Abstract Mahapatra, Debabrata

ABSTRACT

MAHAPATRA, DEBABRATA. Vitamin D Receptor and Xenobiotic Interactions: Insights into the diversity and complexity of molecular interactions and their outcomes (Under the direction of Seth W. Kullman).

The etiology of a significant number of human diseases including hypertension, cardiovascular disease, diabetes, obesity and cancer are in part associated with exposures to environmental contaminants. Recent trends in toxicological research indicate a growing interest in chemicals capable of disrupting the endocrine system. These chemicals comprise a range of natural and synthetic molecules that interact with nuclear hormone receptors altering signaling pathways regulating adverse effects in multiple organ systems, tissue and cell types. While the interaction of endocrine disrupting xenobiotics with nuclear receptors such as the Estrogen receptor (ER), Thyroid receptor (TR), Glucocorticoid receptor (GR) and

Androgen receptor (AR) has been studied in detail, similar research involving xenobiotic interactions with the Vitamin D receptor (VDR) has remained underexplored. This dissertation examines the role of vitamin d receptor as a potential target of xenobiotic induced endocrine disruption and provides new insights into the possible mechanisms underlying the complex molecular interactions between VDR and select VDR agonists and antagonists.

Chapter 1 tests the usefulness and importance of orthogonal assays for validation and confirmation of activity profiles of chemicals generated by the qHTS (Quantitative

Highthroughput Testing) format. A subset of compounds active against the vitamin d receptor was selected from the Tox 21qHTS data set and examined for reproducibility and concordance in orthogonal assay using a luciferase reporter construct. Initial transient transactivation assays were utilized to identify and select VDR agonists, antagonists and synergist. Next we demonstrated the complexity and variability of VDR: xenobiotic interactions at the molecular level through assessments of ligand induced VDR: coregulator interaction including obligate heterodimerization of VDR and RXR (Retinoic-X-Receptor) and recruitment of the VDR coactivator SRC-1 (Steroid Receptor Coactivator-1) in the presence of select list of xenobiotics. Our results indicate that similar to other endocrine receptors such as ER and GR, VDR is susceptible to modulation by a variety of xenobiotics including those that are known endocrine disruptors.

Chapter 2 examines the novel role of trialkyltins including tributyltin (TBT) and triphenyltin (TPT) in modulating the strictly non-permissive interaction between VDR and its obligate heterodimer RXRa in a manner that is conditionally permissive. This alteration in permissivity is observed when cells co-expressing RXR:VDR constructs are exposed to a narrow low dose range of tin compounds resulting in synergistic VDR transactivation only in the presence of the active ligand 1, 25 dihydroxyvitamin D3. Additionally, this synergistic activity is diminished progressively at higher non-toxic doses thus demonstrating a unique dual activity profile. As with other non-permissive heterodimeric interactions such as

RXR:RAR and RXR:TR, RXR:VDR interaction has been reported to be altered by strong RXR agonists such as 9-cis-RA. We report for the first time the ability of organic environmental pesticides (trialkyltins) to disrupt the functionality of RXR:VDR heterodimeric complex that might have profound physiological implications in humans and lower vertebrates as far as endocrine disruption is concerned. Chapter 3 examines the role of yet another well-known non-essential toxic element, cadmium, in disrupting the transcriptional activity of VDR. Cadmium exposure poses serious health risks to humans and other vertebrates. Unlike other nuclear hormone receptors such as the ER where cadmium is a demonstrated receptor agonist, we demonstrate that cadmium inhibits transactivation of ligand induced VDR in multiple cell types. Furthermore, cadmium inhibits vitamin D3 induced endogenous gene expression of a potent downstream gene (CYP24A1) regulated by VDR. Given, the ability of cadmium to disrupt zinc fingers in the DNA binding domain of nuclear receptors we investigated the role of cadmium to inhibit

VDR binding to DNA as a potential mechanism of receptor inhibition. Results from mutation studies and gel retardation assays indicate that cadmium likely prevents binding of VDR to

DNA. Interestingly however, cadmium also facilitated recruitment of coactivator and coregulator when protein:protein interaction was investigated. Taken together these findings clearly suggest two distinct modes of action of cadmium that likely disrupts VDR transactivation. Overall, we demonstrate that cadmium has the potential to disrupt vitamin

D endocrine system.

by Debabrata Mahapatra

A dissertation or thesis submitted to the Graduate Faculty of North Carolina State University in partial fulfillment of the requirements for the degree of Doctor of Philosophy

Comparative Biomedical Sciences

Raleigh, North Carolina

2017

APPROVED BY:

______Dr. Seth W. Kullman Dr. John Godwin Committee Chair

______Dr. Heather Hunt Dr. Heather Patisaul

DEDICATION

This dissertation is dedicated to members of my family especially my wife Anupama for her

extraordinary moral support throughout the years of my graduate studies.

BIOGRAPHY

Debabrata Mahapatra (Deb) was born in the northern state of Bihar, located in the northern

region of India. After graduation from high school in 1994, Deb relocated to Calcutta and

received his Bachelor of Sciences (BS) in Biology followed by a BS degree and a Masters degree in Veterinary Medicine. His long cherished ambition of coming to the US for higher

education in Veterinary Medicine came to fruition when he was admitted to West Texas

A&M for an additional Masters degree in Biological sciences in 2004. Deb got firsthand

experience in teaching advanced anatomy, physiology & genetics to preMed and preVet students. Following his masters in 2006 he joined Texas A&M Veterinary Medical Diagnostic

Lab as a molecular diagnostician. He received training and gained experience from staff

pathologists that helped him gain acceptance into a residency program in Veterinary

Anatomic Pathology at the University of Florida, CVM in 2007. He went on to work at

Disney’s Animal Programs as a fellow in Zoo &Wildlife Pathology. His time spent at Disney

has been the best so far. He received his board certified by the American College of

Veterinary Pathologist (ACVP) in 2011. In 2013 he joined Kullman lab at NCSU to pursue his

PhD in Comparative Biomedical Sc. with a focus in Toxicology.

iii

ACKNOWLEDGMENTS

I would like to acknowledge everyone who have made contributions directly or indirectly towards this work, and helped to build my career as a scientist and as a professional.

First, I would like to thank my graduate mentor Dr. Seth W. Kullman for his guidance and support. I would also like to acknowledge members of my committee Dr. Heather Patisaul,

Dr. Heather Shive and Dr. John Godwin for being extraordinarily supportive and accommodative.

I would also like to thank members of my lab Atlee Watson and Megan Knuth for their unconditional assistance and support whenever I needed them.

Additionally, I would like to acknowledge our administrative assistant Jane Roe for her assistance and advice and members of Drs. Planchart and Mattingly, Dr. Bonner, Dr. Smart and Dr. Tsuji laboratories for their help and advice when needed. Overall, it has been a worthwhile experience this past four years for being a part of educational and collegiate environment.

TABLE OF CONTENTS

LIST OF TABLES……………………………………………………………………………………………………………….viii

LIST OF FIGURES………………………………………………………………………………………………………………ix

GENERAL INTRODUCTION…………………………………………………………………………………………………1

Tox21 in the 21st Century…………………………………………………………………………………………………..1

Vitamin D and Vitamin D Receptor…………………………………………………………………………………….4

Rationale for research initiatives and approaches……………………………………………………………..7

Novel findings and how these address human and environmental health concerns…………..8

References…………………………………………………………………………………………………………………………11

CHAPTER ONE: Insights into the diversity and complexity of molecular

interactions and their outcomes Vitamin D receptor and xenobiotics: Complexity of molecular interactions and outcomes………………………………….14

Abstract……………………………………………………………………………………………………………………………..15

Introduction……………………………………………………………………………………………………………………….17

Materials and Methods………………………………………………………………………………………………………21

Results……………………………………………………………………………………………………………………………….30

Discussion………………………………………………………………………………………………………………………….40

Conclusion………………………………………………………………………………………………………………………….53

References…………………………………………………………………………………………………………………………54

Figures……………………………………………………………………………………………………………………………….60

CHAPTER TWO: Novel role of trialkyltins to promote conditional permissive transactivation of human VDR………………………………………………………………………90

Abstract…………………………………………………………………………………………………………………………….91

Introduction……………………………………………………………………………………………………………………….92

Materials and Methods……………………………………………………………………………………………………..97

Results……………………………………………………………………………………………………………………………..102

Discussion………………………………………………………………………………………………………………………..106

Conclusion……………………………………………………………………………………………………………………….114

References……………………………………………………………………………………………………………………….115

Figures……………………………………………………………………………………………………………………………..119

CHAPTER THREE: New insights into cadmium induced dysregulation

of vitamin D receptor: Potential for endocrine disruption……………………………………………136

Abstract……………………………………………………………………………………………………………………………137

Introduction……………………………………………………………………………………………………………………..138

Materials and Methods…………………………………………………………………………………………………….142

Results……………………………………………………………………………………………………………………………..148

Discussion………………………………………………………………………………………………………………………..153

Conclusion……………………………………………………………………………………………………………………….161

References……………………………………………………………………………………………………………………….162

Figures……………………………………………………………………………………………………………………………..167

GENERAL CONCLUSION……………………………………………………………………………………………………179

References……………………………………………………………………………………………………………………….186

Figures……………………………………………………………………………………………………………………………..187

vii

LIST OF TABLES

CHAPTER 1

Table 1. Tested compounds with names, Cas#, origin, activity and AC50 values…………………………………………………………………………………………………84

Table 2. Summary of Mammalian 2-hybrid (M2H) data across all compounds………………………………………………………………………………………………………………..88

viii

LIST OF FIGURES

CHAPTER 1

Figure 1. Schematic overview of compound selection criteria and experimental workflow………………………………………………………………………………….60

Figure 2. Dose response curves of select VDR agonists identified by transactivation assay……………………………………………………………………………………61

Figure 3. Dose response curves of select VDR antagonists identified by transactivation assay…………………………………………………………………………………….62

Figure 4. Structure based molecular docking using Glide and the human VDR structure……………………………………………………………………………………………65

Figure 5. Protein:protein interaction between VDR with coregulators, coactivators and corepressors in the presence of select agonists…………………66

Figure 6. Protein:protein interaction between VDR with coregulators, coactivators and corepressors in the presence of select antagonists……………69

Figure 7. Heat map showing the variability in the selective preference of compounds for protein interactions…………………………………………………………….72

Figure 8. Endogenous CYP24A1 induction in HL60 cells……………………………………………………..73

Figure 9. Superimposition of predicted binding of calcipotriol and proflavin hydrochloride…………………………………………………………………………….75

Figure 10. 2D predicted binding of calcipotriol…………………………………………………………………..76

Figure 11. 2D predicted binding of proflavin hydrochloride…………………………………………………………………………………………………………………….77

Figure 12. Cell viability assays in HL60 cells………………………………………………………………………..78

Figure 13. Cell viability assays in HeK293 cells……………………………………………………………………80

Figure 14. Cell viability assays in Cos7 cells…………………………………………………………………………83

CHAPTER 2

Figure 1. Biphasic human VDR response curves in transient transactivation assay…………………………………………………………………………………….119

Figure 2. Biphasic human CYP24A1 gene expression………………………………………………………..120

Figure 3. Synergistic responses with zebrafish VDR and Ergocalciferol…………………………………………………………………………………………………………….122

Figure 4. Inhibition of synergistic response by RXR inhibitor LG101208…………………………………………………………………………………………………………..124

Figure 5. Inhibition of synergistic response by RXR AF2 mutant ……………………………………………………………………………………………………………………..126

Figure 6. Recruitment of coactivators and coregulators by trialkyltins……………………………………………………………………………………………………………………128

Figure 7. Dose response curve for dibutyltin (DBT)…………………………………………………………..131

Figure 8. Dose response curves for 9-cis-RA……………………………………………………………………..131

Figure 9. Dose dependent synergy of trialkyltins with vitamin D3…………………………………….132

Figure 10. Coregulator interactions of VDR with trialyltins……………………………………………….133

CHAPTER 3

Figure 1. Inhibition dose response curves of cadmium salts in transient transactivation………………………………………………………………………………………………167

Figure 2. Rescue of Cd induced inhibition by zinc chloride……………………………………………….168

Figure 3. Inhibition of endogenous CYP24A1 gene expression by cadmium salts……………………………………………………………………………………………………………..169

Figure 4. Coactivator and coregulator recruitment by VDR induced by cadmium salts………………………………………………………………………………………………..172

Figure 5. Comparison of inhibitory actions of Cd on wild type and chimeric VDR…………………………………………………………………………………………………………….175

Figure 6. Electrophoretic mobility shift assays and role of Cd inhibition of VDR-DNA binding…………………………………………………………………………………………176

GENERAL CONCLUSION

Figure 1. Targets sites of xenobiotics influencing NR functions…………………………………………187

Figure 2. RXR heterodimer interactions…………………………………………………………………………..188

INTRODUCTION

Toxicity testing in the 21st century

In 2007 the National Research Council1 (NRC) established a set of recommendations for implementation of quantitative high throughput in-vitro assays for toxicity assessments.

These recommendations marked the inception of a historic shift away from traditional in- vivo rodent based toxicity testing to in-vitro quantitative high throughput screening (qHTS) format thus bringing about a new paradigm of toxicology testing in the 21st century. To meet the NRC’s challenge, core regulatory bodies including Environmental Protection

Agency (EPA), National Toxicology Program (NTP), National Institute of Health (NIH)-Center for Advancing Translational Science (NACTS), and the US Food and Drug Administration

(FDA) initiated the collaborative launch of the “Tox 21” program. Another program with similar objectives entitled “ToxCast” was initiated solely by the EPA2,3,4. Overall the goal of these programs has been to establish quantitative cell‐based high‐throughput approaches to screen tens of thousands of compounds, including both industrial chemicals and drugs, for the potential to disrupt biological pathways resulting in toxicity. Both of these screening programs included select sets of human nuclear receptors (NRs) within their respective complements of cell‐based assays. These assays are comprised of differing cell lines stably transfected with an NR (full‐length or ligand binding domain fused to the yeast Gal 4 DNA binding domain) of interest coupled to a reporter system such as luciferase (luc) or β

‐lactamase (β ‐lactamase) and driven by a common response element such as the yeast Gal

4 response element (UAS) or a canonical NR response element. NR assays are conducted in both agonist and antagonist (e.g., in the presence of agonist) modes across multiple concentrations to establish concentration– response curves for each chemical tested.

Computational algorithms are subsequently applied to establish defined parameters including the concentration eliciting half‐maximal activity (AC50) and maximum efficacy

(EMax). These initiatives have now resulted in the generation of an enormous, publicly available compendium of chemical-biological interactions that has enabled researchers to infer predictive public health decisions5,6.

In line with the tox21 objectives, an initial screening study examining the Tox21 pilot phase library of 3,000 environmental chemicals profiled in qHTS (Quantitative High throughput

Screen) format against a panel of 10 nuclear receptors. This study achieved several fundamental things including 1. Identification of novel endocrine compounds that functioned through direct interactions with select NR’’s and 2. Provided a means to compare biological activities of select ligands with NR sequence homology to identify structure activity relationships7. Subsequently, a handful of targeted studies have been conducted that describe specific ligand:receptor interactions that serve as the foundation for establishing defined bioactivity profiles for each NR. For example, profiling ligands for the xenobiotic sensor receptor CAR (Constitutive Androstane Receptor) identified novel molecules with therapeutic potential8. Similarly, a study by Hsu et al9 (2014) identified compounds active against Farsenoid-X- receptor (FXR), a nuclear receptor responsible for the regulation of bile acids and glucose synthesis and metabolism. Assay outcomes were

normalized to cytotoxicity and subject to SAR analysis compared with other NRs including

ERα, AR, PPARγ, and VDR to identify agonist and antagonist unique to FXR. Through this selective process, several FXR‐active (full and partial agonists/ antagonists) structural classes were identified including anthracyclines, avermectins, dihydropyridines, and pyrethroids. Complex SAR analysis was subsequently conducted and each class was recommended for further evaluation with a larger number of tailor‐designed analogs to identify pharmacophores important for FXR binding. Yet another study involving the humans ER was undertaken to identify several classes of environmentally relevant chemicals that affect the function of the estrogen receptor. This study employed two different cell lines and reporter based systems to evaluate and compare the activities of reference compounds with known ER activity10. As evident, these above-mentioned studies were mainly focused on evaluating the feasibility of qHTS assay and the utility of comparisons across assay formats and thus setting standards for future qHTS studies. Also, while these endocrine-screening studies helped to identify, and prioritize chemicals associated with NR transactivation/repression for ER, CAR and FXR, a number of important receptors targets including the TR, PR, GR and VDR were either not explored or not evaluated in depth. In fact, the initial endocrine screen by Huang7 and colleagues (2011) identified just one chemical target of VDR -lithocholic acid, which is already a established endogenous VDR ligand.

Despite a growing interest in understanding the biological role of vitamin D receptor

(VDR), a critically important nuclear hormone receptor, research-investigating VDR as a

potential target of endocrine disruption largely remains underexplored. Given, the emergence of newly identified environmental compounds with potential endocrine disrupting properties against other hormone receptors; a detailed investigation on unknown xenobiotic compounds with the potential to disrupt VDR is warranted. Hence, a

Tox21 library of qHTS compounds screened in VDR -lactamase reporter agonist and antagonist assays was chosen to prioritize and select compounds for screening in orthogonal in vitro assays. From an expanded list of over 90 VDR agonists and 380 VDR antagonists a final subset of 21 agonists and 19 antagonists were selected based on activity in viability assays, potency and efficacy values and curve fit parameters for study in this thesis (see chapter 2). Eventually screening this select set of compounds for reproducibility, confirmation, and gaining mechanistic insights became the primary focus of this research.

Vitamin D and Vitamin D receptor

Vitamin D is an essential secosteroid hormone well known for its metabolic role in bone and mineral homeostatic functions. Research in recent years has however demonstrated its pivotal regulatory roles in additional biological processes including neurodevelopment, neuroprotection, cell proliferation and differentiation, immune function and inflammation among other functions. Vitamin D can be obtained from skin exposure to sunlight or dietary sources. UV rays from sunlight triggers non-enzymatic conversion of precursor 7-dehydrocholesterol to vitamin D3. As an essential dietary component vitamin D is absorbed through the intestine and transported to the liver where

it is hydroxylated at C25 to produce 25 hydroxy vitamin D3- the major circulating form of vitamin D. This is followed by 1 hydroxylation reaction and conversion to the active form-

1-25 di-hydroxy vitamin D3 (1,25D3) (calcitriol) in the kidneys by the anabolic enzyme

CYP27B1. Bound to plasma proteins, 1-25 di-hydroxy vitamin D3 is circulated in the blood to various tissues in the body where endocrine actions are mediated by vitamin d receptor

(VDR)11,12. Similar to other steroid hormones, 1-25 di-hydroxy vitamin D3 has a shorter half-life and optimum blood levels are maintained by a tight feedback mechanism through the action of catabolic enzyme CYP24A1 that is transcriptionally activated by vitamin D3 mainly in the liver. Moreover, other sites such as the skin and hematopoetic tissues are capable of local production of the active form of vitamin D3 through the actions of CYP27B1 where it assumes autocrine/paracrine functions. These sites however, are distinct from the endocrine actions of 1-25 di-hydroxy vitamin D3 such as the small intestine, kidneys, skeleton and parathyroid13,14. Well-defined endocrine function as a steroid hormone is mediated through its high affinity interactions with the vitamin D receptor (VDR), which belongs to NR subfamily NRII113. Ligand binding consequently triggers heterodimerization of

VDR to its cognate partner RXR and together they conform to recognize vitamin D responsive elements (VDREs). These elements optimally constitute a hexametric direct repeat (DR3) of sequence PGGTCA/PGTTCA spaced by three nucleotides with RXR possessing the 5’ and VDR occupying the 3’ half-site1313,15,16. DNA binding is mediated by alpha helices present on the carboxy terminals of the two zinc fingers motifs present on the

N-terminal DNA binding domain. Helix 1 or the P box recognizes specific VDRE’s while the

helix 2 discriminates spacing between half elements and mediates heterodimerization.

Ligand binding to VDR induces conformational changes in the receptors to recruit and create surfaces for coactivator binding. This involves repositioning of the helix 12 of AF2 domain located at the C terminal of VDR to a closed or active position where the liganded receptor stabilizes the RXR:VDR heterodimer with the VDRE and induces a 1800 rotation of

RXR LBD resulting in stabilization of the agonist (holo) conformation12,17,18. Simultaneously, together with helices 3, 4, 5 and 12 a hydrophobic cleft is created that serves as a docking surface for coactivator binding which involves the LXXLL domains of the p160/SRC family of coactivators (i.e. SRC-1) to associate with both heterodimeric partners. This platform also allows for recruitment of additional coactivators such as the histone acetytransferases

(HATs) (i.e.p300/CBP) and histone methyltransferases (HMTs) that subsequently promote

RNA polymerase II docking and gene transcription. Antagonist liganded VDR repression is mediated first by binding of the heterodimer to a negative/non-consensus VDRE and then recruitment of a corepressor (i.e.NCoR-1) and the allosteric locking of the VDR LBD into a corepressor docking conformation. This in turn associates with histone deacetylases leading to chromatin packaging and gene repression12,19,20. Hence, the final outcome of whether a gene is transcribed or repressed largely depends on the precise timing and accuracy of a series of molecular events, which further emphasizes the complex nature of VDR transcription machinery.

Unlike other nuclear hormone receptors, VDR has a near ubiquitous expression across cell lineages13. Consequently, it shares expression with other nuclear hormone

receptors in their respective endocrine tissues such as the thyroid receptor (TR) in thyroid gland and estrogen receptors (ER) in ovaries. Therefore, VDR invariably becomes a target of endocrine disruption when these target tissues (i.e thyroid and ovary) are exposed to chemicals specific for their respective endocrine receptors. In fact, recent epidemiologic evidence suggests that BPA and pthalates, both well known for their endocrine disrupting properties on multiple NRs including TR and ER significantly alter circulating levels of 25

(OH) D in adults in the US21,22. Also, owing to its widespread expression, deficiency of vitamin D affects a variety of organs and systems that can result in increased risk of chronic diseases including common cancers, autoimmune, infectious and cardiovascular diseases and neuropsychiatric disorders in addition to causing growth retardation and skeletal deformities11,23,24.

Rationale for research initiatives and approaches

Although qHTS approaches are capable of rapidly screening compounds within a short time span and requires fewer workforce, these methods do have some obvious limitations. For instance, interpretation of true activity of compounds being tested can be complicated by factors such as non-reproducibility, cytotoxicity, assay sensitivity and specificity, and assay interference from auto fluorescence10,25. Hence, corrective measures to minimize these limitations such as validation with secondary assays have been proposed10,26,27. The primary goal of this dissertation is to validate the qHTS format data set generated against VDR for reproducibility and concordance in a low throughput in-vitro

format using luciferase reporter gene assays. I investigated how structurally and biologically diverse set of chemicals interact with VDR and its heterodimeric partner RXR, facilitate/impede recruitment of coactivators and or corepressors to modulate transactivation. This study is detailed in Chapter 1.

In the course of orthogonal screening process, I identified two groups of potent VDR inhibitors that were found to dysregulate VDR transactivation through unique mechanisms.

These initial rather serendipitous observations prompted detailed investigations into their modes of action. In chapter 2, I discuss how trialkyltins dysregulate transactivation of VDR that are mediated by their unique affinities with RXR. Trialkyltins alter a strictly non- permissive VDR-ligand-heterodimer interaction to a conditionally permissive one within a narrow range of dosage.

In chapter 3, I attempt to shed light on a possible mechanism by which cadmium compounds interact with specific VDR domain and disrupt DNA binding eventually impeding

VDR transactivational activities.

Novel findings and how these findings address human and environmental health concerns

The research presented in this thesis has been focused on a select set of chemicals that have the potential to affect VDR functional activities in-vitro. Core research findings will add to the overall understanding of the underlying mechanisms driving VDR interactions with environmental chemicals at the molecular level.

Below are some salient points that reflect overarching outcomes of this research:

1) Confirmation of qHTS data with orthogonal assays is worthwhile with high

reproducibility and concordance between the assay formats. It helps to minimize the

rates of false negative and false positive thus improving public assurance about the

testing procedures.

2) Cases of non-concordance between the assay results (i.e. Falnidamol hydrochloride

and Fluorescein sodium) relate to false positives obtained in qHTS assays. This

highlights the shortcoming of the qHTS assay.

3) Vitamin D receptor is capable of interacting with a variety of xenobiotics that are

structurally and biologically different including agonists and antagonists.

4) In-vitro transient transactivation and mammalian 2-hybrid data provides a detailed

groundwork for selection of compounds for future testing in in-vivo models.

5) Cheminformatic modeling and simulation studies of VDR ligand interactions reveal

calcipotriol and proflavin hydrochloride as the best-fit agonist and antagonist

respectively. Both share multiple residues within VDR ligand binding pocket with

excellent superimposition despite having opposing roles in VDR transactivation.

Also, novel residues suitable for anchoring and mutagenesis studies in calcipotriol

and the existence of a putative secondary binding site for agonists on VDR are

identified.

6) Activity status of chemicals obtained from dose response curves does not always

reflect activity with respect to recruitment of coregulators, coactivators and or

corepressors. For instance, agonist 9-aminoacridine hydrochloride enhances

recruitment of corepressor NCoR-1 while antagonists such as cadmium salts

recruited coactivator SRC-1 both in the presence of vitamin D3.

7) Similar to other nuclear receptors VDR is clearly a target of endocrine disrupting

compounds especially those containing metals such as tin, cadmium, and arsenic.

Besides, pesticides (Ziram) and therapeutics (Carfizomib) are also capable of

disrupting the function of VDR.

8) Evidence suggests that low dose exposure of VDR over expressed cells in-vitro to

trialkyltins can result in enhanced VDR activity that is several folds higher than that

of vitamin D3 alone. This can have profound physiological implications in humans

and other vertebrates.

REFERENCES

1. NRC. Toxicity Testing in the 21st Century: A Vision and a Strategy. (The National Academies Press, Washington, DC, 2007).

2. Dix, D.J. et al. The ToxCast program for prioritizing toxicity testing of environmental chemicals. Toxicol Sci 95, 5–12 (2007).

3. Collins, F. S., Gray, G. M. & Bucher, J. R. Toxicology. Transforming environmental health protection. Science 319, 906–907 (2008).

4. Kavlock, R,. J., Austin, C., P., &Tice, R.,R. Toxicity testing in the 21st century: implications for human health risk assessment. Risk Anal 29:4 485–497 (2009).

5. Shukla, J.S., Huang, R., Austin, C.P., & Xia, M. The future of Toxicity testing: A focus on in vitro methods using a quantitative high throughput-screening platform. Drug Discov Today 15, 997-1007 (2010).

6. Tice, R. R., Austin, C. P., Kavlock, R. J. & Bucher, J. R. Improving the human hazard characterization of chemicals: a Tox21 update. Environ Health Perspect 121, 756–765 (2013).

7. Huang, R. et al. Chemical genomics profiling of environmental chemical modulation of human nuclear receptors. Environ Health Perspect 119, 1142–1148 (2011).

8. Lynch, C. et al. Quantitative High-Throughput Identification of Drugs as Modulators of Human Constitutive Androstane Receptor. Sci Rep 5, 10405; (2015).

9. Hsu, C.W. et al. Quantitative High-Throughput Profiling of Environmental Chemicals and Drugs that Modulate Farnesoid X Receptor. Sci Rep 4, 6437 (2014).

10. Huang, R. et al. Profiling of the Tox21 10K compound library for agonists and antagonists of the estrogen receptor alpha signaling pathway. Sci. Rep. 4, 5664 (2015).

11. Eyles, D.W., Burne, T.H.J., & Mcgrath, J.J. Vitamin D, effects on brain development, adult brain function and the links between low levels of vitamin D and neuropsychiatric disease. Front Neuroendocrinol 34, 47-64 (2013).

12. Haussler, M. R. & Whitfield, G. K. Molecular Mechanisms of Vitamin D Action. Calcif Tissue Int (2013) 3, 77–98 (2013).

13. Reschly, E.J., & Krasowski, M.D. Evolution and function of the NRII Nuclear hormone receptor subfamily (VDR, PXR, and CAR) with respect to metabolism of xenobiotics and endogenous compounds. Curr Drug Metab 7, 349-365 (2006).

14. Schuster, I. Cytochromes P450 are essential players in the vitamin D signaling system. Biochimica et Biophysica Acta 1814, 186–199 (2011).

15. Shaffer, P., L., & Gewirth, D., T. Structural basis of VDR–DNA interactions on direct repeat response elements. EMBO J 21:2242–2252 (2002).

16. Aranda, A. & Pascual, A. Nuclear Hormone Receptors and Gene Expression. Physiol Rev 81, 1269–1304 (2001).

17. Towers, T., L., Luisi, B., F., Asianov, A. et al. DNA target selectivity by the vitamin D3 receptor: mechanism of dimer binding to an asymmetric repeat element. Proc Natl Acad Sci USA 90: 6310–6314 (1993).

18. Nakajima, S., Hsieh, J-C., MacDonald, P.,N. et al. The C-terminal region of the vitamin D receptor is essential to form a complex with a receptor auxiliary factor required for high affinity binding to the vitamin D-responsive element. Mol Endocrinol 8:159–172 (1994).

19. Jurutka, P.W. et al. Molecular and functional comparison of 1,25-dihydroxyvitamin D(3) and the novel vitamin D receptor ligand, lithocholic acid, in activating transcription of cytochrome P450 3A4. J Cell Biochem 94, 917–43 (2005).

20. Rochel, N., Wurtz, J. M., Mitschler, A., Klaholz, B., & Moras, D. The crystal structure of the nuclear receptor for vitamin D bound to its natural ligand. Mol Cell 5, 173–179 (2000).

21. Kandarakis ED, Bourguignon JP, Giudice LC et al. Endocrine-Disrupting chemicals: An endocrine society scientific statement. Endocrine review 2009, 30, 293-342 (2009).

22. Johns, L. E., Ferguson, K. K., and Meeker, J. D. Relationship between urinary phthalate metabolite and bisphenol A concentrations and vitamin D levels in US adults: National health and nutrition examination survey (NHANES), 2005-2010. J Clin Endocrinol Metab. 101, 4062-4069 (2016).

23. Hollick, M.F. Vitamin D deficiency. NEJM 357, 266-281 (2007).

24. Bouillon, R. et al. Vitamin D and human health: Lessons from vitamin D receptor null mice. Endocrine Reviews (2008). doi:10.1210/er.2008-0004

25. Hsieh, J., H., Sedykh, A., Huang, R., Xia, M., & Tice, R. A data analysis pipeline accounting for artifacts in Tox21 Quantitative High throughput screening assays. J Biomol Screen. 20:7 887-897 (2015).

26. Xia, M. et al. Identification of chemical compounds that induce HIF-1?? activity. Toxicol. Sci. (2009). doi:10.1093/toxsci/kfp123

27. Judson, R. et al. Perspectives on Validation of High-Throughput Assays Supporting 21st Century Toxicity testing. ALTEX 30, 51-56 (2013).

CHAPTER ONE

Confirmation of High-Throughput screening data and novel mechanistic insights into VDR-

xenobiotic interactions by orthogonal assays

Debabrata Mahapatra1, Jill A. Franzosa5, Kyle Roell2, Melaine Agnes Kuenemann2, Keith A.

Houck5, David M. Reif2, Denis Fourches2, and Seth W. Kullman3,4*

1 Comparative Biomedical Sciences, College of Veterinary Medicine, North Carolina State

University, Raleigh, North Carolina, United States of America; 2 Department of Chemistry,

Bioinformatics Research Center, North Carolina State University, Raleigh, North Carolina,

United States of America; 3 Department of Biological Sciences, North Carolina State

University, Raleigh, North Carolina, United States of America, 4 Program in Environmental and Molecular Toxicology, North Carolina State University, Raleigh, North Carolina, United

States of America, 5National Center for Computational Toxicology, Office of Research and

Development, U.S. Environmental Protection Agency, RTP, NC.

ABSTRACT

Quantitative high-throughput (qHTS) testing formats have gained acceptance and widespread recognition in the field of toxicity testing of environmental compounds in recent years. As a result, significant progress has been made with respect to identification of compounds that modulate the function of nuclear receptors in vitro. Recent efforts through Tox21 and ToxCast programs have demonstrated that the Vitamin D receptor (VDR) is activated and/or antagonized by a wide range of structurally diverse chemicals including naturally occurring, synthetic, and environmental chemicals. Vitamin D has gained significant attention in recent years not only for its role in classical bone and mineral homeostatic functions but also for its roles in neurodevelopment, neuroprotection, cell proliferation and differentiation, immune function and inflammation. In this study, we examined the Tox21 qHTS data set generated against VDR for reproducibility and concordance and elucidated functional insights into VDR-xenobiotic interactions. Twenty- one potential VDR agonists and 19 VDR antagonists were identified from a subset of >400 compounds with putative VDR activity generated by Tox21 qHTS library and followed up with additional orthogonal assays. Transient transactivation assay (TT) using a human VDR plasmid and Cyp241A1 luciferase reporter construct revealed 20/21 active VDR agonists and

18/19 active VDR antagonists. Compounds were then examined using a mammalian-2- hybrid assay (M2H) to evaluate VDR interactions with co-activators and co-regulators. With the exception of a select few compounds, VDR agonists exhibited minimal to moderate

recruitment of co-regulators and co-activators whereas antagonists exhibited moderate to marked attenuation of co-activator recruitment by VDR both in the presence and absence of co-regulators. A unique set of compounds exhibiting synergistic activity in antagonist mode and no activity in agonist mode was identified. Additionally, cheminformatics modeling of VDR-ligand interactions were conducted. Overall, the data emphasizes the molecular complexity of VDR transcription machinery in terms of differential and preferential affinities of compounds for co-regulators/co-activators and respective sites of action.

INTRODUCTION

Following National Research Council’s recommendations1 for a shift from traditional low throughput in vivo rodent assays to less expensive in vitro high throughput methods, core regulatory bodies such as the U.S. Environmental Protection Agency (EPA), National

Toxicology Program (NTP), National Institute of Health (NIH), NIH Chemical Genomics

Center (NCGC), US Food and Drug Administration (FDA) responded to the urgency with the initiation of ToxCastTM and Tox21 programs2, 3. These programs were aimed at prioritizing toxicity evaluations through promoting the increasing use of in vitro high throughput screening assays for large numbers of chemicals already in commercial use for which little or no toxicity data was available4, 5. These initiatives have now resulted in the generation of an enormous, publicly available compendium of chemical-biological interactions that has enabled researchers to infer predictive public health decisions.

Within both the Tox CastTM and Tox 21 programs, disruption in nuclear receptor (NR) signaling represents a defined set of molecular targets of interest. Given the role of NR’s in modulating specific endocrine functions, assessing chemical interactions with this superfamily of proteins provides mechanistic data that enables predictive assessments of toxicity pathways related to human disease. Subsequently, targeted cell based in vitro studies have been conducted to identify the selectivity and efficacy of environmentally relevant chemicals that can modify receptor function. For instance, assessments of estrogen receptor alpha (ER) agonists/antagonists demonstrated the feasibility of quantitative high

throughput assays to identify environmental chemicals with the potential to interact with

ERα and revealed the importance of both known and novel ERα active structure classes as agonists/antagonists6. Similarly, structure-activity relationships of FXR-active compounds suggest that this receptor may have multiple modes of action that modulate receptor- coregulator interactions essential to NR transactivation7. Recent studies have also utilized computational modeling based approaches to build predictive models based on structural information and activity data8. Consistent within these approaches is the observation that receptor-ligand molecular interactions are mediated through specific structural determinants that modulate receptor conformation and thus transactivational capacity.

In the wake of the above-mentioned targeted NR studies, and the emergence of newly identified environmental compounds with potential endocrine disrupting properties, we focused our attention to the library of screened compounds that altered the transactivational activity of VDR. Vitamin D has gained much attention in recent years not only for its role in classical bone and mineral homeostatic functions but also for its roles in neurodevelopment, neuroprotection, cell proliferation and differentiation, immune function and inflammation. Vitamin D is unique in that in its native state it is a vitamin or an essential dietary component. However, upon metabolic activation it is converted to 1-25 di-hydroxyvitamin D3 (1,25D3, calcitriol) and serves a well-defined endocrine function as a

9 steroid hormone . Classical transcriptional actions of 1,25D3 are mediated through its high affinity interactions with VDR. Vitamin D receptor is a member of the nuclear receptor superfamily, which is comprised of a large group of ligand-activated transcription factors.

The mechanism of VDR-mediated gene transcription closely resembles that of other steroid hormones usually involving high affinity interactions between ligand and receptor, heterodimerization with RXR, association with a canonical vitamin D response element

(VDRE) within target promoter regions and recruitment of co-regulatory proteins, members of the MED complex and RNA polymerase II to initiate both transactivation and transrepression of gene regulatory networks critical to cellular processes10. Similar to other steroid hormones, 1,25D3 has a short half-life and optimum blood levels are maintained by a tight feedback mechanism through the action of catabolic enzyme CYP24A1. 1,25D3 also serves paracrine/autocrine functions since several target tissues11 are capable of synthesizing the active form of the hormone12. Accordingly, deficiency of vitamin D affects a variety of organs and systems resulting in growth retardation and skeletal deformities, and increased risk of chronic diseases including common cancers, autoimmune, infectious and cardiovascular diseases and neuropsychiatric disorders13, 9.

In this study, we examined the Tox21 qHTS data set generated against VDR (see materials and methods) for reproducibility and concordance in a low throughput format and investigated VDR receptor activity profiles in vitro using luciferase reporter gene assays. We examined how structurally and functionally diverse compounds included in the Tox21 chemical space modify core nuclear receptor functions of VDR with respect to VDR heterodimerization with human RXR, recruitment of coactivator (SRC-1) or corepressor

(NCoR), and the ability to initiate/inhibit transactivation of CYP24. Molecular modeling was

also employed to forecast and study the molecular interactions of the most potent compounds once docked in the VDR binding site.

MATERIAL AND METHODS

Tox21 chemical library

The Tox21 10K compound library

(https://cfpub.epa.gov/si/si_public_record_report.cfm?dirEntryId=246691) was compiled by the Environmental Protection Agency (EPA), the National Toxicology Program (NTP), and the NIH Chemical Genomics Center/ National Center for Advancing Translational Sciences27.

It consists of approximately 8,300 unique samples including drugs, food additives, environmental chemicals, consumer product ingredients and industrial chemicals.

Compound stock preparation was previously described62. Briefly, stock solutions were prepared in DMSO, most at 20 mM, followed by a 3-fold serial dilution series in DMSO resulting in 15 concentrations per compound for testing as described below. Analytical QC information for the library is available at: https://tripod.nih.gov/tox21/samples.

qHTS of VDR Beta-lactamase reporter gene and cell viability assays

The assay description and methods are available from the Pubchem Open

Chemistry Database, bioassay record AID 743241

(https://pubchem.ncbi.nlm.nih.gov/bioassay/743241#section=Top). Briefly, the cell line and the cell culture reagents for the qHTS screening were from Life Technologies (Carlsbad, CA,

USA). The GeneBLAzer® VDR UAS-bla HEK 293T cells stably expressing a VDR-driven beta- lactamase reporter gene under the control of the upstream activator sequence (UAS) was

used. The VDR consisted of the human VDR ligand-binding domain fused to the DNA- binding domain from the yeast GAL4 transcription factor that binds the UAS. Cells were cultured in Minimum Essential Medium (MEM) containing 10% heat-inactivated Fetal

Bovine Serum (FBS), 1 mM sodium pyruvate, 0.1 mM NEAA, 100 U/ml penicillin, and

100mg/ ml streptomycin. Cells were grown humidified incubator at 37°C with 5% CO2 and passaged when ~70- 80% confluent. Prior to screening, GeneBLAzer® VDR UAS-bla HEK 293T

Cells were seeded at 2000 cells/5 µL in black-clear bottom 1536-well plates (Greiner Bio-

One North America, Monroe, NC - CONFIRM) using an 8-tip dispenser (Multidrop/Thermo

Fisher Scientific, Waltham) and incubated for 5 hours (h) at 37º C and 5% CO2. After incubation, the cells received 23 nl of test compounds in DMSO dispensed using a pin tool station (Kalypsys, San Diego, CA) resulting in final concentrations ranging from 109 – 104 M

(15 concentrations). In antagonist mode, 1o, 25-Dihydroxy Vitamin D3 (3 nM, final concentration) was added immediately after test compound addition. For the agonist mode screen, each assay plate contained 38 wells for a dose-titration from 0.1 mM (1:3 Dilutions) of 1α, 25-Dihydroxy Vitamin D3, along with 16 wells of 50 nM and 16 wells of 15 nM 1α, 25-

Dihydroxy Vitamin D3 as a positive control. For antagonist-mode screen, each assay plate contained 48 wells exposed to DMSO and 3 nM 1α, 25-Dihydroxy Vitamin D3 and 16 wells of 92 µM teraoctyl ammonium bromide and 3 nM of 1α, 25-Dihydroxy Vitamin D3. DMSO was used as a vehicle control for compounds. After 16 h (37º C, 5% CO2) of exposure, 1 uL

CCF4 dye was added to each will with a single tip dispenser. Following 2 h of incubation, fluorescence was measured using 405 nM excitation and fluorescent emissions read-outs at

530 nM (channel 1) and 460 nM (channel 2) using an EnVision plate reader (Perkin Elmer,

Shelton, CT, USA). The cytotoxicity effects were measured by the addition of 4 µL of

CellTiter-Glo (Promega, Madison, WI, USA) reagent, followed by an additional incubation for

30 min at room temperature. Luminescence was measured on a ViewLux plate reader using an exposure time of 15 s.

Analysis of qHTS data

The qHTS data, processed using the tcpl data pipeline (Filer et al., 2016), were downloaded from the EPA website (https://www.epa.gov/chemical-research/toxicity- forecaster-toxcasttm-data). Processing of the data in the tcpl pipeline consisted of the following methods. The ratio of the raw plate fluorescence values (channel 2/channel 1) was determined for each well (rval). Ratios were normalized relative to the positive control compounds (agonist mode: 1α, 25-Dihydroxy Vitamin D3, 3 nM, 100%) and DMSO-only wells (0%) by the formula Activity = [(rval-bval)/(pval-bval)] x 100 where rval is the ratio measurement for the treated well, pval median value of the positive control wells, and bval is the median value of the DMSO-only wells. The concentration-response data were then fitted with three separate models using the tcpl methods14. Briefly, these models were a constant model, a Hill model and as gain loss model (the product of two Hill functions with a shared top). Each model fit was compared using an Akaike Information Criterion (AIC) value and the model with the lowest AIC was selected as the winning model. The AC50 and maximum efficacy Emax were determined from the winning model.

Compound acquisition for orthogonal assay screening

Prioritized compounds were either procured under EPA contract EP-D-12-034 from

EvoTec (South San Francisco, CA) or purchased (Sigma-Aldrich Corp., St. Louis, MO, USA). All compounds were serially diluted in DMSO with a final testing concentration ranging from

0.01 to 120 M.

Cheminformatics

Molecular docking was conducted using PDB code = 1S19 X-ray crystal structure preprocessed and curated using the Schrodinger Suite and the Protein Preparation Wizard21 module and the OPLS3 force field. All the missing side chains were generated using Prime63 and protein minimization was performed. The molecular docking procedure was performed using Glide software22-23 with XP scoring functions with a rigid protein and flexible ligand.

The coordinates of the best calcipotriol and proflavin hydrochloride docking pose were subjected to molecular dynamics simulation using DESMOND. Counter-ions were used to neutralize each complex. The whole system was immersed in a periodic TIP3P water orthorhombic box. The molecular dynamics production run was performed for 20 ns. Each recording interval was 1.0 ps for the trajectory and 1.0 for the energy. The NPT ensemble class with a temperature of 300K and a pressure of 1.01325 bar was used.

Heat map

Compound activity patterns for all possible combinations of protein-protein interactions between VDR and its coregulator, coactivator, and corepressor and how individual compounds (agonists and antagonists) affect this interaction. Clustering was performed on the mammalian two hybrid data using the hclust function in R with a

Manhattan distance metric and complete linkage64-66.

Plasmid DNA constructs

The pSG5-Human VDR construct was a gift from Dr. John Moore (GlaxoS- mithKline,

Research Triangle Park, NC). All human coregulator transient transactivation and mammalian 2-hybrid constructs were a gift from Dr. Donald McDonnell (Duke University,

Durham, NC). The CYP24 luciferase reporter, 5XGal4-TATA- Luc mammalian 2-hybrid luciferase reporters, and the pRL-CMV (Renilla luciferase) internal luciferase control were obtained as described previously67,53.

Cell culture

Cell culture media and other necessary reagents were obtained from Life technologies (Carlsbad, CA). Hek293T cells were grown in Minimum Essential Medium

(MEM) containing 10% heat-inactivated Fetal Bovine Serum (FBS), 1 mM sodium pyruvate,

0.1 mM NEAA, 100 U/ml penicillin, and 100mg/ ml streptomycin. Cos 7 cells were cultured in Dubelco’s Modified Eagle Medium (DMEM) containing 10% FBS. Human promyelocytic

leukemic HL60 cells were grown in RPMI suspension media containing 15% FBS and 200nM

L-glutamine. Cells were maintained in a humidified incubator at 37°C with 5% CO2 and passaged when ~70- 80% confluent.

Transient transactivation assay

Full-length VDR constructs were tested in transient transactivation assays with 1α,

25-dihydroxyvitamin D3 (1, 25D3) (EMD Millipore, Billerica, MA) as the positive control.

Experiments were conducted using pSG5hVDR, pRLCMV, and CYP24-Luc expression vectors as previously described67. Hek293T cells were seeded in 96-well plates at 2.5 x 104 cells per well 24 hours prior to transfection. Cells were transfected at 90-95% confluency using

Lipofectamine 2000 (Life Technologies, Grand Island, NY) with DNA diluted in Opti-MEM I

Reduced Serum Medium as per the manufacture’s recommendations. For functional comparisons, 89.7 ng of each pSG5-VDR construct was transiently transfected into Hek293T cells with 19.2 ng CYP24-Luc and 4.5 ng of Renilla luciferase, which serves as an internal luciferase control (Promega Corporation, Madison, WI). Twenty-four hours post-exposure the Dual-Glo Luciferase Assay System (Promega Corporation, Madison, WI) was used to passively lyse the cells and test for luciferase activity following the manufacturer’s protocols. Luciferase activities were measured using a Wallac MicroBet TriLuc Luminometer

(Perkin Elmer Life Sciences, Waltham, MA). Control reactions included empty pSG5 vector and ethanol as a vehicle control. Luciferase readings were normalized to the internal Renilla

control, and VDR response was normalized to an empty vector control. Transient transfection data were used with the ToxCast Analysis Pipeline (TCPL) to generate dose-

14,64 response curves and estimate AC50 values . Curve fitting was performed using Hill, Gain

Loss (GNLS), and constant fit models. AC50 values were generated for antagonists or agonists, respectively, and were chosen from either the Hill or GNLS model using the Akaike

Information Criterion.

Mammalian 2-hybrid assays

Protein-protein interactions between VDR and its heterodimer partner RXR and members of SRC/p160 family of nuclear receptor coactivators and co-repressor (NCoR1) were assessed using a mammalian 2-hybrid system (Clontech, Mountain View, CA). Assays were conducted with chimeric VDRs containing the herpes simplex VP16 activation domain fused to full length human VDR as prey (pVP16-hVDR). NR co-regulators were used as bait for each reaction, and consisted of fusion proteins containing; a complete NR Box of the

SRC-1 (pM-SRC1 aa 241-386), NCOR (pM NCoR-1), or full-length hRXR’ fused to the yeast

Gal4 DNA-binding domain. Assays were conducted in Cos7 cells seeded into 96 well plates twenty-four hours pre-transfection as described above. Cells were transfected with 33.6 ng pVP16-VDR, 33.6 ng pM-coregulator, 126.6 ng 5XGal4-TATA-Luc reporter (containing response elements for the yeast Gal4 DNA binding domain), and 3 ng Renilla using

Lipofectamine 2000 as described above. Controls consisted of transfections containing empty pM, pVP16 or both empty pM and pVP16 vectors. For both assays, experiments were

replicated three times in groups of 3 technical replicate wells. One-way ANOVAs followed by Tukey’s HSD post hoc tests, sigmoidal dose-response calculation with variable slopes followed by nonlinear regression analysis were run in GraphPad Prism version 6 (GraphPad

Software, La Jolla, CA). Note that all assays were conducted in either the presence or absence of co-transfected full length RXR or SRC-1 to assess if exogenous protein expression would further facilitate VDR co-regulator/co-activator interactions.

Cell viability assay

Hek293T, Cos7, and HL-60 cells were seeded in 96 well plates in triplicate at a density of 25,000 cells/well, transfected and dosed with select concentrations of the test

0 compounds and incubated for 18 hours at 37 C/5% CO2. Triton X (0.1%), DMSO (0.1%), and untreated wells served as controls. Resazurin at 1X concentration (20ul) /well was added to the cells and incubated at 370C for 2 hours in the dark. The amount of resorufin produced proportional to the number of viable cells was quantified by using a microplate reader

(described earlier) equipped with a 560 nm excitation / 590 nm emission filter set. Only concentrations that yielded more than 90% viable cells were selected for in-vitro assays.

Real-time PCR

Total RNA was isolated from treated HL60 cells using the Zymo RNA Isolation kit

(Zymo Research Corp, CA USA) and reverse transcribed using a High Capacity cDNA Archive

Kit (Applied Biosystems, Foster City, CA) following the manufacturers’ instructions. CYP24A1

mRNA expression was normalized against that of housekeeping gene GAPDH. Real-time PCR assays were performed in 96-well optical plates on an ABI Prism 7300 Sequence Detection

System with SYBR Green PCR Master Mix. Primers used for GAPDH mRNA expression were designed as follows: [GAPDH-F:5’- CGACCACTTTGTCAAGCTCA- 3’ GAPDH-R: 5’-

GAGGGTCTCTCTCTTCCTCT-3’], while those for CYP24A1 [CYP24A1-F:5’-

TGAACGTTGGCTTCAGGAGAA-3’,CYP24A1-R:5’-AGGGTGCCTGAGTGTAGCATCT-3’] were adopted from Yosuke68 et al 2009. Fold gene induction following treatments were calculated based on the equation: Fold change=2ΔΔCt, where ΔCt represents the differences in cycle threshold numbers between CYP24A1 and GAPDH, and ΔΔCt represents the relative change in these differences between control and treatment groups69. Values were plotted as a percentage and compared to the percentage induction of vitamin D360.

RESULTS

Selection of putative VDR agonists and antagonists for orthogonal screening

Experimental qHTS screening results of Tox21 library of compounds in VDR - lactamase reporter agonist and antagonist assays were used to prioritize and select compounds for screening in our orthogonal in vitro assays. Results from the curve-fitting analysis suggest that the human VDR is activated and/or antagonized by a wide range of structurally diverse chemicals including endogenous, synthetic, and environmental chemicals. In agonist-mode, over 90 compounds activated the VDR reporter gene assay.

Over 380 potential VDR antagonists were also identified with (AC50) values ranging from sub-micromolar to 50 µM. To select the subset of compounds for further in vitro testing, the AC50 values from the VDR beta-lactamase reporter assay and the cell viability assay were used to calculate a ratio (AC50viability/ AC50VDR). The list of active agonists and antagonists was reduced to compounds with ratio values less than 5 and assessed in relation to curve fit parameters and flag criteria which are available via the ToxCast

Dashboard (https://actor.epa.gov/dashboard/). In total, 21 agonists and 19 antagonists were selected for further screening in orthogonal VDR screening assays (Fig. 1).

Ligand-induced receptor transactivation

Selected 21 agonists and 19 antagonists were screened for functional nuclear receptor activities using human VDR expression constructs. Our approach to validate target

receptor interaction incorporated fundamental components of nuclear receptor function including VDR transactivation and co-regulator recruitment. Each of these critical components of NR function were essential in demonstrating that receptor “agonists” or receptor “antagonists” facilitate molecular interactions necessary for target gene induction or gene repression in vivo. Transient transactivation assays in HEK293 cells were utilized to assess selected compounds in either agonist or antagonist modes. AC50 values for VDR agonists were determined relative to 1,25D3 as a positive control. From a total of 21 selected Tox21 agonists, 20 were confirmed to exhibit VDR transactivation activities (Fig. 2,

Table 1). Overall, compounds exhibited a wide spectrum of activity as evident by their AC50 values that ranged from 0.009M for calcipotriol up to 37.41M for novaluron. One compound, falnidamol hydrochloride consistently failed to exhibit any dose response interactions resulting in an ambiguous and inconsistent AC50 value after several repeated trials. This compound however was found to be active in the Tox21 qHTS data set.

The activity of VDR antagonists was assessed through quantifying the inhibition of

1,25D3-induced responses in transient transactivation assays. Eighteen out of a total of 19 compounds were found to be functionally active with the exception of fluorescein sodium.

Activity of compounds ranged widely with AC50 values ranging between 0.01M for phenylarsine oxide up to 7.32M for aristolochic acid. Compounds that contained a transition metal atom (i.e., Cd, Tin, Au, Ar) exhibited both greater potency and efficacy on the VDR activity than those without metals (Fig. 3a). The trialkyltins, ziram (zinc containing pesticide) and other cadmium salts followed a similar trend. Non-metal containing

compounds (i.e., thiram, aristolochic acid and proscillaridin) were relatively less potent and efficacious antagonists (Fig. 3b). A unique group of three non-metal containing compounds, namely dichlone, carfizomib and menadiol were identified that exhibited very weak agonist activity when tested alone but exhibited moderate to marked synergistic activity when tested in antagonist mode in the presence of 1,25D3 (Fig. 3c). These compounds were categorized as antagonists in the Tox21 qHTS database.

Cheminformatics modeling of VDR-ligand interactions

3D molecular docking studies were conducted for each active compound identified in our screening assays, so that we could evaluate and better understand their respective binding modes in the VDR active site. Since the first co-crystalized structure of the VDR

15 receptor with 1,25D3 was reported in 2000 by Rochel , multiple X-ray structures of the VDR receptor in complex with different small molecule ligands16-19 have been published and deposited in the online Protein Data Bank. The thorough analysis of the different holo crystal structures did not reveal any significant conformational and/or structural changes20.

As a result, we decided to select the X-ray structure for the human VDR co-crystalized with calcipotriol19 (PDB code: 1S19).

After the structural preparation and cleaning of the VDR structure using the

Schrodinger suite21 (see Methods), we used Glide22 to dock all our selected compounds in the VDR active site. The XP scoring23 function was utilized to score the intermolecular interactions between VDR and each compound docked in its site. Compounds’ docking

scores (expressed in kcal/mol) and eModel scores (also in kcal/mol) are reported in Fig. 4a.

The compound affording the best docking scores (the lower the better) is indeed calcipotriol: in the procedure of self-docking (i.e., removal of the native ligand from the crystal followed by its blind re-docking into the active site), the binding conformation of calcipotriol was characterized by an excellent docking score as low as -13.3 kcal/mol and eModel equal to -77.54 kcal/mol (these score levels are typical for nanomolar binders).

Moreover, the best predicted docking pose of calcipotriol was found to be reasonably close to the native ligand with RMSD = 2.04Å. This result validated the reliability of the modeling calculations and increased our confidence in the docking results for the other selected compounds. Overall, the molecular docking procedure was able to retrieve as active

(docking score < -7 kcal/mol) most of the agonists present in our dataset (encompassing full and partial agonists). However, the docking procedure was not able to discriminate the experimentally confirmed antagonists as active (best docking score obtained for provaflin hydrochloride equal to -6.71 kcal/mol). Interestingly, the molecular docking procedure predicted both compounds (best agonist calcipotriol and best antagonist proflavin hydrochloride) to share almost the same binding pocket (Fig. 4b, Fig. 9). The two other non- metal antagonists (aristolochic acid and proscillaridin) were not predicted to fit and bind in the VDR binding pocket we used for our docking experiment.

To refine our results, we performed molecular dynamics simulations (MDS – see

Methods for computational protocols) to study the dynamic interactions of the ligands with

VDR simulated over 20 nanoseconds. We chose to run a simulation with the full agonist

calcipotriol and another with the antagonist proflavin hydrochloride. For both simulations, the best predicted binding pose obtained from the molecular docking calculations was used as the starting binding conformation. Interestingly, results of MDS demonstrated that those binding modes are mainly conserved over the entire simulation, underlining the dynamic stability of these two VDR-ligand complexes. Regarding calcipotriol, the predicted binding mode obtained from molecular docking involved the creation of two hydrogen bonds with the VDR receptor (Fig. 10a): one H-bond between His397 and the terminal hydroxyl group, and another H-bond between Ser237 and the methylidene cyclohexanediol. Interestingly,

MDS calculations were able to reveal other critical interactions that are likely to play a role in stabilizing calcipotriol in VDR binding site (Fig. 10b): one H-bond between His305 and the hydroxyl of the terminal 1-cyclopropylmethanol group, another H-bond between Arg274 and the methylidenecyclohexanediol and two other H-bonds between Tyr143, Ser278 and the other hydroxyl group of the methylidenecyclohexanediol of calcipotriol. The persistence

(or conservation ratio) of each of those H-bonds was computed as well: for instance, the H- bond with Ser237 was detected in 96% of the 20,000 MDS frames (i.e., one frame every picosecond). Meanwhile, H-bonds with Tyr143 and Ser278 only appeared in 35% and 45% of the MDS frames respectively. Importantly, we observed that H-bonds with His397 and

His305 (58% and 54%) were switching from one to the other over the simulation.

Meanwhile, the predicted binding mode of proflavin presented a π-π stacking between

Trp286 and the aromatic acridine group, as well as one H-bond between Ser237 and one primary amine group (Fig. 10). Multiple additional interactions were found through MDS

calculations, including H-bonds with Ser278, Ser275, His397, and His305. A π-π stacking interaction between Tyr295 and the aromatic acridine group was also detected through

MDS.

Protein:protein interaction

While transient transfection assays provided us with a global context of chemical receptor transactivation, further functional analysis of ligand induced receptor:coregulator interactions was conducted to gain mechanistic insights into VDR-chemical partnerships.

We conducted Mammalian Two Hybrid (M2H) assays to examine direct protein:protein interactions between VDR and VDR coregulators. With VDR agonists (Fig. 5, Table 1), select ligands facilitated the following: 1) Interaction between VDR:RXR and VDR:SRC-1 both in the presence and absence of co-transfected full length RXR and SRC-1. This includes compounds such as Calcipotriol, Ergocalciferol, LCA, 9 amino acridine monohydrochloride,

Tamoxifen citrate, Triamterene, and 2,2'-methylenebis(6-tert-butyl-4-methylphenol). 2)

Interaction between VDR:RXR only in the presence of co transfected SRC-1 was observed with 7 methyl benzo a pyrene and Lanoconazole. 3) Interaction between VDR:RXR only in the absence of full length SRC-1 was observed with Alpha-Terthiophene, 2,2'- methylenebis(6-tert-butyl-4-ethylphenol), and 4,4'-butylidenebis(6-tert-butyl-m-cresol. 4)

Interaction between VDR:SRC-1 only in the presence of co-transfected RXR was observed with Methyl 3-amino-5,6-dichloropyrazine-2-carboxylate. 5) Interaction between VDR:SRC-1 only in the absence of RXR was demonstrated exclusively by Novaluron. 6) There was a

grouping of VDR agonists that did not facilitate VDR:RXR or VDR:SRC-1 interactions under any condition including 7-(Dimethylamino)-4-methylcoumarin, Disodium 4,4'-bis(2- sulfostyryl) biphenyl, 4-Aminofolic acid, 2,7 Naphthalene disulfonic acid, Cridanimod and

Benzenesulfonic acid. 7) Lastly, except for 9-aminoacridine monohydrochloride none of the agonists promoted recruitment of NCoR in the presence or absence of RXR.

Comparatively, the potential of antagonists to modify 1,25D3 induced VDR:RXR and

VDR:SRC-1 interactions was assessed both in the presence and absence of co-transfected full length coregulators (Fig. 6, Table 1) and facilitated the following interactions: 1) Select compounds consistently attenuated recruitment VDR:RXR and VDR:SRC-1 in both the presence and absence of co-expressed coregulators including Dibutyltin dichloride,

Triphenyltin hydroxide and the Cadmium reference solution. 2) Antagonists that selectively inhibit VDR:RXR interactions only in the absence of full length wild type SRC-1 include

Phenylarsine oxide and Menadiol. 3). Antagonists that selectively inhibited VDR:SRC-1 interactions only in the presence of full length wild type RXR include: Tributyltin chloride,

Proscillaridin and Potassium dicyanoaurate. 4) Antagonists that selectively inhibited

VDR:SRC-1 interactions only in the absence of full length wild type RXR include: Cadmium reference solution and Phenylarsine oxide. 5) There were no antagonists that selectively attenuated VDR:RXR interactions in the presence of full length SRC-1. Unexpectedly, select

VDR antagonists also appeared to enhance some VDR coregulator interactions including: 1)

Antagonists that enhanced recruitment of VDR:RXR and VDR:SRC-1 in both the presence

and absence of co-expressed coregulators including cadmium dichloride and cadmium dinitrate. 2) Antagonists that enhanced the Interaction between RXR only in the presence of co transfected SRC-1 including Proscillaridin and Cadmium acetate dehydrate. 3)

Antagonists that enhanced the interaction between VDR:SRC-1 only in the presence of co- transfected RXR including Carfizomib. 4) Antagonists that enhanced the interaction between VDR:SRC-1 only in the absence of RXR was demonstrated exclusively by

Aristolochic acid. 5) All antagonists as expected recruited corepressor NCoR both in the presence and absence of coregulator RXR.

Clustering of M2H data

In order to visualize VDR functional data in a global context, the mammalian 2- hybrid (M2H) data for all VDR agonists/antagonists tested were visualized as a heatmap Fig.

7. The data resulted in two empirical clusters with CI comprised of VDR:RXR and

VDR:NCoR interactions forming a co-cluster and CII comprised of VDR:SRC-1 interactions forming a solitary subcluster. CI is defined by an overall lower level of activity across the majority of VDR agonists/antagonists. CII exhibits an overall higher assay activity across all a majority of compounds examined. Within each condition (i.e. RXR, SRC-1 or NCoR) the presence or absence of full-length co-regulators to the M2H assay paired together. With

NCoR, addition of full length RXR did not appear to significantly facilitate VDR:NCoR interactions. Conversely addition of full length SRC-1 to VDR:RXR assays and addition of

full length RXR to VDR:SRC-1 assays appeared to have a observable effect. In relation to clustering of VDR active compounds, there appeared to be four predominant subclusters.

Subcluster SI is comprised of both potent agonists and antagonists and appears to cluster based on both VDR:RXR and VDR:SRC-1 interactions. Subcluster SII is comprised of two compounds that strongly recruited NCoR. Subcluster SIII is predominantly comprised of potent antagonists and a few relatively less potent agonists and is driven through VDR:SRC-

1 interactions. Subcluster IV is comprised of VDR agonists that exhibit minimal coregulator recruitment.

Endogenous CYP24A1 induction

The ability of VDR agonists/antagonists to induce or inhibit endogenous expression of CYP24A1, a highly inducible transcriptional target of VDR/1,25D3 was assessed in human myelocytic leukemic (HL60) cells. To ensure consistency in data outcomes between agonists and antagonists, all assays were conducted in the presence of 3nM 1,25D3 and results are expressed as the percentage of gene induction/repression that surpassed baseline 1,25D3 induction alone (Fig. 8a). The concentration of each VDR agonists/antagonist was adjusted to a maximal tolerated dose that did not exhibit HL60 cell cytotoxicity (Fig 12-14). Similar to reporter assays, strong agonists such as calcipotriol, ergocalciferol and lithocholic acid exhibited marked induction compared to weaker agonists such as triamterene, lanoconazole and tamoxifen citrate. However, the induction values of compounds: 4,4'- butylidenebis(6-tert-butyl-m-cresol) and disodium 4,4'-bis(2-sulfostyryl) biphenyl were

higher than expected (Fig. 5a). These compounds exhibited weak agonist activity in transient transfection assay, suggesting the possibility of synergistic activities for these compounds. Of the VDR antagonists 14 of 19 significantly inhibited 1,25D3 induced expression of CYP24A1 (Fig. 8b). Conversely, the five remaining compounds that exhibited antagonist activity in transactivation studies including tazobactam sodium, aristolochic acid, dichlone, chlorambucil and proscillaridin did not exhibit any marked attenuation below

1,25D3 baseline activity in this assay. Among the three potential reverse agonists identified in transient transactivation assays, only menadiol was able to synergize 1,25D3 induced

CYP24A1 expression in HL60 cells (Fig. 8b).

DISCUSSION

The emergence and increased use of qHTS bioprofiling as a tool for cost effective and rapid screening of a large and diverse chemical space has transformed the toxicity- testing landscape in the past decade2. Following National Research Council’s recommendations1, the initiative taken by a group of core governmental agencies informally known as Tox21 and ToxCast, has over the years resulted in generation of compound libraries with critical chemogenomics data regarding their potential to affect biological pathways that was previously unknown. This information now serves as a publicly available source for application of computational and predictive principles for hazard identification and risk assessment, and generating hypotheses on toxicity mechanisms24-26.

Mechanisms associated with toxicity of environmental compounds often involve interaction with units of the transcriptional machinery at the molecular level. In this regard, nuclear receptors are attractive targets for exogenous molecules owing to their characteristic C’ terminal ligand binding domains that are moderately conserved across members of the NR superfamily11. To accommodate the growing need of better understanding NR functions in response to xenobiotic exposures and to satisfy Tox21 objectives, a library of approximately

8,500 environmentally relevant chemicals was screened in qHTS format against a panel of

10 nuclear receptors that included VDR27. Subsequently, a variety of naturally occurring, synthetic, and environmental chemicals active against individual nuclear receptors including

ER, FXR and CAR have been reported7, 28.

The role of VDR as a potential xenobiotic receptor and a target for endocrine disruption remains relatively underexplored. Xenobiotic induced modifications within the vitamin D axis have however been reported, where in vivo exposures result in accelerated

29 synthesis or metabolism of the active form of vitamin D; 1,25D3. Recently, Johns and colleagues (2016) reported epidemiological evidence suggesting that environmental exposure to Phthalates and BPA might alter circulating levels of 1,25D3 in adults in the US although the exact mechanism of VDR disruption has not been elucidated. Both compounds are well known for impacting reproductive processes through AR and ER in addition to putative activities with other NR’s30. Similarly, Nishimura31 et al 2009, demonstrated that

TCDD, a potent AhR agonist, induces up-regulation of vitamin D 1alpha-hydroxylase in kidney, resulting in a 2-fold increase in 1,25D3 serum levels. Less studied is the potential of

VDR to serve as a direct molecular target for xenobiotics and endocrine disruptors. In this study, Tox21 VDR transactivation data was mined to develop a short list of compounds for confirmation in orthogonal assays. We developed a robust selection criteria based on potency and efficacy values, activity in viability assays, curve fit parameters that included shape of dose response curve, flags and sensitivity and selectivity of compounds to VDR.

The objective for confirmation with orthogonal assays was to address key fundamental questions about nuclear receptor function with VDR as a potential target for xenobiotics.

We examined how structurally and functionally diverse compounds may modify (induce or inhibit) core nuclear receptor functions of VDR with respect to its ability to heterodimerize human RXR, recruit coactivator (SRC-1) and corepressor (NCoR-1) and transactivate the

CYP24 promoter. Molecular docking simulations were further conducted to identify key structural interactions between “active” VDR agonists and antagonists and the VDR ligand binding domain.

Transient transactivation assays were performed using full-length human VDR gene reporter construct as opposed to using GAL4-DNA binding domain and NR-ligand binding domain chimeras27 to avoid false negative/positive results. We used a CYP24 promoter fused to a Luciferase reporter in HEK293 cells for this assay. The human CYP24 reporter consists of two DR3 type vitamin D3 response elements (VDREs) located between -140 and -

300 bp upstream of the transcriptional start site of the human CYP24 gene32. Twenty out of

21 agonists were transcriptionally active similar to the qHTS datasets with their activities ranging from an AC50 of 0.009M for Calcipotriol to an AC50 of 37.4M for Novaluron. As anticipated potent agonists such as Calcipotriol exhibit close structural similarities with

1,25D3. However, Ergocalciferol while still structurally similar to 1,25D3 is relatively less potent. Lithocholic acid exhibited an AC50 of 22.29M, which is equivalent to the reported value of 22.39 M in the VDR beta-lactamase qHTS assay27. 9 aminoacridine monohydrochloride, 2,2'-methylenebis(6-tert-butyl-4-ethylphenol) and Tamoxifen citrate and Lanoconazole were each found to have comparatively weaker transactivational activities with higher AC50 values. Interestingly, Tamoxifen citrate is a potent ER repressor

(Huang et al 2011) although it was found to be an active VDR inducer. This again exemplifies the varied nature of xenobiotic interactions with select nuclear receptors that may in fact

facilitate differential biological responses depending upon selective receptor interactions and cell-specific receptor expression.

The transcriptional repression exhibited by identified VDR antagonists was verified to be the outcome of actual chemical-induced VDR inhibition and not artificial interference resulting from cytotoxicity. The use of Dual Glow Luciferase Assay system wherein the expression of an experimental reporter was normalized to that of an internal control reporter which aided in differentiating specific and non-specific cellular responses and offered control over transfection efficiencies between wells (Promega Corp, Madison, WI,

USA). Cell viabilities were measured for all cell lines utilized in this study and aided in determining appropriate dose ranges for test compounds prior to conducting in vitro reporter assays. Only doses that yielded more than 80% cell viability were chosen to be appropriate (Fig 12-14). Antagonists exhibited minimal to marked inhibitory activity with

Fluorescein sodium as an exception, as it did not exhibit any activity in assays utilized in this study. Being a fluorescent tracer that is used extensively in diagnostic medicine33 it is highly likely that was a false positive in the qHTS antagonist assay due to autofluorescence5. The remainder of targeted antagonists was confirmed active in the presence of 3nM 1,25D3.

Metal containing compounds (10/19) listed here in their increasing order of potencies included Phenylarsine oxide, Triphenyltin hydroxide, Potassium dicyanurate, Cadmium reference salt, Cadmium acetate, Tributyltin chloride, Cadmium chloride, Dibutyltin dichloride, Ziram, and Cadmium dinitrate were highly active in repressing VDR transactivation with AC50 values that ranged between 0.01M- 1.33M (See Table 1).

Among these the inhibitory effect of organotins such as the Tributyltin chloride and

Triphenyltin hydroxide, cadmium salts and arsenic-containing compounds are of particular interest, due to well-established linkages to endocrine disruption via activities with other nuclear receptors including PPARy and ER. An important distinction in the effects exist however, in that all three metals tend to have a stimulatory effect on PPARy and ER, 34-37 while they exhibit a potent inhibitory effect on VDR. Nevertheless, functional disruption of a vital endocrine receptors including VDR can result in widespread systemic adverse effects.

Non-metal containing compounds including Chlorambucil, Tazobactam sodium, Thiram,

Proflavin hydrochloride, Aristocholic acid and Proscillaridin additionally had potent inhibitory activity values in the increasing order of listing (AC50 = 0.02M - AC50 = 1.89M). A unique group of reported Tox21 antagonists including Carfizomib, Dichlone and Menadiol did not demonstrate any activity when run in agonist mode. Rather this group of compounds demonstrated a synergistic response in the presence of 1,25D3. Carfizomib is a second-generation irreversible (26S) proteasome inhibitor used as a chemotherapeutic agent against multiple myeloma38. Proteasomes are responsible for protein degradation including nuclear receptors39 and studies have shown that inhibition of proteasome activity can result in increased accumulation and transactivation of nuclear receptors. In fact,

Kongsbak40 and colleagues (2014) have demonstrated up-regulation of VDR protein expression and increased 1,25D3 induced gene activation following proteasome inhibition suggesting that a similar mechanism might be at play with respect to the actions of

Carfizomib. Dichlone is a potent fungicide and pesticide that also induces global DNA

hypomethylation by repressing the action of DNA methyltransferases (DNMTs) suggesting a potential epigenetic role in promoting VDR transactivation41. Water-soluble vitamin K3 or

Menadiol is used to treat coagulopathies associated with obstructive liver disease42. It is unclear how this compound induces VDR transactivation and as such its mode of action could at best be speculated.

Moreover, we applied structure-based 3D docking and molecular dynamics simulations to predict and analyze the binding modes of our experimental hits (including both agonists and antagonists). The excellent docking scores afforded by all agonists and the excellent match between the docking score of calcipotriol with its experimental inhibition potency (IC50 = 8.4 nM) demonstrated the relevance and ability of our analysis to discriminate those active compounds, similar to other studies that have coordinated molecular docking and cell based functional assays to assess VDR activities (Lau et al 2016).

However, our docking model was less suitable to specifically identify VDR antagonists and thus, further investigation using ensemble docking (i.e., combinatorial docking using diverse series of conformations for antagonists as well as a collection of different conformations for the VDR binding site) will be necessary to better account for the flexibility of the binding site. We then conducted molecular dynamic simulations on calcipotriol (agonist) and proflavin hydrochloride (antagonist) to have a better understanding of the dynamic non- covalent interactions of these two compounds once enclosed in the VDR ligand-binding site.

Such analysis was highly critical to determine the relative importance of each residue involved in those interactions. In particular, we demonstrated that the H-bond interaction

between calcipotriol and Ser237 played a major role in the binding abilities of the small molecule ligand, as shown by the high conservation ratio through the MD simulation. In fact, Ser237 could be a suitable candidate for a mutagenesis study. Moreover, both Tyr143 and Ser278 also represent important anchors for calcipotriol but presented lower persistence rates through MDS. With proflavin, MDS results demonstrated that this antagonist forms less stable interactions compared to calcipotriol, as illustrated by the interaction persistence scores (all being lower than 40%) and further confirmed by the higher docking scores and the lower experimental potency. However, molecular dynamic simulations and the superimposition of proflavin and calcipotriol also indicated several shared amino acid contact residues and structural arrangement for both compounds within the VDR LBD. The consistency of orientation between these selected compounds suggest that both VDR agonists and non-metal containing antagonist are capable of dynamic interactions within the receptor, but likely facilitate differential allosteric conformations essential for receptor activation and repression. Interestingly, a recent analysis based on zebrafish VDR43 demonstrated that VDR also presents an alternative binding site when co- crystalized with the agonist lithocholic acid. The structure used in our analysis interestingly presented this alternative binding pocket. So, we decided to evaluate the root mean square fluctuation of the amino acids involved in the second binding site (Ser235, Gln239, Asp149 and Lys240) using MDS in presence of calcipotriol and proflavin hydrochloride. Interestingly, the four amino acids presented smaller fluctuations when VDR was interacting with the agonist calcipotriol (average of RMSF = 0.47Å) than with the antagonist proflavin (average

of RMSF = 0.68Å). Our MDS results might indicate a tendency of VDR to present a second binding site when interacting with an agonist, but longer in depth MDS computations (up to one 10 µs) are needed to validate this hypothesis.

The mammalian 2-hybrid (M2H) assay is a robust tool for studying protein-protein interactions between structural domains or full-length nuclear receptors and other proteins associated with transactivation44. Data outcomes from M2H experiments in this study suggest significantly diverse and complex ligand induced protein:protein interactions with

VDR and VDR coregulators. Results are consistent with the observation that the holo conformation of VDR is ligand-specific and is pivotal for revealing receptor:coregulator interaction domains associated with RXR heterodimerization and recruitment on coactivators and corepressors45. Ligand binding of VDR induces an allosteric shift in receptor conformation, where helix H12 (AF2 domain) rotates and packs tightly over helices H3, H4, and H5, creating a hydrophobic ligand-binding pocket46. The repositioning of H12 creates a

“charge clamp” between the negatively charged glutamate residue (E420 in human VDR) of the AF2 region of H12, and positively charged lysine residue (K246 in human VDR) of H3.

The charge clamp is responsible for coactivator interaction by directly binding with the

LXXLL amino acid motif(s) within the NR box of coactivators47. Small changes in ligand structure appear to affect receptor configurations impacting co-activator binding interface and ultimately varying efficacy and potency of NR transactivation. Supporting this model, our data with Calcipotriol and Ergocalciferol both agonists in transactivation assays, exhibit the ability to fully recruit both VDR:RXR and VDR:SRC-1 interactions similar to 1,25D3.

Comparatively, LCA which functions as a less potent VDR agonist, exhibits attenuated recruitment of coregulators compared to 1,25D3. Previous studies demonstrate that LCA exhibits a selective pattern of ligand-VDR coregulator associations that distinguishes

1,25D3-VDR endocrine from LCA-VDR metabolic functionalities45. Similarly, studies examining 1,25D3 analogs with select 22-alkyl sidechain substitutes also illustrate that slight modifications to ligand structural result in significant alterations in VDR functionality through altering allosteric receptor interactions within helix 1248. In this study Anami and colleagues provide select models of VDR conformations that distinguish between full VDR agonists, partial VDR agonists and VDR antagonists.

In regards to the functionality of partial VDR agonists, multiple modes of activity have been proposed. For instance, ligand mediated phosphorylation with within the AF1 domain can induce receptor transactivation as observed with selective estrogen receptor modulators49. With PPAR, partial agonists have been demonstrated to facilitate suboptimal positioning of receptor conformations resulting in destabilization of helix H12 distinct from conformations induced with full receptor agonists or antagonists50. Partial agonist activity may also result from mixed receptor confirmations where ligands possess both agonist and antagonist properties as previously described for VDR47,51. In this study, partial VDR agonists

(Supplementary Table 2) predominantly exhibited an attenuated recruitment of VDR:RXR and VDR:SRC-1 interactions in the absence of co-transfection with coregulators compared to 1,25D3. This is likely due to an inability of these agonists to re-localize helices H12 and induce a structural transition that triggers the mousetrap-like mechanism stabilizing ligand

binding and co-regulator recruitment52. However, further studies will be needed to identify exact mechanism(s) for each compound tested.

Supplementation of coregulators facilitated selective alterations in receptor transactivation, and protein-protein interaction between VDR:RXR and VDR:SRC-1 with select full/partial agonists. We have previously demonstrated that VDR co-transfections with full-length coregulators enhances protein-protein interactions between VDR, RXR and

53 54 SRC-1 with 1,25D3 as a primary ligand . Thompson and colleagues proposed that the AF2 regions of both VDR and RXR interact with different LXXLL motifs within a single SRC/p160 coactivator. This “bridging” effect of SRC-1 and putatively other coactivators may facilitate stabilization of H12 with less optimal heterodimers. Similarly, both LXXLL motifs of DRIP1 appear to be used by the VDR-RXR heterodimer, suggesting that DRIP1 interacted with the

AF2 regions of both receptors55. Conversely, VDR:RXR heterodimers may exhibit ligand specific protein recruitment with distinct and separate coactivators. For instance, it has been demonstrated that TIF1 can interact with both RXR and VDR, while SUG1 exclusively interacts with VDR56. Differential coactivator recruitment between heterodimer partners may potentially explain our mammalian 2-hybrid data with select ligands. The fact that the cotransfection of either RXR or SRC-1 promotes recruitment between VDR and RXR or

SRC-1 suggests the possibility that coregulators enhance stabilization of H12 through bridging or differential recruitment of p160 family members, which can enhance receptor transactivation. One notable exception however was our observation that the EGFR agonist

falnidamol hydrochloride induced recruitment of RXR both in the presence and absence of co-transfected SRC-1 although it remained inactive in transient transactivation assay.

In comparison, we predicted that VDR antagonists would attenuate 1,25D3 mediated RXR heterodimerization and recruitment of SRC-1. While a reduction in RXR and SRC-1 recruitment was observed with the majority of antagonists as anticipated, other antagonists enhanced select VDR:coregulator interactions including: selective attenuation of either RXR or SRC-1 recruitment or selective recruitment of RXR and SRC-1 both in the presence and absence of cotransfected coregulators. Notable among such interactions were those of phenylarsine oxide and ziram. While the former only inhibited recruitment of RXR and SRC-1 in the absence of cotransfected coregulators, ziram selectively facilitated SRC-1 recruitment only in the absence of cotransfected RXR. Lastly, all antagonists were, as expected, were capable of recruiting NCoR.

Receptor antagonists can be categorized as either “active” or “passive” 57. Active antagonists tend to have bulky structures that destabilize the active confirmation of helix 12 resulting in stearic obstruction of motifs essential for NR-coactivator interactions. By contrast, passive antagonists tend to fit into the ligand-binding pocket but facilitate repositioning of H12 to a stable but non-active configuration. A third mechanism of receptor antagonism has also been proposed where antagonists can facilitate H12 stabilization but destabilize other regions of the receptor including the dimerization interface, impeding the ability to form productive heterodimers with RXR58. It is worth

noting however that non-competitive VDR antagonists have also been identified that function through disruption of VDR-coregulator interactions59. The fact that several of the non-metal containing VDR antagonists identified in this study did not afford a good docking score or were not docked at all, may indicate that several of these compounds function through non-ligand mediated mechanisms that disrupt co-regulator interaction ultimately attenuating or inhibiting VDR transactivation.

We next confirmed the effect outcomes of compounds obtained from transient transactivation assay by endogenous gene activation followed by qPCR in human myelocytic leukemic (HL60) cells, a cell line that expresses VDR and effectively induces CYP24A1 expression in the presence of 1,25D360-61. Because this cell line proved to be highly sensitive with relatively lower cell viabilities at comparable concentrations used for reporter gene assays, the final concentrations for the majority of compounds were adjusted to prevent cytotoxicity and accommodate healthy cell viability values (>80%). Agonists were tested in the presence of 3nM of 1,25D3 such that the combined fold induction surpassed the relative 1,25D3 induction and was expressed as a percentage (Fig. 6a). Expectedly, all agonists further enhanced CYP24A1 expression. Despite weaker transactivation activities, the induction exhibited by 4,4'-butylidenebis(6-tert-butyl-m-cresol and Disodium 4,4'-bis(2- sulfostyryl) biphenyl could be the result of a synergistic response in the presence of 1,25D3.

Most antagonists significantly repressed endogenous CYP24A1 expression. However,

Tazobactam sodium, Aristolochic acid, Dichlone, Chlorambucil and Proscillaridin did not follow a similar pattern which could be attributed to the adjusted low doses used for this

assay as mentioned earlier. Carfizomib being a potent protease inhibitor and antineoplastic agent effective against multiple myeloma38 was predicted to have an inhibitory effect even at low dose (1uM) in HL60 cells despite its reverse agonist response in other cell types and assays. These variations in the CYP24A1 response not only confirm the high sensitivity of this particular cell line to xenobiotic exposure but also provide further evidence to support the role of VDR as a potential target for xenobiotics that is able to bind structurally diverse endogenous and exogenous compounds and modulate activity of other important genes accordingly.

CONCLUSION

Quantitative high throughput chemical screens have been instrumental in identifying compounds that are active towards a variety of nuclear receptors. Those experimental bioprofiles have provided a convenient method of gaining novel information on hundreds of compounds that are potentially toxic, and provide global assessment of ligand interactions with nuclear receptors. In line with the continued surge in scientific interest in dissecting the roles played by NR’s specifically VDR in mechanisms associated with toxicity, we explored the utility of confirming high throughput analysis with orthogonal assays and the potential of VDR as a target of xenobiotics and endocrine disruptors.

Through application of in vitro cell based assays and in silico modeling approaches, we demonstrate the molecular complexity of VDR:ligand interactions and confirm the ability of diverse ligands to directly bind VDR, facilitate differential coregulator recruitment and activate/repress receptor mediated transcription. Each of these processes are likely to contribute to the varied potencies and efficacies observed with VDR ligands examined.

REFERENCES

1. NRC. Toxicity Testing in the 21st Century: A Vision and a Strategy. (The National Academies Press, Washington, DC, 2007).

2. Dix, D.J. et al. The ToxCast program for prioritizing toxicity testing of environmental chemicals. Toxicol Sci 95, 5–12 (2007).

3. Collins, F. S., Gray, G. M. & Bucher, J. R. Toxicology. Transforming environmental health protection. Science 319, 906–907 (2008).

4. Shukla, J.S., Huang, R., Austin, C.P., & Xia, M. The future of Toxicity testing: A focus on in vitro methods using a quantitative high throughput screening platform. Drug Discov Today 15, 997-1007 (2010).

5. Tice, R. R., Austin, C. P., Kavlock, R. J. & Bucher, J. R. Improving the human hazard characterization of chemicals: a Tox21 update. Environ Health Perspect 121, 756–765 (2013).

6. Huang, R. et al. Profiling of the Tox21 10K compound library for agonists and antagonists of the estrogen receptor alpha signaling pathway. Sci Rep 4, 5664; 10.1038/srep05664 (2014).

7. Hsu, C.W. et al. Quantitative High-Throughput Profiling of Environmental Chemicals and Drugs that Modulate Farnesoid X Receptor. Sci Rep 4, 6437 (2014).

8. Huang, et al. Tox21 challenge to build predictive models of nuclear receptor and stress response pathways as mediated by exposure to environmental chemicals and drugs. Front Environ Sci 2, 10.3389/fenvs.2015.00085 (2016).

9. Eyles, D.W., Burne, T.H.J., & Mcgrath, J.J. Vitamin D, effects on brain development, adult brain function and the links between low levels of vitamin D and neuropsychiatric disease. Front Neuroendocrinol 34, 47-64 (2013).

10. Delfosse, V., Marie, A., Balaguer, P., & Bourguet, W. A structural perspective on nuclear receptors as targets of environmental compounds. Acta Pharmacologica Sinica 36, 88- 101 (2015).

11. Reschly, E.J., & Krasowski, M.D. Evolution and function of the NRII Nuclear hormone receptor subfamily (VDR, PXR, and CAR) with respect to metabolism of xenobiotics and endogenous compounds. Curr Drug Metab 7, 349-365 (2006).

12. Schuster, I. Cytochromes P450 are essential players in the vitamin D signaling system. Biochimica et Biophysica Acta 1814, 186–199 (2011).

13. Hollick, M.F. Vitamin D deficiency. NEJM 357, 266-281 (2007).

14. Filer, D.L., Kothiya, P., Setzer R.W., Judson, R.S., & Martin, M.T. The ToxCast pipeline for high-throughput screening data. Bioinformatics 33, 618-620 (2017).

15. Rochel, N., Wurtz, J. M., Mitschler, A., Klaholz, B., & Moras, D. The crystal structure of the nuclear receptor for vitamin D bound to its natural ligand. Mol Cell 5, 173–179 (2000).

16. Matsuo, M. et al. Synthesis of 2alpha-heteroarylalkyl active vitamin d3 with therapeutic effect on enhancing bone mineral density in vivo. ACS Med Chem Lett 4, 671–674 (2013).

17. Verlinden, L. et al. Synthesis, structure, and biological activity of des-side Chain analogues of 1α,25-Dihydroxyvitamin D3 with substituents at C18. Chem Med Chem 6, 788–793 (2011).

18. Hourai, S. et al. Probing a water channel near the A-ring of receptor-bound 1 alpha,25- dihydroxyvitamin D3 with selected 2 alpha-substituted analogues. J Med Chem 49, 5199–5205 (2006).

19. Tocchini-Valentini, G., Rochel, N., Wurtz, J. M., & Moras, D. Crystal Structures of the Vitamin D Nuclear Receptor Liganded with the Vitamin D Side Chain Analogues Calcipotriol and Seocalcitol, Receptor Agonists of Clinical Importance. Insights into a Structural Basis for the Switching of Calcipotriol to a Receptor. J Med Chem 47, 1956– 1961 (2004).

20. Lee, K. Y. et al. Quercetin directly interacts with vitamin D Receptor (VDR): Structural implication of VDR activation by quercetin. Biomol Ther 24, 191–198 (2016).

21. Schrödinger Release 2016-3. Protein Preparation Wizard; Epik. (2016).

22. Friesner, R. A. et al. Glide: a new approach for rapid, accurate docking and scoring. 1. Method and assessment of docking accuracy. J Med Chem 47, 1739–1749 (2004).

23. Friesner, R. A. et al. Extra precision glide: Docking and scoring incorporating a model of hydrophobic enclosure for protein ligand complexes. J Med Chem 49, 6177–6196 (2006).

24. Lock, E.F. et al. Quantitative high-throughput screening for chemical toxicity in a population based in vitro model. Toxicol Sci 126, 578–588 (2012).

25. Judson, R. et al. Perspectives on Validation of High-Throughput Assays Supporting 21st Century Toxicity testing. ALTEX 30, 51-56 (2013).

26. Xia, M. et al. Compound cytotoxicity profiling using quantitative high throughput screening. Environ Health Perspect 116, 284-291 (2008).

27. Huang, R. et al. Chemical genomics profiling of environmental chemical modulation of human nuclear receptors. Environ Health Perspect 119, 1142–1148 (2011).

28. Lynch, C. et al. Quantitative High-Throughput Identification of Drugs as Modulators of Human Constitutive Androstane Receptor. Sci Rep 5, 10405; (2015).

29. Johns, L.E., Ferguson K. K., Meeker, J. D. Relationships between urinary phthalate metabolite and bisphenol A concentrations and vitamin D levels in US adults: National Health and Nutrition Examination Survey (NHANES), 2005-2010. J Clin Endocrinol Metab 101, 10.1210/jc.2016-2134 (2016).

30. Kandarakis, E.D. et al. Endocrine-Disrupting chemicals: An endocrine society scientific statement. Endocrine rev 30, 293-342 (2009).

31. Nishimura, N. et al. Dioxin induced up-regulation of the active form of vitamin D is the main cause for its inhibitory action on osteoblast activities, leading to developmental bone toxicity. Toxicol Appl Pharamacol 236, 301-9 (2009).

32. Chen, K.S., & DeLuca, H.F. Cloning of the human 1,25-dihydroxyvitamin D3 24- hydroxylase gene promoter and identification of two vitamin D-responsive elements. Biochem Biophys Acta 1263, 1–9 (1995).

33. Ayestaray, B., & Bekara, F. Fluorescein sodium fluorescence microscope-integrated lymphangiography for lymphatic supermicrosurgery. Microsurgery 35, 407-10 (2015).

34. Parent, M.E., and Siemiatycki, J. Occupation and prostate cancer. Epidemiol Rev 23, 138–143 (2001).

35. Kanayama, T., Kobayashi, N., Mamiya, S., Nakanishi, T., & Nishikawa, J. Organotin compounds promote adipocyte differentiation as agonists of the peroxisome proliferator-activated receptor/retinoid X receptor pathway. Mol Pharmacol 67, 766– 774 (2005).

36. Grun, F. et al. Endocrine-disrupting organotin compounds are potent inducers of adipogenesis in vertebrates. Mol Endocrinol 20, 2141–2155 (2006).

37. Davey, J.C., Bodwell, J.E., Gosse, J.A., & Hamilton, J.W. Arsenic as an endocrine disruptor: effects of arsenic on estrogen receptor-mediated gene expression in vivo and in cell culture. Toxicol Sci 98, 75–86 (2007).

38. Mcbride, A., Klaus, J.O., & Keith, S.G. Carfizomib: A second-generation protease inhibitor for the treatment of multiple myeloma. American Journal of Health-System Pharmacy. 72, 353-360 (2015).

39. Keppler, B.R., Archer, T.K., & Kinyamu, H.K. Emerging roles of the 26S proteasome in nuclear receptor-regulated transcription Biochimica et Biophysica Acta 1809, 109-118 (2011).

40. Kongsbak, M. et al. Vitamin D up-regulates the Vitamin D Receptor by protecting it from proteasomal degradation in Human CD4+ T Cells. PLoS One 9, 96695; 10.1371/journal.pone.0096695 (2014).

41. Ceccaldi, A. et al. Identification of novel inhibitors of DNA methylation by screening of chemical library. ACS Chem Bio 8, 43-548 (2013).

42. Green, B., Cairns, S., Harvey, R., & Pettit, M. Phytomenadione or Menadiol in the management of elevated international normalized ratio (prothrombin time). Aliment Pharmacol Ther 14, 1685-1689 (2000).

43. Belorusova, A. Y. et al. Structural insights into the molecular mechanism of vitamin D receptor activation by lithocholic acid involving a new mode of ligand recognition. J. Med. Chem. 57, 4710–4719 (2014). 44. Hi, R. & Li, X. Mammalian two-hybrid assay for detecting protein protein interactions in vivo. Methods Mol Biol 439, 327-37 (2008).

45. Jurutka, P.W. et al. Molecular and functional comparison of 1,25-dihydroxyvitamin D(3) and the novel vitamin D receptor ligand, lithocholic acid, in activating transcription of cytochrome P450 3A4. J Cell Biochem 94, 917–43 (2005).

46. Rochel, N., Wurtz, J. M., Mitschler, A., Klaholz, B., & Moras, D. The crystal structure of the nuclear receptor for vitamin D bound to its natural ligand. Mol Cell 5, 173–179 (2000).

47. Savkur, R.S., Bramlett, K.S., Clawson, D., & Burris, T.P. Pharmacology of Nuclear Receptor- Coregulator Recognition (ed. Litwack, G.) 145-183 (Elsevier Academic Press, 2004).

48. Anami, Y.et al. Fine tuning of agonistic/antagonistic activity for vitamin D receptor by 22-alkyl chain length of ligands: 22S-Hexyl compound unexpectedly restored agonistic activity. Bioorg Med Chem 23, 7274-81 (2015).

49. Lipfert, L. et al. Antagonist- induced activation function-2-independent estrogen receptor alpha phosphorylation. Mol Endocrinology 20, 516-33 (2006).

50. Ohashi, M., Oyama, T., & Miyachi, H. Different structures of the two-peroxisome proliferator-activated receptor gamma (PPARγ) ligand-binding domains in homodimeric complex with partial agonist, but not full agonist. Bioorg Med Chem Lett 25, 2639-44 (2015).

51. Nakabayashi, M. et al. Crystal structures of rat vitamin D receptor bound to adamantyl vitamin D analogs: structural basis for vitamin D receptor antagonism and partial agonism. J Med Chem 51, 5320-9 (2008).

52. Singarapu, K.K. et al. Ligand-specific structural changes in the vitamin D receptor in solution. Biochemistry 50, 11025-33 (2011).

53. Kollitz EM, Hawkins MB, Whitfield GK, & Kullman SW. Functional diversification of vitamin D receptor paralogs in teleost fish after a whole genome duplication event. Endocrinology 155, 4641-54 (2014).

54. Thompson, P.D. et al. Distinct retinoid X receptor activation function-2 residues mediate transactivation in homodimeric and vitamin D receptor heterodimeric contexts. J Mol Endocrinol 27, 211–22 (2001).

55. Rachez, C. et al. The DRIP complex and SRC-1/p160 coactivators share similar nuclear receptor binding determinants but constitute functionally distinct complexes. Mol Cell Biol 20, 2718-2726 (2000).

56. vom Baur E. et al. Differential ligand-dependent interactions between the AF-2 activating domain of nuclear receptors and the putative transcriptional intermediary factors mSUG1 and TIF1. Embo J 15, 110-124 (1996).

57. Shiau, A. K. et al. Structural characterization of a subtype-selective ligand reveals a novel mode of estrogen receptor antagonism. Nat Struct Biol 9, 359-64 (2002).

58. Kato, A., Itoh, T., Anami, Y., Egawa, D., & Yamamoto, K. Helix12-Stabilization Antagonist of Vitamin D Receptor. Bioconjug Chem 27, 1750-61 (2016).

59. Nandhikonda, P. et al. Discovery of the first irreversible small molecule inhibitors of the interaction between the vitamin D receptor and coactivators. J Med Chem 55, 4640-51. (2012).

60. Sidhu, P.S., et al. Development of novel Vitamin D receptor coactivator inhibitors. ACS Med Chem Letters 5, 199-204 (2014).

61. Van Driel, M., and van Leeuwen, J.P. Vitamin D endocrine system and osteoblasts. Bonekey Rep 5, 493 10.1038/bonekey.2013.227(2014).

62. Attene-Ramos, M.S. et al. The Tox21 robotic platform for the assessment of environmental chemicals –from vision to reality. Drug Discov Today 18, 716-23 (2013).

63. Schrödinger Release 2015-4. Prime. (2015).

64. Wickham, Hggplot2: Elegant Graphics for Data Analysis. Springer-Verlag New York, 2009.

65. R Core Team (2016). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/.

66. Warnes, G.R. et al. gplots: Varipus R programming tools for plotting data. R package version 3.0.1. http://CRAN.R-project.org/package=gplots.

67. Howarth, D.L. et al. Paralogous vitamin D receptors in teleosts: Transition of nuclear receptor function. Endocrinology 149, 2411–2422 (2008).

68. Yosuke, A., Yoshitake, C., Manabu, M., Kazuo, K., & Makoto, M. Increased nuclear expression and transactivation of vitamin D receptor by the cardiotonic steroid bufalin in human myeloid leukemia cells. J Steroid Biochem Mol Biol 114, 144-151 (2009).

69. Pfaffi, M.W. A new mathematical model for relative quantification in real--time RT-PCR. Nuclei Acids Res 29, 2002-7 (2001).

70. Filer, Dayne L., et al. "tcpl: the ToxCast pipeline for high-throughput screening data." Bioinformatics 33.4 (2016): 618-620

71. Lau, A. J., Politi, R., Yang, G. & Chang, T. K. H. Cell-based and in silico evidence against quercetin and structurally-related flavonols as activators of vitamin D receptor. J. Steroid Biochem. Mol. Biol. (2016). doi:10.1016/j.jsbmb.2016.03.039

FIGURES

Figure 1. Schematic overview of compound selection criteria and experimental workflow.

a b

Figure 2. Dose response curves of select VDR agonists identified by transient transactivation assay (Cyp24-Luc). Dose response curves of a) Vitamin D3 and related active analogs:

Calcipotriol (AC50 = 0.0002uM), 1-25 dihydroxy vitamin D3 (AC50 = 0.65nM), Ergocalciferol

(EC50 = 14.44uM), LCA (EC50 = 16.82uM); b) Less active agonists: 2,2'-methylenebis(6-tert- butyl-4-ethylphenol) (EC50 = 36.76uM), 9 aminoacridine monohydrochloride (EC50 = 12.58uM),

Tamoxifen Citrate (EC50 = 3.84uM), Lanoconazole (EC50 = 20.99uM). Assays were run in

HEK293 cells and data expressed as mean± SEM (n=3).

Figure 3. Dose response curves of select VDR antagonists identified by transient transactivation assay (Cyp24-Luc). Dose response curves of a) Metal containing compounds:

Phenylarsine oxide (IC50 = 0.012), Dibutyltin (IC50= 0.54), Potassium dicyanurate (IC50 = 0.09),

Cadmium acetate (IC50 = 1.16); b) Non-metal containing compounds: Proscillaridin (IC50 =

1.89), Aristocholic acid (IC50 = 0.82), Thiram (IC50 = 0.42), c) Reverse agonists Carfizomib (IC50

= 0.53), Dichlone (IC50 = 0.24), and Menadiol (IC50 = 3.49). Assays were run in HEK293 cells and data expressed as Mean± SEM (n=3).

a) Select VDR antagonists (metals)

b) Select VDR antagonists (non-metals)

c) VDR reverse agonists

Figure 4. Structure-based molecular docking using Glide and the human VDR structure (PDB code 1S19): (a) Docking results for all compounds with their associated XP docking and eModel scores, mechanism and experimental EC50 values; (b) Structural superimposition of the top-predicted binding poses for calcipotriol (pink) and proflavin (grey) docked in the

VDR pocket.

Figure 5. Protein:protein interaction between VDR with RXR, SRC-1 and NCoR in the presence of select agonists: a) Recruitment of coactivator SRC-1 (SRC/p160 family) by VDR in the presence and absence of RXR: b) Recruitment of heterodimerization partner RXR by VDR in the presence and absence of SRC-1 c) Recruitment of corepressors NCoR by VDR in the presence and absence of RXR. Assays were run in Cos7 cells and data expressed as mean± SEM (n=3). Data are normalized to VDR + empty pM vector. Only significant (at least p <0.05) data points are expressed as a percentage.

a) Recruitment of SRC-1

b) Recruitment of RXR

c) Recruitment of NCor-1

Figure 6. Protein:protein interaction between VDR with RXR, SRC-1 and NCoR in the presence of select antagonists: a) Recruitment of coactivator SRC-1 (SRC/p160 family) by

VDR in the presence and absence of RXR: b) Recruitment of coregulator RXR by VDR in the presence and absence of SRC-1 c) Recruitment of corepressors NCoR by select compounds alone in the presence and absence of RXR. Note that corepressor recruitment of vdr by antagonists was also tested in the presence of vitamin D3, however, values were negligible (data not shown). Assays were run in Cos7 cells and data expressed as mean±

SEM (n=3). Data are normalized to VDR + empty pM vector. Only significant (at least p

<0.05) data points are expressed as a percentage.

a) Recruitment of SRC-1

b) Recruitment of RXR

c) Recruitment of NCor-1

CI CII

SII

SIII

SIV

Figure 7. Heat map showing the variability in the selective preference of compounds to enhance or inhibit the ability of VDR to recruit or interact with coregulator (RXR), coactivator (SRC-1) and corepressor (NCoR-1). Higher recruitment values are indicated in green while lower values are in blue.

Figure 8. Endogenous CYP24A1 induction in HL-60 cells by compounds in the presence of

Vitamin D3: a) VDR agonists: data expressed as percentage of vitamin D alone b) VDR antagonists: data are expressed as percentage of vitamin D alone only for compounds exhibiting significant inhibition of at least p<0.05.

a) VDR agonists

1 %

1400 1

6 %

1200 .

4 7

1000 6 6

800 6

n 9

600 .

% 3

400 %

300 2

% %

2 9

250 1

I 8

4 2

7 1

200 1

4 2

1 1

2 1

1 1

p 150 y

C 100

% 80 60 40 20 0

O M M M M M M M M M M M M M M M M M M M M M M S n u u u u u u u u u u u u u u u u u u n u u M 3 0 0 0 0 5 5 5 .8 0 0 0 0 .5 0 0 0 0 0 0 5 5 3 3 3 2 2 l l) e 1 1 2 2 2 2 5 2 2 2 2 0 e n D e l e 1 o d l ) 1 1 1 d 1 1 1 1 t i D l o e s o i l e i d l e a r n o r n e n r io o id id d d c i n y e t tr a i z fe re n r e o tr n c c o ri a c ro n n a i a i e e c h l e a a m a e e l c m m n c t r - p h o h i lo ic ic lu h y u ta l y m l c ip p ic ic n h l l a h p x n o i o a m p - y o c l a c o o v ip o o e c V c c ia ) l h r l y n n d h f b i b if l o o r a ty t d a h fo fo i ro c o o ) h r x y n g T ( u e y C t l l r d o n N l rt a o th a r o h e u u C i ry e c L E z -b -m i - s s y th m y t - m e n t 4 d -4 e i h i t -2 a m e r l- l l n d o L A s a e T - b te y o ty e e n 4 o h n -4 l - t m u z n o lf lp i ) y (6 u a n e u z o h s b -b e l m s A a n t i t- id t a e - r i e b r n r B h n 2 y m m e e l te t i ( p a n -t a - h id is ro l 7 e 6 F (6 p r b y d ( s a c '- lo h li s i N a 4 h t y i b o , c e t b e ,7 n 4 i m u e n 2 i d i n m - D -b e le m u ,6 ( ' l y -a i 5 - ,4 y h 9 d - 7 4 th t o o e e s n m i i -m '- D m ' 2 -a ,2 , 3 2 2 l y th e M Compound +Vit D3 b) VDR antagonists

250 1

n 200

c u

d 150

% 1

4 A

. %

% 4

3 9

% . 5

% %

. 2 100 7

5 3

9 5

4 p

. .

% 5

5 y

9 3

3 .

5 .

. 4

4 4

7 C

3 6

3 % .

% % 50 %

4 1 2

1 1

1 0

O M M M M M M M M M M M M M M M M M M S n u u u u n u M n u n n u u u u M n u n 3 2 2 0 5 5 1 u 0 2 0 0 3 2 2 6 n 0 1 0 M 1 1 1 . 7 6 0 1 0 5 l 1 1 0 5 l 5 D 3 l 1 . ib e 1 5 2 o m 0 2 i 1 D m c m e 2 t e i te te a 5 c h a n lt m a id id e e d a a r e e u in in iu r o r c r d d a r t i id b d n i lo a z u o ri ri n it e Z id id m i h h s fi n a l e c x x m r a o v T c f r a c h lo lo in a o o a la it s la i e a y li c h h M d r e r il f D r c i c c d n lo V m o C i o d i m y i c a r m d h n d m iu rs h s t P u c m ti u h a C o c i m o l in i m n l r a m t iu y t m d i y P b iu is t l d a lt n o d s r m u ty a y e z a s A d b C n h a C a a ri u C e T t C T ib h P o D p P ri T Compound +Vit D3

Figure 9. Superposition of the predicted binding poses of calcipotriol (green) and proflavin

(purple) docked in the VDR pocket (pdb code: 1S19). Both compounds are predicted to bind the same pocket and interact with several common residues: His397, His305, Ser237 and

Ser278 (according to the structure-based docking and MDS results illustrated in Figures 2 and 3).

Figure 10. (a) 2D predicted binding mode of calcilpotriol using molecular docking, (b) 2D predicted binding mode of calcilpotriol using molecular dynamic. Highlighted in red the interaction in common between the molecular docking and the molecular dynamic procedures, highlighted in orange the interaction only detected during the molecular dynamic procedure.

Figure 11. (a) 2D predicted binding mode of Proflavin hydrochloride using molecular docking, (b) 2D predicted binding mode of Proflavin hydrochloride using molecular dynamic.

Highlighted in red the interaction in common between the molecular docking and the molecular dynamic procedures highlighted in orange the interaction only detected during the molecular dynamic procedure.

Figure 12. Figure showing the results of cell viability assay in HL-60 cells for a) VDR agonists and b) VDR antagonists. The concentrations of compounds corresponding to cell viability percentages of more than 80 were considered appropriate for testing in in-vitro assays. The viability percentage of treatments was measured against that of DMSO. Cells treated with

0.1% Triton X acted as a negative control while those treated with Vitamin D3 acted as a positive control. Data were measured as SEM (n=3) and plotted with Graphpad prism.

a) VDR agonists

VDR Agonists HL60 cells

140

120

l i

b 100

l 80

e c

% 40 20 0

X O M M M M M M M M M M M M M M M M M M M M M M n S n u u u u u u u u u u u u u u u u u u n u u o M 3 0 0 0 0 5 5 5 .8 0 0 0 0 .5 0 0 0 0 0 0 5 5 it 3 3 3 2 2 l) l) e 1 1 2 2 2 2 5 2 2 2 2 0 e n r D e l e 1 d l ) 1 1 1 d 1 1 1 1 t i T D l o e o o i l e i d l e e a r n o r n s n r io o id id d d c i n y t tr a i z fe re n e e o tr n c c o ri a c ro n n a i a i e e r h l e a a m a e e l c m m n c t r -c p h o h i lo ic ic lu h y u ta l y l c ip p ic ic n h l l a h p x n o i o a m p m y o c l a c o o v ip o o e c V c c ia ) l- h r l y n n d h f b i b if l o o r a y t d a h fo fo i ro c o o ) h r x y n g T ( t e y C t l l r d o n N l rt a o th a r o u h e u u C i ry e c L E z b -m i - s s y th m y t - m e n - 4 d -4 e i h i t -2 a m e rt l- l l n d o L A s a e T - b e y o ty e e n 4 o h n -4 l -t t m u z n o lf lp i ) y 6 u n e u z o h ( b a -b e l m s A a n t s t- id t a e - r i e i r n r B h n 2 y m m b e l te t i ( p a e t a - h id is o l 7 n - F 6 p r b r y e (6 ( a c '- lo h d s is N a 4 h t li i b o , c e y b e ,7 n 4 i m t e n 2 i d i u n m - D b e le m u ,6 ( '- l y -a i 5 - 4 y h 9 d - 7 , th t o o 4 e e s n m i i -m '- D m ' 2 -a ,2 , 3 2 2 l y th e M [Compound] b) VDR antagonists

VDR Antagonists HL60 cells

140

120

l i

b 100

l 80

e c

% 40 20 0

X O M M M M M M M M M M M M M M M M M M S n u u u u n u M n u u n u u u u M n u n n 3 2 2 0 5 5 1 u 0 2 1 0 3 2 2 6 n 0 1 0 to M 1 1 1 . 7 6 0 1 5 l 1 1 0 5 l 5 ri D 3 l 1 . ib e 1 e 2 o m 0 2 i 1 T D m c m e 2 t e id i te te a 5 c h a n lt m a id id r e d a a r e e u in in iu r o r c r o d a r t i id b d n i lo a z u o l ri n it e Z id id m i h h s fi n a l h e c x x m r a o v T c f r a c h c lo in a o o a la it s la i e a y li c h M d r e r il f D r c i n c d n lo V m o C i o d ti i m y i c a r m d h l d m iu rs h s t P u c m y u h a C o c i m o t in i m n l r a m t iu u t m d i y P b iu is ib l d a lt n o d s r m r ty a y e z a s A d T C n h a C a a u C e T t C ib h P o D p P ri T Compound +Vit D3

Figure 13. Figure showing the results of cell viability assay in HEK293 cells for a) VDR agonists and b) VDR antagonists. The concentrations of compounds corresponding to cell viability percentages of more than 80 were considered appropriate for testing in in-vitro assays. The viability percentage of treatments was measured against that of DMSO. Cells treated with 0.1% Triton X acted as a negative control while those treated with Vitamin D3 acted as a positive control. Data were measured as SEM (n=3) and plotted with Graphpad prism.

a) VDR agonists

VDR Agonists HEK293 cells

140

120

l i

b 100

l 80

e c

% 40 20 0

X O M M M M M M M M M M M M M M M M M M M M M M n S n u u u u u u u u u u u u u u u u u u u u u o M 3 0 0 0 0 0 0 .9 5 0 0 0 0 0 0 0 0 0 0 .5 0 5 it D 3 2 6 4 2 2 4 0 .0 5 2 2 2 3 5 2 2 2 2 1 2 1 r D 1 l e 1 l) l) 0 l) 1 1 1 e d 1 1 1 1 e n T e ro n e o o e l o d d d d i n l e te t i n l e e n id o i i o i c id o y n a a r i o f r e s n r i n c c r a c r n e l tr a m z i e r re e o tr e a a m lo c a u e y i m a a lc t y c h l o h c i h i c l h h x c u t n a m p - p h p ic i n l i a p p o n i o c a l c ip l n n a c o l v i o b o V c i ) -m y o c y o d o h fo o b i r fe lc o r (a l h r l h fo f ri r c l) th a i y o rg T y t d a t l l d o o N y r c x n o t e y C -e u u C y h in r e - o th a E z u m h 4 s is h t y t 2 e L n b - i - e o i m t a - m e t- 4 d l n d n L A s e a m b r l- l ty e e o o h in T - l e y o u z n 4 lf lp z -4 y -t t b n e m u A a ) h 6 u m - e l e s r o t ( b a rt a - y n e s t- d B h in 2 p i m i r i te t d ( o m b e ln - h i is r a 7 e -t a (6 p r b o l n F a c '- l y e (6 is N a 4 h h d s b o , ic t li i e ,7 in 4 d e y b n 2 - m t e e m m ,6 i u n l -a iu 5 D -b le y 9 - -( ' y th d o 7 ,4 h e o n 4 t is i e -m D m m ' -a '- ,2 3 ,2 2 l 2 y th e M Compound b) VDR antagonists

VDR Antagonists HEK293 cells

140

120

l i

b 100

l 80

e c

% 40 20 0

X O M M M M M M M M M M M M M M M M M M S n u u u u n u M u u u u u u u u M u u u n 3 2 2 2 5 5 2 u 2 2 1 1 2 2 2 6 u 7 2 2 to M 1 1 1 . 7 1 6 1 1 1 1 1 1 3 1 1 i D 3 l 1 . e e e l e . l r D c 2 b t d e d d e e m 0 i n T m m e t i a i d ri ri io t t ra id c i n iu h ra n l m r c i d a ta i x e u id i n i o a u a r lo lo a tr Z id r m d i h l s o n c lo h h i e ro x b a a o v T h f iz a li h c c n in c d m ll t s a ic e f y i e a o a i i fl r r o ic n d M d y e r c V m o D a ic h d ti m h in o s a r m C d c l in m iu n s l ro t P u to m ty lt iu ti r h c i m s u y m l la C P a m u i i u t m d y y b d i r m ib u d a n n o a s A r b a C e e z s d T i C h a C ta a D p h T o C ri P P T Compound +Vit D3

a) VDR agonists VDR Agonists COS7 cells

140

120

i l

i 100

l 80

e c

% 40 20 0

X O M M M M M M M M M M M M M M M M M M M M M M n S n u u u u u u u u u u u u u u u u u u u u u o M 3 0 0 0 0 0 0 .9 5 0 0 0 0 0 0 0 0 0 0 .5 0 5 it D 3 2 6 4 2 2 4 0 .0 5 2 2 2 3 5 2 2 2 2 1 2 1 r D 1 l e 1 l) l) 0 l) 1 1 1 e d 1 1 1 e n T e ro n e o o e l o d d d d i id n l e te t i n l e e n id o i i o i c c o y n a a r i o f r e s n r i n c c r a a r n e l tr a m z i e r re e o tr e a a m lo c c u e y i m a a lc t y c h l o h c i h i li l h h x c u t n a m p - p h p ic i n l a p p o n i o c a l c ip l n n a c o fo v i o b o V c i ) -m y o c y o d o h o b i r fe lc o r (a l h r l h fo f ri r c o l) th a i y o rg T y t d a t l l d o in N y r c x n o t e y C -e u u C y h r e - o th a E z u m h 4 s is h t m y t 2 e L n b - i - e o i A t a - m e t- 4 d l n d n L s e a m b r l- l ty e e o 4 o h in T - l e y o u z n lf lp z -4 y -t t b n e m u A a ) h 6 u m - e l e s r o t ( b a rt a - y n e s t- d B h in 2 p i m i r i te t d ( o m b e ln - h i is r a 7 e -t a (6 p r b o l n F a c '- l y e (6 is N a 4 h h d s b o , ic t li i e ,7 in 4 d e y b n 2 - m t e e m m ,6 i u n l -a iu 5 D -b le y 9 - -( ' y th d o 7 ,4 h e o n 4 t is i e -m D m m ' -a '- ,2 3 ,2 2 l 2 y th e M Compound b) VDR antagonists

VDR Antagonists COS7 cells

140

120

i l

i 100

i v

l e

c 60

% 40 20 0

X O M M M M M M M M M M M M M M M M M M S n u u u u n u M u u u u u u u u M u u u n 3 2 2 2 5 5 2 u 2 2 1 1 2 2 2 6 u 7 2 2 to M 1 1 1 . 7 1 6 1 1 1 1 1 1 3 1 1 ri D 3 l 1 . e e e l m e . l D c m e 2 ib t id e id id o te te a d 0 i in T m h t a d r r i r i e c in iu ra n l m r c ri o o d ra ta i x u id d n i lo a o u a l l a t e Z o id b r m i h h s z n c lo h h n i c r x la a o v T c f i a li h c c e in a d o m il it s la i e rf y o c i d y e a c f D r a c i in d M h n r V m o i h d t m i lo s a r m C d c l in m iu n s ro t P u to m ty lt iu ti r h c i m s u y m l la C P a m u i i u t m d y y b d i r m ib u d a n n o a s A r b a C e e z s d T i C h a C ta a D p h T o C ri P P T Compound +Vit D3

Figure 14. Figure showing the results of cell viability assay in Cos7 cells for a) VDR agonists and b) VDR antagonists. The concentrations of compounds corresponding to cell viability percentages of more than 80 were considered appropriate for testing in in-vitro assays. The viability percentage of treatments was measured against that of DMSO. Cells treated with

0.1% Triton X acted as a negative control while those treated with Vitamin D3 acted as a positive control. Data were measured as SEM (n=3) and plotted with Graphpad prism.

Table 1. Tested compounds with their names, Cas numbers, origin, activity, AC50 values obtained from Tox21 qHTS data set and orthogonal assay values, with their corresponding concentration ranges. Note that some of the Tox21 qHTS AC50 values have a range (*).

Compound Name/ Origin Activity Tox21 Derived AC50 Concentration CAS # AC50 range (uM) Agonists 7-(Dimethylamino)-4- EPA Active 2.88- 0.793406607 0.01-15uM methylcoumarin 21.5* (91-44-1) Disodium 4,4'-bis(2- EPA Active 18 0.393480513 0.09-120uM sulfostyryl) biphenyl (27344-41-8) 4-Aminofolic acid EPA Active 11.8 33.73462212 0.09-120uM (Aminopterin) (54-62-6) Ergocalciferol EPA Active 4.28- 14.4456809 0.09-60uM (50-14-6) 9.02* Alpha-Terthiophene EPA Active 15.7 0.304788788 0.09-120uM (1081-34-1) Triamterene EPA Active 5.3-42.1* 3.067190665 0.09-40uM (396-01-0) Novaluron EPA Active 23.4 37.41297387 0.09-120uM (116714-46-6) 2,2'-methylenebis(6- SIGMA Active 52.6 36.76671263 0.09-120uM tert-butyl-4- ethylphenol) (88-24-4) 9 Aminoacridine SIGMA Active 4.25- 12.58629187 0.09-30uM monohydrochloride 11.1* (52417-22-8) 2,2'-methylenebis(6- SIGMA Active 32.1 9.322927048 0.09-40uM tert-butyl-4- methylphenol) (119-47-1)

Table 1 Continued

4,4'-butylidenebis(6- SIGMA Active 20.6 0.293734005 0.09-20uM tert-butyl-m-cresol (96-69-5) Tamoxifen citrate SIGMA Active 33-56* 3.848988065 0.09-20uM (54965-24-1) Methyl 3-amino-5,6- SIGMA Active 8.78 0.445692935 0.09-1.5uM dichloropyrazine-2- carboxylate (1458-18-0) 2,7 Naphthalene EPA Active 14.5 13.78819692 0.09-120uM disulfonic acid (312693-54-2) Cridanimod EPA Active 11.1 13.88549215 0.09-120uM (38609-97-1) 7 methyl benzo (a) SIGMA Active 2.54 10.62870485 0.09-120uM pyrene (63041-77-0) Benzenesulfonic acid SIGMA Active 9.9 0.771910035 0.09-120uM (98-11-3) Falnidamol SIGMA Inactive 3.3 NA 0.09-0.9uM dihydrochloride (1216920-18-1) Lithocholic acid SIGMA Active 5.88- 22.2952456 0.09-50uM (434-13-9) 6.69* Calcipotriol SIGMA Active 0.000294 0.009974181 0.39nM- (112965-21-6) 0.05uM Lanoconazole SIGMA Active 21.1 20.99461043 0.09-120uM (101530-10-3)

Compound Name/ Origin Activity Tox21 Derived AC50 Concentration CAS # AC50 range (uM) Antagonists Dibutyltin dichloride EPA Active 0.0823 0.54221949 0-1uM (683-18-1) Triphenyltin hydroxide EPA Active 0.0929 0.045351184 0-1uM (76-87-9)

Table 1 Continued

Ziram EPA Active 1.39 1.00109703 0-6uM (137-30-4) Fluorescein sodium EPA Inactive 0.356 NA 0-12uM (518-47-8) Cadmium chloride EPA Active 0.171 0.379651184 0-12uM (10108-64-2) Potassium EPA Active 0.0912 0.098536754 0-6uM dicyanoaurate (13967-50-5) Cadmium dinitrate EPA Active 0.167 1.338424806 0-12uM (10325-94-7) Tributyltin chloride SIGMA Active 0.9594 0.200511605 0-1uM (1461-22-9) Thiram SIGMA Active 1.31-0.3 0.428434008 0-12uM (137-26-8) Aristolochic acid EPA Active 7.32 0.826662282 0-12uM (10190-99-5) Proflavin EPA Active 3.41 0.55539424 0-12uM hydrochloride (952-23-8) Tazobactam sodium EPA Active 1.98 0.283347893 0-12uM (89785-84-2) Carfizomib EPA Active 0.7481 0.537666094 0-12uM (868540-17-4) Phenylarsine oxide SIGMA Active 0.097498 0.012834274 0-0.37uM (637-03-6) Proscillaridin SIGMA Active 0.04335 1.897369724 0-12uM (466-06-8) Chlorambucil SIGMA Active 0.000058 0.021010799 0-0.37uM (305-03-3) Cadmium acetate SIGMA Active 0.271 1.166724126 0-12uM dihydrate (5743044) Cadmium reference SIGMA Active NA 0.198439348 0-2.75nM solution (7440-43-9) Dichlone SIGMA Active 0.512- 0.242849625 0-1.5uM (117-80-6) 0.418

Table 1 Continued

Menadiol SIGMA Active 0.979 3.496026921 0-12uM (481-85-6)

Table 2. Summary of Mammalian 2-hybrid (M2H) data across all compounds. Compounds exhibiting statistically significant recruitment or inhibition of coregulators (RXR)/coactivators (SRC-1) have values expressed as percentages. Note that in addition to most agonists, antagonists do facilitate recruitment (FR) or inhibit recruitment (IR) in the presence or absence of co-transfected coregulators/coactivators. For corepressor (NCoR) recruitment data was expressed as fold induction compared to DMSO. NS indicate non-significant induction/inhibition values for coregulator/coactivator/corepressor recruitment.

Agonists pMSRC-1 + pVP16 hVDR pMRXR + pVP16 hVDR pMNCoR + pVP16 hVDR Co Regulator +RXR -RXR +SRC-1 -SRC-1 +RXR -RXR addition Outcome Value Outcome Value Outcome Value Outcome Value Outcome Value Outcome Value Vitamin D3 FR 100 FR 100 FR 100 FR 100 DMSO DMSO Calcipotriol FR 135.9 FR 142.9 FR 133.2 FR 135.4 NS NS Lithocholic acid FR 102.6 FR 30.1 FR 63.5 FR 29 NS NS 7-(Dimethylamino)-4- NS NS NS NS NS NS methylcoumarin Disodium 4,4'-bis(2- NS NS NS NS NS NS sulfostyryl) biphenyl 4-Aminofolic acid NS NS NS NS NS NS Ergocalciferol FR 79.2 FR 97.7 FR 68.6 FR 56.8 NS NS Alpha-Terthiophene NS NS NS FR 3.5 NS NS Triamterene FR 63.7 FR 0.9 FR 3.7 FR 8.4 NS NS Novaluron NS FR 0.9 NS NS NS NS 2,2'-methylenebis(6- FR 8.9 FR 7.2 NS FR 3.1 NS NS tert-butyl-4- ethylphenol) 9 Aminoacridine FR 9.5 FR 14.4 FR 19.9 FR 12.3 FR 53.1 FR 67.7 monohydrochloride 2,2'-methylenebis(6- FR 22.2 FR 8.2 FR 100.7 FR 79.6 NS NS tert-butyl-4- methylphenol) 4,4'-butylidenebis(6- FR 0.91 FR 3.6 NS FR 96.8 NS NS tert-butyl-m-cresol Tamoxifen citrate FR 4.6 FR 6.1 FR 19.4 FR 46.6 NS NS

Table 2 Continued

Methyl 3-amino-5,6- FR 0.26 NS FR 4.2 FR 4.4 NS NS dichloropyrazine-2- carboxylate 2,7 Naphthalene NS NS NS NS NS NS disulfonic acid Cridanimod NS NS NS NS NS NS 7 methyl benzo (a) NS NS FR 1.5 NS NS NS pyrene Benzenesulfonic acid NS NS NS NS NS NS Falnidamol NS NS FR 3.7 FR 5.3 NS NS dihydrochloride Lanoconazole NS NS FR 4.3 NS NS NS Antagonists pMSRC-1NR + pVP16 hVDR pMRXR + pVP16 hVDR pMNCoR + pVP16 hVDR Co Regulator + RXR - RXR +SRC-1 -SRC-1 + RXR - RXR addition Outcome Value Outcome Value Outcome Value Outcome Value Outcome Value Outcome Value Vitamin D3 FR 100 FR 100 FR 100 FR 100 DMSO DMSO Fluorescein sodium NS NS NS NS FR 1.1 FR 1.2 Cadmium chloride FR 241.9 FR 258 FR 292 FR 175 FR 0.5 FR 1.1 Tributyltin chloride IR 24.9 NS IR 16.7 IR 0.5 FR 1.8 FR 2.2 Thiram NS NS NS NS FR 1 FR 1.7 Aristolochic acid IR 47.5 FR 347.5 FR 146.9 FR 402.5 FR 11.4 FR 25.8 Proflavin NS NS NS NS FR 0.8 FR 1.3 hydrochloride Tazobactam sodium NS NS NS NS FR 0.8 FR 1.7 Carfizomib FR 296 NS FR 2663.1 FR 1908.1 FR 0.8 FR 1.1 Phenylarsine oxide NS IR 45.4 NS IR 22.8 FR 1.5 FR 1.2 Proscillaridin IR 25.3 NS FR 190 NS FR 1.6 FR 2.8 Chlorambucil NS NS NS NS FR 1 FR 0.9 Cadmium acetate IR 46.7 IR 51.7 FR 165.6 NS FR 0.5 FR 0.5 dihydrate Cadmium reference IR 0.9 IR 1.3 IR 16 IR 23.7 FR 0.8 FR 0.7 solution Dichlone NS NS NS NS FR 1 FR 1.1 Menadiol NS NS NS IR 57.2 FR 1.1 FR 0.9 Potassium IR 30.9 NS IR 57 IR 14.4 FR 0.7 FR 1.7 dicyanoaurate Cadmium dinitrate FR 168.1 FR 233 FR 270 FR 283 FR 0.5 FR 0.6 Dibutyltin dichloride IR 37.5 IR 20.2 IR 16.7 IR 3.84 FR 0.9 FR 1 Triphenyltin IR 30 IR 55.8 IR 1.8 IR 0.5 FR 1.1 FR 1.5 hydroxide Ziram IR 48.3 FR 189 IR 45.1 IR 25.1 FR 0.4 FR 0.9

CHAPTER TWO

A Novel Role of Trialkyltins to promote conditional permissive transactivation of human VDR

Debabrata Mahapatra1 and Seth W. Kullman1,2

Comparative Biomedical Sciences, College of Veterinary Medicine, North Carolina State

University, Raleigh, North Carolina, United States of America; 2Department of Biological

Sciences, North Carolina State University, Raleigh, North Carolina, United States of America,

3Program in Environmental and Molecular Toxicology, North Carolina State University,

Raleigh, North Carolina, United States of America.

ABSTRACT

Vitamin D receptor (VDR) and its heterodimerization partner RXR (retinoid X receptor) engage in a canonical interaction when bound to the ligand 1,25 dihydroxy vitamin D3. This ligand-mediated interaction facilitates both transactivation and transrepression of a highly selective gene regulatory network. The RXR:VDR interaction is thought to be non-permissive, that is ligands to RXR do not transactivate the receptor heterodimer. This is in contrast to other RXR:NR partnerships such as RXR:PPAR where potent RXR ligands such as organotins can active the receptor heterodimer. Here we report a novel role of organotins wherein both TBT and TPT act as potent VDR antagonists at higher µM concentrations and as VDR synergist at lower concentrations only in the presence of vitamin D3. This biphasic activity is observed both in transient transactivation of

VDR in Cos7 cells and endogenous gene expression of CYP24A1 in HL60 cells. Using functional RXR mutants and potent RXR inhibitors we demonstrate that the synergistic actions of triorganic tins are mediated through RXR. Similarly, the inhibitory actions were also observed to have RXR involvement in the mammalian two-hybrid assay. We propose that trialkyltins modify RXR:VDR interactions in a manner that is conditionally permissive and suggest that this previously unrecognized synergistic role of organotins in altering VDR transcription is likely to have important physiological implications.

INTRODUCTION

Vitamin D has gained much attention in recent years not only for its role in classical bone and mineral homeostatic functions but also for its roles in neurodevelopment, neuroprotection, cell proliferation and differentiation, immune function and inflammation.

Classical transcriptional actions of 1,25D3 are mediated through its high affinity interactions with the vitamin D receptor (VDR). VDR is a member of the nuclear receptor superfamily, which is comprised of a large group of ligand-activated transcription factors. The mechanism of VDR-mediated gene transcription closely resembles that of other steroid hormones involving high affinity interaction between ligand and receptor, heterodimerization with RXR, association with a canonical vitamin D response element

(VDRE) within target promoter regions and recruitment of co-regulatory proteins, members of the Mediator of RNA polymerase II transcription subunit 1 (MED1) complex and RNA polymerase II to initiate both transactivation and transrepression of gene regulatory networks critical to cellular processes1,2 .

The Retinoic Acid Receptor (RXR) functions as an obligate heterodimerization partner for many type II NRs and NR:RXR dimers exhibit multiple functional modalities in relation to ligand‐activated transcription. Permissive heterodimers (RXR:PPAR, RXR:FXR,

RXR:LXR, and RXR:NGFI‐B) can be transactivated in the solitary presence of either RXR ligands or ligands for the heterodimeric partner. In this modality allosteric interaction between either receptor ligand is sufficient to initiate coactivator recruitment and receptor

mediated transcription. Conversely, non‐permissive RXR heterodimers such as those observed with RXR:TR and RXR:RAR heterodimers may only be transactivated through interaction with the NR (partner) ligand. With this modality, formation of the heterodimer precludes interaction of RXR ligands within the RXR LBD, leading to a lack of ability to include conformational activation of the receptor complex. RXR is thus considered a “silent heterodimerization partner.” In some instance, such as the case with RXR:RAR or RXR:LXR, heterodimers may exhibit synergistic increases in receptor transactivation if ligands for both receptors are present3,4,5. Modeling of this interaction suggests that binding of the partner

NR ligand results in allosteric change in the receptor complex that enables subsequent binding of the RXR ligand. In combination, the presence of both ligands enhances the transcriptional potential of the heterodimer. This type of interaction has been termed conditionally permissive heterodimers3.

Historically the RXR:VDR partnership was thought to be non-permissive where addition of 9-cis RA or other strong retinoids to the VDR:RXR heterodimer alone failed to transactivate VDR6. Through functional mutation studies it was determined that within the

RXR:VDR partnership, VDR ligands were sufficient to allosterically modify RXR from the apo to holo confirmation inducing a “phantom ligand effect” that enhances RXR:VDR functionalities including DNA binding, recruitment of NR coregulators and transactivation.

These observations suggest that RXR does indeed act as a major contributor to ligand activated VDR transactivation7. Further in 2006, Sanchez-Martinez8 reported that RXR

ligands can facilitate recruitment of essential RXR:VDR coregulators and potentiate vitamin

D dependent transactivation. These observations suggest that RXR ligands can facilitate allosteric interactions of the RXR:VDR heterodimers that results in a conditional permissive activation of the receptor complex8.

More recently additional RXR ligands have been identified and demonstrated to facilitate permissive and potentially conditional RXR:NR interactions. One such class of compounds is the trisubstituted organotins (i.e. trialkyltins). Tributyltin (TBT) and

Triphenyltin (TPT) are organic trialkyltins that are now considered persistent environmental contaminants of human and animal health concern. These organotins are widely used still in many parts of the world because of their potent biocidal activities against a variety of invertebrate pests such as helminthes, nematodes, arthropods and microorganisms such as fungi9. Bioaccumulation in seafood and shellfish and their subsequent intake by humans allows for a major source of exposure by ingestion. Adverse effects are not uncommon and include immunotoxicity in rodents with a dose dependent loss of thymic weight10. The discovery that these compounds could cause imposex (development of male sex organs in females) in molluscs at very low concentrations qualified them as potential endocrine disruptors11,12, which resulted in renewed scientific interest towards exploring endocrine disrupting effects in higher mammals. In vitro trialkyltins function as potent agonists for both PPAR and RXR. They are demonstrated to promote adipogenesis in preadipocyte murine 3T3-L1 cells and induce adipocyte differentiation in both mouse and human

multipotent stem cells. Conversely, triakyltins suppress osteogenesis in bone marrow multipotent mesenchymal stromal cells13. In vivo trialkyltins are capable of increasing adipose tissue mass in mice14,15 and dysregulating aromatase activity in adipose tissue resulting in diminished estradiol levels and downregulation of ER target genes16. While triakyltins are demonstrated ligands for PPAR, select studies have identified that both TBT and TPT as strong RXR agonists that bind directly with the RXR LBD17. Functional receptor assessment demonstrate TBT and TPT bind RXR and facilitate permissive RXR:PPAR interactions through enhancing DNA binding and recruitment of essential NR coregulators.

The underlying mechanism involves trialkyltin-induced allosteric receptor modifications through the formation of a unique covalent bond with the Cys-432 residue in helix 1117.

Similarly, TBT facilitates permissive transactivation of the RXR:LXR partnership promoting transactivation of LXR regulated genes and proteins associated with cellular cholesterol efflux18.

The observation that trialkyltin’s function as potent RXR agonists facilitating permissive RXR:NR interactions raises the question if these environmental contaminants could further facilitate conditional activation of the vitamin D receptor. Here, we investigated the hypothesis that trialkyltins TBT and TPT can promote synergistic transactivation of human VDR in the presence of 1,25(OH)2D3. Our results demonstrate that trialkyltins alone do not transactivate RXR:VDR heterodimers in vitro. Rather the RXR:VDR heterodimer exhibits a biphasic response to Triakyltins and vitamin D with a synergist interaction occurring with the combination of 0.37uM-3nM trialkyltins and vitamin D and a

potent inhibitory transactivational response occurring at > 400nM of TBT or TPT and 3nM vitamin D. We propose a mechanism where trialkyltins function through RXR to modulate

VDR transcriptional activity.

MATERIALS AND METHODS

Compound acquisition

Trialkyltins and Vitamin D2 were purchased from (Sigma-Aldrich Corp., St. Louis, MO,

USA). All compounds were serially diluted in DMSO with a final well concentration ranging from 0.01uM to 1uM.

Plasmid DNA constructs

The pSG5-Human VDR construct was originally a gift from Dr. John Moore

(GlaxoSmithKline, Research Triangle Park, NC). All human coregulator transient transactivation and mammalian 2-hybrid constructs were a gift from Dr. Donald McDonnell

(Duke University, Durham, NC). The Zf VDR, XREM luciferase reporter, 5XGal4-TATA- Luc mammalian 2-hybrid luciferase reporters, and the pRL-CMV (Renilla luciferase) internal luciferase control were obtained as described previously (Howarth et al 2008 and Kollitz et al 2014).

Cell culture

Cell culture media and other necessary reagents were obtained from Life technologies (Carlsbad, CA). Cos 7 cells were cultured in Dubelco’s Modified Eagle Medium

(DMEM) containing 10% FBS, 1 mM sodium pyruvate, 0.1 mM NEAA, 100 U/ml penicillin, and 100mg/ ml streptomycin. Human promyelocytic leukemic HL60 cells were grown in

RPMI suspension media containing 15% FBS and 200nM L-glutamine. Cells were maintained in a humidified incubator at 37°C with 5% CO2 and passaged when ~70- 80% confluent.

Transient transactivation assay

Full-length VDR constructs were tested in transient transactivation assays with 1α,

25-dihydroxyvitamin D3 (1, 25D3) (EMD Millipore, Billerica, MA) as the positive control or

Ergocalciferol (vitamin D2). Experiments were conducted using pSG5hVDR or Zf VDR, full length wild type human RXR, p/160 wild type SRC-1, pRLCMV, and XREM-Luc expression vectors as previously described by Howarth et al (2008). Cos 7 cells were seeded in 96-well plates at 2.5 x 104 cells per well 24 hours prior to transfection. Cells were transfected at 90-

95% confluency using Lipofectamine 2000 (Life Technologies, Grand Island, NY) with DNA diluted in Opti-MEM I Reduced Serum Medium as per the manufacture’s recommendations.

For functional comparisons, 89.7 ng of each pSG5-VDR construct was transiently transfected into Cos7 cells with 19.2 ng XREM-Luc, 18.3ng both of RXR and SRC-1, 4.5ng of Renilla luciferase, which serves as an internal luciferase control (Promega Corporation, Madison,

WI). Twenty-four hours post-exposure the Dual-Glo Luciferase Assay System (Promega

Corporation, Madison, WI) was used to passively lyse the cells and test for luciferase activity following the manufacturer’s protocols. Luciferase activities were measured using a Fluostar

Omega Luminometer (Cary NC, USA). Control reactions included empty pSG5 vector and ethanol as a vehicle control. Luciferase readings were normalized to the internal Renilla control, and VDR response was normalized to an empty vector control.

Mammalian 2-hybrid assays

Protein-protein interactions between VDR and its heterodimer partner RXR and members of SRC/p160 family of nuclear receptor coactivators were assessed using a mammalian 2-hybrid system (Clontech, Mountain View, CA). Assays were conducted with chimeric VDRs containing the herpes simplex VP16 activation domain fused to full length human VDR as prey (pVP16-hVDR). NR co-regulators were used as bait for each reaction, and consisted of fusion proteins containing; a complete NR Box of the SRC-1 (pM-SRC1 aa

241-386), or full-length hRXR fused to the yeast Gal4 DNA-binding domain. Assays were conducted in Cos7 cells seeded into 96 well plates twenty-four hours pre-transfection as described above. Cells were transfected with 33.6 ng pVP16-VDR, 33.6ng pM-coregulator,

126.6ng 5XGal4-TATA-Luc reporter (containing response elements for the yeast Gal4 DNA binding domain), and 3ng Renilla using Lipofectamine 2000 as described above. Controls consisted of transfections containing empty pM, pVP16 or both empty pM and pVP16 vectors. For both assays, experiments were replicated three times in groups of 3 technical replicate wells. One-way ANOVA followed by Tukey’s HSD post hoc tests, sigmoidal dose- response calculation with variable slopes followed by nonlinear regression analysis were run in GraphPad Prism version 7 (GraphPad Software, La Jolla, CA).

Cell viability assay

Cos7 and HL-60 cells were seeded in 96 well plates in triplicate at a density of 25,000 cells/well, transfected and dosed with select concentrations of the test compounds and

0 incubated for 18 hours at 37 C/5% CO2. Triton X (0.1%), DMSO (0.1%), and untreated wells served as controls. Resazurin at 1X concentration (20ul) /well was added to the cells and incubated at 370C for 2 hours in the dark. The amount of resorufin produced proportional to the number of viable cells was quantified by using a microplate reader (described earlier) equipped with a 560nm excitation / 590nm emission filter set. Only concentrations that yielded more than 90% viable cells were selected for in-vitro assays.

Real-time PCR

Total RNA was isolated from treated HL60 cells using the Zymo RNA Isolation kit

(Zymo Research Corp, CA USA) and reverse transcribed using a High Capacity cDNA Archive

Kit (Applied Biosystems, Foster City, CA) following the manufacturers’ instructions. CYP24A1 mRNA expression was normalized against that of housekeeping gene GAPDH. Real-time PCR assays were performed in 96-well optical plates on an ABI Prism 7300 Sequence Detection

System with SYBR Green PCR Master Mix. Primers used for GAPDH mRNA expression were designed as follows: [GAPDH-F:5’- CGACCACTTTGTCAAGCTCA- 3’ GAPDH-R: 5’-

GAGGGTCTCTCTCTTCCTCT-3’], while those for CYP24A1[CYP24A1-F:5’-

TGAACGTTGGCTTCAGGAGAA-3’,CYP24A1-R:5’-AGGGTGCCTGAGTGTAGCATCT-3 ] were adopted from Yosuke et al (2009). Fold gene induction following treatments were calculated based on the equation: Fold change=2ΔΔCt, where ΔCt represents the differences in cycle threshold numbers between CYP24A1 and GAPDH, and ΔΔCt represents the relative change in these differences between control and treatment groups21. Values

100

were plotted as a percentage and compared to the percentage induction of vitamin D322 and plotted on Graphpad prism.

101

RESULTS

Trialkytins result in a biphasic dose response curves

Activities of TBT and TPT to modulate VDR transactivation was assessed in transient transactivation assays utilizing a XREM- luciferase reporter construct19,20. Titration of TBT (0-

1M) or TPT (0-1M) in the absence of 1,25(OH)2D3 was insufficient for induction of VDR transactivation. Conversely, both TBT and TPT in the presence of 3nM 1,25(OH)2D3 exhibited an atypical biphasic transactivational response in the reporter system (Fig 1). At nanomolar concentrations there was a significant synergistic induction that peaked between 0.37-0.4M followed by a dose dependent inhibition culminating at 1 M. In case of TBT, synergistic activity started as low as 25nM at which point there was 4 fold induction over baseline. This activity increased progressively and peaked at 400nM with a 12-fold induction in transactivation followed by a sharp dose dependent decline in receptor activity.

TPT exhibited a gradual synergy with a peak reaching twice that of the baseline at 400nM followed by a dose dependent decline. This atypical biphasic activity was only observed in the presence of vitamin D3. Altering the concentration of vitamin D3 did not affect the degree of synergism exhibited by TBT and TPT (Fig 9).

Next, the ability of trialkyltins to mimic the biphasic activity was tested in vitro using

HL60 cells. Here, human CYP24A1 gene expression that is readily induced by vitamin D3 was utilized as an endogenous marker of VDR transactivation. Titration of TBT between 0.01-

0.8M and TPT between 0.01-0.8M exhibited a concentration related biphasic gene

102

expression as observed in transient transfection experiments (Fig 2). In the case of TBT, significant synergistic effects were observed at a low dose of 10nM with a concentration dependent increase culminating at 100nM. Concentrations greater than 100nM (200-

800nM) resulted in a concentration dependent inhibition of vitamin D induced CYP24 induction with negligible CYP24A1 activity occurring at 800nM. Comparatively, synergistic induction with TPT was observed between 50nM and 200nM and inhibitory activities were observed at concentrations greater than 600nM suggesting the efficacy of TPT was significantly less than that observed with TBT. Observed inhibitory effects of TBT and TPT were not due to cytotoxicity as all assays (transient transfection and CYP24 induction in

HL60 cells) were conducted at concentrations that maintained greater than 80% cell viability. Conversely, dibutyltin (DBT) exhibited only an inhibitory dose response but no synergism with vitamin D3 (Fig 7). Interestingly, however, 9-cis-retinoicacid (9cisRA), a potent RXR agonist was able to transactivate VDR at very high concentrations and synergize with vitamin D3 at low concentrations (Fig 8).

Synergistic activity in zebrafish vdr and vitamin D3 analogue

To assess if the observed synergistic activities of TBT and TPT were specific to human

VDR, we assessed if trialkyltins could modulate the transactivational function of full-length zebrafish VDR alpha as an evolutionarily distant VDR ortholog. Consistent with human VDR data, both compounds exhibited significant synergy with zebrafish VDR and the XREM reporter in the presence of 3nM 1,25(OH)2D3 and 0.37M TBT or TPT (Fig 3a). We also

103

investigated if the synergistic active trialkyltins was unique to the presence of 1,25(OH)2D3.

We thus tested the combination of ergocalciferol (Vitamin D2) in the presence of 3nM

1,25(OH)2D3 and 0.37M TBT or TPT and observed similar synergistic activities in transient transactivation assay in Cos7 cells (Fig 3b).

Trialkyltin synergism and RXR dependency

In order to verify the involvement of RXR in aiding synergistic activity of trialkyltins in the presence of vitamin D3, two approaches were investigated. First, a potent RXR inhibitor

LG101208 was added in conjunction with triakyltins and vitamin D3 to determine if synergist activities could be attenuated (See Fig 4a&b). Results for these studies demonstrate that the synergistic increases of 0.37 M and 3nM vitamin D could be significantly reduced to activities comparable to that of vitamin D alone if the reaction was co-incubated with 1.25 M of the RXR antagonist. Second, we utilized an RXR deletion mutant devoid of amino acids (382-403) in the C-terminal AF2 region which disrupts RXR interactions with VDR significantly modulating heterodimer transactivation (Fig 5 a, b).

While significant activity was observed with 3nM 1,25(OH)2D3 with wild type VDR and RXR, the RXR-AF2 mutant significantly attenuated 1,25(OH)2D3 transactivation. Addition of 0.37

M TBT or TPT with 3nM 1,25(OH)2D3 enhanced VDR activity approximate two-fold with the wild type RXR and this activity was greatly diminished with the RXR-AF2 mutant.

104

Trialkyltin inhibition and RXR dependency

Lastly to assess the role of RXR on VDR inhibition at higher concentrations of triakyltins we conducted mammalian two hybrid assays to assess if TBT and TPT could influence the recruitment of VDR:coregulator interactions. In our first assessment, we investigated if the inhibitory effect of triakyltins was mediated by RXR though investigating if VDR:SRC-1 interactions were dependent or independent on the presence of RXR. When mammalian two hybrid was conducted in the presence of co-incubated RXR, a concentration dependent decrease in 1,25(OH)2D3 induced VDR:SRC-1 recruitment was observed with both TBT and TPT (fig 6a, Fig 6c). When the same experiments were conducted in the absence of RXR, no attenuation in 1,25(OH)2D3 induced VDR:SRC-1 recruitment was observed (Fig 6b, Fig 6d). As was observed with transactivation data and

CYP24 induction in HL60 cells, TBT appears to be a potent inhibitor of VDR:SRC-1 interactions with lowest observable effect occurring at 50 nM while TPT lowest observable effect on VDR:SRC-1 interactions occurred at 230 nM.

105

DISCUSSION

Heterodimeric partnerships between members of the nuclear receptor families allow for the integration of distinct ligand activated signaling pathways and combinatorial regulation of a myriad of gene repertoires. These nuclear receptor assemblies have been well studied over the years and commonly involve RXR and a member of the NR superfamily. Based on the nature of ligand induced receptor activation, heterodimeric complexes formed with RXR can be categorized as “permissive” or “non-permissive”. While permissive heterodimers (RXR:PPAR, RXR:LXR, RXR:FXR) can become transcriptionally active either by a rexinoid (RXR ligand) or a partner ligand, non-permissive heterodimers

(RXR:RAR, RXR:VDR and RXR:TR) can only be activated by partner ligands and are unresponsive to rexinoids3,4,5. Non-permissive NR partners including RAR:RXR however when induced by certain retinoids in the presence of cognate NR ligands can lead to receptor superactivation termed conditionally permissive receptor activation23. This phenomenon is not restricted to RAR:RXR heterodimers alone but has been demonstrated in RXR:TR and RXR:VDR partnerships as well. In case of RXR:TR a unique cellular environment of CV-1 cells favoring over expression of coactivators allowed RXR stimulation by both 9- cis-RA and T3 resulting in a permissive interaction24. Similarly, for RXR:VDR association, Martinez8 and colleagues (2006) demonstrated that RXR ligand 9-cis-RA is capable of potentiating vitamin D dependent transcriptional response in colon cancer cells.

106

In all of the above-mentioned reports, RXR was demonstrated to play a pivotal role in facilitating conditional permissivity.

The well known promiscuous nature of RXR is not limited to its role as obligate heterodimer for NRs but also for its ability to bind a variety of ligands5,6. Of particular relevance with regards to our study is the ability of RXR to bind organometallic compounds.

Trialkyltin compounds have been reported to function as potent RXR:PPAR agonists causing adipocyte differentiation from precursor cells14. Also, they promote transcriptional activation and increased production of human chorionic gonadotrophin at very low concentrations through their high affinity binding to the ligand-binding domain of RXR homodimers16,25. Crystal structure of RXR bound to TBT reveals the presence of a covalent bond between the tin atom and the Cys 432 residue of RXRa on helix 11, which accounts for its high affinity binding, stabilization and activation of RXR receptor with respect to the

RXR: PPAR heterodimer17. This raised the question whether triorganic tins could possibly modulate RXR:VDR heterodimer. Earlier reports suggested that RXR acted as a silent partner in a non-permissive relationship with VDR26. However, later studies suggested that

1,25(OH)2D3 activated VDR is able to allosterically modify RXR from an apo to a holo receptor conformation in the absence of RXR ligand which allows RXR to actively recruit coactivators in a manner described as the “phantom ligand effect” 7. Recent investigation has further revealed that potent RXR agonists such as 9cis-RA are capable of inducing formation of a VDR:RXR heterodimers, synergizing vitamin D3 liganded VDR activities.

These results suggest that RXR also function both as a silent/or non-silent partner for VDR

107

heterodimerization promoting receptor transactivation7,8 . Taken together these reports suggest that the RXR:VDR heterodimeric partnership is more flexible and amenable to ligand induced conformational changes than previously thought.

Hence, based on earlier reports of trialkyltin induced nuclear receptor dysfunction we hypothesized that tributyltin chloride (TBT) and triphenyltin hydroxide (TPT) have the potential to modulate human RXR:VDR interactions. We report that both TBT and TPT are capable of modulating RXR: VDR transactivation in the presence of 1,25(OH)2D3 in a manner that is conditionally permissive. Furthermore, we demonstrate a previously unreported dual role of trialkyltins wherein they function as synergists at low nanomolar concentrations and antagonists at higher doses. The resulting dose response curve is nonmonotonic (biphasic) with an inverted U shape or  curve and typical of chemicals having endocrine disrupting properties27,28 (Fig 1). Similar biphasic dose response curves associated with bisphenol A –estrogen receptor (ER) interactions have been reported by Bouskine29 and colleagues (2009). Considering that organotins are known to act as potent EDCs by dysregulating the activities of nuclear hormone receptors including human estrogen receptor (ER) and also disrupting 11--hydroxysteroid dehydrogenase enzyme function leading to increased cortisol levels15,30,31 ,our findings further supports the role of organotins as endocrine disruptor for vitamin D endocrine system.

Synergy associated with low dose TBT and TPT in the presence of ligand activated

VDR was similar in pattern although with slightly dissimilar dose response curves likely because of differences in compound specific potencies (Fig 1). TBT being slightly more

108

potent exhibited synergistic activity at 25nM reaching a max of 3-fold activity at 0.4M over baseline 1,25(OH)2D3 activity whereas; TPT was only able to induce a maximum of two times baseline activity at the same concentration. Similar differences in potencies were reported with significant increases in aromatase activity at 10nM for TBT and in excess of

30nM for TPT 16. Synergistic activity was observed only in the presence of 1,25(OH)2D3 and did not differ with increasing concentrations of the VDR ligand (Fig 9). This suggested that although 1,25(OH)2D3 was crucial for the synergistic effect, its role was likely limited to activation of the VDR that allowed allosteric modification of RXR from an apo to holo- receptor conformation to promote binding of organotins7. Further evidence in support of this hypothesis was obtained when equivalent concentration of vitamin D2 (a potent vitamin D3 analogue) was used instead of 1,25(OH)2D3.

Although molecular events culminating in synergy conferred by trialkyltins in the presence of vitamin D3 are unclear, some mechanistic insight can be gained by examining conditional permissive characteristics of other non-permissive RXR:NR interactions. For instance, in case of RXR:RAR complex, RXR binding and recruitment of coactivators with apo-RAR is possible only when either corepressor binding is poor and easily dissociable or corepressor content of the cellular environment is relatively low thus promoting coactivator recruitment, binding and subsequent transactivation32. Our results indicate that trialkyltins do not promote VDR:SRC-1 coactivator recruitment alone. However, corepressor binding ability of tins or quantification of corepressor content in the cells was not investigated.

Interestingly, for RXR:TR interactions, synergistic activity for both partner ligands 9-cis-RA

109

and T3 was observed despite high levels of corepressors24. Again, in case of RXR:TR interactions, coactivator (TIF2) binding to heterodimer was synergistic in the presence of both ligands (9-cis-RA and T3) only when LxxLL box II was mutated suggesting the significance of other two NR boxes (I & III) in RXR-TIF2 associated synergistic interaction24.

Taken together, it appears that the actions of trialkyltins on nuclear receptors are rather unpredictable and quite variable depending on the cell type, interaction with coactivators, presence and or binding affinities with corepressors.

Our results indicate that both tin compounds show little or no induction in the absence of vitamin D despite being potent RXR agonists. This is in contrast to the effects of

9-cis-RA, another potent RXR agonist that have been reported to induce VDR alone, functioning as a permissive ligand and, in the presence of vitamin D3 promoting a synergistic response8. In fact, our data for 9-cis-RA is similar to the previous report except that it is capable of solely inducing VDR transactivation only at concentration greater than

100M (Fig 8). Although, the functional differences between triakyltins and 9-cis with respect to RXR:VDR heterodimers remains unclear, the manner in which both agonists occupy the ligand binding pocket (LBP) of RXR might offer some explanation. Whereas 9- cis-RA occupies the LBP of RXR completely by an induced fit mechanism brought about by conformational changes in the helices H3, H11, & H12 that are crucial for ligand binding,

TBT only partially occupies the LBP and interacts with only a subset of resides in the binding pocket of RXR17,33. Although, similar to 9cis-RA, TBT is able to restore conformation of a specific Cys 432 residue that is critical for full agonistic activity and stabilization of the active

110

form of RXR17 perhaps it is unable to make enough contacts required for activation of

RXR’s partner receptor VDR. In fact, it has been speculated that TBT might not effectively stabilize the active conformation of PPAR. This is because a critical residue Cys 285 of

PPAR that is analogous to Cys 432 of RXR is located on a different helix (H3) and out of reach of TBT17. This could be a reason further as to why TBT behaves as a very weak PPAR agonist.

Thereafter, the synergistic potential of dibutyltin (DBT) (Fig 7), which is a breakdown product of TBT was put to test and it was clear that DBT was unable to synergize induction of VDR in the presence of vitamin D3 confirming the limitation of such a response only to trialkyltins owing to their slightly bulkier alkyl residues. Organotins such DBT fail to establish sufficient contact points within the RXR-LBD to ensure high-affinity binding and correct positioning with respect to Cys-432 residue within the ligand binding pocket16,17.

Research has shown that the disruptive actions of trialkyltins on nuclear hormone receptors are mediated by its interaction with RXR. TBT mimics the induction of hypothyroidism caused by RXR agonist bexarotene in patients with cutaneous T cell lymphoma thus suggesting the involvement of RXR34. Also, Martinez8 et al (2006) reported similar findings suggesting the involvement of RXR. Our results are consistent with previous reports suggesting the involvement of RXR. We demonstrate the loss of RXR mediated synergistic activity induced by trialkyltins by using a functional mutant from of RXR having deletions in the AF2 region and an RXR antagonist LG101208 compound (Fig 4a&b).

Organic trialkyltins pesticides are well known for their biocidal and endocrine

111

disrupting properties mainly in marine invertebrates9,12. To test our earlier hypothesis of

RXR dependency and to investigate the potential effects in lower vertebrates an evolutionarily related but chronologically distant zebrafish VDR alpha construct was substituted for human VDR without altering experimental conditions. Both TBT and TPT were able to significantly synergize zebrafish VDR alpha transactivation in the presence of

3nM of vitamin D3 suggesting that the RXR mediated effects are valid for VDR ortholog as well. Moreover, it exemplifies the potential for endocrine disruptive effects of trialkytins in lower vertebrates.

Following investigation of synergistic effects of low concentration triakyltins, we next turned our attention to the involvement of RXR with respect to inhibition of RXR:VDR transactivation at higher concentrations of triakyltins. Also, whether the inhibitory actions defined by RXR would likely be manifested by a mechanism that is distinct from its synergistic actions observed at low doses. In fact, biphasic curves have been proposed to be a combination of responses arising from distinct pathways with opposing effects27. Evidence supporting RXR involvement in inhibition was obtained when we examined the role of trialkyltins in recruiting coactivator SRC-1 via the LxxLL motif. Both TBT and TPT appeared to favor recruitment of SRC-1 only in the absence of co-expressed RXRa. Conversely, a dose dependant decrease in SRC-1 recruitment was observed with increasing concentration of

TBT and TPT in the presence of RXR and liganded VDR suggesting an inhibitory effect of RXR

(See supplementary figure 4). Interestingly, the dose range used for our protein:protein recruitment assays were similar to our transactivation assay except that we did not observe

112

a synergistic response. Our findings for the mammalian hybrid assay are in line with those of

Cho31 et al (2012) where interaction between human ER LBD and coactivator SRC-1 was similarly inhibited by tin compounds. Also Cui18 and his colleagues (2011) reported that TBT was unable to induce coactivator association with LXR. However, our results differ from earlier studies that reported the ability of RXR liganded with TBT to actively recruit PPAR coactivator PGC1a17.

113

CONCLUSION

In conclusion, we demonstrate that RXR:VDR heterodimeric interaction is a potential target of disruption by trialkyltins. Both TBT and TPT in low concentrations are capable of modulating this heterodimer from a strictly non-permissive to a conditionally permissive state in the presence of vitamin D3 by interacting with RXR as a potent agonist. At higher concentrations trialkyltins cause inhibition of VDR transactivation that is RXR mediated as well thus generating a biphasic dose response curve typical of chemicals with endocrine disrupting properties. Because both VDR and RXR have ubiquitous tissue distribution, exposure to trialkyltins at low doses may result in either exaggerated or inhibiting hormonal signaling leading to adverse physiological implications spanning multiple systems in humans and lower vertebrates alike.

114

REFERENCES

1. Jurutka, P.W. et al. Molecular and functional comparison of 1,25-dihydroxyvitamin D(3) and the novel vitamin D receptor ligand, lithocholic acid, in activating transcription of cytochrome P450 3A4. J Cell Biochem 94, 917–43 (2005).

2. Delfosse, V., Marie, A., Balaguer, P., & Bourguet, W. A structural perspective on nuclear receptors as targets of environmental compounds. Acta Pharmacologica Sinica 36, 88-101 (2015).

3. Shulman, A. I. & Mangelsdorf, D. J. Retinoid X Receptor Heterodimers in the Metabolic Syndrome. N Engl J Med 3536, (2005).

4. Lefebvre, P., Benomar, Y. & Staels, B. Retinoid X receptors: Common heterodimerization partners with distinct functions. Trends in Endocrinology and Metabolism (2010). doi:10.1016/j.tem.2010.06.009

5. Pérez, E., Bourguet, W., Gronemeyer, H. & De Lera, A. R. Modulation of RXR function through ligand design. Biochimica et Biophysica Acta - Molecular and Cell Biology of Lipids (2012). doi:10.1016/j.bbalip.2011.04.003

6. Aranda, A. & Pascual, A. Nuclear Hormone Receptors and Gene Expression. Physiol Rev 81, 1269–1304 (2001).

7. Bettoun, D. J. et al. Retinoid X receptor is a nonsilent major contributor to vitamin D receptor-mediated transcriptional activation. Mol. Endocrinol. 17, 2320–2328 (2003).

8. Sánchez-Martínez, R., Castillo, A. I., Steinmeyer, A. & Aranda, A. The retinoid X receptor ligand restores defective signalling by the vitamin D receptor. EMBO Rep. 7, 1030–1034 (2006).

9. Piver WT. Organotin compounds: industrial applications and biological investigation. Environmental Health Perspectives. 4:61–79 (1973).

10. N. J. Soneij, A. A. van Iersel, A. H. Penninks, W. Seinen, Toxicol. Appl. Pharmac. 81, 274 (1985).

11. Smith, B.S. Male characteristics on female mud snails caused by antifouling bottom paints. Journal of Applied Toxicology 1: 22–25 (1981).

115

12. Bettin, C., Oehlmann, J. & Stroben, E. TBT-induced imposex in marine neogastropods is mediated by an increasing androgen level. Helgolander Meeresuntersuchungen 50: 299-317 (1996).

13. Baker, A., H. et al. Tributyltin engages multiple nuclear receptor pathways and suppresses osteogenesis in bone marrow multipotent stromal cells. Chem Res Toxicol 28:6 1156-1166 (2015).

14. Kanayama, T., Kobayashi, N., Mamiya, S., Nakanishi, T. & Nishikawa, J. Organotin compounds promote adipocyte differentiation as agonists of the peroxisome proliferator-activated receptor gamma/retinoid X receptor pathway. Mol. Pharmacol. 67, 766–774 (2005).

15. Grun, F. & Blumberg, B. Environmental obesogens: Organotins and endocrine disruption via nuclear receptor signaling. Endocrinology 147, 50–55 (2006).

16. Nakanishi, T. et al. Trialkyltin Compounds Bind Retinoid X Receptor to Alter Human Placental Endocrine Functions. doi:10.1210/me.2004-0397

17. Le Maire, A. et al. Activation of RXR–PPAR heterodimers by organotin environmental endocrine disruptors. EMBO Rep. 10, 367–373 (2009).

18. Cui, H., et al. Tributyltin chloride induces ABCA1 expression and apolipoprotein A-I- mediated cellular cholesterol efflux by activating LXRalpha/ RXR. Biochem. Pharmacol. 81, 819–824 (2011).

19. Howarth, D.L. et al. Paralogous vitamin D receptors in teleosts: Transition of nuclear receptor function. Endocrinology 149, 2411–2422 (2008).

20. Kollitz EM, Hawkins MB, Whitfield GK, & Kullman SW. Functional diversification of vitamin D receptor paralogs in teleost fish after a whole genome duplication event. Endocrinology 155, 4641-54 (2014).

21. Pfaffi, M.W. A new mathematical model for relative quantification in real--time RT- PCR. Nuclei Acids Res 29, 2002-7 (2001).

22. Sidhu, P.S., et al. Development of novel Vitamin D receptor coactivator inhibitors. ACS Med Chem Letters 5, 199-204 (2014).

116

23. Széles, L. et al. Research Resource: Transcriptome Profiling of Genes Regulated by RXR and Its Permissive and Nonpermissive Partners in Differentiating Monocyte- Derived Dendritic Cells. Mol. Endocrinol. 24, 2218–2231 (2010).

24. Castillo, A. I. et al. A permissive retinoid X receptor/thyroid hormone receptor heterodimer allows stimulation of prolactin gene transcription by thyroid hormone and 9cRA. Mol. Cell. Biol. 24, 502–513 (2004).

25. Brtko, J and Dvorak, J. Triorganotin compounds-ligands for rexinoid inducible trnscription factors: biological effects. Toxicol Letters 234, 50-58 (2015).

26. Forman, B., M., Umesono, K., Chen, J., &Evans, R., M. Unique response pathways are established by allosteric interactions among nuclear hormone receptors. Cell 81: 541–550 (1995).

27. Schug, T. T., Janesick, A., Blumberg, B. & Heindel, J. J. Endocrine disrupting chemicals and disease susceptibility. J. Steroid Biochem. Mol. Biol. 127, 204–215 (2011).

28. Conolly, R. B. & Lutz, W. K. Nonmonotonic dose-response relationships: Mechanistic basis, kinetic modeling, and implications for risk assessment. Toxicol. Sci. 77, 151– 157 (2004).

29. Bouskine, A., Nebout, M., Brücker-Davis, F., Banahmed, M. & Fenichel, P. Low doses of bisphenol A promote human seminoma cell proliferation by activating PKA and PKG via a membrane G-protein-coupled estrogen receptor. Environ. Health Perspect. 117, 1053–1058 (2009).

30. Atanasov, A. G., Nashev, L. G., Tam, S., Baker, M. E. & Odermatt, A. Organotins disrupt the 11??-hydroxysteroid dehydrogenase type 2-dependent local inactivation of glucocorticoids. Environ. Health Perspect. 113, 1600–1606 (2005).

31. Cho, E.-M. et al. Organotin Compounds Act as Inhibitor of Transcriptional Activation with Human Estrogen Receptor. J. Microbiol. Biotechnol 22, 378–384 (2012).

32. Germain, P., Iyer, J., Zechel, C. & Gronemeyer, H. Co-regulator recruitment and the mechanism of retinoic acid receptor synergy. Nature 415, 187–192 (2002).

33. Egea PF, Mitschler A, Rochel N, Ruff M, Chambon P, Moras D. Crystal structure of the human RXRa ligand-binding domain bound to its natural ligand: 9-cis retinoic acid. EMBO J 19: 2592–2601 (2000).

117

34. Adeeko, A. et al. Effects of in utero tributyltin chloride exposure in the rat on pregnancy outcome. Toxicol. Sci. 74, 407–415 (2003).

118

FIGURES

Figure 1. Dose response curves showing a biphasic human VDR response in transient transactivation assay. Both TBT and TPT show an initial induction at low concentration followed by inhibition at higher concentration in the presence of vitamin D3. Assays were run in Cos7 cells and data expressed as mean± SEM (n=3). Data are normalized to VDR + empty pSG5 vector.

119

Figure 2. Graph showing a biphasic human CYP24A1 endogenous gene expression in response to low and high doses of a) TBT and b) TPT in the presence of vitamin D3 (3nM).

120

Assays were run in HL60 cells and data expressed as mean± SEM (n=3). Data are normalized to DMSO. Asterisks indicate significance: *** = p < 0.001, ** = p < 0.01, * = p < 0.05.

121

Figure 3. Graph showing synergistic response by a) Zebrafish VDR alpha b) Vitamin

D2/Ergocalciferol (analogue of vitamin D3) following transactivation. Both TBT and TPT show synergy only in the presence of Vitamin D3 (3nM) with a peak at 0.37uM. No synergy is evident in the absence of Vitamin D3. Assays were run in Cos7 cells and data expressed as

122

mean± SEM (n=3). Data are normalized to VDR + empty pSG5 vector. Asterisks indicate significance: *** = p < 0.001, ** = p < 0.01, * = p < 0.05.

123

Figure 4. Graph showing the diminution of synergistic response in human VDR in the presence of RXRa inhibitor LG101208. Both a) TBT and b) TPT show synergy only in the presence of Vitamin D3 (3nM) with a peak at 0.37uM. No synergy is evident by compounds

124

themselves and or in the presence of the inhibitor. Assays were run in Cos7 cells and data expressed as mean± SEM (n=3). Data are normalized to VDR + empty pSG5 vector. Asterisks indicate significance: *** = p < 0.001, ** = p < 0.01, * = p < 0.05.

125

Figure 5. Graph showing diminution of synergistic response by human VDR when a RXRAF2 mutant is used instead of a full length RXRa. Both a) TBT and b) TPT show synergy only in the presence of Vitamin D3 (3nM) with a peak at 0.37uM. No synergy is evident in the

126

absence of vitamin D3. Assays were run in Cos7 cells and data expressed as mean± SEM

(n=3). Data are normalized to VDR + empty pM vector. Asterisks indicate significance: *** = p < 0.001, ** = p < 0.01, * = p < 0.05.

127

Figure 6. Graph showing recruitment of coregulators and coactivators by VDR in mammalian 2-hybrid assay in the presence of trialkyltin compounds. Recruitment of SRC-1 by VDR is repressed by TBT a) only in the presence of RXR and b) not in the absence.

Similarly, recruitment of SRC-1 by VDR is repressed by TPT c) only in the presence of RXR and d) not in the absence. Assays were run in Cos7 cells and data expressed as mean± SEM

(n=3). Data are normalized to VDR + empty pM vector. Asterisks indicate significance: *** = p < 0.001.

128

a) Recruitment of SRC-1 with RXR

b) Recruitment of SRC-1 without RXR

129

c) Recruitment of SRC-1 with RXR

d) Recruitment of SRC-1 without RXR

130

Figure 7. Dose response curve showing a dose dependant inhibition of VDR by DBT in transient transactivation assay.

Figure 8. Dose response curves showing 9cis RA is capable of inducing VDR in transactivation assay alone and in the presence of vitamin D3.

131

Figure 9. Dose response curves showing synergistic human VDR response in transient transactivation assay. Both a) TBT and b) TPT show synergy only in the presence of increasing concentrations of Vitamin D3 (6nM &12nM) with a peak at 0.37uM. No synergy is evident in the absence of vitamin D3.

132

Figure 10. Graph showing recruitment of coregulators and coactivators by VDR in mammalian 2-hybrid assay in the presence of trialkyltin compounds. Recruitment of RXR by vdr is repressed by TBT both a) in the presence of SRC-1 and b) in the absence. Similarly, recruitment of RXR by VDR is repressed by TPT both c) in the presence of SRC-1 and d) in the absence. Assays were run in Cos7 cells and data expressed as mean± SEM (n=3). Data are normalized to VDR + empty pM vector. Asterisks indicate significance: *** = p < 0.001.

133

a) Recruitment of RXR with SRC-1

b) Recruitment of RXR without SRC-1

134

c) Recruitment of RXR with SRC-1

d) Recruitment of RXR without SRC-1

135

CHAPTER THREE

New insights into cadmium induced dysregulation of vitamin d receptor: Potential for

endocrine disruption

Debabrata Mahapatra1 Melaine Agnes Kuenemann2 Denis Fourches3and Seth W. Kullman1,4

1 Comparative Biomedical Sciences, College of Veterinary Medicine, North Carolina State

University, Raleigh, North Carolina, United States of America; Department of Chemistry,

Bioinformatics Research Center, North Carolina State University, Raleigh, North Carolina,

United States of America 2,3Department of Biological Sciences, North Carolina State

University, Raleigh, North Carolina, United States of America, 4Program in Environmental and Molecular Toxicology, North Carolina State University, Raleigh, North Carolina, United

States of America.

136

ABSTRACT

Cadmium is a highly toxic non-essential element and a persistent environmental contaminant, exposure to which causes adverse effects spanning multiple organs, systems and organisms. Given the known endocrine disrupting properties of Cd on a variety of transcription factors and nuclear receptors including Sp1, ER, GR, AR, PR we investigated the potential of Cd to modulate transcriptional functionalities of VDR. In this study we demonstrated a dose dependent inhibitory role of Cd on transient transactivation of human

VDR using two distinct cell based reporter systems. A similar dose dependent repression of endogenous CYP24A1 was observed in vitro when tested in HEPG2, SHSY5Y and HL60 cells.

In sharp contrast to transactivation results, mammalian two hybrid assays demonstrate that

Cd enhances both VDR:RXR and VDR:SCR-1 recruitment over that of vitamin D3 alone. This interaction was synergistic in nature and observed only in the presence of vitamin D3. Gel retardation studies suggest that the inhibitory actions of Cd are likely due to the disruption of zinc finger structure on the DNA binding domain of VDR where Cd replaces Zn and hinders DNA binding to cognate vitamin D response elements (VDREs). These finding demonstrate that Cd has a direct effect on VDR transactivation likely mediated through disruption of receptor:DNA interaction which may have significant physiological and pharmacological implications.

137

INTRODUCTION

Cadmium is a highly toxic environmental contaminant with grave human and environmental health concerns. Human exposure to Cd can occur through cigarette smoking, dietary sources and industrial contamination1,2. Increased body burden from chronic Cd exposure is deleterious mainly due to it’s long half-life and ability to bind metallothioneins3,4. Adverse health effects of Cd exposure affect multiple organs and systems and include the skeletal, renal, nervous, hematopoietic, immune, cardiovascular and reproductive systems4,5,6,7,8,9.

The ability of Cd to exert its effects on a variety of cell types could be attributed to its direct interactions with a myriad of transcription factors, modifying their functional activities or indirectly affecting molecules within select signaling transduction pathways10.

Cadmium is a known endocrine disrupting agent that targets a number of nuclear hormone receptors including the ER, AR and GR9,11,12. Perhaps the best studied of all is the effect of

Cd on the ER owing to its estrogen mimicking properties including the ability to promote breast cancer13. Cd specifically interacts with the hormone binding domain of ER to produce a high affinity complex capable of triggering estrogenic effects in breast cancer cells9,14. The ability of Cd to affect the AR receptor has also been demonstrated through activation of an AR-luciferase reporter system in mouse L cells and Cos-1 cells in addition to activating a chimeric GAL4-AR receptor containing the AR ligand-binding domain. Cd also inhibits binding of androgen to the AR receptor suggesting its inhibitory effect on hormone

138

binding11. One prospective target site for Cd interaction with nuclear receptors is the zinc finger domain present within the DNA binding domain (DBD) of nuclear receptors. Because of strikingly similar coordination chemistry with Zn2+, Cd2+ is thought to be able to replace zinc resulting in dysfunction receptor:DNA interactions15. This mechanism of Cd has been demonstrated for several zinc finger-containing proteins such nuclear hormone receptors

(ER, GR, VDR) and transcription factors including metallothioneins, Sp1, DNA repair helicase

BLM protein3,10,16. These reports highlight the sensitivity of zinc finger contain domains to

Cd and the potential importance of direct effects of Cd on nuclear receptors and transcription factor functions.

Cd is also able to modulate signal transduction pathways associated with the nuclear receptors such as the ER, GR, PR and the AR. Cd promotes ER translocation into the nucleus and favors receptor interaction with transcription factors such as c-jun thus enhancing recruitment of this complex to cyclin D1 and c-myc promoter regions17. Cd may also promote breast cancer progression by target the E-cadherin pathway, especially by enhancing nuclear translocation of B-catenin while upregulating c-jun and cyclin D1 oncoproteins18. These studies on ER also demonstrate the ability of Cd to induce carcinogenesis. In addition, while low dose Cd exposure reduces in-vitro gene transcription of the Glucocorticoid receptor, Cd exposure during pregnancy alters the expression profile of DNMT3a resulting in sex dependent changes in methylation and expression of GR110 promoter19 suggesting its effects on the epigenetic pathways. Cd also induces an increase in both progesterone receptor (PgR) mRNA and PgR protein levels20 whereas reducing the

139

concentration of AR protein and messenger RNA in human prostate cancer cells11. Overall these studies emphasize the ability of Cd to target and adversely affect different cell types, receptor types and various stages of the gene transcription machinery.

Vitamin D is essential for intestinal absorption and assimilation of minerals and elements vital to normal physiological functioning of the body. Vitamin d deficiency however impacts the uptake of essential minerals and trace elements and instead favors uptake of toxic elements including cadmium21. The effects of Cd exposure on vitamin D levels is well documented and reports suggest that Cd diminishes the activity of vitamin D by blocking the renal synthesis/hydroxylation of previtamin D to the active form i.e 1,25 dihydroxyvitamin D3 in humans and rodents alike22,23. This can have deleterious effects on bone from impairment of gut absorption of Ca and derangement of collagen metabolism leading to metabolic bone diseases such as osteomalacia and osteoporosis especially in women5. These defects can manifest clinically into the so-called Itai Itai disease reported in

Japan. Associated clinical deficits with this extreme condition include anemia from lack of erythropoietin synthesis from the kidneys and renal proximal tubular dysfunction with elevated serum PTH levels and phosphaturia from induction of FGF2324,25,26.

The fact that Cd functions as a potent endocrine disruptor for many nuclear hormone receptors prompted us to investigate whether it can directly modulate vitamin d receptor functionality. Hence, we investigated the role of Cd in disrupting the VDR transactivation in in-vitro assays. We demonstrated that Cd inhibits vitamin D3 induced transactivation of VDR in two distinct luciferase gene reporters and cell types. The inhibitory

140

actions of Cd exposure were also prevalent when gene expression of downstream target of liganded VDR (CYP24A1) was assessed in three cell types of different lineages. Additionally, mammalian 2-hybrid assays revealed novel and unique interactions where Cd exposure markedly enhanced both coregulator (RXR) and coactivator (SRC-1) recruitment by liganded

VDR suggesting two entirely different mechanisms of actions for Cd induced disruption of

VDR. We demonstrate that the inhibitory actions of Cd on VDR transactivation are likely due to disruption of zinc fingers on DBD of VDR resulting in inhibition of protein-DNA interaction. Mechanisms underlying the Cd induced protein:protein interactions are unclear and needs further investigation.

141

MATERIALS AND METHODS

Compound acquisition

Cadmium salts were purchased (Sigma-Aldrich Corp., St. Louis, MO, USA). All compounds were serially diluted in DMSO with a final well concentration ranging from

0.01uM to 12uM.

Cell culture

Cell culture media and other necessary reagents were obtained from Life technologies (Carlsbad, CA). Hek293T and HEPG2 cells were grown in Minimum Essential

Medium (MEM) containing 10% heat-inactivated Fetal Bovine Serum (FBS), 1 mM sodium pyruvate, 0.1 mM NEAA, 100 U/ml penicillin, and 100mg/ ml streptomycin. Cos 7 cells were cultured in Dubelco’s Modified Eagle Medium (DMEM) containing 10% FBS. Human promyelocytic leukemic HL60 cells were grown in RPMI suspension media containing 15%

FBS and 200nM L-glutamine. SHSY5Y cells were grown in DMEM/F12 media supplemented with 10% FBS and 200nM L-glutamine and maintained according to (Celli et al 1999). Cells were maintained in a humidified incubator at 37°C with 5% CO2 and passaged when ~70-

80% confluent.

Transient transactivation assay

Full-length VDR constructs were tested in transient transactivation assays with 1α,

142

25-dihydroxyvitamin D3 (1, 25D3) (EMD Millipore, Billerica, MA) as the positive control.

Experiments were conducted using pSG5hVDR, pRLCMV, and CYP24-Luc expression vectors as previously described in Howarth et al 2008. HEK293T cells were seeded in 96-well plates at 2.5 x 104 cells per well 24 hours prior to transfection. Cells were transfected at 90-95% confluency using Lipofectamine 2000 (Life Technologies, Grand Island, NY) with DNA diluted in Opti-MEM I Reduced Serum Medium as per the manufacture’s recommendations. For functional comparisons, 89.7 ng of each pSG5-VDR construct was transiently transfected into Hek293T cells with 19.2 ng CYP24-Luc and 4.5 ng of Renilla luciferase, which serves as an internal luciferase control (Promega Corporation, Madison, WI). Twenty-four hours post- exposure the Dual-Glo Luciferase Assay System (Promega Corporation, Madison, WI) was used to passively lyse the cells and test for luciferase activity following the manufacturer’s protocols. Luciferase activities were measured using a Wallac MicroBet TriLuc Luminometer

(Perkin Elmer Life Sciences, Waltham, MA). Control reactions included empty pSG5 vector and ethanol as a vehicle control. Luciferase readings were normalized to the internal Renilla control, and VDR response was normalized to an empty vector control.

Mammalian 2-hybrid assays

143

fused to full length human VDR as prey (pVP16-hVDR). NR co-regulators were used as bait for each reaction, and consisted of fusion proteins containing; a complete NR Box of the

SRC-1 (pM-SRC1 aa 241-386), or full-length hRXR’ fused to the yeast Gal4 DNA-binding domain. Assays were conducted in Cos7 cells seeded into 96 well plates twenty-four hours pre-transfection as described above. Cells were transfected with 33.6 ng pVP16-VDR, 33.6 ng pM-coregulator, 126.6 ng 5XGal4-TATA-Luc reporter (containing response elements for the yeast Gal4 DNA binding domain), and 3 ng Renilla using Lipofectamine 2000 as described above. Controls consisted of transfections containing empty pM, pVP16 or both empty pM and pVP16 vectors. For both assays, experiments were replicated three times in groups of 3 technical replicate wells. One-way ANOVAs followed by Tukey’s HSD post hoc tests, sigmoidal dose-response calculation with variable slopes followed by nonlinear regression analysis were run in GraphPad Prism version 7 (GraphPad Software, La Jolla, CA).

Note that all assays were conducted in either the presence or absence of co-transfected full length RXR or SRC-1 to assess if exogenous protein expression would further facilitate VDR co-regulator/co-activator interactions.

Cell viability assay

Healthy Hek293T, Cos7, HL 60, cells were seeded in 96 well plates in triplicate at a density of 25,000 cells/well, transfected and dosed with cadmium chloride (Cas# 101018-

64-2) or cadmium dinitrate (Cas# 10325-94-7) between 0-12µM and incubated for 18 hours

0 at 37 C/5% CO2. Triton X (0.1%), DMSO (0.1%), and untreated wells served as controls.

144

Resazurin at 1X concentration (20µl) /well was added to the cells and incubated at 370C for

2 hours in the dark. The amount of resorufin produced proportional to the number of viable cells was quantified by using a microplate reader (described earlier) equipped with a 560 nm excitation / 590 nm emission filter set. Only concentrations that yielded more than 80% viable cells were selected for in-vitro assays.

Construct design

For chimeric Gal4-VDR expression construct the XGalX plasmid vector containing the translation initiation sequence (amino acids 1–76) of the glucocorticoid receptor fused to the DNA binding domain (amino acids 1–147) of the yeast Gal4 transcription factor in the pSG5 expression vector (Stratagene, La Jolla, CA) was used. The DNA binding domain (LBD) and AF2 domain of the human VDR flanked by restriction sites was PCR amplified

(hVDRXGalX F: 5’–TCAGTACCGGCATGATGAAGGAGTTCATT-3’and hVDRXGalX R:5’-

GCAAGCTTGGATCCTCAGGAGATCTCATTGCCAAA -3’) and ligated into the KpnI and HindIII sites of the multiple cloning site within the XgalX vector generating GAL4-DBD-hVDR-

LBDAF2 plasmid.

Real-time PCR

Total RNA was isolated from treated HL60 cells using the Zymo RNA Isolation kit

(Zymo Research Corp, CA USA) and reverse transcribed using a High Capacity cDNA Archive

Kit (Applied Biosystems, Foster City, CA) following the manufacturers’ instructions. CYP24A1

145

mRNA expression was normalized against GAPDH as a housekeeping gene. Real-time PCR assays were performed in 96-well optical plates on an ABI Prism 7300 Sequence Detection

System with SYBR Green PCR Master Mix. Primers used for GAPDH mRNA expression were designed as follows: [GAPDH-F:5’- CGACCACTTTGTCAAGCTCA- 3’ GAPDH-R: 5’-

GAGGGTCTCTCTCTTCCTCT-3’], while those for CYP24A1 [CYP24A1-F:5’-

TGAACGTTGGCTTCAGGAGAA-3’,CYP24A1-R:5’-AGGGTGCCTGAGTGTAGCATCT-3 ] were adopted from Yosuke27 and colleagues (2009). Fold gene induction following treatments were calculated based on 2ΔΔCt, where ΔCt represents the differences in cycle threshold numbers between CYP24A1 and GAPDH, and ΔΔCt represents the relative change in these differences between control and treatment groups28. Values were plotted as a percentage and compared to the percentage induction of vitamin D329.

Electrophoretic Mobility Shift Assays

For nuclear protein preparation confluent cultures of HL60 cells were treated with

3μM and 6μM of both Cd chloride and Cd dinitrate along with 3nM of vitamin D3. Cells were harvested after 18 hrs, washed twice in buffered ice cold PBS and centrifuged for

14000g for 5min. The pellets were treated with protease inhibitor cocktail (ThermoTM Halt

Protease Inhibitor Cocktail kit) before lysis reagents were added for extraction of cytoplasmic and nuclear fractions following manufacturers protocol (NE-PER Nuclear kit,

Cat# 78833). Proteins concentrations were measured using Thermo Scientific TM PierceTM

BCA Protein Assay (Cat # 23225)(Pierce Biotechnology, Rockford IL USA). Recombinant wild

146

type VDR and RXR proteins were expressed and purified according to Kollitz et al. (2015) with modifications. EMSA were carried out in a total volume of 20 μl containing 1μg of both recombinant VDR and RXR proteins. Protein- DNA binding reactions were carried out using a canonical VDRE (5’ – AGC TTC AGG TCA AGG AGG TCA GAG AGC – 3’). Equal volumes of single stranded 5’-Cy5-labeled and unlabeled oligos (Integrated DNA Technologies

Coralville, IA) were resuspended in annealing buffer (10 mM Tris, 50 mM NaCl, 1 mM EDTA) to a final concentration of 250 μM. Samples were heated to 95°C for two minutes and allowed to cool overnight. Annealed oligos were further diluted to a final concentration of 1 pmol/μL. For the DNA-binding reactions, receptors and Cy5 labeled annealed oligos were incubated at 25° C for 3 hours in binding buffer (100 mM KCl, 10 mM HEPES, 1 mM EDTA,

0.1 mg/ml BSA, 4 μg/mL sonicated salmon sperm, 1.0 mM DTT, 1% glycerol) and 100 nM 1,

25D3 or ethanol as a vehicle control. Competition experiments were performed in the presence of a 100-fold molar excess of unlabeled wild type oligos. Protein–DNA complexes were resolved on a 6% non-denaturing acrylamide gel in ice-cold 0.5X TBE buffer (45 mM

Tris base, 45 mM boric acid, 1 mM EDTA, pH 8.0) at 100 volts for 70 minutes. Gels were visualized on Typhoon FLA 7000 using the red fluorescence mode (GE Healthcare Life

Sciences, Pittsburgh, PA).

147

RESULTS

Inhibition of VDR transactivation by Cd salts

Initial studies demonstrated the abilities of cadmium dichloride and cadmium dinitrate to inhibit transient VDR transactivation in-vitro employing two distinct reporter systems. Initial experiments were conducted using a combination of human VDR (psG5) with a XREM-Luc reporter construct in Cos 7 cells. Cells were additionally co-transfected with a pSG5 construct expressing full length human RXR as the VDR obligate heterodimerization partner and a pSG5 construct expressing full length SRC-1 as a VDR:RXR coactivator. Under these conditions exposure to 1,25OH2D3 (3nM) resulted in a consistent

30-35 fold induction of the XREM-Luc reporter. Comparatively, addition of increasing concentrations of either cadmium salt between 0-12µM resulted in a dose dependent attenuation in VDR/1,25OH2D3 mediated induction of XREM-Luc reporter (Fig 1a). To confirm these results a second transient transactivation system utilizing a CYP24-Luc reporter in HEK293 cells was examined. This system relied on endogenous RXR and coactivator expression. As observed with Cos7 cells, 1,25OH2D3 (3nM) resulted in a consistent 23-25 fold induction of the CYP24-Luc reporter and addition of cadmium salts resulted in a dose dependent attenuation in VDR/1,25OH2D3 mediated induction of CYP-Luc reporter (Fig 1b). Inhibition curves show a similar trend when either a XREM-Luc or a

CYP24-Luc reporter is used. However, in case of XREM-Luc, the inhibitory effect was initiated at a concentration of 0.18 µM whereas the inhibition of the CYP24-Luc reporter

148

became apparent at 1.5 µM Cd. As a follow-up to our hypothesis, we tested whether addition of Zn would alter the inhibitory actions of Cd. Interestingly, adding micromolar concentrations of ZnCl2 (1µM) directly to the media concurrent to the addition of Cd salts appeared to significantly rescue the inhibitory actions of Cd salts concentration was greater than 6µM (See Fig 2). Addition of ZnCl2 at this concentration alone did not enhance transactivation nor did it negatively affect the VDR transactivation in the presence of vitamin D3.

Cadmium salts modulate endogenous CYP24A1 gene expression

Based on the ability of cadmium salts to attenuate a transiently expressed exogenous VDR/CYP24-Luc system, we next examined the ability of Cd salts to inhibit expression of endogenous CYP24A1 gene expression (Fig 3 a, b & c). Given the broad distribution of VDR and 1,25OH2D3 metabolism across multiple tissues/cell types (Celli et al

1999, Amano et al 2009, Chiang et al 2015) we examined the effects of cadmium on CYP24 expression in three distinct human cell lines including: hematopoetic (HL60) cells, neuronal

(SHSY5Y) cells and a hepatic (HEPG2) cells. Each of the cell lines are responsive to

1,25OH2D3, exhibiting a range of 10-270 fold induction in CYP24A1 with HL60 cells exhibiting the greatest sensitivity (270 fold). Comparatively, SHSY5Y and HEPG2 cells required an ~ 8 fold higher concentration of 1,25OH2D3 to induce CYP24A1 gene expression

60 and 10 fold respectively. Consistent with the transient transactivation data, cadmium salts exhibited a dose dependent attenuation of 1,25OH2D3 mediated CYP24 induction

149

between 3-12 µM. Sensitivity of cadmium induced CYP24 attenuation was variable between cell lines with the neuronal SHSY5Y cell line exhibiting the greatest sensitivity, 10% at 3 µM, compared to HL60 (25% at 12µM) and HEPG2 (<10% at 12µM) cells.

Co-activator recruitment by Cd salts

Following a potent inhibitory response of Cd salts in altering transient transactivation of VDR, we conducted mammalian 2-hybrid (M2H) assay to examine if cadmium slats impacted the ability of VDR to recruit specific protein:protein interactions.

Specifically RXR as VDRs obligate heterodimerization partner and SRC-1 as a NR coactivator (Fig 4). These studies were conducted with 12uM cadmium, a concentration that produced the maximum inhibitory effect in transient transactivation assay in

Cos7/XREM cell reporter system. In our first assessment, we investigated if the presence of cadmium in the M2H system would attenuate VDR:RXR interactions. Addition of cadmium salts alone had no impact on VDR:RXR interactions (Fig 4a) and addition of cadmium slats in combination with 3nM 1,25OH2D3 also had little effect on VDR:RXR interactions. However, a trend in RXR recruitment was observed with 12µM cadmium dinitrate. When mammalian two hybrid assay was conducted in the presence of co-incubated full length SRC-1 a significant increase in RXR recruitment was observed for both cadmium chloride and cadmium dinitrate suggesting that SRC-1 may play a role in facilitating cadmium induced

VDR:RXR interactions (Fig 4b). To investigate the role of SRC-1 further, M2H assays were conducted to assess direct VDR:SRC-1 interactions. Again addition of cadmium salts alone

150

did not facilitate VDR:SRC-1 protein:protein interactions (Fig4c). Rather, assays conducted in the presence of 3nM 1,25OH2D3 exhibited significant synergistic activity both in the presence and absence of RXR (>= 200%) (Fig 4 c&d) although addition of RXR did not alter the induction parameters significantly. Overall it appeared that Cd enhances VDR:SRC-1.

Cadmium may disrupt VDR function though alterations in the DNA binding domain

We next hypothesized that cadmium salts may form complexes with cysteine residues on the zinc fingers of the VDR protein disrupting its DNA binding capabilities. We designed a chimera protein fusing the yeast Gal4 DNA binding domain with the human VDR ligand-binding domain (LBD) containing a functional AF2 domain. We anticipated this construct would allow for Cd salts to substitute for Zn within the yeast Gal4 zinc fingers without disruption in DNA binding behavior as previously reported (Pan et al 1990).

Transient transfections were then conducted with both full-length wild type hVDR, and the

Gal4-hVDR chimera in the Cos7 /XREM reporter system to evaluate the both efficacy of

1,25OH2D3 and the ability of Cd salts to attenuate 1,25OH2D3 mediated VDR transactivation. Control studies were first conducted to demonstrate that the Gal4-hVDR chimera alone was responsive to 1,25OH2D3 and that Cd salts would not induce transactivation (Fig5a). Next, the dose responsive inhibitory activity of Cd salts on

1,25OH2D3 mediated induction of XREM activity was confirmed using wild type hVDR (Figs

1a,b and 4a). Lastly, we demonstrate that Cd is ineffective at attenuating 1,25OH2D3 mediated induction of XREM activity with the Gal4-hVDR chimera (Fig5b).

151

Next, electrophoretic gel retardation assays (EMSA) were conducted with column purified recombinant His-tag human VDR to examine whether Cd salts were capable of inhibiting VDR:DNA binding activity. Results from this analysis demonstrate a dose dependent decrease in recombinant hVDR-DNA complex formation as observed as a decrease in band intensity with increasing concentrations of Cd (Fig 6). Specificity of the reaction was demonstrated by addition of 100X cold non-labeled oligo to the reaction with a result of reducing VDR-DNA band intensity. The inhibitory effect of Cd was independent of the presence/absence of 100pM of vitamin D3(Fig 6a,b). Effects of Cd on VDR-DNA binding were significant at > 50uM with complete attenuation in VDR-DNA binding at or above

150uM of CdCl2. Addition of molar excess of ZnCl2 or EDTA (1.25mM-0.5mM) did not rescue the inhibitory actions of Cd on DNA binding (Fig 6c). We next examined if HL60 cells treated for 24hr with vitamin D in the presence of Cd would exhibit a similar modulation in VDR-

DNA binding. Nuclear extracts were isolated from HL60 cells treated with 3M and 6M of both Cd chloride and Cd dinitrate with in the presence of 3nM of vitamin D3. Nuclear extracts from 3nM of vitamin D3 treated cells show a characteristic shift when mixed with canonical VDRE oligonucleotides. As above, specificity was demonstrated with 100X unlabeled oligo. Comparatively, EMSA results from HL60 cells treated with both 3nM of vitamin D3 and Cd salts demonstrate marked attenuation of VDR:DNA interactions (fig6b).

152

DISCUSSION

Cadmium compounds are known environmental contaminants with potential harmful effects on a variety of organisms1,23. Human and laboratory animals exposed to environmental Cd have been reported to exhibit a number of adverse effects involving multiple organs and systems. Important among them include anemia from lack of erythropoietin production; autoimmunity from immune function impairment; hypertension, diabetes, and myocardial infarction from cardiovascular effects; impaired neurogenesis, neuronal and axonal differentiation and neuronal cell death from effects on the CNS; infertility and breast cancer from reproductive and endocrine disruptive effects4,5,6,7,8,9.

Pertinent to this study are the observed effects of Cd induced renal damage characterized by decreased serum vitamin D3 levels, increased PTH levels, anemia, tubular proteinuria subsequently leading to metabolic bone disease including fibrous osteodystrophy, osteomalacia and osteoporosis6,22,25.

In this study we have investigated the role of cadmium-induced disruption of vitamin D/vitamin D receptor (VDR) signaling. Cd salts demonstrated a marked ability to inhibit transient transactivation of liganded VDR by directly modulating the VDR transcription machinery. The maximum inhibitory effect of Cd was observed at 12µM in- vitro in Cos 7 and Hek293 cells. These results are in line with similar observations of other nuclear receptors. For instance, Martin11 et al (2002) demonstrated that Cd inhibited binding of androgen to the androgen receptor (AR) receptor in human prostate cancer cells

153

without affecting its affinity. Cd was also demonstrated to inhibit the transcription of activity of E2 induced trout estrogen receptor (ER) in recombinant yeast30. Additionally, in- vivo Cd administration was also found to inhibit glucocorticoid receptor (GR) ligand and

DNA binding in the liver of rats12. Conversely, other studies demonstrate a stimulatory effect of Cd specifically with regards to its interaction with the estrogen receptor (ER). For example, Wilson31 et al (2004) demonstrated that Cd exposure in human breast adenocarcinoma cells resulted in a marked induction of ER transcriptional activities. This observation is consistent with an inducing effect of Cd on reporter gene activity in human breast cancer cells where Cd was found to directly bind with the ER LBD forming coordination complexes that resulted in stabilizing the receptor in its active conformation9.

Likewise, Cd exposure also had a stimulatory effect on progesterone receptor (PR)20. These interactions thus suggest that Cd has the capacity to both attenuate and enhance specific

NR functionalities.

To date however no studies have examined if Cd directly impacts VDR function.

Given the well-described impacts of Cd on kidney function and putative indirect impacts within the vitamin D axis, we sought to examine if Cd influences VDR functionalities. In screening compounds that modulate VDR in vitro transient transactivation (See chapter 2), we observed a strong and dose dependent attenuation of 1,25OH2D3 mediated induction of reporter activity following Cd exposures. This observation was consistent across multiple Cd salts suggesting that modulation of activity was due to Cd content and not individual salts.

We next explored the possibility whether Cd would inhibit the activity of a downstream

154

gene regulated by liganded VDR. We employed 3 different cell types each with a distinct lineage: a monocytic HL60, a neuronal (SHSY5Y) and a hepatic (HEPG2) cell line each capable of endogenous induction of CYP24A1 following treatment with vitamin D332,33,34. We hypothesized that Cd would directly disrupt VDR functionality thus affecting it’s ability to initiate transcription of target genes including expression of CYP24A1 which contains multiple VDRE’s within it promoter35. Expectedly, in all cell lines CYP24A1 induction by

1,25OH2D3 was significantly attenuated when dosed with cadmium salts. Again similar to transient transactivation assay there was a dose dependent attenuation with a maximum inhibitory effect attained at 12M thus supporting the disruptive potential of these compounds. Similar inhibitory effects of Cd exposure were observed with gene expression of GR in fetal rats19 and the concentration of AR protein and mRNA in human prostate cancer cells11. These studies further highlight the inhibitory role of Cd on NR activities.

Recruitment of coactivators is important for histone modification, chromatin remodeling and precedes the recruitment of RNA polymerase and other ancillary factors to initiate transcription36. Although a vast body of information is available on Cd modulation of the ER and other nuclear receptors, our understanding of Cd mediated effects on the coactivator recruitment of nuclear receptors is largely unknown and underexplored. Hence, we explored the role of Cd on VDR and performed mammalian 2-hybrid assays to ascertain the potential of Cd to recruit co-regulators and co-activators required for transcriptional activation of VDR. Based on our observation that Cd attenuates 1,25OH2D3 mediated transcription, we anticipated that we would observe a marker reduction in VDR:RXR and/or

155

VDR:SRC-1 protein:protein interactions. Interestingly however, both cadmium compounds favored recruitment of co-activator SRC-1 by VDR and also the recruitment of RXR in the presence of coexpressed SRC-1 in a synergistic manner that was several folds higher than the recruitment of vitamin D3 alone (Fig 6 a-d). Also, recruitment potential of Cd in this assay was tested at the highest concentration (12uM) at which both VDR transactivation and vitamin D induced CYP24 induction were found to be significantly repressed. Jiao et al., have demonstrated that Cd can increase the binding activity between human VDR and nuclear receptor coactivator 2 (TIF-2). However, an underlying mechanism for this interaction has not been established37. The coactivator promoting activity of Cd is considered unique and in sharp contrast to the recruitment behavior exhibited by similar metal containing compounds that are known VDR antagonists. We previously demonstrated

(refer to Chapter 1) that heavy metal containing compounds such as Tributyltin, Ziram,

Phenylarsine oxide, Potassium dicyanurate containing Sn, Zn, Ar, and Au respectively exhibited marked repression of coactivator and coregulator interactions of liganded VDR.

Also, mechanisms underlying repression of VDR coactivator recruitment particularly with trialkyltin (TBT and TPT) are quite different from that of Cd (Refer to Chapter 2). These findings are consistent with reports where inorganic arsenic has been found to inhibit transcription of GR related promoters by disrupting functions of CARM1 and GRIP1 that are members of p/160 family of coactivators38. The mechanism behind the ability of Cd to promote coactivator interaction of VDR is not clear and warrants future investigation.

However, Cd is known for its affinity for cysteine rich residues on proteins in addition to zinc

156

finger domains on NRs39,40. Qin16 and colleagues (2016) reported that Cd prevents DNA binding of helicase enzyme (BLM) by binding to sulfhydryl groups of solvent exposed cysteine residues and consequently promotes formation of protein aggregates.

Alternatively, in our case, Cd might play a role in stabilizing protein:protein interactions between VDR:RXR and/or VDR:SRC-1. Based on our results it is important to note that Cd enhances recruitment only in the presence of activated/liganded VDR. This implies that allosteric shift in receptor confirmation from ligand binding is required for Cd activity together with repositioning of helix 12 to create a charge clamp thus favoring coactivator interaction with LxxLL motifs41,42. Despite these probabilities, however, questions still remain as to where and how Cd binding to the receptor promotes coregulatory recruitment.

We next focused on the plausible mechanism of Cd dependant VDR inactivation, which might involve disrupting the functionality of zinc fingers present on the DNA binding domain of VDR. The DBD is comprised of two α‐helices containing up to nine cysteines in a defined spacing and orientation across 60–70 highly conserved amino acids. These cysteines form an essential bridge for the formation of two zinc finger motifs “P Box” and “D Box” that are responsible for recognition of receptor‐specific DNA binding half‐sites. Each zinc finger contains four of the invariable cysteines residues, which coordinate tetrahedrally with one zinc ion. Through these zinc fingers, VDR binds to canonical and non-canonical vitamin D response elements (VDREs) comprised of two hexametric half sites

5′‐AG(G/T)TCA‐3′, arranged as either a direct repeat separated by three nucleotides (DR3) or an everted repeat separated by six nucleotides (ER6) on the promoter of target genes.

157

Notably each zinc finger binds one Zn atom in a tetrahedral arrangement to form a complex with four highly conserved cysteine residues. Studies on other nuclear receptors and VDR highlight the importance of Zn for protein binding to target DNA site by bringing about significant alteration in the protein structure43,44,45,46 . Cadmium has the known ability to substitute for Zn in a variety of Zn finger containing proteins including nuclear receptors10,46,47. This is because unlike other toxic divalent metals such as As, Pb, & Cr, Cd shares similar co-ordination chemistry with Zn thus allowing it to replace Zn more easily than other metal ions15. Hence, in order to test the effects of Cd on VDR we designed a yeast Gal4-VDR chimeric construct by swapping the yeast Gal4-DBD with the VDR-DBD. We thus employed this construct as bait to examine if cadmium salts would continue to attenuate 1,25OH2D3 mediated transcription. As hypothesized, impact of Cd salts on

1,25OH2D3 mediated transcription was eradicated, whereas the ability of 1,25OH2D3 to fully transactivate the chimera was retained. It is important to note however, that there are significant structural differences in the yeast and other steroid receptor zinc fingers domains. The Gal4 DBD contains 6 Cysteine residues that bind 2 Zn atoms. When bound to

Cd it coordinates with 6 cys residues two of which form bridging ligands between the two

Cd atoms thus forming a binuclear metal cluster rather than a true zinc finger48,49,50.

Nevertheless, despite the difference in DBD structures, both liganded wild type VDR and

Gal4- chimera were able to transactivate, whereas addition of Cd to wild type human VDR abrogated the transactivation further suggesting the role of Cd in disrupting the Zn finger binding to DNA.

158

Given the structure of human VDR zinc fingers, there is likelihood that the first zinc finger (P box) might be a target of Cd inhibition. Functional deletion studies by Nishikawa51 and colleagues (1995) revealed that deletion of the first zinc finger did not affect heterodimerization with RXR as observed with our studies, whereas the deletion immediately downstream of the second finger (D box) did hamper heterodimer formation.

Since Cd at very low concentrations (<1uM) could not effectively inhibit VDR transactivation suggests that heterodimerization with RXR was not inhibited by Cd further suggesting the importance of the P box in Cd induced inhibition of DNA binding in this study.

The protective nature of Zn on Cd induced toxicity is well known52,53. We explored the possibility of Zn to ameliorate the inhibitory effects of Cd. We were able to demonstrate that addition of 1uM ZnCl2 along with Cd in the presence of vitamin D3 partially rescued the transactivation of VDR in Cos7 cells. Although the exact mechanism is not clear at this time addition of Zn could have stoichiometrically outcompeted Cd to restore zinc finger conformation of the VDR protein thus enabling renewal of transient transactivation.

Additionally, Zn could have prevented Cd induced ROS production and subsequent apoptosis of cells52. Removal of Cd from cultures during early periods of exposures has shown to even reverse the inhibitory effects of Cd on Sp1 binding activity10.

To further support the hypothesis that Cd substitutes for Zn disrupting DNA binding we performed gel retardation assays with both recombinant human VDR and RXR proteins and nuclear extracts (NE) of Cd treated HL60 cells to demonstrate a direct effect of Cd on

VDR:DNA interactions. Results from these studies indicate that Cd significantly attenuates

159

recombinant VDR:DNA binding at concentrations at ~ 50uM and above whereas protein:DNA interactions obtained from NE of treated cells were inhibited at 3 & 6 µM. This concentration range is similar to that used in transactivation and gene expression assays.

Electro spray mass spectrometry studies showed that VDR DBD can bind to DNA only when bound to 2 moles of Zn or Cd per mole of protein in-vitro while the presence of additional

Cd dissociates the DNA protein complex46. These studies suggest that the effects of Cd at higher concentrations can be deleterious for receptor function. Similarly, Guevel30 et al

(2000) observed that Cd abolished ER binding to canonical estrogen response elements

(ERE’s) at concentrations of ~ 100µM30. Additional reports with both GR and ER also suggest that Cd may replace zinc within their respective zinc finger motifs43,54. It is additionally important to note that co-incubation of Cd with molar excess of ZnCl2 or the chelator EDTA did not rescue the inhibitory actions of Cd on DNA binding in our EMSA studies suggesting that Cd binding to DNA was irreversible once Cd is bound to the VDR protein and likely results in a highly stable coordinate interaction with cysteine residues within the P and D box domains. This hypothesis however requires confirmation. Another important observation was that there was no change in inhibitory activity of Cd either in the presence or absence of liganded receptor suggesting that Cd most likely did not induce conformational alterations to the RXR:VDR heterodimer under these in situ conditions.

These results are in contrast to those observed with trout ER where Cd was only able to inhibit E2 activated receptor30.

160

CONCLUSION

In conclusion, we demonstrate that Cd has the ability to dysregulate the function of an important nuclear hormone receptor—vitamin D receptor, further emphasizing the endocrine disrupting properties of Cd on vitamin d endocrine system as a whole. Whereas

Cd exhibits inhibitory actions on the VDR transactivation and downstream endogenous gene target likely through Zn finger mediated disruption of DNA binding, the actions of Cd on protein:protein interactions although novel and inductive in nature and require further investigation. Nevertheless, Cd induced disruption of vitamin D endocrine system could have important physiological and pharmacological implications.

161

REFERENCES

1. McElroy, J.A. et al. Cadmium exposure and breast cancer risk. J. Natl. Cancer Inst. 98 (12), 869-873 (2006).

2. U.S. Environmental Protection Agency, Health Assessment Document of Cadmium. Office of Research and Development. EPA-600/8e81, NTIS Pub. PB82- 115163 (1981).

3. Nordberg, G. et al. Kidney dysfunction and cadmium exposure – Factors influencing dose – response relationships. J. Trace Elem. Med. Biol. 26, 197–200 (2012).

4. Wang, B. & Du, Y. Cadmium and its neurotoxic effects. Oxidative Medicine and Cellular Longevity (2013).

5. Alfvé N, T., Elinder, C.-G., Carlsson, M. D. & Grubb, A. Low-Level Cadmium Exposure and Osteoporosis*. J Bone Miner Res 15:1579–1586 (2000).

6. Horiguchi, H. et al., Hypoproduction of erythropoietin contributes to anemia in chronic cadmium intoxication: clinical study on Itai-itai disease in Japan. Arch Toxicol, 68:10 632–636 (1994).

7. Abu-Hayyeh, S et al. Cadmium accumulation in aortas of smokers,” Arteriosclerosis, Thrombosis, and Vascular Biol 21:5 863–867 (2001).

8. Hanson, M, L. et al. Prenatal cadmium exposure dysregulates sonic hedgehog and Wnt/-catenin signaling in the thymus resulting in altered thymocyte development. Toxicol Appl Pharmacol 242:2136–145 (2010).

9. Stoica, A., Katzenellenbogen, B. S. & Martin, M. B. Activation of Estrogen Receptor- ␣ by the Heavy Metal Cadmium. Mol Endocrinol 14:4 545-553 (2000).

10. Watkin, R. D., Nawrot, T., Potts, R. J. & Hart, B. A. Mechanisms regulating the cadmium-mediated suppression of Sp1 transcription factor activity in alveolar epithelial cells. Toxicology 184, 157–178 (2003).

11. Martin, M. B. et al. Role of cadmium in the regulation of AR gene expression and activity. Endocrinology 143, 263–275 (2002).

162

12. Dundjerski, J., Butorovic, B., Kipic, J., Trajkovic, D., & Matic, G. Cadmium affects the activity of rat liver tyrosine aminotransferase and its induction by dexamethasone. Arch. Toxicol. 70:390–395 (1996).

13. Byrne, C. et al. Metals and breast cancer. J. Mammary Gland. Biol. Neoplasia 18 :1, 63-73 (2013).

14. Martineź -Campa, C.M. et al. Melatonin down-regulates hTERT expression induced by either natural estrogens (17beta-estradiol) or metalloestrogens (cadmium) in MCF-7 human breast cancer cells. Cancer Lett. 268:2 272-277 (2008).

15. Deegan, B., J et al. Structural and thermodynamic consequences of the replacement of zinc with environmental metals on estrogen receptor alpha-DNA interactions. J Mol Recognit. 24:6 1007–1017 (2011).

16. Qin, W. et al. Mechanistic insight into cadmium- induced inactivation of the Bloom protein. (2016). doi:10.1038/srep26225

17. Siewit, C.L. et al. Cadmium promotes breast cancer cell proliferation by potentiating the interaction between ERalpha and c-Jun. Mol. Endocrinol. 24:5, 981-992 (2010).

18. Ponce, E., Louie, M.C., & Sevigny, M.B. Acute and chronic cadmium exposure promotes E-cadherin degradation in MCF7 breast cancer cells. Mol. Carcinog. 54:10 1014-1025 (2015).

19. Castillo, P., Ibáñez, F., Guajardo, A., Llanos, M. N. & Ronco, A. M. Impact of Cadmium Exposure during Pregnancy on Hepatic Glucocorticoid Receptor Methylation and Expression in Rat Fetus. PLoS One 7, 1–9 (2012).

20. Garcia-Morales, P., et al. Effect of cadmium on estrogen receptor levels and estrogen-induced responses in human breast cancer cells. J. Biol. Chem. 269:24 16896-16901 (1994).

21. Schwalfenberg, G. K. & Genuis, S. J. Vitamin D, essential minerals, and toxic elements: Exploring interactions between nutrients and toxicants in clinical medicine. Sci. World J. 2015, (2015).

22. Moon, J. The role of vitamin D in toxic metal absorption: a review J American Col Nutri, 13:6 559–564 (1994).

163

23. Uchida, H., Kurata, Y., Hiratsuka, H. & Umemura, T. The Effects of a Vitamin D– deficient Diet on Chronic Cadmium Exposure in Rats. doi:10.1177/0192623310374328

24. Ogawa, T. et al. Relationship among prevalence of patients with Itai-itai disease, prevalence of abnormal urinary findings, and cadmium concentrations in rice of individual hamlets in the Jinzu River basin, Toyama prefecture of Japan. Int J Environ Health Res 14:4 243– 252 (2004).

25. Tsuritani et al. Impairment of vitamin D metabolism due to environmental cadmium exposure, and possible relevance to sex-related differences in vulnerability to the bone damage. J Toxicol Environ Health. 37:4 519-33 (1992).

26. Kido, S. et al. Molecular Mechanisms of Cadmium-Induced Fibroblast Growth Factor 23 Upregulation in Osteoblast-Like Cells. Toxicol. Sci. 139, 301–316 (2014).

27. Yosuke, A., Yoshitake, C., Manabu, M., Kazuo, K., & Makoto, M. Increased nuclear expression and transactivation of vitamin D receptor by the cardiotonic steroid bufalin in human myeloid leukemia cells. J Steroid Biochem Mol Biol 114, 144-151 (2009).

28. Pfaffi, M.W. A new mathematical model for relative quantification in real--time RT- PCR. Nuclei Acids Res 29, 2002-7 (2001).

29. Sidhu, P.S., et al. Development of novel Vitamin D receptor coactivator inhibitors. ACS Med Chem Letters 5, 199-204 (2014).

30. Guevel, R. L. Inhibition of Rainbow Trout (Oncorhynchus mykiss) Estrogen Receptor Activity by Cadmium. Biol. Reprod. 63, 259–266 (2000).

31. Wilson, V.S., Bobseine, K., &Gray, L.E. Development and characterization of a cell line that stably expresses an estrogen-responsive luciferase reporter for the detection of estrogen receptor agonist and antagonists. Toxicol. Sci. 81: 69–77 (2004).

32. Celli, A., Treves, C. & Stio, M. Vitamin D receptor in SH-SY5Y human neuroblastoma cells and effect of 1,25-dihydroxyvitamin D3 on cellular proliferation. Neurochem. Int. 34, 117–124 (1999).

33. Amano, Y., Cho, Y., Matsunawa, M., Komiyama, K. & Makishima, M. Increased nuclear expression and transactivation of vitamin D receptor by the cardiotonic

164

steroid bufalin in human myeloid leukemia cells. J. Steroid Biochem. Mol. Biol. 114, 144–151 (2009).

34. Chiang, K. et al. Hepatocellular carcinoma cells express 25(OH)D-1a-hydroxylase and are able to convert 25(OH)D to 1a,25(OH)2D, leading to the 25(OH)D-induced growth inhibition. J. Steroid Biochem. Mol. Biol. 154, 47–52 (2015).

35. Chen, K. S. & DeLuca, H. F. Cloning of the human 1,25-dihydroxyvitamin D-3 24- hydroxylase gene promoter and identification of two vitamin D-responsive elements. BBA - Gene Struct. Expr. 1263, 1–9 (1995).

36. Haussler, M. R. & Whitfield, G. K. Molecular Mechanisms of Vitamin D Action. Calcif Tissue Int 3, 77–98 (2013).

37. Jiao, J. et al. Prokaryotic expression of human vitamin D receptor (hVDR) and its binding activities to Cd and Pb. Huan Jing Ke Xu 10, 2469-247 (2010).

38. Barr, F. D., Krohmer, L. J., Hamilton, J. W. & Sheldon, L. A. Disruption of histone modification and CARM1 recruitment by arsenic represses transcription at glucocorticoid receptor-regulated promoters. PLoS One 4, (2009).

39. Yamasaki et al. Effects of amino acid replacements on cadmium binding of metallothionein alpha fragment. Cell Mol Life Sci 53:5 459-465 (1997).

40. Jalilehvand, F., Leung, B., O, & Mah, V. Cadmium (II) complex formation with cysteine and penicillamine. Inorg Chem. 48:13 5758-5771 (2009).

41. Jurutka, P.W. et al. Molecular and functional comparison of 1,25-dihydroxyvitamin D(3) and the novel vitamin D receptor ligand, lithocholic acid, in activating transcription of cytochrome P450 3A4. J Cell Biochem 94, 917–43 (2005).

42. Rochel, N., Wurtz, J. M., Mitschler, A., Klaholz, B., & Moras, D. The crystal structure of the nuclear receptor for vitamin D bound to its natural ligand. Mol Cell 5, 173–179 (2000).

43. Freedman et al. The function and structure of the metal coordination sites within the glucocorticoid receptor DNA binding domain. Nature 334:543–546 (1988).

44. Pike, J., W. Vitamin D receptor: Structure and function in transcription. Ann Rev Nutri. 11, 189-216 (1991).

165

45. Freedman, L, P., & Luisi, B., F. On the mechanism of DNA binding by nuclear hormone receptors: a structural and functional perspective. J Cell Biochem 51:2 140- 150 (1993).

46. Veenstra, D., et al. Zinc induced conformational changes in the DNA-binding domain of vitamin D receptor determined by electrospray ionization mass spectrometry. J Am Soc Mass Spectrom 305, 8–14 (1998).

47. Sunderman, F., W. Cadmium substituted for zinc in finger-loop domains of gene- regulating proteins as a possible mechanism for genotoxicity and carcinogenicity of cadmium compounds. Tox Environ Chem 27, 131-141 (1990).

48. Pan, T. & Coleman, J. E. GAL4 transcription factor is not a ‘zinc finger’ but forms a Zn(II)2Cys6 binuclear cluster. Proc. Natl. Acad. Sci. U. S. A. 87, 2077–81 (1990).

49. Gardner, K., H. et al. Structure of the binuclear metal-binding site in GAL4 transcription factor Biochem. 30, 11292-11302 (1991).

50. Kraulis et al. Structure of the DNA binding domain of zinc Gal4 Structure of the DNA- binding domain of zinc GAL4. Nature (Lond) 356:448-450 (1992).

51. Nishikawa, J., Kitaura, M. & Nishihara, T. Vitamin D receptor contains multiple dimerization interfaces that are functionally different. Nucl acid Res 23, 606–611 (1995).

52. Szuster-Ciesielska, a et al. The inhibitory effect of zinc on cadmium-induced cell apoptosis and reactive oxygen species (ROS) production in cell cultures. Toxicology 145, 159–71 (2000).

53. Amara, S. et al. Preventive effect of zinc against cadmium-induced oxidative stress in the rat testis. J. Reprod. Dev. 54, 129–34 (2008).

54. Predky, P.,F., & Sarkar, B. Effect of replacement of ‘‘zinc finger’’ zinc on estrogen receptor DNA interactions. J Biol Chem 267:5842–5846 (1994).

166

FIGURES

Figure 1. Dose response curves showing inhibition of VDR transactivation by Cadmium salts. a) Inhibition curve using the XREM luciferase reporter in Cos7 cells b) CYP24A1 luciferase reporter in HEK293 cells. Data expressed as mean± SEM (n=3). Data are normalized to VDR

+ empty pSG5 vector.

167

Cd- Zn- D3 interactions

150

n o

i 100

t * c

u *

d * n

I **

% 50

O M M O M M S M M M M M u n S M M M M M u n u u n n n 6 3 u u n n n 6 3 M 1 6 3 3 3 M 1 6 3 3 3 D n e 3 3 3 e 3 D n e 3 3 3 te 3 Z d D D D d D Z t D D D a D ri ri n ra tr n o in in in o i it in in in i i l l m n in m h m m m h a i m m m d a c ta ta ta c it d ta ta ta it d i i i d V d i i i d V C V V V C + V V V C + + + C + + + M + M M M M u M M M u u u 6 u u u 6 6 1 u 6 1 1 e n 1 e e n n t e Z n d t Z Z a d Z ri a tr ri o tr i o l i in l h in d h c d c d d d C d C C C + + M M u u 1 1 n n Z Z

Figure 2. Treatment with micromolar concentrations of ZnCl2 aides in partially rescuing the inhibitory effect of Cd salts on VDR transactivation. Assays were run in Cos7 cells and data expressed as mean± SEM (n=3). Data are normalized to VDR + empty pSG5 vector. Asterisks indicate significance: *** = p < 0.001, ** = p < 0.01, * = p < 0.05.

168

Figure 3. Endogenous CYP24A1 gene expression profiles in three distinct cell lineages following treatment with Cd salts in the presence of vitamin D3. a) In myelocytic HL60 cells

CYP24A1 expression was induced to ~ 270 folds expressed as 100% in the presence of 3nM of vitamin D3 b) In neuronal SHSY5Y cells CYP24A1 expression was induced to ~ 70 folds expressed as 100% in the presence of 24nM of vitamin D3 and c) In hepatic HEPG2 cells

CYP24A1 expression was induced to ~ 10 folds expressed as 100% in the presence of 24nM of vitamin D3. Data expressed as mean± SEM (n=3). Data are normalized to DMSO. Asterisks indicate significance: *** = p < 0.001, ** = p < 0.01, * = p < 0.05.

169

a. CYP24A1 expression in hematopoetic cell line

Endogenous CYP24 A1 expression (HL60 cells)

150 n

o 100

b i

h *

n * I 50 % ** **

0 3 3 3 3 3 3 O M D D D D D D S n M M M M M M M 3 n n n n n n D 3 3 3 3 3 3 3 D 3 3 3 3 3 3 in D D D D D D m in in in in in in ta m m m m m m i a a a a a a V it it it it it it V V V V V V + + + + + + M M M M M M u u u u u u 3 6 2 3 6 2 1 1 e e e id id e t te te r r d ra a a lo lo ri it tr tr h h o n i i c c l i in in h d d d d d c d C C d C d d C C C b. CYP24A1 expression in neuronal cell line

Endogenous CYP24 A1 expression (SHSY5Y cells)

150

100

n 50

b i

h 8

n **** I 6

% **** 4 2 0 O 3 3 3 3 3 3 3 S tD D D D D D D M i M M M M M M D V n n n n n n M 4 4 4 4 4 4 n 2 2 2 2 2 2 4 + + + + + + 2 M M M M M M u u u u u u 3 6 2 3 6 2 e e 1 e e 1 d d e t t e ri ri d ra ra t o o ri it it ra l l o n n it h h l i i n c c h d d i d d c d d d C C d C C d C C

170

c. CYP24A1 expression in hepatoma cell line

Endogenous CYP24 A1 expression (HEPG2 cells)

150 n

o 100

n I 50 % ** ** ** ** 0 O M M M M M M M S n n n n n n n M 4 4 4 4 4 4 4 2 2 2 2 2 2 2 D 3 3 3 3 3 3 3 D D D D D D D it it it it it it it V V V V V V V

+ + + + + + M M M M M M u u u u u u 3 6 2 3 6 2 1 e 1 e e e t te e id id d ra a t r r ri it tr ra lo lo o n i it h h l i in n c c h d d i c d d d d d d C C C C C d C

171

Figure 4. Mammalian 2-hybrid assays showing recruitment of co-activator SRC-1 and coregulator RXRa in the presence of Cd salts. a) Recruitment of RXR in the absence of SRC-1. b) Recruitment of RXR in the presence of SRC-1 c) Recruitment of SRC-1 in the absence of

RXR and d) Recruitment of SRC-1 in the presence of RXR. Data are normalized to VDR + empty pSG5 vector and expressed as a percentage. Asterisks indicate significance: **** = p

< 0.0001, *** = p < 0.001, ** = p < 0.01, * = p < 0.05.

172

a) Recruitment of RXR without SRC-1

250 200

n 150

i t

c 100

u d

n 50

% 5 4 3 2 1 0 3 3 O 3 e te S D id D a D M r M tr M M n lo n i n D 3 h 3 in 3 ic + d + d e d te d id C a r tr C lo M i M h u in u c 2 d 2 i 1 1 d d d C C M M u u 2 2 1 1 pMRXR + Empty vector

b) Recruitment of RXR with SRC-1

400 ** 300

n * o

i 200

c u

d 100

% 5 4 3 2 1 0 3 3 O 3 e te S D id D a D M r M tr M M n lo n i n D 3 h 3 in 3 ic + d + d e d te d id C a r tr C lo M i M h u in u c 2 d 2 i 1 1 d d d C C M M u u 2 2 1 1 pMRXR + SRC-1

173

c) Recruitment of SRC-1 without RXR

300 **** **

200

t c

u 100

% 5 4 3 2 1 0 3 3 O 3 e te S D id D a D M r M tr M M n lo n i n D 3 h 3 in 3 ic + d + d e d te d id C a r tr C lo M i M h u in u c 2 d 2 i 1 1 d d d C C M M u u 2 2 1 1 pMSRC-1 + Empty vector

d) Recruitment of SRC-1 with RXR

300 **

200

n ***

t c

u 100

% 4 2 0 3 3 O 3 e te S D id D a D M r M tr M M n lo n i n D 3 h 3 in 3 ic + d + d e d te d id C a r tr C lo M i M h u in u c 2 d 2 i 1 1 d d d C C M M u u 2 2 1 1 pMSRC-1 + RXR

174

Chimeric GAL4-VDR

20 Gal4-hVDR Luc + 0-12uM Cd dichloride Gal4-hVDR Luc + 3nM D3

n a

h 10

l o

F 5

0 -3 -2 -1 0 1 2 3 uM b)

Figure 5. Comparison of inhibitory actions of Cd salts a) VDR transactivation in the presence of Vitamin D3 (3nM) and Cd salts (0-12µM) alone using the chimeric Gal4-hVDR luciferase reporter. b) XREM-Luciferase reporter and a chimeric Gal4-hVDR Luciferase reporter. Data expressed as mean± SEM (n=3). Data are normalized to VDR + empty pSG5 vector.

175

Figure 6. Electrophoretic mobility shift analysis of recombinant human VDR in response to treatment with Cd dichloride in-vitro. a) Shows Cd induced dose dependant inhibition of protein DNA binding in the presence of 100pM of vitamin D3. WT refers to wild type unlabeled oligos. 100-fold molar excess of unlabeled WT VDRE outcompetes DNA-protein complex formation with Cy5 labeled VDRE. The lane furthest on the right had labeled oligos run alone as negative control. b) Shows Cd induced dose dependant inhibition of protein

DNA binding in the absence of 100pM of vitamin D3. c) Cd induced inhibition at 100μM in the presence of 100pM vitamin D3 is not rescued by increasing concentrations of either

ZnCl2 or EDTA. d) Both Cd chloride (CdCl2) and Cd dinitrate (Cd(NO3)2) at 3μM and 6μM concentrations in the presence of 3nM of vitamin D3 inhibit protein-DNA binding when

12μg of nuclear extract (NE) are added.

176

177

178

GENERAL CONCLUSION

Nuclear receptors comprise one of the largest groups of transcription factors among vertebrates and mediate a wide range of physiological functions that are essential for growth, development, differentiation, metabolism, homeostasis and reproduction among others. Transcriptional activities of these receptors are predominantly driven through ligand dependent interactions with endogenous ligands. However, a growing body of evidence suggests that in addition to endogenous ligands, a diverse group of exogenous chemicals including environmental contaminants, pharmaceuticals and synthetic molecules that also include endocrine disruptors, can function as NR agonists or antagonists. These ligands typically interact within the ligand binding domains (LBD) of the NR or with a heterodimeric partner receptor, resulting in allosteric alterations to receptor helices which in turn facilitates recruitment of receptor coregulators and ultimately initiation or repression of transactivational events1,2,3.

We demonstrate transactivation or transrepression of vitamin D receptor by a variety of structurally diverse xenobiotics. We reveal how structural variations in the exogenous ligands can affect their interaction within the ligand-binding domain of VDR and facilitate allosteric modifications of VDR that result it’s ability to heterodimerization with its obligate partner RXR. We also examine how diverse VDR ligands (agonists/antagonists) influence subsequent recruitment of coactivators and /or corepressors thus affecting the overall transactivational functions of VDR.

179

We next investigated how binding of strong agonists to the LBD of both partner receptors (RXR:NR) can significantly alter the heterodimeric relationship between the

RXR:VDR and affect their transactivational response. Here we examined how strictly non- permissive partners such as RXR:VDR can behave in a manner that is conditionally permissive resulting in superactivation of the receptor heterodimer complex4 (Fig 2). In this regard, we discuss the novel role of potent RXR agonists, trialkyltins to modify the strictly non-permissive RXR:VDR heterodimeric partnership to one that is conditionally permissive.

This partnership is unique from those of other non-permissive to conditionally permissive ligand induced modifications. Unlike its usual role as a silent partner, RXR in this case likely assumes a dominant role and drives both the synergistic transactivation and inhibition of liganded VDR at low and high doses of trialkyltins respectively. Similar role of trialkyltins to synergize transactivation through RXR has been reported however, only in case of a permissive partner interaction as demonstrated with RXR:PPARy5 . This is the first instance that we are aware of that demonstrates conditional activation/repression of a typically non- permissive NR with an exogenous ligand for RXR.

While typical NR ligand interactions occur within the defined LBD of most NRs, recently reports suggest that some alternative mechanisms for ligand-receptor interactions.

For instance, Belorusova6 and colleagues (2014) have demonstrated that zebrafish VDR possess an alternate LBD that binds 1,25(OH)2D3 and may exhibit the ability to initiate VDR transactivation. Through our molecular modeling approach we have confirmed these results with human VDR and suggest it may be a site in VDR:xenobiotic interaction, all be it

180

a weaker site of receptor activation than that of the LBD. We and others additionally describe putative NR:xenobiotic interactions that target other NR domains beyond the LBD.

Pertinent to this observation is the highly conserved role the NR DNA-binding domain

(DBD), which is comprised of two α‐helices containing up to nine cysteines in a defined spacing and orientation. These cysteines form an essential bridge for the formation of two zinc finger motifs “P Box” and “D Box” that are responsible for recognition of receptor‐specific DNA binding half‐sites. Each zinc finger contains four of the invariable cysteines residues, which coordinate tetrahedrically with one zinc ion. The residues in the D box facilitate NR interactions with the phosphate groups of the DNA helix as well as dimerization. Our study with cadmium salts poses a perfect example where Cd binding to and likely replacing the Zn with the DBD zinc fingers adversely affects DNA binding and results in inhibition of VDR transactivation. Interestingly, Cd also enhances coregulator and coactivator recruitment to VDR as observed in our protein:protein interaction assays. The affinity of Cd for cysteine residues in locations other than the DBD has been reported earlier. In one instance Cd induced reporter gene activity by binding to the hormone- binding domain of ER7 where as in another Cd inhibited DNA binding of BLM protein (DNA helicase) by binding to solvent exposed cysteine residues and promoting aggregate formation8. However, at this point it is not clear if and where Cd binds/interact in additional locations within the VDR protein other than the DBD to affect coregulator interactions. This hypothesis requires further investigation.

A cumulative (but not exhaustive) model of differing mechanisms of NR disruptions

181

is represented in Fig 1. Exogenous chemicals target multiple sites and modulate receptor transactivation. Important among those sites include 1) the DNA binding domain of the NRs and promotion or inhibition of protein binding to DNA response elements. A good example of inhibition of protein:DNA binding is one that is exhibited by Cd salts following Zn replacement from zinc fingers. 2) Chemicals also alter expression levels of NRs and their heterodimeric partners (i.e Cd and trialkyltin mediated inhibition of CYP24A1 expression in

HL60 cells). 3) Antagonist binding to NRs inhibit RXR:NR interaction with coactivator and enhance recruitment of corepressor proteins consequently repressing receptor transactivation. 4) Agonist binding to the NRs induce a series of RXR:NR conformational changes including helix repositioning and hydrophobic cleft formations that provide a platform for 5) Coactivator recruitment that consequently leads to receptor transactivation.

Still additional mechanisms by which the activity of NRs can be significantly modified do exist. 6) Coactivator binding following ligand induced receptor activation can affect epigenetic remodeling in terms of unpacking of chromatin by histone acetyltransferases and promoting transcription. Conversely, co-repressor binding to the receptor can have a negative effect on gene transcription. Lastly, ligand induced interference with post- translational modifications of proteins can also have significant impact on the downstream signaling pathways and hence cannot be ignored9.

Toxicity testing of environmental chemicals in the 21st century has evolved dramatically with the emergence and application of in-vitro quantitative high throughput testing formats combined with an array of in-silico tools that aid in predicting and

182

prioritizing chemicals for in-vivo testing. This prolific surge in high-throughput techniques for generating libraries containing bioactivity profiles of compounds has resulted in near complete reliance on in-vitro cell based assays all of which use a variety of cell lineages of cancer origin. While these immortalized cell lines have been instrumental in our understanding of molecular biology, a number of important differences with normal cell lines and tumor tissues do exist. These include among others, variations in growth rate, gene and protein expression profiles (i.e. cell surface receptors and metabolic enzymes), genotype, phenotype, mutation rates, sensitivity or resistance to chemotherapeutic agents10,11 .These inherent variations pose challenges to assay reproducibility and data concordance between laboratories, as well as difficulties in data interpretation and extrapolation when it comes to comparison of activity/exposure outcomes in normal human cells. In fact, reasons behind some of the less potent VDR agonistic activities exhibited by compounds in our study such as tamoxifen citrate (antineoplastic) and lanoconazole (antifungal) might be attributed to cell-line specific features including resistance towards therapeutic agents, lack of metabolic enzymes, and limited expression or absence of cell surface transporters. Further testing of such chemicals employing other cell lines and follow-up gene/protein expression studies is however required to validate this presumption.

Whereas it might not be feasible to test each environmental chemical against a battery of cancer cell types or primary cultures because of labor, expense and time constraints, it might be worthwhile to examine activity in at least two or more cell lines of

183

distinct lineage to identify variations in responses if any that might be cell type specific.

Alternatively, testing, validation and adoption of newer more promising technologies such

3D culture to mimic in-vivo microenvironment, patient-specific inducible pluripotent stem cells (iPSC) to provide normal cell counterparts, and harnessing CRISPR/CAS9 for mechanistic studies could be promoted11,12. Most importantly, development of harmonized protocols, and strict regulations for abiding to them across laboratories conducting toxicity testing of environmental chemicals needs to be urgently implemented. The negligence practiced routinely by the scientific community and complacence on the part of regulatory bodies in failing to correct these floating practices cannot be over emphasized. Adoption of homogeneous protocols similar to GLP techniques followed across industrial laboratories could greatly reduce issues with concordance and reproducibility discussed earlier in addition to markedly diminishing publication of fabricated and suboptimal data in the scientific literature.

Despite shortcomings, animal models will remain as a crucial step for target confirmation and assessing overall organismal toxicity of in vitro validated compounds. Only through exploration of significant improvements in standardized tissue culture practices might we be able to produce in vitro model/s that would more closely recapitulate human exposures to environmental chemicals and their responses. Large-scale incorporation of such harmonized practices might eventually generate valuable data with improved scientific and public confidence and heightened credibility for high throughput toxicity testing.

In summary, this is the first study of its kind that demonstrates VDR’s ability to

184

interact with a variety of exogenous ligands and at the same time showcases the vulnerability of the receptor as a target of endocrine disruption. Future studies involving a combination of chemical structure and in-vitro activity profiles of the tested VDR agonists and antagonists can result in better modeling for predictive outcomes. This can greatly help to prioritize chemicals for in-vivo toxicity studies for unraveling underlying mechanisms and signaling pathways.

185

REFERENCES

1. Pardee, K., Necakov, A. S. & Krause, H. Nuclear receptors: small molecule sensors that coordinate growth, metabolism and reproduction. Subcell. Biochem. 52, 123– 153 (2011).

2. Janosek, J., Hilscherova, K., Blaha, L. & Holoubek, I. Environmental xenobiotics and nuclear receptors - interactions, effects and in vitro assessment. Toxicol. in Vitro 20, 18–37 (2006).

3. Grun, F. & Blumberg, B. Environmental obesogens: organotins and endocrine disruption via nuclear receptor signaling. Endocrinology 147, S50–S55 (2006).

4. Pérez, E., Bourguet, W., Gronemeyer, H. & De Lera, A. R. Modulation of RXR function through ligand design. Biochimica et Biophysica Acta - Molecular and Cell Biology of Lipids (2012). doi:10.1016/j.bbalip.2011.04.003

5. Le Maire, A. et al. Activation of RXR–PPAR heterodimers by organotin environmental endocrine disruptors. EMBO Rep. 10, 367–373 (2009).

6. Belorusova, A. Y. et al. Structural insights into the molecular mechanism of vitamin D receptor activation by lithocholic acid involving a new mode of ligand recognition. J. Med. Chem. 57, 4710–4719 (2014).

7. Stoica, A., Katzenellenbogen, B. S. & Martin, M. B. Activation of Estrogen Receptor- ␣ by the Heavy Metal Cadmium. Mol Endocrinol 14:4 545-553 (2000).

8. Qin, W. et al. Mechanistic insight into cadmium- induced inactivation of the Bloom protein. (2016). doi:10.1038/srep26225

9. Aranda, A. & Pascual, A. Nuclear Hormone Receptors and Gene Expression. Physiol Rev 81, 1269–1304 (2001).

10. Sandberg, R., and Ernberg I. Assesment of tumor characteristic gene expression in cell lines using tissue similarity index (TSI). PNAS, 102, 2052-2057 (2005).

11. Wilding, J. L., and Bodmer, W., F. Cancer cell lines for drug discovery and development, Cancer Res 74, 2377–84. (2014).

12. Niu, N and Wang, L. In vitro human cell line models to predict clinical response to anticancer drugs. Pharmacogenomics 16, 273-285 (2015). 186

FIGURES

Figure 1. Multiple mechanisms for NR disruption. Environmental agents are demonstrated to disrupt receptor transactivation at multiple sites of action including (1) receptor–RXR–

XRE interactions, (2) receptor expression, (3) receptor-corepressor recruitment, binding and dissociation, (4) receptor‐selective agonists or antagonist, (5) receptor-coactivator recruitment, binding or dissociation, (6) modification in receptor-coregulator chromatin acetylase/deacetylase activity.

187

Figure 2. RXR heterodimer interactions. NR/RXR dimers exhibit three functional modalities in relation to ligand‐activated transcription. Permissive heterodimers can be transactivated by ligands for either RXR or the partner receptor. The presence of ligands for both RXR and its partner can lead to additive or synergistic transcriptional activities. Non‐permissive RXR heterodimers may only be transactivated through interaction with the NR partner ligand.

Conditional heterodimers do not singularly facilitate transactivation, but rather they may synergize transactivation only in the presence of the partner ligand.

188