The Activity Spectrum of Vif from Multiple HIV-1 Subtypes against APOBEC3G, APOBEC3F, and APOBEC3H

Cloning.

Twelve viral isolates and five molecular clones from different subtypes were obtained through the NIH AIDS Research and Reference Reagent Program (Table 1). Briefly, viral RNA was extracted from the viral stocks using a Qiagen QIAamp Viral RNA Mini Kit, followed by reverse transcription and amplification using primers located outside the Vif coding region. The PCR fragments were cloned into pSC-A (StrataClone PCR cloning vector; Stratagene), and several clones were sequenced and manually aligned to the published sequences. Clones with Vif sequences identical to the published sequences were used as templates for PCR amplification with Vif primers specific for the 5′ and 3′ regions of each variant and containing NotI and EcoRI restriction sites. PCR fragments were digested with NotI/EcoRI, and the coding regions of the subcloned 17 Vif variants were inserted in pCRV1 as described previously (37, 51). Site-directed mutagenesis of subtype F1 and F2 Vifs was performed using overlapping PCR as described previously (37). The mutated Vifs were cloned into pCRV1 vector and confirmed by sequencing. Primer sequences are available upon request.

Amino-terminally FLAG-tagged A3G, A3F, and A3H hapII in PTR600 expression plasmids have been described previously (10, 28, 30). A carboxyl-terminal triple hemagglutinin (HA) tag was added to A3G and A3H hapII by overlapping PCR, followed by cloning into PTR600. An untagged version of A3H hapII was cloned into PTR600. All DNA preparations were sequenced to confirm the integrity of the APOBEC3 sequences. NCBI reference sequence numbers for the APOBEC3 used are NM 021822.3 (A3G), NM 145298.5 (A3F), and FJ376614.1 (A3H hapII, splice variant SV-183).

Cell culture.

HEK 293T and TZM-bl cells were grown in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 100 U/ml penicillin/streptomycin and 10% fetal bovine serum (FBS). TZM-bl cells were provided by J. C. Kappes and X. Wu through the AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, National Institutes of Health, NIH Reagent program.

Infectivity assay.

To determine the level of infectivity rescue achieved by each Vif variant, viral stocks were produced in the presence and absence of APOBEC3. Briefly, 3.0 × 10⁵ HEK 293T cells were cotransfected with 500 ng of pNL4-3ΔVif, 5 ng of Vif, and 50 ng of PTR600FLAG-A3G, 50 ng of PTR600FLAG-A3F, or 25 ng of PTR600FLAG-A3H expression plasmids, using 4 μg/ml polyethylenimine (PEI; Polysciences, Inc.). The total amount of DNA transfected was 1,000 ng, with pcDNA3.1 serving as filler. Viral supernatants were collected at 44 to 48 h posttransfection, clarified by centrifugation, and stored at −80°C. Forty-microliter supernatants were used to infect 1 × 10⁴ TZM-bl cells in black 96-well plates. Infections were done in triplicate with viral supernatants from three independent experiments for each APOBEC3 protein tested. Infectivity of the virus particles in the TZM-bl cells was assessed at 44 to 48 h postinfection using a Galacto-Star System for detecting β-galactosidase activity (Applied Biosystems), as described previously (10, 37).

Vif expression.

A total of 3.0 × 10⁵ HEK 293T cells were transfected with 200 ng of Vif expression plasmids, 100 ng of PTR600-green fluorescent protein (GFP), and 700 ng of pcDNA (24-well plate; the total amount of DNA transfected was 1,000 ng with 4 μg/ml PEI). SDS lysis buffer (1% SDS, 50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 5 mM EDTA) was added to the transfected cells 48 h later, and the clarified lysates were analyzed by Western blotting.

Degradation assay.

A total of 3.0 × 10⁵ HEK 293T cells in 24-well plates were cotransfected with 500 ng of pNL4-3ΔVif, 100 ng of PTR600-GFP, 20 ng of pCRV1-Vif, and 100 ng of PTR600FLAG-A3G, 100 ng of PTR600FLAG-A3F, or 50 ng of PTR600FLAG-A3H hapII expression plasmids (total amount of DNA transfected, 1,000 ng; filler DNA, pcDNA3.1).

The degradation of tagged and untagged A3H hapII was compared by cotransfecting HEK 293T cells as described above with the following expression plasmids: PTR600FLAG-A3H hapII (50 ng) and pCRV1-Vif (50 ng) or PTR600-A3H hapII (untagged; 100 ng) with pCRV1-Vif (100 ng). The total amount of DNA transfected was 1,000 ng (filler DNA, pcDNA). Transfections were performed with PEI (4 μg/ml). At 48 h posttransfection cells were lysed (in lysis buffer consisting of 1% SDS, 50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 5 mM EDTA), clarified by centrifugation, and analyzed by Western blotting.

Western blotting.

Transfected HEK 293T cell lysates were separated on 10% bis-Tris 26-well gels (Invitrogen) and transferred by semidry transfer onto polyvinylidene difluoride (PVDF) membranes (Thermo Scientific). Membranes were blocked in 1% (wt/vol) milk solution for 1 h and incubated in primary antibody overnight. HIV-1 Vif was detected with a rabbit anti-Vif primary antibody (NIH catalog number 2221) at 1:1,000 dilution and with goat anti-rabbit secondary antibody (1:5,000; Sigma). FLAG-tagged APOBEC3 proteins were detected with a mouse anti-FLAG primary antibody (1:5,000; Sigma), GFP was detected with rabbit anti-GFP primary antibody (1:2,000; Santa Cruz), and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was detected with a mouse anti-GAPDH primary antibody (1:2,000; Santa Cruz). Goat anti-rabbit or goat anti-mouse secondary antibody (both, Sigma) was at a 1:5,000 dilution. Membranes were washed in wash buffer (0.1% Tween in phosphate-buffered saline [PBS]) and incubated in secondary antibody for 2 h. Membranes were developed with SuperSignal West Pico or Femto substrate (Thermo Scientific) in an LAS-3000 imaging system (FujiFilm). Only nonsaturated signals were acquired for further analysis with ImageGauge, version 4.0, software.

Vif-A3H coimmunoprecipitation.

A total of 7 × 10⁵ HEK 293T cells were transfected with pCRV1-Vif expression plasmids (200 ng), PTR600HA-A3H hapII (800 ng), PTR600HA-A3G (800 ng), or PTR600-GFP (800 ng) using Fugene 6 according to the manufacturer's instructions. At 24 h after transfection, clasto-lactacystin β-lactone (10 μM; Sigma) was added to the transfection to prevent potential Vif-mediated A3H degradation. At 48 h after transfection, cells were washed with PBS and lysed on ice for 15 min in lysis buffer (1% Triton X-100 in PBS supplemented with EDTA-free protease inhibitor cocktail [Roche]). Proteins were clarified by centrifugation at 4°C (8,200 × g for 10 min at 4°C) and incubated with anti-HA-coated beads (Easyview; Sigma) for 1 h at 4°C. Beads were extensively washed with lysis buffer supplemented with 75 mM NaCl. Bound proteins were eluted by boiling the beads in sample loading buffer. Cell lysates and bound proteins were analyzed by Western blotting.

A3H polyclonal serum.

Two rabbits were immunized three times with the A3H peptide Gln-Phe-Asn-Asn-Lys-Arg-Arg-Leu-Arg-Arg-Pro-Tyr-Tyr-Pro-Arg (A3H residues 12 to 26). Custom antibody production was performed by Pocono Rabbit Farm and Laboratory, Inc. Rabbit polyclonal serum was used at a 1:2,000 dilution to detect transfected untagged and tagged A3H.

Statistical analysis.

GraphPad Prism, version 5, software was used to perform all statistical tests (minimum, maximum, mean, t tests, and Spearman rank correlation). P values are two sided, and values of <0.05 were considered to be significant.

Genotype and expression of selected Vif alleles.

Inspection of >1,500 Vif alleles deposited at the Los Alamos HIV sequence database revealed considerable variation in Vif sequences. Some subtype-specific motifs such as 8L (subtype C), 45DCXH48 (subtype D), and 91QRK93 (subtype AE) were also identified. We selected 12 viral isolates from patients representing seven HIV-1 subtypes and five replication-competent full-length molecular clones from three different subtypes. The subtype-specific motifs listed above were present in the selected Vif alleles. The patients from whom the viruses were originally isolated lived in various countries around the world, like Brazil, Rwanda, Zimbabwe, and Thailand (Table 1). Analysis of the phylogenic relationships of the selected Vif alleles and their corresponding consensus sequences confirmed their subtypes (Fig. 1A).

At the protein sequence level, the 17 Vif alleles displayed, on average, 15% diversity relative to HIV subtype B molecular clone LAI (maximum, 19%; minimum, 7%). The greatest number of changes were observed for Vif G1 (37/192 residues), while Vif B3 (29/192 residues) was most similar to LAI Vif. Inspection of the N-terminal region revealed polymorphic positions both within and outside the domains known to be relevant for APOBEC3-specific interactions while the HCCH Cullin 5-binding motif and the SLQYLA Elongin C-binding motif were conserved in all 17 Vif alleles (see Fig. S1 in the supplemental material for Vif sequence alignments).

The majority of Vif variants were similarly expressed upon transfection in 293T cells (Fig. 1B and C), with the exception of the subtype C Vif C1, which consistently displayed a 2-fold decrease in signal intensity (Fig. 1C). This difference may reflect lower Vif protein expression or reduced antibody recognition by the polyclonal anti-Vif rabbit serum. All 17 Vif variants encode 192 residues, but some Vif bands migrated differently during gel electrophoresis, likely reflecting variations in protein charges (Fig. 1B, compare Vif F1 to Vif F3 or Vif G1).

Anti-APOBEC3 phenotypes of Vif alleles.

The different Vif proteins were analyzed for rescue of infectivity of NL4-3 with a deletion of Vif (pNL4-3ΔVif) in the presence of A3G, A3F, and A3H hapII. In addition, APOBEC3 degradation by the Vif panel was assessed by cotransfecting the Vifs and APOBEC3, followed by Western blot analysis. Titrations of Vif expression plasmids (5 ng, 100 ng, and 500 ng) and APOBEC3 plasmids (25 ng, 50 ng, and 100 ng) in the presence of NL4-3ΔVif were performed to select assay conditions and plasmid ratios that allowed optimal discrimination of both an increase and a decrease in rescue of infectivity (data not shown). Five nanograms of Vif and 50 ng of A3G or 50 ng of A3F were used in the experiments comparing the anti-A3G/A3F activities of different Vif variants. In addition, similar experiments were performed with Vif-deleted HIV-1 molecular clones pLAI, pMJ4 (subtype C, Vif C3), and p94UG114 (subtype D, Vif D1) to exclude potential isolate and subtype incompatibilities (data not shown). Of note, Capsid p24 production upon transfection was comparable for all viral stocks (data not shown).

Most HIV-1 Vif alleles counteract A3G and A3F restriction.

The infectivity of viral stocks was quantified 48 h posttransfection using TZM-bl reporter cells as described previously (10, 28, 37). An empty pCRV1 plasmid and the NL4-3 Vif mutant C133S, which abrogates cullin 5 binding (C133S), were used as negative controls. Infectivity of NL4-3ΔVif produced in the absence of both APOBEC3 and Vif was set to 100% relative infectivity (Fig. 2A and C; see also Fig. 3A). We chose the widely used molecular clone LAI Vif variant as a reference for comparison to other Vifs (Fig. 1A), and the differences relative to LAI Vif were calculated using a t test (Prism GraphPad).

In the absence of Vif or in the presence of the Vif C133S mutant, A3G strongly reduced infectivity to 1%. The reference LAI Vif rescued infectivity from A3G restriction to 60% ± 5%, and the majority of Vif alleles tested were similarly or more active than LAI Vif (Fig. 2A).

Four Vif variants (NL4-3, B2, C1, and G1) displayed low to moderate activities against A3G (relative infectivity, 35% ± 4.5%, 36% ± 2.1%, 10.3% ± 3.5%, and 35% ± 2.5%, respectively), and only one Vif allele (B3) was fully defective against A3G (relative infectivity, 2.7% ± 1.5%). Interestingly, Vif C1, C2, and C3 were expressed to lower levels than the other Vifs, and correcting infectivity data with the Vif expression levels shown in Fig. 2A and B demonstrated that all subtype C Vifs were relatively more active against A3G.

The ability of the Vifs to degrade the APOBEC3s was determined by transfecting HEK 293T cells with 20 ng of the different Vif expression plasmids and 100 ng of A3G (Fig. 2B) or A3F (Fig. 2D). The rescue of infectivity correlated with the ability to degrade A3G in the producer cell, with the exception of Vif C1, which degraded A3G efficiently but failed to rescue viral infectivity to the same degree.

A3F is less potent than A3G in restricting NL4-3ΔVif complemented with an empty control plasmid or with the mutant Vif C133S (no-Vif and Vif C133S, 10% compared to 1% with A3G). LAI Vif rescued infectivity in the presence of A3F to 64% ± 5%, with all Vif alleles displaying activity against A3F. Vif variants AE1, F2, and F3 were more active than LAI Vif (relative infectivity, 73% ± 8.1%, 85% ± 8.5%, and 71% ± 6.6%, respectively). Vif C1, which displayed low activity against A3G, showed average activity against A3F (relative infectivity, 50% ± 2.0%). Interestingly, Vif B3, which was not active against A3G, displayed some activity against A3F (relative infectivity, 26% ± 3.1%) although its ability to degrade A3F was modest (Fig. 2D). Overall, most Vifs were able to degrade and counteract A3G and A3F restriction.

Most non-subtype B HIV-1 Vif alleles counteract A3H hapII restriction.

More than eight naturally occurring A3H haplotypes have been described, with A3H hapII having potent anti-HIV activity (10, 29, 43). There remains some uncertainty about the extent to which A3H is partially or completely resistant to Vif neutralization (5, 10, 29, 41, 43, 52). We therefore tested whether the different Vif alleles have activity against A3H hapII (Fig. 3A). Based on initial optimization experiments (data not shown), we used less APOBEC3 expression plasmid for the experiments with A3H hapII (25 ng of A3H hapII compared to 50 ng of A3G or A3F). Under these experimental conditions, A3H hapII reduces infectivity to 10% ± 1.0% in the absence of Vif or with the mutant Vif C133S.

Interestingly, 9 out of 17 Vif variants were able to rescue infectivity in the presence of A3H hapII. Of the active Vifs, Vif F1 and F3 were, by far, the most efficient at counteracting A3H hapII (F1, 80% ± 7.5%; F3, 77% ± 8.0%) (Fig. 3A), followed by LAI, two subtype C Vif variants (C1 and C3), and one subtype A Vif variant (A1). Of note, Vif C1, which had low activity against A3G and modest activity against A3F, was active against A3H hapII (35% ± 2.0%). Three out of four subtype B variants failed to counteract A3H hapII restriction, with LAI Vif being the sole exception. LAI Vif rescued viral infectivity in the presence of A3H hapII to 43% ± 6.1%, whereas NL4-3 Vif was defective in rescuing A3H-mediated restriction (11% ± 2.6%). This observation is in good agreement with reports showing that Vif from LAI but not NL4-3 is active against A3H hapII (18, 19).

The ability of Vif variants to degrade the A3H hapII was determined by transfecting HEK 293T cells with 20 ng of the different Vif expression plasmids and 50 ng of A3H (Fig. 3B). Overall, A3H hapII was more resistant to degradation than A3G and A3F (Fig. 3B, compare the no-Vif and Vif C133S control levels to the level of A3H in the presence of other Vifs). The Vif variants LAI, F1, and F2 degraded A3H hapII efficiently (Fig. 3C, e.g., only 20% of the “no Vif” A3H hapII control remaining), whereas the other Vif variants were partially or completely defective with respect to degradation in the producer cell (Fig. 3B and C). The degree of Vif-mediated rescue of viral infectivity in the presence of A3H hapII correlated well with the efficiency to degrade A3H hapII (Fig. 3D).

Adding epitope tags to A3H hapII has been reported to influence Vif-mediated A3H hapII degradation (18). To test whether the N-terminal FLAG tag to A3H hapII affects its susceptibility to Vif, we analyzed 10 Vif alleles with N-terminal FLAG tags and untagged A3H hapII for degradation. We transfected the selected Vif variants and the two A3H expression plasmids into HEK 293T cells and used polyclonal rabbit serum raised against an A3H peptide to detect both A3H hapII variants in the cell lysates by Western blotting (Fig. 3E). The A3H hapII degradation patterns were similar for the N-terminally tagged and untagged A3H hapII proteins (Fig. 3E), indicating that the N-terminal FLAG tag did not affect its sensitivity to Vif degradation.

Taken together, in contrast to the efficient activity of most of the Vifs against A3G and A3F, Vifs from different subtypes displayed considerable differences in counteracting A3H hapII restriction.

The spectrum of anti-APOBEC3 activities varies between Vif variants.

Since multiple APOBEC3 proteins are expressed in HIV-1 susceptible cells, we were interested in a comprehensive assessment of Vif-mediated anti-APOBEC3 activities. Figure 4 provides a heat map representation of the anti-APOBEC3 spectrum of each Vif allele (complete rescue indicated in green and maximum restriction shown in red). Vif variants F1 and F3 showed high activity against all three deaminases tested (broad-spectrum anti-APOBEC3 activity), while Vif B3 displayed low activity against all three APOBEC3s. Interestingly, we also noted A3-specific Vif defects in which one (Vif C1 and Vif F2) or two (Vif C2) specific APOBEC3s were not neutralized. Overall, the patterns of rescue from A3G and A3F were similar among Vifs but did not correlate with the activities of these Vifs against A3H hapII. This indicates that residues in Vif that interact with A3G and A3F are likely different from those that interact with A3H hapII.

Identification of the Vif residues required for A3H hapII degradation.

Subtype F Vif variants were efficient in counteracting A3G and A3F but differed in their abilities to block A3H hapII (Fig. 4). This apparent difference between F1 and F2 with respect to A3H hapII was used to accurately determine the Vif requirements for interacting with A3H hapII. We focused on the N-terminal part of Vif because this region has been previously shown to interact with A3G and A3F. Since Vif F1 and Vif F2 differ at only five residues in the N-terminal region (positions 39, 48, 61, 62, and 63) (Fig. 5A), we generated a series of mutants in which residue 39, residue 48, or residues 61 to 63 were exchanged between the two Vifs (Fig. 5A). We tested the infectivity of NL4-3ΔVif with these Vif mutants in the presence and absence of A3H hapII (Fig. 5B). Infectivity data showed that two residues, 39F and 48H, are responsible for the difference between Vif F1 and Vif F2 and that both are required for efficient anti-A3H hapII activity. Replacing either the phenylalanine (F) at position 39 with a serine (39S) or the histidine (H) at position 48 with an asparagine (48N) rendered Vif F1 inactive against A3H hapII, while the introduction of both residues (F39 and H48) was required to render Vif F2 active against A3H hapII (Fig. 5B). The ability of these mutants to degrade A3H hapII was also assessed (Fig. 5C) and showed that the ability to rescue infectivity correlated with A3H hapII degradation (Fig. 5B and C). Interestingly, mutations at these two positions affected only the activity against A3H hapII, whereas the ability to counteract A3G and A3F was unaffected (data not shown).

Before Vif can mediate the degradation of APOBEC3 by the proteasome, both proteins must first physically interact. We speculated, therefore, that the Vif proteins that failed to counteract A3H hapII might be inefficiently interacting with A3H hapII. We performed coimmunoprecipitation experiments by transfecting Vif F1 and Vif F2 with A3H hapII or A3G in HEK 293T cells. The proteasome inhibitor clasto-lactacystin β-lactone was added to the medium 24 h before lysis to prevent APOBEC3 degradation. Vif F1 was more efficient in precipitating A3H hapII than Vif F2, whereas both Vif F1 and F2 associated efficiently with A3G (Fig. 5D). In addition, we established that the two residues were necessary and sufficient to determine the ability of Vif F variants to associate with A3H hapII by cotransfected Vifs F1 and F2 as well as mutant Vifs M4 and M3 with A3H hapII in HEK 293T cells. Vif F1 and mutant M3 (with residues 39F and 48H) immunoprecipitated with A3H hapII, while defective Vifs F2 and mutant M4 (with 39S and 48N) both failed to associate with A3H hapII. This indicates that a phenylalanine at position 39 and a histidine at position 48 determine the association between subtype F Vifs and A3H hapII (Fig. 5E). These results correlated well with the observed infectivity and degradation phenotypes (Fig. 5B and C).

Alanine scanning of residues 38 to 49 in the amino-terminal region of subtype F Vif.

In addition to the polymorphic nature of the amino acids at position 39 and 48, changes within the region between these two residues may affect the interaction between Vif and APOBEC3. Indeed, several motifs in this region have been described to affect subtype B Vif function against A3G (e.g., ⁴⁰YRHHY⁴⁴, 45E, and 48N) (34, 37). We individually replaced all amino acids between positions 38 and 49 in Vif F1 with alanines and analyzed their effects on rescue of infectivity in the presence of A3G, A3F, and A3H hapII. The alanine scanning of this Vif F1 region showed that distinct positions had differential effects on activity against A3G, A3F, and A3H hapII (Fig. 6). Alanines at position 38, 48, and 49 rendered Vif F1 defective against all three APOBEC3s (Fig. 6A, B, C, and E) but did not affect Vif expression (Fig. 6D). None of the other mutations reduced neutralization of A3F, while other mutations showed APOBEC3-specific effects: 40A completely abrogated anti-A3G and anti-A3H hapII activity, and several substitutions in this region resulted in a specific loss of function against A3H hapII (41A, 40A, and 45A) (Fig. 6C). Position 39, one of the two positions that determined the difference between Vif F1 and F2 with regard to A3H degradation, was more tolerant to being mutated to an alanine (from an aromatic phenylalanine) than to the nucleophilic serine. Western blot analysis showed that the efficiency in rescue of infectivity correlates with the level of APOBEC3 degradation by the respective Vifs. Taking these results together, alanine scanning mutagenesis of the region between residues 39 and 48 in subtype F Vif suggests that the residues necessary for A3H hapII recognition are distinct from those required for Vif-A3G or Vif-A3F interactions.

Next, we analyzed the sequences between positions 38 and 49 of the initial subtype set of Vifs to understand why only some Vifs counteract A3H hapII. Figure 7 shows an alignment of the protein region of interest (positions 38 to 49) of the 17 Vif proteins tested and their respective activities against A3H hapII. Nearly all Vifs that efficiently counteracted A3H hapII carried the 39F and 48H combination, whereas most of the inactive Vifs carried alternative pairs. Only a few exceptions existed: Vif B2 contains the FH pair but lacks anti-A3H activity, which could be caused by the unique glycine (G) at position 41 (41G, all other active Vifs carry 41R). This possibility is further supported by the alanine scanning results, which showed the importance of position 41 for anti-A3H function (Fig. 6C). Vif D2, which carries an LH pair instead of the FH pair, displayed low activity against A3H hapII. The leucine (L) at position 39 shares the same hydrophobic and nonpolar properties as the phenylalanine (F), indicating that some variation at position 39 is tolerated with amino acids of similar properties, like 39A (Fig. 6C).

In summary, a 39F-48H or a 39L-48H pair is required for activity against A3H hapII but not for neutralization of A3G and A3F. In addition, amino acid changes at critical residues between these positions (e.g., 40, 41, and 45) may abrogate anti-A3H hapII function in the presence of the FH/LH pairs.