Cancer-Causing Mutations in a Novel Transcription-Dependent Nuclear Export Motif of VHL Abrogate Oxygen-Dependent Degradation of Hypoxia-Inducible Factor

It is thought that degradation of nuclear proteins by the ubiquitylationsystem requires nuclear-cytoplasmic trafficking of E3 ubiquitinligases. The von Hippel-Lindau (VHL) tumor suppressor proteinis the substrate recognition component of a Cullin-2-containingE3 ubiquitin ligase that recruits hypoxia-inducible factor (HIF)for oxygen-dependent degradation. We demonstrated that VHL engagesin nuclear-cytoplasmic trafficking that requires ongoing transcriptionto promote efficient HIF degradation. Here, we report the identificationof a discreet motif, DXGX₂DX₂L, that directs transcription-dependentnuclear export of VHL and which is targeted by naturally occurringmutations associated with renal carcinoma and polycythemia inhumans. The DXGX₂DX₂L motif is also found in other proteins,including poly(A)-binding protein 1, to direct its transcription-dependentnuclear export. We define DXGX₂DX₂L as TD-NEM (transcription-dependentnuclear export motif), since inhibition of transcription byactinomycin D or 5,6-dichlorobenzimidazole abrogates its nuclearexport activity. Disease-causing mutations of key residues ofTD-NEM restrain the ability of VHL to efficiently mediate oxygen-dependentdegradation of HIF by altering its nuclear export dynamics withoutaffecting interaction with its substrate. These results identifya novel nuclear export motif, further highlight the role ofnuclear-cytoplasmic shuttling of E3 ligases in degradation ofnuclear substrates, and provide evidence that disease-causingmutations can target subcellular trafficking.

INTRODUCTION

Top

INTRODUCTION

Ubiquitylation is a multiprotein pathway that destines markedproteins for degradation by the 26S proteasome (22, 59). Theconjugation of ubiquitin to proteins requires the action ofthree different enzymes: E1 ubiquitin-activating enzymes, E2ubiquitin-conjugating enzymes, and E3 ubiquitin ligases. Theprocess of ubiquitylation begins with the loading of a ubiquitinmolecule onto the E1 ubiquitin-activating enzyme. This is followedby the transfer of ubiquitin from the E1 to the E2 ubiquitin-conjugatingenzyme. Finally, transfer of ubiquitin from the E2 to the lysineresidue of a target substrate is catalyzed by the E3 ubiquitinligase. Selectivity of this pathway relies heavily on E3 ubiquitinligases, which ultimately dictate substrate specificity. E3ubiquitin ligases can act individually or form a multisubunitcomplex that may include a member of the Cullin family of proteinsto covalently modify a vast array of cellular proteins. In viewof the essential role of E3 ubiquitin ligases in regulationof many aspects of cellular functions and biological processes,there is mounting evidence that loss of function or deregulationof E3 ligases contributes to the development of disease.

Degradation of nuclear substrates by the ubiquitylation systemoften requires nuclear-cytoplasmic trafficking of both the E3ubiquitin ligase and the substrate protein (2, 54). One exampleis the ubiquitin-mediated degradation of the p53 tumor suppressorprotein by the Mdm2 (murine double minute 2) E3 ubiquitin ligase(45, 47). Mdm2 shuttles continuously between the nucleus andthe cytoplasm in order to efficiently degrade nuclear p53 (12,53). Cancer-causing point mutations that disrupt nuclear exportof Mdm2 are impaired in mediating proteasomal degradation ofp53 (37). Nuclear export of the ROC1-SCF^Fbw1a E3 ubiquitin ligaseis also required for the proteasomal degradation of the Smad3transcription factor (13). Another example is the cyclin-dependentkinase inhibitor p27^Kip1, which requires nuclear export by Jab1for proteasome-mediated degradation. A mutant form of p27^Kip1that fails to assemble with Jab1 cannot be exported from thenucleus and is not degraded by the proteasome (54, 58).

The von Hippel-Lindau tumor suppressor protein (VHL) is a vitalcomponent of the VBC-Cul2 E3 ubiquitin ligase complex, as itacts as the substrate recognition protein to provide specificityto the degradation process (25, 27, 30, 38, 39, 50). VHL promotesthe recruitment, ubiquitylation, and subsequent proteasomaldegradation of the alpha subunit of hypoxia-inducible factor(HIF) in an oxygen-dependent manner (26, 41). Under conditionsof normal oxygen tension (normoxia), HIF ${alpha}$ is hydroxylated atkey prolyl residues within the oxygen-dependent degradationdomain by prolyl hydroxylases (5, 8, 24, 26). This posttranslationalmodification promotes the interaction between HIF ${alpha}$ and VHL andleads to ubiquitin-mediated degradation of HIF ${alpha}$ (62). Under conditionsof low oxygen tension (hypoxia), prolyl hydroxylation does notoccur, leading to the stabilization of HIF ${alpha}$ , since it fails toassemble with VHL (26, 46, 62). Stabilization of HIF ${alpha}$ resultsin increased transcription of an array of hypoxia-induciblegenes, including vascular endothelial growth factor, glucosetransporter 1, and transforming growth factor ${alpha}$ , among others,that modulate angiogenesis, glycolysis, and growth (11, 18,20, 55, 56). Numerous inactivating mutations of the VHL genelead to the stabilization of HIF ${alpha}$ and are associated with theVHL cancer syndrome, in which afflicted individuals developdifferent tumors, such as renal clear cell carcinoma (RCC),retinal angioma, nervous system hemangioblastoma, and pheochromocytoma(6, 23, 33, 40). Inactivating mutations of VHL often preventassembly with its substrate, HIF, or core components of theE3 ubiquitin ligase, elongins B and C, and Cullin 2, resultingin constitutive activation of HIF ${alpha}$ targets (7, 24, 28, 35, 39,46).

Nuclear-cytoplasmic trafficking is essential for the E3 ubiquitinligase function of VHL and oxygen-dependent degradation of HIF ${alpha}$ (17, 34). Failure of VHL to continuously shuttle between thenuclear and cytoplasmic compartments leads to the stabilizationof HIF ${alpha}$ (17, 34, 42-44). VHL engages in nuclear-cytoplasmic shuttlingdynamics independently of the classical, leucine-rich, nuclearexport sequence (NES) (9, 60) but accumulates in the nucleusupon addition of inhibitors of RNA polymerase II (Pol II) activity(17, 34). Interestingly, the general RNA metabolism and translationinitiation factor poly(A)-binding protein 1 (PABP1) exhibitssimilar transcription-dependent trafficking dynamics as VHL,since it also accumulates in the nucleus upon addition of inhibitorsof RNA Pol II activity (1). These results suggest the existenceof a transcription-dependent nuclear export pathway that isemployed by VHL, PABP1, and perhaps other proteins and thatoperates independently of the classical NES/CRM1 system.

Here we report the identification of a novel and discreet nuclearexport motif, DXGX₂DX₂L. We define this motif as TD-NEM (transcription-dependentnuclear export motif), since it mediates nuclear export of proteinsin a manner that requires ongoing RNA Pol II-dependent transcriptionbut operates independently of the classical NES pathway. Diseasemutations of TD-NEM of VHL alter its ability to be exportedfrom the nucleus and to mediate oxygen-dependent degradationof HIF ${alpha}$ without affecting interaction with its substrate. Theseresults highlight the requirement of nuclear-cytoplasmic traffickingof E3 ubiquitin ligases for degradation of their nuclear substrates,provide evidence that mutations targeting subcellular traffickingcan lead to disease, and identify a novel motif that mediatesefficient nuclear export of proteins.

MATERIALS AND METHODS

Top

MATERIALS AND METHODS

Cell culture, transfections, drug treatments, and hypoxia treatment. 786-0 (VHL-negative) renal carcinoma cells, MCF-7 cells, andNIH 3T3 cells were obtained from the American Type Culture Collection.Cells were maintained in Dulbecco's modified Eagle's mediumsupplemented with 5% fetal bovine serum and 1% penicillin-streptomycinin a 37°C and 5% CO₂ environment. Transient transfectionsin MCF-7 cells were conducted with Effective transfection reagent(Qiagen). Transfected cells were incubated for 24 h before anymanipulations or drug treatment. Stable cell lines were generatedby stably transfecting 786-0 cells with the following Flag-taggedconstructs: F-VHL-GFP, F-VHL(D121G)-GFP, F-VHL(D126Y)-GFP, andF-VHL(G123A)-GFP, followed by G418 selection. Where indicated,cells were treated at 37°C with a final concentration of2 µM actinomycin D (ActD), 10 µM leptomycin B (LMB),or 25 µg/ml 5,6-dichlorobenzimidazole (DRB) for 1 h priorto photobleaching experiments and 10 µM ActD or 10 µMLMB for 3 h prior to live cell fluorescence imaging. Proteasomeinhibitor treatment was performed with MG132 (Calbiochem) ata final concentration of 10 µM for 2 h before the cellswere harvested. Hypoxia treatment was performed in a hypoxicchamber at 37°C under a 1% O₂, 5% CO₂, and N₂-balanced atmosphere.

Expression vectors. Human full-length VHL and deletion, truncation, or point mutantsof VHL were cloned into pcDNA3.1 between an NH₂-terminal Flagtag and a COOH-terminal green fluorescent protein (GFP) tag,as previously described (4, 17, 34). F-GFP and F-GFP-NES werepreviously described by Groulx et al. (16) and Lee et al. (34).The PK-GFP-NLS construct used for the in vivo fluorescence lossin photobleaching (FLIP) nuclear export assay consisted of pyruvatekinase (PK), which does not encode any localization determinant,GFP, and a nuclear localization signal (NLS) derived from thesimian virus 40 large T antigen as previously described (16,29). Human full-length VHL and deletion mutants and the strongNES from the human immunodeficiency virus Rev were insertedinto a F-PK-GFP-NLS construct that was previously describedby Groulx et al. (16), between the Flag tag and PK using theApa1 and Xho1 restriction sites. cDNAs corresponding to VHLresidues 114 to 138 [VHL(114-138)], PABP1(296-317), and cyclinC(158-179), which encode TD-NEM sequences and the full-lengthcyclin C, were inserted into PK-GFP-NLS using Apa1 and Xho1restriction sites. The human full-length PABP1 and the deletionmutant ${Delta}$ 296-317 were fused to GFP-F to produce the GFP-F-PABP1and GFP-F-PABP1( ${Delta}$ 296-317) fusion proteins.

Live cell fluorescence imaging. Images of living cells transiently expressing GFP from experimentswhere photobleaching was not utilized were imaged with an AxiovertS100TV microscope (Carl Zeiss MicroImaging, Inc.) equipped witha 40x, 1.2 C-Apochromat water immersion objective using a digitalcharged-coupled-device camera (Empix). Cell nuclei were stainedwith Hoechst 33342 (Sigma). Images were captured using the NorthernEclipse software package (Empix).

Photobleaching and microscopy. Cells were cultured and transfected directly onto 35-mm disheswith coverslip bottoms (MatTek). Photobleaching and live cellmicroscopy were performed using a confocal microscope (LSM5Pa laser scanning microscope; Carl Zeiss Canada). In all experimentscells were maintained at 37°C in an environmental chamber.A 63x plan Apo oil immersion lens with a 1.4 numerical aperturewas used for bleaching and imaging. Indicated areas were exposedto three rapid pulses of a 488-nm argon laser at 100%, and imageacquisition was at 1% of full laser power. For cytoplasmic FLIPexperiments of cells expressing a GFP-tagged fusion protein,a large cytoplasmic region was initially bleached with threerapid pulses to eliminate the dominant cytoplasmic signal. Thiswas followed by repetitive bleaching in a small region of thecytoplasm and imaged at 30-second intervals. For cytoplasmicFLIP experiments of cells expressing a PK-GFP-NLS-tagged fusionprotein, cells were repeatedly bleached in a small cytoplasmicregion and imaged at 30-second intervals. Small bleached areasfor cytoplasmic FLIPs were kept consistent in terms of sizeand distance from the nucleus. Fluorescence loss in the unbleachedareas was quantified as previously described (43, 51) usingthe following equation: I_rel = (I_t/I₀)·(N₀/N_t), whereI_t is the average intensity of the unbleached nucleus or cellat time point t, I₀ is the average prebleached intensity ofthe nucleus or cell of interest, and N₀ and N_t are the averagenuclear or cellular fluorescence intensity of a neighboringcell in the same field of vision prebleach or at time t, respectively.This calculation accounts for any losses in fluorescence bynormalizing the fluorescence of the cell of interest to thatof a neighboring cell of approximately equal size and fluorescentintensity. Nuclear FLIPs were performed by repetitive bleachingof a small nuclear area and imaging at 15-second intervals.Loss of nuclear fluorescence was quantified using the aboveequation; however, the value for I_t in this case was that ofthe bleached nucleus. For fluorescence recovery after photobleaching(FRAP) experiments, a large nuclear region was photobleachedonce with three rapid pulses and images were collected every5 seconds. Recovery of the fluorescent signal within the bleachedregion was calculated as described by Phair and Misteli (51)and the following equation: I_rel = (I_t/I₀)·(T₀/T_t), whereT_t is the total cellular intensity at time t, T₀ is the totalcellular intensity before bleaching, I₀ is the intensity inthe bleached area before bleaching, and I_t is the intensityof the previously bleached region at time t. Pseudocolor imageswere generated to highlight differences in GFP fluorescence,with red representing high fluorescent intensity and light bluerepresenting low fluorescent intensity. The quantification graphicwas generated by using FLIP/FRAP software. For all bleachingexperiments, 10 data sets were analyzed for each result. Pseudocoloringfor bleaching experiments was achieved by applying the gradientmap function of Photoshop (Adobe) to a montage of picture framesprepared with ImageJ software (National Institutes of Health,Bethesda, MD). The Northern Eclipse (Empix), Excel (Microsoft),and FreeHand (Macromedia) software packages were also used tocapture images, analyze the data, and generate graphs.

Polykaryon assay. MCF-7 cells were transfected to express fluorescently labeledproteins and incubated under standard conditions for 24 h. Usually,40 to 60% of cells presented strong fluorescence as observedby 488-nm fluorescence microscopy. Cells were trypsinized 24h after transfection and mixed with untransfected NIH 3T3 cellsin a ratio of 1 to 10. The cell mixture was plated in 35-mmdishes with coverslip bottoms and incubated overnight understandard cell culture conditions. The confluent cell layer wasvisually inspected for even distribution of fluorescent cellsamong untransfected cells. Cells were washed twice with prewarmedphosphate-buffered saline (PBS) and fused for 2 min by additionof a prewarmed 50% solution of polyethylene glycol (PEG) inPBS (Sigma-Aldrich). PEG was removed thoroughly by four washeswith prewarmed PBS, and cells were then replenished with warmedstandard cell culture medium. Hoechst staining of DNA was usedto identify donor and acceptor cells. Cells were observed underphase-contrast microscopy for fusion events and were monitoredfor the redistribution of nuclear expression of PK-GFP-NLS-taggedproteins.

In vitro nuclear export assay. The in vitro export assay was performed as described by Groulxet al. (16). Briefly, cells were plated and grown on a 35-mmcoverslip plate. Cells were washed with transport buffer (TB)containing 20 mM HEPES pH 7.3, 110 mM KO-acetate (KOAc), 5 mMNaOAc, 2 mM Mg(OAc)₂ and permeabilized at 4°C for 5 minwith TB containing 50 µg/ml digitonin and a protease inhibitormixture (Hoechst stain 33258 [Sigma] was used to monitor thepermeabilization). After several washes with TB at 4°C,cells were incubated for 30 to 45 min at 20°C in the presenceof a standard mixture that included TB, 2 mM ATP, 2 mM GTP,and an ATP-regenerating system (5 mM creatine phosphate and20 units/ml creatine phosphokinase). Where indicated, MCF-7total cell lysate was added to the standard mixture.

Immunofluorescence. Cells were seeded onto coverslips and fixed with prechilledmethanol for 10 min at –20°C followed by prechilledacetone for 1 min at –20°C. Anti-PABP1 monoclonalantibody was used (Upstate). Cells were incubated for 1 h witha primary antibody solution containing 10% (vol/vol) fetal bovineserum and 1% (vol/vol) Triton X-100 at room temperature in ahumidified chamber. Cells were then washed several times inPBS before a 1-h incubation with a secondary Texas Red-labeledantibody (Jackson ImmunoResearch) at room temperature in a darkhumidified chamber. Hoechst stain 33342 (Sigma) was added tovisualize nuclei, and coverslips were mounted using FluoromountG (EMS).

Bioinformatic analysis. Proteins containing a TD-NEM were identified using the emotifand My Genomics Resource Center software.

Immunoprecipitation and immunoblotting. Cells were lysed in lysis buffer containing 0.5% Igepal CA630,100 mM NaCl, 20 mM Tris-HCl (pH 7.6), 5 mM MgCl₂, and 1 mM sodiumorthovanadate with 2 µg/ml leupeptin, 2 µg/ml aprotinin,and 1 µg/ml pepstatin. Cell lysates were incubated withanti-Flag M2 beads (Sigma) overnight while tumbling at 4°C.Beads were washed several times and eluted with Flag peptides(Sigma). For total cell lysates, cells were washed several timesin PBS, lysed with 4% sodium dodecyl sulfate (SDS) in PBS, andboiled for 5 min, and the DNA was sheared by passage througha 19-gauge needle. The protein concentration was quantifiedusing the bicinchoninic acid method (Pierce). Samples were separatedon denaturing polyacrylamide gels. Western blots gels were transferredonto polyvinylidene difluoride membranes and blocked in skimmedmilk powder in PBS containing 0.2% Tween 20 before incubationwith Flag-M2 (Sigma), HIF2 ${alpha}$ (Novus Biologicals), actin (Sigma),and GFP (AVES Lab Inc.) antibodies. Membranes were washed with0.2% Tween-PBS and blotted with a secondary antibody conjugatedto horseradish peroxidase (Jackson ImmunoResearch Laboratories)and detected by using Western Lightning chemiluminescence reagentplus (Perkin-Elmer).

RESULTS

Top

RESULTS

Nuclear export of VHL is mediated by a discreet transcription-dependent motif. VHL engages in a nuclear export pathway that requires ongoingRNA Pol II activity (17, 34). As previously reported, additionof the RNA Pol II inhibitors ActD or DRB resulted in nuclearaccumulation of the highly cytoplasmic VHL-GFP but had no effecton GFP alone or when fused to a strong, classical, leucine-richnuclear export signal (see Materials and Methods) (Fig. 1A)(17, 34). In contrast, LMB, a compound that abolishes CRM1-dependentnuclear export of NES-containing proteins (10, 14, 31, 57),had no effect on the steady-state localization of VHL-GFP butcaused nuclear accumulation of NES-GFP (Fig. 1A). Nuclear accumulationby ActD was due to a considerable decrease in the rate of nuclearexport of VHL-GFP, as demonstrated by cytoplasmic FLIP experiments(Fig. 1B; see also the quantification in panel C) and by cellularfusion experiments (data not shown) (34). In the cytoplasmicFLIP experiments, a small region of the cytoplasm is repetitivelybleached (Fig. 1B), and a decrease in nuclear GFP fluorescenceindicates that the GFP-tagged fusion protein has been exportedfrom the nucleus to the cytoplasm. Furthermore, nuclear FLIPrevealed similar intranuclear dynamics between ActD-treatedand untreated cells, demonstrating that ActD (or DRB) did notcause nuclear retention of VHL-GFP (Fig. 1D; see also Fig. S1in the supplemental material). Loss of GFP fluorescence in theFLIP experiments was not due to fragmentation of the GFP-fusionproteins, as demonstrated by immunoblot analysis (Fig. 1F).Next, it was important to determine whether the signal mediatingnuclear export was encoded by VHL or was present within theE3 ubiquitin ligase complex. We also tested a C-terminal deletionof the ${alpha}$ -domain of VHL ( ${Delta}$ C157), which is required for assemblywith core E3 ubiquitin ligase components elongins B and C andCullin 2 (4, 46). ${Delta}$ C157 retained the ability to engage in transcription-dependentnuclear export, since it accumulated in the nucleus upon additionof ActD or DRB (data not shown) and displayed reduced nuclearexport dynamics in the presence of RNA Pol II inhibitors asdemonstrated by cytoplasmic FLIP analysis (Fig. 1E). These dataindicate the presence of a transcription-dependent nuclear exportsignal between residues 1 and 157 of VHL and suggest that formationof the E3 ubiquitin ligase complex is not required for exportof VHL.

View larger version (43K):
[in this window]
[in a new window]

FIG. 1. Transcription-dependent nuclear export of VHL. (A) ActD alters the steady-state localization of VHL. MCF-7 cells transiently expressing VHL-GFP, NES-GFP, or GFP were either untreated or treated with LMB (10 µM), ActD (8 µM), or DRB (25 µg/ml). Insets are the corresponding Hoechst staining of the cells. Bars, 10 µm. (B and C) Cytoplasmic FLIP reveals that ActD decreases nuclear export of VHL. MCF-7 cells transiently expressing VHL-GFP were treated with ActD (2 µM) or LMB (10 µM) for 1 h. Cells were initially bleached in a large cytoplasmic region (dashed squares) to reduce cytoplasmic signal and then submitted to repetitive bleaching in a small cytoplasmic region (white squares). Cells were imaged between pulses, and the corresponding kinetics for loss of nuclear fluorescence were calculated and plotted on a graph (C) (see Materials and Methods). Bar, 10 µm. (D) Nuclear FLIP analysis was performed on cells treated as for panel B by repetitive bleaching in a small area in the nucleus. The loss of nuclear fluorescence was graphed. (E) ${Delta}$ C157 exports from the nucleus in a transcription-dependent manner. ${Delta}$ C157-GFP-transiently expressing cells were treated and analyzed as described for panel B. (F) Western blot analysis using anti-Flag and anti-GFP antibodies verified that the Flag- and GFP-tagged VHL, ${Delta}$ C157, and GFP fusion proteins used for photobleaching experiments were not fragmented.

Larger GFP-tagged truncation mutants of VHL failed to provideinsight on transcription-dependent nuclear export of VHL, sincethey exhibited export rates similar to GFP-GFP, which diffusesacross the nuclear envelope (Fig. 1F). To further examine thenuclear export properties of VHL, we established a quantitativelive cell nuclear export assay utilizing FLIP technology. Afusion protein consisting of the large and amorphous PK, GFP,and an NLS (29) derived from the simian virus 40 large T antigen(PK-GFP-NLS) was used to ensure that any movement across thenuclear envelope was not due to diffusion (Fig. 2A) (see Materialsand Methods). Also, the presence of such a strong NLS wouldoverride an intrinsic nuclear import signal, ensuring that allfusion proteins being tested exhibited similar rates of nuclearimport. PK-GFP-NLS strictly localized in the nucleus at steadystate, owing to the strong NLS activity (Fig. 2B). PK-GFP-NLSdid not undergo a significant loss of nuclear GFP fluorescenceafter cytoplasmic FLIP, as expected, since this protein failsto export from the nucleus (Fig. 2B and E). Addition of a strong,classical, and LMB-sensitive NES of the human immunodeficiencyvirus Rev (9) to the reporter protein (NES-PK-GFP-NLS) resultedin a rapid loss of nuclear GFP fluorescence during cytoplasmicFLIP, demonstrating that this fusion protein engages in nuclearexport (Fig. 2B and E). The inability of PK-GFP-NLS to exportfrom the nucleus compared to its NES-containing counterpartis not a consequence of unforeseen nuclear retention or a significantdifference in nuclear dynamics between the two molecules asrevealed by nuclear FLIP (see Fig. S2 in the supplemental material).Nuclear fluorescence was also similar for both fusion proteins,indicating that the difference in nuclear export capacity isnot due to a large difference in expression levels (see Fig.S2 in the supplemental material). Full-length VHL was able toconfer nuclear export activity to the PK-GFP-NLS reporter, confirmingthat this protein encodes nuclear export activity (Fig. 2C and E).The kinetics for the loss of nuclear signal of VHL-PK-GFP-NLSwas very different compared to molecules that are able to passivelydiffuse, suggesting that the loss of nuclear signal is not aconsequence of fragmentation of the fusion protein (see Fig.S2 in the supplemental material). ActD or DRB, but not LMB,inhibited nuclear export of VHL-PK-GFP-NLS (Fig. 2F; see alsoFig. S2 in the supplemental material) without affecting intranucleardynamics (see Fig. S2 in the supplemental material), similarto what was observed with VHL-GFP (Fig. 1) and confirming thatinhibitors of RNA Pol II-mediated transcription significantlyalter nuclear export of VHL. Furthermore, these results verifythat the transcription-dependent nuclear export property ofVHL was maintained in the fusion protein. Removal of the exon-2-encodedβ-domain ( ${Delta}$ 114-154-PK-GFP-NLS) abrogated the ability ofVHL to export the reporter protein from the nucleus (Fig. 2D and E).This observation is consistent with previous reports demonstratingthat nuclear accumulation of VHL upon treatment with RNA PolII inhibitors requires the exon-2-encoded β-domain of VHL(4, 34).

View larger version (35K):
[in this window]
[in a new window]

FIG. 2. A transcription-dependent nuclear export sequence is encoded within the exon-2-encoded β-domain of VHL. (A) Schematic diagram of the PK-GFP-NLS fusion protein used for the live cell FLIP nuclear export assay, describing the region where protein or peptide sequences were fused. (B to E) MCF-7 cells transiently expressing the indicated constructs were submitted to cytoplasmic FLIP analysis, in which a small cytoplasmic region (white squares) within specific cells (a dashed circle outlines the cell nucleus) was repeatedly bleached. Kinetics for the loss of nuclear fluorescence from images obtained in panels B, C, and D were calculated and plotted on a graph (E). PK refers to the PK-GFP-NLS reporter construct, and NES, VHL, and ${Delta}$ 114-254 indicate the sequences fused to PK-GFP-NLS. Bar, 10 µm. (F) Cells were transiently transfected with VHL-PK-GFP-NLS and were either treated with ActD (2 µM), DRB (25 µg/ml), or LMB (10 µM) for 1 h or left untreated. Cytoplasmic FLIP was performed as described above to verify nuclear export activity.

We used this in vivo FLIP nuclear export assay to map the regionwithin the exon-2-encoded β-domain that mediates transcription-dependentnuclear export of VHL. Further mutagenesis analysis revealedthat a discrete domain between residues 114 and 131 of VHL isrequired for efficient nuclear export of the PK-GFP-NLS reporter(Fig. 3A and B; see also Fig. S3 in the supplemental material).The deletion mutant ${Delta}$ 114-131-GFP (without PK and NLS) also displayeda markedly reduced rate of nuclear export compared to wild-typeVHL-GFP (Fig. 3C; see also Fig. S3 in the supplemental material).The subcellular localization and the nuclear export dynamicsof ${Delta}$ 114-131-GFP were unaffected by treatment with ActD or DRB(Fig. 3C and data not shown; see also Fig. S3 in the supplementalmaterial) but rather exhibited a similar export rate to ${Delta}$ C157-GFPafter ActD treatment (Fig. 1E). More importantly, residues 114to 138 alone were sufficient to confer efficient nuclear exportproperties to PK-GFP-NLS, which was abolished upon additionof ActD (Fig. 4A and B) or DRB (data not shown). The rate ofexport observed with residues 114 to 138 was similar to thatof ${Delta}$ C157, indicating that these residues alone confer nuclearexport activity to the full-length protein (Fig. 4B). In a cellularfusion assay, this sequence was able to efficiently export thePK-GFP-NLS reporter protein from the donor nucleus to the acceptornuclei compared to the PK-GFP-NLS control (Fig. 4C). Residues114 to 138 also stimulated nuclear export of PK-GFP-NLS in anin vitro nuclear export assay (Fig. 4D and E). We have, therefore,identified a novel and discreet motif that mediates transcription-dependentnuclear export of VHL.

View larger version (30K):
[in this window]
[in a new window]

FIG. 3. Residues 114 to 131 are required for nuclear export of VHL. (A) Map of the nuclear export domain of VHL. The schematic diagram indicates deletion mutants of VHL that were submitted to cytoplasmic FLIP to assess the nuclear export activity. + and - indicate the ability or inability of the fusion protein to engage in nuclear export. (B) MCF-7 cells transiently transfected with PK-GFP-NLS, ${Delta}$ C157-PK-GFP-NLS, or ${Delta}$ 114-131-PK-GFP-NLS were submitted to cytoplasmic FLIP, in which a small cytoplasmic region of a cell was bleached repetitively. The loss of nuclear GFP fluorescence was monitored over time and plotted on a graph. (C) Cells transiently expressing ${Delta}$ 114-131-GFP were initially bleached in a large cytoplasmic region followed by repetitive bleaching in a small cytoplasmic region after being treated for 1 h with 2 µM ActD or left untreated. Kinetics for the loss of nuclear fluorescence were calculated and plotted on a graph.

View larger version (48K):
[in this window]
[in a new window]

FIG. 4. Residues 114 to 138 mediate transcription-dependent nuclear export of VHL. (A and B) MCF-7 cells were transiently transfected to express VHL(114-138)-PK-GFP-NLS. Cells were treated with a final concentration of 8 µM ActD or left untreated for 1 h before being subjected to photobleaching. Cytoplasmic FLIP was performed by repetitively bleaching a small cytoplasmic region (white squares) of a cell (the dotted circle outlines the cell nucleus). Cells were imaged between pulses, and the corresponding kinetics for the loss of nuclear fluorescence were calculated and are graphed (B). The nuclear export kinetics of ${Delta}$ C157-PK-GFP-NLS revealed by a cytoplasmic FLIP analysis were plotted (B). Bar, 10 µm. (C) VHL(114-138) can mediate export in a polykaryon fusion assay. MCF-7 cells (donors) were fused with NIH 3T3 cells (acceptors) using PEG, and the transfer of nuclear fluorescence from donor to acceptor cells was monitored. Donor and acceptor cells were differentiated by a cell-specific Hoechst staining pattern. White arrows indicate donor cells, and yellow arrows indicate acceptor cells. Bar, 10 µm. (D and E) VHL(114-138) can mediate export in an in vitro nuclear export assay. MCF-7 cells transiently expressing the indicated constructs were permeabilized with digitonin, after which they were incubated with transport buffer containing ATP, GTP, and an ATP-regenerating system in the presence of buffer or MCF-7 cell lysate. Relative loss in nuclear fluorescence was calculated and plotted on a graph (E).

VHL and PABP1 share a common transcription-dependent nuclear export motif. VHL and PABP1 are implicated in distinct molecular networksbut share the characteristic of engaging in a nuclear exportpathway that requires ongoing RNA Pol II activity (1, 34). Similarto VHL, the subcellular trafficking dynamics of the highly cytoplasmicPABP1 are affected by treatment with ActD (Fig. 5A and B), butnot LMB (Fig. 5A) (1). Sequence alignment revealed a strikingsimilarity between residues 114 to 138 of VHL and 296 to 317of PABP1 that consisted of DXGX₂DX₂L (Fig. 5C). Consistent witha possible role in nuclear export, removal of residues 296 to317 from full-length PABP1 caused a drastic shift in the steady-statelocalization from exclusively cytoplasmic to highly nuclear(Fig. 5D). This deletion mutant of PABP1 was also insensitiveto ActD treatment compared to wild type (Fig. 5D). Residues296 to 317 of PABP1 were sufficient to mediate nuclear exportof the PK-GFP-NLS reporter as efficiently as residues 114 to138 of VHL in our FLIP nuclear export assay (Fig. 5E; see alsoFig. S4 in the supplemental material). In addition, ActD orDRB, but not LMB, abolished the ability of residues 296 to 317of PABP1 to mediate nuclear export of the PK-GFP-NLS reporter(Fig. 5E; see also Fig. S4 in the supplemental material). Residues296 to 317 of PABP1 were able to efficiently export the reporterprotein from the donor nucleus to the acceptor nuclei in a cellularfusion assay (Fig. 5F), similar to results obtained with residues114 to 138 of VHL (Fig. 4C). This led us to postulate that theDXGX₂DX₂L consensus sequence might act as a common motif fortranscription-dependent nuclear export of proteins. A searchof sequence data banks identified several proteins that containa DXGX₂DX₂L motif. We randomly selected the cell cycle regulator,cyclin C, whose nuclear-cytoplasmic shuttling profile had notbeen previously investigated, to test the activity of its potentialnuclear export sequence. Full-length cyclin C was fused to PK-GFP-NLSto prevent any passive diffusion through nuclear pores. CyclinC was able to confer nuclear export activity to the PK-GFP-NLSreporter in a process that was inhibited by addition of ActD(Fig. 5G). Furthermore, residues 158 to 179 of cyclin C, whichencode the DXGX₂DX₂L motif, were able to export the reporterPK-GFP-NLS from the nucleus in an ActD-sensitive manner (Fig.5H). Further characterization of cyclin C, and other proteins,will reveal whether the DXGX₂DX₂L motif is involved in theirsubcellular trafficking profiles. Nonetheless, we have identifieda novel motif that mediates efficient transcription-dependentnuclear export of VHL, PABP1, and perhaps cyclin C. For thesake of simplicity, the DXGX₂DX₂L motif is referred to as aTD-NEM.

View larger version (55K):
[in this window]
[in a new window]

FIG. 5. VHL and PABP1 share a common transcription-dependent nuclear export motif. (A) PABP1 exports by a transcription-dependent mechanism. MCF-7 cells transiently expressing PABP1-GFP, NES-GFP, or GFP alone were either untreated or treated with 8 µM ActD or 10 µM LMB for 3 h. Insets are the corresponding Hoechst staining of the cells. Bars, 10 µm. (B) Endogenous PABP1 is also sensitive to ActD treatment. MCF-7 cells were either left untreated or were treated with 8 µM ActD. Endogenous PABP1 was detected by immunofluorescence using an anti-PABP1 antibody. Insets show nuclei stained with Hoechst stain. A primary antibody exclusion (no 1^ary Ab) control is also shown. Bar, 10 µm. (C) Schematic diagram depicting a region of alignment between the nuclear export sequence of VHL and PABP1. Conserved residues are indicated in red. (D) MCF-7 cells transiently expressing PABP1( ${Delta}$ 296-317)-GFP or PABP1-GFP were treated as described for panel A. Insets are the corresponding Hoechst staining of the cells. Bar, 10 µm. (E) Residues 296 to 317 of PABP1 encode a transcription-dependent nuclear export sequence. Transfected MCF-7 cells were treated with 2 µM ActD or 10 µM LMB for 1 h before being submitted to cytoplasmic FLIP. The corresponding loss of nuclear fluorescence was monitored, measured, and plotted on a graph. (F) PABP1(296-317) can mediate export in a polykaryon fusion assay. MCF-7 cells (donors) were fused with NIH 3T3 cells (acceptors) using PEG, and the transfer of nuclear fluorescence from donor to acceptor cells was monitored. Donor and acceptor cells were differentiated by a cell-specific Hoechst staining pattern. White arrows indicate donor cells, and yellow arrows indicate acceptor cells. Bar, 10 µm. (G) Full-length cyclin C can export the PK-GFP-NLS reporter in a transcription-dependent manner. Cells were treated with 2 µM ActD for 1 h or untreated, and the loss of nuclear fluorescence after cytoplasmic FLIP was plotted on a graph. (H) Residues 158 to 179 of cyclin C encode a transcription-dependent nuclear export motif. The indicated constructs were transiently expressed in MCF-7 cells and treated the same as for panel G. Cells were subjected to cytoplasmic FLIP, and the corresponding kinetics for loss of nuclear fluorescence were plotted on a graph.

Conserved residues of TD-NEM are essential for mediating efficient nuclear export of proteins. Analysis of the TD-NEM sequences from VHL, PABP1, and cyclinC failed to identify additional conserved residues surroundingthe basal DXGX₂DX₂L motif (Fig. 6A), suggesting that these fouramino acids may encode transcription-dependent nuclear exportactivity. Replacement of the four key residues with alaninesabolished nuclear export activity of residues 114 to 138 fromVHL (Fig. 6A and B; see also Fig. S5 in the supplemental material).Replacement of the aspartic acid, leucine, or glycine residueswith alanine reduced nuclear export activity by approximately50 to 70% compared to the basal sequence (Fig. 6A and C; seealso Fig. S5 in the supplemental material). In addition, replacingglycine 123 with alanine in full-length VHL, tagged to GFP andnot PK-GFP-NLS, shifted the steady-state localization towardsa more predominantly nuclear localization (Fig. 6D) and reducedits ability to export from the nucleus (Fig. 6E), further demonstratingthat DxGx₂Dx₂L is a nuclear export motif. These data demonstratethat all four conserved residues in TD-NEM play a central rolein mediating nuclear export.

View larger version (20K):
[in this window]
[in a new window]

FIG. 6. Identification of key residues that mediate transcription-dependent nuclear export. (A) Sequence alignment depicting conserved residues (blue) in the transcription-dependent nuclear export motifs of VHL, PABP1, and cyclin C. Residues in red indicate alanine substitutions at key residues in the TD-NEM sequence of VHL. (B) Conserved residues in the DXGX₂DX₂L consensus sequence are essential for nuclear export. MCF-7 cells were transiently transfected with VHL(115-130)-PK-GFP-NLS (DxGx₂Dx₂L) or VHL(115-130AAAA)-PK-GFP-NLS (AxAx₂Ax₂A), where the four key residues were replaced with alanines. Cytoplasmic FLIP was performed, and the loss of nuclear fluorescence was plotted on a graph. (C) Single alanine substitutions of key residues within the DXGX₂DX₂L consensus sequence have a differential effect on nuclear export activity. MCF-7 cells transiently expressing the indicated PK-GFP-NLS-tagged constructs were subjected to cytoplasmic FLIP. The graph represents the loss of nuclear fluorescence. (D and E) A single amino acid substitution of G123A in full-length VHL affects steady-state localization and nuclear export activity. Steady-state localization of VHL(G123A)-GFP expressed in MCF-7 cells compared to wild-type VHL-GFP is shown (D). Insets are the corresponding Hoechst staining of the cells. Cells expressing VHL-GFP or VHL(G123A)-GFP were subjected to cytoplasmic FLIP, and the loss of nuclear fluorescence was monitored and graphed (E).

Cancer-causing mutations within TD-NEM abrogate nuclear export of VHL and oxygen-dependent degradation of HIF ${alpha}$ . To study the functional consequence of TD-NEM mutations in thecontext of a full-length VHL protein, we searched for naturallyoccurring point mutations within the key TD-NEM residues thatare associated with VHL disease. Replacement of the first asparticacid residue of the consensus, DXGX₂DX₂L, with glycine (D121G)is a germ line mutation associated with type 2B VHL disease,characterized by a high risk of RCC (Fig. 7A) (15, 52, 61).This mutant is of particular interest, since previous studieshave reported that it maintains its ability to form an E3 ubiquitinligase complex and to bind and ubiquitylate HIF ${alpha}$ in vitro (19).In addition, the second aspartic acid residue of the VHL TD-NEMhas been reported to be replaced with tyrosine (D126Y) in individualsafflicted with polycythemia (Fig. 7A) (48, 49). Stable celllines of D121G-GFP, D126Y-GFP, and G123A-GFP (not PK-GFP-NLS)were generated in VHL-defective 786-0 RCC to further study theeffect of these point mutants on the ability of VHL to be exportedfrom the nucleus and mediate oxygen-dependent degradation ofHIF ${alpha}$ . As previously reported, immunoprecipitation analysis revealedthat the VHL mutant D121G is able to bind as efficiently toHIF2 ${alpha}$ as wild-type VHL (Fig. 7B). Likewise, VHL D126Y was alsoable to efficiently bind HIF2 ${alpha}$ (Fig. 7B). Cytoplasmic FLIP experimentsrevealed a markedly decreased rate of nuclear export of D121Gand D126Y, of approximately 40% and 20%, respectively, comparedto wild-type VHL (Fig. 7C). The function of VHL in oxygen-dependentdegradation of HIF ${alpha}$ has been linked to its ability to exportfrom the nucleus (17). Thus, we decided to examine the effectof nuclear export-defective VHL mutants on oxygen-dependentdegradation of HIF ${alpha}$ . Cells were exposed to hypoxia to promoteHIF ${alpha}$ accumulation, followed by reoxygenation (Fig. 7D). VHL-defectiveRCC 786-0 cells did not display a decrease in HIF2 ${alpha}$ levels (786-0cells express HIF2 ${alpha}$ but not HIF1 ${alpha}$ ) following reoxygenation ofhypoxic cells, as expected, since these cells do not expresswild-type VHL (Fig. 7D). Reintroduction of wild-type VHL causedrapid degradation of HIF2 ${alpha}$ upon reoxygenation of hypoxic cells(Fig. 7D). In contrast, both the D121G and D126Y mutants werenot as efficient in degrading HIF2 ${alpha}$ in reoxygenated cells (Fig.7D). Mutants D121G and D126Y were eventually capable of degradingHIF2 ${alpha}$ after long periods of reoxygenation, consistent with theirability to assemble with HIF ${alpha}$ and mediate ubiquitylation (Fig.7E). These data support the hypothesis that VHL mutants thatare defective in nuclear export display a reduced ability todegrade HIF ${alpha}$ even though they are able to bind to and ubiquitylateHIF ${alpha}$ .

View larger version (59K):
[in this window]
[in a new window]

FIG. 7. Cancer-causing mutations within TD-NEM of VHL abrogate nuclear export and oxygen-dependent degradation of HIF2 ${alpha}$ . (A) Sequence alignment depicting cancer-causing mutations (red) in the key TD-NEM residues of VHL that lead to RCC type 2B (D121G) and polycythemia (D126Y) in humans. G123A, which exhibits a defect in nuclear export as shown in Fig. 6, was also used to test its ability to mediate HIF degradation. (B) D121G and D126Y retain the ability to bind to HIF. Cells stably expressing Flag-tagged VHL-GFP, D121G-GFP, D126Y-GFP, and GFP were placed under hypoxic conditions and treated with 10 µM MG132 for 2 h before being harvested. Cell lysates were immunoprecipitated with anti-Flag beads and immunoblotted with anti-Flag and anti-HIF2 ${alpha}$ antibodies. (C) Cancer-causing mutants D121G and D126Y in TD-NEM decrease the nuclear export activity of VHL. Cells stably expressing VHL-GFP, D121G-GFP, D126Y-GFP, or G123A-GFP or transiently expressing PK-GFP-NLS were submitted to cytoplasmic FLIP as described previously. The loss of nuclear fluorescence was monitored and plotted on a graph. (D and E) Cells expressing D121G or D126Y exhibit a deficiency in HIF degradation. Stable cell lines of VHL-GFP, D121G-GFP, D126Y-GFP, and the VHL-defective cell line 786-0 were incubated for 20 h under hypoxic conditions before being reoxygenated by placing them in a normoxic environment, for the indicated time. Cells were lysed with 4% SDS and submitted to Western blot analysis using an anti-HIF2 ${alpha}$ antibody. Levels of VHL or its mutant counterparts were monitored using anti-Flag antibody, and actin was used to ensure equal loading of lysates. (F) G123A retains the ability to bind to HIF. Cells stably expressing the VHL point mutant, G123A, were treated the same as described for panel B. (G and H) Cells expressing G123A exhibit a deficiency in HIF degradation. G123A stably expressing cells were treated in the same manner as described for panels D and E. (I) D121G, D126Y, and G123A stable cells express higher normoxic HIF levels compared to wild-type VHL. Stable cell lines incubated under normoxic conditions were lysed with 4% SDS and submitted to Western blot analysis using anti-HIF2 ${alpha}$ and antiactin antibodies. HIF2 ${alpha}$ levels were normalized to actin and values, calculated relative to HIF2 ${alpha}$ levels in 786-0 cells, were plotted on a graph.

In light of these results, we were interested in testing ifthe nuclear export-defective G123A, although not a known disease-causingmutation, would also exhibit a decreased efficiency in degradingHIF ${alpha}$ (Fig. 7A). Stably expressed G123A maintained its abilityto interact with HIF2 ${alpha}$ (Fig. 7F) but displayed a reduced nuclearexport activity (Fig. 7C). The G123A mutant displayed a strikingdefect in degradation of HIF2 ${alpha}$ upon reoxygenation of hypoxiccells (Fig. 7G and H). We noticed that 786-0 cells expressingthe nuclear export-defective mutants D121G, D126Y, and G123Agenerally displayed higher levels of HIF2 ${alpha}$ than wild-type VHL,but levels were still lower than with the VHL-defective 786-0cells (Fig. 7I). This may be explained by the fact that thenuclear export-defective mutants are still able to bind andubiquitylate HIF ${alpha}$ and partially able to be exported from thenucleus. Nonetheless, the data shown in Fig. 7 suggest thatnuclear export of VHL is required for efficient oxygen-dependentdegradation of HIF ${alpha}$ .

TD-NEM mediates efficient transcription-dependent nuclear export of proteins. Efficient nuclear export of proteins is required for cellularhomeostasis and survival. The classical NES/CRM1 pathway wasthe first general and discreet motif identified which mediatesnuclear export of a wide array of different proteins (9, 60).The herein-described TD-NEM is a discreet motif that is sensitiveto drugs that inhibit RNA Pol II activity but operates independentlyof the classical NES pathway, since it is insensitive to LMB,as demonstrated in our live cell nuclear export assay (Fig.8A and B). This is in contrast to what is observed with theclassical NES, whose activity is inhibited by LMB but not byRNA Pol II inhibitors (Fig. 8C and D). VHL and PABP1 TD-NEMdisplay nuclear export activities nearly as efficient as theRev NES, which is thought to be the strongest NES yet characterized(Fig. 8E) (21). Based on these results, we suggest that TD-NEMis an efficient nuclear export motif.

View larger version (56K):
[in this window]
[in a new window]

FIG. 8. TD-NEM is a novel and efficient transcription-dependent nuclear export motif. (A to D) TD-NEM, contrary to the classical NES, mediates nuclear export in an ActD-sensitive but LMB-insensitive manner. Cells transiently expressing PK-GFP-NLS-tagged TD-NEM of VHL or NES of the Rev protein were either untreated or treated with ActD or LMB as previously described before being submitted to cytoplasmic FLIP. White squares indicate the bleached area, and the dotted circles outline the nucleus of the cell of interest. The loss of nuclear fluorescence was monitored, calculated, and plotted on a graph (B and D). (E) TD-NEM is an efficient nuclear export motif. Cells transiently expressing PK-GFP-NLS-tagged TD-NEM of VHL or PABP1, or the Rev NES, one of the strongest export signals, were submitted to cytoplasmic FLIP in order to compare the efficiency of nuclear export. (F) Model of nuclear export mediated by the TD-NEM consensus sequence compared to the classical NES (3). Nuclear export of TD-NEM is abrogated by RNA Pol II inhibitors, such as ActD and DRB, whereas NES activity is abrogated by LMB.

DISCUSSION

Top

DISCUSSION

We report the identification of a novel and discreet motif,DXGX₂DX₂L, referred to as TD-NEM, which mediates efficient transcription-dependentnuclear export of VHL. This motif was also common among otherproteins, including PABP1, where it mediated its respectivetranscription-dependent export from the nucleus. Nuclear exportby TD-NEM requires ongoing RNA Pol II-mediated transcriptionand operates independently of the classical CRM1/NES-mediatedexport pathway (Fig. 8F). Key residues within TD-NEM of VHLare targeted by naturally occurring point mutations associatedwith renal carcinoma and polycythemia in humans. These disease-causingmutations, which alter the dynamic profile of VHL, restrainoxygen-dependent degradation of HIF ${alpha}$ without affecting interactionbetween substrate and E3 ubiquitin ligase. Our results highlightthe essential role of nuclear-cytoplasmic dynamics in proteinfunction and provide evidence that mutations targeting subcellulartrafficking can lead to disease.

It has become increasingly evident that degradation of nuclearproteins by the ubiquitylation pathway requires nuclear-cytoplasmictrafficking of the E3 ubiquitin-ligase as well as the substrateprotein. Efficient degradation of nuclear proteins, such asp53, Smad3, and HIF ${alpha}$ , is tightly linked to the ability of theE3 ligase to engage in nuclear export (13, 17, 34, 37). Thus,VHL, the ubiquitylation component of an E3 ubiquitin ligasecomplex that mediates oxygen-dependent degradation of HIF ${alpha}$ , wasan ideal subject for studying the functional role of nuclear-cytoplasmictrafficking of E3 ligases. We have previously shown that nuclear-cytoplasmictrafficking of VHL requires ongoing RNA Pol II-mediated transcription(17, 34). The initial goal of this study was to identify thesequence that mediates transcription-dependent nuclear exportof VHL. We stumbled upon a sequence that may potentially mediatenuclear export of several proteins, including the mRNA nuclearexport factor PABP1. VHL and PABP1 share the distinct abilityto engage in constitutive and highly dynamic nuclear-cytoplasmictrafficking, utilizing a pathway that requires ongoing RNA PolII activity. We have identified a new functional domain, TD-NEM,which is present in both VHL and PABP1 and mediates their transcription-dependentexport from the nucleus. The rate of nuclear export of TD-NEMis approximately 70 to 80% of that observed for the Rev NES,perhaps the strongest LMB-sensitive NES so far identified, suggestingthat this motif is highly efficient in mediating nuclear egressionof molecules. Several lines of evidence support a role for TD-NEMas a nuclear export motif. First, removal of TD-NEM from full-lengthVHL and PABP1, or a single amino acid substitution of glycineto alanine, altered the steady-state distribution from mostlycytoplasmic to nuclear, accompanied by a reduction in sensitivityto ActD or DRB treatment and the rate of nuclear export. Also,TD-NEM alone from VHL and PABP1 was sufficient to confer transcription-dependentnuclear export properties to a large reporter protein (i.e.,PK-GFP) in multiple living cells and in vitro assays. Basedon these data, we suggest that TD-NEM represents a new, andpossibly ubiquitous, nuclear export motif that operates independentlyof the known CRM1-NES pathway.

In this study we have shown that transcription-dependent nuclearexport of VHL and PABP1 is mediated by a simple and linear sequence,DXGX₂DX₂L. Each of the four conserved residues of TD-NEM wasfound to be required for full activity of the export signal.It is not surprising that a mere four residues can act as atransport signal, considering that the classical NES, amongothers, relies on a consensus sequence of only a few conservedresidues (9, 60). It would be of interest to examine the flexibilityof this sequence by testing if naturally occurring conservativepermutations of its key residues, such as aspartic acid to glutamicacid or leucine to isoleucine, retain activity. There were noother apparent conserved residues between the core TD-NEM ofVHL, PABP1, and the putative TD-NEM of cyclin C; however, thesethree independent TD-NEMs displayed slightly different nuclearexport activities, suggesting a functional role of the nonconservedresidues in modulating the activity of TD-NEM. These observationsare similar to the classical NES, which can display differentexport activities dependent on the surrounding amino acid context(32). Whether other substitutions or subtle differences of thisTD-NEM, such as different spacing between the key residues,retain activity also remains to be tested. If functional combinationsare found, this would simply provide additional evidence tosupport a ubiquitous nature of TD-NEM.

Discovery of LMB as an inhibitor of NES-mediated nuclear exporthas aided in both deciphering the components of this pathwayand identifying many proteins that are exported from the nucleus(10, 14). Similarly, our use of transcriptional inhibitors,such as ActD and DRB, has led to the discovery of a nuclearexport pathway that relies on the TD-NEM export signal. We envisionthat the discovery of TD-NEM through inhibitors of RNA Pol II-dependenttranscription will help uncover another class of proteins thatundergo nuclear export and the mechanism by which this occurs.It is still unclear as to how these drugs inhibit nuclear exportof TD-NEM-containing proteins, but it is most likely throughdirect or indirect effects on either primary or secondary componentsof this pathway. It is well known that LMB directly interactswith and inhibits the action of CRM1, the exporter of NES-containingproteins (31). It will be interesting to see if ActD and DRBhave a direct effect on a possible TD-NEM exporter or if theycause a secondary effect through transcriptional inhibition.

We have previously shown that perturbing the nuclear-cytoplasmictrafficking profile of VHL is detrimental to its ability tomediate oxygen-dependent degradation of HIF ${alpha}$ . Here we reportthat naturally occurring TD-NEM mutations D121G and D126Y, whichlead to RCC and polycythemia, respectively, abrogate nuclearexport of VHL. It has been puzzling as to how VHL cancer-causingmutations, such as D121G, retain the ability to bind and ubiquitylateHIF ${alpha}$ in vitro yet are able to develop classical tumors associatedwith VHL disease (19). We showed that D121G and D126Y maintainthe ability to bind to HIF2 ${alpha}$ , consistent with previously publisheddata. Expression of these mutants leads to an extended HIF2 ${alpha}$ stability following reoxygenation of hypoxic cells, providinga correlation between the efficiency to mediate degradationof HIF ${alpha}$ and nuclear export activity. We cannot exclude the possibilitythat other aspects of the ubiquitylation pathway may be affectedin vivo; however, we did predict a defect in HIF2 ${alpha}$ degradationwith G123A, a mutant that is restrained in its ability to exportfrom the nucleus but retains its ability to bind to HIF2 ${alpha}$ . TheG123A mutant further supports that the TD-NEM of the β-domainof VHL is not involved in HIF ${alpha}$ binding or E3 ubiquitin ligasecomplex formation but plays an essential role in HIF ${alpha}$ degradationby mediating nuclear export. We achieved low-expressing stablecell lines of the nuclear export-defective mutants, though expressionwas still higher than endogenous VHL. This raises the possibilitythat the nuclear export mutants may be partially rescued byoverproduction and may have more pronounced defects in physiologicalsettings. The differences in HIF ${alpha}$ stability observed with thenuclear export-defective mutants may translate into differenttypes of VHL disease, as observed with type 2A and type 2B mutantswith respect to HIF ${alpha}$ binding (36). Whether the prolonged presenceand activity of HIF ${alpha}$ provide an explanation for how patientswith these mutations develop tumors remains to be tested. Nonetheless,our data illustrate that the dynamic profile of a protein isinstrumental for functional integrity and provide evidence thatmutations targeting subcellular trafficking can abrogate proteinfunction.

In conclusion, we propose that TD-NEM is a novel nuclear exportmotif which utilizes a pathway that requires RNA Pol II-mediatedtranscription. There is emerging evidence that a large classof proteins is able to export by utilizing an NES-independentpathway. It would be interesting to test whether these proteinsencode a TD-NEM and if they engage in transcription-dependentnuclear export. Future work aimed at elucidating the mechanismby which RNA Pol II inhibitors abolish the nuclear export activityof TD-NEM should yield interesting information as to the processinvolved in TD-NEM-mediated nuclear export and perhaps uncoveranother general nuclear export pathway.

ACKNOWLEDGMENTS

This work is supported by a grant from the Canadian Institutesof Health Research to S.L. S.L. is the recipient of the NationalCancer Institute of Canada Harold E. Johns Award. K.M. is arecipient of a Canadian Institutes of Health Research fellowship.

FOOTNOTES

* Corresponding author. Mailing address: Department of Cellular and Molecular Medicine, Faculty of Medicine, University of Ottawa, 451 Smyth Road, Ottawa, Ontario, Canada K1H 8M5. Phone: (613) 562-5800, ext. 8385. Fax: (613) 562-5636. E-mail: slee{at}uottawa.ca

Published ahead of print on 29 October 2007.

Supplemental material for this article may be found at http://mcb.asm.org/.

Present address: Department of Cell Biology, Harvard University,Boston, MA 02115.

REFERENCES

Top