HIV-1 Viral Infectivity Factor (Vif) Alters Processive Single-stranded DNA Scanning of the Retroviral Restriction Factor APOBEC3G^*

Protein Expression and Purification

Recombinant baculovirus production for expression of GST-A3G, GST-A3G D128K, GST-Vif_IIIB, GST-Vif_HXB2, or GST-nucleocapsid protein in Sf9 cells was carried out using the transfer vector pAcG2T (BD Biosciences) as described previously (12, 19). Vif_IIIB was cloned from the pcDNA-hVif vector encoding the Vif protein of HIV-1 NL4-3 that was obtained from the AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, National Institutes of Health from Dr. Stephan Bour and Dr. Klaus Strebel (65). Vif_HXB2 was cloned from pHIV-gpt(WT), a gift from Dr. David Kabat (Oregon Health and Science University). Site-directed mutagenesis was used to create the A3G D128K clone. Cloning primers for Vif variants and the site-directed mutagenesis primers were obtained from Integrated DNA Technologies and are listed in supplemental Table S1. Sf9 cells were infected with recombinant virus at a multiplicity of infection of 1, except for GST-Vif_IIIB and GST-Vif_HXB2, which were infected at a multiplicity of infection of 4 or 2, respectively. Recombinant baculovirus-infected Sf9 cells were harvested after 72 h of infection, except for GST-Vif_IIIB and GST-Vif_HXB2, which were harvested after 48 h of infection. Cells were lysed in the presence of RNase A, and the proteins (A3G, A3G D128K, and nucleocapsid protein) were purified as described previously (22) to obtain protein that was cleaved from the GST tag and 95% pure. The Vif variants were eluted from the glutathione-Sepharose resin (GE Healthcare) as described previously (12) with the GST tag to maintain their stability. Eluted GST-Vif variants were dialyzed against 100 mm Tris, pH 7.5, 250 mm NaCl, 10% glycerol, and 1 mm DTT and were 85% pure. Protein fractions were stored at −80 °C. HIV-1 reverse transcriptase p66/p51 (66) was generously provided by Dr. Stuart F. J. Le Grice (NCI, National Institutes of Health).

Model HIV-1 Replication Assay

The model HIV-1 replication assay, which measures A3G-induced mutagenesis of ssDNA during reverse transcription of an RNA template, was performed as described previously (19). A synthetic (+)RNA genome that contains a PPT, 120 nt of the catalytic domain of the HIV-1 protease (prot), and lacZα (248 nt) was synthesized in vitro. The PPT enabled second strand ((+)DNA) synthesis to occur. The lacZα was used as a reporter gene for mutations by blue/white screening. The originating sequence of the HIV-1 protease gene is from a clone obtained through the AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, National Institutes of Health, p93TH253.3 from Dr. Feng Gao and Dr. Beatrice Hahn (67). The RNA template (25 nm) was annealed to a 24-nt DNA primer (19) and incubated with nucleocapsid protein (1.5 μm), RT (1.2 μm), and dNTPs (500 μm) in RT buffer (50 mm Tris, pH 7.4, 40 mm KCl, 10 mm MgCl₂, and 1 mm DTT) in the presence or absence of 100 nm A3G or A3G D128K. A3G, RT, and nucleocapsid protein were added to the reaction at ratios to the RNA that are estimated to be found in virions (8, 57, 68, 69). Reactions containing A3G or A3G D128K were conducted in the presence or absence of 175 nm Vif_IIIB or Vif_HXB2. Synthesized DNA was PCR-amplified using Pfu Turbo C_x (Agilent Technologies) and cloned into a pETBlue-1 vector backbone that allows the experimentally synthesized lacZα to be used for α-complementation (19). Twenty-five mutated clones for each condition tested were analyzed. DNA sequencing was carried out at the National Research Council of Canada (Saskatoon, Saskatchewan). A t test was used for statistical analysis of sequences.

Deamination Assays

The 118-nt ssDNA substrate was obtained from Tri-Link Biotechnologies, and the sequence is listed in supplemental Table S1. A3G and A3G D128K (10-20 nm) were incubated with 100 nm fluorescein (F)-labeled ssDNA and 700 nm Vif (IIIB or HXB2) in RT buffer. The substrate to A3G ratio was kept high to ensure that reactions were carried out under single hit conditions, i.e. <15% substrate consumed (70), where a single A3G could interact with a particular ssDNA substrate once at most. Reactions were incubated at 37 °C for 2.5–40 min. Deaminations were detected by resolving DNA that had been treated with uracil-DNA glycosylase and heated under alkaline conditions on a 10% (v/v) denaturing polyacrylamide gel as described previously (12). Gel pictures were obtained using a Typhoon Trio (GE Healthcare) multipurpose scanner, and analysis of integrated gel band intensities was performed using ImageQuant software (GE Healthcare). Under these conditions, a processivity factor can be determined by comparing the total number of deaminations occurring at two sites on the same DNA substrate with a calculated theoretical value of the expected deaminations that would occur at those two sites if the events were uncorrelated, i.e. not processive (see Ref.12). The specific activity was calculated from single hit condition reactions by determining the picomoles of substrate consumed/minute for a microgram of enzyme. The addition of DNA was used to start the deamination reactions. Prior to the addition of DNA, the reaction components were prewarmed to 37 °C.

Steady State Rotational Anisotropy Assays

Steady state fluorescence depolarization (rotational anisotropy) was used to measure protein-nucleic acid and protein-protein binding affinities using an F-labeled binding partner. A QuantaMaster QM-4 spectrofluorometer (Photon Technology International) with a dual emission channel was used to collect data and calculate anisotropy. Measurements were made at 21 °C. Samples were excited with vertically polarized light at 495 nm (6-nm band pass) and vertical and horizontal emissions were measured at 520 nm (6-nm band pass). Apparent dissociation constants (K_d) were obtained by fitting to a rectangular hyperbola or sigmoidal curve using SigmaPlot 11.2 software.

A3G, A3G D128K, Vif_IIIB, and Vif_HXB2 were tested for their ability to bind the ssDNA substrates 118-nt ssDNA and prot (−)DNA, which are listed in supplemental Table S1 and described in the text. Reactions were 50 μl and contained F-labeled ssDNA (50 nm) in RT buffer, and A3G (0–700 nm) or Vif (0–200 nm) was titrated into the reaction. Alternatively, A3G or D128K was mixed with a Vif variant at an equimolar ratio before titration (0–350 nm) into the reaction.

For protein-protein binding, A3G or A3G D128K was labeled with fluorescein using the Fluorescein-EX Protein Labeling kit (Invitrogen). The F-A3G or F-D128K (50 nm) was used in a reaction mixture (50 μl) that contained RT buffer and a titration of Vif_IIIB or Vif_HXB2 (0–200 nm).

Decreased A3G-induced Mutagenesis in the Presence of Vif

A reconstituted model HIV-1 replication system was used to determine whether Vif could inhibit the deamination ability of A3G. In this assay, we used an RNA construct that contains (from the 5′- to 3′-end) a PPT, 120 nt of the HIV-1 prot active site, and the lacZα gene. The PPT enables second strand synthesis to occur. The prot and lacZα sequences are used to characterize the spectra of mutations. The prot sequence is further used to gauge the ability of A3G-induced deaminations to inactivate the protease enzyme. As such, the sequence is analyzed in the (+)strand orientation where C→U deaminations are detected as G→A mutations. A3G was able to cause an approximately 17-fold increase in the population mutation frequency and an 11-fold increase in the clone mutation frequency as compared with RT alone (compare Table 2 and supplemental Table S2).

A plot of the distribution of A3G-induced mutations recovered from the sequenced clones demonstrates a mutational gradient in the 5′ → 3′ orientation, meaning that deaminations were biased toward the 5′-end of the nascently synthesized cDNA (Fig. 1A). This has been observed previously for A3G in vivo due to HIV-1 replication dynamics in which (−)DNA furthest from the PPT is single-stranded the longest and incurs the most deaminations (7, 10, 11). To determine whether Vif could influence the deamination activity of A3G, we added Vif at excess (∼2-fold more than A3G). It can be seen by visual inspection of the mutation spectra that addition of Vif_IIIB or Vif_HXB2 caused a decrease in A3G-induced mutagenesis (Fig. 1A) with Vif_HXB2 (Fig. 1B) having a greater inhibitory effect than Vif_IIIB (Fig. 1C). The decrease in the A3G-induced mutations when Vif_HXB2 (Fig. 1B) or Vif_IIIB (Fig. 1C) was present was greater in the prot likely because of the shorter time that this region is single-stranded in comparison with the lacZα due to the proximity to the PPT. The clones were also binned for analysis based on the number of mutations per prot-lacZα (Fig. 1, D–F). A3G was able to induce a widespread number of mutations. We found a range of 3–12 mutations per prot-lacZα occurring for A3G (Fig. 1D). The data are far less spread when A3G-induced mutagenesis occurred in the presence of Vif_HXB2 where we found only three to five mutations per prot-lacZα ∼55% of the time and 0–2 mutations per prot-lacZα ∼35% of the time (Fig. 1E). For Vif_IIIB, the reduction in A3G-induced mutations is demonstrated by a shift of mutations down to three to eight mutations per prot-lacZα (Fig. 1F). The data were analyzed further by comparing the frequency of mutations in heavily mutated sites in the presence of A3G and the in absence and presence of Vif (Table 1). From this analysis, we found that Vif_HXB2 caused six of the eight heavily mutated sites to be mutated at a lower level (Table 1, A3G + Vif_HXB2). The detrimental effect of Vif_IIIB on A3G-induced mutagenesis was only significant in the prot region (Table 1, A3G + Vif_IIIB).

The more potent effect of Vif_HXB2 was further evident when the mutation frequencies were examined. Here we scored the number of white colonies in the population (population mutation frequency) or the number of mutations per base pair (clone mutation frequency) (Table 2). We found that Vif_HXB2 caused a significant decrease in both the population (1.4-fold) and clone mutation frequency (2-fold) from A3G alone (Table 2). The influence of Vif_IIIB was not as large with the population mutation frequency remaining the same as for A3G alone and the clone mutation frequency decreasing 1.5-fold (Table 2). Altogether, the data indicate that both Vif variants we tested can inhibit A3G-induced mutagenesis, consistent with previous results that tested Vif_IIIB (59) or Vif_HXB2 (58). Furthermore, the effect of Vif_HXB2 on A3G was more potent than Vif_IIIB, which is consistent with how these Vif variants function with respect to inducing degradation of A3G and other APOBEC3 enzymes (36).

Vif-mediated Inhibition of A3G Can Promote Sublethal Mutagenesis in a Model HIV-1 Replication System

Here we assessed the potential ability of each Vif variant to cause A3G to induce a sublethal amount of mutagenesis in the prot (supplemental Table S3). After determining the protease amino acid sequence of the mutated clones, we can infer whether the protease would retain activity or be inactivated by drawing on data from Loeb et al. (71) in which an extensive mutagenesis of the protease was conducted. Concomitant with the site to site decreases in mutation frequency (Table 1), we saw a decrease in the number of clones with amino acid changes of the protease in the presence of the Vif variants (supplemental Table S3).

When A3G-catalyzed deaminations occurred in the presence of Vif_HXB2, the types of mutations incurred were not changed but rather decreased with most mutations seen with A3G alone being absent in the presence of Vif_HXB2 (supplemental Table S3). In the presence of Vif_IIIB, there were decreases in mutations at nine of the 13 sites mutated for A3G alone and changes in the mutation frequency, i.e. increased mutation or a novel site in four of the 13 sites mutated for A3G alone (supplemental Table S3). For example, the most heavily mutated site for native A3G is at amino acid 42 (Trp to stop; 44%). However, in the presence of Vif_HXB2 and Vif_IIIB, the frequency of this mutation dropped to 0 and 16%, respectively (supplemental Table S3). It is interesting that some of the sites that were mutated only in the presence of Vif_IIIB, albeit at a low level, caused protease inhibitor resistance (4% for D30N and M46I; supplemental Table S3).

To determine whether each individual model provirus synthesized in our assay would code for an active or inactive protease, we analyzed each clone individually (Fig. 2). A3G-induced mutagenesis was able to cause inactivation of the prot 60% of the time and left the prot active only 8% of the time (Fig. 2A). A significant amount of clones (32%) had no mutations in the prot region. A3G-induced mutagenesis in the presence of Vif_HXB2 changed this distribution so that 68% of clones had no mutations in the prot and 24% had mutations that inactivated the prot (Fig. 2B). Again, 8% of clones were mutated but still coded for an active prot (Fig. 2B). The same trend was found for A3G in the presence of Vif_IIIB where slightly more clones were not mutated (40%; Fig. 2C) in comparison with A3G alone (32%; Fig. 2A). These results are in agreement with the analysis of the total population of mutations (supplemental Table S3) where there was an overall decrease in mutations rather than an alteration of the mutational hot spots.

Vif Alters the Processive Scanning Mechanism of A3G

To investigate how Vif_IIIB and Vif_HXB2 inhibit A3G-induced mutagenesis and why these variants differ in their effect (Fig. 1 and Tables 1 and 2), we conducted assays measuring A3G-catalyzed deaminations on synthetic DNA substrates. Specifically, the assay can measure the processive scanning behavior of A3G by using a 118-nt ssDNA substrate with two 5′-CCC deamination motifs spaced 61 nt apart (Fig. 3A, schematic). Fig. 3A shows a characteristic result of A3G in this assay. A3G deaminated both the 5′- and 3′-proximal CCC motifs, but there was a 2-fold preference for the 5′-motif due to an inherent catalytic orientation specificity of the active site (22). Of importance is the presence of a double deamination band, which indicates that a deamination occurred at both the 5′-C and 3′-C on the same ssDNA (Fig. 3A, 63 nt). The processivity factor is calculated as a ratio of the intensity of this band to the calculated expected value of deaminations that would occur at both 5′- and 3′-proximal motifs if the enzyme were not processive (see “Experimental Procedures”). The processivity factor of 9.1 (Fig. 3A, below gel) means that A3G is 9 times more likely to undergo a processive deamination of both CCC motifs than to deaminate each of the motifs in separate enzyme-ssDNA encounters. Vif_HXB2 and Vif_IIIB were able to decrease the processivity of A3G by at least 1.5-fold from 9.1 to 4.1 or 5.6, respectively (Fig. 3A).

To discern how the Vif variants are able to decrease the processivity of A3G, we used an assay that can make a distinction between the two scanning modes of A3G. A3G has been shown to scan ssDNA by sliding and microscopic dissociations and reassociations termed jumping (12, 19). This bifunctional processive mode was identified using an experiment where a complementary DNA oligonucleotide was annealed between two CCC motifs on an ssDNA substrate (Fig. 3B, schematic). The partially dsDNA to which the enzyme cannot effectively bind acts as a block to the sliding component of A3G processivity (12, 72). However, A3G can still transverse the block by jumping. As such, A3G incurred a ∼2-fold decrease in processivity (Fig. 3, compare A and B, processivity factors of 9.1 and 4.5, respectively). When Vif_HXB2 was added to the reaction with the partially dsDNA substrate, the processivity factor of A3G decreased another 2-fold from 4.1 to 1.1 (Fig. 3, A and B, Processivity factor), and there was a visible decrease of the double deaminations to almost none (Fig. 3B, 63 nt). This indicates that Vif_HXB2 can inhibit the jumping component of A3G so that it is no longer able to transverse the dsDNA. A3G was still able to slide because processive deaminations occurred on fully ssDNA (Fig. 3A). Vif_IIIB did not affect the ability of A3G to jump because the processivity factor in the presence of the dsDNA is 5.5 (Fig. 3B, Processivity factor), which is similar to the processivity of A3G in the presence of Vif_IIIB on fully ssDNA (processivity factor of 5.6; Fig. 3A). However, Vif_IIIB does appear to affect A3G processivity because it decreased the processivity factor on ssDNA 1.5-fold (processivity factor decreased from 9.1 to 5.6; Fig. 3A), suggesting that Vif_IIIB may inhibit A3G sliding. Similar results were found when we used a complimentary RNA oligonucleotide as a block (Fig. 3C). An RNA/DNA hybrid block would be what A3G encounters on HIV-1 (−)DNA during viral replication and suggests that the decrease in A3G-induced mutagenesis seen in the HIV-1 replication assay (Fig. 1) was due to a partial inhibition of A3G processivity by the Vif variants (Fig. 3, A–C).

To further test this hypothesis, we examined the sequences of clones from the HIV-1 replication assay. We can analyze the spatial proximity of mutations in the clones and relate it to processivity where clustered mutations are indicative of scanning by local sliding and distantly spaced singleton mutations are indicative of scanning by jumping (19). Areas of the 368-nt (+)DNA sequence (containing prot and lacZα), which contains multiple G residues, were used to analyze the frequency of clustered A3G-induced G→A mutations. If the sliding component of A3G ssDNA scanning mechanism is retained, A3G should be able to induce mutagenesis at multiple G residues in a close region, i.e. 10 nt or less. The mean clustered mutation frequency for A3G alone (0.48; Fig. 4) is similar to that for the situation when Vif_HXB2 was present (0.33; p value, 0.32; Fig. 4). These data are consistent with Fig. 3, A–C, which shows that A3G retained the ability to slide, but not jump, in the presence of Vif_HXB2. In the presence of Vif_IIIB, A3G was less capable of inducing mutations that are closely spaced (0.14; p value, 0.02; Fig. 4). This indicates that A3G was less able to scan ssDNA by sliding when Vif_IIIB was present and is in agreement with processivity data obtained from synthetic substrates (Fig. 3, A–C).

Inhibitory Mechanism of Vif Is Mediated by an Interaction with A3G

A3G is known to physically interact with Vif (30–32), which suggests that Vif could alter the processivity of A3G through a protein-protein interaction. That the two Vif variants affected A3G processivity differently supports this notion (Figs. 3 and 4). To investigate this possibility, we used the A3G D128K mutant (referred to as D128K), which is insensitive to Vif-mediated degradation and has been shown to have a disrupted physical interaction with Vif (42–44, 73–75). Our rotational anisotropy data confirm the previous findings as we found no detectable interaction between F-D128K and the Vif variants (Table 3). However, we found that Vif_HXB2 and Vif_IIIB interacted similarly with F-A3G. The measured apparent K_d values of 90 nm (Vif_HXB2) and 78 nm (Vif_IIIB) for F-A3G indicate that a strong interaction can occur between these proteins (Table 3).

We tested the ability of D128K to induce mutations in a model HIV-1 replication system in the absence and presence of the Vif variants. The mutational spectrum for D128K alone shows peaks for highly mutated sites in the same locations as for A3G (compare Fig. 1A and Fig. 5A). The intensity of the mutations among the sites differed slightly likely because of the stochastic nature of A3G deamination (18). The mutation frequencies are nearly identical for D128K and A3G at the population level (D128K, 0.96 (Table 4); A3G, 0.87 (Table 2)) and clone level (D128K, 2.2 × 10⁻² (Table 4); A3G, 2.2 × 10⁻² (Table 2)). Overall, the results are consistent with D128K behaving similarly to A3G in this assay. When both Vif_IIIB and Vif _HXB2 were added to the reaction with D128K, we observed no decrease in the mutagenic ability of D128K (Fig. 5, A–F, and Table 4) in contrast to A3G (Fig. 1, A–F, and Table 2). In addition, we found no effects on D128K processive scanning by sliding by the addition of Vif_HXB2 or Vif_IIIB to the model HIV-1 replication assay as the mutation cluster frequency (Fig. 6) and frequency of mutations (Fig. 5 and Table 4) remained constant in all three experimental conditions. This observation is in agreement with data obtained on the 118-nt substrate where neither Vif_HXB2 nor Vif_IIIB was found to change D128K processivity (data not shown). Altogether, the data support a model in which an alteration in the processivity of A3G in the presence of Vif is caused by a Vif-A3G interaction.

Vif Affects the Specific Activity of A3G and D128K

Despite evidence that a Vif-A3G interaction was responsible for the observed decrease in A3G-induced mutagenesis (Table 3 and compare Figs. 5 and 1), Vif has been shown to bind DNA and RNA molecules (54, 76, 77). Therefore, we assessed whether the Vif variants could additionally alter A3G deamination activity, i.e. decrease specific activity, by binding to the ssDNA substrate. Rotational anisotropy data show that A3G could bind to the substrate used in the processivity assays (Fig. 3A, 118-nt substrate) with an apparent K_d of 203 nm (Table 3), which is ∼3-fold higher than the apparent K_d values of the Vif variants for this substrate (K_d of Vif_HXB2, 67 nm; K_d of Vif_IIIB, 79 nm; Table 3). We also examined the binding of A3G and Vif to the prot (−)DNA, which was part of the deamination substrate for the HIV-1 replication assay (Fig. 1). We observed A3G binding to the prot (−)DNA with an apparent K_d of 270 nm (Table 3), which is 10-fold (K_d of Vif_HXB2, 27 nm; Table 3) and 3-fold (K_d of Vif_IIIB, 92 nm; Table 3) higher than the binding of Vif variants to this substrate. Therefore, in both our experimental systems, the apparent K_d values of Vif for DNA and F-A3G (K_d of F-A3G from Vif_HXB2 and Vif_IIIB, 90 and 78 nm, respectively; Table 3) are similar, suggesting the possibility that Vif could compete with A3G for the ssDNA substrate and decrease the specific activity of A3G.

To test this hypothesis, we used D128K, which does not interact with Vif (Table 3). There was no effect of Vif on D128K-induced mutagenesis (Fig. 5), suggesting that Vif interactions with the DNA in the HIV-1 replication assay are inconsequential with regard to processivity (Fig. 6). Notwithstanding, a decrease in specific activity could be occurring but cannot be observed in the HIV-1 replication assay because deamination events are dependent on and limited by other factors such as the rates of RNA degradation and (+)DNA synthesis. To determine whether Vif can decrease the specific activity of A3G or D128K, we used the 118-nt ssDNA substrate used previously for the processive deamination assay (Fig. 3A). The experiment involved preincubation of Vif and D128K or A3G together, and then the reaction was started by the addition of the 118-nt ssDNA substrate. On the 118-nt ssDNA substrate, D128K had a 4-fold higher specific activity (40 pmol/μg/min; Table 5) than A3G (10 pmol/μg/min; Table 5). However, Vif (HXB2 and IIIB) can decrease D128K specific activity 80-fold (D128K + Vif_HXB2, 0.5 pmol/μg/min; Table 5) or 50-fold (D128K + Vif_IIIB,0.8 pmol/μg/min; Table 5) in comparison with an 8-fold (A3G + Vif_HXB2, 1.2 pmol/μg/min; Table 5) or 6-fold (A3G + Vif_IIIB, 1.8 pmol/μg/min; Table 5) decrease in A3G activity. These data suggest that there was a significant amount of Vif binding to the 118-nt ssDNA substrate in the presence of D128K, causing a large decrease in specific activity. To investigate this further, we determined the apparent K_d of a D128K and Vif_HXB2 mixture and found that the value (K_d of 72 nm; Table 3) was almost identical to that of Vif_HXB2 alone (K_d of 67 nm; Table 3). These data support the hypothesis that Vif has a dominant role in ssDNA binding in the presence of D128K that had an ∼3-fold higher apparent K_d than Vif (K_d of 183 nm; Table 3). Similar results were found with a D128K and Vif_IIIB mixture binding to the 118-nt ssDNA substrate (K_d of 51 nm; Table 3). In contrast, an A3G and Vif_HXB2 or A3G and Vif_IIIB complex had an apparent K_d (A3G and Vif_HXB2, K_d of 158 nm; A3G and Vif_IIIB, K_d of 135 nm; Table 3) that was higher than that of Vif_HXB2 or Vif_IIIB alone (Vif_HXB2, K_d of 67 nm; Vif_IIIB, K_d of 79 nm; Table 3). Based on these data, it appears that Vif molecules that interact with A3G may not be able to bind the ssDNA substrate efficiently through the Vif DNA binding domain because it is near the A3G interaction residues (78). This would lower the effective DNA binding concentration of Vif and result in an increase in the apparent K_d value (Table 3). However, the Vif variants may still contact the ssDNA because the apparent K_d values of the Vif·A3G complex are slightly less than that of A3G alone (K_d of 203 nm; Table 3).

In contrast to A3G, which had a decrease in processivity in the presence of Vif (Figs. 3 and 4), we did not observe a decrease in D128K processivity on the 118-nt substrate (data not shown) or in the HIV-1 replication assay (Figs. 5 and 6) in the presence of the Vif variants despite the decrease in specific activity (Table 5). Together with binding data (Table 3), the results indicate that Vif can bind ssDNA and compete with D128K for substrate access, which results in a decrease in D128K specific activity (Table 5). This activity was lessened in the presence of A3G (Table 5), which bound Vif with high affinity (Table 3). Therefore, the data support a model in which the inhibition of processivity and specific activity by Vif are two different functions mediated by an A3G-Vif or ssDNA-Vif interaction, respectively.