Toxicogenomic Investigation into False Positive Responses in the Local Lymph Node Assay (LLNA)

Transcription

1 Toxicogenomic Investigation into False Positive Responses in the Local Lymph Node Assay (LLNA) Darrell Boverhof, Ph.D. Toxicology & Environmental Research and Consulting (TERC) The Dow Chemical Company Midland, MI 1

2 Acknowledgements DOW David Adenuga Michael Woolhiser Bhaskar Gollapudi Lindsay Sosinski Rachel Golden Study supported by: CEFIC-LRI University of Manchester, UK Ian Kimber Rebecca Dearman The Hamner Institutes for Health Sciences Russell Thomas Michael Black 2

3 Background- Allergic Contact Dermatitis GPMT (Guinea pig maximization test) LLNA Proliferation/memory T-cells 3

4 Background- Local Lymph Node Assay Assay which detects the contact sensitization potential of chemicals (sensitization phase) Dorsal surface of ears Sensitization / Lymph Node Proliferation 3 H-thymidine Incorporated into DNA of dividing cells Treat Days 1, 2, 3 Rest Days 4, 5 Inject 3 H-thymidine Day 6 5 hours Remove Lymph Nodes Day 6 Auricular Nodes Reported as a stimulation index (SI) Ratio Treatment/Vehicle Cutoff for positive response SI >= 3 Measure 3 H-thymidine Incorporation Prepare Single, Cell Suspension 4

5 The Problem of False Positive Chemistries in the LLNA

6 The Problem of False Positive Chemistries in the LLNA Can differential gene expression responses (toxicogenomics) in the lymph node be applied to distinguish true sensitizers from false positives in the LLNA? Functional insights into different mechanisms Development of molecular classifiers to distinguish between these two classes 6

7 Study Approach Critical Elements: Chemical Selection Selection of false positive chemistries # of Chemistries Dose All chemicals tested at equipotent doses in LLNA Time Examined multiple time points Comprehensive Anchor gene expression responses to traditional LLNA endpoint 2 Phases- Development and Confirmation 7

8 Study Approach- Comprehensive Phase 1 9 Sensitizers 7 False positives Functional evaluation of differential gene expression Whole genome array Molecular classifier development and assessment Phase 2 6 Sensitizers 6 False Positives Confirmation of functional gene expression responses Focused QRTPCR arrays Evaluation of classifier performance 8

9 Phase 1 Test Materials Sensitizer Class Test Material Vehicle Dose Dinitrochlorobenzene (DNCB) Acetone 0.10% Hexylcinnamic aldehyde (HCA) Acetone 25% Isoeugenol Acetone 10% Sensitizers para -phenylene diamine(ppd) Acetone 1% Hydroquinone (HQ) Acetone 0.25% Methyldibromo glutaronitrile (MDBGN) Acetone 20% Toluene diisocyanate (TDI) Acetone 0.04% Trimellitic anhydride (TMA) Acetone 0.65% Ammonium Hexachloroplatinate (AHCP) DMSO 0.70% Oleic acid DMSO 50% Maleic acid DMSO 11.50% Presumptive False Positives Sodium lauryl sulphate (SLS) DMSO 25% Tetraethylene glycol Monotetradecyl ether (TGME) Acetone 20% Polyaminofunctional siloxane Acetone 45% N-decylphenol polyethyleneglycol ether (DPP) Acetone 35% Hexadecan-1-ol Ethoxylated (EO2) C16 (HDE) Acetone 30% Equipotent doses (SI 6-9) calculated based on results from screening LLNA 9

10 Phase 1- Stimulation Index response 10

11 Phase 1- Toxicogenomic evaluation Isolate Auricular nodes Isolate RNA Test materials grouped into a 3 class coding scheme Vehicles controls Sensitizers False positives Cy3 Reverse Transcription Hybridize on array Data filtered on expression (± 1.5 FC) and 2 way ANOVA linear contrasts (FDR < 0.05) Commonly regulated genes Responses unique to Sens and FP Data Extraction and Analysis Functional evaluation- Gene Ontology 11

12 Toxicogenomic evaluation Commonly Expressed Genes Genes similarly regulated by sensitizers and false positives relative to controls Functional Categories involved in Cell Proliferation Initiation of mitosis Cell cycle regulation The metaphase checkpoint Chromosome condensation in prometaphase Spindle assembly and chromosome separation o Indicates both Sensitizers and False positives induce a LN proliferative response o Offer no ability to discriminate between sensitizers and false positives 12

13 Untreated Acetone DMSO AHCP TDI TMA DNCB HCA HQ Isoeugenol PPD MDBGN DPP HDE Siloxane TGME Maleic Oleic SLS Normalized Mean Intensity (Log 2 ) Toxicogenomic evaluation Sensitizer-Specific Genes 4 Genes differentially regulated by Sens relative to both FPs and Controls 2 0 Fxyd4 Thbs4 Cphx Lgals7 Il Vehicles Sensitizers False Positives Key functional categories 1. Positive regulation of immune system process 2. Leukocyte activation and migration 13

14 Toxicogenomic evaluation False Positive-Specific Genes Genes differentially regulated by FPs relative to both Sens and Controls Key functional categories 1. Acute inflammatory response 2. Innate defense response (Neut/Mac markers Mpo, Lcn2) 14

15 Phase 2 Test Materials Sensitizer Class Test Material Conc. DNCB 0.05% Benzoquinone (BZQ) 0.10% Sensitizers 2-hydroxyethyl acrylate (2-HEA) 20% Phenyl acetaldehyde (PA) 20% Citral 30% Propyl Gallate (PG) 3% Presumptive False Positives DPP 35% Benzalkonium Chloride (BZC) 1% Squalene 50% Methyl Oleate (MO) 50% Hexaethylene glycol monododecyl ether (HGME) 25% Linolenic Acid (LA) 50% Equipotent doses (SI 6-9) calculated based on results from screening LLNA 15

16 Acetone DNCB BZQ 2-HEA PA Citral PG DPP BZC Squalene MO HGME LA Stimulation Index Phase 2- Stimulation Index response SI=3 0 Sensitizers False Positives

17 DNCB BQ 2-HEA PA Citral Propyl galate DPP Benzalk Cl Squalene Me oleate HGME Linoleic acid Fold change to Veh Phase 2 Sensitizer-Specific Genes Cphx Il21 Lgals7 Thbs Sensitizers Key functional categories 1. Positive regulation of immune system process 2. Leukocyte activation and migration False Positives 17

18 DNCB BQ 2-HEA PA Citral Propyl galate DPP Benzalk Cl Squalene Me oleate HGME Linoleic acid Fold change to Veh Phase 2 False Positive-Specific Genes Cd5l Dao Defa-rs2 Il12b Syn Sensitizers False Positives Key functional categories 1. Acute inflammatory response 2. Innate defense response (Neut/Mac markers Mpo, Lcn2) 18

19 Development of Statistical Classifiers- Phase 1 Gene Expression Microarrays Total of 50 iterations Partition Data into 5 Sets Set 1 Set Aside Select Features Predict Hold- Out Set Build Model Averaged parameter values provide broadbased evaluation of predictive performance Accuracy Sensitivity Specificity AUC Repeat 10X Tested in a total of 84 optimized models 19

20 Development and Evaluation of Statistical Classifiers PHASE 1 30 genes PHASE 2

21 Summary/Conclusions Sensitizer- and False Positive-specific gene expression responses were identified Sens- antigen-mediated T-cell response. FPs- consistent with a pro-inflammatory response Class-specific gene expression responses were reproducible in an independent experiment Molecular classifiers were developed that had very good performance Genes that made up the classifier were consistent with those identified through the functional analysis Approach could be used as part of a WoE analysis for suspected false positives 21

22 Acknowledgements DOW David Adenuga Michael Woolhiser Bhaskar Gollapudi Lindsay Sosinski Rachel Golden Study supported by: CEFIC-LRI University of Manchester, UK Ian Kimber Rebecca Dearman The Hamner Institutes for Health Sciences Russell Thomas Michael Black 22