Determinants of Efficient Degradation of APOBEC3 Restriction Factors by HIV-1 Vif

ABSTRACT

The APOBEC3 deoxycytidine deaminases can restrict the replication of HIV-1 in cell culture to differing degrees. The effects of APOBEC3 enzymes are largely suppressed by HIV-1 Vif that interacts with host proteins to form a Cullin5-Ring E3 ubiquitin ligase that induces ⁴⁸K-linked polyubiquitination (poly-Ub) and proteasomal degradation of APOBEC3 enzymes. Vif variants have differing abilities to induce degradation of APOBEC3 enzymes and the underlying biochemical mechanisms for these differences is not fully understood. We hypothesized that by characterizing the interaction of multiple APOBEC3 enzymes and Vif variants we could identify common features that resulted in Vif-mediated degradation and further define the determinants required for efficient Vif-mediated degradation of APOBEC3 enzymes. We used Vifs from HIV-1 NL4-3 (IIIB) and HXB2 to characterize their induced degradation of and interaction with APOBEC3G, APOBEC3G D128K, APOBEC3H, and APOBEC3B in 293T cells. We quantified the APOBEC3G-Vif and APOBEC3H-Vif interaction strengths in vitro using rotational anisotropy. Our biochemical and cellular analyses of the interactions support a model in which the degradation efficiency of Vif_IIIB and Vif_HXB2 correlated with both the binding strength of the APOBEC3-Vif interaction and the APOBEC3-Vif interface, which differs for APOBEC3G and APOBEC3H. Notably, Vif bound to APOBEC3H and APOBEC3B in the natural absence of Vif-induced degradation and the interaction resulted in ⁶³K-linked poly-Ub of APOBEC3H and APOBEC3B, demonstrating additional functionality of the APOBEC3-Vif interaction apart from induction of proteasomal degradation.

IMPORTANCE APOBEC3 enzymes can potently restrict the replication of HIV-1 in the absence of HIV-1 Vif. Vif suppresses APOBEC3 action by inducing their degradation through a direct interaction with APOBEC3 enzymes and other host proteins. Vif variants from different HIV-1 strains have different effects on APOBEC3 enzymes. We used differing Vif degradation capacities of two Vif variants and various APOBEC3 enzymes with differential sensitivities to Vif to delineate determinants of the APOBEC3-Vif interaction that are required for inducing efficient degradation. Using a combined biochemical and cellular approach we identified that the strength of the APOBEC3-Vif binding interaction and the APOBEC3-Vif interface are determinants for degradation efficiency. Our results highlight the importance of using Vif variants with different degradation potential when delineating mechanisms of Vif-induced APOBEC3 degradation and identify features important for consideration in the development of HIV-1 therapies that disrupt the APOBEC3-Vif interaction.

INTRODUCTION

The APOBEC3 (A3) deoxycytidine deaminases can act as intracellular restriction factors against replication of HIV-1 (referred to as HIV) (1). A3 enzymes that are encapsidated into budding HIV virions can restrict HIV replication in the next target cell by deaminating cytosine in minus strand single-stranded DNA (ssDNA), which forms mutagenic uracils and results in numerous C/G→T/A transition mutations that can inactivate the virus (2–4). In CD4⁺ T cells it appears that four of the seven A3 members—A3D, A3F, A3G, and A3H haplotype II (referred to as A3H)—are primarily responsible for HIV restriction (5). Nonetheless, HIV can successfully infect cells where A3 enzymes are highly expressed due to the viral infectivity factor (Vif) protein (6, 7). Vif acts as a substrate receptor for a Cullin5-Ring E3 (Cul5-E3) ubiquitin ligase complex, which can induce polyubiquitination (poly-Ub) and degradation of A3 enzymes (8, 9). This process is mediated by Vif binding to host Cullin5 and the elongin B/C heterodimer (EloB/C) through specific motifs in Vif that mimic human SOCS2 (10–14). Vif also binds with the transcription cofactor CBFβ for thermodynamic stability (15, 16). Within this E3 ligase complex, Rbx2 recruits an E2 ubiquitin-conjugating enzyme to induce ⁴⁸K-linked poly-Ub of A3 enzymes, which is concomitant with their proteasomal degradation (8, 17–20). Vif interacts with A3s through its N-terminal domain (NTD) in distinct regions for A3G (⁴⁰YRHHY⁴⁴), A3H (³⁹F, ⁴⁸H), or A3C, A3D, and A3F (¹⁴DRMR¹⁷) in conjunction with secondary binding sites (21–28). The positively charged surfaces of Vif interact with negatively charged amino acids on the A3 enzyme, with some contribution from hydrophobic amino acid interactions, depending on the interface (23). There are three distinct structural motifs on the A3 enzymes that interact with a corresponding Vif region. These can be categorized into three classes: A3G-, A3H-, or A3C/A3F/A3D-like (23).

A3G is a double Z-domain enzyme that primarily interacts with Vif through residues ¹²⁸DPD¹³⁰ on predicted loop 7 in the NTD (29). The ¹²⁸D amino acid was identified as essential for Vif-mediated degradation and mediation of a cross-species barrier to simian immunodeficiency virus (SIV) by mutating ¹²⁸D to ¹²⁸K, as found in African green monkey or rhesus macaque A3G (30–33). Multiple research groups found that Vif did not interact or induce degradation of the A3G D128K mutant (30–32). However, if the ¹²⁸D was mutated to an alanine, Vif-mediated degradation occurred, demonstrating that the overall charge of the A3G-Vif interface was of importance and not only the amino acid per se (30, 31, 33). The overall interface of A3G that interacts with Vif was studied through a MAPPIT analysis, and it was found that Vif_IIIB could interact not only with ¹²⁸D located on A3G structural loop 7 but also with other adjacent structures, such as helix 6 (34). The MAPPIT analysis was in agreement with other studies that identified larger regions of Vif binding (35–37). Further, a study by our group found that Vif_IIIB and Vif_HXB2 interacted with different regions of the A3G NTD beyond ¹²⁸D (37). This enabled each Vif variant to cause distinct changes in A3G enzymatic activity (37).

For A3H, Vif variant differences in the ability to induce A3H degradation are widespread and have been extensively studied (22, 38). The data from cell culture and HIV-1-infected patients demonstrates that amino acids 39 and 48 in Vif are important for both the ability of Vif to induce degradation of and interact with A3H (22). For example, Vif_LAI which contains ⁴⁸H, can induce degradation but not Vif_IIIB that contains a ⁴⁸N. It was concluded that Vif_IIIB is unable to degrade A3H due to an inability to efficiently associate with the enzyme (38). For A3H, the interface that interacts with Vif occurs on predicted helix 4, with ¹²¹D being the primary determinant in Vif-mediated degradation efficiency (23, 39–41). Zhen et al. found that a D121K mutant was unable to interact with Vif and concluded that similar to the example of A3G D128K a lack of Vif-induced degradation correlates with a lack of an A3-Vif interaction (40).

Vif interacts with the A3C/A3F/A3D interface through A3 helices 2, 3, 4, and β-strand 4 (42–45). For A3F and A3D that contain two Z domains, the structural motif is at their C-terminal domain (CTD). Extensive mutagenesis studies by several groups on A3F demonstrated that mutation of conserved residues on these helices cause less or abolish Vif-mediated A3F degradation (42–44). Mutation of A3F ²⁶³L, ²⁶⁴S, or ³²⁴E that are involved in Vif-mediated degradation (42) did not decrease the binding affinity of Vif for A3F (44, 45), suggesting that for A3F the larger Vif binding interface is not as easily disrupted as for A3G and A3H. However, Kitamura et al. found that an A3F E324K mutant had a partial decrease in the A3F-Vif interaction (42), and Siu et al. found no interaction of an E324Q mutant with Vif (45), demonstrating that there is experimental variability between labs and that the relationship between a lack of Vif-mediated degradation and the A3-Vif interaction still requires further investigation.

Despite a large amount of data regarding A3 and Vif interactions, the biochemical basis of how Vif induces A3 substrate poly-Ub and degradation remains to be determined. Studies which examined differences in Vif variants and found specific nonconserved amino acids that were responsible for the ability to induce degradation suggest that the interaction strength with Vif is a primary determining factor in successful Vif-mediated degradation (22, 38). However, these studies used qualitative coimmunoprecipitations to characterize these interactions. Therefore, to better understand the mechanistic role of Vif as the Cul5-E3 ubiquitin ligase substrate receptor, we undertook a study to define the parameters for efficient Vif-induced degradation of an A3 enzyme, using both biochemical and cellular experiments. We hypothesized that the interaction strength between the A3 and Vif was not the sole determinant of degradation efficiency and that the A3-Vif interface also plays a role. To test this hypothesis, we used Vif variants from HIV NL4-3 (IIIB) and HXB2 and A3 enzymes that are either efficiently degraded by these Vifs (A3G) or not efficiently degraded by both Vifs (A3G D128K, A3H, and A3B) (22, 29–33, 41, 46, 47). We used cellular experiments to determine Vif-induced degradation efficiency and A3-Vif interactions in cells in conjunction with an in vitro quantitative method to determine the binding strength of A3G, A3G D128K, and A3H with Vif variant heterotetramers (Vif/CBFβ/EloB/C), the most stable form of Vif (48–50). Our biochemical and cellular analyses of the interactions support a model in which the degradation efficiency of Vifs correlated with the both the A3-Vif binding strength and A3-Vif interface. Further, Vif_IIIB bound to A3H and A3B in the natural absence of degradation and induced their ubiquitination by a ⁶³K-linked poly-Ub chain. Our data suggest that the consequences of an A3-Vif interaction and determinants of Vif-induced degradation are more complex than originally identified.

MATERIALS AND METHODS

Plasmid constructs.

Human pcDNA3.1-APOBEC3G-HA was obtained through the NIH AIDS Reagent Program (19). Site-directed mutagenesis was used to construct the A3G D128K mutant. An A3H haplotype I clone (NCBI accession no. BC069023) was obtained from Open Biosystems and using site-directed mutagenesis was converted to A3H haplotype II-RDD. The A3H was PCR amplified and subcloned into pcDNA3.1 with or without a C-terminal 3X-hemagglutinin (HA) tag. The human phAPOBEC3B-HA plasmid was obtained through the NIH AIDS Reagent Program (46). Site-directed mutagenesis was used to generate A3B mutants. The A3B NTD (amino acids 1 to 194) and CTD (amino acids 195 to 387) expression plasmids were constructed by PCR and subcloning of the desired nucleotide sequences into a modified pcDNA3.1-3X HA plasmid. The Vif_IIIB was expressed from the partially codon-optimized pcDNA-HVif plasmid obtained through the NIH AIDS Reagent Program (51). A corresponding partially codon-optimized Vif_HXB2 was constructed by gene synthesis (GenScript) and subcloned into pcDNA3.1. The sequence is available upon request. Site-directed mutagenesis was used to generate Vif_HXB2 mutants. Codon-optimized HA-Ub and HA-⁴⁸K-only Ub expression plasmids were provided by Wei Xiao (University of Saskatchewan). The ubiquitin genes from these plasmids were used as a template in a PCR, which added an N-terminal V5 tag to the Ub. The V5-Ub was subcloned into pcDNA3.1. All constructed plasmids were verified by DNA sequencing. Primers were obtained from Integrated DNA Technologies and are listed in Table S1 in the supplemental material.

Vif-mediated degradation assay.

The 293T cells were maintained in Dulbecco modified eagle medium (Invitrogen) supplemented with 10% fetal bovine serum (PAA) in the presence of 5% CO₂ at 37°C. Transfections used 6 × 10⁵ cells per well in a six-well plate and 0.6 to 1 μg of A3 expression plasmid in the presence of 0 to 1 μg of Vif expression plasmid. Empty pcDNA3.1 was used to equalize transfected DNA amounts. The expression of all A3s was similar (data not shown). The Polyfect transfection reagent (Qiagen) was used according to the manufacturer's instructions. For experiments requiring a proteasome inhibitor, MG132 (14 μM) was added 12 to 16 h after transfection. At 40 h posttransfection, cells were washed in phosphate-buffered saline (PBS) and harvested in Laemmli sample buffer (58 mM Tris [pH 6.8], 5% [vol/vol] glycerol, 2% [wt/vol] sodium dodecyl sulfate [SDS], 1.5% [wt/vol] dithiothreitol [DTT]) for analysis by immunoblotting.

Coimmunoprecipitations.

The 293T cells (2.5 × 10⁶ cells per 75-cm² flask) were transfected with 8 μg of total DNA. For experiments examining the Vif and A3 interaction, 4 μg of an A3 or Vif expression plasmid was used alone or in combination. For cellular ubiquitination assays, the cells were cotransfected with either V5-Ub or V5-⁴⁸K only-Ub and A3-HA expression plasmids in the absence or presence of a Vif variant expression plasmid. Equal amounts of each plasmid was used and empty pcDNA3.1 was used to equalize the DNA transfected. Polyfect transfection reagent (Qiagen) was used according to the manufacturer's instructions. After 12 to 16 h of transfection, the cells were treated with MG132 (14 μM). At 40 h posttransfection, the cells were washed with PBS and lysed in coimmunoprecipitation buffer (50 mM Tris-Cl [pH 7.4], 1% Nonidet-P40, 0.1% sodium deoxycholate, 10% glycerol, 150 mM NaCl) supplemented with EDTA-free protease inhibitor (Roche) and 14 μM MG132. Clarified supernatants were precleared with protein A-agarose-conjugated normal rabbit IgG (2 μg; SC-2345) in the presence of RNase A (10 μg; Roche) and then incubated with protein A-agarose-conjugated polyclonal rabbit HA antibody (2 μg; Sigma) at 4°C for 2 h. Resin was washed at room temperature in coimmunoprecipitation buffer three times. The samples were then resuspended in Laemmli sample buffer and heated at 100 °C for 5 min prior to SDS-PAGE.

Immunoblot analysis.

For qualitative immunoblotting, proteins were detected by chemiluminescence after the nitrocellulose membranes were probed with antibodies. The antibodies used were monoclonal mouse HA (1:2,000; Sigma), monoclonal mouse A3H (catalog no. 12155; 1:1,000 [obtained through the NIH AIDS Reagent Program]) (41), HIV-1 Vif monoclonal antibody (catalog no. 6459; 1:600; obtained through the NIH AIDS Reagent Program) (52), or monoclonal mouse α-tubulin (1:1,000; Sigma). After incubation with secondary antibodies, the blots were visualized with X-ray film using Super Signal West Pico chemiluminescence substrate (Thermo Scientific).

For quantitative immunoblotting, nitrocellulose membranes were incubated concurrently with primary antibodies for the A3 protein and the loading control. Experiments with cell lysates used a combination of polyclonal rabbit HA (1:1,000; Sigma) and monoclonal mouse α-tubulin (1:1,000; Sigma), monoclonal mouse HA (1:2,000; Sigma) and polyclonal rabbit α-tubulin (1:1,000 to 1:500; Thermo Scientific), or monoclonal mouse A3H (catalog no. 12155; 1:500 [obtained through the NIH AIDS Reagent Program]) and polyclonal rabbit α-tubulin. For immunoblotting of virions, monoclonal mouse HA (1:2,000; Sigma) and rabbit antiserum to HIV-1 p24 (catalog no. 4250; 1:2,000 [obtained through the NIH AIDS Reagent Program]) was used. Secondary antibodies were also used concurrently and were a combination of goat anti-rabbit IRDye 680RD and goat anti-mouse IRDye 800CW or of goat anti-mouse IRDye 680RD and goat anti-rabbit IRDye 800CW (Li-Cor Biosciences). The detection of Vif was done in parallel and used HIV-1 Vif monoclonal antibody (catalog no. 6459; 1:600), followed by incubation with goat anti-mouse IRDye 680RD. To ensure that the bands were not saturated, which would prevent accurate quantification, images were obtained and analyzed with Odyssey software that detects and prevents analysis of bands with saturated pixels. The percent A3 remaining from degradation assays was calculated by first normalizing each sample lane to the corresponding alpha-tubulin control. This enabled different blots to be compared to each other. Normalized values were then converted to relative amounts of A3 remaining by setting the no-Vif condition at 100% and calculating the relative percentage of A3 remaining.

Single-cycle infectivity assays.

VSV-G-pseudotyped HIV pNL4-3 eGFP Δvif viruses were produced in the presence or absence of an A3 or Vif expression plasmid, as previously described (53). The cotransfection molar ratio of A3 enzymes in pcDNA3.1 to the pHIV eGFP Δvif was 0.33:1, and that of of A3 enzyme to Vif was 1:1. The infection levels in 293T cells were determined by flow cytometry by detecting enhanced green fluorescent protein (eGFP) fluorescence at 48 h postinfection, and data were normalized to HIVΔvif infections in the absence of A3 enzymes.

Protein expression and purification.

The glutathione S-transferase (GST)-A3H was produced using recombinant baculovirus-infected Sf9 cells. The recombinant baculovirus was constructed by using the pAcG2T transfer vector (BD Biosciences) as described by Chelico et al. (54). The A3H haplotype II template used for PCR and subcloning was identical to that used for the pcDNA3.1 vector. The plasmid construction of GST-A3G and GST-D128K has been previously described (37, 54). GST-A3H, GST-A3G, and GST-A3G D128K were expressed and purified as described previously to obtain protein that was cleaved from the GST tag and 95% pure (Fig. 1) (55).

FIG 1.

Assessment of the purity of proteins. The purity of A3G, A3H, Vif_IIIB/CBFβ/ELOB/C, and Vif_HXB2/CBFβ/ELOB/C was assessed by resolving 3.5 μg of protein using SDS–15% PAGE and staining with Coomassie blue.

The Vif heterotetramer (Vif/CBFβ/EloB/C) was obtained by coexpression of all components in Escherichia coli. The Vif heterodimer (Vif/CBFβ) was obtained by coexpression of Vif and CBFβ in E. coli. The elongin B (residues 1 to 187) and elongin C (residues 17 to 112) cloned in the pACYC-Duet plasmid were a gift from Alex Bullock (University of Oxford). The CBFβ isoform 2 (residues 1 to 182, NCBI accession no. NP_001746) was cloned into MCS1 of pET-Duet at the BamHI and HindIII sites. Vif_HXB2 and Vif_IIIB carried in a pAcG2T baculovirus transfer vector as described previously (37) were subcloned into the BglII and XhoI sites of the MCS2 of pET-DUET (Novagen). The codon-optimized sequences of Vif_HXB2 and Vif_IIIB were also cloned into the NcoI and HindIII sites of the MCS1 of pRSF-Duet and pCDF-Duet, respectively, to supplement Vif expression. Protein expression in E. coli BL21(DE3) cells was induced with 0.2 mM IPTG (isopropyl-β-d-thiogalactopyranoside) for 16 h at 16°C. Cells were lysed by sonication in ice-cold lysis buffer (20 mM Tris-Cl [pH 8.0], 300 mM NaCl, RNase A [80 μg/ml; Qiagen]) supplemented with EDTA-free protease inhibitor (Roche) and lysozyme (1 mg/ml; EMD Chemicals). Clarified supernatants were transferred to Talon metal affinity resin (Clontech Laboratories) and incubated for 45 min at 4°C. The affinity resin was then extensively washed with buffer containing increasing amounts of imidazole (20 mM Tris-Cl [pH 8.0], 300 mM NaCl, 5 to 10 mM imidazole). Affinity resin-bound proteins were eluted with elution buffer (20 mM Tris-HCl [pH 8.0], 300 mM NaCl, 200 mM imidazole, 10% glycerol). The eluted proteins were loaded onto a Superdex 200 Increase (10/300) column (GE Healthcare) equilibrated with buffer containing 20 mM Tris-Cl (pH 8.0), 300 mM NaCl, 10% glycerol, and 1 mM DTT. Peak fractions corresponding to the Vif/CBFβ/EloB/C heterotetramer or Vif/CBFβ heterodimer were identified based on comparison to molecular mass standards (Bio-Rad) and confirmed by SDS-PAGE. The Vif heterotetramer was further purified using a DEAE Fast Flow column (GE Healthcare). The proteins were loaded onto the column in low-salt buffer (20 mM HEPES [pH 7.0], 75 mM NaCl, 10% glycerol, 1 mM DTT) and eluted using a salt gradient. The Vif heterotetramer eluted from the column at ∼500 mM NaCl. The complexes were estimated to be 95% pure by SDS-PAGE (Fig. 1).

To determine the oligomerization state of A3G, A3H, and Vif_HXB2 heterotetramer, 100 to 150 μg of purified enzyme was loaded onto a Superdex 200 Increase (10/300) column equilibrated with buffer containing 20 mM Tris-Cl (pH 8.0), 300 mM NaCl, 10% (vol/vol) glycerol, and 1 mM DTT. For determination of the A3G- or A3H-Vif_HXB2 heterotetramer complex, 100 μg of either A3G or A3H plus the Vif_HXB2 heterotetramer were premixed and incubated at room temperature for 3 min before being loaded onto the column. The Bio-Rad gel filtration standard set was used to generate a standard curve from which the molecular masses and oligomerization state was calculated.

Steady-state fluorescence depolarization.

The affinity of the interaction between A3G, A3G D128K, or A3H and Vif/CBFβ/EloB/C was determined by using steady-state fluorescence depolarization (rotational anisotropy). The purified A3G, A3G D128K, or A3H was labeled with fluorescein using a fluorescein-EX protein labeling kit (Invitrogen) and used as the binding substrate for Vif/CBFβ/EloB/C. Reactions (50 μl) were conducted in buffer (50 mM Tris [pH 7.5], 40 mM KCl, 10 mM MgCl₂, 1 mM DTT) and contained 50 nM fluorescein-labeled A3 enzyme and increasing amounts of Vif/CBFβ/EloB/C (0 to 2,800 nM). A QuantaMaster QM-4 spectrofluorometer (Photo 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). An apparent dissociation constant (K_d) was obtained by fitting to a hyperbolic decay or sigmoidal curve equation using SigmaPlot 11.2 software.

In vitro deamination assay.

The 118-nucleotide (nt) ssDNA substrates were obtained from Tri-Link Biotechnologies and are listed in Table S1 in the supplemental material. Substrates contained either two 5′CCC motifs (A3G) or two 5′CTC motifs (A3H) and were used at 100 nM. For A3G, 20 nM enzyme was incubated with 700 nM Vif (IIIB or HXB2) during the reaction. For A3H, 50 nM enzyme was incubated with 1750 nM Vif (IIIB or HXB2) during the reaction. Different amounts of enzyme were used to obtain similar reaction rates for A3G and A3H, but the molar ratio of Vif to A3 enzyme was maintained at 35:1. The reaction buffer used was identical to that used for steady-state fluorescence depolarization. All reactions were carried out under single hit conditions (∼10% substrate consumed), where a single A3 could interact with a particular ssDNA substrate once at most (56). Reaction mixtures were incubated at 37°C for 1.5 to 30 min. Deaminations were detected by resolving DNA that had been treated with uracil-DNA glycosylase and heated under alkaline conditions on a 10% (vol/vol) denaturing polyacrylamide gel. Gel pictures were obtained by 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 (54).

RESULTS

Interaction of A3G with HIV Vif_IIIB and Vif_HXB2.

A3G degradation is induced by all HIV Vif variants (referred to as Vif) tested in the literature (22, 41, 47, 57), including the Vif_IIIB and Vif_HXB2 used in our study (Fig. 2A). Figure 2A demonstrates the almost equivalent abilities of Vif_IIIB and Vif_HXB2 to induce proteasomal degradation of A3G in a degradation assay where A3G and partially codon optimized Vif expression plasmids (see Materials and Methods) were cotransfected into 293T cells. However, Vif_IIIB induced slightly less degradation of A3G than Vif_HXB2 (Fig. 2A, HA blot, lanes 2 and 4). Evidence of the ability of Vif_HXB2 and Vif_IIIB to use different amino acid contacts on A3G beyond the key ¹²⁸D amino acid exists from biochemical studies that examined the effect of Vif on A3G DNA scanning and processivity (37). This biochemical data prompted us to test whether the different A3G-Vif interactions have an impact on the strength of the interaction and whether this influenced the efficiency of Vif-induced degradation.

FIG 2.

Degradation efficiency and interaction of Vif_IIIB and Vif_HXB2 with A3G. (A) Degradation of A3G-HA by Vif_IIIB and Vif_HXB2. 293T cells were transiently transfected with A3G-HA expression plasmid with or without cotransfection of a Vif_IIIB or Vif_HXB2 expression plasmid. The proteasome inhibitor MG132 was added where indicated. Cell lysates prepared in Laemmli sample buffer were analyzed by immunoblotting with antibodies against α-tubulin, HA, and Vif. One representative qualitative immunoblot is shown from three independent experiments. (B) Coimmunoprecipitation of A3G-HA with Vif_IIIB or Vif_HXB2. A3G-HA and Vif variants were either transfected alone or in combination, followed by the addition of MG132. Cells were lysed in coimmunoprecipitation buffer, immunoprecipitated (IP) with anti-HA antibody, and immunoblotted with antibodies against α-tubulin, HA, and Vif. Cell lysates show the expression of α-tubulin, HA, and Vif. One representative qualitative blot is shown from three independent experiments. (C) Steady-state rotational anisotropy between A3G and Vif_IIIB/CBFβ/EloB/C or Vif_HXB2/CBFβ/EloB/C (labeled Vif_IIIB or Vif_HXB2). Anisotropy results were normalized to a start value of 0.14 and an end value of zero for comparison. The actual anisotropy start value was between 0.14 and 0.12. The Vif variant heterotetramer was titrated into a binding reaction with fluorescently labeled A3G to determine an apparent dissociation constant (K_d). A change in anisotropy indicates an interaction between A3G and the Vif heterotetramer. The decrease in anisotropy indicated that the initial fluorescently labeled complex had decreased in size and/or undergone a conformational change that increased the speed of rotation (sketch above panel C). Error bars represent the standard deviations of the mean from three independent experiments. (D) Size exclusion chromatography was used to determine the oligomerization state of an A3G-Vif_HXB2/CBFβ/EloB/C complex compared to A3G alone and the Vif_HXB2/CBFβ/EloB/C heterotetramer alone. Peak fractions were identified by the UV absorbance of the chromatogram and confirmed by resolving samples by SDS-PAGE. The graph represents the quantification of the Coomassie blue-stained bands shown in the SDS-PAGE gels. Comparison to molecular mass standards was used to calculate apparent molecular masses of peak fractions. The presence of EloB/C in fractions containing Vif_HXB2/CBFβ was confirmed by staining with Bio-Rad Oriole stain (gel not shown).

We confirmed previous coimmunoprecipitation results demonstrating that A3G interacted with both Vif variants (Fig. 2B, Vif blot, lanes 4 and 5) (29–32, 34). To obtain a quantitative value for these interactions we used rotational anisotropy. Rotational anisotropy measures the apparent change in rotation speed of a fluorescently labeled molecule due to a change in shape and/or molar mass of the molecule upon binding an unlabeled partner (58). We fluorescently labeled A3G and titrated Vif heterotetramer (Vif/CBFβ/EloB/C) into the binding reaction. The binding of the fluorescently labeled A3G to the Vif heterotetramer will result in a change in rotation speed (anisotropy) until the fluorescently labeled A3G population is saturated with its binding partner. From these measurements, an apparent dissociation constant (K_d) can be calculated. Since Vif must be coexpressed with CBFβ for stability (48), we would be unable to specifically label the Vif and therefore chose to fluorescently label A3G. Usually, upon a fluorescently labeled molecule interacting with a binding partner, the increase in size of the complex causes the rotation speed to decrease, which results in an increase in rotational anisotropy. However, with A3G we found that there was a decrease in anisotropy, indicative of an increase in rotation speed (Fig. 2C). We hypothesized that this was due to Vif disrupting the dimers of A3G (55, 59) and interacting with a monomer of A3G (Fig. 2C, sketch) since the key residues for A3G dimerization (¹²⁶FW¹²⁷) are adjacent to ¹²⁸D (55, 60). The existence of this stoichiometric complex (A3G/Vif/CBFβ/EloB/C) was confirmed with size exclusion chromatography (Fig. 2D). First, A3G and Vif_HXB2 heterotetramer were run individually on the size exclusion column. Consistent with previous studies, A3G had an apparent molecular mass of a dimer (Fig. 2D, fraction 25, 74 kDa) and the Vif_HXB2 heterotetramer was a 1:1:1:1 complex (Fig. 2D, fraction 26, 67 kDa) (55, 61). The A3G-Vif_HXB2 heterotetramer complex resolved in three peak fractions with apparent molecular masses ranging from 157 to 96 kDa (Fig. 2D, fractions 22 to 24). If a Vif_HXB2 heterotetramer were to interact with a monomer of A3G, the expected apparent molecular mass would be 111 kDa, which is within the range of the peak fractions (Fig. 2D, fractions 23 and 24). Equivalent results were found with Vif_IIIB (data not shown). As a result, the apparent dissociation constants obtained were a measure of the ability of the Vif heterotetramer to disrupt A3G oligomerization, which was simultaneous with an interaction between the two proteins. A proportion of the Vif_HXB2 heterotetramer interacts with a dimer of A3G (expected apparent molecular mass of 141 kDa, fraction 22). This is due to the size exclusion chromatography being run with a 1:1 ratio of A3G:Vif heterotetramer to enable resolution of the different species. According to the rotational anisotropy data, we would require a 2-fold excess of Vif heterotetramer to completely disrupt the A3G dimer population in steady state (Fig. 2C, 50 nM A3G saturates at 100 nM Vif heterotetramer). However, we could not determine whether an A3G monomer is more readily degraded than wild-type A3G. The monomeric F126A/W127A A3G mutant (55) is degraded by both Vif variants but ∼2-fold less than A3G (data not shown). This is likely due to the caveat that the mutations are adjacent to ¹²⁸D, which may affect the A3G-Vif interaction.

Consistent with efficient degradation by both Vifs, the apparent K_d of the A3G and Vif heterotetramer complexes were in the low nanomolar range (Fig. 2C, Vif_IIIB, apparent K_d = 46 ± 7 nM; Vif_HXB2, apparent K_d = 11 ± 2 nM). However, the apparent K_d of Vif_IIIB and A3G was ∼4-fold higher than the apparent K_d of Vif_HXB2 and A3G (Fig. 2C). This difference, albeit only subtle, is consistent with results from our degradation assays, where we observed more residual A3G in the presence of Vif_IIIB than Vif_HXB2 (Fig. 2A, HA blot, lanes 2 and 4) and suggests that small changes in the binding strength influence degradation efficiency. Altogether, these data are consistent with other studies that demonstrated that the determining factor for the ability of Vif to degrade an A3 enzyme is a physical interaction.

HIV Vif_IIIB maintains an interaction with A3H in the absence of inducing degradation.

To examine whether determinants of the A3-Vif interaction are similar for other A3s, we studied A3H for comparison. A3H has been shown to interact with different amino acids on Vif than A3G and is differentially sensitive to Vif variants from HIV NL4-3 (IIIB) and LAI (22, 62). The Vif_HXB2 used in our study and Vif_LAI sequences are identical. Accordingly, our results were consistent with previous results, and we observed that A3H was not sensitive to Vif_IIIB-mediated degradation but was sensitive to Vif_HXB2 degradation (Fig. 3A, HA blot, lanes 2 and 3). This difference in the degradation efficiency of each Vif for A3H was not influenced by the HA tag since the same results were obtained with antibody to native A3H (Fig. 3B). However, more lysate was needed to observe residual A3H in the presence of Vif_HXB2 due to the lower sensitivity of the native A3H antibody in comparison to the antibody to the HA tag (Fig. 3B) (41).

FIG 3.

A3H interacts with both Vif_IIIB and Vif_HXB2 in cells and in vitro. (A) Degradation of A3H-HA by Vif_IIIB and Vif_HXB2. 293T cells were transiently transfected with A3H-HA expression plasmid with or without cotransfection of a Vif_IIIB or Vif_HXB2 expression plasmid. Cell lysates prepared in Laemmli sample buffer were analyzed by immunoblotting with antibodies against α-tubulin, HA, and Vif. (B) Vif-mediated degradation assay was carried out as in panel A, but 293T cells were transfected with a native A3H expression plasmid and antibody to native A3H was used for detection of A3H from blots with either 30 or 60 μg of cell lysate loaded per lane. (A and B) Representative qualitative immunoblots are shown from three independent experiments. (C) Coimmunoprecipitation of A3H-HA and Vif_IIIB or Vif_HXB2. A3H-HA and Vif variants were either transfected alone or in combination, followed by the addition of MG132. Cells were lysed in coimmunoprecipitation buffer, immunoprecipitated (IP) with anti-HA antibody, and immunoblotted with antibodies against α-tubulin, HA, and Vif. Cell lysates show the expression of α-tubulin, HA, and Vif. A representative qualitative blot is shown from three independent experiments. (D) By comparison to a standard curve, size exclusion chromatography was used to determine the oligomerization state of A3H as described for A3G in Fig. 2D. (E) Steady-state rotational anisotropy between A3H and Vif_IIIB/CBFβ/EloB/C or Vif_HXB2/CBFβ/EloB/C (labeled as Vif_IIIB or Vif_HXB2). The Vif variant heterotetramer was titrated into a binding reaction with fluorescently labeled A3H to determine an apparent dissociation constant (K_d). Error bars represent the standard deviations of the mean from three independent experiments. (F) Size exclusion chromatography was used to determine the oligomerization state of an A3H-Vif_HXB2/CBFβ/EloB/C complex compared to A3H alone and the Vif_HXB2/CBFβ/EloB/C heterotetramer alone as described for A3G in Fig. 2D. The presence of EloB/C in fractions containing Vif_HXB2/CBFβ was confirmed by staining with Bio-Rad Oriole stain (gel not shown).

To establish whether these observations regarding Vif-mediated degradation correlated with the A3H-Vif interaction strength, we used coimmunoprecipitation. We found that both Vif_IIIB and Vif_HXB2 could coimmunoprecipitate with A3H-HA (Fig. 3C, Vif blot, lanes 4 and 5), suggesting that a lack of Vif-induced degradation did not always correlate with a lack of an interaction. To quantify this interaction, we used rotational anisotropy. First, we determined the oligomerization state of A3H by size exclusion chromatography. In contrast to other single Z-domain A3 enzymes (42, 63), we found that A3H existed as polydisperse forms in solution (Fig. 3D). Based on the apparent molecular masses, A3H can exist as a monomer (apparent molecular mass of 22 kDa), dimer (apparent molecular mass of 44 kDa), and trimer/tetramer (apparent molecular mass of 67 kDa). The dimeric A3H was the most predominant form (Fig. 3D, fraction 28). Similar to A3G, we found that the rotational anisotropy of fluorescently labeled A3H decreased upon the addition of the Vif heterotetramer (Fig. 3E), suggesting that Vif disrupted A3H oligomerization. We found that in the presence of the Vif_HXB2 heterotetramer, the A3H peak shifted from Fraction 28 to Fractions 22 to 24 (Fig. 3F). The A3H in fractions 22, 23, and 24 was approximately equal, suggesting that Vif_HXB2 interacted with A3H monomers (fraction 24), A3H dimers (fractions 23 and 24), and A3H tetramers (fraction 22). Thus, the observed decrease in rotational anisotropy (Fig. 3E) may be a combination of a change in A3H oligomerization and structural conformation, both of which can cause the molecules to rotate faster (58). Notably, the apparent K_d of Vif_HXB2 for A3H was ∼60-fold lower than that of Vif_IIIB for A3H (Fig. 3E, Vif_HXB2, apparent K_d = 7 ± 2; Vif_IIIB, apparent K_d = 442 ± 89). Binka et al. (22) showed that mutations F39S and H48N can disrupt the interaction of Vif_LAI with A3H. Since Vif_IIIB has ³⁹F but lacks a histidine at position 48 and instead encodes an asparagine at position 48 (Fig. 4A), the binding data essentially measure the contribution of Vif amino acid 48 in the binding strength of the A3H-Vif interaction. The interaction between A3H and Vif_IIIB was not induced by transfection conditions (38) or washing conditions during the coimmunoprecipitation since lysates containing Vif only did not immunoprecipitate (Fig. 3C, lanes 1 and 2). Rather, A3H exemplified that although Vif_HXB2 bound A3H with higher affinity than Vif_IIIB, both Vifs bound A3H in the low nanomolar range, but this was not sufficient for complete Vif-mediated degradation of A3H (compare Fig. 3A and E). Therefore, the interaction strength appeared not to be the only determinant.

FIG 4.

The A3-Vif interface affects the efficiency of Vif-mediated degradation. (A) Alignment of Vif_IIIB and Vif_HXB2 amino acid sequences from amino acids 16 to 53. Residues that differ between the Vif variants are shown in lowercase. Individual residues of Vif_HXB2 that were mutated to those of Vif_IIIB are shown in red. The ⁴⁰YRHHY⁴⁴ motif highlighted in blue is the A3G interaction motif. (B) Quantitative immunoblotting of 293T cell lysates from a Vif-mediated A3H degradation assay using untagged A3H. Mutants of Vif_HXB2 are labeled above the blot. 293T cells were transiently transfected with untagged A3H expression plasmid with or without cotransfection of a Vif_IIIB or Vif_HXB2 expression plasmid. Cell lysates prepared in Laemmli sample buffer were analyzed by immunoblotting with antibodies against α-tubulin and A3H on one blot and antibody against Vif on a blot done in parallel. Nonsaturated signals were quantified using Odyssey software. The percent A3H remaining from degradation assays was calculated by first normalizing each sample lane to the corresponding α-tubulin control and then setting the no-Vif condition at 100% and calculating relative amounts of A3H remaining. (C) Quantitative immunoblotting of 293T cell lysates from a Vif-mediated A3G degradation assay, conducted as described for panel B but based on detection of A3G-HA through the HA tag. (B and C) Representative qualitative blots are shown from three independent experiments. For the panel B and C plots, the bars above the red hatched lines indicate less degradation of A3H or A3G occurred compared to the wild-type Vif_HXB2. (D) Quantitative immunoblotting of 293T cell lysates from a Vif-mediated degradation assay, conducted as described for panel B but based on detection of A3H-HA or A3G-HA through the HA tag with a titration of transfected Vif_HXB2 expression plasmid. Error bars represent the standard deviations of the mean from three independent experiments. A representative immunoblot is shown.

We investigated whether another determinant could be the A3-Vif interaction interface. Binka et al. (22) found that A3H interacted with Vif_LAI at a different residue than A3G, and we found a similar result with A3H and Vif_HXB2. By mutating Vif_HXB2 at single amino acid positions to those of Vif_IIIB (Fig. 4A), we determined by quantitative immunoblotting of lysates from a degradation assay that ⁴⁸H was the key residue responsible for degradation of A3H by Vif_HXB2 (Fig. 4B). Mutations in this region of Vif_HXB2 did not affect A3G degradation (Fig. 4C), a finding consistent with previous studies (21, 22). To determine whether the A3-Vif interface could influence the degradation efficiency, we focused on the interaction of A3H and A3G with Vif_HXB2. Although Vif_HXB2 could interact with A3H (Fig. 3E, apparent K_d = 7 ± 2 nM) as tightly as A3G (Fig. 2C, apparent K_d =11 ± 2 nM), the degradation efficiency was not equivalent (Fig. 4D). Quantification of the degradation of A3H and A3G by Vif_HXB2 demonstrated that A3H was degraded 1.5- to 4 -fold less than A3G over a range of Vif_HXB2 plasmid transfection levels (Fig. 4D). These data are consistent with other studies that showed A3H is less susceptible to Vif_LAI than A3G (22, 41). The data demonstrate that mechanistically there is a difference in the degradation efficiency of Vif_HXB2 for A3G and A3H. However, this does not appear to influence HIV restriction, since both A3G and A3H are unable to effectively restrict HIV_LAI (38).

Vif_HXB2 can induce degradation of A3G D128K.

To further test whether the difference in Vif-mediated degradation efficiency is influenced by the A3-Vif interface, we used the A3G D128K mutant. D128K is resistant to transiently overexpressed Vif_IIIB-mediated degradation (30, 31, 33) but only partially resistant to transiently overexpressed Vif_HXB2-mediated degradation (Fig. 5A, HA blot, lanes 2 and 4). Vif_HXB2 can induce degradation of ∼40% of the transiently overexpressed D128K (Fig. 5B, 1,000 ng of Vif_HXB2 plasmid). That the degradation was not strong was consistent with results from others (32) and coimmunoprecipitation data in which neither Vif_IIIB nor Vif_HXB2 could coimmunoprecipitate with the D128K-HA (Fig. 5C, Vif blot, lanes 4 and 5). However, using rotational anisotropy, which is a more sensitive method, we found that D128K and Vif_HXB2 did interact with each other, albeit at low affinity (Fig. 5D, apparent K_d of 1,347 ± 125 nM). The interaction strength of Vif and D128K was ∼120-fold less than with A3G, thus necessitating titration of 4-fold more Vif heterotetramer to determine the apparent K_d (compare x axis, Fig. 5D and Fig. 2C) and consistent with the suboptimal induced degradation (compare Fig. 5B and Fig. 4D). There was no interaction detected when A3G D128K was incubated with increasing amounts of Vif_IIIB heterotetramer (Fig. 5D). Although Vif_HXB2 could induce degradation of D128K, this did not affect the ability of D128K to restrict viral replication in the presence of Vif. Single-cycle infectivity assays with cotransfected ΔVif HIV, Vif (IIIB or HXB2), or A3 (A3G or D128K) expression plasmids showed that D128K could restrict HIV equally under all conditions, in contrast to A3G (Fig. 5E). D128K was degraded equally well by Vif_HXB2 in cells that were not infected (Fig. 5B) or infected (Fig. 5F) with HIV. In the presence of Vif_HXB2 there was less D128K encapsidated into virions compared to the no-Vif or Vif_IIIB condition, but D128K was still able to restrict the virus replication effectively (Fig. 5E and Fig. 5F, virion HA blot). Despite the degradation of D128K by Vif_HXB2 not affecting viral restriction, the data demonstrate that mechanistically the different interfaces of A3G/D128K and Vif_HXB2 and A3H and Vif_HXB2 influences the degradation efficiency. D128K could be degraded by Vif_HXB2 despite an interaction in the micromolar range (Fig. 5A, B, and D), whereas A3H could not be degraded by Vif_IIIB and was degraded by Vif_HXB2 only 2-fold more than D128K, despite a low nanomolar binding affinity (Fig. 4D and Fig. 3E).

FIG 5.

D128K interacts with Vif_HXB2. (A to C) Degradation of D128K-HA by Vif_IIIB and Vif_HXB2. 293T cells were transiently transfected with D128K-HA expression plasmid with or without cotransfection of a Vif_IIIB or Vif_HXB2 expression plasmid, and the proteasome inhibitor MG132 was added where indicated. Cell lysates prepared in Laemmli sample buffer were analyzed by immunoblotting with antibodies against α-tubulin, HA, and Vif. (A) One representative qualitative immunoblot is shown from three independent experiments. (B) Quantitative immunoblotting of 293T cell lysates from a Vif_HXB2-mediated D128K-HA degradation assay was performed as described in Fig. 4C. One representative quantitative immunoblot is shown from three independent experiments (C) Coimmunoprecipitation between D128K-HA and Vif_IIIB or Vif_HXB2. D128K-HA and Vif variants were either transfected alone or in combination, followed by the addition of MG132. Cells were lysed in coimmunoprecipitation buffer and immunoprecipitated (IP) with anti-HA antibody and immunoblotted with antibodies against α-tubulin, HA, and Vif. Cell lysates show the expression of α-tubulin, HA, and Vif. One representative qualitative blot is shown from three independent experiments. (D) Steady-state rotational anisotropy between D128K and Vif_IIIB/CBFβ/EloB/C or Vif_HXB2/CBFβ/EloB/C (labeled Vif_IIIB or Vif_HXB2). The Vif variant heterotetramer was titrated into a binding reaction with fluorescently labeled D128K to determine an apparent dissociation constant (K_d). (E) Virus infectivity was measured by eGFP expression in 293T cells infected with HIVΔvif that was produced in the absence or presence of A3G, D128K, Vif_IIIB, or Vif_HXB2. The results are shown normalized to the no-A3 condition. Error bars represent the standard deviations of the mean from three independent experiments. (F) Quantitative immunoblotting was used to determine the levels of A3G and D128K expressed in cells and encapsidated into HIVΔvif virions. The loading control for cell lysates was α-tubulin and for virions was p24. A representative immunoblot is shown. A cell lysate blot was analyzed to determine the band intensity for the HA tag in the presence of Vif_IIIB or Vif_HXB2, relative to no Vif. Error bars represent the standard deviations of the mean from three independent experiments.

HIV Vif induces a nondegradative posttranslational modification of A3H in 293T cells.

For A3H and Vif_IIIB, the interaction maintained in the absence of A3H degradation may be a consequence of maintaining a lower affinity interaction to decrease the evolutionary barrier when HIV adapts to rare hosts that express a stable A3H haplotype (62). Alternatively or in addition to this, Vif_IIIB may exert degradation-independent inhibitory effects on A3H through this interaction. Degradation-independent effects of Vif on A3G have been characterized such as inhibition of virion encapsidation, depression of A3G mRNA translation, and inhibition of deamination activity of virion encapsidated A3G molecules that escape Vif-mediated suppression (19, 37, 64–69). We did not find that Vif_IIIB/CBFβ or Vif_HXB2/CBFβ could alter the deamination activity or processive scanning of A3H, as was previously found for A3G (Fig. 6A) (37). This may be because the primary interaction residue of A3H (¹²¹D) is predicted to be located on helix 4, not on loop 7 as for A3G, and is on a face of A3H distal to the amino acids predicted to be involved in A3 enzyme processivity (23, 70).

FIG 6.

Degradation-independent effects of Vif on A3H. (A) Processivity of A3H (left gel) and A3G (right gel) were tested on 118-nt ssDNA substrates (100 nM) that contained an internal fluorescein (F) label (yellow star) and two deamination motifs separated by different distances. The A3H substrates had 5′CTC motifs and the A3G substrates had 5′CCC motifs. The two target cytosines were spaced 63 nt apart. Single deaminations of the 5′C and 3′C were detected as the appearance of labeled 100- and 81-nt fragments, respectively; double deamination of both C residues on the same molecule resulted in a 63-nt labeled fragment (5′C and 3′C). The standard deviation of the mean (S.D.) was calculated from three independent experiments. (B and C) Detection of Vif-mediated poly-Ub of A3H and A3G. 293T cells were cotransfected with A3G-HA or A3H-HA in the presence of V5-Ub or V5-⁴⁸K only Ub and MG132, with or without Vif_IIIB or Vif_HXB2. The cells were then lysed in coimmunoprecipitation buffer, immunoprecipitated (IP) with anti-HA antibody, and immunoblotted with V5 and HA antibodies. Ubiquitin chain formation is denoted as Ub_n or ⁴⁸K Ub_n. A nonspecific band detected with the V5 tag is denoted as NS. Verification of expression of α-tubulin, HA, and Vif proteins in cell lysates used for IPs is shown. A representative qualitative blot is shown from two independent experiments. (D and E) Vif-mediated poly-Ub of alternate Ub chain types. Cell lysate obtained in from the experiment shown in panel B was immunoprecipitated with antibody to the HA tag as described in panel B. (D and E) Immunoblotting was carried out with antibody specific to ¹¹K-linked Ub chains (D) and antibody specific to ⁶³K-linked Ub chains (E). Ub chain formation is denoted as ¹¹K Ub_n and⁶³K Ub_n. The results of a representative qualitative blot are shown from two independent experiments. (F) Vif does not mediate ⁶³K-linked poly-Ub on D128K. The 293T cells were cotransfected with D128K-HA in the presence of V5-Ub and MG132, with or without Vif_IIIB or Vif_HXB2. Cell lysates were immunoprecipitated as described in panel B and blotted with antibodies specific to ⁶³K-linked Ub chains and HA. Cell lysates were also analyzed for the presence of α-tubulin, HA, and Vif. A representative qualitative blot is shown. The cross-reacting immunoglobulin heavy chain (HC) and light chain (LC) are labeled on panels E and F.

Since the major mechanism of A3 inhibition by Vif is polyubiquitination and it is likely that Vif primarily exists in host cells as the substrate receptor for the Cul5-E3 ubiquitin ligase complex, we investigated whether A3H was polyubiquitinated in the presence of Vif_IIIB but by a polyubiquitination type that did not result in degradation. For example, ⁶³K-linked ubiquitin chains have a variety of nondegradative functions in DNA repair and cell signaling by mediating protein-protein interactions (71). To examine ubiquitination, we cotransfected 293T cells with V5-Ub and HA-tagged A3 alone or in the presence of Vif_IIIB or Vif_HXB2, followed by immunoprecipitation with HA antibody and immunoblotting for the V5 tag (ubiquitin). Our results showed that both Vif_IIIB and Vif_HXB2 promoted polyubiquitination (poly-Ub) on A3G (Fig. 6B), in agreement with previous reports (8, 9) and consistent with Vif-mediated degradation (Fig. 2A). We also found that Vif_HXB2 and Vif_IIIB both induced poly-Ub of A3H (Fig. 6B). Whereas the poly-Ub of A3H by Vif_HXB2 is consistent with ⁴⁸K-linked poly-Ub, the poly-Ub of A3H by Vif_IIIB suggested that this interaction resulted in an alternate poly-Ub chain that did not induce substrate degradation.

To investigate this further, we first determined the ⁴⁸K linked poly-Ub status of the A3 enzymes to ensure the chain formation specificity of our experimental system. To observe this, we constructed a mutant plasmid (⁴⁸K only) bearing mutations at all lysines of V5-Ub except lysine 48. This was cotransfected with A3G-HA or A3H-HA in the presence or absence of Vif_IIIB or Vif_HXB2. Consistent with Vif-mediated degradation profiles (Fig. 2A and Fig. 3A), our results confirmed the Vif_HXB2 mediated ⁴⁸K-linked poly-Ub on both A3G and A3H and the Vif_IIIB mediated ⁴⁸K-linked poly-Ub on A3G but not on A3H (Fig. 6C).

To determine the type of ubiquitination chain identified on A3H after interaction with Vif_IIIB (Fig. 6B), we investigated alternate poly-Ub chain types. Specifically, we cotransfected cells with V5-Ub, A3H-HA, or A3G-HA in the presence or absence of Vif_IIIB or Vif_HXB2, followed by immunoprecipitation with antibody to the HA tag and immunoblotting with a ubiquitin chain-specific antibody. We wanted to test in parallel whether multiple types of poly-Ub chains could be formed, such as ⁶K, ¹¹K, ²⁷K, ²⁹K, or ⁶³K, but we could only obtain antibodies for ¹¹K- and ⁶³K-linked poly-Ub. We did not detect ¹¹K-linked poly-Ub (Fig. 6D); however, we did find that Vif_IIIB could mediate ⁶³K-linked poly-Ub on A3H but not on A3G (Fig. 6E). Since we performed this experiment in the presence of MG132 that inhibits proteasomal degradation, we also found strong Vif_HXB2-mediated ⁶³K-linked poly-Ub on A3H. However, despite using MG132 to block degradation of A3G, we did not observe Vif-mediated ⁶³K-linked poly-Ub on A3G (Fig. 6E). These data suggest that it is not the Vif variant but the A3-Vif interaction that determines the poly-Ub chain specificity of the Vif Cul5-E3 ubiquitin ligase. In support of this reasoning, we did not detect ⁶³K-linked poly-Ub of A3G D128K in the presence of Vif_HXB2 (Fig. 6F).

A3B interacts with and is posttranslationally modified by HIV Vif in 293T cells.

Similar to A3H, A3B has been reported to be resistant to degradation by Vif_IIIB and partially sensitive to Vif_HXB2-mediated degradation (46, 47). Therefore, A3B could be used to determine whether Vif can induce ⁶³K-linked poly-Ub on other A3 enzymes for which it does not efficiently induce ⁴⁸K-linked poly-Ub and proteasomal degradation. We confirmed previously published results that Vif_IIIB cannot induce degradation of A3B and Vif_HXB2 can cause partial degradation of A3B through the proteasome (47) (Fig. 7A, HA blot, lanes 2 and 4). Further, using coimmunoprecipitation, we found that both Vif variants could interact with A3B (Fig. 7B, lanes 4 and 5). The degradation abilities of Vif_HXB2 and Vif_IIIB for A3B were differentiated by exchanging individual Vif_HXB2 residues to those of Vif_IIIB and determining degradation efficiency by quantitative immunoblotting (Fig. 7C). The Vif_HXB2 interface with which A3B interacted appeared to be different from both A3H (Fig. 4B) and A3G (Fig. 4C). However, amino acid 48 was important for both A3B and A3H degradation (compare Fig. 7C and Fig. 4B), suggesting that if ⁶³K-linked poly-Ub of A3 enzymes by Vif was a common mechanism dependent on the type of A3-Vif interface, then A3B could be susceptible. In agreement with the degradation assay (Fig. 7A), we found that A3B was only modified by ⁴⁸K-linked ubiquitins in the presence of Vif_HXB2, but not Vif_IIIB (Fig. 7D). Both Vifs could induce ⁶³K-linked poly-Ub of A3B (Fig. 7E).

FIG 7.

A3B interacts with Vif_IIIB and Vif_HXB2. (A) Vif-induced degradation of A3B. 293T cells were transiently transfected with A3B-HA expression plasmid with or without cotransfection of a Vif_IIIB or Vif_HXB2 expression plasmid, and the proteasome inhibitor MG132 was added where indicated. Cell lysates prepared in Laemmli sample buffer were analyzed by immunoblotting with antibodies against α-tubulin, HA, and Vif. A representative qualitative immunoblot is shown from three independent experiments. (B) Coimmunoprecipitation of A3B-HA with Vif variants. A3B-HA and Vif variants were either transfected alone or in combination, followed by the addition of MG132. Cells were lysed in coimmunoprecipitation buffer and immunoprecipitated (IP) with anti-HA antibody and immunoblotted with antibodies against α-tubulin, HA, and Vif. Cell lysates show the expression of α-tubulin, HA, and Vif. A representative qualitative blot is shown from three independent experiments. (C) Quantitative immunoblotting of 293T cell lysates from a Vif-mediated A3B degradation assay. A representative immunoblot is shown from three independent experiments. Mutants of Vif_HXB2 used in the experiment are labeled above the blot. Plotted results are normalized to A3B expression in the absence of Vif and compared to conditions where A3B was cotransfected with Vif_IIIB, Vif_HXB2, or Vif mutants as indicated on the plot. Error bars represent the standard deviations of the mean from three independent experiments. (D and E) 293T cells were cotransfected with V5-Ub and A3B-HA in the presence of MG132 with or without Vif_IIIB or Vif_HXB2. (D) Vif_HXB2 promoted ⁴⁸K-linked poly-Ub on A3B. To detect ubiquitination specific to A3B, lysates prepared in coimmunoprecipitation buffer were immunoprecipitated with antibody specific to ⁴⁸K-linked Ub chains, and proteins were resolved by SDS-PAGE with subsequent immunoblotting against the HA tag. Ub chain formation is denoted as ⁴⁸K Ub_n. (E) Both Vif variants promoted ⁶³K-linked poly-Ub of A3B. The lysates used in panel D were immunoprecipitated through the HA tag, followed by immunoblotting with antibody specific to ⁶³K-linked Ub chains. The Ub chain formation is denoted as ⁶³K Ub_n. The cross-reacting immunoglobulin heavy chain (HC) and light chain (LC) are labeled on the blot. (D and E) Verification of expression of proteins in lysates used for IP is shown only in panel D. One representative blot is shown from two independent experiments.

The C-terminal domain of A3B is required for Vif_HXB2-mediated degradation.

We further studied A3B to determine which residues within A3B interact with Vif_HXB2 to enable Vif-mediated degradation. Insights into determinants for degradation can be used as a tool to establish how Vif acts as a substrate receptor for ⁴⁸K-linked poly-Ub of A3s by the Cul5-E3 ubiquitin ligase complex. To find the region of A3B that was responsible for binding with Vif variants, we constructed expression vectors for only the NTD or CTD of A3B. We then performed coimmunoprecipitations of the NTD (Fig. 8A) or CTD (Fig. 8B) with Vif_IIIB or Vif_HXB2. Our results showed that Vif_IIIB interacted with stronger affinity than Vif_HXB2 with the A3B NTD (Fig. 8A, Vif blot, lanes 4 and 5). In contrast, both Vif_IIIB and Vif_HXB2 efficiently coimmunoprecipitated with the CTD of A3B (Fig. 8B). We have been unable to purify large amounts of A3B to enable its labeling with fluorescein and use in rotational anisotropy to obtain quantitative data on the A3B-Vif interaction. To determine the domain responsible for degradation of A3B by Vif_HXB2, we conducted a degradation assay in 293T cells. The data showed that neither Vif_HXB2 nor Vif_IIIB could induce degradation of the A3B NTD (Fig. 8C, HA blot). This is consistent with a low-affinity interaction of Vif_HXB2 with the A3B NTD (Fig. 8A, Vif blot, lane 5) and the inability of Vif_IIIB to induce degradation of full-length A3B (Fig. 7A). Vif_HXB2 did mediate proteasomal degradation of the A3B CTD (Fig. 8D), suggesting that Vif-mediated degradation of A3B results in polyubiquitination on the CTD. The lack of Vif_IIIB-induced degradation of A3B CTD is consistent with the degradation assay of the full-length protein (Fig. 7A).

FIG 8.

The C-terminal domain of A3B is required for Vif_HXB2-mediated degradation. (A and B) Coimmunoprecipitation of the NTD (A) and CTD (B) of A3B-HA with Vif_IIIB or Vif_HXB2. A3B(NTD)-HA or A3B(CTD)-HA and Vif variants were either transfected alone or in combination, followed by the addition of MG132. Cells were lysed in coimmunoprecipitation buffer, immunoprecipitated (IP) with anti-HA antibody, and immunoblotted with antibodies against α-tubulin, HA, and Vif. Cell lysates show the expression of α-tubulin, HA, and Vif. A representative qualitative blot is shown from three independent experiments. (C and D) Vif induced the degradation of NTD-HA and CTD-HA. The 293T cells were transiently transfected with either A3B(NTD)-HA (C) or A3B(CTD)-HA (D) expression plasmids with or without cotransfection of a Vif_IIIB or Vif_HXB2 expression plasmid, and the proteasome inhibitor MG132 was added where indicated. Cell lysates prepared in Laemmli sample buffer were analyzed by immunoblotting with antibodies against α-tubulin, HA, and Vif. (A to D) A qualitative representative blot is shown from three independent experiments. (E) Sequence alignment of amino acids 252 to 302 of A3B with equivalent residues in A3F (top, slash marks indicate intervening amino acids) and amino acids 120 to 133 of A3B and A3G (bottom). The amino acids denoted in red on the alignments are important for Vif-mediated degradation of A3G or A3F and a selection of these residues were mutated in A3B to determine whether they would decrease Vif_HXB2-mediated degradation. (F) Quantitative immunoblotting of 293T cell lysates from a Vif_HXB2 degradation assay was performed as described in Fig. 4D. A representative immunoblot is shown. Plotted results are based on detection of A3B-HA and normalized to A3B expression in the absence of Vif and compared to conditions where A3B and A3B mutants (as indicated on the plot) were cotransfected with Vif_IIIB or Vif_HXB2. Bars above red hatched line indicate less degradation of the A3B mutant occurred in comparison to wild-type A3B. Error bars represent the standard deviations of the mean from three independent experiments.

After determining that Vif_HXB2 is binding with A3B through primarily the CTD, we wanted to determine which amino acids were involved in Vif-induced degradation. Based on the regions of the A3F CTD that interact with Vif, we performed a sequence alignment with A3B (Fig. 8E). A3B has similarity in its amino acid sequence to A3F (Fig. 8E, ²⁵⁹L, ²⁶⁷L, ²⁷³Y, and ²⁹⁶F). These amino acids were mutated in an attempt to block Vif_HXB2-mediated degradation. We also mutated residues that aligned with the A3G NTD interaction site with Vif as a control (Fig. 8E). Since the A3B NTD did not interact with Vif_HXB2 with high affinity (Fig. 8A, Vif blot, lane 5), we did not expect this region of A3B to influence Vif_HXB2-mediated degradation. Using quantitative immunoblotting, we found the least amount of Vif_HXB2-mediated degradation of A3B when NTD residue ¹²⁸E or CTD residue ²⁹⁶F were mutated (Fig. 8F). Although degradation required the CTD (Fig. 8C and D) and Vif _HXB2 did not interact strongly with the A3B NTD (Fig. 8A), in the full-length enzyme Vif_HXB2 may be able to interact with either the NTD or CTD (Fig. 8F) to induce polyubiquitination of the CTD. This may provide a reasoning for why the full-length A3B is degraded less well than the CTD (HA blots, Fig. 7A, lane 4, and Fig. 8D, lane 4). Further study of the A3B-Vif interface is required since mutagenesis in either the NTD or CTD could not fully recover A3B steady-state levels (Fig. 8F).

DISCUSSION

Due to evidence that different Vif variants have different abilities to degrade specific A3 enzymes (22, 47), we undertook an analysis of various A3-Vif interactions to identify the determinants of an optimal Vif substrate and how the Vif variants differ in their ability to induce A3 degradation. This study is a step toward a better understanding of the A3-Vif relationship at a biochemical level. Based on the enzymes used in our study, we found that there was no clear relationship between the absence of Vif-induced degradation correlating with a lack of an A3-Vif interaction, except for D128K. Rather, the data support a model in which the strength of the interaction between the Vif variant and the A3 enzyme in addition to the A3-Vif binding interface are determinants in Vif-induced degradation efficiency.

Basis for more efficient A3 degradation by Vif_HXB2.

Overall, Vif_HXB2 appears to maintain higher affinity interactions than Vif_IIIB with A3 enzymes. For example, in the absence of the ¹²⁸D residue, Vif_HXB2 could maintain an interaction with D128K, whereas Vif_IIIB could not (Fig. 5D). Vif_HXB2 could interact with A3G and A3H with 4- and 60-fold higher affinities, respectively, than Vif_IIIB (Fig. 2C and Fig. 3E). Further, the presumably lower-affinity interaction of Vif_HXB2 with the residues of A3B that are only partially conserved with either A3G or A3F still enabled partial degradation of A3B (Fig. 8E and F).

These differences in Vif_IIIB and Vif_HXB2 may relate to the extended A3 binding interface beyond the primary residues important for inducing A3 degradation. The extended A3G binding interface was found to be different for Vif_IIIB and Vif_HXB2 (37). Experiments examining how Vif variants affected A3G processivity showed that Vif_HXB2 primarily interacts with loop 7 and Vif_IIIB interacts with loop 7 and helix 6 (37). Therefore, it is conceivable that Vif_HXB2 could interact with D128K due to stronger interactions with the ¹²⁹PD¹³⁰ residues on loop 7 than Vif_IIIB, which maintained the ability of Vif_HXB2 to induce degradation (Fig. 5A, B, and D). The Vif from SIV of sooty mangabey monkeys is able to interact with A3G primarily through ¹²⁹P (72). Vif_IIIB could not maintain a measurable interaction with D128K (Fig. 5C and D), suggesting that it stabilizes itself on A3G through low-affinity interactions with helix 6.

The data with A3H and D128K demonstrated that the A3-Vif interface influences degradation efficiency. D128K degradation occurred despite an apparent K_d for Vif_HXB2 heterotetramer in the micromolar range (Fig. 5B and D). In contrast, Vif_HXB2 degradation of A3H was only 2-fold more than for D128K despite a nanomolar range K_d (Fig. 3E and Fig. 4D). Notably, previously both Vif variants were found to not interact with D128K in a similar in vitro assay that used GST-Vif (37). However, our current data with the Vif heterotetramer (Vif/CBFβ/EloB/C) suggest that in the previous study the GST tag nonspecifically disrupted the Vif_HXB2 and D128K interaction or that further stabilization of Vif with CBFβ/EloB/C was required for the interaction with D128K. However, the GST-Vif produced similar results when interacting with A3G as those obtained here (37), presumably due to the tighter interaction of Vif with wild-type A3G (Fig. 2C).

Vif interacts with A3H and A3B in the absence of degradation.

Unlike A3G, A3H and A3B are resistant to Vif_IIIB-mediated degradation in their wild-type forms and maintain an interaction with Vif_IIIB in the absence of degradation (Fig. 3C, 3E, and 7B). In agreement with our results, a study that used A3B showed it interacted with Vif in cells and that Vif could alter the cellular localization of A3B, in the absence of degradation (47). In contrast, a previous study found that A3H did not associate with Vif_IIIB using coimmunoprecipitation (38). However, quantification of the binding interaction of A3H with each Vif variant showed that there was a 60-fold difference in the interaction strength, demonstrating there was a biochemical difference in the interactions that resulted in degradation versus no degradation (Fig. 3E).

The identification of the ability of Vif to induce ⁶³K-linked poly-Ub of A3H and A3B, but not A3G (Fig. 6E and Fig. 7E), exemplifies the different interactions with of A3G versus A3H and A3B with Vif. In the case of A3H, we reasoned that because some circulating HIV-1 strains are unable to induce A3H degradation and must evolve on a person-to-person basis during HIV-1 infection (62), Vif_IIIB maintains an interaction with A3H to decrease the evolutionary barrier to achieve A3H degradation. Alternatively, Vif_IIIB may inhibit A3H activity by a degradation-independent mechanism, which may be less effective than inducing proteasomal degradation but still prevent HIV-1 inactivation during the time in which Vif evolves to induce A3H degradation.

Our current data support the former hypothesis since we did not find evidence of a degradation-independent mechanism of Vif to inhibit A3H activity (Fig. 6A). Further, although we identified that Vif can induce ⁶³K-linked poly-Ub chains on A3H (Fig. 6E), we were unable to identify a functional consequence as of yet. The ⁶³K-modified A3H contained in 293T cell lysates did not demonstrate a decreased ability to deaminate ssDNA (data not shown) and was encapsidated into virions (data not shown). One possible function of the ⁶³K-linked poly-Ub A3H is to alter viral budding. A ⁶³K-linked poly-Ub chain on Gag or a cellular protein in the vicinity of the emerging viral bud is necessary to stimulate virus release (73, 74). HIV may take advantage of ⁶³K-linked poly-Ub A3H, which can associate with Gag through an RNA intermediate or directly (75), to promote virus release. This may aid in virus budding and titrate out the effect of the A3H during viral adaptation akin to how Vif mutants unable to degrade A3G can acquire mutations that enable the production of more virus particles to minimize A3G packaging (76, 77). This may occur for A3H since it binds Alu RNA with an apparent K_d that is 3-fold higher than A3G (A3H K_d of 717 ± 29 nM; A3G K_d of 246 ± 46 nM [data not shown]). However, the influence of ⁶³K-linked poly-Ub on A3H restriction activity during a viral infection remains to be tested further.

The exact mechanism determining ⁴⁸K-linked versus ⁶³K-linked poly-Ub was not identified by our study, but our data can exclude some possibilities. The data exclude that the binding strength is a factor in modification of A3 enzymes by ⁶³K-linked ubiquitin chains since both Vif_HXB2 and Vif_IIIB could induce modification of A3H and had a 60-fold difference in their binding affinities (Fig. 3E and 6E). The data also exclude the oligomeric state of the A3 since both Vif_HXB2 and Vif_IIIB could induce a structural change in A3H and A3G (Fig. 2C and D and Fig. 3E and F). Our data are in support of the hypothesis that Vif interacts with A3H and A3B in a manner that is different than A3G. That we could not detect ⁶³K-linked poly-Ub chains on A3G in the presence of MG132 (Fig. 6E) strongly supports the conclusion that the different A3-Vif interfaces of A3H and A3B versus A3G (Fig. 4B and C and Fig. 7C) (21, 22) can affect the structure of the Cul5-E3 ubiquitin ligase and induce recruitment of an E2 capable of catalyzing ⁶³K-linked ubiquitin chains. In the Cullin5 scaffold in which Vif interacts, Rbx2 recruits the E2 and is reported to interact with at least four E2 enzymes, one of which can catalyze ⁶³K-linked ubiquitin chains (nextprot, http://www.nextprot.org/db/entry/NX_Q9UBF6/interactions).

Conclusion.

The results of our study have identified key differences in Vif_IIIB and Vif_HXB2 that determine their efficiency in inducing A3 degradation and identified a novel consequence of the interaction of Vif with A3H and A3B in the absence of degradation. These lab isolates of Vif are useful for studying mechanisms, but further confirmation of our findings with clinical isolates of Vif is required. Nonetheless, the data support a model in which the A3-Vif interaction strength and interface are important for degradation efficiency. The data show that the interaction of Vif with A3H and A3B is more complex than originally identified and highlight the importance of using Vif variants with different degradation profiles when delineating mechanisms of Vif-induced A3 degradation. Altogether, the features of the A3-Vif interactions identified here are important for consideration in the development of HIV-1 therapies that disrupt the A3-Vif interaction.