The extensive genetic diversity of HIV-1 provides the foundation for its ability to rapidly escape from antiretroviral drugs in treated HIV-1-infected individuals. Virologic failure occurs when viruses with drug-resistance-associated amino acid substitutions in the targeted viral regions (reverse transcriptase, protease, integrase or envelope) are selected and outgrow the more susceptible strains. Viral properties such as the error rate of the reverse transcriptase enzyme  and recombination [2–4] underlie the rapid emergence of drug-resistant variants. In addition, a novel mechanism mediated by human APOBEC3 (A3) cytidine deaminases (apolipoprotein B mRNA-editing catalytic polypeptide-like 3) has been recently proposed to contribute to viral diversity in HIV-1-infected individuals [5,6] and possibly facilitate antiretroviral escape.
APOBEC3 molecules are cellular enzymes that inhibit various types of endogenous and exogenous retroviruses [7–9]. Several members of APOBEC family (A3A, A3B, A3C, A3DE, A3F, A3G and A3H) show varying degrees of antiviral potency with A3G and A3F generally regarded as exerting the strongest HIV-1 restriction [7,10–12]. APOBEC3 exert their HIV-1 restriction by, at least partially, catalyzing cytidine deamination in the HIV-1 minus-strand DNA during new target cell infection resulting in proviral guanosine (G) to adenosine (A) substitutions in the positive sense DNA strand [13–15]. These substitutions predominantly occur in a GG-to-GA (A3G) or GA-to-AA (A3F) dinucleotide context [8,12,14].
Viral replication in human cells expressing APOBEC3 molecules requires counteracting the antiviral activity of these deaminases by the viral protein Vif. HIV-1 Vif induces polyubiquitylation of APOBEC3 proteins and targets them for degradation by the proteasome, thereby preventing their incorporation into virions [16–18]. The N-terminal region of HIV-1 Vif is important for the specific association with APOBEC3 proteins suppression with several Vif-APOBEC3 interacting domains recently being mapped [19–22].
Polymorphisms in HIV-1 Vif could be important in vivo  since some Vif allelic variants display less anti-APOBEC activity than others [6,23]. Because some HIV-1 drug resistance mutations result from G-to-A substitutions in an A3G/A3F preferred dinucleotide context, recent investigations have focused on the possible contribution of partially active Vif mutants to viral escape from antiretroviral drugs [24,25]. Mulder et al.  demonstrated that partially active Vif alleles result in increased viral diversity and that cytidine deamination coupled with recombination can facilitate the emergence of antiretroviral drug resistance in cell culture.
We hypothesized that partially active Vif alleles should be found more frequently in patients failing antiretroviral treatment if partially defective Vif alleles contribute to the emergence of drug resistance in vivo. In this study we investigate Vif polymorphisms in circulating plasma viruses derived from HIV-1-infected individuals failing antiretroviral treatment and compared them to the Vif diversity found in antiretroviral-naive patients. We show that certain Vif mutations are found expressly in the pretreated patient population and that a specific mutation (K22H) abrogates its antiviral activity in single cycle and spreading infections. We examined its consequences on HIV-1 replication, the G-to-A mutation frequency and the appearance of drug-resistance mutations.
Materials and methods
Plasma samples from 92 HIV-1-infected antiretroviral-experienced patients, for whom HIV-1 genotypic resistance was initiated because of highly active antiretroviral treatment (HAART) failure, were retrospectively analyzed. Resistance testing was performed by sequencing protease and reverse transcriptase regions following the French National consensus technique (www.hivfrenchresistance.org). As control group for this study, we analyzed 65 plasma samples from HIV-1-infected patients who were treatment-naive and for whom HIV genotype testing was performed at the time of diagnosis. This study was approved by the ethical committee of the ANRS AC11.
Amplification and sequencing of patient-derived Vif alleles
RNA isolation, cDNA synthesis, and PCRs were performed as described previously . For the first-round PCR, we used primer Vif1, 5′-cgggtttattacagggacagcag-3′ (position 4899-4922 HxB2) and primer Vif2, 5′-cacccaattctgaaatgaa-3′ (position 5774-5793 HxB2). The second-round PCR was performed using primer Vif3, 5′-tggaaaggtgaaggggca-3′ (position 4956-4974 HXB2) and primer Vif4, 5′-ctaggaaaatgtctaacagcttca-3′ (position 5642-5666 HXB2). The PCR product was purified with an Amicon Microcon-100 centrifugal filter device (Millipore) for sequencing. Vif genes were sequenced in both directions.
LAI Vif was amplified from pLAI.2 (NIH Cat#2532) by PCR and cloned into the expression vector pCRV1 (Vif Eco+ 5′-atatgaattcgccatggaaaacagatggcaggt-3′, Vif Not– 5′-agtcatagcggccgcctagtgtccattcattgtatg-3′). The K22H mutation was introduced into LAI Vif by site-directed mutagenesis (Vif22H+ 5′-agaacatggcacagtttagt-3′, Vif22H- 5′-actaaactgtgccatgttct-3′). Additionally Vif mutation K22H was introduced into the molecular HIV-1 clone pNL4-3 by site-directed mutagenesis using QuickChange II XL site-directed mutagenesis kit (Stratagene).
HEK 293T were maintained in complete medium which consisted of Dulbecco's modified Eagle's medium supplemented with heat-inactivated fetal bovine serum (10%), 2 mmol/l glutamine, 170 mmol/l penicillin, and 40 mmol/l streptomycin at 37°C and 5% CO2. Viral stocks were generated by transfecting 5 × 106 293T cells with plasmids using Lipofectamine Plus reagent (Invitrogen). HIV-1 p24gag antigen concentration in viral supernatants was determined after 48 h using Vidas Ag p24 II (Biomérieux).
Assessment of single-cycle infectivity using reporter cell lines
Viruses were generated by co-transfecting 500 ng of the NL4-3 delta Vif expression plasmid (NIH Cat#2481), 50 ng of Vif LAI or Vif mutant K22H expression plasmid and 100 ng of N-terminally FLAG-tagged human A3G into 293T cells. HEK 293T cells were plated at 1 × 105 cells per well in a 24-well plate and the transfection was done using polyethylenimine. Forty-four h later, the viruses were harvested and 20 ul of the viral supernatant was used to infect TZM-bl reporter cells plated in a 96-well plate at 1 × 104 cells per well. Infections were done in duplicate and β-galactosidase activity was measured 48 h after infection with the Galacto-Star β-galactosidase Reporter Gene System (Applied Biosystems).
Spreading viral infections in T-cell lines
MT-2 T cells, which express A3G (termed ‘nonpermissive’) and Sup-T1 T cells, which express little or no A3G (termed ‘permissive’) were used for the multiple-round infection experiments. Cells were cultured in complete RPMI 1640 culture medium at 37°C and 5% CO2. Viral stocks containing 25 ng of p24gag antigen were added to 1 × 106 SupT-1 and MT-2 cells. Viruses produced were used to infect 5 × 105 SupT-1 and MT-2 cells for a second round of infection. Culture supernatants were collected every 1–3 days over 5 days and p24 gag antigen was quantified. Each experiment was performed in duplicate. Two independent experiments were performed.
Proviral DNA isolation, cloning and sequencing
Total DNA of infected cells was extracted using QIAamp blood DNA minikit (QIAGEN, Courtaboeuf, France). To assess the frequency of proviral mutations, a 835-bp fragment encompassing a fragment of reverse transcriptase gene was amplified [primer 2pols, 5′-ccatacaatactccagtatttgc-3′ (position 2712-2734 HxB2); PL1m 5′-cctgcttctgtatttctgctattaagtcttttg-3′ (position 3514-3546 HxB2)]. Purified PCR products were cloned into the plasmid vector pCR4-TOPO (Invitrogen). Recombinant plasmid DNA was transformed into One Shot TOP10 chemically competent Escherichia coli. Clones DNA were reamplified by PCR for sequencing (546 nt-long-sequenced region). The PCRs and the cloning were repeated three times.
Degradation assay and Western blot
Vif activity with respect to A3G degradation was determined by co-transfection of 200 ng WT A3G-flag and 100 ng WT LAI Vif or K22H LAI Vif in 293T cells using polyethylenimine. 44 h later, the cells were lyzed with 1% SDS lysis solution and the lysates were run a 10% Bis-Tris gel (NuPAGE, Invitrogen). The proteins were then transferred onto a PVDF membrane (Thermo Scientific) and probed with anti-A3G C17 polyclonal antibody (NIH Cat# 10082) and anti-Vif polyclonal antibody (NIH Cat# 2221). The membranes were then incubated with horseradish peroxidase-conjugated secondary antibodies and then developed with SuperSignal West Pico Chemiluminescent Substrate (Pierce).
Comparisons between percentages were performed using the chi-squared or the Fisher's exact test when appropriate and comparisons between nominal and continuous variables were tested using the Mann–Whitney test. The 95% confidence intervals (CIs) were computed using the binomial distribution. Analyses were performed using Statview. Differences in the G-to-A mutation frequencies were analyzed using the Hypermut 2.0 program (http://www.hiv.lanl.gov/content/sequence/HYPERMUT/hypermut.html). A sequence was considered hypermutated if it registered a P value of less than 0.05 on the Fisher's exact test that compared the number of G-to-A changes in APOBEC3 versus control contexts.
Patient characteristics, antiretroviral treatment and drug-resistant variants
A total of 157 HIV-1 plasma viruses subtype B derived from infected patients (92 failing HAART and 65 treatment naive) were analyzed. The patient characteristics are listed in Table 1. The two groups were well balanced for sex, age and CD4+ cell count (CD4+ in HAART treated: 337 cells/μl, naive: 335 cells/μl). Differences between the two groups were noted with respect to plasma viremia (HAART-treated: HIV-1 RNA 4540 copies/ml; naive: 89505 copies/ml).
Patients failing treatment had been prescribed a median of eight different antiretroviral drugs since HAART initiation. All patients received nucleoside/nucleotide reverse transcriptase inhibitor (NRTI), non-nucleoside reverse transcriptase inhibitor (NNRTI) and protease inhibitor in their history. At the time of plasma sample collection, the frequency for mutations in reverse transcriptase and protease leading to reduced susceptibility to NRTI, NNRTI and protease inhibitor was 94.5, 53.3 and 64.1% respectively. We focused on the codon changes underlying drug resistance-associated mutations, in order to specifically screen G-to-A changes in GA/GG dinucleotides context. Such G-to-A changes result in G16E, D30N, M36I, M46I, G73S in PR and D67N, M184I and G190S in reverse transcriptase. Overall, these mutations accounted for 14.6% of all the observed HIV-1 drug-resistance mutations in the dataset. D67N, M46I and M36I were the most prevalent G-to-A mutations (48.9, 38 and 32.6%, respectively) followed by G190S, G73S, G16E and D30N (4–15%).
HIV-1 Vif diversity among pretreated and antiretroviral-naive patients
A total of 157 full-length Vif sequences, amplified from plasma viral RNA, were available for analysis. We compared the polymorphisms in the Vif alleles derived from individuals experiencing virologic treatment failure (n = 92) with those of therapy-naive individuals (n = 65). Overall, the Vif proteins were highly polymorphic with at least half of the positions within the coding region being variable (96/192 residues in therapy-naive patients and 102/192 residues altered in pretreated patients). Using the entropy program (http://www.hiv.lanl.gov/cgi-bin/ENTROPY) to compare the Vif amino acids substitutions between the two populations we identified four single positions that were more polymorphic in pretreated patients: K22 (P = 0.01), Y30 (P = 0.03), L153 (P = 0.05) and K157 (P = 0.05; Fisher's exact test).
Because functional Vif domains are conserved  and mutations in these domains may control interactions with APOBEC3 enzymes, we focused our analysis on the genetic variation within these regions. Figure 1 shows the polymorphisms in six functional domains of Vif (Vif/APOBEC3 interacting domains and the BC-box motif). These regions were either very conserved or equally polymorphic (i.e. R17K, R41K, Y44F, D78E) in both patient populations with the exception of residue K22. Analyzing polymorphisms at position K22, we show that, whereas K22N is equally found in both populations (Fig. 1) and K22I, K22Q and K22T occasionally detected, substitution K22H was detected in 12% (11/92) of the patients failing HAART compared to only 1.5% (1/65) in treatment-naive patients (P = 0.015; Fisher's exact test). Taken together, we identified four positions in Vif that are polymorphic in patients failing HAART (K22, Y30, L153, K157) with a specific substitution at one of these positions (K22H) being found 10 times more often in viruses derived from patients experiencing virologic failure.
Association of Vif K22H with G-to-A mutations in reverse transcriptase and protease derived from pretreated patients
Lysine (K) at position 22 is known to be important for A3G neutralization  and we next aimed to determine whether the K22H mutation in Vif (VifK22H) is associated in vivo with a higher rate of G-to-A mutations. For this purpose, we analyzed the protease-reverse transcriptase (PR-RT) regions of the 11 patients failing HAART and harboring VifK22H viruses and compared them with the 59 PR-RT sequences from pretreated patients harboring Vif K22 WT viruses (VifK22 WT).
The PR-RT nucleotide sequences derived from VifK22H viruses displayed a 21% increase in G-to-A mutations (P = 0.023; Mann–Whitney test). We observed a trend towards increased number of G-to-A mutations in an A3G context (GG-to-AG) compared to sequences from VifK22 WT variants (35% increase, P = 0.049; Mann–Whitney test) (Fig. 2a). Interestingly, when focusing only on the drug-resistance-associated mutations, we noted that 72% (n = 8/11) of VifK22H variants but only 44% (n = 26/59) of VifK22 WT variants harbored at least two drug-resistance-associated mutations in a GA/GG dinucleotide context. Specifically, G16E and M36I were more prevalent (P = 0.022 and P = 0.046) in VifK22H viruses than in VifK22 WT variants (Fig. 2b).
Determination of the anti-A3G activity of Vif mutant K22H
We next assessed directly the anti-A3G activity of Vif mutant K22H using a single cycle infectivity assay. Figure 3a shows that Vif mutant K22H fails to completely neutralize A3G (K22H: average RLU 279; No Vif: average RLU 399). As a control, we used wild-type Vif which, in the presence of A3G, resulted in the production of viruses with 50% infectivity (average RLU 39797) of viruses produced in the absence of A3G (average RLU 70848). This failure of Vif K22H to restore viral infectivity in the presence of A3G correlated with the lack of A3G degradation as assessed by Western blotting of transfected cell lysates (Fig. 3b). The Western blot also reveals that Vif K22H was less expressed than wild type and that the lysine-to-histidine substitution at position 22 affected the migration of the mutant protein (Fig. 3b). Taken together, the Vif mutant K22H is, at least partially, defective for A3G neutralization.
Replication kinetics of K22H Vif mutant NL4-3
We next examined the impact of the Vif K22H mutation on Vif function in the context of the full-length virus. WT NL4-3 and K22H Vif mutant NL4-3 viral stocks were used to infect permissive Sup-T1 cells resulting in the production of WTA3G- and K22HA3G- viruses, and MT-2 nonpermissive cells resulting in the production of WTA3G+ and K22HA3G+. In the second round of infection, measurement of the viral p24gag antigen showed that viruses WTA3G- and K22HA3G- produced on the Sup-T1 cells exhibit similar replication abilities (Fig. 3c). In contrast, using viruses produced in MT-2 cells, K22HA3G+ replication was significantly reduced compared to WTA3G+ since the first days of infection. At day 5, fitness differences between WTA3G+ and K22HA3G+ mutant ranged between 6-fold and 17-fold in two independent experiments (Fig. 3c). These results suggest that NL4-3 Vif mutant K22H fails to neutralize A3G, findings which are in good agreement with the single cycle infectivity results.
Viral diversity in T cells infected with K22HA3G- and K22HA3G+ Vif mutants
To determine the impact of the Vif K22H mutation on A3G-driven mutagenesis of the proviral archives, we collected Sup-T1 and MT-2 cells at the end of the spreading infection experiments, extracted proviral DNA PCR-amplified, cloned and sequenced a 546 nt-long portion of the RT region. A total of 183 clones were analyzed, 30 derived from proviral DNA WTA3G-, 31 from DNA K22HA3G-, 60 from proviral DNA WTA3G+, and 63 from proviral DNA K22HA3G+. As expected, proviral clones derived from WTA3G- and K22HA3G- infections showed very few G-to-A mutations since these infections were conducted in Sup-T1 permissive cells. Among the WTA3G+ proviral clones, only two (1.5%) harbored a single G-to-A mutation (Fig. 4a, 4b).
In contrast, proviral clones derived from K22HA3G+ were highly variable with 57% of the clones (n = 36) harboring at least one G-to-A mutation with the majority of these clones exhibiting a wide range of G-to-A substitutions (range 1–34, mean 9.6) (Fig. 4a, b). Using the Hypermut 2.0 program, 28.5% (n = 18) were scored as heavily hypermutated (P < 0.05; Fisher's exact test). The observed substitutions were almost entirely restricted to G-to-A mutations in a GG or GA dinucleotide context. All severely mutated sequences encoded one or several premature stop codons. These stop codons occurred in the context of a tryptophan codon (TGG-to-TGA) at positions 71, 88, 153, 212, 229, 239 of reverse transcriptase. However, four of the modestly mutated sequences (n = 18, not characterized as ‘hypermutated’ by Hypermut program) encoded an open reading frame (see next paragraph for drug resistance in these clones).
We focused next on drug-resistance mutations in reverse transcriptase that may occur as result of A3G deamination. Mutation M184I in reverse transcriptase associated with high level resistance to the 2′,3′-didéoxy-3′-thiacytidine (3TC)  was detected in 25% (n = 16) of the proviral clones derived from K22HA3G+ infection but not in a single WTA3G+ proviral sequence. Most of the proviruses encoding M184I also harbored stop codons although three of these sequences encoded an intact protein open reading frame with one sequence displaying only one mutation, M184I (Fig. 4b). Interestingly, one clone contained one more drug-resistance-associated mutation resulting from G-to-A change in an A3F favored context: G190E in reverse transcriptase which is known to confer resistance to NNRTIs (Fig. 4b). Taken together, K22H increases considerably the proviral diversity and creates a pool of proviruses encoding specific resistance associated mutations. The emergence of drug-resistance-associated mutations was, under these experimental conditions, fully independent of any drug exposure.
Our understanding on the effects of APOBEC3 on HIV-1 adaptation and diversification in infected patients is currently incomplete [25,27,28]. This study investigates whether mutations in Vif (and indirectly endogenous expression levels of APOBEC3G) facilitates emergence of HIV-1 drug resistance mutations in vivo.
We found four polymorphic positions in Vif (K22, Y30, L153, K157) to be associated with antiretroviral treatment failure by comparing Vif sequences from 92 patients failing antiretroviral treatment and from 65 antiretroviral-naive patients (Fig. 1). We focused on the K22H mutation since it was ten times more frequent in patients failing HAART and, moreover, it was the only one of the four polymorphic positions mapping to a known functional domain of Vif. Future studies will dissect the putative contributions of Vif polymorphisms at positions Y30, L153 and K157.
In addition to genotyping the Vif alleles, we employed a functional test to determine the anti-A3G activity of the K22H mutant and demonstrated that this mutant fails to completely neutralize A3G with evident lack of A3G degradation, suggesting that its interaction with A3G is at least partially altered. It is well established that HIV-1 Vif inactivates A3G and A3F through distinct motifs and that single-nucleotide substitutions in these domains may alter partially or totally the cytidine deaminases suppression . In particular, at position 22, it has been demonstrated that changes in AA can have an impact on Vif function [22,23,29]. Substituting alanine (A) or arginine (R) at this position in Vif retained the ability to interact with A3G and A3F proteins, whereas aspartate (D) or glutamate (E) substitutions abolished interaction with A3G but not A3F [22,29]. Here, we show that histidine (H) 22 also alters Vif function against A3G.
We monitored viral replication of a NL4-3 virus harboring mutation K22H in permissive A3G(-) (Sup-T1) and nonpermissive A3G(+) (MT-2) cells. In good agreement with other studies , we show that this attenuated Vif variant displayed reduced fitness when passaged repeatedly in A3G (+) cells (MT2) but not in A3G (-) cells (SupT1). Contrary to WT NL4-3, infection of MT-2 cells by Vif K22H resulted in a proviral population characterized by a wide spectrum of G-to-A mutations, predominantly in a GG-to-AG context. GG dinucleotides are the specific target for A3G, suggesting that A3G is mainly responsible for the G-to-A mutations.
As a consequence of G-to-A substitutions, a quarter of the K22H proviruses recovered from MT2 cells encoded the 3TC resistance-associated mutation M184I in reverse transcriptase. Similarly, sequences derived from HIV-infected patients  as well as other in-vitro studies  have described M184I in the reverse transcriptase region of G-to-A hypermutated proviruses in the absence of any lamivudine exposure. Although premature stop codons detected in most of these clones likely result in nonfunctional proteins, it is interesting to point to the fact that 18.7% of the M184I carried clones harboring no or only one supplemental substitution (Fig. 4). Furthermore, we analyzed the in-vivo impact of Vif K22H mutation on G-to-A changes observed in sequences derived from pretreated patients failing HAART. Reverse transcriptase and protease bulk sequencing of plasma variants with VifK22H showed a higher frequencies of G-to-A mutations compared to those from VifK22 WT variants and, importantly, we show that these Vif variants harbored significantly more G-to-A-encoded drug-resistance mutations in protease. However, the increased frequency G-to-A mutations observed in sequences derived from plasma HIV-1 K22H RNA was less obvious than what we observed in proviral DNA recovered from in-vitro spreading infections by K22H virus. This observation is in agreement with a recent study describing that proviral DNAs contained the highest level of G-to-A mutations and viral RNA genome contained the least G-to-A mutations . We also need to consider that the population sequencing approach used for genotypic drug resistance testing fails to detect minority populations and we cannot exclude that a small proportion of the plasma viruses K22H carry a higher number G-to-A mutations and/or other G-to-A drug-resistance mutations.
In conclusion, the evidence that patients failing antiretroviral treatment frequently harbored a specific Vif variant (K22H) suggests an important role of APOBEC3 proteins in development of HIV-1 drug resistance in vivo. Patients harboring viral strains encoding Vif K22H could be more likely to develop resistance to certain antiretroviral agents as this mutation is able to decrease the efficacy of Vif function and to promote the selection of some G-to-A drug-resistance mutations. Future studies are needed to establish the predictive value of Vif mutations for treatment failure and establish a causal correlation between Vif function and drug escape during HIV-1 infection.
The research leading to these results has received funding from the Agence Nationale de Recherche sur le SIDA (ANRS), the Association de Recherche en Virologie et Dermatologie (ARVD) and the European Community's Seventh framework Program (FP7/2007-2013) under the project ‘Collaborative HIV and Anti-HIV Drug Resistance Network (CHAIN)’. Viviana Simon is a Sinsheimer Scholar and this work was supported by the NIH/NIAID grant R01 AI064001.
Author contributions: S.F.,V.S.,V.C. and A.-G.M. designed research, were involved in interpretation of the data and wrote the paper.
S.F., I.M., M.B., S.B., M.W. and S.S. performed research.
A.S. and C.K. were responsible for patient management.
V.S., V.C., A.-G.M., S.F., I.M., M.B., S.B., S.S., M.W., A.S. and C.K. did the critical revision of the article. All authors approved the final version of the article.
1. Mansky LM, Temin HM. Lower in vivo mutation rate of human immunodeficiency virus type 1 than that predicted from the fidelity of purified reverse transcriptase. J Virol 1995; 69:5087–5094.
2. Moutouh L, Corbeil J, Richman DD. Recombination leads to the rapid emergence of HIV-1 dually resistant mutants under selective drug pressure. Proc Natl Acad Sci U S A 1996; 93:6106–6111.
3. Jetzt AE, Yu H, Klarmann GJ, Ron Y, Preston BD, Dougherty JP. High rate of recombination throughout the human immunodeficiency virus type 1 genome. J Virol 2000; 74:1234–1240.
4. Levy DN, Aldrovandi GM, Kutsch O, Shaw GM. Dynamics of HIV-1 recombination in its natural target cells. Proc Natl Acad Sci U S A 2004; 101:4204–4209.
5. Pace C, Keller J, Nolan D, James I, Gaudieri S, Moore C, Mallal S. Population level analysis of human immunodeficiency virus type 1 hypermutation and its relationship with APOBEC3G and vif genetic variation. J Virol 2006; 80:9259–9269.
6. Piantadosi A, Humes D, Chohan B, McClelland RS, Overbaugh J. Analysis of the percentage of human immunodeficiency virus type 1 sequences that are hypermutated and markers of disease progression in a longitudinal cohort, including one individual with a partially defective Vif. J Virol 2009; 83:7805–7814.
7. Sheehy AM, Gaddis NC, Choi JD, Malim MH. Isolation of a human gene that inhibits HIV-1 infection and is suppressed by the viral Vif protein. Nature 2002; 418:646–650.
8. Bishop KN, Holmes RK, Sheehy AM, Davidson NO, Cho SJ, Malim MH. Cytidine deamination of retroviral DNA by diverse APOBEC proteins. Curr Biol 2004; 14:1392–1396.
9. Esnault C, Heidmann O, Delebecque F, Dewannieux M, Ribet D, Hance AJ, et al
. APOBEC3G cytidine deaminase inhibits retrotransposition of endogenous retroviruses. Nature 2005; 433:430–433.
10. Wiegand HL, Doehle BP, Bogerd HP, Cullen BR. A second human antiretroviral factor, APOBEC3F, is suppressed by the HIV-1 and HIV-2 Vif proteins. EMBO J 2004; 23:2451–2458.
11. Zheng YH, Irwin D, Kurosu T, Tokunaga K, Sata T, Peterlin BM. Human APOBEC3F is another host factor that blocks human immunodeficiency virus type 1 replication. J Virol 2004; 78:6073–6076.
12. Liddament MT, Brown WL, Schumacher AJ, Harris RS. APOBEC3F properties and hypermutation preferences indicate activity against HIV-1 in vivo. Curr Biol 2004; 14:1385–1391.
13. Harris RS, Bishop KN, Sheehy AM, Craig HM, Petersen-Mahrt SK, Watt IN, et al
. DNA deamination mediates innate immunity to retroviral infection. Cell 2003; 113:803–809.
14. Yu Q, Konig R, Pillai S, Chiles K, Kearney M, Palmer S, et al
. Single-strand specificity of APOBEC3G accounts for minus-strand deamination of the HIV genome. Nat Struct Mol Biol 2004; 11:435–442.
15. Miyagi E, Opi S, Takeuchi H, Khan M, Goila-Gaur R, Kao S, Strebel K. Enzymatically active APOBEC3G is required for efficient inhibition of human immunodeficiency virus type 1. J Virol 2007; 81:13346–13353.
16. Sheehy AM, Gaddis NC, Malim MH. The antiretroviral enzyme APOBEC3G is degraded by the proteasome in response to HIV-1 Vif. Nat Med 2003; 9:1404–1407.
17. Marin M, Rose KM, Kozak SL, Kabat D. HIV-1 Vif protein binds the editing enzyme APOBEC3G and induces its degradation. Nat Med 2003; 9:1398–1403.
18. Conticello SG, Harris RS, Neuberger MS. The Vif protein of HIV triggers degradation of the human antiretroviral DNA deaminase APOBEC3G. Curr Biol 2003; 13:2009–2013.
19. Dang Y, Davis RW, York IA, Zheng YH. Identification of 81LGxGxxIxW89 and 171EDRW174 domains from human immunodeficiency virus type 1 Vif that regulate APOBEC3G and APOBEC3F neutralizing activity. J Virol 2010; 84:5741–5750.
20. He Z, Zhang W, Chen G, Xu R, Yu XF. Characterization of conserved motifs in HIV-1 Vif required for APOBEC3G and APOBEC3F interaction. J Mol Biol 2008; 381:1000–1011.
21. Russell RA, Pathak VK. Identification of two distinct human immunodeficiency virus type 1 Vif determinants critical for interactions with human APOBEC3G and APOBEC3F. J Virol 2007; 81:8201–8210.
22. Chen G, He Z, Wang T, Xu R, Yu XF. A patch of positively charged amino acids surrounding the human immunodeficiency virus type 1 Vif SLVx4Yx9Y motif influences its interaction with APOBEC3G. J Virol 2009; 83:8674–8682.
23. Simon V, Zennou V, Murray D, Huang Y, Ho DD, Bieniasz PD. Natural variation in Vif: differential impact on APOBEC3G/3F and a potential role in HIV-1 diversification. PLoS Pathog 2005; 1:e6.
24. Mulder LC, Harari A, Simon V. Cytidine deamination induced HIV-1 drug resistance. Proc Natl Acad Sci U S A 2008; 105:5501–5506.
25. Albin JS, Harris RS. Interactions of host APOBEC3 restriction factors with HIV-1 in vivo: implications for therapeutics. Expert Rev Mol Med 2010; 12:e4.
26. Malet I, Delelis O, Valantin MA, Montes B, Soulie C, Wirden M, et al
. Mutations associated with failure of raltegravir treatment affect integrase sensitivity to the inhibitor in vitro. Antimicrob Agents Chemother 2008; 52:1351–1358.
27. Berkhout B, de Ronde A. APOBEC3G versus reverse transcriptase in the generation of HIV-1 drug-resistance mutations. AIDS 2004; 18:1861–1863.
28. Pillai SK, Wong JK, Barbour JD. Turning up the volume on mutational pressure: is more of a good thing always better? (a case study of HIV-1 Vif and APOBEC3). Retrovirology 2008; 5:26.
29. Dang Y, Wang X, Zhou T, York IA, Zheng YH. Identification of a novel WxSLVK motif in the N terminus of human immunodeficiency virus and simian immunodeficiency virus Vif that is critical for APOBEC3G and APOBEC3F neutralization. J Virol 2009; 83:8544–8552.
30. Kieffer TL, Kwon P, Nettles RE, Han Y, Ray SC, Siliciano RF. G–>A hypermutation in protease and reverse transcriptase regions of human immunodeficiency virus type 1 residing in resting CD4+ T cells in vivo. J Virol 2005; 79:1975–1980.
31. Russell RA, Moore MD, Hu WS, Pathak VK. APOBEC3G induces a hypermutation gradient: purifying selection at multiple steps during HIV-1 replication results in levels of G-to-A mutations that are high in DNA, intermediate in cellular viral RNA, and low in virion RNA. Retrovirology 2009; 6:16.