computational toxicology: new approaches to improve environmental health protection presentation to...
TRANSCRIPT
:Computational Toxicology New Approaches to Improve
Environmental Health Protection
Presentation to the 4th International Academic Conference on Environmental and Occupational Medicine
Kunming, ChinaOctober 17, 2006
Computational Toxicology: New Approaches to Improve
Environmental Health Protection
Presentation to the 4th International Academic Conference on Environmental and Occupational Medicine
Kunming, ChinaOctober 17, 2006
Are you ready for the revolution?
D Butler, Nature Feb 15 2001; 409, 758 - 760Are you ready for the revolution?
D Butler, Nature Feb 15 2001; 409, 758 - 760
“small experiments driven by individual investigators will give way to a world in which multi-disciplinary teams….emerge as the key players…..in the era of systems biology in which the ability to create mathematical models describing the function of networks of genes and proteins is just as important as traditional lab skills.”
“the research teams that will be most successful….are those that switch effortlessly between the lab bench and the suite of sophisticated computational tools.”
DNA
mRNA
Protein
Metabolites
Transcription
Translation
Metabolism
Molecular Biology
Enabling TechnologiesEnabling Technologies3
“…to integrate modern computing and information technology with molecular
biology to improve Agency prioritization of data requirements and risk assessment
of chemicals”
“…to integrate modern computing and information technology with molecular
biology to improve Agency prioritization of data requirements and risk assessment
of chemicals”
www.epa.gov/comptox
What’s It All AboutWhat’s It All About
Digitization Legacy data
Dispersed data
Scale Chemicals
Biological space
Levels of biological organization
Quantifying Physiology, biochemical pathways and networks, biology
Data mining and management
The Source to Outcome ContinuumThe Source to Outcome Continuum
Source/Stressor Formation
Environmental Conc.
External Dose Target Dose
Biological Event
Effect/Outcome
Historically the problem has been approached one chemical at a time, one
stage at a time, with little progress in predicting across the stages and across chemicals. Current demands on the EPA are making this an untenable approach.
Computational Toxicology was initiated to provide new thinking to overcoming the
bottlenecks.
The Source to Outcome ContinuumThe Source to Outcome Continuum
GENOMICS, PROTEOMICS, and METABONOMICS:IDENTIFY/CHARACTERIZE
Source/Stressor Formation
Environmental Conc.
External Dose Target Dose
Biological Event
Effect/Outcome
COMPUTATIONAL METHODS:PREDICT/PRIORITIZE
ToxicologyGenomics
SystemsBiology
Informatics
HighThroughput
Screens
8
Toxicity of conazoles in Integrated Toxicogenomics Study
Toxicity of conazoles in Integrated Toxicogenomics Study
Chemical Liver
ToxicityLiver
TumorsThyroid
DisruptionTesticular Toxicity
Myclobutanil
NO NO NO YES
Propiconazole
YES YES NO NO
Triadimefon
YES YES YES YES
Common MOA across conazoles, tissues, cancer/non-cancer?
Use combination of toxicology and genomics to answer these questions.
Enriched for: Azole antifungals, p<4.9E-11
SP_Hepatomegaly, early gene expression:SV0412215R5RU:
-4 -3 -2 -1 0 1 2 3
-3
-2
-1
0
1
2
Sca
lar
pro
du
ct s
core
s fo
r P
XR
ac
tiv
atio
n s
ign
atu
re
Scalar product scores for hepatomegaly signature
MYCLOBUTANILPROPICONAZOLETRIADIMEFON
10Iconix genomics signatures combine with their reference Iconix genomics signatures combine with their reference database to correctly group EPA compounds with other database to correctly group EPA compounds with other
conazole antifungalsconazole antifungals
Martin, et al Toxicol. Sci. in press
Flu PropiMyclo Tri
A: Liver CYPs
CYP2g1
CYP1a2
CYP3a3
CYP CMF1b
CYP2c
CYP4F4
CYP4A2
CYP3a9
Flu PropiMyclo Tri
B: Liver CAR/PXR
CYP1a2
CYP3a3
Alas1
Ahr
Por
CYP3a9
Udpgtr2
Flu PropiMyclo Tri
C: Testis CYPs
CYP2j4
CYP11b3
CYP2j3p1
CYP2d2
CYP2e1
CYP11
Flu Propi Myclo Tri
Gtf2a2
Hsd17b4
Psbp1
Sult4a1
Nr1d2
Nr3c1
SOCS-2
Nr1d1
Nr4a2
Hsd11b1
D: Testis Steroidogenesis
Up
Down
CYP and CAR/PXR are differentially expressed in conazole treated rat livers
Toxicological Information GapsToxicological Information Gaps
>10,000 Chemicals in Need of Evaluation
>10,000 Chemicals in Need of Evaluation
Antimicrobials 750
EDCs 1500
HPVs 3300
Pesticidal Inerts 3310
MPVs 5400
H2O CCLs 7324
Large Data Gaps Exist for Many Chemical Types
Large Data Gaps Exist for Many Chemical Types
Bioactivity Profile of Environmental Chemicals
0
20
40
60
80
100
120
-2 -1 0 1 2 3 4 5 6 7 8
Log Conc (nM)
Ac
tiv
ity
Su
m
Specific Mechanism
Unknown/Multiple Targets
Non-Toxic
NOEC
Primary Toxicology
Target
Target Gene
Family
Gene Family Promiscuity
Non-specific Assay Interference
HTS Target Assay Panels
Goal: assesses broad biological effect patterns and correlate with the patterns of known toxicants for forecasting potential for hazard
ToxCast – A New EPA Research Program (Dix et al, Toxicol Sci. 2006 in press)
ToxCast – A New EPA Research Program (Dix et al, Toxicol Sci. 2006 in press)
Physical-Chemical Indicators
LogP pKA ……
Chemical 1
Chemical 2
Chemical 3
.......
Bio-Computational Indicators
PASS
Molecular Receptor Docking
#1
……
Biochemical Based Indicators
GPCRsHMG-CoA reductase
……
Cellular Based Indicators
H4IIE H295R ……In Vitro Omics Indicators
Primary Heptatocyte
s
Stem Cells
……Model Organisms
Response 1 Response 2 ……
Chemical Grouping
Bin 1
Bin 2
Bin 3
….
….
Bin ….
Increasing Biological Relevance
Increasing Cost
Pesticides as Proof of ConceptPesticides as Proof of Concept
~800 Registered in the United States
Wealth of Toxicological Information Developmental, reproductive, chronic, etc
Represent broad range of chemistries Azoles, carbamates, pyrethroids, triazines, etc.
Designed with biological activity in mind Receptor binding, enzyme inhibition, cytoskeletal,
etc.
Have created a library of 400 for study
Molecular Docking Experiments & virtual-HTS
A
Target
+
Complex
A
Affinity grid for hPPAR-
PDB ID: 1I7GPDB ID: 1I7G
nM M mM
Affinity increase
Lig
and
s in
mu
ltip
le c
har
ge
stat
esL
igan
ds
in m
ult
iple
ch
arg
e st
ates
PPAR- agonist Fluorocarbons
PPAR Receptor and Molecular Docking
Docking Animation: PFOAs & PPAR
ToxCast – Beyond Proof of Concept
ToxCast – Beyond Proof of Concept
Availability of a science-based system to categorize chemicals of like properties and activities
Increasing confidence as database grows
Once operational, Mode of Action leads for new chemicals
Provide EPA Program Offices with a relatively inexpensive predictive tool box that heretofore has been seriously lacking
Improve the efficiency and effectiveness of the use of animals in hazard identification and risk assessment
ToxCast
Assays
ToxCastDatabase
HazardPrediction
ChemicalChemical
ChemicalChemical
ToxCast
Assays
PrioritizeChemical
YES
NO
No FurtherTesting
LOW
FurtherScreening
And Testing
HIGH
MEDIUM
ToxCast
Assays
ToxCastDatabase
HazardPrediction
ChemicalChemical
ChemicalChemical
ToxCast
Assays
PrioritizeChemical
YES
NO
No FurtherTesting
LOW
FurtherScreening
And Testing
HIGH
MEDIUM
Chemical
Procurement-
Management
Metabolomic
Profiling
Genomic
Profiling
Biochemical
Assays
Cell-Based
Screening
High-Content
Analysis
Microelectronic
Monitoring
Model
Organisms
Toxicity
Predictor
ToxCastDatabase
Chemical
Procurement-
Management
Metabolomic
Profiling
Genomic
Profiling
Metabolomic
Profiling
Genomic
Profiling
Biochemical
Assays
Cell-Based
Screening
High-Content
Analysis
Microelectronic
Monitoring
Model
Organisms
Toxicity
Predictor
ToxCastDatabase
Drinking Water
Contaminants
IRIS
Ecotox, A
ster, Teratox
TS
CA
Su
bstance
Registry List
Pe
stic
ide
A
ctiv
es
Pe
stic
ide
O
the
r In
gre
die
nts
High P
roduction
Volu
me
Informa
tion S
ystem
EPA’s Chemical Data Islands
Green C
hemistry
Air
To
xic
o-
ge
no
mic
s
SMILES
Chemical Structures
Chemical NamesCAS Registry Nos.
Drinking Water
Contaminants
IRIS
Ecotox, A
ster, Teratox
TS
CA
Su
bstance
Registry List
The world of chemical data
Pe
stic
ide
A
ctiv
es
Pe
stic
ide
O
the
r In
gre
die
nts
High P
roduction
Volu
me
Informa
tion S
ystem
EPA’s Chemical Data Islands
Green C
hemistry
Air
To
xic
o-
ge
no
mic
s
SMILES
Chemical Structures
World Wide WebNational
Toxicology Program
European Chemicals Bureau (SIDS)
National Library of Medicine
Chemical NamesCAS Registry Nos.
PubChem
Chemical structures
Drinking Water
Contaminants
IRIS
Ecotox, A
ster, Teratox
TS
CA
Su
bstance
Registry List
The world of chemical data
Pe
stic
ide
A
ctiv
es
Pe
stic
ide
O
the
r In
gre
die
nts
High P
roduction
Volu
me
Informa
tion S
ystem
EPA’s Chemical Data Islands
Green C
hemistry
Air
To
xic
o-
ge
no
mic
s
SMILES
PubChemWorld Wide Web
National Toxicology
Program
European Chemicals Bureau (SIDS)
National Library of Medicine
Chemical NamesCAS Registry Nos.
Chemical structures
Chemical Structures
DS
ST
ox
ChemicalChemicalstructuresstructures
http://www.epa.gov/ncct/dsstox
Capturing the Legacy DataCapturing the Legacy Data
Correlating Domain OutputsCorrelating Domain Outputs
Cell based Assays
Nuclear Receptors
Toxicology Endpoints
Physical chemical propertiesProfile Matching
27
ToxCast Training Set – Pesticide Actives
ToxCast Training Set – Pesticide Actives
correlations correlationsToxRef DB
ToxicologicalProfiles
DSSTox
Chemical Structural Features
27
HTS Profiles
ToxCastDB
ArrayTrackDB
Genomics Profiles
Myclobutanil ? ? + + - -
Propiconazole ? ? - - + -
Triadimefon ? ? + - + +
Rat Thyroid Tumor
HTS Profile Correlation
Genomic Profile CorrelationChemical
MaleFertility
Testicular Atrophy
Mouse Liver Tumors
ToxCast Chemical Classificationof Environmental Chemicals
ToxCast Chemical Classificationof Environmental Chemicals
correlations correlationsToxRef DB
ToxicologicalProfiles
DSSTox
Chemical Structural Features
Chemical Classification Based on Known or Predicted Toxicity
Profiles
Tox
Correlation
Present
Liver
Carcinogen
Kidney
Carcinogen DNT NT
Liver
Enzymes Etc….
Chemical A + - - - + - -
Chemical B + - - + - - -
Chemical C + + - - - + -
Chemical D + + - - - + -
Chemical E + - - + + - -
Chemical F - - - - - - -
28
HTS Profiles
ToxCastDB
ArrayTrackDB
Genomics Profiles
Gas Exchange
Arterial
Brain
GI Tract
Liver
Kidney
Fat
Slowly Perfused
Rapidly Perfused
Metabolism Ven
ous
Rats:
pnd10: 20 g
Adult: 200 g (60 day)
Aged: 450 g (2 yr)
Predicted Venous Concentrations in rats – 6 hr 500 ppm
Predicted Venous Concentrations in rats – 6 hr 500 ppm
Modeling uses age-appropriate physiological and chemical specific (e.g., partition
coefficients, metabolism rates) parameters
0 5 10 15 20 250
50
100
150
200
250
PND10AdultAged
time (hrs)
Ve
no
us
co
nc
en
tra
tio
n (
mg
/L)
Methyl ethyl ketone
0 5 10 15 20 250
5
10
15
20
25
30
35
PND10AdultAged
time (hrs)
Ve
no
us
co
nc
en
tra
tio
n (
mg
/L) Trichloroethylene
The Virtual Liver:A multiscale , computational model
Molecular CircuitryMolecular Circuitry
Hanahan and Weinberg, Hallmarks of Cancer, Cell, 100, 57-70, 2000.
Reduction of uncertainty in risk assessment
Reduction of uncertainty in risk assessment
(Policy-based approach)
Risk
RfC Range ofuncertainty
(Mechanism-based approach)
Information
ToxRefDB
ArrayTrack
VirtualLiver
NCCT
ToxCast
34
AcknowledgementsAcknowledgements
DSSTox Ann Richard
ToxCast and RefToxDB David Dix Keith Houck Richard Judson Matt Martin
Molecular Modeling Jim Rabinowitz Rocky Goldsmith Tom Transue
Life Stages Hugh Barton Chester Rodriquez
Virtual Liver Jerry Blancato Rory Conolly Imran Shah