Immunology and Genetics of Autoimmune Biliary Disease

A dissertation submitted to the Graduate School of the University of Cincinnati in partial fulfillment of the

requirements for the degree of

Doctor of Philosophy (PhD)

Immunology Graduate Program, College of Medicine

2015

By

Wenting Huang

BS, Fudan University, 2007

MS, University of Miami, 2009

Committee Chair: William M. Ridgway, MD Dissertation abstract

Primary biliary cirrhosis (PBC) is a rare, incurable, organ specific autoimmune hepatobiliary disease, being one of top five indications of liver transplantation in USA. PBC is considered as a multifactorial autoimmune disease resulting from a combination of genetic and environmental factors, and is characterized by loss of tolerance to mitochondrial proteins and progressive destruction of cholangiocytes lining the biliary ducts, which leads to cirrhosis, liver function failure and death. Autoreactive T cells, regulatory T cells (Tregs), and the target cells

(cholangiocytes) have all been implicated in PBC etiology. PBC is usually asymptomatic until an advanced stage, and thus the early pathogenesis of the disease remains unclear. In this dissertation, we investigated two mouse models with spontaneous onset of PBC-like autoimmune hepatobiliary disease, aiming to understand the early genetic and immunological events contributing to the disease.

In dominant-negative TGFβ receptor type II transgenic mice (dnTGFβRII), we have previously shown that CD8 T cells can transfer autoimmune biliary disease in an adoptive transfer model In the current dissertation, we have found that the terminally differentiated CD8 T cells (KLRG1+

CD8 T cells) are increased in dnTGFβRII liver. To address whether dnTGFβRII CD8 T cell- mediated disease is intrinsic or extrinsic, we perform mixed bone marrow transfer (dnTGFβRII and B6) and find that mixed bone marrow chimeric mice are protected from biliary disease. This protective effect of B6 bone marrow arises from B6 Tregs, as we show using an adoptive transfer model. B6 Tregs-mediated disease protection is associated with significantly decreased numbers of hepatic dnTGFβRII KLRG1+ CD8 T cells. DnTGFβRII Tregs are defective in suppressing effector CD8 T cell proliferation compared to B6 Tregs. DnTGFβRII CD8 T cells are significantly more cytotoxic to cholangiocytes than B6 CD8 T cells and B6 Tregs can eliminate cholangiocyte toxicity mediated by dnTGFβRII CD8 T cells. Therefore autoimmune biliary disease requires defects in both the T effector and regulatory compartments; an intrinsic T cell

i effector defect in dnTGFβII is not sufficient to mediate autoimmune biliary disease in the setting of intact immune regulation.

In the NOD.Abd3 model, which carries a B6-derived chromosome one allele (Abd3) on NOD background, we show that clinical disease requires the NOD genetic background. Using bone marrow transfers, we demonstrate that autoimmune hepatobiliary disease requires abnormalities in both target tissue cells and hematopoietic cells. RNA-seq of common bile duct from 2-week- old NOD.Abd3 show that the abnormality of target cells arises from defective expression of

Polycystic kidney and hepatic disease 1 (Pkhd1). Loss of Pkhd1 expression is the result of DNA insertion in the intron 35 of the gene. These results suggest that loss of functional Pkhd1 on the

NOD background initiates early bile duct abnormalities, leading to an autoimmune response that ultimately produces clinical autoimmune biliary disease (ABD) in NOD.Abd3 congenic mice.

In summary, our immunological and genetics studies of these two PBC animal models further our understanding of the disease etiology and produce new insights which may ultimately result in novel therapeutic treatment options.

ii iii

Acknowledgements

I would never be able to finish my dissertation without the guidance of my advisor and my committee members as well as support from my friends and family.

I would like to express my deepest gratitude to my advisor, Dr. William Ridgway for his excellent guidance on my research as well as great patience and caring whenever I was having difficulties, especially when my health condition was not in a good shape and when I had to face the stress from thesis writing and the newborn baby during the last several months.

It is also my earnest gratitude to my parents for their unfailig love throughout my lifer, their understanding of my hard work and for taking care my baby during the hardest time of thesis writing. I would like to thank my husband for helping with the baby care, although he is working at far south coast. I would like to thank my lovely son for bringing the good luck to my research: the discovery of the genetic defect in NOD.Abd3 model followed right after the news of my pregnancy!

I thank David, Kritika, Kyle and Yuehong for all their help on my research and my life during the past five years. It is really my honor to be in this lab and I really enjoy the free and warm atmosphere in the lab!

iv

Table of Contents

Dissertation abstract ...... i Acknowledgements ...... iv

Chapter 1 General introduction ...... 1 The morphological and functional heterogeneity of cholangiocytes ...... 1 PBC etiology ...... 3 Autoantigens in PBC ...... 4 Molecular mimicry of PDC-E2 ...... 5 Genetics of PBC ...... 6 Epigenetics of PBC ...... 8 Animal models of PBC ...... 9 NOD congenic strains...... 9 Dominant negative TGFβ type II receptor (dnTGFβRII) mouse model ...... 12 Xenobiotic-induced model ...... 13 TGFβ and its signaling ...... 14 TGFβ ...... 14 TGFβ-Smad signaling ...... 16 Non-Smad signaling ...... 17 TGFβ signaling and microRNA biogenesis...... 17 Breakdown of liver tolerance in PBC ...... 18 Breakdown of innate immune tolerance in PBC ...... 19 Breakdown of adaptive immune tolerance in PBC ...... 20 Dysregulated adaptive immune system in PBC ...... 21 Cholangiocytes as an active player in PBC ...... 23 Summary ...... 25 References ...... 26

Chapter 2 Murine autoimmune cholangitis requires two hits: cytotoxic KLRG1+ CD8 effector cells and defective T regulatory cells ...... 35 Abstract ...... 37 Introduction ...... 39 Materials and methods ...... 40

v

Results ...... 44 Discussion ...... 48 Conclusion ...... 52 Figure ...... 53 References ...... 65

Chapter 3 Aberrant Pkhd1 expression causes autoimmune biliary disease on the NOD genetic background ...... 69 Abstract ...... 72 Introduction ...... 73 Materials and methods ...... 75 Results ...... 82 Discussion ...... 89 Figure and table ...... 94 References ...... 107

Chapter 4 Summary, discussion and future directions ...... 110 Summary, discussion and future directions of dnTGFβRII model ...... 110 Working model of dnTGFβRII ...... 111 Future directions of dnTGFβRII model ...... 112 Summary, discussion and future directions of NOD.Abd3 model...... 115 Working model of NOD.Abd3 ...... 116 Future directions of NOD.Abd3 mouse model ...... 117 Summary statement ...... 123

vi

Chapter 1 General introduction

The biliary tract is a complex three-dimensional tree-like interconnected system of tubular conduits in the gastrointestinal system that stores as well as drains bile secreted from liver into the duodenum to digest fat. It includes bile ducts inside and outside the liver and gallbladder [1]. The epithelium of bile ducts is lined with cells called cholangiocytes which demonstrate changes in morphological features, secretory function and proliferative and apoptotic responses to liver injury [2, 3]. Cholangiopathies are a group of chronic liver diseases with a common central target, the cholangiocytes, showing heterogeneous pathogenesis and a variable natural history. These diseases can be classified into various categories including genetic, immune-mediated, idiopathic, and malignant. Among them, primary biliary cirrhosis (PBC) which is the focus of this dissertation, is one of the most common forms of cholangiopathy. Other common conditions include primary sclerosing cholangitis, polycystic liver disease, cystic fibrosis involving the liver, biliary atresia and cholangiocarcinoma [4]. Between 1988 and 2015, about 19% of all liver transplants performed in the United States are for cholangiopathies, representing nearly two thirds of all liver transplants performed for chronic hepatitis C during the same period. Due to lack of effective medical therapies, cholangiopathies progress with liver transplantation as the end-stage therapy. Thus, the cholangiopathies account for substantial morbidity, mortality and economic burden with an annual expenditure for transplant for these conditions being about $400 million in

2011 [4]. Current understanding of the cholangiopathies is limited due to the complex interacting factors of genetics, epigenetics and environment. This dissertation mainly focuses on PBC and aims to present new discoveries on the animal models of this disease, shedding light on a better insight of the disease pathogenesis.

The morphological and functional heterogeneity of cholangiocytes

There are two distinct populations identified in the cholangiocytes isolated from normal rat liver by counterflow elutriation: small (about 8 um in diameter) and large (about 14 um in diameter)

1 cholangiocytes. There is a strong relationship between the size of the cholangiocytes and the diameter of ducts with small cholangiocytes lining the small ducts (<15 um in diameter) whereas large cholangiocytes line the large ducts (>15 um in diameter) [5]. Furthermore, the nucleus-to- cytoplasm ratio of cholangiocytes is inversely correlated to the size of the ducts and the shape of the cells in small ducts differs from that in large ducts [6]. Microarray analysis in small and large cholangiocytes from normal mice shows a striking heterogeneity in gene expression along the biliary epithelium with 230 genes differentially expressed in the two populations of cholangiocytes [7], confirming the presence of two populations of cholangiocytes lining the biliary tree and emphasizing the difference in their secretory functions.

Cholangiocytes actively participate in up to 40% of the bile flow mainly through the activation of

- - an array of ion channels, transporters and exchangers (Cl /HCO3 anion exchanger2, AE2) leading to the active secretion of bicarbonate in bile [8]. The activation of basolateral secretin receptor

(SR) induces an increase in cAMP level in cholangiocytes resulting in the phosphorylation of cystic fibrosis transmembrane conductance regulator (CFTR) and thus pumping Cl- out of the cell

- - - [9]. Once a favorable Cl gradient across the plasma membrane is reached, the Cl /HCO3 anion exchanger2 (AE2) is activated [10] which leads to the secretion of bicarbonate into bile [11].

Consistent with the gene and protein profiling the small and large cholangiocytes, the core machinery required for bicarbonate secretion is constitutively expressed in large cholangiocytes but missing in small cholangiocytes [3, 5, 12].

Besides the difference in morphology and secretory function, large and small cholangiocytes differ in specific proliferative or apoptotic responses shown in several liver injury models. For example, in a model of cholestasis induced by bile duct ligation (BDL), only large cholangiocytes proliferate in response to damage [13, 14] in a estrogen dependent manner [15, 16]. The proliferation of large cholangiocytes after BDL is also dependent on VEGF signaling as neutralizing VEGF in vivo decreases cholangiocytes proliferation [17]. In fact, large cholangiocytes not only proliferate but also acquire a neuroendocrine-like phenotype

2 characterized by the expression of neuroendocrine markers, the secretion of various cytokines, growth factors, neuropeptides and hormones as well as enhanced response to circulating hormones and neuropeptides, allowing cholangiocytes to interact with other cell types including immune cells recruited in the liver due to damage [18]. On the other hand, in a model with administration of carbon tetrachloride (CCl4) to BDL or normal rats, large cholangiocytes are selectively destroyed while small cholangiocytes proliferate to replenish the loss of large cholangiocytes. The protection from damage of small cholangiocytes is mainly due to the lack of the cytochrome enzyme P4502E1 that initiates CCl4-induced liver damage [19, 20]. Interestingly, the differentiation of small cholangiocytes into large cholangiocytes seems to be Ca2+ dependent as the administration of GABA to BDL rats which induces elevated intracellular level of Ca2+ selectively damages large cholangiocytes and promotes small cholangiocytes to proliferate and acquire typical morphological features of large cholangiocytes and these results can be inhibited by Ca2+ chelators [21]. Taken together, these features indicate that small cholangiocytes may represent a stem cell compartment in the liver contributing to the cholangiocyte population of large bile ducts [22].

PBC etiology

Primary biliary cirrhosis (PBC) is a chronic autoimmune liver disease characterized by slow progressive destruction of bile duct cells in the liver, the infiltration of innate and adaptive immune cells around the intrahepatic bile duct leading to liver failure and liver transplantation as the end-stage therapy [23]. PBC is a rare disease with an annual incidence of 3.3-58 cases per million-population and point prevalence rate of 19.1-402 cases per million-population [24]. The serological hallmark of the disease is the presence of anti-mitochondrial antibodies which are mainly directed against pyruvate dehydrogenase complex E2 subunit (PDC-E2). These antibodies are present in nearly 95% of the PBC patients. The exclusive presence of these antibodies in the

PBC patients makes them a good criterion to diagnose the disease. However, the titer of the

3 antibody is not associated with the severity of the disease (eg. High titer of the antibody at the early and advanced stages and the high titer in patients with liver transplantation with or without recurring PBC). In PBC patients, the majority of liver infiltrating T cells are reactive to PDC-E2 and there is over 100-fold increase in PDC-E2 specific CD4 T cells and 10-fold increase in PDC-

E2 specific CD8 T cells in the liver compared to those in the peripheral blood mononuclear cells

[25, 26], suggesting the significance of adaptive immune system in this organ-specific autoimmune disease. Unique apoptotic features of cholangiocytes may contribute to the presentation of an apotope, an immunogenic apoptosis fragment, to the immune system, causing unique tissue damage in PBC [27-29]. Perpetuation of inflammation may result in senescence of cholangiocytes, contributing to irreversible loss of bile duct. Ursodeoxycholic acid (UDCA), a naturally existing bile acid, is currently the only drug approved by Food and Drug Administration to treat PBC. It helps to increase the hydrophilic properties of the bile acid pool and produce bicarbonate-rich choleresis, which protects hepatocytes from damage [30]. Recent studies with both human patients and murine models have largely advanced our understanding of etiopathology of this autoimmune disease. Genetics plays an important role as shown by two independent genome-wide association studies [31, 32] as well as the concordance rate in monozygotic twins as high as 60% [33]. In addition, epigenetics and environmental factors are thought to contribute to the disease.

Autoantigens in PBC

The autoantigens in PBC are the 2-oxoacid dehydrogenase complexes (2-OADC) which is a multienzyme complex essential in mitochondrial respiratory chain. The enzyme family includes three complexes: PDC, the 2-oxo glutarate dehydrogenase complex (OGDC) and the branched change 2-oxoacid dehydrogenase complex (BCOADC), all of which are composed of three subunits: E1, E2 and E3. Antimitochondrial antibodies in PBC patients specifically recognized the lipoylated domains of these enzymes. The immunodominant epitopes contain an ExDKA

4 motif with the lipoic acid attached to lysine at position 173 [30]. Among the three 2-OADC constituents, the major autoepitope is the inner lipoyl domain of PDC-E2, which is the antigen first cloned in 1987 by Gershwin’s group [34], contributing to both cellular and humoral immune responses in PBC. In autoimmunity, although autoantigens don’t elicit a primary immune response by themselves, they can be recognized by effector T cells activated by pathogens through cross-reactive epitopes. These epitope mimics induce activation and proliferation instead of anergy of autoreactive T cells to break the self-tolerance to the autoantigens. In the case of

PBC, the epitope mimic can be either a mimotope from a microbe or a neoantigen generated by xenobiotic-modified self-antigen, activating the autoreactive lymphocytes that leaked out into the periphery. These epitope-specific T cells then recognize the intact PDC-E2 autoantigen released from cholangiocytes [35].

Molecular mimicry of PDC-E2

The term “molecular mimicry” was first coined by Raymond Damian in 1964, who suggested that antigenic determinants of microorganisms may resemble antigenic determinants of their host [36].

Although the concept was first proposed as a defense mechanism of a microorganism from the host to escape the host’s immune response against the microorganism, it has become a popular explanation for microorganisms to elicit autoimmune responses in the host based on a structural similarity between the microorganism or metabolite and self-protein [37]. Molecular mimicry as a trigger for autoimmune diseases has been extensively discussed [38-43]. Two triggers of molecular mimicry, infection and xenobiotics have been heavily investigated for their role in PBC pathogenesis. Bacteria such as E. coli [44], Pseudomonas aeruginosa [45], virus such as herpes simplex virus [46], mouse mammary tumor virus [47, 48] and Epstein-Barr virus [49], parasites

(trypanosomes and Ascaridiagalli [50]) and fungi (Saccharomyces cerevisiae [51]) all have been reported to associate with PBC. Specifically, sera from PBC patients react with human PDC-E2 and Escherichia coli and Novoshphingobium aromaticivornas [52]. Furthermore,

5 lipopolysaccharide (LPS), a component of gram-negative bacterial cell wall, can induce PBC-like liver histology in mice, with or without PDC-E2 [53]. The linear or conformational mimicry between human mitochondrial antigens and microbial proteins can disrupt immune tolerance by inducing cross-reactive antibodies and effector T cells or by promoting epitope spreading, a process which infection accelerates an ongoing autoimmune disease by local activation of antigen presenting cells [54].

Xenobiotics are chemicals that can change the molecular structure of self- or nonself-proteins, resulting in enhanced immunogenicity [55, 56]. Xenobiotics are an important etiological factor to

PBC because environmental chemicals may form reactive metabolites that can modify self- protein in the cholangiocytes to form neoantigens, which can be recognized by autoreactive T cells. Xenobiotic-induced or oxidative modification of mitochondrial autoantigens is a critical step leading to the loss of tolerance in PBC patients, with lipoic acid, an essential component of

PDC-E2 epitope being accessible to chemical modification [30, 57]. Replacement of the lipoic acid moiety with synthetic structures to mimic a xenobiotically modified lipoyl hapten or modification of lipoic acid on a panel of lipoic acid mimics can induce significantly better reactivity with sera from PBC patients than unmodified native domain [55, 58-60]. Moreover several xenobiotics mimicking lipoic acid can induce AMA production in rabbit [58], guinea pig

[61], NOD.1101 [62] and C57BL/6 [63] without the PDC-E2 peptide backbone. It is also reported that oxidative stress-induced liver damage can lead to a transiently higher frequency of AMA, especially in subjects with acetaminophen poisoning [64]. These results indicate xenobiotics as a potential etiological agent of PBC.

Genetics of PBC

It is widely accepted that genetic susceptibility results in a breakdown of immunological tolerance which can enhance the effector of autoimmune responses. Familial occurrence studies show that the relative risk of a family member of a first-degree relative of PBC patients is 50- to

6

100-fold higher than the general population [65]. High concordance rates of PBC in monozygotic and dizygotic twins also demonstrate the significance of genetic factors in the pathology of the disease [33]. The major histocompatibility complex (MHC) is the major compartment involved in antigen presentation, distinguishing self from non-self. Although there is an association between human leukocyte antigen (HLA) molecules and genetic susceptibility to PBC, populations in different races possess different risk and protective haplotypes of HLA molecules and there is no consistent and reproducible association between specific HLA alleles and clinical features of PBC

[30]. Besides HLA loci, recently four genome-wide association studies (GWAS) [31, 32, 66, 67] and two iCHIP-association studies [68, 69] in Europe, North America and Japan have identified dozens of novel disease-associated non-HLA risk loci. Together, these genes can be categorized into three immune functions: antigen presentation and T cell differentiation (class II HLA, IL12,

IL12R, IL7R, CD80, STAT4, TYK2, SOCS1), myeloid cell differentiation (IRF5, IRF8, SP1B and IL7R) and B cell function (SP1B, PLC-L2, IRF8, PLC-L2, CXCR5, IKZF3) [70, 71]. It is interesting to note that all these associations identified in the genetic studies of PBC with large and well-characterized patient cohorts are related to immune responses, adaptive or innate.

Therefore, the results confirm that PBC is an immune disease at least at the level of genetics. It will be intriguing to discover whether susceptibility loci related to the biology of the key autoantigen, PDC-E2 and to the target cell, cholangiocytes can be found in GWAS performed with more power. An alternative explanation for the absence of these autoantigen and target cell related loci in genetic studies is that they are potentially modulated at epigenetics level or the relevant protein products are targets of xenobiotics, both of which playing important roles in etiology of PBC.

In addition, female predominance and high risk of the disease in female relatives of PBC patients suggest a link between X chromosome and the autoimmunity in the disease pathology of PBC.

This is reasonable because many genes involved in immunological tolerance are located on the X chromosome [72]. Indeed, several major defects in sex chromosomes of PBC patients have been

7 reported, for example, the frequency of X monosomy in female PBC patients is significantly enhanced compared to control females and the frequency of X monosomy increases with age, which may explain why PBC occurs predominantly in middle-aged women. Interestingly, X monosomy is more frequent in peripheral T and B cells than innate immune cells such as monocytes and natural killer cells [73, 74], suggesting a specific defect in cell division during proliferation and/or differentiation of hematopoietic system. Besides the higher frequency of X chromosome monosomy in PBC patients, it was recently found that Y chromosome loss is higher in PBC males compared to healthy male controls and this phenomenon also increases with age. In female PBC patients, loss of one X chromosome may lead to haploinsufficiency of X-linked loci, thus escaping the X-chromosome inactivation process. In male PBC patients, Y chromosome loss may represent an analogue to what happens in female, causing an imbalance for the alleles shared with X chromosome [75]. Therefore, sex chromosome abnormalities might well constitute the common feature of the genetic susceptibility to PBC.

Epigenetics of PBC

Epigenetic modifications are stable and heritable gene expression patterns without any alterations to the DNA sequence. There are four main known epigenetics mechanisms: 1) methylation of

DNA on specific cytosine residues to silence gene expression; 2) post-translational modification of histone tails of nucleosomes (eg. Acetylation, methylation, ubiquitination, etc.) which my activate or inactivate gene transcription depending on the histone modified and the nature of the modification; 3) active remodeling of chromatin by remodeling complexes that can either enhance or suppress gene expression and 4) silencing gene expression by small noncoding RNAs

[76]. Epigenetics may contribute to the PBC pathology as shown in the case of CD40 and CD40 ligand which play a central role in T cell priming, B cell maturation and immunoglobin class- switch recombination. Although no gene mutations were detected in the sequence of CD40 ligand from PBC patients [77], the DNA methylation of CD40 ligand promoter is significantly lower in

8

CD4 T cells from PBC patients compared with controls and such decreased methylation inversely correlated with the serum IgM level in PBC patients [78], strongly suggesting an epigenetic factor in the pathogenesis of PBC. In another study, two X-linked candidate genes CLIC2 and PIN4 were consistently downregulated in an affected twin of discordant pairs, with no differential methylation detected in promoter region, suggesting a mechanism involving epigenetic factors more complex than methylation differences at X-linked promoters in the context of female predominance of PBC [79]. Additionally, small non-coding RNAs designated microRNAs are modulator of gene expression and play critical roles in various immune functions and immune cell development [80]. PBC patients show a unique serum [81] and liver [82] microRNA expression profile compared to normal controls, with the majority of microRNAs significantly downregulated. This is consistent with what is found in the PBC murine model, dominant negative type II TGFβ receptor (dnTGFβRII) [83]. However, it remains to be elucidated what role microRNAs play in the breakdown of the immune tolerance in PBC.

Animal models of PBC

PBC as a chronic autoimmune liver disease, and usually takes years to show manifestations in patients. Given such slow progression, it is hard to study the etiologic mechanisms of the disease directly from patients. Fortunately, in the past two decades, several murine models were developed that resemble the pathophysiological alterations of PBC and they can be divided into two categories: spontaneous models and induced models. Although none of these models perfectly resemble the complex scenario of PBC, they are all valuable in elucidating the etiopathology of the human disease and potentially shedding new lights on therapeutic approaches.

NOD congenic strains

The NOD mouse is a widely used animal model for autoimmune type 1 diabetes (T1D) and the development of NOD congenic strains identifies the genetic basis of T1D in NOD [84]. The

9

NOD.c3c4 mouse was the first mouse model described to resemble PBC manifestations including the presence of anti-PDC-E2 antibodies (with around 60% incidence at the age of three to four months), CD3+, CD4+, CD8+ and eosinophil infiltration around intrahepatic biliary ductules, progressive biliary obstruction, non-suppurative destructive cholangitis (NSDC), granuloma formation and fibrosis. This mouse model is a serendipitous finding during the exploration of insulin dependent diabetes (Idd) resistant alleles on chromosomes 3 and 4 derived from B6/B10 introgressed on non-obese diabetic mice. While fully protected from diabetes, NOD.c3c4 develops PBC-like disease. The adaptive immune system plays a critical role in the development of the liver disease in this model as shown by the fact that transferring CD8+ splenocytes cells from diseased NOD.c3c4 mice to non-diseased NOD.c3c4-scid mice will cause severe lymphocytic infiltration in the liver and depletion of T cells by anti-CD3 antibody successfully ameliorates the disease in NOD.c3c4 [85, 86]. Another NOD congenic strain which carries smaller regions of the B10-derived diabetes protective loci on chromosome 3 and 4, called

NOD.ABD also shows similar abnormal liver histology [87]. The incidence of anti-PDC-E2 antibodies in NOD.ABD is lower than that of NOD.c3c4 probably because of the presence of

NOD Idd9.3 on chromosome 4 instead of an antibody-enhancing B10 Idd9.3 allele as in

NOD.c3c4. Adoptive transfer of NOD.ABD splenocytes causes severe autoimmune biliary disease (ABD) in NOD.c3c4-scid with high proportion of recipients developing anti-PDC-E2 antibody (70% in male recipients and 100% in female recipients) but without biliary epithelial hyperplasia, a scenario more similar to human PBC. However, NOD.ABD splenocytes couldn’t transfer ABD to NOD.scid . Furthermore, NOD splenocytes couldn’t transfer ABD to NOD.c3c4- scid. These results indicate that the disease onset requires abnormality in both target tissue, (the bile ducts) and the immune cells. As in dnTGFβRII mice, the ABD in these NOD congenic mice is mediated by CD8 T cells. Yang et al showed that NOD.ABD has more memory/effector CD8 T cells than MHC-matched B10 control and NOD.ABD but not NOD CD8 T cells can transfer

ABD to NOD.c3c4-scid and such disease transferring effect can be eliminated by co-transfer of

10

NOD.ABD Tregs [87]. On the other hand, however, the adoptive transfer study wasn’t able to distinguish whether the ability of NOD.ABD CD8 T cells to transfer ABD is due to a T cell activation mediated by certain genetic defect, i.e. a genetic region on NOD.ABD CD8 T cells; or due to the presence of autoreactive CD8 T cells in the NOD.ABD T cell repertoire which are lacking in NOD mice. It has also been widely noted that this mouse model has dramatic enlargement of common bile duct and intrahepatic biliary cyst which are never seen in human

PBC. One thing to note is that in a screening of genome-wise SNPs of NOD.c3c4 using a 5K

SNP chip, another B6/B10-derived genetic region has been found on chromosome one. To elucidate the role of such B6-derived fragment in the pathogenesis of PBC-like disease on the model, NOD congenic strain NOD.Abd3 carrying 1 Mb B6/B10-derived chromosome one region was developed. This new mouse line will be fully discussed in chapter three. Briefly, with this model, we identified the disease-causing gene on the target tissue, Polycystic kidney and hepatic disease 1 (Pkhd1) with a novel mutation. The gene is located only 2.5 Mb proximal to the 1 Mb

B6/B10-derived chromosome one region (Abd3 allele) and the mutation, which is a 6kb insertion in the intron between exon 35 and exon 36, doesn’t exist in any of the original parental NOD or

B6/B10 strains but arose during the breeding of these parental strains. The mechanism for which the mutation occurs is postulated to be related to certain transposon elements. Moreover, we have addressed the unsolved problem left over from the NOD.ABD splenic CD8 T cell adoptive transfer model by establishing bone marrow chimera with NOD.Abd3 reconstituted with

NOD.Abd3 or NOD bone marrow. Through this approach, we are able to raise NOD.Abd3 and

NOD immune cells in the same environment and demonstrate that the ABD-causing ability of

NOD.Abd3 immune cells is not due to different T cell repertoire because NOD bone marrow (BM) cells raised in lethally irradiated NOD.Abd3 didn’t cause disease as shown in the BM transfer study. In fact the clinical disease of NOD.Abd3 recipients of NOD BM at 120 days after BM transfer was ameliorated compared to that at the time of BM transfer (see chapter 3).

11

Dominant negative TGFβ type II receptor (dnTGFβRII) mouse model

This mouse model was first published by Dr. Flavell’s group in 2000 using a recombination vector overexpressing a truncated type II TGFβ receptor lacking the intracellular signaling domain under the CD4 promoter. Since the promoter didn’t include a suppressor in CD8 T cells, the transgene is expressed in both CD4 and CD8 cells [88]. The mouse develops autoimmune diseases in multiple organs and in liver, the manifestations resemble human PBC including the presence of anti-mitochondrial antibodies [89]. DnTGFβRII mice have elevated pro- inflammatory cytokines such as TNFα, IFNγ, IL6 and IL12p40. dnTGFβRII mice deficient in

IL12p40 have decreased disease manifestations including histological autoimmune cholangitis and intrahepatic pro-inflammatory cytokines [90] suggesting the importance of Th1 cells in the disease pathology. NKT cells exacerbate the autoimmune biliary disease [91] while B cells seem to be protective to the disease with enhanced autoimmune cholangitis in dnTGFβRII Igμ-/- mice

[92] although they are 100% AMA positive. However, the role of B cells in PBC of dnTGFβRII mouse is still controversial because in another study depleting B cells with anti-CD20 antibody in dnTGFβRII mice ameliorates the autoimmune cholangitis [93]. Interestingly, in a study of altered microRNA expression profile in peripheral blood mononuclear cells from PBC patients, the pathway analysis show that the predicted target genes of these microRNAs are involved in TGFβ signaling [81]. In the liver of PBC patients, β-Spectrin, an adaptor protein in TGFβ signaling pathway recruiting Smads to the receptor [94], is mislocalized in cytoplasm instead of vicinity of cell membrane compared to normal control [95]. Disruption of β-Spectrin in mice leads to disruption of TGFβ signaling in mice [94] and phenotypes include portal inflammation with T cell proliferation and absent intrahepatic bile ducts [95]. Therefore, the association of the mislocalization of β-Spectrin with PBC clearly suggests the significance of TGFβ signaling pathway in the PBC pathogenesis.

Similar to PBC patients, dnTGFβRII mice have an elevated CD8/CD4 ratio [89]. To study the significance of the T cell in the disease pathology in the mouse model, dnTGFβRII mice were

12 crossed to Rag1-/- mice and the dnTGFβRII Rag1-/- mice were found to have dramatically reduced autoimmune cholangitis. Furthermore, similar to CD8 T cells of NOD.ABD mice in disease pathology, splenic CD8 T cells from diseased dnTGFβRII mice can transfer autoimmune cholangitis into B6 Rag1-/- while splenic CD4 can only transfer inflammatory bowel disease into

B6 Rag1-/- recipients. Interestingly, unlike the adoptive transfer system using NOD.ABD

(donor)/NOD.c3c4-scid (recipient), splenic loss of tolerance in dnTGFβRII itself is sufficient to cause autoimmune biliary disease in B6.Rag1-/- recipients with no requirement of specific abnormality in target tissue.

Xenobiotic-induced model

Xenobiotic-modified lipoyl moiety of the major mitochondrial autoantigen PDC-E2 is hypothesized to lead to the loss of tolerance in PBC [55]. As mentioned in previous paragraphs, sera from PBC patients can react rigorously to the mimics of inner lipoyl domain of PDC-E2 [60].

6-bromohexanoate is among the first xenobiotic agents identified to break tolerance and induce

AMAs in both rabbits and guinea pigs [58, 61, 96]. More complex quantitative structure-activity relationship analysis identified 2-octynoic acid (2-OA), as a more reactive hapten recognized by

AMAs. Although not found in nature, the methyl and ethyl esters of this compound are widely used in cosmetic products and many common food flavorings. Administration of xenobiotic, 2-

OA, coupled to bovine serum albumin (BSA) but not BSA alone to C57BL/6 causes production of AMA and human PBC-like liver histopathology [63]. Moreover, NOD.1101 (NOD.B6 Idd10

Idd18r2) mice also develop high titer of AMAs, portal inflammation and autoimmune cholangitis similar to human PBC after administration of 2-OA coupled with BSA [62]. NOD.1101 mouse shares a portion of the B6-derived chromosome 3 congenic region with NOD.c3c4 mouse with

B6-derived insulin dependent diabetes (Idd) loci 10 and 18r2 introgressed onto the NOD background but doesn’t develop spontaneous autoimmune biliary disease as seen in NOD.c3c4 mouse and is partially protected from diabetes compared with the NOD parental strain [62].

Interestingly, like in dnTGFβRII mouse deficient in IgM [92], B cells are indicated to be

13 protective to the PBC-like liver disease in NOD.1101 as B cell depletion by anti-CD10 and CD79 monoclonal antibodies leads to exacerbated autoimmune cholangitis, higher T cell infiltrates and pro-inflammatory cytokines [97]. Furthermore, immunization of NOD.1101 with the addition of

α-galactosylceramide (α-GalCer), an invariant NKT cell activator leads to a profound exacerbation of autoimmune cholangitis, increased level of AMAs as well as fibrosis, a feature not previously reported in other murine models of PBC [98]. Both gain and loss of function of

NKT cells modulate PBC implicating innate immunity in the pathogenesis of PBC. These data may explain the relative failure of immunosuppressive medication to alter PBC because of their ineffectiveness against innate immunity [99].

TGFβ and its signaling

TGFβ

Transforming growth factor beta (TGFβ) was first identified by two independent groups as a growth-stimulatory molecule for its ability to induce colony formation of mouse embryo-derived fibroblastic AKR-2B cells [100] as well as the transformation of normal rat kidney (NRK) cells

[101] using soft agar assay. These studies were important findings after the discovery of Todaro and De Larco [102] that a “factor” secreted by virally transformed cells can cause phenotypic transformation of NRK cells. Although TGFβ was discovered during an intensive exploration of autocrine factors secreted by cancer cells to promote transformation, it was soon apparent that

TGFβ is not only a growth promoting factor but also can be a growth inhibitory factor [103] and can regulate diverse developmental and homeostatic processes and is mutated in numerous human diseases. The pleiotropic effects of TGFβ largely depend on the cell type and context [104]. Later studies of TGFβ on inflammation and immunity further demonstrate that it is a multifunctional regulatory molecule. Although TGFβ was first described to inhibit IL2-dependent proliferation of

T cells [105] and IL2-dependent B cell proliferation, Ig secretion and differentiation [106], it was

14 soon discovered that TGFβ can both stimulate and inhibit IgA secretion by B cells [107] and it can both activate and deactivate macrophages [108, 109].

The mammalian TGFβ superfamily includes over 35 structurally related polypeptide growth factors which can be phylogenetically divided into two main groups: TGFβ/Activin branch and bone morphogenetic protein (BMP)/growth and differentiation factor (GDF) branch., including activing/inhibin family, bone morphogenetic proteins (BMPs), growth differentiation factors

(GDFs), the TGFβ family and glial cell line-derived neurotrophic factor (GDNF) family. TGFβ has profound effects on immune system and participates in cell proliferation, differentiation, migration and apoptosis [110].

There are currently five known isoforms of TGFβ with TGFβ-1, -2 and -3 isolated from mammalian tissues [111] whereas TGFβ-4 and -5 isolated from chick and Xenopus, respectively

[112]. For the three mammalian TGFβ isoforms, the amino acid identity of the same isoform across different species is >97% and the identity among different isoforms is 71%~76% [112,

113]. Therefore, the presence of multiple forms of TGFβ and the conserved sequences of individual isoforms suggests that these isoforms mediate distinct and evolutionarily conserved processes [112]. Indeed, although TGFβ isoforms have a similar receptor-binding pattern and have similar effects in most in vitro biological assays, recent studies suggest that the

TGFβ isoforms can also have selective actions in different situations. For example, TGF-β1 and -

β3, but not -β2, strongly inhibit the growth of some endothelial cells [114, 115] and hematopoietic cells [116]. Phenotypes of knockout of specific TGFβ isoform do not overlap, also indicating there is non-redundant effect among the three isoforms. For example, targeted disruption of TGFβ-1 gene in mouse results in multifocal inflammatory disease [117] and TGFβ-

3 knockout mice have abnormal lung development and cleft palate indicating defects of epithelial-mesenchymal interaction [118]; the TGFβ-2-null mice show multiple developmental defects which are lethal at perinatal stage and the phenotype is not overlapping with those shown in TGFβ-1 or -3-null indicating numerous non-compensated functions between the TGFβ

15 isoforms [119]. More interestingly, it has been recently reported that TGFβ-3-induced Th17 cells are functionally and molecularly distinct from TGFβ-1-induced Th17 cells and have a molecular signature that defines pathogenic effector Th17 cells in autoimmune disease [120].

TGFβ-Smad signaling

As a prototype of the TGFβ protein family, TGFβ-1 is the isoform predominantly expressed in immune system. TGFβ is synthesized in an inactive form and is activated by dissociation from latency-associated protein (LAP) that keeps TGFβ inactive and latent-TGFβ-binding protein

(LTBP) that targets the inactive TGFβ to specific locations in the extracellular matrix [121].

Activation of TGFβ involves cellular recognition of extracellular matrix-associated LTBPs and subsequent recognition of the associated latent TGFβ by integrins, proteolytic events including plasminogen-mediated activation of TGFβ [122]. Activated TGFβ in dimeric form binds to heterodimeric receptor complex consisting of type I and type II trans-membrane serine/threonine kinase subunits. Despite a large number of members in TGFβ superfamily, the number of receptors of these ligands is very limited: only seven type I receptors and five type II receptors encoded in the mammalian genome, with the type I receptor funneling the signal into either

TGFβ-Smad pathway (R-Smad2/3) or the BMP-Smad pathway (R-Smad1/5/8) [123]. Binding of

TGFβ dimer with the type I and type II receptor complex initiates intracellular signaling by triggering the phosphorylation and activation of type I receptor via the type II receptor [124, 125].

Two other cell-surface TGFβ-binding proteins of known structure are betaglycan (also called type

III receptor) and endoglin [126, 127]. Betaglycan is a membrane-bound proteoglycan having a large extracellular domain and a very short cytoplasmic tail with no apparent signaling motif, however it is involved in the presentation of TGFβs to type I and II receptors for signaling [128].

Endoglin is a disulfide-linked homodimeric glycoprotein showing about 70% structural homology to betaglycan at the transmembrane and short cytoplasmic regions of each subunit with high expression level in vascular endothelial cells [127, 129]. TGFβ receptors have different binding affinities for the TGFβ isoforms. In general, type I and type II receptors bind more efficiently to

16

TGFβ-1 and TGFβ-3 than to TGFβ-2 [130]. The intracellular signaling of TGFβ is largely executed through Smad proteins. The activated type I receptor phosphorylates receptor Smads, called “R-Smad”, which form heteromeric complexes with the common mediator Smad4. This complex then translocates to the nucleus and binds specifically to DNA sequences containing a

Smad-binding element (SBE) in cooperation with other DNA binding partners including transcriptional factors to trigger specific downstream gene transcription. The R-Smads can also interact with transcription co-activators or co-suppressors to modulate transcription outcome. On the other hand, TGFβ signaling can also be negatively regulated by the inhibitory Smads, Smad 6 and Smad7 which are the transcription target of R-Smads though either competing with R-Smads to bind with the type I receptor or recruiting ubiquitin ligase complexes to degrade the type I receptor via proteasome [123].

Non-Smad signaling

In addition to Smad-dependent pathways, binding of TGFβ ligands to type I and type II receptors can initiate intracellular signaling through a Smad-independent manner. For example, the type I receptor can activate TRAF6 to induce numerous kinase cascades such as Erk, Jun and p38 Mark pathways [131]. The cross-talk between Smad-dependent and Smad-independent pathways as well as the integration of these pathways in a higher order network with other pathways, such as

Wnt pathway, Hedgehog pathway and Hippo pathway [132] in diverse biological conditions further demonstrates the universal and major importance of TGFβ signaling. Therefore, the manipulation of TGFβ signaling and the interpretation of the outcome in various experimental settings needs careful clarification.

TGFβ signaling and microRNA biogenesis

TGFβ signaling is recently found to directly modulate the biogenesis of a selective set of microRNAs including miR21 through ligand-specific SMAD signal transducer recruited with

DDX5 in the DROSHA microprocessor complex [133]. Interestingly, in the dnTGFβRII mouse model which has decreased TGFβ signaling in CD4+ and CD8+ T cells, the microRNA expression

17 profile is extensively disrupted with the majority being downregulated [83]. Argonaute (Ago) proteins as the core proteins of miRNA-induced silencing complex (miRISC) during microRNA biogenesis are inducibly ubiquitinated and thus post-transcriptionally down-regulated during T cell activation, reducing the microRNA abundance [134]. It is known that dnTGFβRII mice have overall activated CD4 and CD8 effector T cells profile compared to normal control [88] and thus may have skewed microRNA expression profile mediated by decreased level of Ago proteins. It is interesting to know whether the activation status of T cells in dnTGFβRII mouse model can be tuned to a lower level by manipulating the microRNA biogenesis which could be a potential therapeutic approach for PBC patients.

Breakdown of liver tolerance in PBC

The liver is an immunological organ that displays a considerable tolerance effects, as initially recognized through spontaneous liver allograft acceptance between different species with complete mismatch of MHC molecules [135]. It is acknowledged that non-paranchymal liver cells, including Kupffer cells, liver sinusoidal endothelial cells (LSEC), hepatic stellate cells, resident dendritic cells (including myeloid and plasmacytoid dendritic cells) and T regulatory cells all participate in liver tolerogenicity and mediate local and systemic tolerance to self and foreign antigens through secreting anti-inflammatory cytokines such as TGFβ and IL-10 and expressing negative co-stimulatory molecules like PD-L1 [136]. Other mechanisms including microchimerism; soluble major histocompatibility complex and regulatory T cells can also participate in the tolerance induction. The breakdown of immune tolerance in PBC liver can be categorized into two causes: the breakdown of innate immune tolerance and adaptive immune tolerance, which are closely interconnected with each other.

Both human patients and murine models of PBC have demonstrated that CD8 T cells play an important role in the etiopathology of this autoimmune liver disease. However, in the normal physiological condition, the CD8 T cells in the liver, scrutinizing various antigens brought in

18 through gut/portal vein circulation or self-antigens directly or cross presented by different hepatic cell types, are mainly tolerant instead of being effectors. Such tolerance of CD8 T cells is thought to be achieved by various mechanisms with different hepatic cell types involved. A detailed understanding of CD8 T cell phenotype and function in connection with other cell types in the liver is therefore crucial to elucidate the mechanism(s) how liver CD8 T cells become autoreactive in PBC.

Breakdown of innate immune tolerance in PBC

Several mechanisms involving different hepatic cell types have been proposed to elucidate the breakdown of innate immune tolerance. As a unique interface bridging innate and adaptive immunity, activated and mature DCs can prime robust adaptive immune responses (although immature DCs can result in immune tolerance ) [137]. Peripheral and intrahepatic DC number as well as their antigen presentation and cross presentation function are found to be increased in

PBC patients compared to control [26, 138, 139]. Circulating monocyte and liver macrophage have also been implicated in the pathogenesis of PBC due to hypersensitivity to infectious stimuli via increased TLR4 expression, higher level of pro-inflammatory cytokine [140] and decreased level of negative TLR4 modulator RP105 on monocytes [141] as well as increased TLR3 expression on macrophages around the portal tract [140]. In addition, NK cells and NKT cells may also play an important role in early breakdown of tolerance in PBC. Liver lymphocytes are enriched with NK cells and NKT cells. Peripheral NK cells enter the liver and differentiate into liver-specific NK cells, which express TRAIL and have enhanced cytotoxicity. NK cells in the liver can be activated by a variety of cytokine such as IFNα, IFNγ, IL-12, IL-18, IL-2 or IL-15 or by NKG2D ligands that are expressed on activated Kupffer cells and hepatic stellate cells, and damaged hepatocytes. Immature NKT cells enter the liver and can be activated by several cytokines or lipid antigens presented by CD1d on DCs and stellate cells. Both types of cells can produce copious amounts of cytokines when activated, playing important roles in liver injury, fibrosis, and repair [142]. Increased number of NK cells and NKT cells are found in both

19 peripheral blood and liver of PBC patients and these cells express more perforin and have higher cytotoxic activity against autologous cholangiocytes compared to healthy controls [143-145].

Moreover, in three different mouse models of PBC (dnTGFβRII [91], 2-OA-BSA induced model

[98] and N.aromaticivorans induced model [52]), NKT cells are found to be involved in disease exacerbation. Data from human and murine models strongly suggest a critical role for NKT cells in breaking down the immune tolerance in PBC.

Breakdown of adaptive immune tolerance in PBC

All hepatic cell types express MHC-I molecules and are therefore potential antigen presenting cells (APCs) for CD8 T cells. However, CD8 T cells activated by hepatocytes die sooner than those activated by splenic dendritic cells in vitro. Such premature death of CD8 T cells is due to insufficient expression of IL-2 and survival genes such as bcl-xl as well as lack of efficient costimulation [146, 147]. Similarly, transgenic CD8 T cells activated by LSEC secreted significantly lower amount of IL-2 and IFNγ and had lower cytotoxicity compared to CD8 T cells activated by splenocytes [148]. These results indicate that death by neglect may be a mechanism of tolerance for CD8 T cells activated by liver APCs. Recently, a study by Tay et al utilizing recombinant adenoassociated viral vectors to manipulate antigen expression level within the liver shows that a threshold of intrahepatic antigen expression is the dominant factor determining the fate of CD8 T cells recognizing antigen in vivo, with high-level antigen expression leading to T cell exhaustion and low-level antigen expression leading to T cell expansion and execution of effector functions [149]. More importantly, the activation-induced cell death and exhaustion of intrahepatic CD8 T cells can be reversed when these cells are transferred to an environment with low level of antigen expression. Thus, the primary hepatic CD8 T cells activation is dynamic and depends on antigen expression level in the microenvironment these cells encounter, explaining a mechanism why intrahepatic CD8 T cells responses range from deletional tolerance to full effector function in various physiological and pathological conditions. Cholangiocytes, the cellular target of autoreactive CD8 T cells in PBC patients and murine models, constitutes about

20

3% of total cells in the liver [150]. Therefore, the expression level of autoantigen with intact epitope is potentially low. Given by this theory, it would be interesting to know whether increasing the expression level of the autoepitope can tolerize CD8 T cells which are otherwise fully activated.

Dysregulated adaptive immune system in PBC

The enrichment of autoreactive CD4 T cells and CD8 T cells in the portal tracts of the liver as well as around the damaged bile ducts of PBC patients strongly suggests the role of adaptive immune system in the development of PBC [25, 151-154]. Indeed, there is approximately 100- fold increase in PDC-E2-specific precursor CD4 T cell frequency and 10-fold increase in the frequency of PDC-E2-specific CD8 T cells in the liver versus peripheral blood in the PBC patients[25, 155]. More interestingly, although autoreactive T cells are found in both autoimmune disease and healthy controls, their activation states, co-stimulation signal requirements are significantly different [156-158]. It has been shown previously that PBMC from PBC patients are pro-inflammatory regardless of whether co-stimulation-competent or –incompetent APC were used, suggesting they are activated/memory cells, whereas PBMC from healthy control become anergic in the lack of costimulatory signal. The anergic autoreactive T cell clones produce IL-10 but no IFNγ or IL-4 and exhibit regulatory functions in an antigen-dependent, cell contact- independent manner [159]. Therefore, the qualitative difference between autoreactive T cells from PBC patients and those from healthy controls may be dependent on the co-stimulation signals provided by APC.

On the other hand, CD4+Foxp3+ T regulatory cells (Tregs), a subset of CD4+ T cells, play an important role in the maintenance of peripheral tolerance and downregulation of immune responses. In PBC patients and their first-degree female relatives, there is a significantly lower frequency of Tregs in PBMC compared to normal control. In a mouse model deficient in IL-2 receptor α (CD25), which is highly expressed on Tregs and important for Treg development, there is a 100% penetration of AMA development [160]. Interestingly, a child born with genetic

21 deficiency of IL-2 receptor α has several clinical manifestations of PBC [161]. In addition, the

Scurfy mouse which has a mutation in Foxp3, the transcription factor essential for Treg development, lacks Tregs and develops high titers of serum AMA also at 100% penetration [162].

These results clearly indicate that Treg dysfunctionality can be a major characteristic in the loss of tolerance in PBC. Using dnTGFβRII mouse model, in which the selective deficiency of TGFβ signaling in CD4+ and CD8+ cells, we will learn how the peripheral tolerance is broken down at the view of Tregs because Tregs also depend on TGFβ for their regulatory function [163].

The role of autoantibodies and the B cells in PBC pathogenesis is more complex because except for one study [164], most studies suggest no correlation between disease severity and the level of serum AMA [165-167]. B cells actively participate in immune responses through antibody production, antigen presentation, T helper cell polarization and spleen architecture modeling.

However, certain groups of B cells can also act as immune suppressor and facilitate immune tolerance. There is a higher number of B cells in the liver of PBC patients compared to normal controls and CD19+CD69+ activated B cells with AMA-producing cells are enriched in the liver of PBC patients [168]. A unique coronal arrangement of CD38+ plasma cells has been detected only around intrahepatic bile ducts of PBC patients but not in healthy control, suggesting a role of plasma cells in the specific destruction of intrahepatic bile ducts in PBC [169]. In addition, in dnTGFβRII mouse model of PBC, depletion of B cells using anti-CD20 monoclonal antibody ameliorates the disease [93]. However, other studies show that B cells may have a protective role for PBC. For example, the treatment of anti-CD20 and anti-CD79 antibodies to deplete B cells in the 2-OA-BSA-induced mouse model exacerbates the autoimmune cholangitis [97] while the adoptive transfer of CD19+ cells from dnTGFβRII mice into Rag1-/- mice leads to decreased hepatobiliary disease [93]. These controversial results indicate that there may exist distinct B cell subpopulations that have completely opposite role in the PBC pathogenesis, either promoting or suppressing the disease.

22

Cholangiocytes as an active player in PBC

Cholangiocytes are thought not to be an innocent victim in PBC, instead playing an active role in the pathogenesis of PBC [170]. They are proved to contribute to the defense mechanism against infection because they can recognize pathogen-associated molecule patterns (PAMPs) via specific membrane toll like receptors (TLRs) [171-173], resulting in chemokine production and thus activating both innate and adaptive immune system contributing to the defense mechanisms against infectious agents . In addition, biliary epithelial cells are unique in secreting immunoglobulins A of the secretory type (sIgA) [174] and may be involved in tissue-specific autoimmune disease by processing anti-PDC-E2 IgA, contributing to the pathogenesis of PBC

[175]. In PBC patients, cholangiocytes act as non-professional APCs through expressing class II

MHC molecules which is not detected in normal controls, [176, 177] and costimulatory molecules such as CD86 [178].Cholangiocytes from PBC patients can phagocytose apoptotic neighboring cholangiocytes, potentially involved in presenting novel mitochondrial peptides associated with class II MHC molecules, which is also not detected in normal control [179, 180].

As a tissue specific autoimmune disease with PDC-E2 subunit as the predominant autoantigen, one major problem of PBC pathology remains enigmatic that the mitochondrial self-protein is ubiquitously expressed, only cholangiocytes are the target cells recognized by autoreactive lymphocytes in the disease. Recent studies have shown that cholangiocytes are different from other cell types in terms of glutathione modification of mitochondrial proteins when undergoing apoptosis. These studies strongly support that rather than an innocent victim, cholangiocytes play an active role in the pathogenicity of PBC [27, 28, 181] due to their unique biochemical property in processing PDC-E2 during apoptosis, compared to other cell types including epithelial cells.

Odin et al have reported that the lack of glutathionylation of PDC-E2 during apoptosis in cholangiocytes but not found in other cell types may cause the protein to survive immunologically intact and recognizable by autoreactive lymphocytes. Lleo et al found that the immunologically intact PDC-E2 is translocated to biliary apoptotic bodies and in the presence of

23

AMA attract macrophage from PBC patients, causing an intense pro-inflammatory environment

[28]. However, it is not clear whether there is any genetic difference in cholangiocytes between

PBC patients and healthy controls. Although cholangiocytes from PBC patient express higher level of WAF1, p53, Fas, FasL, granzyme B, peroferin and TRAIL and show more DNA fragmentation, suggesting more apoptosis of cholangiocytes in PBC patients compared to normal control [182-186], such differences are most likely secondary to the attack by the autoreactive lymphocytes. Current literature suggest that there is probably no genetic difference in cholangiocytes between PBC patients and normal control [187] because PBC may recur after liver [188] and the genetic susceptible loci found in a recent GWAS of PBC are not necessarily associated with cholangiocytes [32]. However, it should be noted that in the relatively large cohort study done by Charatcharoenwitthaya et al in Mayo Clinic [189], the cumulative incidence of recurring PBC is 21-37% of patients at 10 years after liver transplantation, and 43% at 15 years after liver transplantation, with the median time to recurrence of 3-5.5 years, indicating that there are cases with no disease recurrence and therefore there might be a heterogeneity in the quality of donor cholangiocytes. Furthermore, even in the cases where recurrence does happen, the severity of the recurrent PBC is significantly less compared to the primary condition [188]. Therefore, it is reasonable to hypothesize that there may exist certain predisposing cholangiocyte phenotype in

PBC compared to normal control, together with the unique biology of cholangiocyte, making these cells in PBC patients susceptible to autoimmune attack. This hypothesis is supported by our recent finding in NOD.Abd3 mouse model of PBC which is deficient in a gene crucial to cholangiocyte proliferation and biological functions, resulting in dilation of bile ducts. In this mouse model, it has been proved that both abnormal immune system and target tissue are required for the autoimmune biliary disease. Although bile duct dilation is not seen in PBC patients, we can’t exclude a possibility of structural compromise in the cholangiocytes of these patients, potentially induced by environmental factors such as the metabolites derived from chemicals and drugs modifying the self-proteins.

24

Summary

PBC as a chronic liver disease shows a great variance in clinical features and natural history between patients. Although much progress has been achieved in understanding the basic mechanisms of the loss of tolerance in this disease, very little is known about the relationship between the variable clinical course and its immune tolerance state. Furthermore, it is still unclear when and where the breach in immune tolerance starts in PBC. The two mouse models, dnTGFβRII (chapter 2) and NOD.Abd3 (chapter 3) as the foci of this dissertation provide a good opportunity to explore these questions. In particular, is the liver disease in dnTGFβRII completely mediated by CD8 T cells or do there other immune cells participate in the loss of tolerance to the target cells? How do these cells interact with each other to break the immune tolerance? We will focus on the immune manifestations in dnTGFβRII model to address the loss of tolerance in adaptive immunity. On the other hand, as NOD.Abd3 shows abnormality in target tissue at an early age and previous data suggests a genetic factor contributing to the disease, we ask what the genetic factor is and how it potentially links the target cells and immune system, leading to the break of immune tolerance to the target tissue at the early stage of the disease. The findings in these two mouse models will provide us with new insights in disease pathology as well as improved treatment.

25

References

1. Johnson, L.R., Physiology of the gastrointestinal tract. 4th ed. 2006, Amsterdam ; Boston: Elsevier Academic Press. 2. Kanno, N., et al., Functional heterogeneity of the intrahepatic biliary epithelium. Hepatology, 2000. 31(3): p. 555-61. 3. Glaser, S.S., et al., Morphological and functional heterogeneity of the mouse intrahepatic biliary epithelium. Lab Invest, 2009. 89(4): p. 456-69. 4. Lazaridis, K.N. and N.F. LaRusso, The Cholangiopathies. Mayo Clin Proc, 2015. 90(6): p. 791-800. 5. Alpini, G., et al., Morphological, molecular, and functional heterogeneity of cholangiocytes from normal rat liver. Gastroenterology, 1996. 110(5): p. 1636-43. 6. Benedetti, A., et al., A morphometric study of the epithelium lining the rat intrahepatic biliary tree. J Hepatol, 1996. 24(3): p. 335-42. 7. Ueno, Y., et al., Evaluation of differential gene expression by microarray analysis in small and large cholangiocytes isolated from normal mice. Liver Int, 2003. 23(6): p. 449- 59. 8. Strazzabosco, M., A. Mennone, and J.L. Boyer, Intracellular pH regulation in isolated rat bile duct epithelial cells. J Clin Invest, 1991. 87(5): p. 1503-12. 9. Kanno, N., et al., Regulation of cholangiocyte bicarbonate secretion. Am J Physiol Gastrointest Liver Physiol, 2001. 281(3): p. G612-25. 10. Martinez-Anso, E., et al., Immunohistochemical detection of chloride/bicarbonate anion exchangers in human liver. Hepatology, 1994. 19(6): p. 1400-6. 11. Banales, J.M., et al., Bicarbonate-rich choleresis induced by secretin in normal rat is taurocholate-dependent and involves AE2 anion exchanger. Hepatology, 2006. 43(2): p. 266-75. 12. Han, Y., et al., Recent advances in the morphological and functional heterogeneity of the biliary epithelium. Exp Biol Med (Maywood), 2013. 238(5): p. 549-65. 13. Alpini, G., et al., Molecular and functional heterogeneity of cholangiocytes from rat liver after bile duct ligation. Am J Physiol, 1997. 272(2 Pt 1): p. G289-97. 14. Alpini, G., et al., Heterogeneity of the proliferative capacity of rat cholangiocytes after bile duct ligation. Am J Physiol, 1998. 274(4 Pt 1): p. G767-75. 15. Alvaro, D., et al., Estrogens stimulate proliferation of intrahepatic biliary epithelium in rats. Gastroenterology, 2000. 119(6): p. 1681-91. 16. Alvaro, D., et al., Effect of ovariectomy on the proliferative capacity of intrahepatic rat cholangiocytes. Gastroenterology, 2002. 123(1): p. 336-44. 17. Gaudio, E., et al., Vascular endothelial growth factor stimulates rat cholangiocyte proliferation via an autocrine mechanism. Gastroenterology, 2006. 130(4): p. 1270-82. 18. Alvaro, D., et al., Proliferating cholangiocytes: a neuroendocrine compartment in the diseased liver. Gastroenterology, 2007. 132(1): p. 415-31. 19. LeSage, G.D., et al., Acute carbon tetrachloride feeding selectively damages large, but not small, cholangiocytes from normal rat liver. Hepatology, 1999. 29(2): p. 307-19. 20. LeSage, G.D., et al., Acute carbon tetrachloride feeding induces damage of large but not small cholangiocytes from BDL rat liver. Am J Physiol, 1999. 276(5 Pt 1): p. G1289-301. 21. Mancinelli, R., et al., GABA induces the differentiation of small into large cholangiocytes by activation of Ca(2+) /CaMK I-dependent adenylyl cyclase 8. Hepatology, 2013. 58(1): p. 251-63. 22. Turner, R., et al., Human hepatic stem cell and maturational liver lineage biology. Hepatology, 2011. 53(3): p. 1035-45. 23. Invernizzi, P., C. Selmi, and M.E. Gershwin, Update on primary biliary cirrhosis. Digestive and Liver Disease, 2010. 42(6): p. 401-408.

26

24. Podda, M., et al., The limitations and hidden gems of the epidemiology of primary biliary cirrhosis. J Autoimmun, 2013. 46: p. 81-7. 25. Shimoda, S., et al., Identification and precursor frequency analysis of a common T cell epitope motif in mitochondrial autoantigens in primary biliary cirrhosis. J Clin Invest, 1998. 102(10): p. 1831-40. 26. Kita, H., et al., Identification of HLA-A2-restricted CD8(+) cytotoxic T cell responses in primary biliary cirrhosis: T cell activation is augmented by immune complexes cross- presented by dendritic cells. J Exp Med, 2002. 195(1): p. 113-23. 27. Lleo, A., et al., Apotopes and the biliary specificity of primary biliary cirrhosis. Hepatology, 2009. 49(3): p. 871-9. 28. Lleo, A., et al., Biliary apotopes and anti-mitochondrial antibodies activate innate immune responses in primary biliary cirrhosis. Hepatology, 2010. 52(3): p. 987-998. 29. Kaplan, M.M. and M.E. Gershwin, Primary biliary cirrhosis. N Engl J Med, 2005. 353(12): p. 1261-73. 30. Wang, L., et al., Breach of tolerance: primary biliary cirrhosis. Semin Liver Dis, 2014. 34(3): p. 297-317. 31. Hirschfield, G.M., et al., Primary biliary cirrhosis associated with HLA, IL12A, and IL12RB2 variants. N Engl J Med, 2009. 360(24): p. 2544-55. 32. Liu, X., et al., Genome-wide meta-analyses identify three loci associated with primary biliary cirrhosis. Nat Genet, 2010. 42(8): p. 658-60. 33. Selmi, C., et al., Primary biliary cirrhosis in monozygotic and dizygotic twins: genetics, epigenetics, and environment. Gastroenterology, 2004. 127(2): p. 485-92. 34. Gershwin, M.E., et al., Identification and specificity of a cDNA encoding the 70 kd mitochondrial antigen recognized in primary biliary cirrhosis. J Immunol, 1987. 138(10): p. 3525-31. 35. Gershwin, M.E., et al., Primary biliary cirrhosis: an orchestrated immune response against epithelial cells. Immunol Rev, 2000. 174: p. 210-25. 36. Damian, R.T., Molecular Mimicry: Antigen Sharing by Parasite and Host and Its Consequences. The American Naturalist, 1964. 98(900): p. 129-149. 37. Blank, M., O. Barzilai, and Y. Shoenfeld, Molecular mimicry and auto-immunity. Clin Rev Allergy Immunol, 2007. 32(1): p. 111-8. 38. Chastain, E.M. and S.D. Miller, Molecular mimicry as an inducing trigger for CNS autoimmune demyelinating disease. Immunol Rev, 2012. 245(1): p. 227-38. 39. Fae, K.C., et al., How an autoimmune reaction triggered by molecular mimicry between streptococcal M protein and cardiac tissue proteins leads to heart lesions in rheumatic heart disease. J Autoimmun, 2005. 24(2): p. 101-9. 40. Ang, C.W., B.C. Jacobs, and J.D. Laman, The Guillain-Barre syndrome: a true case of molecular mimicry. Trends Immunol, 2004. 25(2): p. 61-6. 41. Burroughs, A.K., et al., Molecular mimicry in liver disease. Nature, 1992. 358(6385): p. 377-8. 42. Benoist, C. and D. Mathis, Autoimmunity. The pathogen connection. Nature, 1998. 394(6690): p. 227-8. 43. Ribeiro, F.M., et al., Lupus and leprosy: beyond the coincidence. Immunol Res, 2015. 61(1-2): p. 160-3. 44. Varyani, F.K., J. West, and T.R. Card, An increased risk of urinary tract infection precedes development of primary biliary cirrhosis. BMC Gastroenterol, 2011. 11: p. 95. 45. Kita, H., et al., Analysis of TCR antagonism and molecular mimicry of an HLA-A0201- restricted CTL epitope in primary biliary cirrhosis. Hepatology, 2002. 36(4 Pt 1): p. 918- 26.

27

46. Fujiwara, K. and O. Yokosuka, Frequent detection of immunoglobulin M anti-herpes simplex viral antibody in patients with primary biliary cirrhosis. Hepatology, 2012. 56(1): p. 395. 47. Xu, L., et al., Does a betaretrovirus infection trigger primary biliary cirrhosis? Proc Natl Acad Sci U S A, 2003. 100(14): p. 8454-9. 48. Selmi, C., et al., Lack of immunological or molecular evidence for a role of mouse mammary tumor retrovirus in primary biliary cirrhosis. Gastroenterology, 2004. 127(2): p. 493-501. 49. Morshed, S.A., et al., Increased expression of Epstein-Barr virus in primary biliary cirrhosis patients. Gastroenterol Jpn, 1992. 27(6): p. 751-8. 50. Uzoegwu, P.N., H. Baum, and J. Williamson, The occurrence and localization in trypanosomes and other endo-parasites of an antigen cross-reacting with mitochondrial antibodies of primary biliary cirrhosis. Comp Biochem Physiol B, 1987. 88(4): p. 1181-9. 51. Sakly, W., M. Jeddi, and I. Ghedira, Anti-Saccharomyces cerevisiae antibodies in primary biliary cirrhosis. Dig Dis Sci, 2008. 53(7): p. 1983-7. 52. Mattner, J., et al., Liver autoimmunity triggered by microbial activation of natural killer T cells. Cell Host Microbe, 2008. 3(5): p. 304-15. 53. Ide, T., et al., An experimental animal model of primary biliary cirrhosis induced by lipopolysaccharide and pyruvate dehydrogenase. Kurume Med J, 1996. 43(3): p. 185-8. 54. Agmon-Levin, N., B.S. Katz, and Y. Shoenfeld, Infection and primary biliary cirrhosis. Isr Med Assoc J, 2009. 11(2): p. 112-5. 55. Long, S.A., et al., Immunoreactivity of organic mimeotopes of the E2 component of pyruvate dehydrogenase: connecting xenobiotics with primary biliary cirrhosis. J Immunol, 2001. 167(5): p. 2956-63. 56. Long, S.A., J. Van de Water, and M.E. Gershwin, Antimitochondrial antibodies in primary biliary cirrhosis: the role of xenobiotics. Autoimmun Rev, 2002. 1(1-2): p. 37-42. 57. Leung, P.S., et al., Environment and primary biliary cirrhosis: electrophilic drugs and the induction of AMA. J Autoimmun, 2013. 41: p. 79-86. 58. Leung, P.S., et al., Immunization with a xenobiotic 6-bromohexanoate bovine serum albumin conjugate induces antimitochondrial antibodies. J Immunol, 2003. 170(10): p. 5326-32. 59. Naiyanetr, P., et al., Electrophile-modified lipoic derivatives of PDC-E2 elicits anti- mitochondrial antibody reactivity. J Autoimmun, 2011. 37(3): p. 209-16. 60. Rieger, R., et al., Identification of 2-nonynoic acid, a cosmetic component, as a potential trigger of primary biliary cirrhosis. J Autoimmun, 2006. 27(1): p. 7-16. 61. Leung, P.S., et al., Induction of primary biliary cirrhosis in guinea pigs following chemical xenobiotic immunization. J Immunol, 2007. 179(4): p. 2651-7. 62. Wakabayashi, K., et al., Induction of autoimmune cholangitis in non-obese diabetic (NOD).1101 mice following a chemical xenobiotic immunization. Clin Exp Immunol, 2009. 155(3): p. 577-86. 63. Wakabayashi, K., et al., Loss of tolerance in C57BL/6 mice to the autoantigen E2 subunit of pyruvate dehydrogenase by a xenobiotic with ensuing biliary ductular disease. Hepatology, 2008. 48(2): p. 531-40. 64. Leung, P.S., et al., Antimitochondrial antibodies in acute liver failure: implications for primary biliary cirrhosis. Hepatology, 2007. 46(5): p. 1436-42. 65. Parikh-Patel, A., et al., The geoepidemiology of primary biliary cirrhosis: contrasts and comparisons with the spectrum of autoimmune diseases. Clin Immunol, 1999. 91(2): p. 206-18. 66. Nakamura, M., et al., Genome-wide association study identifies TNFSF15 and POU2AF1 as susceptibility loci for primary biliary cirrhosis in the Japanese population. Am J Hum Genet, 2012. 91(4): p. 721-8.

28

67. Mells, G.F., et al., Genome-wide association study identifies 12 new susceptibility loci for primary biliary cirrhosis. Nat Genet, 2011. 43(4): p. 329-32. 68. Liu, J.Z., et al., Dense fine-mapping study identifies new susceptibility loci for primary biliary cirrhosis. Nat Genet, 2012. 44(10): p. 1137-41. 69. Juran, B.D., et al., Immunochip analyses identify a novel risk locus for primary biliary cirrhosis at 13q14, multiple independent associations at four established risk loci and epistasis between 1p31 and 7q32 risk variants. Hum Mol Genet, 2012. 21(23): p. 5209- 21. 70. Carbone, M., et al., Implications of genome-wide association studies in novel therapeutics in primary biliary cirrhosis. Eur J Immunol, 2014. 44(4): p. 945-54. 71. Kar, S.P., et al., Pathway-based analysis of primary biliary cirrhosis genome-wide association studies. Genes Immun, 2013. 14(3): p. 179-86. 72. Bianchi, I., et al., The X chromosome and immune associated genes. J Autoimmun, 2012. 38(2-3): p. J187-92. 73. Invernizzi, P., et al., Frequency of monosomy X in women with primary biliary cirrhosis. Lancet, 2004. 363(9408): p. 533-5. 74. Invernizzi, P., et al., X chromosome monosomy: a common mechanism for autoimmune diseases. J Immunol, 2005. 175(1): p. 575-8. 75. Lleo, A., et al., Y chromosome loss in male patients with primary biliary cirrhosis. J Autoimmun, 2013. 41: p. 87-91. 76. Zhou, V.W., A. Goren, and B.E. Bernstein, Charting histone modifications and the functional organization of mammalian genomes. Nat Rev Genet, 2011. 12(1): p. 7-18. 77. Higuchi, M., et al., Analysis of CD40 ligand gene mutations in patients with primary biliary cirrhosis. Scand J Clin Lab Invest, 1998. 58(5): p. 429-32. 78. Lleo, A., et al., Immunoglobulin M levels inversely correlate with CD40 ligand promoter methylation in patients with primary biliary cirrhosis. Hepatology, 2012. 55(1): p. 153- 60. 79. Mitchell, M.M., et al., Epigenetic investigation of variably X chromosome inactivated genes in monozygotic female twins discordant for primary biliary cirrhosis. Epigenetics, 2011. 6(1): p. 95-102. 80. Xiao, C. and K. Rajewsky, MicroRNA control in the immune system: basic principles. Cell, 2009. 136(1): p. 26-36. 81. Qin, B., et al., Analysis of altered microRNA expression profiles in peripheral blood mononuclear cells from patients with primary biliary cirrhosis. Journal of Gastroenterology and Hepatology, 2013. 28(3): p. 543-550. 82. Padgett, K.A., et al., Primary Biliary Cirrhosis is Associated With Altered Hepatic microRNA Expression. Journal of autoimmunity, 2009. 32(3-4): p. 246-253. 83. Ando, Y., et al., Overexpression of microRNA-21 is associated with elevated pro- inflammatory cytokines in dominant-negative TGF-beta receptor type II mouse. J Autoimmun, 2013. 41: p. 111-9. 84. Kachapati, K., et al., The B10 Idd9.3 locus mediates accumulation of functionally superior CD137(+) regulatory T cells in the nonobese diabetic type 1 diabetes model. J Immunol, 2012. 189(10): p. 5001-15. 85. Irie, J., et al., NOD.c3c4 congenic mice develop autoimmune biliary disease that serologically and pathogenetically models human primary biliary cirrhosis. The Journal of Experimental Medicine, 2006. 203(5): p. 1209-1219. 86. Moritoki, Y., et al., B cells promote hepatic inflammation, biliary cyst formation, and salivary gland inflammation in the NOD.c3c4 model of autoimmune cholangitis. Cellular Immunology, 2011. 268(1): p. 16-23. 87. Yang, G.X., et al., CD8 T cells mediate direct biliary ductule damage in nonobese diabetic autoimmune biliary disease. J Immunol, 2011. 186(2): p. 1259-67.

29

88. Gorelik, L. and R.A. Flavell, Abrogation of TGFbeta signaling in T cells leads to spontaneous T cell differentiation and autoimmune disease. Immunity, 2000. 12(2): p. 171-81. 89. Oertelt, S., et al., Anti-Mitochondrial Antibodies and Primary Biliary Cirrhosis in TGF-β Receptor II Dominant-Negative Mice. The Journal of Immunology, 2006. 177(3): p. 1655-1660. 90. Yoshida, K., et al., Deletion of interleukin-12p40 suppresses autoimmune cholangitis in dominant negative transforming growth factor β receptor type II mice. Hepatology, 2009. 50(5): p. 1494-1500. 91. Chuang, Y.H., et al., Natural killer T cells exacerbate liver injury in a transforming growth factor beta receptor II dominant-negative mouse model of primary biliary cirrhosis. Hepatology, 2008. 47(2): p. 571-80. 92. Moritoki, Y., et al., B cells suppress the inflammatory response in a mouse model of primary biliary cirrhosis. Gastroenterology, 2009. 136(3): p. 1037-47. 93. Moritoki, Y., et al., B-cell depletion with anti-CD20 ameliorates autoimmune cholangitis but exacerbates colitis in transforming growth factor-beta receptor II dominant negative mice. Hepatology, 2009. 50(6): p. 1893-903. 94. Tang, Y., et al., Disruption of transforming growth factor-beta signaling in ELF beta- spectrin-deficient mice. Science, 2003. 299(5606): p. 574-7. 95. Mishra, B., et al., Loss of cooperative function of transforming growth factor-β signaling proteins, smad3 with embryonic liver fodrin, a β-spectrin, in primary biliary cirrhosis. Liver International, 2004. 24(6): p. 637-645. 96. Amano, K., et al., Xenobiotic-induced loss of tolerance in rabbits to the mitochondrial autoantigen of primary biliary cirrhosis is reversible. J Immunol, 2004. 172(10): p. 6444- 52. 97. Dhirapong, A., et al., B cell depletion therapy exacerbates murine primary biliary cirrhosis. Hepatology, 2011. 53(2): p. 527-35. 98. Wu, S.J., et al., Innate immunity and primary biliary cirrhosis: activated invariant natural killer T cells exacerbate murine autoimmune cholangitis and fibrosis. Hepatology, 2011. 53(3): p. 915-25. 99. Leung, P.S., et al., Animal models of primary biliary cirrhosis: materials and methods. Methods Mol Biol, 2012. 900: p. 291-316. 100. Moses, H.L., et al., Transforming growth factor production by chemically transformed cells. Cancer Res, 1981. 41(7): p. 2842-8. 101. Roberts, A.B., et al., New class of transforming growth factors potentiated by epidermal growth factor: isolation from non-neoplastic tissues. Proc Natl Acad Sci U S A, 1981. 78(9): p. 5339-43. 102. de Larco, J.E. and G.J. Todaro, Growth factors from murine sarcoma virus-transformed cells. Proceedings of the National Academy of Sciences of the United States of America, 1978. 75(8): p. 4001-4005. 103. Tucker, R., et al., Growth inhibitor from BSC-1 cells closely related to platelet type beta transforming growth factor. Science, 1984. 226(4675): p. 705-707. 104. Roberts, A.B., et al., Type beta transforming growth factor: a bifunctional regulator of cellular growth. Proc Natl Acad Sci U S A, 1985. 82(1): p. 119-23. 105. Kehrl, J.H., et al., Production of transforming growth factor beta by human T lymphocytes and its potential role in the regulation of T cell growth. J Exp Med, 1986. 163(5): p. 1037-50. 106. Kehrl, J.H., et al., Transforming growth factor beta is an important immunomodulatory protein for human B lymphocytes. J Immunol, 1986. 137(12): p. 3855-60.

30

107. Coffman, R.L., D.A. Lebman, and B. Shrader, Transforming growth factor beta specifically enhances IgA production by lipopolysaccharide-stimulated murine B lymphocytes. J Exp Med, 1989. 170(3): p. 1039-44. 108. Wahl, S.M., et al., Transforming growth factor type beta induces monocyte chemotaxis and growth factor production. Proc Natl Acad Sci U S A, 1987. 84(16): p. 5788-92. 109. Tsunawaki, S., et al., Deactivation of macrophages by transforming growth factor-beta. Nature, 1988. 334(6179): p. 260-2. 110. Blobe, G.C., W.P. Schiemann, and H.F. Lodish, Role of Transforming Growth Factor β in Human Disease. New England Journal of Medicine, 2000. 342(18): p. 1350-1358. 111. Derynck, R., et al., A new type of transforming growth factor-beta, TGF-beta 3. EMBO J, 1988. 7(12): p. 3737-43. 112. Massague, J., The transforming growth factor-beta family. Annu Rev Cell Biol, 1990. 6: p. 597-641. 113. Derynck, R., et al., Intron-exon structure of the human transforming growth factor-beta precursor gene. Nucleic Acids Res, 1987. 15(7): p. 3188-9. 114. Jennings, J.C., et al., Comparison of the biological actions of TGF beta-1 and TGF beta- 2: Differential activity in endothelial cells. Journal of Cellular Physiology, 1988. 137(1): p. 167-172. 115. Merwin, J.R., et al., Vascular cells respond differentially to transforming growth factors beta 1 and beta 2 in vitro. The American Journal of Pathology, 1991. 138(1): p. 37-51. 116. Ohta, M., et al., Two forms of transforming growth factor-[beta] distinguished by multipotential haematopoietic progenitor cells. Nature, 1987. 329(6139): p. 539-541. 117. Shull, M.M., et al., Targeted disruption of the mouse transforming growth factor-[beta]1 gene results in multifocal inflammatory disease. Nature, 1992. 359(6397): p. 693-699. 118. Kaartinen, V., et al., Abnormal lung development and cleft palate in mice lacking TGF- beta 3 indicates defects of epithelial-mesenchymal interaction. Nat Genet, 1995. 11(4): p. 415-21. 119. Sanford, L.P., et al., TGFβ2 knockout mice have multiple developmental defects that are non-overlapping with other TGFβ knockout phenotypes. Development (Cambridge, England), 1997. 124(13): p. 2659-2670. 120. Lee, Y., et al., Induction and molecular signature of pathogenic TH17 cells. Nat Immunol, 2012. 13(10): p. 991-999. 121. Dubois, C.M., et al., Processing of transforming growth factor beta 1 precursor by human furin convertase. J Biol Chem, 1995. 270(18): p. 10618-24. 122. Hyytiainen, M., C. Penttinen, and J. Keski-Oja, Latent TGF-beta binding proteins: extracellular matrix association and roles in TGF-beta activation. Crit Rev Clin Lab Sci, 2004. 41(3): p. 233-64. 123. Massague, J., TGF[beta] signalling in context. Nat Rev Mol Cell Biol, 2012. 13(10): p. 616-630. 124. Aoki, C.A., et al., Transforming growth factor beta (TGF-beta) and autoimmunity. Autoimmun Rev, 2005. 4(7): p. 450-9. 125. Wan, Y.Y. and R.A. Flavell, 'Yin-Yang' functions of transforming growth factor-beta and T regulatory cells in immune regulation. Immunol Rev, 2007. 220: p. 199-213. 126. López-Casillas, F., et al., Structure and expression of the membrane proteoglycan betaglycan, a component of the TGF-β receptor system. Cell. 67(4): p. 785-795. 127. Cheifetz, S., et al., Endoglin is a component of the transforming growth factor-beta receptor system in human endothelial cells. Journal of Biological Chemistry, 1992. 267(27): p. 19027-30. 128. López-Casillas, F., J.L. Wrana, and J. Massagué, Betaglycan presents ligand to the TGFβ signaling receptor. Cell. 73(7): p. 1435-1444.

31

129. Gougos, A. and M. Letarte, Primary structure of endoglin, an RGD-containing glycoprotein of human endothelial cells. Journal of Biological Chemistry, 1990. 265(15): p. 8361-8364. 130. Cheifetz, S., et al., The transforming growth factor-β system, a complex pattern of cross- reactive ligands and receptors. Cell. 48(3): p. 409-415. 131. Mu, Y., S.K. Gudey, and M. Landstrom, Non-Smad signaling pathways. Cell Tissue Res, 2012. 347(1): p. 11-20. 132. Attisano, L. and J.L. Wrana, Signal integration in TGF-β, WNT, and Hippo pathways. F1000Prime Reports, 2013. 5: p. 17. 133. Davis, B.N., et al., SMAD proteins control DROSHA-mediated microRNA maturation. Nature, 2008. 454(7200): p. 56-61. 134. Bronevetsky, Y., et al., T cell activation induces proteasomal degradation of Argonaute and rapid remodeling of the microRNA repertoire. The Journal of Experimental Medicine, 2013. 210(2): p. 417-432. 135. Tiegs, G. and A.W. Lohse, Immune tolerance: what is unique about the liver. J Autoimmun, 2010. 34(1): p. 1-6. 136. Karimi, M.H., B. Geramizadeh, and S.A. Malek-Hosseini, Tolerance Induction in Liver. International Journal of Organ Transplantation Medicine, 2015. 6(2): p. 45-54. 137. Banchereau, J., et al., Immunobiology of dendritic cells. Annu Rev Immunol, 2000. 18: p. 767-811. 138. Akbar, S.M., et al., Peripheral blood T-cell responses to pyruvate dehydrogenase complex in primary biliary cirrhosis: role of antigen-presenting dendritic cells. Eur J Clin Invest, 2001. 31(7): p. 639-46. 139. Harada, K., et al., Significance of periductal Langerhans cells and biliary epithelial cell- derived macrophage inflammatory protein-3alpha in the pathogenesis of primary biliary cirrhosis. Liver Int, 2011. 31(2): p. 245-53. 140. Takii, Y., et al., Enhanced expression of type I interferon and toll-like receptor-3 in primary biliary cirrhosis. Lab Invest, 2005. 85(7): p. 908-20. 141. Honda, Y., et al., Altered expression of TLR homolog RP105 on monocytes hypersensitive to LPS in patients with primary biliary cirrhosis. J Hepatol, 2007. 47(3): p. 404-11. 142. Gao, B., S. Radaeva, and O. Park, Liver natural killer and natural killer T cells: immunobiology and emerging roles in liver diseases. J Leukoc Biol, 2009. 86(3): p. 513- 28. 143. Panasiuk, A., D. Prokopowicz, and J. Zak, Peripheral blood T, B lymphocytes and NK cells in primary biliary cirrhosis. Rocz Akad Med Bialymst, 2001. 46: p. 231-9. 144. Chuang, Y.H., et al., Increased killing activity and decreased cytokine production in NK cells in patients with primary biliary cirrhosis. J Autoimmun, 2006. 26(4): p. 232-40. 145. Kita, H., et al., Quantitation and phenotypic analysis of natural killer T cells in primary biliary cirrhosis using a human CD1d tetramer. Gastroenterology, 2002. 123(4): p. 1031- 43. 146. Bertolino, P., G.W. McCaughan, and D.G. Bowen, Role of primary intrahepatic T-cell activation in the /`liver tolerance effect/'. Immunol Cell Biol, 2002. 80(1): p. 84-92. 147. Bertolino, P., et al., Death by neglect as a deletional mechanism of peripheral tolerance. Int Immunol, 1999. 11(8): p. 1225-38. 148. Limmer, A., et al., Efficient presentation of exogenous antigen by liver endothelial cells to CD8+ T cells results in antigen-specific T-cell tolerance. Nat Med, 2000. 6(12): p. 1348-1354. 149. Tay, S.S., et al., Antigen expression level threshold tunes the fate of CD8 T cells during primary hepatic immune responses. Proceedings of the National Academy of Sciences, 2014. 111(25): p. E2540-E2549.

32

150. Tabibian, J.H., et al., Characterization of cultured cholangiocytes isolated from livers of patients with primary sclerosing cholangitis. Laboratory investigation; a journal of technical methods and pathology, 2014. 94(10): p. 1126-1133. 151. Shimoda, S., et al., HLA DRB4 0101-restricted immunodominant T cell autoepitope of pyruvate dehydrogenase complex in primary biliary cirrhosis: evidence of molecular mimicry in human autoimmune diseases. J Exp Med, 1995. 181(5): p. 1835-45. 152. Shimoda, S., et al., Mimicry peptides of human PDC-E2 163-176 peptide, the immunodominant T-cell epitope of primary biliary cirrhosis. Hepatology, 2000. 31(6): p. 1212-6. 153. Shimoda, S., et al., Biliary epithelial cells and primary biliary cirrhosis: the role of liver- infiltrating mononuclear cells. Hepatology, 2008. 47(3): p. 958-65. 154. Shimoda, S., et al., CD4 T-cell autoreactivity to the mitochondrial autoantigen PDC-E2 in AMA-negative primary biliary cirrhosis. J Autoimmun, 2008. 31(2): p. 110-5. 155. Kita, H., et al., Quantitative and functional analysis of PDC-E2-specific autoreactive cytotoxic T lymphocytes in primary biliary cirrhosis. J Clin Invest, 2002. 109(9): p. 1231- 40. 156. Scholz, C., et al., Expansion of autoreactive T cells in multiple sclerosis is independent of exogenous B7 costimulation. J Immunol, 1998. 160(3): p. 1532-8. 157. Lovett-Racke, A.E., et al., Decreased dependence of myelin basic protein-reactive T cells on CD28-mediated costimulation in multiple sclerosis patients. A marker of activated/memory T cells. J Clin Invest, 1998. 101(4): p. 725-30. 158. Viglietta, V., et al., GAD65-reactive T cells are activated in patients with autoimmune type 1a diabetes. J Clin Invest, 2002. 109(7): p. 895-903. 159. Shimoda, S., et al., Autoreactive T-cell responses in primary biliary cirrhosis are proinflammatory whereas those of controls are regulatory. Gastroenterology, 2006. 131(2): p. 606-18. 160. Wakabayashi, K., et al., IL-2 receptor alpha(-/-) mice and the development of primary biliary cirrhosis. Hepatology, 2006. 44(5): p. 1240-9. 161. Aoki, C.A., et al., IL-2 receptor alpha deficiency and features of primary biliary cirrhosis. J Autoimmun, 2006. 27(1): p. 50-3. 162. Zhang, W., et al., Deficiency in regulatory T cells results in development of antimitochondrial antibodies and autoimmune cholangitis. Hepatology, 2009. 49(2): p. 545-52. 163. Marie, J.C., et al., TGF-β1 maintains suppressor function and Foxp3 expression in CD4(+)CD25(+) regulatory T cells. The Journal of Experimental Medicine, 2005. 201(7): p. 1061-1067. 164. Tanaka, A., et al., The clinical significance of IgA antimitochondrial antibodies in sera and saliva in primary biliary cirrhosis. Ann N Y Acad Sci, 2007. 1107: p. 259-70. 165. Invernizzi, P., C. Selmi, and M.E. Gershwin, Update on primary biliary cirrhosis. Dig Liver Dis, 2010. 42(6): p. 401-8. 166. Invernizzi, P., et al., Comparison of the clinical features and clinical course of antimitochondrial antibody-positive and -negative primary biliary cirrhosis. Hepatology, 1997. 25(5): p. 1090-5. 167. Metcalf, J.V., et al., Natural history of early primary biliary cirrhosis. Lancet, 1996. 348(9039): p. 1399-402. 168. Moritoki, Y., et al., B cells and autoimmune liver diseases. Autoimmun Rev, 2006. 5(7): p. 449-57. 169. Takahashi, T., et al., Plasma cells and the chronic nonsuppurative destructive cholangitis of primary biliary cirrhosis. Hepatology, 2012. 55(3): p. 846-55. 170. Selmi, C., P.L. Meroni, and M.E. Gershwin, Primary biliary cirrhosis and Sjogren's syndrome: autoimmune epithelitis. J Autoimmun, 2012. 39(1-2): p. 34-42.

33

171. Nakamura, M., et al., Increased expression of Toll-like receptor 3 in intrahepatic biliary epithelial cells at sites of ductular reaction in diseased livers. Hepatology International, 2008. 2(2): p. 222-230. 172. Trussoni, C.E., et al., Lipopolysaccharide (LPS)-Induced Biliary Epithelial Cell NRas Activation Requires Epidermal Growth Factor Receptor (EGFR). PLoS ONE, 2015. 10(4): p. e0125793. 173. Harada, K., et al., Endotoxin tolerance in human intrahepatic biliary epithelial cells is induced by upregulation of IRAK-M. Liver Int, 2006. 26(8): p. 935-42. 174. Sakisaka, S., et al., Functional differences between hepatocytes and biliary epithelial cells in handling polymeric immunoglobulin A2 in humans, rats, and guinea pigs. Hepatology, 1996. 24(2): p. 398-406. 175. Fukushima, N., et al., Characterization of recombinant monoclonal IgA anti-PDC-E2 autoantibodies derived from patients with PBC. Hepatology, 2002. 36(6): p. 1383-92. 176. Van den Oord, J.J., R. Sciot, and V.J. Desmet, Expression of MHC products by normal and abnormal bile duct epithelium. Journal of Hepatology, 1986. 3(3): p. 310-317. 177. Ayres, R.C., et al., Intercellular adhesion molecule-1 and MHC antigens on human intrahepatic bile duct cells: effect of pro-inflammatory cytokines. Gut, 1993. 34(9): p. 1245-9. 178. Tsuneyama, K., et al., Expression of co-stimulatory factor B7-2 on the intrahepatic bile ducts in primary biliary cirrhosis and primary sclerosing cholangitis: an immunohistochemical study. J Pathol, 1998. 186(2): p. 126-30. 179. Allina, J., et al., T cell targeting and phagocytosis of apoptotic biliary epithelial cells in primary biliary cirrhosis. J Autoimmun, 2006. 27(4): p. 232-41. 180. Ballardini, G., et al., Aberrant expression of HLA-DR antigens on bileduct epithelium in primary biliary cirrhosis: relevance to pathogenesis. Lancet, 1984. 2(8410): p. 1009-13. 181. Odin, J.A., et al., Bcl-2–dependent oxidation of pyruvate dehydrogenase-E2, a primary biliary cirrhosis autoantigen, during apoptosis. The Journal of Clinical Investigation, 2001. 108(2): p. 223-232. 182. Koga, H., et al., Nuclear DNA fragmentation and expression of Bcl‐2 in primary biliary cirrhosis. Hepatology, 1997. 25(5): p. 1077-1084. 183. Harada, K., et al., Enhanced apoptosis relates to bile duct loss in primary biliary cirrhosis. Hepatology, 1997. 26(6): p. 1399-1405. 184. Harada, K., et al., Cell-kinetic study of proliferating bile ductules in various hepatobiliary diseases. Liver, 1998. 18(4): p. 277-284. 185. Harada, K., et al., Increased expression of WAF1 in intrahepatic bile ducts in primary biliary cirrhosis relates to apoptosis. Journal of Hepatology. 34(4): p. 500-506. 186. Takeda, K., et al., Death receptor 5 mediated-apoptosis contributes to cholestatic liver disease. Proceedings of the National Academy of Sciences, 2008. 105(31): p. 10895- 10900. 187. Liaskou, E., G.M. Hirschfield, and M.E. Gershwin. Mechanisms of tissue injury in autoimmune liver diseases. in Seminars in immunopathology. 2014: Springer. 188. Williams, R. and M.E. Gershwin, How, why, and when does primary biliary cirrhosis recur after liver transplantation? Liver Transplantation, 2007. 13(9): p. 1214-1216. 189. Charatcharoenwitthaya, P., et al., Long‐term survival and impact of ursodeoxycholic acid treatment for recurrent primary biliary cirrhosis after liver transplantation. Liver Transplantation, 2007. 13(9): p. 1236-1245.

34

Chapter 2 Murine autoimmune cholangitis requires two hits: cytotoxic KLRG1+ CD8 effector cells and defective T regulatory cells

35

Murine autoimmune cholangitis requires two hits: cytotoxic KLRG1+ CD8 effector cells

and defective T regulatory cells

Wenting Huang*, Kritika Kachapati*, David Adams*, Yuehong Wu*, Patrick S.C. Leung†, Guo-

Xiang Yang†, Weici Zhang†, Aftab A. Ansari‡, Richard A. Flavell#, M. Eric Gershwin†, and

William M Ridgway*

*Division of Immunology, Allergy and Rheumatology, University of Cincinnati College of Medicine,

Cincinnati, OH 45174;

†Division of Rheumatology, Allergy and Clinical Immunology, University of California at Davis, Davis,

CA 95616

# Department of Internal Medicine Yale University School of Medicine, New Haven, CT 06520

‡Department of Pathology, Emory University School of Medicine, Atlanta, GA 30322;

Corresponding author: William M. Ridgway; Telephone: 513-558-4701; Fax: 513-558-3799; Email: [email protected]

Abbreviations: dominant negative transforming growth factor β receptor type II (dnTGFβRII); primary biliary cirrhosis (PBC); bone marrow chimera (BMC); T regulatory cell (Treg); pyruvate dehydrogenase E2 complex (PDC-E2); central memory T cell (CMT); effector T cell and effector memory T cell (ET/EMT); wild type (WT); mean fluorescence intensity (MFI); interferon gamma (IFNγ), intrahepatic cell (IHC)

36

Abstract

Primary biliary cirrhosis (PBC) is an enigmatic disease mediated by autoimmune destruction of cholangiocytes in hepatic bile ducts. The early immunological events leading to PBC are poorly understood; clinical signs of disease occur very late in the pathological process. We have used our unique murine model of PBC in dominant-negative TGF-β receptor type II transgenic mice to delineate critical early immunopathological pathways, and previously showed that dnTGFβRII

CD8 T cells transfer biliary disease. Herein we report significantly increased numbers of hepatic dnTGFβRII terminally differentiated (KLRG1+) CD8 T cells, a CD8 subset previously shown to be enriched in antigen specific cells during hepatic immune response to viral infections. We performed bone marrow chimera studies to assess whether dnTGFβRII CD8 mediated disease was cell intrinsic or extrinsic. Unexpectedly, mixed (dnTGFβRII and B6) bone marrow chimeric

(BMC) mice were protected from biliary disease compared to dnTGFβRII single bone marrow chimerics. To define the protective B6 cell subset, we performed adoptive transfer studies, which showed that co-transfer of B6 Tregs prevented dnTGFβRII CD8 T cell mediated cholangitis. Treg mediated disease protection was associated with significantly decreased numbers of hepatic

KLRG1+ CD8 T cells. In contrast, co-transfer of dnTGFβRII Tregs offered no protection, and dnTGFβRII Treg cells were functionally defective in suppressing effector CD8 T cells in vitro compared to wild type B6 Tregs. In vitro cholangiocyte cytotoxicity assays demonstrated significantly increased numbers of cytotoxic hepatic dnTGFβRII KLRG1+ CD8 cells compared to

B6. Protection from disease by B6 Tregs was associated with elimination of hepatic dnTGFβRII

CD8 mediated cholangiocyte cytotoxicity. These results emphasize that autoimmune cholangitis requires defects in both the T effector and regulatory compartments, and that an intrinsic T cell effector defect is not sufficient to mediate autoimmune biliary disease in the setting of intact immune regulation. These results have important implications for understanding the early pathogenesis of human PBC.

37

Keywords: KLRG1+ CD8 cells, Primary biliary cirrhosis, T Regulatory cells

38

1.0 Introduction

Primary biliary cirrhosis is an organ specific autoimmune disease in which biliary ductules are the target of autoimmune mediated destruction, resulting ultimately in cirrhosis and liver failure [1, 2]. PBC is associated with a high prevalence of autoantibodies to mitochondrial antigens, most specifically pyruvate dehydrogenase PDC-E2 [3, 4]. In humans, PBC pathogenesis is shrouded in mystery, since most patients develop clinical symptoms long after the initiation of the disease process. This “gap” between onset and clinical detection has frustrated efforts to understand the early events leading to disease. The dnTGFβRII mouse was originally developed by one of us (R.A.F.) using a dominant negative type II TGFβR transgene expressed on the CD4 promoter [5] leading to transgene expression in both CD4 and CD8 cells. These cells are characterized by a significant increase in the effector and effector memory activated T cell subset

(CD44high CD62Llow), showing that loss of TGFβ receptor signaling induces abnormal activation and disruptions in the balance of normal T cell subsets and their absolute numbers [6]. Notably, not all TGFβ receptor signaling is lost, which allows mice to survive for almost normal lifespans

(compared to complete TGFβRII knock out in both CD4 and CD8 cells, wherein mice only survive a few weeks) [7].

The dnTGFβRII mice exhibit autoimmune cholangitis; ~100% of mice develop the PDC-E2 autoantibodies seen in human PBC, and show histological destruction and damage of biliary ductules [6]. In addition, dnTGFβRII mice develop autoantibodies to gp210 and sp100, which are also considered highly specific to PBC [8]. These features establish the dnTGFβRII mice as a model of PBC. An important finding was our demonstration that adoptive transfer of minimacs purified dnTGFβRII CD4 cells led to inflammatory bowel disease but not liver pathology, while transfer of CD8 cells led to biliary disease but not inflammatory bowel disease [9]. This focused our attention on the pathological role of CD8 cells. It is known that the liver is an immune organ

39 with high numbers of resident CD8 cells, in particular it has been suggested that the liver acts as a

“graveyard” for senescent CD8 cells [10, 11]. This hepatic lymphoid function suggested that non- specific effects of derangement in CD8 T cell processing in the liver could be one mechanism of biliary immune mediated disease. An alternative explanation, favored by the development of

PDC-E2 specific autoantibodies in these mice, is that antigen specific T cells are actively involved in the biliary immune process.

To distinguish between these possibilities, we recently constructed dnTGFβRII mice with T cell repertoires confined to a specific (foreign) antigen (Ova) [12]. Unexpectedly, when the CD8 repertoire was confined to a non-hepatic antigen, biliary disease was abolished. In addition, these

T cells could not transfer biliary disease, even though they still showed the characteristic T cell activation abnormalities (greatly increased CD44high and CD62Llow )[12]. Thus an increased population of abnormally activated CD8 cells alone is insufficient for biliary disease: an antigen specific population of T cell is likely involved in the pathogenesis.

These results left unresolved whether the biliary antigen specific, activated CD8 T cells mediated disease in an intrinsic manner (i.e. the T cells were abnormally affected by the decreased TGFβR signaling such that no external immune influences could decrease their pathogenicity) or an extrinsic manner (i.e. in addition to defects in T effector function, immune regulatory defects contributed to the disease). This distinction is important because if the disease is solely a result of an intrinsic defect of CD8 T cells, therapeutic strategies directed to enhancing immune regulation might not work. This paper studies the role of intrinsic or extrinsic factors in the pathogenesis of primary biliary cirrhosis by using BMC and T cell transfer studies with analysis of the phenotype and function of CD8 T cell and T regulatory cell subsets.

2.0 Materials and methods

40

2.1 Mice

B6 (CD45.2) mice, B6.Rag1-/- mice, B6.Cg-Foxp3tm2Tch/J (hereafter, “B6.Foxp3EGFP” ) mice and

B6.SJL-Ptprca mice (hereafter referred to as B6.CD45.1) were purchased from The Jackson

Laboratory. dnTGFβRII mice [6] were maintained as described previously [13]. Mice were maintained under specific pathogen-free conditions and handled in accordance with the institutional animal care guidelines of the University of Cincinnati School of Medicine.

2.2 Bone marrow chimera construction

Groups of (B6 CD45.1×CD45.2) F1 recipient mice were irradiated with 1100-1200 rad.

B6.CD45.1 and dnTGFβRII (CD45.2) mice were bone marrow donors. Mature CD4+, CD8+ and

CD90+ cells were depleted from the bone marrow cells by miniMACS (Miltenyi biotec). Mixed bone marrow chimera (mBMC) were derived by injection of a 1:1 mixture of dnTGFβRII

(CD45.2) and B6. CD45.1 donor bone marrow. Single BMC chimeras received marrow cells from either dnTGFβRII (CD45.2) or B6 (CD45.2) alone. Recipient mice were given water treated with antibiotic (neomycin trisulfate salt hydrate) for 2 weeks after transfer. Recipients were harvested 120 days after bone marrow transplantation (or at the time they became ill).

2.3 Histopathology

Livers were isolated and fixed in 10% formalin, then paraffin-embedded. Samples were stained with hematoxylin and eosin, and scored blindly using microscopy. Scores were based on the severity of portal inflammation. Score 0: 0~5% of portal ducts infiltrated; score 1: 5~25%; score

3: 50~75%; and score 4: 75~100% of the liver section shows the portal duct area infiltrated by leukocytes.

2.4 CD8 and Treg co-transfer study

For transfer studies, B6.Foxp3EGFP, B6 and dnTGFβRII mice served as donors, and B6.Rag1-/- mice served as recipients. 1x106 miniMACS enriched B6 or dnTGFβRII splenic CD8+ cells were transferred to recipients, and in some experiments 0.5x106 FACS-sorted splenic CD4+GFP+ cells

41

(from B6.Foxp3EGFP mice) or 0.5x106 FACS-sorted dnTGFβRII splenic CD4+CD25+ cells were transferred into Treg co-recipients.

2.5 Flow cytometry

Flow cytometric analysis of intrahepatic cells (IHC) was performed on cells obtained by perfusion of liver with 5 mL of EGTA injected through the portal vein followed by 5 mL of

Collagenase IV (Sigma-Aldrich) for 15 min. For absolute cell counts, splenocytes and IHC were counted using a hemocytometer. For surface molecule staining of conventional T cells, cells were incubated with 2.4G2 Fc block for 10 min at 4°C followed by the indicated antibodies (from BD

Biosciences, BioLegend or eBioscience). FACS was performed on LSRII or LSR-Fortessa (BD) and analyzed using FlowJo (Tree Star, version 7.6.5).

2.6 Treg suppression assay

A total of 100,000 miniMACS enriched splenic CD8+ cells or FACS-sorted splenic CD4+CD25- cells from either B6 or dnTGFβRII mice were cultured with 20,000 anti-CD3/CD28-coated beads

(Invitrogen) in the presence of 50,000 FACS-sorted splenic CD4+CD25+ cell from B6 or dnTGFβRII in a criss-cross manner, along with positive and negative controls. Triplicate wells were used for each condition. The cells were cultured at 37°C in 5% CO2 and pulsed with 1 uCi

[3H] thymidine on day 3 for 16 hours, then harvested and counted using a β-scintillation counter.

2.7 CD8 T cell cytotoxicity assay

CD8 T cell cytotoxicity against cholangiocytes was performed using CytoTox 96®

NonRadioactive Cytotoxicity Assay Protocol (Promega) measuring lactate dehydrogenase (LDH) release of the target cells in the culture as described [14]. In brief, 100,000 miniMACS sorted

CD8+ intrahepatic cells were co-cultured with a total of 10,000 murine cholangiocyte cell line [15]

+ or naive CD8 T cells were FACS- sorted and cultured with cholangiocytes at 10:1 ratio. After 5 hr incubation at 37°C and 5% CO2, culture supernatant was transfer to a 96 plate bottom well plate and the released LDH in the supernatant was measured with a 30-minute coupled enzymatic assay; absorbance was recorded

42 at 490nm using a VICTOR X4 2030 Multilabel Reader (PerkinElmer). LDH release was measured in experimental, media alone, “volume correction” control, target cell maximum, target cell spontaneous and effector cell spontaneous wells, as described by the manufacturer. The percent cytotoxicity was calculated as: % Cytotoxicity = [(Experimental - Effector Spontaneous -

Target Spontaneous)/(Target Maximum - Target Spontaneous)] × 100 as described by the manufacturer. NB: the above calculation makes a negative percent cytotoxicity value possible when the experimental LDH release is less than the control LDH release.

2.8 Statistics

All statistical analysis was performed using either the unpaired t test or the Mann–Whitney U test in GraphPad Prism 5 (version 5.01; GraphPad).

43

3.0 Results

3.1 Mice reconstituted with both dnTGFβRII and B6 bone marrow show significant protection

We addressed the issue of intrinsic versus extrinsic CD8 mediated biliary disease by constructing mixed bone marrow (mBMC) mice from allotypically variant strains. B6 CD45.1 bone marrow was mixed in equal proportions with dnTGFβRII CD45.2 bone marrow; single BMC controls included B6 (CD45.2) donors and dnTGFβRII (CD45.2) donors (Fig 1a). While recipients of dnTGFβRII donor cells developed significant disease, as expected (comparable to non-chimeric dnTGFβRII mice), mBMC recipients were protected from biliary disease, with histological scores no different from B6 recipients (Fig 1b, c). Protection from disease in the mBMC recipients was not due to differences in reconstitution efficiency since the absolute number of splenocytes was the same between single and mixed chimeric mice (Fig 1d). The B6 origin cells also did not outgrow cells of the dnTGFβRII origin, since the frequencies and absolute numbers of spleen and liver lymphoid cells expressing CD45.1 vs. CD45.2 allotypes in the F1 recipients were comparable (Fig 1e).

3.2 The dnTGFβRII and mBMC intrahepatic cell populations show expansion of dnTGFβRII origin effector memory and terminally differentiated (KLRG1+ CD127-) CD8 cells

To understand disease protection in the mBMC, we characterized the hepatic CD8 T cell subsets in dnTGFβRII mice and in the BMCs. We have previously shown that dnTGFβRII CD8 cells showed an increase in hepatic CD44highCD62Llow (ET/EMT) T cells [9]. We extended this analysis to assess whether CD44highCD62Llow CD8 cells were further differentiated into terminally differentiated cells (KLRG1+CD127-, associated with antigen specific CD8 cells in the setting of persistent antigen) or effector memory (KLRG1- CD127+, associated in hepatic viral models with resolution of infection) [16, 17]. The dnTGFβRII-origin intrahepatic cells (IHCs)

44 have a massive increase in terminally differentiated KLRG1+CD127- CD44highCD62Llow CD8 cells compared to hepatic B6 CD8+ T cells (Fig 2a). We used the same approach to analyze CD8 hepatic T cells in the bone marrow chimeric mice (representative FACS gating shown in Fig 2b).

As expected, the recipients of only B6 or dnTGFβRII bone marrow showed normal and increased numbers of hepatic CD8 cells, respectively (Fig 3a). In the mixed BMC recipients, however, the total number of dnTGFβRII origin CD8 T cells was significantly greater than the B6 origin CD8

T cells; these findings are comparable to the numbers found in the dnTGFβRII single BMC (Fig

3a). There were increased numbers of naïve B6-origin CD8 T cells compared to dnTGFβRII- origin in the mBMC (Fig 3b). In contrast, in the CD8 effector/memory, central/memory, and terminally differentiated subsets the number (Fig 3c-f) and percent (not shown) of dnTGFβRII- origin hepatic cells was also significantly greater than B6-origin cells and comparable to single dnTGFβRII BMC and non-chimeric dnTGFβRII mice. Although the number of dnTGFβRII- origin cells in effector/memory and terminally differentiated CD8 subsets was significantly increased compared to the number of B6-origin cells in the mBMC mice (Fig 3c-f), these mice did not have significant biliary disease (Fig 1a, b). This raised the possibility that B6 regulatory cells controlled the dnTGFβRII effector CD8 cells in the mBMC.

3.3 DnTGFβRII Tregs are defective in suppressing CD8 T effector cells in vitro

We began to address the role of Tregs in cholangitis by quantitating dnTGFβRII vs. B6 Treg numbers and function. DnTGFβRII mice actually had significantly higher numbers of hepatic

CD4+Foxp3+ Tregs but no difference in the number of splenic Tregs (Fig 4a). Foxp3 MFI in both hepatic and splenic Tregs of dnTGFβRII mice was not significantly different from B6 mice (Fig

4b). There was no difference in the percent Foxp3+ cells in the CD4+CD25+ subset in these strains,

(~95% of CD4+CD25+ cells were Foxp3+, Fig 4c), so we sorted CD4+CD25+ Tregs from both dnTGFβRII and B6 mice for an in vitro suppression assay against both B6 and dnTGFβRII splenic CD4+ (CD25-) and CD8 T effector cells (Fig 4d, left and right panels respectively). The

45 dnTGFβRII CD4+CD25- and CD8 T cells had significantly increased proliferation compared to

B6 at the same level of CD3/CD28 stimulation (Fig 4, legend). When tested against B6 CD4 effectors, B6 Tregs were significantly more suppressive than dnTGFβRII Tregs (P=0.01) although there was no difference between the Tregs in suppression of dnTGFβRII CD4 effectors

(Fig 4d, left). However, the B6 Tregs were significantly more suppressive than dnTGFβRII

Tregs against both B6 CD8 (P=0.04) and dnTGFβRII CD8 T effectors (P=0.01, Fig 4d, right).

The reduced suppression by dnTGFβRII CD4+CD25+ Tregs cannot be attributed to a reduced number of Foxp3+ Tregs in the CD4+CD25+ subset (Fig 4a). Thus the dnTGFβRII Tregs showed a defect in the ability to suppress CD8 (and to a lesser degree, CD4) T effector cells compared to the B6 Tregs. We next directly tested the efficacy of dnTGFβRII vs. B6 Treg function in vivo by adoptive transfer studies.

3.4 B6 , but not dnTGFβRII, T regulatory cells can prevent adoptive transfer of cholangitis by dnTGFβRII CD8 cells

We co-transferred dnTGFβRII CD8 cells along with B6 or dnTGFβRII Tregs into B6 Rag1-/- recipients (Fig 5a). As shown in Fig 5b and 5c, B6 Tregs significantly suppressed the transferred disease, which was indistinguishable from transfer of B6 CD8 T cells alone. In striking contrast, co-transfer of dnTGFβRII CD4+CD25+ Tregs had no effect on the disease mediated by dnTGFβRII CD8 cells (Fig 5b, c). The B6 Treg recipients had no decrease in percentage of lymphoid cells in the liver or spleen compared to the dnTGFβRII recipients (data not shown).

However, total numbers of both lymphoid cells and hepatic CD8 cells were significantly greater in the dnTGFβRII CD8 and dnTGFβRII CD8 plus dnTGFβRII Treg recipients compared to the dnTGFβRII CD8 plus B6 Treg recipients, suggesting that the B6 Tregs decreased the proliferation or accumulation of dnTGFβRII CD8 cells (Fig 5d, e). The percentage of CD8 T cells did not differ between the dnTGFβRII and B6 Treg recipients (data not shown).

46

3.5 Transfer recipients of dnTGFβRII CD8 plus B6 Tregs have decreased numbers of KLRG1+

CD8 cells compared to dnTGFβRII CD8 recipients

We analyzed the differentiation status of the dnTGFβRII CD8 cells in the transfer recipients; a representative FACS analysis figure is shown (Fig 6a). Mice receiving dnTGFβRII CD8 plus B6

Tregs had decreased numbers of ET/EMT CD8 cells (Fig 6c), KLRG1+ CD127- CD8 cells (Fig

6d) and KLRG1-CD127+ CD8 T cells (Fig 6e), and conversely increased numbers of hepatic CD8

CMT cells (Fig 6b), compared to dnTGFβRII CD8 or dnTGFβRII CD8 plus dnTGFβRII Treg recipients. These results strongly suggest that the protective effect of the B6 Tregs was associated with a decrease in the overall expansion of the dnTGFβRII CD8 T cells (resulting in significantly decreased absolute numbers of CD8 T cells, Fig 5d), an increase in CMT CD8 cell numbers, and decreased absolute numbers of CD8 differentiated effector subsets including

KLRG1+ CD8 cells.

3.6 Mechanism of B6 Treg protective effect: dnTGFβRII KLRG1+ CD8 T cells mediate cholangiocyte cytotoxicity; B6 Tregs eliminate dnTGFβRII CD8 cytotoxicity

Finally, we sought a mechanism for the B6 Treg protective effect. We tested dnTGFβRII vs. B6 intrahepatic CD8 cell mediated cytotoxicity against a cholangiocyte cell line [15]. The dnTGFβRII intrahepatic CD8 cells showed a titratable increase in cholangiocyte cytotoxicity compared to B6 CD8 (Fig 7a). Cholangiocyte cytotoxicity was largely mediated by the hepatic

KLRG1+ CD8 subset (Fig 7b; dnTGFβRII CD8 cell cytotoxicity: 43.9% ± 21.89%; dnTGFβRII

KLRG1+ cytotoxicity: 47.0% ± 10.0%). When we isolated hepatic dnTGFβRII CD8 cells from recipients of dnTGFβRII CD8 cells plus B6 Tregs (Fig 5), however, CD8 cytotoxicity was completely eliminated (Fig 7c), and this correlated with the lack of disease in these recipients.

These findings suggest that protective, wild type Tregs can correct any “intrinsic” dnTGFβRII

CD8 cellular defects and act by eliminating the capacity of the KLRG1+ CD8 cells to destroy cholangiocytes.

47

4.0 Discussion

These results are significant in several respects. We demonstrate here that the intrinsic molecular signaling defect of dnTGFβRII CD8 cells is insufficient to mediate disease in the presence of normal B6 cells. More specifically, B6 Tregs abrogated the pathogenicity of dnTGFβRII CD8 cells. Therefore while dnTGFβRII CD8 cells are necessary for the biliary disease, in the presence of normal (B6) Tregs no disease occurs, which indicates that dnTGFβRII Tregs are not sufficient for protection. DnTGFβRII and B6 mice have equal numbers of splenic Tregs, but that the dnTGFβRII Tregs show functional defects. Intriguingly, the dnTGFβRII Tregs appear particularly defective in regulating CD8 cells. This might explain why CD8 T cells can mediate biliary disease in dnTGFβRII mice. B6 Tregs act, at least in part, by eliminating the cytotoxic capacity of KLRG1+ CD8 cells, thus preventing autoimmune cholangiocyte damage. Therefore defects in both the effector and regulatory T cell subsets are necessary for the disease to progress.

How do our findings of abnormal hepatic CD8 effector and KLRG1+ CD8 cells, and defective

Tregs in the dnTGFβRII PBC model, relate to what is known about human PBC? It has been shown that PDC-E2 reactive, cytotoxic CD8 cells are increased in peripheral blood of PBC patients [18]. Antigen specific CD8 cytotoxic cells are also increased ten-fold in PBC portal tracts

[19]. Moreover, Tsuda et al. subsequently showed increased numbers of CD8+ effector memory cells in PBC peripheral blood; these cells expressed increased amounts of perforin and granyzme, suggesting that they were a cytotoxic subset [20]. Our results extend these findings and suggest that human PBC hepatic CD8 populations should be studied to verify the presence of KLRG1+ cytotoxic CTLs. In terms of Tregs, abnormalities have also been found in human PBC. Lan et al. found abnormalities of PBC Tregs in both peripheral blood and liver of PBC patients [21]. While they did not detect a functional defect, the ratio of CD8:Foxp3+ cells was significantly higher in

PBC portal tracts [21]. Bernuzzi et al. recently extended these findings by showing functional

48 defects in novel CD8 regulatory cells[22]. Thus our findings of defective CD8 and Treg lineages have support in the human PBC literature and suggest pathways for further investigation of the human immunology.

B6 Tregs did not entirely eliminate the effects of the intrinsic signaling defects of dnTGFβRII

CD8 cells. For example, B6 Tregs did not completely eliminate the activated KLRG1+CD127-

EM/EMT CD8 T cells in the liver. Rather, they decreased the number of KLRG1+ cells, which predominantly mediate cholangiocyte cytotoxicity. CD4+CD25+Foxp3+ Tregs can limit CD8 T cell expansion and differentiation through modulating IL-2 homeostasis [23]. In a study investigating how human Tregs control different CD8 T cell subsets, Nikolova et al. found that

Tregs can inhibit CD8 memory cell differentiation to cytokine-producing effector cells and can reduce apoptosis of CD8 memory cells compared to effector cells, partially through decreased

PD-1 expression on CD8 T cells after activation [24]. A study of repopulation of CD8 T subsets under lymphopenic conditions showed CD4+CD25+ Tregs have minimal suppressive effect on

CMT, mainly acting on naïve and ET/EMT compartments [25]. Moreover, we demonstrate that

B6 Tregs were capable of completely eliminating dnTGFβRII KLRG1+ CD8 mediated cholangiocyte toxicity. This raises the question of how the B6 Tregs mediate suppression compared to the dnTGFβRII Tregs. Our assay measured suppression mediated by cell:cell contact and this suggests the dnTGFβRII Tregs could have a defect in cell contact mediated suppression.

Lan et al. showed that Tregs in human PBC have increased expression of GITR [21]. It has been shown that a combination of increased GITR-R on T cells with increased GITR on Tregs can cause increased resistance of CD8 cells to suppression[26]. The GITR-GITR ligand axis could thus potentially integrate the findings of defects in both the CD8 and Treg lineages, and represents a logical future pathway to explore to explain how the defects in cell mediated suppression seen in dnTGFβRII mice. In summary B6 Tregs in our model may act by preventing the accumulation or differentiation of effector and terminally differentiated CD8 T cells, thus

49 preventing cholangiocyte damage mediated by these CD8 subsets. The selective defect of dnTGFβRII Treg suppression with respect to CD8 T cells is of interest, and requires further exploration in future studies.

Finally, we show that dnTGFβRII mice, and B6.Rag1-/- or BMC recipients receiving dnTGFβRII cells, spontaneously accumulate enhanced numbers of hepatic KLRG1+ CD8 T cells. We show that dnTGFβRII KLRG1+ CD8 cells are cytotoxic to cholangiocytes. B6 Tregs decrease the number of KLRG1+ cells in transfer studies, and also prevent disease and cholangiocyte cytotoxicity. It was previously shown in the dnTGFβRII model that after stimulation with cognate antigen or via an infectious agent, splenic antigen specific KLRG1+ cells increase— however it was not previously shown that KLRG1+ cells could accumulate spontaneously, nor have hepatic KLRG1+ populations been previously studied [27]. KLRG1+ CD8 cells have been studied in models of viral immunity. Multiple studies have shown that the KLRG1+ CD8 subset is enriched in antigen specific cells directed towards the infectious agent [16, 17, 28]. Bengsh et al. showed that KLRG1+ cells are increased in the presence of an ongoing immune response to persistent hepatitis B virus [16]. Cush et al. showed that ~75% of cells specific for murine γ- herpesviru are KLRG1+ cells that express IL-7 and IL-15 receptors [17]. IL-15 (in addition to IL-

2) is critical for the survival of KLRG1+ CD8 cells, and thus modulation of IL-2, IL-15 or their receptors could affect the accumulation of these cells [29, 30]. The accumulation of KLRG1+ cells is also driven by IL-12; mice lacking IL-12 signaling fail to develop KLRG1+ CD8 cells

[31]. IL-12 acts to create a gradient of T-bet expression; high T-bet drives KLRG1+ survival [32].

So, decreased KLRG1+ cell numbers could reflect the actions of the Tregs to reduce the IL-12 driven autoimmune process which is critical to the dnTGFβRII model [33, 34]. Excess IFNγ can also drive the expression of frankly autoreactive KLRG1+ CD8 T cells, as shown by the increased numbers of these cells in mice expressing the sanroquin mutation of Roquin [35]. Finally, recent studies in tumor immunology have shown that KLRG1+ CD8 cells are cytotoxic to tumor tissue

50

[36-38], supporting the hypothesis that KLRG1+ CD8 cells may mediate hepatic autoimmunity by causing cholangiocyte cell death. This literature suggests that the increased numbers of KLRG1+

CD8 cells in single dnTGFβRII bone marrow chimeric mice, and the dnTGFβRII –origin CD8 cells in the mixed bone marrow chimeric mice, should be accompanied by cytokine abnormalities such as increased IL-12 and IFNγ. Furthermore, cholangiocyte apoptosis, and resultant biliary apotopes, play a major role in PBC pathogenesis [39, 40]. Therefore our finding of increased cytotoxic KLRG1+ cells suggests we should find enhanced cholangiocyte apoptosis selectively in these mice. Cytokine and apoptosis abnormalities will be a future focus of investigation in these mice. In summary, spontaneous accumulation of large numbers of cytotoxic KLRG1+ CD8 cells in the dnTGFβRII model implies that they accumulate after responding to their biliary autoantigen, consistent with our recent, surprising observation that the dnTGFβRII CD8 cells must have an autoantigenic specificity to mediate disease [12]. The studies shown here and published recently strongly suggest that the dnTGFβRII defect encourages the expansion of antigen specific autoreactive T cells. This might be due to effects on either central or peripheral T cell tolerance. Since we do not yet know the identity of the relevant biliary autoantigen(s), we cannot yet directly prove this point. However, future studies will focus on the TCR repertoire in the KLRG1+ CD8 subset and attempt to show if this subset selectively transfer autoimmune biliary disease.

51

5.0 Conclusion

The dnTGFβRII model does not behave like a classic knockout model with complete loss of function of the selected gene. The effects of complete loss of dnTGFβRII signaling in CD4 and

CD8 cells are drastically different than in our dnTGFβRII model, mediating death in 3-4 weeks

[7]. The dnTGFβRII mouse develops a chronic disease, requiring multiple arms of the immune system, consistent with the idea that a complex disease such as PBC must involve multiple defects in the immune system. Therefore the dnTGFβRII model may actually closely mimic the pathogenic events in human PBC that impact both regulatory and effector arms of the immune system with the final effect of biliary autoimmunity. This has important implications for human

PBC. First, it suggests that therapeutic approaches might be able to either decrease effector immunity (immunosuppression) or enhance regulatory immunity. Second, it raises the possibility that PBC might be a “two hit” disease—requiring defects in both effector and regulatory immunity [41]. Requiring at least “two hits” might explain why the disease is somewhat rare. In addition, it might explain the slow progression of disease over a long silent period: initial defects in one arm of the immune system may be insufficient to drive aggressive clinical disease. These findings raise new hypotheses to be tested in understanding and treating human PBC.

52

Figure legends

Figure 1

Reduced autoimmune biliary disease in mixed BMC mice compared to single dnTGFβRII

BMC mice. (a) Schematic showing bone marrow chimera construction. (b,c) Representative histology sections (b, yellow arrows indicate bile ducts surrounded by infiltrated leukocytes) and mean blinded histology scores (c) from mBMC (n=9), single B6 BMC (n=4) and single dnTGFβRII BMC (n=5). (d) Total lymphocyte number in spleen and liver among recipients of mBMC (n=10), B6 BMC (n=7) and dnTGFβRII BMC (n=5) (not all mice that were studied by

FACS had liver histology). The number of days after bone marrow transplantation ranged from

41d to 158d (mice receiving dnTGFβRII often became sick and had to be sacrificed earlier). (e)

Comparison of WT-origin lymphocyte number and dnTGFβRII-origin lymphocyte number in the spleen and liver of mBMC (n=10). Error bar indicates S.D. Statistical analysis by Mann Whitney

U test (c) and unpaired T test (d,e).

Figure 2

Flow cytometric analysis of hepatic T cells: increased KLRG1+ CD8 cells in dnTGFβRII liver. (a) Representative FACS plots comparing hepatic CD8 T cell subsets between B6 and dnTGFβRII. Female B6 and dnTGFβRII aged over 120d were used for comparison. Circles indicate the KLRG1+ CD8 subset. (b) Representative FACS plots of mBMC and single (B6 or dnTGFβRII) BMC controls showing the 8-color gating strategy to compare hepatic CD8 T cell subsets among different groups.

Figure 3 mBMC recipients have increased numbers of dnTGFβRII-origin CD8 effector cells but do not develop biliary disease. Cell numbers of total hepatic CD8 T cells (a) and hepatic CD8 T

53 cell subsets (b-f) differentiating WT-origin and dnTGFβRII-origin cell populations in mBMC

(n=10), B6 BMC (n=7) and dnTGFβRII BMC (n=5) recipients. Error bar indicates S.D.

Statistical analysis by unpaired T test.

Figure 4

DnTGFβRII Tregs are defective in suppressing B6 and dnTGFβRII CD8 T cells.

(a, b) Absolute number (a) of CD4+Foxp3+ Tregs, and (b) MFI of Foxp3 in the liver and spleen from B6 (n=3) and dnTGFβRII (n=3) mice. (c) Comparison of percentage of Foxp3+ cells within splenic CD4+CD25+ T cells between B6 and dnTGFβRII. One representative of three experiments.

(d) Proliferation of B6 (shaded white) or dnTGFβRII (shaded dark) CD4+CD25- (left, n=7 experiments) or CD8+ (right, n=5 experiments) cells stimulated with CD3/CD28 beads in the presence of B6 or dnTGFβRII Tregs. Suppression is shown as a percent proliferation of effector cells alone. The mean CPMs (± SEM) of effector cells alone stimulated with CD3/CD28 beads:

B6 CD4: 12,355 ± 998; B6 CD8: 12,155 ± 694; dnTGFβRII CD4: 110,736 ± 13,406, and dnTGFβRII CD8: 41,416 ± 2657. Statistical analysis by unpaired T test.

Figure 5

Transfer studies confirm that B6, but not dnTGFβRII Tregs can prevent adoptive transfer of autoimmune biliary disease mediated by dnTGFβRII CD8 cells. (a) Schematic showing transfer study design of B6 CD8, dnTGFβRII CD8, dnTGFβRII CD8 plus B6 Tregs or dnTGFβRII CD8 plus dnTGFβRII Tregs, as described in Methods, into B6.Rag1-/- recipients. (b)

Representative liver histology slides at low (left) and high (right) magnification for each transfer group. (c) Mean liver histology scores among recipients of dnTGFβRII CD8 plus B6 Treg (n=7),

B6 CD8 alone (n=8) dnTGFβRII CD8 alone (n=10), or dnTGFβRII CD8 plus dnTGFβRII Tregs

(n=3). (d,e) The absolute number of total liver lymphocytes (d) and liver CD8 T cells (e) are shown for recipients transferred with dnTGFβRII CD8 and B6 Tregs (n=3), B6 CD8 alone (n=6)

54 and dnTGFβRII CD8 alone (n=4) or dnTGFβRII CD8 plus dnTGFβRII Tregs (n=3) (not all individual mice that had liver histology slides were also studied by FACS). The experimental setup was the same as in (b).Error bar indicates S.D. Statistical analysis by Mann Whitney U test

(c) and unpaired T test (d,e).

Figure 6

B6 Tregs reduce the total number of CD8 cells and KLRG1+ CD8 cells while preventing biliary disease. (a) Representative FACS plots showing intrahepatic lymphocyte CD8 T cell subsets from recipients transferred with dnTGFβRII CD8 T cells and B6 Tregs (n=3), B6 CD8 T cells alone (n=6) dnTGFβRII CD8 T cells alone (n=4) or dnTGFβRII CD8 plus dnTGFβRII

Tregs (n=3), as described in Methods. The absolute number of CMT (b), ET/EMT (c), terminal effector cell (d) and effector memory cells (e) are shown. Error bars indicate S.D. Statistical analysis by unpaired T test.

Figure 7

Intrahepatic dnTGFβRII KLRG1+ CD8 cells are cytotoxic to cholangiocytes; B6 Tregs eliminate dnTGFβRII mediated cholangiocyte toxicity.

(a) MiniMACS-sorted liver CD8 T cells from dnTGFβRII or B6 were co-cultured with immortalized murine cholangiocyte cells (mCL) at titrated ratios ; n=3/group. Killing assay was performed as described in methods. Error bars are shown only above the curve. *: P < 0.05; *** p

<0.001 by unpaired T test. (b) CD8+, CD44high CD62Llow CD127-KLRG1+ cells were sorted from dnTGFβRII liver and cultured with murine cholangiocyte cells at a ratio of 100,000(CD8) to

10,000 (mCL) (left column) or 10,000 (CD8) to 1,000 (mCL) (middle column) while 10,000 naïve (CD8+ CD44low CD62Lhigh dnTGFβRII cells were cultured with 1,000 mCL (right column).

Killing assay was then performed as above; experiment performed n=3 for each group. (c)

MiniMACS-sorted liver CD8 T cells (100,000 cells/well) from 1) dnTGFβRII, 2) B6 or 3)

55

B6.Rag1-/- mice reconstituted with dnTGFβRII CD8 T cells and B6 Tregs (as in Fig 5) were co- cultured with mCL at 10:1 ratio. n=3~4/group. Error bar indicates S.D. Statistical analysis by unpaired T test.

56

57 58 59 60 61 62 63 Acknowledgements: Supported by NIH grant R01 DK090019, VA grant BX000827-01A1

Author roles: WMR, MEG: study concept and design; obtained funding, analysis and interpretation of data; drafting of the manuscript; critical revision of the manuscript for important intellectual content; study supervision AA: critical revision of the manuscript for important intellectual content RF: material and intellectual support WH, KK, DEA, YW, PSCL, GY, WZ: acquisition of data; analysis and interpretation of data; drafting of the manuscript; critical revision of the manuscript for important intellectual content; statistical analysis.

64

References

1. Gershwin, M.E., et al., Primary biliary cirrhosis: an orchestrated immune response against epithelial cells. Immunological Reviews, 2000. 174(1): p. 210-225. 2. Kaplan, M.M. and M.E. Gershwin, Primary biliary cirrhosis. N Engl J Med, 2005. 353(12): p. 1261-73. 3. Gershwin, M., et al., Identification and specificity of a cDNA encoding the 70 kd mitochondrial antigen recognized in primary biliary cirrhosis. The Journal of Immunology, 1987. 138(10): p. 3525-3531. 4. Coppel, R.L. and M.E. Gershwin, Primary biliary cirrhosis: the molecule and the mimic. Immunological Reviews, 1995. 144: p. 17-49. 5. Gorelik, L. and R.A. Flavell, Abrogation of TGFbeta signaling in T cells leads to spontaneous T cell differentiation and autoimmune disease. Immunity, 2000. 12(2): p. 171-81. 6. Oertelt, S., et al., Anti-Mitochondrial Antibodies and Primary Biliary Cirrhosis in TGF-β Receptor II Dominant-Negative Mice. The Journal of Immunology, 2006. 177(3): p. 1655-1660. 7. Li, M.O., S. Sanjabi, and R.A. Flavell, Transforming growth factor-beta controls development, homeostasis, and tolerance of T cells by regulatory T cell-dependent and - independent mechanisms. Immunity, 2006. 25(3): p. 455-71. 8. Yang, C.Y., et al., Epitope-specific anti-nuclear antibodies are expressed in a mouse model of primary biliary cirrhosis and are cytokine-dependent. Clin Exp Immunol, 2012. 168(3): p. 261-7. 9. Yang, G.X., et al., Adoptive transfer of CD8(+) T cells from transforming growth factor beta receptor type II (dominant negative form) induces autoimmune cholangitis in mice. Hepatology, 2008. 47(6): p. 1974-82. 10. Crispe, I.N., et al., Cellular and molecular mechanisms of liver tolerance. Immunological Reviews, 2006. 213(1): p. 101-118. 11. Crispe, I.N., The Liver as a Lymphoid Organ. Annual Review of Immunology, 2009. 27(1): p. 147-163. 12. Kawata, K., et al., Clonality, activated antigen specific CD8 T cells and development of autoimmune cholangitis in dnTGFbetaRII mice. Hepatology, 2013. 13. Santibañez, J.F., M. Quintanilla, and C. Bernabeu, TGF-β/TGF-β receptor system and its role in physiological and pathological conditions. Clinical Science, 2011. 121(6): p. 233- 251. 14. Hayashi, T., et al., Ex vivo induction of multiple myeloma–specific cytotoxic T lymphocytes. Blood, 2003. 102(4): p. 1435-1442. 15. Mano, Y., Duct formation by immortalized mouse cholangiocytes: an in vitro model for cholangiopathies. Laboratory investigation, 1998. 78(11): p. 1467-8. 16. Bengsch, B., et al., Analysis of CD127 and KLRG1 Expression on Hepatitis C Virus- Specific CD8+ T Cells Reveals the Existence of Different Memory T-Cell Subsets in the Peripheral Blood and Liver. Journal of Virology, 2007. 81(2): p. 945-953. 17. Cush, S.S. and E. Flaño, KLRG1+NKG2A+ CD8 T Cells Mediate Protection and Participate in Memory Responses during γ-Herpesvirus Infection. The Journal of Immunology, 2011. 186(7): p. 4051-4058. 18. Kita, H., et al., Identification of HLA-A2-restricted CD8(+) cytotoxic T cell responses in primary biliary cirrhosis: T cell activation is augmented by immune complexes cross- presented by dendritic cells. J Exp Med, 2002. 195(1): p. 113-23. 19. Kita, H., et al., Quantitative and functional analysis of PDC-E2-specific autoreactive cytotoxic T lymphocytes in primary biliary cirrhosis. J Clin Invest, 2002. 109(9): p. 1231- 40.

65

20. Tsuda, M., et al., Fine phenotypic and functional characterization of effector cluster of differentiation 8 positive T cells in human patients with primary biliary cirrhosis. Hepatology, 2011. 54(4): p. 1293-1302. 21. Lan, R.Y., et al., Liver-targeted and peripheral blood alterations of regulatory T cells in primary biliary cirrhosis. Hepatology, 2006. 43(4): p. 729-737. 22. Bernuzzi, F., et al., Phenotypical and functional alterations of CD8 regulatory T cells in primary biliary cirrhosis. Journal of Autoimmunity, 2010. 35(3): p. 176-180. 23. McNally, A., et al., CD4+CD25+ regulatory T cells control CD8+ T-cell effector differentiation by modulating IL-2 homeostasis. Proceedings of the National Academy of Sciences, 2011. 108(18): p. 7529-7534. 24. Nikolova, M., et al., Regulatory T cells differentially modulate the maturation and apoptosis of human CD8+ T-cell subsets. Blood, 2009. 113(19): p. 4556-4565. 25. Almeida, A.R.M., et al., CD4+CD25+ Treg regulate the contribution of CD8+ T-cell subsets in repopulation of the lymphopenic environment. European Journal of Immunology, 2010. 40(12): p. 3478-3488. 26. Stephens, G.L., et al., Engagement of Glucocorticoid-Induced TNFR Family-Related Receptor on Effector T Cells by its Ligand Mediates Resistance to Suppression by CD4+CD25+ T Cells. The Journal of Immunology, 2004. 173(8): p. 5008-5020. 27. Sanjabi, S., M.M. Mosaheb, and R.A. Flavell, Opposing Effects of TGF-² and IL-15 Cytokines Control the Number of Short-Lived Effector CD8+ T Cells. Immunity, 2009. 31(1): p. 131-144. 28. Thimme, R., et al., Increased Expression of the NK Cell Receptor KLRG1 by Virus- Specific CD8 T Cells during Persistent Antigen Stimulation. Journal of Virology, 2005. 79(18): p. 12112-12116. 29. Rubinstein, M.P., et al., IL-7 and IL-15 differentially regulate CD8+ T-cell subsets during contraction of the immune response. Blood, 2008. 112(9): p. 3704-12. 30. Tripathi, P., et al., STAT5 is critical to maintain effector CD8+ T cell responses. J Immunol, 2010. 185(4): p. 2116-24. 31. Wilson, D.C., S. Matthews, and G.S. Yap, IL-12 signaling drives CD8+ T cell IFN- gamma production and differentiation of KLRG1+ effector subpopulations during Toxoplasma gondii Infection. J Immunol, 2008. 180(9): p. 5935-45. 32. Joshi, N.S., et al., Inflammation directs memory precursor and short-lived effector CD8(+) T cell fates via the graded expression of T-bet transcription factor. Immunity, 2007. 27(2): p. 281-95. 33. Yoshida, K., et al., Deletion of interleukin-12p40 suppresses autoimmune cholangitis in dominant negative transforming growth factor beta receptor type II mice. Hepatology, 2009. 50(5): p. 1494-500. 34. Tsuda, M., et al., Deletion of interleukin (IL)-12p35 induces liver fibrosis in dominant- negative TGFbeta receptor type II mice. Hepatology, 2013. 57(2): p. 806-16. 35. Chang, P.P., et al., Breakdown in repression of IFN-gamma mRNA leads to accumulation of self-reactive effector CD8+ T cells. J Immunol, 2012. 189(2): p. 701-10. 36. Curran, M.A., et al., Systemic 4-1BB activation induces a novel T cell phenotype driven by high expression of Eomesodermin. J Exp Med, 2013. 210(4): p. 743-55. 37. Curran, M.A., et al., Combination CTLA-4 blockade and 4-1BB activation enhances tumor rejection by increasing T-cell infiltration, proliferation, and cytokine production. PLoS ONE, 2011. 6(4): p. e19499. 38. Hirschhorn-Cymerman, D., et al., Induction of tumoricidal function in CD4+ T cells is associated with concomitant memory and terminally differentiated phenotype. J Exp Med, 2012. 209(11): p. 2113-26. 39. Kahraman, A., G. Gerken, and A. Canbay, Apoptosis in immune-mediated liver diseases. Dig Dis, 2010. 28(1): p. 144-9.

66

40. Lleo, A., et al., Apotopes and the biliary specificity of primary biliary cirrhosis. Hepatology, 2009. 49(3): p. 871-9. 41. Ambrosini, Y.M., et al., The multi-hit hypothesis of primary biliary cirrhosis: polyinosinic-polycytidylic acid (poly I:C) and murine autoimmune cholangitis. Clin Exp Immunol, 2011. 166(1): p. 110-20.

67

FOOTNOTE PAGE

Address correspondence to: William M. Ridgway, HPB room 356, 231 Albert Sabin

Drive, University of Cincinnati College of Medicine, Cincinnati, Ohio, 45267-0563 phone: (513) 558 5551 fax: (513) 558 3799

68

Chapter 3 Aberrant Pkhd1 expression causes autoimmune biliary disease on the NOD genetic background

69

Aberrant Pkhd1 expression causes autoimmune biliary disease on the NOD genetic background

Wenting Huang*, Daniel B. Rainbow †, Yuehong Wu*, David Adams*, Pranavkumar Shivakumar‡,

Leah Kottyan§, Mehdi Keddache¶, Satwica Yerneni¶, Bruce Aronow||, Rebekah Karns||, Jorge

Bezerra‡, M. Eric Gershwin#, Linda S. Wicker†, and William M Ridgway*

*1 Division of Immunology, Allergy and Rheumatology, University of Cincinnati College of

Medicine, Cincinnati Ohio, 45267;

† Juvenile Diabetes Research Foundation/Wellcome Trust Diabetes and Inflammation Laboratory,

Department of Medical Genetics, Cambridge Institute for Medical Research, University of

Cambridge, Cambridge CB2 0XY, UK.

‡Division of Pediatric Gastroenterology, Hepatology and Nutrition, Cincinnati Children's Hospital

Medical Center, Cincinnati OH, 45229

§Department of Pediatrics, Cincinnati Children's Hospital Medical Center, Cincinnati OH, 45229

¶Division of Human Genetics, Cincinnati Children's Hospital Medical Center, Cincinnati OH,

45229

||Division of Biomedical Informatics, Cincinnati Children's Hospital Medical Center, Cincinnati

OH, 45229

70

#Division of Rheumatology, Allergy and Clinical Immunology, University of California at Davis,

Davis CA 95616;

Corresponding author: William M Ridgway; Telephone: 513-558-4701; Fax: 513-558-3799;

Email: [email protected]

71

Abstract

Several nonobese diabetic (NOD) congenic strains, (including “NOD.c3c4” and “NOD.ABD” mice) develop spontaneous autoimmune biliary disease (ABD) that requires an adaptive immune system, is transferable by T lymphocytes, and features anti-PDC-E2 autoantibodies. Here we demonstrate that NOD ABD is linked to a B6-derived genetic region on chromosome one, and identify the genetic cause of this disease as aberrant transcription of Polycystic kidney and hepatic disease 1 (Pkhd1). We have created a novel congenic mouse expressing aberrant Pkhd1 and designate it NOD.Abd3. NOD.Abd3 common bile duct (CBD) is morphologically abnormal at two weeks of age, and demonstrated upregulation of genes involved in cholangiocyte injury/morphology and downregulation of immune-related genes. RNAseq data revealed an abnormal transcription pattern of Pkhd1: upregulation of exons 1-35 and essentially no expression of exons 36-67. PKHD1 is responsible for the human disease, autosomal recessive polycystic kidney disease (ARPKD), and various mutations of Pkhd1 cause biliary abnormalities in mice, but mutations of Pkhd1 have not previously been associated with an autoimmune disease.

Abnormal Pkhd1 transcription in NOD.Abd3 resulted from a genomic insertion event in intron

35, which created a novel alternative polyadenylation site. Clinical biliary disease was eliminated, and abnormal biliary histology decreased, by backcrossing the Pkhd1 mutation onto the B6 genetic background. Neither NOD nor NOD.Abd3 bone marrow caused disease when transferred into NOD recipients, whereas NOD.Abd3 but not NOD bone marrow transferred biliary disease into NOD.Abd3 irradiated recipients. This plus our previously published results clearly shows that the pathogenesis of ABD requires expression of Abd3 in both the immune system and in the target tissue in the context of the NOD genetic background. We interpret these findings to mean that loss of functional Pkhd1 on the NOD background produced early bile duct abnormalities, initiating an autoimmune response that ultimately produced clinical autoimmune biliary disease in NOD.Abd3 congenic mice.

72

Introduction

Autoimmune biliary disease (ABD) in humans is manifest most commonly as primary biliary cirrhosis (PBC) and primary sclerosing cholangitis (PSC) in adults [1-3] and biliary atresia (BA) in children [4, 5]. The biliary epithelial cell (BEC; cholangiocyte) is the primary autoimmune target in these diseases [6, 7]. Several animal models of these diseases have been established. For

PBC, the NOD.c3c4 [8, 9], NOD.ABD [10], and dnTGFβRII mice [11] develop spontaneous

ABD, whereas infection of neonatal Balb/c mice produces an autoimmune reaction very similar to BA [7, 12]. The NOD.c3c4 strain arose in a project to refine Insulin Dependent Diabetes (Idd) loci by introgressing B6 Idd regions onto the NOD genetic background [8]. NOD.c3c4 mice were completely protected from diabetes, but developed ABD characterized by hepatosplenomegaly, wasting, abdominal swelling, liver function abnormalities, and eventually death from obstructive liver disease. Histologically their livers showed substantial lymphocytic infiltration, nonsuppurative destructive cholangitis (NSDC), and macrophage aggregation in the bile ducts; all features similar to human PBC. In addition, the NOD.c3c4 mouse spontaneously develops anti-

PDC antibodies, which are highly specific for human PBC. In contrast to human PBC, the

NOD.c3c4 mice also developed common bile duct dilation and inflammation, which more closely resembles BA. Finally, they developed extensive proliferation of intrahepatic bile ductules, far exceeding the ductule proliferation seen in stage II human PBC [8]. Transfer of splenocytes from

NOD.ABD donors with severe disease, however, resulted in overwhelming inflammation, NSDC, high titer anti-PDC-E2 antibodies, and severe illness in NOD.c3c4-scid recipients—in the absence of any additional significant ductule proliferation [10]. We concluded that the NOD.c3c4 model was a useful model for understanding autoimmune biliary disease.

We have previously identified several mechanisms of NOD autoimmune biliary disease. First,

NOD.c3c4-scid mice do not develop clinical disease and have much diminished hepatic

73 histological abnormalities, indicating that the adaptive immune system is critical to disease pathogenesis [10]. Second, CD8+ T cells from NOD.ABD donors, but not NOD donors, transferred disease into NOD.c3c4-scid recipients [10]. Finally, NOD-scid recipients did not get any ABD in adoptive transfer, even when receiving 20 million splenocytes from NOD.ABD donors that rapidly caused overwhelming disease in NOD.c3c4-scid recipients [10]. These results showed that the genetic background of the target tissue was critical to disease pathogenesis, however it was not completely clear whether disease required expression of the B6/B10 genetic components in the adaptive immune system. An alternate possibility was that NOD splenocytes could not cause disease because their T cells were not expanded, because they were never exposed to the BEC auotantigen(s). This issue is addressed in the current paper by constructing bone marrow chimeric mice.

We used a congenic mapping approach to better understand the genetic origin of ABD in the

NOD.c3c4 model [10]. Although regions on chromosomes 3 and 4 were initially linked to disease, a 5K SNP chip analysis showed a small B6 region on chromosome one in all strains that developed disease. In order to assess the role of this region we constructed a new congenic strain, herein designated NOD.Abd3 that had a 1.0 Mb B6 region on chromosome one on the NOD background. The NOD.Abd3 mouse developed ~100% penetrant ABD in the absence of the original chromosome 3 and 4 regions. We undertook the studies in the present paper to understand how the chromosome one genetic region is necessary and sufficient to mediate biliary disease, and to further characterize the role of genetic background and the chromosome one region in the target tissue (biliary epithelium) and hematopoietic system. We discovered the causative gene for ABD, Pkhd1, located 2.5 Mb upstream of the introgressed Abd3 B6 region.

74

Materials and methods

Animals and genotyping:

NOD.Abd3 (line 8706), NOD, NOD.CD45.2 congenic mice (strain 6908), B10.H2bg7, B6 mice and corresponding F1, F2 and BC2i strains were bred and housed under specific pathogen free conditions, and all procedures were conducted according to approved protocols of the University of Cincinnati College of Medicine Animal Care and Use Committee. NOD.Abd3 were crossed with B6 and then 1) the F1 offspring were intercrossed to get the F2, or 2) twice to B6 to generate Backcross 2 (BC2) mice, which were then further intercrossed to generate BC2 intercrossed (BC2i) generation.

NOD.Abd3 backcrossed mice were genotyped as follows using markers outside the B6 congenic region, which genotype as NOD for NOD.Abd3 mice. The proximal marker is rs13475762 with the forward primer rs13475762_F = TTCCCCCTTTTAATATTTTGCAT and reverse primer rs13475762_R = CAGGGAGGCAGTGATTTAGC. The distal marker is rs32040516 with forward primer rs32040516_F = TGAGCCATCTGACAGACCAG and reverse primer rs32040516_R = TGGATGGCCATGACAAAAA. PCR product was digested with restriction enzyme ACC1 (for proximal marker) or MNL1 (for distal marker) (New England Biolabs) at

37ºC for 2 hours and run PCR products on a 4% agarose gel with 1X TBE buffer. The size of the digested PCR product of proximal markers is 87 bp for NOD and the size of the digested PCR product of distal markers is 182 bp.

Histology:

Livers and/or common bile ducts were isolated from mice and immediately fixed in 10% formalin before being embedded in paraffin. Samples were stained with hematoxylin and eosin, and scored blindly using microscopy for duct epithelial hyperplasia and lymphocytic infiltration semi- quantitatively.

75

Histopathology and disease scoring

Histological liver score: duct epithelial hyperplasia:

0 = 0~3 ductules per portal triad across the section

1 = a.) at least 2 times the normal diameter of ductule lumen in 10% ~ 25% of the entire section.

b.) no fewer than 3 small ductules around one triad in 25% ~ 50% of the entire section.

2 = enlarging ductules across 25% ~ 50% of the entire section. The average diameter of a ductule

lumen section is about 2~4 times of the normal ductual diameter

3 = larger ductules across >50% of the entire section. The average diameter of ductule lumen

section is about 4~5.5 times of the normal ductule diameter

4 = dilated, diffuse, torturous ducts (“cysts”) in much of the section; this is the endstage disease.

The average diameter of ductule lumen section is more than 5.5 times the normal ductual

diameter

Histological score: leukocytic infiltration of portal areas:

0 = none or a few cells in <25% of entire section;

1= small numbers of cells in multiple areas (25% ~ 50%), including some clusters of immune

cells;

2 = small numbers of cells (a few cells thick infiltrate) diffusely (>50% of entire section); or

patchy moderate infiltrates (25%~50% of entire section);

3 = moderate numbers of cells diffusely (>50% of entire section) or patchy large infiltrates (larger

cluster than what is described in “2”, 25%~50% of entire section);

4 = diffuse or significant sections of large cellular infiltrates.

76

Common bile duct (CBD) and liver gross pathological scoring

CBD dilation and the quantitation of liver abnormalities were examined and scored separately.

The clinical CBD score was measured as the diameter (in mm) of the maximum CBD width. The clinical liver score was assigned based on tissue induration (hardness) and the presence of cysts on the surface of liver: 0: soft and no cysts; 1: mild induration, no cysts; 2 indurated with cysts on one lobe surface; 3 indurated with cysts on two to three lobes; 4 indurated and cysts on all four lobes.

Bone marrow Chimeras

2~3-month-old NOD (CD45.1) or NOD.Abd3 (CD45.1) mice were irradiated with 1200 Rads.

15~20 million bone marrow cells from 4-month-old NOD.CD45.2 or 2~3-month-old NOD.Abd3

(CD45.1) donors were extracted without RBC lysis. Mature CD4, CD8 and CD90 cells were removed using magnetic beads (Miltenyl Biotech, California) and the bone marrow was injected i.v. into the irradiated recipient mice. Recipient mice were given water treated with antibiotic

(neomycin trisulfate salt hydrate) for two weeks after transfer. The recipient mice were sacrificed

120 days post bone marrow transfer or when they developed abdominal swelling indicating severe hepatobiliary disease (approximately 2.5 months for NOD.Abd3 recipients of NOD.Abd3 bone marrow). Tissues were analysed for anatomical and histological biliary/liver disease.

CD45.1 or CD45.2 quantification was done for the recipients of different donor cells and the bone marrow reconstitution rate was from 87.6% to 98.9 %.

Quantitative real time PCR of mRNA

Extrahepatic common bile ducts were isolated from 2-week-old NOD.Abd3 and NOD mice and stored in RNAlater solution (Qiagen). For RNA extraction from CBD, the tissue was first removed from RNAlater solution and disrupted in lysing/binding buffer provided in the mirVanaTM miRNA Isolation Kit (Life Technologies) using a homogenizer (Qiagen) and then

77

RNA was extracted from the tissues using mirVana™ miRNA Isolation Kit (Life Technologies) and converted to cDNA by random primers (High capacity cDNA Reverse Transcription kit,

Applied Biosystems). Relative gene expression of Pkhd1 was performed using selective mouse

Taqman gene expression assays (Mm00467747_m1, Mm00467728_m1,

Mm01233737_m1, Life Technologies) and endogenous control assay Rn18S (Mm03928990_g1,

Life Technologies)

RNA-seq profiling of common bile ducts and data analysis

Total CBD RNA was extracted from 2 week old NOD or NOD.Abd3 mice using mirVana miRNA isolation kit (Life Technologies, CA) from derived. Two samples were sequenced for each strain; each of the two NOD.Abd3 samples have RNA from three individual CBDs and the two NOD samples have RNA from three or five individual CBDs respectively. RNA concentration, purity (rRNA 28 s:18 s ratio), and quality (RNA Integrity Number (RIN)) was assessed using the 2100 Bioanalyzer (Agilent, Santa Clara, CA) at the DNA Sequencing and

Genotyping Core of Cincinnati Children’s Hospital Medical Center. The RNA samples had RIN of at least 7.9 automatically designated by the software of the Agilent bioanalyzer in a scale of 1 to 10 with 10 being completely intact.

The RNA sequencing library was generated with the Illumina TruSeq RNA preparation kit and subsequently sequenced on the Illumina Hi-Seq 2000, using single-end, 50-bp read specifications with a read depth of at least 10 million (Illumina, San Diego, CA).

RNA-seq analysis was carried out using Bowtie [13] and Tophat2 [14]. RNA-seq BAM files generated using the Bowtie-Tophat2 pipeline were analyzed for the expression of known and unknown genes/transcripts using the Cufflinks2 pipeline [15] and included removal of reads that did not map uniquely to the mm9 ENSEMBL genome or ENSEMBL-annotated genes.

Workflows included filtering to remove duplicate reads, and those with post-aligned read metrics

78 mapping quality below 40. Our samples showed normal 3’/5’ read distribution ratios as compared to hundreds of other samples run in the CCHMC genomics core. Transcript/isoform and gene summarized expression tables were filtered to identify entities whose expression was at least 5 FPKM (fragment per kilobase of exon per million fragments mapped, Cufflinks) or 5

RPKM (reads per kilobase per million reads mapped, GeneSpring) in at least one sample.

Differentially expressed gene signatures were identified using Audic Claverie tests (P < 0.05) and

Students t test (FDR<0.05) followed by a two -fold change requirement. Gene Ontology and other enrichment and biological network analysis was carried out using ToppGene

(http://toppgene.cchmc.org) [16], and ToppCluster (http://toppcluster.cchmc.org) [17] followed by ToppCluster output of xgmml data files that were network-analyzed in Cytoscape yielding both similar and complementary results as compared to the Toppfun functional enrichment analysis.

PCR of Pkhd1 exons and intron35

For different exons of Pkhd1, the following primer pairs were designed using Primer3

(http://biotools.umassmed.edu/bioapps/primer3_www.cgi) to amplify representative exon sequences in Pkhd1 using genomic DNA as template:

Exon16_forward primer (5’→3’): GCACTGCTGACTGGTTTGAC

Exon16_reverse primer (5’→3’): ACCAGGTGTTGTGGATCTGG

Exon34_forward primer (5’→3’): CTCCATGGGGACAGAGAGAC

Exon34_reverse primer (5’→3’): GAAGAGGCCTGAGAACGATG

Exon40_forward primer (5’→3’): CATTCAGTCCTTCCCAGAGG

Exon40_reverse primer (5’→3’): TCATAGCTCCCACCAGTGC

79

Exon61_forward primer (5’→3’): GACAATCCCCAAATCACACC

Exon61_reverse primer (5’→3’): GCGCTTTCGTTTTCTTTCAC

For Pkhd1 intron 35, the following primer sets were designed by Primer3 or PCRTiler v1.42[18]:

Primer 1_forward primer (5’→3’): TTTCCCTTCCAGGCATCA

Primer 1_reverse primer (5’→3’): TGGTAGCCACAGATGGCTTT

Primer 2_forward primer (5’→3’): AACCCCTAATTCCCAGGTCA

Primer 2_reverse primer (5’→3’): TGAGCACGTGCATGCTATGA

Primer 3_forward primer (5’→3’): GCAGAGACCCAAGCTGCTG

Primer 3_reverse primer (5’→3’): TGAGCTGCGTGTCAGTGAA

Primer N1_forward primer (5’→3’): GTCCCCAATGGAGGAGCTAA

Primer N1_reverse primer (5’→3’): AGCCCAGACAGAAGGACAGC

Primer N2_forward primer (5’→3’): GAGGCAGAACCTGGGTCAC

Primer N2_reverse primer (5’→3’): GGTGTTCCAGGGGAGAGC

Primer N3_forward primer (5’→3’): TGTAACCCAGCCCTGTTCA

Primer N3_reverse primer (5’→3’): ACAATGCGAGGCAACCAT

Primer N4_forward primer (5’→3’): GGGAATTTGCAGGGGACTC

Primer N4_reverse primer (5’→3’): GTTTCTGCCTGGCCTTGG

Primer N5_forward primer (5’→3’): TGACTGGAATCCTCTGAGTCCT

80

Primer N5_reverse primer (5’→3’): TGCTTCCAACTGATGCCTTT

Primer N6_forward primer (5’→3’): CCCTGGGTTTGCTCCTCT

Primer N6_reverse primer (5’→3’): AGGGGTGGGAACCAGAAA

Primer N7_forward primer (5’→3’): TGGATACTTCATCCCTCCTTAAAA

Primer N7_reverse primer (5’→3’): ACACAAGCTCTGGGGGCTAC

Primer N7_forward primer and Primer 3_reverse primer were used with PrimerStar GXL DNA

Polymerase (Clontech Laboratories, Inc., CA) for long-range PCR to discover the inserted DNA fragment in Pkhd1 intron 35 in NOD.Abd3. Long-range PCR products from NOD and NOD.Abd3 were then sent for Next-Generation Sequencing at Cincinnati Children’s Hospital Medical Center

DNA Sequencing and Genotyping Core.

Statistical analysis

GraphPad Prism 5 (GraphPad Software, San Diego, CA) software package was used to perform unpaired, two-tailed Student t test for comparison of the expression level of genes and microRNAs, as well as Mann–Whitney U test and/or Wilcoxon signed rank test to estimate significance levels of live histology rank scores.

81

Results

A 1.0 Mb introgressed region on chromosome one is necessary and sufficient for autoimmune biliary disease

We have previously published that a 5k SNP chip analysis revealed a small region of B6 genetic material on chromosome one in the NOD.c3c4 mouse [10]. To assess the significance of this region we developed a chromosome one congenic mouse (strain 7825 in Fig1A) and designated the c1 region as “Abd3” (chromosome 3 and 4 regions previously described in NOD.ABD [10] were termed Abd1 and Abd2, respectively) . Abd3 alone on the NOD background resulted in a highly penetrant autoimmune biliary disease. We genotyped our previously published NOD.c3c4 and NOD.ABD strains and found that they carried the Abd3 region. Therefore Abd3 is sufficient for ABD (Fig 1). We created several additional congenic strains to narrow the Abd3 region. In each case, the chromosome one region was necessary for disease (Fig1B). The smallest congenic interval, 1.0 Mb in strain 8706, was both necessary and sufficient to develop autoimmune biliary disease on the NOD background; we designated this strain as “NOD.Abd3”.

Abd3 is a recessive allele and clinical autoimmune biliary disease requires the NOD genetic background

To understand the interaction between the Abd3 allele and the NOD background on the disease, we performed backcross-intercross studies. We first crossed NOD.Abd3 with B6.Pl mice (a variant B6 strain originally used to generate the founder NOD congenics) or NOD. The F1 offspring had no liver histological abnormality (Fig 2A) compared to NOD.Abd3 mice, clearly demonstrating that Abd3 is a recessive allele. This also confirmed that the Abd3 allele differs from the original B6.Pl allele, since these two regions were not complementary for disease.

(NOD.Abd3 x B6) F1 mice also did not develop disease (data not shown). To reduce the proportion of NOD background in the mice, we produced F2 mice (average 50% NOD background), and backcross 2 intercrossed (BC2i) mice (average 12.5% NOD background). The

82 visible pathological CBD enlargement and anatomically visible liver abnormalities seen in Abd3 homozygous animals are dramatically reduced in F2 and BC2i offspring that express homozygous

Abd3 but with a reduced NOD background (Fig 2B and C). Therefore reducing NOD genetic background and increasing B6 genetic background ameliorates ABD in Abd3 homozygous animals, strongly emphasizing the importance of the NOD genetic background for the presence of

ABD.

We performed histological analysis of the same crosses with similar results. Animals lacking the

Abd3 allele, or heterozygous for the allele in F2 and BC2i show no histological lesions, whereas homozygous animals show reduced histology scores for duct proliferation and leukocytic infiltration compared to NOD.Abd3 mice (Fig 2D and E). Both the clinical and histological analyses demonstrate that the NOD background interacts with the B6-derived Abd3 region to cause autoimmune biliary disease.

Bone marrow chimera studies demonstrate that expression of Abd3 in both target tissue and immune cells is necessary for disease

We next dissected the level at which the NOD background was acting in the autoimmune process.

We had previously shown that NOD.ABD splenocytes or CD8 cells could not transfer disease into NOD.scid mice, and conversely that NOD splenocytes or CD8 cells could not transfer disease into NOD.c3c4-scid mice. These results suggested that the Abd3 locus acted on both the target tissue and the immune system. An alternate explanation, however, was that the NOD CD8

T cell repertoire has the potential to cause ABD, but that the disease-mediating T cell clones were not expanded since they were never exposed to their tissue autoantigen (i.e. NOD mice lack abnormal BECs). To distinguish between these possibilities we performed bone marrow chimera studies. NOD.Abd3 irradiated recipients were reconstituted with either NOD or NOD.Abd3 bone marrow, and conversely, NOD recipients were reconstituted with NOD.Abd3 bone marrow. The results clearly reproduced the transfer study findings: the recipients must express NOD.Abd3 in

83 biliary tissue to develop disease. NOD.Abd3NOD recipients had no clinical abnormality in

CBD or liver (Fig 3A, right side, bar 7 and 8) and no duct epithelial hyperplasia and minimal lymphocytic infiltrates (Fig 3B, right side, bar 7 and 8). In addition, NOD BM did not cause significant disease in NOD.Abd3 recipients—both the clinical gross liver score and histological duct epithelial hyperplasia score at 120 days post-BM transplant were significantly less than

NOD.Abd3NOD.Abd3 recipients (Fig 3A and B). NOD.Abd3 recipients already had a certain amount of disease at the time of radiation (Fig 3A, B; bars 1 and 2). In the NOD.Abd3

NOD.Abd3 transfer the NOD.Abd3 bone marrow repopulated the mouse and caused the disease to progress. In the NOD NOD.Abd3 mice, the NOD bone marrow not only did not further advance the disease, but in fact the disease score was less than in 10 week old NOD.Abd3 mice

(Fig 3A, B, bars (1, 2) compared to bars (5,6)). Therefore, there is clearly a difference in the ability of the hematopoietic system to cause disease between the two strains; the Abd3 allele must be expressed in both hematopoietic system and target tissue in order for autoimmune biliary disease to develop.

Early cholangiocyte damage response and abnormal immune activation in NOD.Abd3 common bile ducte

In the NOD mouse, islet inflammation begins as early as 3 weeks [19] and is associated with developmental processes of the pancreatic islets. We examined the CBD of two week old mice and found disrupted biliary epithelial architecture in NOD.Abd3 compared to the smooth lumen in the CBD from NOD and NOD.c3c4-scid mice (Fig 4A-C). In the liver, no significant histological difference was observed between NOD.Abd3 and NOD control at this very early time point (not shown). These findings are consistent with our previously published results showing that hepatic inflammation was not detectable until ~ 8 weeks in NOD.ABD mice, and that NOD.c3c4-scid mice had much decreased hepatic disease [8, 20]. To further understand the striking difference between NOD and NOD.Abd3 CBD architecture at 2 weeks, we performed RNAseq on CBD

84 from 2 week old mice. 255 genes were significantly differentially expressed between the strains

(Fig 5A). Cluster enrichment network analysis of these 255 genes based on biological processes, molecular functions, mouse knockout phenotypes and biological pathways showed that many of the genes upregulated in 2wk NOD.Abd3 CBD were extensively involved in cholangiocyte injury

(reflecting the tissue target cell in ABD) whereas genes involved in immune response (both adaptive and innate) were downregulated in NOD.Abd3 CBD (Fig 5B). These results establish a very early tissue abnormality interacting with an abnormal immune response that drives the development of NOD.Abd3 biliary disease.

Bile duct abnormality in NOD.Abd3 caused by aberrant expression of Pkhd1

A major unresolved problem in our NOD congenic mouse models has been the genetic cause of the abnormal biliary tissue. We extensively investigated possible candidate genes within the Abd3 boundaries and did not find convincing expression or sequence differences between NOD and

NOD.Abd3 (data not shown). We therefore hypothesized that genetic differences/mutations centromeric to the Abd3 region could be linked to it (telomeric linkage was excluded by the recombinants shown in Fig 1). By comparing RNAseq reads from 0-24 Mb on chromosome one between NOD and NOD.Abd3, we found a single major genetic difference between these strains:

Polycystic Kidney and Hepatic Disease 1 (Pkhd1) was aberrantly expressed in NOD.Abd3 compared to NOD CBD samples. Pkhd1 is located ~2.5 Mb upstream from the proximal border of our Abd3 region. Pkhd1 is the causative gene for the monogenic human autosomal recessive polycystic kidney disease (ARPKD) [21]. In both human and murine models of ARPKD with mutated Pkhd1, disease manifestations vary due to mutation location, but mutations mediating biliary-hepatic disease (in absence of renal pathology) produce abnormalities very similar to

NOD.Abd3 [22-25]. RNAseq-reads aligned to the reference genome showed that NOD.Abd3 lacked expression of exon 36 to exon 67 of Pkhd1 compared to NOD controls (Fig 6A).

Moreover, the read-density aligned to each exon of Pkhd1 indicated that the expression level of

85 the exon 1-35 of Pkhd1 in NOD.Abd3 CBD at 2 weeks of age is higher than that in NOD control; in contrast, the read-density of all 67 Pkhd1 exons in NOD was relatively constant (Fig 6B).

To rule out the possibility that truncated expression of Pkhd1 was due to genomic DNA deletion of exons 36-67, we designed primers to amplify genomic DNA for proximal and distal exons before and after the observed transcript cutoff site. PCR showed product bands for all representative exons in NOD.Abd3 comparable to NOD and B6 (Fig 6C), clearly demonstrating that missing expression of Pkhd1 exons 36 to 67 in NOD.Abd3 was not caused by a genomic

DNA deletion event. We then tested Pkhd1 expression level using qRT-PCR. Expression of exons 15/16 in 2wk NOD.Abd3 was nearly two times of that in NOD and B6 (Fig 6D), consistent with the RNAseq results (Fig 6B).There was no significant difference in expression of these exons between NOD and B6 (Fig 6D), and the expression level within either NOD or B6 was relatively consistent (data not shown). Although, to our knowledge, the expression of Pkhd1 in immune cells has never been reported previously, the Ensembl database shows that RNAseq reads from spleen are aligned to Pkhd1 [26]. We hypothesized that deficient exon expression pattern of Pkhd1 could be detected not only in CBD, the target tissue but also in immune cells, explaining why both affected systems were necessary for ABD. However, we were unable to detect the Pkhd1 expression in either bulk splenocytes and intrahepatic immune cells (IHC) or

MACS-sorted CD4+ or CD8+ cells from spleen and liver (data not shown).

Details of the Pkhd1 genetic mutation in NOD.Abd3 and potential mechanism of truncated transcription

Pkhd1 is ~2.5 Mb upstream from the Abd3 allele (Fig 7A). The RNAseq data as well as the qRTPCR result revealed a novel loss of Pkhd1 transcription after exon35, and we sought to understand how this transcription deregulation occurs in NOD.Abd3. The alignment of the

RNAseq reads from 2wk NOD.Abd3CBD and NOD CBD identified a unique, transcribed intron region immediately after exon35 (the second longest Pkhd1 intron) in NOD.Abd3 but not in NOD

86 control (Fig 7B). We found no mutations at the splice donor site (“GT”) after exon35 or splice acceptor site (“AG”) before the start of exon 36. On the other hand, there were multiple microsatellite repeats in the uniquely transcribed intron region, which suggested that it could be a

3’UTR-like region in Pkhd1.

To further evaluate the unique 3’ UTR-like region in intron 35 of NOD.Abd3 Pkhd1, we designed primer pairs to perform tiled PCR starting from the end of the 3’UTR-like region in intron 35 towards exon 35 (Fig 8A). The first two pairs of tiled PCR primers at the end of the novel 3’UTR-like region of intron 35 were able to amplify in NOD but not in NOD.Abd3 (Fig

8B). We then designed 7 primer pairs (every 5kb) from exon 36 towards the end of the novel 3’

UTR like region in intron 35 (Table 1). Each of these primer pairs amplified in both NOD and

NOD.Abd3 (Fig 8C). Lastly, we used the forward primer of primer pair N7 (which started on chr1:

20470616, C57Bl/6 genome assembly NCBI37/mm9) and the reverse primer of primer pair 3

(which ended on chr1: 20473540, C57Bl/6 genome assembly NCBI37/mm9) (Fig 8A) and found that the PCR product size in NOD was about 3kb (which was similar to reference target sequence). In striking contrast, the PCR product size in NOD.Abd3 was about 9kb (Fig 8D).

Therefore, the mutation within Pkhd1 was a DNA insertion of about 6 kb.

Using Next-Generation Sequencing technology, we sequenced the PCR product through de novo reads assembly. The assembled PCR product sequence of NOD could be fully aligned with the reference genome while only part of the three nodes of assembly sequence of NOD.Abd3 matched with NOD sequence. The node-2 assembled sequence in NOD.Abd3 Pkhd1 intron 35 was searched against the NCBI chromosome database and had a best hit with the intronic sequence of a mouse gene E3 Ubiquitin Ligase Precursor on chromosome 11 (Fig 8E).

Given these results of genetic characterization of Pkhd1 in NOD.Abd3, we propose the following potential mechanism: the 6kb DNA insertion in intron 35 of NOD.Abd3 Pkhd1 potentially disrupts normal recognition of splicing sites in this intron and forces the exposure of an alternative polyadenylation site within the intron, thus giving rise to a novel shortened transcript

87 which ends after exon35 (Fig 8F). Given that qRT-PCR and RNAseq shows an upregulation of the first 35 exons of Pkhd1 in NOD.Abd3 CBD, this alternative polyadenylation (APA) site must predominate over the normal polyadenylation site (after exon67) in NOD.Abd3.

88

Discussion

This work advances our understanding of NOD congenic autoimmune biliary disease in several ways. First, we prove conclusively that the ABD disease is caused by a region on chromosome one. This region, designated “Abd3”, acts in a recessive fashion and is necessary and sufficient for the autoimmune biliary disease. Next, using genetic breeding approaches, we show that this chromosome one region must interact with the NOD background to mediate biliary disease. No clinical disease is present when the Abd3 locus is backcrossed to the B6 background. Some histological abnormality is present, but in the absence of the NOD genetic background this does not rise to the level of clinical disease. Third, we used bone marrow chimera analysis to show that both the target tissue and the hematopoietic system must express the chromosome one region for

ABD to develop. These findings support our previously published transfer studies, showing that

NOD.scid recipients developed no ABD even with adoptive transfer of a huge number of

NOD.ABD splenocytes (sufficient to cause severe, overwhelming disease in NOD c3c4-scid recipients) [10]. In the bone marrow studies, irradiated NOD recipients of NOD.Abd3 bone marrow developed no clinical disease. Furthermore, expression of Abd3 in the donor hematopoietic system was necessary for disease in the bone marrow chimeras. Fourth, ABD arises remarkably early in the common bile ducts of young mice (already apparent at the age of 2 weeks), and this is associated with a strong, significant, and widespread differential expression of genes largely related to cholangiocyte response and the immune system. In contrast, NOD.c3c4- scid mice have minimal inflammation in CBD at the two week time point, indicating that the adaptive immune system is playing a role even at this early time point.

Finally, we show here that NOD ABD is caused by abnormal expression of Pkhd1. Human

PKHD1 causes the disease ARPKD. Pkhd1 is a large gene with 67 exons (4059 amino acids) encoding fibrocystin, a membrane-associated receptor-like protein with a single transmembrane domain, a small intracellular domain and a large extracellular domain contains multiple repeats of

IPT (immunoglobin – plexin-transcription factor like) domains and two G8 domains (predicted to

89 contain 10 beta strands and an alpha helix) [27, 28]. Fibrocystin has domains that are similar to those found in plexin and hepatocyte growth factor families; these domains function in cellular adhesion and proliferation [23]. Pkhd1 is also extensively alternatively spliced, and spliced differently in different cell types, but a role for alternately spliced products has not been demonstrated [29, 30]. The human and mouse protein sequences share 73% identity [21, 27].

During embryogenesis, fibrocystin is widely expressed in epithelial derivatives, including neural tubules, gut, pulmonary bronchi, renal and hepatic cells. By embryonic day 15 fibrocystin is strongly expressed in bile duct cholangiocyte cilia, and in the epithelial cells in the kidney collecting tubes [29, 31-33]. Monoclonal antibodies also detect fibrocystin in human pancreatic duct, islets, testis and adrenal gland [34].

Although renal involvement is the most prevalent manifestation of ARPKD, NOD.Abd3 mice have normal renal structure throughout life. Consistent with this, three prior mouse models with engineered deletions of Pkhd1 have no renal abnormalities. Mice with targeted deletion of Pkhd1 exons 3, 4 or 40 had no kidney abnormality, but demonstrated intrahepatic bile duct proliferation with progressive cyst formation and associated periportal fibrosis [22, 23, 30]. In addition,

Pkhd1ex4del mice develop splenomegaly which is also apparent in our model. However, none of the previous models involved an autoimmune response nor did they require, as far as we know,

Pkhd1 expression in hematopoetic cells. Disease expression in other models was also apparently unaffected by the genetic background. Hence our model is unique; we show limited histological disease (and no clinical disease) when the Pkhd1 mutant is expressed on a predominant B6 background. At this point, we cannot fully explain our findings, since we have been unable to verify Pkhd1 expression in the immune cells we have tested. It is possible that the upregulated

Pkhd1 exons 1-35 plays a role in immune cells; another possibility is that our mutant Pkhd1 affects alternate splicing in some immune cell subset and that we need to use the correct primers to detect this. Pkhd1 expression in the thymus could affect the immune repertoire. Finally, we cannot exclude that the linked Abd3 region plays some role in the pathogenesis of these mice.

90

Although we have not shown any differentially expressed genes from this region, we will only be able to formally disprove a role of the 1.0 Mb Abd3 region by producing a recombinant between

Abd3 and Pkhd1 and testing the effect of pure mutant Pkhd1 expression (on the NOD background) on disease.

Bone marrow and splenocyte transfer studies show that the Abd3 allele / Pkhd1 must be expressed in both immune system and target tissue to get the disease. However, we do not know how Pkhd1 affects immune cells nor what immune cells are affected, as we haven’t detected any transcription of Pkhd1 using realtime PCR primers for exon15/16, exon 35/36 and exon 60/61 in adaptive immune cells including CD4 and CD8 T cells in either spleen or liver of the mice.

Nevertheless, the results of the bone marrow transfer studies clearly demonstrate that NOD.Abd3 immune cells are pathogenic, while NOD immune cells are not. NOD T cells are unable to cause disease even when raised in an Abd3+ host with abnormal cholangiocytes. Taken together with a similar result in our transfer studies [20], this suggests that the Pkhd1 mutation must be expressed both in the target tissue and in the hematopoietic tissue. However, since the BM transfers involved adult recipients, an alternate explanation is that the hematopoietic system needs to be exposed to the aberrant cholangiocytes much earlier in development. The requirement of the

NOD genetic background (which is highly disposed to autoimmunity) for disease manifestation, in contrast, supports the idea that the immune system perpetuates and enhances the disease process in NOD.Abd3 mice. Previous knockout models of Pkhd1 used B6, 129S/SvEv or mixed genetic backgrounds: Pkhd1del2 [24], Pkhd1del3-4 [35], Pkhd1del4 [22], Pkhd1del40 [23], Pkhd1lacZ

[25]. Among these rodent models, only Pkhd1del3-4 develops any intrahepatic inflammation/ cholangitis. Future studies will be directed to discover the critical immune subsets and the mechanism of Pkhd1 mutation on the immune system.

The mechanism by which Pkhd1 expression is altered in our model appears unique, with increased expression of exons 1-35 and suppressed expression of exons 36-67. Pkhd1del3-4 and

Pkhd1lacZ [25] mice appear to be completely null alleles. In addition, Pkhd1 transcripts of various

91 size at normal physiological condition are largely due to exon skipping, usage of alternative splicing sites and inclusion of novel exons [36]. In contrast, the transcription of Pkhd1 in

NOD.Abd3 CBD provides clear evidence for an alternative polyadenylation event with a novel

3’UTR-like region in IVS35 as reflected by RNAseq (Fig 7A), resulting in a fully-processed transcript of exons 1-35. Our qRTPCR result suggests that the alternative polyadenylation (APA) site within IVS35 predominates over the usual Pkhd1 polyadenylation site (found after exon67).

One of the mechanisms of using an intronic pre-mature APA site includes the loss of inhibition of such APA site mediated by U1snRNP [37]. Such protection of pre-mRNA from premature cleavage by U1 is independent of its role in splicing [38]. The 6kb inserted DNA fragment in the

NOD.Abd3 Pkhd1 intron 35 is likely derived via a transposition event (from the intronic region of the E3 ubiquitin ligase precursor on chromosome 11) which disrupts the transcription machinery and exposes the intronic APA sites, allowing production of a premature transcript of Pkhd1 in

NOD.Abd3 CBD. Minimal transcription up to exon 61 can still be detected in NOD.Abd3 which indicates the existence of the full length transcript and therefore possibly minimal wild type protein product, however severe hepatobiliary disease is 100% penetrant in these mice. This suggests there must be a threshold of full length protein amount in the hepatobiliary system for normal CBD development [27]. The upregulated expression of the first 35 exons in NOD.Abd3 could be explained by the shortage of the full length transcript, which could reinforce a forward feedback loop on the Pkhd1 de novo transcription. In addition, under normal conditions, PKHD1 in human encodes for a complex and extensive array of splice variants, most of which are thought to be fully processed and have a Poly-A tail [27], and the protein products are predicted to vary in size, and whether they are membrane bound or soluble. A critical amount of full length protein is probably essential to be further processed as a predominant protein with different sizes and in a tissue/cell specific manner, as observed in PCK rats [33]. The PCK rat model of ARPKD is also a spontaneous germline mutant with a splicing mutation at intron 35 of Pkhd1 (IVS35-2A →T, the splice acceptor site “AG” is mutated to “TG”) resulting a shortened transcript skipping exon36,

92 but affecting both renal and hepatobiliary system. The large size and multiple exons of Pkhd1 produce a complicated set of transcript variants in both normal and diseased conditions, making it a challenge to understand the mechanisms of its transcriptional and post-transcriptional regulations. As the first mouse model with a natural history of deficient Pkhd1, NOD.Abd3 clearly has different molecular mechanism of transcription regulation of Pkhd1 from the PCK rat and the other rodent models.

In summary, we have identified a novel mutation in Pkhd1 as the key genetic factor causing the biliary tissue abnormality and activating the adaptive immune system in the context of NOD genetic background. Although we have clearly demonstrated that both the target tissue

(cholangiocytes) and the immune system must express the Pkhd1 mutation, we do not yet understand how these systems interact. Massive abnormalities involving hundreds of both tissue specific (cholangiocyte specific) and immune origin genes are affected by 2 weeks of age, indicating a complex, orchestrated event producing this organ specific autoimmune disease. Our mouse model of spontaneously mutated Pkhd1 is thus valuable for exploring both polycystic hepatobiliary disease and autoimmune biliary disease.

93

Figure Legends

Figure 1. Genetic map of the NOD Abd3 region defining the Abd3 locus

A. All strains that developed ABD share a previously undetected B6 region on chromosome 1

(bottom, “Abd3”). Congenic mapping produced a strain (strain 7825) that expressed only Abd3 on the NOD background, but still developed ABD. B. The Abd3 interval on chromosome one was further narrowed to a 1 Mb region according to the presence or absence of ABD in multiple congenic strains. This locus contains three genes and two microRNAs based on NCBI mouse genome assembly build 38.1. C. Genotyping result of chromosome one of line 8706 was confirmed by 5K mouse SNP chip (Affymetrix). The schematic representation of the Abd3 locus is shown at the bottom of the panel. Gray bar indicates B6-derived interval and white bar indicates NOD genome. SNP genotyping results are displayed in the middle of the panel, showing the number of SNPs between B6 and NOD across the Abd3 locus. The corresponding genes within the Abd3 locus shown on the top of the panel are lined up with the 1 Mb interval.

Figure 2. Abd3 homozygosity on the NOD genetic background is necessary for development of autoimmune biliary disease.

A. No significant liver histology abnormalities in NOD.Abd3 backcrossed to B6.Pl or NOD

(NOD: n= 10; NOD.Abd3: n = 10; NOD.Abd3 × B6PL: n = 16; NOD.Abd3× NOD: n = 10). B,C.

Clinical scores (see methods) of NOD.Abd3 (n = 24), homozygous F2 (n = 17), and homozygous

BC2i mice (n=10) ). D,E histological scores of NOD (n = 10), NOD.Abd3 (n=10), F2 and BC2i with 0, 1, or 2 copies of Abd3 (F2 with 0 copy of Abd3: n = 14; F2 with 1 copy of Abd3: n = 25;

F2 with 2 copies of Abd3: n = 17; BC2i with 0 copy of Abd3: n = 9; BC2i with 1 copy of Abd3: n

= 18; BC2i with 2 copies of Abd3: n = 10). All mice aged 120d. Mann-Whitney U test

94 or Wilcoxon signed rank test was performed for statistical significance. **P < 0.01; ***P < 0.001.

Error bars displayed as SEM.

Figure 3. Bone marrow chimera studies show that Abd3 expression is necessary in both the recipient (non-hematopoetic tissue) and in the donor (bone marrow) in order to recapitulate disease

Clinical score (A) and histological (B) scores of: NOD.Abd3 at 10 weeks (n = 7); NOD.Abd3 recipients of either NOD.Abd3 (n = 3, recipients were 8 to 10 weeks old at the time of BMT) or

NOD bone marrow (n = 4, recipients were 9 to 10 weeks old at the time of BMT), and NOD recipients of NOD.Abd3 bone marrow (n = 6, recipients were 8 to 14 weeks old at the time of

BMT). **: p<0.01, by Mann-Whitney U test or wilcoxon signed rank tests. Error bars represent

SEM.

Figure 4. Early development of abnormal biliary system in NOD.Abd3 mice

A-C. 2-week-old NOD.Abd3 mice show substantial dilation of common bile duct (A) compared to age-matched NOD control mice (B) and NOD.c3c4-scid mice (C) . Black arrow indicates peribiliary glands; green arrow indicates hyperplasia of biliary epithelium, consequently merging with peribiliary glands. Histology figures are representative 6 CBD samples of NOD.Abd3 , 6

CBD samples of NOD and 5 NOD.c3c4.Scid mice at 2 weeks.

Figure 5. RNAseq of 2 week old NOD.Abd3 common bile duct demonstrates upregulation of cholangiocyte damage response genes and down-regulation of immune-related genes.

95

A. Heatmap of 255 genes with at least two fold-change (see methods) and significantly different expression in NOD.Abd3 compared to NOD CBD. B. Significantly differentially expressed genes from A were used to build cluster enrichment network. Red hexagon represents genes; squares represents biological process (cyan), molecular function (sky blue), mouse phenotype (light blue), pathways (green) and transcription binding site (purple); triangle represents miR-30a. The left cluster represents the 2x upregulated genes and is highly enriched in cholangiocyte damage response genes; the right cluster represents 2X downregulated genes and is highly enriched in immune-related genes.

Figure 6 Abnormal exon expression pattern of Pkhd1 in NOD.Abd3 CBD

A. RNAseq reads aligned to reference C57BL/6 genome (mm9) show that expression of exons 36 to 67 of Pkhd1 is absent in the CBD of 2wk NOD.Abd3 (upper two tracks) compared to NOD

(lower two tracks). Each track represents one sample. Black arrow indicates transcription direction. B. NOD.Abd3 samples upregulated expression of exons 1-35. Pkhd1 partition read density shown for two NOD.Abd3 samples (red and blue dots) and two NOD samples (grey and brown dots). C. PCR using genomic DNA shows the presence of representative exons 16, 34, 40 and 61 in NOD, B6 and NOD.Abd3 at DNA level. D. qRT-PCR confirms that NOD.Abd3 CBD upregulates exon 15/16 expression and lacks exon 60/61, 35/36 expression compared to NOD or

B6. (N = 3/group, *: P<0.05 by student T test).

Figure 7 RNAseq data reveals early termination of Pkhd1 transcription in NOD.Abd3

A. Genomic organization of Pkhd1 and Abd3 alleles. Blue and red boxes indicate Pkhd1 and

Abd3 allele, respectively. The Pkhd1 gene and the Abd3 region are ~2.5 Mb apart. Note: Pkhd1 is

96 on the reverse strand, so the Abd3 allele is upstream of Pkhd1. B. Magnified view of read- alignment from 2wk CBD RNAseq data in the intron between exons 35 and 36 of Pkhd1 between

NOD and NOD.Abd3 samples shows unique expression in NOD.Abd3. Red, green and blue dashed lines indicate the position of exon35, 36 and the predicted pseudogene Gm15795, respectively across all four RNAseq tracks.

Figure 8 Mutation characterization of Pkhd1 in NOD.Abd3 and potential mechanism of deficient NOD.Abd3 Pkhd1 exon expression pattern

A .Schematic showing strategy of PCR primer design in intron 35 of Pkhd1. B. Primer 1 and 2 failed to amplify using NOD.Abd3 genomic DNA while primer 3 successfully amplifies using both NOD.Abd3 and NOD genomic DNA. C. Primers N1, N2, N3, N4, N5, N6 and N7 were all able to amplify the PCR product with expected product size using NOD.Abd3 and NOD genomic

DNA. D. Long-range PCR using primer N7 forward primer and primer 3 reverse primer showed distinct product size on NOD.Abd3 and NOD genomic DNA. E. Schematic representation of assembled long-range PCR product sequence in NOD and NOD.Abd3 with relative position of the primer pair. Three nodes of de novo assembly were generated for NOD.Abd3 with two small gaps present in between node-1/node-2 and node-2/node-3. Part of node-1 and node-3 can be matched with NOD assembled sequence, leaving the rest of the assemblies about 6 kb in middle uncovered. Genomic positions of the primers and the sequences are based on C57Bl/6

NCBI37/mm9 assembly. F. Schematic representation of potential mechanism of truncated Pkhd1 transcript in NOD.Abd3 CBD. APA: alternative polyadenylation site.

97

Table One. PCR result summary of various primer pairs

Primer set or Starting Ending Amplification of Amplification of exon position position* position* NOD.Abd3 NOD

Exon 36 20,440,302 20,440,458 NA NA

Primer N1 20,440,373 20,440,692 Yes Yes

Primer N2 20,445,607 20,445,925 Yes Yes

Primer N3 20,450,397 20,450,733 Yes Yes

Primer N4 20,455,542 20,455,846 Yes Yes

Primer N5 20,460,303 20,460,626 Yes Yes

Primer N6 20,465,591 20,465,989 Yes Yes

Primer N7 20,470,616 20,470,976 Yes Yes

Primer 1 20,472,908 20,473,208 No Yes

Primer 2 20,473,134 20,473,447 No Yes

Primer 3 20,473,238 20,473,540 Yes Yes

Exon 35 20,493,023 20,493,173 NA NA

* Genomic coordinates: chr1; genomic assembly: NCBI37/mm9 Yes: single band with expected PCR product size; NO: no band shown on the gel; NA: no PCR was performed.

98

Figure 1

A B

Mb Chromosome 3 The Abd3 region 20 D3M it93 D3M it167 Chr 1 Line Line Line Line Line Chr 1 Line 30 Mb 7825 8728 8706 8727 8730 Mb 8706 22 22.7 40 22.8 RS13475762 50 23 4933415F23Rik 22.9

4933415F23Rik 60 RS3659806 O gfrl1 23.0 B3gat2 RS37589619 Sm ap1 23.1 RS31654465 70 RS31505493 24 1110058L19Rik 4921533L14Rik 23.2 Col9a1 miR-30a RS31814523 Col19a1 23.3 miR-30c-2 80 m t-Atp6 Lm brd1 Abd3 RS31334391 O gfrl1 1.00 Mb 23.4 90 25 D3M it41 23.5 D3M it213 Idd10 D1m it211 Bai3 100 23.6 Idd18 Abd2 26 23.7 110 rs38124092 D3M it370 23.8 B3gat2 120 Sm ap1 23.9 Mb Chromosome 4 __ 24.0 70 ABD disease status +++

80 D4Dil30 D4Dil31 Abd1 (Interferon 90 D4M it144 region) D4M it166 C 100 Line 8706 congenic region on Chromosome 1

chr1:22852173..23843298 110 22900k 23000k 23100k 23200k 23300k 23400k 23500k 23600k 23700k 23800k 120 Ensembl Predictions 4933415F23Rik SNORA17 Ogfrl1 B3gat2 130 Gm6420 140 mmu-mir-30a mmu-mir-30c-2 80 150 D4M it42 NOD vs B6 NGS_SNP_Graph D4M it59 160 40

170 SNPs per 10 Kb SNPs Mb Chromosome 1 0 20 rs13475762 Line 8706 rs3659806 25 Abd3 30 rs6237824 rs4137502 Region between in and out marker No SNPs between 35 NOD. NOD. NOD. in and out marker c3c4 ABD Abd3 (strain 7825)

99 Figure 2

NOD A NOD.Abd3 NOD.Abd3 x B6PL 4 NOD.Abd3 x NOD

3

2

1

*** *** Liver histologoy scores histologoy Liver 0 Duct epithelial Leukocytic hyperplasia infiltration B C

5 5 *** *** 4 4

3 *** 3 ***

2 2 Clinical score Clinical 1 score Clinical 1

0 0 F2 NOD.Abd3 F2 NOD.Abd3 BC2i NOD.Abd3 BC2i NOD.Abd3 CBD CBD liver liver CBD CBD liver liver

NOD NOD D F2 - 0 copy of Abd3 E BC2i - 0 copy of Abd3 F2 - 1 copy of Abd3 BC2i - 1 copy of Abd3 F2 - 2 copies of Abd3 BC2i - 2 copies of Abd3 4 NOD.Abd3 4 NOD.Abd3

3 3 *** ** *** 2 2

1 1

Liver histologoy scores histologoy Liver Liver histologoy scores histologoy Liver 0 0 Duct epithelial Leukocytic Duct epithelial Leukocytic hyperplasia infiltration hyperplasia infiltration

100 Figure 3

A * 4 CBD liver 3

2

Clinical score 1

0

Time post BMT N.A. 2.5~4 mo 4 mo

10 wk Abd3

Abd3 BM -> 8~10NOD wk Abd3 BM -> 9~10Abd3 wk Abd3 BM -> 8~14 wk NOD

** B * * 4 ductal epithelial hyperplasia leukocytic infiltration 3

2

1

Liver histology score 0

Time post BMT N.A. 2.5~4 mo 4 mo

10 wk Abd3

Abd3 BM -> 8~10NOD wk Abd3 BM -> 9~10Abd3 wk Abd3 BM -> 8~14 wk NOD

101 Figure 4

NOD.Abd3 NOD NOD.c3c4-Scid

Extrahepatic bileA duct B C

102 Figure 5

A B

abnormal inflammatory response Biosynthesis of amino acids

inflammatory response secretion by cell S-adenosylmethionine metabolic process abnormal neutrophil morphology lipid digestion abnormal gland physiology

increased inflammatory response abnormal intestinal lipid absorption acute inflammatory response abnormal digestion Complement and coagulation cascades abnormal hepatobiliary system physiology retinal dehydrogenase activity enzyme inhibitor activity miR-30 response to wounding Gene regulation logic in retinal ganglion cell developm ... Biological oxidations serine-type endopeptidase activity abnormal pH regulation SLC9A2 CBS HSDL2 Complement and Coagulation Cascades EEF1A2 NGFR FMO2 PNLIP UGT2B10 UPK1B VIP Integrin cell surface interactions CA4 SULT1C2 CYP2C19 PNLIPRP2 SNCG GAP43 ENPEP GSTA3 SLC2A2 MID1 HPX ATP6V0A4 SULT1A3 PAQR5 CERS3 VCAM1 RTN1 PYCR1 REG1B GAL3ST1 GSTA4 UCHL1 FRZB MAT1A ALDOC STMN2 SCNN1A AKR1B10 ALDH1A1 ASS1 GAMT SYCN V$ER_Q6_01 HSD11B1 CYP2S1 ANXA13 FKBP11 REG1A CD37 ADAMTSL3 FKBP5 APOA2 HTR3A CPN1 ASNS KRT7 SPDEF SERPINC1 V$LHX3_01 ANXA8 GNMT FGB CARTPT TSPAN15 steroid binding AGR2 APOC1 WGGAATGY_V$TEF1_Q6IL17RE 2X_Upregulated_ABD A4GNT C3 BEX2 SYT1 GCNT3 DPEP1 SCG2 PTGER3 TF ERO1LB lipid binding HPN SLC26A3 KLF5 GC PLA2G1B HABP2 ANG SLC26A9 CDO1 SYT7 TGANTCA_V$AP1_C ADAP1 CTRL ABCC3 ELF3 abnormal stomach epithelium morphology AMBP F5 HP SCG5 NOXO1 CELA3B CREB3L4 YTAATTAA_V$LHX3_01 MAP3K6 ALDH1A2 FZD9 S100A9 ALB CFI LCP1 LCN2 Wnt antagonist SFRP3 inhibits the differentiation of mo ... CUZD1 FOXA3 RAPGEFL1 PLAUR CORO1A AHSG RAC2 NCF4 FGG ETV4 REG3G FGA 2.5 CRIP1 SFN FZD5 SERPINH1 CXCL17 SDC1 KRT19 XDH RSPO3 CXCR4 NPYAQP5 MSLN FLRT3 CRLF1 ADAM23 NNAT F2RL1 EVPL CA9 F3 SEMA3E FGF13 PTCH1 HHIP IHH MGP TESC TFCP2L1 LGALS4 2X_downregulated_ABD GJB3 SERPINB5 EMR1 1.5 LAMB3 ANGPTL1 GPNMB MIA CLDN6 HMP19 TACSTD2 abnormal intestinal epithelium morphology WNT7A SMOC2 ETV5 SNAP25 CLDN23 WNT7B epithelial cell proliferation SERPINI1 V$HNF3_Q6 VTN V$PAX2_02 V$MYOD_01 epithelium development cell adhesion Hedgehog Signaling Pathway V$STAT5B_01 1.0 morphogenesis of a branching epithelium V$ER_Q6 gland development V$STAT_01

hedgehog family protein binding NOD NOD biological adhesion CLDN2 Expression morphogenesis of an epithelium TGCCAAR_V$NF1_Q6 Class B/2 (Secretin family receptors) Leukocyte transendothelial migration Hedgehog signaling pathway V$STAT5A_01 abnormal monocyte morphology

0.7

Abd3 NOD. Abd3 NOD.

0.5

103 Figure 6

A

NOD.Abd3_S1

NOD.Abd3_S2

NOD_S1

NOD_S2

transcription direction

B Pkhd1 transcript exon 36 exon 35 exon 67 exon 1

C D NOD.Abd3 2wk CBD 2.5 NOD 2wk CBD * B6 2wk CBD

2.0

NOD

B6

Abd3 NOD. NTC 1.5 Exon 16 1.0 Exon 34

Relative expression Exon 40 (normalized to Rn18S) 0.5 * * Exon 61 0 Ex15/16 Ex60/61 Ex35/36

104 Figure 7

A

62,529 bp B 20431.68K 20437.93K 20444.18K 20450.44K 20456.69K Pkhd1 transcript (UCSC transcripts)

exon 36 exon 35

Gm15795

intron 35_3’UTR like NOD.Abd3 S1

intron 35_3’UTR like NOD.Abd3 S1

NOD S1

NOD S2

Cytoband

qA2 qA3 qA4 qA5 qB qC1.1 qC1.3 qC2 qC3 qC4 qC5 qD qE1.1 qE2.1 qE2.3 qE3 qE4 qF qG1 qG3 qH1 qH2.1qH2.3 qH3 qH4 qH5 qH6

105 Figure 8 A

B D NOD.Abd3 NOD NOD.Abd3 NOD 10kb 8kb 1 2 3 1 2 3 3kb

500bp 500bp 400bp 400bp 300bp 300bp 200bp 200bp 100bp 100bp

C N1F/N1R N3F/N3R N4F/N4R N5F/N5R N6F/N6R N7F/N7R N2F/N2R

600bp

500bp 500bp

NO DNA template DNA NO

NOD Abd3 NOD. Abd3 NOD.

NO DNA template DNA NO

NOD

Abd3 NOD.

NOD

NO DNA template DNA NO

NOD

NO DNA template DNA NO Abd3 NOD. NOD Abd3 NOD. Abd3 NOD.

NO DNA template DNA NO NOD template DNA NO template DNA NO Abd3 NOD. 400bp 400bp NOD 300bp 300bp 200bp 200bp 100bp 100bp

E

Primer position on Chr1 20,473,540 (+6)- 20,470,616 (-5)

Position on Chr1 20,473,534 20,470,621 2914 Positoin on assembled 1 73 2540 2548 NOD sequence 2841

gap1 gap2 Position on assembled 1 2476 2617 1 5157 1 316 617 NOD.Abd3 sequence Node-1 Node-2 Node-3

Sequence not matched with assembled NOD sequence

F

Splicing U1 snRNP Splicing APA NOD ex36 ex35

Splicing U1 snRNP Splicing APA NOD.Abd3 ex36 ex35

106

References

1. Kaplan, M.M. and M.E. Gershwin, Primary biliary cirrhosis. N Engl J Med, 2005. 353(12): p. 1261-73. 2. Invernizzi, P., C. Selmi, and M.E. Gershwin, Update on primary biliary cirrhosis. Digestive and Liver Disease, 2010. 42(6): p. 401-408. 3. Eaton, J.E., et al., Pathogenesis of primary sclerosing cholangitis and advances in diagnosis and management. Gastroenterology, 2013. 145(3): p. 521-36. 4. Mack, C.L., The Pathogenesis of Biliary Atresia: Evidence for a Virus-Induced Autoimmune Disease. Semin Liver Dis, 2007. 27(03): p. 233-242. 5. Brindley, S.M., et al., Cytomegalovirus-specific T-cell reactivity in biliary atresia at the time of diagnosis is associated with deficits in regulatory T cells. Hepatology, 2012. 55(4): p. 1130-1138. 6. Rong, G., et al., Epithelial cell specificity and apotope recognition by serum autoantibodies in primary biliary cirrhosis. Hepatology, 2011. 54(1): p. 196-203. 7. Shivakumar, P., et al., Effector Role of Neonatal Hepatic CD8+ Lymphocytes in Epithelial Injury and Autoimmunity in Experimental Biliary Atresia. Gastroenterology, 2007. 133(1): p. 268-277. 8. Irie, J., et al., NOD.c3c4 congenic mice develop autoimmune biliary disease that serologically and pathogenetically models human primary biliary cirrhosis. The Journal of Experimental Medicine, 2006. 203(5): p. 1209-1219. 9. Koarada, S., et al., Genetic Control of Autoimmunity: Protection from Diabetes, but Spontaneous Autoimmune Biliary Disease in a Nonobese Diabetic Congenic Strain. The Journal of Immunology, 2004. 173(4): p. 2315-2323. 10. Yang, G.X., et al., CD8 T cells mediate direct biliary ductule damage in nonobese diabetic autoimmune biliary disease. J Immunol, 2011. 186(2): p. 1259-67. 11. Oertelt, S., et al., Anti-Mitochondrial Antibodies and Primary Biliary Cirrhosis in TGF-β Receptor II Dominant-Negative Mice. The Journal of Immunology, 2006. 177(3): p. 1655-1660. 12. Shivakumar, P., et al., Obstruction of extrahepatic bile ducts by lymphocytes is regulated by IFN-γ in experimental biliary atresia. The Journal of Clinical Investigation, 2004. 114(3): p. 322-329. 13. Langmead, B., et al., Ultrafast and memory-efficient alignment of short DNA sequences to the human genome. Genome Biology, 2009. 10(3): p. 1-10. 14. Kim, D., et al., TopHat2: accurate alignment of transcriptomes in the presence of insertions, deletions and gene fusions. Genome Biology, 2013. 14(4): p. R36. 15. Trapnell, C., et al., Differential gene and transcript expression analysis of RNA-seq experiments with TopHat and Cufflinks. Nat. Protocols, 2012. 7(3): p. 562-578. 16. Chen, J., et al., Improved human disease candidate gene prioritization using mouse phenotype. BMC Bioinformatics, 2007. 8(1): p. 392. 17. Kaimal, V., et al., ToppCluster: a multiple gene list feature analyzer for comparative enrichment clustering and network-based dissection of biological systems. Nucleic Acids Research, 2010. 38(Web Server issue): p. W96-W102.

107

18. Gervais, A.L., M. Marques, and L. Gaudreau, PCRTiler: automated design of tiled and specific PCR primer pairs. Nucleic Acids Research, 2010. 38(suppl 2): p. W308-W312. 19. Bouma, G., et al., Evidence for an enhanced adhesion of DC to fibronectin and a role of CCL19 and CCL21 in the accumulation of DC around the pre-diabetic islets in NOD mice. European Journal of Immunology, 2005. 35(8): p. 2386-2396. 20. Wu, S.J., et al., Innate immunity and primary biliary cirrhosis: activated invariant natural killer T cells exacerbate murine autoimmune cholangitis and fibrosis. Hepatology, 2011. 53(3): p. 915-25. 21. Ward, C.J., et al., The gene mutated in autosomal recessive polycystic kidney disease encodes a large, receptor-like protein. Nat Genet, 2002. 30(3): p. 259-269. 22. Gallagher, A.-R., et al., Biliary and Pancreatic Dysgenesis in Mice Harboring a Mutation in Pkhd1. The American Journal of Pathology, 2008. 172(2): p. 417-429. 23. Moser, M., et al., A mouse model for cystic biliary dysgenesis in autosomal recessive polycystic kidney disease (ARPKD). Hepatology, 2005. 41(5): p. 1113-1121. 24. Woollard, J.R., et al., A mouse model of autosomal recessive polycystic kidney disease with biliary duct and proximal tubule dilatation. Kidney Int, 2007. 72(3): p. 328-336. 25. Williams, S., et al., Kidney cysts, pancreatic cysts, and biliary disease in a mouse model of autosomal recessive polycystic kidney disease. Pediatric Nephrology, 2008. 23(5): p. 733-741. 26. Flicek, P., et al., Ensembl 2014. Nucleic Acids Research, 2014. 42(D1): p. D749-D755. 27. Onuchic, L.F., et al., PKHD1, the Polycystic Kidney and Hepatic Disease 1 Gene, Encodes a Novel Large Protein Containing Multiple Immunoglobulin-Like Plexin- Transcription–Factor Domains and Parallel Beta-Helix 1 Repeats. The American Journal of Human Genetics, 2002. 70(5): p. 1305-1317. 28. Xiong, H., et al., A Novel Gene Encoding a TIG Multiple Domain Protein Is a Positional Candidate for Autosomal Recessive Polycystic Kidney Disease. Genomics, 2002. 80(1): p. 96-104. 29. Nagasawa, Y., et al., Identification and characterization of Pkhd1, the mouse orthologue of the human ARPKD gene. J Am Soc Nephrol, 2002. 13(9): p. 2246-58. 30. Bakeberg, J.L., et al., Epitope-tagged Pkhd1 tracks the processing, secretion, and localization of fibrocystin. J Am Soc Nephrol, 2011. 22(12): p. 2266-77. 31. Masyuk, A.I., et al., Biliary exosomes influence cholangiocyte regulatory mechanisms and proliferation through interaction with primary cilia. Am J Physiol Gastrointest Liver Physiol, 2010. 299(4): p. G990-9. 32. Zhang, M.-Z., et al., PKHD1 protein encoded by the gene for autosomal recessive polycystic kidney disease associates with basal bodies and primary cilia in renal epithelial cells. Proceedings of the National Academy of Sciences of the United States of America, 2004. 101(8): p. 2311-2316. 33. Masyuk, T.V., et al., Defects in cholangiocyte fibrocystin expression and ciliary structure in the PCK rat1. Gastroenterology, 2003. 125(5): p. 1303-1310. 34. Ward, C.J., et al., Cellular and subcellular localization of the ARPKD protein; fibrocystin is expressed on primary cilia. Human Molecular Genetics, 2003. 12(20): p. 2703-2710.

108

35. Garcia-Gonzalez, M.A., et al., Genetic interaction studies link autosomal dominant and recessive polycystic kidney disease in a common pathway. Human Molecular Genetics, 2007. 16(16): p. 1940-1950. 36. Boddu, R., et al., Intragenic motifs regulate the transcriptional complexity of Pkhd1/PKHD1. Journal of Molecular Medicine, 2014. 92(10): p. 1045-1056. 37. Elkon, R., A.P. Ugalde, and R. Agami, Alternative cleavage and polyadenylation: extent, regulation and function. Nat Rev Genet, 2013. 14(7): p. 496-506. 38. Kaida, D., et al., U1 snRNP protects pre-mRNAs from premature cleavage and polyadenylation. Nature, 2010. 468(7324): p. 664-668.

109

Chapter 4 Summary, discussion and future directions

As two spontaneous mouse models of PBC, dnTGFβRII and NOD.Abd3 mice gave us new insights into disease pathology in this dissertation, but in different ways. In our dnTGFβRII studies, we focused on the immune manifestations by which specific CD8 subpopulations contribute to the autoimmune biliary disease, and the interaction of effector and regulatory compartments in adaptive immune system. In the NOD.Abd3 studies, we discovered the key gene controlling the defect in target tissue and a potential molecular mechanism mediating the defective expression of this gene.

Summary, discussion and future directions of dnTGFβRII model

In dnTGFβRII mice, the overexpression of the dominant negative form of type II receptor of

TGFβ is restricted to CD4 and CD8 cells, therefore the adaptive immune system is clearly the major source contributing to the profound disease effect in terms of autoimmune cholangiopathy.

To determine whether the pathogenic CD8 T cells mediated disease intrinsically or extrinsically, we developed bone marrow chimeras (BMC). Firstly, and surprisingly, mixed BMC (mBMC) constructed from dnTGFβRII and B6 donor BM demonstrated significantly less disease compared to single dnTGFβRII BMC. We reported the novel observation that dnTGFβRII mice accumulate large numbers of terminally differentiated hepatic KLRG1+ CD8 T cells. To understand the mechanism of B6 mediated protection, we performed adoptive transfer using dnTGFβRII CD8 effector T cells and B6 vs dnTGFβRII T regulatory cells. Co-transfer of B6 Tregs prevents dnTGFβRII CD8 T cell mediated cholangitis, consistent with the mBMC results, and disease protection was associated with significantly decreased numbers of KLRG1+ CD8 T cells.

However, dnTGFβRII Tregs were unable to decrease cholangitis. We next demonstrated that dnTGFβRII Treg cells were functionally defective in suppressing effector CD8 T cells compared to wild type B6 Tregs, emphasizing that autoimmune cholangitis requires defects in both T effector and regulatory compartments.

110

Working model of dnTGFβRII

Based on the results on CD8 T cells and Tregs in dnTGFβRII mice, we argue that the defects introduced by the transgene affect both CD8 T effector cells and Tregs so that CD8 T effector cells keep proliferating and differentiating into functional terminal effector cells (Figure 1), which is the major source of cholangiocyte cytotoxicity. On the other hand, defective Tregs in dnTGFβRII are not able to restrain such proliferation and effector function of CD8 T effector cells. Moreover, when put in a normal regulatory immune system, dnTGFβRII CD8 T effector cells are not sufficient to cause autoimmune biliary disease. Therefore, we propose a “two hit” model in the dnTGFβRII mice: the defects in both effector and regulatory pathways are necessary for the autoimmune biliary disease.

TGFβ-1 TGFβ-1 TGFβRI dnTGFβRII TGFβRI dnTGFβRII

CD8 Treg

CD8 ET/EMT

cytotoxicity Terminal effector CD8

Figure 1 Working model of dnTGFβRII

111

Future directions of dnTGFβRII model

Based on our above model, we hypothesize that KLRG1+ CD8 cells cause biliary autoimmunity but only if Tregs are defective. Therefore we will define the KLRG1+ CD8 T cell phenotype and its ability to transfer disease.

Antigen specific CD8 T cells undergo clonal expansion, development of effector cells, contraction, and eventual establishment of memory cells. CD8 T cells that become memory cells are called memory precursor effector cells (MPECs), whereas CD8 T cells that die during contraction are short-lived effector cells (SLECs) [1]. MPECs are KLRG1- CD127+; SLECs are

KLRG1+CD127-. Both populations are similar functionally, but different in memory potential and survivability [2-4]. The KLRG1+ CD8 T subset has many antigen specific cells in during anti- viral responses; but when the antigen is cleared the KLRG1+ subset declines [5]. Extrinsic signals such as cytokines are required for control of CD8 effector expansion and memory. TGFβ- mediated apoptosis of this KLRG1+ CD8 T effector cells occurs during clonal expansion and contraction [6], therefore insufficient TGFβ signaling may contribute to excess KLRG1+ accumulation in the dnTGFβRII mice. IL-12 promotes SLEC differentiation by increasing expression of the transcription factor T-bet [7]. Thus, an IL-12 inflammatory response is a regulator of SLEC formation [3, 8].

The role of KLRG1+ CD8 cells in autoimmunity has not been studied in biliary autoimmunity.

IFNγ mediated accumulation of KLRG1+ CD8 SLEC-like cells was found in the Roquinsan/san mouse, which develops lupus [9]. We have also observed specific accumulation of SLEC in the liver of dnTGFβRII mice (Chapter 3). We will define the KLRG1+ CD8 T cell surface phenotype and determine whether such cells are transient or long lived. We will show whether KLRG1+

CD8 T cells can transfer cholangitis into Rag1–/– recipients. These studies will provide novel information not only for understanding pathogenic CD8 T cell effector mechanisms, but also for potential therapeutic intervention.

112

Do dnTGFβRII KLRG1+ CD8 T cells cause cholangitis? Are “two hits” involving both CD8

T effector cells and Tregs necessary to develop biliary disease? Our hypothesis is that

KLRG1+ CD8 SLECs can cause biliary autoimmunity, but not in the presence of normal Tregs.

By adoptive cell transfer studies, we will demonstrate whether the dnTGFβRII KLRG1+ effector

CD8 T cells survive aberrantly (i.e. resisting contraction or exhaustion). We will also test whether KLRG1- effector CD8 T cells turn into to KLRG1+ effector cells. In addition, we will test for functional abnormalities in dnTGFβRII Tregs. Finally, we will examine the “two hit” model of biliary autoimmunity and the pathogenic role of Tregs using an inducible Treg defect in an adoptive transfer model. Approach: we will sort KLRG1+ and KLRG1– CD8 T cells (1x106) from spleen of dnTGFβRII mice and transfer into 6 week-old B6 Rag1–/– mice. Recipient liver will be harvested and scored for pathological changes 8 weeks after transfer. Cholangitis will be evaluated with H&E; intrahepatic cells will be assayed for CD8 populations as in Chapter 3. This approach will test whether dnTGFβRII KLRG1+ CD8 T cells that have already developed can induce cholangitis as in whole CD8 T cell transfer experiments [10]. Control recipients will receive KLRG1– effector CD8 T cells. These experiments will demonstrate whether KLRG1+

CD8 T cells are generated from KLRG1– CD8 memory T cells and will be correlated with the extent of liver disease. We predict that dnTGFβRII KLRG1+ CD8 T cells will accumulate predominantly in recipient liver and that KLRG1– effector cells will transform to KLRG1+ cells and induce cholangitis.

To determine the defects in dnTGFβRII Tregs which are critical for induction of cholangitis, we will quantify the function of dnTGFβRII vs. B6 Tregs by purifying GFP+Foxp3+ cells from both B6.Foxp3 GFP and dnTGFβRII Foxp3-GFP mice [11]. We will first test Treg function, including in vitro and ex vivo proliferation, expression of surface markers, and functional suppression [12, 13]. Next, using TGFβRIIfl/fl mice [14], we will test a “two hit” model of biliary autoimmunity: whether a “second hit” to normal Tregs allows induction of disease in the presence of abnormal dnTGFβRII CD8 cells. This can be achieved by crossing TGFβRIIfl/fl with

113

FoxP3EGFP-Cre-ERT2 mice [15] to produce TGFβRIIfl/fl FoxP3EGFP-Cre-ERT2 mice, that contain Tregs with inducible loss of TGFβ signaling. We will first prove that non-induced Tregs have normal suppressive function. Then we will induce the TGFβ signaling defect and assess disease and Treg function. We will use variable induction times to test when a Treg defect must occur to mediate biliary disease, as it is possible that normal Tregs at the start of the disease process could prevent pathology.

Although CD4+Foxp3+ Tregs accumulate in dnTGFβRII liver, they are defective at suppressing

CD8 T cell proliferation and effector function. One possibility is that they actually acquire pathogenic T cell effector function. Indeed, it is not uncommon for Treg to convert into pathogenic effector cells. For example, CD4+Foxp3+ cells transdifferentiate into Th17 cells and become pathogenic in autoimmune arthritis [16]. To study this possibility, we will measure the expression of pro-inflammatory cytokines such as IFNγ, TNFα, IL-17. We will also adoptively transfer dnTGFβRII Tregs alone into B6.Rag1-/-, monitor the Foxp3 expression of the transferred

Tregs and determine the disease in the recipients.

Potential limitations/alternative approaches: KLRG1+ autoimmune effector CD8 T cells may be short lived, as seen in viral immunity (i.e. if the increased numbers seen in dnTGFβRII are due to increased production, not increased survival), and thus transfer of KLRG1+ cells may cause less biliary disease compared to the control (total CD8) group. However, continual generation of

KLRG1+ effector cells, resulting from defective TGFβ signaling may be involved in biliary cell damage. It is possible that normal Tregs at the start of the transfer will prevent disease even if

Tregs are rendered defective later in the process. However, we can test this by inducing the Treg defect immediately after the adoptive transfer is initiated.

114

Summary, discussion and future directions of NOD.Abd3 model

In the study of NOD.Abd3, NOD background is critical for the ABD as shown by our breeding strategy, and Pkhd1 was discovered to cause the defect in the target tissue. It is interesting to note that two previous models with targeted deletion of Pkhd1 exons resulting in partial Pkhd1 expression also develop biliary disease without kidney involvement similar to our model [17, 18], however, neither shows apparent immune-mediated disease etiology potentially due to their non-

NOD background. We do see a dramatic disease reduction when NOD background is decreased while B6 background is increased during genetic analysis. It is very likely that if the Pkhd1 mutation found in NOD.Abd3 is on a complete B6 background, there will be no ABD, further emphasizing the importance of NOD background which is prone to autoimmune disease. Due to the lack of evidence of Pkhd1 expression in adaptive and innate immune cells, we do not know how the defective Pkhd1 causes abnormalities in the immune system. However, it is possible that the difference in immune system between NOD.Abd3 and NOD is mediated by factors yet-to-be- known located in Abd3 locus. Moreover, the BMT studies strongly suggest that there is a fundamental difference in immune system between NOD.Abd3 and NOD. Surprisingly, the ABD in NOD.Abd3 recipients of NOD bone marrow is decreased, suggesting that without Abd3 allele, the NOD immune system is unable to cause ABD. These results give us a novel hint in PBC pathology that instead of immune system alone, the interaction of both target tissue and immune system is important to the etiology of autoimmune cholangiopathy. Although current genome- wide association studies of human PBC don’t reveal any abnormality of genetic region adjacent to PKHD1, the homologous gene of the defective gene found in NOD.Abd3 or any gene encoding proteins functionally related to PKHD1, such as Polycystic kidney disease 2 (PKD2) [19], the possibility of a potential defect in the target cell cholangiocytes as PBC etiology still cannot be ruled out. To address this question, it will be beneficial to conduct an RNA-seq study using cholangiocytes from PBC patients and healthy controls. Alternatively, the finding that Pkhd1 as a causative genetic defect in target tissue of NOD.Abd3 mouse model doesn’t necessarily mean that

115

it will be responsible for human PBC disease, in which the fundamental difference of the target cells from PBC patients and controls may lie in the environmental triggers such as long-term exposure to drugs and chemicals that can modify the self-protein to produce neo-antigen and/or facilitate the leakage of autoepitope to the adaptive immune system, given the unique environment of liver.

Working model of NOD.Abd3

Defective biliary expression of Pkhd1 due to early transcription termination mediated by a 6kb intronic insertion could dysregulate cholangiocyte development at early stage, potentially leading to exposure of the major autoantigen of PBC, PDC-E2 to the surface of cholangiocyte in the absence of apparent cholangiocyte apoptosis. Either Pkhd1, by an unknown mechanism, or a separate genetic defect potentially located in B6-derived Abd3 allele in conjunction with the NOD genetic background which is biased towards autoimmunity, is responsible for the abnormality in immune system, accounting for the loss of tolerance to cholangiocytes in the middle to late stage of the disease (Figure 2). NOD.Abd3 is a model requiring interaction of target cells and immune system to develop ABD.

NOD + Pkhd1 (hypomorphic)

+ Abd3 allele

Early stage T cell Middle to late stage

Cholangiocytes

Figure 2 Working model of NOD.Abd3

116

Future directions of NOD.Abd3 mouse model

We have reported that abnormalities in both target tissue and hematopoietic system are required for disease onset in the bone marrow transfer model using NOD.Abd3 and NOD donor and recipients, and we pinpoint the genetic defect in the target tissue is within the IVS 35 of Pkhd1.

The next step is to find the defect in the immune system. cDNA of Pkhd1 is over 12 kilobase in mouse and is predicted to go through extensive alternative transcription [20, 21]. Because we haven’t detected the expression of Pkhd1 in T cells, B cells or dendritic cells from either

NOD.Abd3 or NOD mice using real-time PCR probes spanning selective sequential known exons of Pkhd1, we hypothesize that genetic defect of NOD.Abd3 immune cells compared to NOD is reflected by the difference in alternative transcription of Pkhd1 utilizing exon skipping.

Approach: to detect any potential transcripts in immune cells, we will design one set of primer every 250bp on the full length transcript which is 12,935bp. That will yield 52 sets of primers.

These primers will then be used to amplify Pkhd1 cDNA from immune cells in a step-wise manner. 250bp interval is small enough to capture any exon used in this case. This approach will allow us to understand whether there is a difference in the alternative transcription of Pkhd1 using exon skipping in immune cells from NOD.Abd3 and control mice.

Potential limitations and alternative approach: However, if the potential transcription of the gene in immune cells is completely different from what is known in the kidney or biliary cells, eg. usage of novel exons, then this approach probably won’t be able to detect the potential transcripts in immune cell. If that is the case, an RNA-seq of immune cells will be a possible solution.

Another scenario is that Pkhd1 is truly not expressed in any immune cells. This may indicate the genetic defect in NOD.Abd3 immune cells compared to NOD is not related to Pkhd1. Given that

NOD.Abd3 carries a B6-derived region on chromosome one containing two miRNAs that is associated with hepatobiliary development in zebrafish, our alternative hypothesis of the

117

genetic defect in NOD.Abd3 immune cells is that Abd3 allele is responsible for the abnormality of NOD.Abd3 immune cells.

Approach: 1. An RNA-seq will be conducted for bone marrow cells, matured CD4, CD8 T cells from 10-week-old NOD.Abd3 and NOD mice. This approach will not only detect the expression of all the known genes on immune cells from B6-derived or the corresponding NOD-derived chromosome one region from NOD genetic background, but can also uncover any potentially novel transcript on this region. 2. To separate any interacting effect between Pkhd1 and Abd3 allele which are only 2.5 Mb apart, it is essential to have NOD congenic strains with only the mutated Pkhd1 or only Abd3 allele. Due to the two loci is closely located, one approach is to setup a large number of breeding pairs and type each generation of offspring for a recombination event between the Pkhd1 locus and the Abd3 locus. An alternative approach, however, is to utilize

CRISPER/Cas9 technology [22, 23] to cut out the 6 Kb insertion in the IVS 35 of Pkhd1 from

NOD.Abd3 by injecting two guiding RNAs specific to the two ends of the inserted sequence and repair template to the zygotes of NOD.Abd3. The new strain which is still homologous for Abd3 allele will be called NOD.Abd3* carrying only the B6-derived chromosome one region (Abd3 allele) but not the mutated Pkhd1. NOD.Abd3* should not have autoimmune liver disease as seen in NOD.Abd3 because the abnormalities in target cells and immune system are potentially contributed by distinct genetic loci and the genetic defect controlling the target tissue abnormality does not exist in NOD.Abd3*. Therefore, NOD.Abd3* can be considered as an equivalent to the

NOD recipient of NOD.Abd3 bone marrow. Bone marrow transfer will be performed using

NOD.Abd3 and NOD.Abd3* as donor and lethally irradiated NOD.Abd3 as recipient. If the genetic defect in NOD.Abd3 immune cells relies entirely on Abd3 allele, then the disease outcome of the NOD.Abd3 recipients of NOD.Abd3 and NOD.Abd3* bone marrow is expected to be the same.

118

These two approaches will allow us to determine the possibility that abnormalities of target tissue and immune system in NOD.Abd3 attribute to different genetic loci and whether the potential defect for the immune system arm is on Abd3 allele.

Potential limitations and alternative approach:

If the genetic defect of NOD.Abd3 immune system is not located within Abd3 allele, RNA-seq data can at least narrow the number of the target genes for us to compare the difference between

NOD.Abd3 and NOD splenic and hepatic T cell subsets.

We report the defective expression of Pkhd1 in NOD.Abd3 bile duct and this has led us to consider how the change of such integral membrane protein will alter the biology of cholangiocytes and turn them into a “non-innocent victim” in the autoimmune cholangiopathy.

We would like to know whether NOD.Abd3 cholangiocytes expose the autoantigen PDC-E2.

Although it has been postulated that Pkhd1 is associated with cell-cell adhesion and may play a role in cell proliferation in various targeted mutation model, the detailed function of Pkhd1 for cholangiocytes is currently known. Does the defective expression of Pkhd1 on cholangiocytes directly facilitate the exposure of the autoantigen? Mitochondrial antigens usually get exposed to cellular surface when the cell goes through apoptosis and the bile duct epithelial cells translocate immunologically intact PDC-E2 to apoptotic bodies and create an apotope [24] in PBC patient and there is an intense pro-inflammatory cytokine production in the in vitro culture of BEC apotopes, monocyte-derived macrophages from PBC patients and AMAs [25]. PDC-E2 is found intact in apoptotic blebs during cholangiocyte apoptosis and thus apoptotic cholangiocytes are considered as the source of immunogenic PDC-E2 in PBC [24, 26]. The unmodified PDC-E2 in

PBC is dependent on lack of glutathiolation instead of lack of degradation [26]. Glutathione is a major component of antioxidative system and the plasma total glutathione in PBC patients is significantly reduced compared to control [27]. However, in the parental strain of NOD.Abd3,

NOD.c3c4, the intrahepatic cholangiocytes are impaired in Fas expression and are Proliferating

119

cell nuclear antigen(PCNA)-positive and Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL)-negative, indicating they proliferate without apoptosis [28], consistent with the observation in NOD.Abd3 liver histology which shows a profound cholangiocyte hyperplasia.

Moreover, recent studies suggest that Pkhd1 may negatively regulate cell apoptosis and NFκB activity [29, 30]. Therefore, it is intriguing to understand whether NOD.Abd3 cholangiocytes present the autoantigen to the cell surface with the defective Pkhd1 expression.

Approach: To address this question, we will isolate intrahepatic cholangiocytes from NOD.Abd3 and NOD control, treat them with or without biliary salts as apoptosis inducing reagent [24, 31].

The localization of PDC-E2 will then be investigated using a previously described mouse monoclonal antibody against PDC-E2, clone 2H-4C8 [24, 32]. This question is important because it will allow us to understand what role the fibrocystin, the protein encoded by Pkhd1, plays in the process of autoantigen leakage, which is critical for the recognition of autoantigen-specific T cells and production of AMAs, the two essential components in the PBC pathology.

Potential limitations and alternative approach: Unfortunately, there is a paucity of reliable immunoreagents of endogenous fibrocystin [21, 33-35], so it is unable to determine whether there is a co-localization between fibrocystin and PDC-E2.

The unmodified PDC-E2 in PBC is dependent on lack of glutathiolation instead of lack of degradation [26]. Glutathione is a major component of antioxidative system and the plasma total glutathione in PBC patients is significantly reduced compared to control with oxidative stress as an early feature in PBC patients [27]. Given multiple enzymes involved in oxidation and reduction are dysregulated as revealed in our RNA-seq data from 2-week-old NOD.Abd3 bile duct compared to NOD control(data not shown), it is reasonable to speculate the red-ox status of

NOD.Abd3 cholangiocytes is altered. With no obvious evidence of cholangiocyte apoptosis in

NOD.Abd3, we would like to study the mechanism in which the immunogenic autoantigen is generated. We hypothesized that disrupted expression of Pkhd1 in cholangiocytes is

120

associated with altered cellular red-ox status, setting up the environment for immunologically intact PDC-E2.

Approach: Because glutathione is among the most abundant substance in cytosol and participates in reduction and oxidation of the highly prevalent thiol bonds (those containing sulfur) in cysteine of many proteins [36], we will measure glutathione pool as redox status. Three distinct measurements will be performed [37]: (1) Total glutathione, the sum of reduced (GSH) and oxidized glutathione (GSSG) in fresh whole blood; (2) Total glutathione (GSH+GSSG) in plasma.

Fresh whole blood will be added to equal volume of 5,5′-dithio-bis(2-nitrobenzoic acid) (DTNB) and centrifuged immediately. DTNB reacts rapidly with free sulfhydryl groups and prevents oxidation of GSH during sample preparation. (3) GSSG in plasma. Fresh whole blood will be added to equal volume of N-ethyl-maleimide (NEM) and centrifuged as before. NEM chemically captures GSH and prevents auto-oxidation to GSSG. NEM will be removed by solid phase extraction prior to assay. The reduction of DTNB by NADPH will be catalyzed by GSH or GSSG and glutathione reductase and the resulting absorbance measured at λ=412 nm.

In addition, we ask whether autoreactive CD8 T cells attack the cholangiocytes in NOD.Abd3. T cells infiltrate the portal triads of NOD.Abd3 liver and CD8 T cells cause autoimmune cholangitis in the transfer model. However, we don’t have direct evidence showing CD8 T cells damage cholangiocytes in NOD.Abd3. We would like to know whether NOD.Abd3 cholangiocyte release any cellular signal (eg. cytokines) as CD8 T cell chemoattractants and whether there is an active interaction between CD8 T cells and cholangiocytes in NOD.Abd3. In answering these questions, we will be able to further our understanding in the difference of CD8 T cell cytotoxicity between

NOD.Abd3 and NOD control.

Approach: 1. Cholangiocyte killing assay will be conducted by co-culturing hepatic CD8 T cell from NOD.Abd3 or NOD with autologous or allogeneic intrahepatic cholangiocytes for 5 hours.

Controls with only CD8 T cells or cholangiocytes in media treated with or without lysing buffer

121

will be set up. At the end of the culture, the Promega CytoTox 96® Assay will be applied to quantitatively measure lactate dehydrogenase (LDH), a stable cytosolic enzyme that is released upon cell lysis. Released LDH in culture supernatants will then be measured with a 30-minute coupled enzymatic assay that converts a tetrazolium salt (INT) into a red formazan product. The amount of red color is proportional to the number of lysed cells. Visible wavelength absorbance data will then be collected using a standard 96-well plate reader. We expect to see that only in

NOD.Abd3 CD8 T cells co-cultured with autologous cholangiocyte, there is a CD8 T cell- mediated killing of cholangiocyte. 2. In a similar cell culture setting but with CD8 T cells and cholangiocytes separated by a permeable membrane, CD8 T cells will be labeled with fluorescent marker before culture and their movement as defined by distance between initial and final position as well as the movement direction in the cell culture will be recorded using digital camera. 3. The supernatant of cell culture for NOD.Abd3 and NOD cholangiocytes will be collected and subjected to a comprehensive profiling of cytokines and chemoattractants to establish a relationship between the movement of CD8 T cells and the cell signal released by the cholangiocytes.

As uncovered by RNA-seq data of common bile duct from NOD.Abd3 and NOD, there is a significant increase of IL-17RE transcript in NOD.Abd3 bile duct compared to NOD (data not shown). IL-17RE, an orphan receptor of IL-17 receptor family has recently been identified as the functional receptor for IL-17C, a pro-inflammatory cytokine [36, 38, 39]. IL-17RE is highly expressed on intestinal epithelial cells and critical for IL-17C mediated mucosal immune response against pathogen infection or colitis [39, 40]. However, IL-17RE is also expressed in lymphocyte compartment by Th17 cells [38]. It will be interesting to understand what role IL-17RE plays in the autoimmune cholangiopathy of NOD.Abd3. Is cholangiocyte the major resource of IL-17RE?

How is IL-17RE induced on cholangiocyte? Is it through IL-6/TGFβ in the presence of IL23?

Does the Il-17C/IL-17RE autocrine pathway participate in the autoimmune cholangitis? The

122

answers to these questions can also give us insights on the etiopathology as well as the treatment of human PBC.

Summary statement

Although human PBC doesn’t show genetic defects in TGFβ signaling or Pkhd1, our work may model PBC immunopathology in at least certain populations of patients. It may be necessary for both a specifically pathogenic CD8 subpopulation and dysfunctional Tregs to develop in order for

PBC to progress. In other patients, certain genetic, epigenetic or environmental factors can contribute to the alteration of cholangiocyte biology which makes the autoantigen exposure and/or modification extremely susceptible. Further investigation of these abnormalities will eventually allow the treatment of PBC to be based on each individual and become more customized. Currently, most of the treatment for PBC involves UDCA and immunosuppression regimens. However, over 40% of patient show incomplete respond to UDCA [41] and not all patients respond to immunosuppressive regimens [42]. Targeted therapeutic treatment to a specific cell population, either cholangiocytes or immune cells, can potentially lead to new therapeutic approaches in PBC.

123

References

1. Kaech, S.M., E.J. Wherry, and R. Ahmed, Effector and memory T-cell differentiation: implications for vaccine development. Nat Rev Immunol, 2002. 2(4): p. 251-262. 2. Joshi, N.S. and S.M. Kaech, Effector CD8 T Cell Development: A Balancing Act between Memory Cell Potential and Terminal Differentiation. The Journal of Immunology, 2008. 180(3): p. 1309-1315. 3. Joshi, N.S., et al., Inflammation Directs Memory Precursor and Short-Lived Effector CD8(+) T Cell Fates via the Graded Expression of T-bet Transcription Factor. Immunity, 2007. 27(2): p. 281-295. 4. Sarkar, S., et al., Functional and genomic profiling of effector CD8 T cell subsets with distinct memory fates. The Journal of Experimental Medicine, 2008. 205(3): p. 625-640. 5. Thimme, R., et al., Increased Expression of the NK Cell Receptor KLRG1 by Virus- Specific CD8 T Cells during Persistent Antigen Stimulation. Journal of Virology, 2005. 79(18): p. 12112-12116. 6. Sanjabi, S., M.M. Mosaheb, and R.A. Flavell, Opposing effects of TGF-β and IL-15 control the number of short-lived effector CD8(+) T cells. Immunity, 2009. 31(1): p. 131- 144. 7. Cui, W., et al., Effects of Signal 3 during CD8 T cell priming: Bystander production of IL-12 enhances effector T cell expansion but promotes terminal differentiation. Vaccine, 2009. 27(15): p. 2177-87. 8. Obar, J.J., et al., Pathogen-induced inflammatory environment controls effector and memory CD8+ T cell differentiation. J Immunol, 2011. 187(10): p. 4967-78. 9. Chang, P.-P., et al., Breakdown in Repression of IFN-γ mRNA Leads to Accumulation of Self-Reactive Effector CD8+ T Cells. The Journal of Immunology, 2012. 189(2): p. 701- 710. 10. Yang, G.X., et al., Adoptive transfer of CD8(+) T cells from transforming growth factor beta receptor type II (dominant negative form) induces autoimmune cholangitis in mice. Hepatology, 2008. 47(6): p. 1974-82. 11. Bettelli, E., et al., Reciprocal developmental pathways for the generation of pathogenic effector TH17 and regulatory T cells. Nature, 2006. 441(7090): p. 235-8. 12. Kachapati, K., et al., The B10 Idd9.3 locus mediates accumulation of functionally superior CD137(+) regulatory T cells in the nonobese diabetic type 1 diabetes model. J Immunol, 2012. 189(10): p. 5001-15. 13. Lan, R.Y., et al., Liver-targeted and peripheral blood alterations of regulatory T cells in primary biliary cirrhosis. Hepatology, 2006. 43(4): p. 729-37. 14. Leveen, P., et al., Induced disruption of the transforming growth factor beta type II receptor gene in mice causes a lethal inflammatory disorder that is transplantable. Blood, 2002. 100(2): p. 560-8. 15. Rubtsov, Y.P., et al., Stability of the regulatory T cell lineage in vivo. Science, 2010. 329(5999): p. 1667-71. 16. Komatsu, N., et al., Pathogenic conversion of Foxp3+ T cells into TH17 cells in autoimmune arthritis. Nat Med, 2014. 20(1): p. 62-68. 17. Moser, M., et al., A mouse model for cystic biliary dysgenesis in autosomal recessive polycystic kidney disease (ARPKD). Hepatology, 2005. 41(5): p. 1113-1121. 18. Gallagher, A.-R., et al., Biliary and Pancreatic Dysgenesis in Mice Harboring a Mutation in Pkhd1. The American Journal of Pathology, 2008. 172(2): p. 417-429. 19. Kim, I., et al., Fibrocystin/Polyductin Modulates Renal Tubular Formation by Regulating Polycystin-2 Expression and Function. Journal of the American Society of Nephrology, 2008. 19(3): p. 455-468.

124

20. Nagasawa, Y., et al., Identification and Characterization of Pkhd1, the Mouse Orthologue of the Human ARPKD Gene. Journal of the American Society of Nephrology, 2002. 13(9): p. 2246-2258. 21. Boddu, R., et al., Intragenic motifs regulate the transcriptional complexity of Pkhd1/PKHD1. Journal of molecular medicine (Berlin, Germany), 2014. 92(10): p. 1045- 1056. 22. Zheng, Q., et al., Precise gene deletion and replacement using the CRISPR/Cas9 system in human cells. Biotechniques, 2014. 57(3): p. 115-24. 23. Yang, H., H. Wang, and R. Jaenisch, Generating genetically modified mice using CRISPR/Cas-mediated genome engineering. Nat. Protocols, 2014. 9(8): p. 1956-1968. 24. Lleo, A., et al., Apotopes and the biliary specificity of primary biliary cirrhosis. Hepatology, 2009. 49(3): p. 871-9. 25. Lleo, A., et al., Biliary apotopes and anti-mitochondrial antibodies activate innate immune responses in primary biliary cirrhosis. Hepatology, 2010. 52(3): p. 987-998. 26. Odin, J.A., et al., Bcl-2–dependent oxidation of pyruvate dehydrogenase-E2, a primary biliary cirrhosis autoantigen, during apoptosis. The Journal of Clinical Investigation, 2001. 108(2): p. 223-232. 27. Aboutwerat, A., et al., Oxidant stress is a significant feature of primary biliary cirrhosis. Biochimica et Biophysica Acta (BBA) - Molecular Basis of Disease, 2003. 1637(2): p. 142-150. 28. Nakagome, Y., et al., Autoimmune cholangitis in NOD.c3c4 mice is associated with cholangiocyte-specific Fas antigen deficiency. Journal of autoimmunity, 2007. 29(1): p. 20-29. 29. Mangolini, A., et al., NF-κB activation is required for apoptosis in fibrocystin/polyductin-depleted kidney epithelial cells. Apoptosis, 2010. 15(1): p. 94-104. 30. Sun, L., et al., Down-regulation of PKHD1 induces cell apoptosis through PI3K and NF- κB pathways. Experimental Cell Research, 2011. 317(7): p. 932-940. 31. Drudi Metalli, V., et al., Bile salts regulate proliferation and apoptosis of liver cells by modulating the IGF1 system. Digestive and Liver Disease. 39(7): p. 654-662. 32. Migliaccio, C., et al., Monoclonal Antibodies to Mitochondrial E2 Components Define Autoepitopes in Primary Biliary Cirrhosis. The Journal of Immunology, 1998. 161(10): p. 5157-5163. 33. Ward, C.J., et al., Cellular and subcellular localization of the ARPKD protein; fibrocystin is expressed on primary cilia. Human Molecular Genetics, 2003. 12(20): p. 2703-2710. 34. Wang, S., et al., The Autosomal Recessive Polycystic Kidney Disease Protein Is Localized to Primary Cilia, with Concentration in the Basal Body Area. Journal of the American Society of Nephrology, 2004. 15(3): p. 592-602. 35. Menezes, L.F.C., et al., Polyductin, the PKHD1 gene product, comprises isoforms expressed in plasma membrane, primary cilium, and cytoplasm. Kidney Int, 2004. 66(4): p. 1345-1355. 36. Lushchak, V.I., Glutathione Homeostasis and Functions: Potential Targets for Medical Interventions. Journal of Amino Acids, 2012. 2012: p. 26. 37. Sies, H. and T.P. Akerboom, Glutathione disulfide (GSSG) efflux from cells and tissues. Methods Enzymol, 1984. 105: p. 445-51. 38. Chang, Seon H., et al., Interleukin-17C Promotes Th17 Cell Responses and Autoimmune Disease via Interleukin-17 Receptor E. Immunity, 2011. 35(4): p. 611-621. 39. Ramirez-Carrozzi, V., et al., IL-17C regulates the innate immune function of epithelial cells in an autocrine manner. Nat Immunol, 2011. 12(12): p. 1159-1166. 40. Khader, S.A. and R. Gopal, IL-17 in protective immunity to intracellular pathogens. Virulence, 2010. 1(5): p. 423-427.

125

41. Wang, L., et al., Breach of tolerance: primary biliary cirrhosis. Semin Liver Dis, 2014. 34(3): p. 297-317. 42. Imam, M.H., J.A. Talwalkar, and K.D. Lindor, Clinical management of autoimmune biliary diseases. J Autoimmun, 2013. 46: p. 88-96.

126