HIV-1 Vpr Reprograms CLR4DCAF1 E3 Ubiquitin Ligase to Antagonize Exonuclease 1-Mediated Restriction of HIV-1 Infection

HIV-1 polymerase reverse transcribes the viral RNA genome into imperfectly double-stranded proviral DNA, containing gaps and flaps, for integration into the host cell chromosome. HIV-1 reverse transcripts share characteristics with cellular DNA replication intermediates and are thought to be converted into fully double-stranded DNA by cellular postreplication DNA repair enzymes. Therefore, the finding that the HIV-1 accessory protein Vpr antagonizes select postreplication DNA repair enzymes that can process HIV-1 reverse transcripts has been surprising. Here, we show that one such Vpr-antagonized enzyme, exonuclease 1, inhibits HIV-1 replication in T cells. We identify exonuclease 1 as a member of a new class of HIV-1 restriction factors in T cells and propose that certain modes of DNA “repair” inhibit HIV-1 infection.

The human immunodeficiency virus type 1 (HIV-1) accessory proteins Vpr, Vif, and Vpu reprogram cellular E3 ubiquitin ligases to direct key host cell proteins that restrict HIV infection for proteasome-dependent degradation. In particular, Vif usurps a cellular RING ubiquitin ligase assembled on a Cullin 5 scaffold (CRL5) to antagonize APOBEC3 family cytidine deaminases, which are capable of inducing lethal G-to-A hypermutation in viral cDNA (4). Vpu hijacks a CRL1 E3 ubiquitin ligase complex to antagonize the CD4 receptor (5), thereby preventing superinfection, and to deplete select proteins that are involved in interferon-inducible antiviral pathways (6). Overall, how Vif and Vpu exploit cellular E3 enzymes to facilitate HIV-1 infection is relatively well understood.
In contrast to those of Vif and Vpu, the biochemical mechanisms mediating Vpr function in HIV-1 infection are less clear. Vpr hijacks a multisubunit CRL4 DCAF1 E3 ubiquitin ligase complex comprising the Cullin 4 (Cul4) scaffold, DDB1 linker protein, DDA1 core subunit (7), and DCAF1 substrate receptor, to which Vpr binds (8)(9)(10). Several tentative target proteins of the Vpr-CRL4 DCAF1 E3 were recently identified, four of which are bona fide Vpr-recruited substrates (11)(12)(13). The latter includes proteins identified by unbiased screens and have well-established roles in postreplication DNA repair, such as uracil-DNA glycosylase 2 (UNG2) (14), MUS81 (15), and helicase like transcription factor (HLTF) (16,17), or in epigenetic control of gene expression, such as ten eleven translocation methylcytosine dioxygenase 2 (Tet2) (11), which, notably, is also involved in DNA damage response and repair (18). More specifically, UNG2, a nuclear isoform of UNG, excises uracil from DNA that results from misincorporation of dUMP by DNA polymerase or from cytosine deamination, thus initiating base excision repair (19). dUTP concentrations are relatively high in macrophages, and uracil incorporation may flag viral DNA for restrictive processing by cellular DNA repair enzymes in these cells (20,21). MUS81 and HLTF process branched DNA structures, such as those found at stalled replication forks. The MUS81-EME1 (or -EME2) heterodimer is a DNA structure-selective endonuclease involved in homologous recombination and replication fork processing and repair (22,23). HLTF is a DNA helicase which catalyzes strand reversal at damaged replication forks (24).
The recently revealed narrow clustering of biochemically validated, Vpr-antagonized targets within postreplication DNA repair machinery is intriguing, as it suggests that DNA repair enzymes might impinge on HIV-1 replication to a much larger extent than currently appreciated. Noteworthy, many of the previously reported effects of HIV-1 Vpr on HIV-1 replication, including the modulation of the levels of reverse transcription intermediates, the facilitation of nuclear import of HIV-1 preintegration complex, the modulation of HIV-1 mutation rate in macrophages (25), and the modulation of innate sensing of HIV-1 infection (26), might be indirect consequences/manifestations of the counteraction of DNA repair processing of HIV-1 DNA by Vpr.
Vpr also elicits hallmarks of cellular responses to damaged DNA, probably by activating the DNA damage checkpoint controlled by ataxia telangiectasia and Rad3related protein (ATR) kinase (27). Significantly, Vpr interaction with CRL4 DCAF1 E3 has been genetically linked to its effects on DNA damage checkpoint activation (7,28,29) and innate sensing of HIV-1 infection via the cytosolic DNA sensor cyclic GMP-AMP synthase (cGAS) (26). Yet, the so-far-identified targets of Vpr antagonism do not fully explain the reported cellular and virologic effects of Vpr expression, suggesting that additional substrates of the Vpr-CRL4 DCAF1 E3 may exist.
Here, we carried out a focused screen aiming to more fully define the spectrum of postreplication DNA repair proteins antagonized by HIV-1 Vpr via CRL4 DCAF1 E3. Our studies identify exonuclease 1 (Exo1) as a new target of HIV-1 Vpr antagonism. Exo1 is a multifunctional nuclease involved in multiple DNA repair pathways, including nucleotide excision repair (NER), resection of DNA ends during single-and double-stranded break repair, and homologous recombination processes (30). We show that (i) Exo1 levels are depleted in HIV-1-infected CD4 ϩ T cells in a Vpr-dependent manner, (ii) Vpr loads Exo1 onto the CRL4 DCAF1 E3 complex in vivo and in vitro, and (iii) the ability to antagonize Exo1 is a conserved function of Vpr from the main groups of HIV-1 strains and their ancestor simian immunodeficiency viruses from chimpanzees (SIVcpzs). To-gether, our studies identify a key postreplication DNA repair enzyme as a target of HIV-1 Vpr, reinforce the finding that Vpr usurps CRL4 DCAF1 to antagonize major DNA repair mechanisms, provide indirect support of the model in which the cellular postreplication DNA machinery restricts HIV-1 replication, and provide a stage for systematic dissection of the roles of Vpr-targeted DNA repair enzymes in the HIV-1 replication cycle.

RESULTS
Targeted screen for HIV-1 Vpr-CRL DCAF1 E3 substrate proteins involved in postreplication DNA repair identifies Exo1 as a Vpr target. We set up a targeted screen aiming to identify novel proteins that are involved in postreplication DNA repair and are directed for proteasome-mediated degradation via HIV-1 Vpr bound to CLR4 DCAF1 E3. To this end, we searched for proteins that were depleted in U2OS cells inducibly expressing a well-characterized HIV-1 NL4-3 Vpr protein (U2OS-iH1vpr). Levels of approximately 30 proteins, including DNA damage sensors, nucleases, and helicases involved in major postreplication DNA repair pathways, such as base and nucleotide excision repair, DNA strand break processing, and recombination, were characterized. More specifically, lysates prepared from U2OS-iH1vpr cells induced with doxycycline to express Vpr or prepared from noninduced control U2OS-iH1vpr and parental U2OS cell populations were characterized by immunoblotting. As shown in Fig. S1 in the supplemental material, these experiments revealed that exonuclease 1 (Exo1) levels are specifically depleted in U2OS-iH1vpr cells following induction of Vpr expression but not in control cell populations that did not express Vpr, as were the levels of UNG2, as expected. Thus, our targeted screen tentatively identified Exo1 as a novel target of the CRL4 DCAF1 E3 subverted by HIV-1 Vpr.
HIV-1 infection depletes Exo1 levels in human CD4 ؉ T cells in a Vpr-dependent manner. To validate Exo1 as bona fide Vpr target, we began by testing whether Exo1 is depleted in HIV-1-infected CD4 ϩ T cells. CD4 ϩ T cells were isolated from human peripheral blood mononuclear cells by positive selection, activated with CD3/CD28 beads, and challenged 24 h later with single-cycle HIV-1 NL4-3-derived viruses possessing either an intact (NL4-3.GFP.Rϩ) or disrupted (NL4-3.GFP.R-) vpr gene and expressing a green fluorescent protein (GFP) marker protein (16). Two days postinfection, the productively infected cells were isolated by cell sorting for GFP fluorescence, and Exo1 levels in lysates prepared from the sorted cells were assessed by immunoblotting. As shown in Fig. 1A, Exo1 levels were depleted in cells infected with HIV-1 harboring the intact, but not the disrupted, vpr gene. The infected cell lysates were also blotted for HLTF, MUS81, and UNG2, previously validated direct substrates of HIV-1 Vpr-CRL4 DCAF1 E3 involved in postreplication DNA repair (16,17,31). The extent of Exo1 depletion in cells infected with HIV-1 expressing Vpr was comparable to that of HLTF and more pronounced than that seen for MUS81.
Next, we examined the kinetics of Exo1 depletion by Vpr and compared them to those of other Vpr-recruited substrates of Vpr-CRL4 DCAF1 E3. To this end, U2OS-iH1vpr cells were induced with doxycycline to express Vpr and collected at various times postinduction. The levels of Vpr targets in cell lysates were subsequently characterized by immunoblotting. Figure 1B shows that Exo1 levels were depleted with kinetics similar to those seen for HLTF, in line with the data from primary CD4 ϩ T cells. We conclude that HIV-1 infection depletes Exo1 levels in infected CD4 ϩ T cells in a Vpr-dependent manner to an extent similar to that seen for previously validated targets of Vpr-CRL4 DCAF1 E3.
Exo1 is a conserved target of HIV-1 and SIVcpz lineage Vpr. To assess the generality of our finding, we next tested Vpr proteins from the main groups of HIV-1 and closely related SIVcpzs, the latter persisting in chimpanzees (32). U2OS cell populations were engineered to inducibly express synthetic consensus Vpr proteins representative of HIV-1 groups M, N, and O as well as those representative of Vpr proteins encoded by two distinct populations of SIVcpzs isolated from two chimpanzee subspecies: Pan troglodytes troglodytes and Pan troglodytes schweinfurthii. The cells were induced to express Vpr, cell lysates were prepared 24 h postinduction, and Exo1 levels were characterized by immunoblotting (Fig. 1C). We found that all tested consensus HIV-1 and SIVcpz Vpr proteins depleted Exo1 levels, similar to what occurred with HIV-1 NL4-3 Vpr. We conclude that Exo1 is a common target of Vpr proteins of the HIV-1/ SIVcpz lineage viruses.
Exo1 depletion by Vpr is cell cycle phase independent. Vpr induces cell cycle arrest at the G 2 -phase DNA damage checkpoint. To exclude the possibility that Exo1 levels are physiologically downregulated in G 2 and the observed Exo1 depletion is an indirect consequence of the accumulation of Vpr-expressing cells at the G 2 checkpoint, we asked whether Vpr can deplete Exo1 outside the G 2 phase, as we showed previously for HLTF (16). To address this, U2OS-iH1vpr cells were synchronized at the G 1 /S border by double-thymidine block. Vpr expression was then induced with doxycycline 8 h following the initiation of the second thymidine treatment, and cells were harvested at different time points after doxycycline induction for determination of Exo1 levels by immunoblotting. As controls, we used parental U2OS cells and U2OS cells inducibly expressing Vpr with the H71R mutation [Vpr(H71R)], which disrupts Vpr binding to the DCAF1 substrate receptor subunit of the CRL4 DCAF1 E3 complex (7). DNA profiles of cells collected at the time of initiation of the doxycycline treatment and at the 24-h time point clearly show that the cells remained arrested near G 1 /S transition throughout the duration of the experiment (Fig. 2). Exo1 levels were depleted within 6 h of the doxycycline induction of wild-type Vpr, similar to what occurred with those of HLTF. Significantly, the Vpr(H71R) mutant did not have such an effect. The finding that Vpr can deplete Exo1 at the G 1 /S border demonstrates that Exo1 depletion is not an indirect consequence of the Vpr-induced G 2 arrest. The observation that Vpr(H71R) does not deplete Exo1 tentatively links the depletion to Vpr subversion of the CRL4 DCAF1 E3 and ubiquitin proteasome system.
Vpr depletes Exo1 by a proteasome-dependent mechanism that is independent of cyclin F (CCNF) and ATR. Next, we asked whether Vpr targets Exo1 for degradation by a proteasome-dependent mechanism. To test this, U2OS-iH1vpr and control cells were cultured with doxycycline in the presence or absence of the MG132 proteasome inhibitor. Twenty-four hours postinduction, cell lysates were prepared and Exo1 levels, along with those for MUS81 and HLTF, were revealed by immunoblotting. As shown in Fig. 3A, proteasome inhibition prevented the depletion of Exo1 levels in Vpr-expressing cells. Exo1 can be degraded by a proteasome-dependent mechanism in response to DNA damage and/or replication stress. In particular, following UV irradiation, Exo1 degradation is mediated by the SCF-CCNF E3 complex (33); further, following DNA doublestranded break induction, Exo1 is rapidly degraded by the ubiquitin-proteasome system via an ATR-controlled mechanism (34). Since Vpr expression induces hallmarks of the cellular response to damaged DNA (35,36), we asked whether any of the above-described mechanisms mediates the Vpr-dependent depletion of Exo1.
To address the role of the SCF-CCNF, U2OS-iH1vpr cells were subjected to RNA interference (RNAi) targeting CCNF (F) or nontargeting (NT) RNAi and 2 days later induced to express Vpr. It is evident from the data shown in Fig. 3B that CCNF depletion did not restore Exo1 levels in cells expressing Vpr. Importantly, RNAi-mediated CCNF depletion did inhibit Exo1 degradation in response to hydroxyurea (HU)-induced replication stress. The latter control indicates that RNAi to CCNF was deep enough to disrupt SCF-CCNF E3-mediated Exo1 degradation.
We then asked whether ATR signaling is required for Vpr-induced Exo1 degradation. U2OS-iH1vpr cells were induced with doxycycline in the absence or presence of the VE-821 ATR kinase inhibitor. As shown in Fig. 3C, VE-821 treatment did not prevent HIV-1 Vpr Antagonizes Exonuclease 1 ® Exo1 depletion in Vpr-expressing cells. Of note, the VE-821 treatment noticeably suppressed ATR-mediated CHK1 phosphorylation on serine S345 (CHK1P) (37). To corroborate the finding that VE-821 indeed inhibited ATR in this experiment, U2OS cells were exposed to HU, which activates the ATR-controlled checkpoint, leading to a much more robust CHK1 phosphorylation than that induced by Vpr, in the presence or absence of the same VE-821 concentration. Data in Fig. 3D clearly show that VE-821 prevented the HU-induced ATR-mediated CHK1 phosphorylation in this experimental setting. We conclude that the canonical degradation mechanisms linked to cellular response to damaged DNA are not likely to mediate HIV-1 Vpr-induced Exo1 degradation. These findings strengthen the possibility that Exo1 is a direct target of the hijacked by HIV-1 Vpr CRL4 DCAF1 E3 ubiquitin ligase.
Vpr recruits Exo1 to CRL4 DCAF1 E3 for polyubiquitinaltion. We next asked whether HIV-1 Vpr connects Exo1 to the endogenously expressed DCAF1 and DDB1 subunits of the CRL4 DCAF1 E3 complex ( Fig. 4A and B). FLAG-tagged Exo1 was expressed in HEK293T cells, alone or with hemagglutinin epitope (HA)-tagged wild-type HIV-1 NL4-3 Vpr or the Vpr(F69A) variant, which does not bind DCAF1 (12). Anti-FLAG-Exo1 immune complexes were isolated from lysates prepared from the transfected cells and immunoblotted for Exo1, DDB1, DCAF1, and Vpr. We found that wild-type Vpr promoted the assembly of a protein complex containing Exo1 and the DCAF1-DDB1 module of the E3 complex. In contrast, no such effect was observed with the Vpr(F69A) mutant, which does not bind DCAF1, even though this mutant associated with Exo1 to a degree similar to that of wild-type Vpr. Low levels of DCAF1 and DDB1 that coprecipitated with Exo1 in the presence of Vpr(F69A) were similar to those seen when Exo1 complexes were formed in the absence of Vpr. Whether the latter reflects a residual nonspecific association of the immune complexes with Exo1 or a low level of Exo1 binding to DDB1-DCAF1 in the absence of Vpr is not yet known. Regardless, our data clearly show that wild-type Vpr specifically recruits Exo1 to the CRL4 DCAF1 E3 complex, via its binding to the DCAF1 subunit.
To corroborate and extend the above findings, we asked whether Exo1 is a Vprdependent substrate of the CRL4 DCAF1 E3 ubiquitin ligase, using an in vitro ubiquitination system reconstituted with recombinant proteins (13,31). As shown in Fig. 4B, CRL4 DCAF1 E3 ligase in complex with Vpr mediated polyubiquitination of Exo1. In contrast, no polyubiquitinated species of Exo1 were observed upon incubation of Exo1 with E3 ligase alone. These data provide further support to a model in which Vpr directly loads Exo1 onto the CRL4 DCAF1 E3 ligase for polyubiquitination and subsequent degradation by the proteasome.
Exo1 depletion alone does not explain Vpr's ability to arrest cells in the G 2 cell cycle phase. HIV-1 Vpr activates the DNA damage checkpoint that is controlled by the ATR kinase, leading to cell cycle arrest in the G 2 phase (27). This effect might reflect Vpr antagonism of a protein(s) mediating postreplication DNA repair (16). Since the exact mechanism leading to checkpoint activation by Vpr is not yet known, we asked whether Exo1 depletion by Vpr plays a role. To this end, U2OS-iH1vpr cells were induced to express Vpr or subjected to RNAi with small interfering RNAs (siRNAs) targeting two distinct regions in Exo1 mRNA. The cells were harvested 48 h later, and Exo1 levels in cell lysates were assessed by immunoblotting and cell cycle profiles characterized by flow cytometry. As shown in Fig. 5, Exo1 levels were reduced following induction of Vpr expression, and the cells exhibited an altered cell cycle distribution with decreased G 1 and S and increased G 2 /M cell populations, characteristics of the HIV-1 Vpr-induced cell cycle perturbations. Remarkably, even though Exo1 levels following RNAi were considerably lower than those seen in Vpr-expressing cells, RNAi-mediated Exo1 depletion was not sufficient to copy the Vpr-induced cell cycle arrest phenotype. Thus, Vpr-mediated Exo1 depletion alone appears insufficient to mediate deregulation of the cell cycle resulting from Vpr expression.

Vpr targets Exo1's C terminus via the Exo1 PCNA-interacting protein (PIP) binding motif.
To determine whether Vpr targets a known functional element within the Exo1 molecule, we next mapped the Exo1 region that mediates the effect of Vpr. A set of N-and C-terminal Exo1 deletion mutants were generated by (i) removing the N-terminal nuclease domain, (ii) removing a C-terminal region that is unique to the longer (L) of the two Exo1 isoforms, or (iii) deleting/disrupting domains that mediate binding with the mismatch repair proteins MLH1, MSH2, and MSH3, as well as additional regions whose functional roles have not yet been established (Fig. 6A) (30). Since several of the deletions removed the native Exo1 nuclear localization signal (NLS) and Exo1 interaction with Vpr and CRL4 DCAF1 E3 probably occurs in the nuclear compartment, we restored the nuclear targeting of the variant proteins by fusing them to the well-characterized simian virus 40 (SV40) large T antigen NLS.
As shown in Fig. 6B, the Exo1(1-805) variant, corresponding to the shorter Exo1 isoform, was fully sensitive to Vpr expression, whereas the Exo1(1-784) mutant was resistant. This indicates that Vpr targets both Exo1 isoforms for degradation and that Exo1 sequence between residues 784 and 805 was required for the effect of Vpr. Analyses of additional mutants, shown in Fig. 6C, revealed that the N-proximal Exo1 region spanning the nuclease domain, MSH3, and the MLH1 binding site N-proximal segment was dispensable. This narrowed down the Exo1 region required for susceptibility to Vpr to residues located between positions 594 and 805 (Fig. 6C, left). This region comprises part of the Exo1 MSH2 binding element and a putative PCNA-binding motif (PIP box, residues 788 to 795) (38). Interestingly, Exo1 variants with C-terminal deletions that disrupt the integrity of the PIP box (1 to 790) (Fig. 6D), or completely delete it (1 to 784, 1 to 749) (Fig. 6B), were not depleted by Vpr. To directly address the possibility that Vpr targets Exo1 via its PIP box, we tested the effect of alanine substitution for hydrophobic residues in the PIP box (788QIKLNELW¡788AIKANEAA) in the context of full-length Exo1 protein [Exo1.1-846(PIP)]. Significantly, the levels of Exo1.1-846(PIP) mutant were not downregulated by Vpr (Fig. 6C, right). We conclude that the PIP box is necessary for Exo1 binding by Vpr and may constitute the Vpr binding site.
To further validate that the PIP box is directly targeted by Vpr, we subjected two MBP-Exo1 fusion peptides (residues 749 to 805 and 749 to 784) to in vitro ubiquitination assays with CRL4 DCAF1 E3 reconstituted with or without Vpr (Fig. 6E)   As shown in Fig. 7A, the intact wild-type vpr gene conferred an approximately 10-fold replication advantage in CEM.SS_iExo1r cells subjected to the control nontargeting RNAi over a 7-day period, compared to the vpr.Q8* condition (compare box 1 with box 4). We estimate that Vpr stimulated HIV-1 replication in CEM.SS T cells by approximately 20% to 40% per infection cycle. Of note, detection of such a modest, yet reproducible, effect was made possible by selecting an early assay endpoint, well before meaningful depletion of target cells and because HIV-1 DNA levels were quantified relative to cellular DNA.
The HIV-1(vpr.wt) virus replicated to higher levels in Exo1-depleted CEM.SS_iExo1r cells than in control NT cells over a 7-day period (ϳ3-fold, P Ͻ 0.00005) (Fig. 7A, right, boxes 1 and 2). This effect was also noted for HIV-1(vpr.Q8*) though less pronounced and not statistically significant using one-way analysis of variance (ANOVA) (ϳ1.5-fold, P Ͼ 0.05) (Fig. 7A, right, boxes 4 and 5). Of note, the Exo1-depleted and control CEM.SS_iExo1r cells proliferated at similar rates; this excluded the possibility that differences in HIV-1 DNA levels reflected a difference in the proliferation rates of the cells subjected to nontargeting versus Exo1-targeting RNAi (Fig. S2). Importantly, ectopic expression of the RNAi-resistant Exo1 variant restored restriction of HIV-1 infection in the endogenous Exo1-depleted CEM.SS_iExo1r cells (Fig. 7A, left and right,  Doxϩ, boxes 3 and 6). This ruled out a potential involvement of an off-target effect of RNAi on Exo1 upon HIV-1 replication. We conclude that Exo1 restricts HIV-1 replication and that this restriction is partially counteracted by HIV-1 Vpr.
Significantly, it is evident from the data shown in Fig. 7A (left panel, box 5 versus box 1) that Exo1 depletion did not restore replication of HIV-1(vpr.Q8*) to levels seen for HIV-1(vpr.wt). Furthermore, Exo1 depletion relieved the replication of HIV-1(vpr.Q8*) to a much smaller extent than that seen for HIV-1(vpr.wt), the former remaining largely restricted (Fig. 7A, right, ϳ1.5-fold versus ϳ3.0-fold, respectively). These data are best explained by a model in which HIV-1 replication in CEM.SS T cells is restricted not only by Exo1 but also by one or more additional mechanisms which are also counteracted by Vpr. In such a case, in Exo1-depleted cells, vpr-deficient HIV-1(vpr.Q8*) remains restricted by these additional mechanisms, whereas HIV-1(vpr.wt) is relieved to a larger extent due to the presence of functional vpr, which counteracts them.

DISCUSSION
Viral accessory virulence factors hijack host cell E3 ubiquitin ligases to disable innate/intrinsic defenses and thereby provide a more permissive environment for virus replication. Here, we show that Vpr proteins of HIV-1 and its ancestor SIVcpz lineage reprogram CRL4 DCAF1 E3 to antagonize the postreplication DNA repair enzyme Exo1. We show that Vpr recruits Exo1 to the CRL4 DCAF1 E3 complex for ubiquitination and subsequent proteasome-dependent degradation, and we corroborate this model by excluding the possibility that previously identified CRL4 DCAF1 E3-independent mechanisms are involved in Vpr-mediated Exo1 degradation. Finally, our data show that Exo1 exerts a modest inhibitory effect on HIV-1 replication in T cells, consistent with the model in which Vpr reprograms CRL4 DCAF1 E3 to counteract this restriction. Notably, Exo1 is the fourth postreplication DNA repair enzyme, along with the previously described UNG2, MUS81, and HLTF, that is antagonized by HIV-1 Vpr via CRL4 DCAF1 E3 (13,31,39). Together, these findings reveal that HIV-1 Vpr extensively remodels the cellular postreplication DNA repair machinery by impinging on multiple repair pathways and thereby promotes HIV-1 replication.
Our studies identify Exo1 as a direct target of HIV-1 Vpr for recruitment to CRL4 DCAF1 E3 and subsequent proteasome-dependent degradation. Several lines of evidence support direct binding of Vpr with Exo1. First, we find that levels of known Exo1 interaction partners, including the mismatch repair proteins MSH2, MSH3, and MLH1 (30) and subunits of the MRN complex (40), which recruits Exo1 for DNA end resection, are unaltered in Vpr-expressing cells (see Fig. S1 in the supplemental material), consistent with the possibility that Vpr targets Exo1 alone, rather than in a complex with any of the above proteins. Second, Exo1 regions that comprise binding sites for its known interaction partners, with the exception of PCNA, are also dispensable for the effect of Vpr. Regarding PCNA, our data clearly show that Exo1 PIP box mutants resist depletion by Vpr in cell-based assays (Fig. 6). Although this finding allows for the possibility that PCNA bridges Vpr binding with Exo1, we also show that Vpr can load full-length Exo1 as well as the PIP box onto the CRL4 DCAF1 E3 complex for polyubiquitination in the absence of PCNA in a defined in vitro system. Furthermore, we observed previously that PCNA levels are unaltered in HIV-1 Vpr-expressing cells under conditions of Exo1 depletion (16); this again favors the possibility that Vpr targets Exo1 independently of PCNA. Overall, our evidence supports the model in which Vpr binds Exo1 directly, via its PIP box, for loading onto the CRL4 DCAF1 E3 complex.
The Exo1 PIP box mediates Exo1 loading at sites of DNA damage by PCNA (38). In this context, it is interesting to note that HIV-1 Vpr captures UNG2 and HLTF via their DNA binding domains by mimicking DNA (13,39). Thus, it appears that Vpr loads its target DNA repair proteins onto the CRL4 DCAF1 E3 complex by binding to their structured elements that mediate recruitment to sites of DNA damage. Notably, the strategy of antagonizing innate immune proteins by mimicking binding of their natural functionally important ligands minimizes the chances of selecting escape mutations, which would likely compromise the binding surface functionality.
The results of our spreading infection assays in Exo1-depleted CEM.SS T cells clearly show that Exo1 inhibits HIV-1 replication. It is noteworthy that, upon Exo1 depletion, the increase in replication of HIV-1(vpr.Q8*), lacking functional vpr, was relatively small compared to that seen for HIV-1(vpr.wt), indicating that vpr-deficient HIV-1 remained largely restricted. This finding would not be expected if Exo1-mediated restriction were the only Vpr-counteracted mechanism inhibiting HIV-1 replication in these cells. Although we cannot exclude the possibility that this was due to incomplete Exo1 depletion by RNAi, overall, our evidence points to a role for additional restrictions on HIV-1 replication that are counteracted by Vpr.
One could speculate, based on the available data, that Exo1 restricts an early step(s) in HIV-1 infection. Vpr is incorporated into HIV-1 virions in producer cells and, thus, is present in target cells during early postentry events (41). Notably, HIV-1 plus-strand reverse transcripts are primed from multiple primers (42,43) and contain 5= flaps, gaps, and RNA/DNA hybrids, which resemble canonical substrates processed by Exo1 5=-3= exonuclease, flap nuclease, and/or RNase H (30). There are no known checkpoints that would prevent incompletely reverse transcribed HIV-1 genomes from being imported into the interphase nucleus, where they can be accessed by cellular postreplication DNA repair machinery. Hence, we speculate that virion-carried Vpr antagonizes Exo1 to prevent processing of HIV-1 reverse transcripts and/or integration intermediates, as such processing (specifically by Exo1) would interfere with the ordered progression of HIV-1 reverse transcription and/or repair at provirus integration sites and thereby directly inhibit early steps in HIV-1 infection. It is also possible that Vpr depletes Exo1 in producer cells to prevent Exo1 incorporation into the HIV-1 virion for restricting the infection in the target cell. We do not favor this possibility, however, because Exo1 levels are not detectable by immunoblotting in HIV-1 virions produced in the absence of Vpr (M. C. Shun and J. Skowronski, unpublished data).
The failure of RNAi-mediated Exo1 depletion to restore replication of vpr-deficient HIV-1(vpr.Q8*) to levels seen for HIV-1(vpr.wt) is evidence that Exo1 is not the only Vpr-antagonized protein inhibiting HIV-1 infection in CEM.SS T cells. Obvious candidates are suggested by previous findings that HIV-1 Vpr antagonizes HLTF and MUS81-EME1(EME2) postreplication DNA repair enzymes and TET2, also via CRL4 DCAF1 E3. It is worth noting that the former antagonisms are conserved functions of Vpr proteins of HIV-1/SIVcpz lineages (16), like the antagonism of Exo1. Unlike Exo1, HLTF and MUS81 process generic branched DNA structures, such as those transiently generated during HIV-1 reverse transcription (42,43). Hence, they could partner with Exo1, each processing a distinct set of HIV-1 replication intermediates, to restrict infection. Regardless of the exact mechanism, the coordinated removal of a subset of enzymes that process branched DNA structures via Vpr-CRL4 DCAF1 E3 supports the notion that they are also likely to impinge on HIV-1 replication.
It is conceivable that the restrictions exerted by the DNA repair enzymes antagonized by Vpr may be more robust in specific cell types and/or when reverse transcription is suboptimal, due to exceedingly low canonical dNTP concentrations, such as in monocyte-derived cell lineages and/or quiescent/minimally activated CD4 ϩ T cells, or when other constraints are placed by intrinsic-recognition HIV-1 reverse transcripts. Such conditions, once identified, might provide a more robust experimental system(s) for follow-up virologic/mechanistic studies. Since the Vpr antagonism of postreplication DNA repair enzymes is conserved across HIV-1/SIVcpz clades, there is little doubt that cellular DNA repair machinery plays an important, yet poorly understood, role in HIV-1 replication.

MATERIALS AND METHODS
Ethics statement. This study was approved by the University Hospitals Institutional Review Board (IRB) (IRB number 12-09-23).
Expression constructs and genes. HIV-1 group M, N, and O, P. troglodytes troglodytes SIVcpz, and P. troglodytes schweinfurthii SIVcpz consensus vpr genes were previously described (16). FLAG-AU1tagged vpr cDNAs were subcloned into the pLVX TetOne Puro vector (Clonetech). The human Exo1 cDNA was amplified by PCR from an EST clone (MHS6278-202826750; Open Biosystems) and subcloned into the pCG vector with an N-terminal myc, HA, or FLAG-AU1 epitope tag. Exo1 deletion mutants were constructed by PCR. Those lacking the native NLS sequence were tagged at their N terminus with a peptide comprising the FLAG-AU1 epitope tag followed by the SV40 NLS sequence and a GA di-amino acid linker (MADYKDDDDKGDTYRYIGAGARPKKKRKVGAR). An RNAi-resistant Exo1r gene was constructed by introducing silent mutations into Exo1 sequences targeted by Exo1mir30E.33 (ACGACAAGCCAATCT TCTT¡ACGCCAGGCTAACCTCCTC) and Exo1mir30E.34 (AGATGTAGCACGTAATTCA¡AGACGTTGCTCGAA ACTCT) and cloned between BamHI and XhoI sites in the pEasiLv-puro polylinker.
Yan et al.