CARTILAGE OLIGOMERIC MATRIX PROTEIN PARTICIPATES IN THE PATHOGENESIS OF LIVER FIBROSIS

Abstract

Background & Aims

Liver fibrosis is characterized by significant accumulation of extracellular matrix (ECM) proteins, mainly fibrillar collagen-I, as a result of persistent liver injury. Cartilage oligomeric matrix protein (COMP) is largely found in the ECM of skeletal tissue. Increased COMP expression has been associated with fibrogenesis in systemic sclerosis, lung fibrosis, chronic pancreatitis, cirrhosis and hepatocellular carcinoma. We hypothesized that COMP could induce fibrillar collagen-I deposition and participate in matrix remodeling thus contributing to the pathophysiology of liver fibrosis.

Methods

Thioacetamide (TAA) and carbon tetrachloride (CCl₄) were used to induce liver fibrosis in wild-type (WT) and Comp^−/− mice. In vitro experiments were performed with primary hepatic stellate cells (HSCs).

Results

COMP expression was detected in livers from control WT mice and was up-regulated in response to either TAA or CCl₄-induced liver fibrosis. TAA-treated or CCl₄-injected Comp^−/− mice showed less liver injury, inflammation and fibrosis compared to their corresponding control WT mice. Challenge of HSCs with recombinant COMP (rCOMP) induced intra- plus extracellular collagen-I deposition and increased matrix metalloproteinases (MMPs) 2, 9 and 13, albeit similar expression of TGFβ protein in addition to Tgfβ, Tnfα and tissue inhibitor of metalloproteinases-1 (Timp1) mRNAs. We demonstrated that COMP binds collagen-I; yet, it does not prevent collagen-I cleavage by MMP1. Last, rCOMP induced collagen-I expression in HSCs via CD36 receptor signaling and activation of the MEK1/2-pERK1/2 pathway.

Conclusion

These results suggest that COMP contributes to liver fibrosis by regulating collagen-I deposition.

Graphical Abstract

INTRODUCTION

A key event participating in the pathogenesis of liver fibrosis is the significant increase in extracellular matrix (ECM) deposition in response to persistent liver damage of various etiologies. Indeed, the presence of fibrosis is considered a prognostic marker for progression of chronic liver injury eventually resulting in cirrhosis and hepatocellular carcinoma in many patients [1, 2]. During fibrogenesis the liver undergoes significant changes in the quality, quantity, composition and arrangement of the ECM components with a major increase in fibrillar collagen, primarily collagen-I [3], disrupting the hepatic architecture and compromising the normal liver function [4]. Matrix metalloproteinases (MMPs) and tissue inhibitors of metalloproteinases (TIMPs) are at the forefront of ECM remodeling and changes in their physiological balance also lead to disruption of the ECM dynamics hence contributing to scarring [5].

Although several cell types participate in the development of liver fibrosis, hepatic stellate cells (HSCs), located in the space of Disse between the sinusoidal endothelial cells and the hepatocytes, are the primary cells responsible for fibrillar collagen-I deposition. Following the onset of liver injury, they differentiate into myofibroblasts and acquire contractile, migratory, pro-inflammatory and pro-fibrogenic features [6]. HSC activation is driven by multiple mediators such as reactive oxygen species, chemokines, growth factors, matrix stiffness, matricellular proteins and damage-associated molecular patters, which are also secreted by neighboring cells and signal to drive scarring by HSCs in an autocrine and/or paracrine fashion [7, 8].

Cartilage oligomeric matrix protein (COMP), also referred to as thrombospondin-5, is a glycoprotein mainly found in the ECM of cartilage, synovium, ligaments and tendons [9]. It has five identical subunits linked by disulfide bonds forming a large protein of 524 kDa [10]. COMP can bind collagen I, II and IX [11–13] with high affinity via its C-terminal domain, as well as other ECM components such as fibronectin [14], matrilins [15], proteoglycans [16] and heparin [17].

Increased COMP has been associated with fibrogenesis in systemic sclerosis [18], skin keloids [19, 20], vascular atherosclerosis [21], lung fibrosis [22] and chronic pancreatitis [23] as well as with other conditions such as rheumatoid arthritis [24, 25], osteoarthritis [26, 27], pseudoachondroplasia [28], acute trauma [29] and systemic lupus erythematosus [30, 31]. Recently, COMP has been proposed as a novel non-invasive marker for assessing cirrhosis and the risk of hepatocellular carcinoma [32] and as a biomarker of liver fibrosis in chronic hepatitis C [33]. However, to date, the role of COMP and its mechanism of action in the context of liver fibrosis have not been fully defined. Since COMP could promote scarring, our aim was to study the potential involvement of COMP in the development of liver fibrosis and to dissect its ability to modulate pathological collagen-I deposition, thereby contributing to the onset and/or progression of liver fibrosis.

MATERIALS AND METHODS

General methodology

Details on general methodology such as western blot analysis, H&E staining, collagen-I immunohistochemistry (IHC), Sirius red/fast green staining and measurement of ALT and AST activities have been described in our previous publications [34, 35]. The source of commercially available Abs can be found in Supplementary Table 1. The collagen-I antibody was generated and provided by Dr. Schuppan (University of Mainz, Mainz, Germany). The qPCR primers used are shown in Supplementary Table 2.

Cell culture and treatments

Due to the almost complete homology in all the proteins from our study in rat and mice, all the in vitro experiments were carried out with rat HSCs due to their higher yield. Primary rat HSCs were isolated as previously [34, 36]. HSCs were seeded on 6-well plates (300,000 cells/well) in DMEM/F12 supplemented with 10% FBS, fungizone, penicillin and streptomycin. The medium was replaced with serum-free medium 12 h before treatment with 0–25 nM of human recombinant COMP (rCOMP) (R&D Systems, Minneapolis, MN). In some experiments, 1 μM of PD98059 (a mitogen-activated protein kinase MEK1/2 inhibitor) (Calbiochem, San Diego, CA) and 33 nM of mouse CD36 neutralizing Ab Clone FA6-152 or mouse IgG1 kappa isotype control Ab Clone MOPC-21 (StemCell Technologies Inc, Vancouver, BC, Canada), were added to the cells 1 h prior to incubation with rCOMP. The optimal concentration for collagen-I induction was determined experimentally.

Mice

Comp^−/− mice (C57BL/6J) and their wild-type (WT) littermates were donated by Dr. Åke Oldberg (Lund University, Lund, Sweden) [37]. The targeting vector constructed to generate these mice disrupted exon 8 in the Comp gene. Colonies were established by intercrossing Comp^+/− mice and control littermates (referred to as WT) were used in all experiments. Comp^−/− mice do not show any anatomical, histological or ultrastructural abnormalities [37].

Induction of liver fibrosis

Two models of liver fibrosis were used in 10 wks old male Comp^−/− and WT mice. In the first model, mice were given thioacetamide (TAA, Sigma, St. Louis, MO) at a dose of 300 mg/L in the drinking water or equal volume of water in the control group for 4 mos. In the second model, mice were injected carbon tetrachloride (CCl₄, Acros Organics, New Jersey, NJ) dissolved in mineral oil (MO) at a dose of 0.5 ml/kg i.p. or equal volume of MO in the control group twice a week for 1 mo. Thus, 4 groups (n=6 mice/group) were included in each model: WT + Water or MO; WT + TAA or CCl₄; Comp−/− + Water or MO and Comp−/− + TAA or CCl₄. Mice were sacrificed under ketamine/xylazine anesthesia 48 h after TAA withdrawal or after the last injection of CCl₄ to avoid an acute response over chronic liver injury. All mice received humane care according to the criteria outlined in the Guide for the Care and Use of Laboratory Animals prepared by the National Academy of Sciences and published by the National Institutes of Health. Protocols were approved by the Icahn School of Medicine at Mount Sinai Institutional Animal Care and Use Committee.

Please see the Supplementary Material for additional information.

Statistical analysis

Data are expressed as mean ± standard deviation. Statistical comparisons among groups and treatments were performed using the paired Student’s t-test and the two-factor analysis of variance (ANOVA). All the in vitro experiments were performed in triplicate at least four times. A representative image or blot is shown in all figures.

RESULTS

COMP expression is up-regulated in the TAA and in the CCl₄ models of liver fibrosis in WT but is absent in Comp^−/− mice

To determine whether COMP expression increased in liver fibrosis, WT and Comp^−/− mice were either administered TAA or water for 4 mos or injected CCl₄ or MO for 1 mo; both are well-established models of drug-induced liver fibrosis [34, 35]. Cytochrome P450 activity, the key enzyme involved in the metabolism of these two drugs was similar in WT and Comp^−/− mice (Supplementary Fig. 1A–1B). IHC analysis revealed positive staining for COMP in WT mice, which was increased in zones 1 and 3 by either TAA treatment or CCl₄ injection, yet it was totally absent in Comp^−/− mice (Fig. 1A–1B). To identify the source of hepatic COMP, frozen liver sections were immunostained with COMP or arginase-I (hepatocyte marker), desmin (HSC marker) or F4/80 (macrophage marker). Co-localization studies demonstrate that COMP is mostly expressed in hepatocytes but also to a minor extent in Kupffer cells and HSCs (Fig. 1C) and it is induced by TAA (Fig. 1D) or CCl₄ in WT but not in Comp^−/− mice (not shown). This was also confirmed by qPCR (not shown).

Figure 1. COMP expression is up-regulated in the TAA and in the CCl₄ models of liver fibrosis in WT but is absent in Comp^−/− mice.

IHC analysis in liver sections shows minimal COMP positive staining in WT mice given either water or mineral oil (MO) and significant induction of COMP staining in WT mice treated with TAA or injected with CCl₄. No staining is present in Comp^−/− mice (A and B). Double immunofluorescence shows localization of COMP mostly in hepatocytes but also to a lesser extent in Kupffer cells and HSCs (C) and induction of COMP in TAA (D) or CCl₄-injected WT (not shown) but not in Comp−/− mice. White arrows indicate co-localization of COMP. Hepatocytes were identified by arginase-1, HSCs by desmin and Kupffer cells by F4/80 staining; n=6/group.

TAA-treated and CCl₄-injected Comp^−/− are protected from liver injury compared to control WT mice

Since COMP was induced following TAA treatment or CCl₄ injection, to further investigate the role of COMP in liver fibrosis, liver sections were processed for H&E staining. TAA-treated or CCl₄-injected Comp^−/− showed less periportal and pericentral necrosis and inflammation compared to their corresponding control WT mice (Fig. 2A–2B). Moreover, TAA-treated and CCl₄-injected Comp^−/− had less hepatic macrophage infiltration (Supplementary Fig. 2A–2B) yet, there was no difference in the number of neutrophils (not shown). Importantly, rCOMP did not induce migration of macrophages in vitro (Supplementary Fig. 3A).

Figure 2. TAA-treated and CCl₄-injected Comp^−/− are protected from liver injury compared to control WT mice.

H&E staining reveals more centrilobular and periportal necrosis and inflammation in TAA-treated and CCl₄-injected WT compared to Comp−/− mice (A–B). Serum ALT and AST activities are increased in TAA-treated and in CCl₄-injected WT compared to Comp^−/− mice. Results are expressed as mean ± standard deviation; n=6/group, *p<0.05 and **p<0.01 for TAA-treated or CCl₄-injected vs. corresponding control mice; ^•p<0.05, ^••p<0.01 and ^•••p<0.001 for Comp^−/− vs. WT mice (C–D).

The expression of inflammation markers was significantly reduced in TAA-treated and CCl₄-injected Comp^−/− compared to WT mice (Supplementary Fig. 3B–3C). Serum ALT and AST activities were lower in TAA-treated and CCl₄-injected Comp^−/− compared to WT mice (Fig. 2C–2D). Overall, these data suggest that ablation of Comp protects mice from both TAA and CCl₄-induced liver injury.

TAA-treated and CCl₄-injected Comp^−/− are protected from liver fibrosis compared to control WT mice

To evaluate the potential role of COMP in liver fibrosis, the Sirius red/fast green staining for 4-hydroxyproline, a surrogate of total collagen content, and IHC for fibrillar collagen-I were performed. TAA-treated and CCl₄-injected Comp^−/− exhibited less fibrosis compared to WT mice, as shown by both whole-slide scanning and quantification of the Sirius red/fast green staining (Fig. 3A–3B) and collagen-I IHC (Fig. 3C–3D). Likewise, western blot and qPCR analysis from total liver lysates confirmed that there is less collagen-I protein and Col1a1 mRNA expression both in TAA-treated or CCl₄-injected Comp−/− compared to control WT mice (Fig. 3E–3F). Overall, these results suggest that COMP plays a key role in both TAA and CCl₄-induced liver fibrosis contributing to pathological scar formation.

Figure 3. TAA-treated and CCl₄-injected Comp^−/− are protected from liver fibrosis compared to control WT mice.

Liver sections stained with Sirius red/fast green show more total collagen deposition in TAA-treated or CCl₄-injected WT compared to Comp^−/− mice (A–B). IHC analysis confirmed lower collagen-I staining in TAA-treated or CCl₄-injected WT compared to Comp−/− mice (C–D). Western blot and qPCR show that collagen-I and Col1a1 mRNA expression were reduced in Comp^−/− mice under both treatments. The quantification was referred to the WT control which was assigned a value of 1. Col1a1 mRNA expression was normalized to that of Gapdh (E–F). Results are expressed as mean ± standard deviation; n=6/group, *p<0.05, **p<0.01 and ***p<0.001 for TAA-treated or CCl₄-injected vs. corresponding control mice; ^*p<0.05, ^**p<0.01 and ^***p<0.001 for Comp^−/− vs. WT mice.

rCOMP up-regulates intra- and extracellular collagen-I deposition without altering HSCs viability, inducing phenotypic changes or stimulating cell proliferation

Since the in vivo data suggested a potential role for COMP in the development of liver fibrosis, next we studied how COMP could regulate the HSC pro-fibrogenic behavior. First, we analyzed if a challenge with rCOMP could alter HSC migration, phenotype and/or proliferation rate. Following incubation of primary HSCs with rCOMP for 24 h, light micrographs were acquired and the MTT assay was done. HSCs viability, phenotype, migration and proliferation remained similar in rCOMP-treated than in control non-treated cells (Fig. 4A–4B, Supplementary Fig. 4).

Figure 4. rCOMP up-regulates intra- and extracellular collagen-I deposition without altering HSCs viability, inducing phenotypic changes or stimulating cell proliferation.

Light micrographs show that rCOMP does not induce phenotypic changes (A). The MTT assay shows that rCOMP does not stimulate HSCs proliferation at 7 d of culture when incubated with 0–25 nM rCOMP for 24 h (B). Quiescent (3 d) and activated (6 d) primary HSCs and immortalized rat HSCs were cultured with 0–25 nM rCOMP for 6 and 24 h. Western blot analysis shows that rCOMP up-regulates intra- plus extracellular collagen-I protein expression in primary HSCs peaking at 6 d of culture (C). Western blot analysis confirms the pro-fibrogenic effect of rCOMP in a rat HSC line (D). A value of 1 was assigned to the 6 h time-point in the absence of rCOMP and the following time-points were referred to it. Total collagen-I values were calculated as intracellular collagen-I/calnexin plus extracellular collagen-I/calnexin ratios. The same amount of protein was loaded in all lanes. *p<0.05 for rCOMP vs. control.

We then asked whether rCOMP could up-regulate collagen-I expression in HSCs. Western blot analysis showed that rCOMP increased intra- and extracellular collagen-I protein in primary rat HSCs in a time- and dose-dependent manner peaking at 6 d of culture and between 6 and 24 h post rCOMP challenge (Fig. 4C). Similar effects were observed in a rat HSC line (Fig. 4D). Collectively, these results along with the in vivo findings suggest that intracellular and/or secreted COMP stimulates collagen-I production by HSCs and regulates the fibrogenic response to liver injury.

rCOMP binds collagen-I in vitro

Recent studies have shown that COMP binds matricellular proteins [38]. To better understand how COMP up-regulates intra- and extracellular collagen-I and dissect if it binds fibrillar collagen-I, rCOMP and rat-tail fibrillar collagen-I were incubated for 2 h at room temperature. Silver nitrate staining (Fig. 5A) along with COMP and collagen-I western blots (Fig. 5B) demonstrated that rCOMP has the ability of binding collagen-I. Yet, whether the binding could condition collagen-I stability determining the net accumulation of collagen-I remained undefined.

Figure 5. rCOMP binds to collagen-I in vitro.

Silver nitrate staining shows that rCOMP and collagen-I bind in vitro (A). COMP and collagen-I western blots confirm the binding (B). Arrowheads point at the shift in molecular weight due to the formation of the protein complex.

rCOMP increases MMP13, 2 and 9 proteins but not Timp1 mRNA in rat HSCs

Since collagen-I deposition was increased in HSCs under rCOMP treatment, the expression of several MMPs -known to participate in ECM remodeling- was analyzed in the HSCs culture medium. After rCOMP treatment, MMP13 was significantly up-regulated at 24 h (Fig. 6A). Gelatin zymography also revealed up-regulation of MMP2 and 9 at 24 h, both with collagenolytic activity, likely accounting for the extracellular collagen-I down-regulation at the later time-point (Fig. 6B). However, Timp1 mRNA expression, the inhibitor of MMP1 and 13, remained quite similar in rCOMP-treated compared to control HSCs (Fig. 6C). Therefore, the changes in MMPs protein expression and Timp1 mRNA could barely account for the effects on collagen-I protein expression under rCOMP treatment.

Figure 6. rCOMP increases MMP13, 2 and 9 but not Timp1 in rat HSCs.

Primary rat HSCs were incubated with 0–25 nM rCOMP for 6 and 24 h. Western blot analysis reveals that rCOMP up-regulates MMP13 protein expression in HSCs. A value of 1 was assigned to the 6 h control (A). Gelatin zymography shows that rCOMP increases the activity of MMP2 and 9 in a time-dependent manner. The activities were quantified by densitometry and are given under the blots (B). rCOMP minimally changes Timp1 mRNA expression in HSCs. All data were normalized to the expression of Gapdh and results are expressed as mean ± standard deviation (C). rCOMP binds to collagen-I but does not prevent collagen-I degradation in vitro. The rCOMP and collagen-I protein complex was incubated with MMP1 for either 2 or 24 h. Silver nitrate staining shows that rCOMP binds to collagen-I but does not preclude from collagen-I degradation in vitro (arrowheads indicate the presence of cleavage products). In all panels, *p<0.05, **p<0.01 and ***p<0.001 for rCOMP vs. control.

The binding of rCOMP to collagen-I does not prevent collagen-I cleavage by MMP1

To further elucidate whether the binding of rCOMP to collagen-I could preclude from collagen-I degradation, the rCOMP and collagen-I complex was incubated with MMP1, the MMP with the greatest activity towards collagen-I cleavage, for either 2 or 24 h. Silver nitrate staining showed that the binding of rCOMP to collagen-I does not block collagen-I degradation by MMP1 (Fig. 6D) therefore suggesting that COMP does not increase collagen-I stability and likely that other mechanism(s) accounted for the lower collagen-I protein content observed in the fibrotic Comp^−/− mice.

rCOMP does not alter Tnfα and Tgfβ mRNAs or TGFβ protein expression in HSCs

Since TNFα is known to inhibit collagen-I synthesis [39], we analyzed its expression by qPCR; yet, no major changes were observed in Tnfα mRNA following rCOMP treatment (Supplementary Fig. 5A). Because TGFβ, a well-defined pro-fibrogenic factor [40], is also secreted by HSCs [41], we asked if the rCOMP-mediated collagen-I increase could be a consequence of autocrine TGFβ production and subsequent signaling to HSCs. However, qPCR and western blot analysis revealed that rCOMP neither changed Tgfβ mRNA nor TGFβ protein expression in HSCs (Supplementary Fig. 5B–5C). In addition, qPCR and western blot analysis showed similar expression in Tgfβ mRNA, TGFβ and the ratio of pSmad2/3 to Smad2/3 proteins in total liver from TAA-treated and CCl₄-injected Comp^−/− compared to WT mice (Supplementary Fig. 5D–5E). Taken together, these results suggest that the pro-fibrogenic effects of rCOMP likely occur via TNFα and TGFβ-independent mechanisms and that other pathways may be involved.

rCOMP up-regulates collagen-I expression in HSCs via CD36 receptor signaling and activation of the MEK1/2-pERK1/2 pathway

Last, to further dissect the molecular mechanism underlying the rCOMP-mediated collagen-I up-regulation in HSCs, we analyzed the activation of several protein kinases known to regulate collagen-I expression such as p38 MAPK [42], ERK1/2 [43], JNK [44] and Akt1/2/3 [45] as well as their phosphorylation state. HSCs challenged with rCOMP showed increase in pERK1/2 (Fig. 7A). To validate that the pro-fibrogenic effect of rCOMP on collagen-I involved this signaling pathway, HSCs were pre-incubated with PD98059, a MEK1/2 inhibitor, prior to rCOMP treatment. Analysis of collagen-I expression revealed that blocking MEK1/2-pERK1/2 signaling prevented the collagen-I increase mediated by rCOMP (Fig. 7B). Finally, since several studies report that COMP is able to signal via Cd47 [46], Cd36 [47], Rage [48] and α_vβ₃-Integrin [49] we analyzed their mRNA expression. Cd36 mRNA was up-regulated upon TAA- or CCl₄-induced liver fibrosis in WT but not in Comp^−/− mice (Fig. 7C). No changes were observed in the other receptors (not shown). Hence, we analyzed if blocking CD36 could condition the HSC response to COMP treatment. Incubation of HSCs with a CD36 neutralizing Ab prevented the rCOMP-mediated up-regulation of MEK1/2-pERK1/2 and its downstream target collagen-I (Fig. 7D–7F). In summary, these findings suggest that the COMP-CD36 axis participates in the pathogenesis of liver fibrosis via activation of MEK1/2-pERK signaling to increase collagen-I protein expression in HSCs.

Figure 7. rCOMP up-regulates collagen-I expression in HSCs via CD36 signaling and activation of the MEK1/2-pERK1/2 pathway.

Rat HSCs were pre-incubated for 1 h with the MEK1/2 inhibitor PD98059 and treated with 0–25 nM rCOMP, which phosphorylates ERK1/2 (A). rCOMP up-regulated collagen-I expression via MEK1/2-pERK1/2 in HSCs (B). In (A) and (B) *p<0.05 and **p<0.01 for rCOMP vs. control; •p<0.05, ••p<0.01 and •••p<0.001 for co-treated vs. rCOMP. Cd36 mRNA expression was up-regulated by either TAA treatment or CCl₄ injection in WT compared to Comp^−/− mice. Results are expressed as mean ± standard deviation; n=6/group; **p<0.01 for TAA or CCl₄-injected vs. corresponding control mice, ^••p<0.01 and ^•••p<0.001 for Comp^−/− vs. WT mice (C). HSCs were pre-incubated for 1 h with a CD36 neutralizing Ab or its matching isotype control IgG and cells were treated with 0–25 nM rCOMP for 0–24 h. pERK1/2, total collagen-I protein and Col1a1 mRNA were lowered by the CD36 neutralizing Ab upon rCOMP treatment. *p<0.05 for rCOMP vs. control; •p<0.05 for co-treated vs. rCOMP. Col1a1 mRNA expression was normalized to that of Gapdh (D–F).

DISCUSSION

Although COMP has been proposed to participate in various inflammatory processes [20, 22], to date its role in the pathogenesis of liver fibrosis and the mechanism whereby it regulates scarring remained undefined. Thus, in this study, we investigated the potential pathophysiological role of COMP in the wound-healing response to persistent liver injury using WT and Comp^−/− mice and two well-established models of liver fibrosis along with mechanistic studies in HSCs plus in vitro reconstituted systems.

First, we observed that COMP is expressed in the liver from WT mice and is significantly induced following either TAA administration or CCl₄ injection. Immunofluorescence together with IHC analysis revealed that hepatocytes are the main source of COMP in the liver and that Kupffer cells and HSCs express it but to a much lesser degree. Moreover, both in vivo models of liver fibrosis demonstrated a protective effect for Comp ablation since under TAA-treatment or CCl₄-injection there was reduced liver injury, hepatic macrophages, expression of inflammation markers and fibrosis compared to WT mice.

Next, we investigated the molecular mechanism for the protective role of Comp ablation for the onset of liver fibrosis. Previous reports indicated that the carboxy-terminal domain of COMP can bind collagen-I with high affinity and catalyze collagen fibrillogenesis [50], suggesting that it may play a role in modulating pathological collagen-I deposition, stability and fiber organization by forming intermolecular cross-links. Our in vitro experiments supported the pro-fibrogenic effect of COMP, since intra- and extracellular collagen-I was up-regulated in HSCs upon challenge with rCOMP. This effect was independent of phenotypic changes or differences in HSCs viability or proliferation. Both silver nitrate staining and western blot analysis demonstrated that rCOMP binds collagen-I. However, if this interaction could strengthen the collagen-I fibrils perhaps precluding them from accessibility to the cleaving action of MMPs remained unknown.

In view that an imbalance in ECM remodeling contributes to liver fibrosis, that MMPs are the key enzymes regulating ECM degradation and that HSCs are a key source of these proteins in the liver [51], we examined the effect of COMP on the expression of these enzymes critical for ECM dynamics. rCOMP not only up-regulated MMP13 protein expression in HSCs but also stimulated MMP2 and 9 activities in a time-dependent manner even though the expression of their inhibitor Timp1 remained barely changed. The elevated MMP13, 2 and 9 protein levels observed in HSCs under rCOMP treatment are consistent with reports that these zymogens increase in the early stages of liver fibrosis [52]. Our studies proved that the binding of COMP to collagen-I did not preclude MMP1, with the highest affinity for collagen-I degradation, from cleaving it. Thus, it is likely that COMP could lead to a net increase in collagen-I despite the up-regulation of MMPs due to activation of other pathways mostly involved in collagen-I synthesis but not on degradation. Hence, we looked into alternative mechanisms involved.

TGFβ is one of the key pro-fibrogenic factors with direct ability to stimulate collagen-I synthesis in HSCs [53] whereas TNFα has been identified as a transcriptional repressor of the Col1a1 [54] and Col1a2 genes [55]. Consequently, we evaluated the expression of these factors as possibly accounting for the effects of COMP on HSCs. However, following rCOMP treatment, TGFβ protein along with Tgfβ and Tnfα mRNAs remained similar to those of control non-treated HSCs suggesting that the pro-fibrogenic effect of COMP on HSCs was likely TGFβ- and TNFα-independent. In addition, the ratio of pSmad2/3 to Smad2/3 expression and Tgfβ mRNA were analyzed in livers from TAA and CCl₄-treated mice confirming our previous in vitro findings.

Subsequent studies revealed that while following rCOMP treatment other protein kinases such as Akt1/2/3 (not shown) appeared phosphorylated only the MEK1/2-pERK1/2 pathway was involved in the pro-fibrogenic effect of rCOMP since pre-treatment with PD98059 blunted it. Several receptors have been described to mediate the action of COMP. Among them, CD36 is expressed in HSCs and it was up-regulated in response to TAA or CCl₄ treatment in WT mice. Thus, to further determine if the enhanced collagen-I expression under a rCOMP challenge was mediated via CD36 signaling, we incubated rat HSCs with a CD36 neutralizing Ab. Blockage of CD36 signaling prevented MEK1/2-pERK1/2 activation and the collagen-I up-regulation induced by rCOMP, thus suggesting a role for CD36 signaling in the pro-fibrogenic effects of COMP (Fig. 8, working model).

Figure 8. The profibrogenic effects of rCOMP are mediated via CD36 signaling and activation of the MEK1/2-pERK1/2 pathway to increase collagen-I.

rCOMP increases collagen-I expression in HSCs via CD36 receptor signaling and activation of the MEK1/2-pERK1/2 pathway since pre-incubation with a CD36 neutralizing Ab and with PD98059, a MEK1/2 inhibitor, blocks the increase in collagen-I by rCOMP.

In conclusion, COMP, a protein mostly produced by hepatocytes could play an important role in the mechanism whereby pathological collagen-I deposition occurs in HSCs via CD36 receptor signaling through the MEK1/2-pERK1/2 pathway hence contributing to the pathogenesis of liver fibrosis.

List of Abbreviations

ALT

alanine aminotransferase

AST

aspartate aminotransferase

CCl₄

carbon tetrachloride

COMP

cartilage oligomeric matrix protein

Comp^−/−

cartilage oligomeric matrix protein global knockout mice

CV

central vein

ECM

extracellular matrix

GAPDH

glyceraldehyde-3-phosphate dehydrogenase

H&E

hematoxylin and eosin

HSC(s)

hepatic stellate cell(s)

IHC

immunohistochemistry

IOD

integrated optical density

MMP(s)

matrix metalloproteinase(s)

MO

mineral oil

rCOMP

recombinant COMP

TAA

thioacetamide

TGFβ

transforming growth factor-β

TIMP1

tissue inhibitor of metalloproteinase-1

WT

wild-type