HIV-1 Vif can directly inhibit apolipoprotein B mRNA-editing enzyme catalytic polypeptide-like 3G-mediated cytidine deamination by using a single amino acid interaction and without protein degradation.

The human apolipoprotein B mRNA-editing enzyme catalytic polypeptide-like 3G (APOBEC3G), also known as CEM-15, is a host-cell factor involved in innate resistance to retroviral infection. HIV-1 viral infectivity factor (Vif) protein was shown to protect the virus from APOBEC3G-mediated viral cDNA hypermutation. The mechanism proposed for protection of the virus by HIV-1 Vif is mediated by APOBEC3G degradation through ubiquitination and the proteasomal pathway. Here we show that in Escherichia coli the APOBEC3G-induced cytidine deamination is inhibited by expression of Vif without depletion of deaminase. Moreover, inhibition of deaminase-mediated bacterial hypermutation is dependent on a single amino acid substitution D128K that renders APOBEC3G resistant to Vif inhibition. This single amino acid was elegantly proven by other authors to determine species-specific sensitivity. Our results show that in bacteria this single amino acid substitution controls Vif-dependent blocking of APOBEC3G that is dependent on a strong protein interaction. The C-terminal region of Vif is responsible for this strong protein-protein interaction. In conclusion, our experiments suggest a complement to the model of Vif-induced degradation of APOBEC3G by bringing to relevance that deaminase inhibition can also result from a direct interaction with Vif protein.

The human apolipoprotein B mRNA-editing enzyme catalytic polypeptide-like 3G (APOBEC3G), also known as CEM-15, is a host-cell factor involved in innate resistance to retroviral infection. HIV-1 viral infectivity factor (Vif) protein was shown to protect the virus from APOBEC3G-mediated viral cDNA hypermutation. The mechanism proposed for protection of the virus by HIV-1 Vif is mediated by APOBEC3G degradation through ubiquitination and the proteasomal pathway. Here we show that in Escherichia coli the APOBEC3Ginduced cytidine deamination is inhibited by expression of Vif without depletion of deaminase. Moreover, inhibition of deaminase-mediated bacterial hypermutation is dependent on a single amino acid substitution D128K that renders APOBEC3G resistant to Vif inhibition. This single amino acid was elegantly proven by other authors to determine species-specific sensitivity. Our results show that in bacteria this single amino acid substitution controls Vif-dependent blocking of APOBEC3G that is dependent on a strong protein interaction. The C-terminal region of Vif is responsible for this strong proteinprotein interaction. In conclusion, our experiments suggest a complement to the model of Vif-induced degradation of APOBEC3G by bringing to relevance that deaminase inhibition can also result from a direct interaction with Vif protein.
The viral infectivity factor (Vif) 1 protein of human immunodeficiency virus type 1 (HIV-1) plays a dramatic importance in viral infectivity (1,2). Vif is a basic protein of 23 kDa required in the virus-producing cells during the late stages of infection to enhance viral infectivity 10-to 1000-fold (3)(4)(5). HIV-1 vifdefective virus can replicate in permissive cells such as Jurkat and SupT1 cells, but cannot replicate in non-permissive cells such as macrophages, primary human T cells, and some restrictive T cell lines (4 -7). Previous studies found that non-permissive cells contain an anti-viral cellular factor capable of suppress the HIV infectivity (8,9), recently identified as CEM15/ APOBEC3G, for which the antiviral action is overcome by Vif (10 -12). APOBEC3G is a virion-encapsidated cellular protein that deaminates dC to dU in minus-strand viral cDNA during reverse transcription (13)(14)(15)(16), preferentially at CCCA sequences (17). The uracil-containing cDNA may activate a cellular uracil-DNA-glycosylase causing the failure of reverse transcription, characteristic of vif-defective virus (18) and impair the integration of the provirus in the host genome (19,20). Furthermore, if the reverse transcription is completed at low efficiency and the resulting proviral double-stranded cDNA is integrated in the cell genome, the massive C-U conversion in the minus strand leads to a massive G to A hypermutation of the proviral plus-strand cDNA (14 -16). With analogy to APO-BEC1 and AID, substitution mutations in the catalytic domain of APOBEC3G result in an increase of viral infectivity and decrease in dC-to-dT mutations. The exact mechanism how Vif counteracts APOBEC3G action is not clear. It has been shown that Vif inhibits APOBEC3G translation or intracellular halflife (21,22). Recent evidences showed that Vif interacts with APOBEC3G as part of a Vif⅐Cul5⅐Sk-p1-Cullin-F-box complex resulting in the polyubiquitination and proteasomal degradation of APOBEC3G (23,24). The increase degradation and/or reduced level of APOBEC3G expression by Vif reduces its incorporation into virions (22,25) and consequent absence during reverse transcription in the target cell, allowing the virus to replicate.
APOBEC3G presents homology for two distinct but homologous cytidine deaminases: activation-induced cytidine deaminase (AID) and apolipoprotein (Apo) B editing cytidine deaminase (APOBEC1) (26 -28). Cytidine deamination by APOBEC1 begins with activation of zinc-bound water in the active center of the enzyme (28). The catalytic center of the cytidine deaminases has three zinc ligands that in APOBEC1 are histidine in position 61 (His-61), cysteines in positions 93 and 96 (Cys-93 and Cys-96), and a residue that mediates proton shuttling during catalysis, the glutamic acid in position 63 (Glu-63) (Fig.  1). Muramatsu et al. (26) showed that AID is necessary for somatic hypermutation and class switch recombination contributing to the diversification of antibodies. In addition, AID also mutates dC residues in Escherichia coli DNA and in sin-gle-stranded DNA in vitro. Therefore, the amino acid sequence alignment of APOBEC3G, APOBEC1, and AID may provide clues to their function.
Mariani et al. (13) described previously that Vif binds to the anti-viral host-cell factor APOBEC3G, although it was uncertain if the binding was transient or stable, or capable of altering APOBEC3G function. Our studies focused on this key question and investigated if Vif binding to APOBEC3G may affect deaminase function independently of the ubiquitin/proteasome pathway. To examine this hypothesis, we have expressed both APOBEC3G and Vif in E. coli and regulated the expression of its target gene to quantify cytidine deaminase activity. In addition, we aimed to evaluate in bacteria if the mechanism involved in a putative APOBEC3G-Vif interaction could be mediated by the single-amino acid substitution in APOBEC3G involved in species-specific restriction of Vif function.

EXPERIMENTAL PROCEDURES
Plasmids and Constructs-Plasmid ptacKanL94P, a generous gift from Dr. Michel C. Nussenzweig, has a mutation at codon 94 that produces a kanamycin-sensitive (Kan s ) phenotype. Mutating CCA to TCA or CTA restores resistance. Transcription of kanL94P is controlled by the IPTG-inducible tac promoter; this promoter is repressed in the absence of inducer by the expression of lacl q (29). The plasmid pASK-AID E.coli expresses mouse AID, modified according to the E. coli codon preference to produce AID E.coli . The plasmid pASK-APOBEC3G was constructed by subcloning human APOBEC3G gene amplified by PCR from pTrc99A-APOBEC3G, a kind gift of Dr. M. S. Neuberger (30). The hemagglutinin (HA) tag was introduced at the C terminus of apobec3g gene and cloned in pASK-IBA7 (IBA, Germany) using KpnI and XhoI restriction sites. pASK-IBA7 constructs are under the control of tet promoter, which is repressed in the absence of AHT by tet repressor. The pDHC29-Vif was obtained by PCR amplification of vif gene derived from HIV-1 NL43 , subcloning in pCR2.1-TOPO (Invitrogen), digested with NotI and XhoI and cloned by compatible ends in pDHC29, a kind gift from Dr. Gregory J. Phillips (31). The HA tag was introduced at the C terminus of vif gene. Constructs APOBEC3G-E67Q, APOBEC3G-D128K, and Vif-C114F were generated by site-directed mutagenesis according to the manufacturer's protocol (Stratagene). The APOBEC3-G-E67Q mutant was obtained using the oligonucleotides 5Ј-AACTTA-AGTACCACCCACAGATGAGATTCTTCCACTG-3Ј, 5Ј-CAGTGGAAG-AATCTCATCTGTGGGTGGTACTTAAGTT-3Ј (the substitution mutation is underlined). The APOBEC3G-D128K mutant was obtained using the oligonucleotides 5Ј-GCCTCTACTACTTCTGGAAGCCAGATTACC-AGGAGGC-3Ј, 5Ј-GCCTCCTGGTAATCTGGCTTCCAGAAGTAGTAG-AGGC-3Ј (the substitution mutation is underlined). The Vif-C114F mutant was obtained using the oligonucleotides 5Ј-CCAACTAATTCATC-TGCACTATTTTGATGCATTTTCAGAATCTGCTATAAGAAATACCA-3Ј, 5Ј-TGGTATTTCTTATAGCGAATTCTGAAAATGCATCAAAATAG-TGCAGATGAATTAGTTGG-3Ј (the substitution mutation is underlined). All mutations were confirmed by sequencing using the external primer 5Ј-AAATCGAAGGGCGCCGAG-3Ј. The plasmid replication origins of ptacKanL94D, pASK-IBA-7, and pDHC29 were pSC101, pUC and pST19, respectively, for compatibility in E. coli.
Monitoring Cytidine Deamination Activity and Protein Expression-This procedure has been adapted from Ramiro et al. (29). Briefly, BH156 UDG-deficient strain (dcm-6, thr1, hisG4, leuB6, rpsL, ara14, supE44, lacY1, tonA31, tsx78, galK2, xyl5, thi1, mtl!, unh-1), a kind gift from Dr. Michel C. Nussenzweig, was transformed with ptacKanL94D followed by independent transformations with the following plasmid combinations: pASK-APOBEC3G plus pDHC29, pASK-AID E.coli plus pDHC29, pASK-APOBEC3G plus pDHC29-Vif, pASK-APOBEC3G plus pDHC29-Vif-C114F, pASK-APOBEC3G-E67Q plus pDHC29, pASK-APOBEC3G-E67Q plus pDHC29-Vif, pASK-APOBEC3G-D128K plus pDHC29, and pASK-APOBEC3G-D128K plus pDHC29-Vif. For selection of ptacKanL94P, cell cultures were grown in selective medium containing 100 g/ml spectinomycin (Spc), reflecting the presence of aad9 gene. For selection of pDHC29-derived constructs described above, cell cultures were grown in selective medium containing 34 g/ml chloramphenicol (Clo). For selection of pASK-derived constructs described above, cell cultures were grown in selective medium containing 100 g/ml ampicillin (Amp), reflecting the presence of bla gene. For the Rif R assay, 10 -12 independent colonies were grown in selective SOB medium (20 g tryptone, 5 g yeast extract, 0.5 g NaCl, 250 mM KCl, per liter, pH 7) supplemented with 20 mM MgCl 2 to an optical density of 0.6 at 660 nm and induced with 0.2 g/ml AHT (Sigma) plus 1 mM IPTG for 3 h at 37°C. Cell cultures were centrifuged, washed, and spread in LB plates containing 100 g/ml Amp, 100 g/ml Spc, and 34 g/ml Clo (cell viability plates) or 100 g/ml Amp, 100 g/ml Spc, 34 g/ml Clo, and 100 g/ml rifampicin (Rif) (Rif R plates). The mutation frequencies were calculated as the ratio of Rif R colonies to the total number of viable cells, identified for each individual colony. For the Kan R assay, we proceeded in a similar manner. Cultures were spread in LB plates containing 100 g/ml Amp, 100 g/ml Spc, and 34 g/ml Clo (cell viability plates) or 100 g/ml Amp, 100 g/ml Spc, 34 g/ml Clo, and 50 g/ml kanamycin (Kan R plates). The mutation frequencies were calculated as the ratio of Kan R colonies to the total number of viable cells, identified for each individual colony. Three independent assays were performed for each Rif R -and Kan R -induced mutation assay. For protein expression of APOBEC3G, AID E.coli , and Vif constructs, 10 ml of cell cultures was disrupted by sonication in PBS, plus complete protease inhibitor mixture (Roche Applied Science). The APOBEC3G and Vif lysates were subject to Western blot with HRP-conjugated anti-HA monoclonal antibody 3F10 (Roche Applied Science). Similar lysates were immunoprecipitated with anti-HA affinity matrix (Roche Applied Science) and anti-Vif polyclonal serum plus protein A. AID E.coli was immunoprecipitated with anti-FLAG beads. Samples were separated on SDS-PAGE gels under reducing conditions, transferred to a Hybond-C Extra membrane (Amersham Biosciences), and blotted with HRP-conjugated anti-HA monoclonal antibody 3F10 (Roche Applied Science) and HRP-conjugated anti-FLAG monoclonal antibody.
Co-immunoprecipitation Assays-To evaluate protein-protein interactions in vivo, we performed co-immunoprecipitation assays. BH156 cells were transformed with seven different plasmid combinations: pDHC29 and pASK-APOBEC3G, pDHC29-Vif and pASK-APOBEC3G, pDHC29 and pASK-APOBEC3G-E67Q, pDHC29-Vif and pASK-APOBEC3G-E67Q, pDHC29 and pASK-APOBEC3G-D128K, pDHC29-Vif and pASK-APOBEC3G-D128K, and pDHC29-Vif-C114F and pASK-APOBEC3G. Cells were grown in SOB medium supplemented with 20 mM MgCl 2 and respective antibiotics for plasmid selection. Protein expression was induced at OD 660 nm ϳ 0.6 with 1 mM IPTG and 0.2 g/ml AHT during 3 h. Cells were disrupted by sonication in PBS and supplemented with complete protease inhibitor mixture (Roche Applied Science). A mixture of protein A-Sepharose TM beads (Amersham Biosciences) with the BH156 lysate was prepared and incubated on a rotator during 4 h at 4°C. In parallel, a mixture for each cell lysate was prepared with anti-Vif polyclonal serum, kindly provided by Dr. Dana Gabuzda, and incubated on ice during 3 h (protein expression mix). Following incubation, the mix of protein A-Sepharose TM beads was added to the protein expression mix and incubated overnight on a rotator at 4°C. When applicable, samples were washed four times with low ionic buffer (0.15 M NaCl, 0.05 M Tris, pH 7.5, 0.1% Nonidet P-40, and 1ϫ with 0.15 M NaCl, 0,05 M Tris, pH 7.5) or high ionic buffer (0.5 M NaCl, 0.05 M Tris, pH 7.5, 0.1% Nonidet P-40, 1% Triton X-100, and 1ϫ with 0.5 M NaCl, 0.05 M Tris, pH 7.5). Beads were boiled in Laemmli buffer, and proteins were resolved by standard SDS-PAGE techniques, transferred to a Hybond-C Extra membrane (Amersham Biosciences), and blotted with HRP-conjugated anti-HA monoclonal antibody 3F10 (Roche Applied Science).
Phage Display of APOBEC3G-To display APOBEC3G protein on M13 phage, phagemid pComb3X expressing apobec3g gene in fusion with gene III of M13 was transformed into E. coli strain ER2537. A fresh colony of APOBEC3G-HA was grown overnight at 37°C in SOB medium containing 100 g/ml ampicillin. 1 ml of cell culture was inoculated in 100 ml of SOB medium containing 100 g/ml ampicillin. Cells were grown at 37°C until OD 660 nm ϭ 0.6 and infected with 2 ml of M13 helper phage (10 12 plaque-forming units/ml) during 1.5 h at 37°C. Kanamycin was added to a final concentration of 50 g/ml, and growth continued overnight. Following centrifugation at 3000 ϫ g, phages were precipitated from cell supernatant with 4% polyethylene glycol 8000 and 3% NaCl during 30 min on ice, centrifuged at 15,000 ϫ g during 30 min, and resuspended in 1 ml of PBS. The titer of phages displaying APOBEC3G-HA was determined. Bacteria transformed with pComb3X/APOBEC3G was used to asses the expression of the deami-nase. Briefly, transformed ER2537 cells were induced with 0.5 mM IPTG for 12 h at 37°C after cells reached OD 660 nm ϭ 0.6. Bacterial lysates were resolved by standard SDS-PAGE techniques, transferred to a Hybond-C Extra membrane (Amersham Biosciences) and blotted with anti-HA monoclonal antibody 3F10 (Roche Applied Science).
Phage-ELISA Measurements-Phage-ELISA was used to determine the relative affinity and specificity of APOBEC3G phage for HIV-1 Vif protein.
To analyze relative binding affinities of phage-displayed APOBEC3G, ELISA plates (Nunc) were coated with 1 g of HIV-1 Vif protein (33) or soy milk and incubated overnight at 4°C. Wells were blocked for 1 h at 37°C with 0.5% soy milk in PBS. Phages were diluted 1:100 and incubated for 1 h at 30°C. After washing, HRP-conjugated anti-M13 monoclonal antibody (Amersham Biosciences) was used for detection. The results were measured by optical density at 405 nm.
To identify Vif protein domains involved in APOBEC3G-Vif interaction, a competitive-phage ELISA assay using 100 M HIV-1 consensus B Vif (15-mer) peptides was used. HIV-1 consensus B Vif (15-mer) peptides were provided by the National Institutes of Health AIDS Research and Reference Reagents Program. ELISA plates (Nunc) were coated overnight at 4°C with 300 ng of purified recombinant HIV-1 Vif protein (33). Wells were blocked for 1 h at 37°C with 0.5% soy milk in PBS. APOBEC3G phages were incubated 30 min at 30°C following by 2 h at 4°C with 100 M of each Vif peptide in low ionic buffer or high ionic buffer as described for co-immunoprecipitation and added to coated wells. Phages were washed with PBS/0.05% Tween 20, and HRP-conjugated anti-M13 monoclonal antibody (Amersham Biosciences) was used for detection. The results were measured by optical density at 405 nm.

APOBEC3G Induces Cytidine Deamination in E. coli-To
examine whether APOBEC3G induces rapid mutation in E. coli, an initial study was performed to analyze its induced mutation activity using rpoB gene as a selectable marker. For efficient measurement we used mouse AID E.coli as a positive control of cytidine deaminase (kindly provided by Dr. Nussenzweig) (29). Both mouse AID E.coli and human APOBEC3G were cloned in pASK under the control of the tetracycline (tet) pro-moter, inducible with anhydrotetracycline (AHT). Mutator activity of the expressed proteins was monitored by assessing the frequency of rifampicin-resistant (Rif R ) colonies, which resulted from deamination of 1 of the 11 dC residues in the rpoB gene encoding the ␤-subunit of RNA polymerase (12,29). To control the mutator efficiency of the human APOBEC3G deaminase, a putative active site mutant APOBEC3G-E67Q was constructed. This mutation was identified by ClustalW multiple alignment between AID, APOBEC1, APOBEC3G, and the described previously active site mutant of AID E.coli ( Fig. 1) (29). All experiments were performed in BH156 UDG-deficient (ung-1) E. coli. The absence of UDG activity in the cells facilitates subsequent analysis, once it prevents the removal of uracil in U-G mismatch resulting from cytidine deamination, allowing the increase of mutation frequency caused by the activity of deaminase expression. In this assay a population of 10 -12 individual colonies of bacteria alone, and expressing AID, APOBEC3G, and APOBEC3G-E67Q were assayed in the presence of tet inducer during 3 h after the cells reached OD 660 nm of 0.6. This density corresponds to three to four divisions, allowing a reproducible evaluation during the logarithmic bacterial growth curve. Mutation frequencies were calculated as the ratio of Rif R colonies to the total number of viable cells for each individual colony. In Fig. 2A, we observed the following mean frequencies for the number of Rif R colonies when AID E.coli and APOBEC3G were expressed: 398.9 ϫ 10 Ϫ9 and 538.9 ϫ 10 Ϫ9 , respectively. These results showed that both deaminases had similar mutation efficiency in this target. The mutation capability of both AID E.coli and APOBEC3G was 10 to 15 times higher than that of bacterial cells alone (mean frequency, 34.65 ϫ 10 Ϫ9 ) and the putative active site mutant, APOBEC3G-E67Q (mean frequency, 54.4 ϫ 10 Ϫ9 ). The results obtained with APOBEC3G-E67Q mutant were consistent with other reports where a residual deaminase activity may result  1-191). ClustalW multiple sequence alignment shows identities in dark shading. APOBEC3G contains a duplicated active site, a linker, and a pseudoactive site. The different domains are identified based on homology to APOBEC1. The properties common to cytidine deaminases are indicated as follows: zinc ligand amino acids (}), residues that mediate proton shuttling during catalysis (*) (30). from the usage of an alternative proton donor other than the AID E58 homologue (12,29).
Because AID-mediated cytidine deamination activity in E. coli was linked to transcription (29), we tested whether APOBEC3G-induced mutation in bacteria followed a similar induced transcription phenotype. We used an inactive allele of the kanamycin resistance gene (kanL94P), in which the TTG codon encoding a leucine is altered to CCA encoding a proline (29,35). The inactive kan gene is under the control of isopropyl-␤-D-thiogalactopyranoside (IPTG)-inducible tac promoter (ptacKanL94P). Deaminase activity was measured by reversion to kanamycin resistance (Kan R ) resulting from single Cto-T mutation in codon 94 to either CTA (leucine) or TCA (serine) after induced transcription by IPTG. The frequency of AID E.coli -induced cytidine deamination in the kan resistance gene, obtained by counting the number of kanamycin-resistant colonies, was about 40 times higher (mean frequency, 957.6 ϫ 10 Ϫ6 ) than BH156 alone (mean frequency, 63.53 ϫ 10 Ϫ6 ). This difference was higher than a 5-fold difference between APOBEC3G (mean frequency, 267.55 ϫ 10 Ϫ6 ) and the activesite mutant APOBEC3G-E67Q (mean frequency, 55.1 ϫ 10 Ϫ6 ) and BH156 alone (mean frequency, 63.53 ϫ 10 Ϫ6 ) (Fig. 2B). Furthermore, when AID E.coli -induced mutation was compared with that of APOBEC3G, its mutation frequency was three to four times higher. This increase may have resulted from the higher expression level of AID E.coli that is modified at the nucleotide sequence to the E. coli codon preference (29) (Fig.  2C). The expression of each protein construct was analyzed during the cytidine deamination assay, and all proteins were detected at similar levels (Fig. 2C). These results do not reflect the Rif R data where both deaminases had similar rates of mutation, which can be explained by the use of a low effi-FIG. 2. APOBEC3G and AID induce rapid mutation in E. coli DNA and its mutation activity is linked to transcription. Expression of AID E.coli and APOBEC3G induce mutator phenotype. The mouse aid gene, modified according to the E. coli codon preference to produce AID E.coli . The mutator phenotype of APOBEC3G is dependent on the putative catalytic site Glu-67. The BH156 UDGdeficient strain was sequentially transformed with ptacKanL94P, plus AID E.coli or APOBEC3G or APOBEC3G-E67Q as described under "Experimental Procedures." Several (10 -12) individual colonies were grown to exponential phase, and protein expression was induced during 3 h with AHT plus IPTG, to express deaminases and target kanL94P gene. A, Rif R frequency resulting from the expression of AID E.coli , APOBEC3G, and APOBEC3G-E67Q constructs to rpoB gene target. Circles represent the Rif R mean frequency of individual starting colonies; horizontal bars represent mean values that are given under x-axis. B, Kan R frequency resulting of the expression of AID E.coli , APOBEC3G, and APOBEC3G-E67Q constructs to kanL94P gene target. Circles represent the Kan R mean frequency of individual starting colonies; horizontal bars represent mean values that are given under x-axis. C, expression of AID E.coli , APOBEC3G, and APOBEC3G-E67Q proteins in BH156 UDG-deficient cells. After 3 h of induction with AHT and IPTG, cells were disrupted, and whole extracts were immunoprecipitated with FLAG-beads for AID E.coli , and anti-HA beads for APOBEC3G and APOBEC3G-E67Q. Proteins were resolved by standard SDS and analyzed by Western blot with correspondent antibodies. ciency deaminase hotspot for APOBEC3G when compared with AID E.coli in kanL94P (CCA). Sequencing of Kan R colonies showed C-to-T mutations in codon 94, on average half to CTA (leucine) and half to TCA (serine) (data not shown). These data indicate that APOBEC3G has cytidine deaminase activity similar to AID E.coli in different hotspot sequences, and that protondonor mutation E67Q eliminates deaminase activity of APOBEC3G.
HIV-1 Vif Expression Inhibits APOBEC3G Deaminase Activity in E. coli-HIV-1 Vif protein was previously shown to induce ubiquitination and subsequent degradation of APOBEC3G, probably through intimate co-interaction of Vif-APOBEC3G and retargeting to the SCF-like E3 ubiquitin ligase complex in mammalian cells (23,24). Although the putative end-result of Vif-APOBEC3G binding is deaminase degradation, the effect of this direct interaction on the activity of APOBEC3G was not yet analyzed. The bacteria system described above was used to assess this hypothesis, because APOBEC3G activity in mammalian cells is difficult to dissociate from proteasomal degradation, due to the constitutive presence of ubiquitin-proteasome machinery. To determine whether co-expression of Vif with APOBEC3G could affect its deaminase activity, the vif gene from HIV-1 NL43 was amplified by polymerase chain reaction and a hemagglutinin tag was added at the C terminus. The vif gene was cloned in pDHC29 plasmid under the control of IPTG-inducible lac promoter, similar to the control expression of kanL94P gene. As different origins of replication were necessary to maintain plasmid stability in bacteria together with ptacKanL94P and pASK-IBA7, pDHC29 contains a different ori derived from pST19 (31). The bacteria BH156 containing ptacKanL94P, pDHC29, pDHC29Vif, or pDHC29Vif-C114F, together with pASK-APOBEC3G or pASK-APOBEC3G-D128K, was induced with IPTG and AHT, as described under "Experimental Procedures." As observed in Fig. 3A, the background deamination of kanL94P-induced transcription resulted in a mean mutation frequency of 63.5 ϫ 10 Ϫ6 . When Vif was expressed alone, a small increase in the Kan R mutation frequency was observed (mean value, 89.4 ϫ 10 Ϫ6 ). In contrast, a higher increase in Kan R mutation frequency (mean value, 327.3 ϫ 10 Ϫ6 ) was observed when APOBEC3G was expressed alone (BH156 transformed with ptacKanL94P, pDHC29, and pASK-APOBEC3G). Concomitant expression of Vif protein and APOBEC3G (BH156 transformed with ptacKanL94P, pASK-APOBEC3G, and pDHC29Vif) reduced to background levels the Kan R phenotype (mean value, 77.9 ϫ 10 Ϫ6 ), as shown in Fig. 3A. The relative magnitude of cytidine deamination reduction when Vif was expressed with APOBEC3G is consistent with a hypothesis that Vif acts as a direct inhibitor of deaminase activity. To confirm that the expression of Vif protein was essential for the inhibition of APOBEC3G activity in E. coli, a functional inactive mutant of Vif protein, Vif-C114F, was used as a negative control (36). A cytidine deamination assay was performed as described previously for the co-expression of APOBEC3G with wild-type Vif. As shown in Fig. 3A, co-expression of Vif-C114F and APOBEC3G did not affect its deaminase activity. In this assay the mean value of Kan R reversion during co-expression was 442.9 ϫ 10 Ϫ6 . This value was higher than that obtained with APOBEC3G expression alone (mean frequency, 327.3 ϫ 10 Ϫ6 ), showing that Vif-C114F does not inhibit deaminase function of APOBEC3G. These results correlate with the non-protective role of C114F mutation in Vif against viral infection. Moreover, these results also showed that co-expression of Vif-C114F with APOBEC3G might stabilize its activity of cytidine deamination. Concomitant expression of APOBEC3G and Vif constructs showed that all proteins had similar steady-state levels (Fig. 3B). Therefore, the results from this assay indicate that the expression of an active Vif protein in the E. coli context inhibits APOBEC3G activity, without affecting its steady-state level.
A Single Amino Acid Substitution D128K Renders APOBEC3G Resistant to Vif Inhibition-It was previously shown that a single D128K substitution rendered human APOBEC3G resistant to Vif-induced depletion and consequently maintained its capacity to inhibit HIV-1 replication in the presence of Vif (37)(38)(39)(40). In addition, this single amino acid difference in APOBEC3G protein controlled species specificity of HIV-1 viral infectivity factor. Therefore, this mutation determines the functional interaction of APOBEC3G with Vif. These results suggested a model where Vif counteracts the inhibitory effects of APOBEC3G by enhancing its degradation. Nevertheless, questions remain whether the D128K mutation in human APOBEC3G affects its deaminase activity or if deaminase-blocking activity by Vif can be overcome by this single amino acid substitution, independently of the ubiquitinproteasome pathway. To gain insight into this question, expression of APOBEC3G-D128K was induced alone and in presence of Vif. As shown in Fig. 4A, when APOBEC3G-D128K was expressed alone the mean frequency of Kan R mutants obtained was 268.5 ϫ 10 Ϫ6 . Comparison of this value with that of bacterial cells alone (mean frequency, 63.5 ϫ 10 Ϫ6 ) and wild-type APOBEC3G (mean value, 327.3 ϫ 10 Ϫ6 ), showed that the mutant D128K deaminase activity was rather similar to the wildtype enzyme. However, a small reduction on the D128K mutant deaminase activity was observed, reflecting a possible structural alteration that affects its enzymatic activity. When APOBEC3G-D128K and Vif were co-expressed in bacteria, a small increase in the rate of cytidine deamination was observed (mean frequency, 313.8 ϫ 10 Ϫ6 ) compared with APOBEC3G alone (mean frequency, 327.3 ϫ 10 Ϫ6 ). This small increase in Kan R mutants may result from a stabilization of APOBEC3G-D128K activity by Vif. Future studies are necessary to evaluate this hypothesis. Results obtained with co-expression of APOBEC3G-D128K and HIV-1 Vif were in contrast with those of APOBEC3G co-expression with HIV-1 Vif, where a dramatic reduction in cytidine deamination was observed. Similar mutation rates of APOBEC3G and APOBEC3G-D128K/Vif mutation were not due to differences in deaminases expression, because all APOBEC3G constructs were detected equally (Fig.  4B). Therefore, these results strongly suggest that expression of HIV-1 Vif can block APOBEC3G-mediated cytidine deamination, and this effect may be controlled by a single amino acid substitution (D128K) in the enzyme.
Vif-mediated Block of APOBEC3G Involves a Strong Protein Interaction That Is Dependent on a Single Amino Acid Substitution-As shown previously by several authors, Vif and APOBEC3G can interact in mammalian cells, and this phenotype was linked to a Vif-dependent degradation of deaminase by proteasomes (37)(38)(39)(40). This functional result was dependent on the single amino acid substitution D128K. To determine whether Vif-mediated block of cytidine deaminase activity was due to this specific physical interaction, co-immunoprecipitation assays were performed (Fig. 5). As described under "Experimental Procedures," BH156 cells alone, BH156 with pDHC29 and pASK-APOBEC3G, BH156 with pDHC29-Vif and pASK-APOBEC3G, BH156 with pDHC29 and pASK-APOBEC3G-E67Q, BH156 with pDHC29-Vif and pASK-APOBEC3G-E67Q, BH156 with pDHC29 and pASK-APOBEC3G-D128K, BH156 with pDHC29-Vif and pASK-APOBEC3G-D128K, and BH156 with pDHC29-VifC114F and pASK-APOBEC3G were grown and induced with IPTG and AHT to express Vif and APOBEC3G constructs, respectively. Cells were disrupted by sonication, and protein expression was analyzed after immunoprecipitation with anti-HA affinity matrix (Roche Applied Science) (Fig.  5). Similar extracts of bacteria were used to evaluate the presence of APOBEC3G-Vif complexes and how different mutations in each protein may interfere in this interaction. Each lysate was incubated with rabbit anti-Vif polyclonal serum, kindly provided by Dr. Dana Gabuzda. To gain insight into the possible interaction forces involved in this putative APOBEC3G-Vif complex, immunoprecipitated proteins were isolated by affinity absorption to protein A-Sepharose beads (Amersham Biosciences), washed with a low ionic and high ionic strength buffer and eluted with a SDS-containing buffer. After gel electrophoresis and transfer onto nitrocellulose membrane, co-immunoprecipitated proteins were probed with anti-HA monoclonal antibody (Roche Applied Science). As shown in Fig. 5, a low amount of nonspecific immunoprecipitated proteins in BH156 cells was observed. When Vif was immunoprecipitated in the presence of wild-type APOBEC3G and washed independ- FIG. 3. Vif blocks human APOBEC3G-mediated cytidine deaminase activity. A, the BH156 UDG-deficient strain with ptacKanL94P was sequentially transformed with pDHC29 (column 1), pDHC29-Vif (column 2), pDHC29 plus APOBEC3G (column 3), pDHC29-Vif plus APOBEC3G (column 4), and pDHC29-Vif-C114F plus APOBEC3G (column 5). Individual colonies (10 -12) were grown to exponential phase and protein expression was induced during 3 h with AHT and IPTG. Kan R mean frequencies resulting of cytidine deaminase expression constructs to kanL94P gene target were obtained. Circles represent the Kan R mean frequency of individual starting colonies; horizontal bars represent mean values that are given under x-axis. B, protein expression of Vif, Vif-C114F, and APOBEC3G alone or co-expressed in BH156 UDG-deficient cells. After 3 h of induction with AHT and IPTG for APOBEC3G and Vif constructs, respectively, cells were disrupted, and whole extracts were immunoprecipitated with anti-HA beads and anti-Vif polyclonal serum plus protein A beads. Proteins were resolved by standard SDS and analyzed by Western blot with HRP-conjugated anti-HA monoclonal antibody 3F10 (Roche Applied Science). A background band of ϳ30 kDa is constantly present when Vif is immunoprecipitated with anti-Vif polyclonal serum. Cell lysates of Vif, Vif-C114F, and APOBEC3G alone and co-expressed in BH156 UDGdeficient cells were subject to a Western blot with HRP-conjugated anti-HA monoclonal antibody 3F10 (Roche Applied Science). As observed, cell lysate loads have similar amounts of proteins recognized by anti-HA antibody in Western blot. ently with low and high ionic strength buffers, a distinct band of ϳ45 kDa indicating the presence of deaminase was retrieved. Similar results were obtained when Vif was co-expressed with APOBEC3G-E67Q. These data indicated the pres-ence of a stable complex between Vif and APOBEC3G that was not affected by the E67Q catalytic disrupting mutation in APOBEC3G. Co-immunoprecipitation of HIV-1 Vif and APOBEC3G-D128K protein with low ionic wash buffer re- FIG. 4. The single amino acid substitution D128K renders APOBEC3G resistant to Vif blocking. A, expression of APOBEC3G induces a Kan R mutator phenotype that is abolished when Vif is co-expressed. In contrast, expression of APOBEC3G-D128K induces a Kan R mutator phenotype and is not inhibited by the co-expression of Vif or Vif-C114F. The BH156 UDG-deficient strain with ptacKanL94P was sequentially transformed with pDHC29 (column 1), pDHC29-Vif (column 2), pDHC29 plus APOBEC3G (column 3), pDHC29-Vif plus APOBEC3G (column 4), pDHC29 plus APOBEC3G-D128K (column 5), pDHC29-Vif plus APOBEC3G-D128K (column 6). Individual colonies (10 -12) were grown to exponential phase, and protein expression was induced during 3 h with AHT and IPTG. Kan R mean frequencies resulting of cytidine deaminase expression constructs to kanL94P gene target were obtained. Circles represent the Kan R mean frequency of individual starting colonies; horizontal bars represent mean values that are given under the x-axis. B, protein expression of Vif, APOBEC3G, and APOBEC3G-D128K alone or co-expressed in BH156 UDG-deficient cells. After 3 h of induction with AHT and IPTG for APOBEC3G and Vif constructs, respectively, cells were disrupted, and whole extracts were immunoprecipitated with anti-HA beads and anti-Vif polyclonal serum plus protein A beads. Proteins were resolved by standard SDS and analyzed by Western blot with HRP-conjugated anti-HA monoclonal antibody 3F10 (Roche Applied Science). A background band of ϳ30 kDa is constantly present when Vif is immunoprecipitated with anti-Vif polyclonal serum. Cell lysate of vector alone, Vif, APOBEC3G, and APOBEC3G-D128K alone and co-expressed in BH156 UDG-deficient cells were subject to a Western blot with HRP-conjugated anti-HA monoclonal antibody 3F10 (Roche Applied Science). As observed, cell lysate loads have similar amounts of proteins recognized by anti-HA antibody in Western blot. trieved similar quantities of the blocking-inactive APOBEC3G-D128K compared with APOBEC3G. In contrast, when high ionic strength wash buffer was used, the amount of APOBEC3G-D128K retrieved by Vif co-immunoprecipitation was reduced to ϳ50%, relative to APOBEC3G. The residual interaction observed with D128K mutant of deaminase was consistent in all experiments, indicating an alternative mechanism of Vif-APOBEC3G interaction or otherwise an inherent problem to the bacterial system used. To evaluate if a nonfunctional Vif protein could interact with APOBEC3G, mutant Vif-C114F was co-expressed with deaminase and co-immunoprecipitation was performed with anti-Vif polyclonal serum. As shown in Fig. 5, Vif-C114F was immunoprecipitated similarly with both salt concentrations used showing competence for the serum to analyze Vif-interacting proteins. Co-immunoprecipitation of Vif-C114F and APOBEC3G with low ionic wash buffer showed that mutant Vif-C114F bound APOBEC3G at similar quantities compared with wild-type Vif. In contrast, when high ionic strength wash buffer was used, a decrease was observed in the amount of APOBEC3G or APOBEC3G-D128K retrieved by Vif-C114F and Vif, respectively. These results show a strong physical interaction that occurs between Vif and APOBEC3G, which is partially dependent on a single amino acid substitution D128K. Moreover, a residual interaction is maintained at high salt concentrations, indicating the presence of putative strong interaction domains in Vif protein that ultimately mediate Vif-APOBEC3G binding.
Vif C-terminal Peptides Strongly Inhibit Vif-APOBEC3G Interaction in High Salt Stringency-To search for HIV-1 Vif domains that strongly interact with APOBEC3G protein, a competitive assay using HIV-1 Vif consensus B (15-mer) peptides was performed. APOBEC3G was cloned in the phagemid vector pComb3X (32), expressed, and displayed in fusion with pIII on the surface of M13 phage (Fig. 6A).
A phage-ELISA assay was performed to test if APOBEC3G displayed on the phage surface could specifically interact with Vif-C114F in BH156 UDG-deficient cells. As described bellow, proteins were immunoprecipitated after 3 h of induction with AHT and IPTG for APOBEC3G and Vif constructs, respectively. Proteins were resolved by standard SDS and analyzed by Western blot with HRP-conjugated anti-HA monoclonal antibody 3F10 (Roche Applied Science). A, co-immunoprecipitation of APOBEC3G constructs with anti-Vif rabbit polyclonal serum and washes with low ionic buffer. APOBEC3G constructs were analyzed by Western blot with HRP-conjugated anti-HA monoclonal antibody 3F10. B, immunoprecipitation of Vif constructs with anti-Vif rabbit polyclonal serum and washes with low ionic buffer. Vif constructs were analyzed by Western blot with HRP-conjugated anti-HA. A background band of ϳ30 kDa is constantly present when Vif is immunoprecipitated with anti-Vif polyclonal serum. C, co-immunoprecipitation of APOBEC3G constructs with anti-Vif rabbit polyclonal serum and washes with high ionic buffer. APOBEC3G constructs were analyzed by Western blot with HRP-conjugated anti-HA monoclonal antibody 3F10. D, immunoprecipitation of Vif constructs with anti-Vif rabbit polyclonal serum and washes with high ionic buffer. Vif constructs were analyzed by western-blot with HRPconjugated anti-HA monoclonal antibody. A background band of ϳ30 kDa is constantly present when Vif is immunoprecipitated with anti-Vif polyclonal serum. E, cell lysates of each combination expressed alone and co-expressed in BH156 UDG-deficient cells were subject to a Western blot with HRP-conjugated anti-HA monoclonal antibody 3F10 (Roche Applied Science). As observed, cell lysate loads have similar amounts of proteins recognized by anti-HA antibody in Western blot.
Vif protein. Phages displaying APOBEC3G protein bound with high relative affinity to HIV-1 Vif, confirming the co-immunoprecipitation results described above. Negative controls, including binding of naked M13 to Vif, and binding of phage-displayed APOBEC3G to soy milk blocking agent exhibited a reduced binding, confirming a specific APOBEC3G-Vif interaction (Fig. 6B).
To characterize the HIV-1 Vif domains that are responsible FIG. 6. Phage-displayed APOBEC3G interacts specifically with HIV-1 Vif and is inhibited by HIV-1 Vif peptides. A, expression of pComb3X-APOBEC3G in E. coli strain ER2537. Bacterial cells transformed with phagemid expression vector were grown to OD 660 nm ϳ 0.6 and induced with 0.5 mM IPTG during 3 h. Cells were disrupted, and whole extracts were resolved by standard SDS-PAGE and analyzed by Western blot with HRP-conjugated anti-HA monoclonal antibody 3F10 (Roche Applied Science). Negative control expression with bacteria alone is also shown. ELISA of phage-displayed APOBEC3G against recombinant purified HIV-1 Vif protein (33) and soy milk blocking reagent. Binding of naked M13 phage against Vif protein and soy milk is also showed. Binding of phages to target proteins was detected by HRP-conjugated anti-M13 monoclonal antibody (Amersham Biosciences). The results were measured by optical density at 405 nm. B, competitive ELISA of phage-displayed APOBEC3G against Vif protein using HIV-1 consensus B Vif (15-mer) peptides covering the complete Vif sequence. Recombinant Vif protein was added to a 96-well plate. Phage-displayed APOBEC3G was preincubated with 100 M of independent peptides in low or high ionic buffer as described in the "Experimental Procedures." After incubation, excess phages were washed off, and HRP-conjugated anti-M13 monoclonal antibody (Amersham Pharmacia Biotech) was used for detection. The results were measured by optical density at 405 nm. As shown in the figure, at low ionic conditions peptides from regions 13-39, 73-99, 133-147, and 169 -192 can inhibit APOBEC3G binding to Vif. At high ionic conditions, peptides from regions 85-99 and 169 -192 can inhibit APOBEC3G binding to Vif. C-terminal peptides from 169 -192 consistently reduced APOBEC3G-Vif binding with both ionic conditions. for APOBEC3G-Vif interaction, a competitive phage-ELISA using HIV-1 consensus B Vif (15-mer) peptides covering the complete Vif sequence was performed. Vif peptides were preincubated with phage-displayed APOBEC3G to avoid protein conformational constraints and epitope masking of APOBEC3G bound to the plastic support of ELISA plates. As shown in Fig. 6C the independent addition of Vif peptides exhibited different inhibition levels of APOBEC3G-Vif interaction. Vif peptides spanning regions 13-39, 73-99, 133-147, and 169 -192, exhibited similar relative efficacy to inhibit APOBEC3G-Vif interaction when preincubated at low salt conditions with phage-displayed APOBEC3G, favoring low energy interactions. Conversely, peptides encompassing all other regions of Vif protein failed to inhibit APOBEC3G-Vif interaction. Surprisingly, some of the non-inhibiting peptides induced an increase in binding affinity of the two proteins, probably due to conformational alterations of APOBEC3G protein.
When high salt conditions were used, only Vif peptides from region 85 to 99 and 169 to 192 showed relative ability to inhibit APOBEC3G-Vif interaction. Additionally, some peptides that inhibited the binding of phage-displayed APOBEC3G to Vif at low salt conditions had no effect when high salt conditions were used. This result was more prominent in presence of peptides overlapping region 13-39, where protein-protein interaction was observed. Because high salt conditions selected for strong Vif peptides-APOBEC3G interaction, these results showed that only a partial region of Vif protein was consistently able to inhibit APOBEC3G-Vif interaction. Therefore, these data may support the conclusion that APOBEC3G-Vif binding is specifically mediated by a strong interacting Vif domain encompassing regions 85-99 and 169 -192. DISCUSSION The system described in the results aimed to co-express Vif protein with APOBEC3G in conditions where degradation by the ubiquitin-proteasome pathway or kinase regulation could not occur. Our goal was to evaluate if APOBEC3G deaminase activity was inhibited by Vif interaction.
Due to the potential link between APOBEC3G-induced cytidine deamination and transcription, we expressed APOBEC3G in E. coli and regulated the expression of its target gene by using inducible promoters. The data presented in this study demonstrate that HIV-1 Vif protein interacts directly with APOBEC3G and blocks its mediated cytidine deaminase activity. Our studies show that a single D128K substitution can render APOBEC3G resistant to Vif-mediated block of cytidine deaminase activity. Our results strongly support a new mechanistic function of HIV-1 Vif protein, complementary to the model where Vif counteracts the inhibitory effects of APOBEC3G by enhancing its degradation via ubiquitin-proteasome pathway. Our findings that Vif interacts with APOBEC3G and inhibits its deaminase activity may indicate an alternative mechanism to eliminate innate immunity to HIV-1. Recent results from Strebel and co-workers (41,42) support our conclusion that Vif-induced APOBEC3G destruction may not be necessary for the production of fully infectious virus. Nevertheless, a synergistic effect can also be hypothesized to more efficiently inhibit the anti-viral activity of APOBEC3G. This mechanism may occur in conditions where deaminase degradation is not functional or is not possible.
Previous data showed that Vif prevents viral incorporation of APOBEC3G, probably by reducing its intracellular half-life (22,39,40). Nevertheless, it was reported that Vif-induced degradation of APOBEC3G may be dependent on the expression level of these proteins, but no data is available to define its optimal physiological threshold (22,37,39,40). Therefore, if the optimal expression levels of Vif or APOBEC3G are not met, it is conceivable that some APOBEC3G may escape Vif-induced degradation and incorporate directly into viral particles through nucleocapsid interactions (43,44). Because Vif can be present in the viral particle, a co-localization with APOBEC3G in the virus would be sufficient to block cytidine deamination and prevent initial massive hypermutation during reverse transcription. This hypothesis is supported by previous results showing that Vif may also interact with Gag and incorporate into viral particles (45,46).
We have shown that inhibition of APOBEC3G deaminase activity by Vif is overcome by a homologous amino acid substitution from aspartic acid to lysine at position 128 of APOBEC3G. This mutation was described previously to control the species-specific ability of HIV-1 Vif protein to bind and inactivate the host defense factor (37)(38)(39)(40). Our results are partially in agreement with these reports, because in our assay Vif interaction with APOBEC3G-D128K appeared to be reduced but not eliminated. Nevertheless, the work of Pathak and co-workers (37) shows that D128K mutation does not interfere with APOBEC3G-Vif interaction and, therefore, is consistent with our results. An explanation may be envisioned; for example, Vif interaction with APOBEC3G may induce an alosteric conformational change and expose Asp-128 that synergistically binds Vif and promotes enzyme inhibition. This explanation is supported by our results were the D128K mutation does not appear to be important for APOBEC3G activity even in absence of Vif expression. Moreover, the restriction for a neutral or positively charged amino acid in this position suggests a putative alosteric interaction with the surface of Vif. This explanation is also in agreement with another report showing that a fragment of APOBEC3G from amino acids 54 -124 is sufficient for binding Vif (47). In addition, the interaction of Vif with the Asp-128 residue of APOBEC3G may need a conformational protein domain instead of a linear protein contact, because the non-functional Vif mutant C114F is unable to efficiently bind and block APOBEC3G activity. Therefore, the lack of aspartic acid at position 128 in APOBEC3G may contribute to reduce the binding affinity for HIV-1 Vif, but is not sufficient to eliminate protein interaction. An alternative explanation that cannot be excluded is that the bacterial system used in our work may lack a co-factor necessary for Vif-APOBEC3G-D128K interaction.
In conclusion, the interaction of APOBEC3G-D128K and HIV-1 Vif does not block deaminase activity and avoids APOBEC3G degradation through ubiquitination and proteasomal pathway (37)(38)(39)(40). In this sense, it is conceivable that a dual functional mechanism may co-exist when HIV-1 Vif interacts with APOBEC3G and simultaneously inhibits and exposes the structural elements that serve as a recognition signal for proteasomal degradation. Because ubiquitin-proteasomal pathway of protein degradation does not exist in bacteria, the predominant effect observed in our study was enzymatic activity inhibition. This link between enzyme function and degradation pathway must be explored with HIV-2 or SIVmac239 Vif proteins, which do not use Asp-128 in APOBEC3G as a target amino acid for triggering its degradation.
Our findings that APOBEC3G-induced cytidine deamination in E. coli can occur by induced transcription may indicate that viral deamination can also occur during transcription. These results provide an interesting insight for variability that may occur in genes actively transcribed. The interaction/blocking role of Vif in this setting may derive from natural co-factors that may complement or negatively regulate its activity (48,49). This speculation is supported by our results showing a consistent small increase in cytidine deamination activity of APOBEC3G and APOBEC3G-D128K in the presence of Vif-C114F and Vif, respectively, indicating a possible complementary role of this protein.
Our results in the bacterial system showed the existence of a strong protein-protein interaction between APOBEC3G and Vif that was not totally overcome by the mutant APOBEC3G-D128K. By using Vif peptides to compete for APOBEC3G-Vif interaction, it is remarkable that several regions of Vif are involved in APOBEC3G interaction. Nevertheless, only two consecutive regions at the C terminus are responsible for a strong interaction with APOBEC3G. These results are in contrast with those of Wichroski et al. (50) that showed isoleucine at position 9 of Vif as the unique responsible amino acid for APOBEC3G interaction. Our results also showed that the Nterminal Vif peptides are important for the inhibition of APOBEC3G-Vif interaction, but not as dramatic as reported by Rana and co-workers (50). However, there are experimental differences between our system and that of Wichroski et al. (50) that may account for this difference. Because we used an in vitro competitive assay to characterize the binding of Vif to APOBEC3G, it is conceivable that in vivo the N-terminal region of Vif may be involved in the regulation of APOBEC3G binding. The C-terminal region of Vif is constituted by positively charged amino acids that can easily interact with a negative domain of APOBEC3G and be responsible for the strong protein-protein binding. If the N-terminal region of Vif regulates this interaction, the C-terminal domain of Vif could function as an anchor to APOBEC3G. Further studies in vivo are necessary to dissociate the role of N-terminal and C-terminal domains of Vif for APOBEC3G binding.
In conclusion, our experiments suggest a complement to the model of Vif-induced degradation of APOBEC3G by bringing to mind that deaminase inhibition can also result from a direct interaction with Vif protein. Therefore, new research efforts are necessary to differentiate the conditions where these mechanisms occur and how they are regulated.