HIV-1 and HIV-2 exhibit divergent interactions withHLTF and UNG2 DNA repair proteinsKasia Hreckaa,1, Caili Haoa,1, Ming-Chieh Shuna, Sarabpreet Kaura, Selene K. Swansonb, Laurence Florensb,Michael P. Washburnb,c, and Jacek Skowronskia,2

aDepartment of Molecular Biology and Microbiology, Case Western Reserve School of Medicine, Cleveland, OH 44106; bStowers Institute for MedicalResearch, Kansas City, MO 64110; and cDepartment of Pathology and Laboratory Medicine, University of Kansas Medical Center, Kansas City, KS 66160

Edited by Stephen P. Goff, Columbia University College of Physicians and Surgeons, New York, NY, and approved May 20, 2016 (received for review March29, 2016)

HIV replication in nondividing host cells occurs in the presenceof high concentrations of noncanonical dUTP, apolipoprotein BmRNA-editing, enzyme-catalytic, polypeptide-like 3 (APOBEC3)cytidine deaminases, and SAMHD1 (a cell cycle-regulated dNTPtriphosphohydrolase) dNTPase, which maintains low concentra-tions of canonical dNTPs in these cells. These conditions favorthe introduction of marks of DNA damage into viral cDNA, andthereby prime it for processing by DNA repair enzymes. Accessoryprotein Vpr, found in all primate lentiviruses, and its HIV-2/simianimmunodeficiency virus (SIV) SIVsm paralogue Vpx, hijack theCRL4DCAF1 E3 ubiquitin ligase to alleviate some of these conditions,but the extent of their interactions with DNA repair proteins hasnot been thoroughly characterized. Here, we identify HLTF, a post-replication DNA repair helicase, as a common target of HIV-1/SIVcpz Vpr proteins. We show that HIV-1 Vpr reprogramsCRL4DCAF1 E3 to direct HLTF for proteasome-dependent degrada-tion independent from previously reported Vpr interactions withbase excision repair enzyme uracil DNA glycosylase (UNG2) andcrossover junction endonuclease MUS81, which Vpr also directsfor degradation via CRL4DCAF1 E3. Thus, separate functions ofHIV-1 Vpr usurp CRL4DCAF1 E3 to remove key enzymes in threeDNA repair pathways. In contrast, we find that HIV-2 Vpr is unableto efficiently program HLTF or UNG2 for degradation. Our find-ings reveal complex interactions between HIV-1 and the DNA re-pair machinery, suggesting that DNA repair plays important rolesin the HIV-1 life cycle. The divergent interactions of HIV-1 andHIV-2 with DNA repair enzymes and SAMHD1 imply that theseviruses use different strategies to guard their genomes and facil-itate their replication in the host.

HIV | Vpr | postreplication DNA repair | SAMHD1 | restriction

Nondividing memory T cells and myeloid cells are the maintargets of primate lentiviruses during the initial weeks of the

acute, in vivo infection (1–4). Infection of these cells is inhibitedby intrinsic and innate antiviral mechanisms, several of which con-verge on reverse transcription of the viral RNA genome. One suchrestriction is imposed by SAMHD1, a cell cycle-regulated dNTPtriphosphohydrolase that, in G1-phase leukocytes, maintains theconcentrations of canonical dNTPs below the threshold required forefficient reverse transcription (5–8). Another is caused by a rela-tively high concentration of noncanonical deoxyuridine triphosphatecompared with canonical TTP. dUTP is a substrate for HIV reversetranscriptase, which leads to uracil incorporation into viral cDNA.HIV reverse transcripts are heavily uracilated in macrophages(9, 10). Moreover, viral cDNA is a substrate for apolipoprotein BmRNA-editing, enzyme-catalytic, polypeptide-like 3 (APOBEC3)-family editing enzymes, which convert cytidine to uridine in theminus strand of HIV reverse-transcription intermediates (6, 7, 11,12). The latter two mechanisms flag viral cDNA for processing bycellular DNA repair enzymes.The restriction mechanisms that target lentivirus genome rep-

lication are counteracted by accessory virulence proteins Vif, Vpx,and Vpr, which usurp specific cellular E3 ubiquitin ligases to direct

antiviral proteins for degradation by the proteasome (13). Inparticular, Vif loads antiviral APOBEC3 family proteins onto aCRL5 E3 ubiquitin ligase for polyubiquitination and subsequentproteasome-dependent degradation (14). Vpx, encoded by theHIV-2/simian immunodeficiency virus (SIV) SIVsm lineage ofprimate lentiviruses, and closely related Vpr proteins from asubset of SIV viruses isolated from various primate species,counteract SAMHD1-mediated restriction (6, 7, 11). Specifically,Vpx binds to the DCAF1 substrate receptor subunit of theCRL4DCAF1 E3 ubiquitin ligase and loads SAMHD1 onto thisenzyme, thereby targeting it for degradation (15). Although HIV-1does not counteract SAMHD1 directly, it was proposed to bypassthe SAMHD1-imposed restriction as a result of a more efficientreverse transcriptase that can synthesize viral cDNA in a low-dNTPenvironment (16). The failure of the aforementioned viral coun-termeasures flags HIV cDNA for processing by DNA repairenzymes, which, if not successfully completed in a low-dNTPenvironment set up by SAMHD1, could lead to the initiation ofan innate response to viral nucleic acids, increased HIV-1 mu-tation rate, and/or inhibition of HIV-1 infection (17–19).Vpr, a paralogue of Vpx found in all primate lentiviruses, coor-

dinates interactions with postreplication DNA repair machinery,whose role for the replication cycle of primate lentiviruses is notwell understood. Early studies revealed that HIV-1 Vpr modulatesmutation rates in plasmid shuttle vectors in model systems (20,

Significance

In nondividing host cells, HIV is targeted by intrinsic antiviraldefense mechanisms that introduce marks of damage into viralcDNA, thereby tagging it for processing by cellular DNA repairmachinery. Surprisingly, our findings reveal that the two maintypes of HIV exhibit very different interactions with enzymesinvolved in DNA repair. HIV-1, but not HIV-2, efficiently removesselect DNA repair enzymes, whereas HIV-2 increases dNTP supplyin infected cells by removing SAMHD1 (a cell cycle-regulated dNTPtriphosphohydrolase) dNTPase. Our findings imply that increasingdNTP supply during viral cDNA synthesis or repair, or blockingcDNA processing by DNA repair enzymes, are alternative strate-gies used by HIV-2 and HIV-1 to guard their DNA genomes andfacilitate their replication/persistence in the host.

Author contributions: K.H., C.H., M.-C.S., S.K., S.K.S., L.F., M.P.W., and J.S. designed re-search; K.H., C.H., M.-C.S., S.K., and S.K.S. performed research; L.F. and M.P.W. contrib-uted new reagents/analytic tools; K.H., C.H., M.-C.S., S.K., S.K.S., L.F., M.P.W., and J.S.analyzed data; and J.S. wrote the paper.

The author declares no conflict of interest.

This article is a PNAS Direct Submission.

Data deposition: The complete MudPIT mass spectrometry dataset (raw files, peak files,search files, as well as DTASelect result files) can be obtained from the MassIVE databasevia ftp://MSV000079605@massive.ucsd.edu with password KHJS60144.1K.H. and C.H. contributed equally to this work.2To whom correspondence should be addressed. Email: jacek.skowronski@case.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1605023113/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1605023113 PNAS | Published online June 22, 2016 | E3921–E3930

MICRO

BIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

May

31,

202

0

21). Vpr, like Vpx, binds the DCAF1 subunit of the CRL4DCAF1

E3 complex and hijacks this enzyme (22). This interaction isassociated with the induction of DNA repair foci and activationof the serine/threonine kinase ATR-controlled DNA damagecheckpoint; the latter usually reflects the presence of, or failureto repair, damaged DNA at replication forks (23–25). Al-though the exact mechanism by which Vpr mediates the in-duction of replication stress is unclear, it was recently linkedto the Vpr-mediated recruitment of SLX4-SLX1/MUS81-EME1structure-specific endonucleases to DCAF1, resulting in activationof the MUS81-EME1 endonuclease and, surprisingly, DCAF1-and proteasome-dependent MUS81 degradation (26). Sepa-rately, HIV-1 and HIV-2 Vpr were reported to bind a baseexcision repair enzyme, uracil DNA glycosylase (UNG2),which can restrict HIV-1 infection in cells with high concen-trations of dUTP (19, 27–29). HIV-1 Vpr was shown to tar-get UNG2 to the ubiquitin proteasome pathway via theCRL4DCAF1 E3 complex and thereby disrupt UNG2-initiated baseexcision repair in HIV-1–infected cells (30–32).The fact that cellular restriction factors flag HIV cDNA for

processing by DNA repair enzymes, taken together with theconsiderable complexity of cellular repair machineries, raises thepossibility that Vpr coordinately engages DNA repair pathways.Unveiling additional targets in these pathways would lead to amore integrated model for the role of Vpr and DNA repair inthe HIV life cycle. Here, we performed a proteomic screen forVpr-interacting DNA repair proteins, aiming in particular toidentify novel substrates of the reprogrammed by HIV-1 VprCRL4DCAF1 E3 ubiquitin ligase (CRL4DCAF1-H1.Vpr). We showthat human HLTF DNA repair helicase is such a target of HIV-1Vpr. Our findings reveal that HIV-1 Vpr impacts three distinct typesof DNA repair transactions and illustrate the complexity of HIV-1Vpr interactions with cellular postreplication DNA repair machin-ery. We also show that HLTF and UNG2 are common targets ofVpr proteins from all main groups of HIV-1, but, somewhat un-expectedly, not of those found in HIV-2 A or B. Thus, our findingssupport the possibility that the two main types of HIV use verydifferent strategies to stabilize their DNA genomes. HIV-2 increasesdNTP supply in infected cells by removing SAMHD1, whereasHIV-1 inactivates DNA repair enzymes, possibly to compensate forits inability to drive up dNTP concentrations in primary target cells.

ResultsProteomic Screen for DNA Repair Proteins Associated with HIV-1 Vpr.We searched for candidate DNA repair proteins that are targetedby HIV-1 Vpr by purifying Vpr–protein complexes from T cells andcharacterizing their composition by multidimensional proteinidentification technology (MudPIT) (33). In brief, we constructed aCEM.SS T-cell line harboring a doxycycline-inducible HIV-1 NL4-3Vpr allele tagged with a double HA-FLAG-epitope (CEM.SS-iH1.Vpr). Vpr expression was induced with doxycycline for 6 h, and Vpr,together with its associated proteins, was purified by successiveimmunoprecipitations via the HA and FLAG tags (6, 34). As ex-pected, the most abundant cellular polypeptides found to be asso-ciated with Vpr were DCAF1, DDB1, and DDA1, subunits of theCRL4DCAF1 E3 complex, which Vpr binds via its DCAF1 subunit(Table S1) (34, 35. Analysis of the MudPIT datasets using theDatabase for Annotation, Visualization and Integrated Discoverybioinformatics resource (36, 37) identified 21 Vpr-associated pro-teins that have been linked to DNA replication and/or repair. Noneof these proteins was detected in MudPIT datasets obtained forpurifications from control CEM.SS T cells that did not express Vpr.Among the proteins identified (Table S1) were several that mediatepostreplication DNA repair, including UNG2, MSH6, a mismatchrepair protein, RFC clamp loader, and PCNA, replication factorsthat are central to DNA replication and repair (38–40), and HTLF,one of two mammalian homologs of yeast RAD5 DNA helicasethat controls the postreplication error-free DNA repair pathway

(41, 42). The presence of multiple proteins involved with DNArepair was not surprising, as HIV-1 Vpr and DCAF1 were reportedto colocalize with DNA repair foci in chromatin (23). The presenceof UNG2, a known, specific substrate of the CRL4DCAF1-H1.Vpr

E3, in MudPIT datasets indicates that our experimental approachdetects cellular proteins that Vpr recruits to the CRL4DCAF1 E3ubiquitin ligase and thereby directs for proteasome-dependentdegradation.

HIV-1 Vpr Down-Regulates HLTF, a Postreplication DNA Repair Helicase.To assess whether any of the 21 identified DNA repair proteins is apotential substrate of CRL4DCAF1-H1.Vpr E3, we first tested theirlevels in CEM.SS-iH1.Vpr and/or U2OS-iH1.Vpr, the latter alsoharboring a doxycycline-inducible HIV-1 NL4-3 Vpr transgene (Fig.S1). Of note, U2OS cells retain many of the cell cycle regulationcharacteristics of normal cells and are commonly used for cell cycle/DNA repair/replication studies. Interestingly, the levels of endoge-nous HLTF were much lower in CEM.SS-iH1.Vpr and U2OS-iH1.Vpr cells that had been arrested by Vpr at the DNA damagecheckpoint in the G2 phase of the cell cycle compared with controlasynchronously dividing cells that did not express Vpr (Fig. S1).Significantly, HLTF was not depleted in control cells arrestedin late S/G2 phase by etoposide or in early M phase by nocodazoletreatments. These observations are consistent with the possibilitythat HLTF, a DNA repair protein expressed in natural target cellsof HIV-1 infection (Fig. S2), is a specific target of HIV-1 Vpr.

HIV-1 Vpr Down-Regulates HLTF Independently of Cell Cycle Position.Vpr activates the ATR-controlled DNA damage checkpoint,thereby arresting cells in G2 phase (24). The possibility existed thatHLTF down-regulation is an indirect consequence of Vpr-inducedcell cycle perturbations. Hence, to demonstrate that HLTF de-pletion by Vpr is independent of cell cycle phase and ATR acti-vation, additional experiments were performed.First, we asked whether Vpr can deplete HLTF in U2OS-iH1.

Vpr cells outside of the G2 phase. U2OS-iH1.Vpr were syn-chronized in late G1/early S phase by double-thymidine block,and Vpr expression was induced at 8 h into the second thymidinetreatment (Fig. 1A). Cells were harvested at 0, 6, 12, and 24 hafter induction for flow cytometry analysis of cell cycle positionand for immunoblot analysis of HLTF levels. As shown in Fig.1B, U2OS-iH1.Vpr cells remained at the G1/S border throughthe duration of the experiment. Significantly, endogenous HLTFlevels became severely depleted within 6 h of the induction ofVpr expression (Fig. 1C).To assess whether the Vpr effect on HLTF was linked to its

interaction with the CRL4DCAF1 E3 ubiquitin ligase, we next testedthe Vpr(H71R) variant that does not bind DCAF1 (32). Significantly,this mutant did not detectably modulate HLTF levels even at thelate 24-h time point. These findings link the ability of Vpr to depleteHLTF to its interaction with CRL4DCAF1 E3 Ub ligase.Excess thymidine stresses replication forks (43), potentially

contributing to the observed Vpr-mediated HLTF depletion. Toexclude this possibility, we characterized HLTF levels across thecell cycle in asynchronously dividing U2OS-iH1.Vpr cells. Thecells were cultured in the presence or absence of doxycycline for6 h, stained with a vital stain, Vybrant DyeCycle Green, to revealtheir DNA content, and then sorted into highly enriched G1, S,and G2/M populations (Fig. 2A). Whole-cell extracts preparedfrom the sorted cells were analyzed by immunoblotting forCyclinA2, HLTF, and UNG2, the previously identified specificsubstrate of the CRL4DCAF1-H1.Vpr E3 ubiquitin ligase (Fig. 2B).CyclinA2 was detected in only the S and G2/M extracts, asexpected, thus confirming the purity of the sorts. HLTF levels weresimilar in G1, S, and G2/M extracts from uninduced U2OS-iH1.Vpr cells, indicating that HLTF is expressed in all cell cycle phases.Significantly, HLTF levels were much lower in the induced U2OS-iH1.Vpr cells, regardless of their cell cycle position. The levels of

E3922 | www.pnas.org/cgi/doi/10.1073/pnas.1605023113 Hrecka et al.

Dow

nloa

ded

by g

uest

on

May

31,

202

0

UNG2 were similarly depleted following induction of Vpr ex-pression. Finally, HLTF depletion by Vpr was not inhibited byculturing the cells in the presence of caffeine, indicating that thedepletion was not a G2/M DNA damage checkpoint-mediatedresponse (Fig. S3).Together, the aforementioned findings demonstrate that Vpr

down-regulates HLTF levels independently of the cell cycle po-sition and of G2 DNA damage checkpoint, thus further sup-porting the possibility that HLTF is a substrate for the HIV-1Vpr-modified CRL4DCAF1 E3 Ubiquitin ligase.

HIV-1 Vpr Targets HLTF Independently of MUS81. HIV-1 Vpr was re-cently reported to down-modulate the MUS81 subunit of a struc-ture-specific endonuclease in a DCAF1-dependent manner (26).Therefore, we characterized MUS81 levels in control and Vpr-expressing U2OS-iH1.Vpr sorted cell populations. Strikingly, theendogenous MUS81 levels were not detectably altered following abrief, 6-h-long pulse of Vpr expression. This contrasted with therobust depletion of HLTF in the same time frame (Fig. 2B). Thesefindings implied that the effects of HIV-1 Vpr on HLTF andMUS81 are likely to be independent of each other. We furtherexplored this by carefully comparing the time course of HLTF andMUS81 depletion following induction of Vpr expression in U2OS-iH1.Vpr cells. As shown in Fig. 3A, depletion of HLTF levels wasalmost complete at the 6-h time point. In contrast, MUS81 levelswere down-regulated at a much slower rate, and 24–48 h were re-quired for complete depletion, which coincides with the accumu-lation of Vpr-expressing cells at the DNA damage checkpoint inlate S/G2 phase.

A similar analysis was performed with the HIV-1 Vpr(W54R)variant, which is defective for UNG2 loading onto the CRL4DCAF1

complex but retains binding to DCAF1 (28, 30). This variantretained full ability to deplete HLTF and MUS81 levels (Fig. 3B).We conclude that Vpr exerts its effects on HLTF and MUS81 in amanner independent of UNG2, even though all three require theinteraction with DCAF1 (26, 30).The possibility still remained that lower MUS81 levels in Vpr-

expressing cells reflected a cellular response to HLTF depletion,rather than a direct effect of Vpr. Therefore, we asked whetherHLTF and MUS81 levels are correlated in the absence of Vpr.Parental U2OS cells were subjected to RNAi targeting of HLTFor MUS81, and cell extracts were prepared 48 h later and analyzedby immunoblotting. Notably, HLTF knockdown led to elevatedMUS81 levels, suggesting that the latter was to compensate for theloss of HLTF (Fig. 3C). In contrast, MUS81 depletion did not havea detectable effect on HLTF levels. This evidence supports thepossibility that HLTF and MUS81 are specific and independenttargets of the CRL4DCAF1-H1.Vpr E3 ubiquitin ligase.

HIV-1 Selectively Depletes Only the HLTF Homolog of Yeast RAD5 DNARepair Protein. Mammalian cells possess two Rad5 homologs,HLTF and SHPRH, which serve distinct roles in postreplicationDNA repair (41). Experiments were performed to assess whetherVpr selectively depletes HLTF or, instead, targets both Rad5homologs. As shown in Fig. 3, SHPRH levels remained largely

Fig. 1. HIV-1 Vpr depletes HLTF outside of G2 phase. (A) Experimental strategy.U2OS cells harboring the indicated HIV-1 NL4-3 Vpr transgenes were synchro-nized by double thymidine block and then kept arrested at the G1/S border forthe duration of the experiment. HIV-1 Vpr expression was induced by doxycy-cline during the second thymidine treatment, and cells were collected at theindicated time points for FACS analysis of cell cycle position (B) and immunoblotanalysis of the indicated proteins (C). Transcription factor II D (TFIID) was used asa loading control. The cell cycle profile of an asynchronously dividing cell pop-ulation that did not undergo double thymidine block is shown in B and labeledwith “A”.

Fig. 2. HIV-1 Vpr depletes HLTF independently of cell cycle position.(A) Purity of G1, S, and G2/M cell populations isolated by cell sorting.The DNA profiles of U2OS (mock) and U2OS-iH1.Vpr (HIV-1 Vpr) G1-, S-, andG2/M-phase cells isolated by sorting for DNA content are shown overlaidwith that of the parental asynchronously growing unsorted population (la-beled “A”). (B) Cell extracts prepared from asynchronously dividing (“A”),G1, S, and G2/M populations were analyzed by immunoblotting with anti-bodies reacting with HLTF, UNG2, MUS81, CyclinA2, FLAG-tagged Vpr, andTFIID loading control. Asterisk indicates a nonspecific background bandrevealed by the α-UNG2 antibody.

Hrecka et al. PNAS | Published online June 22, 2016 | E3923

MICRO

BIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

May

31,

202

0

unaltered with a small, at most approximately twofold, decreaseat the 24- and 48-h time points. We conclude that Vpr selectivelytargets the HLTF homolog of yeast Rad5 helicase.

HIV-1 Vpr Targets HLTF for Proteasome-Dependent Degradation viaCRL4DCAF1 E3. We next tested whether Vpr directs HLTF forproteasome-dependent degradation. U2OS-iH1.Vpr cells weretreated with doxycycline in the absence or presence of MG132proteasome inhibitor for 9 h. As shown in Fig. 4A, proteasomeinhibition stabilized HLTF levels in Vpr expressing cells whilehaving no detectable effect on HLTF in control cells. We con-clude that Vpr activates a proteasome-dependent mechanism todown-modulate HLTF levels.Next, we asked whether HIV-1 Vpr can recruit HLTF to the

endogenously expressed CRL4DCAF1 E3 complex (Fig. 4 B and C).FLAG-tagged HLTF was expressed alone or coexpressed withHIV-1 NL4-3 Vpr or the Vpr(H71R) variant in HEK 293T cells.Detergent extracts were prepared from the transfected cells, andHLTF immune complexes were analyzed by immunoblottingfor HLTF, Vpr, and the DCAF1 and DDB1 subunits of theCRL4DCAF1 E3. We found that HIV-1 Vpr directed the assembly ofa protein complex containing HLTF as well as DCAF1 and DDB1.In contrast, neither the Vpr(H71R) variant, which retained a reducedbinding to HLTF, nor HLTF alone, nucleated such a DCAF1-containing complex. These observations indicate that Vpr mediatesspecific recruitment of HLTF to the CRL4 E3 complex that usesthe DCAF1 substrate receptor, to which Vpr binds. Of note, HLTFcomplexes formed in the absence of Vpr contained low levels ofDDB1, which suggests that HLTF may physiologically participate inDDB1 complexes that do not contain DCAF1.Overall, these findings indicate that HIV-1 Vpr specifically

binds to HLTF and recruits it to the DCAF1-DDB1 module ofthe CRL4DCAF1 E3 complex, thereby directing HLTF for deg-radation by proteasome.

HIV-1 Vpr Interaction with HLTF Defines Previously UnidentifiedFunctional Elements in both Proteins. HLTF comprises an N-ter-minal DNA-binding HIRAN domain and a C-terminal RINGdomain possessing E3 Ubiquitin ligase activity (Fig. 5A). To mapthe HLTF region that mediates the effect of Vpr, we analyzed theactivity of HIV-1 NL4-3 Vpr toward a set of HLTF deletion mu-tants by using a transient coexpression assay in HEK 293T cells.This assay faithfully reproduced the Vpr-induced proteasome-dependent HLTF down-modulation (Fig. S4). As shown in Fig. 5B,an HLTF fragment lacking the N-proximal 151 residues, includingthe HIRAN domain [HLTF(152–1009)], as well as those lacking theC-distal ∼700 residues comprising the RING domain andHELICc/DEXDc helicase components, were fully responsive toHIV-1 Vpr, thus tentatively mapping the Vpr target site to HLTFresidues 152–299. This region appears to serve as a linker be-tween the HIRAN domain and the N-proximal component of thehelicase ATP-binding domain, and, as such, is not known to possessanother function.We next mapped the residues in HIV-1 Vpr that are required

for the effect on HLTF levels. Because Vpr and its paralogue Vpxuse their N-terminal regions to recruit novel protein substrates toCRL4DCAF1 E3 (15, 30, 44), we constructed a set of HIV-1Vpr point mutants in the N-proximal region and screened them byusing the previously described transient, dose-dependent HLTFdown-modulation assay. As shown in Fig. 5C, a double amino acidsubstitution E24R,R36P [Vpr(E24R,R36P)] diminished the ability ofVpr to down-modulate HLTF, but not UNG2. These observationsrevealed that the E24R,R36P mutation selectively disrupts Vprbinding to HLTF, but does not grossly interfere with binding toDCAF1 or recruitment of UNG2 to the CRL4DCAF1 E3 complex.To corroborate this possibility, we characterized the binding of

the Vpr(E24R,R36P) variant to HLTF and DCAF1. FLAG-HLTF wascoexpressed with HA-Vpr or HA-Vpr(E24R,R36P) in HEK 293T cells,and Vpr immune complexes were analyzed by immunoblotting. The

Fig. 3. Vpr selectively targets the HLTF homolog of RAD5, and this effect is independent from interactions with UNG2 and MUS81. (A and B) Time course of HLTF,SHPRH, MUS81, and UNG2 depletion by HIV-1 Vpr. U2OS-iH1.Vpr and control cells were treated with doxycycline, and cell extracts prepared at the indicated timeswere analyzed by immunoblotting. Lamin B and TFIID provided loading controls. (C) HLTF, MUS81, and UNG2 levels are not coregulated. U2OS cells were subjected tocontrol nontargeting RNAi (scr) or RNAi targeting HLTF or MUS81, and cell extracts were immunoblotted for HLTF, MUS81, UNG2, and TFIID.

Fig. 4. HIV-1 Vpr recruits HLTF to the DDB1-DCAF1 module of the CRL4DCAF1 E3 complex for proteasome-dependent degradation. (A) Vpr targets HLTF forproteasome-dependent degradation. U2OS-iH1.Vpr cells (HIV-1 Vpr) and control U2OS cells (mock) were treated with doxycycline or not treated in theabsence or presence of MG132 proteasome inhibitor (1 μg/mL) for 9 h. HLTF, MUS81, and Vpr levels in cell lysates were revealed by immunoblotting. TFIIDprovided a loading control. (B) Schematic representation of HIV-1–mediated recruitment of HLTF onto the CRL4DCAF1 E3 complex. The placement of HA andFLAG epitope tags on Vpr and HLTF, respectively, is indicated. (C) Vpr recruits HLTF to the DCAF1-DDB1 module of the CRL4DCAF1 E3 complex. FLAG-HLTF wastransiently coexpressed with myc-tagged HIV-1 Vpr or Vpr(H71R) in HEK293T cells as indicated. Endogenous DCAF1, DDB1, and ectopic Vpr and HLTF wererevealed in HLTF immune complexes and in detergent extracts by immunoblotting.

E3924 | www.pnas.org/cgi/doi/10.1073/pnas.1605023113 Hrecka et al.

Dow

nloa

ded

by g

uest

on

May

31,

202

0

Vpr and Vpr(E24R,R36P) variants efficiently coprecipitated the en-dogenously expressed DCAF1-DDB1 module of CRL4DCAF1 E3complex (Fig. 5D). Vpr(H71R), which was analyzed in parallel as anegative control, did not associate with DDB1-DCAF1, and onlyweakly associated with HLTF, as expected. Significantly, only WTVpr, but not the Vpr(E24R,R36P) variant, efficiently coprecipitatedHLTF. Notably, to our knowledge, the E24 and R36 residues havenot been reported to be important for any other known Vpr in-teraction with cellular DNA repair machinery. These findings to-gether confirm that Vpr recruits HLTF to DCAF1 independent ofits previously reported interactions with other DNA repair proteins.

HLTF, MUS81, and the Induction of G2 Arrest by Vpr.We asked whetherVpr-mediated depletion of HLTF disrupts postreplication DNArepair sufficiently to modulate the DNA damage checkpoint suchthat cells are arrested in the G2 phase of the cell cycle. To this end,we tested the HIV-1 Vpr(E24R,R36P) variant for its ability to arrestcells in G2 phase and determined whether HLTF and/or MUS81are required for the induction of G2 arrest by Vpr. U2OS-iH1.Vpr(E24R,R36P) cells were induced to express HLTF degradation-defective Vpr(E24R,R36P) variant, and their cell cycle profile weredetermined 2 d later. As shown in Fig. 6A, HIV-1 Vpr(E24R,R36P)retained the ability to arrest U2OS cells in G2 phase. Similar ex-periments performed with U2OS HLTF-KO cells [U2OS.HLTF.KO-iH1.Vpr (45)] revealed that WT NL4-3 Vpr arrested these cellsin G2 phase (Fig. 6B, panel 12). Thus, HLTF deficiency per se is notsufficient to trigger DNA-damage checkpoint, as reported in pre-vious studies (45, 46), nor is HLTF required to mediate cell cyclearrest in G2 phase by Vpr.As the data shown in Fig. 3C suggest a possible compensatory

interaction between MUS81 and HLTF, we next tested the effectof RNAi-mediated MUS81 knockdown on cell cycle distribution

of U2OS.HLTF.KO cells in the absence of Vpr expression (Fig. 6Cand Fig. S5). Interestingly, MUS81 knockdown in U2OS.HLTF.KOcells was associated with an altered cell cycle profile with an in-crease in the G2-phase population compared with that for cellssubjected to nontargeting siRNA, even in the absence of Vpr (P <0.01) (Fig. 6C). This observation indicated that a combined MUS81and HLTF deficiency alone could lead to the accumulation of cellsin G2 phase.Next, we investigated whether MUS81 is required for the in-

duction of G2 arrest by HIV-1 Vpr in U2OS-iH1.Vpr cells. MUS81levels were depleted by RNAi 48 h before induction of Vpr ex-pression, and cell-cycle profiles were determined 2 d after induction.Strikingly, MUS81 depletion did not diminish the ability of Vpr toarrest cells in G2. To the contrary, it slightly exacerbated the Vprphenotype in these cells, but the difference was not statisticallysignificant. Notably, a similar effect was seen in HLTF-KO cells, inwhich the difference conferred by the MUS81 knockdown wasstatistically significant (P < 0.01). These observations reveal that thepresence of MUS81 is probably not required for the induction ofG2 arrest by Vpr. This evidence supports the model in which rep-lication stress resulting from combined depletion of MUS81 andHLTF contributes to the ability of Vpr to arrest cells at the DNAdamage checkpoint in G2 phase.

HLTF and UNG2 Are Common Targets of HIV-1 and SIVcpz Vpr Proteins.Experiments were performed to establish whether HLTF is a com-mon target of primate lentiviral Vpr proteins. We focused on Vprproteins from HIV-1 and HIV-2, which represent two evolutionarydivergent branches of primate lentiviruses that adapted to replicatein human cells. Fig. 7A shows an amino acid sequence alignment forconsensus Vpr proteins derived from the main groups of HIV-1(M, N, O) and SIVcpz viruses, isolated from two chimpanzee

Fig. 5. HIV-1 Vpr interaction with HLTF does not involve known functional elements in both proteins. (A) Summary of mutations in HLTF and their effects onVpr-mediated down-modulation of HLTF levels. Schematic representation of the HLTF protein and HLTF deletion mutants shows the location of the HIRAN,RING, SNF2, and discontinuous DEXDc and HELICc helicase domains. The HLTF region mediating Vpr sensitivity is boxed. (B) HIV-1 Vpr does not act on knownfunctional domains of HLTF. FLAG-HLTF and its variants, indicated in A, were transiently coexpressed with increasing doses of HA-tagged HIV-1 Vpr in HEK293T cells, and HLTF and Vpr levels in cell extracts were revealed by immunoblotting. α-Tubulin provided a loading control. (C) The Vpr N terminus is requiredfor depletion of HLTF levels. HIV-1 Vpr or Vpr(24R,36P) variant was coexpressed at increasing doses with HLTF (Upper) or UNG2 (Lower) in HEK293T cells, and cellextracts were analyzed as described earlier. (D) The E24R,R36P mutation selectively disrupts Vpr binding to HLTF. HA-Vpr and its variants were coexpressedwith FLAG-HLTF in HEK 293T cells. The Vpr(H71R) variant that does not bind DCAF1 was used as a negative control. Endogenous DCAF1, DDB1, and ectopic Vprand HLTF were revealed in Vpr immune complexes and detergent extracts by immunoblotting.

Hrecka et al. PNAS | Published online June 22, 2016 | E3925

MICRO

BIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

May

31,

202

0

subspecies (Ptt, Pts) from which HIV-1 originated after cross-speciestransmissions (47), as well as Vpr proteins fromHIV-2 groups A andB. Of note, to account for the considerable amino acid sequencedivergence at the C terminus of HIV-2 group A Vpr proteins, twodistinct consensus Vpr proteins, termed A1 and A2, were de-rived. Inspection of the alignment revealed that residues E24 andR36, which we found to be important for HLTF depletion byHIV-1 NL4-3 Vpr, are conserved in all consensus HIV-1 andSIVcpz Vpr sequences (Fig. 7A). Similarly, residue W54, which isrequired for UNG2 loading onto CRL4DCAF1 E3, was also pre-sent in all HIV-1/SIVcpz consensus Vpr sequences. These ob-servations predicted that HLTF as well as UNG2 are commontargets of HIV-1 and SIVcpz Vpr proteins.To test this prediction, we synthesized vpr genes for the con-

sensus Vpr proteins shown in Fig. 7A and characterized their abilityto deplete HLTF and UNG2. As shown in Fig. 7B, all tested HIV-1and SIVcpz Vpr proteins down-modulated HLTF and UNG2.Their activities were comparable to those of the NL4-3 Vpr that weused in the studies described earlier. Remarkably, even theSIVcpzPts Vpr, which is ∼30% divergent from HIV-1 M Vprin the core region, was quite active against HLTF.

HIV-2 Vpr and Vpx Do Not Modulate HLTF Nor UNG2 Levels. Sequencealignment, shown in Fig. 7A, revealed that the key residues re-quired for robust interaction with HLTF, as well as UNG2, arenot conserved in consensus HIV-2 Vpr proteins. Indeed, theconsensus HIV-2 groups A and B Vpr proteins were inactivetoward HLTF and UNG2 despite being expressed to levels muchhigher than that of the NL4-3 Vpr (Fig. 7C). Importantly, theHIV-2 consensus Vpr proteins bound DCAF1 in coimmuno-precipitation assays, similar to HIV-1 NL4-3 Vpr, and thereforeappeared to be functional (Fig. S6A). We conclude that only theHIV-1 and SIVcpz, but not HIV-2, Vpr proteins efficiently directHLTF and UNG2 for proteasome-dependent degradation viaCRL4DCAF1-H1.Vpr E3 ubiquitin ligase.

HIV-2/SIVsm lineage viruses encode a Vpr paralogue termedVpx, which was lost during adaptation of ancient SIV to chim-panzees and is absent in the present HIV-1/SIVcpz branch of pri-mate lentiviruses (48). Vpr and Vpx proteins usurp CRL4DCAF1

E3 and are highly adaptable, as revealed by their shared abilityto reprogram this ubiquitin ligase toward a common SAMHD1substrate in many SIV lineages (11). Therefore, we asked whetherHIV-2 Vpx has acquired the ability to program HLTF or UNG2degradation. As shown in Fig. 7D, neither the HIV-2 A nor B groupconsensus Vpx protein decreased HLTF or UNG2 levels. Of note,the Vpx proteins were functional as judged by their ability todeplete SAMHD1 (Fig. S6B). We conclude that HIV-2 lineageVpr/Vpx proteins do not endow CRL4DCAF1 E3 with specificitytoward HLTF or UNG2.

HLTF and UNG2 Are Depleted in HIV-1–Infected Cells. To corroboratethe aforementioned findings, we characterized HLTF and UNG2levels in the context of HIV infection. As shown in Fig. 8 A and B,challenge of Jurkat T cells with VSV-G pseudotyped, Vpr-loadedHIV-1 virus-like particles (VLPs) or infection with a VSV-Gpseudotyped single-cycle HIV-1 NL4-3.GFP.R+ carrying an intactvpr gene led to the depletion of HLTF and UNG2. Importantly,infection of primary CD4+ T cells with NL4-3.GFP.R+ virus wasalso associated with a profound depletion of HLTF levels, indicat-ing that endogenous Vpr expression is sufficient to remove HLTFfrom natural HIV-1 target cells (Fig. 8C). In contrast, these effectswere not seen in cells infected with control capsid-normalized HIV-1 VLP produced in the absence of Vpr, or with vpr-deficient NL4-3.GFP.R– HIV-1 (Fig. 8 A and B).Next, we asked whether HIV-1 and HIV-2 indeed exert di-

vergent effects on HLTF and UNG2. Jurkat T cells were infectedwith two doses of normalized, single-cycle HIV-1 NL4-3.GFP.R+,and HIV-2 Rod having intact vpr and vpx genes (Fig. 8 D and E).HIV-1 infection led to a dose-dependent depletion of HLTF andUNG2 levels, as expected. In contrast, HIV-2 did not exert acomparable robust effect, even though the levels of both proteins

Fig. 6. The effects of HLTF and MUS81 depletion on HIV-1 Vpr-induced G2 arrest. (A) Vpr(E24R,R36P) arrests U2OS cells in G2 phase. Cell cycle profiles of U2OScells induced with doxycycline [Dox (+)], or not [Dox (−)] to express HIV-1 NL4-3 Vpr, Vpr(E24R,R36P), or Vpr(H71R) for 48 h. The percentage fraction of cells in G1,S, and G2 phase is indicated, and panel numbers are shown in upper right corner. The abscissa is DNA content shown on a linear scale. The ordinate is DNAsynthesis shown on a logarithmic scale (B) HIV-1 Vpr arrests U2OS.HLTF.KO cells in G2 phase. Cell-cycle profiles of U2OS.HLTF.KO cells induced with doxycycline[Dox (+)] or not [Dox (−)] to express HIV-1 NL4-3 Vpr for 48 h. (C) Vpr arrests MUS81-depleted U2OS and U2OS.HLTF.KO cells in G2 phase. Schematic rep-resentation of cell cycle profiles of parental (mock) U2OS (HLTF), U2OS.HLTF.KO cells (HLTF.KO), or their derivative cell lines carrying doxycycline-inducibleHIV-1 Vpr transgenes (HIV-1 Vpr). The cells transfected with siRNA targeting MUS81 (MUS81) or nontargeting (scr) and induced (or not) with doxycycline2 d later as indicated. Cell cycle profiles were determined 2 d after induction, and cells in G1, S, and G2/M phases were quantified as shown in A. Each barrepresents averaged results from four replicates. The significance of the observed differences in G2-phase populations between the indicated conditionsshown above the bars was calculated using an unpaired two-tailed Student’s t test with Welch’s correction (n = 4; *P < 0.05, **P < 0.01, ***P < 0.001, and****P < 0.0001). Representative results of three independent experiments are shown.

E3926 | www.pnas.org/cgi/doi/10.1073/pnas.1605023113 Hrecka et al.

Dow

nloa

ded

by g

uest

on

May

31,

202

0

appeared to be somewhat lower in cells infected with the highestdose of HIV-2. In parallel, similar experiments were performed inTHP-1 cells (Fig. 8F). Of note, these cells express SAMHD1 at highlevels, but HLTF levels were not detectable by immunoblotting. Wefound that HIV-1 readily depleted UNG2 while having no effect onSAMHD1 levels. In contrast, HIV-2 depleted SAMHD1 in a Vpx-dependent manner, as expected, whereas UNG2 levels were notaffected. To corroborate these findings, we tested two ancestorSIVmac251 and SIVmac239 viruses, each encoding an uninter-rupted vpr gene. As shown in Fig. 8G, infection of THP-1 cellsresulted in almost complete depletion of SAMHD1 but had nodetectable effect on UNG2 levels in these cells. We conclude thatHIV-1/SIVcpz lineage viruses counteract HLTF and UNG2 DNArepair proteins via the CRL4DCAF1 E3, whereas the HIV-2/SIVmaclineage viruses tested do not possess a comparable ability.

DiscussionWhen replicating in primary, nondividing, target cells, primatelentiviruses are targeted by innate immune mechanisms that in-troduce marks of damage into viral cDNA, thereby flagging it forprocessing by host-cell DNA repair enzymes. Hence, it is notsurprising that lentiviral accessory proteins intersect with cellularmechanisms that modulate DNA repair, presumably to moder-ate the negative impact of DNA repair on virus genome stabilityand/or priming an innate response to viral nucleic acids. InHIV-1, the interactions with postreplication DNA repair processesare coordinated via the CRL4DCAF1 E3 ubiquitin ligase, which ishijacked by Vpr (22, 26, 34). The present study aimed to gain a

more comprehensive description of the interactions between HIV-1Vpr and cellular DNA repair pathways. To this end, we performeda proteomic screen, which led to the identification of HLTF DNAhelicase as a specific target of the CRL4DCAF1-H1.Vpr E3 ubiquitinligase. Of note, HLTF was also identified as HIV-1 Vpr target byLahouassa et al. (49).Our evidence supports the possibility that HLTF is a direct

target of HIV-1 Vpr for recruitment to the CRL4DCAF1 E3 andproteasome-dependent degradation. First, Vpr rapidly depletesHLTF independently of cell cycle position, and this is not adownstream effect of DNA damage checkpoint activation. Second,coimmunoprecipitation experiments show that Vpr bridges HLTF tothe CRL4DCAF1 E3 complex. Third, Vpr-mediated depletion doesnot require the HLTF RING domain, which was shown to mediateproteasome-dependent HLTF degradation in response to DNAdamage (41). These latter findings indicate that HLTF depletion is nota consequence of other Vpr interactions with DNA repair pathways.Overall, our studies reveal a striking complexity of Vpr interac-

tions with cellular DNA repair machinery. This and previous reportstogether demonstrate that HIV-1/SIVcpz Vpr uses the CRL4DCAF1

E3 to remove, from infected cells, three key enzymes involvedin distinct DNA repair pathways (15, 26, 32). Significantly, therecruitment of HLTF and UNG2, as well as Mus81 and UNG2, toCRL4DCAF1 E3 is mediated via different surfaces of the HIV-1 Vprmolecule. Although we have not yet separated HLTF and MUS81binding surfaces on Vpr, RNAi studies revealed that expression ofthese two DNA repair proteins is not coordinated, thereby implyingthat Vpr depletes their levels independently of each other. Thus, the

Fig. 7. HLTF and UNG2 are depleted by HIV-1 and SIVcpz but not by HIV-2 Vpr proteins. (A) Alignment of consensus sequences of Vpr proteins from main HIV-1 (M,N, O), SIVcpz (Ptt, Pts), and HIV-2 (A1, A2, B) groups. Sequences are shown in one letter code. Residues required for binding to HLTF (E24, R36), UNG2 (W54), andDCAF1 (H71) are boxed. (B) HIV-1 and SIVcpz Vpr programs HLTF and UNG2 for degradation. FLAG-HLTF or FLAG-UNG2 were transiently coexpressed with increasingdoses of the indicated consensus HA-epitope–tagged HIV-1 Vpr proteins. Levels of HLTF and Vpr in cell extracts were revealed by immunoblotting. α-Tubulin providedloading control. (C and D) Consensus HIV-2 Vpr and Vpx proteins do not modulate HLTF or UNG2 levels. FLAG-HLTF or FLAG-UNG2 were coexpressed with the in-dicated HA-epitope–tagged consensus HIV-2 Vpr or Vpx proteins. HLTF, UNG2, and Vpr/Vpx were analyzed as indicated earlier.

Hrecka et al. PNAS | Published online June 22, 2016 | E3927

MICRO

BIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

May

31,

202

0

effects on UNG2, HLTF, and MUS81 most likely reflect three in-dependently selected functions of HIV-1 Vpr.Viral accessory virulence factors usually redirect cellular E3

ubiquitin ligases to remove cellular proteins that exhibit direct orindirect antiviral activity (50). Hence, the concerted removal ofseveral postreplication DNA repair enzymes by Vpr is consistentwith the notion that host cell DNA repair machinery restrictssteps in HIV-1 replication in host cells. Indeed, the key role for adUTP-mediated antiretroviral host defense pathway is evidentfrom the fact that many retrovirus families captured a hostdUTPase gene during viral evolution (51, 52). Although HIV-1does not encode dUTPase, its Vpr protein effectively preventsUNG2-initiated uracil excision repair (31). This alternative mech-anism may allow HIV-1 to alleviative negative consequences ofUNG2-mediated uracil excision and downstream pathways for theintegrity of HIV-1 cDNA (19).The finding that HIV-1/SIVcpz Vpr specifically removes HLTF

from infected cells suggests that it negatively impacts HIV-1 repli-cation. Possible scenarios are suggested by known HLTF functions.In particular, HLTF remodels replication forks that are stalled byDNA lesions into four-way branched DNA structures, therebyproviding an undamaged DNA template that allows for error-freebypass of the lesion and resumption of DNA replication. In-terestingly, fork-like branched DNA structures resembling replica-tion forks are transient intermediates in plus-strand displacementsynthesis that is carried out by retroviral reverse transcriptase (53,54). Recognition and processing of such intermediates by HLTFcould impede an ordered synthesis of the cDNA copy of the HIV-1RNA genome. Although integration competent HIV-1 preintegra-tion complexes form in the cytoplasm, the completion of reversetranscription does not appear to be a prerequisite for HIV-1 entryto the nucleus (55), where partially reverse-transcribed viral cDNAcan be accessed by DNA repair proteins. Alternatively, HLTF couldmodulate processing of the integrated proviral DNA, as it is knownto control the choice between two pathways of postreplication re-pair of DNA lesions at the replication fork: error-free, by templateswitching, and error-prone, through recruitment of mutagenictranslesion DNA polymerases (45, 56). As these two distinct modesof DNA repair, error-free and error-prone, have different muta-genic outcomes upon repair of chromosomal DNA, they would

also have different consequences for the fidelity of HIV-1 provirusrepair, which provides another tentative rationale for the removal ofHLTF by HIV-1 Vpr.Notably, Vpr also disrupts one of the Holliday junctions

resolvases by depleting the MUS81 component of SLX1-SLX4/MUS81-EME1 (2) complex. This interaction was proposed to beassociated with untimely activation of MUS81-EME1 (2) nucle-ase activity to prevent an innate response to products of abortiveHIV-1 reverse transcription (26). Somewhat unexpectedly, our datashow that depletion of MUS81 levels does not alleviate the ability ofVpr to arrest cells in G2. Moreover, MUS81 and HLTF depletion,although each insufficient on its own, together appear to facilitateG2 arrest, consistent with the previously reported roles of theseproteins in replication fork maintenance and restart, and with re-cent genetic studies of the Vpr–MUS81 interaction (45, 57, 58).Thus, it appears that HIV-1 Vpr must perturb additional pathwaysto trigger DNA damage checkpoint leading to G2 arrest.The findings described here provide compelling evidence that

HIV-1 Vpr disrupts select DNA repair pathways via CRL4DCAF1

E3 ubiquitin ligase. This in turn prompts speculation that Vpracts to delay the repair until after integration of the HIV-1 provirusinto the host cell chromosome, an environment in which repair isconventionally handled by alternative postreplication DNA repairpathways. Alternatively, Vpr may block the repair through thesepathways altogether and thereby evade DNA repair-mediated re-striction. Such scenarios, as well as the timing of the action ofHLTF, UNG2, and other aspects of DNA repair impacted by Vpr,will be assessed in future studies.Our comparative studies of the two main types of HIV un-

expectedly revealed that HIV-1/SIVcpz Vpr proteins potentlydeplete HLTF and UNG2, whereas HIV-2 and their ancestralHIV-2sm lineage Vpr do not possess such robust abilities. Sig-nificantly, HIV-2/HIVsm viruses antagonize SAMHD1 dNTPasevia their Vpx proteins (6, 7) whereas HIV-1/SIVcpz do notpossess this ability. We propose that HIV-2 does not need toantagonize cellular DNA repair mechanisms in a manner com-parable to that seen with HIV-1, because it effectively counter-acts SAMHD1, which leads to an increase in cellular dNTPconcentrations. Through this mechanism, HIV-2 avoids mark-ing its cDNA for processing by DNA repair enzymes and the

Fig. 8. Divergent effects of HIV-1 and HIV-2 on HLTF and UNG2 levels. (A and B) HIV-1 infection depletes HLTF and UNG2 levels. (A) Jurkat T cells were infected withHIV-1 VLP transcomplemented (Vpr) or not (−) with HIV-1 Vpr (HIV-1 VLP; Left) or a single-cycle HIV-1 NL4-3.GFP.R+ (vpr+), or R− (vpr−) virus (HIV-1; Right). The levels ofHLTF, UNG2, and α-tubulin loading control in extracts prepared from infected and control (marked “c”) cells 48 h post infection were revealed by immunoblotting.(B) HIV-1 VLP and HIV-1 NL4-3.GFP.R+ and R– viruses used in A were normalized by immunoblotting for p24 capsid. Vpr was revealed with an α-HA antibody (HIV-1VLP) or α-Vpr antibody (HIV-1). (C) HIV-1 depletes HLTF andMUS81 in primary CD4+ T cells. CD4+ T cells were activated with α-CD3/α-CD28 beads and transduced withHIV-1 NL4-3.GFP.R+ (or R–), and GFP-positive cells were isolated by cell sorting 2 d after infection and analyzed for expression of the indicated proteins by immu-noblotting. Twofold serial dilutions of control CD4+ T whole-cell lysates provided quantification standards. (D and E) Divergent effects of HIV-1 and HIV-2 on HLTF andUNG2 levels. Jurkat T cells were infected with HIV-1 NL4-3.GFP.R+ (HIV-1) or HIV-2 ROD with intact vpr and vpx genes expressing an RFP marker. Two days afterinfection, reporter gene expression was revealed by FACS. The abscissa is GFP/RFP fluorescence shown on a logarithmic scale. The ordinate is forward scatter shown ona linear scale (D), and UNG2, SAMHD1, HLTF, and Lamin B loading control in cell extracts were visualized by immunoblotting (E). (F and G) THP-1 cells were infectedwith HIV-1 NL4-3.GFP.R+ [or R− (−)] or HIV-2 ROD expressing Vpr and Vpx (vpr/vpx) or only Vpr [vpr/(−) or not infected (marked as “c”)] (F), or with SIVmac251 VLP orSIVmac239 (G) as indicated, and extracts were analyzed by immunoblotting. Asterisk indicates a nonspecific band revealed by the α-SAMHD1 antibody.

E3928 | www.pnas.org/cgi/doi/10.1073/pnas.1605023113 Hrecka et al.

Dow

nloa

ded

by g

uest

on

May

31,

202

0

resulting repair-mediated restriction and/or provides sufficientdNTP supply to support faithful repair in nondividing G1 cells,unlike HIV-1. Regardless, the evidence presented here impliesthat distinct lineages of primate lentiviruses use different strat-egies to manage their genomes, protecting them from the ac-quisition of DNA damage and/or from processing by DNA repairenzymes when replicating in dNTP-poor cellular environments.The emergence of SIVcpz is thought to have involved re-

combination between two SIV lineages from Old World Monkey(OWM) species (59), resulting in a deletion of the vpx gene andthe ensuing loss of SAMHD1 antagonism by Vpx. SAMHD1antagonism by Vpx is well-conserved in distinct SIV strainsisolated from OWM (11, 60), indicating the importance ofSAMHD1-exerted restriction on replication of primate lentivi-ruses, and thereby raising the question of how the loss ofSAMHD1 counteraction was compensated for in the HIV-1/SIVcpz lineage. One such mechanism was suggested by theprevious finding that cross-species transmission of SIV tochimpanzees was associated with an expansion of the range ofABOBEC3 proteins targeted by Vif (48). Our studies reveal thatthe loss of Vpx in the HIV-1/SIVcpz lineage was also associatedwith the acquisition/enhancement of an ability to disrupt DNArepair by HIV-1 Vpr via HLTF and UNG2, and suggest thatthe latter was to compensate for the inability to antagonizeSAMHD1. This possibility makes sense mechanistically, as re-verse transcription, taking place in the dNTP-poor environmentof nondividing target cells, leads to incorporation of uracil andpossibly other marks of damage into viral cDNA, thereby flag-ging it for processing and restriction by UNG2 and DNArepair enzymes.

Materials and MethodsCell Lines. HEK293T, U2OS, and U2OS.HLTF.KO cells (45) were maintained inDMEM (Life Technologies) supplemented with 10% (vol/vol) FBS, 2 mML-glutamine, and penicillin/streptomycin in 5% (vol/vol) CO2 at 37 °C. CEM.SS(61) (NIH AIDS Reagent Program), Jurkat, and THP-1 cells were maintainedin RPMI 1640 medium supplemented as described earlier. CEM.SS-iH1.Vpr,U2OS-iH1.Vpr, U2OS-iH1V.Vpr(E24R,R36P), and U2OS.HLTF.KO-iH1.Vpr cellsharboring doxycycline-inducible HIV-1 NL4-3 vpr or other variant vprtransgenes were engineered by using lentiviral Tet-On 3G Inducible Ex-pression System (Clontech) and maintained in the aforementioned mediasupplemented with G418 (200 μg/mL) and puromycin (2 μg/mL). Ex-pression was induced by addition of doxycycline (Sigma-Aldrich) to theculture medium.

Immunoblotting and Antibodies. Typically, whole-cell (34), cytoplasmic, or chro-matin extracts (62) were separated by SDS/PAGE and transferred to PVDF mem-branes for immunoblotting. Proteins were detected with appropriate primaryantibodies and immune complexes revealed with HRP-conjugated antibodiesspecific for the Fc fragment of mouse or rabbit IgG (Jackson ImmunoResearchLaboratories) and enhanced chemiluminescence (GE Healthcare), or with fluo-rescent antibodies to mouse or rabbit IgG (KPL) and Odyssey Infrared Imager(Licor). SI Materials and Methods includes the list of antibodies used.

Transfections, Immunoprecipitations, and DDB1-DCAF1 Recruitment Assay.Transfections of HEK 293T cells and immunoprecipitations were performedas described previously (34, 63). For HLTF recruitment assays, HEK 293T cells,at 2 × 107 cells in four 10-cm plates per condition, were cotransfected withpCG plasmids expressing FLAG-tagged HLTF and appropriate HA-taggedHIV-1 Vpr proteins in combinations. Whole-cell extracts were immunopre-cipitated with FLAG-M2 beads (Sigma-Aldrich), and immune complexes wereeluted by competition with FLAG-peptide under native conditions. HIV-1 Vprimmune complexes for MudPIT analyses were purified as we describedpreviously (6, 63).

Cell Synchronization and Cell Cycle Analyses. U2OS-iH1.Vpr cells were syn-chronized in early S phase by double thymidine block. To reveal cell-cycleprofiles, aliquots of 1 × 105 cells were pulse-labeled with 5-ethynyl-2′-deoxyuridine (EdU; 10 μM) for 60 min, and the incorporated EdU was de-tected by using a Click-iT Plus EdU Alexa Fluor 647 or Alexa Fluor 488 FlowCytometry Assay Kit (Life Technologies). Cells were then stained with 2μg/mL 7AAD (Life Technologies) to reveal their DNA content, and analyzedwith an LSRFortessa flow cytometer (BD Biosciences) and FlowJo software. Atotal of 10,000 events were collected for each sample. Statistical analyses ofcell cycle profiles based on quantification of cell populations in G1, S,and G2/M phases in FlowJo were performed in GraphPad Prism software,with P values calculated by unpaired two-tailed Student’s t test withWelch’s correction.

Isolation of Primary HIV-1–Infected CD4+ T Cells by Cell Sorting. CD4+ T cellsobtained from human peripheral blood mononuclear cells by negative se-lection using EasySep hCD4+ T Cell Enrichment Kit (Stemcell Technologies)were plated in 96-well plates at 1 × 106 cells per well and activated withDynabeads Human T-Activator CD3/CD28 (Invitrogen) in the presence of IL-2(30 U/mL) in full RPMI 1640 medium [supplemented with 10% (vol/vol) heat-inactivated FBS and antibiotics (63)]. Two days later, the cells were infectedwith HIV-1 NL4-3.GFP.R+ or NL4-3.GFP.R− virus and transferred into wells ofa 24-well plate in a final volume of 500 μL per well of full RPMI 1640 mediumwith IL-2 (30 U/mL). Two days post infection, cells were pooled and resus-pended in PBS + 1% BSA at 8 × 106/mL. Dynabeads were removed and liveGFP-positive cells were isolated by sorting on a FACSAria. Whole-cell lysateswere prepared from sorted and from uninfected control cells.

Isolation of G1, S, and G2M Cells by Sorting. Cells (8 × 106) were stained with30 μM Vybrant DyeCycle Green stain (Life Technologies), and G1, S, and G2/Mpopulations (∼1 × 106 cell each) were sorted based on DNA content on aFACSAria. Aliquots (105 cells) of the purified populations were reanalyzed byFACS to assess their purity. The cells were lysed (62, 64), and cytoplasmic orchromatin extracts were analyzed by immunoblotting. Detailed experi-mental procedures are described in SI Materials and Methods.

ACKNOWLEDGMENTS. We thank Jinwoo Ahn, Michael Emerman, NicolasManel, and Karlene Cimprich for sharing reagents; Chuanping Wang fortechnical assistance; Karlene Cimprich for stimulating discussions; Teresa Brose-nitsch for critical reading of the manuscript and editorial help; Angela Gronen-born for support; and Tomek Swigut for his help with statistical analyses. CEM.SScells were obtained through the NIH AIDS Reagent Program, Division of AIDS,National Institute of Allergy and Infectious Diseases, NIH, from Dr. Peter L. Nara.This work was supported by NIH Grants AI077459 and AI100673 and aP50GM082251 subcontract (to J.S.). The Flow Cytometry Facility at Case WesternReserve University is supported by Center for AIDS Research Grant P30 AI036219.

1. Mattapallil JJ, et al. (2005) Massive infection and loss of memory CD4+ T cells inmultiple tissues during acute SIV infection. Nature 434(7037):1093–1097.

2. Nishimura Y, et al. (2005) Resting naive CD4+ T cells are massively infected andeliminated by X4-tropic simian-human immunodeficiency viruses in macaques. ProcNatl Acad Sci USA 102(22):8000–8005.

3. Shingai M, et al. (2015) The expression of functional Vpx during pathogenic SIVmacinfections of rhesus macaques suppresses SAMHD1 in CD4+ memory T cells. PLoSPathog 11(5):e1004928.

4. Zhang Z, et al. (1999) Sexual transmission and propagation of SIV and HIV in restingand activated CD4+ T cells. Science 286(5443):1353–1357.

5. Goldstone DC, et al. (2011) HIV-1 restriction factor SAMHD1 is a deoxynucleosidetriphosphate triphosphohydrolase. Nature 480(7377):379–382.

6. Hrecka K, et al. (2011) Vpx relieves inhibition of HIV-1 infection of macrophagesmediated by the SAMHD1 protein. Nature 474(7353):658–661.

7. Laguette N, et al. (2011) SAMHD1 is the dendritic- and myeloid-cell-specific HIV-1restriction factor counteracted by Vpx. Nature 474(7353):654–657.

8. Lahouassa H, et al. (2012) SAMHD1 restricts the replication of human immunodefi-ciency virus type 1 by depleting the intracellular pool of deoxynucleoside triphos-phates. Nat Immunol 13(3):223–228.

9. Kennedy EM, et al. (2011) Abundant non-canonical dUTP found in primary human

macrophages drives its frequent incorporation by HIV-1 reverse transcriptase. J Biol

Chem 286(28):25047–25055.10. Yan N, O’Day E, Wheeler LA, Engelman A, Lieberman J (2011) HIV DNA is heavily

uracilated, which protects it from autointegration. Proc Natl Acad Sci USA

108(22):9244–9249.11. Lim ES, et al. (2012) The ability of primate lentiviruses to degrade the monocyte re-

striction factor SAMHD1 preceded the birth of the viral accessory protein Vpx. Cell

Host Microbe 11(2):194–204.12. Refsland EW, Harris RS (2013) The APOBEC3 family of retroelement restriction factors.

Curr Top Microbiol Immunol 371:1–27.13. Simon V, Bloch N, Landau NR (2015) Intrinsic host restrictions to HIV-1 and mecha-

nisms of viral escape. Nat Immunol 16(6):546–553.14. Yu X, et al. (2003) Induction of APOBEC3G ubiquitination and degradation by an HIV-1

Vif-Cul5-SCF complex. Science 302(5647):1056–1060.15. Ahn J, et al. (2012) HIV/simian immunodeficiency virus (SIV) accessory virulence factor

Vpx loads the host cell restriction factor SAMHD1 onto the E3 ubiquitin ligase com-

plex CRL4DCAF1. J Biol Chem 287(15):12550–12558.

Hrecka et al. PNAS | Published online June 22, 2016 | E3929

MICRO

BIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

May

31,

202

0

16. Lenzi GM, Domaoal RA, Kim DH, Schinazi RF, Kim B (2015) Mechanistic and kineticdifferences between reverse transcriptases of Vpx coding and non-coding lentiviruses.J Biol Chem 290(50):30078–30086.

17. Doitsh G, Greene WC (2016) Dissecting how CD4 T cells are lost during HIV infection.Cell Host Microbe 19(3):280–291.

18. Nguyen XN, Barateau V, Wu N, Berger G, Cimarelli A (2015) The frequency of cytidineediting of viral DNA is differentially influenced by Vpx and nucleosides during HIV-1or SIVMAC infection of dendritic cells. PLoS One 10(10):e0140561.

19. Weil AF, et al. (2013) Uracil DNA glycosylase initiates degradation of HIV-1 cDNAcontaining misincorporateddUTP and prevents viral integration. Proc Natl Acad SciUSA 110(6):E448–E457.

20. Jowett JB, Xie YM, Chen IS (1999) The presence of human immunodeficiency virustype 1 Vpr correlates with a decrease in the frequency of mutations in a plasmidshuttle vector. J Virol 73(9):7132–7137.

21. Mansky LM (1996) The mutation rate of human immunodeficiency virus type 1 isinfluenced by the vpr gene. Virology 222(2):391–400.

22. Romani B, Cohen EA (2012) Lentivirus Vpr and Vpx accessory proteins usurp thecullin4-DDB1 (DCAF1) E3 ubiquitin ligase. Curr Opin Virol 2(6):755–763.

23. Belzile JP, Abrahamyan LG, Gérard FC, Rougeau N, Cohen EA (2010) Formation ofmobile chromatin-associated nuclear foci containing HIV-1 Vpr and VPRBP is criticalfor the induction of G2 cell cycle arrest. PLoS Pathog 6(9):e1001080.

24. Roshal M, Kim B, Zhu Y, Nghiem P, Planelles V (2003) Activation of the ATR-mediatedDNA damage response by the HIV-1 viral protein R. J Biol Chem 278(28):25879–25886.

25. Zeman MK, Cimprich KA (2014) Causes and consequences of replication stress. NatCell Biol 16(1):2–9.

26. Laguette N, et al. (2014) Premature activation of the SLX4 complex by Vpr promotesG2/M arrest and escape from innate immune sensing. Cell 156(1-2):134–145.

27. Bouhamdan M, et al. (1996) Human immunodeficiency virus type 1 Vpr protein bindsto the uracil DNA glycosylase DNA repair enzyme. J Virol 70(2):697–704.

28. Selig L, et al. (1997) Uracil DNA glycosylase specifically interacts with Vpr of bothhuman immunodeficiency virus type 1 and simian immunodeficiency virus of sootymangabeys, but binding does not correlate with cell cycle arrest. J Virol 71(6):4842–4846.

29. Sleigh R, Sharkey M, Newman MA, Hahn B, Stevenson M (1998) Differential associ-ation of uracil DNA glycosylase with SIVSM Vpr and Vpx proteins. Virology 245(2):338–343.

30. Ahn J, et al. (2010) HIV-1 Vpr loads uracil DNA glycosylase-2 onto DCAF1, a substraterecognition subunit of a cullin 4A-ring E3 ubiquitin ligase for proteasome-dependentdegradation. J Biol Chem 285(48):37333–37341.

31. Eldin P, et al. (2014) Vpr expression abolishes the capacity of HIV-1 infected cells torepair uracilated DNA. Nucleic Acids Res 42(3):1698–1710.

32. Schröfelbauer B, Yu Q, Zeitlin SG, Landau NR (2005) Human immunodeficiency virustype 1 Vpr induces the degradation of the UNG and SMUG uracil-DNA glycosylases.J Virol 79(17):10978–10987.

33. Washburn MP, Wolters D, Yates JR, 3rd (2001) Large-scale analysis of the yeast pro-teome by multidimensional protein identification technology. Nat Biotechnol 19(3):242–247.

34. Hrecka K, et al. (2007) LentiviralVpr usurps Cul4-DDB1[VprBP] E3 ubiquitin ligase tomodulate cell cycle. Proc Natl Acad Sci USA 104(28):11778–11783.

35. Le Rouzic E, et al. (2007) HIV1 Vpr arrests the cell cycle by recruiting DCAF1/VprBP, areceptor of the Cul4-DDB1 ubiquitin ligase. Cell Cycle 6(2):182–188.

36. Huang W, Sherman BT, Lempicki RA (2009) Systematic and integrative analysis oflarge gene lists using DAVID bioinformatics resources. Nat Protoc 4(1):44–57.

37. Huang W, Sherman BT, Lempicki RA (2009) Bioinformatics enrichment tools: Pathstoward the comprehensive functional analysis of large gene lists. Nucleic Acids Res37(1):1–13.

38. Fishel R (2015) Mismatch repair. J Biol Chem 290(44):26395–26403.39. Iyama T, Wilson DM, 3rd (2013) DNA repair mechanisms in dividing and non-dividing

cells. DNA Repair (Amst) 12(8):620–636.40. Krokan HE, Bjørås M (2013) Base excision repair. Cold Spring Harb Perspect Biol 5(4):

a012583.41. Lin JR, Zeman MK, Chen JY, Yee MC, Cimprich KA (2011) SHPRH and HLTF act in a

damage-specific manner to coordinate different forms of postreplication repair andprevent mutagenesis. Mol Cell 42(2):237–249.

42. Unk I, Hajdú I, Blastyák A, Haracska L (2010) Role of yeast Rad5 and its human or-thologs, HLTF and SHPRH in DNA damage tolerance. DNA Repair (Amst) 9(3):257–267.

43. Bolderson E, Scorah J, Helleday T, Smythe C, Meuth M (2004) ATM is required for thecellular response to thymidine induced replication fork stress. Hum Mol Genet 13(23):2937–2945.

44. Schwefel D, et al. (2014) Structural basis of lentiviral subversion of a cellular protein

degradation pathway. Nature 505(7482):234–238.45. Kile AC, et al. (2015) HLTF’s ancient HIRAN domain binds 3′ DNA ends to drive rep-

lication fork reversal. Mol Cell 58(6):1090–1100.46. Ortiz-Bazán MA, et al. (2014) Rad5 plays a major role in the cellular response to DNA

damage during chromosome replication. Cell Reports 9(2):460–468.47. Gao F, et al. (1999) Origin of HIV-1 in the chimpanzee Pan troglodytes troglodytes.

Nature 397(6718):436–441.48. Etienne L, Hahn BH, Sharp PM, Matsen FA, Emerman M (2013) Gene loss and adap-

tation to hominids underlie the ancient origin of HIV-1. Cell Host Microbe 14(1):

85–92.49. Lahouassa H, et al. (2016) HIV-1 Vpr degrades the HLTF DNA translocase in T cells and

macrophages. Proc Natl Acad Sci USA 113(19):5311–5316.50. Mahon C, Krogan NJ, Craik CS, Pick E (2014) Cullin E3 ligases and their rewiring by

viral factors. Biomolecules 4(4):897–930.51. Baldo AM, McClure MA (1999) Evolution and horizontal transfer of dUTPase-encod-

ing genes in viruses and their hosts. J Virol 73(9):7710–7721.52. Elder JH, et al. (1992) Distinct subsets of retroviruses encode dUTPase. J Virol 66(3):

1791–1794.53. Boone LR, Skalka AM (1981) Viral DNA synthesized in vitro by avian retrovirus par-

ticles permeabilized with melittin. II. Evidence for a strand displacement mechanism

in plus-strand synthesis. J Virol 37(1):117–126.54. Huber HE, McCoy JM, Seehra JS, Richardson CC (1989) Human immunodeficiency virus

1 reverse transcriptase. Template binding, processivity, strand displacement synthesis,

and template switching. J Biol Chem 264(8):4669–4678.55. Burdick RC, Hu WS, Pathak VK (2013) Nuclear import of APOBEC3F-labeled HIV-1

preintegration complexes. Proc Natl Acad Sci USA 110(49):E4780–E4789.56. Achar YJ, Balogh D, Haracska L (2011) Coordinated protein and DNA remodeling by

human HLTF on stalled replication fork. Proc Natl Acad Sci USA 108(34):14073–14078.57. DePaula-Silva AB, et al. (2015) Determinants for degradation of SAMHD1, Mus81 and

induction of G2 arrest in HIV-1 Vpr and SIVagmVpr. Virology 477:10–17.58. Sarbajna S, Davies D, West SC (2014) Roles of SLX1-SLX4, MUS81-EME1, and GEN1 in

avoiding genome instability and mitotic catastrophe. Genes Dev 28(10):1124–1136.59. Bailes E, et al. (2003) Hybrid origin of SIV in chimpanzees. Science 300(5626):1713.60. Fregoso OI, et al. (2013) Evolutionary toggling of Vpx/Vpr specificity results in di-

vergent recognition of the restriction factor SAMHD1. PLoS Pathog 9(7):e1003496.61. Nara PL, et al. (1987) Simple, rapid, quantitative, syncytium-forming microassay for

the detection of human immunodeficiency virus neutralizing antibody. AIDS Res Hum

Retroviruses 3(3):283–302.62. Méndez J, Stillman B (2000) Chromatin association of human origin recognition

complex, cdc6, and minichromosome maintenance proteins during the cell cycle:

Assembly of prereplication complexes in late mitosis. Mol Cell Biol 20(22):8602–8612.63. Srivastava S, et al. (2008) Lentiviral Vpx accessory factor targets VprBP/DCAF1 sub-

strate adaptor for cullin 4 E3 ubiquitin ligase to enable macrophage infection. PLoS

Pathog 4(5):e1000059.64. Yan J, et al. (2015) CyclinA2-cyclin-dependent kinase regulates SAMHD1 protein

phosphohydrolase domain. J Biol Chem 290(21):13279–13292.65. Connor RI, Chen BK, Choe S, Landau NR (1995) Vpr is required for efficient replication

of human immunodeficiency virus type-1 in mononuclear phagocytes. Virology

206(2):935–944.66. Nègre D, et al. (2000) Characterization of novel safe lentiviral vectors derived from

simian immunodeficiency virus (SIVmac251) that efficiently transduce mature human

dendritic cells. Gene Ther 7(19):1613–1623.67. Wilson IA, et al. (1984) The structure of an antigenic determinant in a protein. Cell

37(3):767–778.68. Evan GI, Lewis GK, Ramsay G, Bishop JM (1985) Isolation of monoclonal antibodies

specific for human c-myc proto-oncogene product. Mol Cell Biol 5(12):3610–3616.69. Chesebro B, Wehrly K, Nishio J, Perryman S (1992) Macrophage-tropic human im-

munodeficiency virus isolates from different patients exhibit unusual V3 envelope

sequence homogeneity in comparison with T-cell-tropic isolates: Definition of critical

amino acids involved in cell tropism. J Virol 66(11):6547–6554.70. Zhang Y, Wen Z, Washburn MP, Florens L (2010) Refinements to label free proteome

quantitation: How to deal with peptides shared by multiple proteins. Anal Chem

82(6):2272–2281.71. Wechsler T, Newman S, West SC (2011) Aberrant chromosome morphology in human

cells defective for Holliday junction resolution. Nature 471(7340):642–646.

E3930 | www.pnas.org/cgi/doi/10.1073/pnas.1605023113 Hrecka et al.

Dow

nloa

ded

by g

uest

on

May

31,

202

0