MECHANISMS OF GENETIC RESISTANCE TO DIOXIN-INDUCED LETHALITY

by

Ivy D. Moffat

A thesis submitted in conformity with the requirements

for the degree of Doctor of Philosophy

Graduate Department of Pharmacology and Toxicology

University of Toronto

© Copyright by Ivy D. Moffat, 2008 Mechanisms of genetic resistance to dioxin induced lethality Ivy D. Moffat Doctor of Philosophy, 2008 Graduate Department of Pharmacology University of Toronto

THESIS ABSTRACT

Dioxins are environmental contaminants that raise concern because they are potent and persistent. The most potent dioxin congener, 2,3,7,8-tetrachlorodibenzo-p-dioxin

(TCDD), causes a wide variety of biochemical and toxic effects in laboratory animals and in humans. Major toxicities of TCDD are initiated by their binding to the AH receptor

(AHR), a ligand-activated transcription factor that regulates expression of numerous genes. However, the specific genes whose dysregulation leads to major toxicities such as wasting, hepatotoxicity, and lethality are unknown. The objective of this thesis research was to identify the molecular mechanisms by which dioxins cause lethality. To this end, a powerful genetic rat model was utilized – the Han/Wistar (Kuopio) rat which is highly resistant to dioxin toxicity due to a major deletion in the AHR’s transactivation domain

(TAD) leading to 3 potential AHR variant transcripts. We found that insertion-variant transcripts (IVs) are the dominant forms of AHR expressed in H/W rats, constitutively and after TCDD treatment. Gene expression array analysis revealed that the total number of TCDD-responsive genes in liver was significantly lower in H/W rats (that carry the

TAD deletion) than in dioxin-sensitive rats (that carry wildtype AHR). Genes that are well-known to be AHR-regulated and dioxin-inducible − such as CYP1 transcripts − remained responsive to TCDD in H/W rats; thus the TAD deletion selectively interferes with expression of a subset of hepatic genes rather than abolishing global AHR-mediated

ii responses. Genes that differed in response to TCDD between dioxin-sensitive rats and dioxin-resistant rats are integral parts of pathways known to be disrupted by dioxin treatment such as protein synthesis/degradation, fatty acid transport/metabolism, and apoptosis. These genes are worthy candidates for further mechanistic studies to test their role in major dioxin toxicities. Numerous differentially-regulated genes were downregulated; however, microRNAs, which downregulate mRNA levels in other systems, likely play no role in downregulation of mRNAs by dioxins in adult liver and are unlikely to be involved in hepatotoxicity. Findings in this research support the hypothesis that H/W rats are resistant to TCDD lethality because the TAD deletion prevents the AHR from dysregulating specific mRNA transcripts but not hepatic miRNAs.

iii ACKNOWLEDGEMENTS

The foundation and support for these research projects was laid by many amazing people to whom I owe a world of gratitude. A very special “thank you” to:

[ Dr. Allan Okey, there are not enough words to express my gratitude. Though opportunity, advice, and independence you have trained many high caliber scientists; I am privilege to be among these fortunate scientists. [ Raimo Pohjanvirta for a highly productive collaboration, exhaustive knowledge of molecular and biochemical physiology, and fruitful scientific discussions. [ PhD advisory committee: Dr. Riddick, Dr. Harper, and Dr. Grant for their genuine interest and guidance since the beginning of this research. Thank you for challenging this research and polishing this scientist. [ Amazing friends who are more like family. Michael Dorr, Lucy Reed, Howard Kim, Heather Runions, Jenny & Katherine Clark, Martha Weber, and Jane Macaulay thank you for sharing in my best times and making me laugh in the worst. [ Paul Boutros, a valued colleague, for numerous discussions and statistical advice. [ Trine Celius for valuable conversations both scientific and other, you made the thesis writing experience more enjoyable. [ The chocolate manufacturing industry and to Paracelsus who stated “it is the dose that makes the poison”, a concept to guide research and life. [ The Canadian Institutes of Health Research, the Academy of Finland, the Natural Sciences & Engineering Council, and the University of Toronto for financial support of this research.

While this research focused on the genetics of the Han/Wistar (Kuopio) rat, it was powered by the genetics of my parents. In particular, the ‘curiosity’ gene and the ‘persistence’ gene to which I owe my father a great deal. I am also grateful to all my family members, who allowed me the independence to follow my dreams in the ‘big city’, but were always there when times got tough.

iv TABLE OF CONTENTS

THESIS ABSTRACT ...... ii ACKNOWLEDGEMENTS ...... iv TABLE OF CONTENTS ...... v LIST OF PUBLICATIONS ...... viii ABBREVIATIONS...... ix LIST OF TABLES ...... xi LIST OF FIGURES ...... xii CHAPTER 1: General Introduction ...... 1 Dioxins and dioxin-like compounds ...... 1 Effects of dioxins in humans...... 3 Acute effects of dioxins in laboratory animals ...... 4 Adaptive response...... 4 Wasting syndrome ...... 5 Hepatotoxicity...... 5 Tumor promotion...... 6 Acute lethality...... 6 The AH Receptor: Transcription factor and essential mediator of dioxins toxicities..... 7 Repertoire of ligands which bind to the AHR ...... 13 Structure of the AHR...... 14 The AHR mediates dioxin toxicities...... 16 Expanding the AHR-mediated transcriptome in relation to TCDD lethality and associated toxicities ...... 30 Potential model systems to identify genes relevant to dioxin lethality and associated toxicities...... 20 Humans...... 20 Animal models...... 21 The TCDD-resistant H/W rat model...... 23 AIMS OF MY THESIS RESEARCH ...... 28 SPECIFIC AIMS OF EACH THESIS PROJECT ...... 29

CHAPTER 2: Aryl Hydrocarbon Receptor (AHR) Splice Variants in the Dioxin- Resistant Rat: Tissue Expression and Transactivational Activity .. 36 ABSTRACT...... 37 INTRODUCTION ...... 38 MATERIALS & METHODS ...... 41 RESULTS ...... 48

v Insertion variants were the predominant AHR splice-variant transcripts constitutively expressed in dioxin-resistant rats...... 48 Dioxin treatment increases expression of AHR splice-variant transcripts...... 49 Functional differences in intrinsic transactivation activity among the splice variants ...... 51 AHR structures predicted in silico: comparison between species and within species...... 54 DISCUSSION...... 59 Constitutive expression of splice variants ...... 59 TCDD effect on expression of splice variants ...... 60 TAD structure in relation to AHR function...... 61 SIGNIFICANCE AND IMPACT OF THIS PROJECT ...... 63

CHAPTER 3: Dioxin Lethality: Aryl Hydrocarbon Receptor (AHR)-Mediated Gene Expression in a Resistant Rat Model...... 64 ABSTRACT...... 65 INTRODUCTION ...... 66 MATERIALS AND METHODS...... 68 RESULTS ...... 75 Global differences in gene expression between dioxin-sensitive and dioxin-resistant rats in response to TCDD...... 75 1. Genes that display a Type II rather than a Type I response to TCDD...... 77 2. Genes whose altered expression, as detected by the arrays, was confirmed by real-time RT-PCR...... 98 3. Genes whose response to TCDD potentially is regulated by the AHR ...... 103 4. Genes that are components of biological processes/pathways associated with phenotypes altered by TCDD exposure...... 106 DISCUSSION...... 110 Type II responses: those that differ between the resistant rat collective and the sensitive rat collective ...... 111 Wasting syndrome ...... 112 Hepatotoxicity...... 117 SIGNIFICANCE AND IMPACT OF CHAPTER...... 126

CHAPTER 4: Micro-RNAs in Adult Rodent Liver Are Refractory To Dioxin Treatment ...... 127 ABSTRACT...... 128 INTRODUCTION ...... 129 MATERIALS AND METHODS...... 131 RESULTS ...... 139 Ahr-null mouse model ...... 140 Dioxin-resistant rat model ...... 150 Cell culture models...... 158 Predicted mRNA targets for candidate miRNA regulation ...... 160

vi miRNAs predicted to affect the AHR...... 160 AH response elements upstream of miRNA coding regions ...... 161 DISCUSSION...... 162 Constitutive miRNA expression is affected only modestly by AHR-genotype in adult rodent liver...... 162 miRNAs in adult rodent livers are refractory to dioxin treatment...... 163 Alternative potential AHR-mediated mechanisms for downregulation of mRNAs ...... 165 SIGNIFICANCE AND IMPACT OF THIS PROJECT ...... 167

CHAPTER 5: Integrating Discussion ...... 168 Conclusions...... 168 Chapter 2...... 168 Chapter 3...... 169 Chapter 4...... 169 Enduring Questions...... 171 Key Thesis Research Findings...... 172 I. AHR structure is the key determinant of susceptibility to TCDD lethality...... 172 II. Lethality from dioxin-like chemicals likely results from dysregulation of subsets of dioxin-responsive genes under the control of the AHR ...... 173 III. Alteration of hepatic miRNA expression levels are NOT a determinant of susceptibility to TCDD lethality ...... 176 IV. Other potential mechanisms by which the restructured TAD structure selectively increases resistance to TCDD lethality in the H/W rat ...... 178 Key implication of thesis research findings: Receptor-mediated mechanism for risk assessment of dioxins...... 182 What is the risk to human health from dioxins? ...... 182 Which animal species provide the best model for assessing human responses to dioxins: dioxin-sensitive species or dioxin-resistant species?...... 183 Complications of using animal models to assess the risk of dioxin to humans...... 184 FUTURE PERSPECTIVES...... 187 OVERALL SIGNIFIANCE OF THESIS RESEARCH...... 196 REFERENCES ...... 197 APPENDICES ...... 222 Appendix 1.1. Genes reported from literature to be regulated by the AHR or that exhibit altered expression following AHR ligand exposure, “Candidate gene approach” ...... 222 Appendix 1.2. Genes reported from literature to be regulated by the AHR or that exhibit altered expression following AHR ligand exposure, “Microarray approach” ...... 224 Appendix 3.1. Quality assessment and quality control...... 228 Appendix 3.2. Sequences of primers and probes used for real-time RT-PCR ...... 229 Appendix 5.1. Heat map of genes that differ in basal expression between rats with the AHRH/W genotype vs. rats with wildtype AHR...... 230 Appendix 5.2. Do basal levels of candidate gene expression correspond to differences in an animal’s sensitivity to dioxin toxicity...... 231

Addendum: Copyright releases for published papers

vii

LIST OF PUBLICATIONS

by I.D Moffat related to this thesis

Moffat ID, Roblin S, Harper PA, Okey AB and Pohjanvirta R (2007b) Aryl hydrocarbon receptor (AHR) splice variants in the dioxin-resistant rat: tissue expression and transactivational activity. Mol Pharmacol 72:956-966.

Moffat ID, Boutros PC, Tuomisto J, Pohjanvirta R and Okey AB. Dioxin lethality: aryl hydrocarbon receptor-regulated gene expression in a resistant rat model.

Moffat ID, Boutros PC, Celius T, Linden J, Pohjanvirta R and Okey AB (2007a) Micro- RNAs in Adult Rodent Liver are Refractory to Dioxin Treatment. Toxicol Sci 99:470-487.

Boutros PC, Moffat ID, Franc MA, Tijet N, Tuomisto J, Pohjanvirta R and Okey AB (2004) Dioxin-responsive AHRE-II gene battery: identification by phylogenetic footprinting. Biochem Biophys Res Commun 321:707-15.

Okey AB, Franc MA, Moffat ID, Tijet N, Boutros PC, Korkalainen M, Tuomisto J and Pohjanvirta R (2005) Toxicological implications of polymorphisms in receptors for xenobiotic chemicals: The case of the aryl hydrocarbon receptor. Toxicol Appl Pharmacol 207:S43-S51.

Pastorelli R, Carpi D, Campagna R, Airoldi L, Pohjanvirta R, Viluksela M, Hakansson H, Boutros PC, Moffat ID, Okey AB and Fanelli R (2006) Differential expression profiling of the hepatic proteome in a rat model of dioxin resistance: correlation with genomic and transcriptomic analyses. Mol Cell Proteomics 5:882-94.

Pohjanvirta R, Niittynen M, Linden J, Boutros PC, Moffat ID and Okey AB (2006) Evaluation of various housekeeping genes for their applicability for normalization of mRNA expression in dioxin-treated rats. Chem Biol Interact 160:134-49.

Tijet N, Boutros PC, Moffat ID, Okey AB, Tuomisto J and Pohjanvirta R (2006) Aryl hydrocarbon receptor regulates distinct dioxin-dependent and dioxin-independent gene batteries. Mol Pharmacol 69:140-53.

viii ABBREVIATIONS

AHR Aryl hydrocarbon receptor Ahrb1 Mouse "responsive" Ahr allele Ahrd Mouse "non-responsive" Ahr allele AHRE AH response element (also known as DRE or XRE-I) AHRE-II AH response element-II (also known as XRE-II) AHRhw Han/Wistar (Kuopio) rat resistant AHR allele AhRR Aryl hydrocarbon receptor repressor AHRwt Wildtype AHR allele from dioxin sensitive rats a.a. Amino acid ANOVA Analysis of variance ARNT Aryl hydrocarbon receptor nuclear translocator bHLH- Basic-helix-loop-helix Period-ARNT-Single-minded PAS bp Nucleotide base pairs BP8 cells Rat hepatoma cell line that does not express any detectable AHR; derived from hepatoma 5L cell line bw Body weight C57BL/6 “Responsive” mouse strain characterized by: Ahrb1 allele CMV Cytomegalovirus CYP Cytochrome P450 (1A1; 1A2, 7A1, etc.) CYP1A1 Cytochrome P4501A1 CYP1A2 Cytochrome P4501A2 CYP1B1 Cytochrome P4501B1 DBA/2 "Non-responsive" mouse strain characterized by: Ahrd allele DRE Dioxin response element (see AHRE) DV Deletion variant form of the AHR from H/W(Kuopio) rats wt/hw wt/hw F1 L-E X H/W cross characterized by: AHR , "B" gene FBS Fetal bovine serum FDR False discovery rate GO Gene ontology GST-Ya Glutathione S-transferase Ya subunit H/W Han/Wistar (Kuopio) rat strain HAH Halogenated aromatic hydrocarbon IVs Insertion variant forms of AHR-H/W L-E Long-Evans(Turku AB) rat strain Line-A Dioxin-resistant line characterized by: AHRhw/hw, "B" genewt/wt Line-B Dioxin-resistant line characterized by: AHRwt/wt, "B" genehw/hw Line-C Dioxin-sensitive line characterized by: AHRwt/wt, "B" genewt/wt LIV Long insertion variant form of AHR-H/W LnA Line-A rat strain LnB Line-B rat strain LnC Line-C rat strain MC 3-methylcholanthrene NQO1 NAD(P)H:quinone oxidoreductase

ix nt Nucleotides PAH Polycyclic aromatic hydrocarbon PAS Per/ARNT/AHR/Sim protein family PCR Polymerase chain reaction P-S-T-rich Proline- serine- threonine-rich subdomain of AHR PUNS Primer-Unigene Selectivity Testing Q-rich Glutamine-rich subdomain of AHR RNAi RNA interference RT Reverse transcription RT-PCR RT-PCR SD Sprague Dawley rat strain siRNA Small inhibitory RNA SIV Short insertion variant form of AHR-H/W t½ Half life TAD Transactivation domain (of the AHR) TCDD 2,3,7,8-tetrachlorodibenzo-p-dioxin (“dioxin”) UGT UDP-glucuronosyltransferase UGT1A1 Uridine diphosphate glucuronosyltransferase 1A1 UGT1A6 Uridine diphosphate glucuronosyltransferase 1A6 WT Wildtype AHR allele from dioxin sensitive rats αMEM αMinimal essential medium

x LIST OF TABLES

CHAPTER 1 Table 1.1. Human health concerns associated with dioxins as reported by the National Academies of Science-Institute of Medicine (USA)...... 4 Table 1.2. Examples of AHR interactions with proteins involved in regulation of gene expression...... 11 Table 1.3. Comparison of the pharmacogenetics of dioxin responses in vivo...... 22 Table 1.4. Comparison of L-E vs. H/W rat responses after treatment with TCDD...... 24

CHAPTER 2 Table 2.1. Primer and probe sequences for real-time quantitative RT-PCR measurements of rat WT AHR and rat splice-variant transcripts...... 43 Table 2.2. Predicted secondary AHR protein structures of transactivation domains in different species and strains...... 57

CHAPTER 3 Table 3.1. Responses of classic AH gene battery members to TCDD...... 79 Table 3.2. Type I responses to TCDD...... 80 Table 3.3. Type II responses to TCDD...... 90 Table 3.4. Gene Ontological (GO) analysis of the TCDD effect on transcript expression in dioxin-resistant and/or dioxin sensitive rats ...... 108

CHAPTER 4 Table 4.1. Real-time RT-PCR measurements of microRNA levels ...... 137 Table 4.2. microRNAs affected by AHR-genotype alone, TCDD treatment alone, or the interaction of AHR and TCDD in the Ahr-null mouse model as measured by Exiqon arrays...... 142 Table 4.3. Summary of differences in microRNA levels between dioxin-sensitive L-E rats vs. dioxin-resistant H/W rats measured by array...... 152 Table 4.4. microRNAs affected by strain, TCDD, or feed restriction in the rat model measured by Exiqon arrays...... 153

CHAPTER 5 Table 5.1. AHR structure is a determinant of susceptibility to TCDD lethality...... 181 Table 5.2. Standards for ‘permissible’ daily intake of dioxins in various jurisdictions 182

xi LIST OF FIGURES

CHAPTER 1 Figure 1.1. Dioxin sources and distribution...... 2 Figure 1.2. General AHR mechanism...... 8 Figure 1.3. microRNAs (miRNAs) are negative regulators of gene expression...... 12 Figure 1.4. Selected AH receptor ligands ...... 13 Figure 1.5. Structure of the AHR...... 15 Figure 1.6. Domain structure of the AHR and the region deleted in the transactivation domain of the H/W(Kuopio) rat due to alternative splicing ...... 26 Figure 1.7. Genetic differences in sensitivity to acute TCDD lethality in rat strains and rat lines...... 27 Figure 1.8. Overview of thesis research projects...... 35

CHAPTER 2 Figure 2.1. Domain structure of the AHR and the region deleted in the transactivation domain of the H/W(Kuopio) rat due to alternative splicing ...... 40 Figure 2.2. Constitutive AHR transcript expression levels in 5 tissues from dioxin- resistant (H/W, LnA and F1) and dioxin-sensitive (L-E and LnC) rats...... 49 Figure 2.3. Expression levels of AHR splice-variant transcripts in rat liver after treatment with TCDD ...... 50 Figure 2.4. Effect of rat AHR TAD polymorphism on intrinsic transactivation activity . 52 Figure 2.5. Effect of rat AHR TAD polymorphism on intrinsic transactivation activity . 53 Figure 2.6. Inter- and intra-species comparison of predicted secondary protein structure of the TAD ...... 55 Figure 2.7. Schematic representation of differences in predicted TAD protein structures between rat WT AHR and the rat splice-variants...... 58

CHAPTER 3 Figure 3.1. Experimental design...... 69 Figure 3.2. Comparison of expression profiles between strains ...... 72 Figure 3.3. Venn diagrams...... 76 Figure 3.4. Type I TCDD-responsive genes from livers of dioxin-resistant and dioxin- sensitive rats: Real-time RT-PCR measurement of mRNA levels for selected genes...... 100 Figure 3.5. Type II TCDD-responsive genes from livers of dioxin-resistant and dioxin-sensitive rats: Real-time RT-PCR measurement of mRNA levels for selected genes...... 102 Figure 3.6. Ahr-null mouse model: measurement of selected mRNA levels by real- time RT-PCR ...... 105 Figure 3.7. Functional analysis ...... 107 Figure 3.8. Summary of proposed responses to TCDD after 3 or 19 h exposure to TCDD...... 125

xii

CHAPTER 4 Figure 4.1. Experimental design for Exiqon MiRCury LNA miRNA array ...... 133 Figure 4.2. Overlap of effects in the mouse model...... 141 Figure 4.3. Measurement of selected miRNA levels in the Ahr-null mouse model by real-time RT-PCR ...... 148 Figure 4.4. Measurement of selected miRNA levels from livers of dioxin-resistant and dioxin-sensitive rats by real-time RT-PCR...... 155 Figure 4.5. Measurement of selected miRNA levels from thymus and kidney of dioxin-resistant rats and dioxin-sensitive rats by real-time RT-PCR ...... 156 Figure 4.6. Measurement of selected miRNA levels in the rat 5L cell line and the mouse Hepa-1 cell line by real-time RT-PCR ...... 159

CHAPTER 5 Figure 5.1. Proposed mechanism of AHR-mediated dioxin lethality and associated toxicities...... 170 Figure 5.2. Animal models support a sublinear threshold model ...... 185 Figure 5.3. Strategy to identify gene responses linked to lethality & hepatic toxicity... 190 Figure 5.4. Pharmacologic screen to filter candidate genes for plausibility that their altered expression is relevant to TCDD toxicity...... 191

xiii

CHAPTER 1: General Introduction

Dioxins and dioxin-like compounds

Labeled in popular media as the "most toxic substance known to man", dioxins are highly persistent trace environmental contaminants. Dioxins first commanded wide attention in the 1960’s when they were revealed as a contaminant of ‘Agent Orange’, an herbicide used in the Vietnam War (Michalek et al., 1996). Since then, numerous accidents have introduced dioxins into the environment; most notable was the 1976 industrial accident in Seveso, Italy (Baccarelli et al., 2005). More recently, dioxins were in the news with the poisoning of President Viktor Yushchenko of Ukraine in 2004 (Schecter et al., 2006). Dioxin is the generic name for a large family of environmental contaminants that share a similar chemical structure and a common mechanism of toxic action. This family includes halogenated aromatic hydrocarbons (HAHs) comprised of 75 polychlorinated dibenzo-p-dioxins (PCDDs), 135 polychlorinated dibenzofurans (PCDFs), and 209 polychlorinated biphenyls (PCBs) congeners. The prototype and most potent dioxin congener is 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD). Dioxins enter the environment from multiple sources including as unintentional byproducts of many forms of combustion and from reservoir sources of previous environmental releases such as the industrial accident in Seveso (Figure 1.1). The long half-life of TCDD in soil, water, air, and subsequently our food supply is cause for concern. Once in the environment dioxins bioaccumulate and biomagnify putting numerous vertebrate species at risk. The current average daily human intake of dioxins in Canada is ~ 2 to 4 pg/kg (Feeley and Grant, 1993; van Leeuwen et al., 2000) . While these intake levels are lower than the tolerable daily intake (TDI) level which the Canadian government considers acceptable (10 pg/kg/day) (Feeley and Grant, 1993), they are much higher than the controversial (Cheng et al., 2006; Starr, 2003) level of 0.006 pg/kg/day set by the United States Environmental Protection Agency (USEPA) (US-

1 EPA, 2004). Therefore, further research is necessary to determine what levels of dioxin constitute a significant risk to human health and what levels produce toxic effects in other vertebrate organisms in the ecosystem. While dioxin emissions and subsequent dioxin intake levels are decreasing with better remediation technologies (Kulkarni et al., 2007), the body burdens of dioxins do not change as rapidly due to the long elimination half-life of TCDD. There is legitimate concern among the general public and government regulatory agencies that TCDD has major adverse effects on human health and health of other species (Birnbaum, 1994).

Incineration Sources Combustion Sources

•Municipal waste •Cement kilns

•Hospital waste •Wood burning •Hazardous waste •Diesel vehicles •Swage sludge •Coal fired utilities

Industrial Sources Reservoir Sources •Pulp & paper Dioxins in the •Biochemical processes Environment •Chemical manufacturing •Photolytic processes

•Metal industry •Forest fires •Accidental releases (“Agent orange”, Seveso)

Food Supply (Animal & animal products, soil, water, inhalation) Bioaccumulation/Biomagnification

Laboratory Humans animals ?

Toxic Outcomes

Figure 1.1. Dioxin sources and distribution. Modified from (Kulkarni et al., 2007).

2

Effects of dioxins in humans

In 1997 the International Agency for Research on Cancer (IARC) classified TCDD as “carcinogenic to humans (‘IARC group 1’)” (IARC, 1997). This classification was based on “limited evidence of carcinogenicity to humans” derived from workers in industrial accidents, sufficient evidence of carcinogenicity in experimental animals, and extensive mechanistic information indicating that TCDD acts through the aryl hydrocarbon receptor (discussed in subsequent sections). The group I classification has been controversial and is both challenged and supported in literature (Cole et al., 2003; Steenland et al., 2004). In 2000 the United States Environmental Protection Agency’s draft re-evaluation of dioxins suggested that human cancer risk might be as high as 1 in 100 for the most extensively dioxin-exposed individuals (Kaiser, 2000a; Kaiser, 2000b). This risk estimate is highly controversial (Starr, 2001) but serves to illustrate the magnitude of concern about dangers from dioxin-like compounds.

In 2004, the National Academies’ Institute of Medicine update on their report, “Veterans and Agent Orange” concludes that there is “sufficient evidence of an association” between herbicide exposure (contaminated with dioxins) and five ailments and “limited or suggestive evidence for an association” between herbicide exposure and seven ailments (Table 1.1) (Stone, 2007). In addition, Schecter and colleagues (Schecter et al., 2006) provide an overview of greater than 32 clinical manifestations of dioxin toxicity including: reproductive and developmental abnormalities, immune deficiencies, skin disease, neurobehavioral disorders, various types of cancer, endocrine disruption, and thyroid disorders.

3 Table 1.1. Human health concerns associated with dioxins as reported by the National Academies of Science-Institute of Medicine (USA)

Sufficient Evidence of an association Chronic lymphocytic leukemia Soft-tissue sarcoma Non-Hodgkin’s lymphoma Hodgkin’s disease Chloracne Limited or Suggestive evidence of an association Respiratory cancer Prostate cancer Multiple myeloma Early-onset transient peripheral neuropathy Porphyria cutanea tarda Type 2 diabetes mellitis Spina bifida in offspring of exposed individual

Note: This report was used to evaluate and compensate USA military veterans exposed to “Agent Orange” during the Vietnam War (Stone, 2007).

Acute effects of dioxins in laboratory animals Concerns about toxic effects of dioxins are not restricted to human health. In laboratory animals, extraordinarily low exposure levels to TCDD cause a wide spectrum of toxic responses. Only a few end points relevant to the present thesis are discussed.

Adaptive response TCDD induces expression of xenobiotic metabolizing enzymes that catalyze metabolic processing of lipophilic chemicals to water-soluble derivatives thereby facilitating their elimination (Gu et al., 2000; Nebert et al., 2004; Okey, 1990). TCDD itself is poorly metabolized and thus can cause chronic and sustained induction of cytochrome P450s, leading to oxidative stress, formation of reactive oxygen species, lipid peroxidation and DNA damage (Hassoun et al., 2001; Hassoun et al., 2002; Nebert et al., 2000; Shertzer et al., 1998; Slezak et al., 2000).

4

Wasting syndrome Within the first few days following a single dose of TCDD a wasting syndrome is observed in laboratory animals characterized by progressive weight loss (up to 50%) and hypophagia (Pohjanvirta and Tuomisto, 1994; Seefeld et al., 1984a; Seefeld et al., 1984b). Depending on the TCDD dose, the decrease in feed intake can be dramatic and irreversible, or temporary. After a sublethal dose of TCDD animals resume feeding within 1-2 weeks; however, their body weight never returns to that of untreated animals. A lethal dose of TCDD produces irreversible hypophagia and weight loss. This wasting syndrome is believed to occur due to a lowering of the body weight set-point (Pohjanvirta and Tuomisto, 1994) and contributes to lethality starting at 2-3 weeks after TCDD exposure. While wasting accompanies death, wasting per se is likely not the only cause of death since maintenance of body weight by parenteral nutrition does not prevent mortality (Gasiewicz et al., 1980). Many pathways have been proposed to explain TCDD-induced wasting, including gluconeogenic pathways (Weber et al., 1991), c-Src- mediated pathways (Vogel et al., 2003), stress-response pathways (Matsumura, 2003), and complex adaptive pathways including disturbances of lipid, carbohydrate, and nitrogen metabolism (Fletcher et al., 2005). Despite extensive research, mechanisms which mediate the wasting syndrome are not fully understood.

Hepatotoxicity TCDD causes significant hepatotoxicity. In most laboratory animals, hepatotoxicity is characterized by hepatocellular hypertrophy, multinucleated hepatocytes, steatosis, and inflammatory cell infiltration often accompanied by scattered focal necrosis (Pohjanvirta and Tuomisto, 1994). After the initial hypertrophy, approximately 1 week following a lethal TCDD dose the liver begins to atrophy (reviewed in (Pohjanvirta and Tuomisto, 1994)).

5

Tumor promotion Tumor promotion is a critical end point used for dioxin risk assessment. TCDD is not genotoxic or mutagenic and therefore is not a direct initiator of cancer. Instead, TCDD appears to promote tumors mainly by disrupting the normal balance between cell proliferation and programmed cell death (Bock and Kohle, 2005; Schrenk et al., 2004).

Acute lethality TCDD is extremely toxic to many animal species, giving TCDD the label “the most toxic substance known to man”. After an acute lethal dose of TCDD animals do not succumb immediately; death is delayed while the wasting syndrome, hepatotoxicity, and other toxic responses progress. This time lapse before death varies from one week up to 8 weeks (Pohjanvirta and Tuomisto, 1994). The ultimate cause of death, as well as the key target tissue(s) responsible for lethality remains elusive. In addition, wide inter- and intra-species differences exist in the lethal potency of TCDD (see species-comparison of LD50s, Table 1.3). A particular animal species may even be resistant to TCDD lethality but sensitive to other endpoints such as xenobiotic metabolizing enzyme induction.

The lack of thorough mechanistic understanding of TCDD lethality and associated toxicities when combined with striking variability among species in response to TCDD greatly complicates dioxin risk assessment. Thus, continued research to discover mechanisms of major dioxin toxicities will greatly benefit all organisms which share our ecosystem.

6

The AH Receptor: Transcription factor and essential mediator of dioxins toxicities

Given the exceptional toxicity of dioxins to a wide range of organisms it clearly is of great practical importance, as well as fundamental scientific value, to understand the mechanisms by which dioxins exert their effects. A little over 30 years ago a radiolabeled analogue of TCDD was found to specifically bind with high affinity to the aryl hydrocarbon receptor (AHR) from mouse liver (Okey et al., 1979; Okey et al., 1980; Poland et al., 1976). Subsequently, the AHR has been cloned from many vertebrate species as have AHR orthologs from invertebrates (reviewed in: (Hahn, 1998; Hahn, 2002; Hahn et al., 2006)). Phylogenetic evidence indicates that the AHR arose over 450 million years ago (Hahn, 2002; Hahn et al., 2006; Thomas et al., 2002), possibly as a strategy for animals to adapt to toxicants in plant food sources (Gu et al., 2000). Following discovery of the AHR much research has been devoted to elucidating the AHR’s signaling pathway, ligands, structure, and role in mediating TCDD lethality and associated toxicities.

The conventional AHR signaling pathway

In the absence of a ligand, the AHR is found in the cytoplasm complexed with 2 molecules of Hsp90 and two additional proteins: p23 and immunuophilin-like X- associated protein 2 (XAP2; also known as ARA9 or AIP1) (Figure 1.2). Hsp90 and p23 stabilize the AHR and maintain it in a conformation accessible for ligands. XAP2 performs numerous roles; it stabilizes the complex by binding to both Hsp90 and the AHR, regulates AHR cellular location, protects the AHR from degradation and represses AHR transcriptional activity (Harper et al., 2006; Hollingshead et al., 2004; Kazlauskas et al., 1999; Kazlauskas et al., 2000; Ma and Whitlock, 1997; Meyer and Perdew, 1999; Petrulis and Perdew, 2002; Pongratz et al., 1992). Other chaperones such as p60 & hap70 have been proposed but their functional roles have not been established (Nair et al., 1996; Petrulis and Perdew, 2002).

7

Other D Hsp 90 TCD p proteins 2 23 Xap AHR

AHRR Ligand Binding & ARNT Translocation AHR ARNT AHRE

specific mRNAs

P450s & Phase II enzymes OTHER GENES: growth / CYP1A1, 1A2, 1B1, 2S1; differentiation / metabolism Gstya, Ugt1a6

“CLASSIC” Altered activation or

DIOXIN TOXICITY detoxication of xenobiotics

Figure 1.2. General AHR mechanism. Ligands bind the AHR in cytoplasm and trigger “transformation” of the receptor into a DNA- binding protein. The ligand-AHR complex translocates into the cell nucleus where it releases chaperone proteins such as hsp90, then dimerizes with a new partner protein, ARNT. The ligand- AHR-ARNT complex binds specific DNA sequences known as AH responsive elements (AHREs) located in the 5’-flanking region of target genes. This provides a platform for recruiting multiple co-activator proteins that increase or decrease gene transcription. (Modified from: (Okey, 2007) )

8 TCDD is highly lipophilic and therefore can enter the cell by simple diffusion. Since TCDD is not significantly metabolized it is the parent compound itself which activates the AHR. TCDD binding to the cytosolic AHR induces a change in AHR conformation exposing the AHR nuclear localization signal (NLS) (Henry and Gasiewicz, 2003; Ikuta et al., 2004; Lees and Whitelaw, 1999). The ligand-AHR complex then translocates into the cell nucleus (Okey et al., 1980) where it dissociates from its partner proteins. The ligand-AHR complex then heterodimerizes with a new partner protein, the AH receptor nuclear translocator (ARNT) (Hankinson, 2005; Kazlauskas et al., 2001; Probst et al., 1993; Tomita et al., 2000). The ligand-AHR-ARNT complex then binds to specific DNA sequences known as AH responsive elements (AHREs) located in the 5’-flanking region of target genes. The AHRE is composed of 2 half-sites, TNGC and GTG, recognized by the AHR and ARNT respectively (Denison et al., 1988; Swanson et al., 1995). AHR-complex binding to a target gene provides a platform for recruiting multiple co-activator and/or co-repressor proteins to regulate transcription of specific genes. Table 1.2 provides examples of co-activators, co-repressors, and other proteins which may interact with the AHR to alter target gene expression.

Upregulation of gene expression. The mechanism that increases gene expression of CYP1A1 is the most completely elucidated. Following AHR-complex binding, a process ensues involving DNA bending, nucleosomal disruption, and interaction with transcription factors and coactivators (reviewed in (Hankinson, 2005)).

Downregulation of gene expression. Little is known about mechanisms by which dioxin- like chemicals suppress mRNA levels (Riddick et al., 2004). It is known that the aryl hydrocarbon receptor repressor (AHRR) is regulated by ligand activation of the AHR in some tissues (Mimura et al., 1999) and in cell culture (Haarmann-Stemmann et al., 2007). The AHRR plays a role in repressing the transcription of AHR-regulated genes – like CYP1A1 – by out-competing the AHR for heterodimerization with ARNT and subsequently blocking AHR-mediated transcription by binding to AHREs and/or forming a co-repressor complex (Haarmann-Stemmann et al., 2007). The AHRR itself cannot transactivate gene expression (Karchner et al., 2002; Mimura et al., 1999). Therefore,

9 AHRR can block upregulation of gene expression and provide negative feedback regulation of the AHR.

One plausible new mechanism by which transcript levels might be reduced by the AHR is through the action of microRNAs (miRNA). miRNAs are evolutionarily conserved, small (~22-nt), non-coding transcripts that suppress levels or activity of target mRNAs by multiple mechanisms including triggering mRNA degradation, blocking translation, or modifying chromatin structure to silence transcription (Figure 1.3). miRNAs regulate diverse biological processes ranging from embryonic development to fat storage, insulin secretion, drug-metabolism, apoptosis, cell growth, tumorigenesis, and death (Gaur et al., 2007; Grimm et al., 2006; Tsuchiya et al., 2006). These same processes are also altered by dioxin exposure. Hypothetically, dioxins could upregulate or downregulate specific microRNAs which, in turn, could alter degradation, efficiency of translation or transcription of target mRNAs.

Attenuation of AHR signaling. Two mechanisms to attenuate AHR signaling have been identified. First, ligand-activated AHR is a target for the ubiquitin-proteasome degradation pathway after it binds to the AHRE (Chen et al., 2005b; Ma et al., 2000; Pollenz, 2007; Pollenz and Buggy, 2006). Secondly, AHR-mediated upregulation of AHRR (discussed above) provides a negative feedback to attenuate AHR signaling (Mimura et al., 1999). Ligand-dependent degradation along with upregulation of AHRR may be the cell’s way of protecting itself from the consequences of excessive stimulation by high concentrations of potent agonists.

AHR-independent TCDD responses. While the majority of TCDD toxicities are mediated by the AHR (see following “AHR mediates dioxin toxicities” section), it is possible that a small fraction of responses may be AHR-independent. TCDD was suggested to activate a few signal transduction proteins by an AHR-independent mechanism (Park et al., 2003; Park et al., 2005; Tan et al., 2002). However, it not known if these molecules cross-talk with the AHR-ARNT signaling pathway. For example, early reports of apparent AHR- independent TCDD regulation of MAP kinase signaling (Tan et al., 2002) were

10 subsequently found to, in fact, enhance activity of ARNT and other cofactors of AHR- ARNT signaling (Tan et al., 2004). Therefore, this thesis research will focus on AHR- mediated signaling events involved in dioxin toxicities.

Table 1.2. Examples of AHR interactions with proteins involved in regulation of gene expression.

AHR interaction with Response Reference

Interactions which may enhance gene transcription General Transcription Factors: Pol II, TBP, TFIIB, TFIIF Reviewed in (Hankinson, 2005) , , , , Co-activators: Src-1 NcoA2 p/CIP p300, RIP 140 CARM1, PRMT1, Brg-1, Brahma

Mediators: Med220, CDK8

Interactions which may inhibit gene transcription SMRT (silencing mediator of retinoic acid and thyroid hormone receptor) (Nguyen et al., 1999) Transcription factors ARNT ARNT dimerization with other proteins may reduce (Hofer et al., ARNT pools available to dimerize with AHR and 2004) thus reduce AHR signaling

Rb Interaction appears necessary for maximal AHR (Ge and (retinoblastoma) activity and potentiates repression of E2F- Elferink, 1998; Marlowe et al., dependent transcription via TCDD-induced AHR 2004; Puga et binding to the E2F promoter which leads to cell al., 2000a) cycle arrest

Unknown protein Together bind AHRE-II (CATGN6CT/ATG) in 5’- (Boutros et al., region of target genes to regulate expression 2004; Sogawa et al., 2004) Steroid hormone signaling pathways E.g. Estrogen E.g., activated AHR-ARNT complex can bind ER, (Ohtake et al., (ER) in the absence of ER ligands to activate 2003) transcription of ER-mediated genes

11

Pre-miRNA

Transcription Pri-miRNA

Mature miRNA within RISC Regulate mRNA

mRNA degradation Translational repression Transcriptional repression

Active mRNA mRNA Chromatin

Histone methylation

Silent Chromatin

Figure 1.3. microRNAs (miRNAs) are negative regulators of gene expression. microRNAs are transcribed in the nucleus, but are not translated. This pri-miRNA is then processed into a 70-nt stem-loop structure known as pre-miRNA by a nuclease Drosha before export to the cytoplasm. In the cytoplasm, pre-miRNA are further processed mature ~22 nt miRNAs by the endonuclease Dicer which also initiates the formation of the RNA- induced silencing complex (RISC). This complex mediates gene silencing by multiple mechanisms including triggering mRNA degradation, blocking translation, or modifying chromatin structure. Greater than 94% of known rat mRNAs are potential targets of miRNAs. (Source: http://www.ambion.com/techlib/resources/miRNA/mirna_fun.html

12 Repertoire of ligands which bind to the AHR Since the AHR’s discovery as a ‘TCDD receptor’ the repertoire of compounds which bind to the AHR has greatly expanded. This promiscuous receptor binds structurally diverse ligands (Figure 1.4). Two major classes of exogenous ligands include the planar PAHs and HAHs. PAHs include 3-methylcholanthrene, benzo[a]pyrene, benzanthracenes, and benzoflavones (Denison and Nagy, 2003; Poland and Knutson, 1982). HAHs include chlorinated or brominated dioxins (TCDD), dibenzofurans, and biphenyls. Structure-activity studies with HAH congeners show that toxicity is proportional to the congener’s affinity for the AHR (Poland and Knutson, 1982; Safe, 1986). TCDD has the highest affinity and remains the most toxic of the known AHR ligands; TCDD is 30,000 times more potent than benzo[a]pyrene to induce CYP1A1 enzyme expression (Nebert et al., 2000). Despite the discovery of new AHR ligands, TCDD remains the prototype AHR ligand to which all subsequent ligands are compared.

Halogenated aromatic hydrocarbon Plant

2,3,7,8-tetrachlorodibenzo-p-dioxin indolo[2,3-b]carbazole

Polycyclic aromatic hydrocarbon

3-methylcholanthrene benzo[a]pyrene

Figure 1.4. Selected AH receptor ligands. Modified from (Okey et al., 1994)

13 Several plant constituents also act as AHR ligands (Denison and Nagy, 2003; MacDonald et al., 2001) – for example, cabbage and other cruciferous vegetables contain abundant indole carbinols that are converted in the GI tract into AHR agonists (Chen et al., 1995; Jellinck et al., 1993; Kleman et al., 1994). While the indole carbinols have a high affinity for the AHR, they do not cause toxicity (Pohjanvirta et al., 2002). Indigo dyes are also AHR agonists (Adachi et al., 2001; Guengerich et al., 2004) whereas resveratrol, a compound found in wine, grapes and other plant materials, is an antagonist (Casper et al., 1999; Chen et al., 2004; Revel et al., 2001; Schneider et al., 2001).

Although no genuine endogenous AHR ligand has been confirmed, many candidates have emerged, including bilirubin (Phelan et al., 1998; Seubert et al., 2002; Sinal and Bend, 1997), tryptophan metabolites (Bittinger et al., 2003; Heath-Pagliuso et al., 1998), arachidonic acid metabolite lipoxin A4 (Schaldach et al., 1999), a carboxylic methyl ester derivative from lung (Song et al., 2002), several prostaglandins (Seidel et al., 2001), retinoids (Gambone et al., 2002; Soprano et al., 2001; Soprano and Soprano, 2003), indigo and indirubin isolated from human urine (Adachi et al., 2001; Sugihara et al., 2004; Sugihara et al., 2007), a modified low-density lipoprotein (McMillan and Bradfield, 2007), metabolically susceptible endogenous activator (s) (Chiaro et al., 2007), and the proposed antagonist 7-Ketocholesterol (Savouret et al., 2000).

Structure of the AHR When the primary structure of the AHR was determined by cDNA cloning it was revealed that the AHR is a member of the basic-helix-loop-helix/Per-Arnt-Sim (bHLH/PAS) superfamily. Members of the bHLH/PAS superfamily play multiple roles as sensors of environmental change and mediators of developmental signals (Gu et al., 2000). Like other bHLH/PAS proteins, the AHR is composed of modular domains that can function independently (Figure 1.5).

14

hsp90 binding

bHLH PAS domains Transactivation Domain COOH NH2 A B Acidic Q-Rich P-S-T-Rich

O

Cl Cl

O DNA Cl O Cl

binding

dimerization with ARNT

Figure 1.5. Structure of the AHR. Modified from (Okey, 2007)

The highly conserved N-terminal half of the AHR contains the bHLH-PAS domain. The PAS region affords specificity for dimerization with ARNT and harbors most of the ligand binding domain (LBD) (Fukunaga and Hankinson, 1996; Fukunaga et al., 1995; Pongratz et al., 1998). Both the bHLH and the PAS domains are responsible for interacting with Hsp90 (Antonsson et al., 1995). The bHLH domain mediates AHR translocation into the nucleus and its subsequent binding to DNA (Ikuta et al., 2004). Translocation of the AHR into the nucleus is required for dioxin toxicities (Bunger et al., 2003). The C-terminal half of the AHR contains the transactivation domain (TAD) which mediates transcription of target gene expression. The TAD is comprised of three distinct subdomains: acidic, glutamine (Q)-rich, and proline-serine-threonine (P-S-T)-rich regions (Ma et al., 1995). In mouse, each of these three subdomains exhibits varying levels of transactivation activity individually. In humans, each subdomain has little activity in isolation; however, combining the subdomains significantly increases levels of transcriptional activation, consistent with synergistic activity among the subdomains (Ma et al., 1995; Sogawa et al., 1995). In addition, the AHR’s Q-rich subdomain is necessary and sufficient for in vitro interaction with coactivator proteins RIP140 and Src-1 as well as with Rb (Ge and Elferink, 1998; Kumar and Perdew, 1999; Kumar et al., 1999). Thus, this region of the AHR is critical for transactivating gene expression and interacting with

15 proteins important for regulating expression of target genes. Marked inter-species and intra-species differences exist in AHR C-terminal mRNA sequences and protein sequences but the sequence, per se, of the transactivation domain is not always a faithful predictor of sensitivity to dioxins (Korkalainen et al., 2000; Korkalainen et al., 2001).

The AHR mediates dioxin toxicities It is firmly established that toxic effects of dioxin-like chemicals are initiated by their binding to the AHR. Three AHR knockout mouse models in the mid-1990s demonstrated that the AHR is required for major forms of TCDD toxicity. In all three models TCDD fails to induce AHR target genes like Cyp1a1 and Cyp1a2. Ahr-null mice also are highly resistant to TCDD toxicities including, acute lethality, hepatotoxicity, teratogenicity, and reproductive defects when compared to mice with wildtype AHR (Fernandez-Salguero et al., 1995; Mimura et al., 1997; Schmidt et al., 1996). During the past four years Bradfield and colleagues provided further evidence that it is the AHR’s transcriptional regulatory function that leads to toxicity. They showed that in mice in which a mutation has been introduced into the nuclear translocation/DRE(AHRE)- binding domain of the AHR, the AHR does not translocate into the nucleus (Bunger et al., 2003). These AHR-NLS mice are as resistant to TCDD toxicities as Ahr-null mice. This critical finding indicates that nuclear translocation of the AHR is required for most, if not all, AHR-mediated responses. Moreover, if mice are engineered to be hypomorphic for the AHR’s dimerization partner, ARNT, they too are highly resistant to TCDD (Walisser et al., 2004b). This indicates that both partners of the heterodimerization transcription factor complex are essential components of the toxic response to dioxins. This composite of experiments convincingly demonstrates that the normal physiologic role of the AHR, as well as, its role in mediating toxicities is due to the AHR’s transcriptional regulatory function. Therefore it is productive to further clarify how the structure of the transactivation domain affects AHR function and to identify the key genes that are dysregulated by TCDD as a means to complete our understanding of the mechanism of dioxin toxicity.

16 Expanding the AHR-mediated transcriptome in relation to TCDD lethality and associated toxicities

The best-characterized AHR target genes altered by TCDD exposure are those encoding Phase I and Phase II enzymes (Nebert et al., 2000). Enzyme induction acts as an adaptive response that, on balance, facilitates biotransformation, detoxication, and clearance of lipophilic foreign compounds (reviewed in (Nebert et al., 2004; Okey, 1990)). CYP1 enzyme induction accompanies toxicity in animals exposed to TCDD and previously was not considered necessary for toxicity (Poland and Knutson, 1982). Ahr- null mice do not exhibit induction of CYP1 enzymes when treated with TCDD (Fernandez-Salguero et al., 1996; Schmidt et al., 1996). However, recent Cyp1a1 or Cyp1a2 knockout experiments in mice show that these mice are highly resistant to TCDD-induced thymic atrophy, hepatic toxicity, wasting, and lethality (Smith et al., 2001; Uno et al., 2004). The Cyp1a1-null and Cyp1a2-null experiments indicate that these P450s contribute to dioxin toxicity in rodents. However, CYP1A1 or 1A2 upregulation alone is not sufficient to cause classical dioxin toxicity; CYP1A1 is upregulated identically in dioxin-sensitive rats and dioxin-resistant rats (discussed below). Moreover, AHR ligands such as benzo[a]pyrene, β-naphthoflavone, or indolo[3,2-b]carbazole increase expression of CYP1A1 and 1A2 without provoking dioxin-like toxicity (Pohjanvirta et al., 2002; Poland and Knutson, 1982). Therefore, upregulation of CYP1A1 or 1A2 may modify dioxin toxicity, but CYP1 induction in itself is not sufficient for dioxin toxicity. Some genes under AHR control in addition to CYP1A1 or 1A2 must be dysregulated in order to provoke major dioxin toxicities.

Which additional AHR-mediated genes are responsible for major dioxin toxicities? Conventional biochemical and molecular methods over the past three decades have revealed numerous genes that are regulated by the AHR or that exhibit altered expression following TCDD exposure (Appendix 1.1). Of these genes, only one candidate gene (p27Kip-1), thus far, revealed itself to be a reasonable candidate gene to mediate of dioxin toxicity, in this case in thymus (Kolluri et al., 1999). Kip1, a cyclin/cdk inhibitor, is induced by TCDD via the AHR pathway leading to decreased proliferation of

17 thymocytes and potential thymic atrophy. However, Kip1 is unlikely to explain other major toxicities such a wasting, hepatotoxicity, and death. Despite the many genes that have been cataloged as being TCDD-responsive, studying one candidate gene at a time has not been successful in fully elucidating the mechanism(s) of dioxin lethality and associated toxicities.

Perhaps the reason the “death gene” has not been identified after 30 years of research is that lethality requires altered expression of more than one AHR-mediated gene. TCDD exposure may alter expression of specific AHR co-regulated genes which contribute to the varied toxic endpoints. But, which set of co-regulated genes matter to dioxin-lethality? Characterizing a set of AHR co-regulated genes which matter to TCDD- toxicity using the one-candidate-gene-at-a-time approach seems an impossible task, especially given the 30 year track record using this approach.

Fortunately, new technologies have matured which allow simultaneous interrogation of numerous gene responses to TCDD. First, new computational algorithms can predict whether a gene has an AHRE and therefore may be regulated by the AHR. Second, gene expression arrays can simultaneously measure responses of thousands of transcripts representing known and novel genes, following virtually any experimental treatment. Enhancing these technologies is the expanding availability of high quality sequence and better-annotated gene information from human, mouse, and rat genome projects, allowing more annotated genes to be simultaneously characterized.

Numerous research groups have applied this rapidly-evolving technology to the field of dioxin toxicology (Appendix 1.2). An astounding number of genes harbor potential AHR binding sites and, in fact, many of these genes respond to AHR ligands (Boutros et al., 2004; Kel et al., 2004; Sun et al., 2004). However, a large proportion of genes whose expression is increased or decreased by dioxins probably are not primary transcriptional events in the mechanism of lethality or other major toxic responses in mammalian species:

18 (a) some dioxin-responsive genes may be altered as the result of toxicity rather than being involved in initiating pathogenesis – several previous gene-array studies assessed gene expression days or weeks after TCDD exposure;

(b) some dioxin-responsive genes elicit biochemical or cellular changes that are not sufficient to cause lethality; these responses are likely adaptive changes or biomarkers of exposure;

(c) some dioxin-responsive genes whose expression is altered in vitro may not correspond to in vivo responses; cells in culture generally are resistant to TCDD toxicity even though they exhibit many transcriptional responses to dioxins; furthermore, in vitro and in vivo expression profiles often are not in accord;

The key message from these studies – Without a strategy to discriminate genes that are mechanistically responsible for specific toxic endpoints from all other dioxin- responsive genes we are just cataloging dioxin-responsive genes. Substantial power can be achieved by combining large-scale gene expression approaches with model systems whose altered AHR structure correlates with susceptibility to specific endpoints of dioxin exposure. This strategy will be critical to elucidating the mechanisms underlying lethal effects of dioxins.

19 Potential model systems to identify genes relevant to dioxin lethality and associated toxicities

A characteristic feature of TCDD toxicity is wide variation in sensitivity within and among species. This variable response is highly endpoint-dependent and thus is a potentially useful model to combine with large-scale gene expression approaches to elucidate underlying mechanisms of TCDD lethality.

Humans To elucidate AHR-mediated mechanisms of TCDD lethality, humans would be a poor choice as a model. First, there are no recorded instances of human fatalities from acute dioxin exposure. Structurally, human AHR possesses the prerequisites for dioxin sensitivity; however, humans appear to be resistant to most TCDD toxicities when compared with standard laboratory species such as mice or guinea pigs. A valid LD50 for humans is unknown; therefore, it may not be reasonable to assume that humans are less sensitive than other laboratory animals such as the H/W rat, DBA/2J mouse, or hamster. To further understand the mechanistic role of the AHR in human responses to TCDD, a “AHR-humanized” mouse model was engineered by knocking human AHR (hAHR) into a C57BL/6J mouse background (Moriguchi et al., 2003). Mice that express human AHR were less susceptible to teratogenicity from TCDD than were DBA/2J mice even though the hAHR and the native AHR in DBA/2J mice have similar affinity for TCDD (Table 1.3). While this response agrees with epidemiologic evidence that humans are a TCDD- resistant species, it is unclear if this response is a true measure of human AHR function or if it reflects an inability of human AHR to function normally in the mouse context. Secondly, alteration in human AHR structures does not clearly correlate with altered susceptibility to TCDD lethality. Several polymorphisms in the AHR gene have been detected, especially in exon 10 which encodes the TAD (Harper et al., 2002). Except for an arginine to lysine substitution at codon 554, none of the identified polymorphisms alone appear to alter receptor function. The Arg554Lys is significantly associated with survival in patients with soft tissue sarcoma (Berwick et al., 2004). A combination of the Arg554Lys and a second Val570Ile polymorphism were found to

20 impair CYP1A1 induction (Wong et al., 2001). However, these polymorphisms are not known to correlated with TCDD lethality and, of course, it would be unethical to attempt to determine the effect of AHR polymorphisms on severe forms of dioxin toxicity directly in humans. Thus, we do not have a clear method to distinguish which TCDD- induced gene responses are relevant to toxicity from among all TCDD responses using a human model directly.

Animal models Animal models allow direct investigation of the role of variability in AHR structure on responses to TCDD. Marked inter-species and intra-species differences in sensitivity to dioxin exist, so care must be taken to select the appropriate model to study. The largest inter-species difference in sensitivity to TCDD exists between the guinea pig and hamster while, the largest intra-species difference exists between the Han/Wistar (Kuopio) (H/W) rat strain versus the Long-Evans (Turku AB) (L-E) rat strain. The molecular and genetic basis for the wide species variability in TCDD toxicities resides mainly in AHR polymorphisms (Table 1.3). Thus, characterization of altered AHR structure and subsequent response to TCDD, in diverse species, could greatly aid in understanding mechanisms of dioxin lethality and associated toxicities.

21

Table 1.3. Comparison of the pharmacogenetics of dioxin responses in vivo.

Mouse Rat Guinea Pig vs. Hamster (C57BL/6 vs. DBA/2J) (L-E vs. H/W)

Genetic Distinct species Both Inbred Inbred vs. Outbred background

LD50 (µg/kg) ~ 1-2 vs. 3,000-5,000 180 vs. 2600 ~10-50 vs. >9,600

TCDD lethality ~ 3,000-fold ~ 10-fold ~ 1,000-fold difference

Inheritance Not applicable Susceptibility is a Resistance is a dominant trait; dominant trait; 1 gene involved 2-3 genes involved

TCDD ~7-fold difference; ~ 2-fold difference; Similar; elimination (t½) 94 vs. 11-15 days ~11 days 20.9 vs. 21.9 days

AH receptor Hamster’s AHR is DBA/2 strain: Ala-to- 106 kDa vs. 98 larger & contains Val substitution at kDa; twice as many codon 375 of LBD H/W: 2 mutations glutamine residues (Ema et al., 1994; Okey et in TAD of AHR (Korkalainen et al., 2000; al., 2005; Okey et al., (see discussion 1989; Poland et al., 1994) Korkalainen et al., 2001) below)

AHR affinity no apparent ~ 10-fold difference; Similar affinity for for TCDD or differences DBA/2 strain: Ahrd AHREs & ligand; AHREs allele ‘low affinity’ L-E: 2-3 fold higher for TCDD concentration of TCDD binding sites

CYP1A enzyme Effective dose similar; ~ 10-fold difference Similar induction magnitude variable in effective dose

Total body fat ~4-fold difference ~2-fold difference Similar

Modified and updated from (Pohjanvirta and Tuomisto, 1994).

22 After comparison of the pharmacogenetics of the dioxin response in available animal models, the rat model is the most suitable model to study the mechanistic role of AHR genetic variability in lethal responses to TCDD since the rat model:

(1) is a within-species model; avoids confounding effects of differences in cross-species genetics (2) has a very large magnitude of difference in lethality and associated toxicities between L-E vs. H/W rat strains (3) specific TCDD toxicity endpoints (lethality, hepatotoxicity, and wasting) differ between H/W and L-E rats, while non-lethal endpoints are conserved (CYP1A1 enzyme induction); in the mouse model all endpoints are altered (4) the TCDD-resistant H/W rat has a deletion in its AHR within a key region which mediates gene transcription

The TCDD-resistant H/W rat model The ~1,000-fold difference in lethality between L-E vs. H/W rats represents the largest intra-species difference known. In addition to acute lethality, responses that differ most dramatically between H/W and other rat strains include (Pohjanvirta et al., 1993), wasting syndrome (Pohjanvirta et al., 1987), increased serum tryptophan (Unkila et al., 1994b), hepatotoxicity (Pohjanvirta et al., 1989a; Simanainen et al., 2002; Simanainen et al., 2003), liver tumor promotion (Viluksela et al., 2000), hyperbilirubinemia and accumulation of biliverdin in the liver (Niittynen et al., 2003). These differences between H/W and L-E rats are termed ‘Type II’ responses. Despite the H/W rat’s extraordinary resistance to acute TCDD lethality, H/W rats are sensitive to specific biochemical responses to TCDD and do respond in a magnitude similar to sensitive rats. These responses are termed ‘Type I’ responses and include: CYP1A1 induction (Pohjanvirta et al., 1988; Viluksela et al., 1998b), thymic atrophy (Pohjanvirta et al., 1989a) and fetotoxicity (Huuskonen et al., 1994). The similarities and differences between L-E and H/W rats are further summarized in Table 1.4.

23 Table 1.4. Comparison of L-E vs. H/W rat responses after treatment with TCDD.

H/W L-E Reference

Similar (“Type I”) responses Pharmacokinetics: disposition, Similar (e.g. t½ Similar (e.g. t½ (Pohjanvirta et al., 1990b) elimination, & metabolism 21.9 days) 20.8 days) AHR binding affinity No differences; 2.4-5.6 nM (Pohjanvirta et al., 1999) DNA binding of AHR complex No differences (Pohjanvirta et al., 1999) Enzyme induction (CYP & UGT) Similar dose: 0.1-1 µg/kg (Pohjanvirta et al., 1988; Viluksela et al., 1998b) Thymic atrophy Similar dose (Pohjanvirta et al., 1989a) Teratogenicity Similar dose (Huuskonen et al., 1994) Rapid aversion to select foods Similar dose (Tuomisto et al., 2000) Serum thyroid hormone levels Reduced by Similar dose (Pohjanvirta et al., 1989a) Serum melatonin levels Reduced by Similar dose (Linden et al., 1991) ARNT splice variants levels Similar before & after TCDD (Korkalainen et al., 2003) AHRR expression levels Similar before & after TCDD (Korkalainen et al., 2004) PEPCK Reduced by Similar dose (Viluksela et al., 1999)

Different (“Type II”) responses Constitutive AHR mRNA levels Similar to LnC 2-fold greater (Pohjanvirta et al., 1998) sensitive rat AHR binding capacity (Bmax) 23-38 fmol/mg 59-84 fmol/mg (Pohjanvirta et al., 1999) AHR gene 2 mutations wildtype (Pohjanvirta et al., 1998) AHR mRNA 3 variants wildtype (Pohjanvirta et al., 1998) AHR protein 98 kDa 106 kDa (Pohjanvirta et al., 1999) Hepatic ARNT expression levels Similar to 3-fold greater (Pohjanvirta et al., 1993; sensitive SD rat Unkila et al., 1994b) Acute lethality, LD50 (male) >9600 µg/kg 17.7 µg/kg (Pohjanvirta et al., 1987) Wasting syndrome (Anorexia) Transient Irreversible (Pohjanvirta et al., 1989a) Liver damage (↑ASAT, ↑ALAT) Mild Severe (Viluksela et al., 2000) Tumor promotion (chronic Mild 100-fold more (Niittynen et al., 2007; exposure) sensitive Pohjanvirta et al., 1993; Viluksela et al., 1998a) Rank order of potency of HAHs hexa > hepta > tetra, penta > (Pohjanvirta et al., 1989b) (chlorination) penta > tetra hexa > hepta Centrally administered TCDD susceptible (Unkila et al., 1993) Serotonin turnover Increased (Unkila et al., 1994a) Hyperbilirubinemia Increased (Unkila et al., 1994b) Serum tryptophan Increased (Niittynen et al., 2007; Pohjanvirta and Tuomisto, 1994) Steatosis (fatty degeneration) Mild Severe (Pohjanvirta et al., 1989a) Serum free fatty acids Increased (Niittynen et al., 2007)

24

Based on breeding studies, the AHR is the key difference between H/W versus L- E rats that is important for dioxin lethality (Tuomisto et al., 1999). The AHR from the L- E rat is 106 kDa, identical to that reported for other common laboratory rat strains, whereas the AHR from the H/W rat is only 98 kDa (Pohjanvirta et al., 1999). Molecular analysis of AHR gene structure in the dioxin-resistant H/W rat strain revealed two point mutations (Figure 1.6) (Pohjanvirta et al., 1998). The first mutation within exon 10 leads to a Valine to Alanine substitution. Since this is a conservative substitution and located in a hypervariable region it is unlikely to play a major role in lethality differences. Instead, the second point mutation at the intron/exon-10 boundary is critical because it destroys a splice site. This destruction leads to use of 3 alternative cryptic splice sites, potentially creating 3 alternative transcripts and 2 protein products that are smaller than the wildtype AHR in dioxin-sensitive rat strains. Genetic studies strongly implicate the restructured TAD of the H/W rat’s AHR as the key determinant of resistance to TCDD lethality (Pohjanvirta et al., 1999; Tuomisto et al., 1999).

Dioxin resistance in the H/W rat segregates genetically with the AHR locus as a dominant trait (Tuomisto et al., 1999). The F1 offspring of a mating between L-E and H/W rats express both wildtype and H/W AHR proteins and are resistant to acute TCDD lethality and associated toxicities (Pohjanvirta et al., 1999; Pohjanvirta et al., 1998).

Backcrossing of F1 rats with L-E rats, combined with phenotyping for TCDD sensitivity/resistance by TCDD challenge and subsequent AHR genotyping, generated three additional rat lines (Figure 1.7). The first line, Line-A (LnA) is only slightly more sensitive to TCDD than the H/W rat due to the minor contribution of the wildtype form of an unknown ‘gene B’ (genotype AHRhw/hwBwt/wt). The AHR is the main determinant of resistance whereas ‘gene B’ makes only a minor contribution (Tuomisto et al., 1999). The second line, Line-C (LnC) is sensitive to TCDD toxicities and genotypically (AHRwt/wt Bwt/wt) is identical to L-E rats. A third line, Line-B (AHRwt/wt Bhw/hw) is slightly more

25

abc * Genomic 5’ 12345 67 89 10* * 11 3’ structure Point mutation (1520) Point mutation (2454) 3 alternative splice sites (*)

mRNA Dioxin Sensitive: AHR-WT Protein

2454 AHRE LBD TAD bHLH PAS Q-Rich NC423 aa 45 aa 1 to 9 10 11 AAA (n) 808

mRNA Dioxin Resistant: AHR-H/W Protein

DV a 2326 DV AAA 1 to 9 10 11 (n) N 380 aa 45 aa C -129 nt 497 766 SIV 2454 (V to A) b 1 to 9 10 11 AAA(n) +29 nt IV N 423 aa 7 C LIV 2454 c AAA 1 to 9 10 11 (n) 497 808 +155 nt (V to A)

= translation termination

Figure 1.6. Domain structure of the AHR and the region deleted in the transactivation domain of the H/W (Kuopio) rat due to alternative splicing. The ligand-binding domain (LBD), dimerization domain and AHRE-binding domain are near the N-terminus, in a conserved region that does not differ between H/W rats and rats with WT AHR. An exonic mutation at nucleotide 1520 from the start codon causes a conservative amino acid substitution that is not believed to significantly affect receptor function. A second mutation at nucleotide 2454 in the intron-10/exon-10 boundary disrupts the normal splice site, leading to potential use of the 3 cryptic splice sites “a”, “b”, and “c” and giving rise to three possible mRNAs and 2 possible protein products. The alternative splice variant termed DV results from use of splice site “a” which leads to a deletion of 129 nucleotides from exon-10 producing a protein lacking 43 amino acids. The IV splice variant results from use of splice sites “b” or “c”. These splice sites lead to an addition of 29 nt (SIV, short insertion variant) or 134 nt (LIV, long insertion variant) from intron-10; however, identical addition of a translation termination site produces identical protein products. The IV protein has a net loss of 38 amino acids (insertion of 7 amino acids from the intron and deletion of 45 amino acids encoded by the last exon). Presented as published (Moffat et al., 2007).

26 resistant than the sensitive L-E rat due to the contribution of the resistant form of ‘gene B’. These new rat lines and their prototype parental strains (L-E and H/W) span a range of global genetic similarity but lie at extremes in the spectrum of TCDD sensitivity phenotypes. Together these rat strains and lines offer the opportunity to discriminate AHR-mediated signaling events that are key to dioxin toxicity from irrelevant strain- specific events.

Line-C Line-A (AHRwt/wtBwt/wt) Line-B (AHRhw/hwBwt/wt) wt/wt hw/hw (AHR B ) F (L-E x H/W) Long-Evans 1 Han/Wistar (Turku AB) (Kuopio)

SENSITIVE RESISTANT

10100 1,000 10,000

LD50 (µg/kg)

Figure 1.7. Genetic differences in sensitivity to acute TCDD lethality in rat strains and rat lines. Long-Evans (Turku AB) (L-E) and Line-C represent a rat strain and a rat line that are sensitive to TCDD lethality (as are most common laboratory rat strains). Han/Wistar (Kuopio) (H/W) rats and Line-A rats are extraordinarily resistant to lethal effects of TCDD. Line-A and Line-C were derived by our collaborators in Finland from a multi-generation breeding study designed to establish the number of genes involved in dioxin toxicity; they were obtained by initially crossing L-E and H/W rats, then selecting for sensitivity/resistance by TCDD challenge. This study revealed a second gene (“Gene B”, identity unknown) that exerts some influence on susceptibility to TCDD but the AHR is most important (Tuomisto et al., 1999). Line-A and Line-C both are homozygous for the wildtype “Gene-B”.

27 AIMS OF MY THESIS RESEARCH

LONG-TERM OBJECTIVE: To determine the molecular mechanism by which dioxins cause lethality. GENERAL HYPOTHESIS: Lethality from dioxin-like chemicals results from dysregulation of a subset of dioxin-responsive genes under control of the AH receptor.

RATIONALE: It is firmly established that toxic effects of dioxin-like chemicals are initiated by their binding to the AHR. The AHR functions as a ligand-dependent transcription factor which regulates expression of specific genes. Because toxicity depends upon the AHR and because the AHR’s function is to regulate gene expression, it is highly probable that TCDD lethality and other major toxicities are due to dysregulation of a specific gene or set of genes that lie under AHR control.

OVERALL RESEARCH STRATEGY: H/W rats are resistant to lethality from TCDD because the TAD deletion prevents the AHR from dysregulating specific genes that are essential in the mechanism of toxicity. Since the AHR transactivation function is essential for major forms of dioxin toxicity and since H/W rats with the TAD deletion are highly resistant to dioxin toxicity, it is logical for my thesis research to use the dioxin- resistant H/W rat model to identify mechanisms that trigger dioxin toxicity. In experiments reported in CHAPTER 2, I characterized tissue expression and transactivational activity of variant forms of AHR in dioxin-resistant rats versus dioxin- sensitive rats. I then used large-scale gene array techniques to compare the sets of mRNAs and microRNAs that are TCDD-responsive in dioxin-sensitive rats versus the sets that respond in genetically dioxin-resistant rats (CHAPTERS 3 & 4) (Figure 1.8). Together these studies investigate whether altered AHR structure can affect the AHR’s ability to regulate gene expression in relation to TCDD lethality.

28 SPECIFIC AIMS OF EACH THESIS PROJECT

CHAPTER 2: Aryl hydrocarbon receptor (AHR) splice variants in the dioxin-resistant rat: tissue expression and transactivational activity

Rationale ƒ The AHR is the known mediator of TCDD toxicity; therefore, any alterations in AHR structure or AHR expression levels could alter the receptor’s ability to drive transcription of genes central to TCDD toxicities ƒ AHR expression levels (transcript, protein & activity) may be influenced by its own ligand ƒ The dioxin-resistant H/W rat can potentially express 3 alternatively-spliced transcripts and 2 possible protein products of the AHR ƒ It is unknown which splice variant is expressed in dioxin-resistant rats and whether TCDD affects expression of the AHR splice variants in a way that differs among the variant transcripts

Goals To determine: (1) which AHR splice variants are expressed in tissues of dioxin-sensitive and dioxin- resistant rats (2) if TCDD affects expression of the AHR splice variants in a way that differs among the variant transcripts (3) if the altered AHR TAD structure affects transactivation function

Outline This chapter describes quantitation of AHR splice variants both in the constitutive state and after TCDD treatment in multiple tissues from 2 dioxin-sensitive rat strains/lines and 3 dioxin-resistant rat strains/lines. Subsequently, to determine if variant forms of the AHR have different abilities to transactivate gene expression we measured

29 intrinsic transactivation activity among the splice variants. Lastly, to determine if AHR TAD structure predicts responsiveness to TCDD, we compared the predicted mRNA and protein secondary structures in the TAD among multiple species and strains which vary in their susceptibility to TCDD lethality.

Contribution to theme of the thesis Since the AHR is the first essential component in the dioxin toxicity cascade, understanding which form of AHR is expressed in sensitive versus resistant rats and whether splice-variant expression is altered by TCDD is critical to discerning whether altered TAD structure affects gene transcription and ultimate susceptibility to dioxin lethality.

30 CHAPTER 3: Dioxin lethality: aryl hydrocarbon receptor-regulated gene expression in a resistant rat model

Rationale ƒ Toxic effects of dioxins are initiated by binding to the AHR ƒ The AHR is a ligand-dependant transcription factor which regulates expression of specific genes ƒ Specific transcriptional events that lead to dioxin lethality have not been identified ƒ Dioxin-resistant rats express an AHR with a large portion of its TAD deleted ƒ We hypothesize that dioxin-resistant rats are resistant to lethality from TCDD because the TAD deletion prevents the AHR from dysregulating specific genes that are essential in the mechanism of toxicity.

Goals To determine which: (1) mRNA expression responses are altered after a short-term TCDD exposure (2) mRNAs are affected by TCDD differently in dioxin-sensitive rats (expressing wildtype AHR) versus dioxin-resistant rats (expressing AHR with the TAD deletion) (3) co-regulated genes are part of biological processes & pathways relevant to lethality (4) TCDD-responsive genes potentially are regulated by the AHR

The ultimate goal is to define a hierarchy of AHR-mediated dioxin-responsive genes likely to be central to lethality and associated toxicities by dioxins.

Outline This chapter first describes quantitation of early gene expression responses to TCDD using the Affymetrix RAE230A GeneChip®, then describes classification of TCDD responses as “Type I” or “Type II” based on the dioxin-resistant rat model. Subsequently, to predict biological processes and pathways leading to dioxin lethality, we performed in silico Gene Ontological and pathway analyses on Type II responsive co- regulated genes. Further in silico analysis was used to predict AHREs within regulatory

31 regions of candidate genes. Lastly, select candidate gene responses to TCDD were measured by real-time RT-PCR to discern if TCDD-altered mRNA expression levels in the rat model are genuine; additionally we tested whether changes in expression of selected genes are dependent on the AHR using the Ahr-null mouse model.

Contribution to theme of the thesis Our hypothesis states that H/W rats are resistant to lethality from TCDD because the TAD deletion prevents the AHR from dysregulating specific genes that are essential in the mechanism of toxicity. This project is designed to discern which genes respond differently to TCDD between dioxin-sensitive versus dioxin-resistant rats to identify candidate genes that may be responsible for dioxin lethality.

32 CHAPTER 4: Micro-RNAs in Adult Rodent Liver are Refractory to Dioxin Treatment

Rationale ƒ Large-scale transcriptomic studies reveal that transcript levels of many genes are decreased (downregulated) by TCDD ƒ Little is known about mechanisms by which dioxin-like chemicals suppress mRNA levels ƒ microRNAs (miRNAs), recently emerged as powerful negative regulators of mRNA levels in several systems ƒ miRNAs regulate diverse biological processes, including fat storage, insulin secretion, drug-metabolism, apoptosis, cell growth, tumorigenesis, and death. These same processes are altered by dioxin exposure ƒ the mRNA level of at least one AHR-regulated gene (Cyp1b1) previously has been reported to be decreased by an miRNA Hypothetically, dioxins could upregulate or downregulate specific microRNAs which, in turn, could alter degradation, efficiency of translation or transcription of target mRNAs.

Goals To determine if: (1) AHR genotype itself affects constitutive expression of microRNAs in mouse or rat (2) TCDD affects microRNA levels and, if so, if this response is dependent on the AHR (3) TCDD affects microRNA levels differently in animals that are sensitive to dioxin toxicity versus those that are dioxin-resistant Outline This chapter describes our assessment of the in vivo effect of TCDD on microRNA levels in adult liver at multiple time points after TCDD treatment using two different microRNA array platforms along with quantitative RT-PCR. In addition we used qRT-PCR to test the effect of TCDD on microRNA levels in thymus, lung, and kidney in vivo as well as the effect in hepatoma cells from mouse and rat in culture. We focused on hepatic microRNAs because liver displays a broad spectrum of mRNAs that

33 are downregulated by dioxins or by AHR genotype (Tijet et al., 2006) and because liver is a prime site of dioxin toxicity (Niittynen et al., 2007). As an adjunct to laboratory measurements, we used bioinformatic techniques to: (a) predict which mRNAs are likely to be targets for specific microRNAs: (b) determine if specific miRNA sequences contain AHR binding sites; and (c) assess whether mRNA for the AH receptor itself might be a target for any microRNA.

Contribution to theme of the thesis Our overall hypothesis states that H/W rats are resistant to lethality from TCDD because the TAD deletion prevents the AHR from dysregulating specific genes that are essential in the mechanism of toxicity. This project is designed to discern if microRNAs are involved in the mechanism of downregulation of gene expression by dioxins as well as to determine if TCDD affects miRNA expression differently between dioxin-sensitive versus dioxin-resistant rats.

34

Other D TCD proteins

aptAerIP2 AHR Ch AHRR Ligand Binding & ARNT Translocation AHR ARNT AHRE Transcription

?

microRNAs ?? specific mRNAs r 4 Chapte

Adaptive: Cyps mRNAs r 3 Developmental Chapte

Dioxin Toxicity

Figure 1.8. Overview of thesis research projects.

Chapter 2: Aryl hydrocarbon receptor (AHR) splice variants in the dioxin-resistant rat: tissue expression and transactivational activity

Chapter 3: Dioxin lethality: aryl hydrocarbon receptor-regulated gene expression in a resistant rat model

Chapter 4: Micro-RNAs in adult rodent liver are refractory to dioxin treatment

35

CHAPTER 2: Aryl Hydrocarbon Receptor (AHR)

Splice Variants in the Dioxin-Resistant Rat: Tissue

Expression and Transactivational Activity

Ivy D. Moffat, Steven Roblin, Patricia A. Harper,

Allan B. Okey, and Raimo Pohjanvirta

As published in: Mol Pharmacol 72:956-966 (2007)

Supplementary Data: http://molpharm.aspetjournals.org/cgi/content/full/mol.107.037218/DC1

In this project, the roles of collaborators were:

Collaborator Contribution

Ivy D. Moffat Study Design, in vivo & in silico experiments

Steven Roblin In vitro experiments

Patricia A. Harper Co-principal Investigator

Allan B. Okey Co-principal Investigator

Raimo Pohjanvirta Treatment & supply of tissues; Principal Investigator

36

ABSTRACT

The AHR locus encodes the aryl hydrocarbon receptor (AHR), a transcriptional regulator of multiple drug-metabolizing enzymes and mediator of toxicity of dioxin-like chemicals. The Han/Wistar(Kuopio) rat strain (H/W) is remarkably resistant to lethal effects of 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) due to a point mutation in the exon/intron 10 boundary in AHR genomic structure which leads to use of 3 alternative cryptic splice sites, potentially creating 3 alternative transcripts and 2 protein products. The deletion variant (DV), which lacks 43 amino acids in the transactivation domain, has the highest intrinsic transactivation activity in vitro; amino acids 766-783 suppress transactivation function. However, DV expression levels in H/W rats in vivo are low in liver, lung, thymus, kidney and testis; insertion-variant mRNAs (IVs) are the dominant mRNA forms in H/W rats where wildtype AHR mRNA is undetectable. In dioxin- sensitive rat strains and lines that are homozygous for wildtype AHR alleles, wildtype AHR mRNA is the most abundant transcript but some IV transcripts are detectable. TCDD treatment in vivo increases transcript levels for both the DV and IVs in H/W rats and increases wildtype transcript levels in dioxin-sensitive rats but does not alter which transcript forms are expressed. In silico modeling indicates that the DV mRNA has lost considerable secondary structure whereas at the protein level the transactivation domain of the IV in the dioxin-resistant H/W rat has greater α-helical content and a more hydrophobic terminus than wildtype AHR which may produce a protein conformation that is less amenable to interaction with other regulatory proteins.

37

INTRODUCTION

The AHR locus encodes a ligand-activated transcription factor, the aryl hydrocarbon receptor which plays key roles in: 1) adaptive metabolism of xenobiotics; 2) developmental and physiological signaling; and 3) toxic responses to dioxin-like environmental pollutants (Fernandez-Salguero et al., 1996; Nebert et al., 2000; Okey et al., 2005; Walisser et al., 2004b). Agonist ligands such as TCDD convert the AHR into a heterodimeric complex with the ARNT protein which then regulates expression of specific genes by binding directly to AH responsive elements (AHREs) (Denison et al., 1988; Hankinson, 2005) or indirectly via binding to adaptor proteins at the AHRE-II site (Boutros et al., 2004; Sogawa et al., 2004). The AHR can either enhance gene transcription or inhibit it (Riddick et al., 2004; Tijet et al., 2006). The AHR is a member of the bHLH-PAS protein family (Gu et al., 2000) and, like other transcriptional activators, is composed of modular domains that function independently. The highly conserved N-terminus contains the bHLH-PAS domain which mediates nuclear localization, heterodimerization with ARNT and AHRE site- recognition. The C-terminal domain contains a modular transactivation domain (TAD) comprised of three distinct subdomains, acidic, glutamine (Q)-rich, and proline-serine- threonine (P-S-T)-rich regions (Supplemental data Figure S1). Marked inter-species and intra-species differences exist in the AHR C-terminal mRNA and protein sequences but the sequence, per se, of the transactivation domain is not always a faithful predictor of sensitivity to dioxins (Korkalainen et al., 2000; Korkalainen et al., 2001). The Han/Wistar(Kuopio) (H/W) rat strain is extraordinarily resistant to lethal effects of TCDD with an LD50 that is more than 1000-fold higher than in dioxin-sensitive rat strains such as Long-Evans (L-E) (reviewed in: (Pohjanvirta and Tuomisto, 1994)). Dioxin resistance in the H/W rat segregates genetically with the AHR locus and is a dominant trait (Pohjanvirta and Tuomisto, 1994; Tuomisto et al., 1999). Our molecular analysis (Pohjanvirta et al., 1998) of AHR gene structure in the dioxin- resistant H/W rat strain revealed a point mutation at the exon/intron-10 boundary which

38 leads to use of 3 alternative cryptic splice sites, potentially creating 3 alternative transcripts and 2 protein products that are smaller than the wildtype (WT) AHR from dioxin-sensitive rat strains (Figure 2.1). The restructured TAD C-terminus in H/W rats appears to exert selective effects on gene transcription. Despite the large deletion in the TAD of H/W rats, CYP1A1 and several genes that are well-known to be AHR-regulated and dioxin-inducible continues to respond normally to induction by TCDD (Okey et al., 2005). However, expression array experiments in our laboratories indicate that several other genes respond differently in rats with the H/W AHR genotype than in rats with WT AHR (M. Franc, I. Moffat, P. Boutros, J. Tuomisto, J.T. Tuomisto, R. Pohjanvirta, A. Okey, in preparation; and I. Moffat, P. Boutros, J.T. Tuomisto, R. Pohjanvirta, A. Okey, in preparation). Potentially those genes whose response to TCDD differs between sensitive strains and resistant strains are central to the mechanism of dioxin toxicity. The dramatic difference in dioxin susceptibility and the variant AHR gene structure between rats with WT AHR versus H/W rats, offers a unique opportunity to better characterize the impact of the AHR’s TAD structure in regulating gene expression and dioxin toxicity. Our overall hypothesis is that H/W rats are resistant to lethality from TCDD because the TAD deletion prevents the AHR from dysregulating specific genes that are essential in the mechanism of toxicity. Our goals in the current study were to: (1) determine which AHR splice variants are expressed in tissues of dioxin-sensitive and dioxin-resistant rats; (2) determine if TCDD affects expression of the AHR splice variants in a way that differs among the variant mRNAs; (3) determine if the altered TAD structure affects transactivation function.

39

abc * Genomic 5’ 12345 67 89 10* * 11 3’ structure Point mutation (1520) Point mutation (2454) 3 alternative splice sites (*)

mRNA Dioxin Sensitive: AHR-WT Protein

2454 AHRE LBD TAD bHLH PAS Q-Rich NC423 aa 45 aa 1 to 9 10 11 AAA(n) 808

mRNA Dioxin Resistant: AHR-H/W Protein

DV a 2326 DV AAA 1 to 9 10 11 (n) N 380 aa 45 aa C -129 nt 497 766 SIV 2454 (V to A) b AAA 1 to 9 10 11 (n) +29 nt IV N 423 aa 7 C LIV 2454 c AAA 1 to 9 10 11 (n) 497 808 +155 nt (V to A)

= translation termination

Figure 2.1. Domain structure of the AHR and the region deleted in the transactivation domain of the H/W(Kuopio) rat due to alternative splicing. The ligand-binding domain (LBD), dimerization domain and AHRE-binding domain are near the N-terminus, in a conserved region that does not differ between H/W rats and rats with WT AHR. An exonic mutation at nucleotide 1520 from the start codon causes a conservative amino acid substitution that is not believed to significantly affect receptor function. A second mutation at nucleotide 2454 in the exon-10/intron-10 boundary disrupts the normal splice site, leading to potential use of the 3 cryptic splice sites “a”, “b”, and “c” and giving rise to three possible mRNAs and 2 possible protein products. The alternative splice variant termed DV results from use of splice site “a” which leads to a deletion of 129 nucleotides from exon-10 producing a protein lacking 43 amino acids. The IV splice variant results from use of splice sites “b” or “c”. These splice sites lead to an addition of 29 nt (SIV, short insertion variant) or 134 nt (LIV, long insertion variant) from intron-10; however, identical addition of a translation termination site produces identical protein products. The IV protein has a net loss of 38 amino acids (insertion of 7 amino acids from the intron and deletion of 45 amino acids encoded by the last exon). (Modified from: (Okey et al., 2005; Pohjanvirta et al., 1998)).

40

MATERIALS & METHODS

Animal Treatments and Isolation of Total RNA. We measured expression of each AHR splice variant and the effect of TCDD on expression of the variants in two dioxin- sensitive rat strains/lines (L-E and Line-C), two dioxin-resistant rat strains/lines (H/W and Line-A) and in F1 offspring from the L-E x H/W cross. L-E is the prototype dioxin- sensitive strain homozygous for AHRWT and H/W is the prototype dioxin-resistant strain homozygous for AHRH/W (Okey et al., 2005; Pohjanvirta and Tuomisto, 1994). Line-A (LnA, resistant) and Line-C (LnC, sensitive) lines were produced by multiple generations of crosses beginning with L-E and H/W rats, combined with phenotyping for dioxin sensitivity/resistance by TCDD challenge (Tuomisto et al., 1999). All animals were from the breeding colony of the National Public Health Institute, Division of Environmental Health, Kuopio, Finland. At age 10-12 weeks male rats were given 100 µg/kg TCDD or the corn oil vehicle by gavage, then euthanized by decapitation after 19 or 96 h. There were 4 rats per treatment group. Total RNA was extracted from liver using Qiagen RNeasy kits according to the manufacturer’s instructions. For kidney, lung, testis, and thymus, total RNA was isolated using TRIzol reagent (Invitrogen, Carlsbad, CA) and subsequently treated with DNase (MBI Fermentas, Burlington, Canada). RNA quality was assessed with an Agilent Bioanalyzer prior to further experiments.

Real-Time Quantitative RT-PCR. Three allele-specific primer/probe sets were designed using IDT SciTools Primer Quest software (http://scitools.idtdna.com/Primerquest/) to amplify mRNA encoding the 2 variant forms of the H/W receptor protein in addition to the WT receptor protein. Since the identical stop codon is used in SIV mRNA and LIV mRNA, both mRNA variants encode the identical protein product. Thus a single PCR primer set was used to quantitate SIV mRNA + LIV mRNA and simply termed IVs. See Table 2.1 for primer and probe sequences. The PUNS (Primer-UniGene Selectivity Testing) program (http://okeylabimac.med.utoronto.ca/PUNS; (Boutros and Okey, 2004)) was applied to ensure that primers were specific to the intended transcript sequence. Probes were

41 labeled with a reporter fluorescent dye FAM (6-carboxyfluorescein) at the 5′-end and a quencher-fluorescent dye Iowa Black FQTM at the 3’-end. Primers and probes were synthesized by IDT (Coralville, IA). Specificity of each primer/probe set was confirmed by: (1) sequencing of purified PCR products (MinElute, Qiagen, Mississauga, Canada) amplified from liver by real-time PCR; (2) positive real-time PCR amplification from a construct containing the specific cDNA of one variant; and (3) negative real-time PCR amplification from constructs containing the other AHR cDNAs. Total RNA (2 µg) was

reverse transcribed using oligo-dT primer, p(dT)15 (Roche Applied Science, Laval, Canada) and Superscript II RNA polymerase (Invitrogen), according to the manufacturer's instructions. Real-time PCR was performed on a Stratagene MX4000 real-time PCR system using primers, probe, and Brilliant QPCR Mastermix (Stratagene, Cedar Creek, TX) according to the manufacturer’s instructions. PCR conditions were established as follows: after 10 min at 95°C, 40 cycles were performed at 95°C for 30 sec/each and 59°C for 1 min. A 10-fold dilution series of each purified PCR product (WT or IVs or DV) was used as an external standard to assess transcript levels for each of the 3 forms of the AHR mRNA. All dilution series were performed in triplicate. Total levels of AHR receptor transcript were compared between strains/lines by a one-way analysis of variance followed by Bonferroni post hoc analysis. Differences between total transcript levels were considered significant when p<0.05. For variables with two factors (treatment and AHR-variant type) comparisons were made within each strain by two-way ANOVA. For significant comparisons (p<0.05), Bonferroni post hoc analysis for detecting deviating groups was employed.

42

Table 2.1. Primer and probe sequences for real-time quantitative RT-PCR measurements of rat WT AHR and rat splice- variant transcripts.

Wildtype AHR Insertion-Variant AHR Deletion-Variant AHR

Sequence NM_013149 AF082125a/AF082126b AF082124 Forward ATGGTCAGTCCTCAGGCGTACTA ATGGTCAGTCCTCAGGCGTACTA AAC TCA CAG TCA GCC ATG TTT CAG Reverse AAT GCT CGG ACT CTG AAA CTT GC TCC CTG TAG AAA GCC CTT ATC TTG C ATATCAGGAAGAGGCTGGGCTTC Probe CCATGTCCATGTACCAGTGCCAGGCAGG CCATGTCCATGTACCAGTGCCAGGCAGG CCAGGCGAGGGAGGTGAGCAGCAGTC Product size 141 151 130

a Alternatively spliced short insertion variant b Alternatively spliced longer insertion variant

43

Expression Constructs Containing the AHR TAD and Reporter-Gene Constructs for Assessment of Transactivation Function. We generated expression constructs for the TADs of WT and the variant receptor forms, IV and DV. The TAD of each AHR variant was cloned from its respective full-length AHR expression construct and inserted into pFA-CMV (Stratagene) in-frame with the Gal4-DNA binding domain to create Gal4- AHRTAD chimeras. Deletion constructs were created in a two-step process using Gal4- AHRTAD-WT as a template. For example, to create Gal4-AHRTAD-WTΔaa766-773:

Step 1. PCR was used to amplify a region representing nucleotides 1270-2297 (where nucleotide 1 is the A of the ATG translation start site of WT rat AHR). In this fragment an artificial BamH I site was added immediately prior to nt 1270 and an artificial Nco I site was added immediately after nt 2297. Step 2. PCR was used to amplify a region representing nt 2314-2563. In this fragment, an artificial Nco I site was introduced immediately prior to nt 2314 and an artificial Hind III site immediately after nt 2563. The fragments from step one and step two were digested with BamH I/Nco I or Nco I/Hind III, respectively, ligated together at the Nco I site and inserted into pFA-CMV in-frame with the Gal4-DNA binding domain.

A similar process was used to create Gal4-AHRTAD-WTΔaa766-783, where the Nco I site was introduced at nt 2344 and for Gal4-AHRTAD-WTΔaa766-800, where the Nco I site was introduced at nt 2386. pFR-LUC (Stratagene) was used as a reporter of transactivation activity of the Gal4-AHRTAD chimera. pFR-LUC consists of five Gal4- DNA binding elements upstream of a basic TATA transcriptional promoter which drives the transcription of a firefly luciferase gene. The plasmid pRL-TK (Promega, Madison, WI) encoding renilla luciferase was used as a control for transfection efficiency.

In vitro Assay for Intrinsic TAD Activity. Rat hepatoma cells (5L cell line) were seeded in 12-well plates (1.0 x 105 cells/well) in αMEM medium supplemented with 10% FBS and incubated for 24 h. Cells were transfected with expression constructs and

44 reporter constructs using Lipofectamine Plus reagent (Invitrogen). Transfection conditions were as follows (per well): 2 µl Lipofectamine, 2 µl Plus reagent, 0.15 µg of one of the three Gal4-AHRTAD chimeras, 0.115 µg pFR-LUC and 0.035 µg pRL-TK in 900 µl of αMEM. Cells were exposed to transfection complex for 4 h; medium then was changed to FBS-supplemented growth medium and incubated for 20 h. Cells were harvested and assayed for both firefly and renilla luciferase activity using the Dual Luciferase Assay (Promega). Significant differences were determined by t-test.

Measurement of TAD Chimeric Protein Levels by Immunoblotting. Rat hepatoma cells (5L) were seeded in 60-mm plates (4.0 x 105 cells/plate) in αMEM medium supplemented with 10% FBS and incubated for 24 h. Cells were transfected (per plate) with 5 µl Lipofectamine and 5 µl Plus reagent (Invitrogen) and mixed with 0.875 µg of one of the three Gal4-AHRTAD chimeras in 1500 µl of exon-10 αMEM. Cells were exposed to transfection complex for 4 h; medium then was changed to FBS-supplemented growth medium and incubated for 20 h. Cells were scraped from the plate in 1.5 ml of ice cold PBS and centrifuged at 3000 x g for 1 min. Pelleted cells were lysed in 300 ml of ice cold immunoprecipitation assay buffer supplemented with protease inhibitor cocktail (Sigma-Aldrich, Oakville, Ontario) and incubated for 45 min with end-over-end mixing. Lysate was then centrifuged at 15,000 x g for 15 min and supernatant retained. Protein concentration in the supernatant was determined by the Bradford assay (Biorad, Hercules, CA). Fifty µg of total protein was added to LDS sample buffer containing reducing agent (Invitrogen). Proteins were separated via SDS-PAGE on pre-cast 4-12% NuPAGE Novex Bis-Tris gel (Invitrogen) then transferred to an Immobilon PVDF membrane (Millipore, Billerica, MA). After transfer, the membrane was washed twice in TRIS buffered saline containing 0.1% Tween 20 (TBS-T), then incubated in TBS-T containing 5% milk (TBST+M) for 1 h at room temperature. Anti-Gal4 antibody and anti-β-actin (Santa Cruz, Santa Cruz, CA) antibody were diluted in TBST+M and incubations were performed in TBST+M at 4°C overnight. Membranes were washed in 5 changes of TBS-T buffer over 30 min. Membranes were exposed to appropriate alkaline phosphatase-conjugated secondary antibody (Santa Cruz) for 1 h in TBST+M, then washed as described above. Antibody conjugates were visualized by ECF (enhanced

45 chemofluorescent) substrate assay as described in the manufacturer’s protocol (GE Healthcare, Chalfont St. Giles, Buckinghamshire, UK). Band intensities were determined using a Molecular Dynamics Storm 860 scanner and ImageQuant software (GE Healthcare).

In Silico Prediction of AHR Structure and Comparison Among Species and Strains. The minimum free energy of RNA formation (ΔG kcal/mol) and 20 suboptimal secondary RNA structures were predicted for each full-length AHR (WT, SIV, DV) using a modified thermodynamics-based Zucker algorithm and the nearest neighbor parameters implemented in RNASTRUCTURE (version 4.4, default parameters (Mathews et al., 2004)). Next, base pairing probabilities within each AHR sequence were predicted using the partition function algorithm within RNASTRUCTURE. The color-annotated base pairing probabilities for each sequence were overlaid on each of the 20 suboptimal structures. Finally, an optimal AHR structure was selected for each AHR type based on two criteria: (1) conserved predicted N-terminal TAD secondary RNA structure among all AHR-types since differences in primary RNA structures occur within the C-terminus; and (2) predicted structure with the greatest abundance of highly-probable base pairings. BioEdit software (version 7.0.5.3 (Hall, 1999)) was used to align AHR protein sequences from L-E rat (P41738), H/W rat (H/W-IV AAC35170/AAC35169, H/W-DV AAC35168), C57BL/6J mouse (AAL89728), DBA/2J mouse (AAL89732), guinea pig (AAR27312), hamster (AF275721), and human (P35869). Mean hydrophobicity profiles were generated using the Cornette scale mean method (Cornette et al., 1987) implemented in BioEdit. Cornette method assigns hydrophobicity values to the 20 amino acids based upon a compilation of experimental data from the literature. A window of 9 amino acids was moved along each TAD sequence, the hydropathy scores were summed along the window, and the average (the sum divided by the window size) was taken for each position in the sequence. The mean hydrophobicity value was plotted for the middle residue of the window. The consensus secondary protein structures obtained for these proteins were predicted using NPS@ (MLRC, DSC, and PHD algorithms using default parameters, Network Protein Sequence Analysis, (Combet et al., 2000)). NPS takes into account multiple predictions; its alignments are 70% correct for a three-state description

46 of secondary structure. This quality is obtained by a 'leave-one out' procedure on a reference database of proteins sharing less than 25% identity. Differences in post- translational modifications of the three TADs were predicted using PhosphoMotif Finder (Peri et al., 2003) which contains known kinase/phosphatase substrates and binding motifs curated from published literature. It reports the presence of any literature-derived motif in the query sequence.

47 RESULTS

Insertion Variants were the Predominant AHR Splice-Variant Transcripts Constitutively Expressed in Dioxin-Resistant Rats. In untreated dioxin-resistant H/W and LnA rats, the IVs were by far the predominant AHR transcripts constitutively expressed in all tissues examined – liver, lung, thymus, kidney and testis (Figure 2.2). A low level of DV transcript was present in each tissue but WT transcript was undetectable in any tissue from dioxin-resistant H/W or LnA rats. Virtually all AHR transcripts in dioxin-sensitive L-E and LnC rats represent the WT receptor. Some IVs mRNAs were present in L-E and LnC rats but at very low levels (Figure 2.2). In liver of F1 offspring from the L-E x H/W cross, each of the three transcripts was expressed (Figure 2.2); IV transcript levels were equal to WT transcript levels and a very low amount of DV was detected. Phenotypically, F1 rats are highly dioxin-resistant, but not as resistant as the H/W parent (Pohjanvirta and Tuomisto, 1994). The lower level of expression of the WT AHR in F1 rat livers compared with L-E rat livers cannot account for the sensitivity difference because the AHR expression level also is low in LnC rats (Figure 2.2) but their sensitivity is almost equal to that of L-E rats. Sensitivity to dioxin toxicity probably is not related to differing levels of total AHR transcripts between sensitive rats and resistant rats. Liver, kidney, and testis from sensitive L-E rats did express ~2.7 to 8.7-fold higher total AHR transcript levels than resistant rats but these higher levels were not observed in the sensitive LnC rats. In lung and thymus, total AHR transcript levels did not differ between sensitive L-E rats and resistant H/W or LnA rats; however, sensitive LnC rats expressed slightly higher transcript levels in lung and thymus than did any of the other strains/lines (Figure 2.2). We attempted to quantitate the specific protein products encoded by each splice variant. However, because the structures are highly similar between the DV and the IVs, we were not able to discriminate between these variant proteins either by two- dimensional gel electrophoresis or by immunoblotting with antibodies raised against peptide sequences that differ between the DV and IV (data not shown).

48 35 H/W-DV 30 H/W-IVs WT

) 25 -8 20

15

10

5

0 H/W LnA F1 LnC L-E H/W LnA LnC L-E Liver Lung

5 20 Testis

) 4

-9 ( expanded scale ) 15 3 (x 10 2 10 1

Transcript per( x10 RNA copies of total µg 5 0 H/W LnA LnC L-E

0 H/W LnA LnC L-E H/W L-E H/W LnA LnC L-E Thymus Kidney Testis

Figure 2.2. Constitutive AHR transcript expression levels in 5 tissues from dioxin- resistant (H/W, LnA and F1) and dioxin-sensitive (L-E and LnC) rats. Transcript levels for WT, IVs and DV were measured by real-time RT-PCR with primers specific to each variant as described in Materials and Methods (mean ± S.D., n=4). F1 rats are the offspring of a H/W x L-E mating.

Dioxin Treatment Increases Expression of AHR Splice-Variant Transcripts. In livers from dioxin-resistant H/W rats, both the IVs and the DV transcripts were significantly increased by treatment with TCDD. The magnitude of upregulation was higher for the DV than for the IVs; however, the IVs remained the most abundant transcripts in H/W liver after TCDD treatment (Figure 2.3).

Wildtype AHR transcript levels in dioxin-sensitive L-E rat livers also were significantly increased at 19 h or 96 h post-TCDD treatment. In L-E rats there was no significant effect of TCDD on transcript levels for the IVs which remained at very low levels. The DV transcript was not detectable in control L-E rats nor after TCDD treatment (Figure 2.3).

49

Feed-restricted-control L-E rats were included to ensure that changes in AHR tra nscript levels were due to TCDD-treatment per se and not the result of decreased feed intake which occurs in dioxin-sensitive strains within 96 h after TCDD exposure. AHR transcript levels did not differ between feed-restricted controls and corn-oil controls. After TCDD treatment, total hepatic AHR transcript levels increased 2-fold in dioxin- sensitive L-E rats and 3.6-fold in the resistant H/W strain (Figure 2.3). Despite the larger TCDD-induced increase in AHR levels in H/W rats compared with L-E rats, the absolute total mRNA levels were 2-fold lower in resistant H/W rats than in sensitive L-E rats (Figure 2.3).

H/W Dioxin-Resistant Rat L-E Dioxin-Sensitive Rat )

-8 40 H/W-DV

10 x H/W-IVs d 30 WT d

20 c c bc bc 10 b b b

Transcript copies per Transcript copies a ( RNA total of µg 0 ctrl 19 96 ctrl 19 96 FRC Hours post-TCDD (100 µg/kg) Hours post-TCDD (100 µg/kg)

Figure 2.3. Expression levels of AHR splice-variant transcripts in rat liver after treatment with TCDD. Liver was taken from H/W and L-E rats at 19 h or 96 h after TCDD treatment and transcript levels measured as described in Materials and Methods. Since the identical stop codon is used in H/W’s SIV mRNA and LIV mRNA both mRNA variants encode the identical protein product. Thus PCR primers common to SIV mRNA and LIV mRNA were designed and simply termed IVs. Control animals received corn oil. The L-E data set includes an additional control group (FRC) whose food intake was restricted to the same level as TCDD-treated animals to determine if any changes in transcript expression were due to TCDD treatment per se versus the possibility that the changes were due to reduced food intake caused by TCDD treatment in the dioxin- sensitive L-E rats. Bars (mean ± S.D., n=4) with non-identical letters significantly differ from one another (ANOVA, Bonferroni post-hoc analysis p<0.05).

50 Functional Differences in Intrinsic Transactivation Activity Among the Splice Variants We thought it important to determine whether the variant forms of AHR that contain different structures in their transactivation domains have different abilities to transactivate gene expression. To remove the influence of other AHR domains we created chimeric constructs in which the isolated AHR TAD is linked to the Gal4 DNA- binding domain so that the Gal4-AHRTAD construct drives expression of a firefly luciferase reporter gene. The Gal4 system tests the intrinsic transactivation ability of the TAD rather than the response of the full-length receptor to ligand-induced activation.

We compared intrinsic activities of the TAD from the three rat AHRs: WT, DV and IV. Immunoblotting confirmed that expression levels for the Gal4-AHRTAD proteins were equivalent for all three constructs when transfected into rat 5L hepatoma cells (Figure 2.4; left panel). The intrinsic transactivation activity was significantly higher for the DV than for the WT or IV; the intrinsic activity of the IV TAD was slightly lower than that of the WT TAD (Figure 2.4; right panel). Since intrinsic transactivation activity was higher for the TAD derived from the DV than from the wildtype rat AHR, we sequentially deleted portions of the TAD from the WT AHR to determine which specific regions were responsible for the difference. Deletion of amino acids 766-773 significantly increased activity above that of the WT receptor. Deletion of a further segment up to aa783 brought the activity up to the level of the DV. Extending the deletion up to aa800 did not result in a further increase in transactivation activity (Figure 2.5).

51

n.s *** 1.2 3

1.0

0.8 2 0.6 ** 0.4 1

Luciferase Activity 0.2

Gal4-TAD proteinexpression 0.0 0 Gal4 V T V V -I M -W -D -C D /W W A A H / β-Actin F T - H p D - A D T A IV V T T W - -D - W D / /W A -H H T D - A D T A T

Figure 2.4. Effect of rat AHR TAD polymorphism on intrinsic transactivation activity. The left-hand panel shows protein expression levels of Gal4-AHRTAD chimeras in the transfected 5L hepatoma cell line. Total protein was separated via SDS-PAGE. Immunoblots were performed for AHR (upper band) and β-actin (lower band) and densities measured with ImageQuant software. The slower migration of the band for the WT chimera is due to the fact that the size of the TAD protein domain is larger for the WT AHR than for either the IV or DV. Expression values are represented as a fraction of expression in cells transfected with the Gal4- AHRTAD-WT normalized to corresponding β-actin (n=3; significance calculated by one-way ANOVA, Bonferroni post-hoc analysis; error bars represent SD). The right-hand panel shows intrinsic activity of the Gal4-AHRTAD chimera from WT AHR versus the alternative AHR splice variants DV or IV in rat 5L hepatoma cells. Firefly luciferase was first normalized to renilla luciferase for each sample. Luciferase activity is represented relative to activity in cells transfected with Gal4-AHRTAD-WT set to 1.0 (mean ± S.D., n=3; unpaired t-test, *=p<0.01, **=p<0.001, ***=p<0.0001).

52 3.0 *** *** *** 2.5

2.0 *

1.5

Immunoblot of AHR protein

Luciferase Activity 1.0 Gal4

0.5 β-Actin

T 3 3 0 V 7 8 0 W 7 8 D - -7 - - - D 6 6 6 W A 7 7 7 / 0.0 T 7 7 7 H Δ Δ Δ - 3 3 0 V T 7 8 0 V T T D M W 7 7 8 -D WT A C - - - - - W W - D 6 6 6 W - - T A A 7 7 7 / D D D F T 7 7 7 -H A A p Δ Δ Δ T A T T T D T T A -W -W -W T D D D A A A T T T

ÅMVSPQAYYAGAMSMYQCQAGPQHTPVDQMHYSPEIPGSQAFLSKFÆ TAD-WT

ÅM GAMSMYQCQAGPQHTPVDQMHYSPEIPGSQAFLSKFÆ TAD-WTΔaa766-773 766 773 VSPQAYYA ÅM GPQHTPVDQMHYSPEIPGSQAFLSKFÆ TAD-WTΔaa766-783 783 766VSPQAYYAGAMSMYQCQA

ÅMGSQAFLSKFÆ TAD-WTΔaa766-800 766VSPQAYYAGAMSMYQCQAGPQHTPVDQMHYSPEIP 800 ÅMFÆ TAD-H/W-DV 766 808 VSPQAYYAGAMSMYQCQAGPQHTPVDQMHYSPEIPGSQAFLSK

Figure 2.5. Effect of rat AHR TAD polymorphism on intrinsic transactivation activity. Deletions of increasing size were created in the TAD of the WT AHR as described in Materials and Methods and illustrated in the bottom panel. The resulting TAD constructs were assayed for intrinsic transactivation activity in rat 5L hepatoma cells as described in the legend to Figure 2.4. The top-left panel indicates that the level of protein expressed was equivalent for all TAD constructs. In the top-right panel luciferase activity is represented relative to activity in cells transfected with Gal4-AHRTAD-WT as in the legend to Figure 2.4.

53

AHR Structures Predicted In Silico: Comparison Between Species and Within Species. Secondary structure of mRNA is, itself, a potential mechanism of translational regulation by cis-acting factors since secondary structure can affect mRNA localization and degradation. To determine if the alternatively-spliced AHR variants possess altered secondary mRNA structures, we compared predicted secondary RNA structures for the TADs of the SIV and DV variants to that of WT. The least energetically-favorable TAD structure is the DV which has a predicted free energy of minus 274 kcal/mol compared with minus 296 for the SIV and minus 327 for the WT. The loss of 129 nucleotides from the DV TAD leads to a significant loss of secondary structure such as hairpin loops and bulges (Supplemental Data Figure S2A). The SIV has acquired 29 nucleotides from intron-10 and this is predicted to create a small hairpin loop (Supplemental Data Figure S2B). Moreover, in the IVs, 138 nucleotides are deleted from the carboxy terminus which is predicted to result in considerable loss of structure (Supplemental Data Figure S2C). In addition to examining the predicted secondary structure of mRNAs for the splice variants we examined how variations in mRNA sequence might affect subsequent protein structure in the TAD of multiple species and strains. As shown in Figure 2.6, predicted secondary structures are similar for the TADs of rat, mouse, guinea pig, and human whereas the TAD from a highly dioxin-resistant species, hamster, contains a large additional segment of amino acids predicted to be involved in formation of α-helices (Figure 2.6A). Our in silico analysis (Figure 2.6) reveals that the TAD of the rat AHR protein has abundant amino acids that favor formation of α-helices in the acidic region and in the amino-terminal end of the Q-rich region of the WT AHR, as well as in each H/W splice variant (Figure 2.6B). In contrast, the carboxyl-terminal half of the Q-rich subdomain is relatively devoid of secondary structure (Figure 2.6). There is a conservative valine-to-alanine amino acid substitution at position 497 of the H/W AHR but this has no impact on predicted secondary protein structure (data not shown).

54

“Acidic” “Q-Rich” “P- S- T-Rich” RAT WT

539 B IV

539

DV 539 Mouse C57BL/6J

535

DBA/2J 535

Guinea Pig

537

Human 545

Hamster A 538 Amino acid position in TAD sequence

Figure 2.6. Inter- and intra-species comparison of predicted secondary protein structure of the TAD. Consensus secondary AHR protein structures for transactivation domains were predicted using Network Protein Sequence Analysis (NPS@, (Combet et al., 2000)). Tall bars represent amino acids contributing to α-helix structure(s). Short bars represent amino acids contributing to β- strand(s). A: Hamster is predicted to have a high abundance of α-helices in its TAD. B: Restructured region of the H/W rat. The approximate boundaries for the major TAD subdomains, based on human nomenclature, are displayed. Note that first amino acid number marks the beginning of the TAD for each species. Subsequent numbering of amino acids refers only to the position within the TAD domain, not the full-length AHR protein.

55

The DV protein is predicted to lack most of the C-terminal α-helix (Figure 2.6B). Inter-species comparisons indicate that this predicted α-helical region is conserved among rat, mouse, human, guinea pig and hamster. The insertion of seven amino acids into the IV TAD is predicted to lead to a small increase the α-helical content as can be seen in Figure 2.6B while the deletion of the last 45 amino acids is predicted not to result in any loss of secondary protein structure. The predicted TAD protein structures of some highly-resistant species/strains such as H/W rat and hamster appear to be more ordered and to form fewer ambiguous states than AHRs of dioxin-sensitive species (Table 2.2). While dioxin-resistant DBA/2J mice have predicted secondary TAD protein structure that is virtually identical to that of dioxin-sensitive C57/BL6J mice, dioxin resistance in DBA/2J mice is primarily attributable to structural variation in the ligand-binding domain (rather than the TAD) which leads to lowered affinity for TCDD (Okey et al., 2005), but lack of homogeneity in predicted TAD structure between mouse strains does not negate possible involvement of the TAD in resistance to TCDD.

Compared with the WT protein, hydrophobicity profiles (Supplemental data Figure S3) revealed deletion of hydrophobic as well as hydrophilic regions from the TAD domain of the DV protein. While the terminal hydrophobic region predicted to exist in the WT TAD protein is lost from the IV TAD protein, a region of greater hydrophobicity is inserted at the terminus (Supplemental data Figure S3). Aside from 2 hydrophilic regions inserted within the hamster TAD and 3 small hydrophobic regions within the guinea pig TAD, the hydrophobicity profiles of rat, mouse, guinea pig, human, hamster are very similar (Supplemental data Figure S3).

56 Table 2.2. Predicted secondary AHR protein structures of transactivation domains in different species and strains

Protein Dioxin Alpha helix Random Ambiguous Extended LD50 µg/kg (%) Coil (%) states (%) strand (%) (male)

Rat L-E WT 18 19.1 74.0 1.59 5.4 H/W > 9,600 IV 19.9 72.2 0.36 7.6

DV 18.4 74.3 1.47 5.9 Mouse C57BL/6 (Ahrb1) 128 19.6 70.9 1.11 8.49 DBA/2J (Ahrd) 2,600 19.4 73.3 0.64 6.7 Guinea Pig 1-2 15.2 76.1 2.27 6.5

Hamster 3,000-5,000 33.2 62.6 0.52 4.7

Human Unknown 19.9 72.6 0.33 7.2 (resistant)

Post-translational modifications, particularly phosphorylation status, are known to affect AHR function (Long and Perdew, 1999; Mahon and Gasiewicz, 1995). Therefore we searched all TAD protein domains derived from each splice variant for amino acid motifs known (from curated literature) to be targets for kinase/phosphatase activities. In the DV, 15 potential post-translational-modification motifs are deleted from the TAD whereas 20 potential motifs are deleted in the IV (Figure 2.7; Supplemental Table S1). The 7 amino acids inserted from former intronic sequence into the IV do not add any unique post-translational modification sites. No SUMOylation sites are predicted to exist within the TAD for IV, DV, or WT receptors.

57

539 595 723 760 853 Acidic Q-Rich P- S- T- Rich

WT SQSAM**VSPQAYYAGAMS** MYQCQ AGPQHTPVDQMHYSPEIPG * * SQAFLSK****** FQSPSILNEAYSADLSSIGHLQTAAHLPRLAEAQPLPDITPSGFL* * ** DV SQSAM* FQSPSILNEAYSADLSSIGHLQTAAHLPRLAEAQPLPDITPSGFL** *** * ** IV SQSAM**VSPQAYYAGAMS** MYQCQ AGPQHTPVDQMHYSPEIPG ** **SQAFLSK IRAFY RE

Repressor binding ?

Figure 2.7. Schematic representation of differences in predicted TAD protein structures between rat WT AHR and the rat splice-variants. Predicted post-translational kinase/phosphatase substrates (represented as asterisks) curated from published literature were identified using PhosphoMotif Finder (Peri et al., 2003). Note: the segment inserted into the IV had no unique posttranslational modifications. See Supplemental Table S1 for specific binding motifs and their substrates. Consensus secondary AHR protein structure was predicted using NPS@ as described in Materials and Methods. The unshaded box represents β-strands and the shaded box represents α-helices.

58 DISCUSSION The deletion in the TAD of the H/W rat AHR provides a natural window onto the relationship between AHR structure and AHR function. To better understand how changes in the TAD impact the H/W rat’s responsiveness to dioxin we compared IV and DV AHR variants to WT receptor for constitutive in vivo expression levels, in vivo expression after dioxin treatment, intrinsic transactivation ability, and predicted mRNA and protein structures. Despite the substantial alteration in AHR structure in H/W rats, genes that are well- known to be AHR-regulated and dioxin-inducible such as CYP1A1, CYP1A2, CYP1B1, ALDH3A1, NQO1 and UGT1A1 continue to be highly responsive to induction by TCDD (Okey et al., 2005). The TAD deletion in H/W rats appears to selectively prevent TCDD from dysregulating genes that are essential to dioxin toxicity rather than causing a blanket failure of transactivation.

Constitutive Expression of Splice Variants Our experiments show that the DV of the H/W AHR has, in fact, higher intrinsic transactivation activity than either the WT or the IVs. However, DV transcript levels are very low in dioxin-resistant H/W and LnA rats and are undetectable in L-E and LnC rats. Thus, despite high intrinsic transactivation function, the DV is unlikely to be a significant mediator of responses to TCDD in vivo. Dioxin resistance behaves as a dominant trait in crosses between dioxin-sensitive strains and dioxin-resistant strains. F1 offspring are resistant to TCDD lethality and they express both the WT AHR protein and the smaller proteins that result from the TAD deletion (Pohjanvirta et al., 1999). Our current experiments show that F1 rats express all three mRNA splice variants. However, expression from the single copy of the WT AHR allele from the sensitive parent is not sufficient to confer dioxin sensitivity in F1 offspring. Rather, the AHR from the resistant H/W parent appears to act in a “dominant negative” fashion to create resistance to TCDD lethality (Pohjanvirta and Tuomisto, 1994; Tuomisto et al., 1999). Our current experiments reveal virtually no DV transcript in tissues of homozygous AHRWT/AHRWT L-E or LnC rats and very low levels of IV transcripts; therefore these

59 animals receive no “protection” from the potential dominant-negative influence of the DV or IV. Conversely, H/W rats and LnA rats (homozygous for AHRH/W/AHRH/W), predominantly express IV transcripts; no WT AHR is detectable in any tissues examined in H/W or LnA indicating that the mutation at the exon/intron -10 boundary completely disrupts the normal splice site in these dioxin-resistant animals. Total AHR abundance might affect susceptibility to TCDD. Our earlier studies indicated that AHR mRNA and protein levels are higher in lung and liver of dioxin- sensitive L-E rats than in dioxin-resistant H/W rats (Franc et al., 2001a; Pohjanvirta et al., 1999; Viluksela et al., 2000). Our current study confirms that total constitutive AHR mRNA levels are higher in liver of L-E rats than in H/W or LnA rats. Mice that are hypomorphic for AHR expression are resistant to TCDD toxicity (Walisser et al., 2004a). However, the dioxin-sensitive LnC rat expresses total AHR mRNA levels that are no higher than in dioxin-resistant H/W rats, indicating that higher AHR levels are not invariably associated with high susceptibility to TCDD toxicity. Because the biochemical and toxic actions of dioxin-like compounds affect many tissues other than liver, we measured splice variant levels in lung, thymus, kidney and testis. Tissue-dependent regulation of alternative splicing is a common occurrence in transcriptional regulation (Le et al., 2004; Nakahata and Kawamoto, 2005) and is employed by members of the bHLH-PAS family such as ARNT2 (Korkalainen et al., 2003) and HIF-1α (Drutel et al., 2000). We found that the expression pattern in liver for WT and H/W variants was recapitulated among other tissues although total mRNA levels varied among tissues.

TCDD Effect on Expression of Splice Variants Previous reports on the effect of TCDD on AHR expression levels in rodent tissues in vivo have been contradictory. TCDD increases AHR protein in liver (Sloop and Lucier, 1987) but there also are reports of AHR depletion in multiple rat tissues (Pollenz et al., 1998; Roman and Peterson, 1998; Sommer et al., 1999). We previously found that exposing H/W and L-E rats to 5 µg/kg TCDD produced a 2- to 3-fold increase in total cytosolic AHR protein levels whereas 50 µg/kg TCDD led to depletion at one day post- TCDD followed by recovery in H/W but not L-E rats (Franc et al., 2001a). Our current

60 study had the capability to quantitatively differentiate variant forms of AHR transcripts. There was no difference in which splicing products were generated constitutively versus the splicing products formed in response to TCDD. After TCDD treatment, IV transcripts remained predominant in resistant rats while the WT transcript was dominant in livers of dioxin-sensitive L-E rats. In our current study, exposure to 100 µg/kg TCDD for 19 h or for 96 h significantly increased total AHR mRNA levels in both dioxin- sensitive (2-fold) and dioxin-resistant (3.6-fold) rat livers. The extraordinary resistance of the H/W rat to lethal effects of TCDD cannot be attributed to alteration of the level of any individual splice variant nor to a failure to up-regulate the total pool of AHR transcripts. However, after TCDD treatment, total AHR transcript levels in H/W rats were ~2-fold lower than in L-E rats. In agreement with our previous study (Franc et al., 2001a), sensitivity to dioxin toxicity is not likely to be attributable to differential regulation of AHR levels by the potent agonist ligand, TCDD.

TAD Structure in Relation to AHR Function We used deletion analysis and reporter-gene assays to identify which amino acids in the AHR TAD account for the higher intrinsic activity of the DV. This was supplemented by modeling in silico to predict secondary structures of mRNA and protein products for each splice variant with the goal of identifying features that might be responsible for differences in transactivation function among the variants from dioxin- sensitive vs. dioxin-resistant rats. Our systematic deletion of the region that is divergent between WT AHR and the DV indicates that amino acids 766-783 of the WT AHR are responsible for its lower intrinsic transactivation activity. Previously Kumar et al. showed that deletion of the entire P-S-T- rich subdomain of human AHR TAD enhances transcriptional activity (Kumar et al., 2001), suggesting that the elevated intrinsic transactivation ability of the DV in our in vitro experiments is due to deletion of amino acids that suppress TAD function. There are many ways in which alteration of TAD structure might affect levels or function of mRNA or protein. At the mRNA level, loss of structure may flag mRNA for degradation leading to the low cellular levels that we observed for the DV mRNA. At the protein level, altered TAD structure, encoded by different splice-variant mRNAs may

61 impair the receptor’s ability to interact with essential regulatory proteins such as co- regulators or phosphatases. Hydrophobicity content and secondary structure are indices of a protein’s accessibility to interact with other proteins. Hydrophobic regions have a low probability of being on the surface of a protein and amino acids that lie within a structured region often are not accessible. Previous studies indicated that an increased content of hydrophobic amino acids and/or α-helical secondary structures in the AHR TAD impair AHR protein-protein interactions (Watt et al., 2005) and enhance transactivation (Jones and Whitlock, 2001), perhaps interrupting repressor protein binding. In the IV protein, our modeling predicts that the altered amino acid sequence increases both α-helical content and hydrophobicity of the TAD terminus compared to WT AHR. Overall, modeling suggests that the TAD terminus of the IV protein adopts a conformation that is less accessible to interactions with other proteins. Noteworthy, in the IV TAD there is loss of a predicted Src kinase-substrate motif and a second such motif adjacent to the inserted α-helix may be obstructed. Mice lacking c-Src kinase activity are resistant to lethality of TCDD but are fully responsive to CYP1A1 induction, as are H/W rats (Dunlap et al., 2002; Matsumura et al., 1997). Loss of AHR phosphorylation by Src kinase may inhibit particular AHR functions relating to dioxin lethality but not alter the AHR’s ability to regulate CYP1A1. In summary, we found that IV transcripts are the dominant AHR splice variants expressed in tissues of dioxin-resistant rats and remain the dominant transcripts after TCDD treatment. Since very little DV mRNA is present in vivo and since its structure is the energetically least-favorable form, it is unlikely that a significant amount of DV protein is synthesized or that the DV plays a role in biochemical or toxic responses. Our deletion analysis indicates that amino acids 766-783 within the AHR’s TAD are critical for suppression of TAD function. Modeling in silico predicts that insertion of 7 a.a. combined with deletion of 45 a.a. from the TAD terminus results in increased secondary structuring that selectively impairs the receptor’s ability to interact with other proteins, subsequently attenuating transcription of genes that are involved in dioxin toxicities while leaving transcription of non-lethal AHR-mediated genes unaltered.

62

SIGNIFICANCE AND IMPACT OF THIS PROJECT

The AHR is the first essential component in the dioxin toxicity cascade. Therefore establishing that sensitive rats express the wildtype AHR while resistant rats express the insertion variant AHR TAD isoform, and that the isoform expression pattern is unaltered by TCDD is critical to discerning whether altered TAD structure alters transcription and ultimate susceptibility to dioxin lethality.

To this end, this project found the AHR structure to be a key determinant of susceptibility to TCDD lethality. Specifically, the altered AHR TAD IV isoform of the H/W may act as a dominant-negative regulator to create resistance to TCDD lethality. Further, the predicted secondary structure of the AHR TAD is a likely determinant of a species’ susceptibility to TCDD lethality. Phylogenetic comparison of AHR TADs may aid in extrapolation of dioxin risk across species and aid in the refinement of permissible levels of dioxin exposure to humans.

63

CHAPTER 3: Dioxin Lethality: Aryl Hydrocarbon

Receptor (AHR)-Mediated Gene Expression in a

Resistant Rat Model

Ivy D. Moffat, Paul C. Boutros, Jouni Tuomisto,

Raimo Pohjanvirta, and Allan B. Okey

In this project, the roles of collaborators were:

Collaborator Contribution

Ivy D. Moffat Sample preparation, RT-PCR analysis, GO & pathway analysis, and establishment of candidate gene biological plausibility

Paul C. Boutros Expression array data analysis & AHRE search

Jouni Tuomisto Development of LnA & LnC rats

Raimo Pohjanvirta Treatment & supply of tissues; Co-principal Investigator

Allan B. Okey Principal investigator

64 ABSTRACT

Major toxicities of 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) result from dysregulation of gene expression mediated by the aryl hydrocarbon receptor (AHR). Dioxin-like chemicals alter expression of numerous genes in liver but the specific genes whose dysregulation leads to toxicities such as wasting, hepatotoxicity, and lethality have not been identified. We searched for genes that are most likely to be key to dioxin toxicity by using gene expression arrays to contrast hepatic gene expression after TCDD treatment in dioxin-sensitive rats (that carry wildtype AHR) with gene expression in H/W(Kuopio) rats which are highly resistant to dioxin toxicity due to a major deletion in the AHR’s transactivation domain (TAD). The total number of TCDD-responsive genes was smaller in rats with the AHRH/W genotype than in rats with wildtype AHR. However, genes that are well-known to be AHR-regulated and dioxin-inducible such as CYP1A1, CYP1A2 and CYP1B1 remained fully responsive to TCDD in AHRH/W rats; thus the TAD deletion selectively interferes with expression of a subset of hepatic genes rather than abolishing global AHR-mediated responses. Genes in the following functional categories differ in response to TCDD between dioxin-sensitive rats and dioxin-resistant rats: fatty acid oxidation, metabolism (xenobiotic, alcohol, amino acid, and fatty acid), phosphate transport, regulation of steroid biosynthesis, nitrogen compound catabolism, and generation of precursor metabolites and energy. Many of these differentially- responsive genes are integral parts of pathways such as: protein degradation and synthesis, fatty acid metabolism and synthesis, cytokinesis, cell growth, and apoptosis which may be part of mechanisms which lead to TCDD-induced wasting, hepatotoxicity, and death. These differentially-responsive genes are worthy candidates for further mechanistic studies to test their role in mediating or protecting from major dioxin toxicities.

65 INTRODUCTION

Extensive evidence demonstrates that virtually all toxic effects of TCDD and related dioxin-like compounds are mediated by the aryl hydrocarbon receptor (AHR), which functions as a ligand-dependent transcription factor (reviewed in (Okey, 2007)). Major toxicities from dioxin exposure including: thymic atrophy, teratogenesis, hepatotoxicity, wasting syndrome, and death are dependent on the AHR, its dimerization partner, ARNT, and require that the AHR have a functional nuclear translocation/transactivation domain (Bunger et al., 2003; Mimura et al., 1997; Peters et al., 1999; Walisser et al., 2004b). Dioxin binding converts the AHR into an activated ligand:AHR:ARNT complex that regulates transcription either by binding directly to an AH-responsive element-I (AHRE-I) (Denison et al., 1988; Hankinson, 1995; Ma, 2001; Whitlock, 1993) or via indirect interaction with AHRE-II motifs (Boutros et al., 2004; Sogawa et al., 2004) located in regulatory regions of target genes. Dioxin toxicities probably arise from AHR- mediated dysregulation of specific genes (Okey et al., 2005). Microarray technologies have accelerated identification of numerous genes that depend on the AHR for constitutive expression or for response to TCDD in vivo (Boverhof et al., 2005; Boverhof et al., 2006; Fletcher et al., 2005; Hayes et al., 2005; Ovando et al., 2006; Slatter et al., 2006a; Tijet et al., 2006), but the key target genes whose dysregulation by dioxin leads to lethality remain unknown. The Han/Wistar (Kuopio) (H/W) rat provides an opportunity to identify which genes, out of the many that respond to dioxins, actually are critical to lethality. H/W rats are extraordinarily resistant to acute lethality from TCDD with an LD50 >10,000 μg/kg compared with ~10 μg/kg for Long-Evans rats (Turku/AB) (L-E) (Pohjanvirta and Tuomisto, 1994). Resistance is associated with a mutation which deletes 38 or 43 amino acids from the AHR transactivation domain (TAD) in H/W rats (Pohjanvirta et al., 1998; Tuomisto et al., 1999). We postulate that this TAD deletion renders the AHR in H/W rats incapable of dysregulating those genes that lie in pathways leading to lethal effects of TCDD. Our strategy for identifying the genes whose dysregulation leads to lethality is to use gene expression profiling to contrast the sets of genes that respond to TCDD in

66 dioxin-resistant rat strains/lines with the sets of genes that respond in dioxin-sensitive rat strains/lines (Okey et al., 2005). H/W rats are the prototype dioxin-resistant strain. L-E rats are the prototype dioxin-resistant strain. Dioxin resistance in rats segregates genetically with the AHR locus and is a dominant trait (Pohjanvirta and Tuomisto, 1994; Tuomisto et al., 1999). To increase the genetic power in our study we also incorporated two specialized rat lines, Line-A (LnA, dioxin-resistant) and Line-C (LnC, dioxin- sensitive) that were produced by multiple generations of crosses beginning with L-E and H/W rats, combined with phenotyping and selection for dioxin sensitivity/resistance by TCDD challenge (Tuomisto et al., 1999). Together H/W and LnA rats constitute the “resistant collective” whereas L-E and LnC rats constitute the “sensitive collective”. We focused on hepatic gene expression because liver displays a broad spectrum of mRNAs that are responsive to dioxins and/or the AHR genotype (Tijet et al., 2006) and because liver is a prime site of dioxin toxicity which displays many phenotypic differences between sensitive and resistant rat strains (Niittynen et al., 2007). We chose a dose of 100 µg/kg TCDD because this produces hepatotoxicity, wasting, and death in rats in the sensitive-collective but no deaths in the resistant-collective. We first assessed global gene expression responses to TCDD using Affymetrix GeneChips® followed by real-time RT-PCR quantitation for candidate mRNAs to identify genes whose expression differs between the dioxin-sensitive collective and the dioxin-resistant collective. In addition we assessed expression of some genes in an Ahr-null mouse model to determine if their response to TCDD was AHR-dependent and used bioinformatic techniques to predict biological processes, pathways and AHRE binding sites for co-regulated genes.

67 MATERIALS AND METHODS

Models in vivo: treatment and RNA isolation

Dioxin-resistant H/W rat model: Liver tissues were from rats in which we previously determined mRNA expression levels for wildtype AHR and for the H/W AHR splice variants (Moffat et al., 2007). Two dioxin-sensitive rat strains/lines, L-E and LnC (which express wildtype AHR), and two dioxin-resistant rat strains/lines, H/W and LnA, (that have a deletion in the AHR transactivation domain, (Pohjanvirta et al., 1998)) were from breeding colonies of the National Public Health Institute, Division of Environmental Health, Kuopio, Finland. All animals were males 10-12 weeks old. They were housed in groups of 4 rats (an entire treatment group per cage) in suspended stainless-steel wire- mesh cages with pelleted R36 feed (Lactamin, Stockholm, Sweden) and tap water available ad libitum. The temperature in the animal room was 21 ± 1°C, relative humidity 50 ± 10%, and lighting cycle 12/12 h light/dark. The study plans were approved by the Animal Experiment Committee of the University of Kuopio and the Kuopio Provincial Government. Liver was harvested between 8:30 and 11:00 from rats treated by gavage with a single 100 µg/kg dose of TCDD or with corn oil vehicle for 3 or 19 h. We focused on these early time-points following TCDD exposure to try to capture primary changes in AHR-mediated gene expression prior to the onset of overt toxicity. There were 12 animals for H/W and L-E (4 control; 4 3h-TCDD-treated and 4 19h-TCDD-treated) and 8 animals for LnA and LnC (4 control; 4 19h-TCDD-treated).

Dioxin-resistant Ahr-null mouse model: Liver tissues were from mice in which we previously mapped AHR-dependent and dioxin-dependent gene batteries by transcriptomic analysis (Tijet et al., 2006). Briefly, male Ahr-null (Ahr-/-) mice in a C57BL/6J background (10 weeks old) and C57BL/6J mice carrying wildtype (Ahr+/+) (15 weeks old) were given a single dose of 1000 µg/kg TCDD or corn oil vehicle by gavage. Liver was harvested 19 h after treatment. We tested 3 TCDD-treated and 3 control mice in the Ahr-/- groups and 4 TCDD-treated and 4 control mice in the Ahr+/+ groups.

68 Total RNA was extracted from both rat and mouse livers using Qiagen RNeasy kits according to the manufacturer’s instructions (Qiagen, Mississauga, Canada). Total RNA yield was quantified by UV spectrophotometry and RNA integrity was verified using an Agilent 2100 Bioanalyser (Agilent Technologies, Santa Clara, CA).

Affymetrix GeneChip® analysis We used an incomplete three-factor experimental design (Figure 3.1). The factors are: strain (four strains: L-E, H/W, LnA, and LnC), animal status (TCDD-treated or vehicle-treated), and time (3 h or 19 h exposure to TCDD). The design is incomplete due to lack of treated animals at 3 h for the LnA and LnC strains and lack of control animals at 3 h for all 4 strains.

L-ET03 L-E L-EC19 Sensitive L-ET19 Collective

LnC LnCC19 L-CT19

LnA LnAC19 LnAT19

Strain or Line Rat Resistant H/WT03 Collective

H/W H/WC19 H/WT19

Control TCDD-treated Groups Groups

Figure 3.1. Experimental Design. An incomplete, three-factor design was used to study the effects of strain, TCDD-exposure and time. Two primary factors were assessed: TCDD exposure (control, 3, or 19 h after 100 μg/kg TCDD treatment) and strain/line (L-E, LnC, LnA, and H/W). In total, mRNA profiles for ten separate experimental conditions were assessed by microarray: control and TCDD-treated animals 19 h after treatment for all 4 strains/lines (L-EC19, L-ET19, LnCC19, LnCT19, LnAC19, LnAT19, H/WC19, H/WT19) and TCDD-treated animals 3 h after treatment for two strains (L- ET03, H/WT03). At each condition, four separate animals were profiled, each on an individual RAE230A microarray. In total, this leads to ten n=4 conditions, for a 40-array experimental design.

69 Sample labeling and hybridization to Affymetrix GeneChips® were according to the GeneChip® Expression Analysis Technical Manual 2001 (Affymetrix, Santa Clara, CA). Briefly, double-stranded cDNA was generated from 5 μg of total RNA using 24mer oligodeoxythymidylic acid primer with a T7 RNA polymerase promoter site added to the 3’end (Superscript cDNA synthesis; Invitrogen, Carlsbad, CA). After second-strand syntheses, in vitro transcription was performed with the Enzo IVT BioArray High Yield RNA transcript Labeling Kit (Enzo Diagnostics; Farmingdale, NY) to produce biotin- labeled cRNA. Twenty μg of the cRNA product was then fragmented and hybridized for 18 h to RAE230A Affymetrix GeneChips® which contain 15,924 ProbeSets. Each microarray was washed and stained with streptavidin-phycoerythrin and scanned at a 6- μm resolution with an Agilent GeneArray 2500 Scanner (Agilent Technologies). The raw array data were visually examined for spatial heterogeneity. All arrays appeared robust and all arrays were retained for subsequent analyses. Array data were then loaded into the R statistical environment (v2.3.1) using the affy package (v1.10.0) of the BioConductor open-source library (Gentleman et al., 2004). Distributional homogeneity was assessed on both Probe and ProbeSet levels (Appendix 3.1A & 3.1B); no aberrant arrays were identified. We further assessed ProbeSets for their apparent RNA quality and did not observe significant inter-array difference (Appendix 3.1C). Next, the array data were pre-processed with the GCRMA algorithm (v2.4.1, (Irizarry et al., 2003)). Raw and processed array data will be deposited in the Gene Expression Omnibus (GEO) repository at NCBI and an accession number will be available prior to publication.

Statistical analysis of array data

Since an incomplete experimental design (Figure 3.1) was used, a full three-factor linear model could not be fit. Therefore, independent pair-wise contrasts between treated and control animals were calculated for each strain. Gene-lists derived separately for each strain were then compared using the limma package (v2.7.10) in the R statistical environment (v2.3.1) with a condition-specific design matrix and within-strain/pairwise- contrasts. For each pair-wise contrast, we employed empirical Bayes moderation of standard error (Smyth, 2003) and false-discovery rate control (Storey and Tibshirani,

70 -2 2003). We then applied a threshold of pmoderated < 10 to each contrast to generate a ProbeSet-list. To identify the number of genes significantly affected by TCDD we plotted the number of differentially expressed ProbeSets against the moderated p-value for all six experimental conditions (Figure 3.2A). Further, to visualize patterns of differential

expression, we selected all ProbeSets that showed differential expression (pmoderated < 0.01

in at least one strain) and subjected their fold-change values (in log2 space) to hierarchical clustering in the R statistical environment (v2.3.1). Correlation (1 – R) was used as a distance metric for clustering and within-row scaling was performed (Figure 3.2B).

Functional analysis of responsive genes Annotation: Each ProbeSet was reannotated by a novel method developed in our laboratory (Boutros et al., manuscript in preparation). Briefly, the target sequence for each ProbeSet was BLASTed against consensus UniGene cluster sequences for UniGene build Rn.156 using stringent parameters, including a word-size = 7. The resulting hits were then parsed in a relational database using BioPerl-based scripts (Stajich et al., 2002) and each ProbeSet was annotated with the highest e-value hit.

Gene ontological analysis (GO): To determine if genes that are perturbed by TCDD fit into functionally-coherent Gene Ontology (GO) categories, genes that were dysregulated by TCDD within each strain were identified, then clustered across strains. Specifically, we employed GO analysis using the high-throughput version of the GoMiner tool (Zeeberg et al., 2003). False-discovery rates (FDRs) were estimated using the maximum of 5000 permutations and only rat-specific annotations were used. The cumulative FDR was calculated as the product of the contrast-wise FDRs and used as a threshold criterion. GO terms passing this threshold criterion were subjected to hierarchical clustering as described above, but without row-wise scaling. To determine which functional groups of genes were enriched in resistant rats vs. sensitive rats, GO analysis was performed as above (but without clustering) on the sensitive-collective as well as on candidate genes from the resistant-collective (including both Type I and Type II responses).

71 A)

LE 3h

LE 19h

HW 3h LnC 19h affected by TCDD by affected HW 19h ProbeSets of Number LnA 19h

Significance of expression (moderated p-value; -log10 ) B)

( moderated p-value < 0.01 ) 0.01 < p-value ( moderated

affected TCDD by ProbeSets

Figure 3.2. Comparison of expression profiles between rat strains/lines. (A) The number of ProbeSets affected by TCDD treatment in each rat line or strain is compared at levels of statistical-significance (moderated p-values; -log10) ranging from 1 to 10. (B) Hierarchical clustering with within-row scaling of all ProbeSets that responded at a moderated p- value of < 0.01 in at least one rat strain or line. Green indicates ProbeSets that were induced (upregulated) by TCDD; red indicates ProbeSets that were repressed (downregulated) by TCDD.

72 Transcription-factor binding-site analysis To determine if Aryl Hydrocarbon Response Elements (AHREs) play a role in the patterns of differential expression observed, we annotated each ProbeSet with AHRE binding-site data wherever the available genomic data permitted. Each UniGene cluster was associated with an Entrez Gene ID where matching was possible and this gene ID was associated with a genomic location on build rn4 of the rat genome using the UCSC Genome Browser database (tables: REFFLAT and REFLINK) and custom Perl scripts (v5.8.8) (Karolchik et al., 2003). Sequence from -10 kbp to +5 kbp relative to the smallest annotated transcriptional start-site was extracted and searched for AHRE-I and AHRE-II motifs as described previously (Boutros et al., 2004) as well as for antioxidant response elements (AREs) using a defined consensus motif (Huang et al., 2000). Each base from each identified motif was then associated with a PhyloHMM conservation score, and the average PhyloHMM score for each motif was calculated. These PhyloHMM scores provide a measure of conservation that accounts for phylogenetic relationships across the different animal species used in the analysis. Scores range between 0 (minimal conservation) and 1 (strong conservation) (Siepel and Haussler, 2004). Data used in this study were based on alignments of the following mammalian genomic assembles: rn4, mm8, hg18, canFam2, bosTau2, monDom4, galGal2, xenTro1, danRer3 and were downloaded from the UCSC genome browser database and parsed with custom Perl scripts (v5.8.8). For each Entrez Gene ID, we report the total number of motifs of each type present in searched region, along with the maximum average PhyloHMM score for each motif type.

mRNA quantitation by Real-Time RT-PCR Total RNA (2 µg) was reverse-transcribed into cDNA using oligo-dT primer, p(dT)15 (Roche Applied Science, Laval, QC, Canada) and Superscript II RNA polymerase according to the manufacturer’s instructions (Invitrogen, Carlsbad, CA). Real-time PCR was performed on an MX4000 system (Stratagene, La Jolla, CA) using in-house designed primers and 5’fluorogenic probes to amplify from 250 nanograms of cDNA (as described below) and Applied Biosystems gene expression assays to amplify

73 from 100 nanograms of cDNA as described by manufacturer (Applied Biosystems, Forest City, CA).Appendix 3.2 provides sequences for all primers/probe sets used. −ΔΔCt Normalized expression (NE) was calculated using NE= 2 , where Ct is the threshold cycle to detect fluorescence. PCR amplification efficiency was determined from a 10-fold serial dilution of a pool of cDNA; efficiency ranged from 90-110% for all genes examined. The data were normalized to either Actb or Gapdh, genes that we previously found to be suitable as normalization standards in dioxin studies (Pohjanvirta et al., 2006). In the rat model, significant differences in mRNA levels were determined using a t-test (two-sided, unequal variance). To define significant differences in mRNA levels in the Ahr-null mouse model, analysis of variance (ANOVA) followed by Bonferroni after hoc tests were performed using GraphPad version 4.0 (GraphPad Software Inc.). Differences between treatment groups were considered significant when: * p<0.05, ** p<0.01, *** p<0.001.

In house primers and probes: Each rat sequence that matched a reported ProbeSet was identified by BLAST analysis using standard blastn of the non redundant (all known) rat database with default settings: “expect 100, word size 7, no filter”. Each sequence was aligned and selected to avoid areas with potential splice-variants, polymorphisms and pseudogenes in the primer design (Altschul et al., 1990). Primers and probes were designed using Primer 3 software (http://frodo.wi.mit.edu/cgi- bin/primer3/primer3_www.cgi) and the IDT SciTools Primer Quest program (http://scitools.idtdna.com/Primerquest/) to ensure no hetero- or homo-dimerization and span an intron to prevent amplification from any potential contaminating genomic DNA. The PUNS (Primer-UniGene Selectivity Testing) program (Boutros and Okey, 2004) also was applied to ensure primer specificity. Probes were labeled with a reporter fluorescent dye 6-carboxyfluorescein at the 5’-end and a quencher-fluorescent dye Iowa Black FQ at the 3’-end. Primers and probes were synthesized by Integrated DNA Technologies, Inc. (Coralville, IA). PCR conditions were: after 10 min at 95°C, 40 cycles were performed at 95°C for 30 sec/each and 60°C for 1 min.

74 RESULTS Ensembl and GenScan predict 22,000 and 28,904 genes, respectively, in the rat genome (www.ensembl.org/Rattus_norvegicus). We used Affymetrix RAE230A GeneChips® containing 15,924 ProbeSets representing 12,660 UniGene clusters to determine the effect of TCDD on gene expression in dioxin-resistant rats versus dioxin- sensitive rats.

Global differences in gene expression between dioxin-sensitive and dioxin-resistant rats in response to TCDD One major question we sought to answer was: does the deletion in the AHR transactivation domain in H/W rats cause a decrease in the total number of genes that respond to TCDD? The total number of ProbeSets affected by TCDD varied substantially among rat strains (Figure 3.2A). More ProbeSets were affected by TCDD treatment in dioxin-sensitive L-E rats than in any other rat strain; this was true both at 3 h and 19 h after TCDD-treatment and across multiple levels of statistical significance (Figure 3.2A). Dioxin-resistant H/W and LnA rats (which carry the AHRH/W genotype) exhibited the smallest number of dioxin-responsive ProbeSets at 19 h (Figure 3.2A). Thus it appears that the transactivation-domain deletion reduces the total number of genes that respond to TCDD. Hierarchical clustering of all ProbeSets that responded significantly to TCDD

(pmoderated <0.01) in at least one rat strain indicated that gene expression also varies strongly depending on the duration of TCDD exposure; 3-h exposures distinctly cluster separately from the 19-h exposures (Figure 3.2B). To determine if there is significant overlap in responses to TCDD across rat strains or dependent on duration of TCDD exposure, we generated a series of Venn diagrams (Figure 3.3). The dioxin-resistant collective (Figure 3.3A) exhibited a fewer number of dioxin-responsive ProbeSets (1157) than the dioxin-sensitive collective (1802) (Figure 3.3B), again indicating that the transactivation-domain deletion in rats with the AHRH/W genotype reduces the total number of genes that respond to dioxin. The sensitive- collective displayed the greatest overlap across conditions, with 30% of dioxin- responsive ProbeSets being altered in two or more conditions in the dioxin-sensitive

75 collective compared to only 20% in the resistant-collective; this difference is statistically significant (p = 1.80 x 10-8; two-sided proportion test). In both dioxin-sensitive L-E rats and in dioxin-resistant H/W rats, more ProbeSets were affected by TCDD after 3 h exposure than after 19 h (Figure 3.3).

Resistant Collective Sensitive Collective

H/W H/W L-E L-E 3-h 19-h 3-h 19-h

64045 157 708155 386

A) B) 119 55 30 105 24 237

125 173

LnA 19-h LnC 19-h Resistant 19-h (159)

C) 51069 16

1 39 19 0 393 H/W 3-h L-E 3-h (770) 277 29 33 608 (1006)

16 140

Sensitive 19-h (355)

Figure 3.3. Venn diagrams. Entries in each cell indicate numbers of ProbeSets that were significantly affected (either up or down) by TCDD treatment for each rat line or strain and at 3 h and 19 h after TCDD. The number of Type-II-response ProbeSets is derived from panel C which indicates ProbeSets whose expression differs between the collective of dioxin-sensitive rats and the collective of dioxin- resistant rats. Shaded circles represent dioxin-resistant rat responses to TCDD. Circles with dashed lines represent gene expression responses after 3 h TCDD exposure, while solid lines represent responses after 19 h TCDD exposure.

76 After assessing global differences in gene expression between sensitive and resistant rats we applied the following stratified criteria to identify candidate genes that may be involved in the mechanism of dioxin toxicity:

1. Genes that display a Type II rather than a Type I response to TCDD We classified TCDD-responsive ProbeSets into one of two categories using the scheme previously developed by Simanainen et al. (Simanainen et al., 2002) for toxic endpoints: Type-I responses are defined as responses to TCDD that are similar between dioxin-sensitive and dioxin-resistant rat strains/lines. Type-II responses are defined as responses that are different between dioxin-sensitive and dioxin-resistant rats. For the current study, genes that exhibited a statistically significant difference in the response to dioxins between the sensitive-collective and the resistant-collective are termed Type-II genes. Figure 3.3C displays the degree of overlap or of separation of dioxin-responsive ProbeSets between the sensitive and resistant collectives.

Type I responses: The validity of the arrays to identify effects of TCDD on gene expression is supported by the fact that: (1) multiple genes that are well-established as being induced by dioxins were upregulated (i.e. CYP1A1, CYP1A2, and CYP1B1) (Table 3.1); (2) there is overlap between our Type-I responses and responses previously reported to be affected by AHR-ligands in array experiments (Boverhof et al., 2005; Fletcher et al., 2005); and (3) numerous genes that are represented on the array by multiple ProbeSets from different regions of the transcript (e.g. Cyp51 and Mvk) show responses to TCDD that are similar for all ProbeSets for that transcript. In total, 345 ProbeSets exhibited Type-I responses at 3 h but only 137 ProbeSets were classified as Type-I at 19 h post TCDD. Thirty-nine ProbeSets displayed a persistent response, i.e. they were significantly affected at 3 h and the response still was significant at 19 h after TCDD exposure (Figure 3.3C and Table 3.2; Full dataset available online). Within the Type-I category, a similar number of ProbeSets was upregulated (180) as was down-regulated (165) after 3 h TCDD exposure.

77 The strong upregulation of CYP1A1, CYP1A2, and CYP1B1, demonstrates clearly that the AHR TAD deletion in resistant rats does not prevent canonical AHR-regulated genes from being induced by TCDD. Since these genes display a Type-I response they are unlikely to be central to the mechanism of lethality but likely represent an adaptive response. The arrays also detected many novel genes not previously known to be affected by TCDD. For some genes the magnitude of the Type-I responses was very large and ranged (in log2 space) from about 10 for CYP1A1 (i.e. a 1024-fold increase) to -7.2 for CYP7A1 (Table 3.2).

Type II responses: Type-II responses were observed for a total of 1084 ProbeSets after 3 h and 241 ProbeSets after 19 h TCDD exposure. Ninety six ProbeSets displayed persistent Type-II responses, i.e. they were classified as Type-II at both 3 h and at 19 h after TCDD (Figure 3.3C and Table 3.3; Full dataset available online). There were 179 ProbeSets up-regulated compared with 245 ProbeSets down-regulated in resistant rats (but not in sensitive rats) versus 252 ProbeSets up-regulated versus 408 ProbeSets down- regulated in sensitive rats (but not in resistant rats) after 3 h TCDD exposure. The arrays detected many genes that previously had been shown to be responsive to dioxins as well as numerous novel genes not previously known to be affected by TCDD. The magnitude of the Type II responses ranged from 5 to -2.5 (in log2 space). Genes that display a Type II response are leading candidates to play a mechanistic role in mediating dioxin lethality.

78 Table 3.1. Responses of genes that are well-known to be AHR-regulated and dioxin-inducible. Expression responses of ProbeSets representing genes known to respond to TCDD via the AHR signaling pathway were measured using Affymetrix RAE230A arrays. Each ProbeSet was annotated to a UniGene cluster and Entrez Gene ID using a BLAST strategy. For each collective (resistant or sensitive) the significance (“S”) of each ProbeSet was classified as “Yes” if all members of the collective was significantly -2 affected by TCDD (pmoderated < 10 ). If a ProbeSet was significantly affected by TCDD in all strains (Type I response gene), the “M” column gives the mean fold-change of differential expression between TCDD-treated and control samples (in log2 space) across all 4 strains of rats. If a ProbeSet was significantly affected by TCDD in only one collective (Type II response gene), then for each collective the mean fold-change (“M” in log2 space) was calculated for all collective members. “Delta” is the difference in magnitude of the fold-changes (in log2 space) between the collective of dioxin-sensitive rat strains and the collective of dioxin-resistant strains. For each gene we identified the total number of AH response elements (AHRE-I and AHRE-II) and the maximum phyloHMM conservation score in the region -10000 to +5000 relative to the transcription start site.

AHRE-I AHRE-I AHRE-I AHRE-II ProbeSet UniGene Entrez Resistant Sensitive Collective Name (core) (extended) (full) S M S M S M/Delta N Score N Score N Score N Score 1370269 at Rn.10352 24296 Yes Yes 10.0 Cytochrome P450, family 1, 13 0.99 8 0.98 3 0.98 1 0 subfamily a, polypeptide 1 1387243 at Rn.5563 24297 Yes Yes 0.7 Cytochrome P450, family 1, 4 0.72 0 NA 0 NA 0 NA subfamily a, polypeptide 2 1368990 at Rn.10125 25426 No 0.5 Yes 1.6 Yes -1.2 Cytochrome P450, family 1, 36 1 11 1 2 0 1 0 subfamily b, polypeptide 1 1368990 at Rn.10125 25426 Yes Yes 8.4 Cytochrome P450, family 1, 36 1 11 1 2 0 1 0 subfamily b, polypeptide 1 1387599 a at Rn.11234 24314 Yes Yes 3.7 NAD(P)H dehydrogenase, quinone 18 0.69 3 0.01 0 NA 1 0 1 1368130 at Rn.105627 25375 Yes Yes 7.6 Aldehyde dehydrogenase family 3, 18 0.46 6 0.33 0 NA 1 0 member A1 1369111 at Rn.26489 24861 Yes -0.3 No -0.1 Yes -0.2 UDP glycosyltransferase 1 family, 11 0.81 3 0.82 2 0.01 1 0 polypeptide A6 1387759 s at Rn.26489 24861 Yes Yes 1.7 UDP glycosyltransferase 1 family, 11 0.81 3 0.82 2 0.01 1 0 polypeptide A6 1369973 at Rn.9938 64677 No 0.5 Yes 1.1 Yes -0.6 Xanthine dehydrogenase 12 0.09 1 0.01 1 0 1 0

79 Table 3.2. Type I responses to TCDD. ProbeSets displaying Type I patterns of differential response to dioxins were identified using Affymetrix RAE230A arrays. Each ProbeSet was annotated to a UniGene cluster and Entrez Gene ID using a BLAST strategy. Features that could not be mapped to a UniGene cluster are designated as ‘NA’. The “Mean” column gives the mean fold-change of differential expression between TCDD-treated and control samples (in -2 log2 space) across all 4 strains for all ProbeSets significantly affected by TCDD in all strains (pmoderated < 10 ). For each gene we identified the total number of AH response elements (AHRE-I and AHRE-II) and the maximum phyloHMM conservation score in the region -10000 to +5000 relative to the transcription start site. A: Responses after 3 h TCDD treatment (55 positive upregulated and downregulated ProbeSets of the total 345 ProbeSets are presented). B: Responses after 3 h and 19 h TCDD treatment (all 39 ProbeSets are presented). C: Responses after 19 h TCDD treatment (37 upregulated and 20 downregulated of the total 137 ProbeSets are presented).

AHRE-I AHRE-I AHRE-I AHRE-II (core) (extended) (full) ProbeSet UniGene Entrez Mean Name N Score N Score N Score N Score

A: Responses after 3 h TCDD treatment 1370269 at Rn.10352 24296 10.0 Cytochrome P450, family 1, 13 0.988 8 0.982 3 0.98 1 0.002 subfamily a, polypeptide 1(Cyp1a1) 1368130 at Rn.105627 25375 7.6 Aldehyde dehydrogenase family 3, 18 0.457 6 0.329 0 NA 1 0 member A1 1369864 a at Rn.9918 25044 6.7 Serine dehydratase 15 1 1 0.001 0 NA 2 0.447 1374446 at Rn.24217 NA 4.2 RM1 mRNA, partial sequence NA NA NA NA NA NA NA NA 1368303 at Rn.25935 63840 3.9 Period homolog 2 (Per2) 32 0.997 6 0.992 2 0 2 0 1374483 at Rn.37600 NA 3.9 Transcribed locus NA NA NA NA NA NA NA NA 1373810 at NA NA 3.8 NA NA NA NA NA NA NA NA NA 1387599 a at Rn.11234 24314 3.7 NAD(P)H dehydrogenase, quinone 18 0.692 3 0.009 0 NA 1 0.002 1 1378016 at Rn.165165 NA 3.3 Transcribed locus NA NA NA NA NA NA NA NA 1370355 at Rn.1023 246074 3.2 Stearoyl-Coenzyme A desaturase 1 18 1 5 0.001 1 0.001 0 NA (Scd1) 1376827 at NA NA 3.2 NA NA NA NA NA NA NA NA NA 1372510 at NA NA 3.1 NA NA NA NA NA NA NA NA NA 1376267 at NA NA 3.0 NA NA NA NA NA NA NA NA NA 1372524 at Rn.23273 NA 3.0 Transcribed locus NA NA NA NA NA NA NA NA

80 1373238 at Rn.23978 360874 2.8 Transcriptional adaptor 1 (HFI1 NA NA NA NA NA NA NA NA homolog, yeast) like 1372600 at NA NA 2.8 NA NA NA NA NA NA NA NA NA 1369718 at Rn.3264 81784 2.8 Signal sequence receptor, gamma 4 0.002 1 0 0 NA 0 NA 1390036 at Rn.157381 303772 2.7 Solute carrier family 16 32 0.422 4 0.205 0 NA 0 NA (monocarboxylic acid transporters), member 6 1398451 at Rn.12417 NA 2.7 Transcribed locus NA NA NA NA NA NA NA NA 1374584 at Rn.165518 NA 2.6 Transcribed locus NA NA NA NA NA NA NA NA 1369958 at Rn.2042 64373 2.6 Ras homolog gene family, member 21 1 2 0.026 0 NA 2 0 B 1369150 at Rn.30070 89813 2.6 Pyruvate dehydrogenase kinase, 9 0.964 2 0 0 NA 1 0 isoenzyme 4 1388756 at Rn.15017 298490 2.5 Phosphopantothenoylcysteine 23 1 5 0.99 0 NA 2 0.085 synthetase 1387809 at Rn.17256 114495 2.4 Mitogen-activated protein kinase 17 1 3 0.013 1 0.075 1 0.001 kinase 6 1368272 at Rn.5819 24401 2.3 Glutamate oxaloacetate 17 1 5 0.007 0 NA 2 0.576 transaminase 1 1372512 at Rn.53006 360953 2.3 Syntaxin 18 20 0.997 2 0.009 1 0.002 2 0.009 1373035 at NA NA 2.3 NA NA NA NA NA NA NA NA NA 1376792 at NA NA 2.2 NA NA NA NA NA NA NA NA NA 1387573 a at Rn.42941 60349 2.2 Nuclear receptor subfamily 5, 9 1 3 1 1 0.009 1 0.005 group A, member 2 1373011 at Rn.162560 619558 2.2 Hypothetical protein LOC619558 15 0.089 4 0.03 0 NA 1 0.001 1370314 at Rn.162702 81826 2.1 Solute carrier family 20 (phosphate 12 0.994 2 0.003 0 NA 2 0.006 transporter), member 1 1377156 at NA NA 2.1 NA NA NA NA NA NA NA NA NA 1370286 at Rn.16393 29642 2.1 Solute carrier family 38, member 2 15 1 3 0.02 1 0 1 0.061 1372292 at Rn.164433 NA 2.1 Transcribed locus NA NA NA NA NA NA NA NA 1387907 at Rn.2135 25262 2.1 Inositol 1,4,5-triphosphate receptor 13 0.998 4 0.997 0 NA 3 0.02 1

81 1372316 at Rn.27 NA 2.0 Transcribed locus, strongly similar NA NA NA NA NA NA NA NA to XP 233838.3 PREDICTED: similar to Expressed sequence AW548124 [Rattus norvegicus]

1374507 at Rn.19289 298652 2.0 Ubiquitination factor E4B, UFD2 NA NA NA NA NA NA NA NA homolog (S. cerevisiae) (predicted) 1369467 a at Rn.10115 24638 2.0 6-phosphofructo-2-kinase/fructose- 14 0.999 3 0.997 0 NA 2 0.727 2,6-biphosphatase 1 1368223 at Rn.7897 79252 2.0 A disintegrin-like and 16 1 8 0.09 0 NA 2 0.774 metallopeptidase (reprolysin Type) with thrombospondin Type 1 motif, 1 1367826 at Rn.10867 83619 2.0 Nuclear factor, erythroid derived 2, 3 0.01 2 0.009 0 NA 1 0.002 like 2 (Nfe2l2) 1376076 at Rn.6259 NA 2.0 Transcribed locus, strongly similar NA NA NA NA NA NA NA NA to NP 076005.1 hypoxia-inducible protein 2 [Mus musculus] 1386987 at Rn.1716 24499 2.0 Interleukin 6 receptor, alpha 13 0.032 2 0.001 2 0.035 0 NA 1370195 at Rn.14789 64630 2.0 Synaptosomal-associated protein NA NA NA NA NA NA NA NA 23 1387665 at Rn.11406 81508 1.9 Betaine-homocysteine 15 1 6 0.993 0 NA 0 NA methyltransferase 1370905 at Rn.10431 259237 1.9 Dedicator of cytokinesis 9 NA NA NA NA NA NA NA NA 1376862 at Rn.19289 298652 1.9 Ubiquitination factor E4B, UFD2 NA NA NA NA NA NA NA NA homolog (S. cerevisiae) (predicted) 1377474 at Rn.8003 NA 1.9 Transcribed locus NA NA NA NA NA NA NA NA 1372181 at NA NA 1.9 NA NA NA NA NA NA NA NA NA 1387087 at Rn.6479 24253 1.8 CCAAT/enhancer binding protein 28 1 3 0.001 1 0 0 NA (C/EBP), beta 1372533 at Rn.164273 NA 1.7 Alpha-mannosidase-like protein NA NA NA NA NA NA NA NA mRNA, 3' UTR 1373233 at Rn.48641 NA 1.7 Transcribed locus NA NA NA NA NA NA NA NA 1398362 at NA NA 1.7 NA NA NA NA NA NA NA NA NA 1390146 at NA NA 1.6 NA NA NA NA NA NA NA NA NA

82 1390850 at Rn.101967 298199 1.6 Adipose differentiation related 7 1 2 0.002 0 NA 1 0.004 protein 1368692 a at Rn.10985 29194 1.5 Choline kinase alpha 14 0.815 3 0.055 1 0.002 0 NA 1367849 at Rn.11176 25216 -1.2 Syndecan 1 24 0.998 4 0.002 0 NA 1 0 1368082 at Rn.9860 24780 -1.2 Solute carrier family 4, member 2 15 1 2 1 0 NA 3 1 1372260 at Rn.995 287061 -1.2 Leucine zipper domain protein 14 1 2 0.002 0 NA 2 0.877 1386817 at NA NA -1.2 NA NA NA NA NA NA NA NA NA 1387020 at Rn.107152 25427 -1.2 Cytochrome P450, subfamily 51 14 1 2 0.997 0 NA 2 0.002 1371462 at Rn.160666 360622 -1.2 Insulin-like growth factor binding 19 1 5 0.013 1 0.001 0 NA protein 4 1372832 at Rn.166377 NA -1.3 Transcribed locus NA NA NA NA NA NA NA NA 1388583 at Rn.54439 24772 -1.3 Chemokine (C-X-C motif) ligand 28 0.721 1 0.224 0 NA 2 0.016 12 1371979 at Rn.41063 300095 -1.3 Sterol regulatory element binding 16 0.976 1 0.059 0 NA 1 0.006 factor 2 (predicted) 1368283 at Rn.3671 171142 -1.3 Enoyl-Coenzyme A, hydratase/3- 13 0.018 3 0.015 0 NA 3 0.01 hydroxyacyl Coenzyme A dehydrogenase 1371469 at Rn.11041 64152 -1.3 Calcium binding protein p22 9 0.996 1 0 1 0.003 1 0.008 1387388 at Rn.11041 64152 -1.3 Calcium binding protein p22 9 0.996 1 0 1 0.003 1 0.008 1388425 at Rn.1638 315594 -1.3 Similar to RIKEN cDNA 12 1 2 0.002 0 NA 1 0.999 D130038B21 1386892 at NA NA -1.3 NA NA NA NA NA NA NA NA NA 1374657 at NA NA -1.4 NA NA NA NA NA NA NA NA NA 1367644 at Rn.3313 25289 -1.4 Adenylate cyclase 6 33 1 5 0.905 2 0.001 2 0 1372069 at Rn.86207 309429 -1.4 Ankyrin repeat domain 15 NA NA NA NA NA NA NA NA 1389906 at Rn.154404 29580 -1.4 Farnesyl diphosphate farnesyl 11 0.373 1 0.001 0 NA 3 0.076 transferase 1 1387119 at Rn.6995 81727 -1.4 Mevalonate kinase NA NA NA NA NA NA NA NA 1387655 at Rn.54439 24772 -1.4 Chemokine (C-X-C motif) ligand 28 0.721 1 0.224 0 NA 2 0.016 12 1371202 a at Rn.9909 29227 -1.4 Nuclear factor I/B NA NA NA NA NA NA NA NA 1367839 at Rn.154404 29580 -1.4 Farnesyl diphosphate farnesyl 11 0.373 1 0.001 0 NA 3 0.076 transferase 1 1372602 at Rn.1935 305234 -1.5 Similar to genethonin 1 20 0.006 4 0.002 0 NA 3 0.038

83 1372601 at Rn.195729 NA -1.5 Activating transcription factor 5 NA NA NA NA NA NA NA NA 1374525 at NA NA -1.5 NA NA NA NA NA NA NA NA NA 1368232 at Rn.6995 81727 -1.5 Mevalonate kinase NA NA NA NA NA NA NA NA 1369633 at NA NA -1.6 NA NA NA NA NA NA NA NA NA 1370583 s at Rn.154810 170913 -1.6 ATP-binding cassette, sub-family 9 0.024 2 0 0 NA 0 NA B (MDR/TAP), member 1A 1387659 at Rn.161897 83585 -1.6 Guanine deaminase 14 0.872 4 0.032 0 NA 1 0 1370465 at Rn.154810 170913 -1.7 ATP-binding cassette, sub-family 9 0.024 2 0 0 NA 0 NA B (MDR/TAP), member 1A 1387165 at Rn.10726 54267 -1.7 V-maf musculoaponeurotic 25 1 4 0.968 0 NA 2 1 fibrosarcoma oncogene homolog (avian) 1368871 at Rn.11081 116667 -1.9 Mitogen activated protein kinase 13 1 1 0.001 0 NA 0 NA kinase kinase 1 1373814 at Rn.33733 362894 -2.0 Similar to mKIAA1002 protein NA NA NA NA NA NA NA NA 1367659 s at Rn.80835 29740 -2.0 Dodecenoyl-coenzyme A delta 14 0.96 3 0.96 1 NA 0 NA isomerase 1374266 at Rn.155488 NA -2.1 Transcribed locus, strongly similar NA NA NA NA NA NA NA NA to NP 083633.2 protocadherin 1 [Mus musculus] 1367982 at Rn.97126 65155 -2.1 Aminolevulinic acid synthase 1 NA NA NA NA NA NA NA NA 1368878 at Rn.10780 89784 -2.1 Isopentenyl-diphosphate delta 9 0.017 1 0.002 0 NA 0 NA isomerase 1368275 at Rn.7167 140910 -2.2 Sterol-C4-methyl oxidase-like 12 1 4 0.555 1 0.556 0 NA 1369658 at Rn.153603 24252 -2.3 CCAAT/enhancer binding protein 30 1 2 1 0 NA 1 0.007 (C/EBP), alpha 1388872 at Rn.167512 NA -2.6 Transcribed locus NA NA NA NA NA NA NA NA 1387022 at Rn.6132 24188 -2.7 Aldehyde dehydrogenase family 1, 3 0.005 0 NA 0 NA 1 0.023 member A1 1368168 at Rn.16933 84395 -2.9 Solute carrier family 34 (sodium 12 0.998 2 0 1 0.005 3 0.001 phosphate), member 2 1372091 at Rn.8267 404280 -3.7 MID1 interacting G12-like protein 12 1 4 1 0 NA 2 0.002

84 1368458 at Rn.10737 25428 -7.2 Cytochrome P450, family 7, 11 0.998 4 0.992 0 NA 2 0.922 subfamily a, polypeptide 1 (Cyp7a1)

B: Responses after 3 h and 19 h TCDD treatment 1370269 at Rn.10352 24296 9.9 Cytochrome P450, family 1, 13 0.988 8 0.982 3 0.98 1 0.002 subfamily a, polypeptide 1 (Cyp1a1) 1368130 at Rn.105627 25375 9.3 Aldehyde dehydrogenase family 3, 18 0.457 6 0.329 0 NA 1 0 member A1 1374446 at Rn.24217 NA 4.0 RM1 mRNA, partial sequence NA NA NA NA NA NA NA NA 1387599 a at Rn.11234 24314 3.9 NAD(P)H dehydrogenase, quinone 18 0.692 3 0.009 0 NA 1 0.002 1 1373810 at NA NA 3.5 NA NA NA NA NA NA NA NA NA 1374584 at Rn.165518 NA 2.8 Transcribed locus NA NA NA NA NA NA NA NA 1376827 at NA NA 2.5 NA NA NA NA NA NA NA NA NA 1390146 at NA NA 2.4 NA NA NA NA NA NA NA NA NA 1378016 at Rn.165165 NA 2.4 Transcribed locus NA NA NA NA NA NA NA NA 1372600 at NA NA 2.3 NA NA NA NA NA NA NA NA NA 1372316 at Rn.27 NA 2.3 Transcribed locus, strongly similar NA NA NA NA NA NA NA NA to XP 233838.3 PREDICTED: similar to Expressed sequence AW548124 [Rattus norvegicus] 1374656 at Rn.6753 252881 2.2 Exocyst complex component 3 7 0.571 2 0.004 1 0.002 1 0 1398451 at Rn.12417 NA 2.2 Transcribed locus NA NA NA NA NA NA NA NA 1368692 a at Rn.10985 29194 2.1 Choline kinase alpha (Chka) 14 0.815 3 0.055 1 0.002 0 NA 1376862 at Rn.19289 298652 1.8 Ubiquitination factor E4B, UFD2 NA NA NA NA NA NA NA NA homolog (S. cerevisiae) (predicted) 1388751 at NA NA 1.8 NA NA NA NA NA NA NA NA NA 1367826 at Rn.10867 83619 1.8 Nuclear factor, erythroid derived 2, 3 0.01 2 0.009 0 NA 1 0.002 like 2 (Nfe2l2) 1383443 at NA NA 1.7 NA NA NA NA NA NA NA NA NA 1374507 at Rn.19289 298652 1.7 Ubiquitination factor E4B, UFD2 NA NA NA NA NA NA NA NA homolog (S. cerevisiae) (predicted) 1398362 at NA NA 1.6 NA NA NA NA NA NA NA NA NA 1377141 at Rn.24906 NA 1.6 Transcribed locus, strongly similar NA NA NA NA NA NA NA NA to XP 573969.1 PREDICTED:

85 similar to RIKEN cDNA 1110018J18 [Rattus norvegicus] 1373667 at NA NA 1.5 NA NA NA NA NA NA NA NA NA 1386908 at Rn.1484 64045 1.4 Glutaredoxin 1 (thioltransferase) 5 0.014 1 0 0 NA 1 0.006 1371592 at Rn.2759 315707 1.3 C-src tyrosine kinase 16 1 2 0.998 1 0 2 0.005 1367906 at Rn.9816 24162 1.3 Acid phosphatase 2, lysosomal 7 0.999 1 0.013 0 NA 0 NA 1387630 at Rn.153081 171400 1.3 ELOVL family member 5, 14 0.021 4 0.024 0 NA 2 0.014 elongation of long chain fatty acids (yeast) 1394117 at NA NA 1.3 NA NA NA NA NA NA NA NA NA 1374620 at Rn.91235 81613 1.2 CEA-related cell adhesion 21 0.162 5 0.006 1 0.001 1 0.009 molecule 1 1367669 a at Rn.41412 64862 1.2 Microtubule-associated protein 1 28 0.259 4 0.005 0 NA 2 0.027 light chain 3 beta 1387027 a at Rn.10706 25476 1.1 Lectin, galactose binding, soluble 9 6 0.044 2 0.001 1 0.001 4 0.002 1387243 at Rn.5563 24297 1.1 Cytochrome P450, family 1, 4 0.72 0 NA 0 NA 0 NA subfamily a, polypeptide 2 1388883 at Rn.104646 361698 1.0 Polymerase (DNA-directed), delta 6 1 1 0.03 0 NA 1 0.72 4 1388348 at Rn.196077 NA 0.9 Transcribed locus NA NA NA NA NA NA NA NA 1369716 s at Rn.64588 25475 0.9 Lectin, galactose binding, soluble 5 8 0.999 4 0.998 1 0.002 4 0.005 1373824 at Rn.86172 NA 0.9 Transcribed locus, strongly similar NA NA NA NA NA NA NA NA to NP 955410.1 craniofacial development protein 1 [Rattus norvegicus] 1388767 at Rn.104071 NA 0.5 Transcribed locus, strongly similar NA NA NA NA NA NA NA NA to XP 217732.2 PREDICTED: similar to ALG-2 [Rattus norvegicus] 1373079 at NA NA -0.3 NA NA NA NA NA NA NA NA NA 1373337 at Rn.7815 NA -0.6 PREDICTED: Rattus norvegicus NA NA NA NA NA NA NA NA glyoxylate reductase/hydroxypyruvate reductase (predicted) (Grhpr predicted), mRNA 1372426 at Rn.62115 310670 -0.9 ADAMTS-like 4 NA NA NA NA NA NA NA NA

C: Responses after 19 h TCDD treatment

86 1368130 at Rn.105627 25375 10.2 Aldehyde dehydrogenase family 3, 18 0.457 6 0.329 0 NA 1 0 member A1 1370269 at Rn.10352 24296 9.8 Cytochrome P450, family 1, 13 0.988 8 0.982 3 0.98 1 0.002 subfamily a, polypeptide 1 1368990 at Rn.10125 25426 8.4 Cytochrome P450, family 1, 36 1 11 1 2 0.001 1 0 subfamily b, polypeptide 1 1387599 a at Rn.11234 24314 4.0 NAD(P)H dehydrogenase, quinone 18 0.692 3 0.009 0 NA 1 0.002 1 1374446 at Rn.24217 NA 4.0 RM1 mRNA, partial sequence NA NA NA NA NA NA NA NA 1373810 at NA NA 3.4 NA NA NA NA NA NA NA NA NA 1389624 s at Rn.6753 252881 3.1 Exocyst complex component 3 7 0.571 2 0.004 1 0.002 1 0 1374584 at Rn.165518 NA 2.9 Transcribed locus NA NA NA NA NA NA NA NA 1368947 at Rn.10250 25112 2.8 Growth arrest and DNA-damage- 11 1 1 0.002 0 NA 1 0.001 inducible 45 alpha 1390146 at NA NA 2.8 NA NA NA NA NA NA NA NA NA 1374656 at Rn.6753 252881 2.6 Exocyst complex component 3 7 0.571 2 0.004 1 0.002 1 0 1370686 at Rn.6753 252881 2.6 Exocyst complex component 3 7 0.571 2 0.004 1 0.002 1 0 1372316 at Rn.27 NA 2.5 Transcribed locus, strongly similar NA NA NA NA NA NA NA NA to XP 233838.3 PREDICTED: similar to Expressed sequence AW548124 [Rattus norvegicus] 1368692 a at Rn.10985 29194 2.5 Choline kinase alpha 14 0.815 3 0.055 1 0.002 0 NA 1388751 at NA NA 2.2 NA NA NA NA NA NA NA NA NA 1376827 at NA NA 2.2 NA NA NA NA NA NA NA NA NA 1372479 at Rn.8739 NA 2.1 Transcribed locus, moderately NA NA NA NA NA NA NA NA similar to NP 064456.1 fibrinogen, beta polypeptide [Rattus norvegicus] 1372600 at NA NA 2.1 NA NA NA NA NA NA NA NA NA 1383443 at NA NA 2.0 NA NA NA NA NA NA NA NA NA 1367673 at Rn.16617 140927 2.0 Selenium binding protein 2 7 0.975 1 0.982 0 NA 5 0.034 1378016 at Rn.165165 NA 2.0 Transcribed locus NA NA NA NA NA NA NA NA 1369854 a at Rn.91235 81613 2.0 CEA-related cell adhesion 21 0.162 5 0.006 1 0.001 1 0.009 molecule 1 1398451 at Rn.12417 NA 1.9 Transcribed locus NA NA NA NA NA NA NA NA 1368180 s at Rn.144550 24422 1.9 Glutathione-S-transferase, alpha 8 0.152 2 0 0 NA 1 0.003 Type2

87 1367691 at Rn.12281 85332 1.8 Protein kinase C, delta binding 15 1 4 0.01 0 NA 1 0.003 protein 1367705 at Rn.1484 64045 1.8 Glutaredoxin 1 (thioltransferase) 5 0.014 1 0 0 NA 1 0.006 1376862 at Rn.19289 298652 1.8 Ubiquitination factor E4B, UFD2 NA NA NA NA NA NA NA NA homolog (S. cerevisiae) (predicted) 1377141 at Rn.24906 NA 1.7 Transcribed locus, strongly similar NA NA NA NA NA NA NA NA to XP 573969.1 PREDICTED: similar to RIKEN cDNA 1110018J18 [Rattus norvegicus] 1373667 at NA NA 1.7 NA NA NA NA NA NA NA NA NA 1368536 at Rn.20403 84050 1.7 Ectonucleotide 3 0.999 0 NA 0 NA 0 NA pyrophosphatase/phosphodiesterase 2 1387759 s at Rn.26489 24861 1.7 UDP glycosyltransferase 1 family, 11 0.811 3 0.817 2 0.007 1 0.001 polypeptide A6 1386908 at Rn.1484 64045 1.7 Glutaredoxin 1 (thioltransferase) 5 0.014 1 0 0 NA 1 0.006 1367826 at Rn.10867 83619 1.7 Nuclear factor, erythroid derived 2, 3 0.01 2 0.009 0 NA 1 0.002 like 2 1383606 at Rn.43919 500707 1.6 Membrane targeting (tandem) C2 0 NA 0 NA 0 NA 1 0.009 domain containing 1 1398362 at NA NA 1.6 NA NA NA NA NA NA NA NA NA 1374507 at Rn.19289 298652 1.5 Ubiquitination factor E4B, UFD2 NA NA NA NA NA NA NA NA homolog (S. cerevisiae) (predicted) 1387630 at Rn.153081 171400 1.5 ELOVL family member 5, 14 0.021 4 0.024 0 NA 2 0.014 elongation of long chain fatty acids (yeast) 1372426 at Rn.62115 310670 -0.9 ADAMTS-like 4 NA NA NA NA NA NA NA NA 1367604 at Rn.4267 338401 -0.9 Cysteine-rich protein 2 22 1 3 0.996 0 NA 2 0 1386881 at Rn.26369 24484 -0.9 Insulin-like growth factor binding 31 1 8 0.92 0 NA 0 NA protein 3 1367755 at Rn.2589 81718 -0.9 Cysteine dioxygenase 1, cytosolic 2 0.006 0 NA 0 NA 2 0.012 1370200 at Rn.55106 24399 -1.0 Glutamate dehydrogenase 1 17 1 4 1 2 0.163 1 0.002 1387567 at Rn.5641 170698 -1.0 Solute carrier organic anion NA NA NA NA NA NA NA NA transporter family, member 1a4 1387307 at Rn.10037 29301 -1.0 Histidine ammonia lyase 27 0.998 10 0.989 1 0.003 1 0.001 1386960 at Rn.1592 29573 -1.0 Solute carrier family 37 (glycerol- 15 0.998 3 0.998 2 0.999 0 NA 6-phosphate transporter), member 4 1389601 at Rn.40435 NA -1.1 CDNA clone IMAGE:7308494 NA NA NA NA NA NA NA NA

88 1369485 at Rn.59648 170570 -1.1 Acyl-CoA thioesterase 12 14 0.128 4 0.044 0 NA 0 NA 1372715 at Rn.115752 364678 -1.1 Sideroflexin 1 15 0.996 5 0.997 0 NA 1 0.008 1370375 at Rn.10202 192268 -1.2 Glutaminase 2 (liver, 23 0.912 3 0.013 0 NA 1 0.012 mitochondrial) 1375247 at Rn.40396 29254 -1.2 Monoglyceride lipase 8 0.431 2 0.009 0 NA 2 0.001 1387228 at Rn.89295 25351 -1.2 Solute carrier family 2 (facilitated 8 0.007 5 0.051 0 NA 3 0.003 glucose transporter), member 2 1371012 at Rn.21425 85255 -1.6 Phytanoyl-CoA 2-hydroxylase 2 13 0.072 3 0.081 1 0.007 0 NA 1368059 at Rn.24561 117024 -1.7 Crystallin, mu NA NA NA NA NA NA NA NA 1370387 at Rn.10489 171352 -1.8 Cytochrome P450, family 3, 3 0.9 1 0.009 0 NA 0 NA subfamily a, polypeptide 13 1389785 at Rn.22552 293653 -2.0 Aspartoacylase (aminoacylase) 3 16 0.341 1 0.001 0 NA 3 0.011 1387959 at Rn.29979 246266 -3.5 Lysophospholipase 15 0.027 4 0.002 0 NA 1 0.012 1373975 at Rn.19133 NA -4.2 Transcribed locus, strongly similar NA NA NA NA NA NA NA NA to XP 347234.2 PREDICTED: similar to thioether S- methyltransferase [Rattus norvegicus]

89 Table 3.3. Type II responses to TCDD ProbeSets displaying Type II patterns of differential response to dioxins were identified using Affymetrix RAE230A arrays. Each ProbeSet was annotated, analyzed, and associated with TFBS data as described in the legend to Table 3.1. For each collective (resistant or sensitive) the significance (“S”) of each ProbeSet was classified as “Yes” if all members of the collective were -2 significantly affected by TCDD (pmoderated < 10 ). For each collective, the mean fold-change (“M” in log2 space) was calculated for all collective members. Only ProbeSets that exhibited a significant difference between the resistant and sensitive collectives are reported. Genes are ordered according to the magnitude of expression differences (“Delta”) between the collective of dioxin-sensitive rat strains and the collective of dioxin-resistant strains. This Delta value is the difference in magnitude of the fold-changes (in log2 space) between the two collectives. A: Responses after 3 h TCDD treatment (Delta cut off 1.1 and -1.1; 90 ProbeSets of the 1084 total ProbeSets are presented). B: Responses persistent after 3 h and 19 h TCDD treatment (15 positive delta and 15 negative delta ProbeSets of the 96 total ProbeSets are presented). C: Responses after 19 h TCDD treatment (20 positive delta and 20 negative delta ProbeSets of the 241 total ProbeSets are presented).

ProbeSet UniGene Entrez Resistant Sensitive Collective AHRE-I AHRE-I AHRE-I AHRE-II Collective Collective Differences Name (core) (extended) (full) S M S M S Delta N Score N Score N Score N Score

A: Responses after 3 h TCDD treatment 1370281 at Rn.197758 NA No 1.9 Yes 5.0 Yes -3.0 Similar to Fatty acid-binding protein, NA NA NA NA NA NA epidermal (E-FABP) (Cutaneous fatty acid- binding protein) (C-FABP) (DA11) 1388108 at Rn.46942 171402 No 2.0 Yes 5.0 Yes -3.0 ELOVL family member 6, elongation of 26 1 5 0.999 2 0.066 1 0.163 long chain fatty acids (yeast) (Elovl6) 1372318 at Rn.147228 NA No 1.7 Yes 4.3 Yes -2.6 Transcribed locus, strongly similar to NP NA NA NA NA NA NA NA NA 035560.1 synuclein, gamma [Mus musculus]

1389189 at Rn.6401 81634 No 0.1 Yes 2.1 Yes -2.0 Actinin, alpha 1 13 0.997 3 0.005 0 NA 1 0.008 1370114 a at Rn.10599 25513 Yes -1.9 No 0.0 Yes -2.0 Phosphatidylinositol 3-kinase, regulatory 13 1 4 0.994 1 0.996 2 0.423 subunit, polypeptide 1 (Pik3r1)

1370051 at Rn.10039 60335 No 0.2 Yes 2.0 Yes -1.9 Transglutaminase 1 9 1 0 NA 0 NA 1 0.001 1372472 at Rn.154513 298848 No 0.0 Yes 1.8 Yes -1.8 Microtubule-associated protein, RP/EB 11 1 3 1 0 NA 3 1 family, member 3

90 1371104 at Rn.801 78968 Yes -2.8 No -1.1 Yes -1.7 Sterol regulatory element binding factor 1 NA NA NA NA NA NA NA NA 1376972 at Rn.165138 NA No 0.7 Yes 2.4 Yes -1.7 Transcribed locus NA NA NA NA NA NA NA NA 1370111 at Rn.44421 54262 No 0.1 Yes 1.8 Yes -1.7 Potassium intermediate/small conductance 20 1 2 1 1 0.992 3 0.998 calcium-activated channel, subfamily N, member 2 1388786 at Rn.7071 NA No 1.0 Yes 2.7 Yes -1.7 Transcribed locus NA NA NA NA NA NA NA NA 1390114 at Rn.8087 360871 No 0.5 Yes 2.2 Yes -1.6 Myelin protein zero-like 1 11 1 4 0.016 1 0.034 1 0.004 1374914 at NA NA Yes -1.8 No -0.3 Yes -1.5 NA NA NA NA NA NA NA NA NA 1371542 at Rn.92961 316531 Yes -1.9 No -0.4 Yes -1.5 Tubulin, alpha 4 14 1 1 0.075 0 NA 3 0.075 1370695 s at Rn.22325 246273 No 0.1 Yes 1.6 Yes -1.5 Tribbles homolog 3 (Drosophila) (Trib3) 16 1 2 0.371 3 0 0 NA 1388426 at NA NA Yes -2.1 No -0.6 Yes -1.5 NA NA NA NA NA NA NA NA NA 1372404 at Rn.2863 366957 No 0.1 Yes 1.6 Yes -1.5 RAS-related C3 botulinum substrate 2 14 0.4 1 0 0 NA 4 0.002 1372973 at NA NA Yes -1.9 No -0.4 Yes -1.5 NA NA NA NA NA NA NA NA NA 1368340 at Rn.15187 171458 No -0.2 Yes 1.2 Yes -1.5 Inositol polyphosphate multikinase 10 0.982 4 0.217 0 NA 0 NA 1373718 at NA NA Yes -1.1 No 0.4 Yes -1.5 NA NA NA NA NA NA NA NA NA 1388641 at Rn.161547 NA No 0.0 Yes 1.5 Yes -1.4 Transcribed locus NA NA NA NA NA NA NA NA 1368086 a at Rn.10211 81681 Yes -1.7 No -0.2 Yes -1.4 Lanosterol synthase 6 0.025 1 0 0 NA 6 0.004 1371400 at Rn.81140 25357 No 0.4 Yes 1.8 Yes -1.4 Thyroid hormone responsive protein (Thrsp) 5 0.323 0 NA 0 NA 1 0

1367652 at Rn.26369 24484 No 0.0 Yes 1.4 Yes -1.4 Insulin-like growth factor binding protein 3 31 1 8 0.92 0 NA 0 NA

1390113 a at NA NA No 0.7 Yes 2.1 Yes -1.3 NA NA NA NA NA NA NA NA NA 1368453 at Rn.162483 83512 Yes -1.4 No -0.1 Yes -1.3 Fatty acid desaturase 2 10 0.03 2 0.003 0 NA 3 0.009 1371211 a at Rn.37438 112400 No 1.0 Yes 2.3 Yes -1.3 Neuregulin 1 27 1 5 1 0 NA 2 0.003 1370959 at Rn.3247 84032 No -0.4 Yes 0.9 Yes -1.3 Procollagen, Type III, alpha 1 5 0 2 0 1 0 1 0.009 1369733 at Rn.112601 84353 No 0.5 Yes 1.7 Yes -1.3 Catenin (cadherin associated protein), beta 1 15 1 5 1 0 NA 0 NA

1368103 at Rn.8398 85264 No -0.1 Yes 1.1 Yes -1.3 ATP-binding cassette, sub-family G 16 0.997 2 0.004 0 NA 1 0.009 (WHITE), member 1 1387134 at NA NA No 1.2 Yes 2.5 Yes -1.2 NA NA NA NA NA NA NA NA NA 1387852 at Rn.81140 25357 No 0.9 Yes 2.1 Yes -1.2 Thyroid hormone responsive protein (Thrsp) 5 0.323 0 NA 0 NA 1 0

1387017 at Rn.33239 29230 Yes -2.2 No -1.0 Yes -1.2 Squalene epoxidase 12 1 3 0.61 0 NA 0 NA

91 1370150 a at Rn.81140 25357 No 0.7 Yes 1.9 Yes -1.2 Thyroid hormone responsive protein 5 0.323 0 NA 0 NA 1 0 (Thrsp) 1371646 at Rn.915 NA No 0.4 Yes 1.6 Yes -1.2 PREDICTED: Rattus norvegicus NA NA NA NA NA NA NA NA phosphogluconate dehydrogenase (Pgd), mRNA 1376051 at Rn.57632 290277 Yes -1.0 No 0.2 Yes -1.2 Crystallin, lamda 1 10 0.971 3 0 1 0.09 4 0.003 1376675 at Rn.103180 361725 No 0.1 Yes 1.3 Yes -1.2 Secretoglobin, family 2A, member 2 2 0.553 2 0.557 0 NA 0 NA 1387799 at Rn.6700 29639 No -0.1 Yes 1.1 Yes -1.2 FXYD domain-containing ion transport 6 0.001 1 0 0 NA 2 0 regulator 2 1375925 at Rn.66757 NA No 0.0 Yes 1.2 Yes -1.2 Transcribed locus NA NA NA NA NA NA NA NA 1368990 at Rn.10125 25426 No 0.5 Yes 1.6 Yes -1.2 Cytochrome P450, family 1, subfamily b, 36 1 11 1 2 0.001 1 0 polypeptide 1 1368474 at Rn.11267 25361 No 0.4 Yes 1.5 Yes -1.2 Vascular cell adhesion molecule 1 6 0.736 0 NA 0 NA 2 0.688 1369181 at Rn.98491 66021 No 0.0 Yes 1.2 Yes -1.1 Cytochrome b-245, beta polypeptide 12 1 1 0.002 0 NA 2 0.003 1369204 at Rn.10945 25734 No -0.1 Yes 1.0 Yes -1.1 Hemopoietic cell kinase 12 0.683 1 0.002 0 NA 1 0.04 1374953 at NA NA No 0.3 Yes 1.4 Yes -1.1 NA NA NA NA NA NA NA NA NA 1367940 at Rn.12959 84348 No 0.5 Yes 1.7 Yes -1.1 Chemokine orphan receptor 1 12 0.023 2 0.002 0 NA 1 0.002 1370694 at Rn.22325 246273 No 0.2 Yes 1.3 Yes -1.1 Tribbles homolog 3 (Drosophila) 16 1 2 0.371 3 0 0 NA 1372558 at NA NA No 0.0 Yes 1.1 Yes -1.1 NA NA NA NA NA NA NA NA NA 1387001 at Rn.4586 116546 No 0.3 Yes 1.4 Yes -1.1 V-ral simian leukemia viral oncogene 15 0.174 5 0.393 1 0.001 1 0.653 homolog B 1386862 at Rn.3318 25673 No 0.3 Yes 1.4 Yes -1.1 Annexin A5 20 1 4 0.003 0 NA 1 0.003 1368174 at Rn.10994 54702 No 0.0 Yes 1.1 Yes -1.1 EGL nine homolog 3 (C. elegans) 10 0.903 2 0.001 0 NA 2 1 1370940 at Rn.10965 115769 Yes -1.5 No -0.5 Yes -1.1 Tight junction protein 2 5 1 1 0.007 0 NA 1 0.013 1367758 at NA NA No 0.1 Yes 1.2 Yes -1.1 NA NA NA NA NA NA NA NA NA 1390431 at Rn.22062 NA No -0.6 Yes -1.6 Yes 1.1 Transcribed locus NA NA NA NA NA NA NA NA 1388267 a at Rn.54397 24567 Yes 2.2 No 1.2 Yes 1.1 Metallothionein 1a (Mt1a) 28 1 4 0.002 2 0.003 2 0.002 1374329 at Rn.164921 NA No 0.0 Yes -1.1 Yes 1.1 Transcribed locus NA NA NA NA NA NA NA NA 1370209 at Rn.19481 117560 Yes 2.2 No 1.1 Yes 1.1 Kruppel-like factor 9 35 1 4 0.889 0 NA 1 0.001 1389179 at Rn.8171 NA No -0.2 Yes -1.3 Yes 1.1 Transcribed locus, strongly similar to XP NA NA NA NA NA NA NA NA 214551.3 PREDICTED: similar to cell death activator CIDE-A [Rattus norvegicus] 1368509 at Rn.15987 113948 Yes 1.4 No 0.3 Yes 1.1 Bardet-Biedl syndrome 2 homolog (human) 19 0.333 6 0.002 0 NA 1 NA

92 1369099 at Rn.10120 58976 No -0.2 Yes -1.3 Yes 1.1 Solute carrier family 30 (zinc transporter), 17 1 5 1 2 0.972 0 NA member 1 1386885 at Rn.6148 64526 No -0.8 Yes -1.9 Yes 1.1 Enoyl coenzyme A hydratase 1, peroxisomal 20 1 5 1 0 NA 1 0 1394234 x at Rn.166202 NA Yes 1.1 No 0.0 Yes 1.1 Transcribed locus NA NA NA NA NA NA NA NA 1369123 a at Rn.10025 24218 No 0.0 Yes -1.1 Yes 1.1 ATPase, Cu++ transporting, beta polypeptide 24 1 4 0.013 1 0.021 0 NA

1379853 at Rn.98778 NA No 0.0 Yes -1.1 Yes 1.1 Transcribed locus, strongly similar to XP NA NA NA NA NA NA NA NA 204090.1 PREDICTED: hypothetical protein LOC72938 isoform 1 [Mus musculus]

1376865 at Rn.49261 NA No -0.5 Yes -1.7 Yes 1.2 Transcribed locus NA NA NA NA NA NA NA NA 1385243 at Rn.164470 NA No -0.4 Yes -1.5 Yes 1.2 Transcribed locus NA NA NA NA NA NA NA NA 1387740 at Rn.14519 85249 No 0.1 Yes -1.1 Yes 1.2 Peroxisomal biogenesis factor 11A 12 1 2 0.978 1 0.001 0 NA 1371038 at Rn.10332 25301 No 0.2 Yes -1.0 Yes 1.2 CCAAT/enhancer binding protein (C/EBP), NA NA NA NA NA NA NA NA gamma 1368435 at Rn.23013 81924 No -0.7 Yes -1.9 Yes 1.2 Cytochrome P450, family 8, subfamily b, 11 1 1 0.003 0 NA 3 0.002 polypeptide 1 1374903 at Rn.8807 306860 No 0.2 Yes -1.0 Yes 1.2 Glucosaminyl (N-acetyl) transferase 2, I- 9 0.031 5 0.022 0 NA 0 NA branching enzyme 1376658 at NA NA No -0.6 Yes -1.8 Yes 1.2 NA NA NA NA NA NA NA NA NA 1374636 at Rn.165782 NA No -0.1 Yes -1.4 Yes 1.3 Transcribed locus NA NA NA NA NA NA NA NA 1371361 at NA NA No -0.2 Yes -1.6 Yes 1.3 NA NA NA NA NA NA NA NA NA 1370371 a at Rn.91235 81613 Yes 2.4 No 1.0 Yes 1.3 CEA-related cell adhesion molecule 1 21 0.162 5 0.006 1 0.001 1 0.009 1388901 at Rn.144288 361810 Yes 2.2 No 0.9 Yes 1.4 FK506 binding protein 5 13 0.353 5 0.02 1 0.041 0 NA 1369248 a at Rn.91239 63879 No -0.2 Yes -1.6 Yes 1.4 Baculoviral IAP repeat-containing 4 6 0.999 2 0.019 0 NA 1 NA 1374962 at Rn.92582 NA Yes 1.7 No 0.3 Yes 1.4 CDNA clone IMAGE:7132213 NA NA NA NA NA NA NA NA 1379044 at NA NA No 0.3 Yes -1.2 Yes 1.5 NA NA NA NA NA NA NA NA NA 1368947 at Rn.10250 25112 Yes 2.4 No 0.9 Yes 1.5 Growth arrest and DNA-damage-inducible 11 1 1 0.002 0 NA 1 0.001 45 alpha 1387116 at Rn.29778 24908 Yes 1.5 No 0.0 Yes 1.5 DnaJ (Hsp40) homolog, subfamily B, 11 1 2 1 0 NA 1 0.007 member 9 1373740 at NA NA Yes 1.6 No 0.1 Yes 1.6 NA NA NA NA NA NA NA NA NA

93 1377713 at Rn.105640 NA No -0.6 Yes -2.3 Yes 1.6 Transcribed locus, strongly similar to XP NA NA NA NA NA NA NA NA 234377.3 PREDICTED: similar to checkpoint suppressor 1 [Rattus norvegicus]

1370147 at Rn.162389 171385 Yes 2.5 No 0.8 Yes 1.7 2-amino-3-carboxymuconate-6- 1 0.001 0 NA 0 NA 1 0.001 semialdehyde decarboxylase 1370750 a at Rn.9758 25663 Yes 3.6 No 1.8 Yes 1.8 Interleukin 1 receptor, Type I 17 0.01 4 0.003 0 NA 3 0.041 1371081 at Rn.42890 252857 Yes 1.6 No -0.2 Yes 1.8 Rap guanine nucleotide exchange factor NA NA NA NA NA NA NA NA (GEF) 4 1375796 at Rn.176886 NA No -0.6 Yes -2.5 Yes 1.8 Transcribed locus, strongly similar to XP NA NA NA NA NA NA NA NA 577096.1 PREDICTED: similar to Ac2-233 [Rattus norvegicus] 1368650 at Rn.2398 81813 Yes 2.0 No 0.1 Yes 1.9 Kruppel-like factor 10 (Klf10) 21 1 5 0.166 0 NA 2 0.003 1388271 at Rn.115549 NA Yes 4.1 No 2.2 Yes 1.9 Transcribed locus, moderately similar to XP NA NA NA NA NA NA NA NA 529783.1 PREDICTED: hypothetical protein XP 529783 [Pan troglodytes] 1376816 at Rn.164344 NA Yes 3.2 No 1.2 Yes 2.0 Transcribed locus NA NA NA NA NA NA NA NA 1368491 at Rn.41640 59296 Yes 2.1 No 0.0 Yes 2.1 Deoxyribonuclease II beta 11 0.149 5 0.002 0 NA 0 NA 1371237 a at Rn.16133 24567 Yes 3.6 No 1.2 Yes 2.4 Metallothionein 1a (Mt1a) 28 1 4 0.002 2 0.003 2 0.002

B: Responses common after 3 h and 19 h TCDD treatment 1373740 at NA NA Yes 2.2 No 0.0 Yes 2.2 NA NA NA NA NA NA NA NA NA 1387188 at Rn.11150 171080 Yes 1.4 No 0.1 Yes 1.3 Solute carrier family 17 (sodium phosphate), 0 NA 0 NA 0 NA 1 0 member 1 1377713 at Rn.105640 NA No -0.6 Yes -1.7 Yes 1.1 Transcribed locus, strongly similar to XP NA NA NA NA NA NA NA NA 234377.3 PREDICTED: similar to checkpoint suppressor 1 [Rattus norvegicus] 1376062 at Rn.11176 25216 No -1.0 Yes -1.8 Yes 0.8 Syndecan 1 (Sdc1) 24 0.998 4 0.002 0 NA 1 0 1376627 at Rn.171954 NA No -0.4 Yes -1.1 Yes 0.7 Transcribed locus NA NA NA NA NA NA NA NA 1372457 at Rn.95344 306487 No -0.5 Yes -1.2 Yes 0.7 Mitochondrial tumor suppressor 1 NA NA NA NA NA NA NA NA 1367849 at Rn.11176 25216 No -0.8 Yes -1.5 Yes 0.7 Syndecan 1 (Sdc1) 24 0.998 4 0.002 0 NA 1 0 1379073 at Rn.41 NA No -0.2 Yes -0.8 Yes 0.6 Transcribed locus NA NA NA NA NA NA NA NA 1388545 at NA NA No -0.4 Yes -1.0 Yes 0.6 NA NA NA NA NA NA NA NA NA 1369854 a at Rn.91235 81613 Yes 2.0 No 1.4 Yes 0.6 CEA-related cell adhesion molecule 1 21 0.162 5 0.006 1 0.001 1 0.009

94 1370553 at Rn.10623 25130 No -0.4 Yes -1.0 Yes 0.6 Epimorphin 15 0.619 2 0.003 1 0.626 0 NA 1371202 a at Rn.9909 29227 No -0.8 Yes -1.4 Yes 0.6 Nuclear factor I/B NA NA NA NA NA NA NA NA 1398759 at Rn.3545 25564 No -1.3 Yes -1.9 Yes 0.6 Transforming growth factor beta 1 induced NA NA NA NA NA NA NA NA transcript 4 (Tgfb1i4) 1371787 at Rn.162119 NA No -0.2 Yes -0.8 Yes 0.6 PREDICTED: Rattus norvegicus similar to NA NA NA NA NA NA NA NA RIKEN cDNA 2310005N03 (LOC502991), mRNA 1368947 at Rn.10250 25112 Yes 2.7 No 2.2 Yes 0.4 Growth arrest and DNA-damage-inducible 11 1 1 0.002 0 NA 1 0.001 45 alpha 1383606 at Rn.43919 500707 No 1.1 Yes 1.5 Yes -0.4 Membrane targeting (tandem) C2 domain 0 NA 0 NA 0 NA 1 0.009 containing 1 1387665 at Rn.11406 81508 No 1.2 Yes 1.7 Yes -0.5 Betaine-homocysteine methyltransferase 15 1 6 0.993 0 NA 0 NA 1371996 at NA NA No 0.4 Yes 0.9 Yes -0.5 NA NA NA NA NA NA NA NA NA 1367652 at Rn.26369 24484 No -0.7 Yes -0.2 Yes -0.5 Insulin-like growth factor binding protein 3 31 1 8 0.92 0 NA 0 NA 1375440 at Rn.22468 360746 No 0.1 Yes 0.6 Yes -0.5 Peptidylprolyl isomerase (cyclophilin)-like 2 4 0.094 0 NA 0 NA 2 0.176 1372292 at Rn.164433 NA No 1.2 Yes 1.7 Yes -0.5 Transcribed locus NA NA NA NA NA NA NA NA 1371646 at Rn.915 NA No 1.1 Yes 1.6 Yes -0.5 PREDICTED: Rattus norvegicus NA NA NA NA NA NA NA NA phosphogluconate dehydrogenase (Pgd), mRNA 1370686 at Rn.6753 252881 No 1.7 Yes 2.2 Yes -0.5 Exocyst complex component 3 7 0.571 2 0.004 1 0.002 1 0 1389624 s at Rn.6753 252881 No 2.0 Yes 2.6 Yes -0.6 Exocyst complex component 3 7 0.571 2 0.004 1 0.002 1 0 1370195 at Rn.14789 64630 No 1.4 Yes 2.0 Yes -0.6 Synaptosomal-associated protein 23 NA NA NA NA NA NA NA NA 1398955 at Rn.99919 363283 No 0.4 Yes 1.0 Yes -0.6 COP9 (constitutive photomorphogenic) 7 1 3 0.996 0 NA 3 0.019 homolog, subunit 8 (Arabidopsis thaliana) 1387105 at Rn.105961 171110 No 0.6 Yes 1.3 Yes -0.7 Zinc finger protein 422 (predicted) 10 0.989 4 0.011 0 NA 2 0.044 1368990 at Rn.10125 25426 No 5.3 Yes 6.7 Yes -1.4 Cytochrome P450, family 1, subfamily b, 36 1 11 1 2 0.001 1 0 polypeptide 1 (Cyp1b1) 1368303 at Rn.25935 63840 No 3.1 Yes 4.5 Yes -1.4 Period homolog 2 (Drosophila) (Per2) 32 0.997 6 0.992 2 0 2 0 1367940 at Rn.12959 84348 No 0.4 Yes 2.0 Yes -1.6 Chemokine orphan receptor 1 12 0.023 2 0.002 0 NA 1 0.002

C: Responses common after 19 h TCDD treatment 1376640 at Rn.194330 NA No 0.1 Yes 2.2 Yes -2.1 Transcribed locus NA NA NA NA NA NA NA NA 1390317 at Rn.18085 NA No 1.0 Yes 3.0 Yes -2.0 Transcribed locus NA NA NA NA NA NA NA NA 1374939 at Rn.194014 NA No 0.4 Yes 2.3 Yes -1.9 Transcribed locus NA NA NA NA NA NA NA NA ` Rn.12959 84348 No 0.3 Yes 2.2 Yes -1.9 Chemokine orphan receptor 1 12 0.023 2 0.002 0 NA 1 0.002 1370823 at Rn.25267 83837 No 0.1 Yes 1.5 Yes -1.4 BMP and activin membrane-bound inhibitor, 13 0.804 5 0.844 0 NA 3 0.145

95 homolog (Xenopus laevis) 1368303 at Rn.25935 63840 No 3.0 Yes 4.4 Yes -1.3 Period homolog 2 (Drosophila) 32 0.997 6 0.992 2 0 2 0 1372980 at Rn.1696 NA No 1.4 Yes 2.2 Yes -0.8 Transcribed locus, strongly similar to XP NA NA NA NA NA NA NA NA 575419.1 PREDICTED: similar to Hypothetical protein MGC30714 [Rattus norvegicus] 1374483 at Rn.37600 NA No 1.6 Yes 2.3 Yes -0.7 Transcribed locus NA NA NA NA NA NA NA NA 1369923 at Rn.124477 58922 No 0.6 Yes 1.4 Yes -0.7 OG9 homeobox gene 18 1 3 0.997 0 NA 3 0.959 1372292 at Rn.164433 NA No 0.8 Yes 1.5 Yes -0.7 Transcribed locus NA NA NA NA NA NA NA NA 1370195 at Rn.14789 64630 No 1.2 Yes 1.9 Yes -0.7 Synaptosomal-associated protein 23 NA NA NA NA NA NA NA NA 1373823 at Rn.119060 NA Yes -1.0 No -0.4 Yes -0.7 Hypothetical protein LOC686524 NA NA NA NA NA NA NA NA 1372619 at Rn.149379 NA No 0.3 Yes 0.9 Yes -0.7 Transcribed locus, moderately similar to NP NA NA NA NA NA NA NA NA 080522.1 mitochondrial ribosomal protein L49 [Mus musculus] 1398955 at Rn.99919 363283 No 0.5 Yes 1.1 Yes -0.6 COP9 (constitutive photomorphogenic) 7 1 3 0.996 0 NA 3 0.019 homolog, subunit 8 (Arabidopsis thaliana) 1387665 at Rn.11406 81508 No 0.9 Yes 1.5 Yes -0.6 Betaine-homocysteine methyltransferase 15 1 6 0.993 0 NA 0 NA 1387105 at Rn.105961 171110 No 0.8 Yes 1.4 Yes -0.6 Zinc finger protein 422 (predicted) 10 0.989 4 0.011 0 NA 2 0.044 1375440 at Rn.22468 360746 No 0.1 Yes 0.6 Yes -0.5 Peptidylprolyl isomerase (cyclophilin)-like 2 4 0.094 0 NA 0 NA 2 0.176 1368509 at Rn.15987 113948 No 0.8 Yes 1.3 Yes -0.5 Bardet-Biedl syndrome 2 homolog (human) 19 0.333 6 0.002 0 NA 1 NA 1373238 at Rn.23978 360874 No 0.9 Yes 1.4 Yes -0.5 Transcriptional adaptor 1 (HFI1 homolog, NA NA NA NA NA NA NA NA yeast) like 1374617 at Rn.32147 287380 No 0.6 Yes 1.1 Yes -0.5 Dehydrogenase/reductase (SDR family) 10 1 0 NA 0 NA 0 NA member 7B 1371412 a at Rn.8180 NA Yes -3.1 No -2.6 Yes -0.5 Neuronal regeneration related protein NA NA NA NA NA NA NA NA 1398759 at Rn.3545 25564 No -1.4 Yes -2.0 Yes 0.7 Transforming growth factor beta 1 induced NA NA NA NA NA NA NA NA transcript 4 1389906 at Rn.154404 29580 No -0.4 Yes -1.1 Yes 0.7 Farnesyl diphosphate farnesyl transferase 1 11 0.373 1 0.001 0 NA 3 0.076 1387094 at Rn.5641 170698 No -2.6 Yes -3.3 Yes 0.7 Solute carrier organic anion transporter NA NA NA NA NA NA NA NA family, member 1a4 1377375 at Rn.3169 NA No -1.6 Yes -2.3 Yes 0.7 Transcribed locus, strongly similar to XP NA NA NA NA NA NA NA NA 231524.3 PREDICTED: similar to Aminoadipate-semialdehyde synthase [Rattus norvegicus] 1376627 at Rn.171954 NA No -0.2 Yes -0.9 Yes 0.7 Transcribed locus NA NA NA NA NA NA NA NA 1371202 a at Rn.9909 29227 No -0.6 Yes -1.3 Yes 0.7 Nuclear factor I/B NA NA NA NA NA NA NA NA 1376974 at Rn.16768 362696 No -0.2 Yes -1.0 Yes 0.7 Tetratricopeptide repeat domain 7 NA NA NA NA NA NA NA NA

96 1373790 at Rn.159812 NA No -1.6 Yes -2.4 Yes 0.8 PREDICTED: Rattus norvegicus carbonic NA NA NA NA NA NA NA NA anhydrase 14 (predicted) (Car14 predicted), mRNA 1367849 at Rn.11176 25216 No -0.8 Yes -1.5 Yes 0.8 Syndecan 1 24 0.998 4 0.002 0 NA 1 0 1368239 at Rn.87059 89787 No -0.3 Yes -1.1 Yes 0.8 Low density lipoprotein receptor-related 22 1 3 1 0 NA 4 0.049 protein 3 1372457 at Rn.95344 306487 No -0.4 Yes -1.2 Yes 0.8 Mitochondrial tumor suppressor 1 NA NA NA NA NA NA NA NA 1377713 at Rn.105640 NA No -0.5 Yes -1.4 Yes 0.8 Transcribed locus, strongly similar to XP NA NA NA NA NA NA NA NA 234377.3 PREDICTED: similar to checkpoint suppressor 1 [Rattus norvegicus] 1373693 at Rn.3658 287805 No -0.5 Yes -1.4 Yes 0.8 G protein-coupled receptor, family C, group NA NA NA NA NA NA NA NA 5, member C 1376062 at Rn.11176 25216 No -1.0 Yes -1.9 Yes 0.9 Syndecan 1 24 0.998 4 0.002 0 NA 1 0 1367896 at Rn.1647 54232 No -0.5 Yes -1.5 Yes 1.0 Carbonic anhydrase 3 11 0.999 3 0.999 0 NA 1 0.415 1393902 at Rn.102679 307092 No -1.1 Yes -2.3 Yes 1.2 Aldo-keto reductase family 1, member C6 1 0.002 0 NA 0 NA 0 NA 1390591 at Rn.16357 266730 Yes 1.8 No 0.2 Yes 1.6 Solute carrier family 17 (sodium phosphate), 2 0 0 NA 0 NA 2 NA member 3 1387188 at Rn.11150 171080 Yes 1.8 No 0.1 Yes 1.7 Solute carrier family 17 (sodium phosphate), 0 NA 0 NA 0 NA 1 0 member 1 1373740 at NA NA Yes 2.5 No 0.0 Yes 2.5 NA NA NA NA NA NA NA NA NA

97 2. Genes whose altered expression, as detected by the arrays, was confirmed by real- time RT-PCR We conducted a broad-scale search for Type I and Type II responses that represent early transcriptional events by using array methods to measure mRNA levels at 3 h and 19 h post-TCDD. Although array methods are becoming very reliable, we further employed real-time RT-PCR as an independent method to evaluate early transcriptional responses to TCDD. To this end, 7 Type-I and 8 Type-II genes (as classified by array analysis) were selected for RT-PCR analysis based on the following criteria: (a) the altered ProbeSet could be mapped to a known gene; (b) the multiple ProbeSets that represent the same gene had similar responses to TCDD; (c) the gene contains AHRE regulatory motifs; and (d) the protein encoded by the gene is plausibly involved in AHR-mediated dioxin lethality. We also selected genes so that some genes that exhibited a low magnitude of response were evaluated by RT-PCR along genes that showed a high magnitude of response.

Type I responses: CYP1A1 is a well established Type-I response gene that is highly inducible via the AHR pathway (Denison and Whitlock, 1995; Nebert et al., 2004; Okey et al., 2005). As expected, both the array (Tables 3.1 and 3.2) and RT-PCR analyses show that CYP1A1 mRNA was highly up-regulated at both 3 h and 19 h after TCDD treatment in all 4 rat strains/lines (Figure 3.4). Therefore, CYP1A1 was confirmed to be a Type-I responder gene. RT-PCR analyses revealed that responses to TCDD for Nrep, Cyp7a1 Chk, Crip2, Per2, and Selenbp2 were similar between dioxin-sensitive and dioxin-resistant rat strains/lines, thus confirming the array classification of these genes as Type I responses (Table 3.2; Figure 3.4). Our previous array analysis (Franc et. al. in preparation) and our current array analysis comparing resistant versus sensitive rat collectives both classified the following genes as Type I responses: Akap11, Itm2b, Ldha, Nfe2l2, Bhmt, RM1, Cdo1, Acaa2, Map1lc3b, Glud1, Ahcy, and Adk. Moreover, our previous RT-PCR analysis (Franc et. al. in preparation) revealed upregulation of expression levels of Nfe2l2, Glud1, Slc2a2, and Sdc1 by TCDD both in sensitive L-E rats and in resistant H/W rats, in agreement with our current array analysis. The arrays detected increased mRNA levels of

98 Mt1a and decreased levels of Sdc1 (2 ProbeSets) and Tgfb1i4 after TCDD treatment in sensitive rats only; some responses for these genes were noted in resistant rats but at the limit of statistical significance. Our previous RT-PCR analysis, using the same rat model, classified Mt1a, Sdc1, and Tgfb1i4 as Type I responses. Therefore, Mt1a, Sdc1, and Tgfb1i4 are likely Type I responders. Thus, 14 genes classified as Type I by array studies were confirmed to be genuine Type I responders by RT-PCR analyses; these Type I genes are unlikely to directly mediate susceptibility or resistance to dioxin lethality. Three genes classified as Type II by the arrays (Mt1a, Sdc1, and Tgfb1i4), were reclassified as Type I responses by the rigorous RT-PCR analyses.

99 Cyp1a1 100 *** *** *** Control 75 *** 3 h TCDD 50 19 h TCDD 25

NDND ND ND 0 H/W LnA L-E LnC Nrep Cyp7a1 100 100 ) *** *** * * 75 75 *** *** **

50 50 25 25

0 0 H/W LnA L-E LnC H/W LnA L-E LnC Chka Crip2 100 100 * *** 75 ** * *** ** 75 50 50

25 25 Normalized mRNAexpression

( percent of maximal expression 0 0 H/W LnA L-E LnC H/W LnA L-E LnC

Per2 Selenbp2 100 100

** *** * 75 75 ***** *** ***

50 50

25 25 0 0 H/W LnA L-E LnC H/W LnA L-E LnC Figure 3.4. Type I TCDD-responsive genes from livers of dioxin-resistant and dioxin-sensitive rats: Real-time RT-PCR measurement of mRNA levels for selected genes. Hepatic RNA was prepared from male adult TCDD-sensitive rats (L-E and LnC; AHRWT/WT) and TCDD-resistant rats (H/W and LnA; AHRH/W/H/W) after treatment with a single dose of 100 µg/kg TCDD (T) or corn-oil vehicle (C) by gavage. Animals were sacrificed at 3 or 19 after dosing. mRNA levels were measured by real-time RT-PCR and normalized to Actb as described in Materials & Methods. For each gene, the mRNA level that was highest for any strain or treatment was set at 100% and all other mRNA levels for that gene are shown as a percentage of the maximal level. All results plotted represent the mean of 4 rats ± standard deviation. Asterisks indicate significant differences in mRNA levels (t-test; two-sided, equal variance as supported by F-test, * p<0.05, ** p<0.01, *** p<0.001). Note: levels of Cyp1a1 in control animals were below detection limits (indicated by ND) and levels of Chka, Crip2, Per2, and Selenbp2 after 3-h TCDD treatment were not measured.

100 Type II responses: Array analyses indicated decreased mRNA expression levels of Cdh2 (2 ProbeSets) and Ghr in the dioxin-sensitive rat collective 19 h after TCDD treatment relative to controls but did not detect any significant effects in the resistant collective (Table 3.3). In agreement with the array results, measurements by RT-PCR also revealed decreased mRNA levels for Cdh2, and Ghr in L-E rats, but not in H/W rats 19 h after TCDD treatment (Figure 3.5).

Both array and RT-PCR analyses detected a significant increase in mRNA levels for Elovl6, Thrsp (represented by 2 ProbeSets on the array) and Trib3 (represented by 2 ProbeSets) in L-E 3 h after TCDD treatment. Increased expression of the genes in L-E and LnC rats 19 h after TCDD treatment was suggested by both the arrays and RT-PCR but was not significant due to higher biological variability. Neither array nor RT-PCR analysis detected any significant response for Elovl6, Thrsp or Trib3 in the resistant rat collective at either duration of TCDD exposure.

RT-PCR measured an increase in mRNA levels of Klf10 and Pik3r1 in dioxin-sensitive L-E rats 3 h after TCDD treatment but not at 19 h, supporting their role as early transcriptional Type II responses. In H/W rats, there was evidence of decreased Pik3r1 mRNA levels by RT-PCR; however, the significance was at threshold level. Arrays indicated that mRNA levels of Klf10 increased and Pik3r1 decreased in H/W rats after 3 h TCDD exposure but not 19 h.

In summary, RT-PCR and array analysis both classified Cdh2, Ghr, Elovl6, Klf10, Pik3r1, Thrsp, and Trib3 as Type II responses. Further, the Type I responses for 7 transcripts CYP1A1, Nrep, Cyp7a1, Chka, Crip2, Per2, and Selenbp2 quantitated by RT- PCR were in strong accord with the array responses.

101 Cdh2

100 * Control 75 3 h TCDD 50 19 h TCDD 25 0 H/W L-E Ghr Elovl6 ** 100 100 *** 75 75 ***

50 50

25 25 0 0 H/W LnA L-E LnC H/W L-E Klf10 Pik3r1 100 *** 100 *** 75 75 50 50

25 25 Normalized mRNA expression expression mRNA Normalized

( percent of maximal expression ) 0 0 H/W LnA L-E LnC H/W LnA L-E LnC Thrsp Trib3 100 100 *** *** 75 75

50 50 25 25

0 0 H/W LnA L-E LnC H/W LnA L-E LnC Figure 3.5. Type II TCDD-responsive genes from livers of dioxin-resistant and dioxin-sensitive rats: Real-time RT-PCR measurement of mRNA levels for selected genes. Animals were treated and mRNA was prepared and quantitated as described in the legend to Figure 4. Note: samples were not available for the 3-h time point for LnA and LnC rats and mRNA levels of Cdh2, and Ghr were measured only for 19 h post-TCDD samples. All results plotted represent the mean of 4 rats ± standard deviation. Asterisks indicate significant differences in mRNA levels (t-test; two-sided, unequal variance, * p<0.05, ** p<0.01, *** p<0.001).

102 3. Genes whose response to TCDD potentially is regulated by the AHR

3a. In silico identification of putative AH response elements (AHREs) Increased expression of genes such as CYP1A1 and Nfe2l2, well-known to be AHR-regulated and dioxin-inducible, involves binding of the TCDD-AHR-ARNT complex to an AHRE-I motif located in the 5’-flanking region. More recently a second motif, AHRE-II, was identified in the rat CYP1A2 gene (Sogawa et al., 2004) and in 36 other genes that are responsive to dioxin-like chemicals (Boutros et al., 2004). To determine if AHRE-I or AHRE-II motifs exist in novel Type I and Type II genes that responded to TCDD in our array study we annotated each ProbeSet with AHRE binding- site data wherever existing rat genomic information permitted, (Tables 3.1, 3.2, and 3.3). Out of 249 putative Type I responsive ProbeSets that could be mapped to the genome, 247 contained core AHRE-I motifs, 227 contained extended motifs and 73 contained full AHRE-I motifs within the 5'-flanking region of genomic sequence. AHRE-II motifs were identified in genes represented by 171 ProbeSets (the full Type I response list with AHRE-motif information will be available on-line). Phylogenetic analysis (phyloHMM) indicates that the AHRE-I motif is highly conserved for Type II genes: out of 606 putative Type II responsive clones that could be mapped to the genome, 596 contained core, 507 contained extended, and 178 contained full AHRE-I motifs within the 5'- flanking region and 401 ProbeSets contained AHRE-II motifs (the full Type II response list with AHRE-motif information will be available on-line).

3b. Gene expression in Ahr-/- mice in vivo To determine if candidate genes from our array experiments on rat liver truly depend on the AHR in order to respond to TCDD we used real-time RT-PCR to evaluate mRNA levels for 4 Type I and 4 Type II genes in livers from Ahr-null mice (Ahr-/-) versus Ahr-wildtype mice (Ahr+/+) (Figure 3.6). It is well-established that major toxicities of TCDD require the AHR (Bunger et al., 2003; Fernandez-Salguero et al., 1995; Mimura et al., 1997; Schmidt et al., 1996; Walisser et al., 2004b). Therefore genes whose mRNA levels are dysregulated by TCDD in Ahr-wildtype mice but not in Ahr- null-mice are the most plausible candidates to be involved in dioxin toxicity. Both the

103 classic AHR-regulated gene, Cyp1a1 (upregulated), and Crip2 (downregulated) were affected by TCDD in mice with wildtype AHR but did not respond to TCDD in Ahr-null mice (Figure 3.6). However, mRNA levels for Chk, Selenbp2, Pik3r1, Elovl6, Klf10, and Thrsp, which all were altered by TCDD in the rat model, were not affected by TCDD in AHR-wildtype mice (Figure 3.6). In the mouse model levels of mRNAs for Chk and Pik3r1 were altered by TCDD independent of the AHR; therefore, regulation of these candidates by TCDD is species-specific. This finding agrees with recent findings of species-specific regulation of gene expression between rat versus mouse (Boverhof et al., 2006). Basally, the AHR appears to suppress the mRNA expression of Chk and increase the expression of Pik3r1, independently of TCDD exposure.

104

Type IControl TCDD Type II Cyp1a1 Pik3r1 100 *** 100 * ** 75 75 50 50

25 25 ND ND ND 0 0 -/- Wildtype Ahr -/- Wildtype Ahr Chk Elovl6 100 *** *** 100

75 75 50 50

25 25 0 -/- 0 Wildtype Ahr Wildtype Ahr -/-

Crip2 Klf10 100 100 *** 75 75

50 * 50

Normalized mRNA expression ( percent ) response of maximal 25 25 0 0 Wildtype Ahr -/- -/- Wildtype Ahr Selenbp2 Thrsp 100 100

75 75 50 50

25 25

0 0 Wildtype Ahr -/- Wildtype Ahr -/- Figure 3.6. Ahr-null mouse model: measurement of selected mRNA levels by real- time RT-PCR. Hepatic RNA was prepared, as described in Materials and Methods, from male adult Ahr-null mice (Ahr-/-) and wildtype C57BL/6J mice (Ahr+/+) after treatment with a single dose of 1000 µg/kg TCDD or corn oil vehicle for 19 h. There were 3 TCDD-treated and 3 control mice in the Ahr-null groups and 4 TCDD- treated and 4 control mice in the wildtype groups. Levels for selected mRNAs, that had been identified in rats as Type I or Type II responses to TCDD by array and RT-PCR analyses, were measured by real-time RT-PCR and normalized to Actb. For each gene, the mRNA level that was highest for any strain or treatment was set at 100% and all other mRNA levels for that gene are shown as a percentage of the maximal level. Error bars represent standard deviation of the mean. Asterisks indicate significant differences in mRNA levels (ANOVA followed by Bonferroni post hoc tests, * p<0.05, ** p<0.01, *** p<0.001). Note that for Cyp1a1 the mRNA level in control animals or in TCDD-treated Ahr-/- mice or control Ahr-/- mice is below the detection limit of the assay; thus there are no bars visible for these groups in this plot (indicated in the plot by ND).

105 4. Genes that are components of biological processes/pathways associated with phenotypes altered by TCDD exposure

Functional Categorization of Responsive Genes by Gene Ontology (GO) Analysis To determine if different combinations of strain and durations of TCDD exposure led to alterations in functionally coherent groups of genes, GO analysis was performed with the candidate gene-list for each rat strain. To test for independence from parameter- -9 selection, thresholds of pcum < 0.05 (Figure 3.7A) and pcum < 10 (Figure 3.7b) both are presented (Figure 3.7). From a broad functional perspective, dioxin-resistant LnA rats were quite similar to the dioxin-resistant H/W rats, while dioxin-sensitive LnC rats were much closer to dioxin-sensitive L-E (Figure 3.7A). Following hierarchical clustering of each dataset, the 3-h time point was dramatically separated from the 19-h time point. Notably, genes involved in sterol biosynthesis are perturbed only at 3 h, whereas those involved in xenobiotic metabolism and amino-acid metabolism are not perturbed until 19 hours (Figure 3.7B). In theory, one would expect to identify both Type I and Type II genes naturally from the heatmap and its dendrogram. However, Figure 3.7 shows that, in a threshold-independent manner, all dysregulated GO bioprocesses were fit into only one of the following categories: (a) variant by time-point but not by strain; (b) variant by strain, but not in sensitive vs. resistant manner; or (c) invariant. Therefore, the analysis did not reveal bioprocesses that are consistently dysregulated in a fashion that differs between sensitive and resistant rats. To determine which functional groups of genes might be enriched in resistant rats vs. sensitive rats GO analysis was performed on all ProbeSets affected by TCDD in the resistant collective and then in the sensitive collective (including both Type I and II responses). There were 19 significant bioprocesses enriched in the resistant collective and 5 in the sensitive collective; 17 of these bioprocesses were common to both resistant and to sensitive collectives (Table 3.4). GO analysis was also performed for Type II responses in H/W vs. L-E classified after 3 h TCDD exposure. Four biological processes were enriched in resistant rats (metabolism: very-long-chain fatty acid, cellular lipid, lipid, and organic acid) while no bioprocesses where enriched in sensitive rats after 3 h TCDD exposure.

106

A)

B)

Figure 3.7. Functional analysis. Gene Ontology (GO) analysis was used to determine if different combinations of strain and time- point led to alterations in functionally coherent groups of genes. The gene-list for each strain was tested for enrichment of each GO category represented on the RAE230A array. False-discovery rates were calculated with 5000 permutations of the dataset using the High-Throughput GoMiner software. The cumulative (joint) FDR for each GO term was calculated and two separate thresholds were used: a relaxed threshold of 0.05 (Panel A) and a tight threshold of 10-9 (Panel B).

107

Table 3.4. Gene Ontological (GO) analysis of the TCDD effect on transcript expression in dioxin-resistant and/or dioxin sensitive rats All transcripts whose expression was significantly alter by TCDD after 3 h or 19 h in resistant rats was subjected to GO analysis. Likewise, all transcripts altered by TCDD in sensitive rats were subjected to GO analysis. GO analysis used the GOMiner software package (Zeeberg et al., 2003), which performs a two-tailed Fisher’s exact test to determine over-representation of GO bioprocesses in the candidate gene list relative to the array as a whole. “Total genes” represent the total number of ProbeSets on the array within this category, counting (without duplication) all the ProbeSets of all of its descendant categories. “TCDD genes” represent the number of unique ProbeSets within the GO category, counting (without duplication) the TCDD affected ProbeSets of all of its descendant GO categories. “Enrichment” is the proportion of TCDD genes in the GO category relative to the expected proportion: the ratio of changed genes in the category divided by the total number of genes in the category, divided by the same ratio for the entire microarray. False Discovery Rate (FDR) is the one-sided Fisher exact p-value corrected for multiple comparisons. Go Categories in bold are unique to resistant collectives or unique to sensitive collectives.

GO Category Total TCDD Enrich FDR genes genes ment Resistant rat GO:0044270 nitrogen compound catabolism 45 10 2.9 4.8E-02 GO:0019395 fatty acid oxidation 27 8 3.9 2.9E-02 GO:0006635 fatty acid beta-oxidation 20 7 4.6 2.6E-02 GO:0006805 xenobiotic metabolism 23 7 4.0 4.0E-02 GO:0009410 response to xenobiotic stimulus 24 7 3.8 4.6E-02 GO:0006817 phosphate transport 10 5 6.6 2.6E-02 GO:0045540 regulation of cholesterol biosynthesis 5 4 10.5 8.4E-03 GO:0050810 regulation of steroid biosynthesis 7 4 7.5 3.9E-02 GO:0000038 very-long-chain fatty acid metabolism 3 3 13.2 2.7E-02 GO:0050875 cellular physiological process 3785 315 1.1 1.0E-03 GO:0008152 metabolism 2492 234 1.2 2.0E-04 GO:0044237 cellular metabolism 2340 222 1.3 5.0E-04 GO:0044238 primary metabolism 2265 204 1.2 1.1E-02 GO:0006629 lipid metabolism 361 60 2.2 0.0E+00 GO:0009058 biosynthesis 539 60 1.5 3.8E-02 GO:0006082 organic acid metabolism 305 53 2.3 0.0E+00 GO:0044255 cellular lipid metabolism 303 53 2.3 0.0E+00 GO:0019752 carboxylic acid metabolism 303 52 2.3 0.0E+00 GO:0008610 lipid biosynthesis 136 26 2.5 4.6E-04 GO:0006631 fatty acid metabolism 124 25 2.7 3.3E-04 GO:0008202 steroid metabolism 105 22 2.8 5.0E-04 GO:0006694 steroid biosynthesis 52 15 3.8 3.8E-04 GO:0016125 sterol metabolism 47 14 3.9 4.0E-04 GO:0008203 cholesterol metabolism 43 12 3.7 2.4E-03 GO:0016126 sterol biosynthesis 21 10 6.3 4.3E-04 GO:0006695 cholesterol biosynthesis 19 8 5.6 1.9E-03

Sensitive Rat

108 GO:0006091 generation of precursor metabolites 215 38 1.7 1.4E-02 and energy GO:0006066 alcohol metabolism 150 32 2.1 3.0E-03 GO:0006520 amino acid metabolism 136 28 2.0 9.4E-03 GO:0009069 serine family amino acid metabolism 16 7 4.3 2.1E-02 GO:0046148 pigment biosynthesis 7 5 7.0 9.4E-03 GO:0050875 cellular physiological process 3785 416 1.1 8.9E-03 GO:0008152 metabolism 2492 313 1.2 0.0E+00 GO:0044237 cellular metabolism 2340 296 1.2 0.0E+00 GO:0044238 primary metabolism 2265 272 1.2 3.6E-03 GO:0009058 biosynthesis 539 80 1.4 9.7E-03 GO:0006629 lipid metabolism 361 66 1.8 0.0E+00 GO:0006082 organic acid metabolism 305 60 1.9 0.0E+00 GO:0019752 carboxylic acid metabolism 303 60 1.9 0.0E+00 GO:0044255 cellular lipid metabolism 303 58 1.9 0.0E+00 GO:0008610 lipid biosynthesis 136 27 1.9 1.6E-02 GO:0006631 fatty acid metabolism 124 26 2.0 9.3E-03 GO:0008202 steroid metabolism 105 25 2.3 2.9E-03 GO:0006694 steroid biosynthesis 52 15 2.8 7.8E-03 GO:0016125 sterol metabolism 47 15 3.1 2.8E-03 GO:0008203 cholesterol metabolism 43 13 3.0 9.3E-03 GO:0016126 sterol biosynthesis 21 10 4.7 1.6E-03 GO:0006695 cholesterol biosynthesis 19 9 4.6 2.7E-03

109 DISCUSSION This gene expression study was intended mainly to identify those genes whose dysregulation by dioxins may be integral to the mechanism of lethality and associated toxicities. The extraordinary resistance of the H/W rat to TCDD toxicity provides a unique opportunity to determine whether altered expression of particular genes is associated with major phenotypic differences in the toxic endpoints of wasting and lethality. Our array experiments revealed that a very large number of genes are responsive to TCDD in rat liver. Far fewer genes were categorized as Type-I responders (i.e., the response in the dioxin-sensitive rat collective was similar to the response of these genes in the dioxin-resistant rat collective) than Type-II responders. Genes that responded in a Type-I fashion include genes that are well-known to be AHR-regulated and dioxin- inducible such as CYP1A1, CYP1A2, CYP1B1, Gsta3, UGT1A6, TiPARP, Nqo1 and Nfe2l2 (also known as Nrf2) (Kohle and Bock, 2007; Nebert et al., 2004). It is clear that the large deletion from the transactivation domain of the AHR in H/W rats does not interfere with the ability of the receptor to drive induction of these well-characterized genes. Because genes that are well-known to be AHR-regulated and dioxin-inducible displayed a Type I response to TCDD they are unlikely to account for the dioxin- resistance phenotype or to be key components in pathways to lethality from TCDD. Rather, transcriptional upregulation of these Type-I responders probably serves as an adaptive mechanism that enhances clearance of potentially toxic xenobiotics (Gu et al., 2000; Nebert et al., 2004; Okey, 1990). Although genes that are well-known to be AHR- regulated and dioxin-inducible may not be directly responsible for major toxicities from TCDD, CYP1A1 can modify TCDD-induced hepatotoxicity and lethality as evidenced by the substantial protection that knockout of the Cyp1a1 gene confers on male mice (Uno et al., 2004) whereas knockout of Cyp1a1 or Cyp1b1 genes does not affect dioxin-mediated teratogenesis (Dragin et al., 2006). As described above, genes that are well-known to be AHR-regulated and dioxin- inducible remain responsive to TCDD in rats with the AHRH/W genotype. However, the total number of genes that respond in animals with this genotype is notably reduced; 236 fewer ProbeSets responded to 3 h dioxin treatment and 196 fewer respond to 19 h dioxin

110 treatment in rats with the AHRH/W genotype than in rats with wildtype AHR. AH receptors from H/W rats do not differ from wildtype AHR in their affinity for TCDD nor in their ability to bind AH response elements on DNA (Pohjanvirta et al., 1999). The predominant form of AHR expressed in rats that carry the AHRH/W genotype is the AHRH/W insertion-variant which has intrinsic transactivation activity that is similar to that of the wildtype AHR (Moffat et al., 2007). The TAD deletion in rats with the AHRH/W genotype does diminish the total number of dioxin-responsive genes in these animals notwithstanding the similarities between the variant AHRH/W and wildtype AHR in affinity for TCDD, binding to AHREs and intrinsic transactivation activity. The genes that are most central to TCDD toxicity may lie among the Type II genes that no longer respond to TCDD or among novel responses to TCDD gained in rats that carry the AHRH/W genotype. Before discussing the Type II genes it should be mentioned that many of the Type I genes which emerged from our array studies and subsequent bioinformatic analyses also are involved in multiple biological processes that are important in toxicology. For example: xenobiotic metabolism (described above), protection from oxidative stress, hypercholesterolemia, amino acid metabolism, nitrogen metabolism, and organic acid metabolism. We will focus the remainder of the discussion on Type II genes since these are the most plausible candidates to explain the mechanistic underpinnings of dioxin-sensitivity/dioxin-resistance.

Type II responses: those that differ between the resistant rat collective and the sensitive rat collective Several toxic outcomes from TCDD exposure are classified as Type II responses because they exhibit a clear difference between resistant vs. sensitive rat strains. Type II toxic endpoints include acute TCDD lethality (Pohjanvirta et al., 1993), wasting syndrome (Pohjanvirta et al., 1987), increased serum tryptophan (Unkila et al., 1994b), hepatotoxicity (Pohjanvirta et al., 1989a; Simanainen et al., 2002; Simanainen et al., 2003), and liver tumor promotion (Viluksela et al., 2000). The following sections discuss genes whose expression is altered by TCDD in a Type II fashion and therefore are candidates for involvement in these manifestations of toxicity.

111 Wasting syndrome A wasting syndrome ensues within the first few days following a single dose of TCDD characterized by progressive weight loss (up to 50%) and hypophagia (Pohjanvirta and Tuomisto, 1994; Seefeld et al., 1984a; Seefeld et al., 1984b). Animals resume feeding within 1-2 weeks if the dose is sublethal but body weight never returns to that of untreated animals. A lethal dose produces irreversible hypophagia and weight loss, probably due to a lowering of the body weight set-point (Pohjanvirta and Tuomisto, 1994). Wasting contributes to lethality starting 2-3 weeks after TCDD exposure but mechanisms that mediate the wasting syndrome are poorly understood. Many pathways have been proposed to explain TCDD-induced wasting, including gluconeogenic pathways (Weber et al., 1991), c-Src-mediated pathways (Vogel et al., 2003), stress- response pathways (Matsumura, 2003), and complex adaptive pathways including disturbances of lipid, carbohydrate, and nitrogen metabolism (Fletcher et al., 2005). Our multi-strain analysis, contrasting resistant vs. sensitive rat collectives, reveals changes in multiple metabolic pathways and, additionally, points to the potential importance of a balance between protein degradation and protein synthesis in manifestations of the wasting syndrome.

Proteolysis via ubiquitin-proteasomal pathways. In wasting diseases (e.g. diabetes, cancer cachexia, and AIDS), multiple steps in the ubiquitin–proteasome system are upregulated (Lecker et al., 2006). Similarly, we found that numerous steps in this system were altered by TCDD treatment. Ubiquitin conjugation. The AHR itself has been proposed to act as a ligand- dependent E3 ubiquitin-protein ligase (Ohtake et al., 2007). Our array experiments indicated that TCDD alters expression of multiple genes that encode components of ubiquitin conjugation. After 3 h exposure to TCDD, transcripts for four E2 ubiquitin- conjugating enzymes were altered: mRNA levels of Ubc2e and Ube2b (required for after- replicative DNA damage repair) were increased in resistant rats whereas levels of Ube2n (involved in after-replicative DNA damage repair) and Ube2g1 were suppressed in sensitive rats. Also in sensitive rats, 3 h TCDD treatment increased transcript levels of Uble1a, Senp2 (a ubiquitin-like protein), and Ubadc1. The human homolog of Ubadc1 is

112 a ubiquitin E3 ligase complex which mediates degradation of p27(Kip1). Noteworthy, fetal thymus cultures of p27Kip1-deficient mice are much less sensitive to TCDD compared to control mice, suggesting that p27Kip1 plays a role in TCDD-mediated thymic atrophy (Kolluri et al., 1999). TCDD induces thymic atrophy in resistant H/W rats as well as in sensitive L-E rats, but it is unknown if TCDD affects thymic expression of Ubadc1 differently between these rat strains. Proteolysis. TCDD treatment altered hepatic levels of multiple transcripts that encode proteolysis functions in a strain-dependent fashion. Out of 9 ProbeSets that encode proteasome subunits, 8 were downregulated in H/W rats only while Psmb9 was significantly upregulated in L-E rats at 3 h after TCDD. The proteasome plays essential roles in regulation of cell growth, metabolism, elimination of misfolded or short-lived proteins, and processing of major histocompatibility complex peptides. Thus, dysregulation of proteasomal subunits at an early time point can have major consequences on downstream pathways and can affect numerous bioprocesses. Regulators of the ubiquitin–proteasome system. Numerous factors that regulate the ubiquitin–proteasome system were affected by TCDD exposure. Insulin or insulin- like growth factor-1 (Igf1) inhibit the ubiquitin–proteasome system. In resistant rats only, transcript levels of Igf1 and a transcript with a strong structural similarity to the rat insulin-like growth factor binding protein acid-labile subunit (Igfals) were significantly decreased after 19 h and 3 h TCDD exposure respectively. In sensitive rats only, expression levels of 2 ProbeSets representing insulin-like growth factor binding protein 3 (Igfbp3) were increased 3 h after TCDD exposure and this increase persisted until 19 h for one ProbeSet. The Igfbp3 protein forms a complex with Igfals and Igf1. Complexed Igfbp3 circulates in plasma, prolonging the half-life of Igf isoforms and altering their interaction with cell surface receptors, perhaps, preventing Igf1 from inhibiting the ubiquitin–proteasome system. Previously, doses of TCDD known to inhibit body weight in female SD rats also inhibited serum Igf1 levels, but measurements were made days (rather than hours) after treatment (Croutch et al., 2005). Igf1, Igfals, and Igfbp transcript levels are decreased in dioxin-sensitive male SD rat livers but not until 7 days after TCDD exposure (Fletcher et al., 2005). Therefore, decreased Igf1 and Igfals in response

113 to TCDD in resistant rats at early TCDD exposure times may serve as a protective mechanism. Conversely, cytokines (i.e. tumor necrosis factor-α (TNF-α)) and the NF-κB transcription factor family activate the system (Lecker et al., 2006). In sensitive rats only, tumor necrosis factor receptor superfamily, member 1a (Tnfrsf1a) levels were increased and member 1b (Tnfrsf1b) levels were decreased only in sensitive rats 3 h after TCDD exposure; Tnfrsf1a transcript levels also were reported to be increased in livers of sensitive SD rats after 6 h TCDD treatment (Fletcher et al., 2005). Tnfrsf1a can activate NF-kappaB (an activator of the ubiquitin–proteasome system), mediate apoptosis (discussed below), and function as a regulator of inflammation (discussed below). In humans, Tnfrsf1b forms a heterocomplex with TNF-α to mediate recruitment of two anti- apoptotic proteins, Birc2 and Birc3, which possess E3 ubiquitin ligase activity. Birc2 is thought to potentiate TNF-induced apoptosis by ubiquitination and degradation of TNF- receptor-associated factor 2, which mediates anti-apoptotic signals. Experiments in Tnfrsf1b-null mice suggest that Tnfrsf1b plays a role in protecting neurons from apoptosis by stimulating antioxidative pathways (Twigger et al., 2007). Tnfrsf1b may also be a candidate gene for metabolic syndrome abnormalities (Glenn et al., 2000).

As a summary of proteolysis, in resistant rats TCDD responses included increased ubiquitin potential, decreased proteasomal subunit expression and no upregulation of the hepatic ubiquitin-proteasomal-dependent proteolysis pathway. In sensitive rats, TCDD responses included both increased and decreased ubiquitin potential, increase of only one proteasomal subunit and activation of the ubiquitin- proteasomal-dependent proteolysis pathway by multiple plausible mechanisms. Of note, transcript levels of an ubiquitin enzyme required for after-replicative DNA damage repair were increased in resistant rats, whereas a different ubiquitin enzyme required for after- replicative DNA damage repair decreased in sensitive rats. Thus, resistant rats may be protected from TCDD-induced wasting by the lack of substantial protein degradation pathway activation along with increased expression of a DNA-damage repair gene.

114 Protein synthesis. As described in the previous section, dioxin exposure affects several components in proteolytic pathways and many of these responses differ between dioxin- sensitive rats and dioxin-resistant rats. An appropriate balance between protein degradation and protein synthesis is vital to all tissues but may be particularly relevant to the wasting syndrome and to hepatotoxicity. Approximately 1 week following a lethal TCDD dose in sensitive rats, the initial liver hypertrophy switches to atrophy (reviewed in (Pohjanvirta and Tuomisto, 1994)). Whether cells hypertrophy or atrophy is influenced by the balance between protein degradation and protein synthesis; hypertrophy occurs when protein synthesis exceeds degradation. We found striking differences between dioxin-sensitive rats and dioxin- resistant rats in the effects of TCDD on expression of genes that encode products involved in protein synthesis. Out of a total of 25 TCDD-responsive transcripts that play a role in protein synthesis, levels increased for 24 in sensitive rats. Conversely, 23 of the 25 TCDD-responsive transcripts related to protein synthesis were decreased in resistant rats. TCDD-responsive transcripts are involved in many areas of protein synthesis. First, the levels of 4 ProbeSets encoding members of the phosphatidylinositol 3 kinase (PI3K) family were altered after 3 h TCDD exposure. Depression of the PI3K/Akt pathway leads to decreased protein synthesis, decreased FoxO transcription factor phosphorylation and subsequent protein degradation (Lecker et al., 2006). Transcript levels of the PI3K regulatory subunit polypeptide 1 (Pik3r1) were decreased by TCDD in resistant rats but levels increased in sensitive rats (as measured by RT-PCR). In addition, levels of the PI3K C2 domain-containing gamma polypeptide (Pik3c2g) increased in resistant rats only. Pik3c2g mediates translocation of proteins into membranes and also may mediate protein-protein interactions. Further, the transcript levels of PI3K regulatory subunit polypeptide 2 (Pik3r2) were decreased in sensitive rats only. Through regulation of protein synthesis, PI3Ks exert an extensive list of subsequent effects including metabolic actions of insulin, activation of second messengers, cell proliferation, oncogenic transformation, cell survival, cell migration, and intracellular protein trafficking. Thus, alteration of PI3K levels can have a large impact on many downstream pathways.

115 Secondly, TCDD