Systems Biology in Small Fish Models D.L. Villeneuve The McKim Conference on Predictive Toxicology Sept , 2007, Duluth, Minnesota

Systems Biology in Small Fish Models 1.What is systems biology or the “systems perspective”? 2.The systems perspective in ecotoxicology. 3.An overview of a systems-oriented small fish research program. 4.A case study 5.Conclusions

The Systems Perspective The experimenter can draw a boundary around it, heuristically The experimenter can conduct defined perturbations within the boundaries The experimenter can, by reasoning generate sensible explanations for the changed behavior. Learning is assessed by how well the understanding enables prediction of changed system behavior in response to defined perturbations. A subset of the world whose behavior, and whose interaction with the world, we believe can be sensibly described. What is a system? Brent 2004, Nature Biotech. 22: Kuipers, B MIT Press, Cambridge

Systems biology is the study of an organism, viewed as an integrated and interacting network of genes, proteins and biochemical reactions which give rise to life. Systems are comprised of parts which interact. Interaction of these parts gives rise to "emergent properties". Emergent properties cannot be attributed to any single parts of the system. Irreducible. To understand systems, and to be able to fully understand a system's emergent properties, systems need be studied as a whole. The Systems Perspective

Non-equilibrium thermodynamics (1930s-40s) dealt with integration quantitatively aimed to discover general principles >> descriptive established connection to molecular mechanisms Investigation of biological self-organization (1950s) examination of how structures, oscillations, and/or waves arise in a steady or homogenous environment Feedback regulation in metabolism (late 1950s) Systems theory in biology (1960s) “search for general biological laws governing behavior and evolution of living matter in a way analogous to the relation of physical laws and non-living matter” Metabolic control analysis (1970s) Approaches to characterize properties of networks of interacting chemical reactions Convergence with “mainstream” molecular biology – high throughput, genome-scale, “data-rich” Westerhoff and Palsson, 2004 Nature Biotech. 22: Wolkenhauer Briefings in Bioinformatics. 2:

The Systems Perspective Shift from reductionism to a holistic perspective Should reduce complexity rather than adding additional layers of complexity Search for organizing principles over construction of predictive descriptions (models) that exactly describe the evolution of a system in space and time. Identification of new concepts and hypotheses that provide a conceptual structure with logical coherence to rival chemistry and physics. Mesarovic et al Syst. Biol. 1:19-27

The Systems Perspective in Ecotoxicology Shift from reductionism to a holistic perspective Historically, ecotoxicology and ecological risk assessment was holistic in focus Apical/integrative endpoints: Survival, growth, reproduction Not feasible to test every chemical, let alone every chemical mixture Increased emphasis on subtle, chronic impacts (e.g., development, behavior, etc.) with long-term population implications. Reductionist in the sense that testing of the universe of chemicals was the paradigm.

Search for organizing principles that underlie biological response to stressors. Develop a conceptual structure with logical coherence that allows us to predict, with reasonable and quantitative certainty, integrated, ecologically-relevant impacts. The Systems Perspective in Ecotoxicology

USEPA – Cincinnati, OH D. Bencic, M. Kostich, I. Knoebl, D. Lattier, J. Lazorchak, G. Toth, R. Wang, USEPA – Duluth, MN, and Grosse Isle, MI G. Ankley, E Durhan, M Kahl, K Jensen, E Makynen, D. Martinovic, D. Miller, D. Villeneuve, USEPA – Athens, GA T. Collette, D. Ekman, J. Kenneke, T. Whitehead, Q. Teng USEPA-RTP, NC M. Breen, R. Conolly USEPA STAR Program N. Denslow (Univ. of Florida), E. Orlando, (Florida Atlantic University), K. Watanabe (Oregon Health Sciences Univ.), M. Sepulveda (Purdue Univ.) USACE – Vicksburg, MS E. Perkins Other partners Joint Genome Institute, DOE (Walnut Creek, CA) Sandia, DOE (Albuquerque, NM) Pacific Northwest National Laboratory (Richland, WA) C. Tyler (Univ. Exeter, UK) Linkage of Exposure and Effects Using Genomics, Proteomics, and Metabolomics in Small Fish Models

11βHSD P45011β. GnRH Neuronal System GABADopamine Brain Gonadotroph Pituitary D1 R D2 R ? ? Y 2 R NPY Y 1 R GnRH R ? Activi n Inhibi n GAB A A R GAB A B R Y 2 R D2 R GnR H Circulating Sex Steroids / Steroid Hormone Binding Globlulin PACAP Follistatin Activi n PAC 1 R Activin R FSH  LH  GP  Circulating LH, FSH FSH RLH R Circulating LDL, HDL LDL R HDL R Outer mitochondrial membrane Inner mitochondrial membrane StAR P450scc pregnenolone 3  HSD P450c1 7 androstenedion e testosterone Cholestero l estradio l Gonad (Generalized, gonadal, steroidogenic cell) Androgen / Estrogen Responsive Tissues (e.g. liver, fatpad, gonads) ARER Blood Compartment 17βHSD P450arom 11-ketotestosterone progesterone17 α -hydroxyprogesterone 17α,20β-P (MIS) 20βHS D Fipronil* (-) Muscimol* (+) Apomorphine (+) Haloperidol (-) Trilostane (-) Ketoconazole (-) Fadrozole (-) Prochloraz (-,-) Vinclozolin (-) Flutamide (-) β-trebolone (+) Ethynyl estradiol (+) Chemical “Probes”

Figure 1. Conceptual Overview of Research Increasing Ecological RelevanceIncreasing Diagnostic (Screening) Utility Phase 1. Fathead minnow 21 d reproduction test Phase 2. Zebrafish genomics proteomics, metabalomics Population modeling Phase 3. Fathead minnow molecular markers metabolomics Systems modeling Molecular mRNA, protein, Enzyme assays Cellular Metabolite profiles Organ Functional and Structural change (Pathology) Individual Altered reproduction or development Population Decreased numbers of animals Levels of Biological Organization Well characterized genome, low eco / regulatory relevance Computational modeling Poorly characterized genome high eco / regulatory relevance →’s Depict the flow of information

A Case Study withFadrozole A Case Study with Fadrozole Aromatase

Ovary Brain

Pregnenolone Progesterone 17  -OH-Pregnenolone 17  -OH-Progesterone CYP17 (hydroxylase) CYP11A 3  -HSD DHEA Androstenedione Testosterone CYP17 (lyase) 3  -HSD 17  -HSD CYP19 17  -estradiol Cholesterol 20  -HSD 17α20β-dihydroxy-4- pregnen-3-one CYP17 (hydroxylase) CYP17 (lyase) estrone CYP19 17  -HSD 11β-OH- Testosterone CYP11B1 11-Ketotestosterone 11βHSD 11-deoxycorticosterone CYP21 corticosterone CYP11B1 aldosterone CYP11B2 11-deoxycortisol CYP21 cortisol CYP11B2 A Case Study with Fadrozole

25 o C, 12 h RIA Androstenedione Testosterone 17  -HSD CYP19 17  -estradiol estrone CYP19 17  -HSD

Brain sGnRH neurons cGnRH neurons dopaminergic neurons sGnRH releasecGnRH releasedopamine release GnRH Pituitary FSH gonadotrophs LH gonadotrophs TestisOvaries Leydig cells Sertoli cells Sperm- atocytes Theca cells Granulosa cells Oocytes FSHLH Liver Sperm mat.oocyte mat. steroids Steroid metab.vitellogenesis Blood supply

transcriptional activation translational activation transcription inhibition dissociation genes mRNA protein receptor pheno- type state transition activated protein association Figure Key GnRH release Gonadotropin expression and secretion Cholesterol transport Steroidogenesis Vitellogenesis and Oocyte maturation Steroid metabolism

Inner mitochondrial membrane Fadrozole transcriptional activation translational activation transcription inhibition dissociation genes mRNA protein receptor pheno- type state transition activated protein association Figure Key

Evidence for compensatory/feedback response to fadrozole. exposure post-exposure

Aromatase A (CYP19A) gene expression exposurepost-exposure

Fig. 1 Pregnenolone Progesterone 17  -OH-Pregnenolone 17  -OH-Progesterone CYP17 (hydroxylase) CYP11A 3  -HSD DHEA Androstenedione Testosterone CYP17 (lyase) 3  -HSD 17  -HSD CYP19 17  -estradiol Cholesterol 20  -HSD 17α20β-dihydroxy-4- pregnen-3-one CYP17 (hydroxylase) CYP17 (lyase) estrone CYP19 17  -HSD 11β-OH- Testosterone CYP11B1 11-Ketotestosterone 11βHSD Cholesterol StAR Cyt b5 LHR FSHR G-protein mediated Signal transduction Pituitary LHβ FSHβ Inner mitochondrial membrane

P450scc (CYP11A) gene expression exposurepost-exposure

exposurepost-exposure FSHR gene expression

Despite evidence for compensation at the molecular and biochemical level, 21 d exposure to fadrozole causes adverse apical effect on reproduction. Under what conditions would compensation be successful? Under what conditions does it fail to prevent adverse effect? Under what conditions does it contribute to or exacerbate the initial impact of the stressor?

A Case Study withFadrozole A Case Study with Fadrozole Supervised – low content analyses Unsupervised – high content analyses

High desnity oligonucleotide microarrays Gonads Brain Liver Fadrozole High-content analyses

Enriched Gene Ontology Categories Brain Cholesterol biosynthesis Cholesterol transport Bile acid synthesis Cytoskeletal components Oxidative phosphorylation Coagulation/wound response Immune response Ion transport Cell adhesion Reproduction Nucleotide biosynthesis Liver Ribosomal proteins Ribosomal RNAs Translation Oocytes/oogenesis Iron transport/plasma proteins Immune response and inflammation Cytoskeletal components Ovary Ribosomal proteins Ribosomal RNAs Extracellular matrix Connective tissue Coagulation/wound response Cytoskeletal components Oxidative phosphorylation Ion transport Embryonic development Cell adhesion Oogenesis

Different reverse engineering algorithms. Dynamic Bayesian Network (Yu et al 2004, Zhao et al 2006 ) Boolean (Hashimoto et al 2004) Information theory (Margolin et al 2006) Reverse engineering gene regulatory architecture

Network Approach Start with complex system Determine players and links in network Calculate network statistics Use network statistics to make predictions about network function Robustness and fragility in ecosystems and gene networks is related to network architecture (sensitivity or insensitivity to network perturbation) In general, more interactions = more plasticity Biological systems – generally many nodes with relatively few links, and relatively few nodes with many links (critical regulatory nodes) Csete & Doyle 2004 Trends in Biotechnol. 22: Zhao et al BMC Bioinformatics 7:386 Csete & Doyle Science. 295:

Consideration of interactions and relationships is at the heart of the systems perspective. Systems models serve as a critical translator between initiating events (modeled by QSAR) and biological outcome. Systems models need not be infinitely detailed. Resolution required defined by the questions to be answered. Toxicity pathways linking exposure to adverse effect include the direct effect of the chemical and a wide variety of indirect secondary, tertiary….and potentially stochastic effects that influence the apical outcome. Molecular and biochemical responses to stressors are both dose and time-dependent, and best represented in three dimensions. Important consideration for biomarker-based bioassay and high-throughput screening. CONCLUSIONS

Modern “omics” tools facilitate unprecedented scale and detail in the descriptive analysis of biological systems. The greater challenge is in defining relationships and the general principles that govern them. Overall architecture of the systems or networks can reveal important attributes related to function. Hypothesis driven and unsupervised analysis of biological systems or networks can reveal critical regulatory nodes that integrate signals and drive component function or phenotype. The ultimate goal of systems ecotoxicology is discovery of generalized or universal principles that govern biological responses to stressors. CONCLUSIONS