|Home | About | Journals | Submit | Contact Us | Français|
HIV-1 viral protein R (Vpr) induces cell cycle arrest at the G2/M phase by a mechanism involving the activation of the DNA damage sensor ATR. We and others recently showed that Vpr performs this function by subverting the activity of the DDB1-CUL4A (VPRBP) E3 ubiquitin ligase. Vpr could thus act as a connector between the E3 ligase and an unknown cellular factor whose ubiquitination would induce G2 arrest. While attractive, this model is based solely on the indirect observation that some mutants of Vpr retain their interaction with the E3 ligase but fail to induce G2 arrest. Using a tandem affinity purification approach, we observed that Vpr interacts with ubiquitinated cellular proteins and that this association requires the recruitment of an active E3 ligase given that the depletion of VPRBP by RNA interference or the overexpression of a dominant negative mutant of CUL4A decreased this association. Importantly, G2-arrest-defective mutants of Vpr in the C-terminal putative substrate-interacting domain displayed a decreased association with ubiquitinated proteins. We also found that the inhibition of proteasomal activity increased this association and that the ubiquitin chains were at least in part constituted of classical K48 linkages. Interestingly, the inhibition of K48 polyubiquitination specifically impaired the Vpr-induced phosphorylation of H2AX, an early target of ATR, but did not affect UV-induced H2AX phosphorylation. Overall, our results provide direct evidence that the association of Vpr with the DDB1-CUL4A (VPRBP) E3 ubiquitin ligase induces the K48-linked polyubiquitination of as-yet-unknown cellular proteins, resulting in their proteasomal degradation and ultimately leading to the activation of ATR and G2 arrest.
Viruses have evolved ways to modulate the host cellular environment in order to promote efficient viral replication and to disrupt elements of innate or acquired immunity. One strategy particularly favored by viruses to achieve these goals is to hijack components of the host ubiquitin-proteasome system in order to induce the degradation, block the degradation, or modulate the expression and activity of specific cellular factors (4, 8, 16). Human immunodeficiency virus (HIV) is no exception to this precept. HIV harbors two extensively studied accessory proteins, viral protein U (Vpu) and viral infectivity factor (Vif), that are usurping the cellular ubiquitin-proteasome system in order to degrade neosynthesized CD4 and the cytidine deaminases APOBEC3F (apolipoprotein B mRNA-editing enzyme, catalytic polypeptide-like 3F) and APOBEC3G (37), respectively. Recently, we and several other investigators demonstrated that a third accessory protein of HIV, viral protein R (Vpr), would also exert its function by usurping the host ubiquitination machinery via the recruitment of an E3 ubiquitin ligase complex composed of VPRBP (viral protein R binding protein, also known as DCAF1), damaged DNA binding protein 1 (DDB1), and cullin 4A (CUL4A) (5, 14, 22, 32, 43, 49, 55).
Ubiquitination is a posttranslational modification that involves the isopeptidic covalent linkage of the C terminus of a small protein called ubiquitin, most commonly to lysine acceptor residues. The conjugation of ubiquitin is performed by E3 ubiquitin ligase complexes, which also control specificity via a direct interaction with substrates (24, 33). The cullin-RING E3 ubiquitin ligases (CRL) structured around the scaffold protein CUL4A (CRL4A) interact with the 126-kDa DDB1 adaptor in order to recruit substrate receptors of the WDxR family (31). Mass spectrometric analyses of CUL4A-DDB1 complexes have revealed physical interactions with at least 30 different WDxR substrates receptors (3, 19, 21, 26), suggesting that CRL4A E3 ligases would likely regulate the function of hundreds of cellular proteins. Surprisingly, however, relatively few cellular proteins have been shown to date to be regulated by these complexes. CRL4A (DDB2) promotes the ubiquitination of histones 2A, 3, and 4 (28, 53); of xeroderma pigmentosum group C protein (XPC) (48); and probably of xeroderma pigmentosum group A protein (XPA) (52) to facilitate UV damage repair. CRL4A (CSA) and CRL4A (DET1-COP1) induce the proteolysis of Cockayne syndrome type B gene product (CSB) (18) and c-JUN (56), respectively. PCNA (proliferating cell nuclear antigen), via the recruitment of a CRL4A (CDT2) ligase, was previously shown to regulate the degradation of the replication licensing factor CDT1 (20, 35) as well as that of the CDK inhibitor p21 (1, 39). The 180-kDa VPRBP/DCAF1 WDxR substrate receptor was identified more than a decade ago as being a Vpr-interacting protein and was found to be expressed at the mRNA level in most tissues (59, 61). However, its normal cellular functions remained elusive until recently. Huang and Chen previously demonstrated that CRL4A (VPRBP) induces the rapid degradation of the tumor suppressor Merlin (NF2 [neurofibromin 2]) following serum stimulation (23). Moreover, the depletion of VPRBP reduced the rate of DNA replication, blocked cells in S phase, and impeded cellular proliferation (22, 38). Consequently, the genetic ablation of VPRBP in mouse (38) and in the evolutionarily distant Arabidopsis thaliana (60) led to embryonic lethality, suggesting an essential role for VPRBP in the cell cycle as well as in development.
The small HIV-1 accessory protein Vpr induces cell cycle arrest in the G2/M phase in various cell types, including transformed cell lines as well as primary lymphocytes (27, 41, 63). Notably, an abnormal accumulation of infected cells in G2 phase can be observed in tissues from patients infected by HIV-1 (63). Although the function of this cell cycle arrest has remained elusive, its molecular mechanism has recently begun to be elucidated. Several investigators reported that Vpr activates the canonical ATR (ataxia telangiectasia and Rad3 related) DNA damage/stress checkpoint (2, 29, 42, 62, 63). The Vpr-mediated activation of ATR is accompanied by the formation of DNA repair foci that include RPA (replication protein A), HUS1, RAD17, BRCA1 (breast cancer 1, early onset), TP53BP1 (tumor protein p53 binding protein 1), and γ-H2AX (phosphorylated histone 2A variant X) (2, 29, 62, 63) and by the activation of the CHEK1 kinase (42). This series of events leads to the inactivation of the CDC2/cyclin B complex, a master regulator of the G2-to-M-phase transition, and ultimately prevents entry into mitosis. In addition, Vpr was recently found to associate with DDB1 and CUL4A via a direct interaction with VPRBP (5, 14, 22, 32, 43, 49, 55). The depletion of VPRBP by RNA interference drastically impaired Vpr-mediated G2 arrest. Similarly, mutations of Vpr in the hydrophobic leucine-rich core region abrogated binding to CRL4A (VPRBP) and concomitantly impaired G2 arrest. In contrast, mutants of Vpr in the C-terminal arginine-rich domain were not compromised for the association with the E3 ligase but nevertheless failed to induce G2 arrest (5, 14, 22, 32, 43, 49, 55). These later mutants also displayed trans-dominant negative activity (14), thus indicating that the recruitment of this E3 ubiquitin ligase is essential but not sufficient to induce G2 arrest. The simplest explanation for these observations is that Vpr would recruit cellular substrates to be ubiquitinated by the complex. The ubiquitination of these as-yet-unknown proteins would lead to their degradation or the modulation of their activity, ultimately resulting in ATR activation and G2 arrest (13). This is not the only instance in which Vpr would be implicated in the degradation of a cellular protein. Vpr directly interacts with and induces the proteasomal degradation of the cellular enzyme UNG2 (uracil-DNA glycosylase 2, also known as CCNO) independently of the induction of cell cycle arrest (43, 44). The Vpr-induced degradation of UNG2 was found previously to be mediated by DDB1 (43) but surprisingly did not require VPRBP (55). In addition, viral protein X (Vpx), a paralog of Vpr present in HIV-2 and some simian immunodeficiency virus (SIV) isolates, is also able to recruit CRL4A (VPRBP) (6, 32, 45, 47). In contrast to Vpr, Vpx does not induce G2 arrest but counteracts a putative cellular restriction factor expressed in macrophages and dendritic cells that targets a postentry step critical for efficient reverse transcription (37). Importantly, the presence of an active proteasome system and the recruitment of CRL4A (VPRBP) by Vpx were previously shown to be required for the inhibition of the restriction factor, suggesting that, similarly to Vpr, Vpx would act via the CRL4A (VPRBP)-mediated ubiquitination and proteasomal degradation of cellular proteins (6, 17, 45, 47).
However, this model, while very attractive, is based solely on the indirect observation that some C-terminal mutants of Vpr retain their interaction with the E3 ligase but fail to induce G2 arrest. Other models for the activity of Vpr have also been proposed and include the inhibition of DDB1 functions (43) or an overall increased activity of CRL4A (VPRBP) (22). Moreover, it is unclear whether this association of Vpr with CRL4A (VPRBP) is the result of an overexpression in transformed cell lines and whether this interaction would occur during the infection of primary lymphocytes. Therefore, we sought to confirm the physiological significance of the Vpr-VPRBP interaction as well as obtain more direct evidence that Vpr is indeed recruiting cellular proteins and inducing their ubiquitination and proteasomal degradation. Herein, we show that Vpr interacts with VPRBP and DDB1 during the infection of primary CD4+ T lymphocytes. In addition, using a tandem affinity approach and the overexpression of a dominant negative tagged ubiquitin mutant, we observed that wild-type (WT) Vpr could associate specifically with unknown ubiquitinated cellular proteins and that this interaction required the recruitment of an active CRL4A (VPRBP) E3 ubiquitin ligase complex. Moreover, C-terminal G2-arrest-defective mutants of Vpr displayed a reduced association with these ubiquitinated proteins. We also provide evidence that Vpr induces the K48-linked polyubiquitination of these cellular proteins, leading to their rapid proteasomal degradation. Finally, these Vpr-induced ubiquitination events were necessary specifically for Vpr to induce the phosphorylation of H2AX, an early marker of ATR-mediated DNA damage/stress checkpoint activation.
HEK293T and HeLa cells were cultured as previously described (58). The anti-hemagglutinin (HA) tag and anti-Myc tag monoclonal antibodies were clones 12CA5 and 9E10, respectively. Rabbit polyclonal antibodies against VPRBP and DDB1 were obtained from Accurate Chemical and Scientific Corporation (Westbury, NY) and Santa Cruz Biotechnology (Santa Cruz, CA), respectively. Vpr was detected by using a rabbit polyclonal antibody directed against a Vpr N-terminal peptide (12). The anti-green fluorescent protein (GFP) antibody was obtained from Molecular Probes (Invitrogen, San Diego, CA), and the anti-phospho-H2AX (Ser139) antibody was clone JBW301 from Upstate (Millipore, Billerica, MA). The rabbit anti-actin antibody was obtained from Sigma-Aldrich (St. Louis, MO).
Plasmids SVCMV-TAP, SVCMV-TAP-Vpr(WT), and SVCMV-TAP-Vpr(Q65R) were described previously (5). Plasmids expressing TAP-Vpr(1-78), TAP-Vpr (1-86), and TAP-Vpr(R87A,R88A) were generated by subcloning the SalI-BamHI fragments from SVCMV-3HA-Vpr(1-78), SVCMV-3HA-Vpr(1-86), and SVCMV-3HA-Vpr(R87A,R88A) (57), respectively, into SVCMV-TAP-Vpr(WT). Plasmids expressing TAP-Vpr(1-84) and TAP-Vpr(1-90) were generated by PCR using a strategy described previously (5). Plasmids expressing Myc-His-tagged ubiquitin K48R (54) and HA-tagged ubiquitin (WT, K0, and K48R) (34) were kind gifts of R. Kopito (Stanford University, Stanford, CA) and T. Dawson (Johns Hopkins University, Baltimore, MD). The plasmid expressing HA-Ub(K63R) was constructed by site-directed mutagenesis using the QuikChange II mutagenesis kit (Stratagene, La Jolla, CA). GFP-expressing plasmid pQBI-25 was obtained from Qbiogene (Carlsbad, CA). Vectors expressing scrambled small hairpin RNA (shRNA) and VPRBP shRNA were obtained from Open Biosystems (Huntsville, AL). The CUL4A dominant negative construct (55) was kindly donated by C. de Noronha (University of Albany, Albany, NY). The construction of the infectious molecular clone HxBru(Vpr−) was described previously (30). HxBru(HA-Vpr) was generated by PCR by adding an HA tag at the N terminus of Vpr, which resulted in the addition of 9 extra amino acids at the C terminus of Vif. The lentiviral vectors WPI and WPI-Vpr as well as the SVCMV-IN-VSV-G expression construct were described previously (5).
HEK293T cells were transfected by using the calcium phosphate precipitation method. Forty-eight hours later, cells were harvested, washed, and lysed in Triton lysis buffer (50 mM Tris-HCl [pH 7.5], 150 mM NaCl, 0.5% Triton X-100, and complete protease inhibitors) or radioimmunoprecipitation assay (RIPA) buffer (8 mM Na2HPO4 [pH 7.2], 2 mM NaH2PO4, 140 mM NaCl, 1% NP-40, 0.05% SDS, 12 mM deoxycholate, and complete protease inhibitors). Tandem affinity purification (TAP) was performed as described previously (5). In some experiments, where indicated, IgG pulldown assays were performed instead of full tandem affinity purification, as described previously (5). Immunoprecipitations were performed by using 50 μl of 50% anti-HA-coupled agarose beads (Sigma-Aldrich, St. Louis, MO) followed by extensive washes in Triton lysis buffer and elution with 100 μg/ml HA peptide. Eluted proteins were separated on a 12.5% SDS-PAGE gel, and Western blot analyses of eluted proteins were performed. Ubiquitinated proteins were revealed by using monoclonal antibodies directed at Myc or HA tags depending on the ubiquitin construct used.
The production and titration of infectious viral particles as well as of the lentiviral vectors expressing GFP (WPI) or coexpressing GFP and Vpr (WPI-Vpr) were performed as described previously (5, 57). Murine leukemia virus-based retroviral vectors expressing VPRBP-targeting shRNA (shVprBP) or scrambled control shRNA (shControl) were produced by the transfection of 15 μg of vector, 12 μg of the packaging construct pCIG3-N (9), and 5 μg of vesicular stomatitis virus G protein (VSV-G)-expressing plasmid pSVCMV-IN-VSV-G in 1.5 million HEK293T cells by using the calcium phosphate precipitation method. Vector-containing supernatants were filtered through a 0.45-μm filter and used immediately.
Peripheral blood mononuclear cells (PBMCs) were extracted, by the Ficoll method, from whole blood obtained from consenting healthy adult donors who gave written informed consent under research protocols approved by the research ethics review board of the Institut de Recherches Cliniques de Montreal. CD4+ T lymphocytes were purified from PBMCs by magnetic negative selection using CD4+ T-cell isolation kit II and the AutoMACS Pro system according to the manufacturer's instructions (Miltenyi Biotec, Auburn, CA). CD4+ T cells were cultured as previously described (57). Ten million activated CD4+ cells were mock infected or infected with HxBru(Vpr−) or HxBru(HA-Vpr) virus at a multiplicity of infection (MOI) of 0.01. Five days after infection, cells were lysed in Triton lysis buffer and subjected to anti-HA immunoprecipitation as described above.
Fifty thousand HeLa cells were seeded onto coverslips in 24-well plates and transfected with Lipofectamine 2000 reagent according to the manufacturer's instructions. Twenty-four hours after transfection, cells were transduced with WPI or WPI-Vpr at a multiplicity of infection of 2.5 in the presence of 8 μg/ml polybrene. Two days later, cells were processed for confocal fluorescence immunohistochemistry as previously described (15).
HEK293T cells were transduced with shControl and shVprBP retroviral vectors in the presence of 8 μg/ml polybrene. Two days after transduction, cells were selected for 10 days with 1 μg/ml puromycin, and single-cell clones were then isolated by the limited-dilution method. Western blots were performed on shControl- and shVprBP-transduced stable clones, and the clone with the most significant decrease in levels of VPRBP expression was selected.
Cell cycle profiles were determined by flow cytometry using propidium iodide staining as previously described (5). Apoptosis was assayed by flow cytometry using annexin V and propidium iodide as a dead cell counterstain as described previously (58). Cell proliferation kinetics were monitored by flow cytometry using a standard carboxyfluorescein succinimidyl ester (CFSE) assay.
The interaction between Vpr and the CRL4A (VPRBP) E3 ubiquitin ligase and its functional importance for Vpr-induced G2 arrest, although strongly established, were, however, demonstrated solely by overexpression studies of transformed cell lines. In order to determine whether Vpr could recruit the CRL4A (VPRBP) E3 ubiquitin ligase under more physiological conditions, primary CD4+ T lymphocytes were infected with HxBru-derived virus expressing hemagglutinin (HA)-tagged Vpr (HA-Vpr) or deleted for Vpr at a multiplicity of infection of 0.01. Five days after infection, cells were harvested and subjected to immunoprecipitation by using anti-HA-conjugated agarose beads, and immunoprecipitated proteins were detected by Western blotting. Endogenous VPRBP and DDB1 could be detected in the lysates of activated primary CD4+ T lymphocytes, thus confirming their expression at the protein level (Fig. (Fig.1).1). The coimmunoprecipitation of VPRBP and DDB1 could be observed for cells infected with HxBru(HA-Vpr) viruses but not for HxBru(Vpr−)- or mock-infected cells (Fig. (Fig.1).1). Therefore, the recruitment of VPRBP and DDB1 by Vpr occurs under conditions mimicking those of an in vivo infection and is thus not the result of an overexpression of Vpr in transformed cell lines.
We next sought to obtain additional data on the mechanisms by which Vpr utilizes this complex to induce G2 arrest. If Vpr indeed acts as a connector between an E3 ubiquitin ligase complex and cellular proteins, and this association leads to the ubiquitination of these substrates, we hypothesized that we should be able to observe Vpr interacting with some of these unknown ubiquitinated proteins. To assess this premise, we developed a tandem affinity purification (TAP) procedure (5) that takes advantage of a Myc-tagged ubiquitin K48R mutant with the unusual properties of acting as a weak chain terminator and stabilizing polyubiquitinated products (7, 54). Of note, the TAP-tagged Vpr (TAP-Vpr) construct used for the procedure is able to induce G2 arrest although to a lesser extent than that induced by untagged native Vpr (see Fig. S1 in the supplemental material) (5). Under these conditions, we were able to detect an interaction between TAP-Vpr and unknown cellular ubiquitinated proteins, as revealed by the anti-Myc-reacting protein smear in the TAP-Vpr pulldown, but not with the TAP tag alone (Fig. (Fig.2A).2A). This association was also detected under stringent extraction conditions, such as RIPA buffer and high-salt buffer (400 mM NaCl) (data not shown). The Myc-tagged ubiquitination signal detected in the presence of pulled-down Vpr was the result of bound cellular ubiquitinated proteins and not of the ubiquitination of Vpr itself, since the heat denaturation of proteins in RIPA buffer prior to the purification of TAP-Vpr complexes abolished this signal. (Fig. (Fig.2B,2B, compare lane 3 to lane 4). For this experiment, we could not perform the full tandem affinity purification procedure because the TAP tag could not bind to calmodulin beads following denaturation. Consequently, these experiments involved an IgG pulldown step solely, hence explaining why the enrichment of ubiquitination observed with Vpr is less important under this particular condition (Fig. (Fig.2B,2B, compare lane 3 to lane 1). Nevertheless, the retention of the ubiquitination signal under denaturing conditions would have indicated that this ubiquitination originated from ubiquitin covalently conjugated to Vpr itself and not to associated proteins. Therefore, in agreement with our model, Vpr is capable of associating with ubiquitinated cellular proteins.
This association of Vpr with cellular ubiquitinated proteins suggests that these proteins might be ubiquitinated by Vpr following the recruitment of the CRL4A (VPRBP) E3 ubiquitin ligase. On the other hand, these cellular proteins might be already ubiquitinated as part of their normal metabolism before their interaction with Vpr and would thus not constitute substrates of the Vpr ubiquitin ligase complex. To distinguish between these two possibilities, we analyzed the effect of knocking down VPRBP on the association of Vpr with ubiquitinated proteins. The transient depletion of VPRBP with small hairpin RNA (shRNA) resulted in a significant decrease in the association of Vpr with cellular ubiquitinated proteins (Fig. (Fig.3A,3A, compare lanes 3 and 4), suggesting that the association of ubiquitinated proteins with Vpr involves the recruitment of the CRL4A (VPRBP) E3 ligase. Moreover, the overexpression of a dominant negative construct of CUL4A (Cul4A DN) produced a similar decrease in the association of TAP-Vpr with ubiquitinated proteins, thus corroborating the results obtained with the depletion of VPRBP and indicating that the CRL4A (VPRBP) E3 ligase must be catalytically active in order to detect these Vpr-associated ubiquitinated proteins (Fig. (Fig.3B,3B, compare lanes 3 and 4). However, these results do not exclude the possibility that this association of Vpr with ubiquitinated proteins might result from the sole association with ubiquitinated components of the E3 ligase complex without necessarily implicating ubiquitinated substrates.
To address this issue, we investigated this association in the context of Vpr mutants defective for their interaction with the E3 ligase or for the putative G2 arrest substrate protein. Surprisingly, the Q65R mutation in Vpr, which results in a strong reduction of the interaction with VPRBP and DDB1 (5, 14, 32, 49), displayed an increased association with ubiquitinated proteins (Fig. (Fig.4A)4A) independently of its higher level of expression. These unexpected results may be explained by nonspecific interactions that might result from the accumulation of this nonfunctional mutant protein. To test whether the association of Vpr(Q65R) with ubiquitinated proteins was independent of the recruitment of CRL4A (VPRBP), we constructed 293T-based stable monoclonal cell lines expressing shRNA against VPRBP or a nontargeting scrambled shRNA control (see Fig. S2A in the supplemental material). Importantly, we did not detect any significant effect of VPRBP depletion on the cell cycle profile (Fig. S2B). We did, however, observe a slight decrease in the growth kinetics of these cells (Fig. S2C), but it did not result in a significant increase in apoptosis (Fig. S2D). The level of stable VPRBP knockdown achieved with this cell line was sufficient to almost completely abolish Vpr-mediated G2 arrest compared to the scrambled control cell line (Fig. S2E). We then used this pair of cell lines to assess the levels of cellular ubiquitinated proteins associated with the Q65R mutant of Vpr. As expected, the increased association of ubiquitinated proteins with Vpr(Q65R) was independent of the recruitment of the E3 ligase given that the depletion of VPRBP did not affect the levels of ubiquitinated proteins associated with this mutant of Vpr (Fig. (Fig.4A,4A, compare lanes 5 and 6). In comparison, with this system, we observed a significant reduction of the association of wild-type Vpr with ubiquitinated proteins in the VPRBP-knocked-down cell line versus the control cell line (Fig. (Fig.4A,4A, compare lanes 3 and 4), corroborating results obtained with transient knockdown experiments. Therefore, we hypothesize that the Q65R mutation, in addition to its effect on binding to VPRBP, might also result in major conformational defects that would lead to this nonspecific association to ubiquitinated proteins. Indeed, we observed that, in contrast to wild-type Vpr, which oligomerizes and displays a nuclear localization, Vpr(Q65R) accumulates in part in the cytoplasm as discrete puncta and also fails to efficiently oligomerize (data not shown). We also investigated the effect of a C-terminal deletion of Vpr, Vpr(1-78), a mutant that fails to induce G2 arrest (see Fig. S1 in the supplemental material) while maintaining the interaction with VPRBP (32). Importantly, Vpr(1-78), which should thus not interact with the putative substrate responsible for G2 arrest, failed to interact with ubiquitinated proteins (Fig. (Fig.4B,4B, compare lanes 2 and 3). Other deletions (starting at positions 84, 86, and 90) or point mutations (R87A, R88A) in the C-terminal domain impaired G2 arrest (Fig. S1) and led to a significant decrease in the association with ubiquitinated proteins (Fig. (Fig.4C)4C) while maintaining their association with VPRBP (data not shown). Our data indicate that these proteins are not solely ubiquitinated UNG2 molecules, given that C-terminal deletions of Vpr retain their interaction with UNG2 (10). Therefore, in agreement with our model, these data further indicate that the putative G2 arrest substrate(s) is ubiquitinated in the presence of Vpr and that the abrogation of Vpr-substrate or Vpr-CRL4A (VPRBP) interactions impairs Vpr-mediated ubiquitination. Overall, these results provide direct interaction-based evidence that Vpr acts as a connector between an E3 ubiquitin ligase complex and substrate proteins.
To determine the fate of proteins that are ubiquitinated by Vpr, we used the tandem affinity purification method in combination with an HA-tagged ubiquitin construct that does not significantly stabilize polyubiquitinated products and does not protect substrates from proteasome degradation (34). With this system, we detected ubiquitinated proteins associated with Vpr (Fig. (Fig.5A),5A), but the ubiquitination signal detected under these conditions did not rely on the recruitment of the E3 ligase complex, since the knockdown of VPRBP did not have any effect on the levels of ubiquitination (data not shown), suggesting that in the absence of an interference with polyubiquitination, the Vpr-targeted substrates might be rapidly degraded. In support of this scenario, a 16-h treatment of cells coexpressing HA-Ub and TAP-Vpr with the proteasome inhibitor MG132 resulted in an increased association of Vpr with cellular ubiquitinated proteins (Fig. (Fig.5A,5A, compare lanes 3 and 4). We also observes an increase in levels of polyubiquitinated proteins with a shorter MG132 treatment (5 h) but to a lesser extent (data not shown). Most notably, this significant increase in levels of HA-polyubiquitinated proteins associated with Vpr following treatment with MG132 required the recruitment of the E3 ligase because the depletion of VPRBP with shRNA drastically reduced the extent of polyubiquitination (Fig. (Fig.5B).5B). Therefore, these results suggest that following polyubiquitination by the Vpr-CRL4A (VPRBP) complex, substrates are rapidly degraded by the proteasome.
To determine whether the observed degradation of Vpr-associated ubiquitinated proteins resulted from classical K48-linked polyubiquitination, we analyzed the effect of the ubiquitin K48R mutation on Vpr-associated ubiquitination. Cells were cotransfected with plasmids expressing TAP or TAP-Vpr with either HA-Ub(WT) or HA-Ub(K48R). In contrast to Myc-tagged Ub(K48R), which acts as a weak chain terminator and stabilizes polyubiquitin products (7, 54), this HA-Ub(K48R) construct was previously shown to potently block ubiquitin chain elongation through lysine 48 (34). We also used the polyubiquitination-null construct HA-Ub(K0), in which all lysines were mutated to arginines, as a control. Following MG132 treatment, the K48R mutation in ubiquitin significantly reduced the levels of ubiquitinated proteins associated with Vpr compared to wild-type ubiquitin (Fig. (Fig.6,6, compare lanes 6 and 8). However, the reduction in the level of ubiquitination did not reach the levels achieved with the K0 mutation (Fig. (Fig.6,6, compare lanes 7 and 8). These data suggest that Vpr induces, at least in part, classical K48-linked polyubiquitination of cellular substrate proteins, leading to their proteasomal degradation. However, given that the levels of ubiquitination observed for the K48R mutant did not reach the levels observed for the polyubiquitination-null K0 mutant, we cannot exclude the possibility that other lysine residues in ubiquitin might be involved in the formation of mixed ubiquitin linkages.
Other investigators demonstrated previously that blocking the activity of the proteasome using small-molecule inhibitors or blocking polyubiquitination via the expression of a dominant negative mutant of ubiquitin [Ub(K48R)] abrogated Vpr-mediated G2 arrest (14, 49). However, caution has to be used when interpreting these results given that the inhibition of polyubiquitination or proteasome function might have pleiotropic effects on checkpoint function without necessarily implicating the direct inhibition of Vpr's activity. Therefore, to evaluate the direct role of K48-linked polyubiquitination in Vpr's activity, we instead monitored the phosphorylation of H2AX (γ-H2AX), an early marker of ATR-mediated checkpoint activation previously implicated in Vpr-mediated G2 arrest (29, 62). Following exogenously induced DNA damages, the phosphorylation of H2AX occurs in the absence of any ubiquitination events (36) and should thus not be affected directly by the K48R mutation in ubiquitin. Indeed, the ectopic expression of HA-Ub(K48R) or HA-Ub(K63R) in HeLa cells did not have any effect on the number of cells displaying γ-H2AX foci following UV irradiation (see Fig. S3 in the supplemental material). To analyze the effect of Ub(K48R) on Vpr-induced γ-H2AX focus formation, we transiently transfected HeLa cells with a plasmid expressing HA-Ub(K48R) or expressing HA-Ub(K63R) (used as a negative control). Twenty-four hours after transfection, cells were transduced with a lentiviral vector expressing GFP alone (WPI) or coexpressing Vpr and GFP (WPI-Vpr). Two days later, cells were processed for immunofluorescence detection by using an anti-HA antibody to detect HA-Ub-expressing cells, anti-GFP (to amplify the GFP signal, a marker of transduced cells), and anti-phospho-H2AX (Fig. (Fig.7A).7A). Cells with more than 10 γ-H2AX foci were considered positive. In mock-transfected cells, transduction with the lentiviral vector expressing Vpr induced a significant increase in the percentage of cells positive for γ-H2AX compared to the control lentiviral vector (83.9% versus 12.3%; P < 0.0001) (Fig. 7A and B). In the presence of HA-Ub(K48R), there was a drastic decrease in the number of Vpr-expressing cells with γ-H2AX foci (25.9% versus 83.9%; P < 0.0005) (Fig. 7A and B). In contrast, the overexpression of HA-Ub(K63R) only weakly altered the ability of Vpr to induce γ-H2AX foci, demonstrating the specific requirement for K48 linkages over other types of linkages. Therefore, the overexpression of the K48R mutant of ubiquitin specifically inhibited the Vpr-induced phosphorylation of H2AX (Fig. 7A and B) without affecting the UV-induced phosphorylation of H2AX (see Fig. S3 in the supplemental material). These observations thus suggest that K48-linked ubiquitination would be especially essential for the activity of Vpr toward early checkpoint activation but not for exogenously induced DNA damages. Therefore, taken together, our results provide direct evidence that Vpr recruits the CRL4A (VPRBP) E3 ubiquitin ligase complex to induce the K48-linked polyubiquitination of one or several as-yet-unknown cellular proteins, resulting in their proteasomal degradation and ultimately leading to the ATR-mediated phosphorylation of H2AX and G2 arrest.
The identification of the substrates targeted by the Vpr-CRL4A (VPRBP) complex represents an important aim not only to fully understand how Vpr activates ATR signaling and promotes G2 arrest but also to comprehend the functional relevance of these biological activities. Given that Vpr induces G2 cell cycle arrest, it is not conceivable to develop approaches that rely on the differential expression pattern of proteins in the presence or absence of Vpr because Vpr cytostatic activity may affect the expression profiles of numerous proteins without necessarily implicating a direct recruitment to the Vpr-CRL4A (VPRBP) E3 ligase. On the other hand, the identification of substrates of E3 ubiquitin ligases by interaction-based proteomic analyses remains a long-standing challenge due to several different inherent and technical problems. Notably, ubiquitinated proteins are present at a low abundance, display a rapid turnover rate, and are subjected to rapid deconjugation (25, 40). Therefore, special care must be taken to enrich and stabilize ubiquitin conjugates by using tagged ubiquitin constructs or proteasome inhibitors. To demonstrate a potential interaction of Vpr with its cognate ubiquitinated substrates, we used a combination of both approaches. First, we used a Myc-tagged ubiquitin K48R mutant with the unusual properties of acting as a weak chain terminator and stabilizing polyubiquitinated products (7, 54) coupled with a highly specific tandem affinity purification procedure (5) in order to enrich Vpr-interacting ubiquitinated proteins. With this method, we were able to show a specific interaction between Vpr and cellular ubiquitinated proteins (Fig. 2A and B). Second, using an HA-tagged ubiquitin construct concomitantly with treatment with the proteasome inhibitor MG132, we were also able to reveal a specific association of Vpr with cellular ubiquitinated proteins (Fig. (Fig.5A).5A). In both cases, a significant part of the Vpr-associated ubiquitinated signal was dependent on the recruitment of an active CRL4A (VPRBP) ligase, since the depletion of VPRBP by shRNA (Fig. (Fig.3A3A and and5B)5B) as well as the overexpression of a dominant negative form of CUL4A (Fig. (Fig.3B)3B) drastically reduced this association. Surprisingly, the Q65R mutation in Vpr, which virtually abrogates the interaction with VPRBP and should thus reduce binding to ubiquitinated proteins, had the opposite effect: it increased the interaction with ubiquitinated proteins (Fig. (Fig.4A).4A). However, we observed that the Q65R mutation led to an accumulation of substantial amounts of Vpr in the cytoplasm and to an inefficient oligomerization of the protein (data not shown), indicating that this mutation has pleiotropic effects on the functions of Vpr and probably induces conformational defects. Therefore, the knockdown of VPRBP and the use of a dominant negative mutant of CUL4A, both of which reduced binding to ubiquitinated proteins, represent a more reliable assessment of the role of the E3 ligase in the association of Vpr with ubiquitinated proteins. Importantly, the deletion of the entire putative substrate-interacting C-terminal domain of Vpr resulted in an abrogation of G2 arrest (see Fig. S1 in the supplemental material) and of the association with ubiquitinated proteins (Fig. (Fig.4B),4B), indicating that these Vpr-associated ubiquitinated proteins are probably not components of the E3 ligase itself. Shorter deletions or point mutations in the C-terminal domain of Vpr also led to an inhibition of G2 arrest (Fig. S1) and to a significant reduction in binding to ubiquitinated proteins (Fig. (Fig.4C),4C), suggesting that at least a significant fraction of these Vpr-associated ubiquitinated proteins would be substrates forcibly recruited to the E3 ligase by Vpr. Given that C-terminal deletions of Vpr retain their interaction with UNG2 (10), our data indicate that these ubiquitinated proteins do not contain detectable levels of ubiquitinated UNG2, therefore excluding the possibility that they are solely ubiquitinated UNG2. These are most likely substrates independently recruited by Vpr rather than an increased ubiquitination of VPRBP's own substrates, because the overexpression of the minimal Vpr-interacting domain of VPRBP was previously reported to increase the association of Vpr with DDB1 as well as the efficiency of Vpr-mediated G2 arrest (32). It is arguably unlikely that this minimal domain, which also contains the WDxR motif responsible for the association with DDB1 (26, 32), would also possess the determinants mediating substrate recognition. Finally, our results do not contradict the previously reported observations that Vpr would increase the neddylation of CRL4A (VPRBP) (22) given that the recruitment of substrates as well as substrate adaptors to CRL complexes, including CRL4A ligases, was shown to markedly promote neddylation (11).
As mentioned above, we found that MG132 stabilized the Vpr-associated HA-tagged ubiquitinated protein (Fig. (Fig.5A),5A), suggesting that in the absence of MG132, Vpr's substrates would be degraded by the proteasome. Indeed, in support of this interpretation, in the absence of proteasome inhibition, the association of Vpr with cellular ubiquitinated proteins was independent of the recruitment of CRL4A (VPRBP) (data not shown), whereas when cells were treated with MG132, the depletion of VPRBP significantly decreased the levels of Vpr-associated ubiquitinated proteins (Fig. (Fig.5B).5B). Moreover, Vpr was found to induce, at least in part, the classical K48-linked polyubiquitination of its substrates (Fig. (Fig.6),6), thus further supporting our evidence that these substrates are degraded by the proteasome given that this type of homopolymeric ubiquitin chain generally leads to proteasomal proteolysis (50, 51).
Several investigators reported previously that Vpr activates ATR in a variety of cell types, including primary CD4+ T lymphocytes. The Vpr-mediated activation of ATR was accompanied by the formation of DNA repair foci that included RPA, HUS1, RAD17, BRCA1, TP53BP1, and γ-H2AX. In contrast to data from other investigators, who used the accumulation of cells in G2/M phase as a marker of Vpr activity (14, 49), we reasoned that using an early marker of checkpoint activation would likely constitute a more direct and less ambiguous strategy to assess the role of Vpr-mediated ubiquitination in its G2 arrest function. The phosphorylation of H2AX represents such a marker. Indeed, Mailand et al. recently reported that the MG132-mediated depletion of nuclear ubiquitin did not impair the phosphorylation of H2AX in response to exogenous genotoxic stresses such as DNA double-strand breaks induced by ionizing radiation (36). Moreover, the overexpression of HA-tagged Ub(K48R) or Ub(K63R) in HeLa cells did not inhibit H2AX phosphorylation following UV irradiation (see Fig. S3 in the supplemental material). Therefore, H2AX phosphorylation appears to be independent of ubiquitination. In contrast, in the case of Vpr-induced H2AX phosphorylation, the overexpression of Ub(K48R) but not Ub(K63R) significantly reduced levels of H2AX phosphorylation (Fig. 7A and B), suggesting that this effect was most probably due to a direct inhibition of Vpr-induced K48 polyubiquitination rather than a pleiotropic inhibition of checkpoint function. These functional data demonstrating the role of Vpr-induced ubiquitination in its G2 arrest activity are also supported by biochemical evidence whereby G2-arrest-defective mutants of Vpr in the putative C-terminal substrate recruitment domain, still competent for associations with CRL4A (VPRBP), failed to interact with ubiquitinated cellular proteins (Fig. (Fig.4B).4B). Therefore, taken together, our results strongly suggest that the Vpr-mediated K48-linked ubiquitination and degradation of one or several putative substrates are responsible for Vpr-induced G2 arrest.
Zimmerman and colleagues previously observed that Vpr was unable to induce checkpoint activation in macrophages due to the absence of ATR in these cells, while gamma irradiation led to ATM (ataxia telangiectasia mutated) activation (63). Importantly, Vpr did not appear to cause DNA double-strand breaks in cycling cells under conditions where ATR was activated (29). Those authors concluded that Vpr likely causes DNA replication stresses rather than direct DNA damages, such as DNA double-strand breaks, that would otherwise activate ATM (63). Thus, it appears that the cellular substrate(s) targeted by Vpr might have important roles in DNA replication and that its degradation by Vpr would cause DNA replication stress, as demonstrated by the formation of RPA foci (29), ultimately leading to ATR activation (42), and the accumulation of cells in G2/M phase. Interestingly, Vpr was previously shown to form nuclear foci that colocalized with DNA repair foci containing γ-H2AX and RPA (29). It would be tempting to speculate that Vpr would recruit CRL4A (VPRBP) onto chromatin to degrade a chromatin-bound component of the DNA replication machinery directly at this site. Alternatively, the degradation of the substrate(s) might not cause DNA replication stresses directly but might somehow mimic signals induced by them. Recently, the forced tethering of DNA repair factors, including ATM, MDC1 (mediator of DNA damage checkpoint 1), and NBS1 (Nijmegen breakage syndrome 1; nibrin), to chromatin was shown to induce the formation of fully competent DNA repair foci in the absence of any DNA damage (46). It is thus conceivable that the ubiquitination and degradation of a DNA repair regulator(s) by Vpr might somehow induce the incorrect recruitment of DNA repair proteins to chromatin in the absence of any DNA replication stress.
In conclusion, using a tandem affinity purification approach, we provide additional and more direct evidence that Vpr recruits the CRL4A (VPRBP) E3 ubiquitin ligase complex to induce the K48-linked polyubiquitination of one or several putative substrates, resulting in their proteasomal degradation. The proteolysis of these putative substrates would lead to the phosphorylation of H2AX, an early target of ATR activation, and, ultimately, G2 arrest. The identification of the cellular proteins degraded by Vpr will be central for an understanding of how Vpr triggers ATR activation and why Vpr induces cell cycle arrest. The tandem affinity purification procedure presented in this study represents a powerful approach to isolate and identify cellular ubiquitinated substrates interacting with Vpr.
We thank T. Dawson and R. Kopito for providing us with plasmids expressing Myc-tagged and HA-tagged ubiquitin. We also thank C. de Noronha for the dominant negative construct of CUL4A, D. Trono for the lentiviral vector WPI and the packaging plasmid psPAX2, and J. Stoye for the packaging plasmid pCIG3-N.
J.-P.B. and J.R. are recipients of studentships from the Canadian Institute of Health Research (CIHR). E.A.C. is the recipient of the Canada Research Chair in Human Retrovirology. This work was supported by grants from the CIHR and FRSQ to E.A.C.
Published ahead of print on 20 January 2010.
†Supplemental material for this article may be found at http://jvi.asm.org/.