The von Hippel–Lindau tumor suppressor gene product represses oncogenic β-catenin signaling in renal carcinoma cells

Abstract

Loss of von Hippel–Lindau (VHL) tumor suppressor gene function occurs in familial and most sporadic clear cell renal cell carcinoma (RCC), resulting in the aberrant expression of genes that control cell proliferation, invasion, and angiogenesis. The molecular mechanisms by which VHL loss leads to tumorigenesis are not yet fully defined. VHL loss has been shown to allow robust RCC cell motility, invasiveness, and morphogenesis in response to hepatocyte growth factor (HGF) stimulation, processes that are known to contribute to tumor invasiveness and metastatic potential. Among the most likely intracellular mediators of these HGF-driven activities is β-catenin, a structural link between cadherens and the actin cytoskeleton, as well as a gene transactivator. We show that reconstitution of VHL expression in RCC cells repressed HGF-stimulated β-catenin tyrosyl phosphorylation, adherens junction disruption, cytoplasmic β-catenin accumulation, and reporter gene transactivation in RCC cells. Ectopic expression of a ubiquitination-resistant β-catenin mutant specifically restored HGF-stimulated invasion and morphogenesis in VHL-transfected RCC cells. VHL gene silencing in non-RCC renal epithelial cells phenotypically mimicked VHL loss in RCC, and HGF-driven invasiveness was blocked by the expression of a dominant-negative mutant of Tcf. We conclude that, unlike many other cancers, where HGF pathway activation contributes to malignancy through the acquisition of autocrine signaling, receptor overexpression, or mutation, in RCC cells VHL loss enables HGF-driven oncogenic β-catenin signaling. These findings identify β-catenin as a potential target in biomarker and drug development for RCC.

Von Hippel–Lindau (VHL) syndrome is an autosomal dominant hereditary neoplastic disorder (1, 2). VHL-associated clear cell renal cell carcinoma (RCC) tumors are malignant and frequently metastatic (2). Defects in the VHL tumor suppressor gene, which is located on the short arm of chromosome 3 (3p25–26), lead to VHL syndrome and also occur in the majority of sporadic clear cell RCC cases (3). Reconstitution of WT VHL expression in RCC-derived cells regulates tumorigenesis in mice (4), confirming a critical role for VHL in RCC. The VHL protein (pVHL) is part of an E3 ubiquitin ligase complex that targets hypoxia-inducible factors (HIFs) for polyubiquitination and proteosomal degradation (5). During hypoxia or when pVHL function is lost, HIFs accumulate and cause broad changes in gene expression that are potentially important in oncogenesis (2, 5, 6). Cultured VHL-negative RCC cells also acquire an abnormal response to hepatocyte growth factor (HGF), manifested as matrix degradation, motility, invasion, and morphogenesis (7). These HGF-driven activities are abolished when WT VHL expression is reconstituted in RCC cells, directly linking loss of VHL function to an invasive tumor phenotype (7).

HGF signaling between mesenchymal and epithelial cells is a major driving force in epithelial growth and differentiation and also regulates transitions between these cell types during kidney development. Inappropriate HGF pathway activation in cancer can resemble these HGF-driven developmental transitions (8–11). Expression of HGF and its receptor, c-Met, persist in adult kidney, but the type of HGF response observed in renal epithelial cells changes dramatically upon completion of development. HGF protects adult kidney tissue from acute toxicity and ischemic stress (12) and counteracts renal fibrosis (13), whereas morphogenic and proliferative responses are minimized. Many intracellular signaling pathways persist through development into adulthood, and it remains unclear which signals are modified or silenced to provide a homeostatic, as opposed to developmental or pathophysiological, HGF response.

Among the known intracellular mediators of HGF signaling in the adult is β-catenin (14–17), which links cadherins to the actin cytoskeleton and also functions as a gene transactivator. β-Catenin and E-cadherin are also expressed during renal development, specifically upon transition of the mesenchyme surrounding the branching ureteric buds to the epithelium that will form the tubules of the nephron (18). This mesenchymal-to-epithelial transition and the ensuing tubule formation involve several Wnt family members acting in an autocrine manner (19, 20), as well as HGF acting in a paracrine mode (21). Dysregulated β-catenin signaling in the adult can be potently oncogenic. Cytoplasmic β-catenin is normally targeted for ubiquitination by a complex of glycogen synthase kinase 3-β, a serine/threonine kinase that phosphorylates β-catenin, adenomatous polyposis coli (APC), and axin, a scaffold protein that holds the complex together. Mutations in the genes encoding APC or β-catenin result in uncontrolled β-catenin signaling and colorectal cancer (22). In both cases, β-catenin bypasses APC-mediated ubiquitination and proteosomal degradation, and it is now known that cytoplasmic β-catenin can be stabilized by a variety of genetic defects (22). Oncogenic missense β-catenin mutants undergo substitutions at positions normally phosphorylated by glycogen synthase kinase 3-β, and thus escape targeting (22). In most non-colon cancers, APC or β-catenin mutations are rare (22, 23). Nonetheless, many of these tumor samples show cytoplasmic and/or nuclear accumulation of β-catenin that occurs through as yet undefined mechanisms (24). We hypothesized that pVHL might negatively regulate β-catenin signaling in mature renal tubule epithelial cells and that VHL loss in clear cell RCC might promote oncogenic β-catenin signaling downstream of HGF, contributing to an aberrantly motile and invasive tumor cell phenotype.

Results

HGF-Induced Invasion and Morphogenesis Are Repressed in Renal Epithelial Cells and Derepressed in VHL-Negative RCC Cells.

3D Matrigel cultures of normal human renal proximal tubule epithelial cells (RPTEC), which have no known defects in the VHL gene or pVHL protein function, were analyzed for HGF-driven cord formation (also referred to as branching morphogenesis). HGF failed to stimulate cord formation in 3D RPTEC cultures (Fig. 5, which is published as supporting information on the PNAS web site). In contrast, a robust HGF-driven morphogenic response was exhibited by the VHL-negative RCC cell line 786-0 (Fig. 5). As shown previously, stable ectopic expression of pVHL in 786-0 repressed HGF-stimulated cord formation (Fig. 5) as well as HGF-stimulated protease production (7). These results suggest that HGF-stimulated morphogenesis in developing renal epithelial cells is repressed at maturity and derepressed during oncogenic transformation. Consistent with a prior study (7), we found that levels of c-Met expression were identical in 786-0 and 786-0/VHL cell lines (data not shown). Moreover, neither cell line showed evidence of HGF mRNA or any abnormalities in ligand-stimulated c-Met autophosphorylation (data not shown), suggesting that pVHL repressed postreceptor HGF signaling. To investigate the possibility that β-catenin mediated HGF-stimulated morphogenesis in VHL-negative RCC cells, we transfected 786-0/VHL cells with an expression plasmid encoding the S37A β-catenin mutant, which cannot be phosphorylated at serine-37 by glycogen synthase kinase 3-β and is thus ubiquitination-resistant. Consistent with the expectation that S37A β-catenin could be mobilized from junctions by HGF but escape proteosomal degradation, a full morphogenic response was restored in 786-0/VHL/S37A cells (Fig. 5).

The RCC cell lines 786-0, 786-0/VHL, and 786-0/VHL/S37A were also tested for HGF-stimulated invasion in vitro by using growth factor-reduced Matrigel-coated filters (Fig. 1). Little invasion was observed for any serum-deprived cell line (Fig. 1 Left). The invasiveness of 786-0 cells was markedly stimulated by serum (Fig. 6 Top Center, which is published as supporting information on the PNAS web site), and restoration of VHL dramatically repressed this effect in both 786-0/VHL and 786-0/VHL/S37A cell lines (Fig. 6 Middle and Bottom). HGF-stimulated invasiveness in 786-0, but not 786-0/VHL, cells and expression of the S37A β-catenin mutant were associated with a dramatic increase in 786-0 cell invasion despite the presence of pVHL (Fig. 1 Right). Importantly, expression of the β-catenin mutant did not increase serum-stimulated invasiveness over that of 786-0/VHL cells (Fig. 6 Center), suggesting that the β-catenin mutant specifically enhanced HGF signaling. Consistent with results obtained in 3D cord formation assays, restoration of β-catenin signaling downstream of c-Met bypassed the repressive effects of ectopic pVHL expression in RCC cells.

Fig. 1.

Expression of the S37A β-catenin mutant rescues HGF-stimulated extracellular matrix invasion in VHL-transfected RCC cells. Bright-field photomicrographs show DiffQuick-stained cells that traversed Matrigel-coated Transwell filters. (Left) Serum-deprived, untreated cells. (Right) Cells treated with HGF (40 ng/ml for 30 h). HGF-stimulated invasion by the RCC-derived cell line 786-0 (Top) is repressed by WT pVHL expression (Middle). HGF-driven invasion is rescued upon supertransfection with the ubiquitination-resistant S37A β-catenin mutant (Bottom). (Magnification: ×4.)

VHL Expression Represses HGF-Induced β-Catenin Mobilization, Cell Shape Change, and Motility.

c-Met activation can disrupt the β-catenin/E-cadherin adherens junction complex in several cell models of tubulogenesis, tumorigenesis, and metastasis, in some cases coincident with β-catenin tyrosyl phosphorylation (14–17). Tyrosyl phosphorylation of β-catenin on residues tyrosine-142 and tyrosine-654 decreases its affinity for α-catenin and E-cadherin, respectively (25–27). In the absence of ubiquitination and degradation, β-catenin freed from cell junctions can translocate to the nucleus and activate gene transcription (15–17, 28). We investigated whether pVHL expression affected β-catenin tyrosyl phosphorylation after HGF stimulation of RCC cells (Fig. 2). Immunoprecipitation with anti-c-Met followed by immunoblotting with either anti-c-Met or anti-phosphotyrosine (pY) confirmed HGF-induced c-Met activation in 786-0 and 786-0/VHL cells (Fig. 2A Left). Immunoprecipitation of β-catenin from these cells followed by immunoblotting with either anti-pY or anti-β-catenin revealed HGF-stimulated tyrosyl phosphorylation of β-catenin in 786-0 but not 786-0/VHL cells (Fig. 2A Center). To determine whether tyrosyl phosphorylation of β-catenin in 786-0/VHL cells was blocked or whether the phosphorylated species was rapidly degraded, cells were treated with HGF in the presence or absence of the proteosomal inhibitor PS341 before lysis and immunoprecipitation with anti-β-catenin (Fig. 2A Right). Immunoblot detection of tyrosyl phosphorylated β-catenin only in the presence of PS341 indicated that this species was indeed rapidly degraded in cells expressing pVHL. Moreover, immunoblotting these samples with anti-β-catenin revealed the marked accumulation of a high-molecular-mass form of β-catenin (likely to represent polyubiquitinated protein) in response to HGF stimulation (Fig. 2A Lower Right, arrow). These results, together with the absence of increased levels of high-molecular-mass β-catenin in VHL-negative cells (data not shown), implicate pVHL in the ubiquitination and proteosomal destruction of tyrosyl phosphorylated β-catenin. Consistent with its resistance to proteosomal targeting, the 786-0/VHL/S37A cell line retained HGF-induced β-catenin tyrosyl phosphorylation despite pVHL expression (Fig. 2B).

Fig. 2.

HGF-stimulated β-catenin tyrosyl phosphorylation is repressed by VHL in RCC-derived cell lines. Serum-deprived cells were left untreated (−) or stimulated with HGF (100 ng/ml for 30 min; +) before lysis and nonionic detergent extraction. (A) 786-0 or 786-0/VHL cell extracts were analyzed by immunoprecipitation (IP) with anti-β-catenin (β-cat) and immunoblotted (Blot) with anti-pY (pY) (Upper), anti-c-Met (Left Lower), or anti-β-catenin (Center and Right Lower). 786-0/VHL cells were pretreated for 16 h with PS341 before HGF stimulation (Right) and lysate preparation for immunoprecipitation and immunoblotting. The arrow on the right indicates a high-molecular-mass form of β-catenin that is likely to be polyubiquitinated and whose accumulation was induced by HGF. (B) Immunoprecipitation of 786-0/VHL/S37A cell extracts with anti-c-Met (Left) or anti-β-catenin (Right) and immunoblotting with anti-pY (Upper), anti-c-Met (Lower Left), or anti-β-catenin (Lower Right).

The association between VHL expression and the rapid degradation of tyrosyl phosphorylated β-catenin was not restricted to 786-0 cells: two other RCC-derived cell lines with complete loss of VHL function, UOK121 and UOK161, each showed HGF-induced tyrosyl phosphorylation of β-catenin (Fig. 7 A and B, respectively, which is published as supporting information on the PNAS web site). Reconstitution of pVHL expression UOK121 blocked HGF-induced β-catenin tyrosyl phosphorylation, as it did in 786-0 (Fig. 7A). Thus, independent of the mechanism of VHL loss of function, HGF stimulated β-catenin tyrosyl phosphorylation, enabling its mobilization from junctions to cytoplasm.

To evaluate the effect of VHL expression on cell shape changes and β-catenin redistribution associated with HGF-stimulated motility, indirect immunofluorescence studies were performed on 786-0 and 786-0/VHL cells in the absence and presence of HGF treatment (Fig. 3). Consistent with prior observations (29), phase contrast microscopy revealed that untreated VHL-negative cells were less well organized as an epithelial monolayer than their VHL-expressing counterparts (data not shown). Immunofluorescence with anti-β-catenin showed that intercellular contacts were less extensive and β-catenin was more randomly distributed in the absence of VHL expression than in its presence (Fig. 3, compare Upper Left and Lower Left). After HGF stimulation, 786-0 cells displayed shape changes typical of HGF-induced scatter, including cell dispersion, trapezoidal cell shape, and a significant redistribution of β-catenin from the cell periphery to cytoplasmic, perinuclear, and nuclear pools or to focal contacts at leading lamellipodia and cell anchor points (Fig. 3 Upper Right). In contrast, 786-0/VHL cells maintained a polygonal epithelial morphology with β-catenin retained predominantly in intercellular contacts (Fig. 3 Lower Right), consistent with a strong inhibitory effect of pVHL on HGF-induced cell shape change, motility, and β-catenin mobilization.

Fig. 3.

HGF-induced cell shape change, motility, and β-catenin mobilization in RCC cells are repressed by VHL. Shown is immunofluorescence microscopy of β-catenin (red) in 786-0 (Upper) and 786-0/VHL (Lower) cell lines. DAPI staining (blue) shows nuclear localization. Cells shown in Left are untreated; cells in Right were treated with HGF (40 ng/ml for 16 h). (Magnification: ×40.)

VHL Expression Destabilizes Cytoplasmic β-Catenin and Blocks Target Gene Transactivation.

High-magnification images of RCC cells treated with HGF highlight the consistent redistribution of β-catenin to cytoplasmic, perinuclear, and nuclear pools in 786-0 (Fig. 8 Left, which is published as supporting information on the PNAS web site) and the repression of these events associated with VHL expression in 786-0/VHL cells (Fig. 8 Right). To quantitate HGF-driven β-catenin mobilization to the cytoplasm in the presence and absence of VHL, we used an E-cadherin–GST fusion protein to capture cytoplasmic β-catenin from nonionic detergent extracts prepared from control and HGF-treated 786-0, 786-0/VHL/S37A, and 786-0/VHL cells (Fig. 4A). This fusion protein is selective for cytoplasmic β-catenin because junctional β-catenin is already associated with endogenous E-cadherin and will not be captured, whereas nuclear β-catenin is not extracted with nonionic detergents (30). SDS/PAGE and immunoblotting of E-cadherin–GST bound β-catenin revealed that HGF stimulation increased the levels of cytoplasmic β-catenin relative to untreated controls in both 786-0 and 786-0/VHL/S37A cell lines, but not in 786-0/VHL cells, consistent with the ability of each cell line to respond to HGF with cord formation and matrix invasion (Fig. 4A).

Fig. 4.

β-Catenin mediates HGF-stimulated branching in VHL-negative RCC. (A) E-cadherin–GST (E-cad-GST) pull-down assays performed on untreated (−) and HGF-treated (+) 786-0 (Left), 786-0/VHL/S37A (Center), and 786-0/VHL (Right) cells immunoblotted for β-catenin (Upper). Lower shows total β-catenin content of each lysate. (B) Coimmunoprecipitation of pVHL and β-catenin in 786-0/VHL cells. Lysates prepared from 786-0 (Upper Left) or 786-0/VHL (Lower Left) cells immunoblotted for β-catenin show equivalent amounts of protein. Lysates were immunoprecipitated with anti-pVHL and immunoblotted with anti-β-catenin or anti-pVHL (Right). (C) Photomicrographs of 3D Matrigel branching assays of control (−) and HGF-treated (+) HEK 293/-VHL (Left) and HEK 293/Scramble (Right) cells. Numerical values below each photo indicate mean TOP-FLASH reporter assay results (fold activation of TOP-FLASH reporter over control FOP-FLASH reporter) obtained from triplicate samples of control or HGF-treated cells. Error bars indicate standard deviation. (Magnification: × 10.) (D) Photomicrographs of 3D Matrigel cultures of HGF-treated HEK 293/-VHL cells (Left) and HEK 293/-VHL/DN TCF cells (Right). (Magnification: ×10.)

To determine whether β-catenin could be found physically associated with pVHL protein in 786-0/VHL cells, coimmunoprecipitation experiments were performed with anti-VHL antibody followed by immunoblotting with anti-β-catenin or anti-VHL (Fig. 4B). No differences in the total amount of β-catenin protein were observed in SDS extracts prepared from 786-0 and 786-0/VHL cell lines (Fig. 4B Left). No β-catenin was detected after immunoprecipitation from 786-0 cells with anti-pVHL (Fig. 4B Upper), which serves as control for the nonspecific capture of β-catenin by anti-pVHL or protein G-Sepharose beads. In contrast, both β-catenin and pVHL were readily detected after immunoprecipitation from 786-0/VHL cells, indicating a physical association between the two proteins (Fig. 4B Lower). These results reinforce the likelihood that pVHL blocks β-catenin signaling before β-catenin nuclear translocation.

β-Catenin-mediated target gene activation was analyzed by using the TOP-FLASH luciferase reporter system. For these studies we used HEK 293 human kidney epithelial cells, which are not RCC-derived and express WT VHL. The effect of VHL silencing on β-catenin-mediated target gene activation was analyzed by stably transfecting HEK 293 cells with a VHL-specific short hairpin RNA expression plasmid (HEK 293/-VHL) or with a plasmid encoding a scrambled VHL-directed short hairpin RNA control oligonucleotide (HEK 293/Scramble). Quantitative RT-PCR analysis showed that VHL mRNA content was reduced 7.5-fold (87%) in HEK 293/-VHL relative to the parental cell line (P < 0.05). HEK 293/-VHL cells grown in 3D Matrigel cultures displayed obvious HGF-induced cord formation as well as increased colony size relative to untreated controls (Fig. 4C Left). In contrast, HEK 293/Scramble cells showed neither response to HGF (Fig. 4C Right). This model system was thus consistent with the HGF responsiveness observed in other VHL-positive and -negative renal epithelial cell lines used. HEK 293/-VHL and HEK 293/Scramble cells were then transiently transfected with either TOP-FLASH or FOP-FLASH reporter plasmids, and HGF-driven β-catenin transactivation activity was analyzed. HGF stimulated activation of the Tcf/LEF-sensitive TOP-FLASH reporter 22-fold over untreated controls in the VHL-silenced cells but had no effect on cells expressing native VHL (Fig. 4C, values below micrographs). To independently confirm that Tcf transactivation was required for HGF-stimulated cord formation in the absence of pVHL, the HEK 293/-VHL cell line was supertransfected with a dominant-negative Tcf expression plasmid (HEK 293/-VHL/DN TCF). As anticipated, cells harboring the mutant Tcf construct (Fig. 4D Right) failed to undergo the HGF-driven morphogenesis in 3D Matrigel cultures, characteristic of HEK 293/-VHL cells (Fig. 4D Left).

In summary, we found that c-Met activation in VHL-negative RCC cells led to the release of β-catenin from cell junctions to cytoplasm and then nucleus, promoting Tcf/LEF-mediated target gene activation, cell contact disruption, motility, matrix invasion, and morphogenesis. In contrast, although HGF-stimulated c-Met activation occurred normally in VHL-positive RCC cells, subsequent β-catenin-mediated signaling events and associated phenotypic changes were blocked.

Discussion

HGF stimulates mitogenesis, motogenesis, and morphogenesis during normal development (28, 31), whereas in adulthood HGF contributes to cytoprotection, wound healing, and occasionally tissue regeneration (12, 13, 32, 33). Clearly many of the HGF-stimulated activities that are essential for development are attenuated or blocked in adulthood. Consistent with this concept, human renal epithelial cell lines from adult (RPTEC) or late-stage embryo (HEK 293) did not respond to HGF with motility or morphogenesis. In VHL-negative RCC-derived cells, however, HGF stimulated robust motility, matrix invasion, and cord formation (ref. 7 and results herein). Moreover, as demonstrated first by Koochekpour et al. (7), repression of HGF-stimulated invasion and cord formation was restored by ectopic VHL expression.

Interestingly, the reciprocal temporal expression patterns of VHL and HIF during renal development point to a role for hypoxia in driving early nephrogenesis and for pVHL in attenuating this process (34–36). VHL mRNA has been found in the mesonephric duct epithelia and mesonephric tubules but not in early metanephric development (34). VHL is then re-expressed upon completion of renal tubulogenesis throughout the tubular epithelium (34). In contrast, the expression patterns of HIF-1α and HIF-2α are consistent with roles in promoting tubulogenesis and vasculogenesis, but upon completion of nephrogenesis HIF-1α and HIF-2α become undetectable, consistent with proteosomal targeting by up-regulated pVHL expression (36). Under normoxic conditions, tight control of HIF levels by pVHL persists in adulthood; when VHL function is lost, aberrant HIF target gene expression and other molecular changes contribute to oncogenesis (2, 5). In this light, the acquisition of HGF-driven invasiveness in clear cell RCC exemplifies the derepression of a robust developmental HGF signaling program in an adult setting.

The absence of autocrine signaling, receptor overexpression, or aberrant ligand-stimulated c-Met activation in RCC-derived cells supports the view that derepression of tubular morphogenesis associated with VHL loss occurs downstream of c-Met. Among the c-Met effectors known to mediate developmental events, to be repressed through ubiquitin targeted proteosomal degradation, and to have significant oncogenic potential, is β-catenin. We show that HGF stimulated the redistribution of β-catenin from peripheral to cytoplasmic, perinuclear, and nuclear pools in VHL-negative RCC cells, consistent with a shift of function from stabilizing intercellular adhesion to gene activation. Other recent studies have also noted aberrant intercellular junctions, diminished E-cadherin expression, and irregular β-catenin distribution in VHL-negative RCC cell lines (37, 38). Ectopic expression of VHL reverted these phenotypic alterations, consistent with our observations (37). Reduced membranous immunoreactivity and abnormal cytoplasmic accumulation of β-catenin have also been reported in immunohistological studies of RCC and correlated with advanced stage and nodal involvement (39–42).

Coincident with the repression of HGF-stimulated β-catenin redistribution in VHL transfectants, cell shape change, motility, and invasion were also blocked. pVHL has been shown to bind to microtubules and directly enhance cytoskeletal stability (43); these activities could also contribute to the inhibition of shape change and motility that we observed. E-cadherin pull-down studies demonstrated increased cytoplasmic β-catenin in HGF-stimulated VHL-negative RCC cells and blockade of this event upon ectopic VHL expression. β-Catenin is known to interact with the F-box proteins β-TrCP or Ebi in E3 ligase complexes; nevertheless, pVHL could be a component of an E3 complex that targets β-catenin for polyubiquitination in renal epithelial cells (44, 45). Hybrid E3 ubiquitin ligases comprised of both SCF (Skp1/Cul1/F-box) and VBC (VHL/EloB/EloC)-Cul2 components have been shown to regulate β-catenin degradation in other cell types (46). Our observation that S37A β-catenin escapes ubiquitination and reverses VHL repression of HGF-stimulated invasion and morphogenesis provides functional evidence that pVHL-associated E3 ligase activity results in β-catenin degradation. Proteosome blockade was required to capture tyrosyl phosphorylated β-catenin and to accumulate polyubiquitinated β-catenin in HGF-treated VHL transfectants, suggesting that β-catenin freed from junctions upon c-Met activation in pVHL-positive cells is normally polyubiquitinated and rapidly degraded. Evidence of β-catenin and pVHL physical association strengthens this conclusion. Although additional studies will be needed to define specific components of the E3 complex(es) that target β-catenin for ubiquitination in normal renal epithelial cells, VHL loss in RCC cells clearly disables the normal ubiquitin-mediated proteosomal degradation of cytoplasmic β-catenin. We further showed that short hairpin RNA silencing of VHL promoted HGF-stimulated morphogenesis and β-catenin target gene reporter activation in nontumor renal epithelial cells and that these activities were blocked by a dominant-negative form of Tcf. Thus, pVHL contributes to the normal ubiquitin-mediated regulation of β-catenin signaling downstream of c-Met in renal epithelial cells, and its loss of function enables c-Met-driven β-catenin target gene transcription and the acquisition of an invasive phenotype in RCC cells.

Oncogenic β-catenin signaling has been demonstrated in colon, breast, prostate, and lung carcinomas as well as melanoma (22, 44). Initial investigations suggest that activating mutations in β-catenin are rare in RCC tumors (39–41), and no inactivating mutations in APC or axin have been reported in RCC (47, 48). Oncogenic β-catenin signaling can also be initiated through aberrant Wnt stimulation or loss of negative repressors of Wnt signaling, such as members of the Dkk family. A recent study demonstrated the striking down-regulation of REIC/Dkk-3 in 15 of 17 (88%) RCC tumor samples (49). Like HGF, Wnt family members are normally expressed in kidney during development and in adulthood (19, 31, 50). Our results suggest that loss of this potential Wnt inhibitor combined with the frequent loss of VHL function in RCC could contribute significantly to tumorigenesis, invasion, and metastasis.

In addition to RCC, patients with VHL syndrome can display severely polycystic kidneys strongly resembling inherited polycystic kidney disease (PKD) (2). Allelic deletions of VHL have been detected in the cysts of VHL patients (2), suggesting that loss of VHL function contributes to this pathology. Evidence linking β-catenin signaling to the molecular pathogenesis of PKD raises the possibility that it may also contribute to the PKD-like phenotype seen in VHL syndrome. Mice carrying a conditional deletion of the APC gene in the renal epithelium displayed multiple kidney cysts and renal adenomas (51). Transgenic mice expressing an oncogenic β-catenin mutant in renal epithelial cells also developed severe polycystic lesions resembling human PKD (52). In the orpk mouse PKD model, mislocalization of β-catenin was observed in renal cysts (53). Mouse models of PKD show that either loss or overexpression of PKD1 give similar phenotypes, suggesting that a critical level of polycystin-1 expression may be needed for normal renal development and homeostasis (54). PKD1 is a β-catenin target gene (55), and HGF has been shown to regulate polycystin-1 expression in a cultured cell model of tubulogenesis and cystogenesis (56). Together these results suggest that VHL loss may enable HGF-driven, β-catenin-mediated polycystin-1 redistribution and/or PKD1 overexpression in cases of VHL syndrome manifesting a PKD-like phenotype.

The recent success of molecularly targeted therapeutics in treating cancers where oncogenic pathways are well defined offers promise for overcoming the resistance to conventional chemotherapy displayed by RCC. Our results reveal that c-Met/β-catenin signaling may contribute to an invasive tumor phenotype in VHL-negative clear cell RCC and to cystogenesis in VHL syndrome, a precursor of RCC in those patients. They also provide a rationale for the assessment of agents that target this pathway for efficacy against RCC.

Materials and Methods

Cell Culture.

Human RPTEC were obtained from Cambrex (Walkersville, MD) and maintained according the manufacturer's instructions. The RCC cell line 786-0, possessing a germ-line VHL frameshift mutation at codon 104, was obtained from American Type Culture Collection (Manassas, VA). UOK121, in which the VHL gene is suppressed by DNA hypermethylation, and UOK161, possessing a missense mutation at VHL codon 703, were derived from human RCC tumors at the Urologic Oncology Branch, National Cancer Institute. The cell lines 786-0/VHL and UOK121/VHL were generated by using the pQCXIX retroviral expression plasmid encoding the WT human VHL gene as described (57). All media and supplements were obtained from Life Technologies (Carlsbad, CA). Recombinant human HGF protein was obtained from R & D Systems (Minneapolis, MN). PS341 (bortezomib) was a gift of Ed Mimnaugh (Urologic Oncology Branch, National Cancer Institute). 786-0/VHL cells were stably transfected with an expression plasmid encoding the β-catenin S37A mutant (gift of Stephen Byers, Georgetown University, Washington, DC) by using FuGENE 6 reagent (Roche Molecular Biochemicals, Branchburg, NJ) according to the manufacturers' instructions and were maintained in G418 (Life Technologies). VHL gene silencing in the human kidney cell line HEK 293 was achieved by using a short hairpin RNA expression plasmid targeting VHL as described in Supporting Materials and Methods, which is published as supporting information on the PNAS web site.

Invasion and Morphogenesis Assays.

Cord formation or branching morphogenesis was examined in 3D cultures of RPTEC, 786-0, 786-0/VHL, 786-0/VHL/S37A, HEK 293/-VHL, HEK 293/Scramble, and HEK 293/-VHL/DN TCF cells in the presence or absence of HGF by using growth factor-reduced Matrigel (BD Biosciences, Bedford, MA) as described (7). Representative wells from duplicate samples were photographed after 72 h by phase contrast light microscopy with a ×10 objective. For invasion assays, 786-0, 786-0/VHL, and 786-0/VHL-S37A were plated in BD Biocoat Growth Factor-Reduced Matrigel Invasion Chambers (BD Biosciences) as previously described (7), and either HGF or FBS was added in the lower chamber. After 30 h cells that had migrated to the bottom surface of the membrane were fixed in situ, stained with DiffQuick Stain Kit (Dade Behring, Newark, DE), and photographed by bright-field microscopy with a ×4 objective.

Immunoprecipitation, Immunoblotting, and Pull-Down Assays.

Analysis of c-Met and β-catenin by immunoprecipitation and immunoblotting was performed as described (58). Briefly, subconfluent cells were serum-deprived for 16 h and treated with HGF for 30 min at 37°C or left untreated. Immunoprecipitates from cleared lysates obtained by using anti-Met or anti-β-catenin antibodies (Santa Cruz Biotechnology, Santa Cruz, CA) were fractionated by SDS/PAGE (PAGE Gel, San Diego, CA), transferred to PVDF membranes, and immunoblotted with anti-Met, anti β-catenin (BD Transduction Laboratories, Franklin Lakes, NJ), or anti-pY (Upstate Biotechnology, Lake Placid, NY) as indicated. Cells were treated with PS341 (50 nM) for 16 h before addition of HGF; cell lysates for immunoprecipitation and immunoblotting were then prepared as described above. Samples for coimmunoprecipitation experiments were prepared by using anti-VHL (BD Transduction Laboratories) and immunoblotting with anti-β-catenin or anti-VHL (Santa Cruz Biotechnology). β-Catenin stabilization was analyzed by pull-down assay using a recombinant GST–E-cadherin fusion protein described previously (30). Cells were treated with HGF for 16 h or left untreated. GST–E-cadherin reagent was added to equal amounts of 786-0, 786-0/VHL, or 786-0/VHL/S37A cell lysates, followed by overnight incubation at 4°C. Protein complexes were captured by using glutathione agarose beads (Santa Cruz Biotechnology) and analyzed by SDS/PAGE and immunoblotting with anti-β-catenin. Whole-cell lysates were immunoblotted similarly for comparison.

Immunofluorescence Microscopy.

786-0 and 786-0/VHL cells cultured on glass coverslips were stimulated with HGF for 16 h or left untreated, then were fixed and permeabilized by using standard procedures. Nonspecific antibody binding was blocked with 1% BSA/PBS for 1 h. Coverslips were incubated with anti-β-catenin (Santa Cruz Biotechnology) diluted in 1% BSA/PBS overnight at 4°C, washed with PBS, and then incubated with Texas red-labeled secondary antibody (Molecular Probes, Eugene, OR). After three PBS washes, nuclei were stained with DAPI (Molecular Probes). Epifluorescence photomicrographs were acquired with a CCD camera using IPLab software (Arlington, VA).

RNA Isolation, Reverse Transcription, and Quantitative Real-Time PCR.

These procedures are described in Supporting Materials and Methods.

Luciferase Reporter Gene Transactivation Assay.

The luciferase reporter constructs TOP-FLASH and FOP-FLASH (Upstate Biotechnology) were used to evaluate Tcf/LEF transcriptional activity. HEK 293 and HEK 293/-VHL cells were transiently transfected in triplicate with each of these reporters by using FuGENE 6 (Roche Molecular Biochemicals). Cells were treated with HGF for 16 h or left untreated. After 24 h, luciferase activity was measured by using the Bright-Glo Luciferase Assay (Promega, Madison, WI) and a Victor3 plate reader (PerkinElmer, Wellesley, MA). Resulting mean values from triplicate samples were normalized to sample protein content.

Abbreviations

HGF

hepatocyte growth factor

RCC

renal cell carcinoma

HIF

hypoxia-inducible factor

VHL

von Hippel–Lindau

pY

phosphotyrosine

RPTEC

renal proximal tubule epithelial cell

PKD

polycystic kidney disease

APC

adenomatous polyposis coli.