The Presence of Human Immunodeficiency Virus Type 1 Vpr Correlates with a Decrease in the Frequency of Mutations in a Plasmid Shuttle Vector

Abstract

The human immunodeficiency virus type 1 (HIV-1) Vpr protein induces cell cycle arrest at the border of G₂ and M similar to the arrest caused by agents which damage DNA. We determined whether the presence of Vpr would affect the ability of cells to repair DNA. We developed a shuttle vector system to analyze the effect of Vpr upon the repair of UV-damaged DNA. Our results demonstrated that the presence of Vpr decreased the rate of deletions in this system. Of note, cells arrested in G₂ by other genotoxic agents also increased the frequency of DNA repair of UV-damaged shuttle vectors. We did not observe any direct effect of Vpr upon the rate of double-strand break repair and/or nucleotide excision repair of genomic DNA in cells. Our results suggest a role for HIV-1 Vpr in altering the frequency of DNA repair, a property which may have importance for HIV-1 replication and pathogenesis.

The genome of human immunodeficiency virus (HIV), the etiological agent of AIDS, contains nine genes, five of which have been determined to be essential for replication in vitro. The remaining four genes, vpr, vpu, vif, and nef, have consequently been dubbed nonessential accessory genes, and they play various roles, either directly, in enhancing the efficiency of the replication cycle of HIV, or indirectly, in enhancing survival in the presence of the immune system in vivo and potentially in the causation of disease (10, 11, 27).

Vpr encodes an 11-kDa protein that is incorporated into the virion via an interaction with the carboxy-terminal portion of the Gag precursor protein (17). Vpr has been proposed to promote nuclear localization of the viral preintegration complex following infection (5, 15), although this role has been disputed (13). This function of Vpr has been suggested to increase the efficiency of replication of HIV in nondividing cells (3, 7). Furthermore, Vpr has been ascribed additional roles in enhancement of transcriptional activity on various promoters, including the HIV long terminal repeat (6), and in the rescue of viral expression from latently infected cell lines (20), although the roles of these effects on the viral life cycle have yet to be determined.

We and others have reported that the HIV type 1 (HIV-1) vpr gene product causes cell cycle arrest in HIV-infected target cells (14, 16, 32, 33). We further proposed that cell cycle arrest induced by Vpr may play a role in pathogenesis by inhibiting the clonal expansion and effective immune response of T cells (16, 31).

Phenotypic characterization of the Vpr-mediated arrest revealed that the cells arrested in the second gap (G₂) phase of the cell cycle, prior to initiation of mitosis (M). The Cdc2 kinase, a major component controlling the transition of cells from G₂ to M, was found to be predominantly in the inactive state (16, 30). This point of arrest can also be triggered in a cell following certain types of DNA damage. Alkylating agents and ionizing radiation cause cells to arrest or delay in G₂ and prevent or postpone the onset of mitosis, depending upon the severity of the damage. This pause in cell cycle progression has been suggested as a period in which repair to the DNA is carried out, after which the cell resumes cell cycle progression into mitosis (21, 24, 26). We identified a phenotypic similarity between Vpr-mediated cell cycle arrest and arrest caused by DNA-damaging agents (30). The attributes identified included growth arrest in the G₂ phase of the cell cycle with hyperphosphorylation of Cdc2 kinase and the ability of methylxanthines to reverse cell cycle arrest by both Vpr and nitrogen mustard. We proposed, therefore, that Vpr may mediate growth arrest either through (i) directly causing DNA damage, (ii) interfering with DNA damage repair processes, or (iii) modulating a DNA damage detection pathway (checkpoint). In this study we determined whether the presence of Vpr influences the rate of DNA damage and repair in cells. We examined the effect of Vpr on both cellular DNA and exogenously introduced nonviral templates.

Our results indicate that the presence of Vpr results in a reduction in the frequency of deletions in a shuttle vector system. Vpr, however, did not detectably affect nucleotide excision repair or double-strand break repair on cellular DNA.

MATERIALS AND METHODS

Cells and viral stocks.

COS cells (African green monkey kidney) were grown in Dulbecco’s modified Eagle’s medium with 10% calf serum (Gemini). Stocks of HIV-1_NL4-3-thy env(−)/vesicular stomatitis virus G protein (VSV-G) and HIV-1_NL4-3-thyvprXenv(−)/VSV-G were prepared and used to infect target cells as previously described (30). The measurement of levels of infection was performed with anti-Thy 1.2 antibodies and flow cytometry as described previously (16). In the case where Vpr was provided in the absence of all other HIV proteins, the vector NLthyΔBgl was replaced with both pHRcmvypyvpr and pCMVdR8.2DVPR, which made up the viral stock pHRVpr. The former was derived from pHRcmvluciferase (25) by removal of the luciferase open reading frame and substitution with the epitope-tagged vpr open reading frame. As a control for this experiment, an additional vector was constructed, pHRcmvthy, which was identical to pHRcmvypyvpr except that the murine thy1.2 gene (29) replaced the vpr open reading frame (giving the viral stock pHRThy). The latter plasmid, pCMVdR8.2DVPR, was used as a packaging plasmid based on pCMVdR8.2D (25), but with the vpr open reading frame deleted. The R80A mutant of Vpr, prepared in the NLthyΔBgl vector, consists of a single-amino-acid change at codon 80 from arginine to alanine and is identical to the mutant previously described (8).

Shuttle vector.

The shuttle vector used in these DNA damage detection experiments was the pBK-CMV (pBK) phagemid vector obtained from Stratagene. Where stated, the plasmid was irradiated in Tris-EDTA buffer (pH 7.5) on parafilm in a Stratalinker UV light irradiator at a rate of approximately 6 J/s. At 16 h postinfection, to allow reverse transcription, integration, and expression of viral proteins, the cells were transfected with the pBK shuttle vector by using Lipofection (Life Technologies) according to the manufacturer’s protocol. The cells were harvested after a further 48 h (allowing maximal episomal replication of the plasmid) and lysed, and a Hirt extraction was performed to recover low-molecular-weight DNA. The recovered DNA was transformed into Escherichia coli and plated on nutrient medium. Colonies were scored after overnight incubation at 37°C and a further 4- to 6-h incubation at 4°C to enhance the blue color. Linearized shuttle vector was prepared by digestion with HindIII, which cuts within the LacZα peptide coding region.

CHEF pulsed-field gel electrophoresis.

The method and reagents used for contour-clamped homogeneous electric field (CHEF) pulsed-field gel electrophoresis were derived from the CHEF genomic DNA plug kit (Bio-Rad). Briefly, cells were harvested with trypsin, washed, and set in a 2% soft agar plug at 2 × 10⁶ to 3 × 10⁶/ml. The plugs were digested overnight at 50°C with proteinase K (1 mg/ml) and then washed four times in wash buffer. The agarose plugs were then set in the electrophoresis gel (1%) and subjected to CHEF pulsed-field electrophoresis in 0.25× Tris-borate-EDTA buffer as follows: initial pulse time, 60 s; final pulse time, 120 s; field strength, 5.2 V/cm; duration, 22.5 h. These settings resolve fragments less than approximately 2 Mbp; the larger fragments will migrate in a “compression” zone at the same point, while unbroken genomic DNA remains at the origin (4, 18).

Unscheduled DNA synthesis (UDS) assay.

Cells were infected as described above. A strong inhibitor of replicative DNA synthesis (hydroxyurea [10 mM]) was added 1 h prior to UV irradiation (30 J/m²). The use of this inhibitor is required because of the potential for high levels of background [³H]thymidine incorporation as the cells progress through DNA synthesis (S) phase. Immediately following UV irradiation, [³H]thymidine was added to the tissue culture medium (50 μCi/ml) and pulsed for a further 2 h in the presence of hydroxyurea. The cells were harvested and lysed on a filter prior to being immersed in scintillation fluid and counted. The values reported are the means of triplicate samples.

Statistical analysis.

Statistical significance (where indicated) was calculated by either the chi-square test or Fisher’s exact test (two-tailed method).

RESULTS

Vpr decreases DNA mutation frequency in a transfected shuttle vector.

To investigate whether Vpr affects the repair of damaged DNA, we utilized a shuttle vector (pBK) containing the lacZα gene under the control of a bacterial promoter as a reporter gene for mutation. When transformed into E. coli, the colonies turn blue in the presence of inducer and substrate. A white colony would indicate a mutation in either the LacZα peptide reporter gene or in its expression control regions. The vector also contained a eukaryotic origin of replication derived from simian virus 40 (SV40) allowing episomal replication in COS cells to sufficiently high levels for detection after extraction and transformation into bacteria. We first determined conditions which would provide a sufficiently high rate of mutation for us to monitor the effects of Vpr. We found that UV irradiation of plasmid DNA at a dose of 2,500 J/m² followed by DNA transfection and recovery after 48 h was sufficient to achieve a mutation frequency of approximately 3%.

We determined the effect of Vpr on the mutation frequency by expressing Vpr in cells prior to transfection with the reporter plasmid. COS cells were infected at a high multiplicity of infection with a pseudotyped virus based on the HIV-1_NL4-3 strain carrying either a wild-type Vpr [HIV-1_NL4-3-thyenv(−)/VSV-G] or a frame-shifted mutant of Vpr, termed VprX [HIV-1_NL4-3-thyvprXenv(−)/VSV-G] that results in truncation of approximately one-third of the protein (16, 29). A sufficient innoculum of virus was used to ensure that more than 90% of the cells were infected, which was verified by Thy 1.2 expression and flow cytometry (data not shown). The mutant protein does not induce G₂ cell cycle arrest. At 16 h after infection, the cells were transfected with the pBK shuttle vector and harvested after a further 48 h. The recovered episomal DNA was transformed into E. coli, and the blue and white colonies were counted. The presence of Vpr reduced the mutation frequency occurring at the lacZ gene locus by approximately 10-fold compared to that of mock- or VprX-infected cells (Table 1). Although the magnitude varied in different experiments, a 10-fold or greater level of reduction of the mutation frequency of the shuttle vector was consistently observed with Vpr.

TABLE 1.

Vpr decreases mutation frequency in UV-damaged shuttle vector^a

Virus stock	Expt 1		Expt 2		Expt 3
	No. of white/blue colonies	% White colonies	No. of white/blue colonies	% White colonies	No. of white/blue colonies	% White colonies
Mock	18/527	3.30	8/1,318	0.60
Vpr	0/2,415	<0.04	0/1,557	<0.06	0/142	<0.7
VprX	17/1,708	0.98	6/888	0.67	25/80	24.0
P value		<0.0001		0.0021		<0.0001

^aCOS cells (2 × 10⁵) were either mock infected or infected with HIV- 1_NL4-3-thyenv(−)/VSV-G or HIV-1_NL4-3-thyvprXenv(−)/VSV-G. At 16 h postinfection, the cells were transfected with 4 μg of pBK-CMV plasmid (UV irradiated at 2,500 J/m² by using Lipofectin. After a further 48 h, the cells were harvested and low-molecular-weight DNA was recovered by Hirt extraction and transformed into E. coli. Blue and white colonies were scored to give mutation frequencies. The observed number of colonies for Vpr-infected cells was found to differ significantly from those for mock- (experiments 1 and 2) and VprX (experiment 3)-infected cells, with the P values given in the bottom row of the table. Experiments 1, 2, and 3 are independent repetitions of the same protocol. 

We confirmed that the effect was due specifically to Vpr by using a point mutant of Vpr (R80A) that is expressed stably yet fails to cause cell cycle arrest (8). We found that VprR80A resulted in a mutation frequency in our shuttle vector (2.6%) similar to that of mock-infected COS cells (2.9%), while wild-type Vpr resulted in a 0.16% mutation frequency (Table 2).

TABLE 2.

Decreased mutation frequency correlates with G₂ arrest phenotype^a

Virus stock	No. of white/blue colonies	% White colonies
Mock	36/1,191	2.90
Vpr	2/1,184	0.16
VprX	10/832	1.20
VprR80A	55/2,055	2.60

^aCOS cells (2 × 10⁵) were either mock infected or infected with HIV- 1_NL4-3-thyenv(−)/VSV-G (Vpr), HIV-1_NL4-3-thyvprXenv(−)/VSV-G (VprX), or HIV-1_NL4-3-thyenv(−)/VSV-G containing an arginine-to-alanine mutation in Vpr at codon 80 (VprR80A). At 16 h postinfection, the cells were transfected with 4 μg of pBK-CMV plasmid (UV irradiated at 2,500 J/m²) by using Lipofectin. After a further 48 h, the cells were harvested and low-molecular-weight DNA was recovered by Hirt extraction and transformed into E. coli. Blue and white colonies were scored to give mutation frequencies. The results observed for Vpr were statistically significant (P < 0.0001) when compared to mock infection. Conversely, those observed for R80A were not significant (P = 0.58). This experiment was repeated twice with similar results. 

We verified that Vpr alone was the cause of the observed reduction in mutation frequency and that it was not a result of other viral proteins present in the NL-Thy vector. We prepared a retroviral vector capable of only Vpr production following infection (pHRVpr). The level of suppression was approximately 10-fold that of the control (pHRThy) and was similar to that obtained with virus carrying the wild-type vpr gene (Table 3).

TABLE 3.

Vpr causes decreased mutation frequency in the absence of all other viral proteins^a

Virus stock	No. of white/blue colonies	% White colonies
Mock	105/9,225	1.10
VprX	44/7,061	0.62
Vpr	7/4,770	0.15
pHRThy	88/5,954	1.40
pHRVpr	2/2,665	0.07

^aCOS cells (2 × 10⁵) were either mock infected or infected with HIV-1_NL4-3-thyenv(−)/VSV-G (Vpr) or HIV-1_NL4-3-thyvprXenv(−)/VSV-G (VprX). At the same time, two additional cultures were infected with a virus capable of producing only Vpr (pHRVpr) or a control producing Thy (pHRThy) and no other viral proteins. At 16 h postinfection, the cells were transfected with 4 μg of pBK-CMV plasmid (UV irradiated at 2,500 J/m²) by using Lipofectin. After a further 48 h, the cells were harvested and low-molecular-weight DNA was recovered by Hirt extraction and transformed into E. coli. Blue and white colonies were scored to give mutation frequencies. The observed counts for Vpr compared to those for Mock infection and those of pHRVpr compared to those for pHRThy achieved significance, with both P values less than 0.0001. 

Vpr reduces the frequency of deletion mutants.

Plasmids were recovered from randomly selected white colonies and analyzed by nucleotide sequence analysis for the type of mutation that abolishes LacZα peptide function. All of the mutations were found to be deletions in the lacZα peptide open reading frame and/or its promoter region (data not shown).

Deletions in DNA may arise when a double-strand DNA break occurs, followed by exonuclease digestion and end rejoining, with the size of the deletion determined by the level of exonuclease activity and the rate of end ligation. Alternatively, a deletion may occur when recombination takes place between two homologous regions of DNA. The size of this deletion is determined by the size of the intervening DNA segment between the homologous regions. Deletion junctions in a random selection of recovered mutant shuttle vectors derived from mock, Vpr-, and VprX-containing cultures were examined by nucleotide sequence analysis for evidence of homologous recombination. We found that the sizes of the deletions varied from one plasmid to the next and that there were no obvious stretches of homology at or near the deletion junctions that would be indicative of homologous recombination (data not shown). This observation is consistent with published work with similar shuttle vector systems (9).

Vpr reduces mutation frequency in the repair of linearized shuttle vector DNA.

The above results indicate that deletion mutants could result from cleavage of DNA followed by nuclease activity. We therefore tested whether Vpr could affect mutation frequency in the shuttle vector cleaved by a restriction endonuclease. The shuttle vector was digested to completion with an enzyme that cuts once within the lacZα peptide coding region. We observed overall a higher frequency of white colonies, as might be expected following transfection of linearized plasmid DNA. As expected, nucleotide sequence analysis of white colonies demonstrated evidence for deletions extending from the cleavage site. We found that the presence of Vpr in this experiment decreased the frequency of mutation by approximately two- to threefold compared to that of either mock-infected or VprX virus-infected cells (Table 4). This result suggested that the presence of Vpr reduces the mutation frequency during the repair and recircularization of linear DNA. One interpretation is that the presence of Vpr facilitates accurate end rejoining of double-strand breaks in DNA. Alternatively, the presence of Vpr may inhibit double-strand break rejoining. Since plasmids with double-strand breaks would not amplify in COS cells, any plasmids bearing mutations originating from a double-strand break would not be detected.

TABLE 4.

Vpr decreases mutation frequency in transfected linearized shuttle vector^a

Virus stock	No. of White/blue colonies	% White colonies
Mock	119/181	40
Vpr	24/80	23
VprX	103/81	56

^aCOS cells (2 × 10⁵) were either mock infected or infected with HIV-1_NL4-3-thyenv(−)/VSV-G (Vpr) or HIV-1_NL4-3-thyvprXenv(−)/VSV-G (VprX). At 16 h postinfection, the cells were transfected with 4 μg of pBK-CMV plasmid that was linearized with HindIII. After a further 48 h, the cells were harvested and low-molecular-weight DNA was recovered by Hirt extraction and transformed into E. coli. Blue and white colonies were scored to give mutation frequencies. The values observed for Vpr compared to those for VprX and mock infection were found to be statistically significant, with associated P values of <0.0001 and 0.0023, respectively. 

Vpr does not directly affect endogenous cellular DNA repair processes.

The above observations are consistent with the idea that the presence of Vpr acts to increase the rate of ligation, thereby preventing the action of nucleases on the free DNA ends; to inhibit cellular exonuclease digestion of free ends; or to inhibit rejoining of double-strand breaks. We tested the possibility that Vpr influences the ability of cellular DNA to repair itself. We measured the rate of endogenous DNA repair either in the presence or absence of Vpr following the generation of double-strand breaks by γ-irradiation. Pulsed-field gel electrophoresis was used to monitor strand breakage and its rate of repair (2, 18). Vpr did not itself cause detectable levels of double-strand DNA breaks (Fig. 1). γ-Irradiation (80 Gy) caused double-strand breakage as measured by the appearance of a lower-molecular-weight band on a pulsed-field gel (Fig. 1). The DNA was repaired completely by 240 min following irradiation, as evidenced by the diminishing intensity of the faster-migrating band over time, as has been previously described (18). In the presence of Vpr, the faster-migrating band diminished at a similar rate, indicating that Vpr did not alter the rate of repair of the double-strand DNA breaks induced by irradiation. Thus, within the limits of this assay, we conclude that Vpr does not detectably enhance or inhibit the rate of repair of double-strand breaks generated by γ-irradiation. Furthermore, the presence of Vpr itself does not induce double-strand breaks in DNA. However, we could not determine from this experiment whether the presence of Vpr affects the rate of mutation at or around the breakpoint junctions.

FIG. 1.

Vpr does not affect rate of endogenous DNA double-strand break repair induced by γ-irradiation. COS cells (6 × 10⁵) were infected with HIV-1_NL4-3-thyenv(−)/VSV-G (Vpr) to give a >90% infection rate as measured by Thy expression at 48 h postinfection. At 18 h postinfection, the cells were exposed to 80 Gy of γ-irradiation and harvested at the indicated time points. The cells were embedded in 1% soft agarose, gently lysed, and subjected to pulsed-field electrophoresis at 5.1 V/cm and 60- to 120-s ramp time for 22 h. Lane Y, Saccharomyces cerevisiae chromosomes were used as markers, with the sizes indicated.

Different repair pathways are utilized by the cell to repair DNA damage, depending upon the nature of the damage to the DNA. Nucleotide excision repair (NER) is the primary mechanism active in repair of cyclobutane pyrimidine dimers, the primary form of damage caused by UV light irradiation. Vpr has been shown to bind to HHR23A (36), a protein that can bind xeroderma pigmentosum-complementing protein group C (XPC). XPC likely plays a role in NER, although its function has not yet been determined (1, 23). A part of normal NER activity is in the resynthesis of the damaged stretch of DNA by using the complementary strand as a template. This has been called UDS and can be measured in cells following UV irradiation by using radioactively labeled nucleotides (12, 34). To test whether Vpr could affect NER, we measured UDS in UV-irradiated cells in the presence and absence of Vpr. UDS was seen as an increase in [³H]thymidine incorporation following UV irradiation. Mock-infected and VprX-infected cells showed approximately a twofold increase in [³H]thymidine incorporation following UV irradiation (Table 5). No significant difference was observed in cells expressing Vpr. Thus, these results indicate that Vpr does not grossly affect the extent of repair synthesis; however, the fidelity of the repair synthesis cannot be assessed by these experiments.

TABLE 5.

Vpr does not affect UDS following UV irradiation of COS cells^a

Virus stock	[3H]-thymidine incorporated (mean cpm)
	No UV irradiation	UV irradiated
Mock	4,508	8,219
Vpr	4,260	8,933
VprX	4,299	9,675

^aCOS cells (2 × 10⁵) were either mock infected or infected with HIV-1_NL4-3-thyenv(−)/VSV-G (Vpr) or HIV-1_NL4-3-thyvprXenv(−)/VSV-G (VprX). At 16 h postinfection, the cells were exposed to UV light (30 J/m²), after which [³H]thymidine (50 μCi/ml) was added for 2 h prior to harvest. Hydroxyurea (10 mM) was added to the culture medium from 1 h prior to UV irradiation until the time of harvest. The harvested cells were lysed, placed on a filter, and counted for levels of [³H]thymidine incorporation. This experiment was repeated twice with similar results. 

Agents that damage cellular DNA can also decrease shuttle vector mutation frequency.

The similarities between Vpr-mediated G₂ cell cycle arrest and the cell cycle arrest induced by DNA-damaging agents suggest that Vpr triggers a normal cellular checkpoint activated in response to damage of the DNA. These checkpoints are normally activated in order to allow cells to repair damaged DNA prior to further progression through the cell cycle (21, 24, 26). It is conceivable that activation of the checkpoint would also activate DNA repair processes which have high fidelity for repair of damaged DNA. Such a scenario has been observed following the activation of p53, which, among other effects, induces cell cycle arrest at the G₁/S phase and simultaneously activates GADD45, a protein involved in NER (19, 35). An increase in resistance to UV light damage in G₂ versus G₁ has been studied in a synchronized cell population. This difference, however, was attributed to an increased time for repair in G₂ until the beginning of the next S phase rather than a specific activation of DNA repair enzymes (28).

Since Vpr alters the frequency of mutation, we determined whether the arrest of cells by other genotoxic agents might have similar effects on the frequency of mutation. COS cells were treated with agents known to cause double-strand DNA breaks and cell cycle arrest in the G₂ phase of the cell cycle. These were γ-irradiation (40 Gy), cisplatin (300 μM; 1 h), and nitrogen mustard (25 μM; 30 min). Treatment with UV irradiation (30 J/m²) was also included, as it generates a different set of lesions, the majority of which are cyclobutane pyrimidine dimers. Compared to mock-treated samples, all treatments caused a decrease in mutation frequency in the transfected shuttle vector ranging from fourfold to ninefold (Table 6). This data suggests that a common mechanism of action may exist for Vpr and other genotoxic agents in the reduction of mutation frequency. One possibility is that both Vpr and DNA damage activate common checkpoint pathways which alter the frequency of DNA repair.

TABLE 6.

Agents that damage cellular DNA can also decrease shuttle vector mutation frequency^a

Treatment and source	No. of white/blue colonies	% White colonies	P value
Irradiation
Mock	47/758	5.8
γ-Irradiation (4,000 rad)	5/309	1.6	0.0024
Mock	156/3,389	4.40
UV light (30 J/m2)	36/3,601	0.98	<0.0001
Chemical
Mock	105/5,681	1.80
Cisplatin (300 μM)	8/3,756	0.21	<0.0001
Nitrogen mustard (25 μM)	6/2,993	0.20	<0.0001

^aCOS cells (2 × 10⁵) were exposed to various DNA-damaging treatments. The chemical treatments were cisplatin (300 μM; 1 h) and nitrogen mustard (25 μM; 30 min). At 19 h postinfection, the cells were transfected with 4 μg of pBK-CMV plasmid (UV irradiated at 2,500 J/m²) by using Lipofectin. After a further 48 h, the cells were harvested and low-molecular-weight DNA was recovered by Hirt extraction and transformed into E. coli. Blue and white colonies were scored to give mutation frequencies. The values for the different agents were compared with those from the mock experiments, and all were found to vary significantly from the control; the associated P values are given in the final column of the table. This experiment was repeated twice with similar results. 

DISCUSSION

Our results demonstrate that the presence of HIV-1 Vpr acts to decrease the frequency of deletion mutations which occur following introduction of a UV-damaged plasmid into cells. Since Vpr induces cell cycle arrest at a G₂ checkpoint, we determined whether Vpr directly affects the repair of damage to endogenous cell DNA. However, we observed no effect of Vpr upon the rate of either double-strand-break repair or NER of cellular DNA. The effects of Vpr on mutation rates were similar to that observed with other agents which induce cell cycle arrest, such as γ radiation and DNA-damaging chemicals.

It is noteworthy that a previous study (22) reported that Vpr could reduce the mutation rate of an HIV-1 shuttle vector by approximately fourfold. The design of the experiments suggested that the mutations that were observed were a result of errors that occurred during reverse transcription. The mutations which we monitored in our experiments were a result of cellular DNA damage repair processes in which reverse transcription is not implicated. The majority of the mutations we observed, both in the presence and absence of Vpr, were deletions, consistent with previous reports of the major types of mutations occurring following transfection of DNA into cells. The most likely mechanism for deletion formation is a double-strand break in the DNA followed by nuclease activity and religation of the free DNA ends. Vpr could decrease the frequency of mutations by (i) increasing the rate of accurate rejoining of DNA following breakage, (ii) decreasing the rate of rejoining of ends resulting from double-strand breaks (resulting in failure to amplify in COS cells), (iii) increasing the fidelity of nucleotide excision repair of UV-damaged plasmids such that double-strand breaks are less likely to occur, (iv) inhibiting the action of nucleases on the broken free ends of DNA, and (v) enhancing the activity of error-free as opposed to error-prone mechanisms of DNA repair. From our experiments, we cannot distinguish among the above possibilities; however, altering the rate of rejoining of double-strand breaks (hypotheses i and ii above) appears less likely, since we did not observe an effect of Vpr on repair of double-strand breaks induced by γ-irradiation of endogenous cellular DNA. If one assumes that the increase in fidelity in our studies is due to mechanisms similar to those in the studies by Mansky (22), as-yet-unknown processes involved in DNA repair common to cellular DNA repair and reverse transcription may be implicated.

We hypothesize that Vpr acts along pathways normally used by cells to induce cell cycle arrest at a G₂ checkpoint. These checkpoints are ordinarily activated in cells following damage by genotoxic agents, such as γ radiation, which induces double-strand breaks in DNA. Failure to arrest at G₂ and entry into mitosis without repair of double-strand breaks would have catastrophic consequences for the ability of the cell to faithfully duplicate its genetic material in daughter cells. Since it is so critical for a cell to faithfully replicate and segregate DNA, it is conceivable that activation of DNA repair processes which are more efficient or act with greater fidelity occurs at the G₂ checkpoint. In this scenario, the mechanism by which Vpr decreases the frequency of mutations is not a direct effect of Vpr but rather a consequence of the cell cycle arrest checkpoint induced by Vpr. A similar decrease in the frequency of mutations observed in cells arrested by other genotoxic agents is consistent with this idea.

We have proposed previously that one important consequence of Vpr action is the crippling of an effective immune response through arresting division of antigen-activated T cells. The results discussed here suggest the hypothesis that enhancing the fidelity of DNA repair may be another function of Vpr important for HIV-1 replication or pathogenesis. As described by Mansky (22), Vpr may act to increase the fidelity of reverse transcription. We demonstrated previously that Vpr packaged within virions is sufficient to induce cell cycle arrest during the initial infection (31). Thus, another point in the viral life cycle where enhanced fidelity of DNA repair could play a role is during integration. Based upon our results, we hypothesize that Vpr may act to facilitate integration by creating an environment within the cell or in the context of the preintegration complex which facilitates integration. One possibility may be to protect the free ends of unintegrated viral DNA from nucleases, akin to our proposed mechanism for reducing deletions in our transfected shuttle vector. Another possibility may be to enhance the rate of ligation of free viral DNA ends to endonuclease-cleaved cellular DNA. Vpr may therefore promote HIV evasion of possible intracellular defenses against exogenous DNA sequences entering the cell. We can test these possibilities by examining the effect of Vpr on the rate of integration and the structures of preintegration viral DNA intermediates.