The Food metabolome
Post on 27-Jan-2015
161 Views
Preview:
DESCRIPTION
Transcript
Glasgow July 2013
THE FOOD METABOLOMETHE FOOD METABOLOME
C. ManachHuman Nutrition Unit, INRA Clermont‐Ferrand, France
Review paper in preparation
COMPLEXITY OF NUTRITIONAL EXPOSURESCOMPLEXITY OF NUTRITIONAL EXPOSURES
Nutrients Non nutrientsNatural
Non-nutrients
DIET
Non nutrientsMan-made
Non-nutrients
Contaminants,Additives,
Agrochemicals,…
Polyphenols,Carotenoids,Phytosterols,
…
Proteins,Carbohydrates,Lipids, Vitamins
Minerals
EXPOSOME
Dietary habits (food choices, shopping places, cooking habits…)
Lifestyle, Environment
Glasgow, July 2013
« We eat other metabolomes » (D. Wishart)
Genetics, Epigenetics,Age, Gender,Microbiota, Physiological status, Medical history
Digestion
Xenobioticmetabolism
Microbialmetabolism
Elimination
Storage
INDIVIDUAL METABOLIC CAPACITYINDIVIDUAL METABOLIC CAPACITY
Glasgow, July 2013
Food MetabolomeAll the metabolites that derive from
the digestion and metabolism of food components
Dietary habits
Metaboliccapacity
THE FOOD METABOLOME DEFINITIONTHE FOOD METABOLOME DEFINITION
Health outcomesClinicaltrialsClinicaltrials Cohorts
Glasgow, July 2013
Metabolomes of foods =Food metabolome ?« Food chemicalome »?
FOOD METABOLOME APPLICATIONSFOOD METABOLOME APPLICATIONS
Age, sex,BMI,Lifestyle,exercice…
GenotypeEnterotype
Food metabolome analysis
Glasgow, July 2013
Segmentation of Poor/High absorbers& metabolizers
New metabolitesNew potentialfood bioactives
Public healthResearch Diet-genotype-health relationships
Monitoring impact of recommendations or policiesMedicine Personalized nutrition
Food intakeNutritional exposures
Dietaryquestionnaires
Dietary assessment
FOOD METABOLOME APPLICATIONSFOOD METABOLOME APPLICATIONS
Food intakeNutritional exposures
Dietaryquestionnaires
Food metabolome analysis
Glasgow, July 2013
Biomarkers of compliance for intervention studies
Validation of dietary questionnaires with biomarkers for a few representative foods
Subject stratification in dietary patterns
Assessment of recent or long-term consumption of a range of foods
Comprehensive and detailed assessment of individual nutritional exposures
Dietary assessment
Usually analyzed using a range of distinct targeted methods(GC‐MS, LC‐UV, LC‐MS in pos or neg mode, NMR, …)
FOOD METABOLOME: AN ANALYTICAL CHALLENGEFOOD METABOLOME: AN ANALYTICAL CHALLENGE
Food metabolome = at least 25,000 compounds
CarbohydratesProteinsLipidsVitaminsMineralsFlavonoidsPhenolic acidsCarotenoidsPhytosterolsChlorophyllsAlkaloidsArtificial colorsFlavoring additives Maillard reaction productsFood contaminants…
Larg
e ra
nge
of c
once
ntra
tions
mM
nM
FOOD METABOLOME: AN ANALYTICAL CHALLENGEFOOD METABOLOME: AN ANALYTICAL CHALLENGE
Food metabolome = at least 25,000 compounds
CarbohydratesProteinsLipidsVitaminsMineralsFlavonoidsPhenolic acidsCarotenoidsPhytosterolsChlorophyllsAlkaloidsArtificial colorsFlavoring additives Maillard reaction productsFood contaminants…
Hydrolysis,Oxidation,Reduction,Methylation,Dehydrogenation,Sulfation, Glucuronidation, Acetylation,Glutathione conjugation,…
Host and microbial biotransformations
Many unknowns
Non‐targeted metabolomics (LC‐MS, GC‐MS, NMR, …)
Larg
e ra
nge
of c
once
ntra
tions
mM
nM
Saito et al., Annu Rev Plant Biol, 2010
TOWARD A MULTIPLATFORM UNTARGETED ANALYSIS OF THE FOOD METABOLOMETOWARD A MULTIPLATFORM UNTARGETED ANALYSIS OF THE FOOD METABOLOME
Glasgow, July 2013
Saito et al., Annu Rev Plant Biol, 2010
TOWARD A MULTIPLATFORM UNTARGETED ANALYSIS OF THE FOOD METABOLOMETOWARD A MULTIPLATFORM UNTARGETED ANALYSIS OF THE FOOD METABOLOME
Glasgow, July 2013
Same approach for the foodmetabolome analysis
1‐Map the analytical coverage of Food Metabolome chemical space
by various platforms
2‐ Optimize methods with wideand complementary coverages
& Define SOPs
FIRST STUDIES TO DISCOVERNEW BIOMARKERS OF FOOD INTAKE
FIRST STUDIES TO DISCOVERNEW BIOMARKERS OF FOOD INTAKE
Glasgow, July 2013
Nootkatone-diol
Limonene-diolProline betaine
DISCOVERY OF BIOMARKERS OF FOOD INTAKEDISCOVERY OF BIOMARKERS OF FOOD INTAKE
m/z 312.21
m/z 144.06CO group
OJ group
m/z 232.09
020
0040
0060
00
020
0040
0060
0080
0010
000
1200
00
500
1000
1500
2000
2500
3000
020
040
060
080
010
0012
00
010
020
030
040
050
050
010
0015
00
050
010
0015
00
050
100
150
200
250
300
CO OR CO ORCO OR
CO OR CO ORCO OR
CO OR CO OR
One month controlled intervention study with orange juice
12 volunteers 500 ml/d Orange juice / Control drinkUsual dietCross-over study, 24h urine D30LC-ESI-Qtof in positive mode
& 105 significant ions (ANOVA BH)
Score plot PLSDA
Pujos‐Guillot et al., J Proteome Res, 2013
Glasgow, July 2013
Hesperetin Naringenin
HCA heatmap
DISCOVERY OF BIOMARKERS OF FOOD INTAKEDISCOVERY OF BIOMARKERS OF FOOD INTAKEShort‐term intervention studies
Citrus Cruciferous vegetablesCocoa drink Almonds Coffee Nuts Red wine Grape juice Whole rye grain Black tea Green tea Milk Soy Salmon Rapsberry Tomato
125 candidate biomarkers(75%= phytochemical metabolites)
16 foods studied Mostly controlled intervention studies
(4-61 subjects)60% acute / 40% medium-term studies
(4 days-12 weeks)>90% used urine samples
(Spots, 24hr urines, or kinetics)
NMR (8 studies), LC-MS (13 studies) or GC-MS (4 studies), including multiplatform analyses (5 studies)
Glasgow, July 2013
Scalbert et al., in preparation
WHAT DID WE LEARN FROM THE FIRST STUDIES?WHAT DID WE LEARN FROM THE FIRST STUDIES?
Urine metabolome well reflects recent food intake,plasma may better reflect long-term dietary habits
Dozens of metabolite had increased level in urine after acute food challengeBut many remain unidentified
Phytochemical metabolites are key discriminants for plant food intake
More putative biomarkers are detected with LC-MS compared to GC-MS or NMR
A small number of subjects (8-20) seems sufficient for biomarker discovery
A standardized diet before the food challenge limits unwanted variation in acute studies and help detecting metabolic changes
Glasgow, July 2013
NEW BIOMARKERS REVEALED BY METABOLOMICSNEW BIOMARKERS REVEALED BY METABOLOMICS
Proline betaine for Citrus Many candidates require further validation
Glasgow, July 2013
Common to many organisms, Not specific for a
given food?Some exceptions
May not besystematically found in the target food, but onlyin certain populations and/or geographiclocations
Host met.Microbiota metabolites
Host met.Microbiota metabolites
Host met.Host met.Microbiota metabolitesMicrobiota metabolites
The natural non‐nutrients and their host metabolitesare more likely to constitute specific
biomarkers of food intake
FOOD METABOLOMICS FOR DISCOVERY OF PLANT FOOD INTAKE BIOMARKERSFOOD METABOLOMICS FOR DISCOVERY OF PLANT FOOD INTAKE BIOMARKERS
Six 24h recalls (1994-2002) +FFQ 2007-2009
Selection of low and high consumers for 20 plant foods or food groups
PhenoMeNEp ALIA 2011‐2013
CorrelationsDistribution of foodconsumption
Coll. S. Hercberg, P. Galan, M. TouvierUREN, Inserm/INRA/CNAM/Paris 13
SU.VI.MAX2 sub‐cohort (210 subjects)
UPLC‐ESI‐Qtof‐MS (mode pos & neg)Morning spot urines
Good discrimination for most foods, especially those consumedfrequently & rich in phytochemicals
Caffeine metabolitesTrigonellineHippuric acidAtractyligenin glucCyclo‐(Isoleu‐Pro)…
Cohort studies
Glasgow, July 2013Fillâtre et al., in preparation
FOOD METABOLOMICS FOR DISCOVERY OF PLANT FOOD INTAKE BIOMARKERSFOOD METABOLOMICS FOR DISCOVERY OF PLANT FOOD INTAKE BIOMARKERS
Six 24h recalls (1994-2002) +FFQ 2007-2009
Selection of low and high consumers for 20 plant foods or food groups
PhenoMeNEp ALIA 2011‐2013
CorrelationsDistribution of foodconsumption
Coll. S. Hercberg, P. Galan, M. TouvierUREN, Inserm/INRA/CNAM/Paris 13
SU.VI.MAX2 sub‐cohort (210 subjects)
UPLC‐ESI‐Qtof‐MS (mode pos & neg)Morning spot urines
Good discrimination for most foods, especially those consumedfrequently & rich in phytochemicals
Caffeine metabolitesTrigonellineHippuric acidAtractyligenin glucCyclo‐(Isoleu‐Pro)…
Cohort studies
68 subjects from the GrainMark study, stratified for consumption of 38 food groups / 4 FFQs over 3 months
Same conclusion in Lloyd et al., AJCN 2013
Glasgow, July 2013
Conduct similar studies in various populationswith different dietary habits
Fillâtre et al., in preparation
COORGR
6-12h
12h-night
0-6h
1st urine d0
6-12h
12h-night
0-6h1st urine d0
1st urine d1
1st urine d1
-25
-20
-15
-10
-5
0
5
10
15
20
25
-34 -32 -30 -28 -26 -24 -22 -20 -18 -16 -14 -12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34
t[2]
t[1]
6
F4
20
F5
6N
F30F5
6N20F5
F2 F420
F5
F2
F5
F4
F3
F1
F2
F4
F3
F1 F5
F2
F4
F3
F1 F5
F2
F4
F3
F1F5
F2
F4
F3
F1
F5
F2
F4
F3
F5
F2
F4
F3
F1
F5
F2
F4
F3
F1
F5
F1
-20
-18
-16
-14
-12
-10
-8
-6
-4
-2
0
2
4
6
8
10
12
14
-15 -14 -13 -12 -11 -10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
t[2]
t[1]
12816H
13214H
13245H
13280H
17774H
13374H
13413H13435H
13457H
13862H13890H
13934H
13950H
15244H
15445H
15817H
15836H
15893H
15935H
16355H
16375H
16405H
16472H
16518H16701H
16725H
16772H16774H
16895H
17159H
17190H
17209H
13911H
17316H
17328H
17469H17477H
17544H17580H
17870H
11864L12521L
12585L
12675L
12756L
13144L
13150L
13204L
15230L13483L
13735L
13766L
13910L
13962L
14274L
14517L
15398L
15420L
15554L15656L
15884L
16387L16467L16543L
16550L
16751L
16886L16947L
16987L
17006L
17049L
17291L
17396L
17472L
17536L
17735L17753L
17877L
17934L
A
ACUTE CONTROLLEDINTERVENTION STUDY COHORT STUDY
Number of discriminant ions
Level of control of the diet
1-MO INTERVENTION STUDY
COMPARISON OF STUDY DESIGNSCOMPARISON OF STUDY DESIGNS
603 significant ions 105 significant ions 19 significant ions
MetaboliteStabilityPharmacokinetics
Lack of specificity
Heterogeneity of the population
Risk of false discovery (Correlations between foods)Validation in
intervention study
Glasgow, July 2013
Pujos‐Guillot et al., J Proteome Res, 2013
Biomarkers of intakeusable in epidemiology
Comprehensive phenotyping of nutritional exposures
COHORT STUDYMEDIUM-TERM STUDY
COMPARISON OF STUDY DESIGNSCOMPARISON OF STUDY DESIGNS
Biomarkers of compliance
ACUTE CONTROLLEDINTERVENTION STUDY
Glasgow, July 2013
DISCOVERY PHASE
Different validations ?ControlledControlled
interventions studies
Cohort studies
3‐days weighed food diaries K‐means Cluster Analysis (33 food groups)
160 Irish subjects
O’Sullivan et al., AJCN 2011
BIOMARKERS OF DIETARY PATTERNSBIOMARKERS OF DIETARY PATTERNS
PLS-DA of 1H-NMR urine data of dietary cluster 1 ( ) compared with cluster 3 ( )
Glasgow, July 2013
It is more difficult to find biomarkers of dietary patterns than biomarkers of food intake
Questionnaire data (Times/wk)
BiomarkerConcentration
SCORING OF FOOD INTAKE BIOMARKERS TO DETERMINE DIETARY PATTERNS SCORING OF FOOD INTAKE BIOMARKERS TO DETERMINE DIETARY PATTERNS
Fish TMAO +?Meat 1-Methyl-Histidine + AnserineMilkCheeseCitrus Proline Betaine +?BerriesApple Phloretin +?Cruciferous veg. S-Methyl-L-cysteine sulfoxide +?Tomato Lycopene +?PotatoRice and pastaWhite BreadWhole bread Alkylresorcinols + ?Chocolate Theobromine + ?ConfectionariesRed wine Resveratrol metab. + ?Coffee Atractyligenin+1-methylxanthineTea 4-O-Methylgallic acid +?
Priority list to be defined with epidemiologists
Glasgow, July 2013
Kits for dietary pattern determination?
List completed in a few years time if we work in a concerted action?
FOOD METABOLOME DATA REPOSITORYFOOD METABOLOME DATA REPOSITORY
Food metabolome studies
Controlled study B
Cohort study A
Controlled study C
Cohort study D
Food Metabolome
Datarepository
Study MetadataMethod descriptionIdentified markers
Annotated raw dataNon-identified markers
Glasgow, July 2013
Candidate biomarkersidentified in Study A
Correlation with coffee intake in all available
studies?
dbNP?Metabolights?
Reporting standardsData formats
Fiehn et al. Metabolomics 2007Metabolomic standards Initiative
BIOMARKER VALIDATION: PROLINE BETAINE AS AN EXAMPLEBIOMARKER VALIDATION: PROLINE BETAINE AS AN EXAMPLE
Heinzmann et al., 2010, Lloyd et al., 2011&2013, Pujos‐Guillot et al., 2013, May et al., 2013
Glasgow, July 2013
Heinzmann et al., 2010; de Zwart et al., 2003; Slow et al., 2005
Found almost exclusively in citrus fruits, with dominance in orange
Associated with citrus intake in 3 acute studies, 2 medium-term interventions , 3 cohort studies
Detected with NMR, LC-QTof, FIE-MSIn morning spot urines, 24hr urine & post-prandial urine kinetics
250 ml orange juice challengeHeinzmann et al., AJCN 2010
Pharmacokinetics data
Training set n=220Validation set n=279
« Excellent biomarker »
ROC curve
Heinzmann et al., AJCN 2010
Validation in INTERMAP-UK cohort
BIOMARKER VALIDATIONBIOMARKER VALIDATION
Glasgow, July 2013
Define a procedure /workflow for validation of biomarkers of intake
Define a validation mark?
Identify the factors affecting the biomarker concentration in biofluids & the content of its precursor in the food source
D. Newly discoveredC. With analytical validation including kinetics and dose-response
relationship in the sample type of interestB. Confirmed in a controlled dietary intervention as well as in cross-
sectional studiesThe number of validating studies could be indicated in a code:
Ex: Proline betaine = B8 ?(found in 3 cohorts and 5 intervention studies)
A. Confirmed to be in accordance with other marker(s) for the same food(s)Adapted from Lars Dragsted’s poster
IDENTIFICATION OF UNKNOWNS IN LC‐MS, THE MAIN BOTTLENECK
IDENTIFICATION OF UNKNOWNS IN LC‐MS, THE MAIN BOTTLENECK
Glasgow, July 2013
IDENTIFICATION OF UNKNOWNS, THE MAIN BOTTLENECKIDENTIFICATION OF UNKNOWNS, THE MAIN BOTTLENECK
Identification workflow (LC-MS)
Find the molecular ion and itsrelated fragments & adducts(MSClust, Camera, MZedDB, …)
Get exact mass with high accuracy(Orbitrap, FT-ICR…)
Elemental formula(Golden rules)
Query compound databases to obtain hypotheses
Analyze standard or compare mass fragmentation in librairies of spectra or literature
Glasgow, July 2013
HMDB
Compound databases
HMDB
Librairies of spectra
In‐house librairies
Definitive or tentative identification
IDENTIFICATION OF UNKNOWNS, THE MAIN BOTTLENECKIDENTIFICATION OF UNKNOWNS, THE MAIN BOTTLENECK
Identification workflow (LC-MS)
Find the molecular ion and itsrelated fragments & adducts(MSClust, Camera, MZedDB, …)
Get exact mass with high accuracy(Orbitrap, FT-ICR…)
Elemental formula(Golden rules)
Query compound databases to obtain hypotheses
Analyze standard or compare mass fragmentation in librairies of spectra or literature
Glasgow, July 2013Definitive or tentative identification
Why?
Host & microbial metabolitesof non-nutrient compounds :
Unknown or not yetincluded in databases
Their standards are lacking
Their mass fragmentation spectra are unknown
It often does not work!!!!
ENRICH DATABASES TO FACILITATE IDENTIFICATIONENRICH DATABASES TO FACILITATE IDENTIFICATION
Food composition databases
30,000 natural food components & additives
7,500 compounds
28,000 compounds, 888 foods
500 polyphenols100 food components
8,500 phytochemicals
Quantitative data on food contents
HMDB
Use in silico prediction tools when no information is available on the metabolic fate of a given compound
Literature
Glasgow, July 2013
Add the known metabolites on non-nutrients in compound databases
Rothwell et al., Database, 2012
HMDB
IN SILICO PREDICTION OF METABOLISMIN SILICO PREDICTION OF METABOLISM
Developed for the pharmaceutical industry. Validation required for dietary compoundsNo tool for prediction of microbial metabolism
Meteor Nexus (Lhasa Ltd) is probably the most powerful tool (477 biotransformations), but costs 5,000€/year
To enrich online and in-house databases with predicted metabolitesTo enrich online and in-house databases with predicted metabolites
To support putative identifications from spectral dataTo support putative identifications from spectral data
META, Metabolexpert, Metabolizer, MetaPrint2D-React, MetaSite, Meteor nexus, SyGMa, TIMES T’jollyn et al., 2011,
Piechota et al., 2013
Tools
Glasgow, July 2013
IN SILICO PREDICTION OF METABOLISM: METEOR NEXUS (LHASA LTD)IN SILICO PREDICTION OF METABOLISM: METEOR NEXUS (LHASA LTD)
Glasgow, July 2013
IN SILICO PREDICTION OF METABOLISM: METEOR NEXUS (LHASA LTD)IN SILICO PREDICTION OF METABOLISM: METEOR NEXUS (LHASA LTD)
Glasgow, July 2013
Tendency to overpredict, Good sensitivity / known metabolites
Good prediction for polyphenols (>80%)Currrently tested for alkaloids and terpenes
Can be used to built in‐house databases for selected foods from knowledge of their
composition
Pujos‐Guillot et al., 2013
Rothwell et al., subm. 2013
Helpful for identification of candidate biomarkers for citrus and coffee intake
Kahweol oxideglucuronide
Limonene 8,9‐diol glucuronideNootkatone 13,14‐diol glucuronide
PHYTOHUBPHYTOHUBAn online database for dietary phytochemicals and their human metabolites
Glasgow, July 2013
(www.phytohub.eu)
Dietary sources
Known metabolites
Predicted metabolites
Spectral data
Physico-chemical data
Links to other databases
1,000 dietary phytochemicals Literature
Literature
expert knowledge on biotransformations
LiteratureExperimental data
Structure developed by INRA, website in collaboration with
Giacomoni et al., in preparation
Should be launched by the end of 2013
An online database for dietary phytochemicals and their human metabolites
Glasgow, July 2013
Dietary sources
Known metabolites
Predicted metabolites
Spectral data
Physico-chemical data
Links to other databases
1,000 dietary phytochemicals
What are the phytochemical precursors & metabolites matching with a monoisotopic mass ?
What are the phytochemical metabolites expected in biological fluids after consumption of a given food?
Open for collaborations for filling and curating the database
PHYTOHUBPHYTOHUB (www.phytohub.eu)
UV spectraEnzymatic reactions (hydrolysis of conjugates,…)H/D exchange experimentsMSn spectral treesIn silico fragmentation Peak collection & preconcentration + NMR, GC-MS…
All the new tools proposed by the metabolomics communityMetFrag, Metfusion, MetiTree, HighChem Mass frontier, mzCloud …
Glasgow, July 2013
Experimental structural elucidation strategies using:
EFFICIENT TOOLS & METHODS FOR STRUCTURAL ELUCIDATIONEFFICIENT TOOLS & METHODS FOR STRUCTURAL ELUCIDATION
Develop projects to synthetize and distribute standards for non commercially available metabolites
Expand in-house libraries of spectra
CONCLUSION: NETWORKING IS ESSENTIAL NOWCONCLUSION: NETWORKING IS ESSENTIAL NOW
To provide rapid access and training to new tools and methodologies
To define current good practices from ring-tests on shared datasets& develop shared pipeline for dietary studies, data analysis and compound identification
To develop of a metabolism prediction tool customized for food compounds
To organize data sharing with a Food metabolome data repository
To avoid redundancy in research and work at commonly defined priorities
To develop a concerted action for biomarker validation
Glasgow, July 2013
Yoann FILLATREJoe ROTHWELLMercedes QUINTANAMathieu RAMBEAU Christine MORANDDragan MILENKOVICBlandine COMTE
JRU1019‐ Human Nutrition Unit
Mathilde TOUVIERLeopold FEZEUNathalie ARNAULTPilar GALANSerge HERCBERG
UREN, Inserm/INRA/CNAM/Paris 13
Charlotte JOLYBernard LYANJean‐François MARTINFrank GIACOMONIEstelle PUJOS‐GUILLOT
THANK YOU VERY MUCH FOR YOUR ATTENTIONTHANK YOU VERY MUCH FOR YOUR ATTENTION
Craig KNOXRoman EISNER
Glasgow, July 2013
top related