Tumor suppression by the von Hippel-Lindau protein requires phosphorylation of the acidic domain.

The tumor suppressor function of the von Hippel-Lindau protein (pVHL) has previously been linked to its role in regulating hypoxia-inducible factor levels. However, VHL gene mutations suggest a hypoxia-inducible factor-independent function for the N-terminal acidic domain in tumor suppression. Here, we report that phosphorylation of the N-terminal acidic domain of pVHL by casein kinase-2 is essential for its tumor suppressor function. This post-translational modification did not affect the levels of hypoxia-inducible factor; however, it did change the binding of pVHL to another known binding partner, fibronectin. Cells expressing phospho-defective mutants caused improper fibronectin matrix deposition and demonstrated retarded tumor formation in mice. We propose that phosphorylation of the acidic domain plays a role in the regulation of proper fibronectin matrix deposition and that this may be relevant for the development of VHL-associated malignancies.

In some anatomical venues, neoplastic transformation occurs upon biallelic inactivation of the von Hippel-Lindau (VHL) 1 gene (1). In their lifetime, patients carrying a mutated VHL gene are predisposed to multiple tumors in a number of organs, including the retina, cerebellum, spinal cord, adrenal gland, and kidney. Despite this phenotypic heterogeneity, VHL patients can be segregated based on their likelihood of developing pheochromocytoma. Type (2,3).
pVHL forms a multiprotein complex (VEC) with elongin C, elongin B, cul2, and Rbx-1 (reviewed in Ref. 4). VEC functions as a ubiquitin-activating enzyme (E3) to target the ␣ subunit of a heterodimeric transcription factor, hypoxia-inducible factor (HIF), for polyubiquitination (5)(6)(7). Polyubiquitin-tagged HIF-␣ is subsequently degraded by the 26 S proteasome. The VHL protein (pVHL) functions as the target recognition moiety of the VEC complex and specifically recognizes prolyl-hydroxylated HIF-␣ subunits (8,9). This post-translational modification of HIF-␣ requires oxoglutarate, iron, and oxygen (8 -10). Thus, ubiquitin-mediated destruction of HIF-␣ occurs selectively under normoxic conditions. Accordingly, biallelic inactivation of the VHL gene through mutation, deletion, or promoter silencing generally leads to the stabilization of HIF-␣ and thereby promotes the up-regulation of numerous HIF target genes, resulting in an inappropriate triggering of the hypoxic response under normal oxygen tension (4). However, type 2C patient mutations do not affect the E3 function of pVHL, and hence, HIF regulation is normal; yet these patients nevertheless develop pheochromocytoma (3). Therefore, we hypothesized that another function of pVHL is essential for its tumor suppressor activity.
The VHL gene produces two wild-type isoforms: a full-length 30-kDa form (pVHL30) and a shorter 19-kDa form (pVHL19) that is generated by alternative translational initiation at Met 54 (11)(12)(13). Both pVHL30 and pVHL19 form an active VEC complex, and both proteins have been shown to suppress tumor formation independently in nude mouse xenograft assays (12)(13)(14). However, analyzing known mutations in VHL patients revealed a number of disease-associated mutations in the acidic domain that would not affect the transcription and function of the pVHL19 isoform (VHL Mutations Database, available at www.umd.be:2020/). Hence, although the functional significance of the extra N-terminal acidic domain of pVHL30 remains unclear, this region in pVHL clearly contributes to its tumor suppressor function.
Here, we show that pVHL30 is phosphorylated within the N-terminal acidic domain by casein kinase-2 (CK2). The phos-phorylation status of pVHL30 did not affect E3 function. Ablation of CK2-specific phosphorylation sites or inhibition of CK2 activity increased affinity for fibronectin, yet resulted in decreased deposition of extracellular fibronectin. Interestingly, cells expressing phospho-defective pVHL delayed the onset of tumor formation in a severe combined immunodeficiency (SCID) mouse xenograft assay. Hence, CK2-mediated phosphorylation represents an HIF-independent tumor suppressor function of the acidic domain of pVHL.
Immunoprecipitation and Immunoblotting-For immunoprecipitations, protein A/G-agarose beads (Santa Cruz Biotechnology, Inc.) and antibody were pre-coupled for 30 min and washed twice with Triton X-100 lysis buffer (20 mM Tris (pH 8.0), 140 mM NaCl, 10% glycerol, and 1% Triton X-100). Samples were added to the pre-coupled beads, and binding was allowed to occur for 90 min at 4°C. The samples were then washed four times with 800 l of Triton X-100 lysis buffer, submitted to SDS-PAGE, and electrotransferred to a polyvinylidene difluoride membrane (Bio-Rad, Mississauga, Ontario, Canada). Specific protein bands on Western blots were visualized using the various indicated antibodies. All primary antibodies were diluted in phosphate-buffered saline (PBS) containing 5% nonfat dry milk and 0.1% Tween 20. Horseradish peroxidase-conjugated goat anti-mouse or goat anti-rabbit antibody (Pierce) was used at 1:20,000 dilution as a secondary antibody, after which enhanced chemiluminescence (PerkinElmer Life Sciences) was performed for detection.
[ 32 P]Orthophosphate Labeling-786-WT or 293T cells were seeded in 6-well plates and grown to 80% confluency. Cells were washed once with phosphate-free medium (ICN Biochemicals, Irvine, CA), followed by growth in phosphate-free medium supplemented with 200 Ci of [ 32 P]orthophosphate and labeling for 4 h. Cells were transferred to ice, washed with ice-cold PBS, and lysed with 500 l of Triton X-100 lysis buffer containing protease inhibitor mixture (Roche Applied Science) and phosphatase inhibitor mixture (0.1 mg/ml Na 3 VO 4 , 0.1 mg/ml ␤-glycerophosphate, and 1 mg/ml Na 4 P 2 O 7 ). These lysates were then submitted to various immunoprecipitations.
In Vitro Kinase Assay-Plasmids encoding wild-type or mutant pVHL were in vitro transcribed and translated using the TNT quickcoupled transcription/translation system (Promega Corp., Madison, WI). 5 units of recombinant CK2␣ (Promega Corp.) with the recommended buffer and 1 l of in vitro translated protein were mixed in a reaction volume of 50 l with 10 Ci of [␥-32 P]ATP (PerkinElmer Life Sciences) for 30 min at 30°C. In vitro translated products were immunoprecipitated to increase the specificity of the signal. The CK2 inhibitors daidzein (Sigma) and 4,5,6,7-tetrabromobenzotriazole (a kind gift from D. Shugar) were used at the indicated concentrations during the incubation with [␥-32 P]ATP.
Gel Filtration Assay-The gel filtration assay was done using a column containing a Bio-Gel A-1.5m (Bio-Rad). The column was equilibrated with PBS (pH 7.4); 200 l of crude cell lysate in Triton X-100 lysis buffer was run at 0.5 ml/min; and 0.5-ml samples were collected. The bed volume of the column used was 25 ml. Fractions were collected and subsequently submitted to trichloroacetic acid precipitation. The precipitated proteins were dissolved in 50 l of 1ϫ sample buffer and used for Western blotting.
Metabolic Labeling-Metabolic labeling was performed as described previously (16). In brief, radioisotope labeling was performed by methi-FIG. 1. pVHL30, but not pVHL19, is phosphorylated. A, 293T cells were transfected with VSV-pVHL30 or VSV-pVHL19 and metabolically labeled with [ 32 P]orthophosphate. Cells were lysed, immunoprecipitated (IP) with anti-receptor protein-tyrosine phosphatasemonoclonal antibody (mAb) (␣-RPTP) as a negative control (left upper panels) or anti-VSV mAb (right upper panels), and immunoblotted (IB) with anti-VSV mAb. Autoradiography (AR) of the immunoblot demonstrated 32 P incorporation (lower panels). B, 293T cells were transfected with HA-pVHL30(WT) or HA-pVHL30(AAA) and metabolically labeled with [ 32 P]orthophosphate. Cells were lysed, immunoprecipitated with anti-receptor protein-tyrosine phosphatase-mAb as a negative control (upper left panel) or anti-HA mAb (upper right panel), and immunoblotted with anti-HA mAb. Autoradiography of the immunoblot demonstrated 32 P incorporation (lower panels). C, 293T cells were metabolically labeled with [ 32 P]orthophosphate. Cells were lysed and immunoprecipitated with anti-HA mAb as a negative control or anti-pVHL mAb, and bound proteins were separated by SDS-PAGE and submitted to autoradiography. onine starvation for 45 min, followed by growth in 5 ml of methioninefree Dulbecco's modified Eagle's medium supplemented with [ 35 S]methionine (100 Ci/ml of medium; Amersham Biosciences, Buckinghamshire, United Kingdom) and 2% dialyzed fetal bovine serum for 3 h at 37°C in a humidified 5% CO 2 atmosphere.
Fibronectin Enzyme-linked Immunosorbent Assay-A fibronectin enzyme-linked immunosorbent assay was performed as described previously (2,16). In brief, cells (2 ϫ 10 3 ) were grown for 4 days in 96-well microtiter plates (Costar, Cambridge, MA) prior to cell removal with PBS containing 2 mM EDTA. Complete removal of cells was visualized by phase microscopy. Plates were blocked with PBS containing 0.1% heat-inactivated bovine serum albumin (Invitrogen) for 1 h at 37°C and subsequently incubated with 10 g/ml rabbit anti-fibronectin antibody in PBS containing 0.1% bovine serum albumin for 1 h. Plates were washed three times with PBS, and alkaline phosphatase-conjugated goat anti-rabbit IgG (0.1 g/ml) was added for 1 h at 37°C. Plates were washed three times with PBS and developed with 1.0 mg/ml p-nitrophenyl phosphate in 0.2 M Tris buffer (Sigma). Reactions were stopped with 3 M NaOH, and the absorbance at 405 nm was measured using a microplate reader.
SCID Mouse Xenograft Assay-Multiple 786-O subclones expressing pVHL(WT), pVHL(AAA), or empty plasmid were grown to ϳ90% confluence in a humidified 5% CO 2 atmosphere at 37°C. Cells were harvested with a solution of 0.25% trypsin and 1 mM EDTA. Cells (2 ϫ 10 6 ) in 50 l of 1ϫ PBS were injected intramuscularly into the left hind legs of male SCID mice (Charles River Laboratories, Inc.). Tumor growth was assessed and measured weekly by carefully passing the tumorbearing leg through a series of holes of decreasing diameter (0.5-mm decrements) in a plastic rod.

RESULTS
The Acidic Domain of pVHL30 Is Phosphorylated by CK2-To determine whether the two naturally occurring pVHL isoforms are phosphorylated, 293T cells were transfected with VSVtagged pVHL30 and pVHL19 and metabolically labeled with [ 32 P]orthophosphate. Cells were then lysed, immunoprecipitated with anti-VSV antibody, resolved by SDS-PAGE, and immunoblotted with anti-VSV antibody (Fig. 1A). Autoradiography of the immunoblot demonstrated that, although VSV-pVHL30 incorporated [ 32 P]orthophosphate, VSV-pVHL19 did not (Fig. 1A). Computer-based scanning of full-length pVHL30 indicated multiple potential phosphorylation sites over the entire open reading frame and three predicted CK2 phosphorylation sites in the acidic domain at Ser 33 , Ser 38 , and Ser 43 (data not shown). Thus, we generated the triple mutant pVHL30(AAA) with Ser-to-Ala substitutions at each of these positions and found that pVHL30(AAA) was no longer capable of incorporating [ 32 P]orthophosphate (Fig. 1B). These results suggest that, in an overexpression system, pVHL30, but not pVHL19, is phosphorylated at one or more of the serine residues within the acidic domain. To determine the physiologic relevance of this finding, we metabolically labeled 293T cells with [ 32 P]orthophosphate. Cells were then lysed, immunoprecipitated with anti-pVHL antibody, resolved by SDS-PAGE, and blotted (Fig. 1C). Autoradiog- E, 293T cells were lysed, and the lysate was submitted to gel filtration. Fractions of different sizes were collected, and after trichloroacetic acid precipitation, the proteins were submitted to Western blot analysis using anti-CK2␣ or anti-pVHL antibody. F, 293T cells were transfected with HA-pVHL30(WT). In vitro kinase assays of anti-HA immunoprecipitations were performed without or with increasing concentrations of the indicated CK2-specific inhibitors (upper panels). The anti-HA immunoblot indicates equal loading (lower panels). TBB, 4,5,6,7-tetrabromobenzotriazole. raphy of this blot clearly showed a specific band at ϳ30 kDa in the anti-pVHL immunoprecipitation, but not in the anti-HA control immunoprecipitation. From these data, we concluded that phosphorylation of pVHL30, as observed, occurs endogenously.
To address directly whether CK2 is responsible for phosphorylation of pVHL30 at the predicted serine residues, we performed an in vitro kinase assay using in vitro translated VSV-pVHL30(WT) and HA-pVHL30(AAA) with or without recombinant CK2␣, the catalytic subunit of CK2 ( Fig. 2A). Autoradiography of the in vitro kinase assay clearly demonstrated that pVHL30(WT), but not pVHL30(AAA), was phosphorylated by CK2␣ ( Fig. 2A). Casein was used as a positive control for the CK2-mediated in vitro kinase assays ( Fig. 2A). To determine the exact site(s) of pVHL30 phosphorylation, single and double Ser-to-Ala mutants were generated and analyzed using the in vitro kinase assay (Fig. 2B). All in vitro translated single Serto-Ala mutants were phosphorylated by recombinant CK2␣. It should be noted, however, that pVHL30(S38A) had a noticeably weaker phosphorylation signal. All double Ser-to-Ala mutants showed significantly weaker phosphorylation profiles compared with their wild-type counterpart (Fig. 2B). These data suggest that all three serine residues can serve as CK2-mediated phosphorylation sites and that phosphorylation of any one site appears to be temporally and spatially independent of the phosphorylation status of the other two sites.
During the course of these in vitro kinase assays, we observed that, even without adding exogenous CK2␣, anti-pVHL30(WT) immunoprecipitates derived from 293T cells exhibited endogenous kinase activity capable of phosphorylating casein (data not shown). Likewise, immunoprecipitates of pVHL30(AAA), which themselves could not be phosphorylated (Figs. 1A and 2, A and B), were also capable of phosphorylating purified casein (data not shown). These data suggest that CK2 exists in a complex with pVHL30 irrespective of its phosphorylation status. To further verify that the pVHL30-associated kinase is CK2␣, 786-O RCC subclones stably expressing HA-pVHL30(WT) (786-WT cells) were lysed and immunoprecipitated with anti-pVHL, anti-CK2␣, or anti-receptor protein-tyrosine phosphatase-(used as a negative control) antibody (Fig. 2C). Bound proteins were resolved by SDS-PAGE and immunoblotted with anti-CK2␣ or anti-pVHL antibody, which showed that CK2␣ associated with pVHL30 (Fig. 2C). To determine whether this interaction occurs in the absence of overexpression, we performed immunoprecipitation using anti-CK2␣, anti-pVHL, or anti-receptor protein-tyrosine phosphatase-(used as a negative control) antibody; an in vitro kinase assay; SDS-PAGE; and visualization by autoradiography. Kinase activity was detected in immunoprecipitates of pVHL and CK2␣, which showed phosphorylation of endogenous pVHL30 (Fig. 2D, first and second lanes). We also observed phosphorylated proteins comigrating with CK2␣ (Fig. 2D), suggesting autophosphorylation of CK2␣. These results suggest that pVHL30 and CK2␣ interact under a physiologic condition. Furthermore, we performed a size gel filtration assay to separate endogenous multiprotein complexes from 293T lysate according to their size. After performing the gel filtration according to size, proteins were concentrated by trichloroacetic acid precipitation and immunoblotted for pVHL and CK2␣. In the fraction containing pVHL30, we did find CK2␣ (Fig. 2E), again supporting the notion that pVHL and CK2␣ are present in the same protein complex.
VEC Complex Formation and Function Are Independent of pVHL30 Phosphorylation-To address whether the pVHL30-(AAA) mutant is capable of forming an E3 complex, we tested the ability of pVHL30(AAA) to bind to Cul2, a component of the VEC complex that acts as a scaffold. 786-WT, 786-AAA, or 786-MOCK cells were lysed, immunoprecipitated with anti-HA antibody, resolved by SDS-PAGE, and immunoblotted with anti-Cul2 or anti-HA antibody (Fig. 3A). This assay showed that pVHL30(AAA) bound Cul2 similarly to pVHL30(WT). Furthermore, immunoprecipitation of HA-pVHL30(AAA) from metabolically labeled 786-AAA cells coprecipitated elongin B, elongin C, and Cul2 (Fig. 4A). These results demonstrate that pVHL30(AAA) forms a VEC complex.
We next asked whether pVHL30(AAA) can ubiquitinate HIF- 1␣(ODD) by performing an in vitro ubiquitination assay. S100 extracts prepared from 786-MOCK cells lacking pVHL and supplemented with in vitro translated empty plasmid failed to support the ubiquitination of 35 S-labeled HA-HIF-1␣(ODD) (Fig. 3B, first lane). However, S100 extracts supplemented with increasing amounts of either in vitro translated HA-pVHL30(WT) (Fig. 3B, second and third lanes) or in vitro translated HA-pVHL30(AAA) (fourth fifth and lanes) ubiquitinated 35 S-labeled HA-HIF-1␣(ODD). As expected, S100 extracts made from 786-WT cells supported the ubiquitination of 35 Slabeled HA-HIF-1␣(ODD) (Fig. 3B, sixth lane). Furthermore, whole cell extract made from 786-AAA cells down-regulated the expression of the HIF target gene GLUT1 under normal oxygen tension similarly to 786-WT cells (Fig. 3C). Whole cell extract made from 786-MOCK cells showed overexpression of GLUT1 under normal oxygen tension. These results indicate that phosphorylation of pVHL30 by CK2 is not required for participation in the VEC complex to ubiquitinate HIF-␣.
Phosphorylation of pVHL30 Regulates Binding of Fibronectin and Secretion into Extracellular Space-Interestingly, HA-pVHL30(AAA) bound strikingly more fibronectin (ϳ3-fold, as measured by densitometry) compared with HA-pVHL30(WT) (Fig. 4A). Furthermore, the addition of increasing amounts of purified recombinant fibronectin during lysis of radiolabeled cells resulted in greater reduction of fibronectin bound to HA-pVHL30(WT) compared with HA-pVHL30(AAA) (Fig. 4, B and  C). For example, whereas we observed no appreciable effect on HA-pVHL30(AAA) binding to fibronectin, HA-pVHL30(WT) binding to fibronectin was reduced by 40% upon the addition of 0.1 g of purified fibronectin, as measured by densitometry (Fig. 4C). The addition of 1 g of purified fibronectin almost completely competed against the binding of endogenous fibronectin to HA-pVHL30(WT), but negligibly affected the binding of endogenous fibronectin to non-phosphorylatable HA-pVHL30(AAA). Collectively, these data imply that dephosphorylation of pVHL30 enhances its ability to bind fibronectin or reduces its ability to release fibronectin.
We next asked whether inhibition of CK2 activity would increase the binding of pVHL30(WT) to fibronectin. During metabolic labeling, 786-WT cells were treated with increasing concentrations of the CK2-specific inhibitor daidzein. Cells were then lysed, immunoprecipitated with anti-HA antibody, resolved by SDS-PAGE, and visualized by autoradiography. As predicted, inhibition of CK2 by daidzein increased pVHL30(WT) binding to fibronectin (Fig. 4B). This effect was not due simply to increased expression of pVHL30(WT) in daidzein-treated cells, as noted by the relatively equal loading of de novo translated pVHL30(WT) (Fig. 4B). Treatment with staurosporine (a general inhibitor of serine/threonine kinases, excluding CK2) had a negligible effect on fibronectin binding (data not shown). These results suggest that specific inhibition of CK2 enhances pVHL30 binding to fibronectin.
We then asked what effect pVHL30 binding to fibronectin has on subsequent deposition of fibronectin into the extracellular space. The relative amounts of fibronectin deposited by 786-MOCK, 786-WT, and 786-AAA cells grown on 96-well plates for 4 days were measured by enzyme-linked immunosorbent assay using anti-fibronectin antibody after the cells were removed by trypsin-free EDTA solution to preserve any fibronectin that was deposited on the plastic. As expected, 786-MOCK cells deposited significantly less fibronectin compared with 786-WT cells. However, the enhanced fibronectin binding by pVHL(AAA) resulted in less fibronectin deposition compared with 786-WT cells (Fig. 5A). Anti-fibronectin immunofluorescence by confocal microscopy demonstrated the expected fibronectin staining in the cytoplasm and along the cell periphery in 786-WT cells (Fig. 5B). However, 786-AAA cells showed excessive perinuclear fibronectin staining as well as punctate staining along the cell cortex (Fig. 5B). Concordantly, 786-AAA cells showed more intracellular fibronectin expression compared with 786-WT cells (Fig. 5C). 786-WT and 786-AAA cells showed similar de novo translation of fibronectin as measured by 35 S metabolic labeling (data not shown). These data suggest that, although 786-AAA cells are capable of producing fibronectin, they are less capable of releasing fibronectin into the extracellular space, which would account for the increased intracellular accumulation of fibronectin.
Phosphorylation of pVHL30 Is Required for Tumor Suppression in SCID Mice-To determine whether phosphorylation plays a role in tumor suppression, multiple 786-O subclones expressing pVHL(WT), pVHL(AAA), or empty plasmid were injected intramuscularly into the left hind legs of SCID mice. Tumor take and size were monitored and measured weekly (Fig. 6). After ϳ3-4 weeks, the mice injected with 786-MOCK cells began to form tumors (15/15), as expected. The mice injected with 786-AAA cells began to develop tumors after 7-8 weeks (15/15), whereas the mice injected with multiple 786-WT subclones were tumor-free (0/14). These results demonstrate that phosphorylation of pVHL30 within the first 53 amino acids have a direct role in tumor suppression. DISCUSSION Not all VHL disease-causing mutations result in the dysregulation of HIF activity (2,3,17). For example, type 2Cassociated mutations such as L188V and K159E have been shown to have wild-type or "normal" E3 activity to target HIF-␣ subunits for oxygen-dependent polyubiquitination. However, every VHL disease-causing mutant tested to date has shown a failure to bind fibronectin, resulting in reduction of extracellular fibronectin matrix assembly (2,3,17). Here, we have shown that inhibition of pVHL phosphorylation, either by Ser-to-Ala substitutions or by treatment with CK2-specific inhibitors, resulted in markedly increased binding of pVHL30 to fibronectin while maintaining wild-type E3 function. However, this increased binding led to reduced fibronectin deposition into the extracellular space. The improper deposition of fibronectin into the extracellular space or interference with its proper function has been correlated with the development of dysplasia in Xenopus, and fibronectin has been shown to influence the malignant behavior of tumor cells in experimental mouse models (18,19). Thus, we postulate that the tumor suppressor function of pVHL30 phosphorylation is related to the regulation of proper fibronectin deposition.
The tumor suppressor function of pVHL that we have described here requires the N-terminal acidic domain and is independent of the regulation of HIF-␣. Multiple studies on the role of stabilized HIF in tumorigenesis using the identical cell system that we have described here have been recently published. Collectively, these previous studies suggest that HIF-2␣ is the effector of tumorigenesis upon loss of functional pVHL. Kondo et al. (20) and Maranchie et al. (21) addressed whether overexpressed stabilized HIF can counteract tumor suppression by pVHL. Kondo et al. (22) have shown that 786-O cells expressing stable HIF-2␣ are able to override pVHL-mediated tumor suppression. However, these previous studies did not address the more physiologically relevant question of whether the suppression of endogenous HIF-2␣ would be sufficient to abrogate tumor development in these patient-derived RCC cells (786-O RCC). Kondo et al. (22) and Zimmer et al. (23) did address this question using HIF-2␣ RNA interference; their data indeed support an oncogenic role for HIF-2␣. However, these experimental data do not exclude further tumor suppressor functions for pVHL. The in vivo experiments described in these previous studies were terminated after 8 -10 weeks, whereas our experiments demonstrated that this time point is only when 786-O cells deficient in CK2-mediated phosphorylation begin to form detectable tumors (Fig. 6). Moreover, tumor growth in the absence of HIF-2␣ is retarded, but not completely inhibited (22). On the basis of these findings and our data, we propose that, although HIF-2␣ is an important regulator of tumor growth, it is not the only effector of tumorigenesis in pVHL-deficient cancers. These data led us to hypothesize that an additional tumor suppressor function is linked to phosphorylation of the acidic domain of pVHL.
The regulation of fibronectin by pVHL is tightly linked to the development of VHL disease. It follows then that there are multiple levels of fibronectin regulation by pVHL30. First, pVHL30 positively regulates transcription of the gene encoding fibronectin (FN1) in an HIF-independent manner (24). Second, as shown here, phosphorylation of the N-terminal acidic domain of pVHL30 mediates the engagement and/or disengagement of fibronectin. Third, modification of pVHL30 by the ubiquitin-like molecule NEDD8 is linked to fibronectin binding, and inhibition of pVHL30 neddylation prohibits fibronectin binding (18). Thus, fibronectin deposition by pVHL seems to be a highly regulated process and is unique to the pVHL30 isoform, as pVHL19 does not bind fibronectin (11).
We have shown here that CK2 exists in a complex with pVHL30 and phosphorylates Ser 33 , Ser 38 , and Ser 43 in the N-terminal acidic domain of pVHL30. CK2 is a heterotetrameric complex composed of two ␤ and two ␣ subunits (25). The two ␤ subunits have no kinase activity, but are thought to have a regulatory role enhancing the catalytic activity of the ␣ subunits (25). Differential gene expression profiling of chemically induced kidney carcinoma cells of Eker rats demonstrated overexpression of CK2␤ subunits, indicating that amplification of CK2␣ function may be involved in renal carcinogenesis (26). In support of this notion, human kidney tumors have been shown to express elevated levels of CK2␣ and CK2␤ (27). Moreover, CK2 localizes and binds to microtubules and has been shown to phosphorylate various microtubule-binding proteins (28 -31). Recently, pVHL has been shown to associate with microtubules (32), which raises the possibility that pVHL30 and CK2 interact on microtubules. Taken together, our data present a novel tumor suppressor function for the acidic domain of pVHL, which might help to elucidate the complex genotype-phenotype correlation in VHL disease.