Keywords

1 Introduction

Interlobular bile ducts are the primary targets in primary biliary cirrhosis (PBC), of which the term was first introduced more than six decades ago [1]. A certain condition comprising genetics, epigenetics, and environmental factors is thought to affect susceptible individuals, resulting in the development of PBC [2–4]. Compatible for this concept, some of the congenic, transgenic, gene-deleted (knocked-out), and chemical xenobiotic-immunized mice have been reported as faithful animal models with accompanying close similarity of features seen in human PBC. In addition to the enthusiastic researches on human PBC, these animals have advanced our understandings of the pathogenesis of PBC.

Five spontaneous models and one faithfully induced mouse model have been reported for human PBC in a decade [5–11]. These models overcome the limitation to access human liver tissues and have enabled to elucidate the pathogenesis of PBC in detail (i.e., contribution of each immune cell population, cholangiocytes, humoral immune components, cytokines/chemokines, and their receptors). Recent findings of PBC mouse models are summarized and compared to the features of PBC (Table 14.1).

Table 14.1 Comparison between human PBC and the mouse models
Full size table

2 dnTGF-βRII Mice

Transforming growth factor (TGF)-β mediates pleiotropic functions on various cells and plays a role in the central negative regulation in autoimmunity. TGF-β receptor II is essential for signal transduction of TGF-β that regulates the activation of lymphocytes. The mouse expressing the dominant-negative form of TGF-β receptor II (dnTGF-βRII) on CD4+ and CD8+ cells demonstrates closely resembling features of human PBC [5, 57]. dnTGF-βRII has a truncated intracellular domain of the normal receptor, resulting in the incapacity to transduce signal in both CD4+ and CD8+ cells after TGF-β ligation. Although anti-mitochondrial antibodies (AMAs) are not sufficient to assure specific models of PBC [58, 59], these mice demonstrate 100 % of serum AMA positivity in a time-dependent manner, and those against PDC-E2, BDOADC-E2, and OGDC-E2 inhibit its enzymatic activity in vitro. In addition, antinuclear antibodies directed against two nuclear proteins, gp210 and sp100, were detectable in all examined sera out of 21 dnTGF-βRII mice at 24 weeks of age [41]. Liver histology demonstrates lymphoid cell infiltrates in portal tracts accompanied with bile duct injury [5]. dnTGF-βRII mice show an increase in the number and frequency of the CD44+ memory phenotype of CD4+ and CD8+ T cells, a decreased CD4+/CD8+ T cell ratio in liver lymphoid cell infiltrates, and an increment of B cells and natural killer T (NKT) cells in the liver [5]. In addition, serum levels of proinflammatory cytokines such as TNF-α, IFN-γ, IL-12p40, and IL-6 are significantly higher in this strain than in control mice.

2.1 T Cell Contribution in dnTGF-βRII Mice

To examine the contribution of T and B cells to the development of PBC-like disease, dnTGF-βRII mice were crossed with recombinase-deficient Rag1−/− mice that lack a diversified B and T cell receptor repertoire, leading to the absence of B and T cells. Rag1−/−-dnTGF-βRII mice do not develop liver pathology, suggesting that a specific condition of T cells with impaired TGF-β-signaling in the presence or absence of B cells is involved in the pathogenesis of PBC-like disease of this mouse [5]. Thus, the pathology developed in this strain could be hypothesized due to either a structural or functional change of TGF-βRII signaling in T cells or following a breakdown of self-tolerance.

To assess the contribution of T and B cells to the development of PBC-like disease, various series of adoptive transfer studies were performed: transferring splenic CD4+ and/or CD8+ T cells derived from dnTGF-βRII mice into Rag1−/− recipients [33, 34]. Whole splenocytes in dnTGF-βRII mice were sufficient to develop features of liver disease similar to human PBC in Rag1−/− recipients, suggesting that the loss of self-tolerance in splenic T and B cells is sufficient to induce cholangitis, and specific abnormality in the biliary targets was dispensable for the onset of the disease, compatible to the “innocent victim” concept of cholangiocytes in human PBC [15, 60]. More importantly, adoptive transfer of CD8+ but not CD4+ T cells into Rag1−/− mice led to cholangiopathy quite similar to PBC livers, emphasizing a pivotal role of CD8+ T cells in the pathogenesis of both human and murine PBCs [18, 33, 61]. In contrast, Rag-1−/− recipients of CD4+ T cells of dnTGF-βRII mice predominantly developed inflammatory bowel disease associated with higher levels of serum interferon (IFN)-γ and tumor necrosis factor (TNF)-α. These data clarified that CD8+ T cells are the primary contributors for bile duct destruction in this model [33]. Of note, breakdown of T cell self-tolerance to liver autoantigens was indispensable to develop cholangiopathy in dnTGF-βRII mice [42]. Ovalbumin (OVA)-specific CD8+ T cell (OT-I) or OVA-specific CD4+ T cell (OT-II) mice were utilized to develop OT-I/dnTGF-βRII/Rag-1−/− and OT-II/dnTGF-βRII/Rag-1−/− mice in which the entire T cell repertoire was substituted for OVA-specific CD8+ and CD4+ T cells and to examine the specificity of autoantigens in dnTGF-βRII mice. Adoptive transfer of CD8+ T cells from dnTGF-βRII mice but not from OT-I/dnTGF-βRII/Rag-1−/− mice or OT-I/Rag-1−/− mice induced cholangitis in Rag-1−/− recipients, confirming that the cholangiopathy is not secondarily due to nonspecifically activated CD8+ T cells in dnTGF-βRII mice.

2.2 B Cell Contribution in dnTGF-βRII Mice

Despite the nearly universal occurrence of serum AMA as well as accumulation of B cells among liver infiltrates [62], the contribution of B cells to the pathogenesis of human PBC had remained unclear [63]. Similarly, B cell self-tolerance is abrogated in dnTGF-βRII mice [5, 41]. Thus, an expectation seemed valid that B cell deficiency should ameliorate liver disease, and dnTGF-βRII mice were crossed with B cell deficient mice (Igμ−/−) and explored for liver inflammation as well as accompanying colitis. Contrarily, genetic B cell deprivation exacerbated both the PBC-like liver disease and colitis [34]. Also, B cell deletion enhanced the expansion of the CD8+ T cell population compared to CD4+ T cells and diminished the hepatic regulatory T cell (Treg) frequency in the CD4+ T cell population. A putative regulatory B cell (Breg) population produces anti-inflammatory cytokine, especially IL-10 [64]; however, dnTGF-βRII hepatic B cells indicated comparable levels of IL-10 mRNA expression to those in B6 mice (Moritoki Y, unpublished data), suggesting that Breg population exists outside of the liver and suppresses liver inflammation. Two major B cell pools, the peritoneal cavity and spleen, were chosen for the B cell sources to examine the B cell suppressive function in the CD8+ T cell adoptive transfer model [33]. B cells from the peritoneal cavity, but not from the spleen of dnTGF-βRII mice, were sufficient to regulate CD8+ T cell induced PBC-like liver disease. These findings revealed the existence of a suppressive/regulatory B cell subset on autoimmune cholangitis in the dnTGF-βRII mice and raised a new concept of regulatory B cells in liver inflammation.

However, since the role of B cells is still controversial and disease phase dependent in autoimmunity [65], the regulatory function may not be universal in B cells for liver inflammation in dnTGF-βRII mice. Hence, to examine the effects of therapeutic B cell depletion, dnTGF-βRII mice were treated by intraperitoneal injection of anti-mouse CD20 antibodies (Abs) every 2 weeks from young (4–6 weeks) and old (20–22 weeks) age and subjected to a comparison with control Ab-treated mice. Sixteen-week anti-CD20 treatment initiated from young age demonstrated a fully depleted serum AMA, a significantly lower incidence of liver inflammation, and a fewer number of activated hepatic CD8+ T cells in dnTGF-βRII mice [35]. However, colitis was significantly exacerbated in anti-CD20-treated mice. In contrast, in the aged mice treated from 20 to 22 weeks of age, anti-CD20 treatment was less effective on either liver or colon inflammation.

To summarize B cell functions in dnTGF-βRII mice, in contrast to genetic B cell depletion to exacerbate liver inflammation, anti-CD20 treatment demonstrated therapeutic efficacy to regulate liver inflammation in young but not in old dnTGF-βRII mice, suggesting time- and disease phase-dependent B cell function in autoimmune cholangitis in this strain (Fig. 14.1). Also, B cell functions for liver inflammation observed in dnTGF-βRII mice are quite similar to those recently reported in the other autoimmune disease [65]; however, B cells are collectively a suppressor population for colitis regardless of disease phase or age in dnTGF-βRII mice. These data suggest that B cells separately contribute to each target organ in autoimmunity. In addition, B cell depletion using anti-CD20 can be a possible therapeutic option for human PBC treatment due to the rare concomitance of inflammatory bowel disease. Several reports for B cell depletion using anti-CD20 rituximab in human PBC patients emerged recently [29, 30, 66, 67]. In an open-labeled study comprised of six patients with PBC, those who were refractory to UDCA treatment and administered rituximab 1,000 mg intravenous infusion on day 1 and 15 demonstrated transient decreases in memory B and T cells, an increase of the CD25+ regulatory subset, an increase in mRNA levels of FoxP-3 and TGF-β, and a decrease in TNF-α in CD4+ T cells. Serum levels of ALP and plasma levels of AMA, IgM, IgA, and IgG were also transiently reduced around 6 months after treatment [29]. Similarly to this, the other study with 14 PBC patients refractory to UDCA indicated a significant reduction of ALP, AMA, and IgM at 6 months after rituximab treatment [30]. However, a case report demonstrated rapid progress in a PBC patient accompanied with coincidental gastric diffuse large B cell lymphoma (DLBCL) after eight doses of rituximab as a part of R-CHOP chemotherapy coupled with 40 Gy radiotherapy, although the patient showed biochemical, immunological, and histological improvements after rituximab treatment [67]. The other case report demonstrated persistent liver dysfunction in a PBC patient where concomitant rheumatoid arthritis was improved after treatment with 7.5 mg/week methotrexate and two doses of 1,000 mg rituximab at 2 weeks apart [66]. Taken together, the therapeutic usage of anti-CD20 seems potent to induce biochemical and immunological responses in PBC, but novel biomarkers are feasible to select adequate patients to be treated with rituximab.

Fig. 14.1
figure 1

Effect of B cell depletion in dnTGF-βRII autoimmune cholangitis. Genetic B cell depletion exacerbated autoimmune cholangitis. However, in contrast, early therapeutic B cell depletion using anti-mouse CD20 monoclonal antibodies ameliorated liver pathology. Further, late B cell depletion exhibits no significant therapeutic efficacy in liver inflammation

Full size image

2.3 Natural Killer T (NKT) Cell Contribution in dnTGF-βRII Mice

Natural killer T (NKT) cells bridge innate and adoptive immunity and exhibit immunoregulatory function in some autoimmune diseases [68, 69]. NKT cells are primed for proinflammatory and anti-inflammatory phenotypes under DC-derived cytokine environment such as IL-12 and IL-10, respectively [70].

The activation of invariant NKT cells is a critical factor to accelerate cholangiopathy in PBC [19, 71]. In dnTGF-βRII mice, since the absolute number and activation marker expression are augmented in TGF-β signaling-deprived CD1d-restricted NKT cells, CD1d−/−-dnTGF-βRII mice were generated to deplete CD1d-restricted NKT cells [32]. CD1d−/−-dnTGF-βRII mice exhibited decreased lymphoid cell infiltrates, milder biliary damage, and higher levels of serum IFN-γ compared to those of the control NKT cell sufficient CD1d+/−-dnTGF-βRII mice. In vivo injection with α-galactosylceramide (α-GalCer) induced increases in liver cell infiltrates and serum IFN-γ in dnTGF-βRII mice. These results suggested CD1d-restricted NKT cells are a primarily proinflammatory subset inducing Th1 cytokine bias in dnTGF-βRII mice [32]. Despite the milder liver inflammation, CD1d deletion did not affect the AMA titer in this strain [32].

In the other study, anti-CD40L treatment for 8 weeks demonstrated a reduction in hepatic NKT cells and activated CD8+ T cells, milder portal inflammation, and diminution of bile duct destruction in dnTGF-βRII mice [44].

2.4 Cytokine/Chemokine Contribution in dnTGF-βRII Mice

Since dnTGF-βRII mice demonstrated an increase in serum levels of proinflammatory cytokines TNF-α, IFN-γ, IL-12p40, and IL-6, partly due to the overexpression of microRNA-21 [5, 40], germline deletion of genes encoding those and related cytokines was extensively performed in dnTGF-βRII mice. All four IL-12 family cytokines, IL-12, IL-23, IL-27, and IL-35, have heterodimeric constructs with α and β chains, and each cytokine shares at least one chain with another member of the family [72]. The IL-12p40 subunit is a common component of both IL-12 and IL-23, which are deprived by the deletion of the IL-12p40 subunit. IL-12p40 deletion demonstrated a marked diminution in the levels of proinflammatory Th1 cytokines including IFN-γ in livers of dnTGF-βRII mice accompanied with reductions in cellular infiltrates around portal areas and bile duct damage [36]. In contrast, IFN-γ depletion indicated no significant effect on the immunopathology of autoimmune cholangitis and colitis in this strain [36] (Yoshida K, unpublished data). These data suggest that IL-12p40 contributing signaling pathways, rather than one of its downstream cytokines IFN-γ, is a major determinant of the autoimmune cholangitis that affects dnTGF-βRII mice [36].

IL-23 comprised of IL-12p40 and IL-23p19 subunits enhances Th17 polarization in autoimmunity [73]. IL-23p19 depletion led to IL-23 deprivation, which improved colitis but not cholangitis in dnTGF-βRII mice [39]. IL-23 supports the differentiation of naïve CD4+ T cells into highly pathogenic Th17 cells to produce IL-17A; however, IL-17A deletion did not affect either cholangitis or colitis in this strain [39]. IL-6 coupled with TGF-β induces Th17 cells from naïve CD4+ T cells. Genetical IL-6 depletion in dnTGF-βRII mice led to a marked improvement in inflammatory bowel disease, but deteriorated biliary pathology [38]. Hepatic levels of IFN-γ and TNF-α were significantly elevated in IL-6-deficient dnTGF-βRII mice while hepatic IL-12p40 was comparable to that of IL-6-sufficient controls [38]. Taken together, the IL-12/Th1 pathway, but not coupled with IL-23/Th17 axis, is essential and sufficient to develop autoimmune cholangitis in dnTGF-βRII mice.

IL-12 heterodimeric cytokine is comprised of IL-12p40 and IL-12p35 subunits. In other words, the IL-12p35 subunit is shared in IL-12 and IL-35 coupled with an IL-12p40 subunit and an Epstein–Barr virus-induced gene 3 (EBI3) glycoprotein, respectively. IL-12p35 deletion in dnTGF-βRII mice induced a similar degree of cholangitis with delayed onset, but not colitis, while AMA titer significantly increased at 6 months of age [43]. The deficiency of IL-12 and IL-35 in the presence of IL-23 did not inhibit but delayed the development of biliary disease, suggesting that the pathogenic role of IL-23 can be enhanced in the absence of a presumably immunoregulatory cytokine IL-35 secreted from Treg [72]. Importantly, IL-12p35 deprivation led to Th1 to Th17 shift with a higher production of IL-6 and IL-17 and reduced mRNA expression of both IFN-γ/STAT1 signaling and an antifibrotic factor HGF, resulting in the development of liver fibrosis in 7 out of 13 examined mice at 24 weeks of age while none of the 14 control dnTGF-βRII mice did so. Of note, liver fibrosis is the essential feature in advanced human PBC livers but has not been previously found in spontaneous PBC mouse models [43]. IL-35 deficiency might hamper the Treg functions including the regulation of both liver pathogenesis and fibrosis. Th17 cells are developed from naïve CD4+ T cells through IL-6 and TGF-β signaling; however, recent studies demonstrate that also Th17 cells can be induced alternatively by way of the effects of IL-6, IL-1β, and IL-23 in the absence of TGF-β [74, 75]. These “alternative” Th17 cells are thought to be more pathogenic than the “classical” Th17 cells [76]. Since the TGF-β signaling pathway is obstructed in dnTGF-βRII T cells, IL-12p35 deprivation is likely to induce more pathogenic Th17 cells.

A genome-wide association study (GWAS) identified that IL-12α (IL-12p35 subunit) and IL-12 receptor β2 gene variants were strongly associated with risk for human PBC [20, 77]. Human liver samples obtained from the patients with PBC (n = 51) and non-PBC (n = 80) control liver diseases were examined immunohistochemically for localized expression of cytokines, their subunits, and corresponding receptors with an extensive panel of antibodies directed to IL-12p70, IL-12p35, IFN-γ, IL-12RB2, IL-23p40, IL-23p19, IL-17, and IL-23R [25]. The expression of IL-12RB2 and the presence of IFN-γ-positive mononuclear cells were both observed around the damaged interlobular bile ducts in PBC livers [25].

Again, IL-12p40 depletion demonstrated significant amelioration of cholangitis in dnTGF-βRII mice, suggesting that IL-12p40 suppression serves as a novel therapeutic option for human PBC. IL-12p40 suppression mediated by monoclonal antibodies or transcription inhibitors has been clinically studied to provide therapeutic benefit in other autoimmune diseases such as psoriasis [78] and Crohn’s disease [79–80].

Detection of antinuclear antibodies (ANAs) was first reported in paternal dnTGF-βRII mice [57] and, thence, extensively studied in dnTGF-βRII mice with concurrent deletions of IL-12p35, IL-12p40, IL-23p19, IL-17, IL-6, IFN-γ, and TNF-α [41]. The changes of autoantibody titers, autoimmune cholangitis and colitis are summarized in Table 14.2. IL-12p40 depletion did not affect serum levels of AMA and anti-SP100, but diminished anti-GP210, suggesting that synthesis and upregulation of IL-12p40 ais not essential for the production of AMA in this model [36, 41]. In contrast, AMA and anti-SP100 were significantly higher, but anti-GP210 was lower in sera of IL-23p19-deleted dnTGF-βRII mice [39, 41]. IL-6 depletion reduced both AMA and anti-GP210 without any change of anti-SP100 titers in dnTGF-βRII mice [38, 41]. Further, higher levels of AMA and similar levels of anti-GP210 and anti-SP100 were detected in IL-12p35-deficient dnTGF-βRII mice [41, 43]. IL-17 depletion did not affect AMA or anti-SP100, but reduced anti-GP210 in dnTGF-βRII mice [39, 41]. IFN-γ deletion did not change titers of any AMA, anti-GP210, or anti-SP100 antibodies [41]. TNF-α depletion did not affect the degree of bile duct damage (Yang GX, unpublished data) or the titers of AMA or anti-SP100, but reduced anti-GP210 antibodies [41]. Deterioration of colitis led to shorter survival in TNF-α-deprived model mice than that of parental dnTGF-βRII mice, resulting in the difficulty of reproduction, and thus the study was discontinued (Yang GX, unpublished data). These results suggest that autoantibody-dependent cytotoxicity is not the major mechanism to induce and promote bile duct damage in the absence of these cytokines.

Table 14.2 Effects on AMA, ANA, cholangitis, and colitis in cytokine-deleted dnTGF-βRII mice
Full size table

2.5 Therapeutic Immunomodulation in dnTGF-βRII Mice

Other than B cell depletion, novel therapeutic immunomodulation has been performed in dnTGF-βRII mice. First, β-glucosylceramide (GC) administration for 18 weeks from 6 weeks of age ameliorates liver inflammation and alleviated cholangitis accompanied with significant reduction of hepatic CD8+ memory T cells in dnTGF-βRII mice [37]. GC is a naturally occurring glycosphingolipid and has been shown to function as a “fine-tuning factor” in several mouse models of immune-mediated disorders [81–84]. AMA is not significantly reduced in GC-treated dnTGF-βRII mice.

Second, therapeutic intraperitoneal administration of hamster anti-mouse CD40L antibodies was examined in dnTGF-βRII mice. Although anti-CD40L treatment reduced AMA titer and delayed the development of cholangitis at 12 weeks of age (after 8 weeks of treatment), however, the severity of which was not affected at 24 weeks of age (after 20 weeks of treatment) [44]. Changes in serum IgM were not indicated in anti-CD40L-treated mice while CD40L promoter methylation inversely correlates with IgM levels in patients with PBC [85]. LPS and IFN-γ stimulated human liver-derived macrophages induced apoptosis of cholangiocytes through a CD154 (CD40L)-dependent manner, which was attenuated by the antagonistic antibodies against CD154 [86]. The efficacy of blocking CD40L has been reported in the other models of systemic and organ-specific autoimmune diseases such as systemic lupus erythematosus, rheumatoid arthritis, and multiple sclerosis [87–90].

3 IL-2Rα−/− Mice

Interleukin-2 receptor α (IL-2Rα)−/− mice demonstrate low frequency but normal suppressive function of Treg [91], similar to the fact that the lower frequency and absolute number of Treg have been observed in PBC patients [92]. Also, PBC-like liver disease was observed in a child with homozygous IL-2Rα deficiency [93]. IL-2Rα−/− mice with C57BL/6 background show 100 % of AMA positivity against PDC-E2, 80 % of ANA positivity, lymphocyte infiltration around portal tracts, and cholangiocyte damage accompanied with intraepithelial CD8+ cells in interlobular bile ducts [6]. Serum levels of IgA were markedly elevated in IL-2Rα−/− mice compared to those of controls [6, 46]. Serum cytokine profiles are quite similar to those of dnTGF-βRII mice, showing elevation of serum levels of TNF-α, IFN-γ, IL-12p40, IL-6, as well as IL-2. Also, similar to liver immunopathology in dnTGF-βRII mice, IL-2Rα−/− mice indicate an increased number and frequency of the CD44+ memory phenotype of CD4+ and CD8+ T cells and decreased CD4/CD8 ratio of liver cellular infiltrates [6].

3.1 T Cell Contribution in IL-2Rα−/− Mice

The role of the interleukin-2 receptor α (IL-2Rα, CD25) and its relationship with Treg has been well demonstrated in murine and human autoimmune diseases such as inflammatory bowel disease as well as PBC [92–94]. Although IL-2Rα−/− mice develop autoimmune cholangitis concomitant with intestinal inflammation while colitis rarely coexists in patients with PBC, it was hypothesized that distinct effector mechanisms would work in selective targeting of autoimmune diseases in the colon and bile duct, and the effects of germline depletion of genes encoding CD4, CD8, or TCR (T cell receptor)-β were evaluated for the severity of colitis and autoimmune cholangitis in IL-2Rα−/− mice. IL-2Rα−/−-CD4−/− mice, in which CD8+ cells are the unique population in TCR+ cells, demonstrated augmented intrahepatic biliary ductular destruction but diminished colitis. In contrast, IL-2Rα−/−-CD8−/− mice lacked biliary ductular destruction but deteriorated colitis, compared with IL-2Rα−/− mice [46]. These results are quite similarly observed in adoptive transfer studies in dnTGF-βRII mice, where CD8+ but not CD4+ T cells are major contributors for biliary damage [33]. Of note, the lack of pathological changes in IL-2Rα−/− TCR-β−/− mice revealed the pivotal role of T cells to induce liver inflammation in IL-2Rα−/− mice, closely similar to the absence of inflammation in Rag-1−/−-dnTGF-βRII mice [5]. Taken together, these results further argue for the key role of CD8+ T cells in the pathogenesis of autoimmune cholangitis. In human PBC, precursors of PDC-E2-specific CTL, recognizing amino acids 159–167 of PDC-E2, are tenfold more frequent in the liver than those in the blood, produce IFN-γ in response to the PDC-E2, and provide cytotoxicity to PDC-E2 peptide-expressing cells [18].

3.2 B Cell Contribution in IL-2Rα−/− Mice

To date, B cell function has not been thoroughly examined in IL-2Rα−/− mice although which sera demonstrated reactivity to both PDC-E2 and nuclear components [6, 46]. Serum reactivity against PDC-E2 was determined in IL-2Rα−/− mice and CD4/CD8-deleted strains and compared to those of control C57BL/6 J mice. Frequencies of serum AMA positivity were varied: 50 % (4/8) in IL-2Rα−/− mice; 0 % (0/8) in IL-2Rα−/−-CD4−/− mice; 75 % (6/8) in IL-2Rα−/−-CD8−/− mice. Of note, bile duct damage was apparent in all (8/8)in IL-2Rα−/−-CD4−/− mice that were all negative for AMA but had an increase in hepatic B cells compared to control mice. In contrast, IL-2Rα−/−-CD8−/− mice demonstrated higher positivity for AMA but lacked any liver inflammation. These results suggest again that CD8+ T cells provide the primary contribution while AMAs themselves are not sufficient to initiate biliary disease in this strain.

Serum levels of IgA were significantly increased in both IL-2Rα−/− and IL-2Rα−/−-CD8−/− mice, possibly corresponding with the degree of colonic inflammation accompanied with an increase in B cell infiltration within the colonic epithelia. IgM levels were reduced in IL-2Rα−/−-CD4−/− mice that had minimal colonic inflammation compared to those in control mice although hyper IgM is one of the features in human PBC patients [46].

3.3 Cytokine/Chemokine Contribution in IL-2Rα−/− Mice

In addition to the severe cellular infiltration in the portal tract, both proinflammatory and T helper 1 (Th1) cytokines, TNF-α, IFN-γ, IL-12p40, IL-2, and IL-17, were elevated in sera from IL-2Rα−/− and IL-2Rα−/−-CD8−/−mice. In contrast, IL-2Rα−/−-CD4−/− mice indicated elevation of those cytokines except IL-17 [46]. In IL-2Rα−/− mice, serum levels of IL-17 peaked around 8–13 weeks of age and hepatic CD4+-positive cells grasped Th17 bias compared to Th1, which was greater than that in splenocytes [47]. Liver non-parenchymal cells supported splenic CD4+ T cells to secrete IL-17 with a tenfold increase than that in CD4+ T cell alone culture [47], suggesting a Th17 induction role in the liver microenvironment in cases of liver autoimmunity and other liver inflammatory diseases.

On the other hand, the CD45RBhi transfer model of colitis has demonstrated that IL-17A is a negative regulator of Th1 development [95]. The barrier function of tight junction in mouse cholangiocytes was disrupted by Th1 and proinflammatory cytokines IFN-γ and TNF-α [96]. These results suggest that Th17 cell population is not only dispensable to develop autoimmune cholangitis, but also may reduce liver damage by inhibiting Th1 shift in autoimmune cholangitis in this model. In humans, Th17-related cytokines IL-23p19 and IL-17 were significantly elevated in sera of patients with PBC compared to those in healthy and chronic hepatitis B-affected subjects [28]. Also, IL-17-positive mononuclear cells are significantly increased in portal tracts in diseased human livers affected by PBC, autoimmune hepatitis, nonalcoholic steatohepatitis, and chronic hepatitis C compared to those with normal livers [47]. IL-12RB2 and IL-23R were intensively expressed in cholangiocytes of PBC patients [25]. Th1/Th17 imbalance was slightly prone to Th17 in inflamed PBC livers while the ratio of Th1 and Th17, balanced in the early stage of PBC, grew into a Th17-weighted shift in advanced stage [25]. The expression of IL-6 and IL-1β was enhanced in damaged bile ducts in PBC patients [27], and IL-6, IL-1β, IL-23p19, and IL-23/IL-12p40 mRNAs were upregulated by toll-like receptor (TLR) ligand stimulation in human cholangiocytes where IL-17 induced the production of IL-6, IL-1β, IL-23p19, and chemokines (CXCL1, CXCL2, CXCL3, CXCL6, and CXCL8 and CCL2 and CCL20) [27]. These results suggest that biliary epithelia support Th17 cell development on the site of inflammation and enhance Th17 cell induced biliary damage in the advanced stage of PBC livers.

4 NOD.c3c4 and NOD.ABD Mice

A congenic variant of the nonobese diabetic (NOD) strain NOD.c3c4 mice has B6- and B10-derived introgressed intervals on chromosome 3 and 4 and is fully protected from diabetes [7, 48]. NOD.c3c4 mice were reported as the first spontaneous model animal developing AMA and liver pathology similar to some features of PBC [7]. These mice demonstrated serum positivity to PDC-E2 up to 50–60 %, ANA positivity of 80–90 %, and autoimmune biliary disease (ABD) comprised of lymphocyte infiltration around portal tracts with nonsuppurative destructive cholangitis and epithelioid granuloma as seen in human PBC livers [7, 19, 48]. However, the morphological feature of biliary ducts is distinct from that of human PBC because intrahepatic and common bile ducts in this strain develop biliary cysts and ductal dilatation, respectively. The biliary cyst formation was affected by T cells since pathological changes of the liver and the common bile duct were ameliorated with a 50 % reduction of liver disease penetrance by single-dose anti-CD3 treatment in NOD.c3c4 mice around 6–10 weeks of age [7]. Also, impaired Fas expression on biliary epithelial cells contributed to the development of the biliary cysts [97].

NOD.c3c4-scid mice, lacking T and B lymphocytes, indicated minimal biliary disease, significantly milder than those of NOD.ABD mice, which was generated from the NOD.c3c4 strain following an intercross with the NOD.B6 Idd10/18 congenic strain, and demonstrate ABD similar to that of NOD.c3c4 mice [7, 49]. NOD.c3c4 and NOD.ABD mice were both examined for adoptive transfer of autoimmune effector mechanisms [7, 49]. ABD was reproducible in both NOD.c3c4-scid and NOD-scid mice by transfer of whole splenocytes or splenic CD4+ T cells from ABD-developed NOD.c3c4 mice [7, 48]. Also, NOD.ABD splenocytes transferred ABD except biliary cyst formation into NOD.c3c4-scid mice, but not into NOD-scid mice [49], suggesting that the primary lesions mediated by the adaptive immune system are biliary duct damage and nonsuppurative destructive cholangitis, and the genetic background of NOD.c3c4-scid mice is indispensable to induce ABD [49]. NOD.ABD splenic CD8+ T cell alone transfer was sufficient to develop ABD, and the degree of which was comparable to that by the CD8+ T cell cotransfer with CD4+CD25− T cells, but ameliorated with CD4+CD25+ Treg.

Autoantibody production is a distinctive feature between NOD.ABD and NOD.c3c4 mice possibly due to the difference of Idd.9.3 allele origins derived from NOD and B10 mice in NOD.ABD and NOD.c3c4 mice, respectively. ANA was detectable in sera of NOD.c3c4 mice but, however, rarely (1/30) detected in NOD.ABD mice although all mice indicated liver inflammation. In contrast, the incidence of serum AMA increased with age and liver disease severity in NOD.ABD mice (up to 44 %) and NOD.c3c4 mice (up to 50 %) [49, 50], suggesting AMA occurrence is secondary to the biliary damage in these strains.

Similar to frequent concomitance of sicca syndrome in PBC, salivary gland inflammation coexists in NOD.c3c4 mice [50]. Genetic B cell depletion ameliorated inflammation in both liver and salivary tissues at 24 weeks, while no difference of which was observed regardless of B cell presence in 8 weeks of age in NOD.c3c4 mice [50], suggesting B cells do not contribute to the initiation of the biliary and salivary diseases in this model.

5 Scurfy Mice

In the context of decreased frequency of CD4+CD25high Treg in human PBC [92], also studied were Scurfy (Sf) mice, which have a Foxp3 gene mutation that results in a deficiency of functional Treg [98, 99]. Sf mice demonstrated serological, histological, and cytokine features characteristic of autoimmune cholangitis as seen in the other mouse models, including expanded CD8+ T cell population, similar to patients with PBC [9]. This report further emphasizes to argue universal Treg requirement to suppress autoimmune diseases including autoimmune cholangitis. Of note, Sf mice develop severe immune dysregulation, as observed in human immune dysregulation, polyendocrinopathy and enteropathy, and X-linked (IPEX) syndrome which affects humans that have mutations in the Foxp3 gene [100]. One out of 12 patients with IPEX syndrome demonstrated enteropathy, skin disease, and elevation of serum total IgM, IgA, and IgA-AMA, but had no signs of cholestasis [101], although IgA and IgA-AMA have been implicated in cholangiocyte apoptosis in PBC livers [102, 103].

6 Ae2a,b −/− Mice

Antigen presenting requires costimulatory signals through CD80/86; however, human and mouse normal biliary epithelial cells (cholangiocytes) lack the ability to express these molecules [104]. It has not been addressed whether cholangiocytes present specific antigen in spontaneous PBC mouse models described above. Since human and mouse cholangiocytes express various toll-like receptors (TLRs) [105], mouse cholangiocytes are involved in immune responses as a first-line defense for microbial infection. Cholangiocytes in patients with PBC expressed similar levels of TLR subtypes, CD40, and human leukocyte antigen DRα (HLA-DRα) and secreted equivalent amounts of chemokines. Cholangiocyte-expressed chemokines enhanced transmigration of liver-infiltrating mononuclear cells (LMNCs) in PBC. Autologous LMNC cocultured with cholangiocytes from patients with PBC produced higher levels of chemokines and enhanced the expression of CD40 and HLA-DRα [60]. Cholangiocytes produced CX3CL1 in the presence of autologous LMNC, TNF-α, and a TLR3 ligand [106]. Unmodified PDC-E2 was localized in apoptotic cholangiocytes, which combined with AMA enhanced production of proinflammatory cytokines including IL-6, TNF-α, and IL-12p40 from monocyte-derived macrophages in PBC patients [107, 108]. These results suggest cholangiocytes are easily targeted on the site of inflammation by PDC-E2-reactive autoimmune cells in PBC livers.

Genetic modification of Cl−/HCO3− anion exchanger 2 (AE2), primarily functioning in cholangiocytes, enabled to develop the other mouse model for human PBC. AE2 is involved in intracellular pH (pHi) regulation and transepithelial acid–base transport including secretin-stimulated biliary bicarbonate excretion. Combination therapy of UDCA and corticosteroid enhanced AE2 gene expression in normal human cholangiocytes [109]. The expression of AE2 gene was attenuated in liver tissues and blood mononuclear cells in patients with PBC [110–112], possibly due to the upregulated expression of microRNA506 to interfere with AE2 mRNA in PBC livers and cholangiocytes compared to those in normal humans [31]. Ae2 gene-disrupted mice (Ae2a,b −/− mice) demonstrated similar features seen in human PBC: enhanced production of IL-12 and IFN-γ, an expansion of CD8+ T cell, and a reduction in Treg populations [8]. Since the major histocompatibility complex class I molecule H2-D1 is markedly upregulated in cholangiocytes, Ae2-deleted cholangiocytes may attract CD8+ T cells and promote their attacks to biliary ducts. Of note, separately from the features of human PBC, Ae2a,b −/− mice exhibit azoospermia, reduced gastric acid secretion, growth retardation, bone abnormalities, and deafness [113].

7 2-Octynoic Acid-Conjugated BSA (2OA-BSA)-Immunized Mice

The serological hallmark of PBC is the presence of antibodies to mitochondrial antigens identified as the E2 subunits of the pyruvate dehydrogenase complex (PDC) and related enzymes [63, 114], leading an expectation that PDC-E2 immunization develops mouse models for PBC; however, previous findings had been controversial [115–117]. Nonetheless, since PBC sera demonstrated high reactivity against 2-octynoic acid (OA)-modified PDC-E2 peptide [3, 118], 2-OA conjugated bovine serum albumin (BSA) was immunized into B6 mice and a congenic variant of the nonobese diabetic (NOD) strain, NOD1101 mice. 2-OA conjugated BSA (2OA-BSA) was a potent immunodominant epitope in both strains to break down B cell tolerance against PDC-E2 and induce PBC-like liver disease, whose pathology was characterized by an abundance of lymphocytes in B6 mice and neutrophils in NOD1101 mice [10, 11]. In this model, B cells were the most dominant population in hepatic mononuclear cells and demonstrated an approximately four- to fivefold increase compared to CD4+ and CD8+ T cells [52]. 2OA-BSA immunization combined with either α-GalCer or a TLR-3 ligand polyinosinic-polycytidylic acid (Poly I:C) induced liver fibrosis [53, 56].

7.1 B Cells in 2OA-BSA-Immunized Mice

Since autoantibody-producing B cells had been believed as contributors in autoimmune cholangitis, it was a surprising fact that B cells negatively suppressed liver inflammation in dnTGF-βRII mice, confirmed in the genetic B cell depletion and CD8+ T cell transfer autoimmune cholangitis model [33, 34]. Thereafter, B cell function has been examined in a few studies using anti-CD20-treated dnTGF-βRII and genetic B cell depleted NOD.c3c4 mice demonstrating amelioration of liver disease [35, 50]. In addition, anti-CD20 and CD79b pretreatment was examined in 2OA-BSA-immunized mice. B cell depleting antibodies were administered intraperitoneally 1 week before the initiation of 2OA-BSA immunization. Bile duct damage, portal inflammation, and granuloma formation were deteriorated in B cell depleted 2OA-BSA-immunized mice, accompanied with elevation of proinflammatory IFN-γ and MCP-1 in sera [54].

7.2 NK and NKT Cells in 2OA-BSA-Immunized Mice

NK and NKT cell depletion using NK1.1 antibodies demonstrated their roles to enhance AMA and cytokine production from T cells [51]. Enzyme-linked immunospot (ELISPOT) assay in response to pyruvate dehydrogenase complex (PDC)-E2 synthetic peptides revealed the reduction of serum AMA- and IFN-γ-producing splenic T cells in anti-NK1.1-treated mice at 6 and 12 weeks after 2OA-BSA immunization. Both anti-NK1.1- and control antibody-treated groups demonstrated mild infiltration of lymphocytes in portal tracts; however, no histological difference was detected [51]. NK and NKT cell contribution is possibly limited to the induction phase of autoimmune cholangitis in the 2OA-BSA-immunized mice model.

Invariant NKT (iNKT) cells are efficiently activated by α-GalCer, resulting in increases of IFN-γ and IL-4 in sera, productions in hepatic iNKT cells, and maturation of DCs in the liver and spleen of C57BL/6 mice [53]. Intravenous administration of α-GalCer prior to 2OA-BSA immunization led to a profound deterioration of liver disease in 2-OA-BSA-immunized mice, including increments in AMA production, CD8+ T cell biliary infiltration, portal inflammation, granuloma formation, bile duct damage, and liver fibrosis, suggesting iNKT cell activation is a critical factor to exacerbate the manifestation of PBC. It is noteworthy again that iNKT cell activation promoted liver fibrosis with higher incidence (10/13) after a 12-week 2OA-BSA immunization compared to that (1/9) of controls.

7.3 Cytokine/Chemokine Contribution in 2OA-BSA-Immunized Mice

2OA-BSA immunization was also examined in C57BL/6 background mice with deletion of cytokines IL-12p40, IL-12p35, IL-23p19, IL-17A, IL-17F, IL-22, and IFN-γ. The Th1/Th17 signaling pathway was examined using gene deletion of IL-12p40 (Th1/Th17 deprivation), IL-12-p35 (Th1 deprivation), and IL-23p19 (Th17 deprivation), revealing that both Th1 and Th17 cytokines are involved, and IL-12p40 is indispensable in the development of autoimmune cholangitis in 2OA-BSA-immunized mice. IL-23p19 depletion did not affect IFN-γ levels in peripheral blood as well as its secretion in mononuclear cells isolated from both the spleen and liver. However, IFN-γ levels of liver-extracted protein were diminished by IL-23p19 deletion.

Bile duct damage and portal inflammation were ameliorated by deletion of IL-17A and IL-22, but not by IL-17F deprivation. Consistently, hepatic levels of inflammatory cytokines IFN-γ, TNF-α, and IL-6 were significantly reduced in IL-17A- and IL-22-depleted mice, but not in IL-17F-deprived mice, suggesting Th17 cytokines IL-17A and IL-22 promote migration and accumulation of proinflammatory Th1 cells into biliary diseased livers. Serum reactivity against PDC-E2 was also reduced in IL-17A-deficient, but not in IL-22-depleted, mice. In contrast, IFN-γ deletion completely abrogated the development of autoimmune cholangitis and significantly decreased serum levels of AMA, suggesting a pivotal role of IFN-γ for the induction of biliary damage in this model [52].

7.4 Therapeutic Immunomodulation in 2OA-BSA-Immunized Mice

Coinhibitory immunoreceptor cytotoxic T lymphocyte antigen 4 (CTLA-4) has been remarked in recent PBC studies [119–122]. CTLA-4 is expressed in T cells and negatively regulates T cell activation through higher-affinity binding to CD80/CD86 than CD28 to inhibit costimulation and transmit inhibitory signals [123, 124]. CTLA-4 immunoglobulin (Ig) is one of the effective treatment options in autoimmune diseases such as rheumatoid arthritis [125] and juvenile idiopathic arthritis [126]. CTLA-4-Ig comprised of the extracellular domain of human CTLA-4 linked to a modified portion of the Fc domain of human IgG-1 and inhibits T cell activation.

Since CTLA-4-Ig binds to both human and mouse CD80/86, its therapeutic effect was examined in 2OA-BSA mice [55]. CTLA-4-Ig pretreatment initiated 1 day before 2OA-BSA immunization completely abrogated liver inflammation, bile duct damage, and development of anti-PDC-E2 autoantibodies. Therapeutic efficacy of CTLA-4-Ig was examined on developed autoimmune cholangitis at 8 weeks after 2OA-BSA immunization. Four weeks of treatment with CTLA-4-Ig significantly diminished bile duct damage and frequency of effector T cells, but did not affect AMA titer in diseased mice, suggesting that an optimized regimen with CTLA-4-Ig is a potential therapeutic candidate for the treatment of PBC patients [55].

8 Summary

The spontaneous model mice reported herein have been developed by single factor disruption using gene modification primarily affecting the immune cells or the target biliary epithelia, except NOD.c3.c4 and NOD.ABD strains, demonstrating certain similarities of immunological, serological, biochemical, and histological features of human PBC, especially hepatic CD8+ T cell expansion. However, of note, since these mice also have some disadvantages of complications distinct to human PBC (Table 14.1), a research objective demands a deliberate selection of a suitable mouse model.

These emerging models have enabled researchers to address issues residing on human PBC, contribute to elucidate pathogenic factors, and ultimately may propose novel therapeutic options for human PBC. Recent works imply potential usages of some biologics such as anti-CD20 (rituximab), anti-IL-12/23p40 (ustekinumab), and CTLA-4-Ig (abatacept) for human PBC.