Uracil DNA Glycosylase Is Dispensable for Human Immunodeficiency Virus Type 1 Replication and Does Not Contribute to the Antiviral Effects of the Cytidine Deaminase Apobec3G

Abstract

It is well established that many host factors are involved in the replication of human immunodeficiency virus (HIV) type 1. One host protein, uracil DNA glycosylase 2 (UNG2), binds to multiple viral proteins and is packaged into HIV type 1 virions. UNG initiates the removal of uracils from DNA, and this has been proposed to be important both for reverse transcription and as a mediator to the antiviral effect of virion-incorporated Apobec3G, a cytidine deaminase that generates numerous uracils in the viral DNA during virus replication. We used a natural human UNG^−/− cell line as well as cells that express a potent catalytic active-site inhibitor of UNG to assess the effects of removing UNG activity on HIV infectivity. In both cases, we find UNG2 activity and protein to be completely dispensable for virus replication. Moreover, we find that virion-associated UNG2 does not affect the loss of infectivity caused by Apobec3G.

Along with proteins produced by the virus, human immunodeficiency virus (HIV) relies on host cellular proteins to successfully complete its replication cycle. Cellular proteins are utilized in parts of the viral life cycle in the virus-producing cell, the viral target cell, and within the virion itself. Virus-directed incorporation of a cellular protein into the human immunodeficiency virus type 1 (HIV-1) virion has been used as evidence that it plays a role in the life cycle of HIV-1 in vivo. One such host protein that is recruited into HIV-1 virions is the cellular uracil DNA glycosylase, UNG2 (23). Viruses from other families, such as herpesviruses and poxviruses, encode their own uracil DNA glycosylase, suggesting a positive role in virus replication (6). However, the life cycles of these viruses differ greatly, and the need for the uracil DNA glycosylase may differ for each virus.

UNG2 is the major initiating enzyme of the base excision repair pathway that removes uracil lesions from the nuclear genome. Uracil lesions arise either by misincorporation of dUMP during DNA replication or by spontaneous deamination of cytidine. DNA glycosylases act in a “quasi-processive” manner along the DNA, sampling each base until a lesion is found (8). Once a glycosylase recognizes an altered base it catalyzes cleavage of the glycosidic bond, releasing the base and leaving an abasic site in the DNA. UNG2 has greater affinity for the abasic site than for the uracil-containing DNA and thus remains bound to the abasic site after the uracil has been removed (8). The abasic site is next processed by apurinic/apyrimidinic endonuclease (APE1) that nicks the DNA backbone on the 5′ side of the abasic site, generating a 5′-deoxyribose phosphate group that is a substrate for DNA repair enzymes.

Both HIV-1 integrase (IN) and the viral accessory protein Vpr have been shown to bind to and mediate incorporation of UNG2 into the virus particle (27, 34), and the presence of UNG2 in the HIV virion has generated hypotheses that there may be a need to remove uracil from the viral genome. Additionally, the mechanism of cytidine deamination by the intrinsic antiretroviral protein Apobec3G generates numerous uracils in the genome (10, 17). Thus, it was hypothesized that virion-incorporated UNG may contribute to the Apobec3G-mediated loss of infectivity by generating abasic sites in the single-stranded viral DNA intermediate. Subsequent cleavage by an endonuclease at those abasic sites would trigger degradation of the viral DNA (10, 28). Furthermore, it has also been reported that Vpr-associated recruitment of UNG2 controls the virus mutation rate and G-to-A hypermutation (18).

Others have examined the role of UNG2 in HIV replication, but the results are not consistent. For example, one recent study found that UNG2 activity was essential for virus growth at the level of reverse transcription (22), and another study found that UNG2 incorporation into virions was specifically needed for infection of macrophages (5). In contrast, another study found that Vpr-mediated degradation of UNG2 was important for virus replication in the presence of Apobec3G (26). Because many of these previous studies relied on transiently overexpressing UNG2 in the virus-producing cell for some of their conclusions (5, 26, 27), we investigated the role of endogenous levels of UNG2 on HIV replication. Thus, we engineered cells to express a potent “humanized” protein inhibitor of UNG catalytic activity, and additionally we used cells in which both copies of the UNG gene are nonfunctional due to inherited mutations. In both experimental designs, we find that UNG2 is dispensable for viral replication in both producer cells and target cells. We also find that UNG2 activity in the virion is not essential for infection of macrophages in a single-round assay. Thus, our data show that UNG2 is not necessary throughout the life cycle of HIV-1 and are different from a previous publication (22) that found UNG2 activity was essential for HIV-1 replication. Moreover, we also examined the role of UNG2 in mediating the antiviral effects of the cytidine deaminase Apobec3G and find that endogenous levels of UNG2 are not important in the magnitude of antivirus activity by Apobec3G. Together, our results imply that uracil excision is not necessary for proper completion of reverse transcription.

MATERIALS AND METHODS

Generation of hUGI-expressing cell lines.

A codon-optimized UGI for expression in human cells (referred to as hUGI) was synthesized and cloned into the HindIII and ClaI sites of the murine leukemia virus-based retroviral expression vector pLGCX and named pLGC-hUGI. pLGCX is a derivative of pLNCX that expresses green fluorescent protein (GFP) instead of a neomycin resistance gene (19). Vesicular stomatitis virus glycoprotein (VSV-G)-pseudotyped viruses of LGCX and LGC-hUGI were generated in 293T cells by transfection with Fugene 6 (Roche). This virus was used to infect 293T and multiple T-cell lines (H9, Jurkat, and SupT1). GFP was used as an indicator of transduction, and positive cells were sorted by multiple rounds of fluorescence-activated cell sorting until greater than 98% of the cells were expressing GFP with a mean fluorescence intensity of greater than 1 log unit above that of untransduced cells.

UNG functional assay.

Cells were disrupted by sonication in HE buffer (10 mM HEPES [pH 7.4], 1 mM EDTA, 1 mM dithiothreitol), and virus was lysed in HE buffer with the addition of 1% Triton X-100. Lysates were incubated in 1× reaction buffer (10 mM Tris [pH 7.4], 1 mM EDTA, 50 mM NaCl) with 0.1 pmol of a 19-base, single-stranded uracil-containing oligonucleotide (5′-CATAAAGTGUAAAGCCTGG-3′) that was labeled at the 5′ end with IR700. Cleavage of the oligonucleotide at the central uracil results in a nine-base oligonucleotide. Reactions were carried out for 15 min at 37°C and stopped by addition of 0.75 volumes of stop buffer (70% formamide, 0.4 N NaOH, 1× Tris-borate-EDTA [TBE]). Samples were heated to 95°C for 15 min and then electrophoresed on an 8.3 M urea, 20% polyacrylamide gel in TBE. The IR700 infrared tag was detected by a LI-COR infrared scanner (LI-COR Biosciences, Lincoln, NE) and quantitated by LI-COR software.

Virus concentration and purification.

293T or 293T-hUGI cells were transiently transfected with HIV-1/LaiΔEnv or HIV-1/LaiΔEnvΔVpr (16) and the supernatant was collected 48 and 72 h posttransfection. Supernatants were cleared first by centrifugation at 350 × g for 5 min and filtered through a 0.2-μm membrane (Nalgene). Cleared supernatants were then layered on a 20% sucrose cushion and subjected to ultracentrifugation at 23,000 rpm in an SW28 rotor for 2.5 h. Virus was resuspended in HE buffer with 1% Triton X-100 and used for either UNG activity assays or Western blot analysis.

Western blotting.

Cell or viral lysates in HE buffer were heated to 95°C for 5 min and subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (Novex NuPAGE bis-Tris gels; Invitrogen) and transferred onto polyvinylidene difluoride membranes using a semidry transfer apparatus. Membranes were blocked for 1 h at room temperature or overnight at 4°C in 1× phosphate-buffered saline (PBS), 0.1% Tween 20, and 10% milk. The blots were then probed with either polyclonal UNG-AB112 (7) (gift of Sal Caradonna) or polyclonal Apobec3G (20) (gift of Jaisri Lingappa) at 1/1,000 dilution in block buffer (1× PBS, 0.1% Tween 20, 10% milk) for 2 h at room temperature. The membranes were then washed three times for 5 min at room temperature in wash buffer (1× PBS, 0.1% Tween 20) and then incubated with secondary antibody conjugated to horseradish peroxidase (goat anti-rabbit or goat anti-mouse immunoglobulin G; Santa Cruz Biotechnology) at a 1:10,000 dilution for 35 min at room temperature. Blots were again washed three times for 5 min in wash buffer. Bound antibody was detected by incubation with ECL Plus Western blotting detection reagent (Amersham) and exposure to Kodak MR film.

Cells.

Peripheral blood mononuclear cells were isolated from used PALL filters (9) and plated at 1.5 × 10⁷ cells per 12-well plate. After 3 days, nonadherent cells were gently washed off. Medium was aspirated and replaced every 2 days for 9 days. Macrophages were then infected with normalized HIV-1 stocks containing 5 ng of p24^gag. UNG-negative cell lines were derived from a patient with a homozygous mutation in the second exon of UNG2 that results in an early stop codon. B cells from the patient were transformed with Epstein-Barr virus (EBV) and were a gift of Anne Durandy and Hans Ochs (12). The wild-type B cells used as a UNG positive control are from an EBV-transformed B-cell line (GM10966) from the Human Genetic Cell Repository (Coriell Institute for Medical Research). B-cell lines were maintained in RPMI medium supplemented with 10% heat-inactivated fetal calf serum (FCS) and 2 mM l-glutamine. Jurkat, H9, and SupT1 T cells were maintained in RPMI medium supplemented with 10% FCS. 293T cells were maintained in Dulbecco's minimum essential medium supplemented with 10% FCS. The MAGI indicator cell line (32) was maintained in Dulbecco's minimum essential medium supplemented with 10% donor calf serum.

Virus infection and transduction.

HIV constructs used were all derived from LAI (16). Luciferase reporter was put in place of the Nef gene (pLai3-Luc2) (35). Lai-LucΔVif was constructed by an NdeI-StuI deletion. Lai-LucΔVpr was generated by a frame shift mutation in the NcoI site. Wild-type HIV or HIV lacking a functional envelope, pseudotyped with the VSV glycoprotein, was generated as follows. Virus-producing 293T cells were plated at a density of 0.3 × 10⁶ cells per well in a six-well plate. The next day, the cells were transiently transfected with 0.5 μg virus-encoding plasmid (with or without Apobec3G plasmid) with Fugene 6. After 48 h, virus-containing supernatants were collected and filtered (0.2 μM). p24^gag was quantitated by enzyme-linked immunosorbent assay (HIV-1 p24 antigen enzyme immunoassay; Beckman Coulter) or the titer of the virus was determined on MAGI cells (32). Viruses normalized by p24^gag or by an equivalent multiplicity of infection were added to target cells in the presence of 5 μg/ml Polybrene and spinoculated at 1,360 relative centrifugal force for 1 h at 30°C. To measure luciferase reporter activity, cells were transferred to a plate with a V-shaped bottom 36 to 48 h postinfection and pelleted by centrifugation at 1,360 relative centrifugal force for 5 min. Supernatants were aspirated and cell pellets were lysed in 80 μl 1× cell culture lysis buffer (Promega, Madison, WI), and 20 μl was used to measure luciferase with the luciferase assay system (Promega, Madison, WI). To monitor spreading infection, wild-type or hUGI-expressing cells were infected at a multiplicity of infection of 0.05, and p24^gag was monitored in the cell-free supernatants at indicated days. To generate virus in UNG^−/− or wild-type control B cells, 293T cells were transiently cotransfected with Lai3Δenv-Luc2 (35) and VSV-G envelope plasmids to generate replication-competent virus additionally containing VSV-G envelope protein. The titer of this virus was determined on MAGI cells, and equivalent infectious units were used to infect UNG^−/− or control B cells. After spinoculation, cells were washed multiple times with 1× PBS to remove any unbound virus. After 3 days, virus-containing supernatants were collected and p24^gag was normalized. Equivalent p24^gag was used to infect SupT1 T cells. Luciferase activity in SupT1 cells was measured 48 h later as described above.

Quantitative real-time PCR of nascent viral cDNA.

Viruses were treated with 20 U/ml TURBO DNase I (Ambion) in 5 mM MgCl₂ for 30 min at 37°C to remove remaining plasmid DNA, then normalized for p24^gag. Viral preparations containing 20 ng of p24^gag were added to 1.5 × 10⁷ SupT1 cells in the presence of 5 μg/ml Polybrene and spinoculated for 1 h at 30°C and then transferred to a CO₂ incubator. Cells were pelleted and low-molecular-weight DNA was purified using the QIAGEN Miniprep spin protocol. DNA was eluted in 60 μl Tris, pH 8.5, and 10 μl was subjected to real-time quantitative PCR using primers MH531 and MH532 and probe LRT-p for late reverse transcription products as previously described (4). Reactions were performed in TaqMan universal PCR mix (Applied Biosystems) on an ABI Prism model 7900HT (Applied Biosystems).

RESULTS

Cell lines expressing the UNG inhibitor hUGI produce HIV virions containing catalytically inactive UNG.

In order to investigate the role of the uracil DNA glycosylase activity of UNG2 in the life cycle of HIV-1, we generated cell lines deficient for UNG2 catalytic activity by introduction of a potent inhibitory protein. The bacteriophage PBS2 contains uracil as a constituent base in its genome and therefore encodes a small protein, uracil DNA glycosylase inhibitor (UGI), that is a competitive, effectively nonreversible, active site inhibitor of bacterial uracil DNA glycosylase (2, 3, 33). When expressed in human cells, UGI binds to and inactivates both UNG2 and UNG1 but not other cellular uracil DNA glycosylases, such as SMUG1, MBD4, or TDG (33). The bacteriophage UGI gene expressed poorly in human cells (data not shown), and we therefore made cell lines that stably express a codon-optimized form of UGI (hUGI), to achieve optimal expression. To specifically measure UNG activity from these stable cell lines, we assayed functional activity on a single-stranded DNA substrate with a central uracil. We found that 293T cells and various T-cell lines (SupT1 cells are shown) expressing hUGI had no UNG activity above our detection limit, indicating at least a 100-fold decrease in uracil DNA glycosylase activity compared with the parent cell lines (Fig. 1A).

FIG. 1.

hUGI inhibits UNG activity in cells and in HIV-1 produced from those cells. (A) Cell lines expressing the UNG inhibitor hUGI have a greater than 100-fold decrease in UNG activity. Increasing amounts of cell lysate for each cell line were incubated with 0.05 pmol of labeled oligonucleotide containing a uracil, as described in Materials and Methods. The amount of uracil DNA glycosylase activity present in the lysates is denoted in picomoles of a 19-mer oligonucleotide cleaved at the uracil base per minute (A, lower). Cleavage results in generation of a 9-mer oligonucleotide. NEG, oligonucleotide in the absence of lysate and therefore our detection limit (depicted on the graphic representation of the data below the panel as a dotted line); POS, oligonucleotide in the presence of recombinant UDG (New England BioLabs) (used as a positive control). Lines with filled squares represent activity of lysates from WT cells, while lines with open circles represent activity of lysates from cells that express hUGI. (B) Western blot analysis of concentrated virus generated in WT or hUGI-expressing cells with antibodies specific for UNG2 or p24 capsid. (C) Concentrated HIV virions from panel B were assayed for UNG activity in vitro as in panel A. oligo, oligonucleotide.

We also tested whether or not UGI inhibits UNG activity in virions. HIV-1 virions were generated, collected, and concentrated from wild-type (WT) or hUGI-expressing 293T cells. Endogenous UNG2 protein (detected by Western blot analysis) was present in virus produced from both WT and hUGI-expressing cell lines in about the same amounts (Fig. 1B; compare the HIV-1 lane in virions from WT with the HIV-1 lane in virions from hUGI). Using functional assays for UNG activity, however, we found that HIV virions produced from 293T-hUGI cells had no detectable UNG activity compared with HIV virions from WT 293T cells (Fig. 1C). Therefore, we conclude that virus produced in the presence of hUGI incorporates a catalytically inactive UNG2.

It has been reported that expression of Vpr prevents incorporation of UNG into the virion (26). However, we found that endogenous UNG2 is incorporated into HIV virions in the presence or absence of Vpr (Fig. 1B; compare HIV-1 virions from WT with HIVΔVpr from WT), and that the level of UNG indicated by UNG functional activity in the virion is not significantly affected by the presence of Vpr (Fig. 1C). Additionally, in both the presence and the absence of Vpr in the virus-producing cell, additional expression of hUGI effectively inhibits UNG activity in HIV-1 virions (Fig. 1C). Therefore, we conclude that Vpr does not modulate the levels of endogenous UNG2 incorporation into HIV-1 virus particles.

The catalytic activity of UNG is dispensable for HIV-1 infectivity and efficient replication.

Using the cell lines expressing hUGI, we evaluated the requirement for having UNG activity in the virion by performing single-cycle replication assays with virus containing either WT or a catalytically inactive UNG. HIV-1 that encodes luciferase in place of the nef gene was collected from transfected WT or hUGI-expressing cell lines and used to infect several different cell lines. Luciferase expression in the target cells was measured after 48 h. We found that UNG activity in the virion was completely dispensable in single-round infections of HeLa cells and two T-cell lines, SupT1 and Jurkat (Fig. 2). Moreover, we infected primary nondividing macrophages with virus produced in WT or UGI-expressing cells in a single-round infection assay and saw no decrease in infectivity in the absence of UNG catalytic activity compared to the infectivity observed in the presence of UNG activity (Fig. 2).

FIG. 2.

UNG is dispensable for infectivity in single-replication assays. Equivalent amounts of virus generated from WT or hUGI-expressing cells were added to the indicated cell types. Infectivity was measured either by expression of a virus-encoded luciferase reporter in the Nef gene (SupT1, H9, and Jurkat cells) or by virus titer in MAGI cells (HeLa cells), measured by number of blue cells per milliliter. Filled boxes represent virus infectivity from virus generated in WT cells, while stippled boxes represent virus infectivity from virus generated in cells that express hUGI. RLU, relative light units.

Single-round replication assays test only the early stages of virus replication through integration and expression. Therefore, we also considered the possibility that UNG catalytic activity may have additional roles in the later stages of the life cycle of HIV-1 that would be evident in subsequent rounds of replication. Thus, we assayed the kinetics of replication-competent HIV-1 infections in several different T-cell lines (Jurkat, SupT1, and H9) that either were WT or expressed huGI. However, we found that the kinetics of virus replication were nearly identical between matched cell lines that either express endogenous, active UNG or have no UNG activity due to stable expression of the UNG inhibitor, hUGI (Fig. 3). Thus, these data indicate that the catalytic activity of UNG is dispensable for HIV replication in multiple cell lines in both single- and multiround assays.

FIG. 3.

UNG is dispensable for efficient HIV-1 replication in a spreading infection. HIV-1 was added to either WT T cells or hUGI-expressing T cells and allowed to spread through the culture. HIV production was monitored by the amount of p24^gag in the culture supernatant as a function of days after infection.

UNG protein is dispensable for HIV-1 replication.

UNG is required in B-cell class switch recombination (CSR) (1, 12, 24). However, it was recently suggested that, while the presence of the protein is required for CSR, the enzymatic activity is dispensable (1). This hypothesis was based on the results that showed that cells from a UNG knockout mouse were deficient at CSR, however, expression of a catalytically inactive UNG or expression of UNG bound to UGI could restore CSR (1). Thus, we examined the possibility that, like CSR, the catalytic activity of UNG might be dispensable for the replication of HIV-1, yet there still may be a requirement for UNG protein independent of its catalytic activity. In order to examine the HIV life cycle in cells where no UNG was present, we used cells from a patient deficient at immunoglobulin class switch recombination. This deficiency is caused by a homozygous 2-bp deletion in the second exon of the UNG gene, which generates an early stop codon (12). EBV-transformed B-cell lines had been generated from this patient and were reported to have no detectable UNG activity in a functional assay (12). We did not observe UNG2 protein by Western blotting (Fig. 4A, inset); however, when we characterized this cell line, we did find these cells have a small amount (∼5%) of detectable UNG functional activity on single-stranded DNA, compared to a control EBV-transformed B-cell line that is WT for the UNG gene (Fig. 4A). It is likely that the low UNG catalytic activity in the UNG^−/− cell line is due to the uracil DNA glycosylase encoded by the EBV. However, when we assayed UNG activity in concentrated virions that were produced from both WT and UNG^−/− cells, we found no detectable catalytic activity in virions generated from UNG^−/− cells (<1% of WT) (Fig. 4B). This indicates that the low level of UNG activity contributed by EBV in the UNG^−/− cells is not incorporated into HIV-1 at detectable levels.

FIG. 4.

Virus production and infectivity in UNG^−/− cells. (A) 10-fold dilutions of cell lysate from WT or UNG^−/− cells were incubated with a labeled U-containing oligonucleotide as described in the legend to Fig. 1 and Materials and Methods. (Inset) Western blot for UNG2 protein in UNG^+/+ B cells versus UNG^−/− B cells. (B) HIV produced in UNG^−/− cells has no detectable UNG activity. Cell-free supernatants were collected from infected cells (WT or UNG^−/−) and assayed for UNG enzymatic activity as in Fig. 1C or subjected to Western blot analysis for UNG2. (C) HIV was produced in either WT or UNG^−/− cells. Equivalent amounts of p24^gag from these cells were used to infect SupT1 T cells. Infectivity was measured by expression of virally encoded luciferase 48 h after infection. (D) VSV-G-pseudotyped HIV-luciferase (LaiΔenvLuc2) was produced in 293T cells, and increasing amounts of virus were used to infect either WT B cells or UNG^−/− B cells. Virus infectivity is measured by expression of virally encoded luciferase. oligo, oligonucleotide; RLU, relative light units.

We then used the WT or UNG^−/− B cells as virus-producing cells and tested that virus for infectivity in a single-round infection assay, in order to determine whether the presence of UNG protein in the virus particle is required for infectivity. Equivalent amounts of p24^gag harvested from the WT or UNG^−/− B cells were used to infect SupT1 cells. We saw no significant difference in infectivity between viruses produced in cells that either express or do not express UNG (Fig. 4C). This indicates that UNG protein is not required in the virus particle in order for that virus to be infectious.

While the data in Fig. 4C evaluated UNG as part of the virion in the producer cells, we also wished to evaluate the requirement for UNG in the virus target cell. Increasing amounts of HIV-1 that had been generated in 293T cells expressing WT levels of UNG were added to either WT or UNG^−/− cells, and the luciferase activity was measured after a single round of infection (Fig. 4D). Again, no notable difference in infectivity was prominent at any multiplicity of infection between WT and UNG^−/− cells, illustrating that there is no requirement for UNG in the virus target cell (i.e., for the early part of the virus life cycle). Together, the lack of necessity for UNG in either the virus-producing cell or the virus target cell demonstrates that UNG is dispensable for HIV-1 infectivity.

The presence of a catalytically active UNG does not make HIV-1 more sensitive to the antiviral effects of Apobec3G.

While UNG does not seem to affect the efficiency of virus replication, we examined the possibility that the action of UNG in the virion may be apparent when Apobec3G has additionally been packaged into the virion. Since Apobec3G is a cytidine deaminase that generates uracil in the viral genome, it has been suggested that virion-incorporated UNG might play a role in this antiviral pathway (10). It has been shown by others that the loss of infectivity caused by Apobec3G is due to both G-to-A hypermutation and degradation or instability of the viral DNA (10, 17, 28). We hypothesized that having UNG in the virion would make the virus more sensitive to the antiviral effects of Apobec3G, since abasic sites in the single-stranded regions of the viral DNA might leave the viral DNA vulnerable to endonucleases that could trigger degradation of the genome or possibly hinder completion of second-strand synthesis.

In order to test this hypothesis, we produced HIV-1ΔVif in WT or hUGI-expressing 293T cells that were cotransfected with increasing amounts of Apobec3G plasmid (Fig. 5A). This virus was then used to infect either WT or hUGI-expressing target cells. If UNG makes the virus more sensitive to the antiviral effects of Apobec3G, one would expect to see a more dramatic loss of infectivity as Apobec3G concentration in the virus-producing cell is increased. If UNG somehow protects the virus from the effects of Apobec3G, one would see the opposite trend. However, when we compared virus infectivity in single-round infections, we saw identical loss-of-infectivity curves over physiologically relevant levels of Apobec3G, regardless of the status of UNG in the virion or in the target cell (Fig. 5A). These data indicate that UNG does not contribute to the loss of infectivity by the antiviral cytidine deaminase Apobec3G.

FIG. 5.

UNG activity does not render HIV-1 more or less sensitive to the antiviral effects of Apobec3G. (A) HIV-1ΔVif that encodes luciferase in place of nef was cotransfected with increasing amounts of Apobec3G in virus-producing cells containing either a WT UNG or the UNG bound to the inhibitor, hUGI. These viruses were then used to infect WT T cells or hUGI-expressing T cells. Virus infectivity is reported as a percentage of infectivity in the absence of Apobec3G. Filled squares, WT/WT as producer/target; open squares, WT/UGI; filled circles, UGI/WT; and open circles, UGI/UGI. (B) Quantitative-real-time PCR analysis of acute HIV-1 infection in SupT1 cells. Equivalent amounts of p24^gag of virus from WT or hUGI-expressing cells (with or without Apobec3G) were added to SupT1 cells. Low-molecular-weight DNA was collected at indicated times and analyzed by quantitative real-time PCR. Copies of viral DNA are reported for 10⁵ cells. Filled squares, WT; filled circles, UGI; open squares, WT plus 100 ng Apobec3G; open circles, UGI plus 100 ng Apobec3G.

In addition, we looked at the accumulation of reverse transcripts in the presence or absence of Apobec3G in both WT cells and hUGI-expressing cells by quantitative real-time PCR (Fig. 5B). Both early (data not shown) and late reverse transcription (RT) products were measured during an acute viral infection. While there appears to be a moderate defect in accumulation of reverse transcripts due to the activity of Apobec3G, the decrease in accumulation of viral DNA that is seen cannot be attributed to the loss of UNG catalytic activity. From these data, we conclude that the status of UNG catalytic activity does not govern the outcome of Apobec3G-mediated deaminated transcripts.

DISCUSSION

Despite the fact that UNG2 binds to at least two HIV-1 virion proteins (IN and Vpr) and is incorporated into budding virions (18, 34), we show here that the presence of UNG activity in the virion has no obvious effect on viral infectivity in any cell type tested, including primary macrophages. We also show that UNG2 protein itself is not necessary for the viral life cycle, since a cell line with endogenous mutations in both alleles of UNG serves both as an effective target for HIV-1 and as an effective producer of fully infectious virus. Moreover, we find that the loss of infectivity caused by Apobec3G is unaffected by UNG2 in the virion and suggest that UNG2 is not necessary for this antiviral pathway.

As an antiretroviral factor, Apobec3G causes potentially catastrophic deamination of the minus-strand cDNA of the HIV-1 genome during reverse transcription. It has been proposed that virion-incorporated UNG could excise uracil bases as they are generated in the minus-strand cDNA as a result of Apobec3G-mediated deamination (10). The resulting phosphodiester backbone missing a base (abasic site) is vulnerable to endonuclease cleavage, and if no complementary strand exists to hold the DNA together, the viral genome might be degraded. Thus, we hypothesized that in the absence of virion-incorporated UNG2, fewer viral cDNAs would be degraded, thereby increasing the amount of integration-competent provirus. Our data show, however, that this is not the case. We see neither an increase in infectivity in the presence of Apobec3G when UNG2 is not incorporated into the virion nor any increase in copies of late reverse transcription products. This suggests that UNG in the virion does not generate abasic sites that trigger degradation of the viral genome by the action of endonuclease cleavage. Moreover, we see no effect of eliminating UNG catalytic activity in both the virion and the virus target cell, illustrating that UNG catalytic activity is not required in mediating the antiviral mechanism of Apobec3G.

However, our data do not rule out the possibility that heavily deaminated DNA containing uracils or abasic sites may have problems completing reverse transcription since, indeed, we do see fewer RT products in the presence of Apobec3G (Fig. 5B). At the same time, this moderate decrease in RT products in the presence of Apobec3G suggests that much of the uracil-containing viral DNA does complete plus-strand synthesis and become competent for integration. Once the viral DNA is double stranded, base excision repair can be initiated either by UNG in the virus target cell or by other cellular uracil DNA glycosylases (see later in the discussion) to repair the uracil lesions, leaving the signature G-to-A hypermutation in the virus coding strand.

Our data regarding the dispensability of UNG in HIV-1 infectivity are in marked contrast to a recent report that stated UNG2 is essential for HIV-1 replication in all cells tested and suggested the role of UNG2 in HIV-1 replication is to act in concert with reverse transcriptase to repair misincorporated uracils (22). Priet et al. also used UGI to inhibit UNG activity or used small interfering RNA to knock down levels of UNG protein and achieved about a 10-fold decrease in UNG2 activity with both methods. In our studies, expression of a codon-optimized version of UGI resulted in much higher levels of enzymatic inhibition (over 100-fold compared to WT), yet we still detect no reduction in virus infectivity in multiple cell lines or primary macrophages. We also see no defect in accumulation of reverse transcripts. However, it should be mentioned that polymerization across uracil is generally not problematic for many polymerases (15, 31), including HIV-1 reverse transcriptase (36). For example, even in the case of severe uracilation of minus-strand cDNA by Apobec3G, we see only a threefold decrease in the amount of reverse transcripts, both early (data not shown) and late (Fig. 5B). Additionally, an equine infectious anemia virus mutant that lacks the virally encoded dUTPase, which presumably controls uracil misincorporation during reverse transcription, does not exhibit any defects in accumulation of viral DNA (30).

That we see no advantage to having UNG in the virion leads to the obvious question: if UNG2 is dispensable for virus replication, why is it incorporated into the virion, and why, reportedly, by two distinct mechanisms? It should be noted that most of the studies to date have shown virion incorporation of UNG2 while overexpressing UNG2 or UNG2 and Vpr in the virus-producing cell. Since both integrase and Vpr bind directly to UNG2, this overexpression may result in an artifactual appearance of high levels of UNG2 in the virion. Ours is the first report of virion-incorporated endogenous UNG2 where accompanied enzymatic activity is also detected. It could be that an association with either Vpr or IN or both at other steps in the HIV-1 life cycle causes low levels of UNG2 to end up in the virion but that it does not have a function in the virion; this is analogous to the presence of cyclophilin A in HIV-1 virions, which turned out not to be important as a virion protein (11, 29).

Nonetheless, it is still possible that UNG2 is indeed specifically recruited into the HIV-1 particle and that its function becomes significant when there are really low levels of all DNA repair and replication proteins, such as in nondividing cells or macrophages. If it is the enzymatic activity that is beneficial, the need for a uracil DNA glycosylase in trans may be relieved when other complementing uracil DNA glycosylases are active in the virus target cell. Since the UNG^−/− cells we used in this study are rapidly dividing, they express other uracil DNA glycosylases that are capable of repairing uracil in the genome, once it is recognized in the context of double-stranded DNA. Human cells express at least five distinct nuclear uracil DNA glycosylases: UNG2, SMUG1, TDG, MBD4, and UDG2 (14). TDG and MBD4 are strictly specific for double-stranded DNA, having specificity for U:G and T:G mispairs in methylated and unmethylated CpG islands. These likely function to repair deaminated cytosine and 5-methyl-cytosine within that context (13). In the life cycle of HIV, uracil occurs in the provirus either from misincorporation of dUMP during reverse transcription or from deamination of the minus-strand cDNA by Apobec3G. By either mechanism, uracil will be opposite adenine in the context of the double-stranded cDNA. Therefore, of all the cellular uracil DNA glycosylases, only UNG, SMUG1, and UDG2, which can recognize uracil in U:A mispairs, are relevant. Both UNG and SMUG1 can excise uracil from single-stranded DNA, yet the catalytic inhibitor UGI exhibits specificity for UNG alone (21, 25). Since we do not detect any glycosylase activity from the hUGI-expressing cell lines or virions produced from these cells in our functional assay using a single-stranded substrate, endogenous SMUG1 is not likely to be a key player for a UNG backup, although a recent article shows transient overexpression of SMUG1 can lead to its incorporation into HIV-1 virions (26). We do, however, detect uracil DNA glycosylase activity on a double-stranded oligonucleotide in our hUGI-expressing cell lines when the uracil is paired with adenine (data not shown), indicating there undoubtedly is compensating activity present. Ultimately, even if UNG does have a role in the HIV life cycle, studies in the UNG^−/− cell lines clearly indicate other cellular factors have redundant activity and if UNG serves a function, regardless of what it may be, it is not essential for HIV-1 replication.