Since the first cases of AIDS were identified in 1981, this disease has led to the death of >20 million people. Despite remarkable medical advances, the incidence of HIV-1 infections continues to rise throughout the world. Current anti-HIV-1 agents target mainly the HIV-1 reverse transcriptase or protease. However, viral resistance and the toxicity associated with inhibitors of these enzymes have created a need for more potent and safer therapies against other viral targets.
Vif is one of only six HIV-1 regulatory proteins1
. To identify small molecules that antagonize HIV-1 Vif function in host cells, we developed a high-throughput screen involving fluorescence-labeled A3G (Supplementary Fig. 1
online). As Vif downregulates A3G levels in human cells9
, we reasoned that Vif-A3G interactions in host cells could be quantitatively monitored by expressing A3G labeled with a fluorescent tag. Our rationale was that when 293T cells are co-transfected with yellow fluorescent protein (YFP)-tagged A3G and HIV-1 vectors with and without Vif (the subgenomic proviral vector pNL-A1 harboring HXB2 strain Vif and the corresponding Vif-deleted vector pNL-A1Δvif
, respectively), YFP fluorescence should be less in cells co-expressing A3G-YFP and wild-type HIV than in cells co-expressing A3G and Vif-deleted HIV. Moreover, a potential Vif inhibitor should reduce Vif downregulation of A3G-YFP, resulting in increased YFP signal intensity.
Using this assay, we screened a diverse library of 30,000 small molecules and identified 537 compounds as primary hits. To exclude false-positive hits, these 537 compounds were retested in duplicate in a secondary screen. This second screen, which evaluated small molecules not only for reproduction of Vif inhibition, but also for inherent fluorescence, ability to increase general transfection and ability to increase nonspecific protein expression (in an red fluorescent protein (RFP)-based assay), yielded 66 compounds. Based on the diversity of their chemical scaffolds, these small molecules were then divided into ten classes. From each of these ten classes, we selected the two or three small molecules exhibiting the best activity in the secondary screen (% Vif inhibition). Structures of these 25 compounds, labeled RN-1 through RN-25, are shown in , and their median inhibitory concentration (IC50) values for inhibition of A3G degradation in 293T cells are shown in .
Figure 1 Small molecules that inhibit HIV-1 Vif. (a) Small-molecule structures. Structures of small molecules isolated from a 30,000-compound library and exhibiting dose-dependent anti-Vif activities. (b) IC50 values. ++++, 0–10 μM; +++, 10 μM–30 (more ...)
HIV-1 replication depends on Vif activity only in host cells that express A3G5
. In contrast to these ‘nonpermissive’ cells, those not expressing A3G do not require Vif for HIV-1 replication and are known as ‘permissive’. To determine whether the small compounds RN-1 to RN-25 inhibit Vif function in the context of viral replication, we examined the antiviral activities of these compounds against wild-type HIV-1 in nonpermissive cells (H9, CEM) expressing A3G and in permissive cells (MT4, CEM-SS), which do not express A3G. We reasoned that specific Vif antagonists with the desired inhibitory profiles would inhibit viral replication in a dose-dependent manner only in nonpermissive cells but not affect replication in permissive cells. Cells were infected with an X4-tropic HIV-1LAI
variant in the presence of varying concentrations of Vif inhibitors, and viral replication was assessed at 2-d intervals by reverse transcriptase activity in culture supernatants (). The viability of cells was also checked every other day, using MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide) assays to evaluate the toxicity of Vif inhibitors. RN-18 neither inhibits cell growth nor exhibits toxicity at 50 or 100 μM (Supplementary Fig. 2
online). The majority of the compounds (RN-5,6,9,10,12,13,15,17,20,21,22,23 and 25) did not exhibit antiviral activity in either MT4 or H9 cells (IC50
values >50 μM). Some compounds (RN-2,3,4,8,24) exhibited antiviral activity in both permissive and nonpermissive cells, indicating that their effects were not related to Vif. Two compounds with closely related structures (RN-18,19) exhibited an antiviral profile consistent with a Vif-specific inhibition in that they affected viral replication only in nonpermissive cells. Because of its greater potency (IC50
values of 4.5 and 10 μM in nonpermissive CEM and H9 cells, respectively, and IC50
> 100 μM in permissive cells), RN-18 was selected for more detailed analysis. Representative inhibitory profiles of RN-18 in permissive and nonpermissive cells are shown in . In the presence of the inhibitor, RN-18, reverse transcriptase activity in the nonpermissive H9 and CEM cells decreased substantially and in a dose-dependent manner ( and Supplementary Fig. 3
online). However, RN-18 did not significantly affect reverse transcriptase activity in the permissive cell lines MT4 or CEM-SS (). RN-18 also exhibited antiviral activity in CEM-SS modified to stably express A3G but did not exhibit antiviral activity in the parental CEM-SS cell line (Supplementary Fig. 4
online). Thus, the ability of RN-18 to inhibit HIV-1 replication in nonpermissive cells, but not in permissive cells, suggests that it is a Vif antagonist.
Figure 2 The small-molecule Vif antagonist, RN-18, inhibits HIV-1 replication in nonpermissive cells but not in permissive cells. (a–d) CD+ T-cell lines H9 (a), CEM (b), MT4 (c) or CEM-SS (d) were treated overnight with varying concentrations of RN-18 (more ...)
To investigate the mechanism and specificity of RN-18 function, we determined the effect of RN-18 on A3G levels in virus-producing cells and in virus particles. To that end, 293T cells were co-transfected with a luciferase reporter HIV-1 (pNL4-3LucR−E−) and hemagglutinin (HA)-tagged A3G expression vectors in the presence or absence of 50 μM RN-18. A3G, Vif and HIV-1 expression levels were analyzed by immunoblotting with antibodies against the HA tag, HIV-1 Vif and the HIV-1 antigen p24, respectively. In cells treated with RN-18, Vif protein levels were downregulated, and A3G expression increased (). Consistent with these findings, A3G incorporation into virions () was enhanced in the presence of RN-18, and Vif levels in virus were decreased, indicating that RN-18 inhibited Vif function and upregulated A3G levels.
Figure 3 The Vif antagonist, RN-18, increases APOBEC3G abundance in HIV-1 producer cells and virions, but does not affect APOBEC3B levels. 293T cells co-expressing HIV-1 (pNL4-3LucR−E−) and various HA-tagged APOBEC3 expression vectors were cultured (more ...)
To determine the effect of RN-18 on A3G abundance and virion incorporation in simian immunodeficiency virus (SIV)-1, we cultured 293T cells co-expressing SIVmac
239 green fluorescent protein Δnef and HA-tagged A3G for 24 h in the presence (+) or absence (−) of 50 μM RN-18. Analysis of total producer cell lysates by immunoblotting with antibodies against the HA tag and SIV-1 antigen p27 showed increased A3G levels in the presence of RN-18 (Supplementary Fig. 5
online). SIV-1 virions from producer cell supernatants were concentrated, lysed and analyzed by immunoblotting. This showed that A3G incorporation in virions was also increased by RN-18 treatment (Supplementary Fig. 5
). Together, these results indicate that the Vif antagonist, RN-18, increases A3G abundance in both SIV-1 producer cells and virions.
The human genome encodes seven APOBEC3 proteins from a single gene cluster located at chromosome 22 (ref. 10
). Similar to A3G, APOBEC3F is a potent inhibitor of HIV-1 replication and is targeted by Vif. Although not as potent as A3G or APOBEC3F, APOBEC3C has been shown to inhibit HIV-1 replication. In contrast, APOBEC3B exhibits modest anti-HIV-1 activity relative to A3G and resists Vif because it does not interact with Vif in cells6,11,12
. To determine the specificity of RN-18 function, we transfected 293T cells with a reporter HIV-1 (pNL4-3LucR−
) vector and various HA-tagged APOBEC3 expression vectors in the presence or absence of 50 μM RN-18 and analyzed protein levels by immunoblotting. RN-18 significantly and quantitatively enhanced levels of A3G, APOBEC3F and APOBEC3C but had no significant effect on APOBEC3B abundance ( and Supplementary Fig. 6
online). Interestingly, Vif protein levels were downregulated in cells expressing A3G, APOBEC3F and APOBEC3C and did not change substantially when APOBEC3B was expressed. These results strongly suggest that RN-18 specifically inhibits Vif-mediated downregulation of APOBEC3 family members by decreasing Vif protein levels when Vif interacts with APOBEC3 proteins.
Our finding that RN-18 inhibited viral replication in nonpermissive cells, but not in permissive cells, suggests that RN-18 is a Vif-specific inhibitor of HIV-1 replication (). To assess whether RN-18 can restore A3G levels in nonpermissive cells in a Vif-dependent manner, we infected H9 and MT4 cells with wild-type HIV-1 () and Vif-deleted (Δvif) HIV-1 () in the presence and absence of RN-18, and analyzed the expression of A3G, Vif, p24 and cyclin T1 (internal control). The immunoblot results show that RN-18 increased A3G abundance by decreasing Vif levels only in nonpermissive H9 cells () and that this increase in A3G levels required Vif (compare ). Furthermore, RN-18 did not increase A3G abundance in uninfected cells (). Together with the results shown in , these results strongly suggest that RN-18 is a specific Vif antagonist that inhibits HIV-1 replication.
Figure 4 The Vif antagonist, RN-18, enhances APOBEC3G expression in nonpermissive cells and in a Vif-dependent manner. (a,b) H9 (nonpermissive) and MT4 (permissive) CD+ T-cells were infected with pNL4-3 LucR−E− (a) or ΔVifpNL4-3 LucR-E (more ...)
APOBEC3G is neutralized by HIV-1 through Vif13–18
, which functions in concert with an E3 ubiquitin ligase complex to mediate the polyubiquitination and rapid degradation of A3G through the proteasome9,15,19–22
. The possibility that RN-18 modulates HIV-1 replication by inhibiting the general proteasome pathway in cells is not supported by our results in and , which indicate that RN-18 specifically antagonizes Vif. Further support against this possibility is the exclusion of compounds that nonspecifically enhanced RFP expression during our secondary screening. To directly address whether RN-18 inhibits a general proteasome pathway, we treated cells with varying concentrations of RN-18 and analyzed levels of the cell cycle inhibitor, p21, expression of which is modulated by the proteasome23
. RN-18 did not affect p21 expression, whereas the proteasome inhibitors MG132, ALLN and lactacystin increased p21 levels (). In addition, pulse-chase kinetic experiments of Vif degradation revealed that RN-18 reduced Vif half-life from ~46 min to ~33 min only when A3G was present (data not shown). To further define the specificity of RN-18 in modulating A3G levels through a mechanism that depends on A3G-Vif interaction, we quantitatively analyzed the effect of RN-18 on the D128K mutant of A3G, which is not targeted by Vif. RN-18 increased abundance of A3G-YFP, but not A3G-YFP (D128K), in HIV-1 producer cells (Supplementary Fig. 7
online). These results indicate that RN-18 is not a general proteasome inhibitor, but enhances A3G levels by specifically downregulating Vif when Vif interacts with A3G. RN-18 could directly disrupt Vif-A3G interactions leading to Vif degradation. Another plausible mechanism is that RN-18 interferes with Vif interaction with an adapter protein in the Vif-A3G–host protein complex making Vif less stable and enhancing its turnover. Further studies are required to understand the details of the molecular mechanism of RN-18 action.
To further understand the mechanism of antiviral function of RN-18, we determined the effect of RN-18 on both the packaging of A3G into virions and the editing function of A3G. First, 293 T cells co-expressing HIV-1 (pNL4-3LucR−
) and various concentrations of A3G-HA (3.75, 7.5, 15 fmoles) were cultured for 24 h in the presence (+) or absence (−) of 50 μM RN-18. Protein contents and viral infections were analyzed as described above. In the presence of RN-18, increasing the amount of A3G vector (7.5–30 fmole) enhanced A3G expression in cell extracts, leading to less infectious virus (Supplementary Fig. 8a,c
online). Control experiments showed that only 3.75 fmole of the A3G vector was sufficient for efficient A3G expression and its antiviral function (Supplementary Fig. 8b,d
). These experiments indicate that the Vif antagonist, RN-18, enhanced A3G abundance in producer cells, resulting in less infectious HIV-1 virions. Similarly, HIV-1 virions obtained from H9 cells treated with RN-18 were less infectious (Supplementary Fig. 9
online). To analyze the editing function of A3G, we used a stable A3G-YFP–expressing cell line to produce wild-type or delta-Vif virions in the presence and absence of RN-18 and sequenced individual clones. This demonstrated that RN-18 enhances cytidine deamination of the viral genome (Supplementary Fig. 10
online). Taken together, these results demonstrate that RN-18 enhances A3G abundance and virion incorporation, which lead to the production of less infectious viruses mediated by the DNA editing function of A3G.
In summary, we have identified a lead compound that specifically antagonizes HIV-1 Vif function and inhibits viral replication by targeting the Vif-APOBEC3 axis. The pharmacological properties of this lead compound can be optimized and improved by analyzing the structure-function relationships governing inhibition of Vif function and target selectivity. Other than their medical applications, these pharmacological probes of Vif function can provide important new insights into the biology of HIV-1.