Potential conflict of interest: Nothing to report.
Liver fibrosis is a major cause of morbidity and mortality worldwide. Platelets are involved in liver damage, but the underlying molecular mechanisms remain elusive. Here, we investigate the platelet-derived chemokine (C-X-C motif) ligand 4 (CXCL4) as a molecular mediator of fibrotic liver damage. Serum concentrations and intrahepatic messenger RNA of CXCL4 were measured in patients with chronic liver diseases and mice after toxic liver injury. Platelet aggregation in early fibrosis was determined by electron microscopy in patients and by immunohistochemistry in mice. Cxcl4−/− and wild-type mice were subjected to two models of chronic liver injury (CCl4 and thioacetamide). The fibrotic phenotype was analyzed by histological, biochemical, and molecular analyses. Intrahepatic infiltration of immune cells was investigated by fluorescence-activated cell sorting, and stellate cells were stimulated with recombinant Cxcl4 in vitro. The results showed that patients with advanced hepatitis C virus–induced fibrosis or nonalcoholic steatohepatitis had increased serum levels and intrahepatic CXCL4 messenger RNA concentrations. Platelets were found directly adjacent to collagen fibrils. The CCl4 and thioacetamide treatment led to an increase of hepatic Cxcl4 levels, platelet activation, and aggregation in early fibrosis in mice. Accordingly, genetic deletion of Cxcl4 in mice significantly reduced histological and biochemical liver damage in vivo, which was accompanied by changes in the expression of fibrosis-related genes (Timp-1 [tissue inhibitor of matrix metalloproteinase 1], Mmp9 [matrix metalloproteinase 9], Tgf-β [transforming growth factor beta], IL10 [interleukin 10]). Functionally, Cxcl4−/− mice showed a strongly decreased infiltration of neutrophils (Ly6G) and CD8+ T cells into the liver. In vitro, recombinant murine Cxcl4 stimulated the proliferation, chemotaxis, and chemokine expression of hepatic stellate cells. Conclusion: The results underscore an important role of platelets in chronic liver damage and imply a new target for antifibrotic therapies. (HEPATOLOGY 2010.)
Liver fibrosis is the sequel of chronic liver diseases and represents the common pathway leading to end-stage liver failure and hepatocellular carcinoma in a significant number of patients. Among chronic liver diseases, hepatitis C infection and steatohepatitis are the main causes of liver cirrhosis and liver transplantation in Europe and the United States.1 Despite advances in the understanding of liver fibrosis, key molecular pathways are still to be defined.2 However, the elucidation of such critical regulatory pathways seems of great clinical importance because the potential for resolution of liver fibrosis has already been demonstrated in different animal models.3
It has recently been appreciated that inflammatory and coagulation pathways are both intrinsically involved in the pathophysiology of liver scarring.2, 4, 5 The importance of the plasmatic coagulation cascade is strongly suggested by an association of progressive liver fibrosis with a genetic host procoagulant status6, 7 and the inhibition of experimental liver fibrosis by anticoagulation therapy with warfarin in vivo.8 Platelets are also known to be actively recruited to the liver where they mediate organ damage in vivo,9, 10 which is partly dependent on the secretion of platelet-derived serotonin.10, 11 The importance of platelets is further supported by the finding that treatment of mice with antiplatelet drugs ameliorates diet-induced liver fibrosis,12 implying a potential treatment option in patients with chronic liver diseases.5 On the other hand, platelet serotonin is needed for proper liver regeneration after partial hepatectomy or ischemia/reperfusion injury.11, 13 Nevertheless, apart from serotonin,10 the molecular mediators underlying the pathophysiological role of platelets in chronic liver diseases are yet to be defined.
Platelets rapidly release high amounts of chemokines upon activation.14 Among these chemokines, chemokine (C-X-C motif) ligand 4 (CXCL4) (also known as platelet factor 4 [PF4]) and CXCL7 are the most abundant and are stored in micromolar concentrations in the α-granules of platelets.15 CXCL4, a 70–amino acid, lysine-rich, 7.8-kDa chemokine, is mainly synthesized by megakaryocytes and comprises 2%-3% of the total released protein from mature platelets.16 CXCL4 avidly binds to glycosaminoglycans, but only a splice-variant of the human chemokine receptor CXCR3 (CXCR3B), which is not present in mice, has been identified as a high-affinity receptor.17 Notably, CXCL4 has already been shown to be an important mediator of atherosclerosis in vitro18 and in vivo,19 supporting its biological relevance in chronic fibroproliferative and inflammatory conditions. Recently, CXCL4 messenger RNA (mRNA) has also been detected at elevated levels in the livers of patients with alcoholic liver disease,20 and CXCL4 serum levels were elevated in patients with viral hepatitis compared to healthy individuals,21 indirectly implying its potential involvement in liver diseases.
In the light of these findings, we hypothesized an important role of the platelet-derived chemokine CXCL4 in human and experimental liver fibrosis. Accordingly, we show here that CXCL4 is up-regulated in human liver fibrosis and that it plays a nonredundant, functional role in experimental liver fibrosis by mediating stellate cell proliferation, migration, and intrahepatic immune cell recruitment.
Serum Samples and Liver Biopsies from Patients with Liver Fibrosis.
Overall, we determined CXCL4 serum concentrations in 100 patients with hepatitis C virus (HCV) infection (Supporting Table 1 and Supporting Methods) and 25 healthy subjects. After histopathological evaluation, we extracted mRNA from paraffin-embedded liver biopsies of the patients with HCV and 42 patients with fatty liver or nonalcoholic steatohepatitis (NASH) with fibrosis as described recently.22
CXCL4 Serum Concentrations and Intrahepatic CXCL4 mRNA in Patients with HCV.
CXCL4 concentrations in serum of patients with mild to moderate and severe fibrosis and a normal platelet count as well as in healthy subjects were determined by enzyme-linked immunosorbent assay (ELISA; R&D Systems, Minneapolis, MN). Total hepatic mRNA from patients with HCV and fatty liver/NASH was reverse-transcribed using Super-Script (Invitrogen). Quantitative reverse transcription polymerase chain reaction (RT-PCR) was carried out for CXCL4 with an Assay on Demand from Applied Biosystems (Hs00236998_m1). Target gene expression was normalized to 18S ribosomal RNA levels.
Transmission Electron Microscopy of Platelets in Liver Fibrosis.
Platelets in livers of HCV-infected patients and their association with extracellular matrix were visualized by transmission electron microscopy. For detailed methods, please refer to Supporting Methods.
Mice with a targeted deletion of the Cxcl4 gene were established on the C57BL/6 background as described19 and backcrossed for at least 10 generations. Control C57BL/6 mice were purchased from Charles River Laboratories (Sulzfeld, Germany). Acute and chronic liver injury was induced as published22 and described in detail in the Supporting Methods.
Expression Analysis of Fibrosis-Related Genes and Chemokines.
Total RNA was isolated from the livers of chronically treated mice (n = 10-12 per strain) by guanidinium thiocyanate-phenol-chloroform extraction 3 days after the last injection of CCl4 or thioacetamide (TAA), respectively. RNA was reversely transcribed using SuperScript (Invitrogen). Quantitative RT-PCR was then carried out for collagen type I alpha 1 (Col1a1), tissue inhibitor of metalloproteinase 1 (Timp1), transforming growth factor beta (Tgf-β), matrix metallopeptidase 9 (Mmp9), Kc (Cxcl1), and interleukin-10 (IL-10) with Assays on Demand on an ABI-Prism 7000 Sequence Detection System (Applera). Assays are available from the Applied Biosystems database (available on the Applied Biosystems web site). All data were normalized to 18S ribosomal RNA levels.
Intrahepatic and Stellate Cell Secreted Chemokine Analysis.
Intrahepatic concentrations of Cxcl4, CCl5, and Kc and chemokine secretion by hepatic stellate cells were determined by ELISA as described in the Supporting Methods.
In Vitro Stimulation of Hepatic Stellate Cells.
Murine stellate cells were obtained and cultured as previously described.23 Chemokine stimulation experiments were performed in the stellate cell line GRX.24 The exact culture conditions are described in the Supporting Methods. Proliferation of cells was determined by [3H]thymidine incorporation.23 Migration of stellate cells toward recombinant murine Cxcl4 was determined in a modified Boyden chamber as described in the Supporting Methods. For stimulation with murine Cxcl4 (R&D Systems), stellate cells were cultured in Dulbecco's modified Eagle medium with 10% fetal bovine serum, followed by serum starvation and treatment with Cxcl4. RNA was isolated from cultures by RNeasy (Qiagen). Quantitative RT-PCR was then carried out for chemokine (C-C motif) ligand 5 (Ccl5, also known as RANTES), Kc (Cxcl1), Timp-1, Tgf-β, and α-smooth muscle actin with Assays on Demand (Applied Biosystems). Ccl5 protein concentration in the supernatant was determined by ELISA (R&D Systems) and collagen content was determined by western blot, as described recently.25
We also stimulated hepatic stellate cells with murine platelet-derived growth factor (PDGF; 20 ng/mL), TGF-β (20 ng/mL), and tumor necrosis factor-α (2 ng/mL) to induce the secretion of Cxcl4 from these cells. Platelets stimulated with adrenalin were used as positive control.
Fluorescence-Activated Cell Sorting Analysis of Intrahepatic T Cells and Platelets.
After induction of fibrosis for 6 weeks, flow cytometry analysis of intrahepatic immune cells was performed as described in the Supporting Methods.
Immunohistochemistry for Intrahepatic Neutrophils and Platelets.
Cryosections of livers were incubated with rat anti-mouse anti-Ly6G antibody for staining of neutrophils or with anti-CD62P (both) for platelets. After washing with phosphate-buffered saline/sodium acid-solution, the sections were incubated for 45 minutes with donkey anti-rat AlexaFluor 488 antibody (Molecular Probes, Invitrogen) or biotinylated goat anti-rat immunoglobulin G (BD Biosciences), respectively.
Data are given as mean ± standard error of the mean. Continuous variables between groups of animals were compared with Student t test with Welsh correction in case of unequal variances. P values of less than 0.05 were considered as statistically significant. All statistical analyses were performed using GraphPad Prism version 5.0 Software (GraphPad Software, La Jolla, CA).
CXCL4 Serum Levels and Hepatic mRNA Expression Are Associated with Progressive Fibrosis in HCV Infection.
We first evaluated an association of CXCL4 with human liver fibrosis. The serum concentration of CXCL4 was significantly increased in HCV-infected patients with moderate to severe (F2/3) liver fibrosis compared to individuals with no or only mild fibrosis (F0/1, P < 0.05) or to healthy controls (mean 37.0 ± 2.3 ng/mL, P < 0.05; Fig. 1A). We next assessed whether these increases in CXCL4 serum levels were also reflected by an increased intrahepatic CXCL4 mRNA expression. Indeed, hepatic CXCL4 mRNA was significantly increased in subjects with progressive fibrosis versus individuals with only mild fibrosis (P < 0.01; Fig. 1B), supporting our serum results and recent findings from patients with alcoholic and viral hepatitis.20, 21 Notably, we could also confirm increased CXCL4 mRNA in patients with NASH and fibrosis compared to subjects with fatty liver without fibrosis (Fig. 1C; P < 0.01). Because platelets are the predominant source of CXCL4, we next assessed whether platelets are morphologically associated with the extracellular matrix in liver fibrosis. Platelets aggregate within damaged liver and are directly adjacent to collagen fibrils in human liver fibrosis (Supporting Fig. 1; Fig. 1D).
Toxic Liver Injury in Mice Leads to Platelet Activation and Augmented Cxcl4 Protein Concentrations.
We next determined whether toxic liver injury is associated with activation of platelets in mice. Administration of CCl4 leads to a significantly increased number of CD62P (P-selectin)-positive platelets compared to control mice (P < 0.01; Fig. 2A,B). Notably, platelets aggregate within the liver early (2 weeks CCl4 treatment) during the fibrogenic response (Fig. 2C). The increased number of activated and aggregated platelets after CCl4 treatment is associated with significantly augmented intrahepatic Cxcl4 protein concentrations (Fig. 2D; both P < 0.01). This increase in Cxcl4 is inhibited by pretreatment of the mice with the antiplatelet drug aspirin (P < 0.05), supporting that platelets are the main source of Cxcl4 in vivo. Of note, treatment with CCl4 or TAA for 2-6 weeks to induce fibrogenesis also resulted in increased hepatic Cxcl4 concentrations (both P < 0.05; Fig. 2E; Supporting Fig. 2A).
Deletion of Cxcl4 Inhibits Liver Fibrosis In Vivo.
Our finding of an association between CXCL4 expression and liver injury in humans and mice stressed the question if Cxcl4 plays a causal role in experimental liver fibrosis. To study the functional relevance, chronic liver damage was induced by two well-established, independent models of fibrotic tissue injury, TAA and CCl4,26 in wild-type and Cxcl4−/− mice. Cxcl4−/− mice displayed significantly reduced liver fibrosis compared to wild-type mice in both models (Fig. 3A-D). Importantly, the difference between wild-type and knockout animals was evident by histological, biochemical (mean hydroxyproline reduction of 16% in the CCl4 and 18% in the TAA model, respectively), and molecular (Col1a1 RT-PCR) parameters of fibrotic tissue injury (all P < 0.05), suggesting a nonredundant effect of Cxcl4 on tissue scarring in vivo.
Deletion of Cxcl4 Is Associated with Altered Expression of Fibrosis-Related Genes and Reduced Hepatic Immune Cell Infiltration.
We next studied mRNA expression of key genes of fibrogenesis in our animals. As depicted in Fig. 4A, the mRNA of Timp-1 and Tgf-β was significantly reduced in Cxcl4−/− mice compared to wild-type mice after chronic liver injury (P < 0.01 and P < 0.05, respectively) with CCl4, whereas expression of Mmp9 and the antifibrotic cytokine IL-10 were significantly increased in (P < 0.05 and P < 0.01, respectively; Fig. 4A,F). Importantly, these changes were confirmed in the second fibrosis model (TAA; Supporting Fig. 2B).
In other models of liver injury, platelets have been shown to be adjacent to damaged hepatocytes and infiltrating inflammatory cells.9 We therefore hypothesized that the activation of platelets in our fibrosis models would lead to local release of Cxcl4 and thereby influence the recruitment of immune cells. Indeed, the mRNA expression (Fig. 4B; P < 0.01) and protein expression (Supporting Fig. 3C) of the neutrophil chemoattractant KC (Cxcl1) was significantly changed in Cxcl4−/− compared to wild-type mice. Functionally, reduced Cxcl1 expression was associated with a diminished infiltration of neutrophils into the livers of Cxcl4−/− mice (Fig. 4C).
Recently, it has also been postulated that Cxcl4 promotes T cell trafficking,27 which might be favored by its interaction with the chemokine Ccl5.28 Therefore, we assessed the infiltration of T cells to the livers in our experimental models and observed reduced numbers of CD3+ T cells, and specifically CD8+ T cells within the livers of Cxcl4−/− mice (Fig. 4D; P < 0.05). A representative fluorescence-activated cell sorting (FACS) blot is depicted in Fig. 4E, which shows the reduced percentage of CD3+CD8+ T cells (right lower panel with dark red line) in an individual Cxcl4−/− mouse compared to a wild-type mouse. The diminished infiltration of T cells in Cxcl4−/− mice was associated with a trend toward lower Ccl5 protein levels in the CCl4 model and significantly reduced Ccl5 levels in the TAA model (Supporting Fig. 3A,B).
Cxcl4 Induces Proliferation, Chemotaxis, and Chemokine Expression in Stellate Cells.
We next investigated the direct effects of Cxcl4 on hepatic stellate cells in vitro. Recombinant murine Cxcl4 significantly induced the proliferation and migration of stellate cells in a dose-dependent manner (Fig. 5A,B). Both aspects are considered important features of stellate cells during the wound-healing response, eventually leading to liver fibrosis. Notably, stellate cells do not secrete relevant amounts of Cxcl4 after stimulation with PDGF, TGF-β, or TNF-α (Supporting Fig. 4A), nor does Cxcl4 induce the expression of matrix-related genes or collagen production (Supporting Fig. 4B).
However, stellate cells have also been shown to regulate the recruitment of immune cells into the injured liver by means of chemokine secretion.29 We therefore assessed whether Cxcl4 can induce Ccl5 and Cxcl1 expression in stellate cells and thereby modify the infiltration of neutrophils and T cells in our models. Cxcl4 indeed stimulated the Ccl5 mRNA (Fig. 5C), protein (Fig. 5D), and Cxcl1 mRNA (Fig. 5E) expression of hepatic stellate cells in vitro, supporting the immunoregulatory role of these cells in liver fibrosis.4
Our results establish the platelet-derived chemokine CXCL4 as a mediator of liver fibrosis in two independent animal models of chronic liver injury in vivo and provide first data for its involvement in human fibrotic liver disease. The findings across two species in our study suggest that CXCL4 belongs to a conserved molecular pathway which plays an important, nonredundant role in liver damage and other chronic inflammatory conditions.19, 27
In contrast to the well-documented role of coagulation pathways, the impact of platelets in liver injury has only recently been recognized.5, 11 However, platelets are among the first cells that are recruited to the site of injury and are thought to play an important role in the wound-healing process. After recruitment and activation, they release an array of proteins, including different mediators which are known to be intrinsically involved in fibrogenesis. These include the platelet-derived growth factors (PDGFs) and TGF-β.30 In line with these results, we found platelet aggregation in early fibrogenesis (Fig. 2C; Supporting Fig. 1) and show that platelets are among the most abundant cells in direct proximity to collagen fibrils (Fig. 1D).
Platelets are also a particularly rich source of chemokines that have been implicated in organ inflammation and injury,31 including the liver.32, 33 However, the specific role of one of the most abundant chemokines in platelets, CXCL4, has not been specifically addressed with regard to liver fibrosis. Thus, we first assessed serum concentrations of CXCL4 in relation to the severity of liver fibrosis in HCV-infected patients and demonstrate a significant increase in patients with moderate to severe fibrosis compared to individuals with only mild fibrosis or healthy controls. These data extend the reported higher concentrations of CXCL4 in patients with viral hepatitis compared to healthy controls.21 Importantly, all of our patients had platelet counts within the normal range, because individuals with severe fibrosis and cirrhosis experience a decrease in CXCL4 serum levels owing to reduced numbers of platelets.34 We also demonstrate an increased hepatic CXCL4 mRNA expression in patients with progressive HCV and NASH-associated fibrosis. These results are in line with recent data of an increased CXCL4 mRNA expression in subjects with alcoholic hepatitis,20 suggesting that CXCL4 might be involved in chronic liver damage of different etiologies.
However, the statistical association of CXCL4 serum levels and intrahepatic mRNA expression with fibrosis cannot differentiate between cause and effect in human diseases. We therefore evaluated the functional role of Cxcl4 in different rodent models of experimental liver fibrosis. For this purpose, we challenged constitutive Cxcl4−/− and wild-type mice with CCl4 and TAA, two models which generate repeated and overlapping phases of injury, inflammation, and wound-healing that mimick the pathological events underlying the development of fibrosis in chronic liver disease. We show that treatment of mice with these agents indeed leads to a significantly increased number of activated platelets, platelet aggregation during early fibrogenesis, and an associated increase in hepatic Cxcl4 expression. Accordingly, deletion of Cxcl4 leads to a significantly reduced degree of fibrotic liver injury in two animal models. These results are comparable with recent reports in which a reduced propensity of Cxcl4−/− mice to atherosclerosis19 and cerebral malaria27 have been demonstrated.
Because platelets are the predominant source of CXCL4,14 our results extend earlier studies which showed an increased injury mediated by platelets in experimental viral liver damage.9, 10 In the latter two studies, a direct effect of platelets and platelet-derived serotonin on the infiltration of T lymphocytes in the liver was demonstrated. We therefore assessed the recruitment of T cells and neutrophils, another important cell population in liver fibrosis,35 in our Cxcl4−/− mice. Deletion of Cxcl4 resulted in strongly reduced infiltration of neutrophils and CD8-positive T cells after liver damage (Fig. 4). These results seem functionally important because neutrophils have already been implicated in the development of liver damage via their production of reactive oxygen species and various hydrolytic enzymes.36, 37 Furthermore, the contribution of intrahepatic T cells to fibrogenesis is widely accepted4 and CD8-positive T lymphocytes have been directly involved in liver fibrosis in adoptive transfer experiments.38
To further investigate the role of stellate cells in Cxcl4-mediated liver fibrosis, we stimulated these cells with recombinant Cxcl4. Although the synthesis of collagen or the expression of matrix-associated genes (including Timp-1, Tgf-β) was not changed in vitro, the proliferation and chemotactic migration of stellate cells was significantly enhanced by Cxcl4. These features are considered as a crucial event in the stellate cell activation process and have been described for other fibrogenic chemokines.39, 40 Because stellate cells have recently been shown to regulate the migration and positioning of leukocytes into the liver,29 we further assessed the Cxcl4-induced chemokine expression in stellate cells and found increased expression of Ccl5 and Kc (Cxcl1) on the mRNA and protein level, respectively. This increased chemokine secretion by Cxcl4 stimulated stellate cells in vitro and the known chemotactic activity of Cxcl4 on T cells27 likely contributes to the inflammatory infiltrate after administration of CCl4 or TAA. Based on this observation, we hypothesize that the recruited immune cells further lead to increased stellate cell activation and the consecutive excess production of collagen by wild-type mice compared to the Cxcl4−/− mice in vivo. The observed secretion of Ccl5 by stellate cells might be especially important because CXCL4 shows heterophilic interactions with this chemokine in humans and mice, thereby augmenting the effects of CCL5 on immune cells in vitro28 and in vivo.41 The evaluation of this interaction in liver fibrosis could be of therapeutical interest as we recently developed a peptide-based therapeutic approach, which targets inflammatory actions of platelet-derived Cxcl4 by disrupting the functional synergism of Cxcl4 and Ccl5 in vivo.41
In conclusion, we have herein identified the platelet-derived chemokine Cxcl4 as a nonredundant mediator of experimental liver fibrosis in vitro and in vivo. Furthermore, we could associate this chemokine with liver fibrosis in humans with different liver diseases. These congruent results in humans and mice reinforce the concept that platelets play an important role in chronic liver damage9, 10, 12 and describe a potential new molecular target for the development of platelet-associated antifibrotic therapies in liver diseases.
The study was supported by grants from the Deutsche Forschungsgemeinschaft (SFB/TRR 57) and the Medical Faculty of the University of Aachen (START and IZKF BIOMAT grants to P.v.H. and H.E.W.).