1. US Army Corps of Engineers BUILDING STRONG ® Systems Biology Investigation to Explore the Computational Toxicology Tool, GO-Modeler Kurt A. Gust Bindu Nanduri Arun Rawat Mitchell S. Wilbanks Michael Quinn Jr. Jeff Chen Shane Burgess Edward J. Perkins

2. Authors K.A. Gust 1, Bindu Nanduri 2, Arun Rawat 3, Mitchell S. Wilbanks 1, Michael Quinn Jr. 4, Jeff Chen 1, Shane Burgess 2, Edward J. Perkins 1 1 US Army, Engineer Research and Development Center 2 Mississippi State University 3 University of Southern Mississippi 4 US Army, Public Health Command

3. "Today’s challenge is to realize greater knowledge and understanding from the data-rich opportunities provided by modern high-throughput genomic technology." Professor Andrew Cossins, Consortium for Post-Genome Science, Chairman.

4. What is the Gene Ontology? “a controlled vocabulary that can be applied to all organisms even as knowledge of gene and protein roles in cells is accumulating and changing”  the de facto standard for functional annotation  assign functions to gene products at different levels, depending on how much is known about a gene product  is used for a diverse range of species  structured to be queried at different levels, eg: ► find all the chicken gene products in the genome that are involved in signal transduction ► zoom in on all the receptor tyrosine kinases  human readable GO function has a digital tag to allow computational analysis of large datasets COMPUTATIONALLY-AMENIABLE ENCYCLOPEDIA OF GENE FUNCTIONS

5. Use GO for…….  Determining which classes of gene products are over-represented or under-represented.  Grouping gene products by biological function.  Relating a protein’s location to its function.  Focusing on particular biological pathways and functions  Hypothesis-testing: GO Modeler.

6. “GO-slim” In contrast, GO-Modeler uses the deep granular information rich data suitable for hypothesis-testing Many people use “GO Slims” which capture only high-level terms which are more often then not extremely poorly informative and not suitable for hypothesis-testing.

7. Step I. GO-based Phenotype Scoring. Gene productTh1Th2TregInflammation IL-21.58 -1.58 IL-40.00 IL-60.00-1.201.20-1.20 IL-80.00 1.18 IL-100.00 IL-120.00 IL-131.51-1.510.00 IL-180.91 IFN-  0.00 TGF-  -1.710.001.71-1.71 CTLA-4-1.89 1.89-1.89 GPR-83-1.69 1.69-1.69 SMAD-70.00 Net Effect-1.29-5.3810.15-5.98 Step III. Inclusion of quantitative data to the phenotype scoring table and calculation of net affect. Underlying GO-Modeler 111SMAD-7 1 GPR-83 1 CTLA-4 10 TGF-  111 IFN-  1111IL-18 ND 1IL-13 ND 1IL-12 011IL-10 11ND IL-8 11IL-6 ND11IL-4 1 ND 1IL-2 InflammationTregTh2Th1Gene product ND = No data Step II. Multiply by quantitative data for each gene product.

8. GO Modeler http://agbase.msstate.edu/cgi-bin/tools/GOModeler.pl

9. Problem Identification Daphnia magna Northern bobwhite Earthworm Fathead minnow Rat Animals may be exposed to soils, water and/or food contaminated with energetic compounds on Army ranges. Song Birds Western Fence Lizard Contamination

10. Problem Identification Toxicology (Parent Compound & Metabolites):  Aberrant Neuromuscular Effects  Anemia and various impacts on Blood Chemistry  Gastrointestinal Impacts / General Impacts on Viscera.  Mortality at High Doses  Result: Increased regulatory concern over RDX, TNT & their breakdown products. 2,4,6-trinitrotoluene (TNT)2-Amino-4,6-dinitrotoluene (2A-DNT)

11. Systems Biol. - Transcriptomics + Proteomics Proteomics Transcriptomics Biological Networks ? Northern Bobwhite 2A-DNT 2A-DNT Exposures with Northern Bobwhite Sub-Acute Exposure – Birds dosed with 2A-DNT for 14d at 0, 50, 125, 225, 550, or 1000 mg/kg/day via oral gavage. Sub-Chronic Exposure -Birds dosed with 2A-DNT for 60d at 0, 0.5, 3, 14 or 30 mg/kg/day via oral gavage, 12 biol reps for each sex. Transcriptomics and Proteomics – Leveraging Sub-Chronic Exp. Proteomics investigated in Liver and Kidney tissues. Four biological replicates were investigated for the 0 and 30 mg/kg/d treatments. Liver tissue was investigated in males and females, and kidney tissue was examined in males only. Transcriptomics investigated in Liver and Kidney tissues. Four biological replicates for both males and females were examined for the 0, 3, 14 and 30 mg/kg/d treatments for each tissue.

12. 2A-DNT Toxicology Figure 1. The number of days survived by northern bobwhite (Colinus virginianus) exposed daily to oral gavages of 2A-DNT (2-amino-2,6-dinitrotoluene; mg/kg-d) for a total of 14 days. Quinn et al 2010, Ecotoxicology Sub-Acute 14d Exposure

13. 2A-DNT Toxicology Sub-Chronic 60d Exposure Increased Liver weights (Brain-normalized) at the highest 2A-DNT dose (30 mg/kg/d) in both males and females. significant reduction in white blood cell counts at the 30 mg/kg/d dose in females Out of 15 blood chemistry investigations: Alanine aminotransferase (ALT) significantly decreased and Triglycerides (TRIG) were significantly increased respectively at intermediate 2A-DNT concentrations (0.5 – 3 mg/kg/d) in males only. Quinn et al 2010, Ecotoxicology

14. Hypotheses Based on Phenotypes  1. Daily oral dosing of 2A-DNT had no effect on genes and molecular pathways involved in Lipid metabolism in liver tissues of Northern bobwhite.  2. Daily oral dosing of 2A-DNT had no effect on peroxisome proliferator- activated receptor (PPAR)-controlled pathways in Northern bobwhite.  3. Daily oral dosing of 2A-DNT had no effect on genes and molecular pathways involved in energy metabolism in Northern bobwhite.  4. Daily oral dosing of 2A-DNT had no effect on genes and molecular pathways involved in immune function in Northern bobwhite.  5. Daily oral dosing of 2A-DNT had no effect on genes and molecular pathways involved in xenobiotic metabolism in Northern bobwhite.

15. GO Modeler – Hypothesis Statements 1. lipid metabolism GO:0006629 lipid metabolic process 2. PPAR controlled pathways GO:0006629 lipid metabolic process (includes GO:0006631 fatty acid metabolic process) GO:0006699 bile acid biosynthetic process GO:0046950 cellular ketone body metabolic process GO:0045444 fat cell differentiation, GO:0002024 diet induced thermogenesis AND/OR GO:0050873 brown fat cell differentiation, GO:0060548 negative regulation of cell death AND/OR GO:0019725 cellular homeostasis, GO:0016567 protein ubiquitination AND/OR GO:0016574 histone ubiquitination, GO:0006094 gluconeogenesis 3. energy metabolism GO:0015975 energy derivation by oxidation of reduced inorganic compounds GO:0015980 energy derivation by oxidation of organic compounds GO:0006119 oxidative phosphorylation 4. immune function GO:0006955 immune response 5. xenobiotic metabolism GO:0006805 xenobiotic metabolic process GO:0009410 response to xenobiotic stimulus 6. liver weight GO:0035265 organ growth GO:0031100 organ regeneration 7. alanine transferase GO:0004021 L-alanine:2-oxoglutarate aminotransferase activity GO:0047810 D-alanine:2-oxoglutarate aminotransferase activity 8. triglyceride GO:0006641 triglyceride metabolic process GO:0034197 triglyceride transport

16. 2nd Generation Multi-tissue Microarray Agilent G2 one-color platform 8 x 15K spot, high-density oligonucleotide Source cDNA library developed using Next Gen Seq Fully Annotated, Open Source Knowledgebase www.quailgenomics.info www.quailgenomics.info Northern Bobwhite Genome Tools Rawat et al 2010 BMC Bioinformatics

17. 2A-DNT Transcriptomics Results Overview Total Differentially Expressed Transcripts (DET) Even Distribution of Increased and Decreased Expr.

18. 2A-DNT Transcriptomics Results Overview L = Liver Tissue 3 = 3 mg/kg/d 14 = 14 mg/kg/d 30 = 30 mg/kg/d 132269 179 345278 270 Total Differentially Expressed Transcripts (DET) Commonality in DET among Doses

19. 2A-DNT Transcriptomics Results Overview L = Liver Tissue 3 = 3 mg/kg/d 14 = 14 mg/kg/d 30 = 30 mg/kg/d Commonality in DET among Sexes was Limited

20. Methods:  Comparative shotgun proteomics  Male and Female liver tissue of Northern bobwhite, 2-ADNT 30mg/kg/d vs Controls.  Pressure cycling technology sample preparation & trypsin digested proteins  Analysis: 1 dimensional liquid chromatography nano-spray tandem mass spectrometry 2A-DNT Proteomics

21. Results Overview:  2,672 proteins identified  Total Differentially Expressed: 2A-DNT Proteomics

22. Transcriptiomics vs Proteomics Target by Target Comparison Limited Overlap of differentially expressed targets. HOWEVER, Syntax differences among annotations still need to be addressed. Commonality is likely UNDER-Represented.

23. GO-Modeler Hypothesis Tests “( )” Reject Null Hypothesis Expression in liver was sex specific Lipid metabolism impacted, but not strongly at transcript level. PPAR controlled pathways were impacted by 2A-DNT Exposure. Transcriptomics Data Female, Liver 30 mg/kg/day Male, Liver 30 mg/kg/day

24. GO-Modeler Hypothesis Tests PPAR controlled pathways Expression in liver was largely sex specific. Marginal Impacts on xenobiotic metabolism and Immune response – Do not Reject Null Hypothesis Significant Impacts: Energy Metabolism, PPAR controlled pathways and Lipid Metabolism – Reject Null Hypothesis Some Parallelism with Transcriptomics Results Proteomics Data

25. Related Work Investigating 2,6-DNT Impacts on Lipid Metabolism Decreased Expression of PPAR Signaling Pathway Decreased Lipid Metabolism  Lipid Inundation in Liver Wintz et al 2006 Toxicol Sci Rawat et al 2010 Physiol Genomics

26. 2A-DNT Transcriptomics Ingenuity Pathway Analysis Many of Top 5 Networks involved in Lipid Metabolism Results Similar to impacts of 2,6-DNT (Rawat et al 2010, Physiol Genomics)

27. 2A-DNT Proteomics Ingenuity Pathway Analysis Many of Top 5 Networks involved in Lipid Metabolism Parallels Results of Transcriptomics Analysis

28. Top Network: Male, Liver 30 mg/kg/day 2A-DNT Transcriptomics

29. Top Network: Female, Liver 30 mg/kg/day 2A-DNT Transcriptomics

30. Summary and Conclusions GO Modeler provides a novel a priori hypothesis testing mechanism utilizing functional annotation. GO Modeler successfully identified impacts related toxicological phenotypes. Provides function-based down-selection of targets focusing on most relevant bio-molecules. Results parallel enrichment identified by Ingenuity Pathway Analysis. Improvements Needed: More Objective Hypothesis Test, Increased Automation, Robust Validation. http://agbase.msstate.edu/cgi-bin/tools/GOModeler.pl