Ran-mediated nuclear export of the von Hippel-Lindau tumor suppressor protein occurs independently of its assembly with cullin-2.

Inactivating mutations of the von Hippel-Lindau (VHL) tumor suppressor gene cause the VHL cancer syndrome and sporadic renal clear cell carcinoma. VHL engages in a nucleocytoplasmic shuttle, which is required for its function. Here, we pursue our investigation to identify mechanisms by which VHL-green fluorescent protein (VHL-GFP) is exported from the nucleus. We show that nuclear export of VHL-GFP in living cells requires ongoing RNA polymerase II activity, and is mediated by mechanisms that are temperature-sensitive and energy-dependent. In vitro nuclear export of VHL-GFP is inhibited by nuclear pore-specific lectins, requires ATP hydrolysis and polyadenylated mRNAs, and occurs with kinetics that are similar to those of proteins containing a nuclear export signal. Biochemical fractionation has revealed that nuclear export of VHL-GFP occurs by way of a Ran-dependent pathway. Size exclusion column chromatography and deletion mutant analysis suggest that VHL-GFP does not require assembly with one of its associated proteins, cullin-2, to engage in nuclear export. These results demonstrate that nuclear export of VHL-GFP is Ran-mediated and ATP hydrolysis-dependent. They also suggest that sequences outside the elongin C binding box may function as a nuclear export domain, potentially providing a novel role for this region of VHL frequently mutated in renal cell carcinoma.

The VHL 1 gene was identified in 1993 as the tumor suppressor gene whose germ line mutations are associated with the inherited von Hippel-Lindau cancer syndrome (1)(2)(3). VHL patients develop a wide variety of tumors including retinal angioma, central nervous system hemangioblastoma, pheochromocytoma, and renal clear cell carcinoma (RCC). Biallelic VHL gene defects are also found in the sporadic form of these tumors, including RCC, the most common type of kidney cancer in human (4,5). It is hypothesized that VHL exerts gatekeeper function in renal proximal tubule cells, which are thought to give rise to RCC (6 -8). VHL-negative RCC form tumors in nude mice that are suppressed by the reintroduction of VHL, consistent with its gatekeeper function (9). In addition, the reintroduction of VHL in VHL-negative RCC restored their ability to regulate levels of hypoxia-inducible mRNAs (10 -13), to form an extracellular matrix (14), to induce differentiation of RCC grown as multicellular tumor spheroids (15), and to exit the cell cycle in low serum (8).
The mechanism by which VHL is able to function as a gatekeeper tumor suppressor still remains unknown. The VHL mRNA produces two proteins from alternative in-frame methionines of 213 and 160 amino acids that do not share sequence similarity with other proteins, thus giving no clues as to their functions (both forms will be referred to as "VHL") (16 -18). VHL assembles into a core heterotetrameric complex with three other associated proteins: elongin B, elongin C, and cullin-2 (VBC/Cul-2) (12, 19 -21). Structural similarity between elongin C and cullin-2 with two components of a yeast E3ubiquitin ligase complex, SKP1 and CDC53, respectively, has led to the suggestion that VBC/Cul-2 may play a role in protein ubiquitination (20 -22). This model is further supported by the recent discovery that VBC/Cul-2 assembles with Rbx1 (23), and acts as an E3-ubiquitin ligase in vitro (24). One potential target of VBC/Cul-2-mediated degradation is the hypoxia-inducible factor-␣ subunit, a protein involved in the regulation of several hypoxia-inducible mRNAs (25). VBC/Cul-2 might regulate specific mRNAs including vascular endothelial growth factor, glucose transporter-1 (Glut-1), and transforming growth factor-␣ through its ability to modulate levels of hypoxia-inducible factor-␣ subunit (10 -13, 25, 26). Disruption of VBC/Cul-2 by tumor-derived mutations results in the abnormal stabilization, and accumulation, of vascular endothelial growth factor, Glut-1, and transforming growth factor-␣ mRNAs, suggesting a role for the complex in the processing of these transcripts (12).
Another intriguing characteristic of VHL is that it is able to localize to the nucleus and the cytoplasm, and shuttle between these two compartments (9,(27)(28)(29)(30)(31)(32). VHL shuttling is sensitive to ongoing RNA polymerase II activity, but is not affected by treatment with leptomycin B, a drug that affects CRM1-mediated nuclear export of protein containing a classical, leucinerich nuclear export signal (NES; Refs. 33 and 34). Likewise, VHL was unable to function as a negative regulator of Glut-1 levels when it was fused to a classical NES and induced to shuttle in a leptomycin B-sensitive, but transcription-insensitive, manner (29). These observations suggested that VHL might export from the nucleus through a yet undescribed pathway. Therefore, we decided to pursue our studies on VHL subcellular trafficking by investigating the mechanisms involved in its regulated nuclear export. We show that VHL nuclear export occurs through a Ran-mediated and ATP hydrolysis-dependent pathway. VHL exports from the nucleus bound to complexes containing cullin-2, but appears to be able to mediate its own nuclear export and that of an inert green fluorescent protein (GFP) reporter. These results provide evidence that VHL contains a nuclear export domain, which might play a role in nuclear export of the VBC/Cul-2 complex.

MATERIALS AND METHODS
Cell Culture and Transfections-The 786-0 (VHL-negative) and HeLa cells were obtained from the American Type Culture Collection (Rockville, MD). The VHL-GFP cell line corresponds to a 786-0 stably transfected with the VHL-GFP fusion protein as described elsewhere (29). All cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% (v/v) fetal calf serum in a 37°C, humidified, 5% CO 2 -containing-atmosphere incubator. Transient transfections of HeLa cells were performed overnight using a standard calcium phosphate method.
Expression Vectors and Constructs-The human VHL cDNA, which codes for a 213-amino acid VHL protein, was subcloned into pcDNA3.1(Ϫ) (Invitrogen) vector. A Flag epitope tag (DYKDDDDK) was added to the N terminus of the VHL cDNA open reading frame. A cDNA coding for an enhanced fluorescence version of the GFP (Fred25; Ref. 35) was subcloned at the C terminus VHL to produce the VHL-GFP fusion protein. A deletion mutant of the last 56 amino acids was also fused to GFP to produce the ⌬C157-GFP fusion protein. The VHL moiety was also replaced by a full-length cullin-2 cDNA (21) to produce cullin-2-GFP. The open reading frame of chicken pyruvate kinase (PK) fused to GFP (PK-GFP), and two GFP in tandem (GFP-GFP), were cloned into pcDNA3.1(Ϫ). The NLS fusion proteins were produced by fusing the strong NLS of simian virus 40 (SV40; Ref. 36) large T antigen (PKKKRKKV; NLS) to the C terminus of GFP. All constructs were verified by DNA sequencing.
In Vitro Nuclear Export Assay-The nuclear export assay was essentially as described (37)(38)(39)(40). Cells were plated on a 35-mm dish with a hole at the bottom replaced by a glass coverslip. Cells were grown to confluence, washed with PBS and incubated 1 min in the presence of transport buffer containing 20 mM HEPES pH 7.3, 110 mM KOAc, 5 mM NaOAc, 2 mM Mg(OAc) 2 , 1 mM EGTA, and 2 mM DTT. The cells were permeabilized at 4°C for 5 min in the presence of transport buffer containing 50 g/ml digitonin and a protease inhibitor mixture including 1 mM phenylmethylsulfonyl fluoride, 0.5 g/ml leupeptin, and 1 g/ml aprotinin. Cells were washed several times with transport buffer and protease inhibitor mixture for 15 min at 4°C. Cells were incubated for the indicated time, generally 30 -45 min, at 20°C in the presence of the standard reaction mixture that included transport buffer, HeLa cytosol, 2 mM ATP, 2 mM GTP, and an ATP-regenerating system (5 mM creatine phosphate and 20 units/ml creatine phosphokinase). Where indicated, ATP and GTP were omitted or replaced with AMP-PNP and GMP-PNP at 2 mM, respectively. Wheat germ agglutinin (WGA) was added, as indicated, at a concentration of 200 g/ml. The overall volume of the reaction mixture was 0.5-1 ml. For the preincubation experiments, permeabilized cells were incubated 15 min at 4°C in the presence of transport buffer, followed by an incubation of 30 min at 20°C in the presence of transport buffer and 2 mM ATP, 2 mM GTP, and an ATP-regenerating system before the addition of the standard reaction mixture. Reactions were stopped by washing the cells several times with ice-cold transport buffer. For biochemical analysis of export, cells were trypsinized, washed in PBS, permeabilized in the presence of transport buffer with 50 g/ml digitonin for 5 min at 4°C supplemented with the protease inhibitor mixture, and subsequently washed several times with cold transport buffer for 15 min at 4°C. Nuclei were aliquoted in 1.5-ml tubes and treated as described above. Export was terminated by centrifugation at 300 ϫ g at 4°C for 2 min to separate nuclei from the supernatant. Nuclei were lysed in the presence of 1% Triton X-100, 20 mM Tris-HCl (pH 8.0), 137 mM NaCl with protease mixture for 30 min at 4°C, or with 4% SDS in PBS. 4ϫ SDS-sample buffer was added to aliquots of nuclear lysates and supernatant, which were separated on SDS-PAGE, and blotted onto nitrocellulose using standard methods. Blots were blocked with 5% milk powder in PBS containing 0.1% Tween 20, incubated with the indicated antibodies at a concentration of 1 g/ml for 16 h at 4°C, washed, and incubated with a horseradish peroxidase-coupled donkey anti-mouse or anti-rabbit antibody (1:5000) for 60 min at room temperature. The enhanced chemiluminescence system was used to detect signal (Amersham Pharmacia Biotech). Blots were also incubated with other antibodies; an anti-p27 antibody at 1:1000 (Transduction Laboratories) and an anti-cyclin B antibody at 1:1000 (Santa Cruz) and a rabbit anti-cullin-2 antibody at 1:5000.
Fluorescence Analysis and Image Processing-Nuclear GFP fluorescence images were captured using a Zeiss Axiovert S100TV microscope with a C-Apochromat 40ϫ water immersion objective, equipped with an Empix digital charge-coupled device (CCD) camera using Northern Eclipse software. Images were manipulated with Northern Eclipse, NIH Image, and Adobe Photoshop software as described elsewhere (29). Briefly, relative nuclear signals were measured as followed. Images were obtained with identical integration time, usually 0.5-3 s at a 2 by 2 binning with an electronic gain of 2. Images were always taken under nonsaturating conditions. Three random images containing between 30 and 45 independent nuclei, which could be observed with Hoechst staining, were captured within 30 -45 s. GFP images were always taken before the Hoechst images to minimize any possible bleaching effect. Hoechst positive regions were used to identify and measure GFP signal in the different nuclei. Reactions were standardized using an arbitrary value of 100%, which corresponded to signal obtained immediately after the 15-min washing step at 4°C. Approximately 90% of nuclei were positive for GFP signal over background, whereas 10% of nuclei showed either weak or no fluorescence signal. Values were average from three independent experiments, always using a new 100% standard in each experiment. For the time lapse imaging, images were captured as described above but with one main difference to eliminate bleaching that occurs during the repeated exposure to 488 nm light. Bleaching was measured at ϳ 2% per picture and taken into account for quantification.
Purification of the Major HeLa Cytosolic Activity for VHL Nuclear Export-One ml of HeLa cytosol was loaded on a Superdex S-200 column equilibrated with 10 mM HEPES (pH 7.3) and 10 mM NaCl using a flow rate of 0.6 ml/min and collecting 25 fractions of 0.5 ml. Half of each other fractions were added to ATP/GTP and an ATP regeneration system to test for nuclear export of VHL-GFP. Aliquots of 50 l of some of the fractions were also separated on SDS-PAGE, blotted on nitrocellulose, and incubated in the presence of anti-Ran antibody (Transduction Laboratories). Recombinant Ran and RanG19V were a kind gift from Dr. Mary Dasso (National Institutes of Health, Bethesda, MD) and added to the export reaction at a concentration of 100 g/ml. For experiments shown in Fig. 5, Ran was produced as described (41). Briefly, bacteria expressing Ran were pelleted and resuspended in 1/20 volume of transport buffer and a mixture of protease inhibitor. The bacterial suspension was lysed by sonication, and the lysate was centrifuged at maximum speed in an Eppendorf microcentrifuge for 30 min. The supernatant was further centrifuged at 80,000 ϫ g for 30 min at 4°C. The supernatant was either used directly or diluted 1:1 in transport buffer immediately before using in transport reaction. Higher dilution of the lysate resulted in a marked loss of nuclear export of VHL-GFP. Bacterial lysate alone did not support VHL-GFP nuclear export in vitro (see Fig. 3A).

RESULTS
Nuclear Export of VHL-GFP in Living Cells-We have previously reported that nucleocytoplasmic shuttling of VHL-GFP is insensitive to leptomycin B treatment raising the possibility that nuclear export of VHL-GFP occurs through a CRM-1independent pathway (29). However, it was unclear if nuclear export of VHL-GFP was the outcome of simple passive diffusion through the nuclear pore complex or if it required energy-dependent mechanisms. The dependence on NTP hydrolysis would suggest that VHL-GFP might utilize a specific, transporter-mediated nuclear export pathway (42). Two independent strategies were developed to test if energy is required for nuclear export of VHL-GFP in living cells. First, transiently transfected HeLa cells were incubated in the presence of the RNA polymerase II inhibitor DRB, which induced VHL-GFP to change its steady state distribution from cytoplasmic to mainly nuclear (Fig. 1A, a and b; see also Ref. 29). Reactivation of RNA polymerase II-mediated transcription, by washing away DRB, caused VHL-GFP to reestablish a predominately cytoplasmic distribution (Fig. 1A, c) without any changes in level of fluorescence (data not shown and Ref. 29). As these experiments are performed in the presence of cycloheximide, the return of VHL-GFP to the cytoplasm can only be explained by the ability of existing molecules to export from the nucleus. The redistri-bution of VHL-GFP to the cytoplasm is inhibited when cells are incubated at 4°C or in the presence of metabolic poisons (Fig.  1A, d and e), suggesting that nuclear export of VHL-GFP is an energy-dependent process. The ability of the VHL-GFP fusion protein to export from the nucleus is mediated by the VHL moiety, since replacing it by another GFP (GFP-GFP) resulted in a fusion protein whose localization was unaffected by DRB or metabolic poisons (data not shown; Ref. 35). The second approach consisted of a strategy by which VHL-GFP was forced to localize exclusively to the nucleus by way of its fusion to the strong, temperature-sensitive, and energy-dependent SV40 nuclear import signal (VHL-GFP-NLS; Fig. 1, B and C (a); Ref. 36). We reasoned that VHL-GFP-NLS would accumulate in the cytoplasm upon incubation of cells at 4°C, or in the presence of metabolic poisons, since the fusion protein would fail to reimport in these conditions. VHL-GFP-NLS localized exclusively in the nucleus even in conditions where NLS activity was inhibited ( Fig. 1C, b and c), suggesting that it is unable to passively diffuse out of the nucleus. The VHL moiety was replaced by another GFP to produce the GFP-GFP-NLS fusion protein whose mass and nuclear localization were similar to that of VHL-GFP-NLS (60 kDa; Fig. 1, B and C (d)). In contrast to VHL-GFP-NLS, GFP-GFP-NLS was readily detectable in the cytoplasm upon incubation at 4°C, or with metabolic poisons, indicating that this fusion protein is able to egress from the nucleus by passive diffusion (Fig. 1C, e and f; Ref. 43). Such as VHL-GFP-NLS, a large fusion protein consisting of pyruvate kinase fused to GFP and NLS (PK-GFP-NLS) remained restricted to the nucleus, regardless of the different treatments ( Fig. 1, B and C (g-i)). Put together, these results suggest that the VHL moiety is able to confer nuclear export activity to a reporter GFP in living cells, which requires ongoing RNA polymerase II activity and energy.
An in Vitro System to Study Nuclear Export of VHL-GFP-The results obtained in living cells suggested that VHL might utilize an energy-dependent mechanism to export from the nucleus. To identify cellular factors involved in this putative process, we used a VHL-GFP in vitro nuclear export assay based on the digitonin-permeabilization system (37)(38)(39)(40)44). This system has been utilized to characterize proteins that contain specific sequences directing their nuclear import or export (38, 44 -49). VHL-negative RCC 786-0 cells that stably express reintroduced VHL-GFP to very low levels, similar to those of endogenous VHL (29), were treated with digitonin causing the complete loss of cytoplasmic signal while retaining nuclear VHL-GFP ( Fig. 2A). The specific nuclear VHL-GFP signal should not be confused with the 786-0 autofluorescence, distinguished as perinuclear dots. To test if VHL-GFP is able to export from nuclei in vitro, permeabilized cells were incubated in standard nuclear export conditions that included transport buffer supplemented with cytosolic extract from HeLa cells, ATP, and GTP, as well as an ATP regeneration system (Fig.  2B). Efficient nuclear export was observed after incubation for 30 min at 20°C, as evidenced by an important loss of nuclear signal (Fig. 2B, a-d; compare a to c and Fig. 2C for quantitation). VHL-GFP failed to export from nuclei when permeabilized cells were incubated in standard conditions for 30 min at 4°C (Fig. 2, B (e-h) and C) or at 20°C with WGA, an inhibitor of signal-mediated nuclear transport (Fig. 2, B (i-l) and C). Both DRB and cordycepin, an inhibitor of polyadenylation of mRNA, partially inhibited the ability of VHL-GFP to export from nuclei ( Fig. 2D), in a manner reminiscent of their effect in living cells. Nuclear export of VHL-GFP was also monitored using a biochemical assay. Loss of nuclear signal was only observed when cells were incubated in standard conditions for 30 min at 20°C but not at 4°C, or with WGA, confirming data obtained with the imaging procedures (Fig. 2E). Lysates were blotted for the nuclear cyclin-dependent kinase inhibitor p27 and cyclin B, a protein that contains a classical NES and whose subcellular distribution at steady state is similar to that of VHL-GFP (50,51). p27, detected as a doublet in 786-0 cells, and cyclin B readily exported from nuclei upon incubation at 20°C, but not at 4°C or with WGA, indicating that the in vitro assay can sustain nuclear export of known exportable substrates. Fig.  3F shows that VHL-GFP accumulated in the supernatant, indicating that the loss of nuclear signal is not due to degradation of the fusion protein. The t 1/2 of VHL-GFP nuclear export was determined to be 6 -7 min by time lapse imaging procedures (Fig. 2, G and H, for quantitation). These results demonstrate that nuclear export of VHL-GFP is an efficient, temperaturesensitive, and signal-mediated phenomenon that occurs in a manner similar to that of substrates that contain characterized nuclear export activities.
Nuclear Export of VHL-GFP Requires Ran and ATP Hydrolysis-Proteins differ considerably in their requirements for soluble cytosolic factors for efficient nuclear export in vitro. The next step consisted of testing if soluble cytosolic factors are required for nuclear export of VHL-GFP. Complete nuclear export of VHL-GFP required the addition of HeLa cytosolic factors although some loss of nuclear signal could be detected in the absence of extract (Fig. 3A). Cytosolic extracts from other mammalian cells, including kidney cells, but not from Escherichia coli stimulated VHL-GFP nuclear export to levels similar to those of HeLa (Fig. 3A). To identify which major cytosolic activity is capable of promoting VHL-GFP nuclear export, HeLa extracts were loaded on a gel filtration column and every other fraction were tested for activity. Fractions in the 10 -60-kDa range showed greater activity compared with higher molecular mass fractions (Fig. 3B). Ran, a small GTP-binding protein of 25 kDa, is a key component of the nuclear/cytoplasmic transport machinery (38, 46, 52, 53) and is lost from cells FIG. 2. In vitro reconstitution of VHL-GFP nuclear export. A, digitonin treatment of 786-0 cells expressing VHL-GFP resulted in the exclusive loss of cytoplasmic signal. Cells were not treated (Ϫdigitonin) or treated with digitonin (50 g/ml) in transport buffer (ϩdigitonin). The corresponding Hoechst signal is shown on the right side. Arrows in b and d point at different nuclei after digitonin treatment to show nuclear VHL-GFP. Notice that the perinuclear dots are due to 786-0 autofluorescence and should not be confused with specific VHL-GFP signal. B, nuclear export of VHL-GFP in vitro. VHL-GFP-expressing 786-0 cells were permeabilized with digitonin, washed for 15 min at 4°C, and incubated in the presence of transport buffer supplemented with HeLa cytosolic extract (2 mg/ml), ATP/GTP (2 mM each) and an ATP regeneration system, for 0 min (a, e, i) or 30 min at 20°C (c), at 4°C (g), or at 20°C but with 200 g/ml WGA (k). Hoechst staining is presented (b, d, f, h, j, l). C, quantitative analysis of nuclear export of VHL-GFP in vitro. CCD camera-captured images were quantified with NIH image and Northern Eclipse softwares. At least 100 nuclei were analyzed in three independent experiments. Average values Ϯ standard errors are presented in relation to signals measured immediately after the 15-min washing step (prior to export; 0Ј). D, effect of DRB and cordycepin on in vitro nuclear export of VHL-GFP. Cells were treated with 25 g/ml DRB or 50 g/ml cordycepin for 2 h at 37°C, permeabilized, and incubated in the presence of complete nuclear export reaction mixture supplemented with DRB or cordycepin for 30 min at 20°C. Quantitative analysis is shown. E, biochemical analysis of VHL-GFP nuclear export. 786-0 cells expressing VHL-GFP were trypsinized, washed in PBS, and permeabilized. Nuclei were incubated in the presence of complete nuclear export reaction mixture for 0 and 30 min at 20°C, at 4°C, or at 20°C with WGA (200 g/ml). Nuclei were collected by centrifugation, washed twice with ice-cold transport buffer, and lysed with 4% SDS in PBS. Blots were incubated with an anti-FLAG M2 antibody (VHL-GFP), an anti-p27 antibody (p27), and an anti-cyclin B antibody (Cyclin B). F, VHL-GFP accumulates in the supernatant fraction during in vitro nuclear export. After incubation of nuclei for 0 and 30 min at 20°C with complete nuclear export reaction mixture, the nuclei were collected by centrifugation and equal aliquots of supernatant fractions (0 and 30 min) were separated on SDS-PAGE, transferred to membrane, and blotted with an anti-FLAG M2 antibody. G, the t 1/2 of VHL-GFP nuclear export in vitro is comparable to that of other exportable substrates. Time lapse imaging of digitonin-permeabilized cells incubated at 20°C in the presence of a complete export reaction mixture in the absence (ϪWGA) or presence of 200 g/ml WGA (ϩWGA) for the indicated time. H, quantitative analysis of the nuclear export rate of VHL-GFP in the presence or absence of WGA. Average and standard error of three independent experiments are shown. upon digitonin treatment and washing at 4°C. We found that fractions that were the most efficient in stimulating nuclear export of VHL-GFP corresponded to the ones containing higher amounts of Ran (Fig. 3B). This observation led us to directly test the involvement of Ran in nuclear export of VHL-GFP. Fig.  3C shows that purified Ran is able to stimulate nuclear export of VHL-GFP to levels similar to those obtained with the HeLa extract. Western blot analysis indicated that VHL-GFP efficiently exported from nuclei and accumulated in the supernatant in the presence of Ran (Fig. 3D). No difference in total amount of VHL-GFP signals could be detected between 0 and 30 min, in either the absence or the presence of Ran. Thus, VHL-GFP is exported from nuclei and there is no intranuclear or cytosolic degradation of the fusion protein in the presence of Ran and NTPs. It has been reported that a preincubation step of permeabilized cells at 20°C in the presence of NTPs depleted nuclei of specific nuclear transport factors required for nuclear export of proteins harboring a classical NES (52). In this assay, we noticed that Ran was sufficient to mediate nuclear export of VHL-GFP, regardless of a preincubation step with ATP (Fig.  3E). These results suggest that Ran is the major cytosolic factor that stimulates in vitro nuclear export of VHL-GFP, which is lost upon permeabilization.
Ran-mediated nuclear export of substrates appears not to require NTP hydrolysis (53). However, we noticed that the presence of NTP is required for efficient nuclear export of VHL-GFP stimulated by HeLa cytosolic extract (Fig. 4A, a and  b; quantitation is presented in Fig. 4B). Omission of ATP essentially inhibited nuclear export (Fig. 4A, c). Loss of nuclear signal was measured when GTP was not added to the export mixture (Fig. 4A, d), although we generally observed that the addition of GTP (Fig. 4A, a) or non-hydrolyzable GTP analogs GMP-PNP (Fig. 4A, e) or GTP␥S (data not shown) was required for optimal VHL-GFP nuclear export. VHL-GFP failed to export when ATP was replaced by AMP-PNP (Fig. 4A, f and g). Similar data were obtained when the extract was substituted with Ran (Fig. 4C). A Ran mutant, RanG19V, which binds GTP but is unable to support hydrolysis, was tested in the assay. RanG19V supported VHL-GFP nuclear export to levels similar to those of wild-type Ran as long as ATP was added to the export reaction mixture (Fig. 4C). Therefore, an ATP hydrolysis step is absolutely required for VHL-GFP nuclear export whereas the presence of GTP, or non-hydrolyzable GTP analogs, is only necessary for optimal nuclear export.
VHL Encodes Sequences That Direct Nuclear Export of GFP-It has been reported that VHL assembles into complexes that contain cullin-2 (8,12). Cullin-2 does not bind directly to VHL but first requires assembly with elongin C, which protein then assures complex formation by binding to the elongin C box on VHL (12). The in vitro biochemical characterization of VHL-GFP nuclear export enabled the analysis of the nucleocytoplasmic distribution of VHL-GFP/cullin-2 complexes. Western blot analysis revealed that about 90% of VHL-GFP, as well as endogenous cullin-2, can be detected in the cytosolic fraction, whereas 10% can be found in the nuclear fraction (Fig. 5A, upper and middle panels) in agreement with data obtained by   FIG. 3. Identification of the small GTPase Ran as the major activity in HeLa cytosolic extract that stimulates in vitro nuclear export of VHL-GFP. A, cytosolic factor requirement for VHL-GFP nuclear export in vitro. Digitonin permeabilized 786-0 cells expressing VHL-GFP were incubated for 30 min at 20°C in the presence of transport buffer supplemented with 2 mM each ATP/GTP, an ATP regeneration system and either with 2 mg/ml (ϩlysate) or without (Ϫlysate) HeLa cytosolic extract, or with 2 mg/ml each of extract from 786-0 RCC cells (ϩkidney lysate), Cos-7 cells, mouse NIH 3T3 cells, and E. coli cells. Quantitative analysis is shown. B, profile of VHL-GFP nuclear export activity of fractions collected from a Superdex-200 column. Twenty-five fractions of ϳ0.5 ml were collected at a flow rate of 0.6 ml/min. Reaction mixtures contained transport buffer, 2 mM each ATP/ GTP, an ATP regeneration system, and 20% of each other fractions. Molecular sizes (MW) for fractions represent values obtained from Amersham Pharmacia Biotech and are based on the flow rate. VHL-GFP nuclear export was measured semiquantitatively on 14 of the 25 fractions in comparison with 4°C transport buffer control and scored as: Ϫ, no or only slight export; ϩ/Ϫ, detectable export; and ϩϩ, export levels similar to those obtained with HeLa extract. An aliquot of 50 l of each fraction was separated on SDS-PAGE, transferred to nitrocellulose, and incubated with an anti-Ran antibody. Ran migrates at about 25 kDa and eluted from the Superdex-200 column in the 10 -60-kDa range, although longer exposure revealed the presence of Ran in fractions 5-12. C, effect of purified recombinant Ran on VHL-GFP in vitro nuclear export. VHL-GFP-expressing 786-0 cells were permeabilized with digitonin and incubated for 30 min with transport buffer, 2 mM each ATP/GTP, an ATP regeneration system, and either without (ϪRan, a) or with 100 g/ml Ran (ϩRan, b) at 20°C or with Ran but at 4°C (ϩRan, 4°C, c). D, biochemical analysis of Ran-dependent VHL-GFP nuclear export in vitro. 786-0 cells expressing VHL-GFP were trypsinized, washed in PBS, and permeabilized with digitonin. Nuclei were incubated at 20°C with transport buffer containing 2 mM ATP/ GTP, an ATP regeneration system without (ϪRan) or with (ϩRan) 100 g/ml recombinant Ran. Reaction time was either 0 or 30 min, as indicated. Nuclei were collected by centrifugation and separated from the supernatant fraction. Equal aliquots of nuclear (N) and supernatant (S) fractions were separated on SDS-PAGE. Blots were incubated with an anti-FLAG M2 antibody, to detect VHL-GFP. E, effect of Ran on VHL-GFP nuclear export after 30-min preincubation of permeabilized cells with ATP/GTP. Permeabilized cells were washed for the standard 15 min at 4°C and then incubated for another 30 min at 20°C in transport buffer containing 2 mM ATP/GTP and an ATP regeneration system. Cells were washed twice with ice-cold transport buffer and incubated for 30 min at 20°C in transport buffer, 2 mM ATP/GTP, an ATP regeneration system, and either without (ϪRan, a) or with (ϩRan, b) at 100 g/ml. digital camera (28). Immmunoprecipitation analysis, using anti-FLAG antibodies directed against the VHL-GFP fusion protein (VHL-GFP is fused to a N-terminal FLAG-tag) revealed that cullin-2 assembled with VHL-GFP in the nucleus and the cytoplasm (Fig. 5A, bottom panel).
Since VHL-GFP/cullin-2 complexes were detectable in the nuclear fraction, we used the in vitro nuclear export system to examine if VHL-GFP exported from nuclei assembled in complexes that contained cullin-2. Cells were permeabilized with digitonin, and the nuclei were collected by centrifugation and washed several times to remove the cytosolic fraction. Nuclei were resuspended in a transport buffer alone or containing an ATP regeneration system and Ran. After the indicated time, the nuclei were collected by centrifugation and separated from the supernatant. Western blot analysis revealed that cullin-2 is exported from the nuclei to the supernatant in a Ran-dependent manner (Fig. 5B). The nuclear and supernatant fractions were immunoprecipitated with anti-FLAG beads as described above (Fig. 5C). VHL-GFP/cullin-2 complexes were found in the supernatant fractions when nuclear export reactions were carried out at 20°C and in the presence of Ran and an ATP regeneration system, suggesting that VHL-GFP is able to export from the nuclei assembled with cullin-2. The nuclear export reactions were also carried out in the presence of a syn-thetic peptide comprising the elongin C-binding box motif of VHL. This peptide is able to prevent the in vitro assembly of VHL with elongin-C/cullin-2 (20, 24), but is unable to cause the disassembly of a preformed VHL-GFP/elongin C/cullin-2 complex (data not shown). We reasoned that the peptide would prevent reassembly of VHL-GFP/elongin C/cullin-2 complexes in the supernatant if VHL and cullin-2/elongin C proteins exported from nuclei independently. The results obtained with the peptide were exactly the same as the ones obtained without the peptide (Fig. 5C and data not shown), suggesting that VHL-GFP is able to export from nuclei assembled in complexes that contained cullin-2. We next wanted to test if a population of nuclear VHL-GFP was able to engage in nuclear export even though it was not assembled to cullin-2. Nuclear lysates were loaded on a size exclusion column, and complexes were separated based on their molecular weight. Anti-Flag immunoprecipitation analysis revealed a population of nuclear VHL-GFP in lower molecular weight fractions, which are lacking cullin-2 (Fig. 5D, top panel, lanes 8 and 9). The two bottom panels show the distribution of VHL-GFP/cullin-2 complexes before nuclear export (Nuclear) and after completion of the export reaction (Supernatant). In both cases, a clear VHL-GFP signal, without a strong corresponding cullin-2 signal, could be observed in fractions 8 and 9. Long exposure time of the autoradiograms revealed some detectable cullin-2 signal in fraction 5 and 8 and VHL-GFP signal in fraction 10 (data not shown). These results suggest that VHL-GFP is able to engage in nuclear export regardless of its assembly with cullin-2.
We then investigated if VHL was able to confer transcriptiondependent nuclear export properties to cullin-2. The subcellular localization of a cullin-2-GFP fusion protein in transiently transfected HeLa cells was observed to be very similar to that of VHL-GFP, but failed to redistribute to the nucleus upon

FIG. 5. Characterization of nuclear export of VHL-GFP/cullin-2 complexes in vitro.
A, subcellular distribution of VHL-GFP, cullin-2, and VHL-GFP/cullin-2 complexes. VHL-GFP stably expressing 786-0 cells were trypsinized, washed in PBS, and treated with 50 g/ml digitonin for 5 min at 4°C. The nuclei were collected by centrifugation (300 ϫ g) and separated from the cytosolic fraction. Nuclei were washed in ice-cold transport buffer and treated with Triton X-100 for 30 min at 4°C. Equal portions of the Triton-soluble nuclear fraction and cytosolic fraction were run on a SDS-PAGE. Total and 786-0 correspond to unfractionated Triton-soluble cellular lysate obtained from 786-0 expressing VHL-GFP, or parental 786-0 cells, respectively. Blots were incubated with an anti-M2 antibody directed toward the N-terminal FLAG-tag of VHL-GFP (upper panel) or with an anti-cullin-2 antibody (middle panel). Fractions were also immunoprecipitated with M2beads. Beads were washed, boiled in 2ϫ SDS loading buffer, and loaded on SDS-PAGE. After transfer, the blots were incubated with the anticullin-2 antibody (lower panel). B, nuclear export of cullin-2 in vitro. Digitonin-treated cells were incubated at 20°C for the indicated time either in the presence of transport buffer alone (buffer) or transport buffer supplemented with Ran and an ATP regeneration system (Ran). Nuclei were collected from the supernatant by centrifugation and lysed in Triton, and equal aliquots of nuclear and supernatant fractions were loaded on a 8% SDS-PAGE. Blots were incubated in the presence of an anti-cullin-2 antibody. Reaction mixture consists of the transport buffer with Ran and an ATP regeneration system, without permeabilized cells. 786-0 refers to permeabilized 786-0 cells. C, analysis of in vitro nuclear export of VHL-GFP/cullin-2 complexes. Permeabilized cells were incubated for 30 min at 20°C either in transport buffer alone (buffer) or transport buffer supplemented with Ran and an ATP regeneration system (Ran/Energy). Nuclei were collected by centrifugation and lysed in Triton, and the nuclear and supernatant fractions were immunoprecipitated with anti-FLAG beads. Beads were washed several times, boiled, and loaded on 8% SDS-PAGE. The top part of the blot was incubated with an anti-cullin-2 antibody (Cullin-2), whereas the bottom part was incubated with an anti-FLAG antibody (VHL-GFP). Heavy chain is detected below the specific VHL-GFP signal. The same experiments were conducted in the presence of 10 -100 g/ml amounts of a peptide comprising the elongin C binding box of VHL as described in Ref. 24. D, a population of nuclear VHL-GFP does not assemble to cullin-2 and is still able to export from nuclei in vitro. Top panel, permeabilized cells lysate were loaded on a Superdex-200 column and fractions were incubated in the presence of anti-Flag beads. Beads were washed, boiled, and loaded on 8% SDS-PAGE. After transfer, filters were incubated with anti-cullin-2 antibody (Cullin-2) or anti-Flag antibody (VHL-GFP). Signal under VHL-GFP is from the heavy chain of anti-Flag antibody. Bottom panels, the panels underneath are either a Nuclei were washed in ice-cold transport buffer and treated with Triton X-100 for 30 min at 4°C (Nuclear; 100 g/lane). Total corresponds to unfractionated, Triton-soluble, cellular lysate obtained from 786-0 cells (60 g/lane). Blots were incubated with an anti-cullin-2 antibody. C, effect of DRB treatment on the subcellular localization of endogenous cullin-2 bound to VHL-GFP in stably expressing 786-0 cells. VHL-GFPexpressing 786-0 cells were treated as described in B. Equal portions of the total cellular extract (Total) Triton-soluble nuclear fraction (Nuclear), and cytosolic fraction (Cytosolic) were immunoprecipitated with M2 beads. Beads were washed, boiled in 2ϫ SDS loading buffer, and loaded on SDS-PAGE. After transfer, the blots were incubated with the anticullin-2 antibody (top panel) or anti-Flag M2 antibody (bottom panel). Different exposure times of the autoradiograms are shown for the total, nuclear, and cytosolic fractions. The band underneath VHL-GFP in the nuclear fraction is the heavy chain of the M2 antibody.
longer exposure of the same autoradiograms as shown above (fractions 5-9; Nuclear) or fractions of the supernatant collected after the nuclear export reaction (fractions 5-9; Supernatant). Notice the presence of VHL-GFP signal, but not of cullin-2, in fractions 8 and 9 in both the nuclear and supernatant conditions. Longer exposure time revealed VHL-GFP signal in fraction 10 and cullin-2 signal in fractions 5 and 8 (data not shown). addition of DRB (Fig. 6A, compared with Fig. 1A). It should be noted that the vast majority of overexpressed cullin-2-GFP is not assembled to endogenous VHL (data not shown). Western blot analysis of nuclear and cytosolic fractions obtained from VHL-negative 786-0 cells confirmed that the subcellular distribution of endogenous cullin-2 was unaffected by arrest of RNA polymerase II activity (Fig. 6B). However, immunoprecipitation analysis revealed that the subcellular distribution of the pool of cullin-2 that is bound to VHL-GFP, in 786-0 cells stably expressing VHL-GFP, was sensitive to ongoing RNA polymerase II transcription (Fig. 6C). Therefore, only cullin-2 that is found assembled with VHL-GFP is sensitive to ongoing RNA polymerase II for its localization, suggesting that VHL is able to confer transcription-dependent nuclear export properties to cullin-2.
Data presented in Figs. 5 and 6 suggested that VHL is able to export from nuclei unassembled to cullin-2. If this were the case, it would be predicted that VHL might encode information that directs its own nuclear export and that of the reporter GFP both in living cells and in vitro. To further investigate this possibility, a C-terminal truncation mutant of VHL, ⌬C157-GFP, which does not bind to elongin BC and cullin-2, was tested for its ability to mediate nuclear export of the reporter GFP in living cells and in vitro (8,12). ⌬C157-GFP showed increased nuclear localization at steady state compared with VHL-GFP and upon addition of DRB accumulated more into the nucleus (Fig. 7A, a and b). Removal of DRB caused ⌬C157-GFP to return to its original distribution (Fig. 7A, c). However, a substantial portion of ⌬C157-GFP remained in the nucleus, in a manner reminiscent of VHL-GFP, when the washout experiments were performed at 4°C or in the presence of metabolic poisons (Fig. 7A, d and e). Like VHL-GFP, ⌬C157-GFP was readily exported from nuclei in vitro at 20°C in the standard export conditions, but not in buffer only (Fig. 7B, a-d). In contrast, no significant difference was observed with PK-GFP-NLS, a fusion protein that is unable to export from nuclei in vitro (Fig. 7B, e and f). The GFP-GFP fusion protein was undetectable in nuclei regardless of the different incubation conditions (Fig. 7B, g and h), or at 4°C (data not shown) demonstrating that this fusion protein is able to passively diffuse out of the nucleus (see Fig. 1C). Nuclear export of ⌬C157-GFP was monitored in the biochemical assay (Fig. 7C). The fusion protein exported only in the presence of HeLa extract and NTPs and accumulated in the supernatant fraction, such as VHL-GFP, demonstrating that the mutant protein was not degraded in the nucleus. Time lapse imaging of ⌬C157-GFP in vitro nuclear export indicated a t 1/2 of 5-7 min (Fig. 7, D and E). These results demonstrate that the first 157 amino acids of VHL are able to confer nuclear export activity to a GFP reporter protein independently of its interaction with BC/Cul-2.

DISCUSSION
An in Vitro Assay to Reconstitute Nuclear Export of VHL-GFP-The VHL tumor suppressor gene product belongs to a group of proteins that require transcription-dependent nuclear/ cytoplasmic shuttling to perform their respective role (54 -65). We have previously shown that nuclear export of VHL required ongoing RNA Polymerase II activity but was insensitive to leptomycin B, a drug that inhibits CRM-1-mediated nuclear export of proteins containing a classical NES (33,34). We, therefore, decided to make use of an in vitro digitonin-permeabilized cell assay to elucidate mechanisms involved in VHL-GFP nuclear export. Nuclear export of VHL-GFP readily occurred when nuclei were incubated with transport buffer supplemented with soluble cytosolic factors and NTPs. Several sets of experiments were performed to demonstrate that the in vitro assay reflected the physiological nuclear export pathway utilized by VHL-GFP. Nuclear export of VHL-GFP occurred in a time-, temperature-, and ATP hydrolysis-dependent manner and was inhibited by WGA, a compound that interferes with signal-mediated nuclear export. Furthermore, VHL-GFP appeared to utilize a very efficient pathway since its nuclear export rate was measured to be similar to that of proteins containing a classical NES (45,47,52). The cell line used for this study expressed VHL-GFP to very low, near physiological FIG. 7. VHL encodes activity that mediates its own nuclear export in living cells and in vitro. A, effect of RNA polymerase II activity on ⌬C157-GFP localization in living cells. HeLa cells were transiently transfected with ⌬C157-GFP (a) and treated with DRB for 2 h at 37°C (b). DRB was removed from the media and cells were incubated at 37°C (c), at 4°C (d), or at 37°C in the presence of 6 mM 6-deoxyglucose, 0.02% sodium azide (e) for 2 h and in the presence of cycloheximide (25 g/ml). B, nuclear export of ⌬C157-GFP in vitro. HeLa cells were transiently transfected with VHL-GFP, and ⌬C157-GFP, PK-GFP-NLS, and GFP-GFP. Cells were permeabilized with digitonin, washed for 15 min at 4°C, and incubated for 30 min at 20°C in transport buffer (Buffer) or in transport buffer containing HeLa cytosolic extract, 2 mM ATP/GTP, and an ATP regeneration system (Lysate, ATP/GTP). C, biochemical analysis of ⌬C157-GFP nuclear export in vitro. HeLa cells were transfected with ⌬C157-GFP. Nuclei were incubated at 20°C with transport buffer (buffer) or complete export reaction mixture that consists of the HeLa cytosolic lysate and an ATP regeneration system (L/E). Reaction time was either 0 or 30 min, as indicated. Nuclei were collected by centrifugation, and separated from the cytosolic fraction. Equal aliquots of nuclei and cytosolic fractions were separated on SDS-PAGE. Blots were incubated with an anti-FLAG M2 antibody to detect VHL-GFP. D, time lapse imaging of ⌬C157-GFP nuclear export in vitro. HeLa cells expressing ⌬C157-GFP were permeabilized and incubated in buffer, or in standard conditions (Lysate ATP/GTP). Low magnification digital images were taken at the indicated times during the export assay. E, quantitative analysis of the nuclear export rate of ⌬C157-GFP. Average and standard errors of three independent experiments are shown. levels (29), explaining why nuclear signal is so weak and comparable to 786-0 autofluorescence. While it is sometimes difficult to distinguish specific signal from autofluorescence in a system in which VHL-GFP is expressed to physiological levels, this does argue that VHL-GFP nuclear export is not a consequence of overexpression of the fusion protein and most likely represents a physiological event. In support of the in vitro data are those obtained with the intact cells, which also demonstrated that VHL-GFP nuclear export is a process that is energydependent. The information required to stimulate nuclear export of the VBC/Cul-2 complex appears to be encoded by VHL since a mutant VHL, which does not assemble with its known binding partners, was able to confer nuclear export properties to a reporter GFP. Put together, these observations suggest that VHL may encode a domain that mediates nuclear export as efficiently as a classical NES.
Factors Required for Efficient Nuclear Export of VHL-The in vitro assay enabled the biochemical reconstitution of the nuclear export pathway utilized by VHL. Our results suggest that at least four independent factors are required for efficient nuclear export of VHL-GFP: an ATP hydrolysis step, the presence of GTP, the synthesis of polyadenylated RNA, and the small GTPase Ran. Omission of exogenous ATP, or addition of a non-hydrolyzable analog, in the export reaction mixture totally inhibited the ability of VHL to export from the nucleus even in cells that maintained active transcription and in the presence of Ran. It is still unclear why ATP hydrolysis is necessary and when exactly this occurs in the nuclear export pathway. ATP hydrolysis has been shown to be required for nuclear export of other exportable substrates, such as REV-1 and glucocorticoid receptor (45,66). Perhaps ATP hydrolysis precedes Ran-dependent nuclear export by giving rise to a transport competent VHL-GFP, which might include its assembly with complexes containing polyadenylated mRNAs. In support of this model are the results obtained with cordycepin, a drug that inhibits polyadenylation of RNA. Cordycepin partially inhibited nuclear export of VHL-GFP, suggesting that the presence of mature, transportable, polyadenylated mRNAs might be important for optimal VHL-GFP nuclear export. It is unlikely that ATP hydrolysis alone is sufficient to stimulate VHL-GFP nuclear egression, even if some loss of nuclear signal was detectable when HeLa lysate, or Ran, was omitted from the reaction mixture. Our interpretation is that enough residual Ran (about 5% in our hands) is retained in permeabilized nuclei to assist in some nuclear export of VHL-GFP.
Proteins that contain leucine-rich NES utilize a CRM-1/Randependent pathway to export from nuclei (67,68). Although VHL does not appear to require CRM-1, partial biochemical fractionation of HeLa cytosolic extract and in vitro reconstitution has revealed that it does export from nuclei through a Ran-dependent pathway. In an effort to identify other molecules that might be involved in nuclear export of VHL-GFP, permeabilized cells were preincubated with ATP, a condition believed to deplete nuclei of transport factors. Using this approach, Kehlenbach and collaborators (52) elegantly showed that Ran-dependent nuclear export of NFAT, a protein that harbors a classical NES, necessitated an additional factor, CRM1, but only when nuclei were preincubated in the presence of ATP. Such molecules do not appear to be required for nuclear export of VHL-GFP, which suggest that Ran is the only ratelimiting soluble cytosolic factor lost upon permeabilization. Despite these observations, the requirement of other factors for VHL-GFP nuclear export cannot be simply excluded. Perhaps other molecules, such as ␤-transporters, are present in sufficient amounts to assist in Ran-dependent nuclear export of VHL-GFP and are not depleted by a prolong incubation with ATP. The identification of these putative factors involved in nuclear export of VHL-GFP will be important as they may play a role in the ability of VHL to function as a tumor suppressor. Regardless of the mechanisms involved in VHL nuclear export, this occurs without Ran-mediated GTP hydrolysis. We found that GMP-PNP could replace GTP and that the GTP hydrolysis-defective RanG19V stimulated VHL-GFP export to levels similar to those of wild-type Ran, consistent with a model in which RanGTP, but not Ran-mediated GTP hydrolysis, is required for nuclear export of proteins (53).
VHL Encodes Sequences That Mediate Its Own Nuclear Export-Immunoprecipitation and column chromatography analyses have revealed at least two independent populations of nuclear VHL-GFP distinguished by the fact that VHL is either assembled or unassembled with cullin-2. The same two populations were also detected in the supernatant after completion of the nuclear export reaction, suggesting that VHL-GFP is able to stimulate nuclear export of the reporter GFP without being assembled to cullin-2. This is highlighted by data obtained with ⌬C157-GFP, a mutant VHL that is unable to assemble with elongin BC/cullin-2. ⌬C157-GFP conferred nuclear export properties to an inert GFP reporter as efficiently as other proteins that contain a classical NES. The fact that another fusion protein (PK) was unable to stimulate nuclear export of GFP, as predicted (49), also argues that VHL encodes sequences that act as a nuclear export domain. Since VHL is able to stimulate its own nuclear export, it is tempting to speculate that it might also have a role in mediating cullin-2 nuclear export. Data presented here support the claim that VHL is, at least in part, able to confer transcription-dependent nuclear export properties to cullin-2. This is accentuated by the observation that, contrarily to VHL-GFP, cullin-2-GFP subcellular localization was unaffected by treatment with DRB. The localization of endogenous cullin-2 in a VHL-negative background was also insensitive to DRB. In contrast, the fraction of cullin-2 that is assembled to VHL-GFP is sensitive to DRB strongly suggesting that the transcription-dependent nuclear export properties of the VHL-GFP/cullin-2 complex is conferred by VHL sequences. Obviously, it is difficult to rule out that VHL, elongin BC, and cullin-2 are all able to engage in nuclear export independently and then reassemble in the supernatant. Some of the experiments were performed in the presence of a peptide comprising of the elongin C binding box, which should normally prevent reassembly of cullin-2/elongin C with free VHL. However, this control does not totally exclude the possibility that cullin-2 might engage in nuclear export independently of its assembly with VHL, perhaps by using a different pathway. If this were the case, cullin-2 would disassemble with VHL in the nucleus, engage in nuclear export, and then reassemble in the cytoplasm. Although we favor a model by which VHL, elongin BC and cullin-2 export out of nuclei as a complex, it will be difficult to resolve this question until we achieve a better understanding on how and when VHL binds to its partners.
Regardless of the way cullin-2 is exported out of the nucleus, we have presented data suggesting that VHL encodes information that directs its own nuclear export and that of a reporter GFP in living cells and in vitro. Nuclear export of VHL-GFP occurs through a signal-mediated pathway that requires Ran, ATP hydrolysis, and ongoing RNA polymerase II activity. Since VHL does not appear to contain leucine-rich sequences or a M9-like domain (61,69,70), fine mapping of these sequences will be required to identify the smallest region responsible for rapid nuclear export of VHL. Several cancer-causing substitutions found in sporadic RCC do not affect the ability of VHL to assemble with elongin BC/cullin-2 and appear to inactivate a yet undescribed VHL tumor suppression function (71). Some of these inactivating substitutions may prevent efficient nuclear export of VHL. It will be important to pursue studies detailing the relationship between the ability of VHL to mediate its own nuclear export, perhaps that of cullin-2, and its tumor suppressor function. The in vitro assay presented here will provide a valuable tool to answer these questions as well as dissect the molecular mechanisms involved in nucleocytoplasmic shuttling of VHL and its associated proteins.