Chemokine Receptor CCR6-Dependent Accumulation of γδ T Cells in Injured Liver Restricts Hepatic Inflammation and Fibrosis

Human Liver Samples

Human liver tissue was acquired either from biopsies for routine clinical purposes or explants of cirrhotic livers obtained during liver transplantation.¹⁵

Mice

C57bl/6 wild-type (WT), congenic CD45.1, Actin-eGFP, Il-17^−/−,¹⁶ and Ccr6^−/− mice⁹ were housed under specific pathogen-free conditions. Experiments were performed with age- and sex-matched animals at 6–12 weeks of age under ethical conditions approved according to German legal requirements.

Induction of Liver Injury and Fibrosis

Mice were injected intraperitoneally (IP) with 0.6 mL/kg body weight of CCl₄ (Merck, Darmstadt, Germany) mixed with corn oil. For induction of liver fibrosis, CCl₄ was injected thrice-weekly for 4 weeks. Mice were sacrificed 48 hours after the last injection. For induction of steatohepatitis, mice were fed with a methionine-choline-deficient (MCD) diet for 3–8 weeks (catalog no. 390439; MP Biomedicals, Solon, OH).

For details on methodology, please see the Supporting Materials.

The Chemokine Receptor, CCR6, and Its Ligand, CCL20, Are Significantly Up-Regulated in Chronic Liver Injury in Mice and Men

To investigate the potential role of CCR6 in CLD, liver samples of patients with different stages of chronic hepatic injury were assessed for expression of CCR6 and its cognate ligand, CCL20. In comparison to healthy controls (n = 5), hepatic messenger RNA (mRNA) expression of both markers was significantly up-regulated in patients with CLDs (n = 50; Fig. 1A,B). CCR6 expression was even higher in cirrhosis than fibrosis (Fig. 1C), whereas highest CCL20 expression was observed in fibrotic livers (Fig. 1D). Of note, CCR6 expression was elevated in patients with viral hepatitis, compared to other disease etiologies, whereas highest CCL20 was observed in primary biliary cirrhosis (Supporting Fig. 1A,B). Immunohistochemistry (IHC) confirmed enhanced CCR6 expression on protein level in liver disease patients and specifically detected CCR6 on lymphocytes in periportal infiltrates (Fig. 1E). In contrast, CCL20 protein expression was remarkably up-regulated by hepatocytes and, to a lesser extent, by biliary epithelial cells (BECs) in diseased versus control livers (Fig. 1E). In patients with cholestatic diseases, larger bile ducts strongly expressed CCL20, especially in regions with large clusters of inflammatory cells and damaged biliary epithelium (Supporting Fig. 1C).

To assess whether CCR6 and CCL20 are also up-regulated in experimental murine fibrosis, c57bl/6 mice were IP injected with CCl₄ thrice-weekly for 4 weeks. CCl₄-challenged mice showed significantly higher hepatic gene expression levels of Ccr6 and Ccl20, compared to controls (Supporting Fig. 2A,B), resembling the results obtained in human samples. Furthermore, murine primary hepatocytes, but not freshly isolated primary hepatic stellate cells (HSCs) and Kupffer cells (KCs)/macrophages, clearly up-regulated Ccl20 upon CCl₄ treatment, compared to cells from untreated livers (Supporting Fig. 2C,D). Together, these results demonstrated that the CCR6/CCL20 pathway is activated in CLD in men and mice, and that injured hepatocytes strongly induce Ccl20 expression, thus likely regulating CCL20-mediated chemotaxis on CCR6-expressing leukocytes upon liver injury.

Ccr6^−/− Mice Develop More Severe Hepatic Fibrosis Upon Chronic Liver Injury

To elucidate the functional relevance of CCR6/CCL20 for liver disease progression, liver fibrosis was induced by repetitive CCl₄ injections in WT and Ccr6^−/− mice. Conventional histology stainings showed severely damaged liver architecture in WT and Ccr6^−/− mice with large areas of necrotic hepatocytes and dense periportal inflammatory infiltrates, alongside strongly elevated aminotransferase activities in serum (Fig. 2A,B). Strikingly, extracellular collagen deposition, as revealed by Sirius Red stainings, was significantly enhanced in Ccr6^−/− mice, compared to WT mice (Fig. 2A). Accelerated fibrosis development in Ccr6^−/− mice was confirmed by increased hepatic concentrations of hydroxyproline, an amino acid abundantly present in collagen matrices, higher area fraction by Sirius Red staining, and increased expression of Collagen1A1 in fibrotic livers of Ccr6^−/−, compared to WT, mice (Fig. 2C).

To confirm the general relevance of the CCR6/CCL20 pathway for hepatic fibrogenesis, mice were also subjected to MCD diet as an independent model of chronic metabolic liver injury, resulting in severe steatohepatitis.¹⁷ Liver injury was enhanced in Ccr6^−/−, compared to WT, mice during MCD diet, as evidenced by histology and liver enzymes (Fig. 2D,E and data not shown). Moreover, Ccr6^−/− mice displayed more severe hepatic fibrosis after 8 weeks of MCD diet, compared to WT animals (Fig. 2D,F). Collectively, these results indicated that CCR6/CCL20 represents a protective chemokine pathway during chronic liver injury and hepatofibrogenesis.

Infiltration of Leukocytes Into Damaged Liver is Enhanced in Ccr6^−/− Mice

Livers of Ccr6^−/− mice showed significantly higher intrahepatic leukocyte accumulation upon injury, compared to WT mice, as evidenced by IHC for CD45 (Fig. 3A). By fluorescence-activated cell sorting (FACS) analysis of intrahepatic immune cell subsets (Fig. 3B), we could not observe differences in neutrophils or macrophages, whereas all major lymphocyte subsets (T, B, and natural killer [NK] cells) were significantly increased in livers of Ccr6^−/− CCl₄-treated mice, compared to WT animals (Fig. 3C,D). Importantly, the relative composition of the different major immune cell populations in injured livers did not significantly differ between WT and Ccr6^−/− mice (Fig. 3B and Supporting Fig. 3).

CCR6 expression had been reported on distinct CD4 T-cell subsets, in particular, Th17 and Treg cells.¹⁸ Thus, we analyzed the expression of the signature cytokines for Th1, Th2, Th17, and Treg cells in the liver, which are interferon (IFN)-γ, IL-4, IL-17, and IL-10, respectively. There was no difference in IFN-γ or IL-10 gene expression and protein concentrations between WT and Ccr6^−/− mice (Fig. 3E,F). Il-4 gene expression was significantly reduced in livers of CCl₄-treated Ccr6^−/− mice, compared to WT mice (Fig. 3E), but not on protein level (Fig. 3F).

More striking, IL-17 expression, but not IL-22, was significantly reduced in livers of Ccr6^−/−, compared to WT, mice, both on mRNA and protein level (Fig. 3E,F). Nonetheless, we also checked for gene and protein expression of Th-lineage-specific transcription factors T-bet (Th1), Gata3 (Th2), FoxP3 (Treg), or Rorγt (Th17; Supporting Fig. 4A–C), indicating that differentiation of intrahepatic Th cells did not change upon the absence of CCR6. Hence, whereas our results demonstrated that Ccr6^−/− mice develop more severe hepatic inflammation after CCl₄ treatment with higher immune cell infiltration, compared to WT mice, this effect appeared to not be related to alterations in the CD4 Th cell compartment.

Infiltration of IL-17-Producing γδ T Cells Is Reduced in the Absence of CCR6

To identify the CCR6-dependent cellular compartment responsible for limiting the inflammatory response in injured liver, we isolated different primary cell types by FACS sorting from murine livers to analyze Ccr6 expression. We could not detect Ccr6 expression in liver-resident cell populations, such as hepatocytes and HSCs, but noted a clear Ccr6 induction in macrophages and CD4 T cells upon injury (Fig. 4A). However, the most remarkable Ccr6 up-regulation was detected on γδ T cells after CCl₄ treatment (Fig. 4A). Moreover, both cell types up-regulated Il-17 after CCl₄ treatment, but, again, expression in γδ T cells was much higher than in CD4 T cells (Fig. 4B). Il-22 was also expressed by both CD4 T cells and γδ T cells, with a tendency toward higher expression in γδ T cells (Fig. 4B).

These analyses hinted at a CCR6-dependent hepatic recruitment of γδ T cells in chronic liver injury. In fact, γδTCR⁺ T cells increased upon CCl₄ treatment in livers of WT and Ccr6^−/− mice (Fig. 4C). However, Ccr6^−/− mice showed significantly reduced intrahepatic γδ T cells, compared to WT mice, after CCl₄ treatment (Fig. 4C).

γδ T cells can be classified into two major subsets according to their surface expression of NK1.1 or CCR6. NK1.1⁺ γδ T cells produce mainly IFN-γ, whereas CCR6⁺ γδ T cells primarily produce IL-17 and share some features with Th17 cells.¹² In healthy livers, approximately half of the γδ T cells were NK1.1⁺ in WT, but only 25% in Ccr6^−/− mice (Fig. 4D). After CCl₄ treatment, the relative amount of NK1.1⁺ γδ T cells decreased significantly in WT mice as a result of an influx of CCR6⁺ γδ T cells (Fig. 4D). In contrast, the NK1.1⁺ subset did not change considerably in Ccr6^−/− after CCl₄ treatment, shifting the balance of both subtypes toward NK1.1⁺ γδ T cells in the injured liver of Ccr6^−/− mice (Fig. 4D). Importantly, in Ccr6^−/− mice, the numbers of IL-17⁺ γδ T cells were significantly lower than in WT mice upon CCl₄ treatment (Fig. 4E). Taken together, these results strongly indicated that the infiltration of IL-17-producing γδ T cells into injured liver depended on the chemokine receptor, CCR6.

IL-17-Producing γδ T Cells Require CCR6 to Traffic to the Liver

Thus, we tested the hypothesis that the IL-17-producing subset of γδ T cells employs CCR6 to migrate to injured liver. Therefore, splenic γδ T cells were isolated from enhanced green fluorescent protein (eGFP) transgenic WT mice and GFP⁻ Ccr6^−/− mice (both from the CD45.2 background), mixed in a ratio of 1:1 and injected intravenously (IV) into congenic WT mice (carrying the CD45.1 allele) that had been treated once with CCl₄ 24 hours before γδ T-cell transfer (Fig. 5A). Twenty-four hours after transfer, γδ T cells were analyzed in liver and spleen. In livers, we found overwhelming numbers of GFP⁺ (=WT) γδ T cells, but only very few GFP⁻ (=Ccr6^−/−) γδ T cells. Very little transferred cells could be found in the spleen with no difference in numbers between WT and Ccr6^−/− T-cells, corroborating that CCl₄ administration induced a strong recruitment signal for transferred γδ T cells to the liver (Fig. 5A). These results indicated that γδ T cells indeed require CCR6 to traffic to liver upon injury.

Adoptive Transfer of WT γδ T Cells, but Not of CD4 T Cells, Limits Hepatic Fibrogenesis in Ccr6^−/− Mice Independent of IL-17

To provide functional evidence that the recruitment of CCR6-expressing γδ T cells is responsible for limiting CCl₄-mediated hepatic injury and fibrogenesis, we reconstituted Ccr6^−/− mice with WT γδ T cells. Therefore, WT and Ccr6^−/− mice were treated with repetitive CCl₄ injections, and mice additionally received WT γδ T cells (100,000–150,000 cells) isolated from spleens of congenic mice (CD45.1) IV once per week (Fig. 5B). After 4 weeks of CCl₄ treatment, a clear population of (transferred) CD45.1⁺ cells could be found in the liver that expressed the γδ T-cell receptor (TCR), but not CD4 (Fig. 5B). Ccr6^−/− mice that received wt γδ T cells showed significantly less hepatic fibrosis than without transfer (Fig. 5C,D). Moreover, the weekly adoptive transfer of WT γδ T cells to Ccr6^−/− mice was capable of fully restoring the fibrosis phenotype to the level of WT mice (Fig. 5C,D). Notably, the adoptive transfer of γδ T cells into WT mice did not significantly affect fibrosis progression, which might be related to the overall low numbers of transferred γδ T cells, compared to the endogenous hepatic γδ T-cell pool already present in fibrotic WT mice.

Next, we investigated the mechanisms by which CCR6-expressing γδ T cells restrict hepatic inflammation and fibrosis. IL-17 was the primary cytokine produced by hepatic CCR6⁺ γδ T cells in injured liver, but IL-17 signaling in macrophages and HSCs is considered profibrogenic in murine liver fibrosis.¹⁹ To test whether CCR6-expressing γδ T cells limit hepatofibrogenesis by IL-17, we reconstituted Ccr6^−/− mice with IL-17-deficient γδ T cells. The fibrosis phenotype of Ccr6^−/− mice injected with Il-17^−/− γδ T cells was fully restored to WT level, similar to adoptive transfer of Il-17⁺ γδ T cells (Fig. 5C,D).

To ensure that accelerated fibrosis in Ccr6^−/− mice was a consequence of reduced γδ T cells, but not of alterations in the CD4 T-cell compartment, we also transferred WT CD4⁺ T cells into CCl₄-challengend WT and Ccr6^−/− mice. As anticipated,² the adoptive transfer of CD4 T cells enhanced collagen deposition in WT mice, but did not further increase the strong fibrosis in Ccr6^−/− mice (Fig. 5C,D), corroborating that CCR6 was specifically important for γδ T cells in vivo. Taken together, our results provided experimental evidence that γδ T cells modulate hepatic inflammation and restrict fibrosis progression, but these effects were unrelated to IL-17 secretion by these cells.

Hepatic γδ T Cells Directly Promote HSC Apoptosis

We hypothesized that γδ T cells interact with HSCs, the major collagen-producing cell type in the liver.¹ We isolated γδ T cells, CD4⁺ T cells, and NK cells from livers of WT mice that had received a single CCl₄ injection and cocultured these cells with GRX cells, a murine HSC cell line, for 48 hours. As a positive control, GRX cells were stimulated with transforming growth factor (TGF)-β1, whereas untreated cells served as negative controls. Hepatic NK cells were included in this experiment because they have been shown to exert antifibrotic functions through killing of activated HSCs.²⁰ GRX cells that were left untreated, incubated with TGF-β, or cocultured with CD4⁺ T cells showed a typical appearance of proliferating, myofibroblast-like cells after 48 hours (Supporting Fig. 5A). In contrast, upon coculture with hepatic γδ T cells, GRX cells grew less dense and had more round-shaped apoptotic myofibroblasts in culture (Supporting Fig. 5A). As expected, stimulation of GRX cells with TGF-β1 resulted in up-regulated expression of Collagen1 and the HSC activation marker PDGF-receptor-β, whereas coculture with NK cells reduced expression of these markers (Supporting Fig. 5B). Coculture of GRX cells with γδ T cells also significantly reduced Collagen1 expression, suggesting that γδ T cells impair HSC-mediated extracellular matrix (ECM) production (Supporting Fig. 5B). This function appeared cell-contact dependent, because coculture with γδ T-cell supernatant did not affect HSC activation markers (Supporting Fig. 5B). Notably, coculture of γδ T cells appeared to moderately increase the fraction of apoptotic cells detected by propidium iodide (PI)-based FACS analysis in GRX culture, similar to effects observed with NK cells (Supporting Fig. 5C).

Next, we performed immunofluorescent double staining for γδ T cells (γδ TCR) and HSC (desmin) in fibrotic mouse livers to assess cell-cell interactions in vivo. In fact, HSCs frequently colocalized with γδ T cells in periportal areas of fibrotic livers (Fig. 6A), whereas hepatic macrophages (identified by F4/80) did not commonly show proximity to γδ T cells. Based on these observations in vitro (GRX cell line) and in vivo (CCl₄ model), we hypothesized that γδ T cells might promote apoptosis of HSCs, similar to the antifibrotic effects of NK cells. In fact, hepatic γδ T cells isolated from fibrotic, compared to control, livers showed increased expression of the apoptosis-inducing protein Fas ligand (FasL; CD95L; Supporting Fig. 5D). Coculture of primary HSCs, freshly isolated from murine livers, with either isolated hepatic γδ T or NK cells, but not with hepatic CD4 T cells, significantly increased the fraction of apoptotic primary HSCs, as evidenced by the sub-G₁ fraction upon PI staining, in culture (Fig. 6B,C). Coculture of HSCs with Il-17^−/− γδ T cells also significantly increased the fraction of apoptotic HSCs (Fig. 6C), demonstrating that this mechanism is IL-17 independent. It had been reported that IL-22, which is expressed by hepatic γδ T cells (Fig. 4B), might be a paracrine factor to induce senescence of HSCs.²¹ However, the effects of hepatic γδ T cells on HSCs were cell-cell contact dependent, because γδ T-cell supernatant (SN) did not increase HSC apoptosis (Fig. 6C). Moreover, specific inhibition by an anti-IL-22 antibody (Ab) did not abrogate γδ T-cell-induced HSC apoptosis (Supporting Fig. 5E). When a monoclonal FasL-blocking Ab was added to the coculture with hepatic γδ T cells, the rate of apoptotic primary HSCs was significantly reduced to baseline levels (Fig. 6B,C).

To assess whether γδ T-cell-induced HSC apoptosis is involved in limiting hepatic fibrosis of WT, compared to Ccr6^−/−, mice, liver sections of CCl₄-treated WT and Ccr6^−/− mice were analyzed for the presence of apoptotic HSCs by costaining for desmin and TUNEL (terminal deoxynucleotidyl transferase dUTP nick end labeling, which marks DNA fragmentation of apoptotic cells). Although the rate of apoptotic HSCs was low, even in WT livers, Ccr6^−/− mice showed tremendously higher numbers of desmin⁺ HSCs in portal areas, consistent with reduced HSC apoptosis during the course of chronic liver injury (Fig. 6D,E). These data strongly support that hepatic γδ T cells promote apoptosis of HSCs, in a FasL-dependent manner, thereby limiting profibrogenic functions of stellate cells.