|Home | About | Journals | Submit | Contact Us | Français|
Previously we described the identification of two compounds (3-amino-5-ethyl-4,6-dimethylthieno[2,3-b]pyridine-2-carboxamide  and 4-amino-6-methoxy-2-(trifluoromethyl)-3-quinolinecarbonitrile ) that interfered with HIV replication through the inhibition of Rev function. We now describe resistant viral variants that arose after drug selection, using virus derived from two different HIV proviral clones, NL4-3 and R7/3. With HIVNL4-3, each compound selected a different single point mutation in the Rev response element (RRE) at the bottom of stem-loop IIC. Either mutation led to the lengthening of the stem-loop IIC stem by an additional base pair, creating an RRE that was more responsive to lower concentrations of Rev than the wild type. Surprisingly, wild-type HIVR7/3 was also found to be inhibited when tested with these compounds, in spite of the fact this virus already has an RNA stem-loop IIC similar to the one in the resistant NL4-3 variant. When drug resistance was selected in HIVR7/3, a virus arose with two nucleotide changes that mapped to the envelope region outside the RRE. One of these nucleotide changes was synonymous with respect to env, and one was not. The combination of both nucleotide changes appeared to be necessary for the resistance phenotype as the individual point mutations by themselves did not convey resistance. Thus, although drug-resistant variants can be generated with both viral strains, the underlying mechanism is clearly different. These results highlight that minor nucleotide changes in HIV RNA, outside the primary Rev binding site, can significantly alter the efficiency of the Rev/RRE pathway.
Cells infected with human immunodeficiency virus (HIV) produce stable species of unspliced and incompletely spliced mRNAs that require the Rev protein for their nuclear-cytoplasmic export and expression (for a review, see reference 29). Rev is a small basic protein of about 116 amino acids that is encoded in the HIV genome (for a review, see reference 44). One well-defined function of Rev in HIV infection is to bind (7, 10, 11, 30, 31, 41, 53) to the Rev response element (RRE) (15, 26, 45) that is found in these unspliced and incompletely spliced mRNAs and promote their nucleo-cytoplasmic export (21, 28, 40) by interacting with the Crm1 cellular export receptor and Ran-GTP (1).
Many previous studies identified stem-loop IIB of the RRE as the primary binding site for a Rev protein monomer and suggested that 6 to 10 subunits of Rev can subsequently bind to the RRE (2, 9, 32, 33, 36, 50). Oligomerization of Rev requires the Rev multimerization domain that has also been shown to be necessary for Rev function (12, 34, 39, 42, 54). Additionally, the molecular details of the interaction of the Rev monomer and its primary binding site have been well characterized through the use of an isolated Rev peptide containing only the arginine-rich binding motif (ARM) and the stem IIB RNA hairpin (3, 4, 48, 49).
Recently, Rev has been shown to form a cooperative high-affinity oligomeric complex with the RRE (13). This complex has an affinity nearly 3 orders of magnitude higher than has been shown before for the monomer interaction with the stem-loop IIB binding site, and its formation appears to be dependent on multimerization sequences in the Rev protein. This study also demonstrated the presence of a second specific binding site for Rev in stem-loop IA of the RRE. Interestingly, in vitro binding studies with the ARM peptide and functional studies in cells with mutants of Rev show that different interactions between the ARM and RNA occur with each of the two binding sites.
Several lines of evidence also suggest that Rev may be required for efficient translation of viral RNAs that contain the RRE (24). Whether this is a consequence of differences in the RNP complex that forms on the mRNA when the Rev RNA export pathway is utilized, compared to other pathways, or whether it is a direct result of the interaction of Rev itself with the translation machinery remains uncertain (6, 8). However, one group reported that HIV mRNA contains a second binding site for Rev near its 5′ end and that Rev binding to this site regulates translation of the mRNA in a manner that is not dependent on the RRE (23).
From the viewpoint of therapeutic development, it is significant that there is no evidence of a cellular Rev orthologue and that most cellular mRNAs appear to be exported by a pathway that is different from that used by Rev. Additionally, several of the interactions that mediate Rev function are completely viral in nature. Targeting these interactions by a therapeutic agent might lead to specific inhibition of viral replication, with only minimal side effects on the cell.
In a previous study, we described two compounds, 3-amino-5-ethyl-4,6-dimethylthieno[2,3-b]pyridine-2-carboxamide (compound 103833) and 4-amino-6-methoxy-2-(trifluoromethyl)-3-quinolinecarbonitrile (compound 104366), which were identified from a screen of 40,000, as inhibitors of Rev-RRE function and viral replication (46). In that study, using a series of reporter assays, we showed that Rev-dependent gene expression was preferentially inhibited by the compounds in cell culture, relative to control genes that did not require Rev. Furthermore, we showed that the compounds did not appear to directly affect Rev-RRE binding. This led us to speculate that some unidentified downstream component of the Rev-RRE pathway might be targeted by these compounds.
The present study describes experiments to select resistance mutations in HIV pNL4-3 that allow growth in the presence of the anti-Rev compounds. We reasoned that analysis of these mutations might shed light on the mechanisms involved in the inhibition of replication by the compounds and also give further insights into the Rev-RRE pathway.
Viral stocks were prepared by transfection of 293T cells using proviral clones, as previously described (27). pCMV Gag-Pol-RRE (where CMV is the simian promoter cytomegalovirus; the construct expresses HIV Gag-Pol proteins in a Rev-dependent manner) was used in transient transfections of 293T cells as previously described (25, 35, 46). SupT1 or Rev-CEM cells were infected as previously described (27), with the specific modifications given in each figure legend. Rev-CEM cells contain a Rev-dependent green fluorescent protein (GFP) reporter (52). They are derived from CEM-SS cells. p24 levels in either the transfected or infected cell supernatants were measured using a p24 enzyme-linked immunosorbent assay (ELISA) that followed a previously published procedure (51).
To isolate viral RNA, 1 ml of viral culture supernatant was centrifuged for 1 h at 30,000 rpm at 4°C in a microcentrifuge. The supernatant was carefully decanted by pipette, and the pellet was suspended in 20 μl of RQ1 DNase 1× buffer (Promega). Five microliters of RQ1 DNase buffer (1 unit/μl) was added, followed by incubation at 37°C for 30 min. Following this incubation, 75 μl of 0.01 M Tris (pH 7.4)–0.001 M EDTA, 10 μl of yeast (Saccharomyces cerevisiae) tRNA (1 μg/μl; Invitrogen), 6 μl of 10% SDS, and 3 μl of proteinase K (10 μg/μl; Sigma) were added. Incubation was then continued at 42°C for 30 min. One hundred microliters of phenol (pH 6.6) and 100 μl of chloroform were then added; the tube was vortexed and then centrifuged at 13,000 rpm for 5 min. The top layer was removed and added to a clean tube. This phenol-chloroform extraction was repeated. The top layer was again removed to a clean tube, and 13 μl of 3 M sodium acetate and 325 μl of 100% ethanol were added. The tube was placed at −80°C for 1 h to precipitate the RNA. The tube then was centrifuged at 13,000 rpm for 15 min at 4°C. The supernatant was decanted, and the pellet was washed twice with 75% ethanol. The RNA pellet was allowed to air dry prior to suspension in 50 μl of 10 mM Tris, pH 7.4. It was stored at −80°C until use.
Viral RNA was amplified into cDNA using a two-step reverse transcriptase PCR (RT-PCR) procedure with nested primers. The first 50-μl reaction mixture contained the following: 5 μl of 10× PCR buffer–MgCl2 solution (Invitrogen); 35.5 μl of diethyl-pyrocarbonate-treated water; 0.4 μl of a mixture of dATP, dCTP, dTTP, and dGTP (2.5 mM each; Promega); 0.125 μl of Taq polymerase (5 units/μl; Invitrogen); 0.2 μl of avian myeloblastosis virus (AMV) reverse transcriptase (22 units/μl; Roche); 0.25 μl of RNasin (40 units/μl; Promega); 1.5 μl of 50 mM MgCl2; 1 μl each of the forward and reverse primers (100 ng/μl); and 5 μl of the 50 μl solution of viral RNA described above. Tubes lacking RNA template or containing RNA template but lacking AMV RT were also prepared as controls. The reaction was run under the following cycling conditions: 42°C for 1 h to allow AMV RT elongation and 95°C for 3 min to inactivate the AMV RT, followed by 30 cycles of 95°C for 1 min, 58°C for 1 min, and 72°C for 2 min for strand separation, annealing, and extension, with a final extension at 72°C for 10 min. The samples were then put at 4°C for storage.
The second 100-μl reaction mixture contained the following: 5 μl of cDNA from the first reaction tube; 10 μl of 10× PCR buffer–MgCl2 solution (Invitrogen); 77 μl of diethyl-pyrocarbonate-treated water; 0.8 μl of a mixture of dATP, dCTP, dTTP, and dGTP (2.5 mM each; Promega); 0.25 μl of Taq Polymerase (5 units/ml; Invitrogen); 3 μl of 50 mM MgCl2; and 2 μl each of the forward and reverse primers (100 ng/μl). Reactions were performed using the following cycling conditions: 95°C for 2 min followed by 30 cycles of 95°C for 1 min, 45°C for 1 min, and 72°C for 2 min for strand separation, annealing, and extension, with a final extension at 72°C for 10 min. The samples were then put at 4°C for storage and further processing.
Primer numbers refer to the index number in our laboratory collection. For amplification of the RRE, the first-round primers used were 396 (nucleotides [nt] 7485 to 7506; 5′-TAAACATGTGGCAGGAAGTAGG-3′) for the forward primer and either 23 (nt 8464 to 8443; 5′-GTTCACTAATCGAATGGATCTG-3′) or 245 (nt 8403 to 8383; (5′-GGCCTGTCGGGTCCCCTCGGG-3′) for the reverse primer. The second-round nested primers used were 1438 (nt 7712 to 7733; 5′-AAGGCAAAGAGAAGAGTGGTGC-3′) for the forward primer and 1439 (nt 8039 to 8018; 5′-GCACAGCAGTGGTGCAAATGAG-3′) for the reverse primer.
For amplification of Rev exon 1, the first-round primers were 299 (nt 5562 to 5579, with 5′ extension; 5′-GCGGGATCCGAACAAGCCCAAGAAGAC-3′) for the forward primer and either 234 (nt 6353 to 6335, with 5′ extension; 5′-GCGAATTCACACAGGTACCCCATAATA-3′) or 128 (nt 7371 to 7352; 5′-TTACAGTAGAAAAATTCCCC-3′) for the reverse primer. The second-round nested primers used were 379 (nt 5603 to 5622; 5′-CAATGAATGGACACTAGAGC-3′) for the forward primer and 479 (nt 7317 to 7300; 5′-TCCCCTCCTGAGGATTGC-3′) for the reverse primer.
For amplification of Rev exon 2, the first-round primers were 366 (nt 8091 to 8115; 5′-TGACCTGGATGGAGTGGGACAGAGA-3′) for the forward primer and 807 (nt 8814 to 8838; 5′-GCTGCTGTGTTGCTACTTGTGATTG-3′) for the reverse primer. The second-round nested primers used were 1557 (nt 7199 to 7221; 5′-GCACATTGTAACATTAGTAGAGC-3′) for the forward primer and 1559 (nt 8911 to 8932; 5′-GCTGTATTGCTACTTGTGATTG-3′) for the reverse primer.
In order to select resistance mutations, we carried out long-term infections in the presence of 0.8 μM compound 104366 by infecting duplicate cultures of Rev-CEM cells with an HIV-1NL4-3 derivative lacking Nef (NL4-3 Nef−). Rev-CEM reporter cells were used in order to visually monitor the infection (52). As the concentration of compound used was near its 50% inhibitory concentration (IC50), we reasoned that some viral replication would occur but that there would remain enough selective pressure on the virus to select resistance mutations. Figure 1 shows that continued passage in the presence of 0.8 μM 104366 largely suppressed viral replication until 47 days, when one of the cultures showed growth (Fig. 1A). To determine if a resistant variant had been selected, virus from day 53 of this infection was then used to infect duplicate cultures of SupT1 cells. The infections were carried out in the presence of the same concentration of compound 104366 and also at a 10-fold higher dose (Fig. 1B). Growth was also examined in the absence of any compound. As a control, the parental virus was also retested (Fig. 1C). As expected, cultures infected with the parental virus in the presence of compound again showed no viral growth. Cultures infected with the day 53 resistant variant showed equally robust replication at both compound concentrations compared to cultures without any compound present.
The resistant viral stock was then grown for 10 passages in the absence of compounds to test for reversion back to the wild-type (wt) virus. To do this, supernatant from the peak growth of each passage was used to infect the next passage (data not shown). Each passage peak was also tested in separate cultures in the presence of the compounds to determine if the viruses remained resistant. For each infection, 125 ng of p24 was used to infect 5 × 105 cells in 5 ml. At each passage the resistant stock failed to show any sign of inhibition, demonstrating that there was no reversion of phenotype. The data for the virus from the last passage are shown in Fig. 1D.
To isolate a viral stock that was resistant to compound 103833, we carried out serial passages in the presence of suboptimal concentrations of drug. To start, SupT1 cells were infected with clone NL4-3 in the presence of either 1 μM (the approximate IC50) or 10 μM compound. As a control, virus was also passaged without any compound in the culture (Fig. 2A). The initial infection produced minimal viral growth in the 1 μM culture and virtually no growth at 10 μM concentrations of compound. In contrast, robust growth was observed in the absence of compound. Supernatant from the peak of the 1 μM culture (day 13) was collected and used to again infect SupT1 cells in the presence of both drug concentrations and with no added drug (Fig. 2B). As before, some growth was seen in the culture containing the 1 μM compound. The peak of viral growth occurred slightly earlier and was at a slightly higher level than in the previous passage. The supernatant from the peak of the second passage (day 10) was again used to infect SupT1 cells for a third passage (Fig. 2C). This virus was now able to replicate to similar levels at both concentrations of compound tested, and replication was virtually identical to the replication of the virus in the absence of compound. (The data from duplicate third-round infections are shown). Supernatant from the peak of infection of one of the cultures with 10 μM compound was then tested for reversion over 10 passages as described above for the 104366 compound. The stock produced from the last passage was tested for resistance as shown in Fig. 2D. Again, the virus remained resistant to inhibition even after passage in the absence of compound.
In order to identify mutations in the resistant variants, viral RNA was isolated from supernatant virus at the peaks of the infection with the higher levels of compound shown in Fig. 1B and and2C.2C. The region covering rev, tat, vpu, env, and the RRE was amplified into cDNA by RT-PCR. For the 104366-resistant variant, the PCR fragments were cloned into Invitrogen TOPO clone pCR4 vectors, and four individual clones were isolated and sequenced. Table 1 illustrates the mutations observed in these analyzed clones. There were no changes within the Rev coding region in any of the clones. Three of the four clones had more than one mutation in either env or the RRE. The mutation A7854G seemed of particular interest because it is located in the RRE near the primary Rev binding site and was present in three of the four clones. It was also the only mutation present in one of the clones. Three out of four of the clones also contained the mutation A7936G, which was located in the loop of stem-loop V. A7854G caused an aspartic acid-to-glycine change in gp41 while A7938G did not result in an amino acid change.
The amplified DNA from the 103833-resistant variant was analyzed by sequencing the bulk PCR product, which identified a single mutation, C7836U, in the RRE. Interestingly, this mutation corresponded to a potential binding partner of A7854 in the RRE structure as shown in Fig. 3A. Both the A7854G mutation identified in the 104366-resistant variant and the C7836U mutation identified in the 103833-resistant variant allowed for the formation of at least one additional nucleotide binding pair at the base of stem-loop IIC in the RRE (Fig. 3B) and possibly two as it seems likely that the G-U right below the altered base pair would also form. Each of these mutations also caused amino acid changes in gp41, i.e., D36G for A7854G and A30V for C7836U.
Fragments containing the mutations listed in Table 1 were built back into a wild-type NL4-3 Nef− proviral background as described in Materials and Methods. To determine if a particular mutation conferred resistance, the resultant proviruses were transfected into 293T cells to produce viruses that were used to infect SupT1 cells in the presence or absence of the compounds. Some of the data from these experiments are shown in Fig. 4. Both the A7854G virus and the C7836U virus grew well in the presence of the compound that was used in their original selection (104366 for A7854G and 103833 for C7836U) (Fig. 4A and C). Each of these viruses was also cross resistant to the other compound (Fig. 4B and D). On the other hand, virus containing the A7936G A8086C U8091C triple mutation appeared to grow less well and remained sensitive (Fig. 4E). Virus containing either the A7854G A7936G double mutation or the A7854G A7936G G8179A C8764U quadruple mutation grew poorly even in the absence of the compounds, making their phenotypes difficult to determine (data not shown). Figure 4F shows data from a control experiment performed at the same time as the experiments shown in panels A to E, which demonstrate that the parental virus was still sensitive to inhibition by either compound.
In addition to allowing extra bases to form at the bottom of stem-loop IIC, the A7854G and C7836U mutations also change an amino in gp41. To distinguish whether the amino acid change or the RRE structural change allowed the virus to grow in the presence of compound, a second site mutation was introduced into each of the resistant mutants. This mutation, C7837A, disrupts the base pairing directly above the base pair formed by the resistance mutation. This is expected to disrupt the stable structure introduced at the base of the stem-loop by the resistance mutation (Fig. 3B). The mutation is synonymous with respect to the amino acid sequence of gp41. Thus, if RNA structure, rather than gp41 amino acid sequence, is a determinant of the resistance phenotype, this second site mutation would be expected to convert the resistant virus containing A7854G or C7836U to a sensitive one.
The data in Fig. 4G and H show that the C7837A mutation does, in fact, restore sensitivity of each resistant virus to each of the compounds (compare Fig. 4A to G and D to H). Furthermore, a comparison of the growth of this these viruses to wild-type NL4-3 (Fig. 4F) shows that these viruses grew equally well in the absence of added compound.
We next carried out transient transfection experiments using NL4-3 proviruses that contained the mutant RREs to determine whether the mutant RREs could convey resistance to the compounds in this context. This transient transfection assay measures viral particle production directly from the transfected proviral DNA and does not require the complete viral replication cycle. Thus, this is a more direct assay of Rev/RRE function, compared to the replication assays described above. Additionally, any functional changes in gp41 which were the result of the amino acid changes caused by the RRE nucleotide changes would not be expected to be a factor in viral particle production as particles efficiently form in this system independently of envelope protein function.
To perform these experiments, 293T cells were transfected with proviral clones carrying either the wild-type RRE, the RRE with the A7854G mutation, or the RRE with the C7836T mutation. Supernatant samples were analyzed after 72 h for p24 production. Figure 5 shows that both compounds efficiently inhibited particle production from the provirus containing the wt NL4-3 RRE and that they failed to inhibit p24 production from the proviruses containing either of the mutant RREs. These data, together with the second-site-mutation data from the RRE variants shown in Fig. 4G and H, clearly demonstrate that the resistance phenotype was caused by structural changes in the RRE and not by changes in envelope protein sequence.
Since RRE structure appeared to be the major contributor to the resistance phenotype, we next examined how the wild-type and mutant RREs differed in their responses to the Rev protein in the absence of compound. To do this we performed dose-response experiments with increasing amounts of a plasmid that expressed NL4-3 Rev and either a NL4-3 proviral clone lacking a functional Rev gene (Fig. 6A) or a subgenomic Gag-Pol-RRE construct that expressed HIV Gag-Pol proteins in a Rev-dependent manner (Fig. 6B). Rev activity was measured as p24 release into the supernatant of transfected 293T cells.
The data in Fig. 6A and B show that, for each of the three RREs tested, there was a differential response to cotransfection with a given amount of Rev-expressing plasmid. Plasmids containing RRE A7854G gave rise to more p24 production than those with the C7836U mutation, and plasmids with the RRE C7836U mutation responded better to smaller amounts of Rev than plasmids with the wild-type RRE. It is striking that the results were similar whether the RREs were tested in the context of the proviral clone (Fig. 6A) or the subgenomic Gag-Pol reporter (Fig. 6B). Thus, we conclude that the mutated RREs are able to function more efficiently with lower levels of Rev. This provides a simple explanation for the resistance phenotype they convey.
Our results show that the A7854G mutation in the context of the NL4-3 viral clone conveys resistance to compounds 103833 and 104366. However, analysis of RRE sequences from other HIV clade B viruses revealed that virtually all other isolates naturally have a G at position 7854, even without selection (37). This was surprising since our previously published data showed that these compounds inhibited Rev function in a variety of assays, all of which utilized RREs from isolates other than NL4-3. Thus, these findings suggested that the A7854G resistance mutation may convey resistance specifically in the NL4-3 context. We decided to examine this issue directly by testing the effects of compound 103833 on the replication of a different infectious clone under the same growth conditions employed in this study for NL4-3. For these studies, we utilized the R7/3 infectious clone (19, 20). R7/3 is a derivative of HXB2 and has an RRE with a G at 7854 (NL4-3 numbering), as well as four other changes in the RRE relative to NL4-3.
R7/3 virus was used to infect SupT1 cells in the presence of either 1 μM or 10 μM 103833. In this experiment, 50-ng equivalents of p24 were used to infect 5 × 106 SupT1 cells, which was 2.5 times less virus than the amount used in the NL4-3 infections. Figure 7 shows that even with this lower concentration of input virus, considerably more replication of R7/3 could be observed in the presence of the compound than in the experiments with NL4-3 (Fig. 2). However, the replication was delayed relative to the control cultures that did not receive compound, indicating that R7/3 was still partially sensitive to compound 103833.
To determine if the virus from the 10 μM culture that replicated to a peak at day 17 had changed from wild type, it was used to reinfect SupT1 cells in both the presence and absence of compound. The day 17 virus again showed delayed growth kinetics in the presence of compound (data not shown). Almost identical results were obtained with virus from a similar infection in the presence of compound 104366 (data not shown).
We repeated the R7/3 selection using a 5-fold lower viral input (10 ng), with the hope that this would allow the compounds to be more effective in controlling virus growth and that we might be able to select resistance mutations. With this lower viral input, the peak of replication in the absence of drug shifted from day 7 to day 17 or 19. Compound 103833 at a concentration of 10 μM virtually abolished viral replication until day 31 (Fig. 7B), and no replication at all was observed in the culture containing 8 μM 104366 (Fig. 7C).
Virus from the peak at day 31 in the culture that received 10 μM 103833 was next retested for the ability to grow in the presence or absence of 10 μM 103833 (Fig. 7D). For comparison the original wild-type virus was also retested (Fig. 7E). The virus that was recovered at day 31 grew similarly in the presence or absence of this compound.
In order to identify possible mutations that caused the resistance phenotype in the R7/3 day 31 viral stock, viral RNA was isolated from particles in the supernatant, and the regions containing rev, env, and the RRE were amplified into cDNA by RT-PCR as described in Materials and Methods. DNA sequencing of the PCR products revealed no mutations in rev or the RRE, but two nucleotide changes within env were found. The first change, A6364C (NL4-3 numbering), mapped to the C1 region of gp120 and was a silent mutation with respect to the envelope protein sequence, changing the alanine codon from GCA to GCC. The second change, A7251C (NL4-3 numbering), was nonsynonymous and changed an aspartic acid (GAT) to alanine (GCT) in the C3 region of gp120 at amino acid 346.
To test if the identified mutations conveyed resistance to compound 103833, we constructed proviral clones derived from R7/3 which contained both mutations or each of the mutations individually. Virus stocks were created by transfection of these clones into 293T cells, and 10 ng of p24 from each viral stock was used to infect 5 × 106 SupT1 cells in the presence or absence of 103833. As a control, 10 ng of p24 virus from a wild-type stock of R7/3 was also tested. The results of these infections are shown in Fig. 8. All four viruses grew well in the absence of compound. Only the virus derived from the proviral clone with the double mutations grew similarly in the absence or presence of compound 103833 (Fig. 8A). The wild-type virus, as well as each of the viruses carrying a single point mutation, was inhibited significantly by compound 103833 (Fig. 8B, C, and D) although the virus carrying the A6364C mutation did show some slight growth in its presence (Fig. 8C). Thus, we conclude that, in the context of the R7/3 virus, the combination of A6364C and A7251C can convey complete resistance to 103833. However, neither mutation by itself appears to be sufficient.
In this study we show that growth of NL4-3 virus in the presence of either 3-amino-5-ethyl-4,6-dimethylthieno[2,3-b]pyridine-2-carboxamide (103833) or 4-amino-6-methoxy-2-(trifluoromethyl)-3-quinolinecarbonitrile (104366) resulted in selection of resistant virus that had changes in the RRE. In each case, a single nucleotide mutation at the base of stem-loop IIC was sufficient to convey resistance, both to the compound used for selection and to the other compound.
Each of the resistance mutations changes an amino acid in the envelope protein and also allows the formation of an additional base pair at the bottom of RRE stem-loop IIC by substitution of a base either on the 5′ or 3′ side of the stem-loop. Once this base pair forms, it is likely that a G:U base pair directly below it would also form, leading to a lengthening of the stem by two additional base pairs. These structural changes in the RRE appear to be the cause of resistance since our experiments show that each mutated RRE functions better at lower levels of Rev than the wild-type RRE. The mutated RREs also mediate drug resistance in a reporter assay that is independent of the envelope protein. An additional experiment showed that mutations that destabilize the base of stem-loop IIC in the resistant mutants, without making additional changes in the envelope protein, restore sensitivity to the compounds in an infection.
It is interestingly that the RRE with the A7854G change appears to be more active than the one with the C7836U change although both add an additional base pair to the bottom of the stem in stem-loop IIC. Since A7854G forms a G-C base pair while C7836U forms an A-U base pair, it is possible that the increased activity results from the more stable stem structure that is formed.
Previous experiments suggested that the two compounds used in this study do not directly affect Rev-RRE binding since they do not inhibit binding in vitro (46). Nevertheless, cells treated with the compounds clearly constitute an environment that compromises Rev function. Thus, any mutation that enables Rev to function better under these circumstances would be expected to cause resistance. In the case of the selected mutations, it would appear that the virus overcomes the block to Rev function by creating an RRE that works more efficiently when Rev levels are low. However, since we do not yet fully understand how the various parts of the RRE interact with themselves and Rev, the basis for this increased efficiency remains unexplained.
The stem-loop IIC region of the RRE that contains the resistance mutations has no known function, but it is immediately adjacent to the stem-loop IIB region of the RRE that functions as the primary Rev binding site (2, 9, 32, 33, 36, 50). The structure of a modified stem-loop IIB has been determined both alone and in a complex with a peptide derived from the Rev RNA binding domain (3, 4, 48, 49). Based on these structures, the binding of the initial Rev monomer is thought to involve recognition of a distinct binding pocket in stem-loop IIB that is formed by noncanonical G-G and G-A base pairs. As the structure of these complexes have been determined with only a minimal modified stem-loop IIB sequence and a Rev peptide, they give little insight into why mutations at the base of stem-loop IIC might affect RRE function. It is possible that this region of stem-loop IIC somehow influences the structure of stem-loop IIB and its interaction with Rev, or alternatively stem-loop IIC may interact directly with Rev, contributing to overall Rev function. In fact, recent studies have shown that Rev oligomerizes on the RRE by using different surfaces of its alpha-helical RNA-binding domain to recognize several low-affinity binding sites (13). This oligomerization appears to involve interactions with the Rev protein dimer interface, as well as regions of the RRE that are distinct from the primary Rev binding site in stem-loop IIB (14). The RRE resistance mutations that we describe here may increase the efficiency by which Rev is able to bind to such secondary RRE binding sites.
Several studies performed in vivo using subgenomic reporter constructs have indicated that sequences throughout the RRE are important for function (16, 17, 31, 41, 43) although others have concluded that stem-loop IIB is the only RRE region required for Rev function (32). Additionally, studies using RNA aptamers, which have been selected for high binding affinity to Rev, have shown the lack of a strict correlation between the affinity of a primary binding sequence for Rev and its function in vivo (47).
Several lines of evidence specifically point to the stem-loop V region of the RRE as a region important for Rev function although it is distant from stem-loop IIB in the RRE secondary structure model. For example, mutations predicted to disrupt the stem of stem-loop V are nonfunctional in both HIV-1 (17) and HIV-2 (18), and oligonucleotides complementary to the stem of stem-loop V are capable of completely disrupting preformed Rev-RRE complexes in vitro (5). These oligonucleotides are also 9-fold more active in blocking Rev-RRE function in vivo than oligonucleotides directed to stem-loop IIB (22). Additionally, previous work from our own group has shown that single-nucleotide changes in stem-loop V of the RRE can overcome resistance to the transdominant negative rev allele, RevM10 (27), by causing a rearrangement of stem-loops III, IV, and V of the RRE into an alternative stable secondary structure (38). Together, all of these studies, as well as ones presented here, highlight the notion that regions in the RRE outside the primary Rev binding site are important for Rev function. They suggest that the RRE is a dynamic structure with long-range interactions capable of structural rearrangements. Such interactions could be affected by the mutations we have identified.
Although the A7854G change in the RRE clearly mediated resistance in the context of NL4-3, we observed that other viruses which naturally possessed this polymorphism still were at least partially sensitive to the two inhibitors used in this study. In particular, we have previously shown that both compounds effectively inhibited Gag protein production, but not Nef, in U1 cells. The growth of the primary isolate 93BR021 in peripheral blood mononuclear cells (PBMCs) and Rev function in the dual luciferase cell line (DLRev) that was used for secondary screening (46) were also affected. In each of these cases, the targeted virus had an RRE with a G at 7854 (NL4-3 numbering) as well as multiple other changes relative to NL4-3 over the length of the extended RRE (12/320 nt for 93BR021, 10/320 nt for the U1 cell virus, and 5/320 nt for the virus in the DL Rev cells) (data not shown).
In the present study, replication of virus derived from the HIV proviral clone R7/3 was shown to be inhibited by both compounds tested despite the fact that the R7/3 RRE contained a G at position 7854. The R7/3 RRE also has four additional changes relative to the NL4-3 RRE. This result, together with the observations discussed in the preceding paragraph, strongly suggest that the overall genetic background of any particular virus contributes to the net Rev function of the virus. Factors in the genetic background that are likely to have an influence include the intrinsic activity of the particular Rev allele expressed and the absolute and relative levels of expression of the various classes of mRNA expressed by a particular viral strain. Virtually any variation in the genetic background that would enable the virus to produce its Rev-dependent protein products more efficiently might be expected to be selected in the presence of the compounds since the compounds compromise Rev function and, as a consequence, essential Rev-dependent protein expression.
Growth of the R7/3 viral clone in the presence of compound 103833 led to the selection of two mutations in the envelope open reading frame that conveyed resistance. One of the mutations was synonymous for the envelope protein sequence but is rarely seen in the HIV sequences database (37). The second mutation changed an amino acid found in R7/3 to the amino acid normally seen in the NL4-3 clone. Neither mutation on its own caused complete resistance although the synonymous change led to partial resistance. While further experiments are needed to understand why the combination of these two mutations cause resistance, the fact that these changes are outside either Rev or the RRE underscores the points discussed above and supports the hypothesis that net Rev function in any given virus is multifactorial. In this particular case, it is tempting to speculate that these changes are operating at the level of RNA structure to create an envelope or Gag-Pol mRNA that can be expressed more efficiently than the wild-type mRNAs in the presence of the inhibitory compound.
We thank Yuntao Wu (George Mason University) for providing Rev-CEM cells and Alan Frankel (University of California, San Francisco) for the R7/3 proviral clone.
This work was supported by National Institutes of Health grants CA097095 and AI054335 to M.-L.H. and AI054213, AI068591, and AI087505 to D.R. E.A.S. was partially supported by an NIH Infectious Disease Training Grant to the University of Virginia (AI007046). H.C. was supported by a fellowship from the Pfizer Initiative in International Health. Salary support for M.-L.H. and D.R. was provided by the Charles H. Ross, Jr., and Myles H. Thaler Endowments at the University of Virginia.
Published ahead of print on 2 February 2011.