Dual Roles of IFN-γ and IL-4 in the Natural History of Murine Autoimmune Cholangitis: IL-30 and Implications for Precision Medicine

Primary biliary cirrhosis (PBC) is a progressive autoimmune liver disease with a long natural history. The pathogenesis of PBC is thought to be orchestrated by Th1 and/or Th17. In this study, we investigated the role of CD4+ helper T subsets and their cytokines on PBC using our previous established murine model of 2-OA-OVA immunization. We prepared adeno-associated virus (AAV)-IFN-γ and AAV-IL-4 and studied their individual influences on the natural history of autoimmune cholangitis in this model. Administration of IFN-γ significantly promotes recruitment and lymphocyte activation in the earliest phases of autoimmune cholangitis but subsequently leads to downregulation of chronic inflammation through induction of the immunosuppressive molecule IL-30. In contrast, the administration of IL-4 does not alter the initiation of autoimmune cholangitis, but does contribute to the exacerbation of chronic liver inflammation and fibrosis. Thus Th1 cells and IFN-γ are the dominant contributors in the initiation phase of this model but clearly may have different effects as the disease progress. In conclusion, better understanding of the mechanisms by which helper T cells function in the natural history of cholangitis is essential and illustrates that precision medicine may be needed for patients with PBC at various stages of their disease process.


Results
The cytokine profiles of 2-OA-OVA immunized model. To understand the roles of Th1/IFN-γ and Th17/IL-17 in the pathogenesis of autoimmune cholangitis PBC, we first determined basal expression levels of IFN-γ , IL-4, and IL-17 in 2-OA-OVA immunized mice. As shown in Fig. 1a, IFN-γ was readily detected in serum and peaked at approximately 4 weeks post 2-OA-OVA immunization, while neither IL-17 nor IL-4 was detectable. There were no detectable levels of IFN-γ , IL-4 and IL-17 at any time points in control mice immunized with adjuvant alone (data not shown). IFN-γ mRNA expression was significantly increased in the liver of mice immunized with 2-OA-OVA, while neither IL-17 nor IL-4 was increased (Fig. 1b). We then determined cytokine expression in hepatic lymphocytes by stimulating cells with phorbol-myristate acetate (PMA) and ionomycin; 2-OA-OVA immunized mice had a higher frequency of IFN-γ expressing lymphocytes compared to the control group immunized with adjuvant without 2-OA-OVA, an increased frequency of IL-17 expressing lymphocytes was also noted, but the frequency was low in both groups. The percentage of IL-4 + in hepatic lymphocytes of both groups was also low (Fig. 1c,d). Strikingly, there was a markedly elevated frequency of IFN-γ expressing CD4 + T cells (Th1 cells) in 2-OA-OVA immunized mice, but no differences in the percentage of IFN-γ expression in hepatic CD8 + T, NK (CD3 − NK1.1 + ), and NKT (CD3 + NK1.1 + ) cells (Fig. 1e). These results suggested that the IFN-γ and Th1 mediated immune response was dominant in 2-OA-OVA immunized mice.
Increased liver inflammation in AAV-IFN-γ treated 2-OA-OVA immunized mice. To investigate the mechanism of how IFN-γ affected autoimmune cholangitis, we administered AAV-IFN-γ before 2-OA-OVA immunization. First, we examined mice for serum IFN-γ following treatment with a single intravenous injection of AAV-IFN-γ or mock virus and then immunized with 2-OA-OVA. As shown in Fig. 2a, IFN-γ was highly expressed in the serum of AAV-IFN-γ injected mice at all time points but basal expression only levels of IFN-γ were noted in controls injected with either 2-OA-OVA and mock virus or normal saline (Fig. 2a). At 5 weeks post 2-OA-OVA immunization, there was a significant increase in the titers of anti-PDC-E2 in AAV-IFN-γ treated 2-OA-OVA immunized mice (Fig. 2b). Total liver lymphocytes were increased 2.5-fold in AAV-IFN-γ treated 2-OA-OVA immunized mice (Fig. 2c). Lymphocyte subsets including T, B, NK, and NKT cells were all significantly increased. There was approximately a two-fold increase of T, B and NKT cells, and a 10-fold increase of NK cells in mice treated with IFN-γ (Fig. 2d). There was a 1.3 fold increase in CD4 + T cells but a 3-fold increase of CD8 + T cells (Fig. 2e). Of note, the numbers of professional antigen presenting cells, including B cells, macrophages and dendritic cells, in the liver were all significantly increased in AAV-IFN-γ treated 2-OA-OVA immunized mice (Fig. 2d,f). The expression of MHC class II on macrophages and B cells were also significantly increased (Fig. 2g). The frequencies of activated CD8 + T cells and NK cells were increased (Fig. 2h). In addition, the frequencies of IFN-γ secreting CD4 + and CD8 + T cells were significantly increased in AAV-IFN-γ treated 2-OA-OVA immunized mice (Fig. 2i).
At 10 weeks post 2-OA-OVA immunization, an increase in lymphocytic infiltrate in the AAV-IFN-γ treated 2-OA-OVA immunized mice could still be found. However, compared with a 2.5 fold increase at 5 weeks, there was only a 1.4 fold increase at 10 weeks (Fig. 3a). Threefold increase of NK cells and 1.5 fold increase of NKT cells were observed following IFN-γ administration. Neither T cells nor B cells were increased after 10 weeks of IFN-γ administration (Fig. 3b). However, the numbers of CD4 + memory T cells were decreased in the AAV-IFN-γ treated 2-OA-OVA immunized mice (Fig. 3c). As expected control mice immunized with 2-OA-OVA without α -GalCer as well as treated with AAV-mock and normal saline induced a slight increase of expression of collagen I and III, there were no changes in the expressions of collagen I and III in IFN-γ treatment (Fig. 4a). However, mice immunized with 2-OA-OVA and stimulated with α -GalCer had an approximate 6-fold expression of collagen I and III compared to naïve mice. 2-OA-OVA/α -GalCer immunized mice treated with AAV-IFN-γ had a lower expression of collagen I and III than controls (Fig. 4b). These results suggest that IFN-γ increased liver inflammation but subsequently led to reduced fibrosis.

IL-30 expression in the liver correlated with IFN-γ expression in the liver. IFN-γ induces IL-30
expression in macrophages and dendritic cells 13,14 and IL-30 inhibits IL-12-, IFN-γ -, and Con A-mediated hepatotoxicity by suppression of endogenous IFN-γ expression 13 . As shown in Fig. 5a, a 10-15-fold increase of IL-30 was found at 5 weeks post 2-OA-OVA immunization, while a 65-fold increase was noted in AAV-IFN-γ treated 2-OA-OVA immunization. However, there was only a 3-fold increase at 10 weeks post 2-OA-OVA immunization and a 26-fold increase in mice treated with AAV-IFN-γ . The expression level of the IL-27 subunit EBI3 was up-regulated in the liver with a similar but much lower pattern compared to p28 (IL-30) (Fig. 5a), suggesting that IL-30, not IL-27, was largely secreted following IFN-γ treatment. In addition, in 2-OA-OVA immunized mice, IL-30 expression in the liver significantly correlated with IFN-γ expression in the liver (r = 0.647, p < 0.01) (Fig. 5b). Taken together, IFN-γ induces IL-30 production which suppresses IFN-γ mediated inflammation.
Administration of IL-4 did not reduce Th1 mediated autoimmune cholangitis but markedly exacerbated liver inflammation and fibrosis. We then investigated whether the Th2 cytokine IL-4 could suppress Th1 IFN-γ mediated autoimmune cholangitis and fibrosis. We treated mice as above with AAV-IL-4 and then immunized with 2-OA-OVA/α -GalCer using the protocol outlined above. At 5 weeks post immunization, there were no increases in liver lymphocyte infiltration (Fig. 6a). However, T cells, including CD4 + and CD8 + T cells, and NK cells were significantly decreased in IL-4 administered 2-OA-OVA/α -GalCer mice. In contrast, B cells in IL-4 administered 2-OA-OVA/α -GalCer mice were significantly increased (Fig. 6b,c). Further, we noted that the numbers of macrophages and granulocytes in the liver of IL-4 administered 2-OA-OVA/α -GalCer immunized mice were greatly increased compared to either the AAV mock or the normal saline control groups (Fig. 6a). Moreover, mice injected with AAV-IL-4 developed a significant induction of fibrosis 10 weeks following immunization (Fig. 7a). Consistent with this data, there were higher expressions of collagen I and III in mice treated with AAV-IL-4 compared to the AAV-mock or normal saline controls (Fig. 7b). These results suggest that administration of IL-4 decreased T cell infiltration but increased macrophages and granulocytes and exacerbated fibrosis.
To investigate whether administration of IL-4 suppressed Th1 cells in this model, we analyzed the Th1 cells in liver at 5 weeks post 2-OA-OVA/α -GalCer immunization. Interestingly, we found that there were no differences   in the frequency of IFN-γ producing CD4 + T cells and the expression level of IFN-γ in CD4 + T cells among three groups of 2-OA-OVA/α -GalCer immunized mice. Surprisingly, the expression level of IFN-γ in CD8 + T cells was increased in IL-4 injected mice. Furthermore, the frequencies and expression level of IFN-γ in NK and NKT cells of IL-4 administered 2-OA-OVA/α -GalCer mice were significantly higher than that of controls (Fig. 8). Hence, administration of IL-4 activates CD8 + T, NK and NKT cells to secrete more IFN-γ and exacerbate cholangitis.

Discussion
The cellular immune response is critical in the natural history of pathology of PBC 3-5,15-21 . CD4 + T cells include at least three distinct subsets based on their cytokine production profile and function: Th1, Th2, and Th17. Th1 cells, characterized by the secretion of IFN-γ in response to IL-12, are a major defense in the eradication of intracellular pathogens, while Th2 cells, characterized by production of IL-4, IL-5 and IL-13, are activators of B cells for IgE production, eosinophil recruitment and mucosal expulsion mechanisms. Th17 cells, which secrete IL-17, IL-21, and IL-22, mediate host defensive mechanisms to various infections, especially extracellular bacterial infections 22 . The pathogenesis of organ-specific autoimmune diseases including multiple sclerosis, rheumatoid arthritis, type I diabetes, and Hashimoto's thyroiditis are thought to be orchestrated and augmented by Th1 and/or Th17 23,24 . In contrast Th2 cells may exert down regulation [25][26][27][28][29] .
Autoreactive CD4 + T cells have been linked to the pathogenesis of PBC. Th1 cells and their signature cytokine, IFN-γ , and Th17 cells and IL-17 are thought to be the main pathogenic mediators in PBC 1-8 . In the 2-OA-OVA mimicry model, the numbers and levels of Th1 cells and IFN-γ are significantly increased and appear to correlate with the onset of disease 30 . Overexpression of IFN-γ enhances liver inflammation by increasing immune cell infiltrates, upregulating MHC class II expression of antigen presenting cells, promoting anti-PDC-E2 production, and activating CD4 + and CD8 + T and NK cells in initiation of disease. However, IFN-γ also protects the portal IFN-γ expression is readily detected in the serum and liver of patients with PBC 1,2,6 . Genome-wide association studies reflect that Th1 related IL-12A and IL-12RB2 variants are strongly associated with PBC 7,8 and previous studies have demonstrated that deletion of IFN-γ in 2-OA-BSA-immunized mice reduced inflammatory portal infiltrates associated with prevention of bile duct damage 30 . In the study herein, we demonstrated that IFN-γ was significantly increased and IFN-γ enhanced liver inflammation in 2-OA-OVA immunization. In addition, we note that IL-12p40 −/− dnTGFβ RII mice manifest a dramatic reduction in histological autoimmune cholangitis and significant decreases in levels of intrahepatic proinflammatory cytokines 31 . These results suggest that Th1 cells and IFN-γ are the dominant contributors in the initiation of the disease by increasing numerous immune cell infiltrates, upregulating MHC class II expression of antigen presenting cells, promoting anti-PDC-E2 antibodies production, and activating CD4 + and CD8 + T cells and NK cells. Indeed our data is consistent with recent data in another murine model that mice with a deletion of the IFN-γ 3' untranslated region adenylate-uridylate-rich element have prolong and chronic expression of IFN-γ and develop a primary biliary cholangiopathy similar to PBC, which likewise proposes a key role of IFN-γ in disease initiation 32,33 .
IFN-γ is regarded traditionally as a critical proinflammatory mediator in several Th1-driven autoimmune disease model systems 23 . However, despite the expected overall proinflammatory and disease-enforcing role in some models, there is also evidence for protection against inflammation in other models [34][35][36][37][38] . The latter data is consistent our thesis that IFN-γ has a dual role in the natural history of an autoimmune disease, particularly a disease with a long natural history 39,40 . Our data also reveals a potential role of IL-30. IL-30, a subunit p28 of IL-27, has been detected in macrophages, dendritic cells, and hepatocytes 13,41 . Induction of high expression of IL-30 occurs with the coordination of activated T cells 42 . In liver injury, IL-30 suppresses the intrinsic ability of CD4 + T cells to produce IFN-γ in acute liver inflammation 13,41 and IL-30 attenuates liver fibrosis by recruiting natural-killer-like T cells to the liver to remove activated hepatic stellate cells 43 . IL-30 has also been shown to inhibit central nervous system autoimmunity via antagonizing Th1 and Th17 responses in experimental autoimmune uveitis 44,45 . More studies are needed to understand how IL-30 inhibits autoimmune cholangitis but we submit that IFN-γ promotes disease by initially activating and recruiting immune cells and thence triggers robust immunosuppressive IL-30 production which reduces the liver inflammation. Consistent with earlier data, mice immunized with 2-OA-OVA without α -GalCer did not induce collagen production and fibrosis while mice injected with 2-OA-OVA and α -GalCer manifest collagen production and fibrosis 9,10 . The administration of IFN-γ did not induce collagen production in 2-OA-OVA immunized mice. In fact, overexpression of IFN-γ decreases collagen production in 2-OA-OVA/α -GalCer immunized mice. α -GalCer is a well-defined potent and specific ligand for iNKT cell activation in both humans and mice. Upon ligation of iNKT cell receptors with α -GalCer, iNKT cells rapidly produce large amounts of cytokines, including IFN-γ and IL-4 11 . IL-4 is not normally considered to play a major role in autoimmune disease, i.e. experimental autoimmune encephalomyelitis (EAE), multiple sclerosis (MS), rheumatoid arthritis (RA), inflammatory bowel disease (IBD), or psoriasis [46][47][48] . However, systemic IL-4 immunotherapy has been shown to improve some Th1/Th17-associated autoimmune diseases [25][26][27][28][29] . In our study, IL-4 levels in 2-OA-OVA immunized autoimmune cholangitis remained at basal values. Nonetheless the administration of IL-4 increased liver inflammation with marked increases in macrophage, granulocyte and IFN-γ secreting CD8 + T, NK and NKT cell populations. Gene transfer of IL-4 in the liver of naïve mice induces hepatitis characterized by hepatocyte apoptosis and a massive macrophage infiltrate 49,50 and IL-4 administration in primates is associated with hepatic damage 51,52 . Overall our data demonstrates Th1 cells and IFN-γ are the dominant contributors in the initiation phase of this model but clearly may have different effects as the disease evolves PBC. Delineating the varying activities of IFN-γ during the course of disease provides insight into the complex role of IFN-γ in autoimmunity and illustrates the complexities of treating a disease like PBC with biologics and the concept that one size may not fit all patients. Mice were sacrificed at 5 weeks post 2-OA-OVA immunization for determining of IFN-γ , IL-4 and IL-17 in liver tissues by quantitative real-time RT-PCR analysis and cytokine expressing immune cells by flow cytometry. In a subsequent set of experiments, and based on our data on the influence of α -GalCer on fibrosis in this model [9][10][11][12] , two μ g of α -GalCer (Funakoshi, Tokyo, Japan) were injected with the first and second 2-OA-OVA immunizations. AAV-IFN-γ and AAV-IL-4 were administered to mice at 3 days before the first mimeotope immunization. Mice were sacrificed at 5 or 10 weeks post-immunization for liver histopathology, definition of mononuclear cell phenotypes, assay of IFN-γ production, and titers of AMAs.

Methods
Isolation of mRNA and real-time PCR. Total RNA from liver specimens was obtained by the TRIzol method (Invitrogen Life Technologies, Carlsbad, CA, USA). The cDNA was generated by oligonucleotide priming using High-Capacity cDNA Reverse Transcription Kits (Applied Biosystems, Foster City, CA, USA). Amplification was performed with SYBR Green MasterMix (Thermo Scientific, USA) using the 7500 Real-Time PCR System (Applied Biosystems). Results were analyzed by 2 −ΔΔCt relative quantification method and normalized to β -actin.
Liver mononuclear cell quantitation and cytokine detection. Livers were perfused with PBS containing 0.2% BSA (PBS/0.2% BSA), passed through a 100 μ m nylon mesh, and re-suspended in PBS/0.2% BSA. The parenchymal cells were removed as pellets after centrifugation at 50 g for 5 minute and the non-parenchymal cells isolated using 40% and 70% Percoll (GE HealthCare Biosciences, Quebec, Canada). After centrifugation, collected cells were washed with PBS/0.2% BSA and the viability confirmed by trypan blue dye exclusion. Subsets of liver mononuclear cells were measured by flow cytometry. Before staining cells, with a previously defined optimal dilution of monoclonal antibodies (Abs), the cells were pre-incubated with anti-CD16/32 (clone 93) to block non-specific FcRγ binding. The following mAbs were used in this study: anti-CD3, anti-CD4, anti-CD8a, anti-CD19, anti-NK1.1, anti-CD69, anti-CD44, anti-CD62L, anti-IA b , anti-CD11b, anti-CD11c, anti-F4/80, anti-Gr1, anti-Ly6G (Biolegend, San Diego, CA, USA). For cytokine detection, liver mononuclear cells were stimulated with phorbol-myristate acetate (PMA, 1 ng/ml, Sigma-Aldrich) and ionomycin (1 μ g/ml, Sigma-Aldrich) for 4 or 18 hours and IFN-γ , IL-4, or IL-17 intracellular staining was performed after cell surface staining. Stained cells were measured with a flow cytometer (BD Biosciences) and analyzed using FlowJo software (Tree Star, Inc., Ashland, OR, USA). Optimal concentrations of the mAbs were used throughout and all assays included positive and negative controls.
Histopathology. Portions of liver were excised and immediately fixed with 10% buffered formalin solution for 2 days at room temperature. Paraffin-embedded tissue sections were then cut into 4-μ m slices for Masson's trichrome staining.
Statistical analysis. Mann-Whitney U analysis was used to determine significant differences between groups. The Spearman test was used to evaluate correlations (Prism 5; Graph-Pad Software, La Jolla, CA, USA). Results are expressed as the mean ± standard error of the mean (SEM).