Functional Role of HIV-1 Virion-associated Uracil DNA Glycosylase 2 in the Correction of G:U Mispairs to G:C Pairs*

Human monocytes/macrophages are target cells for HIV-1 infection. As other non-dividing cells, they are characterized by low and imbalanced intracellular dNTP pool levels and an excess of dUTP. The replication of HIV-1 in this cellular context favors misincorporation of uracil residues into viral DNA because of the use of dUTP in place of dCTP. We have previously reported that the host uracil DNA glycosylase enzyme UNG2 is packaged into HIV-1 viral particles via a specific association with the integrase domain of the Gag-Pol precursor. In this study, we investigated whether virion-associated UNG2 plays a role similar to that of its cellular counterpart. We show that the L172A mutation of integrase impaired the packaging of UNG2 into viral particles. Using a primer-template DNA substrate containing G:U mispairs, we demonstrate that wild-type viral lysate has the ability to repair G:U mismatched pairs to G:C matched pairs, in contrast to UNG2-deficient viral lysate. Moreover, no correction of G:T mispairs by wild-type HIV-1 viral lysate was observed, which argues for the specificity of the repair process. We also show that UNG2 physically associates with the viral reverse transcriptase enzyme. Altogether our data indicate for the first time that a uracil repair pathway is specifically associated with HIV-1 viral particles. However, the molecular mechanism of this process remains to be characterized further.

The control of uracil residue levels in cellular DNA is ensured by a family of uracil DNA glycosylase (UNG) 1 DNA repair enzymes involved in the base excision repair (BER) pathway (for a review see Ref. 1). Uracil in DNA may result from misincorporation of dUTP instead of dTTP or from spontaneous deamination of cytosine. Incorporation into DNA of uracil opposite to adenine is not directly mutagenic. In contrast, incorporation of uracil opposite to guanine is promutagenic and will result, if not corrected, in a G-to-A mutation in the next round of DNA replication.
The uracil DNA glycosylase enzyme recognizes and excises the uracil base from DNA, thus initiating the BER pathway by creating an abasic site. The sugar-phosphate backbone 5Ј of the abasic site is then cleaved by an AP-endonuclease, leaving a 3Ј-OH end. The removal of the baseless sugar residue and insertion of the correct nucleotide are performed by the polymerase ␤ via its lyase and polymerase activities, respectively. Finally, the remaining nick is sealed by the XRCC1-ligase3 complex (2). The nuclei of human cells contain at least five distinct enzymes to excise uracil from DNA, namely UNG2, UDG2, TDG, MBD4, and SMUG1 (3)(4)(5)(6)(7). This redundancy of enzymatic activities required to process uracil residues from DNA argues for the importance of this process in the survival of the cell. In addition to UNGs, all free-living organisms also express the deoxyuridine triphosphatase (dUTPase) enzyme that prevents misincorporation of uracil residues into DNA but through a mechanistically different pathway by acting on the pool of intracellular nucleotides to maintain a low ratio of dUTP to dTTP.
In the viral kingdom, genomes of some DNA viruses, namely pox and herpes viruses, encode both UNG and dUTPase enzymatic activities (8,9). dUTPase and/or UNG minus the mutants of herpes viruses replicate well in cultured dividing cells but are severely impaired in replication, neuroinvasiveness, and reactivation from latency in non-dividing neurons (10 -12). Genomes of ␤-retroviruses such as Mazon Pfizer monkey virus and murine mammary tumor virus and non-primate lentiviruses such as Visna-maedi virus, caprine arthritis-encephalitis virus, feline immunodeficiency virus, and equine infectious anemia virus encode only dUTPase (13). The presence of these virally encoded enzymes in viral particles suggests that they might play an important role in the viral life cycle. Indeed, dUTPase-deficient non-primate lentiviruses replicate well in dividing cells but are impaired in replication in non-dividing macrophages or resting T-cells and display a high level of G-to-A mutations and a loss of invasiveness and pathogenicity in vivo (14 -18). These data strongly argue for the necessity to control the incorporation of uracil residues into proviral DNA in the course of infection of non-dividing cells.
Interestingly, genomes of primate lentiviruses such as human immunodeficiency viruses 1 and 2 (HIV-1 and HIV-2) and simian immunodeficiency viruses encode neither UNG nor dUTPase. Because primate lentiviruses infect non-dividing macrophage cells that are characterized by low intracellular dNTP pool levels with a low dCTP/dTTP ratio and a high level of dUTP pool (19), they have a non-negligible probability to misincorporate uracil residues into their genome during viral DNA synthesis. Therefore, it is likely that they have evolved to control uracil misincorporation into viral DNA. Indeed, we have previously demonstrated that HIV-1 viral particles have the ability to package one of the members of the cellular uracil DNA glycosylase enzyme family, the UNG2 enzyme (20). This packaging occurred via a specific association with the integrase (IN) domain of the Gag-Pol precursor. We have proposed that host UNG2 localized inside HIV-1 viral particles might have a role similar to that played by viral dUTPase encoded by nonprimate lentiviruses (i.e. to prevent the fixation of uracil residues in DNA), avoiding the accumulation of G-to-A mutations in the viral genome. The presence of UNG2 in HIV-1 viral particles highly suggests that HIV-1 might have the ability to control the level of uracil residues misincorporated during viral DNA synthesis.
In this study, we investigated the functional role of virionassociated UNG2 by examining the ability of wild-type or UNG2-deficient HIV-1 viral lysate to process uracil residues from a primer-template DNA substrate containing G:U mismatched pairs. For this purpose, we first demonstrated that leucine residue 172 of IN was critical for UNG2 binding and that recombinant HIV-1 viruses carrying the L172A mutation were impaired for UNG2 packaging into viral particles. We then demonstrated that (i) HIV-1 wild-type viral lysate is competent to correct G:U mispairs to G:C pairs from a uracilcontaining primer-template DNA substrate, (ii) virion-associated UNG2 is a critical component for this process, and (iii) UNG2 and RT can be physically associated. Our data support the notion that HIV-1 virion-associated UNG2 initiates the processing of uracil residues from DNA and that HIV-1 viruses are able to counteract the mutagenic threat of uracil misincorporation into DNA during viral DNA synthesis.

GST Fusion Plasmid Constructions and GST Pull-down Assays-
The IN open reading frame was amplified by PCR from the HIV-1 AD8 molecular clone and subcloned in-frame into the pGEX-5X-2 plasmid vector (Amersham Biosciences) to obtain a resulting plasmid encoding a glutathione S-transferase (GST)-wild-type IN fusion protein. Sitedirected mutagenesis was performed by PCR amplification procedures to obtain IN point mutants. The cDNA encoding the UNG2 sequence (21) was amplified by PCR and cloned in-frame with the GST sequence. The construction of the GST-Gag fusion protein has been described elsewhere (22). All of the constructs were checked by DNA sequencing. GST fusion proteins and His-tagged UNG2 were expressed in Epicurean Coli BL21-codonPlus RIL (Stratagene). The purification of bacterially expressed GST derivatives was performed as reported previously (23). Experimental procedures for GST pull-down assays were done as described previously (20). GST derivatives and recombinant proteins were allowed to interact on a rotating wheel for 2 h at 4°C in 20 mM Hepes, pH 7.6, 250 mM NaCl, 5 mM MgCl 2 , 5 mM dithiothreitol, and 0.5% Nonidet P-40 and then were washed four times with 1 ml of incubation buffer. Anti-UNG2 and anti-IN antibody were kind gifts from G. Slupphaug and D. Trono, respectively. Anti-RT and anti-CA antibodies were purchased from Intracel and Aalto BioReagents, respectively. Primary antibody binding was revealed by horseradish peroxidase-conjugated secondary antibody (DAKO). The horseradish peroxidase activity was detected with ECL Western blotting detection reagents (Amersham Biosciences). Recombinant p66/p51 RT heterodimer, p66/p66 RT homodimer, and p51/p51 RT homodimer were purified until near homogeneity as reported previously (24). Recombinant His 6 -tagged UNG2 was purified until near homogeneity according to Slupphaug et al. (21).
Viral Molecular Clones and Purification of Viruses-Viral molecular clones used in this study were those from HIV-1 NL43 and HIV-1 AD8 . The 3743-bp ApaI-EcoRI fragment of pHIV-1 AD8 was subcloned in pBluescript. An artificial unique ClaI restriction site was then introduced by PCR amplification to modify the fifteenth nucleotide (A versus C) of the IN open reading frame without affecting the codon. To introduce point mutations within the IN sequence, the 708-bp ClaI-HindIII fragment of IN was used as a template, and site-directed mutagenesis was performed using PCR procedures with internal primers containing each of the point mutations and external primers containing ClaI and HindIII restriction sites, respectively. Finally, the ApaI-EcoRI fragment with IN point mutations was cloned back into pNL43 plasmid to obtain HIV-1-mutated IN molecular clones. All of the mutants were checked by sequencing. To obtain the HIV-1 ⌬IN construct, the ClaI restriction site located in HIV-1 IN was filled in and the resulting construct encoded only the five first residues of IN. 293T cells were transfected with proviral DNA in the presence of FuGENE 6 transfectant reagent (Roche Molecular Biochemicals), and 48 h post-transfection, viral particles released in the cell-free supernatant were filtered on 0.45-m pore-size filters and highly purified by ultracentrifugation through 8 -18% Optiprep density gradient as described previously (25). Gradient fractions coinciding with the peak of the reverse transcriptase activity were pooled and normalized for equivalent amounts of CA p24 antigen. Viral lysate was obtained by lysis of purified virions in TNE buffer (10 mM Tris-HCl, pH 7.5, 100 mM NaCl, 1 mM EDTA) in the presence of 0.2% of Triton X-100.
Processing of Uracil-containing Substrate-The 32/70-mer primertemplate DNA substrate was composed of a 5Ј-end labeled 32-mer primer oligonucleotide with a uracil residue at position 32 annealed with a 3-fold molar excess of the unlabeled complementary 70-mer template oligonucleotide. The complete sequence of the 70-mer oligonucleotide is as follows: 5Ј-CACTCTCCATACTTCCAATCTCCACTC-CACCCTCCTCCGTCGACTTCCTGCGCGCTGCGGCTGCGCCTTC-3Ј. In the duplex, the uracil residue is included opposite to guanine within the palindromic sequence recognized by the SalI restriction enzyme. We also used a 32/70-mer primer-template DNA substrate but with a thymine at position 32 so that a G:T mispair was created. The mismatchcontaining substrate processing was assayed as reported previously (26). Viral lysate was incubated with 2 nM of labeled uracil-containing DNA substrate in a final volume of 20 l of a buffer containing 20 mM Hepes, pH 7.6, 60 mM KCl, 10 mM MgCl 2 , 10 M ZnCl 2 , 0.05% Triton X-100, 1 mM dithiothreitol, 5% glycerol, and 100 g/ml bovine serum albumin in the presence of 200 M of each of the four dNTPs. After incubation at 37°C for 30 min and extraction with phenol/chloroform, DNA was recovered by ethanol precipitation in the presence of 3 g of tRNA carrier and subjected to SalI digestion (2 units/point) for 60 min at 37°C in a final volume of 4 l. Four microliters of formamide loading dye was then added, and samples were heated at 75°C for 3 min and resolved by electrophoresis on a 8 M urea-containing 15% polyacrylamide gel in 1ϫ TBE buffer. Radiolabeled DNA products were revealed by autoradiography. (20), we reported that HIV-1 viruses deficient for the expression of IN (HIV-1 ⌬IN) failed to incorporate host UNG2, indicating that IN acts as a carrier to package UNG2 into viral particles. We proposed that virion-associated UNG2 in a manner similar to its cellular counterpart might have a role in controlling misincorporation of uracil residues into DNA during viral DNA synthesis. Because it has been reported that HIV-1 ⌬IN was impaired for efficient viral DNA synthesis (27), we decided to analyze the role of virion-associated UNG2 in the processing of uracil residues by using viruses containing point mutations into IN so that the host UNG2 packaging was altered but not the reverse transcriptase activity.

Mapping of the UNG2 Epitope within IN-In a previous study
To map residues of IN required for association with UNG2, we generated an affinity matrix consisting of a glutathione S-transferase (GST) fused to HIV-1 IN with N-terminal deletions (Fig. 1A) or C-terminal deletions (Fig. 1B) and immobilized on glutathione-agarose beads. Recombinant purified His 6tagged UNG2 was incubated with matrix, and matrix-bound UNG2 was analyzed by SDS-PAGE followed by Western blotting. Similar amounts of GST derivatives were used as judged by Coomassie Blue staining. Results showed that all of the deleted INs still retained their ability to bind UNG2 with the exception of when the region encompassing residues 170 -180 was missing. These data indicate that residues 170 -180 are required for UNG2 binding.
To determine what residues within the region 170 -180 of IN were critical for the UNG2 binding, we generated IN point mutants by alanine-scanning mutagenesis of residues 170 -181. Each of the mutants was expressed as a GST fusion protein, and GST pull-down assays were carried out using recombinant His 6 -tagged UNG2. Results from Fig. 2, left panel, revealed that the mutation of both leucine 172 and lysine 173 residues of IN failed to retain UNG2. All other mutations did not affect the binding of UNG2. To map precisely the residue of IN critical for UNG2 binding, we introduced single point mu-tations on residues belonging to the region 170 -173 of IN. As shown in Fig. 2, right panel, the mutation of residues 170, 171, and 173 has no effect on UNG2 binding. In contrast, the L172A mutation impaired the binding of UNG2. These data indicate that leucine residue 172 is important for the interaction of IN with UNG2, although it cannot be ascertained that leucine 172 is the binding residue of UNG2. Viruses produced in the cell-free supernatant were highly purified by Optiprep gradient velocity centrifugation. Each of the viral lysates was then assayed by Western blotting for the packaging of host UNG2. As shown in Fig. 3, left panel (Fig. 3, right panel). Similar amounts of each of the viral lysates were used as judged by amounts of CA p24 antigen revealed by Coomassie Blue staining of the gel. In addition, Western blotting with anti-CA and anti-IN antibody revealed that Gag and Gag-Pol precursors in viruses carrying IN mutations were processed as well as in wild-type viruses, indicating that mutations did not induce obvious viral protein alterations. These in vivo data are consistent with in vitro data and demonstrate that leucine residue 172 of IN is required to fully package host UNG2 into viral particles. DNA repair enzyme is to initiate the BER process by excising misincorporated uracil residues from DNA. We designed experiments to analyze whether virion-associated UNG2 plays a role similar to that of its cellular counterpart. The uracil-containing DNA substrate (32/70-mer) was composed of a 5Ј end-labeled 32-mer primer oligonucleotide with a uracil residue at position 32 annealed with the unlabeled complementary 70-mer template oligonucleotide (Fig. 4A). In the primer-template DNA substrate, the uracil residue is included opposite to guanine residue within the palindromic sequence recognized by the SalI restriction enzyme. The presence of uracil residue opposite to guanine residue in the SalI sequence has been reported to prevent DNA cleavage by SalI (26). The restoration of the SalI digest implies that the uracil residue in G:U mispairs was corrected to G:C pairs. The 32/70-mer DNA substrate was incubated with increasing amounts (0.3, 0.6, and 1.2 g of CA p24 antigen) of highly purified HIV-1 wild type or L172A/K173A IN mutant viral lysate in the presence of each of the four dNTPs. The substrate was then recovered by ethanol precipitation and subjected to SalI digestion. DNA products were resolved on urea-denaturing gel and analyzed by autoradiography (Fig. 4B). Efficient DNA extension of the primer by increased levels of wild-type or L172A/K173A IN viral lysate was evidenced by the appearance of increased levels of the labeled 70-mer product concomitant with decreased levels of the labeled 32-mer primer. A proportion of the DNA substrate upon incubation with increasing amounts of wild-type viral lysate became susceptible to cleavage by SalI as judged by the appearance of a dose-dependent labeled DNA product with the expected size (27-mer) for the SalI digestion product. This results presumably from a specific repair of G:U mismatched pairs to G:C matched pairs. In contrast, incubation of the DNA substrate with similar amounts of L172A/K173A IN viral lysate failed to generate the 27-mer SalI digestion product. The failure to observe uracil processing upon incubation of the DNA substrate with UNG2-deficient viruses serves as an internal control to certify that cellular proteins in microvesicles did not significantly contaminate our purified virion preparations (28) and to eliminate the possibility that the G:U versus G:C correction was the result of a virion-independent endonucleolytic processing of the substrate. As a control, we incubated a primer-template DNA substrate containing a G:T mismatch instead of a G:U mismatch with increased amounts (0.3, 0.6, and 1.2 g of CA p24 antigen) of wild-type viral lysate, and we subjected it to SalI digestion. As shown in Fig. 4B, right panel, no SalI digest product was observed, indicating the specificity of the repair process. Altogether, our data indicate that uracil residues are corrected from G:U mispairs to G:C pairs in the presence of HIV-1 viral lysate and that virion-associated UNG2 plays a key role in initiating the repair process.

Incorporation of Host UNG2 into IN Point Mutants-We
Physical Association of UNG2 with the Viral Reverse Transcriptase Enzyme-In a previous study, we reported that a G:U mispair buried in the central part of a double-stranded DNA substrate was not corrected to a G:C pair by HIV-1 viral lysate (20). In contrast, we observed here the correction of G:U mispairs located near the free 3Ј-OH end of the primer in the primer-template DNA substrate. The presence of a free 3Ј-OH end in close proximity to the G:U mispair to anchor the viral RT enzyme on the DNA primer seems to be important for uracil processing. We hypothesized that UNG2 and RT might be spatially close to act in a concerted manner to process uracil. Therefore, we examined whether UNG2 and RT have the ability to associate. GST affinity matrix containing HIV-1 IN, HIV-1 Pr55 gag , or UNG2 was incubated with recombinant purified p66/p51 RT heterodimer, and matrix-bound RT was analyzed by SDS-PAGE followed by Western blotting (Fig. 5, left  panel). RT binds to HIV-1 integrase as reported previously (29) as well as to UNG2 but not to GST or GST-Pr55 gag . The possibility that nucleic acids interfered in the interaction RT-UNG2 was ruled out by pretreating the mixture of RT-GST derivatives with micrococcal nuclease and RNase A, which did not decrease amounts of the RT pull-down assay (data not shown). We also tested the ability of p66/p66 RT homodimer (Fig. 5, middle panel) or p51/p51 RT homodimer (Fig. 5, right  panel) to associate with UNG2. The results indicated that each of the homodimers still retained UNG2 association, suggesting that the RNase H domain of RT was probably not involved in the interaction with UNG2. These data indicate that UNG2 and RT physically associated. DISCUSSION We have previously reported that HIV-1 viral particles have the unique property of packaging the host uracil DNA glycosylase DNA repair enzyme UNG2 into viral particles (20). In this paper, we investigated the functional role of virion-associated UNG2 in uracil processing by studying the ability of HIV-1 viral lysate to correct G:U mismatched pairs incorporated into a primer-template DNA substrate. The repair of G:U mispairs to G:C pairs was revealed by the restoration of the cytidine residue instead of the uridine residue in the nucleotidic se- quence recognized and cleaved by the SalI restriction enzyme. We showed that HIV-1 wild-type viral lysate, but not UNG2deficient viral lysate, can process uracil residues when they are present at the free 3Ј-OH end of the primer in the primertemplate DNA substrate. In our early study (20), we showed that HIV-1 viral lysate failed to process uracil residues present in a G:U mispair located in the middle of a blunt doublestranded DNA substrate, probably because of the absence of a free 3Ј-OH end to anchor RT on its primer. In contrast, recombinant UNG2 alone efficiently processed uracil residues in this case. These findings suggested that UNG2, in the context of HIV-1 viral particles, might act in a manner distinct from that of recombinant UNG2. Indeed, we demonstrated that (i) UNG2 is associated and packaged with IN, (ii) UNG2 can associate with RT, and (iii) RT and IN are associated. This latter association has been previously reported to be critical for the initiation of the retrotranscription of viral genome (29). We hypothesized that RT, IN, and UNG2 are physically associated and are part of the viral retrotranscription complex. As a consequence of such a multienzymatic complex, UNG2 would probably not be free in the viral particle and could act only on uracil residues in close proximity of a free 3Ј-OH anchor for the retrotranscription complex. On the basis of this hypothesis, we proposed that DNA polymerization by RT and DNA repair initiated by virion-associated UNG2 are coupled events.
The viral retrotranscription process is a two-step process consisting of the synthesis of the minus-strand DNA using viral RNA as a template followed by the synthesis of the plus-strand DNA using the minus-strand DNA as a template to obtain the proviral double-stranded DNA. One mispairing mutation occurring in the minus-strand DNA of the heteroduplex is mutagenic and is fixed in the plus-strand DNA of the DNA/DNA homoduplex constituting the proviral genome. For example, dUMP residues in the minus-strand DNA will be paired with dAMP residues during the synthesis of the plus-strand DNA, contributing to the accumulation of G-to-A substitutions. We demonstrated that virion-associated UNG2 has the function of preventing accumulation of uracil and, therefore, G-to-A transition in the context of a DNA/DNA duplex. We have not looked at the repair process for RNA/DNA heteroduplexes, which are intermediate products formed during the retrotranscription process, but a previous study (30) have shown that the repair process of a RNA/DNA duplex was effective, albeit with a reduced efficiency, compared with that on a DNA/DNA duplex. Moreover, it has been reported that HIV-1 viruses with an excess of UNG2 packaged by Vpr displayed a more accurate reverse transcription during a unique cycle of infection (31), suggesting that UNG2 has the function of lowering the mutation rate. Taken together, these data support the notion that HIV-1 is able to control the amount of uracil incorporated into its genome during reverse transcription and to prevent the catastrophic G-to-A hypermutations, which have been reported when the dNTP concentration is biased (32,33).
The question arises whether the viral BER process occurred through a pathway similar to that of cellular BER and thus required an AP-endonuclease activity to cleave 5Ј to the abasic site resulting from the action of the DNA glycosylase 2. Western blot analysis of HIV-1 viral lysate revealed no packaging of host AP-endonuclease 1 (34,35) in experimental conditions where host UNG2 was clearly detected in virions (data not shown). Similar negative results (data not shown) were obtained by analyzing viral lysate of virions produced from cells overexpressing either hemagglutinin-tagged AP-endonuclease 1 or hemagglutinin-tagged AP-endonuclease 2, a newly described member of the human AP-endonuclease family (36). Therefore, we speculate that HIV-1 viruses did not contain known AP-endonucleases, suggesting that the processing of uracil residues by HIV-1 viral lysate occurred independently of the presence of a AP-endonuclease activity. An alternative hypothesis would be that the creation of an abasic site by the virally associated UNG2 is followed by the elimination of the baseless nucleotide by the pyrophosphorolytic activity of RT, leaving a free 3Ј-OH end to be used by the polymerase activity of RT. Further studies are in progress to understand the molecular mechanism(s) involved in viral BER.
The fact that the HIV-1 reverse transcription process is mainly accomplished in the cytoplasm of infected cells and the fact that cellular enzymes participating in the DNA repair process are nuclear proteins might explain why HIV-1 particles have to be equipped with their own enzymatic DNA repair activities. In conclusion, we propose that HIV-1 viruses in the manner of non-primate lentiviruses, pox viruses, or herpes viruses which replicate in non-dividing cells, have developed an original strategy to bypass the deleterious consequence of uracil in viral DNA by using IN as a carrier to package host UNG2 into viral particles and by using RT to target UNG2 at the site of nascent DNA synthesis.