Vif Overcomes the Innate Antiviral Activity of APOBEC3G by Promoting Its Degradation in the Ubiquitin-Proteasome Pathway*

Viruses must overcome diverse intracellular defense mechanisms to establish infection. The Vif (virion infectivity factor) protein of human immunodeficiency virus 1 (HIV-1) acts by overcoming the antiviral activity of APOBEC3G (CEM15), a cytidine deaminase that induces G to A hypermutation in newly synthesized viral DNA. In the absence of Vif, APOBEC3G incorporation into virions renders HIV-1 non-infectious. We report here that Vif counteracts the antiviral activity of APOBEC3G by targeting it for destruction by the ubiquitin-proteasome pathway. Vif forms a complex with APOBEC3G and enhances APOBEC3G ubiquitination, resulting in reduced steady-state APOBEC3G levels and a decrease in protein half-life. Furthermore, Vif-dependent degradation of APOBEC3G is blocked by proteasome inhibitors or ubiquitin mutant K48R. A mutation of highly conserved cysteines or the deletion of a conserved SLQ(Y/F)LA motif in Vif results in mutants that fail to induce APOBEC3G degradation and produce non-infectious HIV-1; however, mutations of conserved phosphorylation sites in Vif that impair viral replication do not affect APOBEC3G degradation, suggesting that Vif is important for other functions in addition to inducing proteasomal degradation of APOBEC3G. Vif is monoubiquitinated in the absence of APOBEC3G but is polyubiquitinated and rapidly degraded when APOBEC3G is coexpressed, suggesting that coexpression accelerates the degradation of both proteins. These results suggest that Vif functions by targeting APOBEC3G for degradation via the ubiquitin-proteasome pathway and implicate the proteasome as a site of dynamic interplay between microbial and cellular defenses.

Human immunodeficiency virus 1 (HIV-1) 1 and other lentiviruses encode the vif (virion infectivity factor) gene, which is essential for the production of infectious virus (1,2). Recent studies demonstrate that Vif counteracts the innate antiviral activity of APOBEC3G, also known as CEM15 (3)(4)(5)(6)(7)(8). APOBEC3G, a member of the APOBEC family of cytidine deaminase-editing enzymes, is a potent DNA mutator that can be packaged into budding retroviruses (3,6,8,9). After the initiation of reverse transcription, APOBEC3G edits minusstranded viral cDNA, converting cytosine to uracil and affecting the stability and fidelity of newly synthesized cDNA (4 -8). These mutations may initiate a DNA base excision repair pathway that compromises the structural integrity of the singlestranded viral DNA, resulting in the catastrophic failure of reverse transcription characteristic of Vif-defective (⌬Vif) viruses (10). Furthermore, the massive C to U conversion in the minus strand leads to G to A hypermutation in the viral genome, which affects subsequent stages of the viral life cycle and decreases viral fitness (6,8). Thus, APOBEC3G editing likely contributes to sequence variation within viral populations, particularly the G to A mutational bias characteristic of HIV-1 (11,12).
Vif is required in virus-producing cells during the late stages of infection (10,13). Replication in "non-permissive" cells, such as primary T lymphocytes, macrophages, and certain T cell lines, is strictly dependent on Vif, whereas Vif is dispensable for replication in "permissive" cell lines, such as 293T cells (13,14). APOBEC3G expression is restricted to non-permissive cells (3), whereas its expression in permissive cells confers a non-permissive phenotype (3, 6 -9). In non-permissive cells, ⌬vif viruses can produce virions, but they fail to complete reverse transcription and cannot establish infection (10,15,16). The expression of Vif in trans in virus-producing cells but not target cells rescues this defect (17). Vif forms a complex with APOBEC3G, preventing its viral encapsidation, and thereby protects the viral genome from editing mutations (8). The interaction between Vif and APOBEC3G is species-specific (8), which may play a role in restricting the replication of HIV-1 to humans.
The mechanism by which Vif inhibits APOBEC3G encapsidation into HIV-1 is unclear. In this study we demonstrate that Vif counteracts the antiviral activity of APOBE3G by promoting the ubiquitination and proteasomal degradation of APOBE3G. By destabilizing APOBEC3G, Vif enhances viral infectivity, suggesting that the primary function of Vif is to target APOBEC3G for destruction. APOBEC3G also promotes polyubiquitination and degradation of Vif, suggesting a reciprocal relationship wherein coexpression accelerates the degradation of both proteins. These results suggest that Vif has evolved to target APOBEC3G for degradation and implicate the ubiquitin-proteasome pathway in the dynamic interplay of microbial and host defenses.
Immunoprecipitation and Western Blot Analysis-Hut78 cells and the Hut78-Vif cell line that stably expresses Vif were maintained in RPMI medium with 10% fetal bovine serum or in the same medium with 0.5 g/ml puromycin, respectively. 293T cells were cultured in Dulbecco's modified Eagle's medium with 10% fetal bovine serum and transfected by calcium phosphate with 20 g of DNA or LipofectAMINE 2000 (Invitrogen) with 4 g of DNA. For Western blotting, lysates were prepared 36 -48 h post-transfection in a lysis buffer (50 mM Tris-HCl (pH 7.2), 150 mM NaCl, 10% glycerol, 0.5% Cymal-5, and 1% protease inhibitor mixture (Sigma)). Inhibitors were applied to cells for 6 h before lysis using the following conditions; MG132 (Peptide Intl.) and clastolactacystin ␤-lactone (Boston Biochem) at 50 and 10 M, respectively, and chloroquine (Sigma) at 200 M. Identical amounts of protein were separated by SDS-PAGE, transferred onto polyvinylidene difluoride membranes, and detected by standard techniques. Where indicated, identical amounts of lysate were subjected to immunoprecipitation followed by Western blotting. Co-immunoprecipitations were performed with cells lysed in a detergent buffer (0.5% Nonidet P-40, 50 mM Tris-HCl (pH 7.4), 150 mM NaCl, and a 1% protease inhibitor mixture (Sigma)). In vitro binding studies were performed with GST-Vif purified from bacteria and recombinant APOBEC3G (provided by Immuno-Diagnostics, Inc.). After washing, protein complexes were analyzed by Western blotting. Pulse-chase experiments were performed in 293T cells transfected with pNLX alone (15 g), pNLX⌬Vif (15 g) and pAPOBEC3G:HA (0.1 g), or pNLX (15 g) and pAPOBEC3G:HA (0.1 g) adjusted to 20 g total with an empty vector. Cells were starved for 30 min in Met-Cys-Dulbecco's modified Eagle's medium, labeled for 20 min with 50 Ci/ml [ 35 S]methionine and [ 35 S]cysteine, and chased using 20 mM unlabeled L-methionine and L-cysteine. Cells were lysed in radioimmune precipitation assay buffer and quantitated by the detergent-compatible protein assay (Bio-Rad). Identical amounts of lysate were subjected to immunoprecipitation, resolved by SDS-PAGE, and analyzed by phosphorimaging. Half-life was calculated by modeling exponential decay for early phase degradation of APOBEC3G (0 -45 min) or all time points for Vif (0 -60 min).
Viruses and Infections-Virus was produced from 293T cells transfected with proviral plasmids. Viruses were quantitated by reverse transcriptase assays, and similar amounts were used to infect the reporter cell line Cf2-luc (23). Infection was measured 48 h after infection by performing luciferase assays (Promega). Infections of H9 or SupT1 cells were initiated by co-culture with virus-producing 293T cells (23). After overnight incubation, H9 or SupT1 cells were removed, washed, and cultured in RPMI medium with 10% fetal bovine serum.
Semiquantitative Reverse Transcriptase PCR-Reverse transcriptase PCR was performed as described (24) using total mRNA (1 g) isolated from infected or uninfected H9 and SupT1 cells collected 5 days after infection. The primers used were APOBEC3G forward 5Ј-AAT AGA CCC ATC CTT TCT CGT CGG AAT ACC-3Ј and reverse 5Ј-GTT CAG CAG GAC CCA GGT GTC ATT G-3Ј.

RESULTS
Vif Reduces Steady-state Levels of APOBEC3G-As an initial approach to investigate how Vif counteracts the antiviral activity of APOBEC3G, we examined the effect of Vif on APOBEC3G expression. HA-tagged APOBEC3G was expressed in 293T cells, which do not express APOBEC3G endogenously (3). The 46-kDa HA-tagged APOBEC3G protein, in addition to minor 22-and 24-kDa species that may represent proteolytic degradation fragments of APOBEC3G, were detected by Western blotting (Fig. 1A). In the presence of Vif, the steady-state levels of APOBEC3G were markedly decreased ( Fig. 1A). Similarly, endogenous APOBEC3G was depleted in a Hut78 cell line that stably expresses Vif (Fig. 1B). Co-transfection of increasing amounts of pNLA1.Vif-FLAG together with a constant amount of pAPOBEC3G:HA showed that Vif expression reduced APOBEC3G levels in a dose-dependent manner (Fig. 1C). This effect was specific for APOBEC3G because levels of the control protein tubulin were unchanged (Fig. 1C). A critical Vif:APOBEC3G ratio was required for the reduction of APOBEC3G levels (Fig. 1C, lanes 5 and 6). At lower Vif: APOBEC3G ratios, Vif had no detectable effect on APOBEC3G levels ( Fig. 1C, lanes 2-4). Most subsequent experiments were therefore performed under conditions in which APOBEC3G levels were expected to be limiting. We then used loss-of-function Vif mutants to further explore the effect of Vif on APOBEC3G levels. The highly conserved cysteine residues at positions 114 and 133 and SLQ(Y/F)LA motif at residues 144 -149 are required for Vif function and HIV-1 replication (18, 25). Lysates were subjected to Western blot analysis using anti-HA or anti-Vif antibodies. D, Western blot analysis of APOBEC3G coexpressed with WT Vif or Vif C114S/C133S . 19.5 g of pCDNA1.Vif or pCDNA1.Vif C114S/C133S, or ⌬142-154 were co-transfected with 0.5 g of pAPOBEC3G:HA. Vif phosphorylation site mutants were expressed by co-transfecting 15 g of proviral plasmid with 0.1 g of pAPOBEC3G:HA. A background band is indicated by an asterisk. E, expression of APOBEC3G mRNA is shown in infected and uninfected H9 cells. Reverse transcriptase PCR was performed as described under "Experimental Procedures" using primers that detect APOBEC3G and ␤ 2 -microglobulin. SupT1 cells were included as a negative control.
Thr 96 , Ser 144 , and Thr 188 are highly conserved phosphorylation sites in Vif. Mutation of Thr 96 or Ser 144 impairs but does not abolish viral replication (21,26). In contrast to wild-type Vif, Vif C114S/C133S and the Vif⌬142-154 mutant that removes the SLQ(Y/F)LA motif showed a markedly reduced ability to decrease APOBEC3G levels ( Fig. 1D), suggesting a relationship between the ability of Vif to reduce APOBEC3G levels and to support virus replication. However, the T96A, S144A, and T188A mutants retained the ability to decrease APOBEC3G levels (Fig. 1D). Thus, some mutations that reduce Vif function during viral replication still have the wild-type ability to decrease APOBEC3G levels. We then investigated whether Vif affects APOBEC3G mRNA levels in H9 cells infected with wild-type and vif-deficient HIV-1 by reverse transcriptase PCR (Fig. 1E). APOBEC3G mRNA levels were unaffected by the presence of Vif. Furthermore, infection of H9 cells with either virus did not appear to increase APOBEC3G mRNA levels compared with uninfected cells. Thus, Vif did not alter APOBEC3G mRNA levels, suggesting that Vif decreases APOBEC3G protein levels at a post-translational step.
Vif Accelerates Degradation of APOBEC3G by the Ubiquitin-Proteasome Pathway-The preceding experiments suggested that Vif might reduce APOBEC3G levels by promoting its degradation. Intracellular protein stability is regulated by lysosomal or proteasome-mediated degradation (27). The ubiquitin-proteasome pathway involves the ubiquitination of target proteins and subsequent degradation by the 26 S proteasome. The presence of high molecular mass species of APOBEC3G at ϳ54 and 60 kDa (Fig. 1A) raised the possibility that it is covalently modified by ubiquitin. The proteasome inhibitors MG132 and clasto-lactacystine ␤-lactone in combination markedly reduced the ability of Vif to induce the degradation of APOBEC3G ( Fig. 2A). Similar results were observed over a range of proteasome inhibitor concentrations and combinations. 2 In contrast, the lysosome inhibitor chloroquine had no effect on APOBEC3G stability (Fig. 2C). The expression of APOBEC3G in the presence of myc-tagged ubiquitin resulted in a shift of the higher molecular weight species of APOBEC3G, suggesting that APOBEC3G is ubiquitinated. 2 This was confirmed by Western blot analysis of immunoprecipitated APOBEC3G with anti-myc antibody, which showed a high molecular weight smear typical of polyubiquitination (Fig. 2B, top). Coexpression of Vif dramatically increased polyubiquitination of APOBEC3G (Fig. 2B, top). The effect of Vif on APOBEC3G ubiquitination was most readily detected at high APOBEC3G expression levels (Fig. 2B, middle). However, APOBEC3G ubiquitination was also observed at lower levels of APOBEC3G expression that are destabilized by Vif (Figs. 1A and 2A). These findings indicate that Vif promotes the polyubiquitination and proteasome-mediated degradation of APOBEC3G.
Polyubiquitination typically results in proteolysis of the substrate protein via the proteasome. Treatment of cells with MG132/␤-lactone increased the levels of polyubiquitinated APOBEC3G, suggesting that APOBEC3G is rapidly degraded by the proteasome (Fig. 2C, top). This was further supported by experiments using the ubiquitin mutant K48R. Multiubiquitin chains linked via Lys 48 are the principal signals for proteasomal degradation (27). Ub K48R is a polyubiquitin chain terminator that reduces the efficiency of proteasome-mediated degradation and stabilizes polyubiquitinated substrates (28). When coexpressed with APOBEC3G and Vif, Ub K48R enhanced the accumulation of polyubiquitinated APOBEC3G (Fig. 2C, top). The accumulation of high molecular weight conjugates in the presence of Ub K48R suggests the attachment of ubiquitin to multiple sites within APOBEC3G and/or the incorporation of endogenous wild-type ubiquitin into the mutant polyubiquitin chains. The accumulation of polyubiquitinated APOBEC3G is probably the consequence of disrupted proteasomal targeting signals or the dynamic remodeling of polyubiquitin chains that occur in the presence of the K48R mutant (28,29). In contrast to the effects of proteasome inhibitors, chloroquine had no significant effect. The proteasome inhibitors and ubiquitin mutants did not affect steady-state levels of APOBEC3G in the absence of Vif (Fig. 2C, bottom), suggesting that Vif accelerates the degradation of APOBEC3G by the ubiquitin-proteasome pathway.
Given that Vif decreases APOBEC3G levels and increases its ubiquitination, we investigated the effect of Vif on the half-life of APOBEC3G in virus-producing cells. APOBEC3G was expressed with WT or ⌬Vif proviruses and analyzed by pulsechase metabolic labeling. APOBEC3G displayed a biphasic decay characterized by rapid degradation in the early phase, followed by a slow decay of a relatively stable APOBEC3G population during the late phase (Fig. 2D). The half-life of APOBEC3G during the early phase was significantly decreased from approximately 29 min in the absence of Vif to 18 min in its presence. Proteasome inhibitors restored the half-life of APOBEC3G in the presence of WT virus to that of APOBEC3G with ⌬Vif virus. 2 The stable population of APOBEC3G detected during the late phase appeared to be unaffected by Vif, which may account for the inability of Vif to induce complete degradation of APOBEC3G (Fig. 1A). These results demonstrate two populations of APOBEC3G that differ in their susceptibility to down-modulation by Vif and are degraded at different rates and suggest that Vif decreases the half-life of the susceptible population during early phase decay by promoting its degradation via the ubiquitin-proteasome pathway.
Vif and APOBEC3G Form a Complex-We next investigated whether Vif interacts with APOBEC3G or promotes its degradation through an indirect mechanism. To determine whether Vif and APOBEC3G interact in a complex, we performed reciprocal co-immunoprecipitation assays after expression of Vif and/or HA-tagged APOBEC3G in 293T cells. A significant amount of APOBEC3G could be specifically co-immunoprecipitated with an anti-Vif antibody (Fig. 3A, top). In a reciprocal experiment, Vif was co-immunoprecipitated by an anti-HA antibody (Fig. 3A, middle). No significant immunoprecipitation of either protein was observed when expressed in the absence of each other. Consistent with the preceding experiments, levels of APOBEC3G were reduced when APOBEC3G was coexpressed with Vif (Fig. 3A, bottom). We also performed experiments with purifed recombinant proteins to determine whether Vif and APOBEC3G can interact directly. APOBEC3G was precipitated with GST-Vif but not GST or the truncated GST-Vif-(40 -160) (Fig. 3B). These results suggest that Vif and APOBEC3G interact directly and form a complex in vivo, which accelerates the degradation of APOBEC3G via the ubiquitinproteasome pathway.
Proteasomal Degradation of Vif Is Enhanced by APOBEC3G-To further investigate the relationship between Vif-dependent degradation of APOBEC3G and viral infectivity, we tested the infectivity of WT and ⌬Vif virus produced in the presence of increasing amounts of APOBEC3G. Increasing expression of APOBEC3G dramatically reduced the infectivity of ⌬Vif virus in a dose-dependent manner (Fig. 4A) as previously reported (3). However, a smaller yet significant decrease in the infectivity of WT HIV-1 was also observed (Fig. 4A). Comparing APOBEC3G expression in 293T cells producing WT or ⌬Vif virus demonstrated lower levels of APOBEC3G in the presence of the WT virus. Surprisingly, we found that steady-state levels of Vif were affected by APOBEC3G expression. High expression of APOBEC3G caused a significant reduction in Vif levels in parallel with a reduction in the infectivity of WT virus (Fig.  4A). These data suggest a reciprocal relationship wherein Vif and APOBEC3G influence the steady-state levels of each other.
To investigate the mechanism by which APOBEC3G decreases the steady-state levels of Vif, we examined whether Vif is ubiquitinated. A higher molecular mass 32-kDa form of Vif was observed in transfected 293T cells (Figs. 1A and 4B). The ϳ8-kDa increase in size suggested covalent modification by ubiquitin. Coexpression of Vif with epitope-tagged ubiquitin resulted in a shift in size of the higher molecular weight form (Fig. 4B, top left). Western blotting of immunoprecipitated Vif with an antibody that recognizes the HA epitope tag on ubiquitin confirmed that Vif is covalently modified by the attachment of ubiquitin (Fig. 4B, top right). In the absence of APOBEC3G, the predominant ubiquitinated species of Vif ap-peared to be modified by monoubiquitin (Fig. 4B, top, data not shown). The chain-terminating mutant Ub K48R had no effect on the ubiquitination profile of Vif, consistent with a monoubiquitinated species (Fig. 4B, bottom). Furthermore, treatment of cells with proteasome inhibitors failed to substantially increase the levels of ubiquitinated Vif (Fig. 4B, bottom). These results demonstrate that in the absence of APOBEC3G a minor fraction of Vif is monoubiquitinated. In the presence of APOBEC3G, however, the amount of monoubiquitinated Vif was increased (Fig. 4C). Furthermore, we observed a dramatic increase in polyubiquitinated Vif when APOBEC3G was present. Coexpression of Vif and APOBEC3G with the mutant Ub K48R showed increased amounts of ubiquitinated Vif compared with that observed with WT ubiquitin (Fig. 4C). We next examined the effects of APOBEC3G on the half-life of Vif. Pulse-chase metabolic labeling showed that Vif is a relatively stable protein with a half-life of ϳ90 min (Fig. 4D). However, in FIG. 2. Vif promotes ubiquitin-dependent proteasomal degradation of APOBEC3G. A, Vif and APOBEC3G were expressed in 293T cells as in Fig. 1A. Transfected cells were treated with MG132/␤-lactone or Me 2 SO for 6 h prior to lysis and Western blotting. B, pNLA1.Vif-FLAG (10 g), pAPOBEC3G:HA (5 g), and pRbG4-His 6 -myc-Ub (5 g) were co-transfected into 293T cells as indicated. DNA was equalized to 20 g with an empty vector where needed. APOBEC3G and Vif were immunoprecipitated and Western blotted as in Fig. 1A. Anti-HA immunoprecipitates were probed with anti-myc to detect ubiquitinated APOBEC3G (top). Ubiquitinated species are indicated. C, 293T cells were transfected as in B with the indicated plasmids and treated with chloroquine or MG132/␤-lactone for 6 h before lysis. APOBEC3G and Vif were detected by Western blotting with anti-HA and anti-Vif antibodies, respectively. D, The half-life of APOBEC3G was analyzed by pulse-chase analysis of 293T cells transfected with pAPOBEC3G:HA and pNLX (ࡗ) or pNLX⌬Vif (f) as described under "Experimental Procedures." APOBEC3G was immunoprecipitated with anti-HA antibody and quantitated by phosphorimaging analysis. Protein levels were normalized to t 0 ϭ 100%. Results are representative of four independent experiments. the presence of APOBEC3G the half-life of Vif decreased to approximately 23 min. Taken together, these results indicate that coexpression of APOBEC3G promotes polyubiquitination and proteasomal degradation of a fraction of Vif. Thus, Vif and APOBEC3G demonstrate an interdependent relationship in which both proteins promote the ubiquitination and degradation of each other.

DISCUSSION
APOBEC3G is a DNA-editing enzyme with antiviral activity that is overcome by Vif (5)(6)(7)9). Vif blocks the encapsidation of APOBEC3G into HIV-1 virions (6), but the mechanism of action by which this occurs is unclear. Here we report that Vif counteracts APOBEC3G by targeting it to the ubiquitin pathway for proteasome-mediated degradation. Vif forms a complex with APOBEC3G through a direct physical interaction and enhances APOBEC3G ubiquitination, resulting in destabilization of APOBEC3G. Vif accelerates the degradation of both ectopically expressed and endogenous APOBEC3G and recently has been shown to deplete APOBEC3G levels in HIV-1infected cells (19). The Vif mutants C114S/C133S and ⌬142-154 that cannot support infection fail to induce degradation of APOBEC3G. Mutation of the Thr 96 and Ser 144 phosphorylation sites within Vif impairs viral replication (26) but does not affect APOBEC3G degradation, suggesting that Vif may be important for other functions in addition to inducing degradation of APOBEC3G by the proteasome. These findings suggest that the primary mechanism of action by which Vif neutralizes APOBEC3G is by promoting its degradation via the ubiquitinproteasome pathway, thereby blocking its incorporation into virions. However, our results also imply that an additional mode of action, such as APOBEC3G binding and sequestration, leading to the inhibition of APOBEC3G incorporation into virions (8) or the recruitment of APOBEC3G away from sites of virus assembly may contribute to the functional activity of Vif and the generation of infectious virus.
Ubiquitination is a tightly regulated cellular mechanism that can control the activity, localization, and turnover rates of many cellular proteins (27). The ubiquitin machinery consists of an enzyme cascade that activates ubiquitin and ultimately transfers it to target proteins. An E1 ubiquitin-activating enzyme transfers ubiquitin to an E2 ubiquitin-conjugating enzyme. Ubiquitin can then be directly conjugated to a protein substrate by the E2, but more commonly an E3 ubiquitin ligase is required to covalently attach ubiquitin to the ⑀-amino group of a lysine residue on the substrate. The E3 ubiquitin ligase conveys the exquisite targeting specificity of the ubiquitination machinery. Vif forms a complex with and accelerates the ubiquitination and degradation of a population of APOBEC3G, raising the possibility that Vif may directly ubiquitinate APOBEC3G. However, Vif has no sequence homology to known E3 enzymes. Furthermore, Vif does not appear to have intrinsic ubiquitin ligase activity, because attempts to reconstitute APOBEC3G degradation in vitro using Vif and purified components of the ubiquitin machinery were unsuccessful. 2 Therefore, it seems unlikely that Vif is the enzyme directly responsible for APOBEC3G ubiquitination. This was confirmed by the recent work of Yu et al. (30), which demonstrated that Vif function is dependent on the enzymatic activity of the Elong-inB-ElonginC-Cullin5 ubiquitin ligase complex and suggested that Vif may function as an F-box-like protein by acting as the specificity factor that links APOBEC3G to the ElonginB-Elong-inC-Cullin5 complex. We found that Vif is monoubiquitinated in the absence of APOBEC3G, although monoubiquitination of Vif does not appear to target the protein for degradation. Rather, monoubiquitination may affect its structure, location, or activity (31). We previously demonstrated that mutation of lysines within the Vif C terminus disrupts membrane targeting of Vif and virus infectivity (32). Whether these lysines serve as ubiquitin conjugation sites remains to be determined. On expression of APOBEC3G a fraction of Vif is polyubiquitinated and rapidly degraded; thus, coexpression of APOBEC3G and Vif decreases the stability of both proteins. This finding, together with the observation that Vif and APOBEC3G form a complex in vitro and in vivo (Fig. 3) and the demonstration that Vif mutants that fail to bind APOBEC3G do not decrease APOBEC3G levels (33), raises the possibility that these two proteins may be coordinately degraded as a complex. However, further studies are required to determine whether Vif and APOBEC3G are degraded within the same complex or independently.
Our findings suggest that the ratio between Vif and APOBEC3G is crucial for maximal effects of Vif on APOBEC3G degradation and viral infectivity. In infected cells, Vif is expressed at high levels, approaching those of Gag (34). In contrast, APOBEC3G appears to be expressed at very low levels in non-permissive cells (19). Thus, it appears that HIV-1 conquers the cellular defenses of APOBEC3G by expressing the high levels of Vif that are needed to achieve a Vif:APOBEC3G ratio that leads to APOBEC3G degradation. A difference in Vif: APOBEC3G ratios may explain why our results differ from those previously reported by Mariani et al. (8), who reported that Vif had no significant effect on the steady-state levels or half-life of APOBEC3G. While this manuscript was under review, other studies were published that support our findings of a Vif-dependent decrease in APOBEC3G half-life and steadystate levels (19,30,33,35). Compared with these studies, our study reaches the novel conclusion that there is a reciprocal destabilizing relationship between Vif and APOBEC3G because coexpression of APOBEC3G promotes Vif ubiquitination and reduces its half-life. These results differ from those reported by Stopak et al. (19) and Marin et al. (33) who did not observe that APOBEC3G destabilizes Vif. These discrepancies may be the result of differences in the ratio of APOBEC3G:Vif, which strongly influences the ability to detect this phenomenon. Furthermore, we performed experiments to determine the effect of APOBEC3G on the half-life and ubiquitination of Vif, whereas these other studies did not. Whether the levels of APOBEC3G used in Fig. 4 are ever approached in vivo during certain conditions that increase APOBEC3G expression, such as mitogenic stimulation (19), remains to be determined. None- theless, we observed that only a fraction of Vif is ubiquitinated and destabilized by coexpression of APOBEC3G, possibly because of limiting amounts of APOBEC3G expressed in these cells or the differential localization of a fraction of both proteins.
In the absence of Vif, we found that APOBEC3G has a relatively short half-life of only 29 min during its early phase of decay. The half-life of APOBEC3G is highly dependent on its level of expression; but in the absence of Vif, the half-life of APOBEC3G increases to around 50 min or 5-6 h when we transfect 0.5 or 5 g, respectively, of pAPOBEC3G:HA plasmid instead of 0.1 g. 2 Endogenous APOBEC3G is expressed at very low levels (19), so its half-life is expected to be short under physiological conditions. Given the editing capabilities of APOBEC3G, its rapid turnover may represent a cellular mechanism to tightly regulate the expression of a potential DNA mutator. Inducing expression of APOBEC3G might bolster the antiviral response and decrease the production of infectious virus but may also be deleterious to the overall fitness of the cell. For example, unregulated expression of the related cytidine deaminase APOBEC1 is oncogenic in animal models (36). APOBEC3G displayed a biphasic decay characterized by rapid degradation in the early phase followed by slow degradation during the late phase. The biphasic decay could be because of differential degradation of APOBEC3G populations localized to distinct compartments within the cell. The majority of APOBEC3G is cytoplasmic with a minor fraction localized to the nucleus (19). Vif, a cytoplasmic protein, was capable of FIG. 4. Vif is polyubiquitinated and degraded by the proteasome in the presence of APOBEC3G. A, WT (f) and ⌬Vif viruses (Ⅺ) were produced in 293T cells co-transfected with provirus (3 g) and increasing amounts of APOBEC3G (0, 0.01, 0.02, 0.05, 0.1, or 0.2 g) adjusted with an empty vector to 4 g total. Infectivity was measured on the Cf2-luc reporter cell line. The infectivity of each virus produced in the absence of APOBEC3G was normalized to 100%. APOBEC3G and Vif were detected by Western blot of lysates from virus-producing cells. Data are representative of three independent experiments. B, 293T cells were transfected with pNLA1.Vif-FLAG (10 g) and mycUb (5 g) or HAUb (5 g), and cell lysates were immunoprecipitated with anti-FLAG antibody and then subjected to anti-Vif (top left) or anti-HA (top right) Western blot. Monoubiquitinated Vif is indicated. Vif-FLAG was expressed with untagged WT or K48R Ub in the presence (bottom right) or absence (bottom left) of MG132/␤-lactone. Tubulin was detected by Western blot with an anti-tubulin antibody. C, APOBEC3G, Vif-FLAG, mycUb, or the mutant mycUbK48R was expressed in 293T as in Fig. 2B. Following anti-FLAG immunoprecipitation, Vif and Ub-Vif were detected by Western blot with anti-Vif or anti-myc, respectively. Monoubiquitinated and polyubiquitinated Vif are indicated. D, the half-life of Vif was determined by pulse-chase metabolic labeling in the presence (ࡗ) or absence (f) of APOBEC3G as described under "Experimental Procedures." Vif was immunoprecipitated with anti-Vif serum and quantitated by phosphorimaging analysis. Protein levels were normalized to t 0 ϭ 100%. Results are representative of four independent experiments. reducing the half-life of APOBEC3G only during early phase decay, consistent with our finding that high levels of Vif did not induce complete degradation of APOBEC3G. Nonetheless, virus with full infectivity was produced. Thus, there appears to be a population of APOBEC3G that is resistant to Vif-dependent degradation and does not influence viral infectivity. Whether this represents the nuclear fraction or a cytoplasmic fraction localized away from sites of virus assembly remains to be determined.
We propose that Vif functions as an adaptor molecule, recruiting APOBEC3G to the ubiquitin-proteasome machinery. Other viral proteins are known to subvert the ubiquitination machinery by a similar mechanism. The papillomavirus E6 oncoprotein bridges the tumor suppressor p53 to the ubiquitin ligase E6-AP (37). The HIV-1 Vpu protein scaffolds a ternary complex with CD4 and the E3 ␤TrCP (38). Cellular proteins can also function in an analogous manner with Grb10 recruiting insulin-like growth factor I receptor to the Nedd4 ubiquitin ligase (39). In each case, target proteins are recruited to the ubiquitin machinery by their cognate adaptor, polyubiquitinated, and degraded by the proteasome. Unlike Vif, however, neither E6, Vpu, nor Grb10 is ubiquitinated and degraded on coexpression of the target protein. The interaction between Vif, APOBEC3G, and the ubiquitin machinery implicates the proteasome as a site of dynamic interplay between viral and host defenses and provides a new interface for pharmaceutical intervention.