CXCL10 promotes liver fibrosis by prevention of NK cell mediated hepatic stellate cell inactivation

Abstract

Chemokines, such as CXCL10, promote hepatic inflammation in chronic or acute liver injury through recruitment of leukocytes to the liver parenchyma. The CXCL10 receptor CXCR3, which is expressed on a subset of leukocytes, plays an important part in Th1-dependent inflammatory responses. Here, we investigated the role of CXCL10 in chemically induced liver fibrosis. We used carbon tetrachloride (CCl(4)) to trigger chronic liver damage in wildtype C57BL/6 and CXCL10-deficient mice. Fibrosis severity was assessed by Sirius Red staining and intrahepatic leukocyte subsets were investigated by immunohistochemistry. We have further analyzed hepatic stellate cell (HSC) distribution and activation and investigated the effect of CXCL10 on HSC motility and proliferation. In order to demonstrate a possible therapeutic intervention strategy, we have examined the anti-fibrotic potential of a neutralizing anti-CXCL10 antibody. Upon CCl(4) administration, CXCL10-deficient mice showed massively reduced liver fibrosis, when compared to wildtype mice. CXCL10-deficient mice had less B- and T lymphocyte and dendritic cell infiltrations within the liver and the number and activity of HSCs was reduced. In contrast, natural killer (NK) cells were more abundant in CXCL10-deficient mice and granzyme B expression was increased in areas with high numbers of NK cells. Further detailed analysis revealed that HSCs express CXCR3, respond to CXCL10 and secrete CXCL10 when stimulated with IFNγ. Blockade of CXCL10 with a neutralizing antibody exhibited a significant anti-fibrotic effect. Our data suggest that CXCL10 is a pro-fibrotic factor, which participates in a crosstalk between hepatocytes, HSCs and immune cells. NK cells seem to play an important role in controlling HSC activity and fibrosis. CXCL10 blockade may constitute a possible therapeutic intervention for hepatic fibrosis.

Fig. 1

HSC enriched cell pools express CXCL10 after chronic CCl₄ treatment or upon IFNγ stimulation in vitro. (A) RNase protection assay using total RNA isolated from HSCs of C57BL/6 mice treated with CCl₄ for indicated time points. To prepare a HSC enriched cell pool, livers of 5 animals per time point were subjected to collagenase/pronase digest, followed by density gradient cell separation. The bar graph represents the quantification of relative amounts of CXCL10 mRNA normalized to signal intensities of GAPDH mRNA in each cell pool. Depicted is fold induction with time point d0 set at 1. (B and C) HSC pools have been isolated from 6 CXCL10^–/– or 6 wildtype C57BL/6 mice. (B) Left panel: Oil Red O staining revealed a HSC purity of our preps of >90%. Central panel: Upon IFNγ stimulation (10 ng/ml) for 24 h, CXCL10^–/– HSCs produced no CXCL10. Right panel: Wildtype HSCs secreted CXCL10 upon IFNγ stimulation. Scale bar = 20 μm. (C) Cultured HSCs were left untreated or were stimulated with IFNγ at 10 ng/ml or 100 ng/ml for indicated time points before total RNA was purified. CXCL10 or GAPDH encoding mRNA was analyzed by RPA. Signals have been normalized to the unstimulated wildtype control. Representative of n = 3.

Fig. 2

CXCL10^–/– mice are protected from CCl₄-induced fibrosis. (A) Livers of CXCL10^–/– and wildtype mice have been harvested after 4 weeks of CCl₄ treatment. Cryosections were stained with Sirius Red to visualize collagen fibers (two different magnifications are shown). Note that bridging fibrosis can only be seen in livers of wildtype but not of CXCL10^–/– mice. To localize HSCs, consecutive sections from the identical livers were also stained with antibodies to (B) desmin and (C) αSMA. Total HSC staining (desmin, B) and staining of activated HSCs (αSMA, C) is much weaker in the absence of CXCL10 than in livers of wildtype animals. Scale bar = 100 μm. Representative of n = 8.

Fig. 3

Wildtype and CXCL10^–/– livers are similar in terms of initial HSC numbers and hepatocyte damage by CCl₄. HSCs express CXCR3 and the frequency of CXCR3 + HSCs increases upon stimulation. (A) Desmin staining of liver sections of naïve wildtype and CXCL10^–/– mice. Scale bar = 50 μm. (B) Immunoblotting of equal amounts of total liver homogenates with antibodies to GFAP, desmin and HSP70. Two naïve animals per group were analyzed. (C) ALT and AST activities in sera from wildtype and CXCL10^–/– mice. Sera were collected 4 h or 4 days after first CCl₄ injection or 4 days (*4d) after last CCl₄ treatment. Data represent mean ± SD, n = 5. (D) FACS analysis of HSCs isolated from wildtype control (naïve) or CCl₄-treated (activated) animals. Cell pools isolated from 6 livers per condition were surface stained with an antibody to CXCR3. In parallel, HSCs were identified by intracellular GFAP staining. Numbers indicate percentage of GFAP⁺ cells that were CXCR3⁺.

Fig. 4

CXCL10 stimulates HSC chemotaxis but not proliferation. (A) Freshly isolated HSCs from control or CCl₄-exposed animals were seeded in Transwell chambers. Cells were stimulated with 20 ng/ml CXCL10 or were left untreated (none). Medium was renewed every day. Migration was stopped at indicated time points and cells that had migrated through the Transwell inserts were counted in 10 microscopic fields. Assays were done in duplicates and were repeated 3 times. Data are mean fold induction ± SD (n = 3) with untreated cells set at 1. *P < 0.05 (CXCL10 versus untreated). Consecutive liver sections of CCl₄-treated wildtype or CXCL10^–/– mice were stained with an antibody to the proliferation marker Ki67 (B) or with Sirius Red (C). Ki67 staining was seen in hepatocytes and strongly in areas with high numbers of infiltrating cells (arrowheads). Ki67 expression was low in regions with high fibrosis but low infiltrates (arrows). Proliferation intensity was similar in wildtype and CXCL10^–/– livers. Scale bar = 100 μm.

Fig. 5

Localization of immune cells in livers of CCl₄-treated CXCL10^–/– and wildtype mice. Mice were treated with CCl₄ for 4 weeks before livers were processed for immunohistochemistry. Tissue sections of wildtype (A and C) and CXCL10^–/– (B and D) mice have been stained for B220⁺ cells (A and B), CD4 T cells (C and D), CD19⁺ cells (not shown), and CD11c⁺ cells (not shown). Boxed areas are shown in larger magnification underneath original photomicrographs. Scale bar = 100 μm. (E) Cell numbers were counted in 10 sections of 10 different animals per group from 5 independent experiments. Relative cell numbers have been normalized to the corresponding number of B220⁺, CD19⁺, CD4⁺, and CD11c⁺ cell infiltrations found in livers of CCl₄-treated wildtype mice. Data represent mean ± SEM. *P < 0.05 (CXCL10^–/– versus wildtype).

Fig. 6

Blocking anti-CXCL10 antibody reduces liver fibrosis and number of infiltrating B220⁺ cells and CD4 T cells. Wildtype mice were injected intraperitoneally three times per week with 50 μg or 100 μg blocking anti-CXCL10 mAb and twice a week with CCl₄. After 4 weeks, livers were collected and cryosections were stained with (A) Sirius Red (two different magnifications are shown; scale bar = 100 μm), (B) an anti-αSMA antibody or (C) antibodies to B220 or CD4 (stainings not shown). Left panels (A and B) show liver sections of mice treated with 100 μg anti-CXCL10 mAb. Control mice received 100 μg of an isotype-matched control antibody (A and B, right panel). B220⁺ cells and CD4 T cells were counted in 10 sections of 5 different animals per group from 2 independent experiments. Data represent mean ± SEM. *P < 0.05 (CXCL10^–/– or anti-CXCL10 antibody treated versus isotype control).

Fig. 7

In the absence of CXCL10, increasing numbers of NK cells infiltrate the liver and eliminate activated HSCs by a granzyme B-dependent mechanism. Consecutive tissue sections from livers harvested at week 4 after start of CCl₄ administration were stained for NKp46, αSMA or granzyme B. (A) NK cell numbers were low and αSMA-expressing cells were frequent in livers obtained from wildtype mice (left panels). In contrast, the number of NK cells was 8.12 ± 1.18-fold higher and αSMA expressing cells were scarce in livers of CXCL10^–/– mice (middle panel). Similarly, αSMA expression was low and 7.53 ± 0.94-fold more NK cells were present in mice treated with anti-CXCL10 antibody (100 μg i.p., three times per week) than in wildtype mice. Scale bar = 100 μm. Cell numbers were counted in 4 sections of 5 different animals per group from 2 independent experiments. Relative cell numbers have been normalized to the corresponding number of NKp46⁺ cell infiltrations found in livers of CCl₄-treated wildtype mice. Data represent mean ± SEM. *P < 0.05 (CXCL10^–/– or anti-CXCL10 antibody treated versus control). (B) In wildtype livers, granzyme B staining was rather weak, correlating with the low number of NKp46-expressing cells. In contrast, αSMA-positive cells were highly abundant (upper panel). In CXCL10^–/– livers, αSMA-expressing cells were rare, whereas the number of NK cells was high and granzyme B expression was strong (lower panels). Scale bar = 100 μm.