Applying NMR- and LC/MS-BasedMetabolite Profiling to Predictive

and Investigative Toxicology

Lois Lehman-McKeeman, Ph.D.Discovery Toxicology

Bristol-Myers Squibb Co.

Determining Toxicant Effects at DifferentLevels of Biomolecular Organization

Metabolomics: Phenotype• Altered gene expression (constitutive or induced)• Environmental or dietary influence

– Contribution of intestinal microbiota• Biomarkers of toxicity or disease

– Causally-related to toxicity or result of toxicity

Gene Expression

Proteins MetabolitesMetabolomicsTranscriptomics Proteomics

Integrating Omics Data for HazardIdentification and Risk Assessment

RISK ASSESSMENT

Dose-responseExposure Analysis

Cross-Species AnalysisMechanism(s)

HAZARDIDENTIFICATION

Evaluating ToxicityPredicting Toxicity

Developing HypothesesInvestigating Mechanisms

Application of NMR and MS Methods inMetabolomics

Robertson et al. (Toxicol. Sci. 2011, in press)

Increasing use; uHPLC, GCWorkhorse; matureRapid, non-destructive

Status

Some available; varies withconditions

Databases available(more needed)

Annotation

Minimal; millions of ionsYesDeconvolution required

Overlappingpeaks

Ionization variable; extractionand chromatographic conditions

Yes: protonMinimal sample prep

ConsistencyMuch more than NMRLow (10-5 M)Sensitivity

Requires standardYesQuantitativeMS-basedNMRAttribute

Application of Metabolomics toUnderstand Phenotypic Changes

• Genetic Basis– Male and female mice distinguished by urinary

excretion of trimethylamine and trimethylamine-oxide

– FMO3 only expressed in females: TMAO>>TMA– Males: TMA>>>TMAO

NC H 3

C H 3C H 3

H O NC H 3

C H 3C H 3

+ NCH3

CH3CH3

OGutFlora

FMO3

Characterization of TransporterKnockout Models: Oatps

Oatp1a1 Oatp1a4

WildtypeOatp1a1-/-Oatp1a4-/-

050

100150200250300

E217G E3S TC Digoxin MPP+ CSA

Subs

trat

e U

ptak

e (p

mol

/mg/

min

)

+/- -/-

521 bp

273 bp476 bp

232 bp+/- -/-

Urinary Metabolite Profiles inOatp Null Mice

Serum LC-MS: Bile acidsmuricholic, deoxycholic

– Increased in nulls

Urinary NMRA: Hippurate

– Increased in nullsB, C:phenylpropionic acids

– Decreased in nulls

A B C

Wildtype

Oatp1a1-/-

Oatp1a4-/-

7.5 5.0 2.5 0.0ppm

A B C

7.90 7.80 7.70 7.60 7.50 7.50 7.40 7.30 7.20 7.102.90 2.80 2.70 2.60 2.50

Reciprocal Relationship betweenHippurate and Chlorogenic Acids

Wikoff W R et al. PNAS 2009;106:3698-3703

Distribution and quantity of gut anaerobes increased throughout entire gi tract

Integrating Metabonomic andTranscriptomic Data

Is it possible to integrate omics data to definecausal relationships?– Time course of gene expression changes relative to

effects on metabolites– Integrating global metabonomic outome with single

organ transcriptional data– Static and dynamic effects “merged”

• Multiple time points for metabonomic samples and singletime point for gene expression

• Urine sample collected over 24 hr with gene expressiondetermined only at the end of collection

Phenobarbital: Hepatic Transcriptionaland Urinary Metabonomic Data in Rats

• Phenobarbital: widely usedin transcriptional profilingstudies– Cytochrome P450 enzymes– Phase II enzymes and

xenobiotic transporters– Nuclear receptor activation

and gene expression– Hepatic transcription factors– Cell cycle regulation and cell

proliferation

Toxicol. Sci. 2002 67:219-231

Urinary NMR Metabolomic Profileafter Phenobarbital Treatment

Ascorbic Acid

GulonicAcid

Urinary Metabolomic Profile: MicrosomalEnzyme Inducers

GAAA

Control

GA

AA

DMP 904; 1 Day

DMP 904; 5 Day

PB; 1 Day

PB; 5 Day

Hepatic Ascorbic Acid SyntheticPathway

FEBS Journal Vol. 274: 1-22

Hepatic Transcriptional Profiles InformMetabolite Excretion

HO

HO

H

HHO

CH2OH

C

H

H

O

O

CHO

OHH

HHO

OHH

OHH

COO-

COO-

HHO

HHO

OHH

HHO

CH2OH

OHO

HO

H

HHO

CH2OH

C

O1: Glucuronate Reductase

2: Gulonolactonase

3: L-Gulonolactone Oxidase

+ 5.2 - 1.7

After 5 days of dosing; (p < 0.001)

1 2

3

Glucuronate Gulonic Acid Gulonolactone

- 2.9

Transcriptional Effects on HepaticRecycling of Ascorbic Acid

+ 1.7+ 3.5

+ 1.6

Integrating Metabolomic andTranscriptional Profiling Data

• Metabolomic and transcriptomic data inform eachother– Major urinary metabolomic changes reflected in transcriptional

profiling data• not most affected genes

• Evidence that– Microsomal enzyme induction increases the demand for ascorbic

acid– Ascorbate is required to support microsomal enzyme induction

• Urinary excretion of gulonic and ascorbic acid consistentlyincreased by microsomal enzyme induction in rodents

Skeletal Muscle Fiber Type andBiochemistry

Fiber selective toxicity– Statins: Fast twitch (Type II)– PPARα agonists: slow twitch (Type I)

Variable response to stress or insult– Cell death, atrophy, hypertrophy

CP, glycogenTriglyceridesFuel sourceHighLowGlycolytic capacityLowHighMyoglobinLowHighOxidative capacityLowHighMitochondriaHighLowForce productionLowHighResistance to fatigueFastSlowContraction Type

Type IIType IFiber Type

Preclinical and Clinical Biomarkers ofSkeletal Muscle Toxicity

• Creatine kinase– Used widely clinically; limited utility in preclinical species

• Non-specific markers– AST, Aldolase, LDH

• Muscle-specific proteins– Troponins, myoglobin, myosin light chain 1 (Myl3), fatty

acid binding protein 3 (Fabp3)• Tissue specificity: skeletal and cardiac muscle

– Some expressed in other tissues (liver, kidney)• Short half-life can limit utility

• Combinatorial measurements as best approach

Skeletal Muscle Biomarkers of Toxicity:Fast Twitch Troponin I and Myoglobin

• Cerivastatin: severe skeletal muscle necrosis(Type II) in female rats– FsTnI increased markedly– AST highly variable

• Isoproterenol: cardiac (ventricular) necrosis

25 ± 7*< 6 ± 2ndnd157 ± 17135 ± 19Isoproterenol< 4 ± 1< 5 ± 111 ± 3< 7 ± 1151 ± 10130 ± 21Cerivastatin-M

141 ± 75*nd733 ± 224*< 9 ± 11838 ± 772*114 ± 6Cerivastatin-FTreatedControlTreatedControlTreatedControl

uMB (ng/mg Cr)fsTnI (ng/ml)AST (U/L)

Vassallo et al. Toxicol. Sci 111, 402-412, 2009Cerivastatin dosed for 14 d at 1 mg/kgIsoproterenol dosed to male rats at 0.5 mg/kg (single dose)

3-Methylhistidine

Urine NMR Spectral Characteristics:Cerivastatin

Aranibar et al. Anal Biochem (in press)

Carnosine and MethylhistidineMetabolism

Methylation

Carnosine

Anserine

β-alanine + histidine

H2N CH C

CH2

OH

O

N

N

H3C

β-alanine +

HN CH CH

CH2

OH

OH

HN

N

H2C C

O

H2CH2N

HN CH CH

CH2

OH

OH

N

N

H2C C

O

H2CH2N

H3C

1-Methylhistidine

• Highly specific to skeletal muscle (anserine)– mM levels– Used to study muscle protein turnover

• Rat serum concentrations at µM levels

010002000300040005000600070008000

0100200300400500600700800

3-Methylhistidine

• Post-translational modification of actin and myosin• Rat skeletal muscle concentrations ≥ 500 µM

– 3-MH > 1-MH in smooth muscle, heart

Effect of Skeletal Muscle Toxicity onMethylhistidine Levels

11 ± 3230 ± 18381 ± 45452 ± 34Cerivastatin-M522 ± 88*256 ± 263142 ±666*494 ± 52Cerivastatin-F

TreatedControlTreatedControl

3-MH(nmol/mg Cr)

1-MH(nmol/mg Cr)

Urinary Excretion

2.8 ± 0.13.0 ± 0.214.1 ± 0.815.8 ± 2.1Cerivastatin-M5.5 ± 0.9*1.9± 0.269.7 ± 12.8*6.0 ± 0.9Cerivastatin-F

TreatedControlTreatedControl

3-MH(µM)

1-MH(µM)

Serum

Time Course of Biomarker Changes

Control Day 9 Day 15

** *

The Paradox of Isoproterenol:Methylhistidines and Muscle Biology

• Cardiac necrosis– Myoglobin increased

• Dose-dependent effectson skeletal muscle– Low dose increases muscle

growth– 1- and 3-MH decrease due

to anabolic effect

• Methylhistidines mayreport hypertrophy,atrophy and necrosis

0

100

200

300

400

1-MH 3-MH

ControlIsoproterenol

Urin

ary

MH

(nm

ol/m

g C

r)

* *

Human Relevance of Methylhistidines

• All mammals except humans synthesize (much) anserine– 1-MH may only be applicable to preclinical species

• 3-MH is conserved across all species• Meat consumption can affect constitutive levels

HN CH CH

CH2

OH

OH

HN

N

H2C C

O

H2CH2NHN CH C

H

CH2

OH

OH

N

N

H2C C

O

H2CH2N

H3C

Integrating Omics Data for HazardIdentification and Risk Assessment

RISK ASSESSMENT

Dose-responseExposure Analysis

Cross-SpeciesAnalysis

Mechanism(s)

HAZARDIDENTIFICATION

Evaluating ToxicityPredicting Toxicity

Developing HypothesesInvestigating Mechanisms

Toxicology will progress from predominantly individual chemical studiesinto a knowledge-based science in which computational and informaticstools will play a significant role in deriving new understanding oftoxicant-related disease.

Acknowledgements

BMSJeff VassalloBrenda Lehman

Applied and Investigative MetabolomicsNelly AranibarPetia ShipkovaDon RobertsonMike Reily

Iowa State UniversityJohn Rathmacher

University of Kansas Medical CenterYoucai ZhangCurt Klaassen