Investigating the Mechanisms of Idiosyncratic Drug Reactions: Lessons from Nevirapine-induced Skin Rash in Female Brown Norway Rats

Transcription

1 Investigating the Mechanisms of Idiosyncratic Drug Reactions: Lessons from Nevirapine-induced Skin Rash in Female Brown Norway Rats by Xin Chen A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Graduate Department of Pharmaceutical Sciences University of Toronto Copyright by Xin Chen 2014

2 Investigating the Mechanisms of Idiosyncratic Drug Reactions: Lessons from Nevirapine-induced Skin Rash in Female Brown Norway Rats ABSTRACT Xin Chen Doctor of Philosophy Graduate Department of Pharmaceutical Sciences University of Toronto 2014 Idiosyncratic drug reactions (IDRs) cause significant morbidity and mortality. At present it is impossible to predict who will have such reactions because the mechanisms involved are not understood. This work used nevirapine-induced skin rash in female Brown Norway rats as a model to investigate the mechanisms of IDRs. Specifically, we hypothesized that CD4 + T cells are mediating the rash, and the reactive sulfate metabolite of nevirapine induced the immune response by directly activating antigen presenting cells (APCs). Independent of which chemical species induced the rash (treatment with nevirapine or 12-OHnevirapine), CD4 + T cells isolated from the lymph nodes of animals responded vigorously to the parent drug, but not to 12-OH-nevirapine even though oxidation of nevirapine to 12-OHnevirapine pathway is required to induce the rash. This falsifies the basis for the PI hypothesis, which assumes that what the lymphocytes respond to is what induced the IDR. In the APC activation study it was found that nevirapine, 12-OH-nevirapine, and 12-OHnevirapine sulfate all appeared to activate APCs to some degree. For instance, CD40 was upregulated in RAW264.7 cells and bone marrow-derived dendritic cells, while CD86 was ii

3 upregulated in THP-1 cells. However, the effects were small and not limited to the reactive sulfate. Several interventions were also used to modulate the rash caused by nevirapine to learn more about possible risk factors; however, no results showed that any of them had a significant effect on the rash. This included depleting B cells and treatment with buthionine sulfoximine, retinoic acid, 1-methyl-tryptophan, lipopolysaccharide, imiquimod, or vitamin D. This animal model of an IDR allowed us to test mechanistic hypotheses that would be impossible to test by any other method. It provided insights into the mechanism of this IDR, and by extension, into the mechanisms of other IDRs. iii

4 ACKNOWLEDGMENTS I will never forget that sunny day in February 2007 when I visited Toronto for the first time and met eight supervisors in this top University to decide where I want to spend my next five years of time. Jack was the last one I interviewed with. He greeted me warmly with a father s big smile and asked a question no one else ever cared: Where are you gonna be in ten years from now? I didn t get it the first time. He repeated and elaborated: Describe where you will be and what you will be doing. I clearly remembered my answer: Working in a Pharma, and I don t mind going to other countries. He then started to talk about his education, his career path, and a lot of other things that are not very relevant to sciences, for at least 30 minutes or so. I went back to the hotel and said to myself: This is an interesting and challenging person. Most importantly, he looks like the Santa Claus! It has been 6 and a half years now since I joined Jack s lab and time proved that my decision was right. Throughout my Ph.D, Jack has provided tremendous help and guided me to become an independent scientific investigator and a critical thinker, which have made me stand out from most other people starting my day one in industry. He is probably thanked a million times for his achievement in science and academic training, but I would like to say, it is the non-scientific part that I learned from him will benefit me the rest of my life. He is like a father to us, strict but with love, fair and reasonable to everyone. I truly appreciate what Jack has taught me and thank him for being such a great mentor and role model. Along with great science and great supervisor, there is always a great team or you could call it a family. I sincerely thank all my lab mates for the great moments and conversations during my Ph.D - we shared our happiness and frustration throughout the graduate life. Special thanks to my big brothers Robert, Ervin, Ping and Feng, for their consistent support and inspiration for both better science and personal life. Great thanks to Connie for the amazing food and receipts, and Max for the fun time - he is way too adorable. I also want to thank my friends outside the lab who supported me during my heartbroken and backbroken times: Hui Wang, Carl Song, Yanan Chu, Jing Jing, Yanrong Shi. They made me stronger. iv

5 Lastly, thanks to my uncle s family, my parents, and my fiancékenny Liao for their generous love. I love them. This research was conducted in the spirit of the following (quotes from Jack): Good research always leads to more questions. It is better to get embarrassed early than late. The only reason I know more than you do is because I made more mistakes. It is very important to publish negative results so others know what is not working! You have to be open-minded! The same applies to life. v

6 TABLE OF CONTENTS ABSTRACT... ii ACKNOWLEDGMENTS... iv TABLE OF CONTENTS... vi LIST OF PUBLICATIONS... ix LIST OF ABBREVIATIONS... x LIST OF TABLES... xii LIST OF FIGURES... xiii LIST OF APPENDICES... xvii CHAPTER INTRODUCTION Overview of Adverse Drug Reactions Idiosyncratic Drug Reactions Clinical Manifestations Risk Factors Hypotheses of Mechanisms of IDRs Hapten Hypothesis Danger Hypothesis Pharmacological Interaction (PI) Hypothesis Animal Models NVP-induced Skin Rash in Female BN Rats D-penicillamine-Induced Autoimmune Diseases Investigating the Mechanisms of IDRs vi

7 Detection of Reactive Intermediate Formation and Covalent Binding T Cell Based Assays Co-administration Studies Pharmacogenetics in Clinical Research Rationale of the Present Studies CHAPTER A STUDY OF THE SPECIFICITY OF LYMPHOCYTES IN NEVIRAPINE- INDUCED SKIN RASH ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION CHAPTER INDUCING SKIN RASH IN FEMALE BN RATS BY TOPICAL TREATMENT OF NVP AND/OR 12-OH-NVP INTRODUCTION Topical Treatment of NVP on Sensitized Animals Attempts to Induce a Skin Rash in Naïve Rats by Topical Treatment with 12-OH- NVP MATERIALS AND METHODS RESULTS DISCUSSION CHAPTER FACTORS THAT MAY INFLUENCE THE INCIDENCE AND SEVERITY OF NVP- INDUCED SKIN RASH IN FEMALE BN RATS INTRODUCTION MATERIALS AND METHODS vii

8 4.3. RESULTS DISCUSSION CHAPTER POTENTIAL ACTIVATION OF ANTIGEN PRESENTING CELLS BY NVP AND/OR ITS METABOLITES INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSION CHAPTER CONCLUSIONS AND FUTURE DIRECTIONS Summary Implications and Future Directions REFERENCES APPENDIX viii

9 LIST OF PUBLICATIONS 1. Metushi, I. G., X. Zhu, et al. (2014). "Mild Isoniazid-Induced Liver Injury in Humans Is Associated with an Increase in Th17 Cells and T Cells Producing IL-10." Chem Res Toxicol. 27(4): Ng, W., A. R. Lobach, et al. (2012). "Animal models of idiosyncratic drug reactions." Adv Pharmacol 63: Zhang, X., F. Liu, et al. (2011). "Involvement of the immune system in idiosyncratic drug reactions." Drug Metab Pharmacokinet 26(1): Chen, X., T. Tharmanathan, et al. (2009). "A study of the specificity of lymphocytes in nevirapine-induced skin rash." J Pharmacol Exp Ther 331(3): Dugoua, J. J., M. Machado, et al. (2009). "Probiotic safety in pregnancy: a systematic review and meta-analysis of randomized controlled trials of Lactobacillus, Bifidobacterium, and Saccharomyces spp." J Obstet Gynaecol Can 31(6): Chen, X and J. Uetrecht. Factors that may influence the incidence and severity of nevirapine-induced skin rash in female Brown Norway rats. In preparation. 7. Chen, X and J. Uetrecht. Potential activation of antigen presenting cells by nevirapine and/or its metabolites. In preparation. ix

10 LIST OF ABBREVIATIONS 2-ME β-mercaptoethanol ADR Adverse Drug Reaction ALN Auricular Lymph Nodes APC Antigen Presenting cell AUC Area Under the Curve BCA Bicinchoninic Acid BMDC Bone Marrow-Derived Dendritic Cell BN Brown Norway BSO Buthionine Sulfoximine CD Cluster of Differentiation CFSE Carboxyfluorescein Succinimidyl Ester CO 2 Carbon Dioxide CYP Cytochrome P450 DAPK1 Death-associated Protein Kinase 1 DC Dendritic Cell DMEM Dulbecco's Modification of Eagle's Medium DMSO Dimethyl Sulfoxide ECL Enhanced Chemiluminescence EDTA Ethylenediaminetetraacetic Acid ELISA Enzyme-linked Immunosorbent Assay ELISPOT Enzyme-linked Immunosorbent Spot FBS Fetal Bovine Serum FcγII Fc gamma receptor II FITC Fluorescein Isothiocyanate FOXP3 Forkhead Box P3 G-CSF Granulocyte Colony-stimulating Factor GM-CSF Granulocyte Macrophage Colony-stimulating Factor GRO-KC Growth-related oncogene H & E Hematoxylin and Eosin HEPES 4-(2-Hydroxyethyl)piperazine-1-ethanesulfonic Acid HIV Human Immunodeficiency Virus HLA Human Leukocyte Antigen HMGB1 High Mobility Group Box 1 HPLC High-performance Liquid Chromatography HSPs Heat Shock Proteins IgG Immunoglobulin G LC/MS Liquid chromatography mass spectrometry IDO Indoleamine Dioxygenase IDR Idiosyncratic Drug Reaction x

11 IFN Interferon IgE Immunoglobulin E IL Interleukin IP-10 Interferon gamma-induced Protein 10 IU International Unit LPS Lipopolysaccharide LTT Lymphocyte Transformation Test MFI Mean Fluorescence Intensity MCP-1 Monocyte Chemoattractant Protein 1 MHC Major Histocompatibility Complex MIP-1 Macrophage Inflammatory Protein 1 NNRTI Nonnucleoside Reverse Transcriptase Inhibitor NOD Nucleotide-binding Oligomerization Domain NLRs Nucleotide-binding Oligomerization Domain Receptors (NOD-like receptors) NVP Nevirapine OD Optical Density PBMC Peripheral Blood Mononuclear Cell PBS Phosphate-buffered Saline PE Phycoerythrin PE-Cy7 R-Phycoerythrin-Cyanine Dye 7 PerCP Peridinin Chlorophyll Protein Complex PI Pharmacological Interaction PMA Phorbol Myristate Acetate Poly I:C Polyinosinic: polycytidylic acid RA Retinoic Acid RANTES Regulated on Activation, Normal T cell Expressed and Secreted RPMI Roswell Park Memorial Institute SDS-PAGE Sodium Dodecyl Sulfate Polyacrylamide Gel Electrophoresis SMX Sulfamethoxazole TBST Tris-buffered Saline Tween-20 TCR T Cell Receptor Th T Helper (cells) TLR Toll Like Receptor TNF Tumor Necrosis Factor Treg Regulatory T (cells) TRIM63 Tripartite Motif Containing 63 VEGF Vascular Endothelial Growth Factor xi

12 LIST OF TABLES Table 1. Classification of adverse drug reactions Table 2. Similar characteristics of NVP-induced skin rash in humans and female BN rats Table 3. Percentage of cell types before and after the depletion of lymphocyte subsets with magnetic beads Table 4. Design of NVP and anti-mouse CD20 antibody cotreatment experiments in naïve BN rats Table 5. Design of NVP and anti-mouse CD20 antibody cotreatment experiments in splenectomized female BN rats Table 6. Design of BSO and NVP cotreatment experiments in female BN rats xii

13 LIST OF FIGURES Figure 1. Hapten hypothesis Figure 2. Danger hypothesis Figure 3. PI hypothesis Figure 4. Structures of NVP, 12-OH-NVP, 4-Cl-NVP and 12-Cl-NVP Figure 5. IFN-γ secretion by lymphocytes in response to NVP, 12-OH-NVP, 4-Cl-NVP and 12- Cl-NVP Figure 6. IL-10 secretion by lymphocytes in response to NVP, 12-OH-NVP, 4-Cl-NVP and 12- Cl-NVP Figure 7. Frequencies of lymphocytes responding with the production of IFN-γ when stimulated with NVP or its analogs/metabolites from NVP-rechallenged rats using an ELISPOT assay Figure 8. Cell proliferation in response to NVP, 12-OH-NVP and 4-Cl-NVP as determined by the reduction of alamar blue Figure 9. Concentration of cytokines/chemokines produced by lymphocytes from control (n=4), primary treated (n=6), and rechallenged (n=7) animals in response to NVP as determined by a Luminex assay Figure 10. Production of IFN-γ in response to NVP by lymphocytes from rechallenged rats before and after depletion of CD4 + and/or CD8 + T cells Figure 11. Proposed chemical mechanisms of NVP-induced skin rash by formation of 12-OH- NVP sulfate in the skin Figure 12. H&E staining of skin samples from rats rechallenged with topical NVP (2.5 mg/ml in acetone/olive oil, 1:1/v:v) or vehicle only Figure 13. H&E staining of skin samples from naïve rats treated with topical 12-OH-NVP (15 mg/kg/day in DMSO) or vehicle only xiii

14 Figure 14. Percent of cells stained with CD45RA/B in peripheral blood following anti-mouse CD20 antibody injections Figure 15. Percent of cells stained with CD45RA/B in peripheral blood following anti-mouse CD20 antibody injections and NVP treatment Figure 16. B cell levels in the spleen, ALN, and peripheral blood on D22 of NVP treatment Figure 17. Spleen weight for selected groups on Day 22 of NVP treatment Figure 18. The effect of NVP treatment and anti-mouse CD20 antibody on plasma IgE levels. 75 Figure 19. Plasma levels of NVP and its metabolites in BSO-NVP co-treated or NVP-treated female BN rats Figure hours urinary excretion of NVP and its metabolite in BSO-NVP co-treated and NVP-treated female BN rats Figure 21. Liver weights and glutathione levels in the liver of BSO-NVP co-treated or NVPtreated female BN rats Figure 22. Effect of RA on the incidence of NVP-induced skin rash Figure 23. Effect of RA on plasma levels of NVP and its metabolites Figure 24. NVP and 12-OH-NVP plasma concentrations in animals that received RA (20 mg/k/day in oil, gavage) and an escalating dose of NVP (started from 15 mg/kg/day and escalated to 175 mg/kg/day) Figure 25. Plasma concentrations of cytokines/chemokines during a 21-days treatment course with NVP or RA-NVP determined by a Luminex assay Figure 26. Expression of CD40 on RAW cells after a 24 hours stimulation with 12-OH- NVP sulfate Figure 27. Expression of CD40 on RAW264.7 cells in response to various concentrations of NVP, 12-OH-NVP, or its sulfate metabolite expressed in two different ways xiv

15 Figure 28. Precent CD40 positive cells in aging RAW264.7 cells Figure 29. Representative dot plots and histograms of surface marker staining of THP-1 cells stimulated by various substances Figure 30. Change in cell surface marker expression on THP-1 cells in response to NVP or its metabolites Figure 31. Phenotype of bone marrow-derived cells Figure 32. Representative dot plots and histograms of surface marker staining on BMDCs stimulated by various substances Figure 33. Change in cell surface marker expression on BDMCs in response to NVP or its metabolites Figure 34. Cytokine production by drug-stimulated BMDCs at the end of a 24 hour incubation Figure 35. Covalent binding of DMSO, NVP, 12-OH-NVP, and its sulfate to BMDCs after a 24 or 72 hour incubation Figure 36. Change in cell surface marker expression on BMDCs in response to D-penicillamine or isoniazid Figure 37. Change in cell surface marker expression on RAW264.7 cells in response to D- penicillamine or isoniazid Figure 38. Phenotype of CD4 + CD25 - cells isolated from the spleen and ALNs of naïve female BN rats using magnetic beads and column Figure 39. Proliferation of αβ-tcr + cells measured by CFSE staining Figure 40. Chromatographs of (A) 12-OH-NVP sulfate and (B) 12-OH-NVP after a 24 hour incubation with immature BMDCs generated from naïve female BN rats xv

16 Figure 41. Quantification of 12-OH-NVP sulfate in cell culture medium in the presence or absence of BDMCs over time xvi

17 LIST OF APPENDICES Appendix 1. Supplemental Data of APC activation by 12-OH-NVP sulfate xvii

18 1 CHAPTER 1 INTRODUCTION

19 Overview of Adverse Drug Reactions Primum non nocere ( First, do no harm ) Hippocrates ( BC) It is a Latin phrase that means first, do no harm, a long-held principle in medicine, which unfortunately has never been achieved. As old as Medicine itself, adverse drug reactions (ADRs) unfortunately have been one of the many ways that a patient can come to harm through the practice of medicine. The World Health Organization defines ADRs as harmful, unintended reactions to medicines that occur at doses normally used for treatment [1]. Although many ADRs are mild, some can be very severe and sometimes even life-threatening. They are among the leading cause of mortality and morbidity ahead of pulmonary disease, diabetes, and AIDS etc.; they account for about 7% of hospital admissions [2, 3]. It was estimated that in 1994 overall 2,216,000 (1,721,000-2,711,000) hospitalized patients had serious ADRs, and 106,000 (76, ,000) had a fatal ADR, making these reactions between the fourth and sixth leading cause of death in UK and US [4, 5]. ADRs also represent a huge social and economic problem. Drug-related deaths from ADRs cost more than $136 billion a year, and 10% of drugs have been withdrawn from the market or received a Black Box warning in the past 25 years due to unacceptable safety profiles [2, 6, 7]. They represent a great cost to society because of the longer and advanced care needed for the affected patients. Even when occurring in a small number of people, ADRs can cause a drug to receive a black box warning or even be withdrawn from the market. This has been a big concern to the pharmaceutical industry given the limited numbers of new products in their pipelines. The clinical manifestations of ADRs are complex because they can be present in different organs and mimic other disease processes. Some ADRs are caused by inappropriate use of medications; however, ADRs will probably never be totally eliminated because different people respond to drugs differently [8, 9]. Although randomized controlled trials are the gold standard in testing the efficacy and safety of new drugs, they are not very effective in detecting ADRs, mainly because of the limited sample size and duration of the study. Compared to the highly controlled

20 3 environment in a trial, the real clinical settings are much more dynamic and complex. The inappropriate use of a drug may be caused by inappropriate dosage or duration of treatment, drug-drug interactions, or interactions with the over-the-counter products and off-label use, etc. Table 1. shows the 6-type classification of ADRs adapted from Edwards et al. and Pourpack et al. to showcase the variety and features of different types of ADRs: A (augmented, namely an enhanced pharmacological effect), B (bizarre or idiosyncratic, with unknown mechanisms but most likely involving the immune system), C (chronic or time-related), D (delayed effects), E (end-of treatment effects), and F (failure of therapy) [10]. It is obvious that many types of ADRs are preventable and ideally should not occur. To conquer this clinical and economic problem, many efforts have been applied jointly by regulatory agencies, academia, and the pharmaceutical industry to study the mechanisms of ADRs and develop post-marketing pharmacovigilance programs. However, it is also true to say that many ADRs are uncommon and remain unpredictable, which makes it impossible to set up an effective surveillance system or to prevent such reactions from happening. Most of the unpredictable ADRs fall into the second category, the Type B (Bizarre) reactions, which will be the focus of the present work.

21 4 Table 1. Classification of adverse drug reactions. Adapted from [10, 11]. Category Clinical Characteristics Example Solutions Augmented Predictable Pharmacology related Digoxin toxicity Reduce dose Pharmacological Effect Low mortality Bizarre Unpredictable Uncommon Not related to any known therapeutic effect of the drug Penicillin hypersensitivity Discontinue treatment Chronic Related to the cumulative dose Corticosteroids Reduce dose Delayed Withdrawal Occurs or becomes apparent some time after the use of the drug Occurs soon after withdrawal of the drug Carcinogenesis Reproduction defects β-blocker withdrawal Reintroduce the drug Failure therapy Usually caused by drug-drug interactions Inadequate dosage of an oral contraceptive Increase dosage

22 Idiosyncratic Drug Reactions The definitions of idiosyncratic drug reactions (IDRs) are inconsistent; here we refer to IDRs as those ADRs that do not occur in most patients at any dose used clinically and do not involve the known therapeutic effect of the drug [12]. IDRs accounts for 6-10% of all ADRs and have a typical incidence from 1/100 to 1/ [4, 13]. Although less common than other ADRs, IDRs are unpredictable in nature and often life threatening. The low incidence makes IDRs difficult to detect in clinical trials, and the unpredictable nature makes mechanistic studies on human subjects virtually impossible, and most studies are retrospective. There are also very few animal models for IDRs because IDRs are also idiosyncratic in animals, and only the ones that share similar characteristics with the reactions in humans would be of value. Therefore IDRs represent a big challenge to the pharmaceutical industry, and it is unlikely that much progress will be made in preventing these reactions until their mechanisms are well understood Clinical Manifestations From a broader view, IDRs can affect almost any organ with skin, liver, and blood cells being the most common targets. Some drugs only causes one type of IDR, but many others can affect multiple organs of the same person simultaneously, or affect different organs in different individuals. Some drugs may cause IDRs with similar patterns; there are also some common characteristics shared by many IDRs. However each of the reactions has its own unique characteristics. The most common feature of IDRs is that there is a delay between starting treatment with the drug and the start of the symptoms, while the time to onset on rechallenge is usually shortened. These characteristics provide evidence of an immune-mediated mechanism [14]. Depending on the drug, the delay in onset of symptoms can vary from 1 week to a couple of years. Another common characteristic is that the incidence of IDRs does not appear to increase with dose, and they are often referred as dose independent. However, this is misleading, and no IDR is dose independent. It is true that most patients do not experience IDRs at any dose, but this is due to the relative narrow therapeutic windows used clinically. Given that the mechanism of the IDR does not involve the therapeutic effects of the drug, there is no reason that the dose response curve for the therapeutic effect and the dose response curve for the IDR should be similar. Depending on where the dose-response curve of the IDR falls, the patients could experience an

23 6 adverse reaction well below the therapeutic dose of the drug (IDR dose-response curve falls below the therapeutic dose), or never experience any IDRs if the IDR dose-response curve starts well above the therapeutic window [14] Risk Factors Many factors have been identified as risk factors for IDRs such as age, gender, viral infection, and genetic predisposition, etc. Some of the risk factors are weak and are not true for all IDRs. For example, women are found to be more susceptible to halothane-induced hepatitis and clozapine-induced agranulocytosis, but not other IDRs [15, 16]. The risk of liver toxicity induced by isoniazid is found to be higher in elderly; however, the incidence is higher in infants when induced by valproic acid [17]. Some viral infections such as human immunodeficiency virus (HIV) infections and herpes virus seem to be associated with an increase incidence of IDRs, but most patients develop IDRs without being infected [18, 19]. A strong genetic component was found for some IDRs. For example, Mallal et al in 2002 reported that 50% human leukocyte antigen (HLA)-B*5701 patients treated with abacavir will develop an IDR. A cost-benefit analysis involving 19 countries has shown that prospective screening of this allele can reduce the incidence of abacavir-induced hypersensitivity. However, the cost-effectiveness may depend on patient populations, health care settings as well as the availability of appropriate laboratory assays in those regions [20, 21]. The association between IDRs and HLA genes, i.e., major histocompatibility complex, either MHC I or MHC II, also provides strong evidence that IDRs are immune mediated Hypotheses of Mechanisms of IDRs Although the mechanisms of IDRs are unclear, it is a general consensus that most are immune mediated. There is also overwhelming circumstantial evidence for the involvement of reactive metabolites, although some exceptions appear to exist. Despite all the efforts made to understand the mechanism of IDRs, it is also obvious that one single mechanism does not explain the characteristics of all IDRs. All of the hypotheses discussed below center on an immunological mechanism, and they are not mutually exclusive. One or more might be useful in explaining a specific reaction because IDRs are likely complex cascades of metabolic and immunological events.

24 Hapten Hypothesis The concept of haptens emerged from experiments reported by Landsteiner in He showed that small chemicals were unable to elicit an immune response unless they are chemically reactive and bound to proteins [22]. With more advanced knowledge, the hapten hypothesis has evolved to the following: a hapten refers to a chemically reactive compound or reactive metabolite, which can irreversibly bind to proteins and form drug-modified protein adducts. These adducts can be taken up by antigen presenting cells (APCs) and presented in the major groove of the MHC molecule to T cells. The recognition of the drug-protein adduct by T cell receptors (TCRs) is referred to as signal 1. One example that is consistent with the hapten hypothesis is penicillin-induced allergy [23]. The β-lactam ring of penicillin can irreversibly react with free amino and sulfhydryl groups on proteins and form drug-protein adducts. In some patients, this will induce the production of immunoglobulin E (IgE) antibodies against penicillin-modified proteins, which can stimulate degranulation of mast cells and release histamine, leukotrienes, and other inflammationassociated molecules [24]. These IgE antibodies are pathogenic and can mediate very severe allergic reactions including anaphylaxis. Unlike penicillin and other β-lactam class drugs, most drugs are not chemically reactive, and therefore cannot covalently bind to proteins to form adducts. However, many drugs such as carbamazepine and acetaminophen do form reactive metabolites, which can also act as haptens. Both of these drugs can undergo bioactivation to form quinone metabolites, which are usually detoxified by glutathione formation [25, 26]. When there is an imbalance between the reactive metabolite formation and the detoxification process, the toxic metabolites will bind to other proteins to form haptens [27, 28]. Regardless of whether the parent drug or the reactive metabolite is acting as the hapten, the relationship between covalent binding of a drug to a protein (and the degree of binding) and the risk for developing an IDR is vague. Recently, the β-lactam-albumin conjugates in patient plasma samples were identified, and the profile of drug-protein adducts at specific lysine residues with respect to dose and incubation time was determined [28]. The minimum levels of modification associated with the stimulation of a clinically relevant drug-specific T-cell response

25 8 were characterized using piperacillin-induced immune reactions in patients [29, 30]. This is the first direct evidence of how protein haptenation induced T cell responses at the cellular level. The hapten hypothesis is the mostly widely accepted theory of IDRs, and it has been well documented for contact hypersensitivity and respiratory allergens [31]. However, not all drugs that form reactive metabolites are associated with a significant incidence of IDRs, and it is not clear what determines which drugs will cause IDRs. Perhaps a better understanding of the implications of covalent binding and characterization of the proteins that are typically affected is needed to further support this hypothesis.

26 9 Figure 1. Hapten hypothesis. Reactive chemicals or reactive metabolites covalently bind to a protein and form a drug-protein adduct, which is then picked up by APCs and presented to T cells. The recognition of drugmodified proteins by TCR that leads to an immune response is referred as signal 1.

27 Danger Hypothesis Before introducing the danger hypothesis, we need to briefly review the history of immunological models in order to understand the impact this hypothesis has had on the field. The traditional model to address the specificity of the immune system is known as the selfnonself model that was proposed by Burnet in 1961 [32]. The key principle is that lymphocytes surface receptors can differentiate between the self and nonself factors and only respond to the foreigners. This theory was widely accepted for decades, although it was modified to fit with the results of later experiments. In 1989 Janeway proposed that induction of immune response requires a second signal, and he referred to adjuvants as the immunologist s dirty little secret [33]. The first signal is the recognition of the antigen or antigen-protein complex by TCRs (as described in Hapten Hypothesis); and the second signal is the interaction between costimulatory molecules on APCs and T cells, e.g., interaction between B7 molecules and cluster of differentiation (CD) 28. Although Janeway s theory emphasized that the induction of immune response depends on the recognition of pathogen by APCs not T cells, both the original selfnonself model and Janeway s theory are based on the recognition of foreignness. For almost a half decade, the traditional self-nonself model dominated immunology, although it does not explain many common phenomena. For example, gut bacteria are nonself substances, but we all tolerate them without a problem. A pregnant woman s immune system also does not attack its fetus. If the nonself is the major determinant of an immune response, humans would never be able to reproduce because many of the antigens are not produced until puberty, and therefore would not have induced tolerance in the prenatal period. The self-nonself model also does not explain the mechanisms of autoimmune diseases where no foreign substances are introduced. In 1994, Polly Matzinger presented the Danger model and proposed that it is cell damage rather than nonself that determines whether an immune response will occur [34]. Using this framework, the tissue that is injured or under stress releases danger signals, which then activate APCs leading to upregulation of costimulatory molecules and providing the second signal [35]. The danger hypothesis has significantly changed our perspective on what is involved in the induction of an immune response because it can explain why a wide variety of nonself exposures

28 11 do not trigger an immune response in the absence of significant cell damage as well as how endogenous molecules can induce immune reactions. The danger hypothesis also has had a big impact on our perspective of the mechanisms of IDRs. It could also explain why not all drugs that form reactive metabolites are associated with a high incidence of IDRs: the induction of immune-mediated IDRs may be determined by the ability of reactive metabolites to induce danger signals rather than their ability to form haptenated proteins [36, 37]. The danger hypothesis may also help with animal model development. If it is true, in theory we could introduce some danger signals to overcome the immune tolerance, which has been the greatest challenge to the development of animal models [38]. It is also possible that other factors leading to cell damage, such as viral infection and surgery, may be risk factors for IDRs [39]. As a follow up on the original hypothesis, many efforts have been made to answer two questions: 1) what is the identity of danger signals; 2) what is the link between production of danger signals and the risk of inducing an immune response such as IDRs. Researchers have suggested that hydrophobic biological molecules and stress-induced molecules from damaged cells such as high mobility group box 1 (HMGB1), heat shock proteins (HSPs), and S100 proteins are good candidates to be danger signals [40, 41]. Based on Matzinger s finding, endogenous molecules that are potential danger signals can bind to the same receptors that foreign antigens bind to, i.e., toll like receptors (TLRs) on APCs [35]. Although the danger hypothesis is very attractive, it is so very hard to rigorously test. The range of danger signals is unknown to date, and more evidence is needed to demonstrate to what extent this hypothesis is relevant to the mechanism of immune responses.

29 12 Figure 2. Danger hypothesis. Cells under stress can produce danger signals that can stimulate the costimulatory molecules on APCs and provide signal 2 for T cell priming. The lack of signal 2 leads to immune tolerance. This model does not address signal 1.

30 Pharmacological Interaction (PI) Hypothesis The pharmacological interaction of drug with antigen-specific immune receptors, or the PI hypothesis, proposed that chemically inert drugs can bind directly and non-covalently to the MHC-peptide complex. The non-covalent binding between drug and MHC-peptide complex may not be consequential per se, but a T cell with the appropriate TCR may bind to this drug- MHC complex with a much higher affinity and lead to an immune response [42]. This hypothesis was proposed by Werner J. Pichler based on the observation that T cell clones generated from patients with history of sulfamethoxazole (SMX) proliferated in response to the parent drug in the absence of metabolism [43]. He also found that when exposed to the parent drug and the metabolites (nitroso-smx and SMX-hydroxylamine), the generated T cell clones responded much better to the parent drug than to the metabolites [44]. The PI hypothesis is relatively new and therefore less tested compared to the other hypotheses. However, the use of T cell clones may lead to artifacts, and the response of these cells may not represent the mechanism by which a drug induced an immune response. In fact, more recent studies on SMX from Dean Naisbitt s group showed the opposite results: lymphocytes from patients with a history of SMX-mediated allergic reactions proliferated strongly with the nitroso metabolite but very weakly to the parent drug [45]. Consistent with this observation, the same group also showed that SMX can form metabolism-derived antigenic protein adducts in dendritic cells, which then stimulate dendritic cell signaling and lead to a T cell-mediated immune response [46, 47]. There is an implicit assumption on which the PI hypothesis is based - what the lymphocytes respond to is what initiated the immune response. We were skeptical that this assumption is correct, and we were able to test it with our animal model of nevirapine (NVP)-induced skin rash. The details of this research will be discussed in Chapter 2 [48].

31 14 Figure 3. PI hypothesis. The chemically unreactive parent drug can bind directly and reversibly to the MHC-peptide complex on an APC and be recognized by T cells with a TCR that fits the drug-mhc complex. This model does not address signal 2.

32 Animal Models Given the low incidence of IDRs and their unpredictable nature, mechanistic studies of human subjects are virtually impossible. Since IDRs are often not diagnosed until late, most human studies are retrospective and cannot study the events that led up to the IDR. Therefore, it is very difficult to test mechanistic hypotheses. The most effective tool to study mechanisms of IDRs is to use a valid animal model. However, drug reactions that are idiosyncratic in humans are also idiosyncratic in animals, and the model is useful only if the mechanisms and clinical manifestations mimic what occurs in humans. Therefore, it has been a great challenge to develop animal models, and as a result, very few good models exist to date. Sulfonamide-induced hypersensitivity in dogs mimic clinical manifestations in humans, and it appears to be a good model, but the incidence is about 1%, mostly in large breed dogs, and it is not very practical to study large numbers of large dogs [49]. Overdoses of acetaminophen can lead to acute liver failure, which is characterized by centrilobular hepatic necrosis both in humans and animals [25]. Although it has been extensively studied, acetaminophen-induced idiosyncratic liver injuries in mice is a model of direct hepatotoxicity rather than a model of an IDR [50]. Amodiaquine induces both agranulocytosis and hepatotoxicity in humans. When rats are treated with amodiaquine, it induces a similar immune response including slightly elevated liver transaminase levels and leucopenia, but without histological evidence in liver damage or agranulocytosis [51, 52]. Our group has shown that there is an immune adaptation component in the rodent model, and overcoming immune tolerance appears to be the biggest challenge to develop an animal model that mimics an IDR that occurs in humans [50]. Despite all of the challenges, we are still fortunate to have a few models that are good representatives of mechanisms of human IDRs. Two of these models will be discussed in the following sections: the NVP-induced skin rash in female Brown Norway (BN) rats, and the D- penicillamine-induced autoimmune syndrome in BN rats NVP-induced Skin Rash in Female BN Rats NVP (Viramune ) is a nonnucleoside reverse transcriptase inhibitor (NNRTI) used to treat human immunodeficiency virus-1 infections. Soon after being marketed, it was found to cause

33 16 skin rash and liver toxicity including Stevens-Johnson syndrome and toxic epidermal necrolysis [53]. In early studies, the incidence of skin rash was 16%, and that of clinically evident hepatotoxicity was 1% [53]. However, the incidence is lower at present because patients are started at a lower dose (200 mg once daily) for two weeks followed by the full dose (200 mg twice daily). Our lab found that NVP also causes a skin rash in female BN rats with similar characteristics to the rash that it causes in humans, and the detailed characteristics are listed below in Table 2.

34 17 Table 2. Similar characteristics of NVP-induced skin rash in humans and female BN rats. Characteristic Humans Rats Time to onset Less than 6 weeks, mostly 1~3 weeks [53] Develop red ears in 7 days and skin lesions in 2~3 weeks [54] Dose-response Incidence increases with dose [55] Incidence increases with dose [54] Female sex Increased susceptibility [56, 57] Increased susceptibility [54] NVP plasma levels ~ 5 µg/ml [58] ~20-40 µg/ml [59] Rechallenge Earlier onset and more severe [55] Earlier with red ears in <24 hours and skin rash in a week [54] Lead-in dose treatment A 2-week low dose (200 mg/day) followed by the full dose (200 mg twice daily) decreased the incidence [53] A 2-week low dose (40 mg/kg/day) followed by the full dose (mg/kg/day) prevented the rash [54] CD4 T cell count Low CD4 + T cells count is protective [60] Partial depletion of CD4 + T cells delayed the rash [61] Response of lymphocytes from patients/rats with a rash T cells produce interferon (IFN)-γ when stimulated with NVP [62] T cells proliferate and produce IFN-γ when stimulated with NVP [48]

35 18 Female patients are more susceptible to NVP-induced skin rash than males [56, 57]. Similarly, NVP-induced skin rash in rats is strain- and sex-dependent [54]. When female BN rats are fed a diet containing NVP at a dose of 150 mg/kg/day, they develop red ears in about 7 days and skin rash in days with an incidence of 100% [54]. The incidence in female Sprague Dawley rats was 20%, whereas none of the male rats of either strain developed a rash. However, the blood level of NVP and its 12-hydroxy-metabolite (12-OH-NVP) are also lower than in female BN rats, and if these animals are cotreated with aminobenzotriazole to inhibit cytochrome P450, the incidence of skin rash increases [54]. In addition to the fact that females have a higher incidence of skin rash in both humans and in our model, the characteristics of the skin rash are similar as well. For instance, the onset of rash occurs 2-3 weeks after starting NVP treatment, and the syndrome resolves when the treatment is discontinued. Upon rechallenge with NVP, the onset of rash is accelerated. A 2-week low dose (40 mg/kg/day) followed by the full dose (150 mg/kg/day) of NVP also prevented the rash in female BN rats [54]. A low CD4 + T cell count decreases the risk of rash in humans. Similarly, partial depletion of CD4 + T cells delayed the rash in female BN rats [61]. The fact that the characteristics of NVP-induced skin rash in female BN rats are similar to that in humans suggests that the mechanisms are similar; therefore, we have used this as an animal model to study IDRs. However, it should be noted that the skin rash in humans can vary from a mild rash that resolves despite continued treatment to life-threatening toxic epidermal necrolysis; the rash in rats is more like the milder form of rash in humans. In patients, in addition to the faster onset on rechallenge and the presence of drug-specific T cells, which are clear evidence of immune reactions, NVP-induced skin rash is also reported to be associated with specific HLA genotypes including the MHC II allele HLA-DRB1*01 [63] and HLA-DRB1*0101 [64], as well as the MHC I allele HLA-Cw8 [65]. These associations suggest involvement of the adaptive immune system. A toxicogenomics study also suggested that the genetic predisposition to NVP-induced IDRs varies between different ethnic groups [66]. Likewise, our animal studies also provided overwhelming evidence of an immune-mediated mechanism in rats. First, histology of skin sections revealed the existence of an inflammatory cell infiltration, primarily T cells and macrophages [54, 67]. Second, when rats with a skin rash were removed from NVP until the rash resolved and then rechallenged with NVP, there was a more rapid onset with red ears in less than 24 hours and skin lesions in about 1 week [54]. This suggests an amnestic immune response. Moreover, this sensitivity can be transferred to naïve

36 19 animals with splenocytes or just splenic CD4 + T cells from rechallenged animals, and the onset of the reaction for these naïve recipients followed the same time course as NVP-rechallenged animals [54]. Pretreatment with immunosuppressants, i.e. cyclosporine and tacrolimus, prevented NVP-induced skin rash, and it also led to resolution of the rash during NVP treatment, further supporting an immune mechanism of the rash [61]. One fundamental question is whether an IDR is caused by the parent drug or a reactive metabolite. Jie Chen et al. in 2009 showed that hydroxylation at the 12 position of NVP to form 12-OH-NVP is required to induce a skin rash [59]. Specifically, substitution of the methyl hydrogen atoms at the 12 position of NVP with deuterium decreased the rate of 12-hydroxylation and led to a decreased incidence of skin rash; treatment with the 12-OH-NVP metabolite also causes a rash at a lower dose [59]. This demonstrates that the 12-OH-NVP pathway is responsible for inducing the rash. 12-OH-NVP is not chemically reactive, and it is the same oxidation state as the quinone methide so the rash cannot be caused by oxidation of 12-OH-NVP, but it can be further metabolized to a benzylic sulfate. The most recent data showed that this sulfate metabolite can covalently bind to proteins in the epidermis of rats, where the sulfotransferases are located. Such binding was also present when the 12-OH-NVP metabolite was incubated with homogenized human skin, but not with murine skin [68]. This may explain why mice do not develop a rash following NVP treatment: they lack the sulfotransferase required to form the reactive sulfate metabolite in the skin. Moreover, topical treatment with 2- phenylhexanol, a sulfotransferase inhibitor, prevented the covalent binding as well as the rash, but only where it was applied [69]. These results provide definitive evidence that 12-OH-NVP sulfate formed in the skin is responsible for the rash. There is also a report that 12-mesyloxy- NVP, a synthetic 12-sulfate-NVP surrogate, can form conjugates with glutathione, amino acids, DNA, and haemoglobin [70-72]. This finding is irrelevant because the mesylate is far more reactive than the sulfate, and it is not formed in biological systems. Another question is how this IDR is initiated, which is virtually impossible to study in humans. There are three major hypothesis explaining the initial steps of IDRs: the hapten hypothesis, the danger hypothesis, and the PI hypothesis. The finding that the sulfate binds to skin proteins is consistent with the hapten hypothesis. We also found that treatment of animals led to early (6 hours) changes in gene expression in the skin that are consistent with the danger hypothesis, and there were many more changes after treatment with 12-OH-NVP than with NVP, which is

37 20 consistent with the fact that 12-OH-NVP is the obligate intermediate in the formation of the sulfate reactive metabolite in the skin [73]. Our group also found that T cells from a sensitized animal responded to incubation with NVP by producing cytokines such as IFN-γ and interleukin (IL)-10, which implies that the rash was induced by NVP rather than a metabolite, and this is consistent with the PI hypothesis. However, we already knew that the induction of rash depends on 12-hydroxylation. This assumption was the basis for the PI hypothesis, and it is only with an animal model that this could be tested. NVP-induced skin rash in rats, as one of the very few animal models with characteristics similar to the IDR that occurs in humans, has allowed testing of several hypotheses that could not be tested in humans. It is the first study to use a valid animal model to demonstrate that a reactive metabolite is responsible for an IDR, in this case a reactive metabolite formed in the skin. It appears that both hapten formation and the induction of danger signals are involved, but there are probably many factors that make this drug so effective in inducing an immune response and many details remain to be studied. This thesis will further explore the sequence of events by which a reactive metabolite leads to an IDR using this animal model D-penicillamine-Induced Autoimmune Diseases Penicillamine is used in the treatment of Wilson s disease and rheumatoid arthritis but associated with a relatively high incidence of various autoimmune syndromes, including a lupus-like syndrome, and myasthenia gravis [74]. It also causes an autoimmune reaction in BN rats, with features of lupus in humans such as antinuclear antibodies and immune complex deposition in the kidneys [75, 76]. The reaction is idiosyncratic because it only occurs in BN rats and the incidence is ~50% even though this is a highly inbred strain of rats; and other strains are resistant. The dose-response curve of penicillamine-induced autoimmune disease in BN rats is very unique: the incidence is 0% at a dose of 5 to 10 mg/day; the incidence is between 50% and 80% at a dose of 20 mg/day and not increased by increasing the dose to 50 mg/day. A low dose treatment (10 mg/day) for 2 weeks induces immune tolerance to a dose of 20 mg/day, which can be transferred to naïve animals with spleen cells from a tolerant animal [77]. In contrast, the protection induced by a low dose treatment in NVP-induced skin rash in BN rats is metabolic tolerance, because it could not be transferred with splenocytes, and co-treatment with aminobenzotriazole (a P450 inhibitor) breaks the tolerance [61].

38 21 D-penicillamine is chemically reactive without being metabolized; it reacts with aldehydes on APCs which are normally involved in signaling between T cells and APCs [78]. Rhodes et al. reported that the amines on T cells and the aldehydes on APCs can form a reversible imine bond, which is essential for antigen-specific T cell activation [79]. Similarly, penicillamine can react with aldehydes to form a thioazolidine ring, which is not readily reversible, and lead to activation of APCs and an autoimmune syndrome [78, 80]. Partial depletion of macrophages with clodronate-filled liposomes can decrease the incidence of penicillamine-induced autoimmunity; however, it also inhibits tolerance during low-dose penicillamine treatment [81]. The above studies suggested that macrophages play an important role in both the pathogenic mechanism and mediating the tolerance induced by low dose treatment. Recent studies found that 24 hours following the first dose of penicillamine, a spike in IL-6 was observed only in rats that developed autoimmune syndrome later in the treatment course [82]. The percentage of T helper 17 (Th17) cells was significantly increased, but only in sick animals. IL-17, a characteristic cytokine produced by Th17 cells, was increased in sick animals at both the messenger RNA and serum protein level. Macrophages can be activated by penicillamine and produce IL-6; IL-6 is also the driving force of Th17 cell differentiation [80]. Taken together, these findings suggest that macrophages are directly activated by penicillamine, which in turn induce Th17 cell differentiation and lead to penicillamine-induced autoimmunity. Many immune modulators have been used to manipulate the immune system to influence the incidence of penicillamine-induced autoimmune syndrome in BN rats. For example, the incidence and severity are increased by poly I:C, a synthetic double-stranded polyribonucleotide that functions as a viral RNA analog and can stimulate APCs via TLR 3. Lipopolysaccharide (LPS), which is found in the outer membrane of Gram-negative bacteria and stimulates macrophages via TLR 4, also had similar effects to that induced by poly I:C, but less pronounced [77, 83]. In contrast to poly I:C and LPS, misoprostol (a prostaglandin E analog) and aminoguanidine (an inhibitor of inducible nitric oxide synthase) were both protective [83]. On the other hand, tacrolimus, a T cell immunosuppressant, not only prevented the autoimmune syndrome and induced tolerance to continued treatment, but it could also reverse ongoing disease and prevented recurrence of autoimmunity upon re-exposure to penicillamine [81].

39 22 Similar to the NVP-induced IDR in BN rats, the penicillamine-induced autoimmune syndrome has a delay in onset of symptoms of about 3 weeks. However, unlike the NVP model, the onset of disease is not accelerated on rechallenge of penicillamine. Although the mechanism is obviously immune-mediated, the reason why there is no immune memory is unknown. In addition, given that it is an autoimmune syndrome and the antigens remain, it is not known why it does not persist after the drug is stopped. We believe that IDRs are due to a failure of immune tolerance mechanisms in the patients who develop a reaction; likewise, immune tolerance is the major reason why some animals do not develop IDRs while others do. The fact that penicillamine only causes an IDR in certain BN rats makes this model a perfect tool for comparative studies between rats that do get sick and rats that do not. We were able to show some markers that can predict which rats will develop such reactions and which don t, e.g., the early spike in IL-6. We will continue to use this animal to study the difference between sensitive animals and tolerant animals and hopefully provide more mechanistic clues on how to predict which drugs will cause IDRs in humans Investigating the Mechanisms of IDRs Understanding IDRs has been a real challenge due to the lack of valid animal models. However, our research has progressed rapidly with the development of many advanced assays and growing knowledge of immunology and the human genome. In this section, several state-of-the-art approaches used for mechanistic studies will be discussed to showcase how they have advanced our knowledge of drug hypersensitivity Detection of Reactive Intermediate Formation and Covalent Binding In general, formation of drug-modified proteins following bioactivation of the parent drug to form reactive intermediates is believed to be a required step for most IDRs. Many efforts have been devoted to screen drug candidates for the formation of reactive metabolites such as screening for suicide inhibition of drug metabolizing enzymes or glutathione conjugates. However, these approaches usually cannot detect all reactive metabolites because the pathways are complex [84]. Covalent binding, as suggested by the hapten hypothesis, appears to be a crucial step in the pathogenesis of most IDRs. However it is important to note that many drugs

40 23 that do not lead to IDRs are also known to form reactive metabolites, and the quantity of covalent binding does not necessarily correlate with the incidence of IDRs [84]. It is suggested that the pathogenesis of the binding is determined by the target of binding and the association with cytotoxicity in vivo. However, we have not yet identified any common pattern of binding or unique target proteins associated with most IDRs. A good example of covalent binding that leads to an IDR is the NVP-induced skin rash in female BN rats as discussed previously [69]. This is also the first study to use an animal model to demonstrate that a reactive metabolite is responsible for an IDR, in this case the 12-OH-NVP sulfate formed in the skin. Future characterization of the covalent binding in this model will provide more evidence on how that translated into toxicity and contributed to the initiation of an immune response. As discussed, reactive metabolite formation and covalent binding are probably necessary but not sufficient for most IDRs. Screening drug candidates to avoid chemical structures that are likely to form a reactive intermediate or covalent binding would likely decrease the likelihood of IDRs, however, it also results in potential loss of useful drugs T Cell Based Assays T cell based assays are routinely used in diagnosis of drug hypersensitivity and research in IDRs because they pose no threat to patient safety. The most well known is the proliferation based lymphocyte transformation test (LTT). This test reproduces T cell responses to a drug in vitro, from which one concludes a previous in vivo sensitization [85]. Basically, drug-specific T cells are isolated from patients with a history of drug hypersensitivity and cloned in vitro to increase the sensitivity of the assay. T cell response is then determined by proliferation or cytokine production. A positive LTT demonstrates the specificity of the T cells to the drug tested. It is reported that the LTT has a general sensitivity of 60-70% and a specificity of 93% in detecting β- lactam hypersensitivity [85]. However, numbers are different for different drugs. A large prospective study on isoniazid-induced hepatotoxicity by Warrington showed that the specificity of LTT was 83-90% and the sensitivity was only 50% [86]. This well designed study also addressed the fact that a greater sensitivity was obtained when drug-modified proteins (instead of the drug or its metabolite) were used to stimulate T cells in vitro. Despite that, the LTT is widely accepted as a diagnostic tool. However, the mechanistic relevance of detecting drug reactive T cells is indirect, e.g. it does not explain how the immune response is initiated or if the T cell

41 24 response is a cause or an effect of the IDR. In fact, what T cells respond to may not necessarily be what induced the immune response in the first place. Such assumptions can only be tested by using an animal model. While the LTT is used to measure the response of long lasting antigen-specific T cells, there are other T cell-based assays developed to predict the potential of a drug to cause hypersensitivity. These assays typically utilize isolated human APCs (for example, monocytes-derived dendritic cells) to recognize chemically reactive sensitizers such as 2,4-dinitroclorobenzene [87]. Basically, APCs isolated from humans are exposed to an allergen for 1-2 weeks in the presence of naïve T cells to allow initial antigen presentation. Afterwards, T cells are re-stimulated under the same experimental conditions to allow detectable levels of proliferation or cytokine production. The APCs used in most studies are derived from peripheral blood mononuclear cells (PBMCs) isolated from healthy volunteers. Responder T cells are usually purified by flow cytometry or magnetic beads to exclude natural CD4 + CD25 + FOXP3 + regulatory T (Treg) cells. The elimination of Treg cells lowers the barrier of T cell response to weakly immunogenic allergens and increases the sensitivity of the assay [88]. The critical point of this assay is the antigen presentation process: can APCs successfully deliver a message to T cells. Therefore, the success depends on the availability of the APC as well as their ability to recognize and present the antigens [89]. Allergens can be added either in the form of a reactive metabolite to directly activate APCs or in the form of a hapten-protein conjugate. The modification of cellular/extracellular proteins of APCs by a reactive metabolite is impossible to control, but it is a determinant of the antigen presentation and T cell activation. Therefore, only a positive T cell response is informative, and a negative result cannot be interpreted [90]. One possible explanation of a negative T cell priming assay is that the modification of the APCs did not lead to generation of epitopes that can be recognized by T cells. This could be because the type of APC used did not have the MHC required to present the required peptide, or the wrong antigen was used (i.e. a specific drug-modified protein/reactive metabolite may be needed instead of the parent drug). Another possible explanation for a negative T cell priming assay could be that the TCR repertoire used in the system does not include a TCR that can recognize the chemical [90]. In general, using hapten-protein conjugates has some advantages except that danger signals (such as cytokines) that are produced in vivo may be missing and may need to be added to the assay in order to successfully prime T cells [90].

42 25 T cell based assays provide strong evidence that a specific drug is involved in inducing an immune response and are complementary to most other single cell based in vitro assays. Due to the obvious complexity, these assays are still in development and need to be optimized individually for each compound Co-administration Studies Co-administration of a drug that causes IDRs with other substances helps to identify the risk factors and is useful in animal model development. These substances usually include, but are not limited to, metabolic inhibitors, immunosuppressants, and modulators. Some of them are used to determine the involvement of the immune system. For example, co-treatment with immunosuppressants or immune modulators can provide evidence of the involvement of the immune system and which cells are involved. These modulators can be small molecules such as agonists or antagonists of a specific pathway, or biological products such as neutralizing antibodies. A good example is that the depletion of CD4 cells by anti-cd4 antibodies led to a decreased incidence of NVP-induced skin rash in female BN rats. This suggested that CD4 cells mediate this IDR [61]. TLR agonists such as LPS and poly I:C are also often studied because they facilitate breaking immune tolerance, which has been the biggest challenge in animal model development. Metabolic inhibitors manipulate the production and turnover rate of certain metabolites in the biological system and identify their roles in inducing the IDRs. For example, inhibition of detoxification pathways may result in accumulation of a reactive metabolite that is responsible for an IDR and lead to an increased incidence. Nevertheless, all of these manipulation studies either require an animal model that represents a similar mechanism of IDR as that in humans, or aim to develop such models Pharmacogenetics in Clinical Research Genomic studies of patients has provided strong evidence that genetic predisposition is a risk factor for some IDRs. Identifying these risk factors could significantly decrease the incidence of IDRs and improve the process of drug development. There are mainly two categories of genetic predispositions that are associated with IDRs: polymorphisms involving drug metabolizing enzymes and drug transporter genes, or the immune system in the case of a drug-induced allergic reactions [91]. One example of the first category is the link between deficiency in glutathione synthetase and increased hepatotoxicity of certain drugs such as acetaminophen [92, 93].

43 26 Glutathione conjugation is a major detoxification pathway of many drugs. A deficiency in synthesizing the molecule may cause the accumulation of reactive drug metabolites, which in turn leads to liver toxicity. The most intensively investigated example for the second category is the association between IDRs and the polymorphism of HLA molecules. For instance, a tight association was reported between carbamazepine-induced Steven-Johnson syndrome and toxic epidermal necrolysis and the HLA-B*1502 allele in Han Chinese but not in Caucasians [94, 95]. HLA molecules play a key role in antigen presentation and initiating an immune response. The fact that the polymorphism leads to different incidences in different ethnic groups suggested that immune systems are involved in these reactions. Pharmacogenetics is perhaps the most informative tool to study the mechanisms of IDRs in clinical settings, and it is an emerging area that may allow us to identify potential biomarkers and establish a standard process for drug screening Rationale of the Present Studies IDR is a complex topic and requires comprehensive understanding of drug metabolism, immunology, genetics, and other fields. In the present studies, this question is tackled from different angles by a variety of methods using the NVP-induced skin rash in female BN rats as a model. We aim to test the PI hypothesis and investigate the role of CD4 + T cells, review risk factors, and determine the initial steps of the immune response induced by the metabolite of NVP. These objectives will be discussed in the following chapters.

44 27 CHAPTER 2 A STUDY OF THE SPECIFICITY OF LYMPHOCYTES IN NEVIRAPINE-INDUCED SKIN RASH Chen, X., et al., A study of the specificity of lymphocytes in nevirapine-induced skin rash. J Pharmacol Exp Ther, (3): p /09/ $20.00 THE JOURNAL OF PHARMACOLOGY AND EXPERIMENTAL THERAPEUTICS Vol. 331, No. 3 Copyright 2009 by The American Society for Pharmacology and Experimental Therapeutics / JPET 331: , 2009 Printed in U.S.A. Reprinted with permission of the American Society for Pharmacology and Experimental Therapeutics. All rights reserved.

45 ABSTRACT Background: NVP treatment can cause a skin rash. We developed an animal model of this rash and determined that the 12 hydroxylation metabolic pathway is responsible for the rash, and treatment of animals with 12-OH-NVP also leads to a rash. Objective: To determine the specificity of lymphocytes in NVP-induced skin rash. Materials and Methods: BN rats were treated with NVP or 12-OH-NVP to induce a rash. Lymph nodes were removed and the response of lymphocytes to NVP and its metabolites/analogs was determined by cytokine production (ELISA, ELISPOT, and Luminex) and proliferation (alamar blue assay). Subsets of lymphocytes were depleted to determine which cells were responsible for cytokine production. Results: Lymphocytes from animals rechallenged with NVP proliferated to NVP but not to 12- OH-NVP or 4-chloro-NVP. They also produced IFN-γ when exposed to NVP, significantly less when exposed to 4-chloro-NVP, and very little when exposed to 12-OH-NVP even though oxidation to 12-OH-NVP is required to induce the rash. Moreover, the specificity of lymphocytes from 12-OH-NVP-treated rats was the same, i.e. responding to NVP more than to 12-OH-NVP even though these animals had never been exposed to NVP. A Luminex immunoassay showed that a variety of other cytokines/chemokines were also produced by NVPstimulated lymphocytes. CD4 + cells were the major source of cytokines. Conclusions: The specificity of lymphocytes in activation assays cannot be used to determine what initiated an immune response. This has significant implications for understanding the evolution of an immune response and the basis of the PI hypothesis.

46 INTRODUCTION NVP is a NNRTI used in the treatment of human immunodeficiency virus infections. Although effective, its use has been limited due to its propensity to cause skin rash and liver toxicity. In patients, skin rashes vary from mild erythematous, maculopapular rashes to more severe Stevens- Johnson syndrome or toxic epidermal necrolysis [96, 97]. Our group discovered a novel animal model of NVP-induced skin rash in rats. The characteristics of NVP-induced skin rash in BN rats are very similar to the milder rashes that occur in humans, which suggests that the mechanisms are very similar. Specifically, in both humans and rats there is a 2-3 week delay between starting the drug and the onset of rash, and on re-exposure, symptoms are more severe and accelerated [54, 55]. Females are more susceptible to developing rash than males in both BN rats and humans. Furthermore, the sensitivity to NVP-induced skin rash can be transferred with CD4 + T cells from NVP-rechallenged rats to naïve recipients [61]. Also, partial depletion of CD4 + T cells delayed and decreased the severity of rash while depletion of CD8 + T cells did not prevent the development of NVP-induced skin rash which fits with the observation that the incidence of rash is lower in patients with a low CD4 + T cell count [61]. There is circumstantial evidence that many IDRs involve reactive metabolites of the drugs rather than the parent drug, but there is rarely definitive evidence for the involvement of a reactive metabolite. We recently demonstrated that the NVP-induced skin rash is not caused by NVP itself but requires 12-hydroxylation of NVP, presumably because the 12-hydroxy metabolite is further converted to a more reactive sulfate in the skin. This conclusion was based on experiments in which the 12-methyl hydrogens were replaced by deuterium, which decreases 12- hydroxylation and rash but does not affect other properties of the drug. Furthermore, treatment with 12-OH-NVP also led to a rash [59]. In the present study we used modifications of the LTT to determine the specificity of lymphocytes from animals with NVP-induced skin rash. The structures of the compounds used in these experiments: NVP, 12-OH-NVP and 4-chloro-NVP (4-Cl-NVP) are shown in Figure 4. A chlorine atom and a methyl group are of approximately the same size and so noncovalent binding of 4-Cl-NVP should be similar to that of NVP but the chlorine blocks oxidation to the 12-OH-NVP and subsequent formation of the sulfate, which is the putative immunogen.

47 30 Another analog, 12-chloro-NVP (12-Cl-NVP), which is more reactive than the sulfate of 12-OH- NVP, was eliminated from later studies because of its cytotoxicity. Figure 4. Structures of NVP, 12-OH-NVP, 4-Cl-NVP and 12-Cl-NVP.

48 METHODS Chemicals. NVP was kindly supplied by Boehringer-Ingelheim Pharmaceuticals Inc. (Ridgefield, CT). 12-OH-NVP [98], 12-Cl-NVP [99] and 4-Cl-NVP [100, 101] were synthesized as previously described. Phosphate-buffered saline (PBS, ph 7.4), fetal bovine serum (FBS), 1640 RPMI-HEPES modified, MEM non-essential amino acids solution and Penicillin- Streptomycin liquid were purchased from Invitrogen Canada, Inc. (Burlington, ON). Dimethyl sulfoxide (DMSO), indomethacin, phorbol myristate acetate (PMA) and inomycin were purchased from Sigma Aldrich (Oakville, ON). ß-mercaptoethanol (2-ME) was purchased from Bio-Rad Laboratories (Canada) Ltd. (Missisauga, ON). Rat CD4 and CD8 MicroBeads were purchased from Miltenyi Biotec (Auburn, CA). Antibodies for flow cytometry studies including anti-rat CD4 phycoerythrin (PE) (mouse immunoglobulin G1 (IgG1)), anti-rat CD8a fluorescein isothiocyanate (FITC) (mouse IgG1), mouse IgG1 PE (isotype control) and mouse IgG1 FITC (isotype control) were purchased from Cedarlane Laboratories (Burlington, ON). Rat cytokine/chemokine Luminex bead immunoassay kit, LINCOplex, 24 Plex, was purchased from Millipore (Billerica, MA). Alamar blue solution was purchased from AbD Serotec (Oxford,UK). IFN-γ and IL-10 enzyme-linked immunosorbent assay (ELISA), IFN-γ enzyme-linked immunosorbent spot (ELISPOT) immunoassay kits and anti-rat CD32 (Fc-gamma receptor II or FcγII) antibody were purchased from BD biosciences (Mississauga, ON). Animal care. Female BN rats ( g) were obtained from Charles River (Montreal, QC) and housed in pairs in standard cages with free access to water and Agribrands powdered lab chow diet (Leis Pet Distributing Inc., Wellesley, ON). The animal room was maintained at 22 o C with a 12:12 hour light:dark cycle. After one week of acclimatization period, the rats were either continued on the same diet (control) or switched to a diet mixed with NVP (treatment group). Primary-treated animals refer to rats that were treated with NVP at a dose of 150 mg/kg/day for 21 days. Rechallenged animals refer to rats that have recovered (4 weeks off drug) from primary treatment and then reexposed to the same dose of NVP for 5 days. The amount of NVP mixed with the diet was calculated based on the body weight of the rats and their daily intake of food. All animals were monitored for the development of red ears, skin rash, food intake, and body weight. At the termination of the experiment, rats were killed by carbon dioxide (CO 2 ) asphyxiation. All of the animal studies were conducted in accordance with the guidelines of the

49 32 Canadian Council on Animal Care and approved by University of Toronto s animal care committee. Preparation of single cell suspension from auricular lymph nodes (ALNs). ALNs were excised and put into Petri dishes containing culture medium (50 ml of FBS, 5 ml of MEM nonessential amino acids, 5 ml of antibiotics, 5 ml of diluted 2-ME (35 µl of 2-ME in 100 ml of distilled water) and 435 ml of 1640 RPMI-HEPES-modified medium. ALN cells were teased out of the nodal capsule using the butt end of a sterile 3 ml syringe plunger and filtered twice through a 40 µm nylon mesh cell strainer (BD falcon). The cell viability was assessed in 0.4% Trypan Blue. Determination of cell proliferation using an alamar blue assay. Single cell suspensions made from control, primary-treated, or rechallenged animals were resuspended at a density of 10 6 cells/ml with 1 µg/ml of indomethacin. They were then plated at 200 µl/well in a 96-well plate. The cells were incubated with various concentrations of NVP, 12-OH-NVP, or 4-Cl-NVP dissolved in DMSO for 72 hours at 37 C in a 5% CO 2 atmosphere. Cells were incubated with equal volume (10µL) of PMA (50 ng/ml in DMSO) and inomycin (500 ng/ml in DMSO) served as a positive control. DMSO alone-treated wells served as a negative control. Alamar blue reagent was added to each well in an amount equal to 10% of the volume in the well (20 µl) at 24 hours during the incubation. Optical density (OD) values at 570 nm and 600 nm were read at 72 hours. Alamar blue reagent incorporates an oxidation-reduction indicator that changes color in response to the chemical reduction of growth medium resulting from cell growth. The difference in its reduction between treated and control wells were calculated following the manufacture s instructions. Luminex, ELISPOT and ELISA assays. Single cell suspensions made from control, primarytreated, and rechallenged animals were incubated in the same manner as in the alamar blue assay except that the cells were plated at 2 ml/well in a 24-well plate for the ELISA and Luminex assays, and the addition of alamar blue was omitted. After 72 hours of incubation, the cell culture supernatant was collected and stored at -20 o C. Quantitation of IFN-γ and IL-10 were achieved by using an ELISA assay and a broad screening of cytokines was performed using a Luminex immunoassay kit from Millipore. The cytokines/chemokines measured were: IL-1α, IL-1β, IL-2, IL-4, IL-5, IL-6, IL-10, IL-12p70, IL-18, granulocyte macrophage colony-

50 33 stimulating factor (GM-CSF), growth-related oncogene (GRO/KC), IFN-γ, monocyte chemoattractant protein 1 (MCP-1), tumor necrosis factor α (TNF-α), IL-9, IL-13, IL-17, Eotaxin, granulocyte colony-stimulating factor (G-CSF), leptin, macrophage inflammatory protein 1α (MIP-1α), interferon gamma-induced protein 10 (IP-10), regulated on activation, normal t cell expressed and secreted (RANTES), and vascular endothelial growth factor (VEGF). In the case of the ELISPOT assay, cells were resuspended at 2 X 10 6 cells/ml and 200 µl was added into each well of the 96 well ELISPOT plate. After 72 hours of incubation, the frequencies of cells that produce IFN-γ were analyzed by an automated enzyme-linked immunospot counter (Cellular Technology Ltd., OH). All the immunoassays were performed by following the manufacture s instructions. Depletion of CD4 + and/or CD8 + lymphocytes. In order to identify which cells produced IFNγ, CD4 + and/or CD8 + cells were depleted from the single cell suspensions by labeling CD4 + and/or CD8 + cells with immunomagnetic microbeads coated with anti-cd4 or anti-cd8a antibodies, followed by passage through a magnetic column. The CD4 + and/or CD8 + depleted cells were resuspended at 10 6 cells/ml and 2 ml of cells were incubated with 12.5 µg/ml of NVP in DMSO in the same way as the other immunoassays described above. To examine the effect of the depletion procedure itself on cytokine production, a control experiment was performed in which some CD4 + and/or CD8 + depleted cells were combined with CD4 + and/or CD8 + cells (labeled by the corresponding antibodies) in the same portion as they were before the depletion and then incubated with NVP. A small portion of cells without any processing was also cultured the same way with NVP as a positive control (before depletion). DMSO alonetreated cells served as negative control. The amount of IFN-γ released into culture medium was determined by ELISA. Flow cytometry. Single cell suspensions before and after the depletion of CD4 + and/or CD8 + cells were surface-labeled, and one- or two-color immunofluorescence analysis was conducted. Briefly, cells were resuspended at a density of 2 X 10 7 cells/ml in PBS/3% FBS and 50 µl of these cell suspensions were aliquoted to wells in a 96-well plate. These cells were first incubated with anti-cd32 antibody for 10 minutes at room temperature to reduce nonspecific binding. Then monoclonal antibodies or suitable isotype controls were aliquoted to the appropriate wells and incubated at room temperature for 20 minutes. The cells were washed twice and finally resuspended in 400 µl of the same buffer. Samples were analyzed immediately with a FACS-

51 34 Calibur (Becton Dickinson) using the CellQuest software (Becton Dickinson). FlowJo (Tree Star, Inc.) was used to analyze the difference in CD4 + and/or CD8 + cell populations between depleted and undepleted cells.

52 RESULTS ALNs Cell Specificity as Determined by IFN-γ Production. Cytokine production was used as a measure of T cell activation. It was found that IFN-γ was a sensitive marker of cell activation, and the addition of indomethacin to the culture medium made the assay more sensitive and consistent, presumably because it decreased the inhibitory effects of prostaglandin E production. Drug-specific IFN-γ secretion from lymphocytes gradually increased from 6 hours until reaching maximal concentrations on day 3; no IFN-γ could be detected in cultures in the absence of NVP or in cultures of ALN cells from animals after 21 days of primary NVP exposure in the presence of NVP (data not shown). Analysis of cell culture supernatants after 3 days showed that ALN cells from secondary treatment rats incubated in the presence of NVP or 4-Cl-NVP produced IFN-γ with only a minimal response to 12-OH-NVP and no response to 12-Cl-NVP (Figure 5). 12-Cl-NVP was found to be cytotoxic causing morphologic changes in the cells; therefore, it was excluded from the rest of the studies. Likewise, ALN cells from animals in which the rash was induced by 12-OH-NVP and rechallenged with 12-OH-NVP also produced IFN-γ on exposure to NVP with a smaller response to 4-Cl-NVP and even less to 12-OH-NVP (Figure 5). Again, no IFN-γ production was detected in cells from animals treated for 21 days on primary exposure to 12-OH-NVP (data not shown). The production of IL-10 by these lymphocytes was also quantified and showed similar results. Independent of whether the rash was induced by NVP or 12-OH-NVP, lymphocytes always produced IL-10 on exposure to NVP with a smaller response to 4-Cl-NVP and even less to 12-OH-NVP (Figure 6). The IFN-γ ELISPOT assay showed the same specificity as the ELISA assay. For both control and primary-treated animals, no response was observed (data not shown). The frequency of cells from NVP-rechallenged animals that respond to NVP was approximately 1:4,000 compared to 1:40,000 for the frequency of cells responding to 4-Cl-NVP. Virtually no cells responded to 12- OH-NVP (Figure 7). ALN Cell Specificity as Determined by Cell Proliferation. The alamar blue assay was used to measure cell proliferation in response to NVP, 12-OH-NVP, and 4-Cl-NVP. No increase in proliferation was detected in cells from animals after primary exposure to NVP (data not shown). For rechallenged animals, increased proliferation was detected in NVP-stimulated cells, which was maximal at a NVP concentration of 6.25 µg/ml (Figure 8). No proliferation was detected in

53 36 response to 12-OH-NVP or 4-Cl-NVP. Higher concentrations of all drugs appear to be toxic and led to less reduction of the alamar blue reagent compared to control wells. Screening Cytokines/Chemokines Production by Lymphocytes. The Luminex immunoassay was used to screen for the production of 24 cytokines/chemokines by lymphocytes. In response to NVP, lymphocytes from primary-treated animals produced increased levels of IL-6, IL-10, IL- 17, IL-18, GMCSF, GRO/KC, RANTES, and MIP-1α (Figure 9). In contrast, rechallenged animals had increased production of IL-17, GMCSF, GRO/KC, MIP-1α, TNF-α, IL-10, IL-18, RANTES, and IFN-γ. In general, cells from rechallenged animals produced higher levels of cytokines/chemokines than cells from primary-treated animals; however, their cytokine profiles were slightly different. Primary-treated animals produced more IL-18 and produced IL-6, which was not observed in rechallenged animals. In contrast, rechallenged animals produced TNF-α as well as a large amount of IFN-γ, which were absent in cells from primary-treated animals (Figure 9). Stimulation of cells from both primary and rechallenged animals with NVP actually appeared to decrease the basal production of MCP-1. Production of IFN-γ by CD4 + and/or CD8 + Depleted Lymphocytes To identify which cells were responsible for cytokine production, CD4 + and/or CD8 + cells were depleted from ALN cells and then cultured with NVP. Flow cytometry was used to determine the degree of depletion (Table 3). When only CD4 + cells were depleted, the production of IFN-γ was decreased to levels similar to the negative control (Figure 10). When both CD4 + and CD8 + cells were depleted simutaneaously, IFN-γ production was virtually eliminated. However, when ALN cells were depleted of CD8 + cells the production level of IFN-γ was only slightly decreased. There was no significant difference in the production level of IFN-γ between undepleted lymphocytes and lymphocytes combined after depletion procedure.

54 37 Figure 5. IFN-γ secretion by lymphocytes in response to NVP, 12-OH-NVP, 4-Cl-NVP and 12-Cl-NVP. (A) lymphocytes from NVP-treated animals (n=6), (B) lymphocytes from 12-OH-NVP-treated animals (n=4). CON indicates control animals; for example, CON/NVP indicates that the lymphocytes were isolated from untreated control animals and incubated with NVP. The data are expressed as the mean STD.

55 38 Figure 6. IL-10 secretion by lymphocytes in response to NVP, 12-OH-NVP, 4-Cl-NVP and 12-Cl-NVP. (A) lymphocytes from NVP-treated animals (n=6), (B) lymphocytes from 12-OH-NVP-treated animals (n=4). CON indicates control animals; for example, CON/NVP indicates that the lymphocytes were isolated from untreated control animals and incubated with NVP. The data are expressed as the mean ± STD.

56 39 Figure 7. Frequencies of lymphocytes responding with the production of IFN-γ when stimulated with NVP or its analogs/metabolites from NVP-rechallenged rats using an ELISPOT assay. Cells incubated with a PMA/inomycin mixture served as a positive control while DMSO alonetreated well served as a negative control. The numbers at the left corner of each well represents the number of cells responding to the drug out of a total of 0.4 million cells/well.

57 40 Figure 8. Cell proliferation in response to NVP, 12-OH-NVP and 4-Cl-NVP as determined by the reduction of alamar blue. (A) lymphocytes from untreated control animals, (B) lymphocytes from NVP rechallenged animals. The reduction of alamar blue in control wells (DMSO alone-treated) was set as 1. The percentage difference in reduction of alamar blue in NVP and its metabolites and/or analogtreated wells as compared to the control wells was plotted against the concentration of these drugs in log scale and a decreasing manner. The data represent the mean from 4 rats STD. Statistical significance between treated samples and control samples was determined using the Mann Whitney test; values of p 0.05 were considered statistically significant.

58 41 Figure 9. Concentration of cytokines/chemokines produced by lymphocytes from control (n=4), primary treated (n=6), and rechallenged (n=7) animals in response to NVP as determined by a Luminex assay. The data are separated into panels A and B based on their different range of concentrations. The levels are expressed as the mean SEM. Statistical significance between treatment group and control group was determined using the Mann Whitney test; values of p 0.05 were considered statistically significant.

59 42 Figure 10. Production of IFN-γ in response to NVP by lymphocytes from rechallenged rats before and after depletion of CD4 + and/or CD8 + T cells. Negative control was DMSO alone-treated cells. Combined samples were obtained by combining CD4 + and/or CD8 + depleted cells with CD4 + and/or CD8 + cells (labeled by the corresponding antibodies) in the same portion as they were before the depletion and then incubated with NVP. The data represent the mean from 4 rats SE. Statistical significance between depletion groups and negative control group was determined using the Mann Whitney test; values of p 0.05 were considered statistically significant.

60 43 Table 3. Percentage of cell types before and after the depletion of lymphocyte subsets with magnetic beads. Rat Cell type Before CD4 + CD8 + CD4/CD8 + investigated depletion (%) depletion (%) depletion (%) depletion (%) 1 CD CD CD CD

61 DISCUSSION In the present studies we determined the molecular specificity of lymphocytes from NVPrechallenged animals. Detection of drug-specific IFN-γ secretion by lymphocytes proved to be a sensitive method for the detection of lymphocyte activation. A recent study of a patient with a history of NVP-induced liver toxicity found that the patient s T cells proliferated in response to NVP but not to 12-OH-NVP, 2-OH-NVP, or descyclopropyl-nvp; they did not test the 4-Cl analog [102]. In another study, T cells from a patient with NVP-induced skin rash responded to incubation with NVP by producing IFN-γ (Keane NM et al., abstract MOPEB007, 4 th International AIDS Society Meeting, Australia, 2007). These clinical findings provide further evidence that the immune response in the animal model is similar to that in NVP-induced idiosyncratic reactions in humans. The finding that CD4 + T cells were the source of IFN-γ is also consistent with the observation that depletion of these cells in both the animal model and humans is protective where, at least in the animal model, depletion of CD8 + T cells actually seemed to make the rash worse. Other cytokines were also produced and the pattern was different in cells from animals after primary exposure to NVP than those from rechallenged animals. The rash in animals is mild on initial exposure to NVP, but the reaction is systemic with weight loss and a widespread infiltration of lymphocytes in the skin on rechallenge so it is not surprising that the response and number of responding cells was markedly greater in cells from rechallenged animals [67]. A quite surprising finding was that there was a complete disconnect between what induced the skin rash and the specificity of the T cells. Not only do we know from independent experiments that oxidation of NVP to 12-OH-NVP is required to cause a skin rash, but even when the skin rash was induced by treatment with 12-OH-NVP and the animals had not been exposed to NVP (12-OH-NVP is not converted to NVP), their T cells responded much better to NVP than to 12- OH-NVP. This has significant implications for understanding how an immune response evolves. It is likely that induction of this immune response requires covalent binding of a reactive metabolite NVP derived from 12-OH-NVP because this results in modified protein (Hapten Hypothesis) and/or because it causes cell damage (Danger Hypothesis). However, once induced, there is much more drug present than modified protein and it is possible that some T cells crossreact with parent drug and proliferate, i.e. epitope spreading [12]. NVP and 4-Cl-NVP are more

62 45 lipophilic than 12-OH-NVP and this may lead to higher affinity binding. The basis for why there is a disconnect between what induces the immune response and the specificity of the T cells is speculation; however, what is clear is that the specificity of T cells cannot be used to determine what induced the immune response. These findings also have implications for the PI hypothesis. This hypothesis was originally based on the LTT results from patients with a history of an idiosyncratic reaction to SMX whose cloned T cells showed a response to the parent drug in the absence of metabolism [103]. The unstated assumption is that what T cells respond to is what induced the immune response which no longer can be considered a valid assumption. This does not mean that the p-i hypothesis is wrong in all cases; there are IDRs such as ximelagatran-induced liver failure in which a reactive metabolite does not appear to be involved. Ximelagatran is structurally similar to a small peptide and may be able to initiate an immune response through a p-i type of interaction; in fact there is evidence that it can bind directly but reversibly to MHC, specifically DRB1*07 and DQA1*02 [104]. However, our study does have significant implications for the interpretation of LTTs. Acknowledgments. We thank Boehringer-Ingelheim for their supply of nevirapine. Dr. Jack P. Uetrecht is the recipient of the Canada Research Chair in Adverse Drug Reactions.

63 46 CHAPTER 3 INDUCING SKIN RASH IN FEMALE BN RATS BY TOPICAL TREATMENT OF NVP AND/OR 12-OH-NVP

64 INTRODUCTION 12-OH-NVP is a major metabolite of NVP, and the 12-OH-NVP pathway was proven to be responsible for the skin rash [59]. Moreover, we recently found that it is 12-OH-NVP sulfate that is responsible for covalent binding in the epidermis and the skin rash that occurs when rats are treated with NVP. Specifically, topical treatment of 1-phenyl-1-hexanol, a sulfotransferase inhibitor, prevented both covalent binding where it was applied and the rash [68, 69]. These results indicate that 12-OH-NVP sulfate formed in the skin is responsible for the skin rash, and the proposed chemical activation of NVP is shown in figure 11. Although the culprit metabolite has been identified, the mechanism of how the immune response was elicited is yet to be fully defined. In addition to other attractive hypotheses, such as the danger hypothesis, the data suggest that the initiation of this IDR occurs in the skin. The fact that the covalent binding of 12-OH-NVP sulfate formed in keratinocytes causes the rash suggested that NVP may share a similar mechanism to that of allergic contact hypersensitivity. In the case of contact hypersensitivity, the chemical allergens covalently bind to proteins derived from resident cells in the skin (i.e. keratinocytes, Langerhans cells, and dendritic cells, etc.). The Langerhans and/or dendritic cells then travel to the local draining lymph nodes and present the antigens to T cells, which leads to an immune response. It was found that the reactivity of chemicals and their ability to covalently bind to proteins (hapten formation) are linked to their immunogenicity and sensitization, and this led to tests to identify potential sensitizers [ ]. If this hypothesis is right, it should be possible to induce a skin rash by topical treatment with 12- OH-NVP Topical Treatment of NVP on Sensitized Animals We first started this experiment with sensitized animals because on rechallenge, the immune response is much more vigorous than that of the first-time sensitization, and the onset is also accelerated. When NVP-sensitized female BN rats are rechallenged orally with NVP, the onset of the disease is accelerated and more severe (the animals develop red ears in 1 day and skin rash in 1 week). The dose (oral treatment) needed to induce an immune response is as low as 1/30 of that needed

65 48 for primary treatment. We have also done ear patch tests on pre-sensitized animals, and the results are similar to the oral treatment studies [67]. The ear patch tests also showed a systemic reaction: when one ear of an animal was painted with NVP or 12-OH-NVP (the major metabolite which is responsible for the skin rash), both ears turned red within 24 hours. However, we did not test if the rat would develop skin rash after prolonged treatment. One thing that should be emphasized is that, although NVP needs to be biotransformed to 12-OH-NVP sulfate in the skin to initiate the immune response, the parent drug can induce the allergic reaction on rechallenge without metabolism. It was demonstrated in chapter 2 that the lymphocytes from sensitized animals can recognize and respond to the parent drug in the absence of metabolism on rechallenge. Therefore, the use of 12-OH-NVP sulfate (the culprit metabolite) or 12-OH-NVP, which can be sulfated and form the culprit metabolite by skin resident cells, is not necessary for topical treatment on rechallenge. In our animal model, although the development of red ears is used as the earliest sign of NVPinduced IDRs, we also demonstrated that the redness was a result of vascular dilation. Our goal is to understand the mechanisms of the immune responses in the skin, and to eliminate possible interference of the vascular effect; therefore, we repeated the patch tests and used histology to evaluate the response Attempts to Induce a Skin Rash in Naïve Rats by Topical Treatment with 12-OH-NVP With the success of inducing skin rashes in pre-sensitized animals, we wanted to determine if this would work with naïve animals. The sequence of events is that NVP is first oxidized in the liver to 12-OH-NVP, which goes to the skin where sulfotransferase in keratinocytes forms the reactive sulfate. Therefore, topical administration of NVP would not be expected to cause a rash. If our hypothesis that the skin is where the immune response is initiated and the mechanism is similar to that of contact hypersensitivity is correct, a direct topical 12-OH-NVP treatment should be able to induce the skin rash.

66 49 Figure 11. Proposed chemical mechanisms of NVP-induced skin rash by formation of 12- OH-NVP sulfate in the skin.

67 MATERIALS AND METHODS Chemicals. NVP was kindly supplied by Boehringer-Ingelheim Pharmaceuticals Inc. (Ridgefield, CT). 12-OH-NVP [98] was synthesized as previously described. Acetone was purchased from Sigma Aldrich (Oakville, ON). Animal care. Female BN rats ( g) were obtained from Charles River (Montreal, QC) and housed in pairs in standard cages with free access to water and Agribrands powdered lab chow diet (Leis Pet Distributing Inc., Wellesley, ON). The animal room was maintained at 22 o C with a 12:12 hour light:dark cycle. After a one-week acclimatization period, the rats were either continued on the same diet (control) or switched to a diet mixed with NVP (treatment group). Primary-treated animals refer to rats that were treated with NVP at a dose of 150 mg/kg/day for 21 days. Rechallenged animals refer to rats that have recovered (4 weeks off drug) from the primary treatment and then reexposed to the same dose of NVP for 5 days. The amount of NVP mixed with the diet was calculated based on the body weight of the rats and their daily intake of food. All animals were monitored for the development of red ears, skin rash, food intake, and body weight. At the termination of the experiment, rats were killed by CO 2 asphyxiation. All of the animal studies were conducted in accordance with the guidelines of the Canadian Council on Animal Care and approved by University of Toronto s animal care committee. Inducing skin rash in rechallenged animals by topical treatment with NVP. Six female BN rats received NVP in food (150 mg/kg/day) for 21 days or until the skin rash developed, and then the drug was removed for at least 1 month to allow the animals to recover. After that, part of the skin on the back of three rats was shaved (approximately 3 cm X 3 cm) and painted with NVP (2.5 mg/ml in acetone/olive oil, 1:1/v:v). The others were treated with vehicle solution on a daily basis. As a control for the effects of shaving, an additional area of skin was shaved distant from where the drug or vehicle solution was applied. Attempts to induce skin rash in naïve animals by topical treatment with 12-OH-NVP. To test this hypothesis, 8 female BN rats were shaved on the back (approximately 3 cm X 3 cm). Two of them were painted with 100 µl of 12-OH-NVP suspension in acetone/olive oil (1:1/v:v) at a dose of 1.5 mg/kg/day; 2 were painted with 25 µl of 12-OH-NVP in DMSO at a dose of 15 mg/kg/day. The other 4 control animals were treated with the vehicle used in the treatment groups.

68 RESULTS Topical treatment of NVP-induced skin rash in sensitized animals. The rats that received NVP applied topically to the back developed red ears within 24 hours and skin lesions in the painted area in less than 1 week. Due to the irritation and burn caused by the solvent on the lesion area, the treatment was stopped at approximately day 5. The vehicle control areas in treated animals did not have apparent lesions, but the skin surface was dry and exfoliated. The animals that received vehicle alone had no discomfort or symptoms at all, and the skin looked normal. The vehicle alone appears to cause some edema and infiltration of a few lymphocytes, but the histology is not significantly different from other areas on the same control rats (Figure 12. A and B). The control area of NVP-treated rats (Figure 12. C) showed more lymphocyte infiltration in the dermis and epidermis. It also had a thickened epidermal layer but few dead keratinocytes. The histology of the skin where the topical NVP was applied (Figure 12. D-F) showed massive lymphocyte infiltration everywhere, i.e. epidermis, dermis, and subcutaneous tissue. There was also an infiltration of eosinophils and neutrophils as well as dead keratinocytes.

69 52 A) B) C) D) E) F) Figure 12. H&E staining of skin samples from rats rechallenged with topical NVP (2.5 mg/ml in acetone/olive oil, 1:1/v:v) or vehicle only. A) control rat, control area, 20X magnification; B) control rat, vehicle-application area, 20X magnification; C) treated rat, vehicle-application area. 20X magnification; D-F) treated rat, NVP-application area, 20X, 10X and 100X magnification, respectively.

70 53 Topical treatment with 12-OH-NVP did not induce a skin rash in naïve animals. The naïve animals that were treated for 6 weeks with topical 12-OH-NVP had no evidence of skin rash nor did they appear to have any discomfort. For rats that were treated with 12-OH-NVP in acetone/olive oil (1:1/v:v) at a dose of 1.5 mg/kg/day, the histology (hematoxylin and eosin or H&E staining) of the painted skin showed no difference between control and treated animals for either control or treated areas (results not shown). This may be due to the very limited solubility of 12-OH-NVP in this vehicle solution and the drug seemed to precipitate out on the surface of the skin once the acetone had evaporated. Therefore the amount of absorption may have been minimal. On the other hand, DMSO penetrates skin very well, and it was also the best solvent for 12-OH-NVP by far that we tried. It seemed that it delivered the drug well and no precipitation was observed. Figure 13 shows that DMSO alone caused some skin edema (thicker epidermis) as well as limited infiltration of lymphocytes in the epidermis and dermis compared to the control area on the control rats (A). In 12-OH-NVP-treated rats, there seems to be more lymphocytes in the dermal and subcutaneous areas. An eosinophil infiltration was also observed (C-E). However, overall the histological changes were minimal, and the difference between drug-treated animals and control animals were not significant.

71 54 A) B) C) D) E) Figure 13. H&E staining of skin samples from naïve rats treated with topical 12-OH-NVP (15 mg/kg/day in DMSO) or vehicle only. A) control rat, control area; B) control rat, DMSO-application area; C) treated rat, DMSOapplication area; D-E) treated rat, 12-OH-NVP-application area. A-D): 10X magnification; E) 20X magnification.

72 DISCUSSION Even though there is not much P450 in the skin, and NVP must to be metabolized to 12-OH- NVP in order to induce a skin rash in naïve animals, we were still able to induce a skin rash by topical treatment of the parent drug in sensitized animals. The reason is likely due to the epitope spreading effect discussed in Chapter 2. Once the immune response is initiated, the lymphocytes will cross-react and respond to both the parent drug and 12-OH-NVP. Therefore, the parent drug is enough to trigger the immune response on rechallenge. The fact that we were able to induce a skin rash by topical treatment of NVP suggested that, at least on rechallenge, the immune response is initiated in the skin. The question is then which cells are responsible for this reactivation and what is the mechanism. In contact hypersensitivity, dendritic cells (CD11c + ) and Langerhans cells (CD207 + ) are the most common antigen presentation cells in the skin [109, 110]. They migrate to the draining lymph nodes and maybe the spleen once being activated by antigens. The signals are then presented to T cells in the secondary lymphoid organs and an immune response may be elicited. On the other hand, T cells expressing skin homing receptors could also be attracted by skin residential cells expressing the corresponding ligands, and migrate to the local inflammatory site, then induce an immune response. [111, 112]. In addition to the cell response to the drug, the mechanism involved in inducing the systemic reaction by local topical treatment of sensitized rats may also contain an autoimmune component. Although it is possible that there is sufficient absorption through the skin to produce a significant circulating concentration of NVP, if the amount applied to the skin is decreased, eventually no rash is induced, but there is no dose at which there is only a local response. We have recently found the presence of autoantibodies in the sera of rechallenged animals, which further supports the autoimmune component (Amy Sharma, unpublished observations). Further characterization of the autoantibodies such as the time course and classification will be determined. Although no skin rash was induced by topical treatment of 12-OH-NVP in naïve animals, the fact that the absolute low dose of drugs/metabolites which will get absorbed by the skin is limited and may not be enough to induce an IDR. The histology results from primary exposure experiments also suggested the possibility that 12-OH-NVP may have induced some immune response although the changes were not that significant. Therefore this study does not directly

73 56 prove our hypothesis wrong. However, it is also possible that the successful induction of an immune response in the case of NVP requires more than just the local reaction in skin, and the only way to fully understand the mechanism is to vigorously test each hypothesis.

74 57 CHAPTER 4 FACTORS THAT MAY INFLUENCE THE INCIDENCE AND SEVERITY OF NVP-INDUCED SKIN RASH IN FEMALE BN RATS

75 INTRODUCTION NVP (Viramune) is a NNRTI used in combination therapy to treat human immunodeficiency virus-1 (HIV-1) infections. Soon after being marketed, it was found to cause skin rash and liver toxicity including Stevens-Johnson syndrome and toxic epidermal necrolysis [53]. Early studies showed the incidence of skin rash was 16%, and that of clinically evident hepatotoxicity was 1% [53]. However, the incidence is lower at present. Starting patients at a lower dose (200 mg once daily) for two weeks followed by the full does (200 mg twice daily) reduces the incidence of rash, while treatment with steroids increases the incidence of rash [113]. Also, female patients are more susceptible to skin rash than males [57]. We found that NVP also causes a skin rash in female BN rats and developed a novel animal model of drug-induced idiosyncratic skin rash [54]. Specifically, when female BN rats were fed a mash diet containing NVP at the dose of 150 mg/kg/day, they developed red ears in about 7 days and skin rash in days with an incidence of 100% [54]. The incidence in female Sprague Dawley rats was 20%, whereas none of the male rats of either strain developed a rash. Besides the fact that females are associated with higher incidence in both human and our model, the characteristics of the skin rash are similar as well. For instance, the onset of rash occurs 2-3 weeks after NVP treatment and the syndrome resolves when the treatment is discontinued. Upon the rechallenge with NVP, the onset of rash is accelerated. A 2-week low dose (40 mg/kg/day) followed by the full dose (150 mg/kg/day) NVP also prevented the rash in female BN rats [54]. A low CD4 + T cell count decreases the risk of rash in humans; similarly, partial depletion of CD4 + T cells delayed the rash in female BN rats. The characteristics of NVP-induced skin rash in female BN rats suggest an immune-mediated mechanism, and we believe it is similar to that in humans. In the past few years we have used a variety of modulators to manipulate the incidence/severity of NVP-induced skin rash to better illustrate the mechanisms of this IDR. For instance, we have shown that a partial depletion of CD4 + cells by an anti-cd4 antibody is protective suggesting that CD4 + cells play an important role in this IDR model [54]. On the other hand, a partial depletion of CD8 + cells seemed to make the rash worse. The immunosuppressants, cyclosporine and tacrolimus, also prevented NVP-induced skin rash and the associated increase in serum IgE, which suggests that NVP-induced skin rash is immune-mediated [54]. Agents that were

76 59 demonstrated to change the severity/incidence in another immune-mediated IDR model (penicillamine-induced autoimmunity in the BN rats) including poly I:C, misoprostol, and aminoguanidine, had no effect on NVP-induced skin rash [61, 83]. The first sign of NVPinduced skin rash was red ears, which suggested the involvement of histamine and/or serotonin, and this led us to test a few anti-allergic drugs including cromolyn, ketanserin, and astemizole. However, none of these agents had any effect on this IDR [61]. In this paper, we continued to explore agents that may be able to manipulate the incidence and/or severity of this IDR with the aim to better understand the mechanism. Previous studies have shown that the number of B cells increased much more than any other cells (CD4 + T cells, CD8 + T cells, and macrophages) in ALNs of NVP-treated animals [67]. The number of B cells expressing MHC II also significantly increased. In addition, there was a spike of IgE on day 7 following NVP treatment [67]. All this evidence suggested that B cells may play an important role in our model, and the best way to test this hypothesis is to deplete B cells in vivo to see if it affects the NVP-induced skin rash. Rituximab has been used clinically to deplete B cells for the treatment of many autoimmune diseases including multiple sclerosis and rheumatoid arthritis. Because of its murine/human chimeric nature, it probably will not recognize rat CD20 molecules; however, the anti-mouse CD20 antibody from Genetech Inc. seemed to work in rats and was used to deplete B cells in the present study. Glutathione reacts with electrophilic reactive intermediates derived from exogenous chemicals such as drugs, and glutathione conjugation is an important detoxification pathway. It has been shown that NVP forms a glutathione adduct at the 12 position in human liver microsomes and cytochrome P450 3A4 (CYP3A4) cultures; therefore, it may represent a protective pathway [114]. In the present study buthionine sulfoximine (BSO) was used to deplete glutathione to determine if this would lead to a more severe reaction. Retinoic acid (RA) is a vitamin A metabolite used to treat acne. It is also reported to inhibit Th17 differentiation while promoting regulatory T cell differentiation [115]. In the D- penicillamine-induced autoimmune disease model, in which we think Th17 cells are involved, RA significantly increased the incidence [82]. In contrast, a small (N=2) pilot study showed that RA significantly delayed the onset of NVP-induced skin rash. These preliminary findings led to the investigation of the effect of RA on NVP-induced skin rash.

77 60 Indoleamine dioxygenase (IDO), an intracellular hemoprotein enzyme which catalyses degradation of the essential amino acid tryptophan, appears important to promote immune tolerance induction [116]. The levo form of 1-methyl-tryptophan was reported to inhibit IDO produced by human dendritic cells [117]. Our previous studies have shown a role of regulatory T cells in NVP-induced skin rash (unpublished observation). Treating animals with 1-methyltryptophan might break the tolerance induced by regulatory T cells and result in an earlier onset or a more severe disease or even liver toxicity. Imiquimod and LPS can induce immune responses through toll-like receptors in a non-specific manner. LPS is a major component of the outer membrane of Gram-negative bacteria; it enhances an immune response by stimulating APCs through toll-like receptor 4 and upregulating costimulatory molecules [118]. In another animal model of IDRs, D-penicillamine induced lupus-like syndrome in male BN rats, LPS was able to reverse tolerance in a small percentage of animals [77]. It was interesting to see if LPS would affect the disease progression of NVPinduced skin rash as well. Imiquimod stimulates APCs through toll-like receptor 7 and can induce the synthesis of IFN- and other cytokines in a variety of cell types [119]. Treatment of imiquimod may accelerate the onset of NVP-induced skin rash or make the disease worse. Vitamin D deficiency is associated with higher risk of drug-induced hypersensitivity syndrome and might be a risk factor in the NVP model [120]. It is much easier to treat rats with Vitamin D than to deplete it in vivo, so animals were dosed with vitamin D to see if it has any effect.

78 MATERIALS AND METHODS Chemicals. NVP was kindly supplied by Boehringer-Ingelheim GmbH. Anti-mouse CD20 antibody 5D2 (mouse IgG2a) at a concentration of 17.2 mg/ml was kindly supplied by Genetech. PBS (ph 7.4), and FBS were purchased from Invitrogen Canada, Inc. (Burlington, ON, Canada). Anti-rat CD45RA/B PE antibody (mouse IgG1) and anti-mouse IgG1 PE antibody (isotype control) were purchased from Cedarlane Laboratories (Burlington, ON, Canada). Antirat CD32 antibody was purchased from BD Pharmingen (Mississauga, ON, Canada). Rat IgE ELISA kit was purchased from GenWay (Hayward, CA, USA). Microvette ethylenediaminetetraacetic acid-coated (EDTA-coated) tubes for blood samples were purchased from Sarstedt (Montreal, QC, Canada). Ammonium chloride (NH 4 Cl), potassium bicarbonate (KHCO 3 ), ethylenediaminetetraacetic acid (EDTA), dextran, RA, LPS, and 1-methyl-tryptophan were purchased from Sigma-Aldrich (Oakville, ON, Canada). L-buthionine - (S, R) - sulfoximine and imiquimod were purchased from Toronto Research Chemicals Inc. (North York, ON, Canada). Vitamin D3 was purchased from BioShop Canada Inc. (Burlington, ON, Canada). Rat cytokine/chemokine Luminex bead immunoassay kit, LINCOplex, 24 Plex, was purchased from Millipore (Billerica, MA). Glutathione assay kit was purchased from BioVision (Nountrain View, CA, USA). Animal care. Female BN rats ( g) were obtained from Charles River Canada (Montreal, QC, Canada) and housed in pairs in standard cages with free access to water and Agribrands powdered lab chow diet (Cargill, Inc., Minneapolis, MN). The animal room was maintained at 22 C with a 12:12 h light/dark cycle. After one week of acclimatization, the rats were either continued on the same diet (control) or switched to a diet mixed with NVP (treatment group). The amount of NVP mixed with the diet was calculated based on the body weight of the rats and their daily intake of food. The treatment period was 21 days with a daily dose of approximately 150 mg/kg body weight unless otherwise specified. For B cell depletion experiments, the antimouse CD20 antibody was injected into the tail vein with various dosing schedules. Vehicle control animals received PBS injections. All animals were monitored for the development of red ears, skin rash, food intake, and body weight. At the termination of the experiment, rats were killed by CO 2 asphyxiation. All of the animal studies were conducted in accordance with the Guidelines of the Canadian Council on Animal Care and approved by University of Toronto s animal care committee.

79 62 Cotreatment of NVP and anti-cd20 antibody. 1. Testing the efficacy of anti-mouse CD20 antibody in rats. First of all, the efficacy of this anti-mouse antibody in rats was tested. Three female BN rats received 3 different doses of anti-mouse CD20 antibody by i.v. injection (1 mg, 2 mg or 3 mg per rat). To determine the B cell levels in various tissues, anti-rat CD45RA/B was used for flow cytometry due to the lack of an anti-rat CD20 antibody. 2. Depleting B cells with various dosing regimens of anti-mouse CD20 antibodies. Five different dosing regimens with various injection schedules were tested to see if any of them would change the incidence or severity of NVP-induced skin rash. Twenty eight female BN rats were divided into 7 groups, and their corresponding dosing schedules are shown in Table 4. All groups except the control were treated with NVP from Day 0 to 22 at a dose of 150 mg/kg/day. Each injection of anti-cd20 antibody was 1 mg/rat. 3. Effect of B cell depletion on NVP-induced skin rash in splenectomized rats. The antibody did not deplete B cells in the spleen and so another group of animals was splenectomized to remove this B cell refuge. The rats were anesthetised with isoflurane prior to surgery. The hair was removed from the surgical site and the skin washed with iodine and alcohol. A small incision (2-3 mm long) was made at the lower left abdominal, the spleen was then identified and removed after ligating the vessels. The abdominal musculature was closed with 3-0 interrupted Vicryl sutures and the skin closed with 3-0 Vicryl interrupted sutures. The surgery was performed using sterile technique. After the surgery, animals were heated to keep them warm and monitored for 30 minutes before moving them back to normal cages. In the splenectomy experiments, all animals receiving surgery were closely monitored for their well being and allowed to recover before they were started on another treatment. To test the effect of B cell depletion on NVP-induced skin rash in splenectomized rats, three female BN rats received a splenectomy and were allowed to recover for 2 weeks. They were then treated with NVP mixed in mash diet at a dose of 150 mg/kg/day. Meanwhile, another 3 healthy rats received the same NVP diet and served as a control. On day 5, the splenectomized rats started to develop pink ears, which were not very obvious until day 6 or 7. The control group had developed red ears by the end of day 7. It seems that splenectomy had little effect on the onset (not significant) and no effect on severity of NVP-induced skin rash. Therefore, a controlled study was performed to see if depleting B cells would make a difference. This study involved 5

80 63 groups of rats with 2 in each group. All splenectomised animals were allowed to recover for 1 month before receiving any treatments. The experimental design was outlined in Table 4. Preparation of single-cell suspension from peripheral blood, ALNs, and spleen. Blood was drawn from the tail vein and mixed with an equal volume of saline with 3% of dextran and left at room temperature for 20 minutes. The top straw-colored layer was obtained and mixed with 1 ml of red cell lysis buffer (155 mm NH 4 Cl/10 mm KHCO 3 /0.1 mm EDTA) and left at room temperature for 10 minutes. The cell suspensions were then centrifuged at 1800 rpm for 10 minutes at 4 C. The cell pellet was obtained and resuspended in staining buffer (PBS/3% of FBS) and proceeded to flow cytometry analysis. ALNs were excised and put into staining buffer. ALN cells were teased out of the nodal capsule by using of the butt end of a sterile 3-mL syringe plunger and filtered twice through a 40 µm nylon mesh cell strainer (BD falcon, BD Biosciences). The cell viability and concentration were assessed in 0.4% trypan blue. The splenic cells were prepared in the same way as for the ALNs except that an additional step of red cell lysis was required. Flow cytometry. Single-cell suspensions were surface-labelled, and single immunofluorescence analysis was conducted. In brief, cells were resuspended at a density of 2 x 10 7 cells/ml in staining buffer, and 50 µl of these cell suspensions were aliquoted to wells in a 96-well plate. These cells were first incubated with anti-rat CD32 antibody for 10 minutes at room temperature to reduce nonspecific binding. Then, monoclonal antibodies or suitable isotype controls were aliquoted to the appropriate wells and incubated at room temperature for 20 minutes. The cells were washed twice and finally resuspended in 400 µl of the same buffer. Samples were analyzed immediately with a FACS-Calibur (BD Biosciences) with use of CellQuest software (BD Biosciences). FlowJo (Tree Star, Inc., Ashland, OR) was used to analyze the stained population. Plasma IgE levels. The blood was drawn from the tail vein and IgE levels were determined using the protocol included in the ELISA kit. Cotreatment of BSO and NVP in naïve animals. Sixteen female BN rats were divided into 4 groups with 4 animals in each group and treated with the experimental plan described in Table 6. For metabolic studies, animals in the NVP and BSO-NVP groups were put in metabolic cages one day a week for urine collection, i.e. on days 7, 14, and 21. To avoid the contamination of

81 64 urinary samples with food and ensure adequate uptake of the interventional substances, when they were in metabolic cages, all rats received NVP by gavage and had free access to regular food and tap water. The gavaging dose was pre-optimized based on food and water intake: NVP and BSO were suspended in corn oil at a dose of 75 mg/kg/day and 250 mg/kg/day, respectively. Both drugs were gavaged twice a day with a 6 hour interval. Control and BSO rats received the same amount of corn oil as vehicle control on that day. Plasma samples were taken right before the first dose, and urine samples were collected at the end of the 24 hours. All rats were monitored for red ears and skin rash on a daily basis and euthanized on day 22. Concentrations of NVP and its major metabolites (12-OH-NVP, 3-OH-NVP, 4-COOH-NVP, 2-OH-NVP) were determined using liquid chromatography-mass spectrometry (LC/MS) using the previously reported method by our group [59]. Total glutathione levels in the liver were determined following instructions of the glutathione assay kit. Cotreatment of BSO and NVP in sensitized animals. Eight female BN rats were treated with NVP at a dose of 150 mg/kg/day in food for 21 days followed by 4 weeks of recovery period, then rechallenged with NVP at the same dose for 7 days. Four of these rats received BSO at a dose of 500 mg/kg/day (with 2% of glucose) starting 1 week prior to rechallenge until the end of this study. All animals were monitored for red ears and skin rash. The levels of NVP and its metabolites in the plasma and urine were quantified using LC/MS using the previously reported method by our group [59]. Cotreatment of RA and constant dose of NVP. Female BN rats were co-treated with NVP (150 mg/kg/day) in food and RA at two different doses (5 mg/kg/day or 20 mg/kg/day). RA was suspended in corn oil and administered by oral gavage. Each of the three co-treatment groups had 4 animals and the treatment lasted 21 days. Cotreatment of RA and an escalating dose of NVP. Twelve female BN rats were divided into 3 groups: 1 control, 1 NVP and 1 cotreatment group. The NVP group received a constant dose of NVP at 150 mg/kg/day while the cotreatment group received 20 mg/kg/day of RA and NVP with a starting dose of 150 mg/kg/day in food. The plasma levels of NVP and its major metabolites were determined by LC/MS, and the dose of NVP in the cotreatment group was escalated to approximately 175 mg/kg/day over the 21-day treatment course. Serum cytokine levels were

82 65 determined by using a Luminex multiplex assay, which can simultaneously measure 24 cytokines/chemokines. Cotreatment of NVP with 1-methyl-tryptophan, LPS, imiquimod, or vitamin D. For each of the four test substances, one cotreatment group with 4 female BN rats and one NVP group with 4 animals (treated with 150 mg/kg/day NVP in food) were used. The dose and administration routes are outlined below: 1-methyl-tryptophan at a dose of 500 mg/kg/day in water by oral gavage; LPS in water at a dose of 5 mg/kg/week by i.p.; imiquimod in water at a dose of 30 mg/kg twice per week by oral gavage; vitamin D at a dose of 40 international unit(iu)/rat/day in oil by oral gavage. All animals were monitored for red ears and skin rash.

83 66 Table 4. Design of NVP and anti-mouse CD20 antibody cotreatment experiments in naïve BN rats. Group Day -7 Day 0 Day 4 Day 7 Control (N=4) NVP (N=4) Anti-CD20 D-7 (N=4) Injection Anti-CD20 D-7 &0 (N=4) Injection 1 Injection Anti-CD20 D-7 & 7 (N=4) Injection Injection 2 Anti-CD20 D4 (N=4) Injection 1 -- Anti-CD20 D4 & 7 (N=2) Injection 1 Injection 2 Anti-mouse CD20 was injected i.v. at a dose of 1 mg/rat. Starting on Day 0, all animals except the control group were started on a NVP mash diet at a dose of 150 m/kg/day.

84 67 Table 5. Design of NVP and anti-mouse CD20 antibody cotreatment experiments in splenectomized female BN rats. Group (n=2) Splenectomy NVP Anti-CD20 injection Control PBS on days 1&4 NVP mg/kg/day PBS on days 1&4 Splen-NVP Yes 150 mg/kg/day PBS on days 1&4 Splen-CD20-NVP Yes 150 mg/kg/day 1 mg/rat on days 1&4 CD20-NVP mg/kg/day 1 mg/rat on days 1&4

85 68 Table 6. Design of BSO and NVP cotreatment experiments in female BN rats. Group BSO NVP Control Tap water with 2% glucose Regular food BSO Tap water with 20 mm of BSO & 2% glucose Regular food NVP Tap water with 2% glucose 150mg/kg/day in food BSO-NVP Tap water with 20 mm of BSO & 2% glucose 150mg/kg/day in food Treatment with 20 mm of BSO in 2% glucose resulted in a dose of approximately 500 mg/kg(bw)/day

86 RESULTS Anti-mouse CD20 antibody effectively depleted B cells in peripheral blood. As shown in Figure 14, a single injection of anti-cd20 antibody dramatically decreased B cell blood levels to less than 5% in 3 days. The B cell level started to gradually recover from day 7, and the time required to fully recover appeared to depend on the dose. The above results suggest that: 1) in peripheral blood, B cells were effectively depleted by this anti-mouse CD20 antibody; 2) higher doses (2 mg/rat and 3 mg /rat) did not lead to a better depletion than the lowest dose did; in fact, they appeared less effective, and the level of B cells returned to normal levels faster. B cell depletion had no significant effect on NVP-induced skin rash in normal animals. During the 22 days of NVP treatment, only group Anti-CD20 D4 & 7 rats that received 2 injections of anti-cd20 antibody on days 4 and 7 following NVP treatment seemed to have an altered onset of the disease: they started to develop red ears on day 10, which did not become obvious until day 12, whereas other groups developed red ears on day 7. The skin rash of all rats that received 2 antibody injections seemed less severe than the others. Blood was taken weekly and CD45 + cell levels were determined using flow cytometry. Since there was no change observed in most groups, data for group Anti-CD20 D4 & 7 (which seemed to have an effect) and group Anti-CD20 D-7 & 0 were selected (both groups received two injections at 1 week intervals) as a representative of this study (Figure 15). B cell levels in group Anti-CD20 D-7 & 0 that received the antibody injections on days -7 and 0 gradually increased from 10% (day 0) to 20% (day 20) while the levels in group Anti-CD20 D4 & 7 remained less than 10% during the entire examined period. B cell depletion in other tissues. By day 22, all rats treated with NVP developed a skin rash and were euthanized. At the end of the treatment course, the % B cells was ~20% in the peripheral blood for group anti-cd20 D-7 & 0 and ~3% for group Anti -CD20 D4 & 7, both significantly less than that of the NVP and control groups (Figure 16). No significant difference in B cell blood levels was observed between the control and NVP groups. There are about 10% more B cells in the ALNs of the NVP group (60%) compared to that of control group (50%), and the levels were significantly less in anti-cd20 D-7 & 0 group (40%). Although N=2, the Anti-CD20 D4 & 7 group seemed to have a reduced % B cells as well (29%). On the other

87 70 hand, there was no difference in splenic B cell levels between the NVP group and the antibodytreated groups. Effect of B cell depletion on spleen weights and plasma IgE production. The spleen sizes of the antibody-treated animals were similar to that of the control animals. Animals in the NVPtreated group had enlarged spleens compared to all other groups (Figure 17). A previous study reported a spike in IgE observed in NVP-treated animals on Day 7, which returned back to baseline by day 14 and remained the same throughout the rest of the NVP treatment [61]. The same trend was observed in both the NVP-treated and anti-cd20 D4 & 7 group (Figure 18). Interestingly, the IgE level started to climb on day 4 following NVP treatment in the anti-cd20 D-7 and 0 group and peaked on day 7 before returned to baseline. Depleting B cells in splenectomized rats had no effect on the NVP-induced skin rash. The results of this experiment were the same as the previous one with normal rats: There was a significant decrease in B cell levels in the peripheral blood and ALNs (data not shown). However, there was no change in the onset of symptoms, incidence or severity of this NVPinduced IDR.

88 % B cells B cell percentage in peripheral blood mg/rat 2mg/rat 3mg/rat Control N= Treatment time (days) Figure 14. Percent of cells stained with CD45RA/B in peripheral blood following antimouse CD20 antibody injections. Three female BN rats each received a single dose of anti-mouse CD20 antibody on Day 0 (1 mg, 2 mg, and 3 mg, respectively). Four control rats received PBS injections. The control data represent the average of 4 animals ± SE and each of the other three lines represents one animal.

89 % B cells B cell levels in peripheral blood Control NVP anti-cd20 (D-7 & 0) anti-cd20 (D4 & 7) NVP Treatment time (Days) Figure 15. Percent of cells stained with CD45RA/B in peripheral blood following antimouse CD20 antibody injections and NVP treatment. Control groups (N=4) received regular food while all other groups received a NVP diet at a dose of 150 mg/kg/day starting on Day 0. Group anti-cd20 D-7 & 0 (N=4) received anti-cd20 antibody injections on days -7 and 0; group anti-cd20 D4 & 7 (N=2) received the injections on days 4 and 7. Each injection was 1 mg/rat. Data represent the mean ± STD for group Anti- CD20 D4 & 7 and the mean ± SE for all other groups.

90 % B cells 73 B cell level in various tissues on D * * Control NVP anti-cd20 (D-7 & 0) anti-cd20 (D4 & 7) 0 Spleen Auricular Lymph Nodes Blood Figure 16. B cell levels in the spleen, ALN, and peripheral blood on D22 of NVP treatment. The control group received regular food while all other groups received a NVP diet at a dose of 150 mg/kg/day starting at day 0. Group anti-cd20 D-7 & 0 received anti-cd20 antibody injections on days -7 and 0; group anti-cd20 D4 & 7 received the same injections on days 4 and 7. Each injection was 1 mg/rat. The data represent the average of 2 animals ± STD for group anti-cd20 D4 & 7 and mean of 4 animals ± SE for all other groups. Statistical significance between cotreatment groups and NVP alone group was determined using the Mann Whitney test; values of p 0.05 were considered statistically significant.

91 Spleen weight (g) 0.8 Spleen Weight Control N=4 NVP N=4 Anti-CD20 (D-7 & 0) N=4 Anti-CD20 (D4 & 7) N=2 Figure 17. Spleen weight for selected groups on Day 22 of NVP treatment. The control group received regular food while all other groups received a NVP diet at a dose of 150 mg/kg/day. Group anti-cd20 D-7 & 0 received anti-cd20 antibody injections on days -7 and 0; group anti-cd20 D4 & 7) received the same injections on days 4 and 7. Each injection was 1 mg/rat. The data represent the mean ± STD for group anti-cd20 D4 & 7 and mean ± SE for all other groups.

92 Plasma IgE (ng/ml) Plasma IgE levels * * * D-7 D0 D7 D14 D21 0 Control N=4 NVP N=4 Anti-CD20 (D-7 & 0) N=4 Anti-CD20 (D4 & 7) N=2 Figure 18. The effect of NVP treatment and anti-mouse CD20 antibody on plasma IgE levels. The control group received regular food while all other groups received a NVP diet at a dose of 150 mg/kg/day. Group anti-cd20 D-7 & 0 received anti-cd20 antibody injections on days -7 and 0; group anti-cd20 D 4 & 7 received injections on days 4 and 7. Each injection was 1 mg/rat. The data represent the mean ± STD for group anti-cd20 D4 & 7 and mean ± SE for all other groups. Statistical significance in IgE levels between baseline and different time points was determined using the Mann Whitney test; values of p 0.05 were considered statistically significant.

93 76 BSO decreased the incidence of NVP-induced skin rash in naïve animals but not in sensitized animals. In the primary treatment experiments, 2 (out of 4) rats in the BSO-NVP group developed red ears on day 18, and the other 2 had no symptoms until the end of study (day 21). All animals in the NVP group developed red ears on day 7 and skin rash by day 21. For rechallenge studies, pre-treatment of BSO for 7 days did not change the time to onset or incidence of NVP-induced skin rash. All animals rechallenged with NVP developed red ears on day 1 and skin rash by day 7. Effect of BSO on NVP metabolism on primary treatment. Figure 19 shows that the plasma levels of both 12-OH-NVP and the parent drug were significantly lower in the BSO/NVP cotreatment group compared to that of the NVP group. No significant difference was observed for 2-OH-NVP and 4-COOH-NVP. 3-OH-NVP was not detectable in either group. It seems that the protective role of BSO on NVP-induced skin rash was a result of decreased NVP and/or 12-OH-NVP plasma levels; however, how BSO affected NVP metabolism remains unknown. The urinary excretion of NVP was also lower in the cotreatment group than that of the NVP group, but there was no difference in any other major metabolites (Figure 20). Effect of BSO on liver weight and glutathione levels on primary treatment. A decrease in liver weight and glutathione level was also observed. The increase in liver weight in this study can be explained by the fact that NVP is a P450 inducer. Compared to the NVP group, there is less of an increase in liver weight in the BSO-NVP group, which might be a result of glutathione depletion. The glutathione level in the cotreatment group is lower than that of the NVP and control groups, but higher than that of the BSO group.

94 g/ml g/ml g/ml g/ml g/ml 77 A NVP plasma level B 12-OH-NVP plasma level NVP BSO-NVP NVP BSO-NVP Day Day C 2-OH-NVP plasma level D 3-OH-NVP plasma level NVP BSO-NVP NVP BSO-NVP Day Day E 3 4-COOH-NVP plasma level 2 1 NVP BSO-NVP Day Figure 19. Plasma levels of NVP and its metabolites in BSO-NVP co-treated or NVPtreated female BN rats. NVP was given at 150 mg/kg/day in food and BSO was given at 20 mm/day in tap water with 2% glucose. A) NVP, B) 12-OH-NVP, C) 2-OH-NVP, D) 3-OH-NVP, E) 4-COOH-NVP. The data represent the mean concentration SE (N=4 for each group).

95 g/24 hrs g/24 hrs g/24 hrs g/24 hrs g/24 hrs 78 A Urinary excretion of NVP in 24 hrs 1500 B Urinary excretion of 12-OH-NVP in 24hrs NVP BSO-NVP NVP BSO-NVP Day Day C Urinary excretion of 2-OH-NVP in 24hrs 4000 D Urinary excretion of 3-OH-NVP in 24hrs NVP BSO-NVP NVP BSO-NVP Day Day E Urinary excretion of 4-COOH-NVP in 24hrs NVP BSO-NVP Day Figure hours urinary excretion of NVP and its metabolite in BSO-NVP co-treated and NVP-treated female BN rats. NVP was given at 150 mg/kg/day in food and BSO was given at 20 mm/day in tap water with 2% glucose. A) NVP, B) 12-OH-NVP, C) 2-OH-NVP, D) 3-OH-NVP, E) 4-COOH-NVP. The data represent the mean SE (N=4 for each group).

96 mg/g wet liver tissue liver weight (g) Liver weight 10 * 5 0 Control BSO NVP BSO-NVP 1.0 Total glutathione level * Control BSO NVP BSO-NVP Figure 21. Liver weights and glutathione levels in the liver of BSO-NVP co-treated or NVP-treated female BN rats. NVP was given at 150 mg/kg/day in food and BSO was given at 20 mm/day in tap water with 2% glucose. Tissues were collected at the end of the 21 day treatment. The data represent the mean SE (N=4 for each group). Statistical significance between the cotreatment group and NVP group was determined using the Mann Whitney test; values of p 0.05 were considered statistically significant.

97 80 RA decreased the incidence of NVP-induced skin rash but also decreased the plasma concentrations of NVP and 12-OH-NVP. The incidence of skin rash in RA-NVP co-treated animals was lower than that of NVP-treated group in a dose dependent manner (Figure 22.). All animals in NVP-treated group developed red ears in 7~9 days and skin rash in 2 weeks. Animals in low dose RA-NVP group had no red ears, but 2 out of 4 rats had redness on some part of the skin. High dose RA-treated animals experienced no symptoms at all. The protective effect is dramatic; however, it seems to be a result of the significantly decreased plasma levels of both the parent drug and the 12-OH metabolite (Figure 23.). There is literature evidence that RA can induce CYP3A4 [121], which may also be responsible for metabolizing NVP and 12-OH-NVP. The induction of P450 would accelerate the metabolism of both the parent drug and 12-OH-NVP and lead to lower plasma levels of both substances. However, even the compromised plasma levels seemed high enough to induce a skin rash and therefore this protection may be partially immune-mediated. Two simple ways to test this hypothesis would be by using a P450 inhibitor (for example, aminobenzotriazole) or manipulating the drug plasma levels in the cotreatment group by using a higher dose of NVP. The former involves the co-administration of three substances and can lead to complicated results. Therefore, I took the second path. RA is protective in animals with comparable drug plasma levels with that of the NVPtreated animals. In this experiment, animals in the cotreatment group received an escalating dose of NVP to compensate the reduction of drug plasma levels caused by the induction of P450 by RA. Even though the drug plasma levels were similar in both groups, the RA-NVP cotreatment group had a lower incidence (75%) and the onset of the disease in RA-NVP group was delayed by a minimum of 5 days. The severity of the skin rash and ear redness were also much milder in all rats that received RA. The Luminex multiplex assay results showed that RA prevented the increase in many of the pro-inflammatory cytokines, and this suggests that it may have a regulatory role in NVP-induced skin rash. Cotreatment with 1-methyl-tryptophan, LPS, imiquimod, and vitamin D. None of these four substances had any effect on the incidence or severity of NVP-induced skin rash. There was also no change in the onset of symptoms compared to that of NVP-treated group.

98 81 Figure 22. Effect of RA on the incidence of NVP-induced skin rash. Female BN rats were co-treated with NVP (150 mg/kg/day in food) and RA at two different doses with 4 animals in each group for 21 days. A low dose of RA refers to 5 mg/kg/day and a high does refers to 20 mg/kg/day in oil by oral gavage. NVP group received the standard dose of NVP and corn oil by oral gavage.

99 l/ml l/ml l/ml NVP plasma level NVP RA(low)-NVP RA(high)-NVP OH plasma level NVP RA(low)-NVP RA(high)-NVP * Time of Treatment (Day) * * * * * * * * Time of Treatment (Day) OH plasma level NVP RA(low)-NVP RA(high)-NVP Time of Treatment (Day) Figure 23. Effect of RA on plasma levels of NVP and its metabolites. Female BN rats were co-treated with NVP (150 mg/kg/day in food) and RA at two different doses with 4 animals in each group for 21 days. A low dose of RA refers to 5 mg/kg/day and a high does refers to 20 mg/kg/day in oil by oral gavage. Drug plasma concentrations were determined by LC-MS. The data represent the mean of 4 animals ± SE. Statistical significance between cotreatment groups and NVP alone group was determined using the Mann Whitney test; values of p 0.05 were considered statistically significant.

100 l/ml l/ml NVP NVP RA-NVP Time of Treatment (Day) OH-NVP NVP RA-NVP * Time of Treatment (Day) Figure 24. NVP and 12-OH-NVP plasma concentrations in animals that received RA (20 mg/k/day in oil, gavage) and an escalating dose of NVP (started from 15 mg/kg/day and escalated to 175 mg/kg/day). Drug plasma concentrations were determined by LC/MS. The data represent the mean of 4 animals ± SE. Statistical significance between RA-NVP group and NVP alone group was determined using the Mann Whitney test; values of p 0.05 were considered statistically significant.

101 pg/ml pg/ml pg/ml pg/ml pg/ml pg/ml pg/ml pg/ml pg/ml pg/ml IFN NVP RANVP IL NVP RANVP IL-18 NVP RA-NVP Treatment time (day) Treatment time (day) Treatment time (day) Treatment time (day) Leptin NVP RANVP IL NVP 800 RANVP 600 MCP-1 NVP RA-NVP Treatment time (day) Treatment time (day) Treatment time (day) Treatment time (day) RANTES NVP RA-NVP Treatment time (day) Treatment time (day) Figure 25. Plasma concentrations of cytokines/chemokines during a 21-days treatment course with NVP or RA-NVP determined by a Luminex assay. Female BN animals in the NVP group received a dose of NVP at 150 mg/kg/day; animals in the cotreatment group received RA (20 mg/kg/day in oil by gavage) and an escalating dose of NVP (started from 150 mg/kg/day and escalated to 175 mg/kg/day). The data represent the mean of 4 animals ± SE.

102 DISCUSSION B cells have multiple functions in many autoimmune diseases. In addition to acting as precursors of antibody-producing plasma cells, they also negatively regulate autoimmunity through B10 cell (B cells that produce IL-10) function and serve as critical adjuvants for CD4 + T-cell activation [122, 123]. The increase in both the total number of B cells and MHC-II-expressing B cells suggested their potential involvement in presenting antigens to T helper cells; the presence of IgE following NVP treatment also suggested a role in antibody production [67]. We employed an anti-mouse CD20 antibody to deplete B cells in vivo to determine if it could prevent the rash. It is also possible that B cells are negative regulators or involved in downstream events of this immune response. If that is the case, the depletion would not prevent the animals from getting a rash and could make it worse. The present study showed that this anti-mouse CD20 antibody significantly decreased B cell levels in the peripheral blood and in ALNs; however, it was not effective in the spleen. A delay in the onset of red ears was observed in the group that received 2 antibody injections on days 4 and 7 following NVP treatment, and that also corresponds to a lower level of B cells in the blood and ALNs compared to other groups. These results suggest that B cells may play a role in initiating or mediating the rash. The finding that the spike of IgE on day 7 in the NVP group declined to a baseline level by day 14 is the same as what was previously reported [67]. However, the IgE started to increase on day 0 in group Anti-CD20 D-7 & 0 which received antibody injections on days -7 and 0; this may be the result of the injection of anti-mouse antibody because these animals had not been treated with NVP on day 0. The peak on day 7 for group Anti-CD20 D-7 & 0 and Anti-CD20 D4 & 7 is hard to interpret due to the lack of an isotype antibody serving as a control. We do not know how long and to what level the increase in IgE caused by anti-cd20 injections are. However, the spike on day 7 may be a combination of anti-cd20 injection and NVP treatments. In any case, the goal to test IgE levels was to determine if functions of plasma cells, which are derived from B cells, were reduced by anti- CD20 injections; however, no difference between NVP group and antibody-injected groups was observed. The spleen is the largest lymphoid organ in the body and contains many different immune cells. The reason that splenic B cell levels were not effectively depleted may be due to the anti-mouse

103 86 nature of the antibody. The antibody may be cleared by phagocytic cells such as macrophages. As a result, the formation of plasma cells, which occurs in the spleen and lymph nodes, was not affected. This may explain why we did not observe a significant delay in the onset of red ears. If we could somehow effectively deplete B cells in the spleen it might result in a longer delay of onset or change in incidence. Therefore, we decided to perform a study in splenectomized animals. The removal of spleen can simply avoid the resistance of splenic B cells to antibody clearance, but it may also lead to other effects on the immune response. The spleen clears dead red blood cells and many other old cells. Splenectomy may lead to an accumulation of dead cells, which could potentially release danger signals and in turn accelerate or increase the possibility of developing IDRs on exposure to NVP. Therefore the effect of splenectomy alone on NVPinduced IDRs was determined first. The pilot study showed that following NVP treatment, the ears of splenectomized rats started to turn red on day 5, but it was not obvious until day 7. Such an effect might be due to the short recovery time (2 weeks), but the severity and all other symptoms were similar to that of the healthy rats that received NVP. To make sure the B cells were depleted prior to the development of IDRs, antibody injections were administered on days 1 and 4, although the previous study showed it might have a better effect if given on days 4 and 7. However, the results showed again that depleting B cells in splenectomized rats had no effect on either the onset or severity of NVP-induced skin rash. In summary, depleting B cells using this anti-mouse CD20 antibody did not change the incidence of NVP-induced skin rash in female BN rats. The depletion was not complete, which might be due to the fact that it is an anti-mouse antibody and could be cleared in the spleen. However, even in splenectomized rats, it did not have a significant effect. In contrast, a partial depletion of CD4 + cells can significantly delay or prevent NVP-induced skin rash [61]. Although the evidence is not conclusive, it is likely that B cells are not essential for the induction of a skin rash in our model. The second modulator tested was BSO. It can irreversibly inhibit gamma-glutamylcysteine synthetase, and it has been used to deplete glutathione in rats and mice [124]. However, glutathione depletion inhibits the antigen presenting process and shifts a Th1 immune response pattern to Th2 [125]; therefore, it might change the incidence or severity of skin rash induced by

104 87 NVP treatment in female BN rats. On the other hand, glutathione protects cells against oxidative stress, and its depletion may lead to liver toxicity. The cotreatment of BSO changed the incidence of NVP-induced IDRs in female BN rats; however, this is likely due to decreased plasma levels of parent drug and metabolites. One study suggested that BSO can simultaneously induce cytochrome P450 2E1 and alcohol dehydrogenase in mouse livers [126]. In our case, it might interact with P450 and result in less metabolism and/or more clearance of NVP which would lead to a decreased covalent binding of NVP and, therefore, a delayed onset of the disease. There was no difference in urinary excretion of NVP metabolites that were tested between the BSO-NVP and NVP groups, but the parent drug excretion was significantly less in the latter group. It is not likely that BSO induced P450 in this case because we did not observe higher plasma levels or urinary excretion of the metabolites. On the other hand, BSO may have accelerated the clearance of NVP given that the metabolite excretion levels were similar but that of parent drug was significantly reduced. The mechanisms by which BSO is protective in NVP-induced skin rash is complicated, and it would be quite difficult to determine the exact mechanism of its effects. RA came to our attention originally because of its potential to alter the balance between Th17 cells and Treg cells, and we speculated that it might have an effect on the NVP-induced IDR, which is mediated by CD4 T cells. However, recent reports showed that RA may have both positive and negative regulatory roles. With respect to Th17 cells, it may suppress them at higher concentrations (in vitro) but induce Th17 cells expressing gut-homing receptors at physiological levels (in vivo)[115, 127, 128]. In the present study, RA seems to inhibit the production of many pro-inflammatory cytokines. However, at the same time, it might have also induced P450, which led to reduced drug plasma levels. When the dose of NVP was increased so that the blood levels of NVP were similar RA appeared to be protective. However, because of the complex effects of RA on the immune response which would be very difficult to sort out, we decided not to pursue this further. The present study attempted to manipulate the NVP-induced skin rash from different angles such as suppressing the immune system, breaking immune tolerance, and activating the innate immune system, etc. While most of the modulators do not have any significant effect, the very few that worked seemed to involve different mechanisms than we expected. Based on the BSO

105 88 and RA experiments in which both reagents seemed to have an impact on the metabolism of NVP and resulted in lower drug levels in the plasma, one can never conclude the effect of any modulator on drug-induced adverse reactions without testing its effect on metabolism. Although it was difficult to find something that has a clean effect on the NVP-induced IDR, it is important to continue to explore such risk factors. Only with an valid animal model will we be able to vigorously test all these hypotheses and eventually determine what is going on in humans.

106 89 CHAPTER 5 POTENTIAL ACTIVATION OF ANTIGEN PRESENTING CELLS BY NVP AND/OR ITS METABOLITES Co-author: Amy Sharma (covalent binding study)

107 INTRODUCTION One fundamental question in drug-induced IDRs is whether it is the parent drug or the reactive metabolite that leads to the reaction. In the case of NVP-induced skin rash, we were able to show that 12-OH-NVP pathway was involved in the induction of skin rash in female BN rats [59]. The most recent data further confirmed that this was due to the formation of 12-OH-NVP sulfate in the epidermis, which covalently binds to proteins and induces an immune response [68, 69]. Although we have demonstrated that 12-OH-NVP sulfate causes the rash, we have not studied how this may activate the immune system. In order to elicit an immune response, the drug may form a protein adduct by covalent binding, which is then picked up by an APC and presented to T cells in the presence of co-stimulatory signals. Alternatively, the drug may also directly bind and activate APC similar to what we saw with the D-penicillamine-induced autoimmune disease [78, 80]. A recent paper by Kevin Park s group showed that SMX and its protein-reactive metabolite, nitroso SMX, can stimulate dendritic cell (DC) co-stimulatory signaling. In particular, the response of DCs to the nitroso reactive metabolite is much stronger than that to the parent drug [46]. A similar study was also reported by Reuter et al. where the co-stimulatory molecule CD86 was found to be upregulated on PBMC-derived DCs in an in vitro model for allergic contact dermatitis [129]. In our NVP model, the interaction between reactive metabolite and the APCs has not been studied. We hypothesized that 12-OH-NVP sulfate can directly activate APCs, and this may lead to the idiosyncratic skin rash caused by NVP in female BN rats. To test this hypothesis, APCs from different sources were stimulated by 12-OH-NVP sulfate, 12-OH-NVP and/or the parent drug using the expression of co-stimulatory molecules as a measure of APC activation.

108 MATERIALS AND METHODS Chemicals. NVP was kindly supplied by Boehringer-Ingelheim Pharmaceuticals Inc. (Ridgefield, CT). The synthesis of 12-OH-NVP, 12-OH-NVP sulfate, and preparation of NVP antiserum were described previously [59, 98, 130, 131]. D-penicillamine was purchased from Richman Chemical Inc. (Lower Gwynedd, PA). RPMI 1640 medium, dulbecco's modification of Eagle's medium (DMEM) high glucose medium, PBS (ph 7.4), FBS, penicillin and streptomycin concentrated solution, 2-ME, and normal goat serum were purchased from Invitrogen Canada, Inc. (Burlington, ON, Canada). Microvette EDTA-coated tubes for blood samples were purchased from Sarstedt (Montreal, QC, Canada). Isoniazid, RIPA buffer, DMSO, indomethacin, PMA, inomycin, ammonium chloride (NH 4 Cl), LPS, potassium bicarbonate (KHCO 3 ), EDTA, horseradish peroxidise-conjugated goat antirabbit IgG (H + L chains), and dextran were purchased from Sigma-Aldrich (Oakville, ON, Canada). Anti-rat CD4 allophycocyanin, anti-rat CD25 PE, anti-rat CD80 PE, anti-rat CD86 FITC, anti-rat MHC II PE, anti-rat CD161 FITC, anti-rat pan-macrophage PE, anti-rat αβ-tcr peridinin chlorophyll protein complex (PerCP), anti-rat rat recombinant IL-4, and GM-CSF were purchased from Cedarlane (Mississauga, ON, Canada). FITC hamster anti-mouse CD40, anti-rat CD32, and anti-mouse CD16/CD32 antibodies were purchased from BD Pharmingen (Mississauga, ON, Canada). Anti-human Fc binding receptor, anti-human CD40 FITC, anti-human CD80 PE, anti-human MHC II antibody conjugated with R-Phycoerythrin-Cyanine dye 7 (PE-Cy7), anti-human CD86 conjugated with allophycocyanin, anti-mouse CD80 PE, and anti-mouse CD86 allophycocyanin antibodies were purchased from ebioscience (San Diego, CA, USA). SDS and Tween-20 were obtained from BioShop (Burlington, ON). Stock acrylamide/bis solution, nonfat blotting grade milk powder, and nitrocellulose membranes were purchased from BioRad (Hercules, CA). Amersham enhanced chemiluminescence (ECL) Plus Western Blotting Detection System was obtained from GE Healthcare (Oakville, ON). Protein concentrations were determined using a bicinchoninic acid (BCA) protein assay kit (Novagen, EMD Biosciences Inc.). Animal Care. Female BN rats ( g) were purchased from Charles River (Montreal, QC) and housed in pairs in standard cages with free access to water and Agribrands powered lab chow diet (Leis Pet Distributing Inc, Wellesley, ON). The animal room was maintained at 22 o C with a 12:12 hour light:dark cycle. Rats were killed by CO 2 asphyxiation. All of the animal

109 92 studies were conducted in accordance with the guidelines of the Canadian Council on Animal Care and approved by University of Toronto s Animal Care Committee. Studies of Potential Activation of Freshly Isolated Cells from Naïve Rats and PBMC Monocyte-derived DCs by NVP and Its Metabolites. The first study was to test the response of cells isolated fresh from PBMCs, ALNs, and the spleen of a naïve female BN animal to NVP and the metabolites. However, even the positive control, LPS, a strong stimulant of APCs, did not induce significant changes in the markers that were tested. A possible explanation is that these tissues have diverse cell populations, and the APC abundance might not be high enough to detect any changes. However, even when APCs from ALNs were concentrated using magnetic beads (APCs MACS cell separation kit, Miltenyi Biotec), no upregulation of any markers was observed, but a significant decrease in MHC II expression was observed (shown in Appendix 1). The second attempt was to derive DCs from PBMCs isolated fresh from naïve animals using the method described by Kevin Park s group [46]. Although this seems to be the standard approach for studies involving human cells, it was not practical for rat studies because the yield of cells is too low. General Procedures for Direct Activation of APCs by NVP and Its Metabolites. NVP, 12- OH-NVP, and 12-OH-NVP sulfate were dissolved in 0.5% DMSO at various concentrations: 1.0, 2.0, 3.9, 7.8, 15.6, 31.3, 62.5, and 125 µg/ml. APCs from different sources were populated at a concentration of 1 million/ml (unless specified) in complete medium and incubated with drugs at 37 o C, 5% of CO 2 for 24 hours. At the end of the culture period, expression of co-stimulatory molecules CD40, CD80, CD86, and MHC II on these cells were tested by flow cytometry. Direct Activation of RAW264.7 Cells by NVP and Its Metabolites: The RAW cell line was purchased from American Tissue Culture Collection (ATCC, Manassas, VA, USA) and maintained according to the manufacturer s instruction. The day before the APC activation experiment, the cells were seeded at 0.5 million/ml in a 24-well plate overnight in DMEM high glucose medium containing 10% FBS and antibiotics, allowing the cells to adhere to the culture plate. The next morning, non-adherent cells were washed off and fresh culture medium containing various concentrations of 12-OH-NVP sulfate, 12-OH-NVP, or NVP were added. After 24 hours of stimulation, the cells were harvested and surface markers were tested by flow cytometry.

110 93 Direct Activation of APCs by NVP and Its Metabolites: THP-1 Cells. The THP-1 cell line were purchased from ATCC (Manassas, VA, USA) and maintained according to the manufacturer s instruction. The procedures for stimulating THP-1 cells were the same as stimulating RAW264.7 cells except that RPMI1640 medium was used. Direct Activation of APCs by NVP and Its Metabolites: Bone Marrow-Derived Dendritic Cells (BMDCs). The complete culture medium used to generate BMDCs (referred as the complete medium hereafter) was prepared as follows: RPMI 1640 medium was supplemented with 10% heat-inactivated FBS, 100 U/mL penicillin, 100 µg/ml streptomycin, 50 µm 2-ME, 5 ng/ml recombinant rat GM-CSF, and 5 ng/ml IL-4. Bone marrow was prepared from the femur bones of naïve female BN rats. The bones were placed in a 100-mm dish and washed with 70% alcohol for 2 minutes. After removing the remaining muscle tissue, they were transferred into a new dish with RPMI medium. Both ends of the bones were cut and the bone marrow was flushed out with a syringe and a 25-gauge needle using 10 ml of RPMI medium. The cells were passed through two 40 µm cell strainers and centrifuged at 350 g. To derive DCs from the bone marrow cells, the cell pellet was resuspended in 10 ml of the complete medium and plated on 100-mm petri dishes at a density of 5 X 10 5 cells/ml. On day 3, 10 ml of freshly prepared complete medium was added to the culture. On day 6, the supernatant was removed, and 10 ml of freshly prepared complete medium was added. On day 8 or 9, the cells were harvested by scraping the plate and resuspended at a concentration of 5 X 10 5 cells/ml. The single cell suspension was then seeded at 1 ml per well in a 24-well tissue culture plate and incubated with various concentrations of NVP, 12-OH-NVP, or 12-OH-NVP sulfate. The cells incubated with LPS served as positive control whereas the DMSO-treated cells were used as negative control. Covalent Binding of NVP and Its Metabolites to BMDCs. To determine if 12-OH-NVP and/or its sulfate covalently bind to the BMDCs, incubations of these compounds as well as NVP, or DMSO was incubated with BMDCs. Immature BMDCs were harvested on day 8 and resuspended at a concentration of 1x10 6 cells/ml and a total of 10 x10 6 cells were used for each incubation. The concentration of NVP and its metabolites in the incubations was 62.5 µg/ml, the incubation time was 24 or 72 hours at 37 o C in 5% CO 2. After incubation, the cells were lysed using RIPA buffer, and the proteins were pelleted. The protein concentration was

111 94 determined by a BCA protein assay kit (Novagen, EMD Biosciences Inc., Mississauga, ON), and the proteins were loaded on a sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS- PAGE) gel at a concentration of 40 µg/lane for 24 hours samples and 35 µg /lane for 72 hours samples. SDS-PAGE and western blotting procedures were described previously by Amy Sharma[69]. Specifically, SDS-PAGE gels were hand-cast (stacking gel, 5% bisacrylamide; resolving gel, 8% bisacrylamide) and run at ~110 V. Electrophoresis running buffer (BioRad, Mississauga, ON) consisted of 25 mm Tris, 192 mm glycine, and 0.1% SDS, ph 8.3. Transfer to nitrocellulose membrane (pore size 0.2 µm, BioRad) was performed at 0.13 ma for 90 minutes at 4 ºC using Protean-3 minigel system (BioRad). Tris-glycine transfer buffer consisted of 25 mm Tris, 192 mm glycine, and 20% methanol at ph 8.5. Membranes were washed twice in Tris-buffered saline Tween-20 (TBST) wash solution for 5 minutes. Membranes were then blocked in 5% nonfat milk blocking solution in TBST for 90 minutes at room temperature and washed with three changes of TBST for 5 minutes each. A 1:500 dilution of the primary anti- NVP antiserum and 10% normal goat serum in TBST was added to the membranes and incubated overnight at 4 ºC. After a 20 minute wash on the next day, the membranes were incubated with a secondary goat anti-rabbit horseradish peroxidase antiserum (1:2000) for 90 minutes. Blots were washed 3 times with TBST and incubated with enhanced chemiluminescence stain for 5 minutes and imaged for 3 minutes using a FluroChem CCD imager. Activation of APCs by D-penicillamine and Isoniazid. BMDCs generated by the procedures described above or RAW264.7 cells were stimulated with D-penicillamine or isoniazid at various concentrations for 24 hours: 0.1, 0.4, 1.2, 3.6, 11, 33, 100, and 300 µg/ml, and the expression of surface markers was determined by flow cytometry. D-penicillamine and isoniazid were shown to activate RAW264.7 cells by binding the aldehyde groups on RAW264.7 cells, and therefore it should be possible to use them as positive controls in this study [132, 133]. T Cell Activation Assay. 1. Naïve T Cell Isolation. Autologous T cells were isolated from the spleen and ALNs of female BN rats. Naïve T cells were isolated by negative selection using a CD4 + T cell isolation kit followed by a positive selection of CD25 + cells (Miltenyi Biotec, Anburn, CA, USA). The bead

112 95 isolation procedures were carried out according to the manufacturer s instructions. The phenotype of the isolated cells was determined by flow cytometry using anti-rat CD4 APC and anti-rat CD25 PE antibodies. Isolated naïve T cells were resuspended in RPMI 1640 medium (supplemented with 10% FBS and antibiotics) at X 10 6 cells/ml, and then an equal volume of FBS containing 20% DMSO was added to the cell suspension dropwise on ice. The cells were stored at -80 o C until use. 2. T Cell Co-culture with BMDCs. Immature DCs were harvested on day 8 of differentiation from bone marrow cells and resuspended at a concentration of 4 X 10 5 cells/ml. These cells were seeded at 500 µl per well in 24-well plates. Naïve T cells were thawed, and the RPMI 1640 medium was added dropwise to make 10 times the initial volume. After washing, cells were resuspended at 2 X 10 6 cells/ml and added to BMDCs at 1 ml per well. 10 µl of the stock solutions of NVP, 12-OH-NVP, or 12-OH-NVP sulfate (12.5 mg/ml dissolved in DMSO) were diluted with 490 µl of cell culture medium and then added to the T cell/bmdcs mixture resulting in a final drug concentration of 62.5 µg/ml and a total culture volume of 2 ml. The mixture was incubated for 7-8 days at 37 o C in a 5% CO 2 atmosphere. 3. Carboxyfluorescein Succinimidyl Ester (CFSE) Staining and T Cell Restimulation. After the co-culture period, the cells were labeled with CFSE and re-stimulated before T cell proliferation was assessed. Specifically, the stimulated T cell/apc mixtures were harvested and CFSE staining was carried out according to the manufacturer s instructions (Invitrogen, Canada). The cells were then resuspended in 1 ml of medium, and 100 µl was aliquoted to 96-well plates. Another batch of immature BMDCs prepared in advance were resuspended at 4 X 10 5 cells/ml and added to the T cell/apc cell mixture at a volume of 50 µl per well. Various concentrations of NVP, 12-OH-NVP, and 12-OH-NVP sulfate diluted in 50 µl of medium were added to the culture and incubated for 72 hours at 37 o C in a 5% CO 2 atmosphere. PMA/Inomycin-stimulated cells were used as a positive control. At the end of restimulation period, the proliferation of T cells was tested using flow cytometry. Specifically, anti-rat αβ- TCR + antibody was used to label the T cells, and the proliferation was measured by the reduction in CFSE staining. Flow Cytometry. A PBS buffer containing 3% FBS was used as the medium for flow cytometry. Cells were first incubated with an antibody against the Fc binding receptor for 10

113 96 minutes at room temperature to reduce nonspecific binding. After that, monoclonal antibodies or suitable isotype controls were aliquoted to the appropriate wells and incubated at room temperature for 20 minutes. The cells were washed twice and finally resuspended in 200 µl of the same buffer. Samples were processed immediately with a FACS Canto II flow cytometer (BD Biosciences) with CellQuest software. FlowJo (Tree Star, Inc., Ashland, OR) was used to analyze the samples. ELISA. Rat ELISA kits for IL-1β, IL-12p40, and TNF-α were purchased from R&D system (Minneapolis, MN). Experiments were performed following the manufacturer s instructions. Stability of 12-OH-NVP Sulfate. The sulfate of 12-OH-NVP was synthesized as described by Chen et al. [59]. The stability of 12-OH-NVP sulfate was determined by high-performance liquid chromatography (HPLC) (SHIMADZU LC-10AS). The column was an Ultracarb 100X2 mm, 5 micron, ODS(30), and the mobile phase consisted of 12% acetonitrile in aqueous 2 mm ammonium acetate/1% acetic acid with a ph of 3.7 and a flow rate of 0.4 ml/minutes. To test the stability in cell culture, immature BMDCs harvested on day 8 of differentiation were stimulated with 62.5 µg/ml of 12-OH-NVP sulfate. Aliquots of supernatant were diluted 10X with the mobile phase and analyzed with HPLC at various time points. A parallel experiment was also carried out to determine the stability of the sulfate in the culture medium in the absence of cells. In both studies, 12-OH-NVP was used for a control incubation to represent complete hydrolysis.

114 RESULTS Activation of RAW264.7 Cells by NVP and Its Metabolites. Examples of the flow cytometry dot plots and a histogram are shown in Figure 26. The dot plots showed that there was an increased expression of CD40 on the 12-OH-NVP sulfate-treated cells compared to that of the DMSO vehicle control cells (65.4% versus 15.9%). The histogram also showed that the mean fluorescence intensity (MFI) was greater for the 12-OH-NVP sulfate-treated cells than for the DMSO control cells. The dose response curves in Figure 27 showed that both the percentage of CD40-expressing cells and the expression level per cell (measured by MFI) increased with increasing drug concentrations. Among the three substances, 12-OH-NVP seemed to better activate RAW cells at the concentrations of 32 and 64 ug/ml compared to NVP and 12- OH-NVP sulfate. One problem with RAW cells was that the baseline levels of activation markers increased throughout the experiments (Figure 28). This was independent of the solvent because even in the absence of DMSO, the cells express higher levels of CD40 at higher passages (results not shown). There is literature evidence that over-passaged cells may exhibit reduced or altered functions as a result of selective pressures and genetic drift, and therefore they no longer represent reliable models of their original source material [134].

115 98 A. B. C. D. E. Figure 26. Expression of CD40 on RAW cells after a 24 hours stimulation with 12- OH-NVP sulfate. A) unstained cells, B) cells incubated with 0.5% DMSO and stained with CD40 antibody as a negative control, C) cells incubated with 125 µg/ml 12-OH-NVP sulfate in 0.5% DMSO and stained with CD40 antibody, D) cells incubated with 2 µg/ml LPS as a positive control, E) Histogram of CD40 expression: grey (unstained control), blue (DMSO negative control), red (125 µg/ml 12-OH-NVP sulfate), green (LPS positive control).

116 CD40 MFI % CD40 + Cells 99 A * * * * * * * * * * NVP 12-OH-NVP 12-OH-NVP Sulfate Drug Concentration ( g/ml) B * * * * * NVP 12-OH-NVP 12-OH-NVP Sulfate Drug Concentration ( g/ml) Figure 27. Expression of CD40 on RAW264.7 cells in response to various concentrations of NVP, 12-OH-NVP, or its sulfate metabolite expressed in two different ways. A) percent CD40 positive cells; B) the number of CD40 molecules on cells as measured by MFI. The data represent the mean ± SE of 4 experiments. Unpaired t test, p < The control (solid lines) represent cells treated with 0.5% DMSO (N=4).

117 % CD40 + Cells Cell Passage Figure 28. Precent CD40 positive cells in aging RAW264.7 cells. The data represent the average of 2 experiments. Cells were treated with 0.5% DMSO for 24 hours.

118 101 Activation of THP-1 cells by NVP and Its Metabolites. There was an increase in percentage of cells that express CD86 with increasing concentrations of all three drugs. Surprisingly, the MFI of MHC II expression decreased when the concentration of all three substances increased. No significant change was observed for CD40 and CD80 expression (Figure 30).

119 LPS NVP (62.5 µg/ml) DMSO Unstain Count 102 CD80 CD86 CD40 MHC II Fluorescence Intensity Figure 29. Representative dot plots and histograms of surface marker staining of THP-1 cells stimulated by various substances.

120 ratio of treated versus control cells in % of CD80 cells ratio of treated versus control cells in % of CD86 cells ratio of treated versus control cells in CD40 MFI ratio of treated versus control cells in MHC II MFI 103 A. B NVP 12-OH-NVP 12-OH-NVP Sulfate Drug Concentration ( g/ml) NVP 12-OH-NVP 12-OH-NVP Sulfate Drug Concentration ( g/ml) C. D * * * * * * * * * * * * * * * * * * * NVP 12-OH-NVP 12-OH-NVP Sulfate Drug Concentration ( g/ml) NVP 12-OH-NVP 12-OH-NVP Sulfate Drug Concentration ( g/ml) Figure 30. Change in cell surface marker expression on THP-1 cells in response to NVP or its metabolites. Cells incubated with 0.5% of DMSO were used as the baseline (the dash lines). Data represent the ratio of treated versus DMSO control samples for various surface marker expression. A) the intensity of CD40 expression measured by MFI, B) the intensity of MHC II expression measured by MFI, C) percent CD80 + cells, D) percent CD86 + cells. The data represent the average of 3 experiments ± SE. Unpaired t test, p < 0.05.

121 104 Activation of BMDCs by NVP and Its Metabolites. 1. Phenotype of BM-derived Cells. Figure 31. shows that the majority (>90%) of bone marrow-derived cells express CD11c, which suggests that they are DCs. The presence of other cell types, such as granulocytes and nature killer cells (CD161 + ), macrophages (panmacrophage), and T cells (αβ-tcr) were minimum with percentages less than <0.5%, <5%, and <2%, respectively. 2. Change in Cell Surface Marker Expression. The dose-response curve (Figure 33) shows that CD40 was significantly upregulated with an increasing concentration of all three substances with 12-OH-NVP being the most potent. On the other hand, MHC II was down regulated with increasing drug concentration; the same trend was also observed for CD80. The production of several cytokines by BMDCs was measured by ELISA. Figure 34 shows that there may be a small increase in IL-12p40 and TNF-α levels at higher concentrations of all three substances; however, the differences were not statistically significant. 3. Covalent Binding of NVP and Its Metabolites to BMDCs (Western Blot By Amy Sharma). In order to determine the mechanism of the effect of 12-OH-NVP on BMDCs, we investigated the covalent binding of these substances to BMDCs. Figure 35 showed that at both 24 and 72 hours, the sulfate metabolite binds strongly to proteins extracted from BMDCs; there was also a bit of binding with 12-OH-NVP. No binding was observed for NVP and DMSO samples.

122 MHC II 105 A. CD11c B. C. D. Figure 31. Phenotype of bone marrow-derived cells. Bone marrow cells were differentiated in the presence of 5 ng/ml recombinant rat GM-CSF and 5 ng/ml IL-4 for 6 8 days before use. A) cells stained with MHC II and CD11c antibodies (blue dots) versus unstained cells (red dots), B) cells stained with CD161 antibody, C) cells stained with pan-macrophage antibody, D) cells stained with αβ-tcr antibody.

123 LPS Sulfate (62.5 µg/ml) DMSO Unstain Count 106 CD80 CD86 CD40 MHC II Fluorescence Intensity Figure 32. Representative dot plots and histograms of surface marker staining on BMDCs stimulated by various substances.

124 ratio of treated versus control cells in % CD80 cells ratio of treated versus control cells in % CD86 cells ratio of treated versus control cells in CD40 MFI ratio of treated versus control cells in MHC II MFI 107 A. B. Relative MFI of CD40 (all studies N=6) * * * * * * * * * * * * * * * NVP 12-OH-NVP 12-OH-NVP Sulfate Drug Concentration ( g/ml) NVP 12-OH-NVP 12-OH-NVP Sulfate Drug Concentration ( g/ml) C. D NVP 12-OH-NVP 12-OH-NVP Sulfate Drug Concentration ( g/ml) NVP 12-OH-NVP 12-OH-NVP Sulfate Drug Concentration ( g/ml) Figure 33. Change in cell surface marker expression on BDMCs in response to NVP or its metabolites. Cells incubated with 0.5% of DMSO was used for the baseline (the dash lines). Data represent the ratio of treated versus DMSO control samples for various surface marker expression. A) the intensity of CD40 expression measured by MFI (N=6), B) the intensity of MHC II expression measured by MFI (N=3), C) percent CD80 positive cells (N=3), D) percent CD86 positive cells (N=3). The data represent the average ± SE. Unpaired t test, p < 0.05.

125 pg/ml pg/ml pg/ml 108 A B. NVP 12-OH-NVP 12-OH-NVP Sulfate Drug Concentration ( g/ml) NVP 12-OH-NVP 12-Sulphate-NVP NVP 12-OH-NVP 12-OH-NVP Sulfate Drug Concentration ( g/ml) C. TNF NVP 12-OH-NVP 12-Sulphate-NVP NVP 12-OH-NVP 12-OH-NVP Sulfate Drug Concentration ( g/ml) Figure 34. Cytokine production by drug-stimulated BMDCs at the end of a 24 hour incubation. The data represent the average ± SE of 5 experiments. A) IL-1β, B) IL-12 p40, C) TNF-α.

126 109 Figure 35. Covalent binding of DMSO, NVP, 12-OH-NVP, and its sulfate to BMDCs after a 24 or 72 hour incubation µg/ml of each of the three substances in 0.5% DMSO were incubated with 10 million BMDCs. Immature BMDCs were generated from naïve female BN rats and harvested on day 8 of differentiation. 40 µg/lane for 24 hour samples and 35 µg/lane for 72 hour samples were loaded.