The Cellular Antiviral Protein APOBEC3G Interacts with HIV-1 Reverse Transcriptase and Inhibits Its Function during Viral Replication

Abstract

The cytidine deaminase APOBEC3G (A3G) exerts a multifaceted antiviral effect against HIV-1 infection. First, A3G was shown to be able to terminate HIV infection by deaminating the cytosine residues to uracil in the minus strand of the viral DNA during reverse transcription. Also, a number of studies have indicated that A3G inhibits HIV-1 reverse transcription by a non-editing-mediated mechanism. However, the mechanism by which A3G directly disrupts HIV-1 reverse transcription is not fully understood. In the present study, by using a cell-based coimmunoprecipitation (Co-IP) assay, we detected the direct interaction between A3G and HIV-1 reverse transcriptase (RT) in produced viruses and in the cotransfected cells. The data also suggested that their interaction did not require viral genomic RNA bridging or other viral proteins. Additionally, a deletion analysis showed that the RT-binding region in A3G was located between amino acids 65 and 132. Overexpression of the RT-binding polypeptide A3G_65-132 was able to disrupt the interaction between wild-type A3G and RT, which consequently attenuated the anti-HIV effect of A3G on reverse transcription. Overall, this paper provides evidence for the physical and functional interaction between A3G and HIV-1 RT and demonstrates that this interaction plays an important role in the action of A3G against HIV-1 reverse transcription.

INTRODUCTION

Several host proteins have been identified as intrinsic restriction factors because of their ability to inhibit HIV replication and/or dissemination (2, 31, 41, 52). Among them, the apolipoprotein B mRNA-editing enzyme-catalytic polypeptide-like 3G (APOBEC3G; here referred to as A3G) is the one that restricts HIV-1 replication through more than one mechanism (17, 32, 37, 38, 40). In the absence of the HIV-1 viral infectivity factor (Vif), A3G is incorporated into progeny viruses through its interaction with the nucleocapsid (NC) domain of the Gag protein and/or viral RNA (1, 15, 56, 63). Once these progeny viruses initiate new infection, the incorporated A3G will deaminate the cytidine to uridine in the viral minus-strand DNA during reverse transcription, resulting in hypermutation in the provirus. As a result, the HIV-1 proviral DNA will be no longer functional or degrade rapidly (21, 32, 38, 64). Additionally, the mutated proviral DNA may produce defective or truncated viral polypeptides that represent a significant source of major histocompatibility complex class I (MHC-I)-restricted epitopes to activate HIV-1-specific CD8⁺ cytotoxic T lymphocytes (CTLs) (12).

Reverse transcription catalyzed by HIV-1 reverse transcriptase (RT) is a critical step for HIV-1 to establish its replication cycle. In infected cells, RT employs ${tRNA}_3^{Lys}$ and the polypurine tract as primers and converts the viral genomic RNA into double-stranded viral DNA (23). This process is catalyzed by both the DNA polymerase and RNase H activities of RT (23, 49). Interestingly, in addition to the deaminase activity, A3G has also been shown to directly inhibit HIV-1 reverse transcription by a non-editing mechanism (36, 37, 40). Several reports have indicated that catalytically inactivated A3G mutants still exert antiviral effects to a significant extent (26, 42). A number of studies have elucidated the mechanisms underlying the inhibition effect of A3G on reverse transcription. It has been shown that A3G is able to interfere with multiple steps of reverse transcription, including the inhibition of ${tRNA}_3^{Lys}$ primer annealing through an interaction with NCp7 (18–20), the blocking of strand transfers, which consequently reduces late viral DNA synthesis (34, 40), and the suppression of ${tRNA}_3^{Lys}$ cleavage and removal, which produces aberrant viral 3′ long terminal repeat (LTR) ends (40). Moreover, using purified catalytically active A3G protein, it was shown that all reverse transcriptase (RT)-catalyzed DNA elongation reactions were significantly inhibited by A3G (26). Additionally, endogenous reverse transcription assays in cell-free HIV-1 particles demonstrated that A3G reduces HIV-1 viral DNA levels by inhibiting the elongation of reverse transcripts rather than enhancing degradation (10). Taken together, these studies clearly indicate that A3G is able to inhibit the accumulation of viral DNA independently of its deaminase activity.

Because one action of A3G is to inhibit HIV-1 reverse transcription, a characterization of the mechanisms by which A3G targets RT and inhibits its function is of considerable interest, as it may provide a novel insight into the mechanisms underlying the antiviral effect of A3G. In this study, using a cell-based coimmunoprecipitation (Co-IP) assay, we showed that A3G was able to interact with both subunits of HIV-1 RT, and this interaction was not mediated by viral RNA bridging. Furthermore, we have mapped the RT-binding polypeptide in A3G, which is located between amino acids (aa) 65 and 132. This RT-binding polypeptide was able to disrupt the interaction between wild-type A3G and RT. Intriguingly, the results also showed that the presence of the RT-binding polypeptide was able to attenuate the inhibitory effect of A3G on HIV-1 reverse transcription. Overall, this paper provides evidence to support the notion that A3G directly binds to RT, which is an important step for its action against HIV-1 reverse transcription.

MATERIALS AND METHODS

Plasmid constructions.

The plasmid pAS1B-HA-A3G was constructed previously (8), where the APOBEC3G (A3G) amplified from Homo sapiens cDNA clone IMAGE:1284557, was kindly provided by S. K. Petersen-Mahrt (14, 45). The pAS1B-HA-A3G truncation mutants (65 to 132, 102 to 257, and 195 to 384) were constructed by the PCR-based mutagenesis method with the following primers: A3G₆₅-KpnI-5′, 5′-CGGGGTACCTATCCTTATGACGTGCCTGACTATGCCAGCCACCCAGAG-3′; A3G₁₃₂-BamH-3′, 5′-CGGGATCCCTGGTAATCTGGG-3′; A3G₁₀₂-Xba-I-5′, 5′-CTAGTCTAGAAGGGATATGGCCACGTTC-3′; A3G₂₅₇-PstI-3′, 5′-AACTGCAGTCAATGCGGCCTTCAAGGAA-3′; A3G₁₉₅-XbaI-5′, 5′-GGAGATTTCTAGACACTCGATG-3′; and A3G₃₈₄-BglII-3′, 5′-ATAGATCTATCGATTCAGTTTTCCTGATT-3′. ProLabel-HA-A3G was constructed by subcloning hemagglutinin (HA)-A3G into a ProLabel-C vector (Clontech) with the following primers: HA-KpnI-5′, 5′-CGGGGTACCGCTTCTAGCTAT-3′; and A3G-SmalI-3′, 5′-TCCCCCGGGTCAGTTTTCCTG-3′.

To generate the RT expression plasmid, HIV-1 reverse transcriptase (RT) was amplified from HIV-1 HxBru provirus using mutagenesis primers to create a BamHI site in the front with the start codon removed and a NotI site preceded by the stop codon. The PCR fragment was digested with BamHI and NotI and cloned into the SVCMVin-T7 vector (6) that was digested with the same restriction enzymes to generate the plasmid SVCMVin-T7-RT. The T7-RT truncation mutants (1 to 243, 1 to 323, and 1 to 439) were obtained using the same strategy. The following primers were used: RT-BamH-5′, 5′-CGCGGATCCCCCATTAGTCCT-3′; RT-NotI-3′, 5′-ATTTGCGGCCGCCTATAGTACTTTCCT-3′; RT₂₄₃-ClaI-3′, 5′-CCATCGATCTAAGGCTGTACTGTCCATTTA-3′; RT₃₂₃-ClaI-3′, 5′-CCATCGATCTATTTTGATGGGTCATAATAC-3′; and RT₄₃₉-ClaI-3′, 5′-CCATCGATCTACGTTTCTGCTCCTACTATG-3′.

The HxBru Vif⁻ and HxBruΔRI plasmids were constructed previously (6, 8). In HxBru Vif⁻, the amino acids at positions 21 and 22 of Vif were changed to a stop codon (8). In HxBruΔRI, RT and integrase (IN) gene sequences were deleted, while a 194-bp sequence harboring a central polypurine tract/central termination sequence (cPPT/CTS) cis-acting elements was maintained, as previously described (6). The HxBruΔRI-Vif⁻ provirus was constructed by two-step PCR. First, cDNA that encompassed a region between the ApaI and SalI sites of the provirus was amplified, and the amino acids at positions 21 and 22 of Vif were changed to a stop codon. The amplified PCR fragment was then digested with ApaI and SalI restriction enzymes and used to replace the corresponding region in HxBruΔRI. The mutagenic primers were as follows: Vif Stop-5′, 5′-GAGGATTAGAACATGATGACGTGTAGTAAAACACC-3′; and Vif Stop-3′, 5′-GGTGTTTTACTACACGTCATCATGTTCTAATCCTC-3′.

To generate the HA-Vpx expression vector, the cDNA of Vpx was amplified by PCR from simian immunodeficiency virus SIVmac₂₃₉ and cloned into the pAS1B-HA vector. The primers used for PCR was as follows: Vpx-BglII/XbaI-5′, 5′-CGAAGATCTAGAAGTGATCCCAGGGA-3′, and Vpx-ClaI/BglII-3′, 5′-CGTCAGATCTATCGATTATGCTAGTCCTGG-3′.

Antibodies and reagents.

A rabbit anti-HA antibody (Sigma) and a mouse anti-T7-tagged antibody (Novagen) were used for immunoprecipitation. The antibodies used for the Western Blot (WB) analysis were as follows. Purified rabbit anti-hA3G (catalog no. 10201) and anti-RT (catalog no. 6195) polyclonal antisera were obtained through the NIH AIDS Research and Reference Reagent Program. Mouse anti-HIV-1 p24 and rabbit anti-IN monoclonal antibodies were described previously (5, 6, 60). A horseradish peroxidase (HRP)-conjugated anti-T7 antibody was purchased from Novagen. As secondary antibodies, HRP-conjugated donkey anti-rabbit IgG and sheep anti-mouse IgG were purchased from Amersham Biosciences. A WB-detection ECL kit was purchased from PerkinElmer Life Sciences. NP-40 was purchased from Calbiochem.

Cell lines and transfection.

Human embryonic kidney 293T cells were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum (FCS) and 100 U/ml penicillin and 100 μg/ml streptomycin. The CD4⁺ C8166 T cell lines were maintained in RPMI 1640 medium containing 10% FCS and 100 U/ml penicillin and 100 μg/ml streptomycin. DNA transfection of 293T cells was performed with the standard calcium phosphate DNA precipitation method (62).

Virus production and infection.

To produce Vif⁻ HIV-1 viruses, 293T cells were cotransfected with the corresponding HIV-1 proviral DNA and HA-A3G plasmids. The supernatants were collected at 48 h posttransfection and subjected to ultracentrifugation (35,000 rpm for 1.5 h at 4°C) to pellet the virus. The quantification of virus stocks was determined by Gag-p24 measurements using an HIV-1 p24 enzyme-linked immunosorbent assay (ELISA) kit (purchased from the AIDS Vaccine Program of the Frederick Cancer Research and Development Center) (61). For the WB analysis, viruses were pelleted by ultracentrifugation through a 20% sucrose cushion. The lysed virus samples were resolved by 10% SDS-PAGE followed by WB.

To infect CD4⁺ C8166 T cells, equal amounts of viruses were incubated with a control T cell line or the cell line expression HA-A3G at 37°C for 4 h. The cells were then washed and incubated with fresh medium. At different time points, the virus-containing supernatants were collected and pelleted for WB analysis as described above.

Real-time PCR analysis.

CD4⁺ C8166 T cells were infected with equal amounts of Vif⁻ viruses produced from 293T cells. Heat-inactivated virus (pretreated at 80°C for 30 min) was used as a negative control for infection. Prior to infection, viruses were treated with 340 U/ml DNase (Roche) for 60 min in 37°C to remove residual plasmid DNA. After 2 h of infection, the cells were washed with phosphate-buffered saline (PBS) and cultured in RPMI medium. At 12 h postinfection, the infected cells were collected, and DNA was extracted using a QIAamp DNA blood minikit by following the manufacturer's instructions (Qiagen). The DNA samples were processed for detecting total viral DNA synthesis using the following primers: TD-Gag-Fr, 5′-ATCAAGCAGCCATGCAAATG-3′; TD-Gag-Rv, 5′-CTGAAGGGTACTAGTAGTTCC-3′ and TD-Gag-Rv2, 5′-GTTCCTGAAGGGTACTAGTAG-3′. To evaluate the DNA content of the extracted chromosomal DNA preparations, a detection of the human β-globin gene was performed by real-time PCR as described previously (53).

Co-IP assay of 293T cells and the produced HIV-1 virions.

To detect the interaction of A3G-RT and identify their binding regions, the cell-based Co-IP assay was performed as previously described (37). Briefly, the T7 or T7-RTwt/mutant (where wt is wild-type) plasmid was cotransfected with HA-A3Gwt/mutant into 293T cells for 48 h. Next, 90% of the transfected cells were lysed in 0.5% NP-40 prepared in 199 medium containing a cocktail of protease inhibitors (Roche) for 30 min and clarified by centrifugation at 14,000 rpm for 30 min at 4°C. After centrifugation, the clarified supernatant was subjected to immunoprecipitation using a mouse anti-T7 antibody. The immunoprecipitates were then resolved by 10% SDS-PAGE. The RT-bound proteins were detected by WB using anti-HA or anti-A3G antibodies. In addition, the presence of T7-RTwt/mutant in the immunoprecipitates was detected by an anti-T7 antibody. To detect the expression of HA-A3Gwt/mutant and T7-RTwt/mutant, 10% of the transfected cells were lysed in 0.5% NP-40 199, and the lysates were analyzed by WB with corresponding antibodies. For the ProLabel assay, the ProLabel activity of ProLabel-HA-A3G and the ProLabel-T antigen from the Co-IP samples and cell lysates were measured by using a ProLabel detection kit II (Clontech Laboratories, Inc.).

To detect the interaction between HA-A3G and RT in the virions, Vif⁻ virions were lysed with 0.5% NP-40 in 199 and immunoprecipitated with a mouse anti-HA antibody, followed by WB with an anti-RT antibody to detect A3G-bound RT. In the RNase treatment assay, the viral lysates in 199 medium (containing 0.5% NP-40) were pretreated with 100 μg/ml RNase A (Invitrogen) for 1 h before immunoprecipitation.

RNA isolation and detection of HIV-1 integrase mRNA.

RNA was extracted from the viral lysates using TRIzol reagent (Invitrogen). Isolated RNA was subjected to real-time PCR with Moloney murine leukemia virus reverse transcriptase (Promega). The reverse transcribed DNA was then amplified with Taq DNA polymerase (Fermentas) using the following primers: IN-5′, 5′-GAAAGTAGGAGCTCTAGATGGAATA-3′; and IN-stop-3′, 5′-CTAAACGGATCCATGTTCTAA-3′. The PCR products were electrophoresed through a 1% agarose gel.

RESULTS

The inhibitory effect of A3G on reverse transcription does not require the presence of integrase.

In this study, we first tested the effect of A3G on HIV-1 reverse transcription using real-time PCR (4) to quantitatively measure the accumulation of proviral DNA at 12 h postinfection. First, the HIV-1 Vif⁻ virus and the HIV-1 Vif⁻ virus encapsidated with HA-A3G (HIV-1-Vif⁻+A3G) were produced from 293T cells by cotransfection of the HxBru-Vif⁻ provirus (8) with an HA-tagged plasmid or an HA-A3G expression plasmid. After the virus stocks were collected by ultracentrifugation, the presence of different virus-associated proteins, including HIV-1 reverse transcriptase (RT) and Gag-p24, and encapsidated HA-A3G in each virus stock sample were detected by Western blotting (WB) analysis using corresponding antibodies (Fig. 1A, left panel). Next, equal amounts of HIV-1 Vif⁻ and HIV-1 Vif⁻+A3G viruses were used to infect CD4⁺ C8166 T cells. At 12 h postinfection, the amounts of late reverse transcription (LRT) products were measured as described in Materials and Methods. The level of LRT in HIV-1 Vif⁻ virus-infected cells was arbitrarily set as 100%. The results showed that the LRT level after infection with the HIV-1 Vif⁻+A3G virus was reduced to less than 20%, compared to that in HIV-1 Vif⁻-infected cells (Fig. 1A, right panel). Since the reverse primer TD-Gag-Rv (5′-CTGAAGGGTACTAGTAGTTCC-3′) used for the measurement of LRT products (Fig. 1A) contained a CC pair at the 3′ end, which may be a potential target for A3G deaminase, we also quantified the LRT products by using another reverse primer, TD-Gag-Rv2, which has no CC pair near the 3′ end. Results showed that the two primers, TD-Gag-Rv and TD-Gag-Rv2, detected similar levels of HIV-1 proviral DNA from A3G⁺ and A3G⁻ HIV-1-infected cells (data not shown), suggesting that the decrease in the LRT level in A3G⁺ HIV-1-infected cells was due to A3G-mediated inhibition of HIV reverse transcription, but not through an impact on real-time PCR efficiency. All together, in agreement with previous reports (9, 24, 36, 40), our results indicate that the presence of A3G disrupted HIV-1 reverse transcription in the absence of the Vif protein.

Fig 1.

Inhibitory effect of A3G on HIV-1 reverse transcription. (A) Vif⁻ HIV-1 virions were produced from 293T cells in the presence or absence of A3G. Virus-associated proteins, including HIV-1 RT, Gag p24, and incorporated HA-A3G were analyzed by WB with the corresponding antibodies (left panels). Also, equal amounts of viruses (normalized by amounts of HIV-1 Gag p24) were used to infect CD4⁺ C8166 T cells. Twelve hours postinfection, total DNA was extracted from the infected cells. HIV-1 late reverse transcription (LRT) products were analyzed by real-time PCR (right panels). The level of LRT in HIV-1 Vif⁻ virus-infected cells was set to 100%, with errors bars representing the standard deviations from three independent experiments. (B) Schematic structure of the RT/IN-deleted HIV-1 provirus HxBruΔRI-Vif⁻ and the Vpr-RT-IN and Vpr-RT expression plasmids, which have been previously described (5, 7, 65). PROM, promoter; PCS, protease cleavage site. (C) WB analysis of the RT/IN trans-complemented HxBruΔRI-Vif⁻ viruses, which were produced from cotransfected 293T cells with HxBruΔRI-Vif⁻ and Vpr-RT-IN or Vpr-RT in the presence or absence of A3G. Briefly, viruses were collected, lysed, and loaded into SDS-PAGE gels. The presence of viral proteins and HA-A3G was analyzed by WB with corresponding antibodies. (D) Real-time PCR analysis of the level of LRT in C8166 T cells infected with equal amounts of the trans-complemented HxBruΔRI viruses in the presence or absence of A3G. The level of LRT was measured at 12 h postinfection by real-time PCR as in panel A. The values presented are the means and standard deviations from three independent experiments. Panels A and D show a representative WB image from three independent experiments.

HIV-1 reverse transcription was previously shown to be facilitated by the viral protein integrase (IN) (58), which is known to be able to interact with RT and A3G (36). Therefore, the A3G-induced reverse transcription inhibition may have been caused by the interaction between A3G and HIV-1 IN, which may interfere with the effect of IN on reverse transcription. To test this possibility, we modified a previously described HIV-1 single-cycle replication system (7) by constructing a Vif⁻ RT-IN-deleted HIV-1 provirus (HxBruΔRI-Vif⁻) and a Vpr-RT expression plasmid (Fig. 1B). The viruses produced by coexpression of the HxBruΔRI-Vif⁻ provirus with the Vpr-RT-IN or Vpr-RT plasmids were able to enter cells and process reverse transcription. First, different viruses were produced from cotransfected 293T cells in the presence or absence of HA-A3G. The viral protein compositions and incorporated HA-A3G were detected by WB using specific antibodies as indicated in Fig. 1C. The results showed that HIV-1 IN was detected only in viruses produced from 293T cells cotransfected with the HxBruΔRI-Vif⁻ provirus with the Vpr-RT-IN plasmid (Fig. 1C, middle panel, compare lanes 1 and 2 to lanes 3 and 4). Moreover, the data indicated that the RTp66, RTp51, and Gagp24 levels in the virions were not affected by the presence of IN or HA-A3G (Fig. 1C, upper and middle panels). Equal amounts of viruses (adjusted by p24) were used to infect CD4⁺ C8166 T cells. After 12 h of infection, the level of HIV-1 LRT was quantified by real-time PCR. When the level of LRT in the HIV-1 virus trans-complemented with Vpr-RT-IN-infected cells was arbitrarily set as 100%, the amount of LRT was reduced to 70% when the infecting viruses lacked IN (Fig. 1D, compare bar 3 to bar 1), confirming that the presence of IN increased reverse transcription. Interestingly, our data also showed that regardless of the presence or absence of HIV-1 IN, A3G was able to efficiently inhibit viral DNA synthesis by more than 80% (Fig. 1D), indicating that the inhibitory effect of A3G on reverse transcription does not require the presence of IN.

Interaction between A3G and RT in the virions.

For a better understanding of the mechanism by which A3G affects HIV reverse transcription, we performed a coimmunoprecipitation (Co-IP) assay to detect the possible interaction between RT and A3G in the virus. Briefly, equal amounts of HIV-1 Vif⁻ and HIV-1 Vif⁻+A3G viruses, produced from 293T cells by cotransfected with the HxBru-Vif⁻ provirus with an HA tag or HA-A3G, were lysed and immunoprecipitated with an anti-HA antibody to pull down the HA-A3G protein. The coimmunoprecipitated RT protein was then detected by WB using an anti-RT antibody. Interestingly, the results revealed that both RTp66 and RTp51 were coimmunoprecipitated with A3G (Fig. 2A, lane 2), suggesting that A3G is able to interact with RT. To rule out the possibility that the coimmunoprecipitated RT was due to the variable amounts of viral protein in each viral sample, we also detected the RT level in the lysed virions. The results showed no differences in the RT p66 and p51 levels between the different viral samples (Fig. 2A, compare lane 3 to lane 4).

Fig 2.

A3G interacts with HIV-1 reverse transcriptase in the virus. (A) To test the A3G-RT interaction in the virus, the HIV-1 Vif⁻ provirus was cotransfected with the pAS1B-HA or pAS1B-HA-A3G expression plasmid in 293T cells. Seventy-two hours posttransfection, viruses were purified from the culture supernatants. The viral lysates were immunoprecipitated with an anti-HA antibody followed by WB with anti-RT or anti-HA antibodies (lanes 1 and 2). The expression of virus-associated RT and HA-A3G was also detected by directly loading viral lysates into SDS-PAGE gels followed by WB with the corresponding antibodies (lanes 3 and 4). A nonspecific band below RT p51 was found in each testing sample. (B) The A3G-RT interaction was not mediated by the RNA bridge. HIV-1 Vif⁻ viruses produced from 293T cells in the presence or absence of HA-A3G were either untreated (lanes 1 and 3) or treated with RNase (lane 2) and then lysed. After lysis, 50% of viral samples were subjected to immunoprecipitation with anti-HA antibody. The precipitated samples (left upper panels) and other 25% of viral samples (left lower panels) were then loaded into SDS-PAGE gels followed by WB with anti-RT, anti-IN, and anti-HA antibodies, respectively. Meanwhile, the remaining 25% of viral lysates were subjected to viral RNA isolation followed by reverse transcription. The reverse-transcribed DNA was detected by PCR and analyzed by agarose gel electrophoresis (right panel). NC stands for negative control. (C) To test the effect of A3G on HIV-1 RT processing, the Vif⁻ viruses produced in 293T cells at the various expression levels of HA-A3G were collected, lysed, and loaded into SDS-PAGE gels. The RT, Gag p24, and HA-A3G expression levels were analyzed by WB using anti-RT (top panel), anti-p24 (middle panel), and anti-HA (lower panel) antibodies. The total amounts of RT p66 and p51 from two independent experiments were measured by imaging analysis. The p66/p51 ratios in the viral lysates are shown at the bottom of the RT panel. (D) The Vif⁻ viruses produced in A3G⁻ C8166 cells (lanes 1 and 3) or A3G-expressing C8166 T cells (lanes 2 and 4) after 2 to 3 days of infection were collected, lysed, and loaded into SDS-PAGE gels. The levels of RT and Gag p24 were detected by anti-RT antibody (top panel) or anti-p24 antibody (lower panel). The p66/p51 ratios in lysates from two independent experiments are shown at the bottom of the RT panel. Panels A and B show representative WB results from three independent experiments.

As a member of the APOBEC family, A3G shares the property of RNA binding (28, 44), which elevates the A3G packaging efficiency into virions (30, 51, 56) and enhances the stability of the A3G association with the reverse transcription complex (RTC) (30). Given that both RT and A3G are RNA-binding proteins, we evaluated the possible role of RNA on the A3G and RT interaction. The viral lysates were initially treated with RNase or left untreated and then divided into two parts. One part was subjected to a Co-IP assay to detect the RT/A3G and IN/A3G interaction (Fig. 2B, lanes 1 to 3). The other part was used to test the efficiency of the RNase treatment (Fig. 2B, lanes 4 to 6). Upon RNase treatment, the HIV-1 gene sequence was unable to be amplified, confirming that viral RNA was efficiently digested (Fig. 2B, compare lane 4 to lane 5). Meanwhile, the IN/A3G interaction was reduced upon RNase treatment (Fig. 2B, second panel, compare lane 2 to lane 3), which further confirmed the efficiency of RNase treatment, as IN/A3G interaction was RNA dependent (36). However, the binding of A3G and HIV-1 RT was not significantly affected by the RNase treatment (Fig. 2B, first panel, compare lane 2 to 3), suggesting that their interaction does not require the presence of viral RNA.

In the HIV-1 virion, RT forms a heterodimer with two subunits, p66 and p51 (57). p51 is derived from p66 by proteolytic processing during viral maturation. Although the p66 subunit contains activity sites for both of the catalytic activities of RT, the p51 subunit has been shown to be essential for maintaining the structural stability of the RT heterodimer (27, 46, 49). RT dimerization is believed to be required for RT enzymatic activity (47, 48). Because A3G interacts with both the p66 and p55 subunits of RT in the virus, we wanted to test whether the proteolytic processing of RT was affected in the presence of A3G. The HxBru-Vif⁻ provirus was cotransfected with various amounts of the A3G plasmid into 293T cells. After 48 h, virions were collected, and the amounts of RT p66, p51, p24, and incorporated A3G were determined by WB with the corresponding antibodies (Fig. 2C). The results showed that A3G was packaged into the Vif⁻ virus in a dose-dependent manner (lower panel). Additionally, a comparable amount of RT p66 and RT p51 was presented in each viral sample (upper panel). We set the virus-associated p66/p51 ratio in the absence of A3G as 1.00 and calculated the relative p66/p51 ratio in all of the other samples. As shown at the bottom of the RT panel, the levels of p66/p51 were similar for all of the samples. In addition, we further analyzed the RT proteolytic processing of viruses released from the HIV-1-infected control CD4⁺ T cells and from the infected CD4⁺ T cells that stably expressed a low level of HA-A3G (8). A similar p66/p51 ratio was obtained from HA-A3G and the control cell lines (Fig. 2D). All of these results provide evidence that the A3G-RT interaction has no effect on RT proteolytic processing.

A3G-RT interaction does not require the presence of other HIV-1 proteins.

Having demonstrated that A3G interacts with RT in the viral particles and that the interaction does not require the presence of viral RNA, we further examined their association in 293T cells where no other viral proteins were expressed. The plasmid, T7-RT, was cotransfected with HA-A3G or a ProLabel-A3G (PL-A3G) expression vector into 293T cells (Fig. 3A). PL-T (a simian virus 40 [SV40] large T antigen fused at the C terminus of the PL tag) was cotransfected with T7-RT as a control. Forty-eight hours after transfection, cells were lysed and immunoprecipitated with an anti-T7 antibody. Bound HA-A3G and PL-A3G were determined by WB using an anti-HA antibody and by measuring PL activity, respectively (Fig. 3B and C). The results showed that the immunoprecipitation of T7-RT was able to specifically pull down HA-A3G (Fig. 3B). Additionally, significant PL activity was detected only from the precipitated T7-RT sample in the 293T cells coexpressing T7-RT and PL-A3G but not in the 293T cells coexpressing T7-RT and PL-T (Fig. 3C, left panel), even though similar levels of PL-A3G and PL-T were detected in 293T cells (Fig. 3C, right panel). All of these experiments demonstrated a specific interaction between RT and A3G in the absence of other viral proteins.

Fig 3.

A3G-RT interaction in the absence of other viral proteins. (A) Schematic structure of the T7-tagged RT, HA-tagged, and ProLabel-tagged A3G expression plasmids. ProLabel is a fragment of the split β-galactosidase, which has no enzymatic activity. However, the ProLabel-tagged fusion protein can recombine with a Ω fragment of β-galactosidase to reconstitute an active enzyme which is able to cleave chemiluminescence substrate. (B) To test RT/A3G interaction in the absence of other viral proteins, the pAS1B-HA-A3G plasmid was cotransfected with SVCMVin-T7 or SVCMVin-T7-RT expression plasmid into 293T cells. After 48 h, cells were lysed, and 10% of them (input) were analyzed by WB using anti-HA or anti-T7 antibodies to detect HA-A3G and T7-RT (lower two panels). Meanwhile, the remaining cell lysates were immunoprecipitated by the anti-T7 antibody to pull down T7-RT protein, and the coprecipitated HA-A3G was detected by WB with anti-HA antibody (upper panel). (C) The A3G-RT interaction was also detected by a chemiluminescence-based enzyme complementation (ProLabel [PL]) assay. Briefly, T7-RT and ProLabel-A3G (PL-A3G) were coexpressed in 293T cells. After 48 h, cells were lysed and immunoprecipitated by an anti-T7 antibody. The Co-IP samples (left panel) and cell lysates (right panel) were analyzed by the ProLabel activity assay. Values are averages from three independent experiments, and error bars indicate standard deviations from the means. Statistical significance was calculated using Student's t test, and P values are shown above the bars. RLU, relative light units.

Delineation of the regions in RT and A3G required for their interaction.

To define the region in HIV-1 RT that is required for binding to A3G, we constructed a series of T7-tagged RT deletion mutants, including T7-RT_1-243, T7-RT_1-323, and T7-RT_1-439 (Fig. 4A). 293T cells were cotransfected with HA-A3G and wild-type RT (T7-RT_wt) or each truncated mutant. In parallel, 293T cells cotransfected with HA-A3G and T7-tagged plasmids were used as a negative control. As shown in Fig. 4B, anti-T7 immunoprecipitation was able to pull down a similar level of individual T7-RT wt/mutant, and comparable amounts of A3G were co-pulled down with T7-RT_wt or all of the truncated T7-RT proteins, including a RT-truncated mutant with 318 amino acids removed from its C terminus (RT_1-243) (Fig. 4B, upper panel, lanes 2 to 5). In contrast, no bound A3G could be detected in the HA-A3G/T7-tag-coexpressing sample (Fig. 4B, upper panel, lane 1). To rule out the possibility that this difference was due to various levels of HA-A3G in each sample, we directly measured the amount of HA-A3G in the cell lysates by WB, and the data showed that similar levels of HA-A3G were detected in each sample (Fig. 4B, lower panel). These data suggest that the N-terminal fingers-palm domain of RT (RT_1-243) is sufficient for binding to A3G.

Fig 4.

Mapping the binding domains in both RT and A3G required for their interaction. (A) Schematic representation of the different T7-tagged RT wt/mutant constructs used in the domain-mapping experiments. Full-length T7-RT is shown at the top. The numbers indicate the amino acid positions in RT. The domains are marked in the full-length RT. Different RT truncated mutants are shown at the bottom. (B) Interaction of HA-A3G with T7-RT wt/mut. HA-A3G was cotransfected with T7-RTwt/mut in 293T cells for 48 h. The association of A3G with truncated RT was analyzed by Co-IP with an anti-T7 antibody followed by WB with an anti-HA antibody (upper panel). The immunoprecipitated T7-RTwt/mut was detected by an anti-T7 antibody (middle panel). The HA-A3G expression levels in the cell lysates were evaluated by WB with an anti-HA antibody (lower panel). (C) Schematic representation of the different HA-A3G wt/mutant constructs used in the domain-mapping experiments. Full-length HA-A3G is shown at the top. The numbers indicate the amino acid positions in A3G. The domains marked are the cytidine deaminase domain (CDD), the linker (LINK), and the pseudoactive site (PAS). Different A3G truncated mutants are shown at the bottom. (D) Interaction of T7-RT with HA-A3G wt/mut was detected in 293T cells cotransfected with SVCMVin-T7-RT plasmid and each of the A3G mutants. After 48 h of transfection, the interaction of each A3G truncated mutant with RT was analyzed by Co-IP with anti-T7 antibody followed by WB with an anti-HA antibody (upper panel). The immunoprecipitated T7-RT was detected by anti-T7 antibody (middle panel). The expression levels of HA-A3Gwt/mutant in the cell lysates were detected with an anti-HA antibody (lower panel). The images represent the results of three independent experiments.

Similarly, to map the regions in A3G required for the A3G-RT interaction, we constructed a panel of A3G N- and/or C-terminal deletion mutants based on the locations of the two cytidine deaminase domains (CDDs) in A3G (13, 25, 28), including HA-A3G_65-132, HA-A3G_102-257, and HA-A3G_195-384 (Fig. 4C). Each of these mutants was cotransfected with a T7-RT or T7-tagged plasmid into 293T cells. In parallel, HA-A3Gwt was included in the experiments. As shown in Fig. 4D, the expression of the different A3G mutants varied (lower panel). Interestingly, despite the low level of expression, the mutant HA-A3G_65-132 was able to bind to RT efficiently (Fig. 4D, lane 4). In contrast, the truncated mutants HA-A3G_102-257 and HA-A3G_195-384 did not bind to RT (Fig. 4D, lanes 6 and 8). Collectively, these data indicate that the N-terminal domain of A3G that encompasses amino acids 65 to 132 and contains the first CDD and part of the linker region is the necessary region for A3G to bind to HIV-1 RT.

The RT-binding polypeptide A3G_65-132 inhibits both A3G-RT interaction and the effect of A3G on HIV-1 reverse transcription.

Because A3G_65-132 is able to bind to HIV-1 RT, it is possible that this peptide could compete with wild-type A3G for RT binding. To test this possibility, we performed competition Co-IP experiments with 293T cells by coexpressing T7-RT and HA-A3G_wt with HA-A3G_65-132, HA-A3G_102-257, and HA-A3G_195-384. After the Co-IP assay with anti-T7 precipitation, followed by WB with an anti-HA antibody, we found that the presence of the RT-binding peptide A3G_65-132 significantly disrupted the binding of HA-A3Gwt with RT (Fig. 5A, top panel, lane 3). In contrast, the presence of equal amounts of HA-A3G_102-257 or larger amounts of HA-A3G_195-384 showed no inhibition effect on the A3G-RT interaction (Fig. 5A, lanes 4 and 5). This binding difference among each sample was not due to the variation of HA-A3Gwt and T7-RT expression, because similar amounts of HA-A3Gwt and T7-RT were detected in different samples (Fig. 5A, two middle panels). Taken together, these data indicated that A3G_65-132 did not influence the expression of A3Gwt and RT but specifically competed with HA-A3Gwt to bind to HIV-1 RT.

Fig 5.

A3G_65-132 both disrupts the A3G-RT interaction and attenuates A3G's inhibitory effect on HIV-1 reverse transcription. (A) A3G_65-132 blocks the binding of A3Gwt with HIV-1 RT, HA-A3Gwt, SVCMVin-T7-RT, and each HA-A3G truncated mutant plasmid was cotransfected into 293T cells for 48 h. The interaction between HA-A3Gwt and T7-RT was analyzed by Co-IP with anti-T7 antibody followed by WB with an anti-HA antibody (upper panel). The immunoprecipitated T7-RT was visualized by WB with anti-T7 antibody (second panel). The expression of HA-A3Gwt in the cell lysates was detected with anti-A3G antibody (third panel). HA-A3G mutants were detected with an anti-HA antibody (lower panel). (B) To evaluate the virus incorporation of A3G in the presence of HA-A3G_65-132, the HxBru-Vif⁻ provirus was cotransfected with HA-A3Gwt or HA-A3G_65-132 alone or with the HA-A3Gwt and HA-A3G_65-132 expression plasmids. After 48 h of cotransfection, virions were collected by ultracentrifugation, lysed, and loaded on SDS-PAGE gels. The presence of A3G and viral protein Gag p24 in each viral sample was analyzed by WB with anti-A3G (upper panel) and anti-p24 (lower panel) antibodies. (C and D) To test the effect of A3G_65-132 on HIV-1 reverse transcription and viral replication, equal amounts of viruses (normalized by p24 ELISA) were used to infect CD4⁺ C8166 T cells. Twelve hours postinfection, total DNA was extracted from the infected cells. The level of HIV-1 LRT was analyzed by real-time PCR. Meanwhile, at different time points after infection, the supernatants from the infected cell cultures were collected (C), and viral replication was monitored by measuring the level of the HIV-1 Gag p24 antigen in the supernatant (D). Panels A and B show representative WB results from three independent experiments. The values presented in panel C are the means and standard deviations from three independent experiments. P values determined by the t test are shown above the bars of compared groups. The HIV p24 values presented in panel D are the means and ranges of values from two independent experiments.

Because A3G_65-132 blocks the binding of A3G to RT, we asked whether HA-A3G_65-132 could impact the antiviral activity of A3Gwt. To test this possibility, we produced Vif⁻ HIV-1 from 293T cells by cotransfecting the HxBru-Vif⁻ provirus with HA-A3Gwt or HA-A3G_65-132 alone or with HA-A3Gwt and HA-A3G_65-132 expression plasmids together. After the viruses were harvested, the amount of A3Gwt packaged into the viruses was measured by WB. The results showed that both A3Gwt and HA-A3G_65-132 was able be incorporated into virus (Fig. 5B, lane 2 to 4). Meanwhile, an HA-tagged Vpx (SIVmac₂₃₉) was unable to be packaged into the virus (data not shown) (55), thus demonstrating the specificity of HA-A3G_65-132 incorporation. Interestingly, the association of the HA-A3G_65-132 polypeptide did not appear to affect the packaging of A3Gwt (Fig. 5B, compare lane 3 to 2). To further test the effect of A3Gwt and HA-A3G_65-132 on virus infection, each virus stock shown in Fig. 5B was used to infect C8166 T cells, and after 12 h, the viral reverse transcription was evaluated by real-time PCR. The results showed that in the cells infected by the Vif⁻ viruses containing only A3Gwt, the level of LRT was reduced to 20%, which was similar to the results shown in Fig. 1A. In contrast, after cells were infected by the virus produced in the presence of both HA-A3Gwt and HA-A3G_65-132, the detected LRT level reached approximately 60% of the wild-type virus infection level without A3Gwt (Fig. 5C, compare bar 3 to bar 1). Meanwhile, the effect of HA-A3G_65-132 alone was also tested and showed no significant inhibition on HIV-1 reverse transcription (Fig. 5C, bar 4). Furthermore, we also monitored viral replication after infection by each viral stock indicated in Fig. 5B. Results showed that the virus produced in the presence of A3Gwt lost its infection ability (Fig. 5D). Interestingly, the data also revealed that the virus produced in 293T cells expressing both A3Gwt and HA-A3G_65-132 was able to induce a low-level infection that reached approximately 15% of the wild-type infection level (Fig. 5D). All of these results indicate that the presence of A3G_65-132 polypeptide is able to inhibit the anti-HIV effect mediated by A3G.

DISCUSSION

Several intrinsic anti-HIV factors, including TRIM5α, A3G, tetherin, and SAMHD1, have been identified recently (50, 52, 54). Among them, A3G is a member of the cytidine deaminase family of nucleic acid-editing enzymes which have been found to confer broad protection against retroviruses, including HIV-1, and limit the cross-species transmission of these pathogens (39, 52). It is well known that one of the main functions of A3G is to terminate HIV infection by deaminating cytosine residues to uracil in the viral DNA minus strand during reverse transcription, resulting in G-to-A mutations in the HIV-1 provirus (21, 32, 38, 64). In addition, a number of other studies have reported that A3G also inhibits HIV-1 reverse transcription soon after Vif⁻ virus entry into the cells, and this anti-reverse transcription activity was not correlated with the deaminase activity of the protein (9, 10, 24, 42, 59), indicating that A3G exerts a multifaceted antiviral effect against HIV-1 infection. However, until now, it has been unclear whether A3G requires binding to HIV-1 RT to disrupt its function. In this study, using a Co-IP assay, we have demonstrated that A3G specifically binds to HIV-1 RT in the viruses, and their interaction did not require other HIV-1 proteins or viral RNA. In addition, our deletion analyses indicate that a region of A3G (residues 65 to 132) containing the first CDD and part of the linker region, and the RT fingers-palm domain, were required for their interaction. Interestingly, the A3G_65-132 polypeptide was shown to be able to block the A3G-RT interaction and attenuate the anti-HIV effect of A3G. Thus, this study provides evidence for an interaction between A3G and HIV-1 RT, which may be a necessary step for A3G to inhibit HIV-1 reverse transcription.

Previous studies showed that A3G was able to interrupt several steps of HIV-1 reverse transcription, including the inhibition of ${tRNA}_3^{Lys}$ primer annealing (18–20), the blocking of strand transfers (34, 40), and/or the suppression of ${tRNA}_3^{Lys}$ cleavage and removal (40). By studying the properties of A3G using purified A3G and RT proteins, Iwatani et al. proposed a “roadblock” mechanism to explain the inhibition of A3G on all RT-catalyzed DNA elongation reactions (26). Their study indicated that A3G binds to nucleic acid with a higher affinity than RT, which consequently blocks the movement of RT along the template. Conversely, our study provides evidence for the interaction of A3G with HIV-1 RT, and an RT-binding peptide derived from A3G interfered with the ability of A3G to both bind to RT and inhibit HIV-1 reverse transcription. This observation raises the possibility that A3G binding to RT may be one of the prerequisite steps for its inhibitory effect on HIV-1 reverse transcription. Several mechanisms may account for its inhibitory effect. The binding of A3G to RT could interfere with the access of RT to the substrate or physically inhibit the enzymatic activities of RT. Interestingly, an earlier study reported an interaction between A3G and RT in hepatitis B virus (HBV) (43). This study indicated that A3G binding to HBV RT was required for the incorporation of A3G into HBV nucleocapsids. In HIV-1 infection, several previous studies clearly showed that A3G virus incorporation is mediated by the NC domain of the Gag protein and/or viral RNA (1, 15, 63). We also examined the effect of RT on the incorporation of A3G into the virus, and the results suggest that the content of A3G in the HIV-1 virions was not significantly reduced when RT was deleted in our single-cycle replication virus (data not shown). This finding suggests that the interaction of A3G with HIV-1 RT may occur after A3G is already within the virus. It should be noted that both our study and the study by Nguyen and Hu (43) revealed that A3G binding to both HBV and HIV-1 RT did not require the presence of the viral genomic RNA (Fig. 2A).

Through the deletion analysis, we mapped the RT-binding region in the A3G encompassing residues 65 to 132, which partially overlaps with the region essential for A3G packaging (13, 25, 35). This finding suggests that the RT-binding region may partially overlap with the NC-binding region in A3G. These results led to the speculation that at the initial stage of HIV-1 assembly, A3G may be first associated with the HIV-1 NC and/or RNA. At this stage, the incorporation of A3G does not necessarily require the presence of RT. Consistently, our data also revealed that the presence of the RT-binding polypeptide A3G_65-132 does not significantly impair the incorporation of A3G_wt into the virus (Fig. 5B). During viral maturation and Gag/Gag-Pol proteolytic processing, A3G may be released from Gag and/or the Gag-Pol precursor and targets HIV-1 RT in the virus. This highly regulated order of binding may be necessary for an efficient packaging of A3G and, consequently, its action on viral reverse transcription. This speculation is also supported by the fact that the interaction between A3G and NC does not affect reverse transcription (26). Moreover, our results showed that the RT processing in the virus was not affected by A3G (Fig. 2C and D). These data suggest that A3G may bind to RT after RT is cleaved from the precursor and forms the p66/p51 heterodimer. However, we still do not have direct evidence to support this hypothesis, and further investigation will be required to elucidate these mechanisms.

A number of peptides have demonstrated therapeutic potential against HIV-1 replication, and some of them target HIV-1 RT (3). A peptide derived from HIV-1 NCp7 reduced the viral DNA level in a dose-dependent manner through the inhibition of the NCp7/RT interaction (16). In this study, we identified a polypeptide, A3G_65-132, which could compete with A3Gwt for the interaction with RT. Intriguingly, the infection experiments showed that when CD4⁺ T cells were infected by viruses produced in the presence of both the A3Gwt and A3G_65-132 peptide, the reverse transcription level was significantly higher than that in cells infected with viruses produced in the presence of only A3Gwt (Fig. 5C). In contrast, the viral reverse transcription was not affected when the infecting viruses were produced in the presence of the A3G_65-132 peptide alone. These results clearly indicate that even though the A3G_65-132 peptide was able to bind to RT efficiently, it did not inhibit its function; however, its presence was able to compete with A3Gwt and dominantly interfere with its anti-reverse transcription activity. However, we also could not eliminate the possibility that A3G_65-132 may bind to A3Gwt and disrupt its ability to interact with HIV RT, which will require further investigation. Overall, these observations provide evidence for the requirement of the A3G-RT interaction for the inhibition of A3G on HIV-1 reverse transcription. Also, our study suggests that RT binding through the binding motif itself is not sufficient to disrupt the activity of RT, and the other domains of A3G are also required. It is worth noting that the previously described virion incorporation domain (aa 124 to 127) and a Vif-binding motif (aa 128 to 130) are present in both the RT-binding peptide A3G_65-132 and RT-unbound A3G_102-257 (Fig. 4D), suggesting that these two functional domains may not be sufficient for A3G binding to HIV-1 RT. Additionally, we also observed that even though A3G_65-132 significantly inhibited A3G's effect on HIV reverse transcription (Fig. 5C), the presence of A3G_65-132 could only modestly increase viral production (Fig. 5D). At this point, we still do not know whether the presence of A3G_65-132 had less effect on A3G-induced viral hypermutation. Hence, more studies are required to address these questions. Also, more detailed analysis will be required to define the critical amino acid(s) and/or motifs within A3G_65-132 that are required for its interaction with RT.

Until now, A3G has been reported to bind to the following four HIV-1 proteins: NC, Vif, IN, and RT (29, 35, 36). Among those, the A3G-binding viral proteins Vif, NC, and IN have all been shown to interact with RT and promote the efficiency of reverse transcription to various levels (11, 22, 23, 29, 33). Thus, an unanswered question is whether A3G-mediated inhibition of reverse transcription could be due to the disrupting effect of their associations with RT. In our study, we have ruled out the involvement of IN and Vif, because in the absence of Vif and IN, A3G still significantly inhibited reverse transcription (Fig. 1D). However, we still do not know if A3G binding to RT could interfere with the NC/RT interaction, and this question is yet to be answered. Additionally, it would also be interesting to investigate whether the A3G-RT interaction could contribute to A3G-induced hypermutation during HIV replication, and if this is the case, whether A3G_65-132 could affect the rate of A3G-induced hypermutation. Overall, our results have demonstrated the A3G-RT interaction and its impact on HIV-1 reverse transcription. Further investigation of this interaction will elucidate our understanding of the mechanism underlying the antiviral action of A3G, which will contribute to the design of new anti-HIV strategies.