Journal of Autoimmunity 78 (2017) 19e28

Contents lists available at ScienceDirect

Journal of Autoimmunity

journal homepage: www.elsevier.com/locate/jautimm

Chemokine CXCR3 deficiency exacerbates murine þ autoimmune cholangitis by promoting pathogenic CD8 activation

Hong-Di Ma a, Wen-Tao Ma a, Qing-Zhi Liu a, Zhi-Bin Zhao a, Mu-Zi-Ying Liu a, Koichi Tsuneyama b, Jin-Ming Gao c, William M. Ridgway d, Aftab A. Ansari e, ** * M. Eric Gershwin f, Yun-Yun Fei g, , Zhe-Xiong Lian a, h, a Liver Laboratory, Institute of Immunology and The CAS Key Laboratory of Innate Immunity and Chronic Disease, School of Life Sciences, University of Science and Technology of China, Hefei, Anhui, China b Department of Molecular and Environmental Pathology, Institute of Health Biosciences, The University of Tokushima Graduate School, Tokushima, Japan c Department of Respiratory Disease, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences, Beijing, China d Division of Immunology, Allergy and Rheumatology, University of Cincinnati, Cincinnati, OH, USA e Department of Pathology, Emory University, Atlanta, GA, USA f Division of Rheumatology, Allergy and Clinical Immunology, University of California at Davis School of Medicine, Davis, CA, USA g Department of Rheumatology, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences, Beijing, China h Innovation Center for Cell Signaling Network, Hefei National Laboratory for Physical Sciences at Microscale, Hefei, Anhui, China article info abstract

Article history: CXC Receptor 3 (CXCR3) is functionally pleiotropic and not only plays an important role in Received 18 October 2016 , but also participates in T cell differentiation and may play a critical role in inducing and Received in revised form maintaining immune tolerance. These observations are particularly critical for autoimmune cholangitis 1 December 2016 in which CXCR3 positive T cells are found around the portal areas of both humans and mouse models of Accepted 1 December 2016 primary biliary cholangitis (PBC). Herein, we investigated the role of CXCR3 in the pathogenesis of Available online 24 January 2017 autoimmune cholangitis. We have taken advantage of a unique CXCR3 knockout dnTGFbRII mouse to focus on the role of CXCR3, both by direct observation of its influence on the natural course of disease, as Keywords: Interferon-gamma induced well as through adoptive transfer studies into Rag / mice. We report herein that not only do CXCR3 fi Primary biliary cholangitis de cient mice develop an exacerbation of autoimmune cholangitis associated with an expanded effector þ CD8þT cell gene expression profile number, but also selective adoptive transfer of CXCR3 deficient CD8 T cells induces þ T-bet autoimmune cholangitis. In addition, gene microarray analysis of CXCR3 deficient CD8 T cells reveal an KLRG1 intense pro-inflammatory profile. Our data suggests that the altered gene profiles induced by CXCR3 þ deficiency promotes autoimmune cholangitis through pathogenic CD8 T cells. These data have signif- icance for human PBC and other autoimmune liver diseases in which therapeutic intervention might be directed to chemokines and/or their receptors. © 2017 Elsevier Ltd. All rights reserved.

Abbreviations: PBC, Primary biliary cholangitis; dnTGFbRII, TG, dominant negative transforming growth factor b receptor II; CXCR3, CXC 3; WT, wide type; TGC3, dnTGFbRII CXCR3-/-; MNCs, mononuclear cells; CTL, cytotoxic ; APCs, -presenting cells; DCs, dendritic cells; IFN-g, interferon-g; TNF-a, tumor necrosis factor alpha; IL-6, -6; MIG, monokine induced by gamma-interferon; IP-10, interferon-induced of 10 kDa; I-TAC, interferon-inducible T cell alpha chemoattractant; CXCL9, chemokine (C-X-C motif) ligand 9; CXCL10, chemokine (C-X-C motif) ligand 10; ELISA, enzyme-Linked Immunosorbent Assay; AMAs, anti-mito- chondrial antibodies; Th1, T helper 1; Tem, effector memory T cells; KLRG1, killer cell lectin-like receptor G1; CAMs, cell adhesion molecules; KEGG, Kyoto Encyclopedia of Genes and Genomes; GPCRs, G proteinecoupled receptors. * Corresponding author. Liver Immunology Laboratory, Institute of Immunology and The CAS Key Laboratory of Innate Immunity and Chronic Disease, School of Life Sciences and Medical Center, University of Science and Technology of China, Hefei, 230027, China. ** Corresponding author. Department of Rheumatology, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences, Beijing, 100032, China. E-mail addresses: [email protected] (Y.-Y. Fei), [email protected] (Z.-X. Lian). http://dx.doi.org/10.1016/j.jaut.2016.12.012 0896-8411/© 2017 Elsevier Ltd. All rights reserved. 20 H.-D. Ma et al. / Journal of Autoimmunity 78 (2017) 19e28

1. Introduction Kusatsu, Shiga, Japan). The expression levels of target genes (Cxcl9 and Cxcl10) were normalized to the housekeeping gene b2- CXCR3 is a chemokine receptor highly expressed on effector T microglobulin and the results calculated by DDCt [17]. Known cells and critically involved in both T cell trafficking and the dif- standards were included throughout. Primer sequences were based ferentiation of T cells into distinct functional subsets [1]. CXCR3 is on Primer-Bank (http://pga.mgh.harvard.edu/primerbank) and upregulated following CTL and Th1 cell differentiation [2]. CXCL9 blasted to confirm the target genes using Primer-Blast (http:// (MIG), CXCL10 (IP-10), and CXCL11 are the natural ligands of CXCR3 www.ncbi.nlm.nih.gov/tools/primer-blast). The primer sequences [3e5]. CXCR3 is functionally pleiotropic; it plays an important role used were as follows:b2-microglobulin, 50- CCGAACA- in the chemotaxis of Tregs to sites of inflammation following TACTGAACTGCTACGTAA -30 and 50- CCCGTTCTTCAGCATTTGGA -3’; binding [6,7], but also participates in T cell differentiation [5,8]. Cxcl9, 50- AATGCACGATGCTCCTGCA -30 and 50- AGGTCTTTGAGG- CXCR3 expression on T cells alters the balance between effector and GATTTGTAGTGG -3’; Cxcl10, 50-GCCGTCATTTTCTGCCTCA-30 and 50- þ memory CD8 T cell generation [9]. Finally, the CXCL11-CXCR3 CGTCCTTGCGAGAGGGATC-3’. The primers were synthesized by chemokine axis polarizes naive T cells and repolarizes effector Sangon Biotech (Shanghai, China). þ CD4 T cells into IL-10hi Tr1 cells to induce an immune-tolerizing state [10]. Our lab has been studying the pathogenesis of human 2.3. Enzyme-linked immunosorbent assays (ELISA) PBC and during the course of these studies reported significant increases in the levels of soluble IP-10 (CXCL10) and MIG (CXCL9) in TG and TGC3 mice livers were homogenized in PBS and centri- PBC and also a marked increase in the frequency of CXCR3 fuged for 5 min at 5000 g; supernatants were assayed using a expressing T cells in both blood and liver [11]. An epigenetic study CXCL9/MIG Quantikine ELISA Kit or Mouse CXCL10/IP-10/CRG-2 of DNA from PBC revealed aberrant demethylation of the CXCR3 Quantikine ELISA Kit (R&D Systems, Minneapolis, MN, US). Serum þ promoter in CD4 T cells [12]. antiePDC-E2 (AMA) was measured by ELISA with purified recom- DnTGFbRII mice, transgenic for directed expression of a binant PDC-E2 as previously described [18] and including positive dominant-negative form of TGFb receptor type II (dnTGFbRII), un- and negative controls. der the direction of the CD4 promoter [13], develop autoimmune cholangitis [14]. DnTGFbRII mice mimic several key phenotypic 2.4. Serum features of human PBC, including spontaneous production of AMAs, lymphocytic liver infiltration with periportal inflammation and an Serum concentration of IFN-g, TNF-a and IL-6 from TG and TGC3 inflammatory profile. Furthermore, using adoptive trans- mice were quantitated using a mouse Th1/Th2/Th17 CBA (BD þ fer studies we have demonstrated that the CD8 T cell is pathogenic Biosciences, Franklin Lakes, NJ, US), using FACSVerse flow cytom- effector population [15]. In this study, we have advanced our work eter (BD Bioscience). Data were analyzed by FCAP Array™ v3.0.1 by developing CXCR3 gene deleted dnTGFbRII (TG) mice. We report (BD Bioscience) Software. herein that knocking-out CXCR3 significantly worsens autoimmune cholangitis and that the mechanism is secondary to generation of a 2.5. Liver tissue preparation and histological scoring þ hyper-activated CD8 T cells which promote biliary pathology. These data are significant for human autoimmune disease in which Liver from TG and TGC3 mice were fixed with 4% para- blockade of CXCR3 or its ligands is proposed. formaldehyde, embedded in paraffin and cut into 4 mm sections for hematoxylin-eosin (H&E) staining. Liver portal inflammation 2. Materials and methods scores were “blindly” performed on H&E-stained liver sections using a set of 4 indices by a “blinded” pathologist (K.T.), based on 2.1. Mice the levels of inflammation surrounding middle-large portal tract; the indices were scored as 0, none; 1, mild; 2, moderate; and 3, dnTGFbRII (TG) mice on a C57BL/6 back-ground (B6.Cg-Tg(Cd4- severe inflammation. TGFBR2)16Flv/J) [13] were initially derived from Jackson Labora- tory. The maintenance and genetic monitoring of this colony has 2.6. Preparation of hepatic mononuclear cells and flow cytometric been previously reported [14]. CXCR3 knockout mice on a C57BL/6 analysis background were previously established by gene targeting [16] and were kindly provided by Dr. Bao Lu (Harvard Medical School, Bos- Hepatic mononuclear cells were isolated as described [19].For ton, MA, USA). dnTGFbRII CXCR3-/- (TGC3) mice were obtained by flow cytometry analysis, hepatic mononuclear cells were blocked selectively backcrossing TG mice with CXCR3 knockout mice. 8e14 by incubation with anti-mouse CD16/32 (Biolegend, San Diego, CA, week old female TGC3 mice were used in all experiments. B6/ US) and stained with fluorochrome-conjugated antibodies (Abs) at Rag1-/- mice (Ly5.2) were obtained from Jackson Laboratory. 8e9 4 C in PBS with 0.2% bovine serum albumin for 20 min. All þ week old female Rag1-/- mice were used as recipients in our CD8 T fluorochrome-conjugated Abs, unless otherwise noted, were pur- þ adoptive transfer model. All mice were housed in a specific chased from Biolegend. To identify liver CD8 T cells and different T pathogen-free and controlled environment (22 C, 55% humidity, cell subsets, hepatic mononuclear cells from TG and TGC3 mice and 12-h day/night rhythm) and care provided according to the were stained with PacificBlue-CD3 (17A2), APC-Cy7-CD8a (19517), regulations of animal care at University of Science and Technology V500-CD4 (RM4-5, BD Bioscience), PE-Cy7-NK1.1 (PK136), PE- of China (Hefei, Anhui, China). CXCR3 (CXCR3-173), FITC-CD44 (IM7), PerCP/Cy5.5-CD62L (MEL- 14), APC-KLRG-1 (2F1/KLRG-1). Intracellular transcriptional factor 2.2. RNA preparation, reverse transcription and quantitative real- T-bet was stained by PerCP/Cy5.5-T-bet (4B10) and Alexa 647- time PCR Foxp3 (MF14). Normal IgG isotype controls (Biolegend) were used as controls. Total RNA was extracted from the liver of wide type B6 and TG For intracellular cytokine staining, hepatic mononuclear cells mice using Trizol (Invitrogen, Carlsbad, CA, US). M-MLV Tran- were stimulated in vitro with Cell Stimulation Cocktail (plus protein scriptase (Invitrogen) was used for reverse transcription. Quanti- transport inhibitors) (eBioscience, San Diego, CA, US) in RPMI-1640 tative real-time PCR was performed using Premix Ex Taq (Takara, (Life Technology, Carlsbad, CA, US) with 10% fetal bovine serum (GE H.-D. Ma et al. / Journal of Autoimmunity 78 (2017) 19e28 21

Healthcare Life Sciences, Logan, UT, US) at 37 C, 5% CO2 for 4 h. Cxcl10 P ¼ 0.0060) and the protein level (P < 0.0001). These Cells were thence fixed and permeabilized with BD Cytofix/Cyto- increased chemokine levels were associated with increased fre- perm™ Fixation/Permeabilization Kit (BD Bioscience), and stained quencies (Fig. 1D,E and Fig. S1) and absolute numbers of CXCR3 þ þ with PE-IFN-g (XMG1.2). Stained cells were analyzed using a expressing CD8 T(P< 0.0001) and CD4 T cells (P < 0003) in the FACSVerse (BD Bioscience) flow cytometer. Flow data were calcu- liver of 16 week old female TG mice as compared with WT litter- lated using Flow Jo software (Tree Star, Ashland, OR, US). mates as determined by flow cytometry (P < 0.0001) (Fig. 1C). In addition, there was an age associated increase in the frequencies þ 2.7. Adoptive transfer (Fig. 1D,E and Fig. S1) and numbers of both CXCR3 positive CD8 T þ (P < 0.0001) and CD4 T(P¼ 0.007) cells (Fig. 1C). Adoptive transfer was performed as described by our laboratory [15]. Splenic cells were collected from 7 to 8 week old female TG 3.2. CXCR3 deficiency promotes autoimmune cholangitis þ and TGC3 mice and CD8 T cells were purified by positive selection with anti-CD8 microbeads (Miltenyi Biotech, Bergisch Gladbach, Histological studies of liver from 13 to 14 week old female Germany). 8e9 week old female Rag1-/- mice were used as re- dnTGFbRII CXCR3-/- (TGC3) mice revealed a significant increase of þ cipients. Specifically, 1 106 purified CD8 T cells were injected i.v. inflammatory infiltrates within the portal tract (Fig. 2A) compared into Rag1-/- recipients. 8 weeks after transfer, the livers from re- with age-matched WT dnTGFbRII (P ¼ 0.0005, Fig. 2C). The cipients were harvested and the total number of hepatic mono- increased total number of hepatic mononuclear cells in TGC3 mice þ nuclear cells were calculated and donor-derived CD8 T cells were correlated with histological score (P ¼ 0.0011, Fig. 2B). Further, analyzed by flow cytometer. deletion of the CXCR3 gene led to the spontaneous development of elevated levels of anti-mitochondrial Abs (AMAs)-PDC-E2 2.8. Gene-expression profiling analysis of liver CD8þT cells (P ¼ 0.0370) in TGC3 mice (OD of PDC-E2: 0.301 ± 0.152) compared with age matched TG mice (OD of PDC-E2: 0.180 ± 0.047) (Fig. S2A). þ CD8 T cells from the livers of 13e14 week old female TG and The increased inflammatory response in the liver of TGC3 mice was TGC3 mice were first enriched by positive selection with anti-CD8 associated with increased levels of the inflammatory cytokines IL-6, magnetic microbeads (Miltenyi Biotech) and then purified by IFN-g and TNF-a compared with age matched TG mice (IL-6 þ FACS Aria (BD Bioscience). The CD8 T cell sorting purity was P ¼ 0.0153, IFN-g P ¼ 0.0157, TNF-a P ¼ 0.0275, Fig. 2D). þ greater than 95%. Sorted CD8 T cells were suspended in Trizol (Invitrogen). Total RNA was extracted and purified using an miR- 3.3. CXCR3 deficiency expands the effector memory T cell subset Neasy Mini Kit (Cat.#217004, QIAGEN, GmBH, Germany) and with pathogenic potential checked for a RIN number to inspect RNA integration by an Agilent þ þ Bioanalyzer 2100 (Agilent technologies, Santa Clara, CA, US). Total The levels of absolute numbers of total, CD4 and CD8 T cells RNA was amplified and labeled by Low Input Quick Amp Labeling were all increased in liver from TGC3 mice compared with age Kit, One-Color (Cat.# 5190-2305, Agilent technologies). Labeled matched TG mice (Total T cell P ¼ 0.0111, CD8þTP¼ 0.0143, CD4þT þ þ cRNA, using an RNeasy mini kit (Cat.# 74106, QIAGEN), were hy- P ¼ 0.0249), Fig. 3A and B. Subset analysis of the CD8 and CD4 T bridized with a Gene Expression Hybridization Kit (Cat.# 5188- cells in livers of TGC3 mice revealed a selective increase in the 5242, Agilent technologies). Slides were scanned using an Agilent effector memory population (CD8Tem, P ¼ 0.0007; CD4Tem, Microarray Scanner (Cat.#G2565CA) and data extracted with P ¼ 0.0308) compared with age-matched TG mice (Fig. 3C). How- Feature Extraction software 10.7. Raw data were normalized using ever, the frequency of Foxp3þ Treg cells did not significantly the Quantile algorithm, Gene Spring Software 12.6.1. The tran- change (Fig. S3A). Further, there were higher percentages (Fig. 4A) þ scription profile chip service was provided by Shanghai Biotech- and absolute numbers (Fig. S2B) of KLRG1 expressing CD8 T and þ nology Cooperation (Shanghai, China). Gene classification was CD4 T cells in TGC3 mice. These data suggest that CXCR3 defi- þ based on the annotation in the Kyoto Encyclopedia of Genes and ciency in TGC3 mice led to selective differentiation of CD8 T and þ þ Genomes (KEGG) data base. Heat map and red-green scale schemes CD4 T cells into cytotoxic KLRG1þCD8 T(P¼ 0.0016) and þ were designed using Multiple Experiment Viewer 4.8 software. The terminally differentiated KLRG1þCD4 T(P< 0.0001) cells (Fig. 4B). GEO accession number of the data is GSE84218. The access to the These phenotypic changes were associated with increased levels of þ þ GEO data is http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi? T-bet and IFN-g expressing CD8 T and CD4 T cells in the livers of token¼ovoleiikzhiprad&acc¼GSE84218. TGC3 mice compared with age-matched TG mice (Figs. S3C and 4D). þ The relative fluorescence intensity (RFI) of T-bet (CD8 TP< 0.0001, þ 2.9. Statistical analysis CD4 TP¼ 0.0001, Fig. 4C) and the frequencies of IFN-gþ pro- þ þ ducing (CD8 TP< 0.0001, CD4 TP¼ 0.0002, Fig. 4E) hepatic þ þ All data were presented as the mean ± standard deviation (SD). CD8 T and CD4 T cells in TGC3 mice were significantly increased The significance of differences was determined using a two-tailed compared to TG mice. unpaired Student's t-test in GraphPad Prism. All experiments þ were replicated at least three times. The significance levels are 3.4. CXCR3 deficient CD8 T cells promote autoimmune cholangitis marked * P < 0.05; **P < 0.01; ***P < 0.001. þ 1 106 splenic CD8 T cells from TG mice or TGC3 mice were 3. Results intravenously transferred into Rag1-/- recipients (Fig. 5A). 8 weeks later, livers from recipients were harvested and mononuclear cells þ 3.1. Increased CXCR3 ligands are associated with increased analyzed for the absolute numbers of different donor derived CD8 frequencies of hepatic CXCR3 expressing T cells T cells by flow cytometry. As seen in Fig. 5B, the adoptive transfer of þ CD8 T cells from TGC3 mice led to a marked increase in cellular þ As noted in Fig. 1A and B, the levels of Cxcl9 and Cxcl10 were infiltrates compared with adoptive transfer of CD8 T cells from TG significantly upregulated in the liver of 12e14 week old female mice (P ¼ 0.0244). Phenotypic analysis of the cellular infiltrates dnTGFbRII (TG) mice compared with age and sex-matched wild demonstrated that the predominant phenotype of the donor in- þ type (WT) B6 littermates both at the message (Cxcl9 P ¼ 0.0026, filtrates were composed of CD8 T cells (P ¼ 0.0242, Fig. 5C) that 22 H.-D. Ma et al. / Journal of Autoimmunity 78 (2017) 19e28

Fig. 1. CXCR3 ligands-CXCL9, CXCL10-and CXCR3 expressing T cell number increased in liver of autoimmune cholangitis model. (A) mRNA levels of Cxcl9, Cxcl10 in liver of control B6 (WT) mice (n ¼ 5) and dnTGFbRII (TG) mice (n ¼ 5). (B) CXCL9 and CXCL10 levels in liver homogenates were determined by ELISA in WT (n ¼ 8) and TG (n ¼ 7) mice. (C) þ þ þ þ Number of CXCR3 expressing CD8 T cells (CXCR3þCD3 CD8þNK1.1-) and CXCR3 expressing CD4 T cells (CXCR3þCD3 CD4þNK1.1-) in liver of 16 week old WT mice (n ¼ 4), 8 þ þ week old TG mice (n ¼ 5) and 16 week old TG mice (n ¼ 5) by flow cytometry. Representative FACS plots show CXCR3 expression patterns of CD8 T (D) and CD4 T cells (E). Graphs reflect mean ± SD. **P < 0.01, ***P < 0.001.

þ expressed the memory phenotype (P ¼ 0.0224, Fig. 5D and Fig. S4A) Rag1-/- recipients of CD8 T cells from TG mice (P ¼ 0.0049, and increased frequencies of IFN-g producing cells compared with Fig. 5E). H.-D. Ma et al. / Journal of Autoimmunity 78 (2017) 19e28 23

Fig. 2. CXCR3 deficiency exacerbates autoimmune cholangitis. (A) H&E-stained livers of female TG and dnTGFbRII CXCR3-/- (TGC3) mice at 13e14 weeks of age. (B) Absolute number of hepatic mononuclear cells from TG mice (n ¼ 7), TGC3 mice (n ¼ 7). (C) Scores of portal inflammation in TG (n ¼ 7) and TGC3 mice (n ¼ 9). (D) Concentrations of inflammatory cytokines in serum of TG (n ¼ 10) and TGC3 mice (n ¼ 8). Graphs reflect mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001.

þ 3.5. CXCR3 deficient CD8 T cells display a pro-inflammatory and cells from CXCR3 deficient TG mice (Fig. 6A). The lack of CXCR3 led hyper-activated profile to the over-expression of a number of pro-inflammatory cytokines and cytokine receptors including Cxcl14, Tnfrsf25, Ifng and Csf2rb Based on the data we obtained we next focused on potential (Fig. 6A). The genes related to cell cycling were also upregulated in þ þ þ intrinsic changes of CD8 T cells. Gene expression profiles of CD8 T CD8 T cells from CXCR3 deficient TG mice (Fig. 6B). Finally CXCR3 cells isolated from livers of both the TG and TGC3 mice were deficiency significantly suppressed the activity of leukocyte trans- analyzed by gene microarray. Genes coding for cytokine-cytokine endothelial migration and downregulated the expression of che- þ receptor interaction related genes were over-expressed in CD8 T mokine receptors (Fig. 6C). 24 H.-D. Ma et al. / Journal of Autoimmunity 78 (2017) 19e28

þ þ þ Fig. 3. CXCR3 deficiency increases effector memory T cell numbers. Total cell numbers of T (A), CD8 and CD4 cells (B), CD8Tem (CD3þNK1.1-CD8 CD44 þCD62L-) and CD4Tem þ (CD3þNK1.1-CD4 CD44 þCD62L-) cells (C) in liver of TG (n ¼ 7) and TGC3 (n ¼ 7) mice by flow cytometry. Graphs reflect mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001.

þ 4. Discussion profile. Our proposed model is illustrated in Fig. 7 reflecting CD8 T cell activation in the presence (up) and absence (down) of CXCR3 in Primary biliary cholangitis (PBC) is a progressive autoimmune autoimmune cholangitis. When CXCR3 is absent, the frequency of þ liver disease characterized by portal inflammation and immune- KLRG1þ terminal differentiated effector memory CD8 T cells is mediated destruction of the intrahepatic bile ducts [20e24]. The upregulated, the expression of T-bet is increased and IFN-g pro- þ etiology of PBC remains enigmatic, but is linked by a combination of duction enhanced in activated CD8 T cells. At the same time, þ genetic susceptibility and environmental factors [25e28]. Clearly, CXCR3 deficiency promotes the proliferation of CD8 T cells, which the immune response and biliary destruction is mediated by a results in more severe liver inflammation. multi-lineage innate and adaptive immune response [20,23,29].In Chemokines and chemokine receptors guide the migration of both humans and mouse models, there is a specific enhancement of effector cells into inflammatory sites, which are critical for the þ þ autoantigen-specific CD4 T and CD8 T cells [15,23,30,31].We pathogenesis of autoimmune diseases [3,4,33,34]. However, the have also recently reported that intrahepatic terminally differen- observation in TGC3 mice that inflammatory cells accumulate þ tiated (KLRG1þ) CD8 T cells from dnTGFbRII mice are cytotoxic to within the inflammatory sites of liver in the absence of CXCR3 þ cholangiocytes in vitro [32].Definition of the earliest events that suggests that the pathogenic role of CXCR3 deficient CD8 T cells lead to biliary destruction will be critical for the identification of does not require the chemotaxis of CXCR3 and is likely mediated therapeutic targets. either by other redundant chemokine(s) mediated chemotaxis or We demonstrate herein that CXCR3 deficiency promotes the due to the absence of inhibitory signals normally provided by þ activation and proliferation of pathogenic CD8 T cells and is CXCR3 ligation. Thus, instead of reducing the inflammatory infil- accompanied by intrinsic alteration of the T cell gene expression tration, CXCR3 deficiency aggravates autoimmune cholangitis by H.-D. Ma et al. / Journal of Autoimmunity 78 (2017) 19e28 25

þ þ Fig. 4. CXCR3 deficiency promotes T cell activation. (A) Representative FACS plots demonstrate KLRG1 expression patterns of CD8 T (upper panel, gated on CD3þNK1.1-CD8 þ þ þ þ hepatic MNCs) and CD4 T cells (lower panel, gated on CD3þNK1.1-CD4 hepatic MNCs) from TG and TGC3 mice. (B) The frequency of KLRG1þ cells in liver CD8 T and CD4 T cells þ þ from TG (n ¼ 7) and TGC3 (n ¼ 7) mice are shown. (C) Relative fluorescence intensity (RFI ¼ MFI/MFIiso) of T-bet in liver CD8 and CD4 T cells from TG (n ¼ 4) and TGC3 (n ¼ 5) þ þ mice by intracellular staining. (D) Representative FACS plots demonstrating IFN-g production in liver CD8 T and CD4 T cells from TG and TGC3 mice. (E) Percentage of IFN-g þ þ producing cells in liver CD8 T and CD4 T cells from TG (n ¼ 4) and TGC3 (n ¼ 5) mice. Graphs reflect mean ± SD. **P < 0.01, ***P < 0.001.

þ promoting pathogenic CD8 T cell activity and proliferation. It is 10hi (Tr1) and IL-4hi (Th2) cells, mediated via p70 kinase/mTOR in generally thought that chemokines participate in the process of T STAT3-and STAT6-dependent pathways [10]. cell differentiation two ways. First, APCs, like DCs, may use che- CXCR3 deficiency expands effector memory T cell number and mokines to promote encounters with antigen-specific T cells promotes their pathogenic potential. Gene microarray data dem- þ [35,36]. Second, chemokines expressed in specific lymphoid com- onstrates that CXCR3 deficient CD8 T cells differentially display an partments may influence more globally the positioning of T cells in increased number of pro-inflammatory genes that likely contrib- particular microenvironments, to bring these cells in contact with uted to hyper-activation and proliferation, suggesting that CXCR3 þ the appropriate APCs or accessory cells important for their differ- deficiency affects CD8 T cell activation and proliferation in an entiation [9,37,38]. Chemokine receptors are 7-transmembrane intrinsic way. We hypothesize that CXCR3 deficiency alters the þ GPCRs, and thus may transmit diverse signaling cascades upon signaling pathway in CD8 T cells associated with their activation binding to different ligands to determine T cell fate [10]. Specif- and proliferation. Further, gene profile analysis suggests that a lack ically, CXCL10/CXCR3 interactions drive effector Th1 polarization of CXCR3 leads to the over-expression of pro-inflammatory cyto- via STAT1, STAT4, and STAT5 phosphorylation, while CXCL11/CXCR3 kines and cytokine receptors, i.e. Cxcl14, Tnfrsf25, Ifng and Csf2rb binding induces an immune-tolerizing state characterized by IL- (Fig. 6A), and influences the activation of key signaling pathways in 26 H.-D. Ma et al. / Journal of Autoimmunity 78 (2017) 19e28

þ þ Fig. 5. CXCR3 deficient CD8 T cells promote autoimmune cholangitis in adoptive transfer. (A) Schematic diagram shows the strategy of CD8 T cell transfer. (B) The total þ þ number of liver mononuclear cells from TG derived CD8 T (TG-CD8T) recipients (n ¼ 4) and TGC3 derived CD8 T (TGC3-CD8T) recipients (n ¼ 4) was compared. The absolute þ þ numbers of liver CD8 T cells (C), CD8 Tem cells (D) and IFN-g producing ability of liver CD8 T cells (E) in TG-CD8T recipients (n ¼ 4) and TGC3-CD8T recipients (n ¼ 4) by flow cytometry.

þ CD8 T cells. These effects promote the pathogenic character and inhibits migration and proliferation and induces [44]. þ increases the number of CD8 T cells. Since one form appears to confer cell activation and the other ap- þ Since CXCR3 deficiency promotes pathogenic CD8 T cell ac- pears to inhibit cellular proliferation, it is possible that the inhibi- tivity and proliferation, why is there a high level of CXCR3 and its tory form of CXCR3 is critical for normal liver homeostasis and ligands in not only the murine models of PBC but also patients with contributes to tolerance induction. In its absence the migratory þ PBC? We note that CD8 T cells accumulate in the livers of TG mice cells mediate effector function unchecked. Thus, one needs to þ and in the CD8 T cell transfer model even without CXCR3 distinguish trafficking properties (chemotactic function) from þ expression, implying that the infiltration of pathogenic liver CD8 T signaling and inhibitory/activation function. cells may be due to the activation and proliferation of intrahepatic þ CD8 T cells unique to the specific microenvironment in PBC. þ 5. Conclusion Activated CD8 T cells expand in PBC liver and produce more IFN-g and therefore a positive feedback loop for inflammation. The þ We report herein that CXCR3 deficient TG mice have a signifi- upregulation of CXCR3 expression on CD8 T cells by inflammation þ cant enhancement of autoimmune cholangitis. Further, the mech- suppresses further activation and proliferation of CD8 T cells in þ þ anism of this aggravation is secondary to hyper-activated CD8 T this feedback loop. Thus, CXCR3 expression on CD8 T cells be- cells which promote biliary pathology. Our data has significance not comes an important regulator of liver inflammation. only in human PBC, but in other autoimmune liver diseases in Clearly our data supports the view that CXCR3 is functionally which therapeutic intervention of chemokines and/or their re- pleiotropic. CXCR3 does exist in allelic forms. Human CXCR3 has ceptors are contemplated. three splice variants: CXCR3-A, CXCR3-B and CXCR3-alt [39,40]; CXCR3 ligands have distinct and non-redundant biological roles and likely explain the diversity of cellular behavior triggered by Financial support CXCR3 activation [41]. Activation of CXCR3-A appears to induce chemotaxis and proliferation [42,43], while CXCR3-B activation Financial support provided by the National Natural Science Foundation of China (81130058, 81430034, 81401336, 91542123), H.-D. Ma et al. / Journal of Autoimmunity 78 (2017) 19e28 27

þ Fig. 6. Gene expression profile of CXCR3 deficient CD8 T cells. Differentially expressed genes from TG-CD8T (left) and TGC3-CD8T (right) cells were classified by functional category. Heat maps reflect signal values of the listed genes. Cytokine- interaction related genes are shown in panel (A). Cell cycle associated genes are displayed in panel (B). Genes involved in cell chemotaxis and migration including leukocyte transendothelial migration (up), chemokine signaling pathway (middle) and cell adhesion molecules (CAMs) (bottom) are exhibited in panel (C).

National Basic Research Program of China (973 Program- 2013CB944900), Research Fund for the Doctoral Program of Higher Education of China (RFDP 20133402110015) and National Institutes of Health, grant DK090019 (MEG).

Conflict of interest

The authors no conflict of interest.

Author contributions

Hong-Di Ma, Yun-Yun Fei and Zhe-Xiong Lian designed the ex- periments. Hong-Di Ma conducted most of the experiments and analyzed the data. Wen-Tao Ma, Qing-Zhi Liu, Zhi-Bin Zhao and Mu- Zi-Ying Liu contributed to some of the experiments. Koichi Tsu- neyama scored the H&E-stained liver sections. Jin-Ming Gao pro- vided important experiment materials. William M. Ridgway, Aftab þ Fig. 7. CXCR3 deficiency promoted pathogenic CD8 T cell proliferation and ac- A. Ansari and Eric M. Gershwin helped to discuss about research þ tivity to aggravate autoimmune cholangitis. A schematic diagram of CD8 T cell design and manuscript writing. Hong-Di Ma, Eric M. Gershwin and activation in the presence (up) and absence (down) of CXCR3 in autoimmune chol- angitis. When CXCR3 is absent, the frequency of KLRG1þ terminal differentiated Zhe-Xiong Lian wrote the manuscript. þ effector memory CD8 T cells is upregulated, the expression of T-bet is increased and þ the ability of IFN-g production is enhanced in activated CD8 T cells. At the same time, þ CXCR3 deficiency also promotes the proliferation of CD8 T cells, which exacerbates Acknowledgments inflammation. The authors are grateful for Professors Craig Gerard and Bao Lu for providing CXCR3 knockout mice. 28 H.-D. Ma et al. / Journal of Autoimmunity 78 (2017) 19e28

Appendix A. Supplementary data [20] M.M. Kaplan, M.E. Gershwin, Primary biliary cirrhosis, N. Engl. J. Med. 353 (2005) 1261e1273. [21] S. Shimoda, F. Ishikawa, T. Kamihira, A. Komori, H. Niiro, E. Baba, et al., Supplementary data related to this article can be found at http:// Autoreactive T-cell responses in primary biliary cirrhosis are proinflammatory dx.doi.org/10.1016/j.jaut.2016.12.012. whereas those of controls are regulatory, Gastroenterology 131 (2006) 606e618. [22] A.K. Abbas, J. Lohr, B. Knoechel, Balancing autoaggressive and protective T cell References responses, J. Autoimmun. 28 (2007) 59e61. [23] G.M. Hirschfield, M.E. Gershwin, The immunobiology and pathophysiology of [1] J.R. Groom, A.D. Luster, CXCR3 in T cell function, Exp. cell Res. 317 (2011) primary biliary cirrhosis, Annu. Rev. pathology 8 (2013) 303e330. 620e631. [24] Y. Sun, K. Haapanen, B. Li, W. Zhang, J. Van de Water, M.E. Gershwin, Women [2] S.K. Bromley, T.R. Mempel, A.D. Luster, Orchestrating the orchestrators: che- and primary biliary cirrhosis, Clin. Rev. Allergy Immunol. 48 (2015) 285e300. mokines in control of T cell traffic, Nat. Immunol. 9 (2008) 970e980. [25] K.D. Lindor, M.E. Gershwin, R. Poupon, M. Kaplan, N.V. Bergasa, E.J. Heathcote, [3] G.S. Campanella, B.D. Medoff, L.A. Manice, R.A. Colvin, A.D. Luster, Develop- et al., Primary biliary cirrhosis, Hepatology 50 (2009) 291e308. ment of a novel chemokine-mediated in vivo T cell recruitment assay, [26] G.J. Webb, K.A. Siminovitch, G.M. Hirschfield, The immunogenetics of primary J. Immunol. methods 331 (2008) 127e139. biliary cirrhosis: a comprehensive review, J. Autoimmun. 64 (2015) 42e52. [4] G.S. Campanella, J. Grimm, L.A. Manice, R.A. Colvin, B.D. Medoff, [27] G.J. Webb, G.M. Hirschfield, Using GWAS to identify genetic predisposition in G.R. Wojtkiewicz, et al., Oligomerization of CXCL10 is necessary for endo- hepatic autoimmunity, J. Autoimmun. 66 (2016) 25e39. thelial cell presentation and in vivo activity, J. Immunol. 177 (2006) [28] Y.H. Wang, W. Yang, J.B. Yang, Y.J. Jia, W. Tang, M.E. Gershwin, et al., Systems 6991e6998. biologic analysis of T regulatory cells genetic pathways in murine primary [5] I. Salomon, N. Netzer, G. Wildbaum, S. Schif-Zuck, G. Maor, N. Karin, Targeting biliary cirrhosis, J. Autoimmun. 59 (2015) 26e37. the function of IFN-gamma-inducible protein 10 suppresses ongoing adjuvant [29] M.E. Gershwin, I.R. Mackay, The causes of primary biliary cirrhosis: conve- arthritis, J. Immunol. 169 (2002) 2685e2693. nient and inconvenient truths, Hepatology 47 (2008) 737e745. [6] A. Erhardt, C. Wegscheid, B. Claass, A. Carambia, J. Herkel, H.W. Mittrucker, et [30] H. Kita, S. Matsumura, X.S. He, A.A. Ansari, Z.X. Lian, J. Van de Water, et al., al., CXCR3 deficiency exacerbates liver disease and abrogates tolerance in a Quantitative and functional analysis of PDC-E2-specific autoreactive cytotoxic mouse model of immune-mediated hepatitis, J. Immunol. 186 (2011) T in primary biliary cirrhosis, J. Clin. investigation 109 (2002) 5284e5293. 1231e1240. [7] R. Bonecchi, E. Galliera, E.M. Borroni, M.M. Corsi, M. Locati, A. Mantovani, [31] S. Shimoda, H. Miyakawa, M. Nakamura, H. Ishibashi, K. Kikuchi, H. Kita, et al., Chemokines and chemokine receptors: an overview, Front. Biosci. 14 (2009) CD4 T-cell autoreactivity to the mitochondrial autoantigen PDC-E2 in AMA- 540e551. negative primary biliary cirrhosis, J. Autoimmun. 31 (2008) 110e115. [8] M. Thapa, D.J. Carr, CXCR3 deficiency increases susceptibility to genital herpes [32] W. Huang, K. Kachapati, D. Adams, Y. Wu, P.S. Leung, G.X. Yang, et al., Murine simplex virus type 2 infection: uncoupling of CD8þ T-cell effector function but autoimmune cholangitis requires two hits: cytotoxic KLRG1(þ) CD8 effector not migration, J. virology 83 (2009) 9486e9501. cells and defective T regulatory cells, J. Autoimmun. 50 (2014) 123e134. [9] J.K. Hu, T. Kagari, J.M. Clingan, M. Matloubian, Expression of chemokine re- [33] J.H. Xie, N. Nomura, M. Lu, S.L. Chen, G.E. Koch, Y. Weng, et al., Antibody- ceptor CXCR3 on T cells affects the balance between effector and memory CD8 mediated blockade of the CXCR3 chemokine receptor results in diminished T-cell generation, Proc. Natl. Acad. Sci. U. S. A. 108 (2011) E118eE127. recruitment of T helper 1 cells into sites of inflammation, J. Leukoc. Biol. 73 [10] Y. Zohar, G. Wildbaum, R. Novak, A.L. Salzman, M. Thelen, R. Alon, et al., (2003) 771e780. CXCL11-dependent induction of FOXP3-negative regulatory T cells suppresses [34] I.A. Khan, J.A. MacLean, F.S. Lee, L. Casciotti, E. DeHaan, J.D. Schwartzman, et autoimmune encephalomyelitis, J. Clin. investigation 124 (2014) 2009e2022. al., IP-10 is critical for effector T cell trafficking and host survival in Toxo- [11] Y.H. Chuang, Z.X. Lian, C.M. Cheng, R.Y. Lan, G.X. Yang, Y. Moritoki, et al., plasma gondii infection, Immunity 12 (2000) 483e494. Increased levels of chemokine receptor CXCR3 and chemokines IP-10 and MIG [35] R.S. Friedman, J. Jacobelli, M.F. Krummel, Surface-bound chemokines capture in patients with primary biliary cirrhosis and their first degree relatives, and prime T cells for synapse formation, Nat. Immunol. 7 (2006) 1101e1108. J. Autoimmun. 25 (2005) 126e132. [36] B. Molon, G. Gri, M. Bettella, C. Gomez-Mouton, A. Lanzavecchia, A.C. Martinez, [12] A. Lleo, W. Zhang, M. Zhao, Y. Tan, F. Bernuzzi, B. Zhu, et al., DNA methylation et al., T cell costimulation by chemokine receptors, Nat. Immunol. 6 (2005) profiling of the X reveals an aberrant demethylation on CXCR3 465e471. promoter in primary biliary cirrhosis, Clin. epigenetics 7 (2015) 61. [37] H.L. Tang, J.G. Cyster, Chemokine Up-regulation and activated T cell attraction [13] L. Gorelik, R.A. Flavell, Abrogation of TGFbeta signaling in T cells leads to by maturing dendritic cells, Science 284 (1999) 819e822. spontaneous T cell differentiation and autoimmune disease, Immunity 12 [38] J.R. Groom, J. Richmond, T.T. Murooka, E.W. Sorensen, J.H. Sung, K. Bankert, et (2000) 171e181. al., CXCR3 chemokine receptor-ligand interactions in the lymph node opti- [14] S. Oertelt, Z.X. Lian, C.M. Cheng, Y.H. Chuang, K.A. Padgett, X.S. He, et al., Anti- mize CD4þ T helper 1 cell differentiation, Immunity 37 (2012) 1091e1103. mitochondrial antibodies and primary biliary cirrhosis in TGF-beta receptor II [39] L. Lasagni, M. Francalanci, F. Annunziato, E. Lazzeri, S. Giannini, L. Cosmi, et al., dominant-negative mice, J. Immunol. 177 (2006) 1655e1660. An alternatively spliced variant of CXCR3 mediates the inhibition of endo- [15] G.X. Yang, Z.X. Lian, Y.H. Chuang, Y. Moritoki, R.Y. Lan, K. Wakabayashi, et al., thelial cell growth induced by IP-10, Mig, and I-TAC, and acts as functional Adoptive transfer of CD8(þ) T cells from transforming growth factor beta receptor for , J. Exp. Med. 197 (2003) 1537e1549. receptor type II (dominant negative form) induces autoimmune cholangitis in [40] J.E. Ehlert, C.A. Addison, M.D. Burdick, S.L. Kunkel, R.M. Strieter, Identification mice, Hepatology 47 (2008) 1974e1982. and partial characterization of a variant of human CXCR3 generated by [16] W.W. Hancock, B. Lu, W. Gao, V. Csizmadia, K. Faia, J.A. King, et al., Require- posttranscriptional exon skipping, J. Immunol. 173 (2004) 6234e6240. ment of the chemokine receptor CXCR3 for acute allograft rejection, J. Exp. [41] C. Billottet, C. Quemener, A. Bikfalvi, CXCR3, a double-edged sword in tumor Med. 192 (2000) 1515e1520. progression and angiogenesis, Biochimica biophysica acta 1836 (2013) [17] R. Zhou, H. Wei, R. Sun, Z. Tian, Recognition of double-stranded RNA by TLR3 287e295. induces severe small intestinal injury in mice, J. Immunol. 178 (2007) [42] Q. Wu, R. Dhir, A. Wells, Altered CXCR3 isoform expression regulates prostate 4548e4556. cancer cell migration and invasion, Mol. cancer 11 (2012) 3. [18] S. Moteki, P.S. Leung, R.L. Coppel, E.R. Dickson, M.M. Kaplan, S. Munoz, et al., [43] M. Furuya, T. Yoneyama, E. Miyagi, R. Tanaka, K. Nagahama, Y. Miyagi, et al., Use of a designer triple expression hybrid clone for three different lipoyl Differential expression patterns of CXCR3 variants and corresponding CXC domain for the detection of antimitochondrial autoantibodies, Hepatology 24 chemokines in clear cell ovarian cancers and endometriosis, Gynecol. Oncol. (1996) 97e103. 122 (2011) 648e655. [19] Y. Yao, W. Yang, Y.Q. Yang, H.D. Ma, F.T. Lu, L. Li, et al., Distinct from its ca- [44] S. Kim, M. Bakre, H. Yin, J.A. Varner, Inhibition of endothelial cell survival and nonical effects, deletion of IL-12p40 induces cholangitis and fibrosis in angiogenesis by protein kinase A, J. Clin. investigation 110 (2002) 933e941. interleukin-2Ralpha(-/-) mice, J. Autoimmun. 51 (2014) 99e108.