Decorin interferes with platelet-derived growth factor receptor signaling in experimental hepatocarcinogenesis

Authors


  • The authors declare no conflict of interest.

Correspondence

I. Kovalszky, 1st Department of Pathology and Experimental Cancer Research, Semmelweis University, Üllői út 26., Budapest, Hungary 1085

Fax: +36 1 317 1074

Tel: +36 1 459 1500 Ext. 54449

E-mail: koval@korb1.sote.hu

Abstract

Decorin, a secreted small leucine-rich proteoglycan, acts as a tumor repressor in a variety of cancers, mainly by blocking the action of several receptor tyrosine kinases such as the receptors for hepatocyte, epidermal and insulin-like growth factors. In the present study we investigated the effects of decorin in an experimental model of thioacetamide-induced hepatocarcinogenesis and its potential role in modulating the signaling of platelet-derived growth factor receptor-α (PDGFRα). Genetic ablation of decorin in mice led to enhanced tumor prevalence and a higher tumor count compared with wild-type mice. These findings correlated with decreased levels of the cyclin-dependent kinase inhibitor p21Waf1/Cip1 and concurrent activation (phosphorylation) of PDGFRα in the hepatocellular carcinomas generated in the decorin-null vis-à-vis wild-type mice. Notably, in normal liver PDGFRα localized primarily to the membrane of nonparenchymal cells, whereas in the malignant counterpart PDGFRα was expressed by the malignant cells at their cell surfaces. This process was facilitated by a genetic background lacking endogenous decorin. Double immunostaining of the proteoglycan and the receptor revealed only minor colocalization, leading to the hypothesis that decorin would bind to the natural ligand PDGF rather than to the receptor itself. Indeed, we found, using purified proteins and immune-blot assays, that decorin binds to PDGF. Collectively, our findings support the idea that decorin acts as a secreted tumor repressor during hepatocarcinogenesis by hindering the action of another receptor tyrosine kinase, such as the PDGFRα, and could be a novel therapeutic agent in the battle against liver cancer.

Abbreviations
DAPI

4′-6′-diamidino-phenylindole

Dcn −/−

decorin gene-knockout

EGFR

epidermal growth factor receptor

ERK1/2

extracellular-regulated signal kinase 1/2

HCC

hepatocellular carcinoma

HRP

horseradish peroxidase

IGF-1R

insulin-like growth factor-1 receptor

PDGF

platelet-derived growth factor

PDGFR

platelet-derived growth factor receptor

RTK

receptor tyrosine kinase

TA

thioacetamide

TGF-β1

transforming growth factor-beta 1

VEGF

vascular endothelial growth factor

Introduction

Hepatocellular carcinoma (HCC) is one of the most rapidly spreading cancers in the world. In the majority of liver cancers, chronic inflammation-induced fibrosis or cirrhosis precedes the development of the tumor, although this is not essential for tumor formation. At the same time, the process, resulting in the overproduction of extracellular matrix, favors cancer development in at least two ways, namely (a) accelerated hepatocyte regeneration leads to insufficient DNA damage repair and (b) the pathological matrix impairs the presentation of signals coming from the environment.

The extracellular matrix is a complex, well-organized structure of macromolecules that interact with each other and with the resident cells of the connective tissue. As a result, matrix macromolecules provide structural integrity and influence cell-growth regulation, migration and differentiation. Consequently, in the last decade, research on carcinogenesis has extended in this area of interest, focusing not only on tumor cells but also on the surrounding tumor stroma, including the abnormal synthesis and deposition of proteoglycans.

Decorin is a small leucine-rich proteoglycan of the extracellular matrix containing a single chondroitin sulfate or dermatan sulfate chain and is primarily produced by fibroblasts and myofibroblasts [1, 2]. It is present in a low quantity in the normal healthy liver, around the central veins and in the portal tracts. Together with other matrix proteins the amount of decorin significantly increases during fibrogenesis [3, 4]. Previous studies showed that the decorin protein core is able to bind transforming growth factor-beta 1 (TGF-β1) [5], directly blocking the bioactivity of the growth factor, thereby functioning as protective endogenous agent against fibrosis [6, 7]. By modulating tumor stroma deposition and cell-signaling pathways, decorin is recognized as a promising tumor growth and migration inhibitor [8, 9].

Decorin binds directly to the epidermal growth factor receptor (EGFR) and inhibits its activity as well as the activity of other members of the ErbB receptor tyrosine kinase (RTK) family [10]. These receptors are frequently overexpressed and/or mutated in various cancers, accelerating tumor progression [11]. Moreover, decorin targets EGFR for degradation by caveolar-mediated endocytosis [12]. Besides EGFR, decorin is a known ligand for the hepatocyte growth factor receptor, Met [13]. Such binding down-regulates the receptor and blocks its activity. Furthermore, decorin binds directly to insulin-like growth factor-1 receptor (IGF-1R) [14, 15] and vascular endothelial growth factor (VEGF) receptor type 2 [16], suppressing signaling pathways originating from these receptors. In parallel, decorin inhibits the endogenous VEGF production of tumor cells [17]. This pan-RTK blockage often leads to growth arrest and hinders tumor growth by keeping tumor cells in quiescence. It has been proven that a functional p21WAF1/CIP1, which causes G1 phase arrest, is indispensable for the tumor-repressor action of decorin in most tumor-cell lines [18]. Decorin typically surrounds proliferating tumor cells in the so-called tumor microenvironment [19]. The elevated concentration of decorin around tumor cells may be a form of paracrine defensive mechanism by stromal cells, counteracting the growth of malignant cells on the invasive front of solid tumors [20, 21].

Platelet-derived growth factors and their receptors have crucial roles in the development and maintenance of liver tumors. The levels of both platelet-derived growth factor receptor (PDGFR)α and PDGFRβ represent valuable prognostic markers in patients with HCC [22]. The importance of PDGFRβ is well documented as it represents a target of the multikinase inhibitor sorafenib used in targeted therapy of HCC [23, 24]. It is known that PDGFRα is involved in tumor angiogenesis and in maintenance of the tumor microenvironment and has been implicated in the development and metastasis of HCC [25, 26]. Another study reported that about 70% of HCCs had elevated PDGFRα levels as a result of diverse mechanisms, suggesting that targeting this receptor may be of therapeutic value [26].

Very little is known about the role of decorin in hepatic tumorigenesis. The few limited reports have shown that decorin inhibits the proliferation of hepatoma cell lines in vitro [27], and that expression of the decorin gene is significantly down-regulated in HCC, as shown by gene-expression analyses [28, 29]. Furthermore, in an earlier report, decorin inhibited PDGF-stimulated vascular smooth-muscle-cell functions by binding to the ligand PDGF and preventing PDGFR phosphorylation [30]. Based on these observations, we hypothesized that hepatic decorin, primarily expressed by the nonparenchymal liver cells (stellate cells and fibroblasts), would act in a paracrine manner to hamper the bioactivity of PDGFRα during the course of chemical-induced hepatocarcinogenesis.

Results

Lack of decorin leads to enhanced tumor formation in the liver

By immunostaining, as expected, decorin was detectable only in wild-type mice where it was deposited in periportal connective tissue and around the central veins, confirming the knockout phenotype of the animals (Fig. 1A,B).

Figure 1.

Decorin immunostaining and morphology of TA-induced liver tumors and tumor prevalence. Immunostaining for decorin in representative liver samples from wild-type (A) and decorin-deficient (Decorin−/−) (B) control mice. Nuclei are counterstained with DAPI (blue). Scale bar = 100 μm. Representative images of hematoxylin and eosin-stained normal (C) and tumor-bearing (D) liver tissues induced by treatment with TA. Connective tissue-specific picrosirius staining on untreated control livers (E) and on livers with HCC (F). HE, hematoxylin and eosin staining; N, nodule; PS, picrosirius staining specific for connective tissue; T, tumor. Arrows indicate tumor borders. Asterisks show the same vein on hematoxylin and eosin- and picrosirius-stained sections. Scale bars=100 μm. (G) Bar charts show the ratios of tumor-bearing mice in experimental groups of wild-type and decorin gene knockout (decorin−/−) mice. n = 12. ***< 0.001 (chi-square test). (H) Columns represent the average tumor count per liver in livers exposed to TA. n = 12. ** < 0.01.

Metabolization of thioacetamide (TA) in hepatocytes via cytochrome p450 causes fibrosis and subsequently hepatic cirrhosis. Thus, chronic exposure to TA provokes hyper-regeneration of hepatocytes, initiating hepatocarcinogenesis in the cirrhotic liver [31, 32] (Fig. 1C–F) . Tumors formed after exposure to TA were rich in cytoplasm with strong eosinophil staining, and had a connective tissue capsule (Fig. 1D,F). Ninety-three per cent of decorin-null mice developed macroscopic tumors in their livers, in contrast to 22% of wild-type mice (n = 15, < 0.001) (Fig. 1G). In parallel with the higher tumor prevalence, an elevated number of tumors was detected in mice lacking decorin compared with mice with a wild-type genetic background. We detected a 7.3-fold increase in tumor numbers with an average of 2.2 tumors per liver in decorin knockout (Dcn−/−) mice versus 0.3 tumors per liver in wild-type mice (< 0.01) (Fig. 1H). In conclusion, the lack of decorin sensitized the liver for tumor formation, suggesting that decorin acts as a soluble tumor repressor during experimental hepatocarcinogenesis.

Tumor repressor activity of decorin utilizes p21WAF1/CIP1 in liver cancer

As decorin is known to exert its inhibitor effect on tumor cell proliferation via p21WAF1/CIP1 in most cases, we tested whether p21WAF1/CIP1 would be changed in our experimental hepatocarcinogenesis model. At the mRNA level (Fig. 2A), p21WAF1/CIP1 was induced 140-fold in the wild-type TA-treated livers compared with control samples (< 0.01). In contrast, in the decorin-null livers only a 12.6-fold increase (~9% of wild-type) was detected vis-à-vis Dcn−/− control samples (< 0.05) (Fig. 2A). At the protein level (Fig. 2B,C), p21WAF1/CIP1 was barely detectable in either control group. In TA-treated livers, wild-type samples contained 1.8 times more p21WAF1/CIP1 protein compared with Dcn−/− samples (p < 0.05), as seen on western blots (Fig. 2B) and quantified by densitometric analysis of the bands (Fig. 2C). To determine what cell types express p21WAF1/CIP1 and whether they are affected by the absence or presence of decorin, fluorescence immunostaining specific for p21WAF1/CIP1 was performed (Fig. 2D,E). In wild-type TA-treated livers, stromal cells showed strong nuclear staining, and hepatocytes of the nontumorous tissue also expressed p21WAF1/CIP1 (Fig. 2D). In decorin-null sections, stromal cells of the connective tissue also displayed immunopositivity, but to a lesser extent compared with wild-type connective-tissue stromal cells. Beside, hepatocytes outside the tumor were almost completely negative for immunostaining, in contrast to livers of wild-type mice (Fig. 2D). Within the tumors, both tumor and stromal cells displayed immunopositivity, meanwhile Dcn−/− tumor cells appeared to lack the p21WAF1/CIP1, while stromal cells of the tumor expressed p21WAF1/CIP1 (Fig. 2D,E). These results indicate that lack of decorin in the liver reduces the levels of p21WAF1/CIP1, a powerful cyclin-dependent kinase inhibitor, and presumably would favor growth of the malignant cells.

Figure 2.

Alterations in the p21WAF1/CIP1 level in wild-type and decorin gene knockout (Dcn−/−) mice. (A) Bar charts represent the relative p21 mRNA levels in livers of wild-type (Wt) and decorin knockout (Dcn−/−) mice without treatment (control=CTL) or after treatment with TA. *< 0.05, **< 0.01. (B) Representative image of western blots specific for p21 protein and the β-actin loading control. (C) Relative levels of p21 normalized to β-actin obtained by densitometric analysis of p21 western blots. *< 0.05. (D) Representative images of p21WAF1/CIP1 immunostaining (red) on tumor-bearing liver sections from wild-type and decorin-null (Dcn−/−) mice. Nuclei were counterstained with DAPI. T, tumor, arrows point at the tumor border. Scale bar = 100 μm. (E) p21WAF1/CIP1 immunopositivity (red) captured within the tumors of wild-type and decorin knockout (Dcn−/−) mice. Nuclei are shown in blue (DAPI). Scale bar = 100 μm.

Ablation of decorin results in elevation of activated PDGFRα

As decorin is known to bind to, and block, the activity of several RTKs, we performed a protein array for phospho-RTKs to determine if the lack of the pan-RTK inhibitor decorin would lead to activation of any of these receptors. Four receptors were found to show altered phosphorylation owing to the lack of decorin, and we selected PDGFRα for further examination. The protein array results revealed that Dcn−/− TA-treated samples contained 1.8-times more phospho-PDGFRα than did wild-type samples (< 0.001) (Fig. 3A,B). These data were validated by western blotting followed by band densitometry (Fig. 3C,D). PDGFRα was found to be present in small amounts both in Dcn−/− samples and in wild-type control samples, and no difference between the genotypes was observed. Phosphorylation of the receptor was not detected in either of the control groups. TA treatment increased the PDGFRα level in wild-type and Dcn−/− samples, exerting a 1.8-fold higher phosphorylation level of PDGFRα in decorin-null liver homogenates than that of wild-type ones (< 0.01, Fig. 3D), confirming the results of the phospho-RTK array.

Figure 3.

Changes in PDGFRα and phospho-PDGFRα in TA-induced liver cancer. (A) Representative image of the phospho-RTK array dots of phospho-PDGFRα and the phosphotyrosine-positive control in untreated (CTL) and TA-treated (TA) liver samples of wild-type and decorin gene-knockout (Dcn−/−) mice. (B) Bar charts represent the results of densitometry of array dots showing relative levels normalized to the phosphotyrosine-positive control. ***< 0.001. (C) Representative image of the western blot membrane with PDGFRα and phosphotyrosine immunostaining and with Ponceau-staining as the loading control. Note that for PDGFRα and phosphotyrosine the blots were double-stained; the same band is shown after using chemiluminescence and fluorescence detection. (D) The diagram shows the phosphorylated PDGFRα level relative to the total amount of receptor, normalized to Ponceau staining. CTL, control; P-PDGFRα, phospho-PDGFRα; P-Y, phosphotyrosine; Wt, wild type. **< 0.01.

PDGFRα mainly localizes on nonparenchymal cells of the normal liver

Next, we determined the subcellular localization and expression of PDGFRα in normal and experimental livers. In the absence of any experimental challenge, we did not observe any difference by immunofluorescence staining between decorin-null and wild-type mice in the location, amount or phosphorylation status of PDGFRα (Fig. 4). The receptor mainly localized in the periportal areas (Fig. 4A) and often in a sinusoidal pattern (Fig. 4B). However, there was little colocalization with anti-(phosphotyrosine) IgG in the sinusoid, suggesting that under normal conditions this receptor is not significantly activated. Sections of normal livers from either genotype showed no immunopositivity for the receptor on the surface of hepatocytes. The majority of PDGFRα localized on the cell membrane of nonparenchymal cells, such as fibroblasts and myofibroblasts, as seen at higher magnification (Fig. 4C).

Figure 4.

Localization of PDGFRα in the normal liver. PDGFRα and phosphotyrosine double immunostaining in the periportal area (A) and within a lobule (B) of the liver. Within the parenchyma (C), but not the hepatocytes, the nonparenchymal cells are positive for immunostaining. P-Y, phosphotyrosine. Scale bars: 100 μm for panels A and B, and 50 μm for panel C.

Induced PDGFRα level, and its appearance on hepatocytes in TA-treated livers

Treatment with TA resulted in elevated levels of PDGFRα and its phosphorylated form, as detected by immunostaining. The majority of the receptor was located in cirrhotic septa and in the connective-tissue capsule of the tumors (Fig. 5A). We detected more severe cirrhosis in livers of Dcn−/− mice than in livers of wild-type animals, and this was associated with enhanced expression of both total and phosphorylated PDGFR (Fig. 5A). Within the tumors, the receptor appeared on the surface of tumor cells to a greater extent in decorin-null livers than that of wild-type ones (Fig. 5B). In conclusion, treatment with TA led to the induction of PDGFRα level, increased its activation state, and probably induced the de-novo expression of this receptor in the tumorigenic cells and its localization at the cell surface. Notably, these events were mainly observed in decorin-null liver sections rather than in wild-type samples, suggest-ing that lack of decorin is permissive for liver carcinogenesis.

Figure 5.

Localization of PDGFRα in TA-treated livers of wild-type and decorin-null animals. (A) PDGFRα and phosphotyrosine double immunostaining in cirrhotic septa of wild-type and decorin gene-knockoutl (Dcn−/−) liver sections. Scale bar = 100 μm. (B) Tumor cells in wild-type and Dcn−/− TA-treated livers stained by anti-PDGFRα and anti-phosphotyrosine IgGs. P-Y, phosphotyrosine; T, tumor. Scale bar = 10 μm.

Decorin does not colocalize with PDGFRα but binds directly to its natural ligand, PDGF

Next, we wanted to test the possibility that decorin could directly block the activity of PDGFRα. To this end, we determined whether there was subcellular colocalization of endogenous decorin and PDGFRα in wild-type livers. Double immunostaining of decorin and PDGFRα revealed that decorin and the receptor did not significantly colocalize in most parts of TA-treated wild-type livers, including cirrhotic connective tissue septa (Fig. 6A–C) and tumor stroma (Fig. 6D–F). Only a minor fraction of dual immunofluorescence reaction specific for decorin and the receptor showed colocalization (white arrows in Fig. 6F). These observations suggest that decorin hinders the action of the receptor using a mechanism other than direct binding and down-regulation of PDGFRα in our experimental animal model of hepatocarcinogenesis.

Figure 6.

Colocalization of decorin and PDGFR in tumors of cirrhotic liver. Decorin and PDGFR double immunostaining in sections of wild-type liver exposed to TA. Representative images of the reactions in cirrhotic septa (A–C) and in tumor stroma (D–F). Scale bar = 100 μm.

As colocalization experiments of decorin and PDGFRα failed to provide an explanation for the higher activated level of PDGFRα in decorin-null TA-treated livers, we tested if decorin was able to directly bind the PDGF ligand. Immobilized recombinant human decorin was incubated with PDGF AB, followed by an immunoreaction using an antibody specific for PDGF AB. Indeed, dots of decorin exposed to PDGF AB were visualized with the PDGF AB antibody, reporting that the proteoglycan–PDGF complex was formed (Fig. 7).

Figure 7.

Detection of the interaction between decorin and PDGF AB. Human recombinant decorin (from left to right, the first four columns of dots) or PDGF AB (from left to right, the last column of dots) was immobilized for dot-blot analysis. Dots of a 1 : 2 serial dilution of decorin (from left to right, the first three columns of dots) were incubated with either PDGF AB ligand or with NaCl/Tris (none), and visualized by incubation with PDGF AB-specific antibody. Immobilized decorin was visualized by incubation with anti-decorin Ig, and PDGF AB was visualized by incubation with PDGF AB Ig.

Discussion

At present we understand that extracellular matrix macromolecules form not only an inert, space-filling microenvironment but they also create a vivid playground for cells to interact with each other, as well as with the components residing in the matrix. These interactions modulate the utilization of signals that regulate the behavior of cells controlling key cellular events of physiologic and pathologic processes, namely adhesion, migration, proliferation, differentiation, survival and angiogenesis [33-38]. A soluble matrix molecule that has been shown to be involved in the regulation of all the aforementioned cellular events, and thereby markedly contributes to health and disease, is decorin, a small leucine-rich extracellular matrix proteoglycan [39-41]. The evidence suggests that decorin, among others, represents a potent antitumor molecule [42, 43]. Early studies with decorin-deficient mice have indicated that although the lack of decorin does not lead to the development of spontaneous tumors [44], it is permissive for tumorigenesis [45]. More recently, however, it has been shown that decorin-null mice in a pure C57Bl/6 background have the propensity to develop intestinal tumors, especially when fed a western-type high-caloric diet [46, 47]. Furthermore, ectopic expression of decorin has been shown to cause generalized growth suppression in neoplastic cells of various histologic origin [18]. As a consequence, several other studies supporting the antitumor and antimetastatic activity of decorin have been published [48-53]. Decorin expression in neoplastic tissues in vivo is relatively unexplored in clinical reports, by contrast to the large number of studies on the behavior of decorin in in-vitro tumor cell and xenograft models [10, 54-56]. Regarding liver tumors, we know that decorin inhibits the proliferation of HuH7 [27] and HepG2 [57] hepatoma cell lines. In addition, decorin was found to be suppressed in HCCs, as shown by gene-expression microarray analyses [28, 29]. In line with these reports, our present study provides the first evidence that decorin acts as a tumor repressor in an experimental animal model of primary hepatocarcinogenesis as a lack of decorin is accompanied with significantly higher tumor prevalence and elevated tumor count. In parallel with enhanced tumor formation in decorin-null mice versus wild-type mice, we detected that livers of Dcn−/− mice failed to induce p21WAF1/CIP1 expression both at the mRNA and protein levels when compared with samples from wild-type mice. p21WAF1/CIP1 is a potent cyclin-dependent kinase inhibitor, leading to G1 arrest of the cell cycle [58, 59]. In this way, ablation of decorin in liver cancer leads to impaired cell-cycle regulation, allowing a higher proliferation rate of tumor cells. This observation is in agreement with the well-known fact that decorin utilizes P21Waf1/Cip1 to display its tumor-repressor effect in most cases [18, 57, 60].

Decorin is known to modulate the mechanism of cell growth and to induce signal transduction via growth-factor receptors at the cell surface. Of note, decorin modulates and induces signal transduction along pathways involving the EGFR [11, 12, 61], the IGF-1R [14, 42] and Met [49]. In this study, we performed a phosphoprotein array to test changes in the activation state of RTKs upon decorin deficiency in liver cancer. Our results unveiled a novel player in the signaling of epithelial tumors affected by decorin, namely PDGFRα. Elevated amounts of both total and active PDGFRα were found in Dcn−/− livers exposed to the hepatotoxin, TA, compared with wild-type TA-treated livers, in both the cirrhotic septa and within the tumor stroma. Notably, elevated PDGF signaling has been associated with fibrotic diseases such as pulmonary fibrosis, liver cirrhosis, scleroderma, glomerulosclerosis and cardiac fibrosis [25]. Certainly, in our earlier report the lack of decorin was associated with enhanced fibrosis of the liver [3], providing a fine explanation for higher PDGFRα levels in decorin-null cirrhotic livers. PDGFRα can play important roles, not only in tumor angiogenesis, but also in directly stimulating tumor-cell proliferation [25, 26]. The receptor is not present in adult normal hepatocytes as its expression substantially declines during development with no significant activation or phosphorylation [26]. Indeed, we did not detect the presence of this receptor in either wild-type or decorin-null normal livers. Unexpectedly, in tumor cells of TA-treated livers, PDGFRα appeared in the cell membrane of tumor cells and became activated to a greater extent in Dcn−/− mice than in mice with a normal genetic background. This observation is in a good agreement with earlier reports demonstrating that HCC cell lines and resected tumors exhibit increased expression of both total and activated phosphorylated PDGFRα, with higher levels in more aggressive cell lines and in tumors with poor prognosis [26]. In addition, inhibition of PDGFRα by neutralizing monoclonal antibody has been shown to inhibit survival and proliferation of several human and mouse cell lines, including HCC [26, 62, 63]. Hence, PDGFRα appears to play significant roles, not only in tumor angiogenesis but also in cancer proliferation and survival in HCCs, and its targeting may be effective in the therapy of this malignancy.

To address a potential mechanism of action of decorin we utilized double immunostaining of decorin and PDGFRα. Surprisingly, there was only focal colocalization of the two proteins in a minor fraction of the sections of cirrhotic areas and in tumor stroma of wild-type livers. This observation suggested that the effects of decorin might not occur through a direct protein–receptor interaction, as in the case of the other RTKs. It is well established that almost all binding domains for growth factors are located in the leucine-rich region of the protein core of decorin, which acts as a docking reservoir for RTKs within the extracellular matrix [64]. Such direct interaction was revealed with TGFβ1, for example, resulting in the blockade of the growth factor receptor [5]. Thus, we hypothesized that decorin would bind the natural ligand PDGF, and by this indirect way would sequester the growth factor and attenuate PDGFRα activity. Cell-free binding studies revealed a binding interaction between human recombinant decorin and PDGF AB, confirming our hypothesis of indirect activity. Further studies are needed to clarify if decorin is capable of establishing a direct interaction with PDGFR. Our observation is in agreement with the report showing that decorin sequesters PDGF BB, thereby providing a potential mechanism for inhibition of intimal hyperplasia after balloon angioplasty [30]. Furthermore, our earlier observations of oligoarray hybridization (K. Baghy, R.V. Iozzo and I. Kovalszky, unpublished data) showed that the mRNA levels of PDGF A and B are elevated in Dcn−/− TA-induced tumor samples compared with wild-type samples. Besides PDGF AB and BB we do not yet know whether decorin is able to bind other forms of the PDGF ligand, such as AA or CC, but it would be of importance to examine this in a future prospective study.

Based on our previous results, lack of decorin leads to enhanced efficiency of TGF-β1 during the development of hepatic fibrosis [3]. In line with this observation, TGF-β1 increases PDGF B mRNA levels in a dose-dependent manner [65]; moreover, in-vitro experiments on epithelial-mesenchymal transition with hepatocytes revealed a marked induction of expression of PDGF A and of PDGFRα and -β upon exposure to TGF-β1 [66]. Moreover, PDGF induces its own expression in an autocrine manner via extracellular-regulated signal kinase 1/2 (ERK1/2) [67]. Our earlier studies revealed elevated amounts of active phospho-ERK1/2 in decorin-null livers during fibrogenesis [3]. Taken together, it is possible that in the absence of decorin the higher activity of TGF-β1 causes overexpression of PDGF, which would further trigger its own production (Fig. 8). Additional mechanisms for decorin in affecting PDGF production may occur. For example, the expression of PDGF B is enhanced by Wnt/β-catenin signaling [68], and decorin is a known inhibitor of this important pathway [20, 69]. In this way, we hypothesize that when decorin is not present, the production of the ligand PDGF increases because it is not sequestered by decorin. The accumulated PDGF in the extracellular space would then be free to bind to its cognate receptors. These events would ultimately lead to higher activity of the PDGF receptor (Fig. 8).

Figure 8.

Action of decorin on PDGF signaling in experimental liver cancer. The action of TGF-β is known to up-regulate PDGF ligands. The presence of decorin hinders both Smad-dependent and -independent signaling from the TGF-β receptor, leading to decreased expression of PDGFs. In the extracellular environment, decorin may prevent PDGF from binding to its receptors, resulting in interference with downstream signaling. PDGF may act in an autocrine or a paracrine manner. The changes in ERK1/2 in the presence or absence of decorin can be an outcome of crosstalk between the different growth factor receptors. Decorin utilizes p21WAF1/CIP1 for cell-cycle blockade to display its tumor-repressor effect in most model systems. MEK, mitogen-activated protein kinase/ERK kinase; PI3K, phosphoinositide 3-kinase; PLCγ, phospholipase C gamma; TGF-βR1, TGF-β receptor 1; TGF-βR2, TGF-β receptor 2.

It is known that decorin inhibits the production of VEGF by tumor cells [17] and can directly block VEGFR2 at the same time [16]. It is possible that, similarly to this action, decorin inhibits PDGF production and may directly block the PDGFR in tumors, causing decreased phosphorylation of the receptor. The cellular origin of PDGF production remains unclear as the ligands may also affect the cancer cells following its secretion from stromal cells [70] in a paracrine manner, in parallel with its synthesis by tumor cells (autocrine mode) [66]. The different isoforms can originate from different cell types of the liver [71]. It has been shown, in several studies and in different cancer models, that glycosaminoglycans, such as chondroitin sulfate, play key partner roles in PDGFR effects [72-74] and can contribute to matrix remodeling [75-77]. Taking this knowledge into account, it is possible that either decorin protein core or its DS/CS chain exerts a regulatory effect on PDGF signaling. Further investigations are necessary to identify which constituent of decorin proteoglycan is responsible for its action on PDGF signaling.

In conclusion, we provide novel information regarding a natural stroma-specific product of the liver and its potential role in suppressing tumor growth by hindering the action of PDGFRα in hepatocarcinogenesis, in part by its ability to bind and sequester PDGF AB. Our studies support the potential utilization of decorin as a viable therapeutic agent, either alone or in combination [20, 43, 51]. The fact that decorin is a nontoxic natural biological product, and therefore not immunogenic by itself, provides a rationale for targeted delivery and/or expression of decorin in cancer tissues as a new anti-oncogenic strategy [43, 64, 78, 79].

Materials and methods

Decorin-null mice

All animal experiments were conducted according to the ethical standards of the Animal Health Care and Control Institute, Csongrád County, Hungary (permit no. XVI/03047-2/2008). Decorin-deficient mice were generated as previously described [44]. In brief, inactivation of the decorin gene was achieved by targeted disruption of exon 2 through insertion of a PGK-Neo cassette. Two male and two female C57Bl/6 mice, heterozygous for the decorin gene (Dcn−/−), which were backcrossed into the C57Bl/6 background for nine generations, were bred until homozygosity. The genotype of the offspring was determined using PCR. Tail DNA was isolated using the high-salt method. Subsequently, three primers were applied: one sense primer and one antisense primer that were specific for exon 2, and a primer corresponding to the PGK-Neo cassette. PCR products were analyzed by electrophoresis through a 2% agarose gel.

TA treatment

For induction of liver cancer, we utilized a total of 24, 1-month-old male mice. Twelve wild-type and 12 Dcn−/− mice with a C57Bl/6 background were exposed to TA dissolved in drinking water (150 mg·L−1). To obtain fully developed HCC, the mice were treated with TA for 7 months. Age-matched untreated mice with an identical genetic background served as controls. Mice were killed after 7 months of TA treatment by cervical dislocation in ether anesthesia. At this point, the body weight and the liver weight of the mice were measured and the number of macroscopically detectable tumors was counted. Half of the liver samples were frozen for further processing and the other half were fixed in 10% formaldehyde and embedded in paraffin for histological analysis. Paraffin sections were dewaxed in xylene and stained with hematoxylin and eosin, or processed further for immunohistochemistry. Stained sections were used for histological diagnosis.

Real-time RT-PCR

For RT-PCR, total RNAs were isolated from frozen livers. After homogenization in liquid nitrogen, total RNA was isolated using the RNeasy Mini Kit (Qiagen, Hilden, Germany), according to the protocol provided by the manufacturer. The yield and purity of the isolated RNAs were estimated using an ND-1000 spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA). The integrity and size distribution of the total RNAs purified were analyzed using Experion RNA Chips and the Experion Automated Electrophoresis Station (Bio-Rad, Hercules, CA, USA). cDNAs were generated from 1 μg of total RNA using the M-MLV Reverse Transcriptase kit (Invitrogen by Life Technologies, Carlsbad, CA, USA) according to the instructions of the supplier. Real-time PCR was performed in an ABI Prism 7000 Sequence Detection System (Applied Biosystems by Life Technologies, Carlsbad, CA, USA), using ABI TaqMan Gene Expression Assays for mouse p21WAF1/CIP1 (CDKN1A, Assay ID: Mm00432448_m1) applying 18S rRNA as an endogenous control (Part No: 4319413E), according to the manufacturer's protocol. All samples were run in duplicate in a total volume of 20 μL, containing 50 ng of cDNA, using TaqMan Universal PCR Master Mix (Part No.: 4324018; Applied Biosystems by Life Technologies). The thermal cycle conditions for the reactions were as follows: denaturation for 10 min at 95 °C, followed by 40 cycles of denaturation at 95 °C for 15 s and annealing at 60 °C for 1 min. The results were obtained as cycle threshold (Ct) values. Expression levels were calculated using the inline image method.

Phospho-RTK array and western blotting

Total proteins were extracted from frozen liver tissues. After homogenization in liquid nitrogen, 1 mL of lysis buffer was added to the samples (20 mm Tris (pH 7.5), 2 mm EDTA, 150 mm NaCl, 1% Triton X-100, 0.5% Protease Inhibitor Cocktail (Sigma, St Louis, MO, USA), 2 mm Na3VO4 and 10 mm NaF). After incubation for 30 min on ice, the samples were centrifuged at 15 000 g for 20 min. Supernatants were carefully decanted and the protein concentrations of the supernatants were measured as described by Bradford [80]. The activities of phospho-RTKs were assessed by their relative levels of phosphorylation using the Proteome Profiler Array (R&D Systems, Minneapolis, MN, USA) according to the manufacturer's instructions. The same liver protein samples were used for western blotting. Pooled samples of five livers from the same experimental group were homogenized in lysis buffer (described earlier) and adjusted to 300 μg of protein per 250 μL of lysate. Signals were developed by incubating the membrane in SuperSignal West Pico Chemiluminescent Substrate Kit (Thermo Fisher Scientific Inc., Waltham, MA, USA) and visualization using a Kodak Image Station 4000MM Digital Imaging System (Carestream Health, Inc., Rochester, NY, USA).

For western blotting, 30 μg of total protein was mixed with loading buffer containing β-mercaptoethanol and incubated at 99 °C for 5 min. Denatured samples were loaded onto a 10% polyacrylamide gel and run for 30 min at 200 V on Mini Protean vertical electrophoresis equipment (Bio-Rad). Proteins were transferred to poly(vinylidene difluoride) membrane (Millipore, Billerica, MA, USA) by blotting for 1.5 h at 100 V. Ponceau staining was applied to determine blotting efficiency. Membranes were blocked with 3% (w/v) nonfat dry milk (Bio-Rad) in NaCl/Tris for 1 h followed by incubation with the primary antibodies (p21WAF1/CIP1: ab7960 (Abcam, Cambridge, UK); PDGFR: RD-AF1062 (R&D Systems); and p-Y: RD-MAB1676 (R&D Systems) at 4 °C for 16 h. Mouse β-actin or Ponceau staining served as loading controls. Membranes were washed five times with NaCl/Tris containing 0.5% (v/v) Tween-20, then incubated with appropriate secondary antibodies for 1 h. For p21, anti-rabbit Ig–horseradish peroxidase (HRP) conjugate (P 0448) from DakoCytomation (Glostrup, Denmark) was applied. For PDGFRα and phosphotyrosine double staining, PDGFRα was visualized by anti-goat Ig-HRP conjugate (P0449) and phosphotyrosine was visualized by anti-mouse immunglobulins/biotin (E0443) with a subsequent incubation with Qdot® 525 streptavidin conjugate (Q10141MP; Invitrogen by Life Technologies) for 45 min. Signals were detected using the SuperSignal West Pico Chemiluminescent Substrate Kit (Pierce/Thermo Scientific, Waltham, MA, USA) and visualized using the Kodak Image Station 4000MM Digital Imaging System. For PDGFRα and phosphotyrosine double staining, membranes were visualized for luminescence and then exposed to UV light for detecting fluorescence from Qdots using the same system. The density of the bands was also measured using the Kodak Image Station.

Immunofluorescence

For p21WAF1/CIP1 immunostaining, formalin-fixed paraffin-embedded sections were dewaxed by xylene and ethanol, and antigens were retrieved by incubation in target retrieval solution (DakoCytomation) for 20 min using a pressure cooker. For PDGFRα, phosphotyrosine and decorin reactions, frozen sections of the liver were fixed in ice-cold methanol for 20 min. Next, the slides were washed in NaCl/Pi, then blocked with NaCl/Pi containing 5% (w/v) BSA and 10% nonimmune serum of secondary antibody at 37 °C for 30 min. After washing, sections were incubated with the primary antibody specific for PDGFRα [antibody 1: RD-AF1062 (R&D Systems); and antibody 2: #3174 (Cell Signaling Technology, Danvers, MA)], phosphotyrosine (RD-MAB1676; R&D Systems), decorin (AF1060; R&D Systems) or p21WAF1/CIP1 (ab7960; Abcam), diluted in 1 : 50 in NaCl/Pi containing 1% (w/v) BSA, at 37 °C for 1.5 h or at 4 °C for 16 h. Appropriate fluorescent secondary antibodies (for PDGFRα antibody 1: Alexa Fluor 555 donkey anti-goat IgG: cat. no.: A21432; for PDGFRα antibody 2: Alexa Fluor 555 donkey anti-rabbit IgG: cat. no.: A31572; for phosphotyrosine: Alexa Fluor 488 donkey anti-mouse IgG: cat. no.: A21202; and for decorin: Alexa Fluor 488 donkey anti-goat IgG: cat. no.:A11055; all from Invitrogen by Life Technologies) were applied at room temperature for 30 min. Nuclei were stained with 4′-6′-diamidino-phenylindole (DAPI). Pictures were taken using a Nikon Eclipse E600 microscope with the help of Lucia Cytogenetics version 1.5.6 program, or using a confocal laser scanning microscope (MRC-1024; Bio-Rad, Richmond, CA, USA).

Decorin and PDGF AB interaction

Decorin was purified from the secretions of Chinese hamster ovary cells transfected with a full-length decorin-expressing pcDNA3.1 vector, as described previously [81-83]. For dot-blot, ~ 2 μg of decorin per dot was applied onto a nitrocellulose membrane (Millipore), in a 1 : 2 serial dilution in wells for detecting the interaction between PDGF AB and decorin. Next, the membrane was blocked with 3% nonfat dry milk dissolved in NaCl/Tris. Subsequently, membranes were incubated NaCl/Tris, without or with 4 μg·mL−1 of PDGF AB (ab73228; AbCam), for 18 h at 4 °C. Next, the membranes were washed and probed with primary antibody specific for PDGF AB (ab50201; Abcam) diluted 1 : 500 in NaCl/Tris for 18 h at 4 °C. After washing, HRP-conjugated secondary antibody (anti-goat Ig-HRP conjugate (P0449); Dako) was applied for 1 h. Dot-blot images were visualized using the SuperSignal West Pico Chemiluminescent Substrate Kit (Pierce/Thermo Scientific) and visualized using the Kodak Image Station 4000MM Digital Imaging System.

Statistical analysis

All statistical analyses were performed using graphpad prism 4.03 software (Graphpad Software Inc., La Jolla, CA, USA). Data were tested for normal distribution using the omnibus normality test of D'Agostino & Pearson. Significance of changes were tested using a nonparametric test (the Mann–Whitney U-test) or the Student's t-test, depending on the distribution of the data. The difference between wild-type and Dcn−/− groups in tumor prevalence was tested for significance using the chi-square test. The independent experimental sets were compared for reproducibility. Only reproducible significant changes were considered as significant. Significance was declared at the standard P < 0.05 level.

Acknowledgements

This work was supported in part by Hungarian Scientific Research Fund, grants 67925 and 100904 (to IK); grant 105763 (to KB) and by the National Institutes of Health grant RO1 CA39481 (to RVI). The authors would like to thank András Sztodola and Mónika Borza for their help with animal experiments, Dr Sándor Paku for his assistance with the confocal laser scanning microscope and for Zsuzsa Kaminszky for her technical assistance.

Ancillary