Proapoptotic effects of the chemokine, CXCL 10 are mediated by the noncognate receptor TLR4 in hepatocytes


  • Potential conflict of interest: Nothing to report.

  • This work was supported by grants from the Deutsche Forschungsgemeinschaft (WA 2557/2-1 and SFB/TRR 57 P08; to H.E.W.) and the Else Kröner-Fresenius Stiftung (to H.E.W.).


Aberrant expression of the chemokine CXC chemokine ligand (CXCL)10 has been linked to the severity of hepatitis C virus (HCV)-induced liver injury, but the underlying molecular mechanisms remain unclear. In this study, we describe a yet-unknown proapoptotic effect of CXCL10 in hepatocytes, which is not mediated through its cognate chemokine receptor, but the lipopolysaccharide receptor Toll-like receptor 4 (TLR4). To this end, we investigated the link of CXCL10 expression with apoptosis in HCV-infected patients and in murine liver injury models. Mice were treated with CXCL10 or neutralizing antibody to systematically analyze effects on hepatocellular apoptosis in vivo. Direct proapoptotic functions of CXCL10 on different liver cell types were evaluated in detail in vitro. The results showed that CXCL10 expression was positively correlated with liver cell apoptosis in humans and mice. Neutralization of CXCL10 ameliorated concanavalin A–induced tissue injury in vivo, which was strongly associated with reduced liver cell apoptosis. In vitro, CXCL10 mediated the apoptosis of hepatocytes involving TLR4, but not CXC chemokine receptor 3 signaling. Specifically, CXCL10 induced long-term protein kinase B and Jun N-terminal kinase activation, leading to hepatocyte apoptosis by caspase-8, caspase-3, and p21-activated kinase 2 cleavage. Accordingly, systemic application of CXCL10 led to TLR4-induced liver cell apoptosis in vivo. Conclusion: The results identify CXCL10 and its noncognate receptor, TLR4, as a proapoptotic signaling cascade during liver injury. Antagonism of the CXCL10/TLR4 pathway might be a therapeutic option in liver diseases associated with increased apoptosis. (HEPATOLOGY 2013)

Acute hepatitis, cirrhosis, and hepatocellular carcinoma are associated with acute or chronic loss of hepatocellular integrity, which leads to increased mortality in many affected patients.1 Despite different causes of liver cell injury, a major mechanism leading to hepatic dysfunction is hepatocyte apoptosis.2 Thus, a better understanding of programmed cell death within the liver appears important to develop new therapeutic options for many different disease entities. However, the molecular mediators controlling hepatocyte apoptosis have not been fully deciphered in vivo and in vitro. In recent years, the contribution of chemokines to acute and chronic liver diseases has been reported in patients and in animal models.3 Chemokines are a class of small (8-12-kDa) chemotactic cytokines orchestrating the influx of immune cells into sites of inflammation, but also directly affect the biology of resident cells.4 In liver diseases, a predominant CXC chemokine receptor (CXCR) expressed by infiltrating lymphocytes is CXCR3.5 CXCR3 and its splice variant bind the interferon-gamma-inducible chemokines, CXC chemokine ligand (CXCL)9, CXCL10, CXCL11, and CXCL4, in humans.6 Mice with a genetic deletion of CXCR3 were more prone to liver injury, which is mediated by the loss of antifibrogenic and angiostatic properties of CXCL9 on hepatic stellate cells (HSCs)7 and sinusoidal endothelial cells (ECs).8 On the other hand, deletion of the CXCR3 ligands, CXCL4 and CXCL10, inhibits murine liver fibrosis,9, 10 and the neutralization of CXCL10 ameliorates experimentally induced liver injury in wild-type (WT) mice.9, 11 Furthermore, serum and intrahepatic CXCL10 levels in patients are positively associated with the severity of hepatitis C virus (HCV)-induced liver disease.7, 12 In line with these findings, increased serum levels of CXCL10 have also been identified to be independently associated with early fibrosis recurrence after liver transplantation (LT) for hepatitis C.13 The molecular mechanisms underlying these primarily contradictory observations of CXCL10 and its cognate receptor CXCR3 remain obscure. One possible explanation is that CXCL10 might operate through a noncognate receptor and thereby mediate biological effects independent of CXCR3. Indeed, CXCL10 has been implicated in non-chemokine-receptor-mediated apoptotic effects in cell culture.14-16 However, whether these effects are also true for liver cells and whether they also operate in vivo have not yet been investigated.

Therefore, we have systematically explored whether CXCL10 is functionally involved in hepatocyte apoptosis in vitro and in vivo and whether these effects are mediated through the CXCL10 cognate receptor, CXCR3, or through an alternative signaling pathway.

Patients and Methods

Liver Biopsies From Patients With HCV Infection.

Degree of apoptosis in patients with HCV infection was determined by immunohistochemical (IHC) staining of cleaved caspase-3 (Cell Signaling Technologies, Danvers, MA) in liver biopsies.7 In addition, total RNA was extracted from paraffin-embedded liver biopsies and reverse transcribed using SuperScript (Invitrogen, Carlsbad, CA), as described previously.17 Quantitative reverse-transcriptase polymerase chain reaction (qRT-PCR) was performed for CXCL10 with an Assay-on-Demand (Applied Biosystems, Foster City, CA). Experiments were performed after approval by the local ethics committee, and informed consent was obtained from patients before analysis.


Ab, antibody; Akt, protein kinase B; ALI, acute liver injury; ALT, alanine aminotransferase; AST, aspartate transaminase; BCL-2, B-cell lymphoma 2; BW, body weight; ConA, concanavalin A; CXCL, CXC chemokine ligand; CXCR, CXC chemokine receptor; ECs, endothelial cells; FCS, fetal calf serum; HCV, hepatitis C virus; HSCs, hepatic stellate cells; IHC, immunohistochemically; IP, intraperitoneally; JNK, Jun N-terminal kinase; LT, liver transplantation; mAB, monoclonal antibody; mRNA, messenger RNA; OS, oxidative stress; PAK-2, p21-activated kinase 2; PI3K, phosphatidylinositide 3-kinase; qRT-PCR, quantitative reverse-transcriptase polymerase chain reaction; ROS, reactive oxygen species; SEM, standard error of the mean; TGF-ß, transforming growth factor beta; TLR4, Toll-like receptor 4; TUNEL, terminal deoxynucleotidyl transferase dUTP nick-end labelling; WT, wild type.

Murine In Vivo Experiments.

C57BL/6 WT, CXCR3−/−,7 TLR4−/− as well as Casp8Δhepa mice18 (all back-crossed for more than 10 generations onto the C57BL/6 background) were housed in air-conditioned rooms with a constant temperature of 23°C. Mice were fed a standard laboratory chow (Ssniff) and had access to tap water ad libitum. All in vivo experiments were conducted after approval by the state animal protection board. WT mice (6-8 weeks old, 18-20 g) were injected with CCl4 (0.6 mL/kg body weight [BW]; Merck KGaA, Darmstadt, Germany) intraperitoneally (IP) for 24 hours7 or concanavalin A (ConA; 18 mg/kg BW) intravenously for 6 and 24 hours.19 For blocking experiments, a neutralizing anti-CXCL10 monoclonal antibody (mAb; 100 μg/100 μL; R&D Systems, Minneapolis, MN) or monoclonal rat Immunoglobulin G2A (100 μg/100 μL; R&D Systems) was administered IP to C57BL/6 WT mice concomitantly with or without ConA.11 In a separate experiment, WT and TLR4−/− mice were treated with recombinant CXCL10 (5 μg/100 μL; Biomol GmbH, Hamburg, Germany) for 8 hours.

Quantification of Apoptosis.

The terminal deoxynucleotidyl transferase dUTP nick-end labeling (TUNEL) assay (Roche Applied Science, Penzberg, Germany) was used for the detection and quantification of apoptosis in cryosections from livers of animals. The percentage of TUNEL-positive cells was quantified in a minimum of three independent magnification fields per animal.

Intrahepatic Protein and Gene Expression.

Isolation of total protein and RNA from snap-frozen liver tissue samples were carried out as described previously.17 Intrahepatic CXCL10 concentrations were determined using a mouse CXCL10 enzyme-linked immunosorbent assay Kit (R&D Systems), following the manufacturer's instructions. qRT-PCR was performed for B-cell lymphoma 2 (BCL-2) gene with Assays-on-Demand (Applied Biosystems). ß-actin was used as the reference gene.

In Vitro Experiments.

Isolation and culture of primary hepatocytes and HSCs from mice were performed as described by Taura et al.20 Hepatocytes were cultured on collagen-coated plates in William's E medium (PAA Laboratories GmbH, Pasching, Austria) and 4% heat-inactivated fetal calf serum (FCS). HSCs were cultured in Dulbecco's modified Eagle's medium with 4.5 g/L glucose (PAA Laboratories) and 10% heat-inactivated FCS. For chemokine stimulation, cells were starved in suitable medium containing 0.5% FCS for 16 hours and stimulated with recombinant mouse CXCL10 (100 ng/mL; Biomol). In some experiments, CXCL10 was preincubated for 30 minutes at 37°C with 2 μg/mL of polymyxin B (Sigma-Aldrich, St. Louis, Missouri, USA). Untreated hepatocytes and stellate cells were used as controls. All experiments were performed in triplicates.

Western Blotting Anaylsis.

Western blottings of p21-activated kinase 2 (PAK-2; Santa Cruz Biotechnology, Santa Cruz, CA), cleaved caspase-3, phosphorylated and total protein kinase B (Akt) and Jun N-terminal kinase (JNK) (all antibodies [Abs] from Cell Signaling Technology) were performed as described previously.17

Caspase-3 and Caspase-8 Assay.

The fluorogenic substrates, Ac-DEVD-AFC and Ac-IETD-AFC (both from Biomol), were used to quantify Caspase-3 and Caspase-8 activity, respectively. Fluorescence signals were detected by a fluorometer (GENios; Tecan Group Ltd., Männerdorf, Switzerland) at excitation and emission wavelengths of 400 and 510 nm, respectively.

Assay for Reactive Oxygen Species.

Reactive oxygen species (ROS) levels were measured as previously described.21 In brief, mouse hepatocytes were cultured on collagen-coated glass slides. After 16-hour starvation, cells were incubated with CXCL10 for 8 hours, followed by the addition of 10 μM of carboxy derivative of fluorescein (CM-H2DCFDA; Invitrogen) and staining in phosphate-buffered saline for 30 minutes, according to the manufacturer's instruction. ROS production was visualized by fluorescence microscopy. As a positive control, hepatocytes were preincubated with 5 μM of H2O2 for 1 hour, whereas the negative control (CM-H2DCFDA) was omitted.

Statistical Analysis.

Data are presented as means and standard error of the mean (SEM). Statistical significance was determined by the Student t test. Pearson's correlation was used to measure a linear association between two variables. Statistical analyses were assessed using GraphPad Prism 5 software (GraphPad Software Inc., La Jolla, CA).


CXCL10 Expression Is Associated With Liver Cell Apoptosis in Humans and Mice.

First, we assessed a possible correlation of CXCL10 messenger RNA (mRNA) expression and the number of apoptotic cells in a cohort of HCV-infected patients with different degrees of chronic liver injury.7 In this cohort, the number of cleaved Caspase-3-positive cells was strongly associated with augmented hepatic CXCL10 mRNA expression, suggesting a link between CXCL10 and the degree of apoptotic liver cell death (Fig. 1A). In extension of the human data, we investigated whether there is also a positive correlation between level of CXCL10 and degree of cell death in murine liver injury models. Indeed, augmented intrahepatic CXCL10 protein expression, in response to challenge of WT mice with either CCl4 or ConA, was positively associated with increased numbers of TUNEL-positive liver cells (Fig. 1B), implying species-independent effects of CXCL10.

Figure 1.

Association of CXCL10 expression with liver cell apoptosis. Apoptosis was assessed by counting cleaved caspase-3 positive nuclei (magnification, ×400) after IHC staining of cleaved caspase-3 in liver biopsies of HCV-infected patients. The number of caspase-3-positive cells (expressed as percentage of all nuclei) is strongly correlated to CXCL10 mRNA in the same biopsies (n = 10, R2 = 0.8678, P < 0.001) (A). Administration of CCl4 and ConA for 6 and 24 hours led to a increased number of TUNEL-positive nuclei (magnification, ×200) in WT mice, compared to control (n = 6 mice per group). Increased cell death was correlated with CXCL10 protein levels in these samples (R2 = 0.7540, P < 0.001) (B).

Neutralization of CXCL10 Attenuates ConA-Induced Liver Injury.

Next, we evaluated whether the association between CXCL10 expression and liver cell death is a statistical phenomenon or whether a functional relationship exists between these parameters. To this end, we evaluated direct effects of a neutralizing CXCL10 mAb on ConA-induced acute liver injury (ALI) and related liver cell death. Treatment of WT mice with ConA for 6 hours led to an increase in TUNEL-positive cells, compared to vehicle-treated mice (Fig. 2A). Indeed, neutralization of CXCL10 protected ConA-treated mice from hepatocellular death, as determined by TUNEL assay (Fig. 2A). The protective effects of CXCL10 neutralization also translated into reduced serum levels of alanine aminotransferase (ALT) and aspartate transaminase (AST) (Fig. 2B,C) 6 hours after ConA administration.

Figure 2.

Neutralization of CXCL10 ameliorates ConA-induced liver cell apoptosis. Treatment of WT mice with ConA for 6 hours resulted in an increased number of TUNEL-positive cells, compared to vehicle-treated mice. Neutralization of CXCL10 with an anti-CXCL10 Ab protected ConA-treated mice from hepatocellular injury (TUNEL; magnification, ×100) (A). Amelioration of liver damage in anti-CXCL10-treated animals were further confirmed by reduced ALT (B) and AST (C) levels in serum, compared to vehicle-treated mice. *P < 0.05; **P < 0.01. Data are expressed as means ± SEM of 6 mice per group.

Proapoptotic Effects of CXCL10 Are Mediated by Akt Signaling.

Because the in vivo findings suggested a functional association between CXCL10 and apoptosis, we examined the direct effects of the chemokine on primary hepatocytes and stellate cells from WT mice in vitro. Stimulation of hepatocytes with CXCL10 led to strong morphologic changes of these cells (Fig. 3A). Next, we assessed whether caspases, which play a key role in the execution of apoptosis, are involved in these apparent cytopathic effects of CXCL10. Indeed, incubation of hepatocytes with the chemokine resulted in a time-dependent activation of caspase-3 (Fig. 3B) and caspase-8 (Fig. 3C), supporting a direct involvement of CXCL10 in hepatocyte apoptosis. On a molecular basis, CXCL10 led to increased Akt phosphorylation within 20 minutes, which was sustained through the entire 8-hour culture period (Fig. 3D,E and Supporting Fig. 1A,B).

Figure 3.

CXCL10 induces hepatocyte apoptosis in vitro. Stimulation of hepatocytes with CXCL10 led to morphological changes of these cells (×600 magnification) (A). In this in vitro setting, CXCL10 induced time-dependent activation of caspase-3 (B) and caspase-8 (C). Data are normalized to vehicle-treated control. CXCL10 stimulation of hepatocytes was associated with increased Akt phosphorylation within 60 minutes (D), which was preserved throughout the entire 8-hour culture period (E). *P < 0.05; **P < 0.01. Data are expressed as means ± SEM of three independent experiments.

Having shown that CXCL10 leads to sustained Akt activation in hepatocytes, we next investigated the pathway that may switch traditionally considered prosurvival Akt activation into proapoptotic signals. PAK-2 is a direct downstream effector of Akt and is known to exert apoptotic effects when its cleavage into PAK-2p34 fragments is mediated by caspases.22 In hepatocytes, CXCL10 indeed induced increased levels of activated PAK-2p34 (Fig. 4A and Supporting Fig. 2A). Apart from Akt and caspase-3 activation, CXCL10 also induced long-term phosphorylation of JNK (Fig. 4A and Supporting Fig. 2A), a pathway induced by inflammatory cytokines, such as transforming growth factor beta (TGF-ß) and oxidative stress (OS), leading to hepatocyte apoptosis.23 In contrast, inhibition of the phosphatidylinositide 3-kinase (PI3K)/Akt pathway by Wortmannin completely blunted Akt activation in the presence of CXCL10 (Fig. 4B and Supporting Fig. 2B). PI3K inhibition also abrogated CXCL10-induced caspase-3, JNK, and proteolytic PAK-2p34 activation (Fig. 4B and Supporting Fig. 2B), indicating that PI3K/Akt acts upstream in this apoptotic pathway. Besides JNK and caspase-3 activation, CXCL10 also induced ROS production in hepatocytes (Fig. 5A).

Figure 4.

Intracellular mechanisms governing the effect of CXCL10 in hepatocytes. CXCL10 induced increased levels of active JNK, caspase-3, and PAK-2p34 in hepatocytes (A), whereas inhibition of the PI3K pathway by Wortmannin in WT hepatocytes blunted Akt activation in the presence of CXCL10 (B). Representative blots from three independent experiments are shown.

Figure 5.

CXCL10 mediates its effect by ROS and caspase-8. CXCL10 contributes to OS-induced apoptosis, as determined by fluorescence staining of ROS on hepatocytes (magnification, ×600). As a positive control, hepatocytes were preincubated with H2O2, whereas the negative control (CM-H2DCFDA) was omitted. − ctrl, negative control; ctrl, untreated hepatocytes; + ctrl, positive control (A). CXCL10 also induces Akt and JNK phosphorylation in Casp8−/− hepatocytes, but does not affect caspase-3 and PAK-2p34 activity (B).

To further dissect the signaling pathways of CXCL10-induced hepatocyte apoptosis, we used hepatocytes from mice with hepatocyte-specific knockout of caspase-8 (Casp8Δhepa). Interestingly, CXCL10 induced Akt and JNK phosphorylation in these cells, but did not affect caspase-3 and PAK-2 cleavage (Fig. 5B and Supporting Fig. 3A). Because stellate cells are also known to play an important role during liver injury, we next assessed the effects of CXCL10 on stellate cells. However, treatment of stellate cells with CXCL10 led to no changes in caspase-3 activity (data not shown), suggesting a primarily hepatocyte-specific effect of CXCL10.

Proapoptotic Effects of CXCL10 in Hepatocytes Are Independent of CXCR3.

Next, we assessed the role of its cognate receptor (CXCR3) in this process to further dissect the mechanisms of CXCL10-mediated hepatocyte apoptosis. To this end, we exposed CXCR3−/− hepatocytes to CXCL10 or vehicle. Interestingly, CXCL10 also induced apoptosis in these cells, as evidenced by increased levels of active caspase-3 and caspase-8 (Fig. 6A,B) as well as by prolonged Akt phosphorylation (Fig. 6C and Supporting Fig. 3B). To exclude a contamination of the recombinant CXCL10 by lipopolysaccharide, we preincubated CXCR3−/− hepatocytes with polymyxin B. In fact, this preparation did not change caspase-3 and Akt activation (Supporting Fig. 3C,D), demonstrating a CXCL10-specific effect on hepatocyte apoptosis.

Figure 6.

CXCL10 mediates hepatocyte apoptosis through TLR4, but not CXCR3, signaling. CXCL10 stimulated CXCR3−/− hepatocytes displayed a time-dependent increased level of active caspase-3 (A) and caspase-8 (B). Results were further confirmed by the phosphorylation of Akt (C). In contrast, stimulation of hepatocytes from TLR4−/− mice with CXCL10 did not result in caspase-3 (D) and caspase-8 (E) activation. These results were confirmed by the lack of Akt phosphorylation (F). **P < 0.01; ***P < 0.001. Data are expressed as means ± SEM of three independent experiments.

Importantly, in contrast to CXCL10, the related chemokine (CXCL9) did not affect hepatocyte apoptosis, as evidenced by measurement of caspase-3 activity (data not shown).

CXCL10 Mediates Apoptosis in Hepatocytes Through TLR4 Signaling.

Because CXCR3 is not involved in hepatocyte apoptosis, we became interested whether an alternative receptor could trigger CXCL10-induced apoptosis in hepatocytes. Recently, Schulthess et al.24 identified TLR4 as a receptor for CXCL10 in pancreatic β-cells. First, we confirmed the expression of TLR4 on hepatocytes by PCR analysis (Supporting Fig. 4A). Next, we stimulated TLR4−/− hepatocytes with CXCL10 or vehicle. Indeed, we found no caspase-3 and caspase-8 activation (Fig. 6D,E). These results were confirmed by lack of Akt phosphorylation (Fig. 6F and Supporting Fig. 4B) subsequent to CXCL10 stimulation of these cells. Thus, activation of TLR4 signaling appears essential to trigger CXCL10-induced hepatocyte apoptosis.

CXCL10 Induces TLR4-Mediated Apoptosis Within the Liver In Vivo.

In light of these in vitro data, we hypothesized that systemic administration of CXCL10 might also induce liver cell apoptosis in vivo. Indeed, a single injection of CXCL10 led to a low, but increased, number of TUNEL-positive liver cells, compared to vehicle treatment (Fig. 7A). The apoptotic response in CXCL10-treated animals was also reflected by increased caspase-3 and caspase-8 activity within livers of these animals (Fig. 7B and Supporting Fig. 4C). Moreover, treatment with CXCL10 increased AST serum levels (Fig. 7C) and reduced intrahepatic mRNA expression of the antiapoptotic factor, BCL-2 (Fig. 7D). Importantly, in this experimental setting, TLR4−/− mice were almost completely protected from the proapoptotic effects of CXCL10. In contrast to WT mice, treatment of TLR4−/− mice with CXCL10 neither resulted in augmented cell death (Fig. 7A) nor in caspase-3 or caspase-8 activation (Fig. 7B and Supporting Fig. 4C). In line with these results, lack of TLR4 also triggered no changes in AST and BCL-2 levels after CXCL10 challenge, compared to their vehicle-treated counterparts (Fig. 7C,D), identifying the CXCL10/TLR4 axis as an important chemokine-based apoptotic pathway within the murine liver in vivo.

Figure 7.

CXCL10 induces TLR4-mediated apoptosis within the liver in vivo. A single injection of CXCL10 of WT mice led to an increased number of TUNEL-positive cells (magnification, ×200; insert, ×400), compared to vehicle-treated mice (A). Apoptotic response in CXCL10-treated animals was also reflected by strongly increased hepatic caspase-3 activity (B). Moreover, treatment with CXCL10 enhanced AST levels in serum (C) and reduced intrahepatic mRNA expression of the antiapoptotic factor, BCL-2 (D). In contrast to WT mice, treatment of TLR4−/− mice with CXCL10 led neither to cell death nor to caspase-3 activation. In line with these results, the lack of TLR4 also led to no changes in AST and BCL-2 levels after CXCL10 challenge, compared to their vehicle-treated counterparts (A-D). *P < 0.05; **P < 0.01; ***P < 0.001.


Here, we provide in vitro and in vivo evidence that CXCL10 exerts proapoptotic effects in hepatocytes through its noncognate receptor (TLR4). In clinical studies, hepatocyte apoptosis has been identified as a crucial feature of liver injury in chronic hepatitis C,25 mediated by cytokines expressed by activated macrophages, HSCs, and hepatocytes themselves.2 Notably, CXCL10 is known to be strongly linked to the severity of HCV-mediated liver damage7, 12 and to predict early fibrosis recurrence after LT for hepatitis C.13 In the current study, we could functionally link these two observations and show that an increase of apoptotic cells within livers of HCV-infected patients is strongly correlated with an increased mRNA expression of CXCL10. These findings, together with the knowledge of the involvement of CXCL10 in epithelial,16 pancreatic,15 and β-cell14 injury, motivated our interest to further understand the role of CXCL10 in liver cell apoptosis.

Accordingly, we used different murine liver injury models to validate our results obtained in human samples. Indeed, in ConA- and CCl4-induced ALI, increased CXCL10 expression was associated with increased number of TUNEL-positive cells. To assess whether a functional relationship underlies this association, we treated mice with ConA to induce acute hepatitis26 and inhibited CXCL10 with a neutralizing mAb. In fact, in this experimental setting, antagonism of CXCL10 led to an attenuation of ConA-induced liver injury and cell death, again suggesting a direct effect of CXCL10 on hepatic cells. Notably, these results are in line with earlier findings of the relevance of CXCL10 in CCl4-injured liver models,9, 11 suggesting model-independent hepatoprotective effects of CXCL10 antagonism in vivo.

In contrast to these potentially deleterious effects of CXCL10 in the CCl4 and ConA models, CXCL10 was reported to mediate hepatoprotective effects during acetaminophen-induced ALI.27 These model-dependent effects of CXCL10 warranted a further systematic exploration as to how CXCL10 directly modulates liver cell injury. Therefore, we isolated hepatocytes and stellate cells from WT mice and exposed these cells to CXCL10. This stimulation of hepatocytes with CXCL10 led to an apparent injury of these cells, associated with sustained Akt phosphorylation. Akt is a critical regulator of PI3K-mediated hepatocyte proliferation and survival. However, a reversed proapoptotic effect of Akt has already been shown in epidermal and neuroblastoma cells.28, 29 Indeed, stimulation of hepatocytes with the PI3K inhibitor, Wortmannin, blocked CXCL10-induced phosphorylation of Akt, suggesting that CXCL10 mediates its proapoptotic effects by prolonged Akt phosphorylation.

Current evidence from mouse studies24, 30 implied the Akt downstream effector, PAK-2, as a critical mediator of apoptotic response. The caspase-cleaved form of PAK-2 (PAK-2p34) is known to induce apoptosis, whereas active PAK-2 has been crucially implicated in survival pathways.22 We found elevated PAK-2p34 levels after caspase-3 activation in hepatocytes in response to CXCL10 stimulation. In our present and also in previous studies,31 caspase-3-mediated PAK-2 cleavage was associated with long-term activation of JNK, a pathway known to be activated by inflammatory cytokines (TGF-ß and OS), finally leading to hepatocyte apoptosis.23 In line with these observations, we could show that CXCL10 also caused the generation of ROS, which, in turn, might amplify the JNK signal. Recently, blocking of JNK has been identified to inhibit the CXCL10-induced cleavage of caspase-3 and PAK-2 in β-cells,24 suggesting that JNK is an upstream mediator of caspase-3 and PAK-2. Interestingly, CXCL10 also induced prolonged Akt and JNK activation in caspase-8-deficient hepatocytes, but did not affect the activity of the proapoptotic factors, caspase-3 and PAK-2p34, confirming that caspase-8 is an upstream protease involved in caspase-3 and PAK-2 cleavage32 in response to CXCL10.

Because CXCL10 is considered to mainly mediate its biological effects by binding to the G-protein-coupled receptor, CXCR3, we next investigated the role of this cognate receptor in hepatocyte apoptosis. Interestingly, CXCL10 induced similar apoptosis-related effects in CXCR3−/− hepatocytes as in WT hepatocytes. Moreover, another CXCR3 agonist (CXCL9) did not affect the apoptotic process in WT cells. Such CXCR3-independent effects of CXCL10 have also been shown in ECs.33 Because TLR4 was recently suggested as a noncognate receptor for CXCL10,24 we next assessed the proapoptotic effects of CXCL10 in hepatocytes isolated from TLR4-deficient mice. Indeed, we could not detect changes in Akt and caspase-3 activation in these cells, suggesting that this signaling pathway is also functional in hepatocytes. Because our in vitro findings provided evidence for a direct effect of CXCL10 on hepatocytes, we next investigated the possibility to induce apoptotic liver injury in vivo by systemic CXCL10 application. Indeed, systemic CXCL10 challenge led to an increased number of apoptotic liver cells in WT mice. These results were confirmed by increased activation of caspase-3 and caspase-8 within the liver as well as by elevated AST levels in serum. In contrast to WT mice, TLR4-deficient mice were resistant to CXCL10-induced liver cell apoptosis and injury, identifying the CXCL10/TLR4 axis as the first chemokine-based apoptotic pathway of hepatocytes. Although TLR4 on stellate cells34 and Kupffer cells35 are known to be implicated in distinct features of liver disease,36 the functional role of its expression on hepatocytes has not yet been clearly defined. Here, we provide first evidence that TLR4 signaling in hepatocytes is a prerequisite for the development of liver injury triggered by apoptotic events in response to increased CXCL10 expression.

In summary, our results define CXCL10-induced TLR4 activation as a noncognate chemokine pathway specifically involved in hepatocyte apoptosis. Through TLR4, but not its cognate receptor (CXCR3), CXCL10 induces long-term Akt and JNK activation, which switches toward hepatocyte apoptosis by caspase-3 and PAK-2 cleavage (Supporting Fig. 5). From a clinical perspective, therapeutical modulation of the CXCL10/TLR4 pathway might be a promising target for liver diseases associated with augmented hepatocyte apoptosis.