (2009). Procyanidin dimers A1, A2, and B2 are absorbed without conjugation or methylation from the small intestine of rats. The Journal of Nutrition, 139(8), 1469-1473.
Avorn, J., Monane, M., Gurwitz, J. H., Glynn, R. J., Choodnovskiy, I., & Lipsitz, L. A. (1994). Reduction of bacteriuria and pyuria after ingestion of cranberry juice. JAMA, 271(10), 751-754.
Baba, S., Osakabe, N., Natsume, M., & Terao, J. (2002). Absorption and urinary excretion of procyanidin B2 [epicatechin-(4β-8)-epicatechin] in rats. Free Radical Biology and Medicine, 33(1), 142-148.
Barbosa-Cesnik, C., Brown, M. B., Buxton, M., Zhang, L., DeBusscher, J., & Foxman, B. (2011). Cranberry juice fails to prevent recurrent urinary tract infection: results from a randomized placebo-controlled trial. Clinical Infectious Diseases, 52(1), 23-30.
Beachey, E. H. (1981). Bacterial adherence: adhesin-receptor interactions mediating the attachment of bacteria to mucosal surfaces. Journal of Infectious Diseases, 143(3), 325-345.
Beckonert, O., Keun, H. C., Ebbels, T. M., Bundy, J., Holmes, E., Lindon, J. C., & Nicholson, J. K. (2007). Metabolic profiling, metabolomic and metabonomic procedures for NMR spectroscopy of urine, plasma, serum and tissue extracts. Nature Protocols, 2(11), 2692-2703.
Benjamini, Y., & Hochberg, Y. (1995). Controlling the false discovery rate: a practical and powerful approach to multiple testing. Journal of the Royal Statistical Society. Series B (Methodological), 289-300.
Bingol, K., Li, D.-W., Bruschweiler-Li, L., Cabrera, O. A., Megraw, T., Zhang, F., & Bruschweiler, R. (2014). Unified and Isomer-Specific NMR Metabolomics Database for the Accurate Analysis of 13C–1H HSQC Spectra. ACS Chemical Biology, 10(2), 452-459.
Blatherwick, N. (1914). The specific role of foods in relation to the composition of the urine. Archives of Internal Medicine, 14(3), 409-450.
Blatherwick, N., & Long, M. L. (1923). Studies of urinary acidity II. The increased acidity produced by eating prunes and cranberries. Journal of Biological Chemistry, 57(3), 815-818.
201
Bond, W., Favero, M., Petersen, N., Gravelle, C., Ebert, J., & Maynard, J. (1981). Survival of hepatitis B virus after drying and storage for one week. The Lancet, 317(8219), 550-551.
Bouatra, S., Aziat, F., Mandal, R., Guo, A. C., Wilson, M. R., Knox, C., Bjorndahl, T. C., Krishnamurthy, R., Saleem, F., & Liu, P. (2013). The human urine metabolome. PloS one, 8(9), e73076.
Brindle, J. T., Antti, H., Holmes, E., Tranter, G., Nicholson, J. K., Bethell, H. W., Clarke, S., Schofield, P. M., McKilligin, E., & Mosedale, D. E. (2002). Rapid and noninvasive diagnosis of the presence and severity of coronary heart disease using 1H-NMR-based metabonomics. Nature Medicine, 8(12), 1439-1445.
Burton, L., Ivosev, G., Tate, S., Impey, G., Wingate, J., & Bonner, R. (2008). Instrumental and experimental effects in LC–MS-based metabolomics. Journal of Chromatography B, 871(2), 227-235.
Bylesjö, M., Rantalainen, M., Cloarec, O., Nicholson, J. K., Holmes, E., & Trygg, J.
(2006). OPLS discriminant analysis: combining the strengths of PLS‐DA and SIMCA classification. Journal of Chemometrics, 20(8‐10), 341-351.
Caton, P. W., Pothecary, M. R., Lees, D. M., Khan, N. Q., Wood, E. G., Shoji, T., Kanda, T., Rull, G., & Corder, R. (2010). Regulation of Vascular Endothelial Function by Procyanidin-Rich Foods and Beverages†. Journal of Agricultural and Food Chemistry, 58(7), 4008-4013.
Center, B. (2004). USDA database for the proanthocyanidin content of selected foods. Washington, DC: US Department of Agriculture.
Chambers, M. C., Maclean, B., Burke, R., Amodei, D., Ruderman, D. L., Neumann, S., Gatto, L., Fischer, B., Pratt, B., & Egertson, J. (2012). A cross-platform toolkit for mass spectrometry and proteomics. Nature Biotechnology, 30(10), 918-920.
Chen, P. X., Bozzo, G. G., Freixas-Coutin, J. A., Marcone, M. F., Pauls, P. K., Tang, Y., Zhang, B., Liu, R., & Tsao, R. (2014). Free and conjugated phenolic compounds and their antioxidant activities in regular and non-darkening cranberry bean (< i> Phaseolus vulgaris</i> L.) seed coats. Journal of Functional Foods.
Crozier, A. (2013). Absorption, metabolism, and excretion of (−)-epicatechin in humans: an evaluation of recent findings. The American Journal of Clinical Nutrition, 98(4), 861-862.
Delaglio, F., Grzesiek, S., Vuister, G. W., Zhu, G., Pfeifer, J., & Bax, A. (1995). NMRPipe: a multidimensional spectral processing system based on UNIX pipes. Journal of biomolecular NMR, 6(3), 277-293.
202
Déprez, S., Brezillon, C., Rabot, S., Philippe, C., Mila, I., Lapierre, C., & Scalbert, A. (2000). Polymeric proanthocyanidins are catabolized by human colonic microflora into low-molecular-weight phenolic acids. The Journal of Nutrition, 130(11), 2733-2738.
Deprez, S., Mila, I., Huneau, J.-F., Tome, D., & Scalbert, A. (2001). Transport of proanthocyanidin dimer, trimer, and polymer across monolayers of human intestinal epithelial Caco-2 cells. Antioxidants and Redox Signaling, 3(6), 957-967.
Deziel, B. A., Patel, K., Neto, C., Gottschall‐Pass, K., & Hurta, R. A. (2010). Proanthocyanidins from the American Cranberry (Vaccinium macrocarpon) inhibit
matrix metalloproteinase‐2 and matrix metalloproteinase‐9 activity in human prostate cancer cells via alterations in multiple cellular signalling pathways. Journal of Cellular Biochemistry, 111(3), 742-754.
Di Martino, P., Agniel, R., David, K., Templer, C., Gaillard, J., Denys, P., & Botto, H. (2006). Reduction of Escherichia coli adherence to uroepithelial bladder cells after consumption of cranberry juice: a double-blind randomized placebo-controlled cross-over trial. World journal of urology, 24(1), 21-27.
Donovan, J. L., Lee, A., Manach, C., Rios, L., Morand, C., Scalbert, A., & Rémésy, C. (2002). Procyanidins are not bioavailable in rats fed a single meal containing a grapeseed extract or the procyanidin dimer B3. British Journal of Nutrition, 87(04), 299-306.
Dunn, W. B., Broadhurst, D. I., Atherton, H. J., Goodacre, R., & Griffin, J. L. (2011). Systems level studies of mammalian metabolomes: the roles of mass spectrometry and nuclear magnetic resonance spectroscopy. Chemical Society Reviews, 40(1), 387-426.
Engemann, A., Hubner, F., Rzeppa, S., & Humpf, H.-U. (2012). Intestinal metabolism of two A-type procyanidins using the pig cecum model: detailed structure elucidation of unknown catabolites with Fourier transform mass spectrometry (FTMS). Journal of Agricultural and Food Chemistry, 60(3), 749-757.
Eriksson, L. (2006). Multi-and megavariate data analysis: MKS Umetrics AB.
Feliciano, R. P., Krueger, C. G., Shanmuganayagam, D., Vestling, M. M., & Reed, J. D. (2012). Deconvolution of matrix-assisted laser desorption/ionization time-of-flight mass spectrometry isotope patterns to determine ratios of A-type to B-type interflavan bonds in cranberry proanthocyanidins. Food Chemistry, 135(3), 1485-1493.
Fihn, S. D. (2003). Acute uncomplicated urinary tract infection in women. New England Journal of Medicine, 349(3), 259-266.
203
Fonville, J. M., Richards, S. E., Barton, R. H., Boulange, C. L., Ebbels, T., Nicholson, J. K., Holmes, E., & Dumas, M. E. (2010). The evolution of partial least squares models and related chemometric approaches in metabonomics and metabolic phenotyping. Journal of Chemometrics, 24(11‐12), 636-649.
Foo, L. Y., Lu, Y., Howell, A. B., & Vorsa, N. (2000). A-Type Proanthocyanidin Trimers from Cranberry that Inhibit Adherence of Uropathogenic P-Fimbriated Escherichia c oli. Journal of Natural Products, 63(9), 1225-1228.
Foxman, B., Geiger, A. M., Palin, K., Gillespie, B., & Koopman, J. S. (1995). First-time urinary tract infection and sexual behavior. Epidemiology, 6(2), 162-168.
Fuleki, T., Pelayo, E., & Palabay, R. B. (1994). Sugar composition of varietal juices produced from fresh and stored apples. Journal of Agricultural and Food Chemistry, 42(6), 1266-1275.
Garcia‐Aloy, M., Llorach, R., Urpi‐Sarda, M., Jáuregui, O., Corella, D., Ruiz‐Canela, M., Salas‐Salvadó, J., Fitó, M., Ros, E., & Estruch, R. (2014). A metabolomics‐driven approach to predict cocoa product consumption by designing a multimetabolite biomarker model in free‐living subjects from the PREDIMED study. Molecular Nutrition & Food Research.
Gika, H. G., Theodoridis, G. A., Wingate, J. E., & Wilson, I. D. (2007). Within-day reproducibility of an HPLC-MS-based method for metabonomic analysis: application to human urine. Journal of Proteome Research, 6(8), 3291-3303.
Giusti, M. M., & Wrolstad, R. E. (2001). Characterization and Measurement of Anthocyanins by UV-Visible Spectroscopy: John Wiley & Sons, Inc.
Gonthier, M.-P., Donovan, J. L., Texier, O., Felgines, C., Remesy, C., & Scalbert, A. (2003). Metabolism of dietary procyanidins in rats. Free Radical Biology and Medicine, 35(8), 837-844.
Graham, S. F., Holscher, C., & Green, B. D. (2014). Metabolic signatures of human Alzheimer’s disease (AD): 1H NMR analysis of the polar metabolome of post-mortem brain tissue. Metabolomics, 10(4), 744-753.
Griffin, J. L. (2006). The Cinderella story of metabolic profiling: does metabolomics get to go to the functional genomics ball? Philosophical Transactions of the Royal Society B: Biological Sciences, 361(1465), 147-161.
Griffiths, L. (1964). Studies on flavonoid metabolism. Identification of the metabolites of (+)-catechin in rat urine. Biochemical Journal, 92(1), 173.
Gu, L., House, S. E., Rooney, L., & Prior, R. L. (2007). Sorghum bran in the diet dose dependently increased the excretion of catechins and microbial-derived phenolic acids in female rats. Journal of Agricultural and Food Chemistry, 55(13), 5326-5334.
204
Gu, L., Kelm, M., Hammerstone, J. F., Beecher, G., Cunningham, D., Vannozzi, S., & Prior, R. L. (2002). Fractionation of polymeric procyanidins from lowbush blueberry and quantification of procyanidins in selected foods with an optimized normal-phase HPLC-MS fluorescent detection method. Journal of Agricultural and Food Chemistry, 50(17), 4852-4860.
Gu, L., Kelm, M. A., Hammerstone, J. F., Beecher, G., Holden, J., Haytowitz, D., Gebhardt, S., & Prior, R. L. (2004). Concentrations of proanthocyanidins in common foods and estimations of normal consumption. The Journal of Nutrition, 134(3), 613-617.
Gu, L., Kelm, M. A., Hammerstone, J. F., Beecher, G., Holden, J., Haytowitz, D., & Prior, R. L. (2003). Screening of foods containing proanthocyanidins and their structural characterization using LC-MS/MS and thiolytic degradation. Journal of Agricultural and Food Chemistry, 51(25), 7513-7521.
Gu, L., Kelm, M. A., Hammerstone, J. F., Zhang, Z., Beecher, G., Holden, J., Haytowitz, D., & Prior, R. L. (2003). Liquid chromatographic/electrospray ionization mass spectrometric studies of proanthocyanidins in foods. Journal of Mass Spectrometry, 38(12), 1272-1280.
Gu, Y., Hurst, W. J., Stuart, D. A., & Lambert, J. D. (2011). Inhibition of key digestive enzymes by cocoa extracts and procyanidins. Journal of Agricultural and Food Chemistry, 59(10), 5305-5311.
Gupta, K., Chou, M., Howell, A., Wobbe, C., Grady, R., & Stapleton, A. (2007). Cranberry products inhibit adherence of p-fimbriated Escherichia coli to primary cultured bladder and vaginal epithelial cells. The Journal of Urology, 177(6), 2357-2360.
Haverkorn, M. J., & Mandigers, J. (1994). Reduction of bacteriuria and pyuria using cranberry juice. JAMA, 272(8), 590-590.
Hawkins, D. M., Basak, S. C., & Mills, D. (2003). Assessing model fit by cross-validation. Journal of Chemical Information and Computer Sciences, 43(2), 579-586.
Heinzmann, S. S., Brown, I. J., Chan, Q., Bictash, M., Dumas, M.-E., Kochhar, S., Stamler, J., Holmes, E., Elliott, P., & Nicholson, J. K. (2010). Metabolic profiling strategy for discovery of nutritional biomarkers: proline betaine as a marker of citrus consumption. The American Journal of Clinical Nutrition, 92(2), 436-443.
Hodgson, J. M., Yee Chan, S., Puddey, I. B., Devine, A., Wattanapenpaiboon, N., Wahlqvist, M. L., Lukito, W., Burke, V., Ward, N. C., & Prince, R. L. (2004). Phenolic acid metabolites as biomarkers for tea-and coffee-derived polyphenol exposure in human subjects. British Journal of Nutrition, 91(02), 301-305.
205
Holt, R. R., Lazarus, S. A., Sullards, M. C., Zhu, Q. Y., Schramm, D. D., Hammerstone, J. F., Fraga, C. G., Schmitz, H. H., & Keen, C. L. (2002). Procyanidin dimer B2 [epicatechin-(4β-8)-epicatechin] in human plasma after the consumption of a flavanol-rich cocoa. The American Journal of Clinical Nutrition, 76(4), 798-804.
Hooton, T. M. (2001). Recurrent urinary tract infection in women. International Journal of Antimicrobial Agents, 17(4), 259-268.
Horai, H., Arita, M., Kanaya, S., Nihei, Y., Ikeda, T., Suwa, K., Ojima, Y., Tanaka, K., Tanaka, S., & Aoshima, K. (2010). MassBank: a public repository for sharing mass spectral data for life sciences. Journal of Mass Spectrometry, 45(7), 703-714.
Howell, A. B. (2007). Bioactive compounds in cranberries and their role in prevention of urinary tract infections. Molecular Nutrition & Food Research, 51(6), 732-737.
Howell, A. B., Botto, H., Combescure, C., Blanc-Potard, A.-B., Gausa, L., Matsumoto, T., Tenke, P., Sotto, A., & Lavigne, J.-P. (2010). Dosage effect on uropathogenic Escherichia coli anti-adhesion activity in urine following consumption of cranberry powder standardized for proanthocyanidin content: a multicentric randomized double blind study. BMC Infectious Diseases, 10(1), 94.
Howell, A. B., Reed, J. D., Krueger, C. G., Winterbottom, R., Cunningham, D. G., & Leahy, M. (2005). A-type cranberry proanthocyanidins and uropathogenic bacterial anti-adhesion activity. Phytochemistry, 66(18), 2281-2291.
Ichiyanagi, T., Shida, Y., Rahman, M. M., Hatano, Y., & Konishi, T. (2006). Bioavailability and tissue distribution of anthocyanins in bilberry (Vaccinium myrtillus L.) extract in rats. Journal of Agricultural and Food Chemistry, 54(18), 6578-6587.
Ito, H., Gonthier, M.-P., Manach, C., Morand, C., Mennen, L., Rémésy, C., & Scalbert, A. (2005). Polyphenol levels in human urine after intake of six different polyphenol-rich beverages. British Journal of Nutrition, 94(4), 500-509.
Jepson, R., Craig, J., & Williams, G. (2013). Cranberry products and prevention of urinary tract infections. JAMA, 310(13), 1395-1396.
Kang, J., Choi, M.-Y., Kang, S., Kwon, H. N., Wen, H., Lee, C. H., Park, M., Wiklund, S., Kim, H. J., & Kwon, S. W. (2008). Application of a 1H nuclear magnetic resonance (NMR) metabolomics approach combined with orthogonal projections to latent structure-discriminant analysis as an efficient tool for discriminating between Korean and Chinese herbal medicines. Journal of Agricultural and Food Chemistry, 56(24), 11589-11595.
Kemsley, E. K., Le Gall, G., Dainty, J. R., Watson, A. D., Harvey, L. J., Tapp, H. S., & Colquhoun, I. J. (2007). Multivariate techniques and their application in nutrition: a metabolomics case study. British Journal of Nutrition, 98(1), 1-14.
206
Kontiokari, T., Sundqvist, K., Nuutinen, M., Pokka, T., Koskela, M., & Uhari, M. (2001). Randomised trial of cranberry-lingonberry juice and Lactobacillus GG drink for the prevention of urinary tract infections in women. BMJ, 322(7302), 1571.
Kresty, L. A., Howell, A. B., & Baird, M. (2011). Cranberry proanthocyanidins mediate growth arrest of lung cancer cells through modulation of gene expression and rapid induction of apoptosis. Molecules, 16(3), 2375-2390.
Krogholm, K. S., Haraldsdóttir, J., Knuthsen, P., & Rasmussen, S. E. (2004). Urinary total flavonoid excretion but not 4-pyridoxic acid or potassium can be used as a biomarker for the intake of fruits and vegetables. The Journal of Nutrition, 134(2), 445-451.
Lee, Y. A., Cho, E. J., Tanaka, T., & Yokozawa, T. (2007). Inhibitory activities of proanthocyanidins from persimmon against oxidative stress and digestive enzymes related to diabetes. Journal of Nutritional Science and Vitaminology, 53(3), 287-292.
Li, S., Sui, Y., Xiao, J., Wu, Q., Hu, B., Xie, B., & Sun, Z. (2013). Absorption and urinary excretion of A-type procyanidin oligomers from< i> Litchi chinensis</i> pericarp in rats by selected ion monitoring liquid chromatography–mass spectrometry. Food Chemistry, 138(2), 1536-1542.
Liang, J., Xu, F., Zhang, Y.-Z., Huang, S., Zang, X.-Y., Zhao, X., Zhang, L., Shang, M.-Y., Yang, D.-H., & Wang, X. (2013). The profiling and identification of the absorbed constituents and metabolites of Paeoniae Radix Rubra decoction in rat plasma and urine by the HPLC–DAD–ESI-IT-TOF-MS n technique: A novel strategy for the systematic screening and identification of absorbed constituents and metabolites from traditional Chinese medicines. Journal of Pharmaceutical and Biomedical Analysis, 83, 108-121.
Liebich, H., & Först, C. (1990). Basic profiles of organic acids in urine. Journal of Chromatography B: Biomedical Sciences and Applications, 525, 1-14.
Lin, H. M., Helsby, N. A., Rowan, D. D., & Ferguson, L. R. (2011). Using metabolomic analysis to understand inflammatory bowel diseases. Inflammatory Bowel Diseases, 17(4), 1021-1029.
Lin, S., Chan, W., Li, J., & Cai, Z. (2010). Liquid chromatography/mass spectrometry for investigating the biochemical effects induced by aristolochic acid in rats: the plasma metabolome. Rapid Communications in Mass Spectrometry, 24(9), 1312-1318.
Lindon, J. C., Holmes, E., & Nicholson, J. K. (2006). Metabonomics techniques and applications to pharmaceutical research & development. Pharmaceutical Research, 23(6), 1075-1088.
207
Liu, H., Liu, H., Wang, W., Khoo, C., Taylor, J., & Gu, L. (2011). Cranberry phytochemicals inhibit glycation of human hemoglobin and serum albumin by scavenging reactive carbonyls. Food & Function, 2(8), 475-482.
Liu, H., Zou, T., Gao, J.-m., & Gu, L. (2013). Depolymerization of cranberry procyanidins using (+)-catechin,(−)-epicatechin, and (−)-epigallocatechin gallate as chain breakers. Food Chemistry, 141(1), 488-494.
Liu, Y., Black, M. A., Caron, L., & Camesano, T. A. (2006). Role of cranberry juice on
molecular‐scale surface characteristics and adhesion behavior of Escherichia coli. Biotechnology and Bioengineering, 93(2), 297-305.
Llorach, R., Garrido, I., Monagas, M., Urpi-Sarda, M., Tulipani, S., Bartolome, B., & Andres-Lacueva, C. (2010). Metabolomics study of human urinary metabolome modifications after intake of almond (Prunus dulcis (Mill.) DA Webb) skin polyphenols. Journal of Proteome Research, 9(11), 5859-5867.
Llorach, R., Urpi-Sarda, M., Jauregui, O., Monagas, M., & Andres-Lacueva, C. (2009). An LC-MS-based metabolomics approach for exploring urinary metabolome modifications after cocoa consumption. Journal of Proteome Research, 8(11), 5060-5068.
Loke, W. M., Jenner, A. M., Proudfoot, J. M., McKinley, A. J., Hodgson, J. M., Halliwell, B., & Croft, K. D. (2009). A metabolite profiling approach to identify biomarkers of flavonoid intake in humans. The Journal of Nutrition, 139(12), 2309-2314.
Manach, C., Hubert, J., Llorach, R., & Scalbert, A. (2009). The complex links between dietary phytochemicals and human health deciphered by metabolomics. Molecular Nutrition & Food Research, 53(10), 1303-1315.
Marchesi, J. R., Holmes, E., Khan, F., Kochhar, S., Scanlan, P., Shanahan, F., Wilson, I. D., & Wang, Y. (2007). Rapid and noninvasive metabonomic characterization of inflammatory bowel disease. Journal of Proteome Research, 6(2), 546-551.
McKay, D. L., Chen, C.-Y. O., Zampariello, C. A., & Blumberg, J. B. (2015). Flavonoids and phenolic acids from cranberry juice are bioavailable and bioactive in healthy older adults. Food Chemistry, 168, 233-240.
McMurdo, M. E., Bissett, L. Y., Price, R. J., Phillips, G., & Crombie, I. K. (2005). Does ingestion of cranberry juice reduce symptomatic urinary tract infections in older people in hospital? A double-blind, placebo-controlled trial. Age and Ageing, 34(3), 256-261.
Mennen, L. I., Sapinho, D., Ito, H., Bertrais, S., Galan, P., Hercberg, S., & Scalbert, A. (2006). Urinary flavonoids and phenolic acids as biomarkers of intake for polyphenol-rich foods. British Journal of Nutrition, 96(01), 191-198.
208
Moffett, J. R., & Namboodiri, M. A. (2003). Tryptophan and the immune response. Immunology and Cell Biology, 81(4), 247-265.
Natsume, M., Osakabe, N., Oyama, M., Sasaki, M., Baba, S., Nakamura, Y., Osawa, T., & Terao, J. (2003). Structures of (−)-epicatechin glucuronide identified from plasma and urine after oral ingestion of (−)-epicatechin: differences between human and rat. Free Radical Biology and Medicine, 34(7), 840-849.
Neveu, V., Perez-Jimenez, J., Vos, F., Crespy, V., Du Chaffaut, L., Mennen, L., Knox, C., Eisner, R., Cruz, J., & Wishart, D. (2010). Phenol-Explorer: an online comprehensive database on polyphenol contents in foods. Database, 2010, bap024.
Nicholson, J. K., Lindon, J. C., & Holmes, E. (1999). 'Metabonomics': understanding the metabolic responses of living systems to pathophysiological stimuli via multivariate statistical analysis of biological NMR spectroscopic data. Xenobiotica, 29(11), 1181-1189.
Ottaviani, J. I., Momma, T. Y., Kuhnle, G. K., Keen, C. L., & Schroeter, H. (2012). Structurally related (−)-epicatechin metabolites in humans: assessment using de novo chemically synthesized authentic standards. Free Radical Biology and Medicine, 52(8), 1403-1412.
Ou, K., & Gu, L. (2013). Absorption and metabolism of proanthocyanidins. Journal of Functional Foods.
Ou, K., & Gu, L. (2014). Absorption and metabolism of proanthocyanidins. Journal of Functional Foods, 7, 43-53.
Ou, K., Percival, S. S., Zou, T., Khoo, C., & Gu, L. (2012). Transport of cranberry A-type procyanidin dimers, trimers, and tetramers across monolayers of human intestinal epithelial Caco-2 cells. Journal of Agricultural and Food Chemistry, 60(6), 1390-1396.
Pluskal, T., Castillo, S., Villar-Briones, A., & Orešič, M. (2010). MZmine 2: modular framework for processing, visualizing, and analyzing mass spectrometry-based molecular profile data. BMC Bioinformatics, 11(1), 395.
Prior, R. L., Rogers, T. R., Khanal, R. C., Wilkes, S. E., Wu, X., & Howard, L. R. (2010). Urinary Excretion of Phenolic Acids in Rats Fed Cranberry†. Journal of Agricultural and Food Chemistry, 58(7), 3940-3949.
Psychogios, N., Hau, D. D., Peng, J., Guo, A. C., Mandal, R., Bouatra, S., Sinelnikov, I., Krishnamurthy, R., Eisner, R., & Gautam, B. (2011). The human serum metabolome. PloS one, 6(2), e16957.
209
Pujos-Guillot, E., Hubert, J., Martin, J.-F., Lyan, B., Quintana, M., Claude, S., Chabanas, B., Rothwell, J. A., Bennetau-Pelissero, C., & Scalbert, A. (2013). Mass spectrometry-based metabolomics for the discovery of biomarkers of fruit and vegetable intake: citrus fruit as a case study. Journal of Proteome Research, 12(4), 1645-1659.
Rechner, A. R., Kuhnle, G., Bremner, P., Hubbard, G. P., Moore, K. P., & Rice-Evans, C. A. (2002). The metabolic fate of dietary polyphenols in humans. Free Radical Biology and Medicine, 33(2), 220-235.
Richelle, M., Tavazzi, I., Enslen, M., & Offord, E. (1999). Plasma kinetics in man of epicatechin from black chocolate. European Journal of Clinical Nutrition, 53(1), 22-26.
Rios, L. Y., Bennett, R. N., Lazarus, S. A., Rémésy, C., Scalbert, A., & Williamson, G. (2002). Cocoa procyanidins are stable during gastric transit in humans. The American Journal of Clinical Nutrition, 76(5), 1106-1110.
Robbins, R. J., Leonczak, J., Li, J., Johnson, J. C., Collins, T., Kwik-Uribe, C., & Schmitz, H. H. (2012). Determination of flavanol and procyanidin (by degree of polymerization 1–10) content of chocolate, cocoa liquors, powder (s), and cocoa flavanol extracts by normal phase high-performance liquid chromatography: collaborative study. Journal of AOAC International, 95(4), 1153-1160.
Rudell, D. R., Mattheis, J. P., & Fellman, J. K. (2005). Evaluation of diphenylamine derivatives in apple peel using gradient reversed-phase liquid chromatography with ultraviolet–visible absorption and atmospheric pressure chemical ionization mass selective detection. Journal of Chromatography A, 1081(2), 202-209.
Rzeppa, S., Bittner, K., Döll, S., Dänicke, S., & Humpf, H. U. (2012). Urinary excretion and metabolism of procyanidins in pigs. Molecular Nutrition & Food Research, 56(4), 653-665.
Salo, J., Uhari, M., Helminen, M., Korppi, M., Nieminen, T., Pokka, T., & Kontiokari, T. (2012). Cranberry juice for the prevention of recurrences of urinary tract infections in children: a randomized placebo-controlled trial. Clinical Infectious Diseases, 54(3), 340-346.
Sánchez-Patán, F., Cueva, C., Monagas, M., Walton, G. E., Gibson, G. R., Martín-Álvarez, P. J., Moreno-Arribas, M. V., & Bartolomé, B. (2012). Gut microbial catabolism of grape seed flavan-3-ols by human faecal microbiota. Targetted analysis of precursor compounds, intermediate metabolites and end-products. Food Chemistry, 131(1), 337-347.
Sano, A., Yamakoshi, J., Tokutake, S., Tobe, K., Kubota, Y., & Kikuchi, M. (2003). Procyanidin B1 is detected in human serum after intake of proanthocyanidin-rich grape seed extract. Bioscience, Biotechnology, and Biochemistry, 67(5), 1140-1143.
210
Setchell, K. D., Brown, N. M., Desai, P. B., Zimmer-Nechimias, L., Wolfe, B., Jakate, A. S., Creutzinger, V., & Heubi, J. E. (2003). Bioavailability, disposition, and dose-response effects of soy isoflavones when consumed by healthy women at physiologically typical dietary intakes. The Journal of Nutrition, 133(4), 1027-1035.
Simonetti, P., Gardana, C., & Pietta, P. (2001). Plasma levels of caffeic acid and antioxidant status after red wine intake. Journal of Agricultural and Food Chemistry, 49(12), 5964-5968.
Singleton, V., & Rossi Jr, J. (1965). Colorimetry of total phenolics with phosphomolybdic-phosphotungstic acid reagents. American Journal of Enology and Viticulture, 16(3), 144.
Sobota, A. (1984). Inhibition of bacterial adherence by cranberry juice: potential use for the treatment of urinary tract infections. The Journal of Urology, 131(5), 1013-1016.
Spencer, J. P., Abd El Mohsen, M. M., Minihane, A.-M., & Mathers, J. C. (2008). Biomarkers of the intake of dietary polyphenols: strengths, limitations and application in nutrition research. British Journal of Nutrition, 99(01), 12-22.
Spencer, J. P., Chaudry, F., Pannala, A. S., Srai, S. K., Debnam, E., & Rice-Evans, C. (2000). Decomposition of cocoa procyanidins in the gastric milieu. Biochemical and Biophysical Research Communications, 272(1), 236-241.
Spencer, J. P., Schroeter, H., Shenoy, B., S Srai, S. K., Debnam, E. S., & Rice-Evans, C. (2001). Epicatechin is the primary bioavailable form of the procyanidin dimers B2 and B5 after transfer across the small intestine. Biochemical and Biophysical Research Communications, 285(3), 588-593.
Stothers, L. (2002). A randomized trial to evaluate effectiveness and cost effectiveness of naturopathic cranberry products as prophylaxis against urinary tract infection in women. Canadian Journal of Urology, 9, 1558-1562.
Stoupi, S., Williamson, G., Drynan, J. W., Barron, D., & Clifford, M. N. (2010). A
comparison of the in vitro biotransformation of (–)‐epicatechin and procyanidin B2 by human faecal microbiota. Molecular Nutrition & Food Research, 54(6), 747-759.
Stoupi, S., Williamson, G., Viton, F., Barron, D., King, L. J., Brown, J. E., & Clifford, M. N. (2010). In vivo bioavailability, absorption, excretion, and pharmacokinetics of [14C] procyanidin B2 in male rats. Drug Metabolism and Disposition, 38(2), 287-291.
211
Törrönen, R., McDougall, G. J., Dobson, G., Stewart, D., Hellström, J., Mattila, P., Pihlava, J.-M., Koskela, A., & Karjalainen, R. (2012). Fortification of blackcurrant juice with crowberry: Impact on polyphenol composition, urinary phenolic metabolites, and postprandial glycemic response in healthy subjects. Journal of Functional Foods, 4(4), 746-756.
Tsang, C., Auger, C., Mullen, W., Bornet, A., Rouanet, J.-M., Crozier, A., & Teissedre, P.-L. (2005). The absorption, metabolism and excretion of flavan-3-ols and procyanidins following the ingestion of a grape seed extract by rats. British Journal of Nutrition, 94(02), 170-181.
Walker, E. B., Barney, D. P., Mickelsen, J. N., WALYON, R., & Mickelsen, R. (1997). Cranberry concentrate: UTI prophylaxis. Journal of Family Practice, 45(2), 167-168.
Walsh, M. C., Brennan, L., Pujos-Guillot, E., Sébédio, J.-L., Scalbert, A., Fagan, A., Higgins, D. G., & Gibney, M. J. (2007). Influence of acute phytochemical intake on human urinary metabolomic profiles. The American Journal of Clinical Nutrition, 86(6), 1687-1693.
Westerhuis, J. A., Hoefsloot, H. C., Smit, S., Vis, D. J., Smilde, A. K., van Velzen, E. J., van Duijnhoven, J. P., & van Dorsten, F. A. (2008). Assessment of PLSDA cross validation. Metabolomics, 4(1), 81-89.
Wiklund, S., Johansson, E., Sjöström, L., Mellerowicz, E. J., Edlund, U., Shockcor, J. P., Gottfries, J., Moritz, T., & Trygg, J. (2008). Visualization of GC/TOF-MS-based metabolomics data for identification of biochemically interesting compounds using OPLS class models. Analytical Chemistry, 80(1), 115-122.
Wilson, T., Singh, A. P., Vorsa, N., Goettl, C. D., Kittleson, K. M., Roe, C. M., Kastello, G. M., & Ragsdale, F. R. (2008). Human glycemic response and phenolic content of unsweetened cranberry juice. Journal of Medicinal Food, 11(1), 46-54.
Wishart, D. S., Tzur, D., Knox, C., Eisner, R., Guo, A. C., Young, N., Cheng, D., Jewell, K., Arndt, D., & Sawhney, S. (2007). HMDB: the human metabolome database. Nucleic Acids Research, 35(suppl 1), D521-D526.
Young, J. F., Nielsen, S. E., Haraldsdóttir, J., Daneshvar, B., Lauridsen, S. T., Knuthsen, P., Crozier, A., Sandström, B., & Dragsted, L. O. (1999). Effect of fruit juice intake on urinary quercetin excretion and biomarkers of antioxidative status. The American Journal of Clinical Nutrition, 69(1), 87-94.
Zampariello, C. A., McKay, D. L., Dolnikowski, G., Blumberg, J., & Chen, C. (2012). Determination of cranberry proanthocyanidin A2 in human plasma and urine using LC-MS/MS. The FASEB Journal, 26, 124.128.
212
Zhang, Y., Song, T. T., Cunnick, J. E., Murphy, P. A., & Hendrich, S. (1999). Daidzein and genistein glucuronides in vitro are weakly estrogenic and activate human natural killer cells at nutritionally relevant concentrations. The journal of nutrition, 129(2), 399-405.
213
BIOGRAPHICAL SKETCH
Haiyan Liu was from Xi’an, China. She received her B.S. degree in food safety
and security from China Agricultural University in 2008. She was admitted into a master
program in the Food Science and Human Nutrition Department at the University of
Florida in 2009. She received her M.S. degree in 2011. Afterwards she continued her
study and joined the food science doctoral program. Haiyan received her Ph.D. degree
from the University of Florida in December 2015. She published four research papers
and presented her research at national conferences in 2009, 2011, 2014 and 2015.