Metabolomic Applications and Tools for Dietary and

Can Metabolomics Be Used for Short and Long Term Dietary and Nutritional Assessment?

Preliminary Results From Proof of Concept Projects

Western Human Nutrition Research Center430 West Health Sciences Dr., Davis, CA 95695

Joint Agency Workshop on Dietary Intakeand Nutritional Assessment

John W. Newman

March 31, 2009

Ph2-1Ph2-3

Ph6-5 Ph7

Ph8-2Ph9-3

Ph9-6

Ph9-4

Ph2-2

Ph2-4

Ph9-1

Ph9-2

The Metabolome Is Derived From Both Endogenous and Exogenous Sources

• A variety of studies are underway evaluating this question.

• The MEDE Study – Aberystwyth Univ. / Newcastle Univ., United Kingdom

• Plasma Biomarkers of Whole vs. Refined Grain – USDA, WHNRC, Davis CA.

• Phytochemical Biomarkers of Plant Intake – INRA, Clermont-Ferrand, France

• Circulating Markers of “Fish Oil” Consumption – USDA, WHNRC, Davis CA.

Can Metabolomics Aid Dietary Assessment?

The MEDE StudyMEtabolomics to characterise Dietary Exposure

NuGO and SYSDIET Metabolomics workshop 24 February 2009

Newcastle UniversityPr. John C. MathersDr. Gaëlle FavéMr. Graham HaroldDr. Long Xie

Aberystwyth UniversityPr. John H. DraperDr. Manfred E. BeckmannDr. Shaobo ZhouDr. WanChang LinMs. Kathleen Tailliart

GOAL: Evaluate metabolomics as an alternative tool for dietary assessment

Approach: Controlled Experimental Test Meals

Pre-test dayNo alcoholMinimal physical activityStandardised evening meal

Test day Blood - up to 30 mLUrine - total void(s)Saliva - alternative biofluid

Fasted state Fed state:2, 4, 6 and 8h later

Study design : 2 test days per volunteer, designated A and B

Standardised evening mealChicken breast

with carrots, peas and potatoesChocolate éclair

Still mineral water

‘Metabolomically’ neutral, balanced, attractive

Test breakfastOrange juice

Tea with sugar and milkCroissant

Corn flakes with milk


Individual Metabolite Fingerprints Are Reproducible

-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 x 10-3-1

-0.5

0

0.5

1

1.5

2x 10 -3

DF1 (Tw: 49.52)

M2P-N-210-510 PC-LDA

DF2

(Tw

: 21.

95)

ID-213AID-213BID-214AID-214BID-215AID-215BID-216AID-216BID-217AID-217BID-218AID-218BID-219AID-219BID-220AID-220BID-221AID-221BID-222AID-222BID-223AID-223BID-204AID-204B

• Plasma analysed by positive mode FIE-MS (m/z 100-220) veryHIGH THROUGHPUT

• Discriminating individual by test day

• Satisfactory grouping of A & B visits


Metabolite Fingerprints are Dynamic and Respond to Diet

• Plasma analysed by positive mode FIE-MS (m/z 100-220)

• Discriminated by sample class

•Good state discrimination

• fasting vs. fed

• breakfast vs. lunch

fasting

After breakfast

After lunch

-2 -1.5 -1 -0.5 0 0.5 1 x 10-3-1

-0.8

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

0.8

1x 10-3 M2P-P-100-220 PC-LDA

DF1 (Tw: 6.13)

DF2

(Tw

: 3.3

2)

0A0B2A2B4A4B6A6B8A8B


Changes in the Postprandial Plasma Metabolome with a High vs. Low Glycemic Diet Challenge


Plasma Biomarkers of Grain Consumption

John NewmanTheresa Pedersen

Dmitry GrapovAlison KeenanWilliam Keyes

Nancy KeimWilliam Horn

Sridevi Krishnan

Subjects:20 overweight women (BMI: 25-29.9). 3 day cross-over to diets rich in whole or refined grain

Sample Collection:8hr fasting, 0.5, 2, 3.5, 6, 8 hrs postprandial post-test meal plasma.

Analyses:Clinical lipids, insulin/glucose, leptin, calorimetery LC/MS-based Metabolic Fingerprinting

Plasma Biomarkers of Grain Consumption: Study Design

Diet Glycemic Index Alters Plasma Biomarkers

• Short duration consumption of whole vs. refined grain diets shift postprandial metabolome profiles

1158174196

704

784

328

282293

250 457

457

657

759

787

761

783808

839840

866903

707

759815

783

285

385

-0.2

-0.1

0.0

0.1

0.2

150

200

250

300

350

400

450

500

550

600

650

700

750

800

850

900

950

1000

1050

1100

1150

1200

1250

1300

1350

1400

1450

1500

m/z (Da)

PC1

Load

ings

PC1 Loadings± 1s± 2s

-30000

0-30000 0

PC1 Scores (49% of Variance)

PC2

Scor

es HG-Fasting

LG-Fasting

HG-3.5hr

LG-3.5hr

Reduced postprandial phospholipids with low glycemic meal

Increased postprandial unknowns ([M+H]+ m/z 250, 283, 457) with low glycemic meal

150 – 500 m/z

700 – 900 m/z

NuGO and SYSDIET Metabolomics workshop 2009

Characterizing the Food metabolome to discover new biomarkers of food intake

a proof-of-concept study on Citrus

Hubert J, Llorach R, Martin JF, Pujos-Guillot E, Claude S,Bérard A, Bennetau-Pelissero C, Morand C, Scalbert A, Manach C.

UMR 1019, Unité de Nutrition Humaine Centre Clermont-Ferrand-Theix, France

Daidzein glucuronide

Bergaptolglucuronide

Uroterpenol glucuronide

Hydroxystachydrine

Dihydrogenistein glucuronide

Genistein glucuronide

Terpeneglucuronide

Glyciteinglucuronide

Bergaptol glucuronide

Genistein diglucuronide

PLS-DA

SOYIsoflavones13 potential biomarkers

CITRUSFuranocoumarins

& Terpenes9 potential biomarkers

• New markers of intake revealed by metabolomics

CRUCIFEROUSGlucosinolates

2 potential biomarkers

Markers of food / phytochemical intake in urine

Manach C. (IINRA France) - NuGO and SYSDIET Metabolomics workshop 2009

Correlations between urinary polyphenols and intake of their major food sources

Markers of polyphenol intake in urine

Pearson correlations (p < 0.05)

Ito et al. (2005) BJNMennen et al. (2006) BJNMennen et al. (2007) EJCN

OH OH

OH O

OH

O

OMe

OH

OH

OH

O

H

OH

OH

OH

COOH


Urine metabolome : Principle Components Analysis

PCA Scores plot: all samples, ions p<0.05 ANOVA / beverage

PCA trajectories show the temporal evolution of the urine metabolome

-20

-10

0

10

20

-20 -10 0 10 20 30

t[2]

t[1]

MJ0

MJ0MJ0

MJ0

0-60-6

0-6

0-6

6-12h

6-12h

6-12h

6-12h12-N

12-N

12-N12-N

MJ1

MJ1

MJ1

MJ1

MJ0MJ0MJ0

MJ0

0-6

0-6

0-6

0-6

6-12h

6-12h

6-12h6-12h

12-N

12-N

12-N

12-N

MJ1MJ1

MJ1

MJ1

MJ0MJ0MJ0

MJ0

0-60-60-60-6

6-12h6-12h

6-12h

6-12h

12-N

12-N

12-N12-N

MJ1

MJ1

MJ1

MJ1

0-6h

6-12h

12h-night

1st void d1

1st void d0

OrangeGrapefruitControl

1st void d01st void d1

0-6h

6-12h

12h-night


Volunteers are discriminated by type of beverage

Phenol-ExplorerPhenol-Explorer527 polyphenols in 591 food items

Isoflavones: 128 metabolites

Food composition tables- Nutrients- Phytochemicals

Phytochemical metabolitedatabase

Ph2-1Ph2-3

Ph6-5 Ph7

Ph8-2Ph9-3

Ph9-6

Ph9-4

Ph2-2

Ph2-4

Ph9-1

Ph9-2

Peak Annotation & Food Metabolome Databases Are Needed


6586 metabolites1193 identified in blood472 identified in urine

1932 food components(first pass from FDA Food Additives List)

Needed: Open Access Metabolite/Food Component Databases

UNIVERSITY OF ALBERTA

Oliver Fiehn UC Davis.‘We need annotated and searchable spectral libraries for unknown identification.’

Dr. Davis Wishart, U. Alberta, CanadaDeveloping a database cataloging all of the small molecules within the human body, annotated with observed concentrations.

Agilent Technologies Fiehn GC/MS

Metabolomics RTL Library

• We Need Databases of Metabolites With Searchable Meta Data Which Should Include

• Food Sources

• Spectral Data

UNIVERSITY OF CALIFORNIA, DAVIS

~1050 searchable spectra from 700

compounds

Red Blood Cell Phospholipids: Reporters of compliance and lipid intake


Circulating Biomarkers of Lipid Intake



Charles StephensenGertrud SchusterPatrice Armstrong Alina WhetstoneXiaowen Jiang

Controlled Feeding Shifts RBC Phospholipids in Mice

• Study: Impact of DHA vs. EPA feeding on asthma symptom severity

• Dietary Regimen: 4wks ad lib. with 4 distinct diets.

• Results: RBC phospholipid fatty acids profiles reflected dietary lipids.

Dr. Gertrud Schuster,UC Davis,Dept Nutrition

Human Fatty Acids Profiles of Red Blood Cell Phospholipids

-0.2

0

0.2

0.4

0.6

0.8

1

1.2

C20

:3n3

C18

:1n7

/1n6

C20

:3n6

C18

:0C

22:4

n6C

18:1

n9/1

n7t

C18

:2n6

C16

:0C

20:4

n6C

22:5

n6C

18:2

C16

:1n7

C20

:4n3

C18

:3n6

C22

:6n3

C22

:5n3

C18

:3n3

C20

:5n3

PC1

Load

ings

-2

0

2

4

-5 0 5

PC1 Scores(72% of Variance)

PC2

Scor

es

PlaceboFish Oil

• Study: Impact of Fish Oil on Macrophage Stimulation in 5-LOX polymorphism

• Dietary Regimen: 6wks 0f 6g/day fish oil or placebo.

• Results:– RBC phospholipid are

generally responsive.

– Changes in EPA most dramatic.

Variability in Human RBC Membrane Fatty Acid

Data courtesy of G. Shearer and W. Harris, Univ. South Dakota

•Normalized RBC FAs to arachidonate, then greatest variance observed in:

•Long Chain n3 PUFAs

•trans-fatty acids

•Sub-populations observed based on LC n6/n3 ratios.

•Diet vs. Metabolism?26

%16

%62

%34

%30

%

49%

44%

39%

20%

85%

71%

28%

140%

Histogram of All>20n6/>20n3:Freq. dist. (histogram)

0 1 2 3 4 5 6 7 8 9

-100

102030405060708090

100110120130

Bin Center

Histogram of All>20n6/>20n3:Freq. dist. (histogram)

0 1 2 3 4 5 6 7 8 9

-100

102030405060708090

100110120130

Bin Center

RBC Phospholipid Profile Distribution in Multivariate Space

• With high trans-fat subjects removed, primary separation by:– SAT/MUFA– n6/n3 PUFAs

DHA:SAT Quartiles

• As RBC PUFAs decrease – Elongation Index Increases– LCn6/n3 PUFAs increase (1:1 threshold)

• LCn6/n3 PUFAs increase, the Elongation Index increases.

Interactions between Red Cell LC-PUFA & n6:n3

Limited marker of “relative intake”

Plasma Oxygenated PUFA Metabolites:Novel reporters of compliance and dietary intake


Circulating Biomarkers of Lipid Intake



Gregory ShearerWilliam Harris

PUFAEicosanoids

Leukotrienes

CannabinoidsVanilloids

Steroids

Cholesterol

LC-ω3-PUFA Feeding Increases LC-ω3 Oxygenated Metabolites in Human Plasma

Hydroxy-, epoxy-, and dihydroxy-LCω3-PUFAs in the plasma of subjects (n=10) after 4wks of 4g/day EPA/DHA supplementation.

The Magnitude of Plasma Oxylipid Changes Depend on Baseline Concentrations

The fold-change in plasma hydroxy-PUFAs were inversely proportional to initial concentration.

Reduction in arachidonate metabolites are only observed when basal levels are “high”.

Potential indication of an metabolic “Optimum”.

Human subjects (n=10),

• Preliminary results from Proof-of-Principle studies suggest that Metabolomics can be applied to Diet Assessments.

– An individual’s plasma metabolomic fingerprint is stable but can be altered by diet in the short term.

– Urinary phytochemicals correlate with intake of their major food sources over a 24 hour period.

– Expanded, searchable databases of food composition are needed. • Annotated with

– Dietary source and Biological Significance (for the biologist)– Searchable spectral data (for the chemist)

– Plasma and membrane lipids provide indications of dietary exposure, but homeostatic balances make these risky quantitative intake markers.

– Standardized and validated test diets and nutritional challenges are needed.

– The question is of global interest and international collaborations should be part of the solution.

Summary: Metabolomics for Dietary Intake Assessment

Bruce Kristal

Theresa Pedersen

Augustine Scalbert

David Wishart

John Mathers

John Newman

John DraperLorraine Brennan

Ben van Ommen

Manfred Beckmann

Oliver Fiehn

Susan Woperies

Lars Dragsted

Claudine Manach

TOOLS AND METHODS FOR MASS SPECTROMETRY METABOLOMICS In Nutrition

Mass-spectrometry-based metabolomics in nutrition – Current limitations and recommendations. Scalbert, A., L. Brennan, O. Fiehn, T. Henkemeir, B. Kristal, B. van Ommen, E. Pujos-Guillot, E. Verheij, D. Wshart, S. Wopereies. In Press.

CONTRIBUTORS & Colleagues in Nutritional Metabolomics

Cross correlation matrix for plasma LC-ω3-PUFAs (n=20, r>0.68, p<0.001).

EPA and DHA metabolites were tightly associated

Prostaglandins showed positive associations with epoxides of arachidonate, linoleate, and alpha-linolenate.

Changes in Plasma LC-n3 Oxylipids are Tightly Correlated

Documents

Metabolomic Applications and Tools for Dietary and