Stress Response Pathway Ensemble: A New Paradigm in High Throughput Toxicity Screening

1. Steve Simmons US EPA Office of Research and Development RTP, NC Stress Response Pathway Ensemble: A New Paradigm in High Throughput Toxicity Screening The McKim Conferences September 2008

2. Outline • The Challenge Before Us • Stress Response Pathways as Toxicity Pathways • Stress Response Assays • Implementation to HTS • What does this have to with QSAR?

3. Problem/Challenge • Large number of environmental compounds that currently need characterization and prioritization for further screening • Limited testing resources • Current approaches are slow, laborious, and expensive • Ethical need to reduce animal use in toxicity testing • In vitro –omics approaches too expensive for screening applications and data interpretation problematic • Multiple alternative approaches needed, assay cost to be minimized

4. Rapid, inexpensive, reproducible and predictive assays that allow for effective screening of large numbers of compounds to enable prioritization for further characterization In vitro assays amenable to high-throughput screening, preferably using human cells and tissues that are significantly more economical and practical than –omic methods and traditional in vivo testing methods Solution

5. Toxicity Pathways Chemical Characterization Risk Contexts Toxicity Testing Targeted Testing Toxicity Pathways Population and Exposure Data Dose-Response and Extrapolation Modeling Toxicity Pathways Targeted Testing • Testing conducted to evaluate metabolites, assess target tissues, and develop understanding of affected cellular processes at genomics level • Limited types and duration of in vivo studies, focusing on up to 14-day exposures • More extensive testing for representative compounds in novel chemical classes • Evaluation of perturbations in toxicity pathways rather than apical endpoints • Emphasis on high-throughput approaches using cell lines, preferably of human origin • Use of medium-throughput assays of more integrated cellular responses • Poor definition • Number unknown • Extended research effort Adapted from Toxicity testing in the twenty-first century: a vision and strategy: NRC, July 2007

6. Toxicity Pathways • Are all cellular pathways potential toxicity pathways? • If so, are all pathways created equal, or are some more important than others? • How do you determine importance/priority? • Cover the most chemical space? • Easiest to model?

7. Stress Responses Sense Perturbations Exposure Tissue Dose Biologic Interaction Perturbation Adapted from: Toxicity Testing in the Twenty-first Century: A Vision and a Strategy, National Research Council. 2007. Normal Biologic Function Biologic Inputs Reversible Early Cellular Changes Irreversible Cell Injury Adaptive Stress Response Morbidity and Mortalilty 6

8. Protective signaling pathways activated in response to environmental insults such as chemical toxicity Present in all cells and highly conserved Broad indicators of cellular toxicity Triggered at low doses before more apical effects such as cell death or apoptosis A few (<10) key cellular stress pathways identified Pathways well-characterized and classified mechanistically mode of action info?? Adaptive Stress-Response Pathways

9. Adaptive Stress Response Pathways Oxidative stress DNA damage Heat shock ER stress Hypoxia/Anoxia Inflammation Heavy metal stress Osmotic stress* 8

10. Well-definedMechanisms : MOA information? Oxidative Stress Genotoxic Stress Heat Shock ER Stress Hypoxia Inflammation Metal Response

11. Stress Pathway Architecture Stress Transducers Sensor TF Target Genes Nucleus

12. Example:Nrf2-mediated Oxidative Stress Response Oxidative Stress PKC ERK2 JNK1 p38 Keap1 Nrf2 Hmox1 NQO1 GSTA2 Nucleus

13. Pathway inducer Sensor Tr. Fact. Sens. K/O TF K/O Dbl. K/O Ox stress Oxygen radicals Keap-1 Nrf-2 Emb. leth Viable* viable Genotoxic DNA damage Mdm2 p53 Emb. leth Viable* viable Hypoxia Oxygen deprivation VHL HIF-1 Emb. leth Emb. leth ??? Heat shock Protein denatn. Hsp-90 HSF-1 HSF-3 Emb. leth Viable* ♀ infertile ??? Inflammat. TNF, LPS IkB NFkB Emb. leth Viable* ??? Metals Heavy metals (none) Zn activat. MTF-1 NA Emb. leth ??? Biological Centrality of Stress Responses: * Disease susceptibility and normal functions compromised

14. Stress-Response Pathways • Stress pathways share a common pattern of organization • Stimulation of a stress pathway results in activated transcription factor • The transcription factor serve as a nodal point for multiple “toxicity pathways” • Activated transcription factor up-regulates unique target genes Regulatory elements of target genes can be used to measure pathway activation!

15. Genomic vs. Synthetic Promoters sensitivity vs. specificity Heme Oxygenase-1 Nrf2 AP-1 AP-1 Maf AP-2 -12kb Sp1 CREB CEBPa MTF-1 Nrf2 Sp1 NF-kB AP-1 MTF-1 +1 NF-kB Maf CREB Sp1 Distal Enhancer 2 Distal Enhancer 1 Proximal Enhancer ProximalPromoter Nrf2 Nrf2 Nrf2 ARE ARE ARE Basal Promoter Nrf2 Synthetic MultimerizedResponseElements

16. Constructing Stress-Responsive Reporter Genes Stress Promoters Reporter ORF

17. Why Reporter Genes? • Why not measure stress protein levels? • Stability- rapid turnover • Throughput • Why not measure transcripts by microarray or qPCR? • Cost • Throughput • Remember… we need to assay 1000s of chemicals and establish dose-response information • Reporter genes meet all of the HTS requisites at the right price

18. Miniaturization: 96-, 384-, 1536-well formats minimizes compound requirements and waste lowers screening cost for consumables dose-response and time course data Assay Performance Signal-to-background Coefficient of variance ↑ Signal/Background, ↓ CV = ↑ Z’ score Reagent availability, cost, compatibility w/ library Relevance What Makes a Good HTS Assay?

19. Typical Assay Conditions • Stable cells seeded overnight in multi-well assay plates • Cell treated with compounds for pre-determined time • Luciferase activity normalized to GFP viability per well Inactivation Activation

20. Stress Signatures

22. Throwing Off Pharmaco-philosophy • Disparity between the mandates of pharma industry and regulatory toxicology • Acceptance of false positive/negatives • Known targets vs. unknown mechanisms • $$$ • Much of the efforts to-date to implement HTS tox testing has adopted pharma tools and pharma thinking • Blunt tests for cytotoxicity vs. sensitive assays for “drug-able” targets • Single dose testing (usually determined by solubility) • Overly-conservative criteria for calling “hits”

23. Throwing Off Pharmaco-philosophy • What is needed: • Reduced reliance on loss-of-signal assays • Better understanding of mechanisms/pathways • Sensitive assays to measure mechanistic endpoints • Chemical libraries at higher concentrations • Dose-response information • Establish criteria for determining “actives” • $$$

24. Chemical Library Construction • Chemicals are assembled in groups of several hundreds • Diverse chemical properties including solubility • Pharma libraries use constrained property ranges • Molecular weight • log P • DMSO is currently the solvent of choice • Compound with lowest solubility determines solubility limit for the entire library • This increases chances for false negatives

25. True Negative or NOEL?

26. Abandoning Single-dose Screening • Quantitative HTS (qHTS) employs chemical libraries in 1536-well plate format • Each library constrained to a single 1536-well plate; library is titrated across multiple plates 92uM Titrated positive control Fixed positive control Vehicle control Additional positive control Compound Area 1.2nM

27. Using HTS Assays in Mechanistic Research Caspase 3/7 assay (Gain of signal) EC50 IC50 Cell TiterGlo (ATP) (Loss of signal) Huang et al. Chem. Res. Toxicol. 2008, 21, 659–667

28. Hierarchal Clustering NTP 1408 Chemicals 23 Clusters Estrogenics Antineoplastics Hormone Antagonists Cytostatics Huang et al. Chem. Res. Toxicol. 2008, 21, 659–667

29. “While the goal of the clustering is to generate a hypothesis about a compound’s specific mechanism of action, the broad nature of these cytotoxicity assays likely prevents any detailed understanding of the molecular basis of the toxic effect; the inclusion of or confirmation of activity in other, more mechanistic assays would obviously improve this aspect of the current study.” Stress pathway assays move us a step in this direction Huang et al. Chem. Res. Toxicol. 2008, 21, 659–667

30. Progress to Date Engineered 20+ assays covering most of the key stress pathways; filling in gaps with new assays Miniaturizing assay to 1536-well format Screened two assays (Nrf2 and hsp70) with two libraries: NTP and EPA; preliminary re-clustering Screening additional assays in phased manner New chemical libraries NTP-B 1408: Summer 2009 EPA-B 1408: Summer 2009; EPA-C 1408: Under consideration Adding primary cells models: human hepatocytes, rodent renal proximal tubule cells, etc. 29

31. Beyond In Vitro control 0.2uM 5uM Blechinger SR, Warren JT Jr, Kuwada JY, Krone PH. Developmental toxicology of cadmium in living embryos of a stable transgenic zebrafish line. Environ Health Perspect. 2002 Oct;110(10):1041-6. 125uM

32. In Vitro Alt. Species QSAR Screening and Prioritization Enhanced Predictivity

33. Summary and Conclusions Stress pathway assay ensemble to generate “stress signatures” Clustering by biological response in in vitro assays: structural similarities??? Can this type of information improve QSAR models? Current HTS assays measure typically blunt responses: mechanisms will need further delineation As we move forward with HTS testing, we need to move from the pharma approach to maximize information gains that will useful for toxicology 32

34. Acknowledgements US EPA Neurotoxicology Division Ram Ramabhadran Chun-Yang Fan Jeanene Olin Theresa Freudenrich Helen Carlsen NIH Chemical Genomic Center Chris Austin Jim Inglese Menghang Xia Ruili Huang Sunita Shukla Open Biosystems (Thermo-Fisher) John Wakefield Attila Seyhan The Hamner Institutes Rusty Thomas US EPA, National Center for Computational Toxicology Keith Houck David Dix