The von Hippel-Lindau protein interacts with heteronuclear ribonucleoprotein a2 and regulates its expression.

The product of the von Hippel-Lindau (VHL) tumor suppressor gene, pVHL, functions as a ubiquitin-protein isopeptide ligase in regulating HIF-1 protein turnover, thus accounting for the increased transcription of hypoxia-inducible genes that accompanies VHL mutations. The increased vascular endothelial growth factor mRNA stability in cells lacking pVHL has been hypothesized to be due to a similar regulation of an RNA-binding protein. We report the expression of the GLUT-1 3'-untranslated region RNA-binding protein, heteronuclear ribonucleoprotein (hnRNP) A2, is specifically increased in pVHL-deficient cell lines. Enhanced hnRNP A2 expression was apparent in all cell fractions, including polysomes, where a similar modest effect on hnRNP L (a GLUT-1 and VEGF 3'-untranslated region-binding protein), was seen. Steady state levels of hnRNP A2 mRNA were unaffected. Regulation of hnRNP A2 levels correlated with the ability of pVHL to bind elongin C. Proteasome inhibition of cells expressing wild type pVHL selectively increased cytoplasmic hnRNP A2 levels to that seen in pVHL-deficient cells. Finally, an in vivo interaction between pVHL and hnRNP A2 was demonstrated in both the nucleus and the cytoplasm. Collectively, these data indicate that hnRNP A2 expression is regulated by pVHL in a manner that is dependent on elongin C interactions as well as functioning proteasomes.

Von Hippel-Lindau (VHL) 1 disease is an autosomal dominant cancer syndrome characterized by the predisposition to develop highly vascular tumors, including renal clear cell (RCC) carcinomas, cerebellar hemangioblastomas, retinal angiomata, and pheochromocytomas (1). Both germline as well as sporadic mutations of the VHL gene have been identified in patients afflicted with this disease (2)(3)(4)(5)(6)(7). Sporadic renal clear cell carcinomas are highly associated with mutation or transcriptional silencing of the VHL gene and subsequent loss or inactivation of the remaining VHL allele (6,7). Thus, VHL conforms to Knudson's two-hit model of a tumor suppressor gene, in which gene inactivation occurs as the result of loss of function of both alleles (8).
The human VHL gene encodes a full-length protein of 213 amino acids which migrates with an apparent molecular mass of 30 kDa (9,10). Internal translational initiation from an internal ATG at codon 54 produces a second 18-kDa gene product (11)(12)(13). Both isoforms behave identically in all reports to date. VHL protein is predominantly expressed in the cytoplasmic compartments of most tissue and cell types, although it can shuttle between the nucleus and cytoplasm (10, 14 -18). The localization of pVHL appears to be regulated according to cell density; pVHL is cytoplasmic in confluent cultures, but shuttles to the nucleus under sparse culture conditions (15). Nuclear export of pVHL is reduced by inhibitors of RNA polymerase II and polyadenylation, whereas nuclear import is unaffected, resulting in the localization of pVHL to the nucleus (16,19).
VHL has been demonstrated to form a multimeric complex with two components of the transcriptional elongation factor elongin (elongin B and C) (11,12,14,20), as well as cullin-2 (21,22) and Rbx1 (23). Significantly, the majority of VHL mutants are defective in their ability to bind elongin C, implying functional significance for this interaction in vivo (14,20,24,25). Additional clues to VHL function were provided by the discovery that elongin C and cullin-2 bear homology to yeast proteins (Skp1 and Cdc53), which function as a ubiquitin E3 ligase when complexed with an F-box protein (26 -29). Furthermore, anti-pVHL immunoprecipitates can support E3 ubiquitin ligase activity in vitro if supplemented with exogenous ubiquitin-conjugating enzymes (30,31). As a consequence, a model has evolved in which the pVHL-elongin B/C-Cul-2 complex functions as a ubiquitin E3 ligase, which targets specific substrates for ubiquitin-mediated proteasomal degradation (32).
The absence of pVHL in renal carcinoma cell lines is associated with a hypoxic phenotype under normoxic culture conditions; these cells express increased levels of vascular endothelial growth factor (VEGF), platelet-derived growth factor, and glucose transporter 1 (GLUT-1) (33,34). The overproduction of these genes likely contributes to the hypervascular phenotype characteristic of VHL disease-associated neoplasms (35). The ubiquitin E3 ligase model of pVHL action accounts for the increased transcription of hypoxia-inducible genes (VEGF, platelet-derived growth factor), which occurs under normoxic conditions with pVHL mutations (33, 34, 36 -39). The transcription of these genes is regulated by the levels of the transcription factor hypoxia-inducible factor (HIF)-1 (Refs. 40 and 41, and references therein). Under conditions of normoxia, rapid ubiquitination and proteasomal-dependent degradation of HIF-1␣ is mediated by an oxygen-dependent degradation domain (42). Under hypoxic conditions, the stability of HIF-1␣ protein increases, resulting in enhanced transcription of hypoxia-inducible genes (43,44). Subsequent studies have shown that pVHL and HIF-1␣ directly interact under normoxic conditions, resulting in the ubiquitination and proteasomal targeting of HIF-1␣ by the pVHL-elongin B/C-Cul-2 complex (45,46). Thus, in the absence of pVHL or with mutations that alter its ability to function as part of a ubiquitin E3 ligase complex, HIF-1␣ levels rise, leading to increased transcription of hypoxia-inducible genes such as VEGF (44).
Increased GLUT-1 and VEGF mRNA stability has also been observed in cells under conditions of hypoxia (47). In cells lacking pVHL, VEGF mRNA stability has been demonstrated (33,34). It has been hypothesized that a similar mechanism accounts for the stabilization of GLUT-1 and VEGF mRNA observed in pVHL-deficient cells (45,47). In this model, an RNA-binding protein, instead of a transcription factor such as HIF-1␣, constitutes the target by which pVHL regulates the stability of these mRNA. The 3Ј-UTR of both GLUT-1 and VEGF have been shown to contain AU-rich elements (AURE), which regulate mRNA turnover (48).
Previous work in our laboratory indicated that hnRNP A2 binds a cis-acting instability element in the GLUT-1 3Ј-UTR that plays a role in the post-transcriptional regulation of GLUT-1 expression (49). These data suggested that overexpression of hnRNP A2 might account for the change in GLUT-1 mRNA turnover associated with the absence of pVHL. In RCC cell lines that differ only in their expression of functional pVHL, we observed that hnRNP A2 expression is increased in pVHL-deficient cell lines. This pVHL-dependent reduction in hnRNP A2 expression occurs independently of cell confluence. Northern blotting demonstrated that hnRNP A2 mRNA levels were unaffected, suggesting that pVHL deficiency results in decreased hnRNP A2 protein expression through changes in protein turnover or translation. Proteasome inhibition of cells expressing wild type pVHL resulted in cytoplasmic accumulation of hnRNP A2, but not the closely related and homologous protein hnRNP A1 (50). Finally, an in vivo interaction between pVHL and hnRNP A2 is demonstrated in both proteasomally inhibited cytosolic and untreated nuclear extracts. Collectively, these data indicate that hnRNP A2 expression is regulated by pVHL, and suggest the possibility that hnRNP A2 is a target of the pVHL/elongin B/C/Cul2/Rbx1 ubiquitin degradation machinery.
Preparation of Subcellular Fractions-Cytosolic lysates were prepared using a method characterized by its lack of contamination by nuclear proteins (51). Briefly, cells were washed twice in 1ϫ phosphatebuffered saline, then lysed by gentle resuspension in 1% Triton X-100 lysis buffer (50 L/2 ϫ 10 7 cells) consisting of 10 mM PIPES, pH 6.8, 100 mM KCl, 2.5 mM MgCl 2 , 300 mM sucrose, 1 mM Pefabloc, and 2 g/ml leupeptin and pepstatin A. Samples were incubated for 3 min on ice and then centrifuged for 5 min at 500 ϫ g. The supernatant was aliquoted and stored at Ϫ80°C as the cytosolic fraction. The pellet was resuspended in lysis buffer and then spun through a 30% sucrose cushion.
The nuclear pellet was resuspended in 0.5 nuclear pellet volume in low salt buffer consisting of 10 mM Tris-HCl, pH 7.6, 20 mM KCl, 1.5 mM MgCl 2 , 0.5 M dithiothreitol, 0.2 mM EDTA, 25% glycerol, 2 mM Pefabloc, 1 g/ml each leupeptin and pepstatin A. While vortexing, 1.5 nuclear pellet volume of high salt buffer (containing 0.5 M KCl) was added dropwise. Samples were incubated with agitation for 30 min at 4°C, and then centrifuged for 30 min at 12,000 ϫ g at 4°C (52). The supernatant was aliquoted and stored at Ϫ80°C as the nuclear fraction. For polysome preparations, pRc-9, C162F, and WT-8 cells were homogenized in buffer A (10 mM Tris-HCl, pH 7.6, 1 mM potassium acetate, 1.5 mM magnesium acetate, 2 mM dithiothreitol, 2 g/ml leupeptin and pepstatin A, and 2 mM Pefabloc), and nuclei removed by centrifugation. The supernatant was layered over a 30% sucrose cushion followed by ultracentrifugation at 36,000 ϫ rpm for 5 h at 4°C. Whole cell lysates were prepared by washing cells twice with 1ϫ phosphate-buffered saline, and then solubilizing in 2ϫ Laemmli SDS sample buffer (53).
Immunoprecipitation and Immunoblotting-Coimmunoprecipitation of pVHL with hnRNP A2 from WT-8 proteasomally inhibited cytoplasmic (100 g) or untreated nuclear (400 g) lysates was performed with a mouse monoclonal antibody directed against pVHL (1G32) (10). Immunocomplexes were captured with protein A-Sepharose beads (Amersham Pharmacin) for 2 h at 4°C, and beads were washed six times in 300 mM NaCl. Proteins were resolved by 15% SDS-PAGE and electrotransferred to nitrocellulose membrane in CAPS buffer, pH 11.0, with 15% methanol. Immunoblots were washed with Tris-buffered saline, 0.1% Tween 20, and blocked in 3% bovine serum albumin overnight at 4°C. Membranes were then probed with a mouse monoclonal antibody directed against hnRNP A2 (EF67) and goat anti-mouse HRP-conjugated secondary antibody (Bio-Rad). To demonstrate the reciprocal interaction, CNBr-hnRNP A2-conjugated beads were used to immunoprecipitate from WT-8 lysate, and then immunoblotted with 1G32, followed by a goat anti-mouse HRP-conjugated secondary antibody. CNBr beads were used in lieu of protein-A Sepharose in order to eliminate interfering signal generated by detection of the light chain of the secondary antibody. Depleted lysates represent supernatants of immunoprecipitations. Polysomal lysates were probed with a rabbit polyclonal anti-hnRNP A2 antibody (Act2), followed by goat anti-rabbit HRP-conjugated secondary antibody. Detection of hnRNP A1 protein was accomplished by probing with rabbit polyclonal anti-hnRNP A1 antibody (Act1), followed by goat anti-rabbit HRP-conjugated secondary antibody. As indicated, blots were probed with anti-GAPDH monoclonal antibody (6C5-American Research Products), followed by goat anti-mouse HRP-conjugated secondary antibody to control for loading. Reactive antigens were visualized with Supersignal chemiluminescence substrate (Pierce).
Densitometry-Protein and mRNA bands were quantified by densitometric scanning of autoradiographs and immunoblots using NIH Image software.

Increased hnRNP A2 Expression in pVHL-deficient RCC Cell
Lines-The mechanism by which pVHL deficiency increases GLUT-1 mRNA stability is unknown. It was hypothesized that the effect on GLUT-1 mRNA stability is mediated through a specific GLUT-1 mRNA-binding protein (24). Previous work identified hnRNP A2 as a GLUT-1 mRNA 3Ј-UTR-binding protein (49). For these reasons, we examined hnRNP A2 protein levels in three RCC cell lines, which differ only in their expression of VHL.
Cytoplasmic lysates derived from 786-0 RCC cells, which lack functional VHL protein, pRc-9 cells (786-0 cells trans-fected with empty expression vector), and WT-8 cells (transfected with wild type VHL) were examined for hnRNP A2 expression. Data shown are representative of seven experiments. Immunoblotting demonstrated increased hnRNP A2 levels in cells lacking functional VHL (786-0 and pRc-9) relative to those containing wild type VHL (WT-8) (Fig. 1A). Densitometric analysis of hnRNP A2 expression in these lysates demonstrated 3-5-fold elevated levels of this protein in cells lacking wild type pVHL. The change in cytosolic hnRNP A2 levels among the cell lines did not appear to be due to differences in subcellular distribution, as extracts directly lysed in SDS loading buffer (whole cell lysates) and boiled showed similar patterns of expression (Fig. 1B). The expression of hnRNP A1 and GAPDH was unaffected. This finding indicates that pVHL-dependent regulation of hnRNP A2 was highly specific; hnRNP A1 and hnRNP A2 have 70% homology at the amino acid level (50). Consistent with this interpretation, the expression of other AURE-binding proteins reportedly associated with mRNA turnover such as HuR and AUF1 was unaffected (data not shown). Nevertheless, to ensure that the observed hnRNP A2 overexpression observed in the 786-0 and pRc-9 cell lines was not cell line-specific, another set of RCC cell lines was tested (Fig. 1C). As seen above, UMRC cells transfected with pVHL (WT) exhibited decreased levels of hnRNP A2 relative to the pVHL-deficient parental cell line (PAR).
Increased Expression of hnRNP A2 in pVHL-deficient Cell Lines Occurs Independent of Confluence-Recent reports have described the influence of cell density on the ability of pVHL to mediate biochemical and morphological differentiation of renal carcinoma cell lines (56,57). In these studies, VHL expression promoted renal cell differentiation and growth arrest under conditions of high cell density, in association with down-regulation of integrins and up-regulation of hepatocyte nuclear factor 1␣, a global activator of proximal tubule-specific genes (56). To exclude differences in cell density and growth as possible explanations for the disparity in hnRNP A2 protein expression, all of the experiments delineated above were performed with 100% confluent cells. Subsequently, we examined cytosolic levels of hnRNP A2 at various states of confluence ( Fig. 2). As seen earlier, hnRNP A2 protein was expressed consistently higher in pRc-9 cytosol, independent of cell density. In neither cell line did hnRNP A2 levels vary as a function of cell confluence. Moreover, GAPDH levels are constant between the cell types, regardless of cell density. These data suggest the observed differences in hnRNP A2 protein expression were not a consequence of altered rates of proliferation or differentiation between the cell lines.
pVHL Expression Regulates Polysomal hnRNP A2 and L Levels-Some studies indicate that AURE-mediated mRNA turnover requires either polysomal loading or ribosomal transit (58). Since total cellular levels of hnRNP A2 were decreased by pVHL expression, we wanted to determine if polysomal expression reflected this as well. Sucrose density fractionation of the cytosol showed that polysomal hnRNP A2 levels were markedly decreased by the expression of wild type pVHL (Fig. 3). Data shown are representative of four separate polysome profiles. Previous work had demonstrated that hnRNP L binds the 3Ј-UTR of GLUT-1 at a site distinct from that bound by hnRNP A2 (49). As part of these studies, it was shown that polysomal hnRNP A2 associated with hnRNP L (49). For these reasons, we examined polysomal hnRNP L levels by immunoblotting (Fig. 3). Expression of pVHL was associated with a significant reduction in polysomal hnRNP L levels. Immunoblot analysis of GAPDH levels in pRc-9 and WT-8 cells demonstrated uniform expression of this protein (Fig. 3). In contrast to this polysomal fraction, no consistent pVHL-dependent effect on hnRNP L was seen in the cytosolic, nuclear, or whole cell fractions (data not shown). Interestingly, pVHL was detected in this polysomal fraction from WT-8 cells (Fig. 3).
pVHL Expression Does Not Alter hnRNP A2 mRNA Accumulation-TheVHL-dependent difference in hnRNP A2 protein expression among the 786-0, pRc-9 and WT-8 cell lines was examined by Northern blotting of total cellular mRNA. As demonstrated in Fig. 4, hnRNP A2 mRNA levels did not vary among the cell lines. Densitometric analysis of mRNA likewise failed to indicate differences in hnRNP A2 mRNA among the cell lines. Thus, the elevated expression of hnRNP A2 protein in cells lacking VHL is not due to increased transcription or stability of hnRNP A2 mRNA. This lack of change suggests these differences are due to changes in protein turnover or mRNA translation.
Elongin C Binding by pVHL Plays a Role in Regulating hnRP A2 Expression-The possibility that VHL might regulate hnRNP A2 protein turnover was an attractive hypothesis, given its identification as an E3 ligase (23,30,31). The enzymatic activity of VHL requires complex formation with elongins B, C, Cul2, and Rbx1 (23, 32, 59). We therefore assessed hnRNP A2 expression in whole cell lysates from a 786-0derived RCC cell line stably transfected with a VHL cDNA  Figure shows anti-hnRNP A2 immunoblot of pRC-9 and WT-8 cytosolic lysates. pRc-9 and WT-8 cells were grown to the indicated cell density, cytosols extracted, and protein (25 g) separated by 15% SDS-PAGE. Cytosols were immunoblotted for expression of hnRNP A2 with Act-2. Lysates were also blotted in parallel with anti-GAPDH monoclonal antibody to control for loading. encoding the C162F mutation (Fig. 5). This mutation has been shown to result in decreased elongin C interaction and continued GLUT-1 overexpression (21). Levels of hnRNP A2 were still reduced in C162F RCC cells relative to pRc-9 cells (Fig. 5). However, there was a small but consistent increase in hnRNP A2 expression accompanying C162F expression relative to WT-8 cells, suggesting that hnRNP A2 levels were modestly modulated by pVHL levels in the absence of elongin C interaction.
Proteasomal Inhibition Increases Cytoplasmic hnRNP A2 Expression in a pVHL-dependent Manner-The data with the C162F RCC cell line suggested that the interaction of pVHL with elongin C was necessary to regulate hnRNP A2 protein levels. To test the hypothesis that pVHL regulated hnRNP A2 levels through modulating protein turnover, the effect of proteasomal inhibition on hnRNP A2 gene expression was evaluated (Fig. 6). Overnight treatment with the proteasome inhibitor MG132 increased cytoplasmic hnRNP A2 in WT-8 RCC cells, while pRc-9 RCC cells were unaffected (Fig. 6A). Following proteasomal inhibition, cytosolic hnRNP A2 levels in WT-8 cells approached that seen in pRc-9 cells. Similar results were obtained with a proteasome inhibitor with a different mechanism of action, LLnL (Fig. 6B). Immunoblots were then reprobed with polyclonal antibody directed against hnRNP A1. The specificity of this effect on hnRNP A2 expression is further demonstrated by lack of change in nuclear or cytosolic levels of hnRNP A1 in either cell type (Fig. 6, A and B). Thus, proteasomal inhibition selectively increased cytosolic hnRNP A2 in a VHL-dependent manner, suggesting that VHL mediates proteasomal degradation of hnRNP A2.
Interaction of pVHL and hnRNP A2 in Vivo-To address the possibility that pVHL regulates hnRNP A2 protein turnover, we examined whether VHL physically interacts with hnRNP A2 in vivo, as it does with HIF-1␣ (45,47,60). Due to low levels of hnRNP A2 in the cytosol of WT-8 cells, we examined their association following treatment with MG132. Immunoprecipitates of pVHL from proteasomally inhibited WT-8 cytosol were shown to contain hnRNP A2 (Fig. 7A, left panel). Indeed, hnRNP A2 immunoprecipitation quantitatively depleted pVHL from the cytosol (Fig. 7A, right panel). In contrast, no pVHL was found in hnRNP A1 immunoprecipitates (data not shown). This interaction between pVHL and hnRNP A2 can also be shown in the nuclear fraction of WT-8 cells (Fig. 7B). These data not only imply that VHL and hnRNP A2 interact in vivo in the nucleus and cytosol, but also indicate that, in the context of proteasome inhibition, cytoplasmic VHL is complexed almost entirely with hnRNP A2.

DISCUSSION
In this report, we demonstrate that hnRNP A2 levels in RCC cell lines vary inversely as a function of pVHL expression. Neither altered subcellular distribution nor different rates of cell growth account for the elevated expression of hnRNP A2 observed in pVHL-deficient cells. The effects of pVHL on hnRNP A2 levels were not due to altered sequestration, as equivalent effects were seen with cells directly lysed in SDS-PAGE loading buffer. The levels of hnRNP L, which has been shown to form complexes in vivo with hnRNP A2 (49), were also reduced by pVHL, but this effect was more modest than the alteration in hnRNP A2 protein expression and was restricted to polysomal fractions. Moreover, increased hnRNP A2 protein expression in these cells does not correlate with changes in mRNA levels, implying that pVHL affects the rate of hnRNP A2 protein turnover or translation. Pharmacologic inhibition of proteasome function increased cytoplasmic hnRNP A2 levels, but only in cells expressing pVHL. Immunoprecipitation studies indicate that pVHL is complexed with hnRNP A2 in the nucleus and cytoplasm of these cells. In summary, hnRNP A2 and pVHL interact in vivo, and pVHL-dependent regulation of cytoplasmic hnRNP A2 expression requires functioning proteasomes. These findings have potential relevance to the increased levels of GLUT-1 and VEGF mRNA that occur in the absence of pVHL, as hnRNP A2 and L have been shown to bind functionally relevant cis-acting elements in the 3Ј-UTR of these mRNA (49,61). In addition, these findings suggest a potential model of the tumor suppressor role of the VHL gene.
The pVHL-dependent difference in hnRNP A2 expression is not due to increased transcription or mRNA stability, as levels of hnRNP A2 mRNA levels were unaffected by VHL gene expression in three experiments. Therefore, pVHL most likely regulates hnRNP A2 expression at the level of mRNA translation or protein turnover. However, several lines of data strongly support a mechanism of increased hnRNP A2 protein turnover, including the demonstration that hnRNP A2 and pVHL directly interact in vivo and that hnRNP A2 levels increase with proteasomal inhibition, but only in pVHL-expressing cell lines.
Consistent with this possibility, we demonstrate that C162F cells, in which the ability of pVHL to bind the elongin B/C complex is abrogated, exhibit elevated expression of hnRNP A2 relative to cells expressing wild type pVHL. The interaction of pVHL with elongin C has been shown to be necessary for conferring pVHL function as an E3 ubiquitin ligase (32,59). In this model, the failure of pVHL to interact with components of ubiquitin degradation machinery (elongins B and C, Cul-2, and Rbx1) results in cytosolic hnRNP A2 accumulation, as hnRNP A2 cannot be targeted by pVHL to the proteasome for proteolytic degradation. Such a model would predict (and our data support) that hnRNP A2 and pVHL interact in vivo, and that this association is heightened by proteasome inhibition.
Further supporting a role for altered hnRNP A2 protein turnover is the demonstration that polysomal hnRNP L levels are affected by pVHL expression. Since hnRNP A2 and L have been shown to exist as a complex on the polysomes, these data raise the possibility that pVHL mediates proteasomal targeting of this hnRNP A2/L mRNA complex and degradation. This would explain why the effects of pVHL expression on hnRNP L levels were only consistently seen in polysomal fractions, presumably through its association with hnRNP A2 (49). In this model, only hnRNP L that is associated with hnRNP A2 is degraded by pVHL targeting to the proteasome.
The model that pVHL plays a role in targeting specific mRNA-protein complexes to the proteasome is consistent with both increased expression of hnRNP A2/L and the stability of VEGF and GLUT-1 mRNA that occurs that occurs in its absence (33,62). For example, hnRNP L has been shown to bind the 3Ј-UTR of VEGF and play a role in stabilizing this mRNA (61). A similar function may exist for hnRNP L binding to GLUT-1 mRNA (49). In contrast, hnRNP A2 binding to GLUT-1 mRNA has been correlated with increased mRNA turnover in brain tumor extracts (49,63). The demonstrated relationship between increased hnRNP A2 levels and GLUT-1 mRNA stability in pVHL-deficient RCC cells is not consistent with these data and may reflect tissue-specific differences. Such a tissue-restricted effect would be consistent with the selective tissue pathology associated with VHL disease (17,18).
Validating this model requires demonstration that pVHL mediates hnRNP A2 protein turnover. These studies will be limited by the stability of hnRNP A2 protein, which in prelim- Cytosolic lysate (100 g) derived from MG132-treated WT-8 cells was immunoprecipitated (IP, ip) with ␣-VHL monoclonal antibody (1G32), and then analyzed by immunoblotting with ␣-hnRNP A2 monoclonal antibody (EF67). Input indicates start material, while Depl refers to supernatant unbound to beads. Reciprocal experiment is indicated in right panel. B, coimmunoprecipitation of hnRNP A2 and VHL from untreated WT-8 cells. Nuclear lysate (400 g) extracted from WT-8 cells was immunoprecipitated with 1G32, and then analyzed by immunoblotting with EF67.
pVHL Interacts with and Regulates hnRNP A2 inary studies has a half-life of greater than 12 h in both cell lines (data not shown). Moreover, we have been unable to detect an increase in the M r of hnRNP A2 with proteasomal inhibition expected with ubiquitination (data not shown). The failure to demonstrate ubiquitinated forms of hnRNP A2 does not eliminate the possibility that proteasomal targeting of an mRNA⅐hnRNP A2 complex might occur through ubiquitination of another protein in that complex. Immunoprecipitation of the hnRNP A2 complexes from cells that differ solely in pVHL expression may resolve some of these issues. Independent of the mechanism by which pVHL reduces hnRNP A2 levels, its specificity is notable. No effect is seen on hnRNP A1 levels, which shares identical domain organization, nuclear import/export signals, and considerable amino acid homology (70% overall). The N-terminal RNA binding domains of hnRNP A1 and A2 share 85% homology (50), in accordance with their high affinity binding of polyuridine sequences (64). Despite these similarities, hnRNP A1 and A2 exhibit disparate AURE binding specificity (49). Recombinant hnRNP A2, but not hnRNP A1, was shown to bind the GLUT-1 3Ј-UTR, whereas the reverse was true for the granulocyte macrophage colony-stimulating factor AURE (49). Since pVHL has been demonstrated to play a role in GLUT-1, but not granulocyte macrophage colony-stimulating factor, mRNA turnover, these data are consistent with the interpretation that selective modulation of hnRNP A2 levels by pVHL is associated with its RNA binding specificity in vivo.
Finally, VHL mutations are specifically and highly associated with sporadic and familial forms of clear cell RCC, which arise in proximal tubular epithelial cells (6). It is intriguing that the proximal tubular epithelial cell of the kidney is relatively unique in its high level of cytosolic hnRNP A2 expression (65). 2 The overexpression of hnRNP A2 in cells lacking pVHL is of particular interest, given its association with neoplastic transformation and tumorigenesis in the lung (66,67). Collectively, these data suggest that the association of VHL mutations with clear cell carcinomas of the kidney is due to its critical role in regulating the high ambient levels of cytoplasmic hnRNP A2 in normal proximal tubular epithelium. In this model, the absence of pVHL-elongin C interaction causes hnRNP A2 levels to rise above a threshold level, resulting in renal carcinogenesis. In closing, the demonstration of a specific relationship between pVHL and hnRNP A2 suggests potential correlates with the subcellular distribution and function of pVHL, including tissue-specific tumor suppression and regulation of GLUT-1 mRNA stability.