^tDsrx			.	_	.	jJTOXICOLDGY

• g \Metabolomics for Developing Markers
J of Chemical Exposure and
Distinguishing Toxicity Pathways

Tracy Whitehead, Tim Collette, Drew Ekman, John Kenneke, Wayne Garrison, and Quincy Teng

US EPA ORD/NERUERD, Athens, GA

Abstract

Adetabolomics involves the application of advanced analytical and statistical
tools to profile changes in levels of endogenous metabolites in tissues and
biofluids resulting from disease onset, stress, or chemical exposure. Nuclear
Magnetic Resonance (NMR) spectroscopy-based metabolomics has proven
useful in mammalian systems for distinguishing between sites and mechanisms
of toxicity for tissue-specific toxins. Metabolomics has been characterized as
the true measure of metabolic outcomes suggested by changes in gene and
protein expression; as such, metabolomics provides a connection between these
molecular endpoints and whole organism responses. Although used mostly in
mammalian studies, metabolomics is now finding utility in a wide variety of
other organisms, including aquatic species.

We have developed a research program in metabolomics that involves
numerous partners across EPA, other Federal labs, academia, and the private
sector. Our goals are to (1) develop metabolite-based markers that can be used
by EPA in chemical exposure assessments and (2) develop and test hypotheses
about toxicity pathways for risk assessments. We are focusing this program on
ecologically relevant species—in particular, small fish toxicological models.

For example, to better understand the mode of action of endocrine-disrupting
chemicals (EDCs) in small fish (fathead minnow, zebrafish), we are conducting
metabolomic analyses with multiple tissues (brain, blood, liver, and gonad) and
urine. Initial metabolomic studies were focused on collection of baseline data
for actively spawning male and female fathead minnows. Subsequent work is
focusing on animals exposed to potent EDCs, such as the steroid 17a-
ethinylestradiol (Eiy. We are developing hypotheses about which tissue- and
biofluid-specific metabolite changes will be definitively related to exposure
based on the current understanding of modes of action for these chemicals.
Results will allow testing of these hypotheses to refine understanding of
activity and will help ensure that molecular markers of EDC exposure—another
outcome of this research—are meaningful. While certain metabolites are being
specifically targeted in these studies, we will also discern changes in the
complete metabolic profile using NMR spectroscopic data with statistical
approaches that allow capturing subtle changes in less-abundant metabolites.
These data will be integrated with genomic, proteomic, and whole organism
data from untreated fish and those exposed to known EDCs.

Introduction

The recently developed approach known as metabolomics involves the use of
advanced analytical techniques such as NMR (Nuclear Magnetic Resonance)
spectroscopy (Figure 1) to characterize changes in the levels of cellular
metabolites (e.g. sugars, carbohydrates, amino acids, etc.) that relate to the
toxic mechanism(s) involved in responding to the presence of a given chemical.
This approach has proven to be a powerful tool in the assessment of toxicity or
other physiological alterations in a variety of different organisms (Bailey, 2003;
Coen, 2003; Viant, 2003). We have employed NMR-based metabolomics for
the determination of changes in metabolite profiles in a variety of tissues
obtained from small fish exposed to the endocrine-disrupting compound 17a-
ethinylestradiol .(EEj), the active ingredient in oral contraceptives. This
compound was chosen because it has been detected in aquatic environments
and is considered relevant as an environmental contaminant.

Figure 1 - NMR spectroscopy

NMR spectroscopy is an analytical technique that allows
one to observe the levels of metabolites in a variety of
tissues and biofluids for
metabolomic analyses.

A portion of an NMR spectrum
of a trout liver tissue extract with
several metabolites labeled.
Changes in the levels of these
metabolites (e.g. sugars,
carbohydrates, amino acids, etc.)
can allow one to determine the
nature of toxicity induced by
exposure to a given compound.

Because changes in an organism's metabolite profile often occur in conjunction
with, or as a result of, changes in transcript and protein levels, metabolomics
can provide complementary information to genomic and proteomic studies.
Moreover, metabolomics (specifically NMR-based) is attractive because it
offers some distinct advantages relative to these other techniques that make
such a technique attractive for a number of reasons.

NMR-based Metabolomics for Toxicology

Advantages:

Disadvantages:

¦ no need for sequenced genome

¦ need for metabolite peak assignment

¦ open-ended (no p libelee tion of metabolites)

¦ observed metabolites "p reselected " by & undance

¦ high throughput

<15 mill per satvple ran in most cases

& ¦ when metabolites of interest in bw abundance
> 1 In' p er sanpleran

' low per sample cost
<$2.00 USD

a high initial cost for instrumentation
> $750K USD

¦	high level of structural information

¦	non-destructive

"overlapping" information



a little or no sample preparation
¦ excellent reproducibility (eg., cross-instrument)

>I; Disadvantages that maybe mitigated by
complementary use of other analytical
techniques and/or chemometrics

Traditionally, metabolomics has been utilized in many areas of study (e.g.
disease, strain typing, etc.). In ORD, our focus is on its application to
toxicology. Specifically, we are applying metabolomics to both better
understand the mode(s) of action (MOA) of EDCs as well as to develop markers
of exposure using small fish models such as fathead minnows and zebrafish
(Figure 2).

Figta-e 2 - Smallfish models used in ecometabolomics studies

Zebrafish (Danio rerio)

Fathead Minnow (Pimephales promelas)

While determining a given MOA requires a greater understanding of
biochemical processes, we believe that this understanding is important for
establishing reliable and meaningful biomarkers.

Metabolomics

Toxicology Disease Strain
State	Typing

Mode of Action

Results

Marker of Exposure

Figure 3 - NMR spectra of fathead minnow tissue extracts and biofluids

in)

» ilLi I uUfcA J



iUjj/LJL

- k





jUN

It







Data collected for aqueous extracts from fathead minnow (Figure 3) and zebrafish
tissues have shown that it is possible to identify metabolites that are tissue
specific.

After collecting spectra from a statistically-relevant number of organisms, the
number of metabolites changing in response to a given toxic insult from a large
sample population can be overwhelming without the use of statistical tools that
can quickly identify changes.

Various computational methods designed to simplify the analysis of large
amounts of data (e.g, principal components analysis (PCA) and partial-least
squares discriminant analysis (PLS-DA, Figure 4)) provide the ability to quickly
determine metabolite changes as opposed to assessing changes simply by eye.
While these methods are effective, more- advanced methods are needed to
observe very subtle changes that may carry information crucial to more
accurately defining organism responses.

Figure 4 - PLS-DA atutfysis showing separation between control
zebrafish ttnd EE ^-exposed zebrafish (liver tissue extract)

•

/ 100 ng/L /
( \ • * .'

30 ng

/L (



*"		

100 ng/L

Male





' i ' 2 ' 3 ' J ' 5 ' 6 ' 7 9 ' 10

tpi

Support from researchers in academia, industry, and various State and Federal
agencies has been crucial for the early success of the metabolomics program in
ORD. At present, there are several collaborations that have been established
and that have active projects that are underway.

PBPK	analytical	spectral

modeling	support	analysis, etc.

Conclusions

The potential for metabolomics to supply rapid and accurate information on the
toxicity of compounds of interest to the EPA has been shown here through the
analysis of the effects of endocrine-disrupting compounds (EDCs) on aquatic
species. One goal of the ORD NMR facility is to use metabolomics, in
conjunction with genomic, proteomic, and whole organism data, to develop
markers of EDC exposure in small fish models. To ensure that these markers are
meaningful, we also seek to better understand the modes-of-action of EDCs
within various biological contexts. Prior to achieving this, we have sought to
determine that metabolites can be readily observed in relevant tissues and
biofluids from single animals and that exposure to environmentally-relevant
compounds lead to observable changes for dosed fish. As illustrated here, we
have now made these determinations and are moving forward to determine toxic
modes-of-action and develop biomarkers indicative of exposure. By
collaborating with other EPA researchers (NERL and NHEERL), academic
institutions and other agencies, it is our intention to extend NMR-based
metabolomics into the ecotoxicology realm. As a result of Agency support and
collaborative efforts with the entities mentioned above, this initial study serves
as a major step in the establishment of an integrated metabolomics program at
EPA.

References

1.	Bailey, N.J.C., Oven, M„ Holmes, E„ Nicholson, J.K.,Zenk,M.H. (2003). Metabonomic analysis of the

Phytochemistry. 62: 851-858.

2.	Coen, M.,Lenz, E.M., Nicholson, J.K., Wison, I.D., Pognan, F„ Lindon, J.C. (2003). An integrated metabonomic
investigation of acetaminophen toxicity in the mouse using NMR spectroscopy. Chem. J%s. Toxicol 16: 295-303.

3.	Viant, M.R., Rosenblaum, E.S., and Tj eerdema, R.S. (2003). NMR-based metabolomics: a powerful approach for
characterizing the effects of environmental stressors on orpnism health. Environ. Sci. and Techno!. 37: 4982-4989.

Disclaimer: Although this work was reviewed by EPA and approved for presentation, it may not necessarily reflect official Agency policy. Mention of trade names or commercial products does not constitute endorsement or recommendation for us

epascienceforum

Your Health • Your Environment • Your Future


-------