ORD's
Computational Toxicology
Research Program
Implementation Plan
(FY 2006 - 2008)
April 2006
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
-------
DISCLAIMER
This document has been subjected to internal and external review for clearance.
The Implementation Plan was developed in response to a recommendation of the first
review of the National Center for Computational Toxicology, by the Computational
Toxicology Subcommittee of the Office of Research and Development's Board of
Scientific Counselors (BOSC), conducted during a two-day meeting on April 25-26,
2005. In addition to the BOSC Subcommittee this Plan was reviewed by several ORD
staff, including members of the ORD Science Council. A second BOSC review was held
on June 19-20, 2006 and comments were documented in a letter report to Dr. George
Gray, Assistant Administrator for the Office of Research and Development, dated
December 12, 2006. Information on the BOSC can be found at
http://www.epa.gov/comptox/bosc review/2006/index.html. This Plan does not constitute
an Agency position or policy concerning computational toxicology. Any mention of trade
names does not constitute Agency endorsement.
-------
Table of Contents
Table of Contents ii
List of Figures iv
Acronyms v
Executive Summary vi
I. Introduction and Background 1
II. Office of Research and Development Computational Toxicology Program 3
National Center for Computational Toxicology's (NCCT) Mission 4
Science to Achieve Results Centers 5
Board of Scientific Counselors 6
III. Research Projects 6
IV. Summary of Outcomes from Research 9
V. Partnerships 9
VI. Conclusion 11
Table 1 -Identification of Computaional Toxicology Research Program (CTRP) Projects 13
Table 2 -CTRP Output/Input Table 14
Table 3 -FTE's from ORD's Supporting Multi Year Plans 22
Appendix A 23
Appendix B 25
in
-------
List of Figures
Figure 1 - Component of the Computational Toxicology Research Program within ORD 4
Figure 2 - NCCT Staffing Profile 5
Figure 3 -Alignment of the Five Research Track 7
Figure 4 -Interactions between Long Term Goals 8
Figure 5 -Computational Toxicology Research Program Design 12
-------
Acronyms
BOSC
Board of Scientific Counselors
CPCP
Categorization and Prioritization Community of Practice
CoP
Communities of Practice
CompTox
Computational Toxicology
CTISC
Computational Toxicology Implementation and Steering Committee
CTRP
Computational Toxicology Research Program
DNA
Deoxyribonucleic acid
DSSTox
Distributed Structure-Searchable Toxicity
EBRC
Environmental Bioinformatics Research Center
EPA
Environmental Protection Agency
FTE
Full Time Equivalents
FY
Fiscal Year
LTG
Long Term Goal
MOA
Memorandum of Understanding
MYP
Multi Year Plan
NCCT
National Center for Computational Toxicology
NCEA
National Center for Environmental Assessment
NCER
National Center for Environmental Research
NERL
National Exposure Research Laboratory
NHEERL
National Health and Environmental Effects Research Laboratory
NIH
National Institutes of Health
NIEHS
National Institute of Environmental Health Sciences
NRMRL
National Risk Management Research Laboratory
NICEATM
National Toxicology Program Interagency Center for Evaluation of
Alternative Toxicological Methods
OPPTS
Office of Pesticides, Prevention, and Toxic Substances
ORD
Office of Research and Development
QSAR
Quantitative Structure Activity Relationship
RFA
Request for Application
RNA
Ribonucleic acid
SAB
Science Advisory Board
STAR
Science to Achieve Results
ToxCast
Chemical Prioritization Research Program
-------
Intentionally Left Blank
-------
Executive Summary
This document presents the implementation plan for the Framework for the
Computational Toxicology Research Program (CTRP), a strategy document developed by the
Office of Research and Development (ORD) in 2003. Computational toxicology is defined as the
integration of modern computing and information technology, with molecular biology and
chemistry to improve risk assessment and prioritization of data requirements of chemicals by the
Agency. The program is intended to provide innovative solutions to a number of persistent and
pervasive issues facing The Environmental Protection Agency (EPA) regulatory programs. The
three objectives of the Framework have been translated into long term goals (LTGs) for the
CTRP and the subsequent research has been aligned into five supporting tracks. The LTGs for
the program are: (I) EPA risk assessors use improved methods and tools to better understand and
describe linkages across the source to outcome paradigm; (II) EPA Program Offices use
advanced hazard characterization tools to prioritize and screen chemicals for toxicological
evaluation; and (III) EPA risk assessors and regulators use new models based on the latest
science to reduce uncertainties in dose-response assessment, cross-species extrapolation, and
quantitative risk assessment. The supporting research tracks are: (A) Development of Data for
Advanced Biological Models; (B) Information Technologies Development and Application; (C)
Prioritization Method Development and Application; (D) Providing Tools and System Models
for Extrapolation across Dose, Life Stage, and Species; and (E) Advanced Computational
Toxicology Approaches to Improve Cumulative Risk Predictions. A standing subcommittee of
ORD's Board of Scientific Counselors has been established to provide guidance to the CTRP as
it develops its research agenda.
The research supporting the CTRP flows from three areas. The first component is
composed of the efforts of the National Center for Computational Toxicology (NCCT), an
organizational entity created within ORD in 2005 to accelerate the development and use of
computational tools in the EPA's regulatory operations. The staffing plan for the NCCT includes
computational chemists, computational biologists, and bioinformaticians who will work both on
internal projects and through partnerships with scientists in other ORD Labs and Centers as well
as, external scientists to provide scientific expertise and leadership related to the application of
mathematical and computational tools and models to high priority Agency needs. The projects
located within the NCCT include those developing information databases for chemical toxicity
(DSSTox), building toolboxes (ToxCast) for prioritizing chemicals for toxicology evaluation,
providing computation models that extrapolate exposure and effects across life stages,
developing multi-scale computational models of organ systems that will help provide
understanding of how mechanisms of toxicity determine dose and time-responses, evaluating
issues related to estimation of model parameters in computational models, and integrating
diverse types of information needed to understand the impact of the environment in a broad
sense on the health of individuals.
The second component is a suite of seven projects distributed across ORD that were
funded in FY05 by a competitive process overseen by the Computational Toxicology
Implementation and Steering Committee (CTISC), a cross EPA group formed to initiate the
ORD program. These projects include efforts that bring systems level approaches to the study of
chemical toxicity in small fish models and in amphibian metamorphosis, to the effects of diesel
vi
-------
particles on lung cells, and to factors associated with development of asthma in children. Other
projects are designed to predict metabolites of chemicals, use metagenomic approaches to assess
microbial pollution sources, and evaluate the use of toxicogenomic data in risk assessment of
reproductive toxicants.
The final component of the CTRP is derived from the Science to Achieve Results
(STAR) program of ORD's National Center for Environmental Research (NCER). The STAR
program supported a grant solicitation for investigator directed research in this area of systems
level understanding of the hypothalamic-pituitary-gonadal axis in FY04 that funded two projects
involving small fish models and one on the female rat. In FY05, the STAR program supported
the establishment of two Centers for Environmental Bioinformatics, one at the University of
North Carolina in Chapel Hill and the other at the University of Medicine and Dentistry of New
Jersey. These Centers were funded as cooperative agreements, which allows for close
interactions between scientists within the University-based Centers and EPA scientists. With
project periods of five years, they are intended to provide considerable momentum in the
development and application of bioinformatics tools in the protection of human health and the
environment.
Within this plan, the research issue and relevance, experimental approach, progress to
date and milestones over the next three years are articulated for each of the individual projects.
Collectively the elements of the plan have been developed to provide a sequential series of short
to medium term projects that will advance the utilization of computational tools in hazard and
risk assessment. Particular progress is expected over the next few years in the area of
prioritization tools, as the ToxCast program begins developing fingerprints of chemical activity
based on the collection of broad spectrum, high throughput data for a large number of well
characterized chemicals. Projects developing systems level understanding of biological functions
will require longer investments, but the results in terms of improving the linkages in the source-
to-outcome paradigm and subsequently in the extrapolation of information across dose, time,
chemical and species will reduce some of the inherent uncertainties present in risk assessment as
currently practiced.
vii
-------
I. Introduction and Background
This document lays out the rationale and short to medium term objectives (up to 3 years)
of the ORD new research program in computational toxicology. The emerging field of
computational toxicology applies mathematical and computer models and molecular biological
and chemical approaches to explore both qualitative and quantitative relationships between
sources of environmental pollutant exposure and adverse health outcomes. Recent technological
advances make it possible to develop molecular profiles using
genomic, proteomic, and metabolomic (the "omics") methods
to identify the impacts chemicals may have on living
organisms or the environment. With these tools, scientists can
produce a more-detailed understanding of the hazards and
risks of a much larger number of chemicals. The integration
of modern computing with molecular biology and chemistry
will allow scientists to better prioritize data, inform decision makers on chemical risk
assessments, and understand a chemical's progression from the environment to the target tissue
within an organism and ultimately to the key steps that trigger an adverse health effect.
Currently, risk estimates are most often based on gross outcomes of disease such as
occurrence of cancer, a neurological disorder, or a visible birth defect. It has long been assumed
that these disease outcomes were the result of numerous and crucial alterations at the molecular
level inside living cells. Molecules of DNA, RNA, or endogenous proteins all have crucial
chemical and physical arrangements and interactions. Alterations in these arrangements and/or
interactions might be the first step in a cascade that can then lead to disease, morbidity, and
mortality. Research in genomics, proteomics, metabolomics, and computational toxicology are
expected to result in risk assessments based on specific changes at the molecular level, rather
than just the number of tumors, deaths, overt clinical changes observed in test animals. In the
future, assessments will be based on changes in molecular markers, such as the number of DNA
molecules altered at a crucial site, the change in an allosteric membrane protein that acts as a
receptor, or the change in a regulating protein inside the cell. The key is we will better
understand how those changes lead directly to the types of adverse health effects that have been
the traditional basis of EPA risk assessments and to use this understanding to reduce the
uncertainties in the extrapolation of effects across dose, species and chemicals.
Over the past 15 years the explosive growth of information regarding the structure of
many of the "molecules of life" has occurred. The sequencing of the human genome for
example, has given us a wealth of information only dreamed of in the previous decades. It will
soon be possible to accurately ascertain exactly where a xenobiotic chemical interacts with
regions of the genome and with other endogenous molecules. Binding of a toxic chemical with a
crucial portion of membrane proteins for example, may alter the structure and porosity of the cell
membrane. Such a change might lead to a change in the membrane potential of the cell. If that
cell is a neuron or cardiac cell critical physiologic changes might occur that could lead to overt
disease.
Advancements in the "omics" technologies coupled with the advances in analytic tools
such as microarray techniques will enable us to predict changes and evaluate which changes can
initiate, promote, or adjoin the cascade to disease. Computational systems biology techniques
"Computational toxicology:
integration of modern
computing and information
technology with molecular
biology"
1
-------
will help us integrate the necessary quantitative information related to those changes. The
concepts of computational chemistry and relating structure to activity are not new. For years
chemists and physicists have been relating key chemical and physical properties to molecular
and atomic structures. However, with advances in high performance computing and the
application of high throughput biological screening assays, there is unprecedented opportunity to
conduct this research on a much larger scale of operation.
From the sciences of "omics" technologies arise two possibilities: the first is the broad
spectrum (or high content) interrogation of multiple molecular events within a cell or tissue (e.g.
microarrays) and the second is the application of one or a few molecular targets in a high
throughput screening mode against large numbers of chemicals. Both of these possibilities
present opportunities and challenges. The opportunity is to use the ensuing outputs of these
technologies to shift toxicology from a descriptive to a predictive science and therefore improve
the ability of the EPA to assess hazard and characterize risk. The challenge lies in reducing vast
amounts of data generated by these approaches into useful information and in determining their
biological significance. The EPA's ORD has formulated its research program with this in mind
and will approach the issues through several targeted research tracks. The main thrust of the
research program will be to develop generic approaches which are adaptable to the more specific
types of risk assessment activities that are undertaken in the other Laboratories, Centers, or
Program Offices.
This challenge was given urgency in 2002, when Congress ordered a redirection of $4
million from available EPA funds,''for the research, development and validation of non-animal
alternative chemical screening and prioritization methods, such as rapid, non-animal screens
and Quantitative Structure Activity Relationships (QSAR), for potential inclusion in EPA 's
current andfuture relevant chemical evaluation programs". To fulfill this directive, the EPA
embarked on development of a research program that: (1) was consistent with the Congressional
mandate; (2) complemented and leveraged related on-going Agency sponsored efforts to
consider alternative test methods; (3) further advanced the research to support the Agency's
mission; and (4) would not duplicate the mission and programs in this area conducted by other
agencies. An innovative program, entitled Computational Toxicology (CompTox), was initiated
to target these goals and, in the process, significantly advance toxicology and risk assessment as
currently practiced by the Agency and the broader environmental sciences community. As
recommended by Congress, the proposed approach was developed in consultation with the
Office of Pesticides, Prevention, and Toxic Substances (OPPTS). The research projects funded in
FY02 and FY03 were largely devoted to proof-of-concept demonstrations that the approaches of
computational toxicology could be adapted to the study of endocrine disruptors (chemicals which
perturb the functioning of an endocrine system and which subsequently lead to adverse health
effects in an individual or a population). Early successes of these efforts included refinement of
estrogen receptor ligand binding data for use in development of quantitative structure-activity
models, evaluation of two EPA-developed cells lines for the detection of activity of estrogens
and androgens, and the development of an alternative test method for evaluating effects on
steroidogenesis that avoids the use of animal tissues. As the current document is intended to be
primarily forward-looking, these early projects will not be discussed further in this document
(see the 2004 Activity Report for more details).
2
-------
With increasing attention to and expectations for the CompTox program over the next
several years, ORD developed A Framework for a Computational Toxicology Research Program
in 2003, which provided strategic direction for the program. This
document was the product of a cross functional ORD team of scientists
and was endorsed by the Science Advisory Board (SAB). The Framework
identified three objectives for computational toxicology in the EPA and
these have been translated into three Long Term Goals (LTGs) for the
research program described herein:
• I. Risk assessors use improved methods and tools to better
understand and describe the linkages of the source-to-
outcome paradigm,
• II. EPA Program Offices use advanced hazard characterization tools to prioritize
and screen chemicals for toxicological evaluation and
• III. EPA assessors and regulators use new and improved methods and models
based on the latest science for enhanced dose-response assessment and
quantitative risk assessment.
II. ORD Computational Toxicology Program
With issuance of the Framework in FY04, ORD began the process of implementing a
research program by establishing the Computational Toxicology Implementation and Steering
Committee (CTISC) to help guide the program and communicate progress across the Agency.
Membership on the committee consists of the chair, the Director of NCCT, two representatives
from each of ORD's Laboratories and Centers, which are nominated by their Directors,
appropriate ORD management representatives (e.g., Associate Laboratory Directors or National
Program Managers from aligned programs), an ORD Regional Scientist, and representatives of
the main client offices of the CTRP Program. Membership consists of a three year term,
renewable once, (link to current membership). Recognition of the potential impact of
computational approaches to toxicology has continued to grow over the past several years, and
now research contributions flow from three distinct areas within ORD. The first component is
composed of the efforts of the NCCT, an organizational entity created within ORD in 2005, to
accelerate the development and use of computational tools in the EPA's regulatory operations.
The second component consists of projects funded in FY05 by a competitive process overseen by
the CTISC. The final component of the CTRP is derived from the STAR program of ORD's
NCER. Within the CTRP, the STAR program supports a small number of investigator lead
grants and two larger Centers. The relationship of these components is depicted in Figure 1. The
innermost component (grey oval) represents the efforts taking place within the NCCT. The
location of the inner circle indicates the central role NCCT in providing leadership to the overall
program. In the next outer circle (green oval) are the CTISC funded projects, which represent
research being performed in other Laboratories and Centers. These efforts are largely within the
National Health and Environmental Effects Research Laboratory (NHEERL) and the National
Exposure Research Laboratory (NERL). Smaller efforts are contained within the National Risk
Management Research Laboratory (NRMRL) and the National Center for Environmental
Assessment (NCEA). The third and final component of the CTRP is research supported by
3
-------
NCER through the STAR program, funded through Request for Applications (RFAs). A
particularly important effort of the STAR program was the funding of two academic centers to
support the advancement of bioinformatics in environment health (see below). Greater detail on
the projects in each of the three components is provided in Section III. It is important to
recognize that considerable efforts in computational research and in "omic" research reside
outside those directly or indirectly associated with the NCCT and CTISC. These efforts,
represented by the yellow oval in Figure 1, are part of ORD's overall research in health,
ecological, and risk assessment, and are contained within a variety of Multi-Year Plans (MYPs).
Figure 1 - Component of the Computational Toxicology Research Program within ORD
The National Center for Computational Toxicology. In October 2004, the EPA Science
Advisor and Assistant Administrator for ORD, Dr. Paul Gilman announced the formation of the
NCCT. which began official functions in February 2005. The announcement states:
"The Center will advance the science needed to more quickly and efficiently evaluate the potential risk of
chemicals to human health and the enviromnent. The Center will coordinate and implement EPA's research on
computational toxicology to provide tools to conduct more rapid risk assessments and improve the identification of
chemicals for testing that may be of greatest risk."
NCCT's Mission
The Center's mission is to achieve the goals set forth by ORD and the Framework
document by performing research that integrates modern computing and information technology
with molecular biology to improve Agency prioritization of data requirements and risk
assessment of chemicals. The Center's staffing profile is designed to provide expertise in
systems biology, computational chemistry, and bioinformatics, as outlined in Figure 2.
Recruitments are currently underway for two senior level investigators, one in the area of
computational systems biology and the other in bioinformatics. These positions are expected to
be filled by the end of FY06 and with supporting positions will complete staffing of the NCCT,
as currently envisioned.
CompTox
Research
Program
CompTox
Center
Computational/Omic
Research
4
-------
National Center for Computational Toxicology
(NCCT)
ADMINISTRATIVE SUPPORT
Director: Robert Kavlock
Program & Management Analyst- Karen Dean
Deputy Director: Jerry Blancato Executive Secretary- Sandra Roberts -
Admin Assistant (SEE)- Dorothy Goodson
SYSTEMS MODELING
COMPUTATIONAL CHEMISTRY BIOINFORMATICS
Hugh Barton
Steve Little To Be Named (2)
Michael Breen (Post-doc)
Melissa Pasquinelli (Post-doc)
Elaine Cohen Hubal
Jim Rabinowitz
Rory Conolly
Ann Richard
David Dix
Eric Weber (Detailed to NERL)
Keith Houck
Matthew Martin (EPA Intern)
Chester Rodriquez (Post doc)
Rhyne Woodrow Setzer
To Be Named (2)
Figure 1 - NCCT Staffing Profile
The NCCT scientists will advance the field of computational toxicology by working
across ORD to develop predictive models for screening and testing chemicals and by reducing
the uncertainties associated with extrapolating predicted effects across chemicals, levels of
biological organization and species. They will provide scientific expertise and leadership related
to the application of mathematical and computational tools and models, conduct research to
improve the predictive capabilities of the methods, models and measurements that constitute the
input materials to the computational approaches. In addition, they will conduct and sponsor
research to provide models for fate and transport of chemicals, environmental exposures to
humans and wildlife, delivery of the chemical to the target site of toxicity, molecular and cellular
pathways of toxicity, and ultimately systems-level understanding of biological processes and
their perturbation. A key part of the process of advancing the science will involve developing
partnerships with other government and private organizations so as to best leverage resources
committed to the effort. (Link to current NCCT staffing: http://epa.gov/ncct/organization.html)
STAR Centers
STAR Environmental Bioinformatics Centers: One area of research need, implicit in
many of the research projects contained with the CTRP, is bioinformatics. This rapidly emerging
technology is crucial to the computational toxicology program, and whereas the NCCT will be
adding a senior level bioinformaticist in FY06, there remains a large gap in ORD relative to the
ability to analyze the high volumes of molecular data and to predict potential toxicity, modes of
action, and ultimately risk. To help bridge this gap, NCER has supported the establishment of
two STAR Environmental Bioinformatics Research Centers (EBRC). The Research Center for
Environmental Bioinformatics and Computational Toxicology at the University of Medicine &
Dentistry of New Jersey (UMUDJ), Piscataway, NJ and The Carolina Environmental
-------
Bioinformatics Research Center at the University of North Carolina, Chapel Hill, NC will
operate as cooperative agreements and help facilitate the application of bioinformatics tools and
approaches to environmental health issues supported by the CTRP. It is anticipated that over the
next five years the EBRCs will conduct research to improve the science of bioinformatics and
assist in the analysis and interpretation of data relevant to the protection of human health and the
environment. Descriptions and web links of the two Centers are provided in Appendix A.
Bioinformatics staff of the NCCT will serve as technical liaisons with the EBRCs and help guide
interactions with other ORD investigators.
The STAR program is planning another Request for Applications in the area of
computational toxicology in FY07, and the topic for that call is currently under discussion.
BOSC
Board of Scientific Counselors: To help guide the program, ORD established a standing
panel of Board of Scientific Counselors (BOSC) to provide review and advice to NCCT and
CTRP. The panel first met in April 2005 to review the organization of NCCT, initial plans for
implementation, and progress of the early CTRP work. The panel commented very favorably on
the Center's early progress and the means outlined to achieve its goals (BOSC CompTox
Review). The composition of staff, plans for future hiring, establishment of working
partnerships, and the Center's strategic plan were especially highlighted. Several main
recommendations were made for consideration, two of which are addressed here. The first was to
develop a formal implementation plan for the future. The second was to develop Communities of
Practices (CoPs) within the EPA which can serve as a networking function for interested
scientists. Three such CoPs have been organized and are discussed later in this document. A few
other minor suggestions were also made, which were addressed in the formal ORD response to
the review (ORD BOSC Response). A second site visit of the BOSC is being planned for June
19-20, 2006, which will entail an in-depth assessment of the progress NCCT has made in
executing this implementation plan.
III. Research Projects
As noted above, research projects within the CTRP are composed of projects developed
internally within the NCCT, efforts funded across ORD by the CTISC, and research supported
by ORD's STAR program (in the form of both individual research grants as well as larger Center
grants). Research efforts supporting the three LTGs identified in the CompTox Framework have
been grouped into five research tracks: (A) Development of Data for Advanced Biological
Models; (B) Information Technologies Development and Application; (C) Prioritization Method
Development and Application; (D) Providing Tools and System Models for Extrapolation across
Dose, Life Stage, and Species; and (E) Advanced Computational Toxicology Approaches to
Improve Cumulative Risk Predictions. The alignments of the five research tracks with the LTGs
of the Framework are depicted in Figure 3. The component research projects are identified in
Table 1, with hyperlinks to both summary information on their individual outputs and impacts
(Table 2), as well to more descriptive information on the research issue and relevance, approach,
impact and partnerships in Appendix B. For those projects housed outside the NCCT, Table 3
provides a summary of the supporting information from other MYPs.
6
-------
A. Development of Data
for Advanced Biological
Models
B. Information Technologies
Development and Application
C. Prioritization Method
Development and
Application
LTG III
D. Providing Tools and System
Models for Extrapolation across
Dose, Life Stage, and Species
E. Advanced Computational
Toxicology Approaches to Improve
Cumulative Risk Predictions
Figure 3 -Alignment of the Five Research Track
Track A contains seven projects that are targeted at collecting data to improve various
aspects of the linkages within the source-to-outcome paradigm. Included is a project
characterizing thyroid toxicity in an amphibian model, three projects applying "omic"
technologies to the study of endocrine disruptors in small fish, one project examining gene
expression networks in the rodent uterus exposed to estrogens, one project characterizing the
properties of diesel particles that are toxic to human pulmonary cells, and one project seeking
environmental influences on childhood asthma. These projects are distinguished in that they
have a large data collection effort combined with a computational modeling component. Track B
contains efforts to improve the management of information, both from the chemo-informatics
and bio-informatics prospective, to provide the basis for development of better structure-activity
models and to support interpretation of high throughput data being derived from Track C.
Importantly, Track B is designed to migrate the maximum amount of information into the public
domain so it can be used by others for a variety of purposes. Track C contains four projects
intended to speed the development of data for hazard characterization, and includes components
on predicting xenobiotic metabolism, molecular docking models for in-silico predictions of
ligand-receptor interactions, high throughput data acquisition to fingerprint chemicals for
potential hazard, and development of technologies to identify sources of fecal contaminants in
aquatic environments. Track D contains seven projects that are primarily computational in
nature, with efforts ranging from statistical considerations for model parameter fitting, to
providing generic tools for cross species and cross life stage extrapolations, exploring model
portability and linkage issues via use of software standards, modeling particular cell and organ
processes, applying "omic" data in risk assessment, and developing tools and platforms for use
of "omics" in regulatory decision processes. Finally, Track D contains two projects that are
exploring computational approaches to real-world type situations where exposure is to more than
one form of stressor.
7
-------
The outputs of the various projects within each LTG will also provide support for future
activities in other LTGs through relationships depicted in Figure 4.
Interactions Between LTGs
LTG I: Linkages-Risk assessors will
use improved methods and tools to
better understand and describe the
linkages of the source-to-outcome
paradigm.
LTG II: Prioritization-Regulators and
risk assessors will use advanced hazard
characterization tools to prioritize and
screen chemicals for toxicological
evaluation.
1 1
LTG HI: QRA-Assessors and regulators will use
improved methods and models based on the latest
science for enhanced quantitative risk assessment.
Figure 4 -Interactions between LTGs
For example, in projects under LTG II, involving the use of advanced hazard
characterization tools by EPA Program Offices, early efforts are being directed towards
providing new high throughput techniques for prioritization and for identifying those
environmental stressors that should undergo further testing (see project IIC-3). However,
information gleaned from such high throughput technologies are expected to prove informative
for advancing modeling efforts (e.g. projects IIID-5). Preliminary information about mode-of-
action can be extracted from these studies and the models can be exercised using a variety of
scenarios derived from these studies. The models can help identify mode-of-action scenarios that
would contribute most to potential hazard, and risk models could then target on those pathways.
Additional molecular studies could then be conducted, concentrating on those scenarios
identified as having the greatest potential impact. This next phase of information would be used
to further refine the models and reduce uncertainty. Such iterative approaches between controlled
experiments and advanced modeling techniques, while very useful, have proven to be very
expensive and time consuming. The cost and required time could be reduced using some of the
approaches developed under this LTG. The systems biology project (IIID-5) will serve as a
culmination project whereby information from a variety of sources (high throughput, in-vitro, in-
vivo, and in-silico) is used to formulate a risk assessment. In this project, models can be used to
estimate the probability of adverse effects and the variability and uncertainty associated with
those estimates from available knowledge and data. The models can also be used as testing tools
to estimate how much uncertainty could be reduced with more experimental data and also to
identify the type of needed data. Such linkage between LTGs will help advance the overall
program. As noted, LTG II will make significant contributions to LTG III and will also feed
information to LTG I. Similarly, as toxicity pathways are characterized at multiple levels of
biological organization in LTG I, screening tools can be added to projects in LTG II that will
enable larger numbers of chemicals to be characterized. Finally, furthering understanding of
toxicity pathways from projects in LTG I will advance computational work in LTG III, by
examining how various toxicity pathways interact with one another to modulate activity.
8
-------
IV. Summary of Outcomes from Research
The implementation plan described herein was developed to provide a sequential series of
short to mid term projects that will advance utilization of computational tools in hazard and risk
assessment. In particular, over the next 1 to 3 years we expect to make significant progress in
developing and applying prioritization and screening tools. For example, there will be targeted
application of our current understanding of selected toxic endpoints so predictive models can be
developed, leading to more efficient testing paradigms and reduction in uncertainties in inter-
species extrapolation. These models will be populated with data at the molecular and
biochemical molecular levels and will provide a basis to interpret interspecies homology and
comparative toxicity. Proof-of-concept of ToxCast, in particular, will provide a number of EPA
Program Offices with an extremely useful tool to improve the efficiency and effectiveness of
hazard identification and risk assessment methodologies. New and innovative ways will be
developed to assimilate, evaluate, and use the myriad of data assorted with molecular and
chemical information. Further development and application of DSSTox will be a key component
of these efforts. Computational chemistry will provide in-silico models for predicting complex
interactions of environmental chemicals with important biochemical receptors, which can then
lead to adverse effects. Early examples of modeled target interactions include estrogenic,
androgenic receptors, and acetyl cholinesterase.
Progress will be made at developing computational models and modeling systems that
represent comprehensive descriptions of the underlying biology of adverse impacts caused by
exposure to environmental agents. Computational models describing the relationship between
diesel exhaust particle composition and its genotoxic and inflammogenic properties are
examples. Other projects will develop methods to measure and then describe the mechanisms of
childhood asthma resulting from environmental exposures. The whole systems biology modeling
approach will develop a range of models, from those describing pharmacodynamic connections
between exposure and effects, to those describing complex endogenous pathways, and the
perturbations in such pathways resulting from environmental exposures. Also, ways to
incorporate and use "omics" information in these models will be explored. Finally, attempts will
be made at formulating models of common, but complex, disease processes which are then
exacerbated by exposures to exogenous substances and stressors. A detailed summary of outputs
and impacts resulting from the projects listed in the previous Table 1 can be found in Table 2.
V. Partnerships
Given the broad nature of the challenges facing computational toxicology, the CTRP
must engage collaborative partners both across ORD and outside organizations in order to be
successful. Internally, development of these linkages began at the management level with the
initiation of a Memorandum of Understanding (MO A) with the NHEERL and the NERL. This
MOA provides the NCCT with access to administrative support functions such as oversight of
extramural actions, formal quality assurance procedures, and planning for information
management. Access to these and other functions allow for more efficient and effective use of
FTEs within the Center. Management staff of the NCCT has a regularly scheduled monthly
meeting with counterparts in NERL and NHEERL to ensure smooth operations and to strategize
on common scientific directions.
9
-------
With the successful initiation of a cross functional ORD research program supporting the
Framework and the creation of the NCCT in February 2005, the charter of the CTISC was
redefined to include the following functions:
• Serving in an oversight function over the computational toxicology projects
funded by the activities of the committee and coordinate, as appropriate, with
relevant MYPs associated with the projects.
• Establishing partnerships external to the EPA to help support the Framework as
implemented in the various Laboratories and Centers of ORD.
• Advising NCER on the formulation of ideas for new RFAs, providing suggestions
for scientific peer reviewers, serving on relevancy reviews as appropriate, and
keeping informed on the research progress of the STAR grants program.
• Acting in a consultancy role for activities of the NCCT.
• Providing a forum for communication of computational toxicology activities
across ORD and the Agency.
The ORD's multi-year planning process provides another opportunity for linkage
between the CTRP and related research efforts. Each of the MYPs is led by a National Program
Director with support from staff of the relevant Laboratories and Centers. The NCCT participates
actively in four of the MYP teams. Three contain similar research activities for screening and
prioritizing chemicals, i.e. Endocrine Disrupting Chemicals (EDCs. LTG III), Safe
Pesticides/Safe Products (SP2. LTG I), and Drinking Water Research Program (DW, LTG II)
whereas the fourth has a major focus on the incorporation of biologically based mode-of-action
information into quantitative risk assessment (the Human Health Research Strategy and the
Human Health MYP. LTG II). The Director of the NCCT meets at least quarterly with the
National Program Directors for these MYPs, and the NCCT is attempting to allocate at least 10%
of its available extramural resources to computational toxicology-related projects in other
Laboratories and Centers. Preference in deciding which projects to support will be given to
those efforts that have the active participation of a member of the NCCT.
On the scientific level, the NCCT has initiated three Communities of Practice (CoP) in
the areas of Chemoinformatics, Biological Modeling, and Categorization and Prioritization that
are intended to unite practitioners in the designated fields. Prior to implementation of these CoPs,
the concept of 'adjunct' appointments with researchers outside the Center to create a
collaborative environment was considered, but there was no clear idea of how such appointments
would function. The concept of the CoPs was suggested by the BOSC in April 2005, and has
since been adopted as a primary means of communication and integration of activities across
ORD, the EPA, and outside entities. These efforts will serve to enhance communication and
coordination, develop common standards, promote consistency, evaluate and provide guidance
on best practices, recommend research priorities, and provide training to interested parties.
Copies of the charters for these groups are provided here (Chemo; Modeling; and CPCP).
Extending further outside the confines of ORD and the EPA, the NCCT is also
developing strong linkages with the National Toxicology Program of the NIEHS. The
commonalities between A Framework for Computational Toxicology and the A National
Toxicology Program for the 21st Century: A Roadmap for the Future and the co-location on the
10
-------
RTP federal research campus provided a strong impetus for these interactions. Beginning in mid
2005, the two groups began an exchange of work-in-progress seminars that w
led to establishment of a number of specific events, including efforts that
draw on the complementary strengths of the two organizations. These
include a joint seminar program, collaborating on statistical analysis of
complex dose-response patterns, sharing of workgroup and CoP
memberships, developing a mathematical construct for model-building
based on information from varying levels of biological information, meta
analysis of PBPK model development across chemicals, participating in a class study on
perfluornated compounds, and the offering of technical training courses to parties outside the
organizations. Working relationships are also being developed with the National Toxicology
Programs Interagency Center for the Evaluation of Alternative Toxicological Methods
(NICEATM), which is currently exploring high throughput hazard identification tools. These
efforts extend to opportunities for collaboration and linkages with NIH through the NIH
Roadmap. Accelerating Medical Discoveries to Improve Health are also being sought. This
especially concerns the theme, "New Pathways to Discovery: Molecular Libraries and Molecular
Imaging and Building Blocks Biological Pathways and Networks." Discussions with Dr. Chris
Austin, Senior Advisor to Director of Translational Research of the National Human Genome
Research Institute, have led to consideration for inclusion of the EPA-relevant chemicals and
biological assays within the Molecular Libraries Initiative.
VI. Conclusion
Ultimately the CTRP and NCCT will provide outcomes which will help reduce or
prevent the risk to humans and the environment from environmental stressors. As such, the
program is aligned with overarching goals of the Agency. Figure 5 depicts the design of the
program for achieving these goals. Shown is the progression of the research program over the
next several years. We list the resources and inputs that will help define the program and the
activities that are discussed in this plan. Next, we highlight expected research outputs, means for
their transfer, and example clients and beneficiaries of these outputs. In summary, this plan will
communicate the intent or expected outcomes and how those outcomes will improve risk
assessment and management procedures and policies. This will clearly take several years to
come to fruition. Hence this plan, while it outlines a multifaceted process, has concentrated
mostly on the first years where the activities are commenced and specific outputs are produced.
The desire is that the outputs will help achieve short-term outcomes and will greatly contribute to
the results expected in later years.
This plan should be considered a "living document" to be updated and modified on a
yearly basis. These annual updates will reflect and document progress, any necessary change in
direction, and added objectives as the program matures.
11
-------
Research Program
Research Outputs
Inputs and
Resources
Strategic
guidance
•Research Plans:
client needs,
research
questions,
research
priorities,
outcomes,
multi-year plans
(EDC, SP2, DW,
HH), Research
Strategies
•Partnerships
•ORD
Intramural
Research
Resources
•ORD
Extramural
(STAR)
Research
Resources
•Program-level
funding & FTE
Research
Topics &
ORD conducts
research to:
•Develop
improved
methods and
tools to better
understand and
describe the
linkages of the
source to
outcome
paradigm
•Develop
advanced hazard
ch ar ac teriz ation
tools to prioritize
and screen
chemicals for
toxicologic al
evaluation
•Develop new
and improved
methods and
models based on
the latest science
for enhanced
quantitative risk
assessment
Research
Outputs
ORD research
provides:
•Prioritization and
screening tools with
abbreviated testing
protocols to improve
inter-species
extrapolation and
identify potential
toxicities
•New and innovative
ways to assimilate,
evaluate, and use
molecular and
chemical
information
•In-silico means to
predict complex
interations of
pollutants with
important
endogenous proteins
•Modeling systems
that represent
comprehensive
descriptions of the
underlying biology
of adverse impacts
caused by
envi ronmental
pollutants
Clients
Outreach
&
Transfer
•EPA (OPPTS,
OW, OCHP,
OEI, OAR,
Regions
•Risk assessors
(IRIS, RAF)
•Scientific
•Other Federal
„ conferences
agencies (FDA,
and
AT SDR, NIH,
workshops
NTP, DOE)
•Peer-
•States
reviewed
•Tribes
publications
•International
•Technical
health
and decision
organizations
support
and universities
•Publicly
(WHO)
available
•Regulated
data bases
community
and models
•Academic
community
•T ar geted
communi ties
•American
Public
Short-T erm
Outcomes
Independent experts,
risk assessors, or
others in the field:
•Use ORD's
Screening and
prioritization tools to
identify the greatest
potential toxicity for
large number of
compounds or compound
classes:
•Use ORD's
Advanced data base
systems to study the
relationships between
exposure, chemical
properties, molecular
interactions and toxicity
•Use ORD's
methods and
models to better
characterize the
mechanistic basis for
adverse effects at
molecular level for risk
Assessment purposes
• Use ORD's
methods and models to
characterize
effectiveness of risk
management decisions
Outcomes
In ter mecli a te
Outcomes
•Large number
of compounds or
compound
classes of HPV,
pesticides, and
others that may
pose greatest risk
and warrant
further testing
are more quickly
identified
•Reduction on
use and reliance
of animal tests.
•Exposure to
compounds
expected to
produce greatest
risk is reduced
•Dose-response
models based on
latest sound
science are used
in making risk
man a gement
decisions thereby
improving the
effectiveness and
cost effectiveness
of risk
man a gement
decisions
Long-Term
Outcomes
•Risk to
humans and
the
envi r onmen t
from
envi ronmental
stressors is
reduced or
prevented
Externalities that May Impact the CompTox Program
Congressional appropriations and Administration budget decisions, changes in EPA science priorities, changes in EPA regulatory requirements, availability of investment capital,
consent agreements
Figure 5 -Computational Toxicology Research Program Design
12
-------
Table 1- Organization of Projects in the Computational Toxicology Program (The Roman numeral indicates the Long Term Goal (LTG), theA-E letter indicates the Research Track and
the Arabic numeral identifies the project number. Projects shown in blue are projects largely external to the NCCT; projects in red are primarily conducted within the NCCT; and projects
in green are part of the STAR Program. Note: the STAR Environmental Bioinformatics Centers are not listed here, as they are cross cutting activities).
Long Term
Goal
Research Track
Project Title
I. EPA risk assessors
use improved
methods and tools to
better understand and
describe linkages
across the source-to-
outcome paradigm
A. Development of
Data for Advanced
Biological Models
IA-1 Linkage of Exposure and Effects Using Genomics. Proteomics. and Metabonomics in Small Fish Models
IA-2 Svstems Biologv Modeling of Fathead Minnow Response to Endocrine Disruptors
IA-3 Chemicallv Induced Changes in Gene Expression Patterns along the HPG Axis at Different Organization Levels Using a Small Animal Model CMedaka")
IA-4 A Svstems Approach to Characterizing and Predicting Thvroid Toxicitv Using an .Amphibian Model
IA-5 Estrogen Elicited Gene Expression Network Elucidation in the Rat Uterus
IA-6 Risk Assessment of the inflammogenic and mutagenic effects of diesel exhaust particles: A svstems biologv approach
IA-7 Mechanistic Indicators of Childhood Asthma ("MICA")
II. EPA Program
Offices use advanced
hazard
characterization tools
to prioritize and
screen chemicals for
toxicological
evaluation
B. Information
Technologies
Development and
Application
IIB-1 Integrated Chemical Information Technologies Applied to Toxicologv
C. Prioritization
Method Development
and Application
IIC-1 Simulating Metabolism of Xenobiotic Chemicals as a Predictor of Toxicitv
IIC-2 Modeling Molecular Targets for Toxicitv. a Computational Approach to Understanding Kev Steps in the Mechanisms for Toxicitv and a Tool for Prioritizing
Bioassav Requirements
IIC-3 ToxCast. a Tool for Categorization and Prioritization of Chemical Hazard Based on Multi-Dimensional Information Domains
IIC-4 Development of microbial metagenomic markers for environmental monitoring and risk assessment
III. EPA assessors
and regulators use
new and improved
methods and models
based on the latest
science for enhanced
dose-response
assessment and
quantitative risk
assessment.
D. Providing Tools
and System Models
for Extrapolation
Across Dose, Life
Stage and Species
IIID-1 Statistical Methodologv for Estimating Parameters in PBPK/PD Models
IIID-2 Modeling Toxicokinetics for Cross-Species Extrapolation of Developmental Effects
IIID-3 Development of a portable software language for phvsiologicallv-based pharmacokinetic CPBPIO models
IIID-4 Svstems Modeling of Prostate Regulation and Response to Antiandrogen
IIID-5 Svstems Biologv Model Development and Application.
IIID-6 Use of Toxicogenomics Data in Risk Assessment: Case Studv for a Chemical in the Androgen-Mediated Male Reproductive Development Toxicitv Pathwav
IIID-7 Developing Computational Tools for Application of Toxicogenomics to Environmental Regulations and Risk Assessment
E. Advanced
Computational
Toxicology
Approaches to
Improve Cumulative
Risk Predictions
IIIE-1 Dose-Time-Response Modeling for Evaluating Cumulative Risk of N-Methvl Carbamate Pesticides
IIIE-2 Application of Visual Analvtic Tools to Evaluate Complex Relationships between Environmental Factors and Health Outcomes
Table 1 -Identification of CTRP Projects
13
-------
Table 2 -CTRP Output/Input Table
I. EPA risk assessors use improved methods and tools to better understand and describe linkages across the source to outcome paradigm
A. Development of Data for Advanced Biological Models
Project Number & Title
Outputs/Outcomes 06
Outputs/Outcomes 07
Outputs/Outcomes 08
Expected Impacts
I A-1 Linkage of
Exposure and Effects
Using Genomics.
Proteomics. and
Metabonomics in Small
Fish Models
Development of a conceptual model for the
HPG axis in small fish models as a basis for
focused hypothesis testing of potential
endocrine disrupting chemicals.
Preliminary results on the effects of
model chemicals on the fecundity of
fathead minnows and single gene and
protein expression.
Gene and protein expression as the
basis for extrapolation of the effects of
endocrine disrupting chemicals across
small fish models.
OPPTS - Development and validation of screening
test methods for chemicals impacts the HPG in
humans and wildlife species and for the
incorporation of mechanism-specific data into
OPP's probabilistic risk assessments.
IA-2 Systems Biology
Modeling of Fathead
Minnow Response to
Endocrine Disruptors
To determine and compare gene and protein
expression profiles and physiological and
reproductive endpoints for adult FHM
exposed to a model estrogen 17 alpha-
ethinylestradiol (EE2), androgen (17p-
trenbolone), or their antagonists (ZM
189,154 and flutamide, respectively).
To predict gene expression patterns of
two compounds (zearalenone and
EE2) that are environmental
estrogens.
To develop a computational modeling
framework that integrates exposure
concentration, gene expression, and
proteomic profiles with physiological
endpoints.
This STAR Grant is expected to develop a
computational model and identify 10-15 molecular
and protein biomarkers that are specific and
predictive of adverse effects of exposure to
estrogenic compounds in reproduction of fathead
minnows. This quantitative model will help
improve risk assessment of exposure of wildlife
and by extrapolation, of mammals to endocrine
disrupting compounds.
IA-3 Chemically
Induced Changes in Gene
Expression Patterns along
the HPG axis at Different
Organization Levels
Using a Small Animal
Model CMedaka")
Determine natural backgrounds and
variability of gene expressions, optimize
methods based on findings, develop
methods to identify gene products and begin
exposure studies with model chemicals.
Finalize exposure studies and
identification of genomic expression
patterns based on four model
compounds and perform statistical
comparisons. Quantify gene products
and determine their functionality and
biological relevance. Compare
responses of test chemicals to model
chemicals. Submit final report on
project.
A model will be developed to predict the biological
relevance of the observed changes in gene
expression profiles By identifying the systemic
target sites, and the series of biological events from
gene expression to the manifestation of an adverse
outcome (e.g., reproductive performance),
thresholds at the molecular level that are indicative
of effects on the fitness of the individual, including
survival, growth and reproduction (fertility and
fecundity as well as survival of the offspring) will
be determined. Understanding the potential of
individual chemicals and complex environmental
mixtures to interfere with molecular pathways of
concern will enhance our understanding of the
basic mechanisms of toxicities, and thus, will have
the potential to develop better focused, more rapid
and cost effective models for quantitative risk
assessment. The bioassay can be used to screen for
a wider range of endocrine disruptor effects.
14
-------
I. EPA risk assessors use improved methods and tools to better understand and describe linkages across the source to outcome paradigm (cont.)
A. Development of Data for Advanced Biological Models (cont.)
Project Number & Title
Outputs/Outcomes 06
Outputs/Outcomes 07
Outputs/Outcomes 08
Expected Impacts
IA-4 A Systems
Approach to
Characterizing and
Predicting Thvroid
Toxicity Using an
Amphibian Model
Establish data management system;
Formulate initial systems model for
thyroid function; Develop in-vitro
thyroid and pituitary culture systems.
Characterize molecular changes
associated with TH inhibition;
Characterize specific molecular
responses in thyroid gland culture;
Develop initial QS AR for NIS
inhibition.
Refine systems model for relating
MO A to outcome; Complete first
round QS AR hypothesis for testing
for NIS.
Development of a sufficient understanding of the
HPT so that predictive models can be developed,
testing protocols can be abbreviated, and efforts in
inter-species extrapolation can be improved.
Populate these models with data specifically at the
molecular and biochemical levels which will
provide a basis to interpret interspecies homology
and comparative toxicity.
IA-5 Estrogen Elicited
Gene Expression
Network Elucidation in
the Rat Uterus
Establish estrogenic endocrine elicited dose-
and time-dependent changes in rat uterine
gene expression using four estrogen
receptor ligands; investigate role of ER in
mediating changes in gene expression.
Phenotypically anchor changes in
gene expression to histopathological
outcomes. Develop a computational
model that describes the estrogen
elicited gene expression network;
Submit final project report.
The models developed will identify gene
expression changes most highly associated with
EED elicited histopathological uterine responses.
Examination of multiple environmental endocrine
disruptors with varying potencies will also identify
key regulatory nodes responsible for eliciting these
responses, which could lead to the development of
high throughput endocrine disruptor screening
assays for chemicals in commerce. The data and
resulting models can also be integrated with other
algorithms (i.e. PBPK) to create a more
comprehensive model of the hypothalamic-
pituitary-gonadal axis.
IA-6 Risk Assessment of
the Inflamagenic and
Mutagenic Effects of
Diesel Exhaust Particles:
A systems Biology
Approach
Generate the first four DEP samples;
Conduct chemical characterization of the
DEP samples; Conduct a pilot study
assessing the inflammogenicity and signal
transduction profile of two DEP samples in
human and rodent airway epithelial cells.
Generate eight DEP samples; Identify
overlaps in human and mouse gene
expression patterns associated with
DEP inflammogenicity or
mutagenicity; Perform a bioassay-
directed fractionation on the available
DEPs and determine the distribution
of mass and mutagenicity among the
fractions.
Generate remaining four DEP
samples; Define signal transduction
pathways involved in inflammogenic
and mutagenic responses to DEP
exposure based on gene expression
patterns; Complete biological models
of relevant signaling pathways.
Predictive models that quantitatively describe the
relationship between diesel exhaust particle
composition and its genotoxic and inflammogenic
properties: Well developed and scientifically
based risk assessments for diesel exhaust particles
including the use of genomics and proteomics in
the risk assessment
NOTE: for 4th year (09): Complete a database of
all biological measurements and analytic results,
and the transfer of this data to modelers and risk
assessors for the preparation of manuscripts;
Develop exposure models based on ambient
exposure data, geographic information (locations
of roadways and major sources of PAHs and
metals), and home/school locations; Compare
exposure model estimates with biomarkers of
exposure and early effects.
15
-------
I. EPA risk assessors use improved methods and tools to better understand and describe linkages across the source to outcome paradigm (cont.)
A. Development of Date for Advanced Biological Models (cont.)
Project Number & Title
Outputs/Outcomes 06
Outputs/Outcomes 07
Outputs/Outcomes 08
Expected Impacts
1A-7 Mechanistic
Indicators of
Childhood Asthma
(MICA)
Collect bloods and isolate RNA
from rodent lung and blood
samples and blood; Develop QA
plan for gene expression analysis;
Develop and seek approval for
Intramural Research Protocol
including recommended and
standard operating protocols for
collecting pilot study gene
expression analysis data.
Develop relevant educational
modules appropriate for
schoolchildren and their
parents to enhance participant
recruitment and ensure their
informed consent; Develop
asthma severity score criteria
based on asthma diagnosis,
medicine and FEV1 in school-
based questionnaires and lung
function measurements for
selection of schoolchildren;
Develop standard operating
procedures for the collection,
processing, and analysis of
biological samples from
children; Complete monitoring
of ambient exposure levels
indoor/outdoor.
Develop plan for collection
and analysis of biological
samples; Collect, prepare
process and analyze human
biological samples from 200
children; Prepare
comprehensive data base
NOTE: for 4thyear (09):
Complete a data base of all
biological measurements and
analytic results and the
transfer of this data to
modelers and risk assessors
for the preparation of
manuscripts; Develop
exposure models based on
ambient exposure data,
geographic information
(locations of roadways and
major sources of PAHs and
metals), and home /school
locations; Compare exposure
model estimates with
biomarkers of exposure and
early effects.
Research would be associated with
Annual Performance Goal (APG) #6: By
2012, provide risk assessors and mgrs
with methods and tools for measuring
exposure and predicting effects in
children, including adolescents,
characterizing cancer and non-cancer
hazards and risk to children, and
reducing risks to children in schools
from harmful enviromnental agents;
Enhancement of quantitative risk
assessment, produce better methods and
predictive models for quantitative risk
assessment and to provide a useful tool
for large-scale biomonitoring in humans;
Better determination of the shape of the
exposure-response curve, especially in
the low-exposure region, through the
incorporation of gene expression data
into experimental systems; The
development of more accurate,
biologically-based mathematical
exposure-response models that predict
responses outside the range of
experimental values; The identification
of regulatory metabolic or physiologic
pathways, that may act in concert and
lead to adverse health outcomes, through
the evaluation of multiple health end
points with linkages to gene expression
changes.
16
-------
II. EPA Program Offices use advanced hazard characterization tools to prioritize and screen chemicals for toxicological evaluation
B. Information Technologies Development and Application
Project Number & Title
Outputs/Outcomes 06
Outputs/Outcomes 07
Outputs/Outcomes 08
Expected Impacts
IIB-l Integrated
Chemical
Information
Technologies
Applied to
Toxicants
05: Incorporate DSSTox standard
chemical fields into CEBS; Create
DSSTox database and
documentation files for NTP
Immunotox database; create
chemical index file for the EPA
IRIS programs; Establish
Communities of Practice -
Chemoinformatics Workgroup to
begin to coordinate efforts across
the Agency to inventory, retrieve,
and explore chemical information
data.; Propose plan to link DSSTox
effort with the NLM PubChem
project. 06: Create DSSTox
standard chemical field index files
for public genomics databases and
NTP legacy toxicity data;
coordinate linkages within the
CEBS relational search
enviromnent; Create and publish
DSSTox database and
documentation files for additional
published toxicity databases and
chemical index files for EPA and
NTP programs; Propose plan to
structure-index and quality review
chemical information in EPA data
files currently on the web and to
provide for an EPA-wide structure
browser; Assist with formation of
proposed toxicity chemicals subset
for high-throughput testing in
ToxCast and the NIH Molecular
Libraries Screening Initiative
collaboration.
Assist with incorporation
chemical structure browser
technology into EPA
ArrayTrack, and full with
DSSTox data files and
structure-indexed public
genomics data; Establish
procedures and protocols for
automating the chemical
annotation of new data
submitted to CEBS or EPA
ArrayTrack from the DSSTox
Master chemical list; Continue
expansion of the DSSTox
public toxicity database
inventory; Implement plan to
uniformly structure-index
EPA data files and provide
EPA website structure-
searchability through curated
EPA chemical data files;
Create structure-annotated
database of test results for
toxicity subset chemicals from
the NTP/NIH Molecular
Libraries Initiative.
Advise and assist with the
implementation of chemical
definitions (e.g., assigning a
chemical to a class) and
structure analog searching
capability across chemically
indexed data files, integrated
with toxicogenomics data and
bioinfonnatics capabilities, to
serve as the foundation for
chemoinformatics capabilities
across EPA and NTP data;
Advise and assist with the
development of procedures
and capabilities for deriving
chemical signatures for
predicting toxicity outcomes
from the complete profile of
ToxCast data; Begin to
tabulate and explore data from
NTP/NIH Molecular Libraries
screening collaboration in
relation to other DSSTox
databases; Implement
procedures for expanding the
structure-annotation of EPA
chemical data records and
providing methods for flexible
structure or analog searching
on the EPA website.
Improve capabilities to access, mine, and
integrate useful chemical-biological
activity information from existing and
new data, both within and outside EPA.
These efforts have the potential to
impact a wide variety of EPA program
offices that heavily rely on chemical
information resources, such as the High-
Production Volume Testing Program,
the Premanufacture-Notification
Program in OPPTS, ORD's IRIS
Program, and the Office of Pesticide
Programs.
17
-------
II. EPA Program Offices use advanced hazard characterization tools to prioritize and screen chemicals for toxicological evaluation (cont.)
C. Prioritization Method Development and Application
Project Number & Title
Outputs/Outcomes 06
Outputs/Outcomes 07
Outputs/Outcomes 08
Expected Impacts
IIC-1 Simulating
Metabolism of
Xenobiotic Chemicals as
a Predictor of Toxicity
For priority chemicals, incorporate available
chemical metadata, metabolism data, and
metabolic maps into a searchable database
for data management and
structure/substructure searchable access;
Forecast metabolic pathways for selected
priority chemicals using existing simulator
for liver metabolism; Initiate in-vitro Phase
I liver microsome experiments; incorporate
new laboratory and literature metabolism
data into simulator training sets.
Confirm formation of predicted
metabolites for priority chemicals and
compare observed maps to forecasted
maps; Develop and approach to
evaluate and enhance simulator
performance through improvement of
transformation probability estimates
and expansion of transformation
reaction domain; evaluation of refined
simulator and model predictions in
context of Program Office
prioritization needs.
Provide effects-based prioritization of chemicals
(parent compound and metabolites) relevant to EPA
Program Offices: Provide OPPT and OPP the ability
to prioritize chemical lists (based upon predicted toxic
effects of parent chemical and metabolites) with
reliability estimates for use in chemical evaluations
and to rank chemicals for in-vitro or in-vivo screening
and toxicity testing; provide capability to OPP and
OPPT for predicting bioactive metabolites; develop
searchable metabolism database for OPP and OPPT
use for identification of relevant chemical and/or
substructures of interest for risk assessment; provide
linkage of effects based toxicity model with metabolic
simulation.
IIC-2 Modeling
Molecular Targets for
Toxicity: a
Computational Approach
to Understanding Key
Steps in the Mechanisms
for Toxicity and a Tool
for Prioritizing Bioassav
Requirements
Results of the use of the target-toxicant
paradigm to screen for
estrogenicity/androgenicity; Results of the
affect of the two binding site model on the
cumulative risk of chemicals acting through
the enzyme AChE.
Results of the importance of protein
flexibility in evaluating the interaction
of chemicals with macromolecular
receptors.
Evaluation of the use of the target-
toxicant method as a tool in a diverse
chemical screen; Application of the
target-toxicant approach to other
targets of toxicity.
Help fulfill the Agency need for predictive models for
hazard identification, both the sub areas of QS AR and
other computational approaches and High Throughput
Screening.
IIC-3 ToxCast: A Tool
Develop conceptual framework for ToxCast
(started in 05); establish initial battery of
assays across the information domains,
identify list of chemicals to evaluate proof
of concept for framework and begin data
acquisition.
Report on the utility of statistical
clustering techniques on assay results
from pilot chemicals to group them
according to known toxicity patterns;
revise framework as dictated by
results.
A biologically and chemically based system to begin
to associate chemicals of like properties and activities
will provide a number of EPA Program Offices with
an extremely useful tool that addresses the mission of
improving the efficiency and effectiveness of hazard
identification and risk assessment methodologies
employed by the EPA.
for Categorization and
Prioritization of
Chemical Hazard Base
on Multi-Dimensional
Information Domains
IIC-4 Development of
Microbial Metagenomic
Markers for
Environmental
Monitoring and Risk
Evaluation of metagenomic databases as a
source of molecular markers to assess
human and animal fecal contamination in
surface waters.
Report on the evaluation of PCR-
based host-specific assays to confirm
the presence of animal fecal sources
of pollution in waters impacted by
fecal contamination; Report on fecal
indicator microorganisms and/or
genetic markers from fecal material
whose densities in recreational waters
best correlate with the rates of
illnesses in users of recreational
waters.
Development of assays that can be used in source
identification, environmental monitoring, and risk
assessment and to distinguish between human and
nonhuman fecal contamination in Nation's waterways
and supplies.
18
-------
III. EPA risk assessors and regulators use models based on the best science to reduce uncertainties in the dose-response and cross-species
extrapolations.
D. Providing Tools and System Models for Extrapolation Across Dose, Life Stage, and Species
Project Number & Title
Outputs/Outcomes 06
Outputs/Outcomes 07
Outputs/Outcomes 08
Expected Impacts
IIID-1 Statistical
Methodology for
Estimating Parameters in
PBPK/PD Models
Submission of a journal article outlining
outstanding statistical issues in the analysis
of PBPK models and providing methods for
use of formal statistical methods for more
reliable and rational estimates of key
parameters and for model evaluation.
Joint NIEHS/NCCT workshop on
Statistical Issues in PBPK/PD
modeling to establish expert based
consensus on best practices in
statistical analysis and evaluation of
PBPK/PD models.
Publication of framework for formal
statistical analysis of PBPK/PD
models.
Better characterization of methods for estimating
parameters and quantifying uncertainties in
pharmacokinetic and pharmacodynamic models
predictions will remove one major impediment to
their more general application. This will allow
replacement of default uncertainty factors with
transparent mechanism-based statements of scale and
uncertainty, in turn decreasing the subjectivity and
increasing the transparency of environmental health
risk assessments, impacting several program offices
including OPP, OPTS, and OW.
IIID-2 Modeling
Toxicokinetics for Cross
Species Extrapolation of
Developmental Effects
Model rat pup lactational exposure for
perfluooctanoate (PFOA) using
compartmental approaches.
Develop initial physiologically-based
pharmacokinetic model for PFOA in
rat maternal-fetal-pup unit and
identify data gaps in relation to rat and
human.
Evaluate modeled dosimetry for rat
fetus and pup for a limited number of
prototype compounds to inform the
uncertainty in use of maternal
exposure dose in risk assessments.
Characterization of internal dosimetry will inform the
uncertainties attendant to analyses based upon the
maternal exposure dose. In the presence of
information on the critical window, the models may
directly form the basis of the quantitative risk
assessment by deriving the relevant internal dose
metric for extrapolation to humans. Perfluorinated
compounds, including PFOA, are an important proof
of concept for this research because they are a class of
compounds producing developmental effects of
significant regulatory concern to OPPTS and others.
These compounds are currently under evaluation by
the Agency.
IIID-3 Development of
Portable Software
Language for PBPK
Models
Report on findings of assessing
extensiveness required for the XML
schema.
NCCT-sponsored meeting on
evaluation of the performance of the
completed XML schema and status of
the visualization program.
Report on proposal to have finalized
XML schema included in the next
release of SBML and evaluation of
visualization program.
Enhance the ability of formulating and investigating
quantitative dose-response relationships using mode
and mechanism of action knowledge and data: The
ability to create, transfer, use, augment, and review
PBPK models without the limitation of software
compatibility are a long-standing desire in the PBPK
community. The ability to seamlessly link PBPK
models to biological pathway models is of great
interest, since this gives the modeler a fast and
efficient means of extending the estimations of tissue
dose from the PBPK model to cellular-level
responses.
19
-------
III. EPA risk assessors and regulators use models based on the best science to reduce uncertainties in the dose-response and cross-species
extrapolations, (cont.)
D. Providing Tools and System Models for Extrapolation across Dose, Life Stage, and Species (cont.)
Project Number & Title
Outputs/Outcomes 06
Outputs/Outcomes 07
Outputs/Outcomes 08
Expected Impacts
IIID-4 Systems Modeling
of Prostate Regulation
and Response to
Androgens
Prostrate function model following
castration.
Dose-response relationship with
testosterone and antiandrogen
exposure.
Biologically based model of prostate
androgen dependent gene regulation
incorporating genomics data.
Development and demonstration of a biologically
based model of endogenous endpoints and their
pertubationscaused by exposure to environmental
factors. A true pharmacodynamic model that can be
used by OPPTS and others in mechanistically based
risk assessments.
HID 5 Systems Biology
Model Development and
Application
Journal article on use of biologic models to
ascertain necessary resolution of exposure
measurements; Collaborative groups
formed; Formulation of conceptual model
and writing the mathematics and code for
testing models; Selection and begin model
implementation for endogenous biochemical
system; Start model development of disease
process.
Implementation of model and begin
application for cases with exposures to
known environmental toxicants -
abstracts and presentations; Disease
model coded, exercised, and
evaluated; Determine through
literature search and other means for
examples of exogenous exposure that
impact the disease process.
Journal articles illustrating some uses
of "omics" information in quantitative
models; Summary report for Agency
use on earliest best practices on
"omics" and system biology models;
Journal article for disease model;
Enhancement of disease model to
incorporate exposures to
environmental toxicants.
The latest mechanistic information and the interplay
between exposure, endogenous factors, pre-existing
conditions, and genetic predisposition can be
rationally accounted for by using such models. There
is a great need for quickly illustrating how the latest
molecular information, especially "omic" information
and information coming from high through-put
studies will be used in risk assessments by NCEA and
several program offices, including OPPTS, OW,
OAR.
IIID-6 Use of
Toxicogenomic Data in
Risk Assessment: Case
Study for a Chemical in
the Androgen-Mediated
Male Reproductive
Development Pathwav
Draft of the case study report (includes
scoping exercise complete for case studies;
i.e., progress on case study and defining
approach for integrating TG data into risk
assessment); Discussions with the EPA
chemical assessment team, Regions, and
Program Offices.
Conduct EPA Colloquium presenting
results and lessons learned from the
case study; External peer review draft
of report; Submit manuscript on
project to peer-reviewed journal.
This project was developed in response to a
recommendation from the NCEA sponsored
Genomics and Risk Assessment Colloquium of 2003.
Specifically, it will conduct a case study to provide a
practical attempt to incorporate currently available
toxicogenomics data that would illuminate issues and
the methods development. The results will help EPA
prepare for genomics data availability and submission
by addressing areas of risk assessment where such
data may be particularly useful; analyzing acceptance
criteria for inclusion of toxicogenomics data in risk
assessment; and incorporating approaches for the use
of toxicogenomics in risk assessment.
IID-7 Developing
Computational Tools for
Application of
Toxicogenomics to
Environmental
Regulations and Risk
Assessment
Completion of participation in the
Microarray Quality Control (MAQC) with
FDA and publication of papers describing
best practices- includes description of SPC
Genomics Technical Framework.
Installation of FDA ArrayTrack database for
ORD and Agency use.
Continued development of
ArrayTrack database and analytical
tools for toxicogenomics in
cooperation across Agency, with
FDA, and with the NC and NJ
Environmental Bioinformatics
Centers. Trial incorporations of
toxicogenomic data into risk
assessments in collaboration with
various Program and Regional Office
staff.
Publication of examples and
principles for integrating
toxicogenomic data into risk
assessments in peer-reviewed
scientific journals and contribution of
these principles into Agency science
policy.
Development of these toxicogenomic databases and
tools, and application of these various toxicogenomic
data within Program and Regional Offices will
provide EPA staff with valuable, practical training in
genomics and associated disciplines. As
toxicogenomics grows more important to
environmental science and policy, such activities will
help EPA develop the computational tools and
methods to properly evaluate genomics information.
20
-------
III. EPA risk assessors and regulators use models based on the best science to reduce uncertainties in the dose-response and cross-
species extrapolations, (cont.)
E. Advanced Computational Toxicology Approaches to Improve Cumulative Risk Predictions
Project Number & Title
Outputs/Outcomes 06
Outputs/Outcomes 07
Outputs/Outcomes 08
Expected Impacts
IIIE-1 Dose-Time -
Response Modeling for
Evaluating Cumulative
Risk of N-methvl
carbamates
Science Advisory Panel review (August 05)
of preliminary cumulative risk assessment,
including dose-response modeling; release
and Science Advisory Panel review of
revised cumulative risk assessment;
submission of methods and dose-response
models for publication in the peer-reviewed
literature.
Incorporation of dose-time-response
methodology into the Agency's
BMDS software.
The dose-response results of this work will be used
by OPP in the dose-response assessment portion of
the Agency's N-methyl carbamate cumulative risk
assessment. Methods developed in this analysis could
be used in the future for cumulative risk assessments
for agents with ephemeral acute effects.
IIIE-2 Application of
visual analvtic Tools to
Evaluate Complex
Relationships between
Environmental Factors
and Health Outcomes
Demonstration of potential for application
of VA to evaluate exposure data and
workshop on applying VA to analyze
children's cohort data.
Generic conceptual model of complex
relationships between environmental
factors and human health outcomes.
Demonstration of application of V A to
analyze children's cohort data.
Develop multi-factorial analyses methods to conduct
national-scale regulatory-based risk assessments
(program offices); to conduct community-based risk
screening and remediation (regions and states); to
support epidemiology studies investigating gene-
environment interactions (interagency); and to
characterize exposure and risk for public health
tracking. The results of this effort may be used to
develop concepts and tools for application to the
Detroit Children's Study, the North Carolina Cohort,
and the National Children's Study.
21
-------
FTEs From ORD Multi-year Plans Associated with the Computational Toxicology New Start Projects Initially Funded in FY05
(Note: FTEs distributions as indicated in the funded project proposal).
ID
Title
Lead
L/C/O
FTEs
Multi Year Plan
PM
AT
DW
WQ
ECO
SP2
EDC
HH
CT
HS
HHRA
IA-l
Linkage of Exposure and
Effects Using Genomics,
Proteomics and
Metabonomics in Small
Fish Models
NERL
12
5
5
2
IA-6
Risk Assessment of the
inflammogenic and
mutagenic effects of
diesel exhaust particles: A
systems biology approach
NHEERL
2.6
1.2
0.4
1.0
IA-7
Mechanistic Indicators of
Childhood Asthma
(MICA)
NHEERL
5.6
1.9
0.2
1.2
0.2
1.6
0.2
0.3
IIC-1
Simulating Metabolism of
Xenobiotic Chemicals as
a Predictor of Toxicity
NERL
3.7
1.2
2.5
IA-4
A Systems Approach to
Characterizing and
Predicting Thyroid
Toxicity Using an
Amphibian Model
NHEERL
6.6
1.8
4.7
0.1
IIC-4
Development of microbial
metagenomic markers for
environmental monitoring
and risk assessment
NRMRL
4.6
4.6
IIID-
6
Use of Toxicogenomic
Data in Risk Assessment:
workshop and Case
Studies for a Chemical in
the Androgen-mediated
Male Reproductive
Development Toxicity
Pathway
NCEA
1.6
0.2
1.1
Total
36.7
3.1
0.2
1.2
4.6
5
8.6
6.9
2.6
2.8
0.3
1.1
Table 3 -FTE's from ORD's Supporting MYPs
22
-------
Appendix A
1) The Research Center for Environmental Bioinformatics and Computational Toxicology at the
University of Medicine & Dentistry of New Jersey (UMDNJ), Piscataway, which will bring
together a team of computational scientists with diverse backgrounds in bioinformatics,
chemistry and environmental science, from UMDNJ, Rutgers, and Princeton Universities, and
the US Food and Drug Administration's Center for Toxicoinformatics. The team will address
multiple elements of the source-to-outcome sequence for toxic pollutants as well as develop tools
for toxicant characterization. The computational tools developed through this effort will be
extensively evaluated and refined through collaboration between Center scientists as well as with
colleagues from the three universities and the EPA. Particular emphasis will be on methods that
enhance current risk assessment practices and reduce uncertainties. Researchers will also develop
a web-accessible Environmental Bioinformatics Knowledge Base that will provide a user-
oriented interface to an extensive set of information and modeling resources.
2) The Carolina Environmental Bioinformatics Research Center at the University of North
Carolina, Chapel Hill, will develop new analytic and computational methods, create efficient
user-friendly tools to disseminate the methods to the wider community, and apply the
computational methods to molecular toxicology and other studies. The Center brings together
multiple investigators and disciplines, combining expertise in biostatistics, computational
biology, chemistry, and computer science to advance the field of Computational Toxicology.
Researchers will focus on providing biostatistician support to the Center by performing analyses
and developing new methods in Computational Biology. The Center will also create a framework
for merging data from various technologies in a systems-biology approach.
24
-------
Appendix B
Left Blank Intentionally
25
-------
IA-1
Title: Linkage of Exposure and Effects Using Genomics, Proteomics, and Metabonomics in
Small Fish Models.
Lead: David Bencic, NERL
Research Issue and Relevance: Over the past decade there has been a focused international
effort to identify possible adverse effects of EDCs on humans and wildlife. Effects on
reproduction and development mediated through alterations in the HPG axis have been of
particular concern (USEPA 1998). There is convincing evidence that fish in the environment
are being affected by EDCs both at the individual and population levels (World Health
Organization 2002). Unfortunately, because feral fish populations are simultaneously exposed to
multiple stressors, it is difficult, if not impossible; to accurately assess the role of EDCs in
producing adverse impacts (World Health Organization 2002). As a result, many protocols using
fish have been developed and validated, both nationally and internationally, for regulatory
programs for EDCs (Ankley and Johnson 2004). In the US, a 1996 congressional mandate
directed the US Environmental Protection Agency (EPA) to develop a formal screening and
testing program for EDCs. Among the Tier 1 tests recommended (USEPA 1998) to detect
endocrine disruption of the HPG axis is a 21-d reproduction assay with adult fathead minnows
(Ankley et al. 2001). This approach is also being applied to EDC testing with two other small
fish models (i.e., medaka, zebrafish) in other countries via activities coordinated by the
Organization for Economic Cooperation and Development (OECD; Ankley and Johnson 2004).
While of great utility, an important limitation of these tests is that they require significant
investment in time and resources. Furthermore, many of the effects endpoints measured in the
assays are not diagnostic of specific endocrine-related MOA. An ideal screening assay for EDCs
would quickly identify diagnostic endpoints directly indicative of exposure in adult organisms.
Detection of anomalies at the genomic level would enable screening methods of shorter duration
to identify effects at the molecular level, soon after exposure, before they are manifested at the
population level. This research also directly addresses the Computational Toxicology objective
of providing approaches for prioritizing chemicals for testing.
Approach: We propose to use a combination of whole organism endpoints, genomic,
proteomic, and metabonomic approaches, and computational modeling to (a) identify new
molecular biomarkers of exposure to endocrine disrupting compounds (EDCs) representing
several modes/mechanisms of action (MOA) and (b) link those biomarkers to effects that are
relevant for both diagnostic and predictive risk assessments using small fish models. These
goals will be achieved through a three-phase approach that incorporates expertise across
EPA/ORD and capitalizes on partnerships with other federal and academic laboratories. During
Phase 1, effects of a candidate list of nine compounds having different MOA within the
hypothalamic-pituitary-gonadal (HPG) axis will be characterized using the fathead minnow
(Pimephales promelas) Phase 2 will take advantage of the well characterized zebrafish (Danio
rerio) genome, to identify transcriptome and proteome level changes in addition to metabolite
changes, associated with zebrafish exposure to the same suite of EDCs. Phase 2 data will be
used to identify relevant molecular changes that could (a) serve as diagnostic markers for various
types of EDC exposure and (b) begin to inform a systems-level characterization of the responses
26
-------
to those exposures. In Phase 3, candidate genes/diagnostic markers identified in zebrafish (Phase
2) will be validated in fathead minnows through focused toxicological testing aimed at
examination of changes in specific gene expression and protein levels. In this way, changes at
the genomic, proteomic, and metabonomic level will be linked to one another, linked across
multiple teleost species, and ultimately linked to adverse effects at the individual- and, through
modeling, population-level (i.e. linkage back to Phase 1). This three phase effort will identify
new, potentially cost-effective, diagnostic exposure markers for EDCs, and developing source-
to-outcome linkages critical for effective use of biomarkers for risk assessments. This is one of
three objectives of the EPA/ORD Computational Toxicology program (Kavlock et al. 2004).
Progress to Date: This is New Start in the Computational Toxicology Research Program and
was initiated in January 2005.
Impact: The research described in this proposal will directly and indirectly impact several EPA
Program Offices. The most immediate client for the research will by the Office of Science
Council and Policy (OSCP) within OPPTS, who are charged with developing and validating
screening and testing methods for chemicals that impact the HPG axis of humans and wildlife.
ORD (MED) has been working with OSCP for the past 5 years in developing the fathead
minnow as a model species for testing endocrine disrupting chemicals, both in the US and, under
the auspices of the Organization for Economic Cooperation and Development (OECD), through
the world. Because of this existing relationship with OSCP (and OECD), data from the proposed
work have a significant and timely impact on international testing programs for endocrine-
disrupting chemicals. The Office of Pesticides (also within OPPTS) also will benefit from this
research, in two fashions. First, at least two of the test chemicals to be used for these studies
(prochloraz, ketoconazole) are of direct interest to the Program Office with respect to both
human health and ecological effects. This work also will benefit the Environmental Fate and
Effects Division (EFED) within the Office of Pesticides. EFED is responsible for pesticide
registration/risk assessments for fish and wildlife; in doing this, they are moving to probabilistic
risk assessments that incorporate consideration of mechanism-specific data in the context of
population-level effects, an approach parallel to this proposal.
Partnerships/Collaborations: This research is enabled through an extensive network of EPA
and non-EPA partners. The Ecological Exposure Research Division (EERD) in NERL has
extensive facilities and trained personnel to conduct state-of-the-art molecular biology research
in aquatic animals. The Mid-continent Ecology Division (MED) in NHEERL is recognized
throughout the world for cutting edge aquatic toxicology research with small fish species,
including the fathead minnow. The Ecosystems Research Division (ERD) in NERL will be
responsible for metabonomics on the project; this will be achieved through the use of a new 600
mhz NMR. In addition to EPA partnerships on the project a number of external groups will be
involved. The EPA STAR program recently awarded a grant Dr. Nancy Denslow (University of
Florida) for a proposal entitled "Systems biology modeling of fathead minnow response to
endocrine disruptors". The research proposed by Denslow et al. and that described in the present
proposal are extremely complementary. The University of Florida team includes leaders in the
area of ecotoxicogenomics, and strong systems biology component (contributed to the University
of Florida effort by Dr. Karen Watanabe, Oregon Health and Sciences University). Further, their
proposal has a significant emphasis on effects of estrogens on the HPG axis, which is not
27
-------
covered in this proposal. Hence, combining the efforts of EPA and the University of Florida on
this common research problem will result in synergistic outputs. To facilitate this collaboration,
Dr. Denslow has agreed for the University of Florida award to be converted from a grant to a
cooperative agreement.
Milestones/Products:
FY06
Development of a conceptual model for the HPG axis in small fish models as a basis for
focused hypothesis testing of potential endocrine disrupting chemicals.
FY07
Preliminary results on the effects of model chemicals on the fecundity of fathead
minnows and single gene and protein expression
FY08
Gene and protein expression as the basis for extrapolation of the effects of endocrine
disrupting chemicals across small fish models (fathead minnow and zebrafish)
QA: QA Plan is being evaluated
28
-------
IA-2
Title: Systems Biology Modeling of Fathead Minnow Response to Endocrine Disruptors
Lead: Nancy Denslow, University of Florida
Research Issue and Relevance: The U.S. Environmental Protection Agency (EPA) is
interested in the application of novel technologies, derived from computational chemistry,
molecular biology and systems biology, in toxicological risk assessment. In assessing risk
associated with exposure to a chemical or other environmental stressor, a number of scientific
uncertainties exist along a "source-to-adverse outcome" continuum, beginning with the presence
of the chemical in the environment, the uptake and distribution of the chemical in the organism
or environment, the presence of the active chemical at a systemic target site, and the series of
biological events that lead to the manifestation of an adverse outcome that can be used for risk
assessment. The object of this study to develop a computational model to evaluate molecular and
protein biomarkers in relation to reproductive dysfunction in fathead minnows exposed to
environmental estrogens. The model will incorporate a number of biochemical endpoints along
the entire hypothalamic-pituitary-gonadal axis, direct evaluation of physiological changes and
reproductive endpoints and the pharmacodynamics and kinetic distribution of the contaminants.
The hypothesis that we will test is that genomic and proteomics biomarkers will be diagnostic of
the estrogenic effects of environmental estrogens and that they will provide a global
understanding of mechanisms of action that will relate specifically to reproductive endpoints in
FHM that are adversely affected by exposure to estrogenic compounds.
Approach: Fathead minnows will be exposed to three concentrations each of ethinylestradiol
(EE2), and its antagonist ZM 189,154 and to combinations of the two compounds for 48 hrs or
21 days to measure a battery of biochemical, physiological and reproductive endpoints. Short
exposures will be used for gene expression and proteomics studies while both short and long
exposures will be used to measure the physiologic, biochemical and reproductive endpoints.
These data will be brought together in a predictive computational model for the action of
environmental estrogens. The model will then be tested with an exposure of fathead minnows to
zearalenone (estrogen mimic and positive testor) and trenbolone (nonaromatizable androgen and
negative testor), compounds that are used in the cattle industry.
Expected Results: We expect to develop a computational model and identify 10-15 molecular
and protein biomarkers that are specific and predictive of adverse effects of exposure to
estrogenic compounds in reproduction of fathead minnows. This quantitative model will help
improve risk assessment of exposure of wildlife and, by extrapolation, of mammals to endocrine
disrupting compounds.
Specific Aims:
Specific Aim 1: To determine and compare gene and protein expression profiles and
physiological and reproductive endpoints for adult FHM exposed to a model estrogen 17 alpha-
ethinylestradiol (EE2), androgen (17P-trenbolone), or their antagonists (ZM 189,154 and
flutamide, respectively).
-------
Specific Aim 2: To predict gene expression patterns of two compounds (zearalenone and EE2)
that are environmental estrogens.
Specific Aim 3: To develop a computational modeling framework that integrates exposure
concentration, gene expression, and proteomic profiles with physiological endpoints.
Progress Summary: Endocrine disrupting compounds (EDCs) can target the HPG axis at
different levels of complexity. These levels span from the molecular level to the whole
organism. Our work is centered on understanding how changes at the molecular and protein
levels affect downstream physiological endpoints and how well these biomarkers predict adverse
effects on reproduction. There are many published studies that have shown a direct impact of
EDCs on the reproductive success of fish and wildlife. These compounds include certain
organochlorine pesticides, industrial compounds, pharmaceuticals, plasticizers, surfactants, and
metals. These compounds have been shown to alter fertility, fecundity, egg hatchability, survival
of young, sex ratio, and other reproductive parameters. Although these endpoints have high
ecological value, they do not point to specific mechanisms of action. We are pursuing a
"Systems Toxicology" approach in which we apply molecular tools to understand global
mechanisms of action that are affected by chemical exposure. We will relate these changes to
reproductive endpoints using a novel physiology-based mathematical model.
In Year 1 of the project, significant progress was made in the formulation of a physiologically
based model of the HPG axis in male FHM. Following the suggestions of our proposal
reviewers, a modular approach to model development is being taken so that useful submodels
will be developed in the process of constructing an integrative model representing multiple
scales of biological organization (from molecular-level gene expression to physiological-level
reproductive effects). The first submodel to be developed is a physiologically based
pharmacokinetic (PBPK) model that simulates the disposition of EE2 in male FHM.
Milestones/Products:
FY06
To determine and compare gene and protein expression profiles and physiological and
reproductive endpoints for adult FHM exposed to a model estrogen 17 alpha-ethinylestradiol
(EE2), androgen (17P-trenbolone), or their antagonists (ZM 189,154 and flutamide,
respectively).
FY07
To predict gene expression patterns of two compounds (zearalenone and EE2) that are
environmental estrogens.
FY08
To develop a computational modeling framework that integrates exposure concentration,
gene expression, and proteomic profiles with physiological endpoints.
QA: QA Plan is being evaluated
-------
IA-3
Title: Chemically Induced Changes in Gene Expression Patterns along the HPG Axis at
Different Organizational Levels Using a Small Animal Model (Japanese Medaka)
Lead: John Giesy, Michigan State University
Research Issue and Relevance: The EPA is interested in the application of novel technologies,
derived from computational chemistry, molecular biology and systems biology, in toxicological
risk assessment. In assessing risk associated with exposure to a chemical or other environmental
stressor, a number of scientific uncertainties exist along a "source-to-adverse outcome"
continuum, beginning with the presence of the chemical in the environment, the uptake and
distribution of the chemical in the organism or environment, the presence of the active chemical
at a systemic target site, and the series of biological events that lead to the manifestation of an
adverse outcome that can be used for risk assessment. "Endocrine-disrupting" compounds have
been defined as exogenous agents that interfere with the "synthesis, secretion, transport, binding,
action, or elimination of natural hormones in the body that are responsible for the maintenance of
homeostasis, reproduction, development, and/or behavior". Much of the recent concern and
energy has been focused on compounds that are hormone direct agonists or antagonists,
especially those that interact with the estrogen receptor (ER). Effects consistent with exposure to
ER agonists have been observed in fish exposed to natural hormones and some synthetic
chemicals such as nonylphenol (NP), nonylphenol polyethoxylates (NPEs), and octylphenol
(OP). Because chemicals can cause both direct (receptor-mediated) and indirect effects through
changes in signal transduction pathways, methods are needed that permit the screening of
multiple effects. Furthermore, methods are needed that can screen for these effects
simultaneously in a number of tissues, including during critical windows of development, when
tissues may be small and the amount of material available for testing is small and difficult to
remove from the organism. We propose to develop a screening method to use molecular
techniques such as in situ hybridization, in situ RT-PCR and immuno-histochemical staining
(IHCS) to screen for effects of chemicals on the hypothalamic-pituitary-gonadal (HPG) axis with
a special emphasis on steroidogenic pathways and hormonal control mechanisms along the HPG-
axis in the Japanese medaka. The proposed method will allow for screening of multiple effects in
multiple tissues, even at points in development when the tissues are too small to be accurately
dissected for use in more traditional molecular techniques. The proposed methods will be applied
to a set of "model" and "test" compounds for a set of target genes. Once the methods have been
developed and validated, they can be adapted for use with other genes and/or species of interest
and used to efficiently and completely screen for endocrine disruptor effects beyond simple
receptor binding.
Approach: To investigate the tissue-specific expression of genes in the most efficient manner a
variety of in situ molecular techniques will be used to visualize and quantify mRNA and/or gene
products. To maximize sensitivity and permit multiplexed gene expression quantification
methods will be based on in situ hybridization and a variety of in situ PCR techniques. The aim
of these investigations is to produce the simplest and most 'transportable' techniques so the use of
a common technique for all genes investigated will be a priority for the final protocol.
-------
Expected Results: The proposed research program will establish an integrative and quantitative
small fish model to identify and evaluate the modes of action, key target sites, and biological
relevance of EDCs acting along the HPG-axis. To achieve this we will apply state-of-the-art
molecular techniques in a whole animal systems approach (see approach section). The studies
will be conducted in two phases. In the first phase normal physiological processes along the
HPG-axis will be described in order to establish a natural basis for the evaluation of EDC effects.
Information on the natural variability, reproductive cycling, diurnal pattern, needs, etc. will be
used to optimize the proposed test system. Thus, we will establish a general model system that
can be used in the assessment of single chemical and complex mixture effects on reproductive
endocrinology. In the second phase we will use this system to detect and assess changes in
tissue-specific gene expression patterns (GEPs) caused by the exposure to endocrine active
"model" compounds, and their mixtures. We will identify key genes for each individual exposure
scenario and establish a quantitative RT-PCR system as a rapid screening tool for these genes.
A model will be developed to predict the biological relevance of the observed changes in GEPs.
By identifying the systemic target sites, and the series of biological events from gene expression
to the manifestation of an adverse outcome (e.g., reproductive performance), we will identify
thresholds at the molecular level that are indicative of effects on the fitness of the individual,
including survival, growth and reproduction (fertility and fecundity as well as survival of the
offspring). The proposed model test system will provide a new and powerful approach in hazard
identification and toxicological risk assessment. Understanding the potential of individual
chemicals and complex environmental mixtures to interfere with molecular pathways of concern
will enhance our understanding of the basic mechanisms of toxicities, and thus, will have the
potential to develop better focused, more rapid and cost effective models for quantitative risk
assessment. The bioassay can be used to screen for a wider range of endocrine disruptor effects.
Progress Summary/Accomplishments: A subset of genes has been selected for the development
of whole fish in situ hybridization methods. Beta-actin was selected as the housekeeping gene, and
two isoforms of the aromatase gene, CYP19A and B, were selected as initial target genes due to
the fact that their expression is highly tissue specific and responsive to endocrine disruptors.
cDNA and riboprobes have been successfully developed for these genes and currently are being
tested and optimized using cryosectioned tissues and parafin-embedded sections for whole fish
mounts. The in situ hybridization procedures are being optimized in the laboratory. The
methods have demonstrated a very high level of staining for all probes tested. Although some of
the staining seems to be of nonspecific background, the sense and antisense probes do exhibit the
expected binding profiles. Certain factors that may decrease the nonspecific binding are in the
process of being tested, including high stringency posthybridization washes, high temperature
hybridization washes, and high temperature posthybridization washes. At this point, the probes
in use are digoxygenin-labeled. They are visualized using standard protocols.
Milestones/Products:
FY06
NCER needs to provide
FY07
QA: QA Plan is being evaluated
-------
IA-4
Title: A Systems Approach to Characterizing and Predicting Thyroid Toxicity Using an
Amphibian Model.
Lead: Sigmund J. Degitz, MED/NHEERL
Research Issue and Relevance: The EPA was recently mandated to evaluate the potential
effects of chemicals on endocrine function and has identified Xenopus as a model organism to
use as the basis for a thyroid disruption screening assay. The main objective of this work is to
develop a hypothalamic-pituitary-thyroid (HPT) model for this bioassay. This model will provide
a rational framework to organize and interpret toxicological data from the molecular to the
organismal levels and will serve as a basis for development of predictive tools related to thyroid
toxicity. Recent developments in understanding the molecular events involved in TH
homeostasis and action suggest that thyroid toxicity might be identifiable using appropriate
molecular endpoints in an abbreviated test. This potential improvement, which will lower costs
and reduce animal use, is currently more likely to succeed due to expanding genomic
information. This project will focus on building linkages between early molecular events
associated with exposure and organismal-level effects. Understanding these linkages and their
relative importance will be assessed using the HPT model. One of the major goals of this work
is to populate the HPT model with data specifically at the molecular and biochemical levels
which will provide a basis to interpret interspecies homology and comparative toxicity.
Approach: Development of this assay is progressing, but its widespread use on Agency
chemical inventories will be limited due to limited resources. As a consequence, a strategy to
objectively rank and prioritize the order of chemical testing needs to be developed. One of the
most likely uses for a HPT systems model is to aid in the understanding and discrimination of
different toxic modes of action. As such, these models further enable the development of
quantitative structure activity relationships (QSARs) by providing a basis for sorting chemicals
by mode of action, a necessary step prior to quantifying features of chemical structure associated
with a particular type of toxicity. If these relationships can ultimately be established, then
predictive models can be developed to rank chemicals for future in vivo testing. In vivo testing
for HPT effects will be improved through this research by providing a basis to link early
molecular events to organismal outcomes.
There are four specific aims for this research. The first is to develop an HPT systems model which is
capable of integrating data from different levels of biological organization, molecular to organismal,
into a coherent system. The second specific aim is to develop an understanding of the compensatory
mechanisms at the genomic level involved in TH homeostasis and how they respond to chemical
perturbation. The third specific aim is to use in vitro models to help define the functions of
component systems of the HPT system. Both pituitary and thyroid gland cultures will be used as
experimental approaches to define tissue-specific outputs in response to endogenous and xenobiotic
chemical inputs. The fourth specific aim is to use the emerging knowledge of the HPT system to
develop usable, predictive toxicological tools. Once the input-output relationships are developed for
the thyroid glands in culture, this work will be extended to specifically address chemicals known to
inhibit T4 synthesis via disruption of iodine uptake by the sodium/iodide symporter (NIS).
1
-------
Progress to Date: This is a New Start in the Computational Toxicology Research Program that
began in January 2005. Past efforts have been focused on the development of a thyroid screening
assay for OSCP. This assay, which is based primarily on whole organism responses, is proposed to
last 14-21 days beginning in late pre-metamorphosis or early pro-metamorphosis using the
amphibian, X. laevis. We have worked with three T4 synthesis inhibitors (6-propylthiouracil (PTU),
perchlorate (PERC), and methimazole (METH)), two receptor agonists (T4 and T3), and a deiodinase
inhibitor (iopanoic acid (IOP)) in an effort to test the responsiveness of this model system to
chemicals with different modes of action. We have developed dose-response relationships for each of
these chemicals. Interestingly, we have found that in the case of the T4 synthesis inhibitors, clear
signs of thyroid system disruption, as evidenced by thyroid follicular cell hypertrophy and
hyperplasia, occurs at concentrations below those which effect developmental rate. This observation
suggests that the HPT executed a successful compensatory response that permitted normal
development (Tietge et al, 2004; Degitz et al, 2004). More extreme glandular changes were observed
at higher chemical concentrations accompanied by delayed development.
We have initiated studies to examine the changes in molecular and biochemical endpoints
specific to components of the HPT system. As a prerequisite to understanding the changes associated
with exposure to TH synthesis inhibitors, we have been determining the expression patterns of
several genes throughout normal metamorphosis. TSH and type II deiodinase gene expression
patterns in the pituitary, for example, have been developed and demonstrate patterns of up-regulation
coincident with initiation of metamorphic climax These patterns of gene expression are critically
important as they establish the baseline conditions against which changes induced by chemical
exposure will be compared and interpreted.
Impact: The primary benefit of this work is to develop a sufficient understanding of the HPT so
that predictive models can be developed, testing protocols can be abbreviated, and efforts in
inter-species extrapolation can be improved.
Partnerships/Collaborations: This work in being done in conjunction with systems biology
modeler at the PNNL.
Milestones/Products:
FY06
Establish data management system
Formulate initial systems model for thyroid function
Develop in vitro thyroid and pituitary culture systems
FY07
Characterize molecular changes associated with TH inhibition
Characterize specific molecular responses in thyroid gland culture
Develop initial QSAR for NIS inhibition
FY08
Refine systems model for relating MOA to outcome
Complete first round QSAR hypothesis for testing for NIS
QA:
QA Plan is being evaluated
2
-------
IA-5
Title: Estrogen Elicited Gene Expression Network Elucidation in the Rat Uterus
Lead: Tim Zacharewski, Michigan State University
Research Issue and Relevance: The EPA is interested in the application of novel technologies,
derived from computational chemistry, molecular biology and systems biology, in toxicological
risk assessment. In assessing risk associated with exposure to a chemical or other environmental
stressor, a number of scientific uncertainties exist along a "source-to-adverse outcome"
continuum, beginning with the presence of the chemical in the environment, the uptake and
distribution of the chemical in the organism or environment, the presence of the active chemical
at a systemic target site, and the series of biological events that lead to the manifestation of an
adverse outcome that can be used for risk assessment. Systems biology involves the iterative
development of strategies that integrate disparate physiological and biochemical data into
computational models that are capable of predicting the biology of a cell or organism. In order to
facilitate hazard identification and risk assessment, a comprehensive quantitative understanding
of the molecular, cellular, physiological, and toxicological effects that are elicited following
acute and chronic exposure to synthetic and natural chemicals is required within the context of
the whole organism. The objective of this proposal is to develop a computational model that will
identify and predict critical estrogenic endocrine disruptor elicited changes in gene expression
which play a central role in the observed physiological/toxic effects based on systematic and
quantitative data obtained from comparative in silico, genomic, molecular and histopathological
approaches.
Approach: Gene expression and histopathological changes elicited by ethynyl estradiol,
genistein, bisphenol A, o,p'-DDT will be assessed in ovariectomized immature female Sprague-
Dawley rats. Dose-and time-dependent gene expression changes will be determined using
customized sequence-verified cDNA rat arrays enriched with estrogen responsive genes.
Significant changes in expression will be identified and weighed using Bayes and t-statistic
approaches, and verified by quantitative real-time PCR (QRTPCR), western analysis, in situ
hybridization and/or immunohistochemistry. Chromatin immunoprecipitation (ChIP) assays will
further elucidate the estrogen receptor (ER)-mediated mechanisms of action and causal
relationships between genes. In addition, histopathological assessments will be conducted to
distinguish adaptive and toxic responses using various computational methods including
canonical correlation, and Fisher discriminate analyses. Genetic algorithm (GA)/partial least
squares (PLS) analysis will then be used to integrate this disparate data into a model that can
identify the most relevant genes associated with a histopathological outcome. All data will be
captured in dbZach (http://dbzach.fst.msu.edu). a MIAME-compliant toxicogenomic supportive
database that facilitates data analysis, integration of disparate data, and sharing with other
investigators.
Expected Results: The models developed will identify gene expression changes most highly
associated with EED elicited histopathological uterine responses. Examination of multiple EEDs
with varying potencies will also identify key regulatory nodes responsible for eliciting these
responses, which could lead to the development of high throughput endocrine disruptor
screening assays for chemicals in commerce. The data and resulting models can also be
-------
integrated with other algorithms (i.e. PBPK) to create a more comprehensive model of the
hypothal ami c-pituitary-gonadal axi s.
Milestones/Products:
FY06
NCER to provide
FY07
To predict gene expression patterns of two compounds (zearalenone and EE2) that are
environmental estrogens.
QA: QA Plan is being evaluated
-------
IA-6
Title: Risk Assessment of the Inflammogenic and mutagenic effects of diesel exhaust particles:
A systems biology.
Lead: James Samet, NHEERL
Research Issue and Relevance: Diesel exhaust particulate matter (DEP) is a ubiquitous
ambient air contaminant derived from mobile and stationary diesel fuel combustion. Exposure to
DEP is associated with carcinogenic and immunotoxic effects in humans and experimental
animals. At the cellular level, these health effects are underlain by genotoxic and inflammatory
properties of chemical compounds present in DEP. DEP is composed of elemental, inorganic and
organic compounds that vary widely in composition with the source of the fuel, engine operating
conditions, sampling methods and other parameters. The genotoxic and inflammatory potencies
of DEP also vary with its physicochemical properties, and these differences along with multiple
health effects impede the development of targeted regulatory strategies for mitigating the impact
of DEP exposure on human health. While traditional reductive toxicology approaches are not
likely to succeed in quantifying relationships between DEP composition and its numerous health
effects, generating a database for modeling the toxicological effects of DEP would provide a
framework for quantitative hazard identification. This project proposes a systems approach to
developing and applying predictive computational models that quantitatively describe
relationships between the composition of DEP and its genotoxic and inflammogenic potencies.
Approach: The objectives of this project will be met by conducting research in three phases. In
phase 1 (Specific Aim 1), 16 distinct DEP will be generated using a combination of fuels, engine
types, engine loads and collection temperatures. These DEP will then be characterized through
extensive chemical and physical analyses. In phase 2 (Specific Aims 2 and 3), the inflammogenic
and genotoxic potencies of each of the 16 DEP will be determined quantitatively. Specific
bioassays will measure the expression of the pivotal inflammatory mediator IL-8/MIP-2 in
cultured human and mouse lung cells in response to DEP exposure. Signaling mechanisms that
regulate the expression of IL-8/MIP-2 in response to DEP exposure will also be examined in
order to provide mechanistic insight and support for the models. The genotoxicity of the 16 DEP
will be assayed using bacterial mutagenicity assays. Human, mouse and bacterial gene
expression arrays will be used to provide additional mechanistic insights on patterns of gene
expression induced by DEP. Phase 3 (Specific Aim 4) will utilize the generated data to construct
a series of statistical and mathematical models that quantitatively relate DEP composition, its
inflammogenic and mutagenic effects and the relevant intracellular signaling mechanisms.
Progress to Date: See below
Impact: The information generated by this multidisciplinary research program is intended for
use in risk assessment aimed at mitigating the health effects of DEP exposure, including
quantitative and computational approaches, cross-species extrapolation and endpoint validation.
This proposal is directly responsive to priority research needs identified by NCEA, Office of Air
and Radiation (OAR) as well as the ORD PM Research Program. The research will be conducted
using a custom-designed diesel emission sampling system, leading edge genomics and
1
-------
proteomics technologies and the latest tools in bioinformatics and computational modeling
software. Beyond providing biological plausibility in support of DEP regulations, the usefulness
of the findings from these studies will hinge upon identifying toxicological effects mechanisms
in the context of DEP characteristics. Thus, this project's primary aim is to deliver a set of
predictive models that quantitatively describe the relationship between DEP composition and its
genotoxic and inflammogenic properties. An important impact of the process of developing and
applying this set of models via the systems biology approach will be the generation of novel
mechanistic data with the specific intent of identifying and characterizing critical paths that lead
from DEP exposure to toxicity. It is anticipated that many of these data will be applicable to the
study of other sources of particulate air pollution whose effects include mutagenisis or
inflammatory responses. Moreover, application of the systems biology approach will represent a
case study adaptable to computational studies of exposure and toxicological effects of a broad
range of environmental agents. Although the reductive studies will also yield novel mechanistic
information about DEP toxicity, the actual deliverable product of this research will be a series of
computational models that can be used by client offices in support of assessment and regulatory
efforts. These models will have particular practical application in offices responsible for DEP
assessment and regulatory programs including the Office of Air and Radiation (OAR) and the
National Center for Environmental Assessment (NCEA).
Partnerships/Collaborations: The NRMRL high bay facility is designed for combustion
research. Drs Linak and Gilmour have retrofitted a combustion bay complete with dilution tunnel
and baghouse collection units. In addition, the diesel exhaust may be piped to inhalation
exposure chambers. The facility is equipped with a computer-integrated gas and particle-
monitoring bench. The NRMRL research laboratories and the Environmental Carcinogenesis
Division also operate core organic chemistry laboratories for routine solvent extraction and GC-
MS capability. The laboratories of Drs Samet, Reed and Gilmour are equipped with cell culture
areas, nucleic acid extraction and measurement instruments, ELISA, western blot and real time
PCR capability. Dr DeMarini's laboratories have capabilities for high throughput screening of
bacterial mutagenicity assays using a number of Salmonella Sp substrains, and through other
members of ECD has access to methodologies for eukaryotic DNA damage and repair testing.
Additional core support for genomics and bioinformatics will be available to all investigators
through the NHEERL genomics and proteomics Committee (NGPC). As a faculty member of the
University of North Carolina at Chapel Hill (UNC-CH), Dr. Reed has access to the resources of
the UNC-CH Center for Bioinformatics and the UNC Shared Bioinformatics Resource that
provide access to software packages for the analysis of gene expression profiling data, including
GeneSpring (Silicon Genetics), PathArt (Jubilant Biosys) and GenoMax (InforMax).
Milestones/Products:
FY06
Generate the first four DEP samples
Conduct chemical characterization of the DEP samples
Conduct a pilot study assessing the inflammogenicity and signal transduction profile of
two DEP samples in human and rodent airway epithelial cells.
FY07
2
-------
Generate eight DEP samples
Identify overlaps in human and mouse gene expression patterns associated with DEP
inflammogenicity or mutagenicity
Perform a bioassay-directed fractionation on the available DEPs and determine the
distribution of mass and mutagenicity among the fractions.
FY08
Generate remaining four DEP samples
Define signal transduction pathways involved in inflammogenic and mutagenic responses
to DEP exposure based on gene expression patterns
Complete biological models of relevant signaling pathways
QA:
QA Plan is being evaluated
3
-------
IA-7
Title: Mechanistic Indicators of Childhood Asthma (MICA)
Lead: Jane Gallagher NHEERL; Haluk Ozkaynak NERL
Research Issue and Relevance: The US Environmental Protection Agency (EPA) is interested
in the interplay of environmental and genetic factors on the development and exacerbation of
asthma. The Mechanistic Indicators of Childhood Asthma (MICA) study will use exposure
measurements and markers of environmental exposure and health effects. Based on
epidemiological studies of air pollution and asthma, there is sufficient evidence to justify
investigations that incorporate state-of-the art technologies including genomics and proteomics,
to study how and which genes and environmental factors interact in a way that increases the risk
of worsening asthmatic responses. EPA scientists will use collected markers of exposure and
effects to analyze, characterize, and possibly quantify combined risk factors that relate to asthma
severity from multiple agents/stressors. Our study will also provide information on some key
molecular events associated with chemical exposures, giving risk assessors more reliable data to
assist in defining exposure-response relationships and in making estimates on the range of risks
expected in the population compared to data based on biological monitoring and/or screening
level hazard data.
Approach: Phase 1 of MICA includes an assessment gene expression data collected from
rodent blood and respiratory tissues RNA produced by short-term controlled exposures to
concentrated particulate matter. Mobile units housed with rodent exposure chambers (State
University) were employed in Grand Rapids, Michigan and in the Detroit Metropolitan Area.
Phase 1 will provide 1) information on the reliability of surrogate blood RNA samples to predict
target tissues effects and 2) context for the human gene expression data, collected in phase 2
planned for the summer of 2006. Phase 2 is a children exposure assessment/biomarker study
focusing on three broad classes of particulate associated chemicals: volatile organic compounds
(VOCs), metals, and poly cyclic aromatic hydrocarbons (PAHs). MICA will study 150 asthmatic
and 50 non-asthmatic children. Blood and urinary measures of these chemicals will be compared
to benchmark levels of these chemicals and metabolites from the Center for Disease Control and
Prevention's (CDC) National Exposure Report. MICA consists of clinical measurements
including (a) a skin prick test for allergen sensitivity; (b) analysis of blood, urine, nail clippings,
and DNA; (c) immunological markers, odor testing, lung function and breath analysis; (d) gene
expression and protein tests, viewed in the context of environmental assessments and respiratory
health history.
Progress to Date: New start Jan 2005. Rodent exposures completed. Blood RNA isolated from
two species and two sites in Michigan. Isolation of RNA from matching rodent lung samples
ongoing. Collaborations and/or contracts for over 30 exposure and health effects markers have
been identified.
Impact: 1). MICA provides linkages in the source-to-outcome paradigm
2). Enhancement of quantitative risk assessment, produce better methods and predictive models
for quantitative risk assessment and to provide a useful tool for large-scale biomonitoring in
1
-------
humans.
3). Better determination of the shape of the exposure-response curve, especially in the low-
exposure region, through the incorporation of gene expression data into experimental systems.
4). The development of more accurate, biologically-based mathematical exposure-response
models that predict responses outside the range of experimental values.
5)The identification of regulatory metabolic or physiologic pathways, that may act in concert
and lead to adverse health outcomes, through the evaluation of multiple health end points with
linkages to gene expression changes. •Identification of molecular indicators of exposure and
toxicity that can be applied to other epidemiological studies or risk assessment.
More specifically the Office of Air and Radiation (Office of Indoor Air, Office of Air Quality
Standards and Planning and the Office of Transportation and Air Quality) and the Office of
Environmental Information. Our study will provide information on some key molecular events
associated with chemical exposures, giving risk assessors more reliable data to assist in defining
exposure-response relationships and in making estimates on the range of risks expected in the
population compared to data based on biological monitoring and/or screening level hazard data.
Research will support the Government Performance and Results Act (GPRA) goal "Risk
Assessment for Susceptible Subpopulations", Long-Term Goal 4 of the Human Health Multi-
Year Plan (MYP), Objective 4.4 Science/Research, Sub-Objective 4.4.2 Research. More
specifically, the proposed research would be associated with Annual Performance Goal (APG)
#6: "By 2012, provide risk assessors and managers with methods and tools for measuring
exposure and predicting effects in children, including adolescents, characterizing cancer and
non-cancer hazards and risk to children, and reducing risks to children in schools from harmful
environmental agents to support EPA risk assessment and risk management"
Partnerships/Collaborations: (e.g., where is the data coming from?). This is a multi-
disciplinary proposal requiring the expertise of epidemiologists, modelers, toxicologists, and risk
assessors, and the knowledge of computational data analysis and systems biology from NERL
and NHEERL investigators, EPA Region 5 and potentially Detroit's City Health Department
(Department of Health and Wellness Promotion (a recipient of a CDC grant to conduct asthma
education, training and outreach activities). It is expected the DCHD or other recognized
community leaders will have a role in effectively educate parents of the enrolled children related
to potential asthma triggers and how best to communicate health and research data back to the
community participants.
Milestones/Products:
FY05
• Collect bloods and isolate RNA from rodent lung and blood samples and blood
• Develop QA plan for gene expression analysis
• Develop and seek approval for Intramural Research Protocol including recommended and
standard operating protocols for collecting pilot study gene expression analysis data.
FY06
• Develop relevant educational modules appropriate for schoolchildren and their parents to
2
-------
enhance participant recruitment and ensure their informed consent.
• Develop asthma severity score criteria based on asthma diagnosis, medicine and FEVi in
school-based questionnaires and lung function measurements for selection of schoolchildren.
• Develop standard operating procedures for the collection, processing, and analysis of
biological samples from children.
• Complete monitoring of ambient exposure levels indoor/outdoor
FY07
• Develop plan for collection and analysis of biological samples.
• Collect, prepare process and analyze human biological samples from 200 children.
• Prepare comprehensive data base.
FY08
• Complete a data base of all biological measurements and analytic results and the transfer of
this data to modelers and risk assessors for the preparation of manuscripts.
• Develop exposure models based on ambient exposure data, geographic information (locations
of roadways and major sources of PAHs and metals), and home /school locations.
• Compare exposure model estimates with biomarkers of exposure and early effect.
QA:
QA Plan is being evaluated
3
-------
IIB-1
Title: Integrated Chemical Information Technologies Applied to Toxicology
Lead: Ann Richard, NCCT (FTE .0.8)
Research Issue and Relevance: A central regulatory mandate of the Environmental Protection
Agency, spanning many Program Offices and issues, is to assess the potential health and
environmental risks of large numbers of chemicals released into the environment, often in the
absence of relevant test data. Models for predicting potential adverse effects of chemicals based
primarily on chemical structure play a central role in prioritization and screening strategies yet
are highly dependent and conditional upon the data used for developing such models. Hence,
limits on data quantity, quality, and availability are considered by many to be the largest hurdles
to improving prediction models in diverse areas of toxicology. Generation of new toxicity data
for additional chemicals and endpoints, development of new high-throughput, mechanistically
relevant bioassays, and increased generation of genomics and proteomics data that can clarify
relevant mechanisms will all play important roles in improving future SAR prediction models.
The potential for much greater immediate gains, across large domains of chemical and toxicity
space, comes from maximizing the ability to mine and model useful information from existing
toxicity data, data that represent huge past investment in research and testing expenditures. In
addition, the ability to place newer "omics" data, data that potentially span many possible
domains of toxicological effects, in the broader context of historical data is the means for
optimizing the value of these new data.
The challenges for application of information technologies, including chem-informatics and
bioinformatics, are fourfold: 1) to more efficiently migrate legacy toxicity data from diverse
sources into standardized, electronic, open, and searchable forms into the public domain; 2) to
employ new technologies to mine existing data for coherent patterns that can provide scientific
underpinning for extrapolations; 3) to place a new chemical, of unknown hazard, appropriately in
the context of existing data and chemical and biological understanding; and 4) to integrate data
from different domains of toxicology and newer "omics" experiments to look beyond traditional
means for classifying chemicals, inferring modes of action, and predicting potential adverse
effects.
Approach: In the area of improving data resources for structure-based mining, the NCCT is
supporting further development and expansion of the DSSTox (Distributed Structure-Searchable
Toxicity) database network. The DSSTox project is primarily focused on migrating toxicity data
from diverse areas of study into structure-annotated, standardized form for use in relational
structure-based searching and structure-activity model development. An essential element of this
effort involves bridging understanding and forging productive linkages between the toxicology
domain experts and the data users and modelers by means of focus on clarifying the chemistry
content and summary presentation of the toxicology data. The larger goal of these efforts is to,
in effect, overcome inherent and limiting data constraints in focused domains of toxicological
study (e.g., cancer, developmental toxicity, neurotoxicity, etc) by expanding the searchable and
mine-able data network across both chemical and biological domains.
1
-------
As an extension of the DSSTox project, NCCT researchers are promoting adoption of
standardized chemical structure data fields for public toxicogenomics datasets to enable broader
searchability across these data domains, and to enable integration of these datasets with legacy
toxicity data and other public data. In particular, collaboration of the DSSTox project with the
NIEHS Chemical Effects in Biological Systems (CEBS) project is working towards
incorporation of DSSTox data fields and providing structure-searching capability and linkages
across CEBS data and public genomics data, as well as DSSTox and National Toxicology
Program legacy toxicity databases. Chemical structure and genomic expression patterns provide
common metrics for exploring diverse toxicological effects, and can provide the basis for
development of predictive patterns or signatures of a toxicological effect. Similarly, biological
activity profiles consisting of experimentally determined, or computationally predicted
interaction spectra (receptors, proteins, enzymes) could be viewed as expanded "properties" of
the chemical and could augment structure-based information for enhancing toxicity classification
and prediction algorithms.
Finally, NCCT researchers are taking a lead in efforts to address more fundamental and essential
needs to migrate older paper legacy data (such as within EPA Program Offices such as OPP and
OPPT) into electronic form suitable for incorporation into standardized, searchable relational
databases. New commercial technologies from IBM, SciTegic and others that allow for more
automated structure-annotation, and chemical indexing and retrieval procedures are being
evaluated to facilitate efficient electronic conversion and structured content-annotation of legacy
EPA data. In addition, related issues of quality control of chemical information are being
addressed, and Agency-wide chemical structure-browser capabilities are being explored.
Progress to date:
DSSTox Project Status
The EPA DSSTox website (http://www.epa.gov/nheerl/dsstox/). launched in March 2004,
provides detailed information on DSSTox standard chemical fields, guidance for creating new
DSSTox databases, and links to a wide range of public information resources. A major emphasis
of the DSSTox project is on creating field-delimited, content-enhanced data files for diverse
toxicity endpoints. Five DSSTox databases are currently published on the website and several
others are in progress or currently undergoing review. Toxicity endpoints considered include:
rodent carcinogenicity, mutagenicity, estrogen receptor binding affinity, fish acute toxicity, and
pharmaceutical maximum adverse effect dose levels. Additional toxicity endpoints slated for
DSSTox database publication include: skin sensitization, acute toxicity, nasal and eye irritation,
androgen receptor binding, rodent developmental toxicity, DNA intercalation, pesticide
ecotoxicity, and immunotoxicity. A large emphasis has been placed on the quality review of
chemical information, which has led to the creation of a central DSSTox Master chemical
structure reference data file and detailed quality data review procedures.
CEBS DSSTox Project Status
The DSSTox project is collaborating with the NIEHS National Toxicogenomics Center CEBS
Knowledge-Base project firstly by the incorporation of DSSTox standard chemical fields into the
data dictionary and CEBS data entry system. DSSTox Standard Chemical Fields (SCFs) have
recently been revised to better handle the diverse chemical content of public toxicity databases,
which include all variety of mixtures and less well-defined substances, and to better coordinate
2
-------
with other public data standards efforts, such as the ToxML public toxicity data schema project.
These DSSTox SCFs will be additionally employed to index the largest public genomics
databases, to provide expanded structure-searchability within and outside CEBS data content. A
chemical inventory of these databases has begun and will be followed by attempts to encourage
external public data standards organizations (e.g., MGED) and database sources (ArrayExpress,
NCBI) to adopt more rigorous chemical structure annotations of genomics data. Coordination
with other large public database efforts, such as the National Library of Medicine's PubChem,
NIH Molecular Libraries Initiative, and NIH National Cancer Institute molecular database
projects, will also directly impact on the CEBS DSSTox project collaboration.
EPA Communities of Practice - Chemoinformatics Workgroup Status
As part of the effort to improve our ability to index, search and link chemical information data
files across EPA Labs, Centers, and Program Offices, NCCT Researchers have formed a
"Communities of Practice" Chemo-informatics Workgroup to begin to forge Agency-wide
collaborations and coordination with respect to improving treatment and utility of chemical
structure-related information within EPA data files. Additionally, the National Computer
Center's Scientific Visualization group is evaluating possible solutions for providing Agency
wide structure browsing capability.
Impact: NCCT researchers are involved in efforts that are poised to dramatically improve
capabilities to access, mine, and integrate useful chemical-biological activity information from
existing and new data, both within and outside EPA. These efforts have the potential to impact a
wide variety of EPA program offices that heavily rely on chemical information resources, have
large internal stores of data, and have a need for structure-based data exploration, analog
searching, and improved toxicity prediction models. These include many programs within
OPPTS [e.g., Green Chemistry, Premanufacture-Notification Program (PMN), Office of
Pesticide Programs (OPP), High-Production Volume (HPV) Testing Program] as well as EPA's
Integrated Risk Information System (IRIS) Program, Office of Water, and Office of
Environmental Information. New information technologies that incorporate more flexible and
diverse means for assessing of biological and chemical similarity will also improve the
identification of toxicologically relevant analogs by enhancing the ability to explore data and
quantify associations in diverse chemical and biological domains.
Partnerships/Collaborations:
The DSSTox public toxicity database effort is being coordinated and linked with a number of
other public efforts involving data-mining companies (e.g., LeadScope), non-profits
(International Life Sciences Institute - ILSI) and LHASA, UK (VITIC SAR Toxicity Database
Project), and government research laboratories (NIEHS, FDA) that are promoting controlled
toxicity vocabularies, adopting data standards, and migrating diverse toxicity data into the public
domain. The CEBS-DSSTox collaboration more broadly includes collaborations with FDA's
National Center for Toxicological Research, for adoption and use of a chemical structure
browser, and the NIEHS National Toxicology Program (NTP), for incorporation and structure-
linkage to legacy toxicity data in the NTP data files. Coordination of chemical data information
standards that will greatly facilitate cross-platform structure-searching is also ongoing with the
National Institutes of Standards and Technology and International Union of Pure and Applied
3
-------
Chemists (NIST-IUPAC), as well as the National Cancer Institute's Structure-Browser, National
Library of Medicine (ChemID) and PubChem. Additionally, dialog has begun with the NIEHS
NTP and the NIH Molecular Libraries Initiative to coordinate efforts to use chemoinformatic
tools and DSSTox project capabilities to propose sets of chemicals for high-throughput screening
testing based on chemical structure-activity considerations.
Through current efforts and the enlarged Communities of Practice workgroup, partnerships and
collaborations with scientists across EPA are being forged to better improve chemoinformatic
capabilities across the Agency. Collaborations are on-going with MED-Duluth and the
ASTER/AQUIRE system, and the IRIS project, and are being continued or initiated with
scientists in OPP, OPPT, and OEI to inventory, process, review, and integrate data across various
Agency Programs from a unified chemical structure perspective.
Milestones/Products:
FY05
• Incorporate DSSTox standard chemical fields into CEBS.
• Create and publish DSSTox database and documentation files for NTP Immunotox database;
publish a chemical index file for the EPA IRIS programs.
• Establish Communities of Practice - Chemoinformatics Workgroup to begin to coordinate
efforts across the Agency to inventory, retrieve, and explore chemical information data
records.
• Propose plan to link DSSTox effort with the NLM PubChem project.
FY06
• Create DSSTox standard chemical field index files for public genomics databases and NTP
legacy toxicity, to provide linkages within the CEBS relational search environment.
• Create and publish DSSTox database and documentation files for additional published
toxicity databases (skin sensitization, acute tox, nasal irritation, etc) and publish additional
chemical index files for EPA programs (e.g., HPV, SRS-TSCA).
• Propose plan for coordinated effort to structure-index and quality review chemical
information in EPA data files currently on the web and propose plan to provide for an EPA-
wide structure browser on the inter- and intranet EPA website.
• Assist with formation of proposed toxicity chemicals subset for high-throughput testing in
the NTP/NIH Molecular Libraries Initiative collaboration.
FY07
• Assist with incorporation of NCTR Array Track chemical structure browser technology into
CEBS, and full structure-searching integration with DSSTox data files and all CEBS-linked
genomics data.
• Establish procedures and protocols for automating the chemical annotation of new data
submitted to CEBS from DSSTox Master chemical list.
• Continue expansion of the DSSTox public toxicity database inventory to include varied
databases from the published literature and public sources for a wide range of toxicological
endpoints.
4
-------
• Implement plan to structure-index EPA data files and provide EPA website structure-
searchability through curated EPA chemical data files.
• Create structure-annotated database of test results for toxicity subset chemicals from the
NTP/NIH Molecular Libraries Initiative.
FY08
• Advise and assist with the implementation of chemical definitions (e.g., assigning a chemical
to a class) and structure analog searching capability within CEBS, integrated with
toxicogenomics data and bioinformatics capabilities, to serve as the foundation for
developing chemoinformatics capabilities within CEBS.
• Advise and assist with the development of procedures and capabilities for deriving chemical
signatures for predicting toxicity outcomes from the complete profile of CEBS data, with the
integration of chemical structure information and chemical analogy and structure-activity
relationship concepts.
• Begin to tabulate and explore data from NTP/NIH Molecular Libraries screening
collaboration in relation to other DSSTox databases.
• Implement procedures for expanding the structure-annotation of EPA chemical data records
and providing multiple methods for flexible structure or analog searching on the EPA
website.
QA:
QA Plan is being evaluated
5
-------
IIC-1
Title: Simulating Metabolism of Xenobiotic Chemicals as a Predictor of Toxicity.
Lead: - Jack Jones, NERL
Research Issue and Relevance: The mission of the EPA is to protect human health and the
environment from adverse effects caused by exposure to pollutants in the water, soil, air, and
food. Of the approximately 80,000 chemicals used in the US commerce, relatively few have
undergone extensive testing to allow a thorough evaluation of risk requiring an extensive array of
data such as physicochemical properties, persistence, bioconcentration factor, exposure, uptake,
distribution, metabolism, and toxicity. One of the major uncertainties in evaluating risk is
determining exposure of the chemical stressor to the target organism. A compounding factor is
that biotic and abiotic transformations of the chemical inside the target organism following an
exposure may lead to the formation of reactive metabolites that are toxic. For the vast array of
organic pollutants, the availability of transformation rate data and metabolite identification is
sparse. Experiments to identify metabolic pathways can be analytically demanding and costly,
and are often incomplete with respect to potentially reactive intermediates. Thus, elucidating the
metabolism of a chemical and formation of reactive metabolites is a major challenge in
determining pollutant exposure and toxicity for risk assessments.
Approach: Using computational approaches to screen and prioritize chemicals for risk
assessments and to minimize animal toxicity testing has been a goal of EPA and other regulatory
agencies for years. In silico simulation of biotransformations and descriptions of metabolic maps
have great potential for assessing chemical impacts on human and ecosystem health. Predicted
metabolism may also be useful for guiding strategic studies to identify 'bioactive' intermediates
for toxicity assessment and for pollution prevention by avoiding commercial use of chemicals
forming toxic metabolites. The proposed research will develop a capability for forecasting the
metabolism of xenobiotic chemicals of EPA interest, to predict what chemical metabolites are
the most likely to be formed, and to interface that information with toxic effect models that
predict chemical binding to the estrogen receptor (ER), a well-recognized pathway of toxicity
leading to endocrine disruption. This project will (a) illustrate the importance of considering
metabolic activation in toxic effects modeling to predict not only parent chemical toxic potential
but to identify chemical metabolites of equal or greater potential toxicity than the parent
chemical, and (b) demonstrate an approach to provide this capability for large chemical lists of
risk assessment concern.
Progress to Date: ERD-Athens has initiated in vitro metabolism studies using rat microsomes
to track chemical transformation and metabolite formation for a select group of chemicals (e.g.
conazole fungicides) of importance to OPP. Analytical methods for metabolite identification,
utilizing GC/MS, LC/MS, and NMR spectroscopy, are currently being developed and will be
applied where appropriate to metabolism studies in the current proposal. An advanced analytical
capability at ERD-Athens, a 600 MHz NMR coupled to a LC/MS/MS for metabolite
identification and metabolism profiling, will be available in 2005. The cost of testing chemicals
as reproductive toxicants precludes the possibility of evaluating large chemical inventories
without a robust strategic approach for setting priorities. A systematic approach has been
1
-------
developed at MED-Duluth for efficiently expanding the small chemical training sets used for
QSAR development to adequately cover the chemical lists of EPA concern. Chemical selection
algorithms are being developed and strategic testing is ongoing at MED to develop the ER
binding QSAR for chemical prioritization.
Impact: The development of a metabolic simulator for prediction of chemical metabolic maps
integrated with a quantitative structure-activity relationship (QSAR) toxic effects model will
provide effects-based prioritization of chemicals (parent compound and metabolites) relevant to
EPA Program Offices. One of the primary objectives of the CompTox Program and this
proposal is the development of computational tools for prioritization of chemicals for toxicity
testing with the goal of minimizing dependence on test animals. A primary function of OPP and
OPPT is to provide ecological and human health risk assessment for chemicals to anticipate and
limit adverse outcomes. The significant challenge that EPA Program Offices (OPP and OPPT)
face for performing this prioritization process is due to the large chemical inventories (coupled
with minimal data requirements) for chemical registration and screening. Clearly, computational
tools that provide the necessary information to achieve these Program Office goals are needed.
The following outputs for the proposed research will be of benefit in this regard. (1) Provide
OPPT and OPP the ability to prioritize chemical lists (based upon predicted toxic effects of
parent chemical and metabolites) with reliability estimates for use in chemical evaluations and to
rank chemicals for in vitro or in vivo screening and toxicity testing. (2) Provide capability to
OPP and OPPT for predicting bioactive metabolites. (3) Develop searchable metabolism
database for OPP and OPPT use for identification of relevant chemical and/or substructures of
interest for risk assessment And (4) Provide linkage of effects based toxicity model with
metabolic simulator.
Partnerships/Collaborations: The research will proceed iteratively from modeling to testing to
refining models, so frequent communication among ORD scientists and outside partners will be
essential. ORD investigators will meet at least monthly by conference call and will communicate
much more routinely by email. The lead Pis from the Mid-Continent Ecology Division (MED) in
Duluth, MN, and at the Ecosystems Research Division (ERD) in Athens, GA will oversee daily
research activities at their respective Divisions. Analytical techniques for metabolite
identification at MED and ERD will be initially developed by ERD and shared across groups.
Enhanced development of the metabolism simulator will be the primary function of the
Cooperative Agreement awardee (scientists at Bourgas University) with input and oversight by
the lead Pis from ERD and MED. Dr. Jack Jones will serve as Project Officer for the awarded
Cooperative Agreement and will routinely interact with Bourgas scientists via e-mail and
conference call. Annual on-site meetings will occur between all investigators. Linkage of
metabolic simulator outputs with a QSAR-based toxic effects modeling will be done at MED.
The breadth of expertise of this scientific team will enhance the planning and execution of this
project. There will be frequent interactions with EPA Program Office staff by phone and email to
verify chemical lists, to discuss needs and concerns of all involved, and to communicate
progress. Program Offices managers and NERL and NHEERL senior management will be
regularly informed of research planning and progress.
2
-------
Milestones/Products:
FY06
For priority chemicals, incorporate available chemical metadata, metabolism data, and
metabolic maps into a searchable database for data management and structure/sub structure
searchable access
Forecast metabolic pathways for selected priority chemicals using existing simulator for
liver metabolism
Initiate in vitro Phase I liver microsome experiments; incorporate new laboratory and
literature metabolism data into simulator training sets
FY07
Confirm formation of predicted metabolites for priority chemicals and compare observed
maps to forecasted maps
Develop and approach to evaluate and enhance simulator performance through
improvement of transformation probability estimates and expansion of transformation reaction
domain
Begin evaluation of refined simulator and model predictions in context of Program Office
prioritization needs
QA:
QA Plan is being evaluated
3
-------
IIC-2
Title: Modeling Molecular Targets for Toxicity, a Computational Approach to Understanding
Key Steps in the Mechanisms for Toxicity and a Tool for Prioritizing Bioassay Requirements
Lead: James Rabinowitz, NCCT (FTE 2.5)
Research Issue and Relevance: The Agency frequently encounters situations where it must
make decisions about the potential health and environmental effects of chemicals when all of the
relevant data is not available. One rational approach to this problem is to estimate the relevant
missing information by extrapolating from existing data. Knowledge of key steps in the
potential mechanisms of action provides a template for developing models for this extrapolation.
These models may be used to inform experimental studies and provide a tool for prioritizing data
requirements. One relevant example is the Agency need to make decisions on the potential of
specific anthropogenic chemicals to cause endocrine disruption. Developing molecular models
that can be used for rapid screening of the interaction of environmental chemicals with receptors
in the endocrine system will provide an important tool for selecting priorities.
Approach: For many mechanisms of toxicity the key differential process is the interaction
between the ultimate toxicant and a macromolecular target (receptor-ligand, enzyme-substrate,
DNA-genotoxicant, etc.). Modeling this process on a molecular level provides an approach for
prioritizing chemical information needs. This interaction initiates a cascade of events leading to
the ultimate (adverse) outcome. Computational molecular models of the interaction of a
molecule (potential toxicant or its metabolites) with the relevant target, provides insight into the
capacity of a chemical to initiate the relevant cascade.
A number of recent scientific and technical advances facilitate this approach. First, many
of the relevant targets have been identified experimentally. The molecular structure of some of
these targets has been determined and additional information of this type is likely to become
available in the future. Second, molecular modeling software for the simulation and analysis of
interactions of this type has become more sophisticated in a number of relevant ways and make it
possible to more realistically simulate the processes of toxicity. These advances have resulted
from both basic computational investigations of the structure and dynamics of macromolecules
in biological systems and the requirements of the pharmaceutical industry for the discovery of
new therapeutic agents. Third, computational hardware and visualization techniques have
steadily improved. Increased processing speed and memory have made it possible to include
large segments of macromolecules in classical simulations and even in quantum mechanical
calculations. We are applying molecular modeling methods, fueled by these current scientific
and technological advances, to investigating chemicals for their capacity to cause toxicity
through specific modes of action and using a target-toxicant paradigm.
Initially, this approach is being applied to the study of environmental endocrine
disruption. Crystal structures exist in the literature for many receptors in the endocrine system.
By removal of the ligand computationally these crystal structures are used to create virtual
biomolecular targets for endocrine disruption. The best fit to the target for each of a series of
potential ligands can then be determined by computational methods. The properties of this fit
may then be used as part of a scheme to predict the potential of an environmental agent to cause
endocrine disruption.
1
-------
The ultimate goal of this research is to develop a library of biomolecular targets for
chemical toxicity and methods appropriate for the prediction of the ability of a chemical to
interact with these targets. These targets may then by used as part of a chemical prescreen
Partnerships/Collaborations: Scientists from RTD/NHEERL are providing a data base of
chemicals that interact with receptors in the endocrine system and also expertise on the endocrine
system. Scientists from NTD/NHEERL will provide similar expertise and data and important
environmental targets in the nervous system. Commercial molecular modeling software and
advice are available from Schrodinger Inc. These products have been developed primarily for
the pharmaceutical industry and their use must be adjusted for the needs of this project. Dr. Wei-
tao Yang of the Chemistry Department of Duke University is a collaborator for advanced
molecular modeling methods.
Progress to Date: There is increasing concern about the potential of environmental chemicals to
produce adverse health effects through interaction with the endocrine system. One general
mechanism for disruption of the endocrine system involves competition for steroid hormone
binding sites by xenobiotic chemicals that may fully or partially mimic natural hormones.
Crystal structures of the estrogen and androgen receptors have been used to create
macromolecular targets. Multiple crystal structures available for the estrogen receptor have been
used to demonstrate the importance of including receptor flexibility in determining the best fit
and therefore the chemicals likely have the most effect. Studies on including flexibility are in
progress. In other preliminary studies, methods existing methods are being employed. Using a
data base containing estrogens, androgens and chemicals that bind to other nuclear, it was found
that the results for a single receptor, considered individually, did not correctly order the ligands
according to the ability to bind to that receptor. However, when the data for a series of receptors
is considered simultaneously and the results analyzed to determine which receptor a chemical
interacts with most favorably, all chemicals are classified correctly.
An additional Agency concern, which may be approached with the target-toxicant
paradigm and molecular modeling, is the cumulative risk of some specific pesticides on the
enzyme acetylcholinesterase (AChE). There is some data that indicates that AChE has two
binding sites, the catalytic site and an allosteric site that affects the specificity and efficacy of the
enzyme. In experiments involving just a single toxicant it is not possible to identify the site of
interaction but we have shown that for mixtures of AChE active chemicals the cumulative
toxicity depends on the relative interaction of each chemical with each site. A computational
scheme is being developed for the cumulative toxicity of mixtures that takes advantage of the
two site model.
Relevance: This research addresses the Agency need for predictive models for hazard
identification, both the sub areas of QSAR and other computational approaches and High
Throughput Screening.
Milestones:
FY06 - Results of the use of the target-toxicant paradigm to screen for
estrogenicity/androgenicity; Results of the affect of the two binding site model on the cumulative
risk of chemicals acting through the enzyme AChE.
2
-------
FY07 - Results of the importance of protein flexibility in evaluating the interaction of chemicals
with macromolecular receptors. Report on the importance of protein flexibility in evaluating the
interaction of chemicals with macromolecular receptors.
FY08 - Evaluation of the use of the target-toxicant method as a tool in a diverse chemical
screen. Application of the target-toxicant approach to other targets of toxicity
QA:
QA Plan is being evaluated
3
-------
IIC-3
Title: ToxCast, a Tool for Categorization and Prioritization of Chemical Hazard Based on
Multi-Dimensional Information Domains.
Lead: Keith Houck, NCCT and David Dix, NCCT (FTE 2.7)
Research Issue and Relevance: Across several EPA Program Offices (e.g., OPPTS, OW,
OAR), there is a clear need to develop strategies and methods to screen large numbers of
chemicals for potential toxicity, and to use the resulting information to prioritize the use of
testing resources towards those entities and endpoints that present the greatest likelihood of risk
to human health and the environment. This need could be addressed using the experience of the
pharmaceutical industry in the use of advanced modern molecular biology and computational
chemistry tools for the development of new drugs, with appropriate adjustment to the needs and
desires of environmental toxicology. A conceptual approach named ToxCast has been
developed to address the needs of EPA Program Offices in the area of prioritization and
screening.
Approach: Modern computational chemistry and molecular biology tools bring enabling
technologies forward that can provide information about the physical and biological properties of
large numbers of chemicals. The essence of the proposal is to conduct a demonstration project
based upon a rich toxicological database (e.g., registered pesticides, or the chemicals tested in the
NTP bioassay program), select a fairly large number (50-100 or more chemicals) representative
of a number of differing structural classes and phenotypic outcomes (e.g., carcinogens,
reproductive toxicants, neurotoxicants), and evaluate them across a broad spectrum of
information domains that modern technology has provided (i.e., physical-chemical properties,
predicted biological activities based on existing structure-activity models, biochemical properties
based on high throughput screening assays, cell based organotypic assays, and genomic analysis
of cells or organisms). These domains represent increasing biological relevance, as well as
increasing resource requirements. The ultimate goal of the project would be to mine the
resulting data for association between and among the various domains and the known
toxicological properties of the base set of chemicals in order to provide a structured strategy to
identify potential toxicity pathways, and to prioritize chemicals them for subsequent testing
based on that information.
The underlying hypothesis is that whether is concerned with the off target effects of
drugs, as desired to be understood by the pharmaceutical industry, or toxicity in case of
environmental agents of interest to the EPA, the response is driven by interactions with
biomolecular targets of one form or another. One needs only to identify those receptors of
concern and identify tools for assessing the likelihood of interaction with the chemicals of
concern. In moving from the drug development arena (which can be compared to working along
one or just a few vectors) to the environmental toxicology arena (which can be likened to
working on a matrix instead of a vector), one needs to shift from a specific screening target to a
more global agenda, and it becomes necessary to vastly expand the number of potential
biomolecular targets, be these obtained from in silico assays, biochemical assays, cell based in
vitro assays, surrogate animal models, or short term studies in traditional species. Hence, a wider
net of endpoints and information sources will be applied, at least initially, as the concept
transgresses from a concept to a reality.
1
-------
Of course, a number of hurdles would need to be addressed before launching such an
effort, including: (1) identification of a subset of chemicals for serving as the proof of concept
models; (2) developing a chemical inventory management and distribution system; (3)
identifying an upper cap on the per chemical cost of obtaining screening level data: (4) selecting
assays within the available resources; (5) flexibility, or tiering, of domains based upon pre-
existing knowledge; (6) perhaps initially targeting only a few manifestations of toxicity rather
the all possible ones to decrease the complexity of the task; (7) evaluating the impact of
metabolizing capability, or lack thereof, on the efficiency of the screening assays; (8) developing
a bioinformatic approach to mining the resulting data and identifying signatures of concern; and
(9) carrying out a prospective assessment of the bioinformatic approach using chemicals
currently entering a traditional testing process. These hurdles would be the subject of
considerable discussion as the potential feasibility of this concept proposal is discussed further.
Progress to Date: At this point in time, ToxCast is nearly the end of its conceptual design
phase. The information domains have been identified, and a number of potential contributing
data sources have been investigated (e.g., Iconix, MDS Pharma, CEREP, Amphora, PASS).
Recruitment actions are underway to add two staff members to the NCCT who will be
responsible for the biological and information processing components of ToxCast.
Communications have been established with the NTP/NIEHS which has similar interests and
which is beginning to work with the NIH Molecular Libraries Initiative (see also the DSSTox
implementation plan). Outreach to the OPPTS, ACC, EDF and other external groups have also
begun to help develop a consensus on the specific directions and contents of ToxCast.
Impact: The availability of a biologically and chemically based system to begin to associate
chemicals of like properties and activities will provide a number of EPA Program Offices with
an extremely useful tool that heretofore has been seriously lacking. The tool may be one of the
first broad scale products of the NCCT that addresses the mission of improving the efficiency
and effectiveness of hazard identification and risk assessment methodologies employed by the
EPA.
Partnerships/Collaborations: The NCCT is working to establish partnerships with a number of
external groups that can facilitate development of the information needed in ToxCast. These
groups include the OPPTS, the NTP/NIEHS, the ACC, the EDF and a number of commercial
vendors that market some of the enabling technologies.
Milestones/Products:
FY06 - Develop conceptual framework for ToxCast
FY07 - Establish initial battery of assays across the information domains, identify list of
chemicals to evaluate proof of concept for framework and begin data acquisition
FY08 - Report on the utility of statistical clustering techniques on assay results from pilot
chemicals to group them according to known toxicity patterns; revise framework as dictated by
results.
QA:
QA Plan is being evaluated
2
-------
IIC-4
Title: Development of microbial metagenomic markers for environmental monitoring and risk
assessment.
Lead: Jorge W. Santo Domingo, NERL
Research Issue and Relevance: The microbiological water quality standards established by
EPA depend on culturing fecal indicator bacteria to predict the risks associated with water usage.
For decades this has been the favored approach to microbiological monitoring in spite of the fact
that culture-based methods tend to underestimate the densities and diversity of microorganisms
in environmental samples (Amann et al., 1995). Relevant to public health is the fact that
nonculturable pathogens could be the etiological agents of many waterborne illnesses (Sails et
al., 2002). In addition, nonculturable organisms may be better indicators of fecal contamination.
Moreover, nonculturable bacterial indicators may be useful in the identification of the fecal
sources impacting a particular water system. Identifying the sources responsible for the fecal
pollution of a natural waterway is important in order to reduce the fecal bacterial levels and in
turn reduce the illnesses associated with recreational activities, and food (e.g., fish) and water
consumption (Simpson et al., 2002). Consequently, strict dependence on culturing techniques
continues to be a notable roadblock in the path towards understanding the correlation between
bacterial indicators of fecal contamination and the hazards associated with fecally impacted
waters. Nucleic acid-based approaches can circumvent many of the shortcomings of the culture-
based methods. For instance, the possibility of rapidly and simultaneously monitoring hundreds
of microorganisms and genes relevant to public and environmental health is now becoming a
reality in light of the recent advances in microarray technology (Stahl and Tiedje, 2002). Such a
capability will categorically improve microbial risk assessment and the framework used in the
development of environmental monitoring and risk management tools.
Approach: We propose the development of an innovative metagenomics program to enhance
our current capabilities for environmental microbial monitoring, risk exposure, risk management,
and risk assessment. The general goal of this study is to develop a novel molecular markers
based on fecal microbial genomes to better assess the sources of fecal contamination in natural
water systems. Assuming that the evolutionary pressure for the selection of host specific
populations must involve ecologically driven processes we hypothesize that genes playing a role
in host-microbial interactions and cell-cell recognition (e.g., cell surface proteins, toxins, and
adhesin) are better markers of host specificity. The primary objective of this study is to discover
novel and validate the use of fecal metagenomic sequences asMST host specific markers. By
looking at collective genomes we will substantially increase the number of potential genes and
microbial species that can be used in the development of molecular microbiological assays
(Schloss and Handelsman, 2003). The first phase of this program will focus on constructing fecal
metagenomic libraries specific to animals that are known to be relevant pollution sources of
watersheds in the United States. The following phases will focus on systematically identifying
DNA sequence markers for specific human and bovine fecal sources and testing the specificity of
these markers against waters impacted with different pollution levels and pollution sources. We
will use genomic subtraction of metagenomic libraries to generate unique sequences from the
collective microbial genomes present in animal feces. Due to the complexity of microbial
1
-------
communities, this approach will facilitate the identification of candidate host specific microbial
genes. Real-time PCR assays will be developed to test the host specificity of the latter markers
and to quantify the target genes in environmental waters impacted with different levels of fecal
contamination. A similar approach will be undertaken with samples from ongoing
epidemiological studies. Markers that show host specificity will be included on biochips
containing sequences specific to fecal indicator bacteria and pathogens and challenged with
DNA extracts of different fecal samples and water samples impacted with different levels of
pollution. By combining the epidemiological studies with the host specific assays and
pathogenic markers, we hope to gain better predictive capability to determine outcomes from
scenarios associated with fecally impacted waters.
Progress to Date: See below
Impact: Distinguishing between human and nonhuman fecal contamination in our waterways
has never been more important than in recent years in light of the need by the States to comply
with deadlines associated with the Total Maximum Daily Loadings program (Simpson et al.,
2002). Accurate assessment of the primary sources of fecal pollution is needed in order to
calculate the different risks associated with each of the potential sources impacting water
systems and to correctly implement and evaluate Best Management Practices (BMP) used to
remediate fecally polluted waters. As currently developed, most MST methods depend on
markers that have no ecological meaning. The main goal of this study is to develop assays that
can be used in source identification, environmental monitoring, and risk assessment. In order to
accomplish this goal we will combine a metagenomic-scale approach with high-throughput gene
mining (microarrays) and real-time PCR assays. Therefore, thiis research supports EPA's GPRA
Goal 2 (Clean and Safe Water). The proposed work will provide a rich source of metagenomic
information (i.e., novel molecular markers) for sequenced-based analyses aimed at better
monitoring of fecal pollution and improving the assessment of the outcome associated with
different pollution sources. Consequently, it is anticipated that the results will provide a
molecular-based framework to (1) better predict the relationship between microbial water quality
and public health risks, (2) determine the impact of different microbial pollution sources on
watershed biology, and (3) effectively control or eliminate pollution in our Nation's watersheds.
Partnerships/Collaborations: As this program is relevant to water fecal contamination and
public health issues related to water, we have requested the participation of personnel from
EPA's Office of Water and the National Center for Environmental Assessment as consultants.
The primary points of contact of the latter organizations will be Dr. Robin Oshiro and Dr. Mary
Rothermich. Drs. Oshiro and Rothermich will act as liaisons for their respective organizations.
Their role will be to provide advice on how the data can be utilized by their offices. They will be
invited to participate in mid year review meetings and conference calls. We have also briefed
other personnel from the Office of Water, and they have expressed interested in receiving
updates on the progress of this project. As this project has an immediate impact on problems that
states and tribes are currently facing, we will keep Drs. Ron Landy and Bobbye Smith (Region 3
and 9 ORD-regional science liaisons, respectively) informed on the current progress of this
project. In addition, we will consult with members of EPA's Genomics Task Force Workgroup
on issues pertaining Data Management, Data Analysis, and Data Submission and Methods
Performance. Drs. Rebecca Calderon and Al Dufour will also be consulted on a regular basis as
2
-------
they are managing the epidemiological study that is currently been conducted by members of
NERL and NHEERL.
Milestones/Products:
FY06 - Evaluation of metagenomic databases as a source of molecular markers to assess human
and animal fecal contamination in surface waters
FY07 - Report on the evaluation of PCR-based host-specific assays to confirm the presence of
animal fecal sources of pollution in waters impacted by fecal contamination. Report on fecal
indicator microorganisms and/or genetic markers from fecal material whose densities in
recreational waters best correlate with the rates of illnesses in users of recreational waters.
QA:
QA Plan is being evaluated
3
-------
IIID-1
Title: Statistical Methodology for Estimating Parameters in PBPK/PD Models
Lead: R. Woodrow Setzer, NCCT (FTE 0.8)
Research Issue and Relevance: PBPK/PD models are large dynamic models that predict tissue
concentration and biological effects of a toxicant before PBPK/PD models can be used in risk
assessments in the arena of toxicological hypothesis testing, models allow the consequences of
alternative mechanistic hypotheses to be predicted in highly non-linear systems. Whether for
quantifying the uncertainty of the extrapolation of a dose metric or determining which hypothesis
is better supported by the evidence, and by how much, it is critical that the uncertainties inherent
in such modeling be properly quantified, and established principles of statistical inference be
brought to bear.
However, such quantification is far from simple. Biological systems models generally
involve large numbers of parameters whose values are known with varying degrees of
uncertainty. Some parameters (for example, tissue-specific fractions of cardiac output and
fractions of body mass) can reasonably be thought of as varying among individuals of a
population. Other parameter values have been estimated in in vitro systems, or may have been
calculated from properties of the chemicals involved (for example, partition coefficients in a
physiologically-based pharmacokinetic model). It is typically the case that the values of
parameters in a biological systems model will be derived from the results of several different
experiments, of very heterogeneous designs, and often conducted in different laboratories at
different times. Over the last decade or so a small number of statistical investigators have begun
to explore the issues involved in applying statistical methods to such systems. While there are
examples of the application of statistical methodology to PBPK/PD models in the literature, there
is as yet no coherent approach to this activity that takes into account the special issues that arise
in PBPK/PD modeling. The NCCT is establishing a program to further develop the statistical
methodology to support the rigorous quantification of both the uncertainty of parameter
estimates and of model predictions, with the aim of providing a framework for statistical
applications in PBPK/PD modeling.
Approach: The outstanding issues are a mix of computational and theoretical statistical
problems, and are best investigated through exploring specific models and datasets. Thus, we
need to identify PBPK/PD models for chemicals of interest to the Agency, identify the
outstanding theoretical statistical issues that need to be solved, and set up a computing platform
to handle the large computing loads this exploration will entail. Finally, theoretical statistical
exploration needs to precede hand-in-hand with practical PBPK/PD model development,
ultimately leading to the publication of a practical framework for applying statistical
methodology to PBPK/PD models.
Progress to Date: Investigators in NHEERL, NERL, and CUT Centers for Health Research
have been approached for appropriate models and data sets, and several models have been
identified for further work. A graduate student in biostatistics and his advisor have been
recruited to work on outstanding theoretical issues and apply solutions to models and data sets of
1
-------
interest to the Agency. The literature surrounding this topic has been surveyed, and a literature
review is in progress.
Impact: PBPK/PD models are of increasing interest in the Agency for use in extrapolation from
animals to humans, across routes of exposure, and for integrating acute effects over discrete
episodic exposures. Better characterization of methods for estimating parameters and
quantifying uncertainties in such models predictions will remove one major impediment to their
more general application. This will allow replacement of default uncertainty factors with
transparent mechanism-based statements of scale and uncertainty, in turn decreasing the
subjectivity and increasing the transparency of environmental health risk assessments.
Partnerships/Collaborations: Collaborations with NERL, NHEERL, and the CUT Centers for
Health Research and the Department of Biostatistics in the UNC School of Public Health are
being developed to develop statistical methods for parameter estimation in PBPK/PD models.
Milestones/Products:
FY06 - Submission of a journal article outlining outstanding statistical issues in the analysis of
PBPK models.
FY07 - Joint NIEHS/NCCT workshop on Statistical Issues in PBPK/PD modeling.
FY08 - Publication of framework for the statistical analysis of PBPK/PD models.
QA:
QA Plan is being evaluated
2
-------
IIID-2
Title: Modeling Toxicokinetics for Cross-Species Extrapolation of Developmental Effects
Lead: Hugh Barton, NCCT (FTE 1.4)
Research Issue and Relevance: Animal toxicology studies used to evaluate the potential for
effects due to exposures during developmental periods are extrapolated to humans based
upon the maternal exposure dose. The approach does not address whether the toxicokinetics
are similar across species during the relevant critical developmental window. In this research
program we will develop toxicokinetic models for estimating internal dosimetry during early
life stages to serve as a basis for improved interspecies extrapolation. Absent information
specifying the critical window, estimates of internal dosimetry during the in utero,
lactational, and early post-weaning periods will characterize the uncertainty in evaluations
based upon maternal exposure dose. With information on the critical window, models may
describe toxicokinetic and toxicodynamic processes in the relevant toxicological species and
humans permitting improved quantification.
Approach: This research will incorporate a range of pharmacokinetic modeling methods
including noncompartmental analyses and classical and physiologically-based
pharmacokinetic modeling to describe dosimetry in early life-stages relevant to one-
generation toxicity studies (i.e. maternal exposure resulting in fetal exposure, lactation
exposure of pups after birth, and exposures of growing pups). Perfluooctanoate (PFOA) and
related compounds will be used as initial prototype chemicals due to the importance of
effects observed in offspring, which are used as endpoints in risk assessments. Lactational
exposure will be modeled first because dosimetry during this period has not yet been
evaluated. A physiologically based pharmacokinetic (PBPK) model for PFOA will be
developed to characterize dosimetry during these different lifestages and identify data gaps
that limit the ability of the model to address both rats and humans. Finally, this early life
stage model will be used to evaluate dosimetry during different early life periods for a
selected group of other prototype compounds for which developmental toxicities are
observed. Prototype compounds would be selected to have a range of pharmacokinetic and
physical chemical properties. This analysis will aid in the evaluation of the uncertainties in
the default risk assessment approach based upon maternal exposure.
Progress to Date: Literature reviews have been ongoing to obtain physiological parameters
for growing rat pups. Compartmental pharmacokinetic modeling of PFOA in adult animals,
including pregnant and lactating females, has been completed.
Impact: Biologically based modeling of toxicokinetics and toxicodynamics during
developmental periods can improve chemical risk assessments based upon developmental
effects. At a minimum, characterization of internal dosimetry will inform the uncertainties
attendant to analyses based upon the maternal exposure dose. In the presence of information
on the critical window, the models may directly form the basis of the quantitative risk
assessment by deriving the relevant internal dose metric for extrapolation to humans.
Perfluorinated compounds, including PFOA, are an important proof of concept for this
1
-------
research because they are a class of compounds producing developmental effects of
significant regulatory concern.
Partnerships/Collaborations: There are active collaborations with NHEERL, NERL, and
NTP to collect data for perfluorinated compounds and NHEERL to collect data on conazole
pesticides. Evaluation of published literature to obtain physiological parameters has been an
ongoing collaboration with NCEA. In addition, this project complements ongoing efforts by
others to develop models for human children. Sharing of information and physiological
parameters has been ongoing through an early life physiological parameters database project
at the International Life Sciences Institute with participating by US and Canadian
government agencies, academic institutions, companies and non-profit organizations.
Milestones/Products:
FY06 - Model rat pup lactational exposure for perfluooctanoate (PFOA) using
compartmental approaches.
FY07 - Develop initial physiologically-based pharmacokinetic model for PFOA in rat
maternal-fetal-pup unit and identify data gaps in relation to rat and human.
FY08 - Evaluate modeled dosimetry for rat fetus and pup for a limited number of prototype
compounds to inform the uncertainty in use of maternal exposure dose in risk assessments.
QA:
QA Plan is being evaluated
2
-------
IIID-3
Title: Development of a portable software language for physiologically-based
pharmacokinetic (PBPK) models.
Lead: R. Woodrow Setzer, NCCT (FTE 0.1)
Research Issue and Relevance: The PBPK modeling community has had a long-standing
problem with modeling software compatibility. The numerous software packages used for
PBPK models are, at best, minimally compatible. This creates problems ranging from model
obsolescence due to software support discontinuation to difficulty in linking models created
in different packages.
The objective of this research effort is to create an XML extension to be used in
conjunction with the systems biology markup language (SBML), an XML-based, machine-
readable format for representing models of biochemical reaction networks, including
metabolic networks, cell-signaling pathways, and regulatory networks. This extension will
augment SBML to accommodate PBPK models with reasonable complexity, including full
documentation, PBPK modeling vocabulary, and examples for instructional and
demonstrational purposes. This extension will be constructed to permit seamless linking of
PBPK models to the biological pathway models abundantly available in SBML format. A
secondary goal is to develop a pseudo-generalized visualization program that is capable of
translating an SBML file for a PBPK model into a visual schematic along with equations
representing the schematic in differential equation form, accompanied by parameter values
and references for the parameters.
End-users of a PBPK model, such as risk assessors, would benefit greatly from being able
to use the software of their choice to run the model and to visualize the model itself and
documentation for it, making incorporation of PBPK models in the risk assessment process
easier and more efficient.
Approach: We will begin by coding simple to moderately complex PBPK models in SBML
and testing the results. This initial step will map out the limitations and weaknesses of
SBML for accommodating PBPK models and give insight into the magnitude of the work
involved in creating an appropriate SBML extension. This initial step will be an iterative
process with complexity of the models increasing until all major necessary attributes of the
extension are uncovered. The next step will be to create an initial version of the schema and
extensively test it. An NCCT-sponsored meeting will be held with all interested parties
where the schema will be presented and a plan for moving forward will be presented and
debated. This plan will include formal and informal collaborations and partnerships. Once a
final version of the extension has been developed and fully tested, a formal SBML
workgroup will be formed and a proposal will be written to the SBML Community at Large
for incorporation of the extension into the current version of the base SBML schema. At this
point, SBML would fully accommodate PBPK models. Simultaneously, we will be
developing the visualization program with the objective of using it not only as an aid in the
development and usage of PBPK models, but also as a powerful educational tool.
1
-------
Progress to Date:
A team has been assembled to develop the schema as well as advise and manage the
project. Some necessary features of the extension have been determined and we are currently
coding a simple PBPK model in SBML as part of the process of determining all necessary
attributes. The interest of the PBPK community has been assessed and a list of interested
parties, some possibly for collaboration, has been created.
Impact:
The ability to create, transfer, use, augment, and review PBPK models without the
limitation of software compatibility is a long-standing desire in the PBPK community. The
ability to seamlessly link PBPK models to biological pathway models is of great interest,
since this gives the modeler a fast and efficient means of extending the estimations of tissue
dose from the PBPK model to cellular-level responses, thereby facilitating a quantitative
investigation of dose-response relationships through mode/mechanism of action. By making
the use of PBPK models faster, easier, and more reliable, this product will make the use of
these models more common and widespread in risk assessment.
Partnerships/Collaborations:
Formal collaboration has been established with the Lockheed Martin Corporation through
the U.S. EPA Office of Environmental Information. The possibility of collaborating with
The Aegis Technologies Group (developers of the prominent PBPK modeling software
package, ACSLXtreme) is currently being explored. Contact with original SBML developers
and project managers have been established with positive feedback. Possible collaboration
with CUT Centers for Health Research is also being explored. Groups that have expressed
interest in being immediate clients for the products of this project include EPA/NCEA and
NIEHS/NTP.
Milestones/Products:
FY06 - Report on findings of assessing extensiveness required for the XML schema.
FY07 - NCCT-sponsored meeting on evaluation of the performance of the completed XML
schema and status of the visualization program.
FY08 - Report on proposal to have finalized XML schema included in the next release of
SBML and evaluation of visualization program.
QA:
QA Plan is being evaluated
2
-------
IIID-4
Title: Systems Modeling of Prostate Regulation and Response to Antiandrogen
Lead: Hugh Barton, NCCT (FTE 0.3)
Research Issue and Relevance: The prostate is an androgen-dependent tissue that is an
important site of disease in human males as well as an important indicator of androgen status
in animals. The rat prostate is used for studying antiandrogenic drugs as well as for
evaluation of endocrine disruption (e.g., Hershberger Assay). Pubertal changes in the
prostate have been observed to be as sensitive to environmental antiandrogens as in utero
effects. The goal of this research is to model the biology of prostate androgen function on a
systems level to determine the factors responsible for the dose-response observable with
androgens and antiandrogens in the male rat. This includes investigation of the roles of
positive and negative feedback loops in prostatic response following castration and dosing
with testosterone and/or antiandrogens.
Approach: A biologically-based, systems-level model will be developed describing the
regulation of the prostate by androgens. The model will extend an existing model for the
male rat central axis, which describes feedback between luteinizing hormone and testosterone
production in the testes, to include the prostate and conversion of testosterone to
dihydrotestosterone (DHT). The prostate model will describe binding of androgens to the
androgen receptor, 5a-reductase catalyzed production of DHT, and gene regulation affecting
cell proliferation, apoptosis, and prostatic fluid production. The model will combine
pharmacokinetic models for endogenous hormones (i.e., testosterone, DHT, LH) and
exogenous antiandrogens (e.g., finasteride, flutamide or casodex, vinclozolin), and a
pharmacodynamic model for androgen-dependent prostate functions. Linkages of this model
with genomics data obtained with castration, testosterone treatments, or antiandrogen
treatment will be explored to assist in developing perspectives on how such data would fit
into quantitative risk assessments.
Progress to Date: A model has been constructed that recapitulates the time-dependent
changes in prostate following castration of male rats resulting in decreased serum
testosterone, prostatic androgen receptor, and prostate weight. The model currently poorly
describes the dose-response for less drastic changes in testosterone; this is under
investigation. A classical compartmental model for finasteride, a therapeutic 5a-reductase
inhibitor, has been linked to the biologically-based model for the rat endogenous hormone
function. Opportunities to link the model with genomics data are being explored.
Impact:
This project is designed to evaluate how perturbations of endogenous biological systems
produce observable dose-response behaviors using a widely used endpoint, rat prostatic
response to androgen status. In addition, insights will be obtained concerning the potential
utility of genomic data for dose-response analysis in relation to more traditional toxicological
endpoints.
1
-------
Partnerships/Collaborations: (e.g., where is the data coming from?). The initial modeling
utilizes data published in the peer-reviewed literature. Collaboration with Dr. Mitch Rosen
(NHEERL/RTD) is under development to obtain prostate genomics data. Collaboration with
Dr. Robert Chapin (Pfizer) also is being developed to obtain genomics data following
testosterone implantation.
Milestones/Products:
FY06 - Report on the prostate function model following castration.
FY07 - Report on dose-response with testosterone and/or antiandrogens.
FY08 - Evaluate linkage of the biologically-based model of prostate androgen-
dependent gene regulation with genomics data.
QA:
QA Plan is being evaluated
2
-------
IIID-5
Title: Systems Biology Model Development and Application
Lead: Rory Conolly, NCCT (FTE 2.0)
Research Issue and Relevance: System biology models holistically describe, in a quantitative
fashion, the relationships between different levels of a biologic system. Relationships between
individual components of a system are delineated. System biology models describe how the
components of the system interact to give rise to the physiologic function of the system. For the
realm of toxicology these models will be developed to not only describe such interactions but to
also describe how exposure to toxicants can perturb these interactions and the normal physiology
of the system. A hallmark of these models is that they are designed to allow study of the
multiple components of the system simultaneously. The resolution of such models depends upon
the problem being studied. They can describe interaction between molecules, between molecules
and tissue, organs, and whole systems. They can even extend to interaction between different
species within ecosystems. Most populations, including humans, are simultaneously exposed to
numerous potential toxicants under a myriad of conditions. Further those populations have a
variety of other processes occurring during and with those exposures. Underlying disease,
nutritional factors, and genetic predisposition are just a few examples of underlying factors that
can greatly influence an organism's or population's response to environmental toxicants. System
biology models offer the opportunity to describe and understand some of these mechanisms so
that risk assessments can eventually be based on the most relevant biological information and not
just on default assumptions for which the uncertainty is not easily identified nor quantified. The
models are also excellent tools to help analyze data and test different hypotheses. These models
will use and depend upon complex data such as is generated from genomics, proteomics, and
metabolomics. Iteration between experimental measurements and computational modeling is
necessary to understand the function of complicated biologic systems.
Approach: This project will progress along four broad levels each informing and helping
develop one another. First, and already on-going, are a number of tasks using existing
physiologic pharmacokinetic and pharmacodynamic models to develop and then test different
hypotheses describing the adverse affect that may result from environmental exposures. This
work is being, at this time, applied to humans. Models describing enzymatic changes, such as
cholinesterase inhibition, are being used to show the relative impact of different exposure
scenarios. Further, these models are also being developed and used to help design the most
useful and cost-effective exposure measurement studies. Collaborative work is being performed
to test the suitability of using in-vitro and computationally derived parameters in models such as
these. This is another important aspect as given the complex nature of future models and
exposure scenarios methods to rapidly estimate key physiologic, thermodynamic, and
biochemical parameters will be necessary, especially for those conditions that preclude practical
laboratory measurements. In addition, a number of tasks are underway or being planned that will
build more complicated pharmacodynamic models for this purpose. (See project description on
pharmacodynamic modeling of the prostate as an example.)
Second, a task is being formulated that will begin to build a conceptual model that would
used various types of data, such as pharmacokinetic data, mode of action data, "omics" data, etc.
1
-------
for a specific case. A mathematical model will then be built and implemented. The model will
be used to show the importance of relevant data. That is, the question will be answered "how
much can the risk assessment be improved and uncertainty reduced as more data demanded and
become available?" Some work in this task will begin to develop the mathematical constructs
necessary to incorporate "omic" data and information into quantitative models.
The third task will start to describe a fairly complicated endogenous physiologic or
biochemical process in detail. Organisms have many biochemical processes that help maintain
the homeostasis of the system. Understanding and describing such processes may be crucial to
eventually describing and predicting the adverse effects resulting from perturbation of those
processes resulting from exposure to exogenous substances and factors. Although this task is
still in formative stages examples include the development of kinetic model of the microsomal
oxygenation system in hepatoctyes, or describing the glutathione system in physiologic detail.
After model formulation, implementation, and testing the model will be expanded for use with
pharmacokinetic models to describe what occurs in the endogenous system after exposure to
environmental toxicants. Again, such a model can be used for hypothesis testing as well as for
predictive risk assessments.
The fourth task will seek to develop a model for a disease process affecting many people.
By modeling a disease process with the subsequent pharmacologic exposures a model will be in
place which can then be used to investigate the potential of exacerbation by environmental
exposures. Again, in the very formative stages one such disease state being considered is type 2
diabetes.
These latter two tasks will develop examples of how endogenous processes, disease
states, and exposure to endogenous environmental factors may interact to cause adverse affects
or exacerbate pre-existing conditions. It is obvious how all four of these tasks are related on how
they build upon each other. Further some over arching themes are important to all of them. For
each, eventual application to probabilistic systems will be an anchor. As such, advanced
statistical techniques will be used (as developed here, in other computational toxicology projects
and elsewhere) to account for variability, sensitivity, and uncertainty. Also, as mentioned, these
models are excellent tools for hypothesis testing. That makes them also good tools to keep a
balance between describing the detailed complexities of a biologic system and the parsimony
that is often practically necessary real world risk assessment process.
Progress to Date: In FY 2005 key staff has come into the center including an ST and two
biomedical engineers (one as a post-doc). Further staffing plans are being made. Collaborations
(see section below) are being established. Work on the first task and work in related Center
modeling projects are well underway. Potential collaborators and staff for the third and fourth
tasks have also been identified.
Impact: As described in previous sections, this work will help take risk assessment to the next
higher level. The latest mechanistic information and the interplay between exposure,
endogenous factors, pre-existing conditions, and genetic predisposition can be rationally
accounted for by using such models. There is a great need for quickly illustrating how the latest
molecular information, especially "omic" information and information coming from high
through-put studies will be used in risk assessments. NCEA and several program offices,
including OPPTS, OW, OAR will all be faced with these types of biologic data and currently
have few examples of tools for using such information in a beneficial way. This work will be
2
-------
accomplished over several years will gradually add more and more tools to their arsenal for risk
assessment.
Partnerships/Collaborations: National Health and Ecological Effects Laboratory, the National
Exposure Research Laboratory, the National Institutes of Environmental Health Sciences,
Department of Energy, University of Michigan, Moscow State University (Russia).
Milestones/Products:
FY06 - Journal Article on use of biologic models to ascertain necessary resolution of
exposure measurements. Collaborative groups formed for task 2, 3, and 4
Formulation of conceptual model and writing the mathematics and code for that
model for task 2, above. Selection and begin model implementation for
endogenous biochemical system for task 3. Start model development of disease
process for task 4.
FY07 - Implementation of model for tasks 2 and 3 and begin application for cases with
exposures to known environmental toxicants - abstracts and presentations.
Disease model coded, exercised, and evaluated for task 4. Determine through
literature search and other means for examples of exogenous exposure that impact
the disease process selected for task 4.
FY08 - Journal articles illustrating some uses of "omics" information in quantitative
models. Summary report for Agency use on earliest best practices on "omics" and
system biology models. Journal article for disease model. Enhancement of disease
model to incorporate exposures to environmental toxicants.
This project will continue beyond this three year period and more products relevant to
Agency clients will result.
QA:
QA Plan is being evaluated
Related Tasks:
Systems Biology Modeling
Computational Modeling of Interleukin 1 (IL-1) Mediated Intracellular Signaling
(Collaboration with Jim McDougal, Wright State University)
IL-1 signaling is an important component of the molecular mechanism by which the
acute inflammatory reaction in skin develops in response to a dose of skin irritant. The short-
term goal of this project is to understand how best to develop computational models of signaling
pathways given limitations of the databases describing network topology, protein concentrations,
reaction rate constants, etc. The longer term goal is to develop a computational model that can
provide useful dose- and time-response predictions of the acute inflammatory response in skin.
3
-------
Computational Modeling of Extracellular Signal-Regulated Kinase (ERK) Signaling in
Cerebellar Granular Cells
(Collaboration with Bill Mundy, NTD, and Qiang Zhang, CUT Centers for Health Research)
Extracellular signal-regulated kinase is important for regulating neuronal survival and
growth. ERK its activation is susceptible to perturbation by environmental chemicals. The
objective of this study is to 1) characterize the dynamics of ERK activation in response to BDNF
and NMDA; 2) use computational modeling to promote understanding and dissecting of the
signaling network underlying ERK activation. This project will improve our understanding of
molecular mechanisms which stressors such as BDNF and NMDA perturb neuronal cells,
thereby enriching the database needed for identification of predictive biomarkers.
Modeling of Mammalian Biomolecular Responses: Computational Core in support
of a Superfund Basic Research Project (SBRP) at Michigan State University
(Collaboration with Norb Kaminski, Michigan State University, and Mel Andersen, CUT
Centers for Health Research)
Starting in the summer of 2006, Dr. Conolly will work closely with a Computational
Core housed at the CUT Centers for Health Research in support of the SBRP at Michigan State
University. The initial effort will be to develop computational descriptions of signaling pathways
involved in B cell maturation and which are perturbed by exposure to low levels of TCDD.
Mathematical Model of Steroidogenesis: Molecular Response to Endocrine Disruptor Exposures
(Collaboration with Michael Breen, NCCT, Gerald Ankley and Dan Villeneuve, Mid-Continent
Ecology Laboratory)
Ankley et al. are using a systems biology approach to characterize the HPG axis in small
fish. The project will increase our understanding of the basic biology that is perturbed by
endocrine disrupters. A computational model of steroid hormone biosynthesis, starting from
cholesterol in support of this effort is being developed. This project is expected to provide rich
opportunities for iterative interaction between computational modeling and data collection.
Visualizations of Computational Systems Biology Model Predictions
(Collaboration with Michael Breen, NCCT, and the National Computer Center)
The goal of this project is to develop software tools that facilitate effective interactions
between computational modelers and laboratory experimenters. As an initial project, we are
developing simulated western blots to visualize model-predicted concentration-time histories.
These visualizations will be used to correlate model predictions with literature data typically
shown as western blot images.
Risk Assessment Modeling
Dose-response modeling of formaldehyde carcinogenicity
(Collaboration with Fred Miller, Cary, NC)
An updated version of an existing dose-response model for the carcinogenicity of
formaldehyde is being developed. This effort focuses on refinements directed at specific issues
that were raised after publication of the earlier model.
4
-------
IIID-6
Title: Use of Toxicogenomics Data in Risk Assessment: Case Study for a Chemical in the
Androgen-Mediated Male Reproductive Development Toxicity Pathway
Lead: Susan Euling, NCEA
Research Issue and Relevance: The goal of this project is to address the question, "Can
existing toxicogenomics (TG) data improve Environmental Protection Agency (EPA) chemical
health or risk assessments?" Although genomics data promises to impact multiple areas of
science, medicine, law, and policy, there are only a few areas where genomics data currently has
application (e.g., biomarkers of disease). As the technology continues to advance, EPA will
need to prepare for genomics data availability and submission by: 1) identifying areas of risk
assessment where such data may be particularly useful; 2) developing acceptance criteria for
inclusion of toxicogenomics data in risk assessment; and 3) developing approaches for the use of
toxicogenomics in risk assessment. These needs will likely require an iterative and collaborative
research process between risk assessors and scientists (inside and outside the Agency). At the
NCEA sponsored Genomics and Risk Assessment Colloquium in 2003, one of the
recommendations was to conduct case studies that could provide a practical attempt to
incorporate currently available toxicogenomics data that would illuminate issues and the
methods development. This project is responds to this recommendation.
Approach: To address the question of whether TG data can improve health risk assessments, a
case study will be performed in which TG data for one chemical will be incorporated
qualitatively within the hazard characterization step of a recent or ongoing EPA chemical health
or risk assessment. Integrating TG data into an assessment case study will identify areas that
may be impacted by TG data and contribute to the development of criteria and approaches for
incorporating TG data in assessments. The chemical for the case study will be selected from
among those that affect the androgen-mediated male reproductive development toxicity pathway.
This pathway was selected because it is well-characterized, there are published TG studies, and a
number of genes in the pathway have been identified. Criteria for chemical selection for the case
study will include: 1) a relative abundance of available TG data; 2) a recent or ongoing EPA
assessment; and 3) an interest by EPA Program and/or Regional Offices. Using the most recent
or ongoing (depending on the selected chemical) EPA assessment as a starting point, the team
members will conduct an evaluation of the data presented in the assessment, focusing on the
hazard characterization and dose response sections. The following questions will be considered:
Is the mode of action fully understood for all of the endpoints of concern? Could gene expression
information aid in a further understanding of the MO A? Does the TG data provide insights into
other aspects of the assessment (e.g., dose-response)? The TG data analysis will then be
integrated into the assessment. A report of the case study will be developed and conclusions will
be drawn about whether the TG data strengthened or corroborated the risk assessment,
qualitatively, and the utility of the approach for incorporating TG data for future risk or health
assessments.
-------
Progress to Date: Dibutyl phthalate (DBP) was selected among five candidate chemicals as the
chemical for the case study. This decision was based on DBP having 1) the largest and most
consistent toxicogenomics database including data indicating gene expression changes in genes
known to be in the androgen-mediated male reproductive toxicity pathway, one dose-response
gene expression study with low to high dose, and "phenotypic anchoring" (i.e., linkage between
in vivo alterations and gene expression changes) for some of the gene expression data; 2) an
ongoing IRIS assessment that allows our case study to address some broadly applicable
questions about the use of toxicogenomics in risk assessment; and 3) an external review draft of
the DBP IRIS assessment that our team could utilize. Two subgroups (Data and Approaches
Subgroups) were formed to assess the toxicity data and the toxicogenomics data. The
Approaches Subgroup has assessed the draft DBP IRIS assessment and associated toxicity
studies. The toxicity database of studies with male reproductive and developmental effects has
been assembled. The Data Subgroup has assembled the toxicogenomics database from eight
publications (PCR and microarray studies) and summarized each of the studies. Possible
questions to focus the case study were developed. For each question, its relevancy to the ongoing
IRIS assessment and whether the available toxicogenomics data could address the question were
considered. A collaboration with the STAR funded Bioinformatics Center at the Robert Wood
Johnson Medical School University of Medicine & Dentistry of New Jersey (UMDNJ)
Informatics Institute has been initiated. The collaborative project involves performing a genetic
pathway analysis of the eight toxicogenomics studies. The results of this analysis may suggest
additional pathways affected by DBP. In light of any additional genetic pathways affected by
DBP treatment, the in vivo toxicity data will be further assessed. The team has begun writing
drafts of chapters for the case study report. In addition to toxicogenomics and toxicity database
assessments, our approach to integrating the toxicogenomics data into the DBP assessment is
described. This approach may be useful to future health and risk assessments that utilize
toxicogenomics data.
Milestones/Products:
FY06: Draft of the case study report (includes scoping exercise complete for case studies;
i.e., progress on case study and defining approach for integrating TG data into risk
assessment); Discussions with the EPA chemical assessment team, Regions, and Program
Offices.
FY07: Conduct EPA Colloquium presenting results and lessons learned from the case
study; External peer review draft of report; Submit manuscript on project to peer-
reviewed journal.
QA
QA Plan is being evaluated
-------
IIID-7
Title: Developing Computational Tools for Application of Toxicogenomics to Environmental
Regulations and Risk Assessment
Lead: David Dix, NCCT (FTE 1.0)
Research Issue and Relevance: Toxicogenomics is the study of changes in gene expression,
protein, and metabolite profiles within cells and tissues, complementary to more traditional
toxicological methods. Genomics tools provide detailed molecular data about the underlying
biochemical mechanisms of toxicity, and could represent sensitive and precise approaches for
detecting effects of exposures, or methods for comparing these effects between species or
individuals. Thus genomics, proteomics and metabonomics can provide useful weight-of-
evidence data along the source-to-outcome continuum, when appropriate bioinformatic and
computational methods are applied towards integrating molecular, chemical and toxicological
information. The Interim Policy on Genomics (http://www.epa.gov/osa/spc/genomics.htm)
recognizes that if genomics is to become useful in regulatory decision-making, risk assessment,
and environmental monitoring, the Agency will require the computational methods to handle
such data. Measuring changes in gene expression using DNA microarrays has proven useful for
identifying biological processes and informing hazard identification and mode of action in
toxicological research. Similar microarray data have already arisen in Agency environmental
decision-making, and regulatory applications of genomics are likely to increase. EPA's Science
Policy Council (SPC) paper on the Potential Implications of Genomics for Regulatory and Risk
Assessment Applications at EPA (http://www.epa.gov/osa/genomics.htm) highlights the potential
of toxicogenomics in chemical prioritization and risk assessment. To realize this potential, EPA
must have the ability for proper analysis and storage, as well as the computational tools to
incorporate these types of data into regulatory decisions. Development of these databases and
tools, and application of these various toxicogenomic data within Program and Regional Offices
will provide EPA staff with valuable, practical training in genomics and associated disciplines.
As toxicogenomics grows more important to environmental science and policy, the NCCT will
help EPA develop the computational tools and methods to properly evaluate genomics
information
Approach: To address the need for development of tools for managing and analyzing
toxicogenomics data, the National Center for Computational Toxicology (NCCT) is working
across the Office of Research and Development (ORD), the Program and Regional Offices of
EPA, and with other Federal and extramural partners. The NCCT is coordinating it's
toxicogenomics efforts with the rest of the Agency through the SPC's Genomics Technical
Framework and Training Workgroup. This Workgroup has drafted an Interim Guidance for
Microarray-Based Assays: Regulatory and Risk Assessment Applications at EPA, that
recommends continued collaboration with other federal agencies and stakeholders in developing
management and analysis tools for genomics data, and the execution of a series of case studies of
genomics applications to chemical prioritization or risk assessment. The NCCT intends to
follow these recommendations through a series of projects and partnerships within the Agency,
and with the FDA and the STAR-funded Environmental Bioninformatic Centers (EBC) in NC
and NJ.
1
-------
First, the NCCT will develop a federated database(s) and analytical tools for the
management and analysis of toxicogenomic data for both research and regulatory applications.
This project was initiated in FY2006, and is building on the success of FDA's ArrayTrack
database. It is NCCT's goal that this effort provides a complete data management solution that
addresses requirements unique to scientifically-based risk assessments, confidential and
proprietary data security, public access, and other aspects of regulatory application. Consistency,
scientific and operational robustness, common access, and availability in a scalable environment
are all part of these data management requirements. The expected result is an Agency-wide data
management solution integrating genomics, toxicological, and other key data required for both
research and regulatory applications. This EPA database containing gene expression profiles and
toxicological data for a wide variety of chemicals will facilitate creation of the statistical and
computational methods for predictive toxicology. As part of this effort to develop microarray and
toxicogenomic analysis tools, the NCCT is continuing to participate in the Microarray Quality
Control (MAQC) project. The MAQC is a comprehensive study of microarray quality control
and cross-platform comparison, executed by a consortium of many commercial, government
(FDA, EPA, NIST, NIH), and academic participants. The MAQC objectives include measuring
intra-platform performance; inter-platform comparability; relative accuracy; and concordance of
expression measurements to other technologies (e.g. TaqMan PCR).
The second part of NCCT efforts in toxicogenomics are a series of specific, model
applications of toxicogenomics data to environmental chemical prioritizations or risk
assessments. This includes some of the toxicogenomics elements of the ToxCast program for
chemical prioritization, that are generating genomics and metabolomics data from cell cultures
which can be loaded into the EPA toxicogenomics database. Also, toxicological data from EPA
Program Offices for pesticides and other environmental chemicals will be captured into the
toxicogenomic database for use for both chemical prioritization efforts (i.e., ToxCast) as well as
risk assessment applications of toxicogenomics. Additional risk assessment applications of
toxicogenomics data include ORD-wide work on the conazole fungicides cancer and non-cancer
mode(s) of action; pyrethroid pesticides neurotoxicity; perfluorylalkyl acids (PFAA)
developmental and hepatotoxicity; the antiandrogenicity of phthalates; the immunotoxicity of
diesel particles; and the role of urban air particles in asthma. The NCCT will coordinate across
ORD and the Program and Regional Offices, as well as with the EBC in NC and NJ, on how to
manage and analyze these datasets in ways that maximize their utilization in risk assessments.
Milestones/Products:
FY06: Completion of participation in the Microarray Qaulity Control (MAQC) with FDA
and publication of papers describing best practices- includes description of SPC
Genomics Technical Framework. Installation of FDA ArrayTrack database for ORD and
Agency use.
FY07: Continued development of ArrayTrack database and analytical tools for
toxicogenomics in cooperation across Agency, with FDA, and with the NC and NJ
Environmental Bioinformatics Centers. Trial incorporations of toxicogenomic data into
risk assessments in collaboration with various Program and Regional Office staff.
2
-------
FY08: Publication of examples and principles for integrating toxicogenomic data into
risk assessments in peer-reviewed scientific journals and contribution of these principles into
Agency science policy.
QA
QA Plan is being evaluate
-------
IIIE-1
Title: Dose-Time-Response Modeling for Evaluating Cumulative Risk of TV-Methyl Carbamate
Pesticides
Lead: R. Woodrow Setzer, NCCT (FTE 0.4)
Research Issue and Relevance: EPA's Office of Prevention, Pesticides, and Toxic Substances
(OPPTS) is required by the Food Quality Protection Act to completely reevaluate pesticide
registrations by the end of August, 2006. This evaluation must include the evaluation of
cumulative and aggregate risk of compounds that act through a common mechanism. One such
group of pesticides is the TV-methyl carbamates, whose common mechanism is the reversible
inhibition of acetylcholine esterase (AChE) in nervous tissue. The inhibition of AChE is
generally rapidly reversible, with half-lives in rodents at low doses of 1 - 3 hours or so.
Accounting for episodic exposures of the carbamates, for example through the diet, then requires
a more complex risk assessment model for combining the effects of multiple exposures (for
example, an exposure of chemical A at breakfast, followed by an exposure of chemical B at
lunch), than it would for exposures of compounds with longer-lived common effects. It also
requires the estimation of parameters for models that describe both the dose-response and the
recovery of AChE activity from multiple studies of the same dose response relationship.
Approach: The approach to this problem is to adapt the dose-response model already used for
modeling AChE activity after organophosphate exposure to include a model for recovery; to use
a hierarchical model to account for the variability among studies, so that all the datasets for a
chemical can be used to estimate dose-response parameters; to develop a model that accounts
for AChE inhibition subsequent to multiple non-simultaneous exposures, using the parameter
estimates derived from dose-response modeling.
Progress to Date: A dose-time-response model has been developed that can be used both for
modeling individual chemical dose-time-response data and for predicting AChE inhibition;
preliminary parameter estimates have been generated.
Impact: The dose-response results of this work will be used in the dose-response assessment
portion of the Agency's TV-methyl carbamate cumulative risk assessment. Predictions of AChE
inhibition under realistic exposure scenarios may be used in the hazard characterization portion
of that risk assessment. Methods developed in this analysis could be used in the future for
cumulative risk assessments for agents with ephemeral acute effects.
Partnerships/Collaborations: This work is being done in collaboration with scientists in the
Health Effects Division of OPPTS and from the Neurotoxicology Division of the National
Health and Environmental Effects Research Laboratory.
Milestones/Products:
FY06 - Science Advisory Panel review of preliminary cumulative risk assessment,
including dose-response modeling.
1
-------
FY07 - Release and Science Advisory Panel review of revised cumulative risk
assessment.
FY08 - Submission of methods and dose-response models for publication in the peer-
reviewed literature. Incorporation of dose-time-response methodology into the Agency's
BMDS software.
QA:
QA Plan is being evaluated
2
-------
IIIE-2
Title: Application of Visual Analytic Tools to Evaluate Complex Relationships between
Environmental Factors and Health Outcomes
Lead: Elaine Cohen Hubal, NCCT (FTE 1.0)
Research Issue and Relevance: Characterizing cumulative risk and understanding the complex
relationships between environmental exposures and human health outcomes requires collection
and analysis of a wide range of data. Information on the characteristics of multiple stressors
(chemical, physical, biological and psychosocial), the characteristics of the human receptor
(genetics, health status, life stage, behaviors, social factors, etc.) at multiple levels of
organization (individual, community, population), and the temporal and spatial patterns of
exposures and outcomes must be combined to assess cumulative risk.
Current approaches for designing studies and evaluating collected data tend to focus on a
limited number of environmental factors and/or measures of outcome. As such, it is difficult to
understand the potential impacts of the full range of factors on environmental health. More
holistic approaches for interrogating this multidimensional data space are required to identify
potentially important relationships for further study and analysis. Application of emerging
computational tools will allow us to optimize utility of collected data, improve understanding of
complex exposure-outcome systems, and improve risk assessment.
Approach: In this project, we plan to use engineering principles to develop conceptual and
mathematical models of the human-receptor, source-to-outcome system, and visual analytic tools
to address the significant challenges associate with characterizing cumulative risks. First, we
will apply a systems approach to develop a human-receptor based conceptual framework. We
plan to adapt the strategy for creating conceptual models for complex ecological risk assessments
presented by Suter (1999).
We will also identify and evaluate visual analytic tools required to address analysis needs
for characterizing multi-factorial relationships between environmental factors and human health
outcomes. Visual analytics is a new branch of visualization that merges scientific and
information visualization and includes technologies from other fields, including information
extraction, knowledge management, and statistical analysis. Visual analytics tools can be
developed and applied to represent complex multidimensional data. Large, dynamic and
complex data sets containing text, measurements, and images, can be effectively combined to
reveal significant relationships and trends and to enhance discovery. Visual analytics can be
used for outcome analysis and visualization, to find patterns and subtle relationships in data, and
to infer rules that allow predictive analysis to prevent and mitigate environmental disease.
Finally, we will test and demonstrate VA tools using existing available data. We
propose doing this demonstration using data from children's cohort studies to explore the
potential of visual analytics to facilitate evaluation of the effects of environmental exposures on
child health and development. Another possible demonstration could involve using data
collected for a community-based cumulative risk assessment.
1
-------
Progress to Date:
To date some preliminary (and very conceptual) research has been conducted to consider
how biomonitoring data can be used to characterize cumulative risk and how psychosocial
factors can be incorporated into cumulative risk assessments. In addition, a novel data mining
tool is being tested and visual analytic software packages are being evaluated using extant
children's exposure data to identify relationships between exposure factors and body burden for
a selected set of chemical agents.
Impact:
Across EPA, program and regional offices are being called on to assess cumulative risk
resulting from real-world exposures. The Agency is also being required to identify vulnerable
populations, characterize life-stage risks, and evaluate gene-environment interactions. Multi-
factorial analyses of one form or another are required: to conduct national-scale regulatory-based
risk assessments (program offices); to conduct community-based risk screening and remediation
(regions and states); to support epidemiology studies investigating gene-environment interactions
(interagency); and to characterize exposure and risk for public health tracking (all of the above).
The approaches and tools developed through this research will help the Agency meet the
increasingly complex needs for cumulative risk assessment. In addition, results of this effort
may be used to develop concepts and tools for application to the Detroit Children's Study, the
North Carolina Cohort, and the National Children's Study.
Partnerships/Collaborations:
In this project, the NCCT will take advantage of visual analytic capabilities that are being
developed in the Scientific Visualization Center at the EPA National Environmental Scientific
Computing Center. In our initial evaluation of Visual Analytic software packages, we will build
on statistical analysis of children's exposure data conducted in NERL. We also hope to
collaborate with investigators conducting children's cohort studies to explore the potential of
visual analytics to facilitate evaluation of the effects of environmental exposures on child health
and development. As a first step in developing these collaborations, we presented this research
concept during a breakout session on Computational Toxicology at the 2005 Collaborations for
Children's Environmental Health Research Workshop (part of the EPA/NIEHS Children's
Centers Scientist-to-Scientist Workshop, July 11, RTP). Finally, we are exploring a potential
collaboration with PNNL's National Visualization and Analytics Center.
Milestones/Products:
FY06 - Demonstration of potential for application of VA to evaluate exposure data. Workshop on
applying VA to analyze children's cohort data.
FY07 - Generic conceptual model of complex relationships between environmental factors and human
health outcomes.
FY08 - Demonstration of application of VA to analyze children's cohort data.
QA:
QA Plan is being evaluated and will be inserted in the next version.
2
-------
MEMORANDUM OF AGREEMENT
BETWEEN THE
NATIONAL CENTER FOR COMPUTATIONAL TOXICOLOGY,
NATIONAL HEALTH AND ENVIRONMENAL EFFECTS RESEARCH
LABORATORY, AND
NATIONAL EXPOSURE RESEARCH LABORATORY
Background
On October 7, 2004, the Office of Research and Development (ORD) announced the
establishment of the EPA National Center for Computational Toxicology (NCCT) headquartered
in Research Triangle Park, NC. The Center will provide scientific expertise and leadership
related to the application of mathematical and computational tools and models to high priority
Agency needs. These needs include improving the Agency's data reporting requirements for
environmental fate and transport and for toxicity testing, setting priorities for the acquisition of
those data based upon predictive models, and for understanding toxicity and risks posed by
environmental agents. The tools and models are derived from modern technological advances in
the general areas of computational chemistry, mathematical and systems biology, and similar
systems.
The reorganization package for the NCCT staff identified nineteen full time positions staff,
including 11 who were initially detailed from other ORD National Labs/Centers/Offices. After
implementation of the reorganization, these individuals will be permanently reassigned to the
NCCT until the proposed sunset date in October 2009. Although two administrative positions
exist in the NCCT. many of administrative and research support functions will be provided by
the Program Operation Staffs of the National Health and Environmental Effects Research
Laboratory (NNHERL) and the National Exposure Research Laboratory (NERL). This MOA
outlines the administrative support services that will be the primary responsibility of the NCCT
and those that NHEERL and NERL will provide to the NCCT. NCCT management is
responsible for the decisions and actions implemented by the NCCT. Since good
communication will be an essential element in the success of these interactions, the parties agree
to bi-lateral discussions at least bi-monthly with the senior managements of NERL and NHEERL
to discuss implementation issues and resolve any ambiguities or issues that arise as the
relationships between the organizations develop and mature. This agreement is effective as of
January 31. 2005 and will remain in effect until amended by mutual agreement of the
participants.
Purpose of the Memorandum of Agreement and Services Provided
The delegation of authority for ORD's NCCT exists under the ORD Policies and Procedures
Manual. Chapter 7.4 dated March 11. 2005 and this agreement refers to the administrative
support that will be supplied by NHEERL and NERL. This support includes but is not limited
to: Funds Control. Freedom of Information Act requests. Training Agreements, Senior
-------
Environmental Employment Program, Technical Qualifications Board, Buildings and Facilities,
Capital Equipment. Extramural Management. Technology Transfer, Contractual Support. Health
and Safety, and Quality Assurance.
Administrative Support Services that will be primary responsibilities of the NCCT:
1) Budget
- Monitor funds utilization, prepare budget reports and plans, and prepare justifications
for budget initiatives
Create WCF Service Agreements and Funding Levels
Prepare FTE/PC&B Projections and Workforce Management/Strategies
Initiate reprogrammings to transfer funds to other organizations or across budget
structure dimensions
Create and maintain IRMS Execution Implementation Plans
Develop NCCT budget narratives for OMB and President's budget development (with
ORMA)
2) Records Management
Provide new and/or updated NARA/Agency/ORD guidance
Represent the NCCT's interests with the Project Officer on Center file inventories,
records awareness week, Center training needs, negotiation/funding of records
management contractor
Assist the NCCT staff with interpretation of records guidance, schedules, etc.
3) Timekeeping
4) Travel authorizations, itinerary planning, and preparation and submission of Travel Vouchers:
purchase card holder and authorizing official
5) Ethics
The NCCT Director will serve as the Deputy Ethics Official (DEO).
The NCCT Program/ Management Analyst will serve as the DEO Assistant.
6) Human Resources
Manage all Human Resources activities, including performance, orientation, awards,
training, career development, leadership development, post doc program, form
preparation, and HR liaison with OARM Human Resources Management Division.
7) Federal Managers' Financial Integrity Act (FMFIA)
-------
- Develop and submit NCCT's Mid-year and the Annual Assurance Letter
- Represent NCCT on ORD workgroups related to Management Integrity
- Keep Center up to date on new and/or changed guidance, policies and/or procedures
8) Management Reviews
Conduct Management Reviews of the Center
Prepare review reports
Develop new protocols, review guides, etc. if needed due to the nature of work being
conducted by the Center
9) Extramural Instruments
Prepare procurement documents including those for contracts, interagency agreements
and cooperative agreements.
- Work with Office and Science Policy on Technology Transfer needs.
Administrative Support Services provided to the Center by NHEERL
1) Funds Control Officer
- NHEERL will provide funds certification and data entry into appropriate ORD and
Agency systems. NHEERL commits to provide this function for the duration of FY
2005 and will evaluate the impact of the workload associated with this function for
future FY's. If the workload is determined to be more than the FCO can manage
comfortably, NHEERL will notify NCCT management with sufficient notice to make
other arrangements for this support.
2) Electronic Purchase Card System
NHEERL will allow NCCT to utilize its lotus notes-based electronic purchase card
system. If NCCT chooses to use the system, it will provide funding to support its use
as well as maintenance of this system.
3) Freedom of Information Act requests (FOIA)
NHEERL will coordinate and prepare responses to FOIA requests in accordance with
FOIA regulations. The Center will gather the responsive information, prepare it for
submission, and submit it to NHEERL.
4) NRC and other Training Agreements
- NHEERL will allow the Center access to the training agreements with local
-------
universities and its agreement with the National Research Council.
NCCT will be included in calls for needs associate with those agreements and will
work with NHEERL PO's and coordinators.
NCCT will be responsible for funding any individuals being trained by its staff through
those agreements. An agreement will be reached on the level of funding the Center
will place on particular agreements.
5) Senior Environmental Employment Program (SEEP)
NHEERL's SEEP Coordinator will:
Coordinate recruitment packages (Position Description Form, Requisition Form, Cost
Analysis Worksheet, Commitment Notice and IFMS screen, and memo) for processing
and forward to selected grantee for recruitment.
Request Commitment Notices for renewals of SEEP enrollees.
Provide Center with new/revised guidance, policies and/or procedures regarding the
program.
Forward the quarterly report to the appropriate SEEP Monitor(s) for reconciliation.
The Coordinator will provide information on the reconciliation procedures.
6) Technical Qualifications Board (TQB)
- NHEERL will include NCCT in calls for candidates requiring review through the TQB
process and will provide support for scheduling and holding the TQB review meetings.
- NCCT will provide funding for the professional services contracts for the ad hoc
reviewers used for the review of NCCT's candidates.
7) Building & Facilities Support
The NHEERL Facilities Coordinator will provide support for B&F and R&I processes
as well as the ORD Capital Equipment process.
This includes updating ORD and Agency systems and databases related to space and
facilities.
This includes liaison with OARM on space issues and assistance in facilities design
and renovation.
NCCT funding will be allocated to RC-6 (NCCT) and NCCT will provide the
appropriate funding for its projects.
8) Graphics, Web Development and Photography Support
Support for graphics, photography, and web development and maintenance will be
provided in coordination with NHEERL.
- NCCT will have access to NHEERL support contracts, or NHEERL will assist in
referring NCCT to appropriate available contracts.
- NCCT will provide funding for the contractual support that they require.
-------
Administrative Support Services provided to the Center by NERL;
1) Extramural Management
The appropriate NERL EMSers (Extramural Management Specialists) will provide
advice and guidance on appropriate extramural vehicles: keep NCCT current on
requirements pertaining to extramural policies and procedures; and include NCCT in
calls and due dates related to extramural vehicles and procurement.
The appropriate NERL EMSers will review all NCCT packages that require EMS
review and/or concurrence.
2) Contractual Support Mechanisms
NERL will allow the Center access to various technical support contracts that it
maintains provided the requested effort is within the Statement of Work (SOW) and
that the Center covers all costs associated with that use.
3) Health and Safety
- NERL's Health and Safety Team will support the Center.
4) Scientific Quality Assurance (QA)
- NERL's QA Director will ensure that all Center research projects comply with Agency
QA requirements
5) Information Management (IM)
NERL will provide support in the area of IM to NCCT.
Administrative Support Serv ices provided to the Center by Consolidated Support
The following Administrative Support Services are currently provided to all Laboratories and
Centers at RTP by consolidated services. If there is a change to how the support in these areas is
provided, support for the NCCT will be addressed at that time.
1) Public Affairs and Scientific Communications Support
The ORD Communications Team provides support for media, events, and local
congressional coordination. The Team also provides the interface with ORD and EPA
Communications staff
- NCCT will provide funding on an allocated basis to support the communications
effort.
2) Information Technology
-------
The ORD Consolidated Center's information technology (IT) staff provide a wide range of
IT. and information security (IS) to the labs and centers at RTP. General IT and IS support
for NCCT will be provided by the Consolidated Center. NCCT will provide funding for
IT support for its full-time employees. IT support for NCCT part-time employees will be
provided by the "home" laboratory.
RTP Shared Cost Budget
The RTP Shared Cost Budget covers services that all of RTP uses. For the Center, these
services include the EPA Wellness Center, special emphasis program support.
Management Council initiatives, and supplies for the T.CO.
The RTP Management Council will agree on a formula-based allocation. The Center will
receive a breakout for their portion of the total shared cost budget.
-------
rOM'l.RKFXCTS
/ /
Jmy BLfttc.-.u?!, fkprii1, Direciui Dju
[vnifMiiil t/cna-r'Cofl ( onipataftnnal '1 i>\iv'olou\
J ^ i', ' V ' i I
'•.||fV'' '-»T ¦ "Vrii .1 ..I ..,.^awawg!iB^W5l» , JLm*-Xf#*nLr
lewc1 i\'fniiif,. Depufx DuecHif fin Miiijueinenl tbtl-
I Kpuiiirc Hesizsrch !.uliu^uv-.
...iP...> uii'iihi n.inr
John ,lobv\ l)cput\ rthcviortni \inua«swwt> Date
National I feallh ami I'nvitonmrut.ii F.iYocis Ur.search I-al'isnUorv
x'7^ j:'v
6
-------
Intentionally Left Blank
-------
December 2005
EPA Community of Practice:
Biological Modeling Working Group (BMWG)
The EPA Biological Modeling Working Group is proposed to be formed in Summer 2005
to advance the principals for development and application of dosimetry and other
biologically based models within the Agency. Dosimetry modeling includes multiple
forms of toxicokinetic modeling (e.g., physiologically based toxicokinetic (PBTK)
modeling, compartmental modeling), respiratory tract dosimetry modeling (e.g.,
computational fluid dynamics), and related modeling (e.g., dermal absorption modeling).
The working group will also focus on biologically based response modeling with special
emphasis on using the newest "omics" information in biologically based models . The
goal of this work group is to foster adoption of modeling science by Agency clients in
regulatory decision making. A cross-ORD group that also has representation from
outside of ORD can carry great influence throughout the Agency, helping assure that
ORD efforts are viewed as relevant and important to the Agency mission.
The work group would include representatives of the ORD Labs and Centers with
expertise in this area and have a representative from the Risk Forum. The BMWG would
draw on the expertise of other scientists within ORD and the Agency as needed for
specific issues. The BMWG would ask the Risk Forum to provide broader Agency-wide
comment and review on specific matters as it determined necessary. Committee
members would serve for a period of two years, following which the lab/center would
renew their membership or appoint a new member.
The charter of the BMWG includes the following functions:
• Facilitate communication and co-ordination among the ORD biological modelers
to foster their role as the major technical resource for the Agency in this area.
• Develop consensus positions around issues for how analyses are conducted using
models to provide consistency (e.g., processes for application of uncertainty
factors in non-cancer evaluation, methods for using models to describe population
distributions, durations over which AUC is calculated)
• Develop consensus positions around issues of model evaluation and
documentation to provide assistance to model developers inside and outside the
Agency.
• Develop recommendations for research needs to be addressed by ORD labs and
centers (e.g., methods for statistically evaluating alternative model structures and
characterizing model output uncertainties)
• Facilitate the development of training materials and programs for chemical
managers and risk assessors that assist in the implementation of these
technologies.
• As the science advances develop recommendations and guidance on the use of
omics information to enhance dosimetry and pharmacodynamic models
• Facilitate communication to the larger modeling and scientific on issues for
modeling and applications to risk assessment (e.g., SOT symposia, specialty
meetings)
- 1 -
-------
December 2005
The work group will:
• Hold meetings: quarterly to identify issues/projects to be taken on and discuss
progress
• Organize internal meeting/conference calls around specific issues/projects
with goal of identifying state of the science, research needs, consensus
positions.
• Prepare publications for peer-reviewed literature on specific issues/topics.
• Identify research needs that could be addressed internally or through
extramural contracts or grants.
• Organize outreach to broader modeling and scientific community to
demonstrate EPA's interest and obtain input and participation
• Report regularly (every six months) to the ORD science council on issues,
progress, and outputs
• The group's representative from the Risk Assessment Forum will request time
on the Forum's Agenda to brief regular Forum members of issues, progress,
and outputs.
-------
November 28, 2005
EPA Categorization and Prioritization Community of Practice
The EPA Categorization and Prioritization Community of Practice (CPCP) formed in
December 2005 to advance research into the utility of computational chemistry, high-
throughput screening (HTS) and various toxicogenomic technologies for Agency use.
Modern computational chemistry and molecular biology technologies can provide
information about the physical and biological properties of large numbers of chemicals.
The goal of the CPCP is to advise on the development of a research program within the
National Center for Computational Toxicology (NCCT) generating data from these
technologies and interpreting it in order to categorize chemicals and predict toxicity. If
proven accurate, these toxicity predictions could then be used for prioritization of limited
testing resources towards chemicals and endpoints that present the greatest risk to human
health and the environment.
The primary function of the CPCP is to advise on the NCCT's planning, conduct and
interpretation of a chemical categorization and prioritization research project entitled
"ToxCast". The ToxCast project will be designed to have the ability to predict, or
forecast toxicity. Initially, this would entail a demonstration project based upon a set of
chemicals with a rich toxicological database (e.g., registered pesticides, or the chemicals
tested in the NTP bioassay program). This set of 200 or more chemicals would represent
a number of differing structural classes and phenotypic outcomes (e.g., tumorigens,
developmental and reproductive toxicants, neurotoxicants, immunotoxicants). The
ToxCast project would evaluate chemical properties and effects across a broad spectrum
of information domains: physical-chemical properties, predicted biological activities
based on existing structure-activity models, biochemical properties based on HTS assays,
cell based phenotypic assays, and genomic analysis of cells or organisms. The ultimate
goal of the ToxCast project would be to mine the resulting data for associations between
and among the various domains and the known toxicological properties of the base set of
chemicals, in order to provide a structured strategy to categorize chemicals, identify
potential toxicities and pathways, and to prioritize chemicals for subsequent testing based
on that information.
The Community would include individuals from the Office of Research and
Development (ORD) Labs and Centers, as well as scientists outside of EPA with
expertise in HTS, toxicogenomics, predictive toxicology or bioinformatics. The CPCP
would be chaired by a member of the NCCT, but draw on the expertise of scientists
across ORD and the Agency as needed for specific issues. The Community would ask
ORD's Computational Toxicology Implementation Steering Committee (CTISC) to
provide broader Agency-wide comment and review on specific matters as necessary.
Community members would serve for the period required for planning and execution of
the ToxCast demonstration project. It is expected that the CPCP would work closely
with the Chemoinformatics Community of Practice (chaired by Ann Richard) on issues
relevant to both groups.
The charter of the CPCP includes the following functions:
- 1 -
-------
November 28, 2005
1. Organize periodic meetings or conference calls (at least quarterly) around specific
issues or projects, with goal of identifying state of the science, research needs, and
consensus positions.
2. Organize outreach to broader HTS and toxicogenomics scientific community to
demonstrate EPA's interest and obtain input and participation.
3. Report regularly to the Director of the NCCT and the ORD CTISC on issues,
progress, and outputs.
Specific goals of the CPCP are to advise on the following:
1. Development of key partnerships and collaborations with external groups that can
facilitate development of the information needed in ToxCast. These groups
include the Office of Air and Radiation (OAR), Office of Prevention Pesticides
and Toxic Substances (OPPTS), Office of Water (OW); the NTP/NIEHS and
NIH/MLI; the ACC, CropLife, EDF, and other external groups to help develop a
consensus on the specific directions and contents of ToxCast.
2. Identification of a set of chemicals for the ToxCast demonstration project.
3. Selection of data domains and specific assays based upon pre-existing knowledge
and within the available resources.
4. Selection of key target toxicities for initial focus of the ToxCast proof of concept.
5. The impact of metabolizing capability, or lack thereof, on the efficiency of the
screening assays.
6. Development of a bioinformatic approach to mining the resulting data and
identifying signatures of concern.
7. Reporting the utility of assay results and analysis techniques to categorize pilot
chemicals according to known toxicity patterns; revise methods and approaches as
dictated by results.
8. Expanding the ToxCast project beyond proof of concept, and carrying out a
prospective assessment of the approach using chemicals currently entering a
traditional testing process.
Impact: The availability of a biologically and chemically based system (ToxCast) to
categorize chemicals of like properties and activities will provide a number of EPA
Program Offices with an extremely useful tool that heretofore has been seriously lacking.
ToxCast may be one of the first broad-scale products of the NCCT that addresses the
mission of improving the efficiency and effectiveness of hazard identification and risk
assessment methodologies employed by the EPA.
-2 -
-------
December 2005
EPA Communities of Practice:
Chemoinformatics
The Chemoinformatics workgroup is proposed to be formed in Summer 2005 to
facilitate, coordinate and integrate efforts to address the challenges of chemical structure
annotation (or indexing), retrieval, and mining of chemically-related data and documents,
including newer toxicogenomics and metabonomics data, across EPA Program Offices,
Labs and Centers. Much of EPA's public and internal chemical data records and
databases currently are indexed only by chemical name (imprecise and non-unique) and
CAS registry numbers (proprietary) and are not searchable by the more universal and
informative metric of chemical structure. Additionally, there is unnecessary duplication
of efforts and lack of coordination in the area of chemical structure-annotation and
quality review of chemical information (e.g., structures, CAS) across diverse EPA
databases. Finally, no Agency-wide chemical structure searching capability currently
exists to enable EPA scientists, regulators, and outside parties to efficiently and precisely
locate Agency chemical information based on chemical structure.
The Chemoinformatics workgroup would invite members of various Offices, Labs and
centers currently involved in either chemical database construction or use, to coordinate
ongoing efforts in the area of chemical-structure annotation and structure-searchability of
EPA records and data, and to help chart a path forward for expanded Agency-wide
adoption of capabilities in these areas. One of the first tasks of this workgroup would be
to consider and refine the elements of the draft charter below, and create a finalized
charter for the workgroup.
A draft charter of this Chemoinformatics workgroup is suggested to include the following
functions:
• Facilitate communication and co-ordination among the Agency personnel who are
directly using, developing, or managing chemical information data records,
including those associated with newer toxicogenomics and metabonomics
technologies, particularly in ORD, OPP and OPPT whose duties involve chemical
searching and structure-activity assessments.
• Propose adoption of application-independent standards for chemical structure
representation (including information pertaining to the characterization of
mixtures, racemic chemicals, polymers, and stereochemistry) that will facilitate
the broadest possible utility and compatibility of such information, both within
and outside of EPA.
• Guide the creation of a consolidated EPA database of chemical identification
information (structure, CAS, name), obtained from public sources
(chemfinder.com, PubChem, NLM ChemID) but having undergone additional
levels of quality consistency review, to serve as a common resource for the
structural annotation of EPA records or databases. [Note that physical/chemical
properties can be measured or derivedfrom the accurate chemical structures by a
variety of algorithms and means; hence, such property annotation of chemical
- 1 -
-------
December 2005
information resources will be considered as a separate and distinct task of
chemical annotation not included in the initial charter of this workgroup.]
• Create procedures and guidelines for chemical annotation of Agency data and
records, as an adjunct to EPA Information Quality Guidelines, to improve
consistency, quality, and efficiency related to storage, retrieval and mining of
chemical information within the Agency.
• Exchange information and experiences regarding available public and commercial
approaches and initiatives in the areas of chemical structure representation,
chemical searching, and managing and modeling of chemical data.
• Evaluate and recommend which of the available approaches and initiatives might
be adopted or coordinated to best serve EPA's long-term needs in this area, and
serve as an information resource to other Agency personnel on these matters.
• Evaluate and provide guidance concerning the needs and requirements of an
Agency-wide chemical structure-searching capability, possibly from publicly
available or open-source parties, that would offer low-cost, flexible solutions to
enable EPA scientists or others to locate records or data pertaining to chemical
information on EPA's intranet or internet websites.
• Facilitate the development of training materials and programs for scientists,
chemical managers and risk assessors that assist in the implementation and wider
use of these chemical information resources and technologies.
• Report regularly (every six months) to the ORD science council on issues,
progress, and outputs
Office/Division/Branch managers across various research offices, including ORD, OPP
and OPPT, will be invited to appoint representatives to this workgroup who would
benefit most from, and contribute most to these discussions, as well as who would serve
as a conduit of information to and from their respective Branch/Division/Office.
-2 -
------- |