Behavioral Metabolomics

Center for Biomarker Research and Personalized Medicine

Behavioral Metabolomics

October 21st, 2010By Joseph L McClay

Center for

Biomarker

Research

and

Personalized

Medicine

Presentation Overview

• The “omics” philosophy• Metabolomics as an assay of

biological function• Technologies (MS, NMR)• Neurochemical metabolomics in

rodents• Study of methamphetamine

• Summary • Bioinformatics tools example

Center for

Biomarker

Research

and

Personalized

Medicine

Hierarchies of Order

Oltavi & Barabasi (2002) Science 298, p763

Center for

Biomarker

Research

and

Personalized

Medicine

• Many omics variants:• DNA sequence

• GWAS

• Whole genome sequencing

• Epigenetics• Whole genome methylation

• Gene expression (RNA)• Expression arrays

• microRNA arrays

• Protein• Proteomics

• Metabolites• Metabolomics

• Metabonomics

Center for

Biomarker

Research

and

Personalized

Medicine

The omics “principle”

• Assume you know nothing• Try to measure everything

• Is this a hypothesis-driven approach to science?

• Advantages – new discovery• Disadvantages – false positives, cost

Center for

Biomarker

Research

and

Personalized

Medicine

Law of the Instrument

• “If you have a hammer, everything looks like a nail”

• Omics approaches are very technology driven

• Technology = assays + informatics• Pushing the limits of technology is

extraordinarily expensive• However, there is the opportunity to break

open the complexity of biology

Center for

Biomarker

Research

and

Personalized

Medicine

Metabolomics

• Biochemistry on a large scale

• Examination of all endogenous metabolites (under 1500Da) in a sample

• Several thousand in human metabolome

• Ultimate indicators of biological system response

Center for

Biomarker

Research

and

Personalized

Medicine

Possible applications

• Comparison of tissue-specific metabolic profiles

• Drug effects on metabolism• Personalized medicine

• Developmental effects• Metabolic disturbances in disease /

pathogenesis• In combination with other omics

• For example, GWAS to map quantitative trait loci for individual differences in metabolite leves (mQTLs)

Center for

Biomarker

Research

and

Personalized

Medicine

Technologies – characterizing the mixture

Nuclear Magnetic Resonance Mass Spectrometry

Center for

Biomarker

Research

and

Personalized

Medicine

What are the data like?

• Input is a complex mixture of metabolites

• Integrate across spectrum / identify specific compounds

• Examination of relative peak heights / integrals or compound levels

• So, quantitative in nature (more akin to gene expression than genotype data)

Brain mass spec (Woods et al 2006)

Urinary 1H NMR (McClay et al 2010)

Center for

Biomarker

Research

and

Personalized

Medicine

Methamphetamine

• percentage of past-year MA use among persons 12+ has remained relatively stable

• Estimates ranging from 0.7% in 2002 to 0.6% in 2007

However, admissionsto treatment programshave increased dramatically since themid 1990s

Center for

Biomarker

Research

and

Personalized

Medicine

Rationale for a metabolomics study of methamphetamine in mice

• Behavioral studies and animal models are well worked out

• While some gene expression and other studies have been carried out, to date no metabolomics study

• Returning to the “omics” principles outlined earlier, do we really know all the effects of meth?

• If we can better characterize the effects, we can perhaps see pathways that could mediate the addiction process

• Find candidate compounds for in vivo imaging

Center for

Biomarker

Research

and

Personalized

Medicine

Study design• 8 inbred strains of mice, chosen for

maximum genetic variation• 48 mice per strain• Acute vehicle, 1, 3 or 10mg/kg meth• Chronic vehicle or 3mg/kg meth for 5 days• 1 hour behavioral assessments of

locomotor activity using automated boxes• Followed by sacrifice, brain excision and

freezing in liquid nitrogen• Shipment to Metabolon, RTP, NC• GC and LC mass spectrometry

Center for

Biomarker

Research

and

Personalized

Medicine

Overview schematic

Center for

Biomarker

Research

and

Personalized

Medicine

010

00

200

0ha

ctv

Vehicle 1 mg/kg 3 mg/kg 10 mg/kg

Methamphetamine Dose

hactv

050

010

00

totd

ist



totdist

050

100

150

mo

vno



movno

050

100

mo

vtim

e



movtime

500

550

600

rest

ime



restime

050

010

00

strc

nt



strcnt

050

100

strn

o



strno

050

100

150

strt

ime



strtime

020

040

060

0m

rgd

ist



mrgdist

300

400

500

600

mrg

time



mrgtime

020

040

0ct

rdis

t



ctrdist

010

020

030

0ct

rtim

e



ctrtime

Acute Behavioral Effects, Significant Outcomes

Behavioral pharmacology

Center for

Biomarker

Research

and

Personalized

Medicine

Pharmacometabolomics

• Acute vehicle, acute 3mg/kg meth and chronic 3mg/kg meth for 5 days

• 18 mice per strain, 8 strains total• Test for differences in metabolite

levels between groups• 300 metabolites in total were

identified by Metabolon and tested• False Discovery Rate control

necessary because of large number of tests

Center for

Biomarker

Research

and

Personalized

Medicine

Center for

Biomarker

Research

and

Personalized

Medicine

Acute metabolic effects

Compound Beta p-value

q-

value Primary pathway Class

fructose -0.282 6.3E-06 0.001 Glycolysis, gluconeogenesis Carbohydrate

lactate 0.305 4.4E-05 0.003 pyruvate metabolism Carbohydrate

malate 0.223 0.0001 0.006 Krebs cycle Energy

2-hydroxyglutarate 0.129 0.0001 0.007

succinate 0.191 0.0003 0.015 Krebs cycle Energy

tryptophan 0.156 0.0008 0.025 Tryptophan metabolism Amino acid

fumarate 0.118 0.0009 0.027 Krebs cycle Energy

linoleate (18:2n6) 0.272 0.0027 0.059 Long chain fatty acid Lipid

citrate 0.078 0.0043 0.081 Krebs cycle Energy

sorbitol -0.224 0.0045 0.081 starch, and sucrose metabolism Carbohydrate

glycerophosphorylcholine -0.081 0.0052 0.081 Glycerolipid metabolism Lipid

Center for

Biomarker

Research

and

Personalized

Medicine

Chronic effects – part 1

Compound Beta p-value q-value Primary pathway Classlactate 0.405 1.4E-10 1.3E-07 Glycolysis, gluconeogenesis Carbohydrate

malate 0.305 3.6E-10 1.6E-07 Krebs cycle Energy

citrate 0.140 1.3E-09 3.7E-07 Krebs cycle Energy

2-hydroxyglutarate 0.167 3.8E-09 8.3E-07tryptophan 0.227 6.3E-09 1.1E-06 Tryptophan metabolism Amino acid

alanine 0.356 1.8E-06 0.0003 Alanine metabolism Amino acid

2-aminoadipate 0.156 7.0E-06 0.0007 Lysine metabolism Amino acid

3-hydroxybutyrate 0.320 4.5E-05 0.003 Ketone bodies Lipid

urea -0.149 8.4E-05 0.005 Urea, arginine metabolism Amino acid

maltotriose 0.773 0.0001 0.007 Sucrose metabolism Carbohydrate

choline phosphate 0.102 0.0003 0.013 Glycerolipid metabolism Lipid

gamma-aminobutyrate 0.132 0.0003 0.015 Glutamate metabolism Amino acid

ergothioneine 0.159 0.0004 0.015glycerophosphorylcholine -0.085 0.0005 0.019 Glycerolipid metabolism Lipid

fructose -0.177 0.0005 0.019 Sucrose metabolism Carbohydrate

gamma-glutamyl alanine 0.227 0.0006 0.022 Gamma-glutamyl

serine 0.067 0.0008 0.025 Glycine, serine and threonine Amino acid

glucose -0.177 0.0012 0.035 Glycolysis, gluconeogenesis Carbohydrate

ribose 0.299 0.0015 0.040 Nucleotide sugars Carbohydrate

glycerol 3-phosphate -0.085 0.0016 0.041 Glycerolipid metabolism Lipid

succinate 0.141 0.0018 0.044 Krebs cycle Energy

Center for

Biomarker

Research

and

Personalized

Medicine

Chronic effects – Part II

Compound Beta p-value q-value Primary pathway Class

uridine 0.283 0.002 0.058 Pyrimidine metabolism Nucleotide

adenosine 5'diphosphoribose 1.356 0.003 0.059 Nicotinamide metabolismCofactors and vitamins

nicotinamide 1.120 0.003 0.059 Nicotinamide metabolismCofactors and vitamins

guanosine 5'- monophosphate 0.822 0.003 0.065 Purine metabolism Nucleotide

glutamine -0.076 0.003 0.065 Glutamate metabolism Amino acid

dehydroascorbate 0.648 0.004 0.071 Ascorbate metabolismCofactors and vitamins

ribulose 5-phosphate 1.210 0.004 0.077 Nucleotide sugars Carbohydrate

phenylalanine 0.091 0.004 0.081 Tyrosine metabolism Amino acid

maltose 0.692 0.005 0.081 Sucrose metabolism Carbohydrate

cysteine 1.277 0.005 0.081 Cysteine metabolism Amino acid

butyrylcarnitine -0.127 0.005 0.081 Fatty acid metabolism Lipid

pipecolate -0.118 0.006 0.086 Lysine metabolism Amino acid

inosine 0.848 0.006 0.095 Purine metabolism Nucleotide

Center for

Biomarker

Research

and

Personalized

Medicine

Alternate parameterization•Between group comparison shows the extensive metabolic disruption due to meth administration•However, does not disaggregate acute from chronic meth effects. •For this we need a 2nd parameterization:

Intercept (a) represents the “simplest” condition--acute vehicle (av). Parameter 1 (d1) captures marginal effect of acute meth over acute vehicle. Parameter 2 (d2) captures marginal effect of chronic vehicle injection over “just” acute meth. Parameter 3 (d3) captures marginal effect of chronic meth over chronic vehicle injection + acute meth. We include with a random intercept to account for clustering within strain (u0).

Metabolite level = a + b1*d1 + b2*d2 + b3*d3 + u0 + e

Center for

Biomarker

Research

and

Personalized

Medicine

Results – alternative parameterization

parm Compound Beta p-value q-value Primary pathway Class

d1 2-hydroxyglutarate 0.129 1.20E-04 0.048 Citric acid cycle energy

d3 ergothioneine 0.19 3.00E-04 0.069 Dietary

d3 choline phosphate 0.118 3.50E-04 0.069 Ceramide signaling phospholipid

Center for

Biomarker

Research

and

Personalized

Medicine

Behavioral MetabolomicsSensitization:•Essentially is an increase in response to the same dose of drug after repeated exposure•We are measuring locomotor activity•In locomotor terms, sensitization means that mice will move around more after their dose of drug on the last day, as compared to the first day•However, the automated boxes measure locomotor activity in several ways•Around 20 locomotor activity variables are collected

Center for

Biomarker

Research

and

Personalized

Medicine

Factor analysis

Factor Variance Difference Proportion CumulativeFactor1 6.40 2.04 0.36 0.36Factor2 4.36 2.13 0.25 0.61Factor3 2.23 0.57 0.13 0.74Factor4 1.66 0.24 0.09 0.83Factor5 1.42 0.60 0.08 0.91Factor6 0.82 0.53 0.05 0.96Factor7 0.29 0.10 0.02 0.97Factor8 0.18 0.01 0.01 0.98Factor9 0.17 0.11 0.01 0.99Factor10 0.06 0.01 0.00 1.00Factor11 0.06 0.05 0.00 1.00Factor12 0.01 0.00 0.00 1.00

Variable Factor1 Factor2 Factor3 Factor4-------------+--------------------------------------------------------------------------------rtime_sens~p 0.08 0.34 0.04 -0.08rmovno_sen~p -0.01 0.98 0.05 0.00ractv_sens~p -0.01 0.96 0.04 -0.01ctrtime_se~p -0.06 -0.06 -0.99 0.05ctrdist_se~p 0.84 0.03 -0.23 0.17mrgtime_se~p 0.07 0.06 0.99 -0.05mrgdist_se~p 0.89 0.01 0.34 0.00strtime_se~p 0.58 -0.19 -0.19 0.33strno_sens~p 0.33 -0.09 -0.03 0.80strcnt_sen~p 0.83 -0.10 -0.06 0.19vtime_sens~p 0.03 0.67 0.02 -0.10vmovno_sen~p 0.02 0.97 0.06 -0.03vactv_sens~p -0.02 0.95 0.02 -0.06restime_se~p -0.98 -0.02 0.00 -0.09movtime_se~p 0.98 0.02 0.00 0.09movno_sens~p 0.29 -0.02 -0.12 0.82totdist_se~p 0.97 0.02 0.16 0.07hactv_sens~p 0.90 -0.02 0.05 0.34

4 factors: horizontal/total movement, vertical movement, center/margin time, stereotypy

Create BLUPs for each animal for sensitization, i.e. increase in horizontal movement over course of study

Center for

Biomarker

Research

and

Personalized

Medicine

Results – metabolomics analysis of sensitization

compound beta p-value q-value primary pathway class

malonylcarnitine 0.817 7.74E-05 0.014 Amino acid conj Lipid / Energy

serine -0.084 1.54E-04 0.014 Serine threonine met Amino acid

homocarnosine -0.155 1.56E-04 0.014 Amino acid conj Peptide

ergothioneine 0.2 2.64E-03 0.177 Unknown N/A

histamine 0.777 5.00E-03 0.238 Histidine metabolism Amino acid

NADH 0.167 6.20E-03 0.238Nicotinamide /

energyCofactors and vitamins

In this analysis, we are correlating individual differences in the levels of specific metabolites with individual differences in sensitization tomethamphetamine.

Center for

Biomarker

Research

and

Personalized

Medicine

Summary

• Metabolomics analysis can yield insights into the metabolic sequelae of drug administration

• In this study, we observed extensive and dramatic alterations to neurochemistry following meth administration

• Among specific findings were changes to glutamine / alanine-related metabolites and choline phosphate following chronic adminsitration

• Associations with sensitization implicated histamine and homocarnosine

Center for

Biomarker

Research

and

Personalized

Medicine

Summary (contd)• Previous studies have implicated GABA,

histamine, phospholipids etc in relation to stimulant drug abuse / administration

• This first attempt at neurochemical / behavioral metabolomics appears promising

• Much additional work to be done • Application to other drug / behavior

pairings (e.g. PPI and antipsychotics)

Center for

Biomarker

Research

and

Personalized

Medicine

Many statistical development opportunities

For example, identify subsets of metabolites whose concentrations are always coupled.

Use that to define test statistic:– Multivariate– Eliminates some of

the dynamics

Center for

Biomarker

Research

and

Personalized

Medicine

Acknowledgements

http://www.pharmacy.vcu.edu/biomarker/

CBRPM, School of PharmacyEdwin van den OordDaniel AdkinsShaunna ClarkRenan Souza

Department of Pharmacology and ToxicologyPatrick BeardsleyRob VannSarah VunckAngela Batman (now at Pfizer UK)

Funding: NIDA

Center for

Biomarker

Research

and

Personalized

Medicine

Databases

• What does my metabolite do?• Choline Phosphate• Gamma-glutamyl alanine

• Search databases:• Reactome• KEGG – Kyoto Encyclopedia of Genes

and Genomes• BioSystems @ NCBI

Center for

Biomarker

Research

and

Personalized

Medicine

Web sites

• www.reactome.org• http://www.genome.jp/kegg/• www.ncbi.nlm.nih.gov/biosystems/

Behavioral Metabolomics

Documents