Proteinase-activated receptor 2 (PAR2) in hepatic stellate cells – evidence for a role in hepatocellular carcinoma growth in vivo

Previous studies have established that proteinase-activated receptor 2 (PAR2) promotes migration and invasion of hepatocellular carcinoma (HCC) cells, suggesting a role in HCC progression. Here, we assessed the impact of PAR2 in HCC stromal cells on HCC growth using LX-2 hepatic stellate cells (HSCs) and Hep3B cells as model. PAR2 expression and function in LX-2 cells was analysed by RT-PCR, confocal immunofluorescence, electron microscopy, and [Ca2+]i measurements, respectively. The impact of LX-2-expressed PAR2 on tumour growth in vivo was monitored using HCC xenotransplantation experiments in SCID mice, in which HCC-like tumours were induced by coinjection of LX-2 cells and Hep3B cells. To characterise the effects of PAR2 activation in LX-2 cells, various signalling pathways were analysed by immunoblotting and proteome profiler arrays. Following verification of functional PAR2 expression in LX-2 cells, in vivo studies showed that these cells promoted tumour growth and angiogenesis of HCC xenografts in mice. These effects were significantly reduced when F2RL1 (encoding PAR2) was downregulated by RNA interference (RNAi). In vitro studies confirmed these results demonstrating RNAi mediated inhibition of PAR2 attenuated Smad2/3 activation in response to TGF-β1 stimulation in LX-2 cells and blocked the pro-mitotic effect of LX-2 derived conditioned medium on Hep3B cells. Furthermore, PAR2 stimulation with trypsin or a PAR2-selective activating peptide (PAR2-AP) led to activation of different intracellular signalling pathways, an increased secretion of pro-angiogenic and pro-mitotic factors and proteinases, and an enhanced migration rate across a collagen-coated membrane barrier. Silencing F2RL1 by RNAi or pharmacological inhibition of Src, hepatocyte growth factor receptor (Met), platelet-derived growth factor receptor (PDGFR), p42/p44 mitogen activated protein kinase (MAPK) or matrix-metalloproteinases (MMPs) blocked PAR2-AP-induced migration. PAR2 in HSCs plays a crucial role in promoting HCC growth presumably by mediating migration and secretion of pro-angiogenic and pro-mitotic factors. Therefore, PAR2 in stromal HSCs may have relevance as a therapeutic target of HCC.


Background
Hepatocellular carcinoma (HCC) is the fifth most common malignancy worldwide and the second most frequent cause of cancer-related death [1]. Curative liver transplantation or resection is only possible in approximately 30 % of patients. For advanced HCC, the multikinase inhibitor sorafenib remains the only approved chemotherapy agent with survival benefits for advanced stage HCC patients. However, its use is limited to those with well-preserved liver function [2][3][4].
Up to 90 % of cases of HCC develop on the background of advanced liver fibrosis or cirrhosis, thus representing a typical inflammation-related cancer type. It is characterised by a tumour microenvironment where both tumour cells and stromal cells interact via secretion of cytokines, proteinases and other biological active factors. Key players in the organization of an optimal HCC stromal compartment are activated hepatic stellate cells (HSCs) [5,6]. HSCs are characterised by high proliferation, pronounced matrix proteinase activity and secretion of extracellular matrix (ECM) proteins, as well as of factors including different proteases that aid in creating an immunosuppressive, angiogenic and tumour growthpromoting environment, and thereby critically contribute to HCC initiation and progression [7][8][9][10][11][12][13][14][15][16][17][18].
At present no information is available about the impact of PAR 2 in stromal cells for HCC development and progression. However, human HSCs have been shown to express PAR 2 and upon its activation increase the release of collagen [44] and augment inflammatory and pro-fibrotic pathways through the induction of pro-inflammatory cytokines and extracellular matrix proteins [45]. This indicates that PAR 2 stimulates the activation, proliferation, collagen production, and TGF-β protein production by HSCs [45]. Therefore, we specifically addressed this issue by evaluating the effect of PAR 2 silencing in the HSC cell line LX-2 [46] on tumour growth in a SCID mouse xenograft model, in which a HCC was induced by (co)injection of LX-2 cells and Hep3B liver carcinoma cells.

PAR 2 knockdown inhibits tumour growth in a HCC-SCID mouse model
Activated HSCs are known to promote HCC growth and progression [7][8][9][10][11][12][13][14][15][16][17][18], however, whether HSC-expressed PAR 2 is involved here remains unclear. To analyse this, we employed the human HSC cell line LX-2 in subcutaneous tumourigenicity experiments in a HCC-SCID mouse model. Although PAR 2 expression by HSCs has been reported [44,45], specific data for LX-2 cells in this regard were not available. PAR 2 expression was therefore analysed by PAR 2 -specific reverse transcription-polymerase chain reaction (RT-PCR), confocal immunofluorescence and electron microscopy. Expression was readily detected at both the mRNA (Fig. 1a) and protein level (Fig. 1b). Granular PAR 2 immunoreactivity was prominently visible around the nucleus, and to a lesser extent in the peripheral cytoplasm and the membrane compartment (Fig. 1b). Membrane localization of PAR 2 was also found using scanning electron microscopy techniques and immunogold labeling (Additional file 1: Figure S1). To verify that the PAR 2 protein on LX-2 cells is signalling-competent, [Ca 2+ ] i mobilisation in response to ligand stimulation was used as an index for PAR 2 activation [47]. We observed a strong effect of both the synthetic PAR 2 -AP, 2-furoyl-LIGRLO-NH 2 (10 µM), and trypsin (10 nM) on free intracellular calcium (Fig. 1c). The concentration dependency and data for PAR 2 specificity of [Ca 2+ ] i mobilisation induced by PAR 2 -AP are shown in Additional file 2: Figure S2.
Having demonstrated both PAR 2 expression and function, we went on to study the effect of PAR 2 knockdown in LX-2 cells on the growth of tumour xenografts in vivo. For that purpose, suspensions of cells from the HCC cell line Hep3B and LX-2 cells (LX-2-wt) stably expressing a short hairpin (sh) RNA directed against PAR 2 (LX-2-shPAR 2 ) and LX-2 cells with a non-target control a RT-PCR of PAR 2 expression. Extraction of total RNA from the LX-2-wt cells and synthesis of cDNA was performed as described in the Methods section. PCR reactions without cDNAs were run as a negative control (Primer). Integrity of the cDNA was independently confirmed by amplification of beta-actin (Actin). MW marker, molecular-weight marker. Representative results of three independent experiments are shown. b PAR 2 immunofluorescence was detected using the confocal laser scanning microscope LSM-510 Meta (Carl Zeiss, Germany). Localization of immunofluorescence labelled PAR 2 is shown in permeabilized LX-2-wt cells using SAM-11 (1:100) and a FITC-conjugated anti-mouse IgG (1:200) as secondary antibody. c LX-2-wt cells grown on Lab Tek chambered borosilicate cover glass were loaded with fluo-4-AM as described in Methods. For calcium measurements, an inverted confocal laser scanning microscope LSM 510 was used. Fluorescence was monitored at 488 nm. (a) PAR 2 -AP (10 µM) and (b) trypsin (10 nM) induce Ca 2+ rise in LX-2 cells. (c) Fluorescence images, in pseudocolor, from single LX-2 cells. The sequence shows a fast and transient fluorescence increase from 15 s to 40 s after PAR 2 -AP addition (0 s). Data represent the mean ± SD from calcium measurements in 20 individual cells, respectively. (d-f) PAR 2 knockdown in LX-2 cells inhibits tumour growth in a SCID mouse model. SCID mice were randomized into five groups, each consisting of 8 animals. Hep3B and LX-2 cells were subcutaneously (co)injected at the right flank of the mice [(1): 5 × 10 5 LX-2-wt; (2): 10 5 Hep3B cells; (3): 10 5 Hep3B cells plus 5 × 10 5 LX-2-wt, (4): 10 5 Hep3B cells plus 5 × 10 5 LX-2-shCo cells, (5): 10 5 Hep3B cells plus 5 × 10 5 LX-2-shPAR 2 cells]. d Tumours prepared 16 days after xenotransplantation. Data for body weight and tumour volume are indicated as mean ± SD (n = 8/group); **p < 0.01). e Representative micro CT images from the tumours 16 days after xenotransplantation. For 3D reconstruction of the tumours the software "Imalytics" (Philips GmbH, Aachen, Germany) was used. f Paraffin sections obtained from tumours and stained for hematoxylin/eosin, cytokeratin, Ki67 or CD31. Images were taken at magnification = × 5 (left panel) and at magnification = × 200 shRNA, (LX-2-shCo) were injected into the right flank of SCID mice. After 16 days, tumour formation was evaluated by macroscopic inspection, micro CT analysis and histochemical/immunohistochemical staining. While injection of LX-2-wt cells did not result in tumour development [ Fig. 1d, (1)] and Hep3B cell injection alone yielded only small tumours (tumour volume < 3 mm 3  Histochemical staining of tumour sections from experimental groups (2)-(5) by hematoxylin/eosin showed no gross differences and revealed histological features of a HCC with a trabecular growth pattern, central necrosis and pronounced inflammatory component (Fig. 1f). The epithelial nature of the tumours was confirmed by cytokeratin staining with the pan-keratin antibody MNF 116 [48].
To further characterise the inhibitory effect of F2RL1 silencing in LX-2 cells on tumour growth, immunohistochemical stainings of Ki67 and CD31 were performed to analyse tumour cell proliferation and tumour vascularization, respectively, both being hallmarks of cancer progression and tumour growth [49]. As demonstrated in Fig. 1f, the number of Ki67-positive cells (Hep3B and LX-2 cells) was apparently higher in tumours induced by coinjection of Hep3B cells with LX-2-wt cells (or LX-2-shCo) than with PAR 2 deficient LX-2 cells (quantification by manual counting revealed approximately 2.5-fold higher number of Ki67-positive cells, data not shown), strongly suggesting that the reduced tumour growth mediated by silencing F2RL1 in LX-2 cells may depend on diminished cell proliferation in the tumour microenvironment. Next, we assessed whether the reduced tumour growth with LX-2-shPAR 2 cells correlated with changes in angiogenesis. Therefore, expression of endothelial marker CD31 in sections from tumours induced by coinjection of Hep3B cells with either wt, shCo or shPAR 2 LX-2 cells were compared. As shown in Fig. 1f, PAR 2 knockdown in LX-2 cells resulted in remarkably reduced immunostaining for CD31 indicating that LX-2expressed PAR 2 promoted angiogenesis. Overall these data suggest that PAR 2 expression in HSCs promotes HCC growth by increasing cell proliferation and inducing angiogenesis.

PAR 2 knockdown inhibits the pro-mitotic effect of LX-2 cell culture supernatants on Hep3B cells
Having shown the inhibitory effect of F2RL1 silencing in LX-2 cells on tumour growth in vivo, we sought to reveal whether reduced proliferation and/or increased apoptosis of the tumor cells in presence of PAR2-deficient LX-2 cells may be the underlying mechanism. For this purpose, we stimulated Hep3B cells with culture supernatants (conditioned for 48 h) from PAR 2 depleted LX-2 cells or PAR 1 depleted LX-2 cells as control. After incubation of Hep3B cells with conditioned media from the various LX-2 transfectants for 72 h, vital cell numbers were dramatically lower in Hep3B cultures which had received conditioned medium from PAR 2 depleted LX-2 cells (Fig. 2). This effect resulted from reduced proliferation rather than enhanced apoptosis since the number of non-viable cells, as determined by trypan blue exclusion, was not different between the various cultures (data not shown). These data show that PAR 2 but not PAR 1 expressed by LX-2 cells is critically involved in Based on the results in vivo demonstrating that PAR 2 knockdown in LX-2 cells resulted in remarkably reduced tumour growth in a xenograft SCID mouse model, we set out to analyse in more detail how PAR 2 impacts on LX-2 cells. We first focussed on signalling pathways known to be triggered by PAR 2 activation [24]. PAR 2 -mediated activation of different intracellular effectors was monitored by immunoblotting. Stimulation of LX-2 cells with the PAR 2 agonists trypsin (10 nM) or PAR 2 -AP (10 μM) caused a marked increase of phosphorylated (p)-Src (Fig. 3a), p-Met (Fig. 3b), p-PDGFR (Fig. 3c) and p-p42/p44 MAPK immunoreactivity ( Fig. 3d and e). To define the PAR 2 -initiated phosphorylation sequence, we performed different inhibition experiments. We found in LX-2 cells that trypsin and PAR 2 -AP-induced phosphorylation of Met and PDGFR could be blocked by PP2 (Fig. 3b and c), while the Met inhibitor PHA 665752 and the PDGFR antagonist AG 1296 The cells were treated with PAR 2 -AP (10 μM) or (b-e) trypsin (10 nM) for 3 min and cell lysates were subjected to SDS-PAGE and Western blotting with an anti-phospho-Met antibody and re-probed with an anti-Met antibody. P-Met = phosphorylated Met, t-Met = total Met. c The cells were treated with PAR 2 -AP (10 μM) or trypsin (10 nM) for 3 min and cell lysates were subjected to SDS-PAGE and Western blotting with an anti-phospho-PDGFR antibody and re-probed with an anti-PDGFR antibody. P-PDGFR = phosphorylated PDGFR, t-PDGFR = total PDGFR. Serum-starved LX-2-wt cells were preincubated with d vehicle or the Met receptor tyrosine kinase inhibitor, PHA 665752 (0.1 μM) for 1 h or e with vehicle or the PDGFR inhibitor AG 1296 (1.0 μM) for 1 h. Following stimulation for 10 min with the respective receptor agonist cell lysates were immunoblotted with an anti-phospho-p42/p44 MAPK antibody and re-probed with a p42/p44 MAPK antibody. P-p42/p44 = phosphorylated p42/p44 MAPK, t-p42/p44 = total p42/p44 MAPK. f LX-2-wt cells were depleted of PAR 2 or PAR 1 protein by siRNA transfection as described in Fig. 2. Forty-eight hours after the second transfection, the growth medium was removed and replaced by fresh medium supplemented or not with 10 ng/ml TGF-β1. After 30 min, cells were lysed and immunoblotted with anti-phospho-Smad2/3 antibody and subsequently with a β-actin antibody to verify equal loading. The phosphorylated forms of Smad2 and Smad3 run as closely spaced bands due to the slightly lower electrophoretic mobility of Smad2. a-f Quantification of immunoblots was performed by scanning densitometry (Image J 1.43; National Institutes of Health, Bethesda, Maryland, USA). The data shown in the histograms above the blots are expressed as the fold increase over untreated control (averages ± S.D.) from three independent experiments. *P-value < 0.05 versus non stimulated control were able to inhibit PAR 2 -mediated phosphorylation of p42/p44 MAPK ( Fig. 3d and e).
TGF-β is overexpressed in HCC [50,51] and considered to be a major factor promoting liver carcinogenesis. Specifically, TGF-β drives transformation of HSCs into myofibroblasts [52,53] as well as their migration and invasion [54]. PAR 2 may enhance these pro-oncogenic effects by stimulating TGF-β gene expression and protein production in HSCs [45]. However, based on recent data from our group [55], PAR 2 may also promote TGF-β signaling in a more direct fashion in HSCs/LX-2 cells. Indeed, small interfering RNA (siRNA)-mediated depletion of PAR 2 in LX-2 cells impaired the ability of TGF-β1 to induce Cterminal phosphorylation of Smad2 and Smad3, reflecting activation of the pathway, in response to TGF-β1 challenge ( Fig. 3f).
Taken together, these data indicate that PAR 2 activation stimulates a Src-Met/PDGFR-p42/p44 MAPK signalling axis in LX-2 cells while PAR 2 protein expression is required for activation of TGF-β/Smad signalling.

PAR 2 stimulation in LX-2 cells increases secretion of pro-angiogenic and pro-mitotic factors and proteinases
We next assessed the possibility that PAR 2 activation in LX-2 cells may result in the secretion of pro-mitotic and pro-angiogenic factors, in turn promoting tumour cell growth. For this purpose proteome profiler arrays were employed allowing for simultaneous detection of 55 cytokines and 35 proteinases. Using these arrays for analysis of culture supernatants of LX-2 cells stimulated with PAR 2 -AP or trypsin for 24 h, or non-stimulated LX-2 cells as control, revealed predominantly enhanced levels of IL-8, uPA (Fig. 4a), cathepsins V, X, Z, P, MMPs 1, 3, 7, 13 and ADAMTS1 (Fig. 4b) in supernatants of LX-2 cells treated with PAR 2 -AP or trypsin, compared to control cells. PAR 2 activation increases migratory capabilities of LX-2 cells in a Src-, Met-, PDGFR-, p42/p44 MAPK-and MMPdependent manner HSC locomotion is known to be essential for disease processes in liver including fibrosis and cancer [56,57]. Specifically, activated HSCs infiltrate HCC stroma and peri-tumoural tissue during hepatocarcinogenesis where they secrete substantial amounts of bioactive proteins [58] thereby facilitating HCC development, progression and metastasis. Since PAR 2 activation in LX-2 cells led to activation of signalling pathways driving cell motility (see Fig. 3) and to secretion of pro-invasive proteases (see Fig. 4), we addressed the question of whether these events were associated with increased migration of LX-2 cells. Using a boyden-chamber assay, we found that stimulation of LX-2 cells for 6 h with trypsin or PAR 2 -AP, but not with the reverse PAR 2 -AP (2-furoyl-OLRGIL-NH 2 , PAR 2 -RP), significantly enhanced their migration (Fig. 5a). As further shown in Fig. 5a, the effect of trypsin and PAR 2 -AP on cell migration was absent in LX-2-shPAR 2 cells. Representative results from three independent experiments are shown. *P-value < 0.05 versus non stimulated control, **P-value < 0.05 versus stimulated with PAR 2 -AP and versus non stimulated control It is well known that a migratory phenotype can be associated with epithelial-mesenchymal transition (EMT), which in turn involves morphological changes, such as lamellipodial protrusions and filopodial spikes [59]. Therefore, scanning electron microscopy analyses on the various LX-2 cells were performed. Interestingly, treatment of LX-2 cells with PAR 2 -AP or trypsin induced the formation of filopodia spikes in a significantly larger number of cells when compared with non-stimulated controls (Fig. 5b).
To assess the contribution of the PAR 2 activated signalling pathways (see Fig. 2) for LX-2 cell migration, we again performed inhibition studies with a battery of protein tyrosine kinase and MMP inhibitors. As shown in Fig. 5c

Discussion
During the last years it became evident that PAR 2 plays an important role in the development, progression and metastasis of tumours especially from epithelial origin [30][31][32][33][34][35][36][37][38][39]60]. In case of HCC, studies have focussed on the function of PAR 2 expressed by the tumour cells. This PAR-subtype is known to promote migration and invasion of HCC cells by stimulating various intracellular signalling systems including calcium mobilisation, reactive oxygen species, Met and p42/p44 MAPK [40,41]. Moreover, PAR 2 is involved in the regulation of CD47 + HCC stem cells, contributing to tumour initiation, selfrenewal and metastasis [42]. In addition, it could be shown that the TF/factor VIIa/PAR 2 -signalling cascade stimulates m-TOR signalling in liver carcinoma cells and reduces m-TOR-mediated autophagy induction in HepG2 xenografts in a SCID mouse model, suggesting a role in HCC progression [43].
Recent data indicated that also the stromal compartment plays a crucial role in HCC development and progression and provides an attractive target for HCC therapy [11,14]. In the liver, HSCs, fibroblasts, myofibroblasts as well as immune and endothelial cells represent the major cell types of the hepatocyte stromal microenvironment. Among them, activated HSCs, beside their well-studied role in liver fibrosis and extracellular matrix remodeling [61], are discussed as a key driver for liver cancer development and progression [7][8][9][10][11][12][13][14][15][16][17][18]. HSC activation is regulated by complicated networks of numerous growth factors, proteinases, cytokines, integrins and their cognate receptors including the PAR family receptors [61,62]. PAR 2 has been shown to stimulate activation, proliferation, collagen synthesis and TGF-β production in HSCs and to promote experimental liver fibrosis [45,[63][64][65].
In this study, we provide for the first time evidence that PAR 2 in HSCs stimulates HCC growth. This was concluded from the finding that F2RL1 knockdown in hepatic stellate LX-2 cells inhibited the growth-promoting effect of these cells in a HCC xenograft SCID mouse model, as demonstrated by significantly smaller volumes of the tumours established in mice by coinjection of Hep3B and LX-2 cells depleted of PAR 2 . In addition, immunohistochemical analyses of these tumours revealed lower indices for proliferation (reduced immunoreactivity of Ki67 in both Hep3B and LX-2 cells), and angiogenesis (reduced CD31 in endothelial cells). This suggests a promoting role for HSC-derived PAR 2 in HCC growth through proproliferative, pro-angiogenic and possibly fibrotic effects on tumour and stromal cells in the HCC microenvironment. The pro-mitotic effect of HSC-derived PAR 2 was subsequently shown in cell counting experiments with Hep3B cells stimulated with conditioned media from PAR 2 depleted LX-2 cells. Of note, this growth promoting effect was specific to PAR 2 and was not observed for the related PAR 1 . These data are in agreement with the in vivo data and show that HSCs act on tumour cells through the release of soluble mitogenic factors. We also observed that transfection of PAR 2 siRNA decreased LX-2 cell numbers (data not shown), suggesting the possibility that HSCderived PAR 2 also contributes to tumour growth by promoting HSC proliferation in a cell-autonomous or auto-/paracrine manner. Given that activated HSCs are the major matrix producing cells in the liver, this may secondarily result in enhanced total matrix accumulation and eventually fibrosis.
It is now established that HSCs may facilitate HCC development, progression and metastasis by different processes including migration to the cancer site. More specifically, during cancer development activated HSCs infiltrate HCC stroma and peri-tumoural tissue and secrete substantial amounts of bioactive proteins [58] which may act either directly on the cancer cells or help to establish a microenvironment permissive for HCC progression. Migration of HSCs is known to be driven mainly by PDGF/PDGFR-dependent mechanisms [66][67][68], including a regulatory role for nitric oxide (NO) and the small GTPases, Rac1 and Rho [69,70]. Notably, PAR 2 induced the activation of PDGFR and Met following stimulation with PAR 2 -AP or trypsin, suggesting the possibility that PAR 2 utilizes these pathways to drive the migratory capacity of LX-2 cells. The importance of this PAR 2 -PDGFR-Met crosstalk was further supported by the finding that the PAR 2 -mediated migration of LX-2 cells was blocked by AG 1296 and SU 11274/PHA 665752, respectively. In addition, we provided evidence for a role of Src and p42/p44 MAPK in PAR 2 -PDGFR/Met-induced signalling to PAR 2 -induced invasion in LX-2 cells, since this could be inhibited by PP2 and SL 327/U 0126, respectively, and stimulation of PAR 2 with PAR 2 -AP or trypsin induced phosphorylation of Src and p42/p44 MAPK. Moreover, the PAR 2 -induced effect on p42/p44 MAPK could be blocked by AG 1296 and PHA 665752. These results also demonstrate that the PAR 2 -triggered phosphorylation of Src is upstream while p42/p44 MAPK is downstream of PDGFR and Met activation in LX-2 cells. Our results further indicate that, in addition to a Src-PDGFR/Met-p42/p44 MAPK signalling axis, MMP activity seems to be involved in PAR 2 -mediated invasive effect on LX-2 cells as concluded from the findings that LX-2 cells treated with PAR 2 -AP secrete MMPs 1, 3, 7, 13 and that the MMP inhibitor GM-6001 inhibited the effect of PAR 2 -AP and trypsin on LX-2 cell migration. PAR 2 activation augments inflammatory and pro-fibrotic pathways through the induction of genes encoding proinflammatory cytokines and extracellular matrix proteins [45]. In line with this, our in vitro experiments with LX-2 cells showed that activated PAR 2 not only stimulated cell migration but also the release of IL-8, uPA, cathepsins, MMPs and ADAMTS 1. All these factors are known to contribute to HCC initiation, growth and progression of HCC by degrading and remodeling the extracellular matrix and by stimulating angiogenesis and immunomodulation [71][72][73][74][75][76][77].
In addition to pro-inflammatory cytokines, TGF-β is considered a major factor promoting liver carcinogenesis. TGF-β1 and TGF-β2 are overexpressed in HCC [50,51] and may also drive transformation of HSCs into myofibroblasts [52,53] as well as their migration and invasion [54]. Notably, HSCs and LX-2 cells are sensitive to TGF-β treatment [46,61], and TGF-β gene expression and protein production in HSCs is stimulated by PAR 2 [45]. Prompted by recent observations in other cell types [55], we hypothesised that PAR 2 expression is also required for TGF-β signaling in a cell-autonomous manner. To this end, depletion of PAR 2 but not PAR 1 from LX-2 cells inhibited the ability of TGF-β1 to activate Smad2 and Smad3. This suggests that HSC-derived PAR 2 , in addition to driving tumour cell and HSC proliferation may promote TGF-β sensitivity of HSCs and, as a consequence, matrix production.
Evidence for a crucial role for stromal cell-derived PAR 2 in promoting primary tumour growth came also from mouse models for other solid and stroma-rich cancer types. PAR 2 deficient mice (harbouring PAR 2 deficient stromal cells) exhibit decreased tumour growth and metastasis in a spontaneous polyoma middle T (PyMT) breast cancer model [78]. In a mouse model of pancreatic cancer, stromal PAR 2 promoted primary tumour growth [79]. In tumours resulting from B16 melanoma cells injected subcutaneously, PAR 2 limited primary tumour growth [80]. The role of stromal PAR 2 in metastasis, however, appears to be strongly contextdependent. Stromal PAR 2 limited lymph node metastasis in a spontaneous polyoma middle T (PyMT) breast cancer model [78] and in the pancreatic cancer mouse model [79] but enhanced spontaneous metastasis of B16 melanoma cells injected subcutaneously [80]. Finally, PAR 2 has been shown to play a critical role in stimulating tumour angiogenesis in the context of mammary carcinoma [81].

Conclusion
Our study provides novel insight into the function of PAR 2 in HSCs, which may be involved in HCC growth and progression. PAR 2 induces secretion of proangiogenic and pro-mitotic factors and proteinases, directly stimulates tumour cell growth in vitro through secreted factors and activates various signalling pathways (Src, Met, PDGFR, p42/p44 MAPK) being related to HSC growth and matrix production. It can be hypothesised that these mechanisms are of importance in the HCC microenvironment where both tumoural and stromal cells may produce PAR 2 -activating proteinases, like the kallikreinrelated peptidases [82,83] driving PAR 2 -mediated effects in HSCs on HCC growth.
Numerous studies convincingly demonstrate that HSCs play a critical role in HCC growth and progression [7][8][9][10][11][12][13][14][15][16][17][18]. From our data on LX-2 cells PAR 2 may be concluded as a player in this scenario. This was supported by preliminary data demonstrating PAR 2 function also in primary cultures of human HSCs (R.K. and H.U., unpublished observation). Since we used exclusively Hep3B cells as a HCC cellular model so far, analogous studies with other HCC cell lines and well-characterised primary HCC cultures are warranted to confirm the tumour-promoting function of PAR 2 in HSCs.

Peptide synthesis
The PAR 1 -selective peptide, TFLLRN-NH 2 , PAR 2 -AP, 2-furoyl-LIGRLO-NH 2 and PAR 2 -RP 2-furoyl-OLRGIL-NH 2 , were synthesised by Fmoc strategy on an ABI-Peptide-Synthesizer 433A using H-Rink Amide ChemMa-trix®resin (capacity 0.47 respectively 0.52 mmol/g; PCAS BioMatrix Inc, Canada). The furoyl-group was coupled manually with 5 equivalents of acid. The cleavage of the peptides from resin was performed with trifluoroacetic acid, 5.0 % H 2 O und 3.0 % triisopropylsilane. The peptides were precipitated by diethylether and lyophilized. Crude synthetic peptides were purified using preparative HPLC on a 30 × 250 mm Kromasil C18-column with a flow rate of 40 ml per minute under standard conditions (buffer A: 0.2 % TFA in water, buffer B: 0.2 % TFA in water:acetonitrile, 1:4). The purified peptides were dried by lyophilization and characterised by analytical HPLC and mass spec analysis on a Voyager-DE PRO workstation. For calcium measurement in single cells, an inverted confocal laser scanning microscope (LSM 510, Carl Zeiss, Göttingen, Germany) was used. Fluorescence images were collected by using the 488 nm argon ion laser line and a 505 nm long pass filter. All fluorescence measurements were made from subconfluent areas of the dishes, enabling the ready identification of individual cells. Image data were analysed using the Carl Zeiss AIM software Version 4.2. A selected image in each image set was used as a template for designating each cell as a region of interest.

Reverse Transcription (RT)-PCR analysis
Total RNA was extracted from 1 × 10 7 LX-2 cells (RNeasy ® Mini Kit, Qiagen GmbH, Hilden, Germany), and for RT-PCR analysis the Reverse Transcription System (Cat. No. A 3500) from Promega Corporation (Madison, WI, USA) was used. For PCR amplification the following primer pairs were used: forward primer: 5'-TGGATGAGTTTT CTGCATCTGTCC-3' and 5'-CGTGATGTTCAGGGCA GGAATG-3'. The primers were constructed to generate a fragment of 490 bp for PAR 2 . PCR amplification was performed with Taq polymerase for 32 cycles at 94°C for 30 s, 55°C for 30 s, and 72°C for 30 s, and, finally 72°C for 5 min [87]. Amplified samples were electrophoresed on a 2.0 % agarose gel containing 0.2 μg/ml SybrGreen and visualized under UV transillumination. In addition, PCR products for PAR 2 of LX-2 cells were purified and DNA sequences were determined by 4base lab GmbH, Reutlingen (Germany).

Transient transfection of LX-2 cells with PAR 2 siRNA and TGF-β stimulation
For transient silencing of F2RL1, LX-2 cells seeded on day 1 (2.8 × 10 5 /6-well) underwent two rounds of a 4-h transfection on days 2 and 3 with 50 nM of siRNA to PAR 2 (Invitrogen) and Lipofectamine 2000 transfection reagent (Life Technologies). As control, a siRNA to PAR 1 or a universal control siRNA (both from Life Technologies) was transfected in parallel. Successful knockdown of PAR 2 and PAR 1 expression was verified by quantitative real-time (RT)-PCR as outlined elsewhere [55] (data not shown). For sequence information on siRNAs and PCR primers see [55]. Forty-eight hours after the second transfection, supernatants to be used for Hep3B proliferation assays (see below) were collected, centrifuged briefly to remove dead cells and cellular debris and stored frozen until use. The remaining LX-2 cells were stimulated for 0.5 h with TGF-β1 (10 ng/ml, ReliaTech, Wolfenbüttel, Germany) in normal growth medium followed by lysis in RIPA buffer containing protease (Boehringer Complete, PMSF)/phosphatase (NaF, sodium orthovanadate) inhibitors and phosphoimmunoblotting for Smad2/3 (see below).
For testing the effect of PAR 2 silencing on cell viability, cells were seeded into 24-well plates and after treatment with the respective siRNA the viable cells were counted by trypan blue dye exclusion test.
To generate lentiviral particles, HEK293T cells were maintained in DMEM (Biochrom) supplemented with 10 % FCS. The cells were transiently transfected with the pLKO.1-derived plasmids in combination with pRev, pEnv-VSV-G and pMDLg using polyethyleneimine. After 12, 24 and 48 h, media containing the retrovirus particles were collected. LX-2 cells were infected three times with the particles in the presence of 8.0 μg/ml polybrene. Forty-eight hours after transduction, transduced cells were selected with 2.0 μg/ml puromycin. The efficiency of the knockdown was assessed by quantitative real-time RT-PCR and by calcium measurements. PCR amplification was performed with Maxima Hot Start Taq DNA Polymerase for 40 cycles at 94°C for 30 s, 60°C for 30 s and 72°C for 30 s and finally 72°C for 5 min. Primers were forward: 5'-TGGATGAGTTTTCTGCATCTGTCC-3' , and reverse: 5'-CGTGATGTTCAGGGCAGGAATG-3'. The qPCR was performed on a MasterCycler RealPlex4 (Eppendorf, Hamburg, Germany). All values for PAR 2 expression were normalized to those for the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH) in the same sample.
For calcium measurements the effect of trypsin and the PAR 2 -AP, 2-furoyl-LIGRLO-NH 2 , on [Ca 2+ ] i mobilisation in PAR 2 knockdown-LX-2 cells was investigated as described above.

Immunofluorescence studies
The expression of PAR 2 on protein level was established using the monoclonal anti-PAR 2 antibody SAM-11, generated against a peptide representing amino acids 37 SLIGKVDGTSHVTG 50 of human PAR 2 (Santa Cruz Biotechnology). This antibody has been validated for western blot and immunohistochemical/flow cytometry procedures [88,89].

Analysis of plasma membrane filopodial structures
LX-2 cells were grown on glass coverslips. After incubation with PAR 2 -AP for 4 h, the cells were fixed with 2.5 % glutaraldehyde and subsequently dehydrated as described for detection of PAR 2 immunolabeling. The cells were sputter coated with gold and examined in a LEO 1450 VP scanning electron microscope (Carl Zeiss) at 10 kV acceleration voltage and a working distance of 7 mm. In this study, the cells were scored as positive or negative for filopodial spikes. Filopodia-positive cells were defined as having more than 10 thin processes beyond the cellular edge of the plasma membrane [90]. Quantitative analyses were performed in a blinded manner from 100 individual cells per each group and from three experimental preparations.

Protein assay
Protein was determined using the DC Protein Assay System from BioRad Laboratories according to the manufacturer instructions.
Analysis of Src, Met, PDGFR, p42/p44 MAPK and Smad2/3 phosphorylation Immunoprecipitation (IP) and Western blotting For IP, cleared cell lysates normalized for protein content were treated with 2.0 μg anti-PDGFR antibody for 2-3 h, followed by 20 μl protein A-Sepharose beads overnight at 4°C. Beads were collected by centrifugation, washed three times with lysis buffer containing 2.0 μg/ ml leupeptin, 1.0 μg/ml pepstatin A, 100 μg/ml PMSF, 100 μg/ml aprotinin, 1.0 mM Na 3 VO 4 and 1.0 mM NaF. Ten μl Roti®-Load were added to the beads and the samples were heated for 5 min at 95°C.
Proteins of cell lysates (Src, Met, p42/p44 MAPK and Smad 2/3 detection) or immunoprecipitates (PDGFR detection) were subjected to 10 % SDS-PAGE and transferred to nitrocellulose membranes (BioRad Laboratories). After blocking in 1.0 % BSA/1.0 % skimmed milk for 1 h, the nitrocellulose strips were incubated overnight with the respective first antibody. Strips were washed two times with 0.05 % Tween 20 washing buffer, incubated for 45 min with the respective secondary antibody conjugated to horseradish peroxidase and washed again two times as described above. In all experiments, the immunoblots were stripped and reprobed with antibodies to total protein to confirm equal protein loading. Secondary antibodies were detected by using the chemiluminescence (ECL) Western blotting detection system (Amersham) and exposure to Kodak X-Omat films.
Immunoreactive bands for phosphorylated forms of Src (p-Src), Met (p-Met), PDGFR (p-PDGFR),p42/p44 MAPK (p-p42/p44 MAPK), Smad2 and Smad3 (p-Smad2/3) as well as Src (t-Src), total Met (t-Met), total PDGFR (t-PDGFR), total p42/p44 MAPK (t-p42/p44 MAPK), and β-actin were quantified using the image processing program Image J 1.43 (National Institutes of Health, Bethesda, Maryland, USA). Briefly, nitrocellulose membranes were blocked for 1 h and then incubated with 500 μl of supernatant (24 h after treatment with PAR 2 -AP or vehicle) overnight at 4°C. After washing, the membranes were incubated with the detection antibody cocktail for 2 h and streptavidin-HRP for 30 min. For data evaluation, the membranes were exposed to chemiluminescent reagents (ECL Plus Western Blotting Detection System; Amersham) and finally to X-ray films for 30 s. The developed films were scanned and the average pixel density of the positive signals was analysed using the analysis software Image J 1.43 (National Institutes of Health, Bethesda, Maryland, USA). All images from the arrays were normalized by subtracting the background and inverted to eliminate the background differences. To measure the pixel density, a fixed size rectangular box was generated around each dot blot/band and the pixel density was measured. The same sized rectangular box was used for all bands in all arrays performed. For analysis, the pixel densities of the negative control on each array were subtracted from the pixel densities obtained from each band on the array. The data was further converted and normalized into fold change in expression by dividing the pixel densities of each band by the average pixel densities of the Streptavidin-HRP reference spots (R) located at the three corners of each array.

Stimulation of Hep3B cells with LX-2 cell conditioned media and cell counting
Hep3B cells were seeded in 48-well plates (2 × 10 4 /well). On the next day, the growth medium was removed and replaced by 500 μl of cleared culture supernatant conditioned for 48 h by LX-2 cells transfected with either control siRNA, PAR 1 siRNA or PAR 2 siRNA (see above: "Transient transfection of LX-2 cells with PAR 2 siRNA and TGF-β stimulation"). Following incubation with the LX-2 conditioned media for 72 h, the medium was removed, centrifuged to collect non-adherent cells and the resuspended pellets combined with the detached adherent cells from the respective well. Then, viable and dead cells were counted by trypan blue dye exclusion test.
Cell migration assay LX-2 cell migration was measured using a 48-well boyden-chamber (NeuroProbe, Inc., Gaithersburg MD, USA). 51 μl of the cell suspension (4 × 10 4 cells in RPMI-1640) with or without inhibitor were placed in each upper chamber well and 27 μl of cell culture medium containing the chemoattractant or vehicle in each lower well. Then, incubation for 6 h at 37°C in a humidified incubator with 5.0 % CO 2 was performed to allow cell migration through a porous polycarbonate filter (6.5 mm in diameter, 8.0 μm pore size) precoated with collagen (0.35 %). After the incubation period, the filter was removed, and its upper side was wiped gently with a cotton tip swab to remove non-migrated cells. The migrated cells were fixed on the lower surface of the membrane with 96 % ethanol, stained with Giemsa solution, and counted under a Zeiss Axiolab microscope. Data were acquired from three independent experiments, involving octuplicate measurements for each condition.

Estimation of cell viability
For testing the effect of inhibitors on LX-2 cell viability, cells were seeded into 24-well plates. Following treatment with the respective inhibitor for 24 h, viable and non-viable cells were distinguished by trypan blue dye exclusion method and counted manually. In addition, this test was used to monitor the viability of LX-2 cells from cultures used for animal experiments and that of Hep3B cells incubated with LX-2 culture supernatants.

Animal model
Five to six week-old, male SCID mice (BALB/cJHAN®-Prkdc scid ) were purchased from Harlan Laboratories Ltd. (Venray, Germany) and were handled under governmental and institutional rules and regulations after approval by the respective authorities. Animals were housed under specific pathogen-free conditions with a 14 h light/12 h dark cycle and received food and tap water ad libitum.