Using Metabolomics to Quantify Health Elaine Holmes, PhD, FRSC, FSB Biomolecular Medicine, Imperial College, U.K.

Transcription

1 8 th September 2011 NUGO, Wageningen Using Metabolomics to Quantify Health Elaine Holmes, PhD, FRSC, FSB Biomolecular Medicine, Imperial College, U.K.

2 Summary Background to technology and adaptations for modelling human data Characterizing metabolic consequences of food interventions Considering the mammalian ecosystem and complexity Defining biomarkers of health Future directions of metabolic phenotyping within the nutritional arena

3 Metabonomic Analysis Strategies GC-MS MS Strategy Biofluids and Extracts NMR UPLC-MS Biofluids NMR Strategy 1 H, 13 C, 31 P Intact Tissues (MAS) GC-MS LC-MS n 2 D methods LC-MS n -NMR LC-NMR Etc. Chemometrics and Disease Modelling Other stuff Diffusion analysis C E Metabolite Identification/ Quantitation and Interactions

4 Omics data tend to be highly multivariate, may not be annotated completely, can have missing values, usually require aligning and standardizing in multiple dimensions and are all structured differently NMR transcriptomic LC-MS 2D gel electrophoresis

5 Example of PCA Analysis ( a) 2 Spectrum t[2] Spectrum 2 ( b) p[2] t[1] Spectrum 3 Spectrum 4 Spectrum p[1] Spectrum Spectra scores loadings spectra

6 ACETOACETATE ACETONE B-HYDROXYBUTYRATE

7 A food intervention study to find food biomarkers via metabolic profiling. Compliance; quantitative intake; product quality / differences apple, orange, grapes, grapefruit 03/12/09 Silke Heinzmann

8 NOT ALL PARTS OF THE SPECTRA YIELD EQUAL INFORMATION AND DIFFERENT SPECTRAL AREAS REFLECT DIFFERENT SOURCES OF VARIANCE TO DIFFERENT DEGREES. Inter-individual variation Diet-related variation 0.05 ppm Variance explained (R 2 ) Variance predicted (Q 2 ) X32 3 W121 V226 W323 W394 Z74 X11 Y144 V359 Z262 Z104 Z122 V354 X355 W663 W520 W150 X131 X234 V122 V241 W155 X210 W574 W173 Z25 Z149 W396 V350 W196 Z173 W208 Z30 Z180 W420 W85 Y146 Z293 V379 Z77 X285 V356 V230 W270 W661 W199 V30 X172 Y4 X157 X462 V93 X318 V24 Z27 Z162 V270 Z166 X114 Z46 Z256 W527 W236 Z62 Y196 W296 W471 W561 X326 W334 V50 V116 X13 V330 V102 V65 X223 Y219 X463 W170 Z228 V206 V244 V222 Z93 V367 V31 W223 V382 Z151 V87 Y189 V103 W243 W409 X16 Y13 V394 Y232 W373 W256 X148 X26 X339 Y77 V37 Z325 W565 V201 W668 X295 V119 Z55 W140 W415 W537 W572 Z169 X283 W12 Y88 W219 V209 Y148 W276 X398 W641 W622 V235 W649 Y233 V128 V204 W579 X275 W364 W78 X118 X212 similarity dendrogram Silke Heinzmann / Marc Dumas 0.1 ppm

9 Elucidation of potential food biomarkers from metabolic profiles with pattern recognition Are biomarkers of healthy food intake equivalent to biomarkers of health? ACUTE VS CHRONIC Increased excretion of: trimethylamine-n-oxide taurine methylhistidine creatine choline carnitine Partial least squares discriminant analysis Increased excretion of: proline betaine r=0.71 tartaric acid hippuric acid 03/12/09 Silke Heinzmann

10 Features of a good food biomarker: Detectable (relatively high concentration, low probability of overlap, more than one spectral feature); Long half life; not multiply metabolized proline betaine 3-methylhistidine TMAO creatinine creatinine creatine glycine hippurate urea citrate trigonelline formate aminoacids SCFA, ketoacids 03/12/09 Silke Heinzmann

11 1 H NMR analysis to quantify proline betaine in fruit juices and fruits Analysis of different fruits and fruit juices in a fully relaxed 1 H NMR experiment Juice/fruit description Proline betaine (mg/l) Juice/fruit description Proline betaine (mg/l) Citrus fruit juice (authentic) Orange juice, from concentrate 1316 Orange juice 1189 Orange juice, freshly squeezed 1062 Grapefruit juice 766 Citrus fruit juice (synthetic) Orange soft drink 216 Orange squash (20% orange juice from concentrate) 201 Orange squash 75 Citrus fruit Orange 786 Lime 651 Satsuma 486 Lemon 240 Other fruit juice Pineapple juice 57 Red grape and raspberry juice 46 Pomegranate and blueberry juice 18 Peach, mango, passionfruit juice 17 Apple juice 14 Blackcurrant juice 12 Other fruit Kiwi 66 Grape 51 Melon 34 Banana 28 Strawberry 22 Pear 14 Apricot 10 03/12/09 Silke Heinzmann

12 Proline betaine excretion in individual volunteers Kinetics study (N=6): orange juice consumption and urine sampling [hours] [ppm] 03/12/09 Silke Heinzmann

13 Establishing the relationship between proline betaine excretion and citrus consumption in population studies: Validation of biomarkers Quantify proline betaine concentrations in 24-hr urine samples Ascertain whether the recorded intake of citrus fruit matches the dietary questionnaires data p-value = 1.2x10-24 p-value = 8.8x /12/09 Silke Heinzmann

14 Good predictivity of the biomarker in a free-living population Receiver operating characteristics curve Training Set: AUC: 92.3% 90.6% specificity 86.2% sensitivity Test Set: AUC: 93.5% 92.3% specificity 80.6% sensitivity 03/12/09 Silke Heinzmann

15 Significant differences in nutrient intake of citrus consumers and non-consumers Variable P value Energy, kcal/24-h 0.8% n.s. Total fat, % kcal -9.4% < Total SFA, % kcal -10.1% < Total MFA, % kcal -9.9% < Total PFA, % kcal -10.1% < Omega-3 FA, % kcal -4.0% 0.01 Omega-6 FA, % kcal -11.3% < Trans FA, % kcal -11.8% < Cholesterol, mg/1,000 kcal -11.0% < Keys Score -8.1% < Total carbohydrate, % kcal 8.1% < Starch, % kcal 0.0% 0.95 Total sugars, % kcal 16.7% < Fructose, % kcal 44.4% < Galactose, % kcal 20.0% <0.001 Glucose, % kcal 39.5% < Lactose, % kcal 3.7% n.s. Maltose, % kcal -12.5% n.s. Sucrose, % kcal 2.0% n.s. Fibre, g/1,000 kcal 11.9% < Variable Total protein, % kcal P value n.s. Vegetable protein, % kcal 7.4% < Animal protein, % kcal -4.9% <0.001 Vitamin A, IU/1,000 kcal 29.3% < Beta-carotene, µg/1,000 kcal 35.5% < Retinol, µg/1,000 kcal n.s. Thiamine, mg/1,000 kcal 8.3% < Riboflavin, mg/1,000 kcal 4.5% <0.01 Niacin, mg/1,000 kcal n.s. Pantoth. acid, mg/1,000 kcal 8.3% < Vitamin B6, mg/1,000 kcal 11.0% < Vitamin B12, µg/1,000 kcal n.s. Vitamin C, mg/1,000 kcal 126.4% < Vitamin E, mg/1,000 kcal -1.3% 9.1% 1.8% 0.0% -2.2% n.s. Folic acid, µg/1,000 kcal 18.2% < Proline betaine excretion is correlated with: Healthy eating patterns Low BMI & BP Variable Non-consumers > consumers (stat. sign.) Consumers > non-consumers (stat. sign.) Not significant P value Urinary sodium, mmol/24-h -3.2% 0.05 Urinary potassium, mmol/24-h 16.9% < Magnesium, mg/1,000 kcal 8.8% < Calcium, mg/1,000 kcal 4.6% <0.01 Phosphorus, mg/1,000 kcal 2.1% <0.05 Iron, mg/1,000 kcal 7.4% < Copper, mg/1,000 kcal 11.1% < Selenium, µg/1,000 kcal 14-day alcohol, g/24-h n.s. n.s. Caffeine, mg/1,000 kcal -29.7% < Body mass index, kg/m 2-3.8% < Systolic BP, mm Hg -1.4% <0.01 Diastolic BP, mm Hg -1.5% -10.8% 0.3% n.s. Education, years 9.0% < /12/09 Silke Heinzmann

16 The INTERMAP Population Study 2 x 24 h urine samples collected 3-6 weeks apart from 17 centres spanning Peoples Republic of China (PRC), Japan, US and UK (4,680 people = ca 10,000 samples including quality control split samples). NMR spectra were acquired for each sample 4 visits were made by each person Measurements: blood pressure - seated (2 x for 4 visits) Height (2 visits) weight (2 visits) daily alcohol over previous 7 days questionnaire (2 visits) Dietary intake (including vitamin/mineral supplements) recorded using 24 hr recall method (4 visits) Demographic and medical history questionnaire (1 visit) Investigating the effects of nutrition on adverse population blood pressure levels via metabolic profiling. Study Design

17

18 Large Scale Metabonomic Screening of Human Populations: Identifying Outliers Type 2 Diabetes Cystinuria/lysinuria Normal Onion Skin Approach

19 DIFFERENT PATTERNS OF VARIATION IN HUMAN METABOTYPE

20

21 Correlation between dietary input, metabonomic profiles and ion exchange data that discriminate normal vs clinically obese

22 Mapping and Understanding The Integrated System Response to Interventions. Can we measure lifestyle interventions metabolically- and if so what do they mean?

23 Porcine model of weaning diet and probiotic to explore shaping of the metabolome, immune system and microflora

24 Differential spectral patterns associated with weaning diet and probiotic intervention.

25 Correlation map showing relationship between immune and metabolic parameters.

26 Diet & Prebiotic

27 Bariatric Surgery in a Rat model Jia Li, Nigel Gooderham Collaborators: Prof. Darzi, Mr. Athanasiou and Mr. Ashrafian Dr Marchesi Groups: Sham (N=18) Roux-en-Y gastric bypass (N=18) Time points: Surgery & Weight loss Pre-op, 2, 4, 6 and 8 weeks post-op Samples: Urine and faeces Methods: NMR-based metabolic profiling UPLC-MS analysis on faecal bile acids 454 Sequencing system for microbial composition analysis

28 Bariatric Surgery: RYGB in non-obese rats TYosc TYosc A C sham Tcv RYGB B D FA 8.5 urine PAG Cre TMAO 8-week post RYGB Cre IS PG IS HA PAG HA PG AV PS 5.2 GT AP Suc AP HP FM Crn Crn HP 8-week post sham control ppm 8-week post RYGB Suc ET Ura Gly MT GABA Tyr PT TMA MA PT DE Asp Asp OS 8-week post sham control ppm faeces 0.8 Correlation Coefficients (a.u.) Correlation Coefficients (a.u.) 0

29 Bariatric Surgery: 454 sequence data A Sham 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Rat 1 Rat 2 Rat 3 Rat 4 Rat 5 Rat 6 w2 w8 w2 w8 w2 w8 w2 w8 w2 w8 w2 w8 Firmicutes p (sham vs. RYGB) w2 w *** RYGB 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Rat 7 Rat 8 Rat 9 Rat 10 Rat 11 Rat 12 w2 w8 w2 w8 w2 w8 w2 w8 w2 w8 w2 w8 Bacteroidetes Proteobacteria Actinobacteria * <0.0001*** B week 2 week 8 p (sham vs. RYGB) Sham Actinobacteria w w Bacilli Bacteroidetes * Clostridia 0.012** *** Erysipelotrichi RYGB Gammaproteobacteria Others 0.036* _ <0.0001*** _

30 Phylogenetic relationships in RYGB/sham operated rats GAKP9UA03F8FRPS06 Gammaproteobacteria 1 GAKP9UA03FU57NB11 Betaproteobacteria GAKP9UA03F3DRVB05 Gammaproteobacteria 1 GAKP9UA03GYTSWS06 Gammaproteobacteria 1 GAKP9UA03GFQI9S06 Gammaproteobacteria 1 GAKP9UA03HH8X3S06 Gammaproteobacteria 1 GAKP9UA03HD1XQB07 Gammaproteobacteria 1 GAKP9UA03FNKVSB08 Gammaproteobacteria 1 GAKP9UA03GGDEDB08 Gammaproteobacteria 1 GAKP9UA03G9WZOS06 Gammaproteobacteria 1 GAKP9UA03GLPRBB04 Gammaproteobacteria 1 GAKP9UA03FR0AZB07 Gammaproteobacteria 1 GAKP9UA03GD24WB10 Gammaproteobacteria 1 GAKP9UA03GIKIRB05 Gammaproteobacteria 1 GAKP9UA03GU7OMB10 Gammaproteobacteria 1 GAKP9UA03F8872B04 Gammaproteobacteria 1 GAKP9UA03FXNIJB04 Gammaproteobacteria GAKP9UA03GVPI7B09 Gammaproteobacteria 1 GAKP9UA03F9ZQ1S1 1 Clostridiales GAKP9UA03FSX2DB04 Clostridiales GAKP9UA03GDHMZS1 1 Clostridiales GAKP9UA03GWEZLS01 Clostridiales GAKP9UA03F80K8S04 Clostridia GAKP9UA03GZ4ZYB02 Clostridiales GAKP9UA03G9JSNB08 Clostridiales GAKP9UA03G6VZIS07 Clostridiales GAKP9UA03GUH0GS01 Clostridiales GAKP9UA03GGUUCB09 Clostridia GAKP9UA03F6T2AS07 Clostridiales GAKP9UA03F84BQS02 Clostridiales GAKP9UA03GO9OOS05 Clostridiales GAKP9UA03FMNJVS05 Clostridiales GAKP9UA03GVQGXS08 Clostridiales GAKP9UA03FYO81S02 Clostridiales GAKP9UA03G8ZR0S03 Clostridiales GAKP9UA03FZY1GS02 Clostridiales GAKP9UA03FRHGUS05 Clostridiales GAKP9UA03GKFM4S08 Clostridiales GAKP9UA03G9PHXS05 Clostridiales GAKP9UA03G8OB0B01 Clostridiales GAKP9UA03GBN5CS09 Clostridiales GAKP9UA03GZAH9S01 Clostridiales GAKP9UA03GGB2BS1 1 Clostridiales GAKP9UA03GS7DKS05 Clostridiales GAKP9UA03FL2QLS08 Clostridiales GAKP9UA03GIFUXS08 Clostridiales GAKP9UA03GC6NQS01 Clostridiales GAKP9UA03GKDC0S08 Clostridiales GAKP9UA03GTIS0S09 Clostridiales GAKP9UA03FYZBRS09 Clostridiales GAKP9UA03FTSPXS08 Clostridiales GAKP9UA03F99NXS08 Clostridiales GAKP9UA03GQ8FNS1 1 Clostridiales GAKP9UA03GQ9I3S05 Clostridiales GAKP9UA03GOM51S05 Clostridiales GAKP9UA03FZNC0S08 Clostridiales GAKP9UA03G2JOES01 Clostridiales GAKP9UA03GF4BGS02 Clostridiales GAKP9UA03FXMW1S03 Clostridiales GAKP9UA03G10G1S02 Clostridiales GAKP9UA03GIC96S03 Clostridiales GAKP9UA03HB52MS03 Bacillales Mollicutes GAKP9UA03FOJ6KS06 CMN et rel. GAKP9UA03HB7D4S06 CMN et rel. GAKP9UA03GEJEEB01Actinomycetaceae-Bifidobacteriaceae GAKP9UA03FHVJPS10Actinomycetaceae-Bifidobacteriaceae GAKP9UA03FKKW0B04Actinomycetaceae-Bifidobacteriaceae GAKP9UA03GEABUB12 Bacillales Mollicutes GAKP9UA03FV28WB12 Bacillales Mollicutes GAKP9UA03GTFF3S04 Bacillales Mollicutes GAKP9UA03G19YYB05 Bacillales Mollicutes GAKP9UA03GR60WB03 Bacillales Mollicutes GAKP9UA03FRQG8B03 Bacillales Mollicutes GAKP9UA03GEPE8B02 Bacillales Mollicutes GAKP9UA03FZU02B04 Clostridiales GAKP9UA03GAEWKB01 Clostridiales GAKP9UA03FOY31B12 Clostridiales GAKP9UA03GHY2QB04 Clostridiales GAKP9UA03F7Q9LS04 Bacteroidetes GAKP9UA03GW9V0S05 Bacteroidetes GAKP9UA03FOVCES04 Bacteroidetes GAKP9UA03HCLC6B09 Bacteroidetes GAKP9UA03GJTCBS1 1 Bacteroidetes GAKP9UA03G1ZCKS12 Bacteroidetes GAKP9UA03FZ72FB01 Bacteroidetes GAKP9UA03FYW48B07 Bacteroidetes GAKP9UA03GYC7KS02 Bacteroidetes GAKP9UA03FS94GB01 Bacteroidetes GAKP9UA03HD6UIB10 Bacteroidetes GAKP9UA03FOJHHB10 Bacteroidetes GAKP9UA03GYDIQB10 Bacteroidetes GAKP9UA03GAI96B05 Bacteroidetes GAKP9UA03GX3IBS09 Bacteroidetes GAKP9UA03HERBWB10 Bacteroidetes GAKP9UA03GS2ORS02 Bacteroidetes GAKP9UA03HAR3OB02 Bacteroidetes GAKP9UA03FJ3MGB03 Bacteroidetes GAKP9UA03FYX5RB02 Bacteroidetes GAKP9UA03FUFS7S08 Bacteroidetes GAKP9UA03F81S1S10 Bacteroidetes GAKP9UA03GXJA1S10 Bacteroidetes GAKP9UA03FNH8RS10 Bacteroidetes GAKP9UA03GPHK3S10 Bacteroidetes GAKP9UA03G79A2S10 Bacteroidetes GAKP9UA03GPQDAS10 Bacteroidetes GAKP9UA03FWIGES10 Bacteroidetes GAKP9UA03GOB4QS12 Bacteroidetes GAKP9UA03GFXRUS03 Bacteroidetes GAKP9UA03G1I8TS07 Bacteroidetes GAKP9UA03GJSZGS02 Bacteroidetes GAKP9UA03G0PEYS07 Bacteroidetes GAKP9UA03GJPA3B05 Bacteroidetes GAKP9UA03FPG52S03 Bacteroidetes GAKP9UA03FMCZUS02 Bacteroidetes GAKP9UA03HI8ZCS03 Bacteroidetes GAKP9UA03HA949S01 Bacteroidetes GAKP9UA03GQSFGB08 Bacteroidetes GAKP9UA03FKFGBB08 Bacteroidetes GAKP9UA03HHEVGS08 Bacteroidetes GAKP9UA03GQBB6B02 Bacteroidetes GAKP9UA03GYNHCS04 Bacteroidetes GAKP9UA03GMYA6S01 Bacteroidetes GAKP9UA03G96MSS1 1 Bacteroidetes GAKP9UA03HB45MS01 Bacteroidetes GAKP9UA03FR05ZB02 Bacteroidetes GAKP9UA03GDHAXB1 1 Bacteroidetes GAKP9UA03FKA9BB1 1 Bacteroidetes GAKP9UA03FQTFTB1 1 Bacteroidetes GAKP9UA03HGWQPB1 1 Bacteroidetes Numbers of 454 reads Legend: Treatment group RYGB SHAM Taxonomy: Clostridia Proteobacteria Bacteroidetes * 0

31 The Roux-en-Y gastric bypass (RYGB) operation is a bariatric procedure that achieves its physiological benefits through B.R.A.V.E effects: Bile flow alteration Reduction of gastric size Anatomical gut rearrangement and altered flow of nutrients Vagal manipulation Enteric gut hormone modulation Ashrafian et al Obes Rev. Sep 29 Summary (1) A shift from Firmicutes to Proteobacteria, especially gammaproteobacteria post RYGB. (2) Upregulation of TCA cycle indicates increased energy expenditure. (3) Increased biogenesis of urinary p- cresol and related compounds and host-microbial co-metabolites. (4) A shift from protein degradation to putrefaction

32

33 Determination of the relative suspension growth as a measure of cytotoxicity of fecal water extracts following treatment of L5178Y mouse lymphoblastoid cells for 24 h (mean ± S.E.M of 6 independent samples) Relative suspension growth *** *** NC RYGB-W2 RYGB-W8 Sham-W2 Sham-W8 EMS

34 0.06 O-PLS regression analyses of fecal water and urine extracts against relative suspension growth values obtained from a 24h treatment of L5178Y cells with sham or RYGB operated rat faecal extracts (week 2 and week 8) A cell growth Fecal water tyrosine aspartate covariances tyramine uracil putrescine GABA tyramine GABA putrescine glycine ppm B cell growth Urine 2-oxoglutarate 2-oxoglutarate 0.1 covariances formate IS PAG p-hydroxyphenylacetate p-cresyl glucuronide PAG creatine p-cresyl sulf p-cresyl glu 5-aminovalerate ppm

35 Scatter plots of normalised bacterial levels and relative suspension growth of L5178Y mouse lymphoblastoid cells exposed to fecal water extracts sham 2 wk sham 8 wk RYGB 2 wk RYGB 8 wk

36 Early influences on the development of the gut microbiome Infection Delivery Method Antibiotics Breast vs Bottle Antipyretics - Calpol Weaning Age Exposure to Pets Weaning Strategy Travel Hygiene

37

38 Differential biomarkers associated with preterm birth PCA scores and loadings plots based on urine NMR profiles showing metabolic differentiation of young adults (18-25 yrs) in preterm vs normal birth t[2] Term Preterm b t[1] Neena Modi, Mathew Hyde, Jim Parkinson

39

40 Concluding Remarks 1. Advances in metabolic profiling methods allow its application in a wide variety of biomedical fields. 2. Statistical spectroscopic & chemometric analysis of spectroscopic and related data allow biomarker structure, pathway elucidation and mechanistic insights. 3. Can identify biomarkers of nutrient intake, quantify, and monitor effects on health in animals and humans. 4. Variation in human populations need not be an insurmountable obstacle and can even carry valuable information Personalized Medicine 5. The metabolic looking glass provides a systems biology window on an organism -provides a framework for stratification and

41 ACKNOWLEDGEMENTS Pediatrics Nina Modi Anisha Wijeyeskera Mathew Hyde James Parkinson Northwestern University/ Jeremiah Stamler Martha Daviglus AstraZeneca Ian Wilson Chinese Academy of Science, Wuhan Yulan Wang Imperial College Jeremy Nicholson John Lindon Nigel Gooderham Jia Li Ivan Yap Jonathan Swann Will Edmands Silke Heinzmann Claire Merrifield Cardiff University Julian Marchesi FUNDERS Nestle, NHS, NIH, MRC, AstraZeneca, EU FP7