ABSTRACT
The HIV-1 virion infectivity factor (Vif) targets the cellular cytidine deaminases APOBEC3G (A3G) and APOBEC3F (A3F) for degradation via the host ubiquitin-proteasome pathway. Vif recruits a cellular E3 ubiquitin ligase to polyubiquitinate A3G/F. The activity of Vif critically depends on the cellular core binding factor beta (CBFβ). In this study, we investigated the Vif-CBFβ interaction and the role of CBFβ in the E3 ligase assembly. Vif-CBFβ interaction requires an extensive region of Vif spanning most of its amino terminus and zinc finger region, and cullin 5 (Cul5) binding enhances the stability of the Vif-CBFβ interaction. Our results further demonstrate that CBFβ plays a critical role in facilitating Cul5 binding to the Vif/elongin B/elongin C complex. Vif, with or without bound substrate, is unable to bind Cul5 in the absence of CBFβ. These studies support the notion that CBFβ serves as a molecular chaperone to facilitate Vif-E3 ligase assembly.
IMPORTANCE The host antiviral restriction factors A3G/F inhibit viral replication. The HIV-1 protein Vif targets A3G/F for degradation. This immune evasion activity of Vif is dependent on the cellular factor CBFβ. Multiple regions of Vif are known to be important for Vif function, but the mechanisms are unclear. The studies described here provide important information about the Vif-CBFβ interaction interface and the function of CBFβ in E3 ligase assembly. In particular, our comprehensive Vif-CBFβ interface mapping results help to delineate the role of various Vif regions, determining if they are important for binding CBFβ or A3G/F. Furthermore, our studies reveal an important potential mechanism of CBFβ that has not been shown before. Our results suggest that CBFβ may serve as a molecular chaperone to enable Vif to adopt an appropriate conformation for interaction with the Cul5-based E3 ligase. This study advances our understanding of how CBFβ facilitates the Vif-mediated degradation of APOBEC3 proteins.
INTRODUCTION
The cellular cytidine deaminases APOBEC3G (A3G) and APOBEC3F (A3F) are potent viral restriction factors that hypermutate the viral DNA (1–6) and/or physically block reverse transcription (7–9). The HIV-1 virion infectivity factor (Vif) overcomes this restriction by targeting A3G/F for degradation via the host ubiquitin-proteasome pathway. Vif specifically interacts with a host E3 ubiquitin ligase complex containing the scaffolding protein cullin 5 (Cul5), the adaptor proteins elongin B (EloB) and elongin C (EloC), and the E2-linking protein Rbx2 (5, 10–13). Vif, a 23-kDa protein, assembles the E3 ligase through distinct domains in its C terminus (the Vif C-terminal domain [VifCTD]). Vif binds EloC through its BC box region (residues 144SLQYLA149) (13–16), which mimics the conserved cellular interface of SOCS box proteins. The interaction between Vif and Cul5 is primarily mediated by a novel zinc-binding motif (H108C114C133H139) upstream of the Vif BC box (17–21).
The interaction between Vif and A3G/F occurs primarily through several nonlinear regions within the N terminus of Vif (the Vif N-terminal domain [VifNTD]). While several regions are required for both Vif-A3G and Vif-A3F interactions (22–26), some regions of Vif are required only for the interaction with A3G or with A3F (27). Species specificity for A3G is conferred by 14DRMR17 of Vif, while the same region is necessary for Vif-A3F binding but not species differentiation (27). Degradation of A3G/F is conferred by 21WXSLVK26, 69YXXL72, and 81LGXGXSIEW89; however, 40YRHHY44 is also required for A3G degradation but not A3F degradation (22, 23, 26, 27). These similarities and differences show the versatility of Vif.
An additional cellular binding partner of Vif, core biding factor beta (CBFβ), is required for Vif-mediated APOBEC3 degradation (28, 29). CBFβ forms a heterodimer with RUNX family proteins. These complexes play an essential role in host cell differentiation and development (30, 31). During an HIV-1 infection, the depletion of CBFβ results in the inability of Vif to degrade A3G/F (28, 29) and other Vif-sensitive APOBEC3 proteins (32). This function of CBFβ is achieved through a direct interaction with Vif, leading to increased Vif stability and solubility, inhibition of Vif oligomerization, and enhancement of Vif binding to Cul5 (33–35). The CBFβ-Vif interface involves residues 15 to 126 of CBFβ (35–37). Additionally, the Vif-CBFβ interaction may perturb the transcription of RUNX target genes by sequestering CBFβ away from RUNX proteins (35).
Despite the established importance of CBFβ for Vif function, the mechanism by which CBFβ facilitates Vif-mediated proteasomal degradation of A3G/F is not clear. We carried out biochemical and biophysical studies to determine the role of CBFβ in the assembly of the Vif/E3 ligase complex. Here, we report that the Vif-CBFβ interaction requires an extensive region of the Vif N terminus and the zinc finger region and that the addition of Cul5 increases the stability of the Vif/CBFβ complex. Furthermore, we show that CBFβ is essential to E3 ligase assembly, as in the absence of CBFβ, Vif does not bind Cul5, thus preventing the assembly of the E3 ligase complex.
MATERIALS AND METHODS
Cloning and expression.The Vif sequence from residues 1 to 176 (Vif1-176) was cloned into pMAT9S (38) with an N-terminal 6×His-maltose binding protein (MBP) tag and into pETDuet-1 (Novagen) with an N-terminal 6×His tag. Truncated Vif proteins consisting of Vif residues 1 to 101 (Vif1-101), 4 to 176 (Vif4-176), 5 to 176 (Vif5-176), 14 to 176 (Vif14-176), and 1 to 109 (Vif1-109) were also cloned into pMAT9S. EloB with or without a 6×His tag (residues 1 to 98 and 1 to 118) and EloC (residues 17 to 112) were expressed from pACYCDuet (a gift from Alex Bullock, University of Oxford, Oxford, United Kingdom). CBFβ (residues 1 to 141) was cloned into pCDFDuet (Novagen) with no tag, an N-terminal 6×His tag, an N-terminal glutathione S-transferase (GST) tag, or an N-terminal green fluorescent protein (GFP) tag. Vif (residues 1 to 154) and the human Cul5 N-terminal domain (NTD) were cloned into pETDuet-1 (Novagen) with an N-terminal 6×His tag on Cul5NTD. Mouse Cul5NTD was also cloned into pGEX4T-1 with a thrombin-cleavable N-terminal GST tag. Human Cul5 (residues 1 to 780) and Rbx2 were expressed from pRSFDuet (a gift from John Gross, University of California San Francisco, San Francisco, CA). Vif residues 100 to 154 were cloned into pETDuet-1 (Novagen) with an N-terminal 6×His tag. An A3F C-terminal domain (CTD) with 7 mutated residues, Y196D, C259A, F302K, W310K, Y314A, Q315A, and F363D (A3FCTD7×mut), was cloned into pETDuet-1 (Novagen) with an N-terminal 6×His tag and into pGEX4T-1 with a thrombin-cleavable N-terminal GST tag. The Vif point mutants were made by QuikChange site-directed mutagenesis (Stratagene) using Vif1-176 as the template. Original A3FCTD7×mut was ordered from GeneArt. All proteins were overexpressed in Escherichia coli BL21(DE3) cells at 18°C for 18 h by induction with 0.5 mM isopropyl-β-d-thiogalactopyranoside (IPTG). Vif100-154/EloBC/Cul5, Vif1-154/EloBC/Cul5, and CBFβ/Vif1-154/EloBC/Cul5 were coexpressed as described above.
Protein purification and in vitro ligase assembly.Cells were harvested by centrifugation at 5,000 rpm for 10 min. Cells were resuspended in lysis buffer [50 mM Tris, pH 8.0, 500 mM NaCl, 20 mM imidazole, 0.1 mM Tris(2-carboxyethyl)phosphine (TCEP)] and lysed using a microfluidizer. Cell debris was clarified by centrifugation at 15,000 rpm for 45 min. Individual proteins as well as protein complexes were purified by nickel affinity, anion exchange, and size exclusion chromatography and were analyzed after each step by SDS-PAGE.
To purify the Vif-A3FCTD7×mut complexes, 32 nmol of MBP-Vif4-176/EloBC/6×His-CBFβ was mixed with a 5-fold excess of A3FCTD7×mut (160 nmol). The mixture was run on a preparative S200pg size exclusion column to separate the different complexes. The fractions containing the Vif/EloBC/A3FCTD7×mut complex were collected, concentrated, and run on an analytical S200gl size exclusion column. All peak fractions were analyzed by SDS-PAGE.
Analysis of binding of Vif-A3F complex to Cul5NTD.The Vif/EloBC/A3FCTD7×mut complex was mixed with stoichiometric amounts of Cul5NTD, and binding between the two was tested by size exclusion chromatography. Peak fractions were analyzed by SDS-PAGE.
Size exclusion chromatography analysis of CBFβ dissociation.MBP-Vif4-176/EloBC/6×His-CBFβ and Cul5/Rbx2 were expressed, harvested, and purified as described above. Aliquots of 100 μl of 30 μM MBP-Vif4-176/EloBC/6×His-CBFβ were left at room temperature. Aliquots were run on an S200gl size exclusion column each day for 2 days. The content of each peak was evaluated by SDS-PAGE. Equal molar amounts of MBP-Vif4-176/EloBC/6×His-CBFβ and Cul5/Rbx2 were mixed, and the 6-component complex was purified by size exclusion chromatography. Aliquots of 50 μl of 68 μM MBP-Vif4-176/EloBC/6×His-CBFβ/Cul5/Rbx2 were left at room temperature. Aliquots were run on an S200gl size exclusion column after 1, 2, 3, and 8 days of incubation. The content of each peak was evaluated by SDS-PAGE.
Ni-NTA-GST pulldown.GST-CBFβ and 6×His-MBP-Vif constructs were coexpressed and harvested as described above. Cell lysate was loaded onto a 2-ml or 5-ml GST gravity-flow column (GE Healthcare). The column was washed with 30 column volumes of lysis buffer and subsequently eluted with 50 mM Tris, pH 8.0, 150 mM NaCl, and 10 mM glutathione. Elution fractions were then loaded onto a 0.2-ml gravity-flow Ni-nitrilotriacetic acid (NTA) column (Qiagen), washed with 25 column volumes of lysis buffer, and subsequently eluted with 5 column volumes of Ni-NTA elution buffer (50 mM Tris, pH 8.0, 500 mM NaCl, 400 mM imidazole, 0.1 mM TCEP). The flowthrough, wash, and elution fractions were analyzed by SDS-PAGE.
MBP pulldown.6×His-CBFβ and 6×His-MBP-Vif constructs were coexpressed and harvested as described above. Lysate was loaded onto an Ni-NTA column, and bound proteins were eluted with Ni-NTA elution buffer. The Ni-NTA elution was then loaded onto a 1-ml MBP affinity column (GE Healthcare), washed with 5 column volumes of lysis buffer, and eluted with MBP elution buffer (50 mM Tris, pH 8.0, 150 mM NaCl, 10 mM maltose). The MBP flowthrough and elution fractions were analyzed by SDS-PAGE.
GST pulldown.GST-Cul5 (1.5 nmol) were bound to a 0.2-ml GST gravity-flow column. The column was washed with 20 column volumes of lysis buffer. For each Vif complex (Vif1-154/EloBC, Vif1-176/EloBC/CBFβ or Vif1-154/EloBC/CBFβ), 1.1 nmol was loaded onto the GST column with bound GST-Cul5. The column was then incubated for 2 h before washing with 25 column volumes of lysis buffer. Each complex was eluted with 50 mM Tris, pH 8.0, 150 mM NaCl, 10 mM glutathione. Flowthrough, wash, and elution samples were analyzed by SDS-PAGE.
Plasmids for in vivo analysis of Vif Zn finger mutants.Wild-type A3G-hemagglutinin (HA)- and NL4-3ΔVif-coding sequences were previously reported (29). The full-length Vif from NL4-3 (residues 1 to 192) was constructed with an N-terminal HA tag and inserted between the EcoRI and BamHI sites of VR1012. The Vif C114S and C133A mutant plasmids were made from wild-type (WT) Vif plasmid by site-directed mutagenesis.
Cell lines for in vivo analysis of Vif Zn finger mutants.HEK293T and MAGI-CCR5 (AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, National Institutes of Health) cells were maintained in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and penicillin-streptomycin (D-10 cell medium).
Antibodies for in vivo analysis of Vif Zn finger mutants.The following antibodies were used: anti-Cul5(H-300) rabbit polyclonal antibody (catalog no. sc-13014; Santa Cruz), CAp24 monoclonal antibody (catalog no. 1513; AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, National Institutes of Health), anti-HA rabbit polyclonal antibody (catalog no. 71-5500; Santa Cruz), anti-CBFβ monoclonal antibody (catalog no. sc-166142; Santa Cruz), anti-elongin B (FL-118) rabbit polyclonal antibody (catalog no. sc-1144; Santa Cruz), and anti-β-tubulin monoclonal antibody (catalog no. NMS-410P; Covance).
Coimmunoprecipitation assay.HEK293T cells were transfected with 2 μg of HA-tagged wild-type Vif, HA-tagged Vif Zn finger mutants, or the VR1012 vector as a negative control. Cells were harvested 48 h later and subjected to immunoprecipitation analysis using the anti-HA antibody conjugated to agarose beads. Harvested cells was resuspended in lysis buffer (50 mM Tris-HCl [pH 7.5], 150 mM NaCl, 0.5% [vol/vol] NP-40, and Complete protease inhibitor cocktail tablets), incubated at 4°C for 30 min, and then centrifuged at 13,000 rpm for 30 min. Cell lysates were mixed with anti-HA antibody-conjugated agarose beads (Roche) and incubated at 4°C for 3 h. Samples were then washed 6 times with washing buffer (20 mM Tris-HCl, pH 7.5, 100 mM NaCl, 0.1 mM EDTA, 0.05%[vol/vol] Tween 20). Proteins bound to the beads were eluted with 50 μl of elution buffer (0.1 mM glycine-HCl, pH 2.0) or 4× loading buffer containing 0.08 M Tris, pH 6.8, 2.0% SDS, 10% glycerol, 0.1 M dithiothreitol, and 0.2% bromophenol blue. The eluted proteins were subsequently analyzed by Western blotting.
Virus purification, viral infectivity (MAGI) assay, and A3G degradation.For APOBEC3 (A3G or A3F) degradation and viral infectivity assay, HEK293T cells were transfected with 0.5 μg of NL4-3ΔVif, 0.1 μg of VR1012, or wild-type or mutant Vif and 0.5 μg of A3G-HA in six-well plates. Empty vector VR1012 was used as a negative control. Two days after transfection, cell culture medium was harvested. Empty vector VR1012 was used as a negative control. Cells were reserved for detection of A3G degradation. Virus in cell culture medium was disposed of by centrifugation and filtration through a 0.22-μm-pore-size membrane (Millipore). Virus particles were concentrated by ultracentrifugation. The viral pellets resuspended in lysis buffer (phosphate-buffered saline [PBS] containing 1% Triton X-100 and Complete protease inhibitor cocktail [Roche]) were analyzed by Western blotting.
Viral particle infection.MAGI-CCR5 cells were prepared in 24-well plates in 0.5 ml of D-10 cell medium before infection. Cells at 30 to 40% confluence were infected by removing the medium from each well and adding dilutions of virus in a total volume of 500 μl of complete DMEM with 20 μg of DEAE-dextran per well. After a 2-h infection, 500 μl of complete DMEM was added to each well, and the cells were incubated for 48 h at 37°C in a 5% CO2 incubator. Five hundred microliters of fixing solution (1% formaldehyde, 0.2% glutaraldehyde in PBS) was added after the supernatants were removed. The cells were washed twice with PBS after a 5-min incubation. The staining solution (20 μl of 0.2 M potassium ferrocyanide, 20 μl of 0.2 M potassium ferricyanide, 2 μl of 1 M MgCl2, 10 μl of 40-mg/ml 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside [X-Gal]) was added. Cells were incubated for 2 h at 37°C in a non-CO2 incubator. Staining was stopped by removal of the staining solution, and the cells were thoroughly washed twice with PBS. β-Galactosidase activity is under the control of the HIV-1 long terminal repeat promoter, which is trans-activated in this system; blue dots indicate the presence of integrated virus. The number of blue dots was counted, and viral infectivity was determined after normalizing the amount of input virus in terms of the p24 antigen content. All results represent those for infections done in triplicate.
FCS.Fluorescence correlation spectroscopy (FCS) measurements were made on a homebuilt system centered on an inverted IX-71 microscope (Olympus) as previously described (39). A continuous-emission 488-nm DPSS 50-mW laser (Coherent, Santa Clara, CA) was set to a 5- to 20-mW output power and further adjusted with neutral-density filters to 5 μW power just prior to entering the microscope. Fluorescence was collected through the objective and separated from the excitation laser using a Z488rdc long-pass dichroic filter and a HQ600/200m band-pass filter (Chroma, Bellows Falls, VT). Fluorescence was focused onto the aperture of a 50-μm optical fiber coupled to an avalanche photodiode (PerkinElmer, Waltham, MA). A digital correlator (Flex03LQ-12; Correlator.com, Bridgewater, NJ) was used to generate the autocorrelation curves. Measurements were made in chambered cover glasses (Nunc, Rochester, NY) that were polyethylene glycol treated to prevent protein adsorption, as described previously (39). All measurements were at 20.5 ± 0.5°C. Photon traces were collected continuously, and autocorrelation curves were calculated for every 200 s of measurement. The curves were fit, using lab-written software in Matlab, to an equation for two-component diffusion (equation 1), where the diffusion time of each component was fixed on the basis of measurements and fitting of each individual component to an equation for single-component diffusion (equation 2). The diffusion time of free GFP-CBFβ was calculated from measurements of free purified protein, and the diffusion time of the GFP-CBFβ/MBP-Vif4-176/EloBC complex was measured immediately after the complex was separated from free protein by size exclusion chromatography. s accounts for the radial-to-axial dimensions of the focal volume and was calculated from fits to free Alexa 488. This allowed us to extract the fraction (F) of GFP-CBFβ diffusing freely and as part of the GFP-CBFβ/MBP-Vif4-176/EloBC complex.
(1)
(2)where G(t) is the autocorrelation function, N is the number of molecules, t1 is the diffusion time of component 1, s accounts for the radial-to-axial dimensions of the focal volume and was calculated from fits to free Alexa 488 dye, and t2 is the diffusion time of component 2.
RESULTS
Multiple regions in the N terminus of Vif are critical for interaction with CBFβ.Several regions within the Vif N terminus are required for A3G and/or A3F degradation. As the Vif-CBFβ interaction is essential for A3G/F degradation, we tested whether any of these previously identified regions within Vif are required for Vif-CBFβ binding (Fig. 1A). We utilized double-pulldown experiments with 6×His-MBP-Vif constructs and a GST-tagged CBFβ to assess the ability of the Vif mutants to bind CBFβ (Fig. 1B).
Multiple residues in VifNTD are critical for interaction with CBFβ. (A) Diagram of the domain structure of Vif showing the N-terminal point mutations in the mutants used in the mapping experiments. The Vif N terminus that interacts with A3G/F is shown in gray. (B) Double pulldown of various 6×His-MBP-tagged Vif mutants with GST-tagged CBFβ. The SDS-polyacrylamide gels were stained with Coomassie blue. (C) Size exclusion chromatograms of purified MBP-Vif1-176 alone, CBFβ alone, and the binary MBP-Vif1-176/CBFβ complex and their corresponding SDS-polyacrylamide gels. (D) Size exclusion chromatograms and SDS-polyacrylamide gels of purified MBP-Vif mutants that bound CBFβ. (E) Affinity chromatograms of Vif mutants that do not bind CBFβ. MBP-Vif1-176 (G84 SIEW/4A) and MBP-Vif1-176 (EW88AA) bind an MBP affinity column, while CBFβ flows through. An SDS-polyacrylamide gel of flowthrough and elution samples is shown. Lanes M, molecular mass markers, where the numbers on the left are molecular masses (in kDa). Lanes 1 to 3 correspond to the elution peaks marked in the chromatograms. mAu, milli-absorbance units.
The Vif mutants tested have various binding affinities for CBFβ. For comparison, we first examined the binding of WT Vif residues 1 to 176 (Vif1-176) to CBFβ. The WT Vif-CBFβ interaction was detected using a stringent double-pulldown experiment and was observed as a 1:1 stoichiometric complex, as judged by their band intensities on SDS-PAGE. Of the mutants tested, only Vif 40YRHHY44/5A retained wild-type binding affinity, while two Vif mutants, Vif W21A/W38A and Vif EW88AA, exhibited significantly reduced CBFβ binding. The Vif G84A 86SIEW89/4A mutant presented the most striking effect, with a complete loss of Vif-CBFβ binding. These results suggest that G84 and 86SIEW89 are critical for the Vif-CBFβ interaction, while W21 and W38 also play an important role in the interaction (Fig. 1B).
We further inspected the formation of Vif/CBFβ complexes using chromatographic techniques. We coexpressed 6×His-MBP-tagged WT Vif or its mutants with 6×His-tagged CBFβ and used Ni-NTA, anion exchange, and size exclusion chromatography to purify the complex (Fig. 1CD). We also purified MBP-Vif1-176 or 6×His-CBFβ individually as controls. MBP-Vif1-176 eluted off a size exclusion column at a position corresponding to a large heterogeneous aggregation (Fig. 1C). CBFβ, as expected, eluted as a monodispersed species at a position corresponding to its molecular mass, 18 kDa (Fig. 1C). The WT MBP-Vif1-176/CBFβ complex eluted from a size exclusion column as a monodispersed species at a position corresponding to a molecular mass of a 1:1 MBP-Vif/CBFβ complex (∼79 kDa) (Fig. 1C). Consistent with the double-pulldown assay results, Vif1-176 40YRHHY44/5A also formed a monodispersed, 1:1 complex with CBFβ (Fig. 1D). Small amounts of a Vif1-176 W21A/W38A/CBFβ complex were also detected (Fig. 1D), confirming that the W21A/W38A mutation significantly reduces but does not abolish Vif-CBFβ binding, consistent with the results of the double-pulldown experiment.
The mutant Vif constructs that showed severely impaired CBFβ binding in the pulldown experiment could not be copurified with CBFβ in vitro. These Vif mutants included Vif1-176 EW88AA and Vif1-176 G84A 86SIEW89/4A. While the MBP-Vif mutants bound the MBP column, 6×His-CBFβ flowed through (Fig. 1E). These data confirm that Vif residues G84 86SIEW89 are critical for binding to CBFβ. These residues have previously been shown to be essential for HIV infectivity (22). Our data suggest that the functional importance of these residues may be their role in maintaining the interaction with CBFβ.
Vif residues 5 to 14 are required for CBFβ binding.We further investigated the importance of the Vif N terminus for interaction with CBFβ. We created Vif N-terminal truncation constructs Vif4-176 (residues 4 to 176), Vif5-176 (residues 5 to 176), and Vif14-176 (residues 14 to 176) (Fig. 2A). We used the double-pulldown and copurification assays described above to test their ability to bind CBFβ. The results from the copurification and pulldown assays showed that Vif4-176 and Vif5-176 behave like wild-type Vif and form a monodispersed, 1:1 complex with CBFβ (Fig. 2B and C). In contrast, the results of both assays demonstrated that Vif14-176 is unable to bind CBFβ (Fig. 2C and D). These results indicate that residues between Vif residue 5 and Vif residue 14 are critical for the Vif-CBFβ interaction, either by directly participating in the interaction or by playing an important structural role in enabling the interaction from other regions in Vif.
Vif residues 5 to 14 are critical for the interaction with CBFβ. (A) Diagram of the domain structure of Vif showing the N-terminal truncation constructs used in the mapping experiments. The Vif N terminus that interacts with A3G/F is shown in gray. (B) Size exclusion chromatogram and SDS-polyacrylamide gel of the purified MBP-Vif4-176/CBFβ complex. (C) Double pulldown of 6×His-MBP-Vif5-176 and 6×His-MBP-Vif14-176 with GST-tagged CBFβ. The SDS-polyacrylamide gels were stained with Coomassie blue. MBP-Vif5-176 retained wild-type-like binding to CBFβ, while MBP-Vif14-176 lost the ability to bind CBFβ. (D) MBP-Vif14-176 does not interact with CBFβ, as it binds an MBP affinity column, while CBFβ flows through. An SDS-polyacrylamide gel with flowthrough and elution samples is shown. Lanes M, molecular mass markers, where the numbers on the left are molecular masses (in kDa); B, percent elution buffer.
Vif residues 102 to 109 are critical for CBFβ binding, and the remainder of the zinc finger strengthens the interaction.To further map the CBFβ-Vif interface, we examined Vif constructs containing C-terminal truncations and mutations (Fig. 3A). As Vif1-140 has been shown to retain the WT capacity for binding to CBFβ (34), we investigated if the Vif zinc finger domain (residues 100 to 140) is required for the interaction. Our results indicated that the absence of the Vif zinc finger abolishes the Vif-CBFβ interaction, as little Vif1-101 was recovered from the double pulldown with CBFβ (Fig. 3C). Similarly, Vif1-101 could not be purified with CBFβ in vitro, as MBP-Vif1-101 bound the MBP column and 6×His-CBFβ flowed through, despite the significant overexpression of CBFβ (Fig. 3D). These results demonstrate a requirement for Vif zinc finger residues 100 to 140 for CBFβ binding.
Vif residues 102 to 109 are critical for the interaction with CBFβ, and the remainder of the zinc finger enhances the interaction. (A) Diagram of the domain structure of Vif showing the C-terminal truncation and mutation constructs used in the mapping experiments. The Vif N terminus that interacts with A3G/F is shown in gray. (B) Size exclusion chromatogram and SDS-polyacrylamide gel of the purified Vif1-109/CBFβ complex. (C) Double pulldown of various 6×His-MBP-tagged Vif C-terminal mutants and truncations with GST-tagged CBFβ. The SDS-polyacrylamide gels were stained with Coomassie blue. The mutations and truncations significantly impaired the ability of Vif to bind CBFβ. (D) MBP-Vif1-101 does not interact with CBFβ, as it binds an MBP affinity column, while CBFβ flows through. An SDS-polyacrylamide gel with flowthrough and elution samples is shown. Lanes M, molecular mass markers, where the numbers on the left are molecular masses (in kDa). The unlabeled lane and lanes 1 and 2 correspond to the elution peaks marked in the chromatograms.
Interestingly, we could purify a small fraction of 6×His-Vif1-109 and CBFβ expressed in E. coli as a monodispersed complex eluting off a size exclusion column at a volume corresponding to the molecular mass of the binary complex (∼25 kDa) (Fig. 3B). Furthermore, the results of the double-pulldown experiment showed that Vif1-109 has an intermediate affinity for CBFβ that is significantly less than that of wild-type Vif1-176 but substantially greater than that of Vif1-101 (Fig. 3C). We then investigated what residues in Vif101-109 are involved in the binding of CBFβ. We again used the double-pulldown experiment to test the ability of two Vif mutants, 102LAD104/ANA and 105QLI107/AAA, to interact with CBFβ. Both triple mutations are detrimental to binding to CBFβ (Fig. 3C), confirming the importance of this region in the interaction. Our results provide a potential mechanistic explanation of the previous observation that Vif 105QLI107/AAA abolished HIV-1 infectivity (40).
As discussed above, Vif1-109 retained the ability to bind CBFβ, but with a reduced affinity compared to that of Vif1-176 (Fig. 3C) and Vif1-140 (34). These results indicate that Vif residues 110 to 140 contribute to Vif-CBFβ binding. This region belongs to a noncanonical zinc finger of Vif (residues 101 to 140) important for Cul5 binding (17–21). Thus, we sought to investigate the involvement of two of the zinc coordinating residues of Vif, C114 and C133, in CBFβ binding in vitro and in cell-based studies (Fig. 4). The in vitro results show that MBP-VifC114S and MBP-VifC133S are able to be copurified with 6×His-CBFβ (Fig. 4B). These results indicate that mutation of the two conserved Zn-coordinating cysteines does not abolish CBFβ binding, consistent with our purification results obtained with the Vif1-109/CBFβ complex (Fig. 3B). We further used in vivo cell-based studies to assess whether the Zn finger mutations decreased CBFβ's affinity for Vif. Our coimmunoprecipitation assays showed that VifC114S and VifC133A have a slightly impaired ability to bind CBFβ compared to that of wild-type Vif (Fig. 4C). This study also assessed the Vif Zn finger mutant's ability to bind Cul5. Consistent with multiple published studies, mutation of Vif C114 and C133 abolished coimmunoprecipitation with Cul5 (Fig. 4C) (18, 20). VifC114S and VifC133A are defective at degrading A3G (Fig. 4D), and viruses containing the C114 and C133 Vif mutations are defective in infection (Fig. 4E). This is all consistent with the notion that a defective Vif-E3 ligase assembly prevents A3G/F degradation. Taken together, our results suggest that the overall zinc finger geometry helps to maintain the Vif-CBFβ interaction, likely through a critical interface involving Vif residues 102 to 109.
The zinc finger contributes to the interaction with CBFβ. (A) Diagram of the domain structure of Vif showing the zinc finger mutant constructs used in mapping experiments. The Vif N terminus that interacts with A3G/F is shown in gray. (B) Size exclusion chromatograms and SDS-polyacrylamide gels of the purified MBP-Vif1-176 Zn finger mutants bound to CBFβ. Lanes M, molecular mass markers, where the numbers on the left are molecular masses (in kDa). The unlabeled lanes correspond to the elution peaks in the chromatograms. (C) Effects of mutations in the Zn finger region on Vif function in vivo. Coprecipitation assay with HA-tagged Vif (wild type and Zn finger mutants) testing the ability of Vif constructs to interact with cellular factors in vivo. HEK293T cells were transfected with the wild type or Vif mutants with the HA tag and the VR1012 vector as a negative control. Cells were harvested 48 h later and subjected to immunoprecipitation (IP) analysis using the anti-HA antibody conjugated to agarose beads. Coprecipitated proteins were analyzed by Western blotting. (D) Effects of wild-type Vif and Vif Zn finger mutants on A3G degradation and virion packaging. HEK293T cells were transfected with NL4-3ΔVif plus A3G in the presence of WT or mutant Vif, as indicated. A3G stability and packaging were assessed by Western blotting. hVif, human Vif. (E) Effect of wild-type Vif and Zn finger mutants on antiviral activity of A3G. HIV-1 strains were produced as described for panel D. Virus infectivity was assessed using MAGI-CCR5 indicator cells, and the virus infectivity in the presence of wild-type Vif was set to 100%. Error bars represent the standard deviations from triplicate wells.
CBFβ allows the assembly of a Vif-Cul5 E3 ligase.We investigated the role of CBFβ in the interaction between Vif and Cul5. The inclusion of CBFβ allows Vif to interact with the E3 ligase complex containing Cul5 and EloBC, enabling the purification of a complex containing all 5 proteins. We first used an N-terminal construct of Cul5 (Cul5NTD, residues 1 to 384), as it has been established that Cul5NTD contains all the Vif-interacting elements of the protein (18, 20). When coexpressed, the Vif1-154/EloBC/Cul5NTD/CBFβ complex can be stably purified as a stoichiometric complex (Fig. 5A). The 5 components elute off an anion exchange column together and can be further purified by gel filtration chromatography (Fig. 5A). Consistent with this result, our pulldown experiment showed that GST-Cul5NTD coelutes with Vif1-176/EloBC/CBFβ (Fig. 5A) and Vif1-154/EloBC/CBFβ (data not shown). We also analyzed binding between full-length Cul5/Rbx2 and MBP-Vif1-176/EloBC/CBFβ. Consistent with the results obtained with Cul5NTD, we were able to purify a complex containing all six components, MBP-Vif1-176, EloBC, CBFβ, Cul5, and Rbx2 (Fig. 5B). These results confirm that in the presence of CBFβ, Vif can assemble into the Cul5-containing E3 ligase.
CBFβ is critical for full-length Vif to bind Cul5. (A) Schematic of the complex (Vif/EloBC/CBFβ/Cul5NTD) formed in the experiments (top left), SDS-polyacrylamide gel of a GST pulldown showing that Vif1-176/CBFβ/EloBC binds GST-Cul5NTD (bottom left), and size exclusion chromatogram and SDS-polyacrylamide gel of the purified Vif1-154/CBFβ/EloBC/Cul5NTD complex (right). (B) Schematic of the complex formed (Vif/EloBC/CBFβ/Cul5/Rbx2) in the experiment (left) and size exclusion chromatogram and SDS-polyacrylamide gel of the purified MBP-Vif1-176/CBFβ/EloBC/Cul5/Rbx2 complex (right). (C) Schematic of results showing that Vif/EloBC does not bind Cul5NTD without CBFβ (top left), and SDS-polyacrylamide gel of a GST pulldown assay showing that Vif1-154/EloBC does not interact with GST-Cul5NTD (bottom left). A chromatogram and SDS-polyacrylamide gel showing that Cul5NTD separates from Vif1-154/EloBC on an anion exchange column are shown (right). (D) Schematic of results showing that the Vif/A3F/EloBC complex does not bind Cul5NTD without CBFβ (top). (Bottom left) Size exclusion chromatograms of the purified MBP-Vif4-176/EloBC/A3FCTD complex and purified Cul5NTD; (bottom right) size exclusion chromatogram showing that MBP-Vif4-176/EloBC/A3FCTD does not bind Cul5NTD. They elute as two distinct peaks on a size exclusion column. SDS-polyacrylamide gels of the peaks are shown. Lanes M, molecular mass markers, where the numbers on the left are molecular masses (in kDa); lanes FT, GST flowthrough; lanes E, GST eluate. Lanes 1 and 2 correspond to the elution peaks marked in the chromatograms.
Vif does not assemble into the Cul5-containing E3 ligase in the absence of CBFβ.We next investigated the interaction between Vif and Cul5 in the absence of CBFβ. The results showed that Vif1-154, which is able to bind Cul5 in the presence of CBFβ (Fig. 5A), does not bind Cul5 without CBFβ (Fig. 5C). Vif1-154 retained the ability to form a stable complex with EloBC; however, it did not associate with Cul5NTD (Fig. 5C) in the absence of CBFβ. Specifically, the Vif1-154/EloBC complex clearly separated from Cul5NTD on an anion exchange column (Fig. 5C), even though the Vif1-154/EloBC subcomplex had good solution behavior and existed in a monomeric, monodispersed form, as shown by size exclusion chromatography (data not shown). We further used GST affinity chromatography to show that GST-Cul5NTD is unable to pull down Vif1-154/EloBC, confirming that a Vif/Cul5 complex is not formed (Fig. 5C). These data clearly demonstrate that Vif requires CBFβ to bind Cul5.
The Vif/EloBC/substrate complex cannot bind Cul5 in the absence of CBFβ.We purified a Vif/substrate complex, Vif/EloBC/A3FCTD, to test the influence of the APOBEC3 substrate on the ability of Vif to interact with Cul5 (Fig. 5D). We incubated A3FCTD (residues 185 to 373), which contains the Vif-interacting site of A3F, with purified MBP-Vif4-176/EloBC/CBFβ to assemble the substrate complex of Vif in vitro. Interestingly, this allowed us to purify a monodispersed MBP-Vif4-176/EloBC/A3FCTD complex to homogeneity (Fig. 5D), however, without CBFβ. To test the binding of this Vif/substrate complex to Cul5, we also purified Cul5NTD alone to homogeneity as a monodispersed species (Fig. 5D).
The Vif/substrate complex without CBFβ did not assemble with Cul5 in vitro (Fig. 5D). We incubated the purified MBP-Vif4-176/EloBC/A3FCTD complex with Cul5NTD and examined their interaction using size exclusion chromatography. No interaction was detected, as MBP-Vif4–176/EloBC/A3FCTD and Cul5NTD eluted as two distinct peaks off the size exclusion column at positions corresponding to those of the two individual inputs before mixing (Fig. 5D). This result suggests that in the absence of CBFβ, even if a Vif/substrate complex is formed, it is unable to assemble with Cul5.
The C-terminal half of Vif (VifCTD) associates with Cul5 in the absence of CBFβ.Interestingly, our data, as well as others, clearly show that the C-terminal half of Vif (VifCTD) can assemble with Cul5 in the absence of CBFβ (18, 41). When C-terminal Vif constructs (residues 100 to 154) were coexpressed with E3 ligase components EloBC and Cul5NTD, a monodispersed, monomeric complex containing all four proteins could be purified (Fig. 6) (41). It has been demonstrated previously that the affinity between VifCTD/EloBC and Cul5 is ∼90 nM, whereas the affinity between EloBC and Cul5 is ∼7 μM (33, 35, 41). The interaction between EloBC and Cul5 is significantly weaker than that between VifCTD/EloBC and Cul5, indicating that VifCTD is responsible for the majority of the VifCTD/EloBC interaction with Cul5 (41). This, together with the observation that full-length Vif cannot bind Cul5 without CBFβ (Fig. 5CD), suggests that the N terminus of Vif prevents the binding of Cul5 to Vif and that the addition of CBFβ relieves this Vif N-terminal inhibition.
The C-terminal half of Vif (VifCTD) binds Cul5 in the absence of CBFβ. (A) Schematic of the VifCTD/EloBC/Cul5NTD complex. (B) Size exclusion chromatogram and SDS-polyacrylamide gel of the purified Vif100-154/EloBC/Cul5NTD complex. Lane M, molecular mass markers, where the numbers on the left are molecular masses (in kDa). Lane 1 corresponds to the elution peak in the chromatogram.
The addition of Cul5 stabilizes the CBFβ-Vif interaction in vitro.The lack of a Vif/substrate/CBFβ complex in the experiment described above prompted us to investigate the stability of the Vif-CBFβ interaction in vitro. We used fluorescence correlation spectroscopy (FCS) and size exclusion chromatography to analyze the dissociation of CBFβ from the remainder of the Vif/EloBC/CBFβ complex. FCS measures fluctuations in intensity arising from diffusion of fluorescent molecules through an experimental observation volume (focal volume) (Fig. 7A). The transit time of the molecules reflects their size; smaller molecules diffuse more rapidly, while larger molecules diffuse more slowly. It is therefore a powerful technique to detect the formation or dissociation of molecular complexes. Here we used FCS to observe the dissociation of CBFβ tagged with GFP from the GFP-CBFβ/MBP-Vif4-176/EloBC complex over time. At the initial time point, all the GFP-CBFβ was bound to Vif/EloBC, reflected by a long diffusion time. Over time, the average diffusion time of the sample decreased, reflecting contributions both from the GFP-CBFβ/MBP-Vif4-176/EloBC complex and from free GFP-CBFβ, which diffused much more rapidly. Over 5 h, more than 40% of the starting complex dissociated to free GFP-CBFβ and MBP-Vif/EloBC (Fig. 7A). Consistent with this observation, our size exclusion chromatography analysis showed the breakdown of the majority of the complex in 1 to 2 days at room temperature (Fig. 7B).
The CBFβ-Vif interaction is unstable in vitro, and the addition of Cul5 stabilizes the E3 ligase. (A) Schematic of the FCS setup (top) and the resulting measurements showing GFP-CBFβ/MBP-Vif4-176/EloBC complex dissociation over a 5-h assay time (bottom). (B) Size exclusion chromatogram of the purified MBP-Vif4-176/CBFβ/EloBC complex after incubation at room temperature (RT) for 0 (original), 1, or 2 days. An SDS-polyacrylamide gel of each peak is shown. A schematic of the complex analyzed is shown on the top. This complex is not stable. (C) Size exclusion chromatography of the purified MBP-Vif4-176/CBFβ/EloBC/Cul5/Rbx2 complex after incubation at room temperature for 0 (original), 3, or 8 days. An SDS-polyacrylamide gel of each peak is shown. A schematic of the complex analyzed is shown on the top. Increased stability was observed with the addition of Cul5/Rbx2. Lanes 1 to 3 correspond to the elution peaks marked in the chromatograms. Lanes M, molecular mass markers, where the numbers on the right are molecular masses (in kDa).
The addition of Cul5 and Rbx2 to Vif/EloBC/CBFβ resulted in a significantly more stable complex in vitro. We monitored the dissociation of the purified MBP-Vif4-176/EloBC/CBFβ/Cul5/Rbx2 complex using size exclusion chromatography. In contrast to the observation presented above, the results showed a much more stable complex, with a significant fraction remaining intact even after 1 week at room temperature (Fig. 7C). These observations are consistent with those of a study that found that Vif/EloBC/CBFβ/Cul5/Rbx2 is less prone to precipitation than the Vif/EloBC/CBFβ complex (34). These in vitro stability tests on purified complexes suggest that the Vif-CBFβ interaction may be intrinsically unstable and that it can be enhanced by the E3 ligase scaffolding protein Cul5.
DISCUSSION
It has recently been established that CBFβ plays a critical role in Vif-mediated A3G/F degradation. However, the mechanisms by which CBFβ facilitates A3G/F depletion remain unclear. The studies described herein provide important information about the function of CBFβ in E3 ligase assembly as well as details of the Vif-CBFβ interaction interface. In particular, multiple regions of Vif are known to be important to Vif function, but the mechanisms are unclear. Our comprehensive Vif-CBFβ interface mapping results help to delineate the role of various Vif regions and provide a detailed biochemical foundation for understanding how Vif interacts with CBFβ. Furthermore, our studies on Vif-Cul5 E3 ligase assembly reveal an important potential mechanism of CBFβ that has not been shown before, namely, that it serves as a molecular chaperone to shape Vif in the appropriate conformation to interact with Cul5.
Our mapping studies showed that Vif interacts with CBFβ using an extended interface encompassing most of the viral protein which overlaps with regions implicated in its interaction with other cellular binding partners (Fig. 8A). Regions in the N terminus of Vif are known to be the interaction sites for the A3G/F proteins. Two of these regions were also found to be essential for CBFβ binding in our experiments. For example, 84GXSIEW89 was critical for the Vif-CBFβ interaction (Fig. 1B and E), consistent with in vivo data illustrating that the mutation of 86SIEW89 inhibited A3G/F degradation (22). Similarly, mutation of Vif residues W21/W38 significantly reduced the affinity for CBFβ (Fig. 1B and D), consistent with in vivo studies showing that these residues are important for the interaction with CBFβ (29) and the degradation of A3G/F (23, 25). It is likely that the importance of these Vif regions in APOBEC3 degradation is due to their roles in the Vif-CBFβ interaction, rather than for Vif A3G/F binding. On the other hand, our studies showed that Vif residues 40YRHHY44 are not involved in the Vif-CBFβ interaction, despite their established role in A3G degradation (Fig. 1B and D) (27), indicating that these residues compose a bona fide A3G interaction site. Therefore, our mapping studies help to delineate the specific roles of Vif residues in viral evasion of the APOBEC3 proteins.
Vif-cellular protein interfaces and role of CBFβ in E3 ligase assembly. (A) Schematic of the Vif-CBFβ interaction (top) and critical Vif residues required for the binding (bottom). Vif residues involved in the interactions with other cellular binding partners of Vif are marked. (B) Schematics showing that the full-length Vif does not bind Cul5 in the absence of CBFβ.
Besides the essential N-terminal regions, the C-terminal zinc finger region of Vif is also important for CBFβ binding. Specifically, the first 8 residues of the zinc finger region (residues 102 to 109) are critical for interaction with CBFβ, while the remainder of the zinc finger (residues 110 to 140) also contributes to the Vif-CBFβ interaction (Fig. 3). Previous studies showed that the first 140 residues of Vif can interact with CBFβ (34). Our study further shows that deletion of the zinc finger domain prevents the reconstitution of the Vif/CBFβ complex in vitro (Fig. 3CD). The addition of residues at the beginning of the zinc finger domain (residues 102 to 109) allowed the purification of Vif/CBFβ, albeit with a reduced affinity compared to that of Vif with an intact zinc finger (Fig. 3BC). In vivo coimmunoprecipitation studies showed that mutation of the two critical Vif Zn finger residues (C114/C133) slightly decreased CBFβ binding to Vif (Fig. 4C). These experiments indicate that the zinc finger domain of Vif is involved in CBFβ binding, with residues 102 to 109 playing a critical role.
Our experiments demonstrate that CBFβ is required for the assembly of the Vif/EloBC/Cul5 E3 ligase, specifically, for Cul5 binding to Vif (Fig. 5AB). In the absence of CBFβ, full-length Vif is able to assemble with the adaptor proteins EloBC but is unable to bind Cul5 (Fig. 5C). Interestingly, the C terminus of Vif is capable of binding both EloBC and Cul5 (Fig. 6). This suggests that in the absence of CBFβ, VifNTD likely adopts a conformation preventing Cul5 association and that the binding of CBFβ alters the Vif structure to prime it for Cul5 binding. These observations are consistent with in vivo studies demonstrating that silencing CBFβ inhibits the interaction between Vif and Cul5 (29) and that mutation in Vif 86SIEW89, a region critical for CBFβ association (Fig. 1B and E), eliminates Cul5 binding (22). CBFβ presumably enables both VifNTD and VifCTD to participate in the interaction with Cul5. This is in agreement with studies that show that in the presence of CBFβ, full-length Vif has an affinity to Cul5 (33, 35) higher than that provided by VifCTD (41). These results support the notion that CBFβ serves as a molecular chaperone, enabling Vif to assemble the Cul5-containing E3 ubiquitin ligase (Fig. 8B).
Taken together, the extensive interaction between Vif and CBFβ may serve to increase the solubility and stability of Vif, as well as to shape the viral protein in a conformation that facilitates the assembly of the Cul5-containing E3 ubiquitin ligase. The fully assembled Vif/EloBC/CBFβ/Cul5/Rbx2 E3 ligase exhibits further increased stability in comparison to that of its constituents without Cul5. Significant data from other studies and our studies have ruled out a direct interaction between Cul5 and CBFβ (28, 29, 33); thus, the added stability is likely a result of Cul5 stabilizing Vif. It has been proposed that CBFβ functions to stabilize Vif in vivo (28, 35). Our study demonstrates that, in addition to this stabilization role, CBFβ has a more fundamental function in serving as a molecular chaperone that enables the formation of the Vif/Cul5 complex essential for A3G/F degradation, albeit the complex further enhances the stability of Vif.
ACKNOWLEDGMENTS
We thank Alex Bullock for providing the EloBC plasmid, Bryan Cullen for providing Vif HXB3 DNA, Jeffery Babon for the murine Cul5 clone, and John Gross for providing the human Cul5/Rbx2 plasmid. We thank Xiaoyun Ji and Olga Buzovetsky for experimental assistance.
This work is supported in part by grants from the NIH (AI078831) to Y.X. and the Chinese Ministry of Science and Technology (2012CB911100 and 2013ZX0001-005), the Chinese Ministry of Education (IRT1016), and the Key Laboratory of Molecular Virology, Jilin Province (20102209) to X.-F.Y., as well as from the NSF (0919853) to E.R.
FOOTNOTES
- Received 24 December 2013.
- Accepted 27 December 2013.
- Accepted manuscript posted online 3 January 2013.
- Address correspondence to Yong Xiong, Yong.Xiong{at}yale.edu.
↵* Present address: Leslie S. Wolfe, KBI Biopharma Inc., Durham, North Carolina, USA.
REFERENCES
- Copyright © 2014, American Society for Microbiology. All Rights Reserved.
