Mutations in the HIV-1 envelope glycoprotein can broadly rescue blocks at multiple steps in the virus replication cycle

Significance

HIV-1 adapts over time to bypass blocks imposed by genetic lesions in the viral genome, typically by acquiring compensatory mutations in the defective gene itself. Here we report that HIV-1 can evade replication blocks by acquiring mutations in the envelope (Env) glycoprotein that enhance cell-to-cell transmission. We identified mutations in Env that arose in the presence of the antiretroviral inhibitor Dolutegravir, thereby circumventing restriction. These data, which demonstrate that mutations in Env can provide escape from an anti–HIV-1 drug in vitro, could have broad implications for HIV-1 drug resistance and viral transmission.

Abstract

The p6 domain of HIV-1 Gag contains highly conserved peptide motifs that recruit host machinery to sites of virus assembly, thereby promoting particle release from the infected cell. We previously reported that mutations in the YPX_nL motif of p6, which binds the host protein Alix, severely impair HIV-1 replication. Propagation of the p6–Alix binding site mutants in the Jurkat T cell line led to the emergence of viral revertants containing compensatory mutations not in Gag but in Vpu and the envelope (Env) glycoprotein subunits gp120 and gp41. The Env compensatory mutants replicate in Jurkat T cells and primary human peripheral blood mononuclear cells, despite exhibiting severe defects in cell-free particle infectivity and Env-mediated fusogenicity. Remarkably, the Env compensatory mutants can also rescue a replication-delayed integrase (IN) mutant, and exhibit reduced sensitivity to the IN inhibitor Dolutegravir (DTG), demonstrating that they confer a global replication advantage. In addition, confirming the ability of Env mutants to confer escape from DTG, we performed de novo selection for DTG resistance and observed resistance mutations in Env. These results identify amino acid substitutions in Env that confer broad escape from defects in virus replication imposed by either mutations in the HIV-1 genome or by an antiretroviral inhibitor. We attribute this phenotype to the ability of the Env mutants to mediate highly efficient cell-to-cell transmission, resulting in an increase in the multiplicity of infection. These findings have broad implications for our understanding of Env function and the evolution of HIV-1 drug resistance.

The assembly of HIV type 1 (HIV-1) particles is driven by the expression of the viral Gag polyprotein precursor, Pr55^Gag, which contains several major structural domains required for virus-like particle production, including the p6 domain that promotes membrane scission to release budding virions (1–4). HIV-1 p6 encodes two highly conserved peptide motifs, known as “late domains,” that recruit components of the cellular endosomal sorting complexes required for transport (ESCRT) machinery to sites of virus assembly (5–7). The physiological function of the ESCRT apparatus is to drive membrane-scission reactions that occur in a variety of cellular contexts, including the biogenesis of multivesicular bodies and cytokinesis (5–7). The Pro-Thr/Ser-Ala-Pro (PT/SAP) motif of p6 interacts directly with the ESCRT-I subunit Tsg101 (8–14); the Tyr-Pro-X_n-Leu (YPX_nL, where X is any residue and n = 1–3 amino acids) motif of p6 binds the ESCRT-associated protein Alix (15–21). While the requirement for the p6–Tsg101 interaction in HIV-1 release is well established, the physiological role of p6–Alix binding is less well defined.

Expression of Gag alone is sufficient for the formation of virus-like particles, but the incorporation of the HIV-1 envelope (Env) glycoprotein complex is required for the generation of infectious particles. Env expression on the membranes of both free virions and infected cells promotes viral spread. Productive viral transmission from infected to uninfected cells can occur via two pathways: cell-free infection or cell-to-cell transmission (22–26). The latter pathway, which is thought to be a more rapid and efficient mode of viral propagation than cell-free infection, is initiated by interactions between Env expressed on the surface of the infected cell and CD4 on the surface of the target cell, in the absence of cell–cell fusion, inducing the formation of a virological synapse (VS) (27). Alternatively, when cell-surface HIV-1 Env engages CD4 on target cells, cell fusion can occur, resulting in the formation of multinucleated cells, or syncytia. Several studies have demonstrated the importance of cell-to-cell transmission in vitro in overcoming barriers to cell-free infection, including target cell infectability, virus stability, and defects in virus production (28–30). Additionally, cell-to-cell transmission can allow HIV-1 spread in the presence of broadly neutralizing antibodies (bNabs) (31). Finally, cell-to-cell transmission of HIV-1 has been shown to be less sensitive to antiretrovirals (ARVs) compared with cell-free transmission (29, 32–35). The ability of the virus to evade blocks to infection may in part be attributed to a higher multiplicity of infection (MOI) during cell-to-cell vs. cell-free infection, allowing for a higher percentage of cells to be infected with more than one virus (36). These findings raise the intriguing possibility that HIV-1 could potentially escape the inhibitory activity of antiviral agents through the acquisition of mutations in Env that promote highly efficient cell–cell transmission.

We have previously shown that mutations in the Alix binding site of p6 induce relatively minor defects in Gag processing, virus release, and cell-free particle infectivity, but impose significant delays in replication kinetics in physiologically relevant cell types (37). To further characterize the significance of p6–Alix interactions, we selected for viral revertants that alleviate the replication defects imposed by a panel of mutations in the p6 YPX_nL motif. We identified second-site compensatory changes in both Vpu and Env that rescue replication defects imposed by the mutations in p6. The three Env compensatory mutations that arose can rescue virus replication despite exhibiting severe defects in cell-free particle infectivity. Strikingly, these Env mutations also provide a replication advantage in the context of an integrase (IN) mutant and in the presence of the IN strand-transfer inhibitor (INSTI) Dolutegravir (DTG). De novo selection in the presence of DTG led to the acquisition of at least one additional Env mutation that confers cell-line–independent resistance to DTG in vitro. We attribute the decreased DTG sensititivity of the Env mutants to their ability to efficiently transmit viral material in a cell-associated manner, resulting in an increased MOI during spreading infections.

Results

p6–Alix-Binding Site Mutants Acquire Second-Site Mutations in Vpu and Env.

To further characterize the role of the p6–Alix interaction in HIV-1 replication, we propagated the p6 mutants (Fig. 1A, Left) in culture to select for viral revertants. The Jurkat T cell line was transfected with pNL4-3 WT and p6 mutant proviral clones and virus replication was monitored over time. Consistent with our previous results (37), we observed delayed replication kinetics with all five p6 mutants; the delays relative to the WT ranged from ∼1–2 wk. Virus was collected at days of peak replication and serially passaged in Jurkat cells. By passage three, near-WT replication kinetics were observed for all of the p6 mutants (Fig. 1B). These data suggest that the p6 mutants reverted in culture, perhaps by acquiring second-site compensatory mutations. Viral DNA was isolated from the third passage, amplified by PCR, and sequenced. Unexpectedly, we observed second-site mutations not in Gag but in the Vpu and Env ORFs (Fig. 1A, Right). Two of the p6 mutants, p6-Y36S/L44R and p6-L41A, acquired inactivating mutations in Vpu, M1I, and K31stop, respectively. All five of the p6 mutants acquired substitutions in Env as follows: Env-P81S, Env-A327T, and Env-A556T for p6-Y36A; Env-A556T for p6-Y36S/L44H; Env-I744V and Env-R786K for p6-Y36S/L44R; Env-Y61H and Env-R166I for p6-L41A; and Env-A556T for p6-L41R (Fig. 1A, Right, and SI Appendix, Fig. S1A, Right).

Fig. 1.

Identification of second-site compensatory changes obtained during propagation of p6–Alix binding site mutants. (A) Schematic of the HIV-1 genome indicating the location of the Gag p6–Alix binding site mutations and the Vpu and Env substitutions. Mutations in p6, Vpu, and Env are indicated by underlined residues and amino acid position (NL4-3 numbering). Location of the mutations within the genome are indicated within dashed regions. Labeled domains are defined as follows: CA, capsid; C1–C5, constant region 1–5; FP, fusion peptide; CT, cytoplasmic tail; HR1/HR2, heptad repeat 1/2; MA, matrix; MSD, membrane-spanning domain; NC, nucleocapsid; V1–V5, variable region 1–5. (B) Replication kinetics of the p6–Alix binding site mutants at the third passage. Jurkat T cells were transfected with the indicated pNL4-3 p6 mutant proviral clones and assayed for replication kinetics by measuring RT activity. Virus-containing supernatants were collected at days of peak replication and used to infect new Jurkat cultures. After two rounds of reinfection, cells were collected at days of peak replication, viral genomic DNA was extracted, amplified, and sequenced. Data shown are from one representative selection experiment.

Env Mutants Y61H, P81S, and A556T Rescue Replication-Defective p6–Alix Binding Site Mutants in Jurkat Cells.

To determine whether the selected Vpu and Env substitutions can rescue the replication defects imposed by the p6–Alix binding-site mutations, we constructed pNL4-3 p6/Env and p6/Vpu mutant clones and evaluated their replication kinetics, in parallel with WT and the original p6-mutant clones, in Jurkat cells. The Vpu-inactivating mutations partially rescued the replication-defective p6-Y36S/L44R and p6-L41A mutants (SI Appendix, Fig. S1 C and D). The p6-Y36A replication defect was largely rescued by both Env-P81S and Env-A556T (Fig. 2A). Similarly, the replication defects of both p6-Y36S/L44H and p6-L41R were rescued by Env-A556T, with reverse-transcriptase (RT) peaks occurring at or near the day of peak RT for the WT (Fig. 2 B and D, respectively). In contrast, the replication defective p6-Y36A substitution was not rescued by Env-A327T (SI Appendix, Fig. S1B) and the p6-Y36S/L44R mutations were not rescued by Env-I744V or Env-R786K (SI Appendix, Fig. S1C). The delay in replication exhibited by p6-L41A was rescued by Env-Y61H, but not by Env-R166I (Fig. 2C and SI Appendix, Fig. S1D). Because the Env mutations R166I, A327T, I744V, and R786K did not contribute to rescue of the p6 mutants, they were not analyzed further. Similarly, considering that Vpu mutations often arise during propagation of replication-defective HIV-1 mutants in Jurkat cells, we elected to focus on the Env compensatory mutations.

Fig. 2.

Rescue of replication-defective p6–Alix binding site mutants by compensatory changes in Env. Jurkat T cells were transfected with the indicated pNL4-3 p6 and p6/Env mutant proviral clones and assayed for replication kinetics by measuring RT activity. Individual panels/graphs represent one p6 mutant with its corresponding compensatory changes: (A) p6-Y36A, (B) p6-Y36S/L44H, (C) p6-L41A, (D) p6-L41R. Replication kinetics of WT NL4-3 are indicated by a solid black line, p6 mutants by solid colored lines, and p6/Env mutants by colored line markers of varying shapes. B–D are from one experiment and the WT data are shared across these panels. Data are representative of at least two independent experiments.

Env Compensatory Mutants Display Highly Efficient Replication Kinetics in Jurkat T Cells and Peripheral Blood Mononuclear Cells Despite Severe Defects in Single-Cycle Infectivity and Fusogenicity.

We next determined the replicative fitness of the Env compensatory mutants in the context of WT Gag. We transfected Jurkat cells with pNL4-3 Env mutant proviral clones and observed that the Env compensatory mutants exhibited WT or faster-than-WT replication kinetics (Fig. 3A). The 293T-derived virus-containing supernatants were normalized for RT activity and used to infect the reporter cell line TZM-bl, which contains an integrated luciferase gene under transcriptional control of the HIV-1 long terminal repeat (LTR) (38, 39). The Env mutants displayed approximately two- to sixfold defects in single-cycle infectivity compared with WT (Fig. 3B), a phenotype that is highly discordant with the robust replication fitness of these viruses. At an MOI greater than 1, we also observe defects in infectivity of Env-A556T compared with WT (SI Appendix, Fig. S2A). In contrast to their phenotype in Jurkat cells, in CEM12D7 cells the Env mutants exhibited replication defects relative to WT, consistent with their defects in single-cycle infectivity. To determine whether the infectivity defects of the Env mutants are dependent on the producer or target cell, we used Jurkat-derived virus to infect TZM-bl cells and again observed severe defects in single-cycle infectivity of the three Env compensatory mutants (SI Appendix, Fig. S2B). Finally, we also inoculated Jurkat cells at high viral inputs and observed very inefficient cell-free infection (SI Appendix, Fig. S2C). With spinoculation, a high MOI was achieved, and we again observed defects in single-cycle infectivity of Env-A556T compared with WT (SI Appendix, Fig. S2C). These results establish that the single-cycle infectivity defects conferred by the Env mutants are independent of producer or target cell type and viral input. We also observed that the Env mutants markedly reduced particle infectivity in the context of the original p6 mutations (Fig. 3C), again demonstrating that the Env compensatory mutations rescue replication despite exhibiting defects in cell-free particle infectivity.

Fig. 3.

Enhanced replication kinetics of Env mutants in Jurkat cells are discordant with defective cell-free particle infectivity and impaired fusogenicity. (A) Jurkat T cells were transfected with the indicated proviral clones and replication kinetics were monitored by measuring RT activity. (B) 293T-derived Env mutant viruses were collected 48 h posttransfection, RT normalized, and used to infect TZM-bl cells. Luciferase activity was measured ∼36 h postinfection; data are normalized to WT. Data from at least three independent experiments are shown as means ± SD. (C) Infectivity of the indicated mutants was analyzed as in B. Data from at least three independent experiments are shown as means ± SD. (D) 293T cells were cotransfected with the indicated Env mutant expression vectors and an HIV-1 Tat expression vector at a ratio of 10:1. Twenty-four hours posttransfection, 293T cells were removed and overlaid onto TZM-bl or Jurkat-1G5 cells with serial dilutions in duplicate. Twenty-four hours postoverlay, luciferase was measured as above and normalized relative to WT Env-expressing cells. Data from three independent experiments per reporter cell line are shown as means ± SD. (E) Cell-to-cell transmission of the indicated mutants was measured by infecting Jurkat donor cells with VSV-G–pseudotyped pBR43IeG-Env mutant viruses, normalizing for GFP⁺ cells, and inoculating target Jurkat cells. The accumulation of GFP⁺ cells during a 48-h coculture was measured. Data from three independent experiments were normalized to WT and plotted as means ± SD; ns, not significant. *P < 0.05, **P < 0.01, and ***P < 0.001.

An additional interesting feature of the rescuing Env mutants is that they did not form syncytia during a spreading infection in Jurkat cells. To quantify the fusogenic activity of the Env mutants, we cocultured 293T cells coexpressing Env and Tat with TZM-bl or Jurkat-1G5 reporter cell lines. Fusion of the Env-expressing 293T cells with the CD4/CXCR4-expressing TZM-bl or Jurkat-1G5 cells leads to Tat-mediated transactivation of the LTR-luciferase in the reporter cell and subsequent luciferase expression. The relative fusogenicity of the Env mutants paralleled their single-cycle infectivity; Env-Y61H, P81S/A327T, and A556T were all significantly defective in cell–cell fusion relative to WT (Fig. 3D). In this experiment, the Env A327T substitution is included with the P81S mutation; however, we have shown that the A327T mutation does not contribute to phenotypic differences in replication kinetics or single-cycle infectivity (SI Appendix, Fig. S3 A and B). Thus, the reduced fusogenicity of the Env mutants in these quantitative fusion assays correlates well with their inability to form syncytia in spreading infections.

Finally, given that the Env compensatory mutants do not enhance cell-free infectivity of the p6–Alix binding site mutants, we asked if they might affect cell-to-cell transmission. We infected Jurkat cells with vesicular stomatitis virus-G glycoprotein (VSV-G)–pseudotyped pBR43IeG Env mutant viruses; this NL4-3–based construct expresses GFP from an internal ribosome entry site (IRES) downstream of Nef. The infected donor Jurkat cells were normalized for GFP expression and cocultured with target Jurkat cells. Efficiency of Jurkat-to-Jurkat cell-to-cell transmission was measured as the increase in GFP⁺ cells above input 48 h postcoculture. For this assay, we focused on the Env-A556T mutant, which displays the greatest defect in fusion capacity (Fig. 3D). We found that not only did p6-Y36A alone exhibit a statistically significant decrease in cell-to-cell transmission compared with WT, but the Env-A556T mutant rescued this defect (Fig. 3E). These data indicate that the Env compensatory mutants are able to overcome poor cell-free infectivity by enhancing cell-to-cell transmission.

To investigate the replication fitness of the Env compensatory mutants in physiologically relevant cells, we infected peripheral blood mononuclear cells (PBMCs) from three different donors in duplicate (Fig. 4 and SI Appendix, Fig. S4) and monitored replication kinetics as above (Fig. 4 A–C and SI Appendix, Fig. S4 A–C). Replication in Jurkat cells was analyzed in parallel (Fig. 4D and SI Appendix, Fig. S4D). To overcome the poor first-round infectivity inherent to the Env mutants, viruses were pseudotyped with the VSV-G glycoprotein. All replication downstream from the initial round of virus entry, reverse transcription, and integration would then fully depend on HIV-1 Env. In general, the Env mutants are capable of replicating with WT kinetics in PBMCs; this is particularly evident with mutants Y61H and P81S. However, we did observe donor-to-donor variability in the ability of Env-A556T to replicate in PBMCs (Fig. 4 A–C and SI Appendix, Fig. S4 A–C). “Donor 2” PBMCs supported near-WT levels of Env-A556T replication (Fig. 4B and SI Appendix, Fig. S4B), whereas in PBMCs from “donor 1” and “donor 3,” replication of Env-A556T was impaired (Fig. 4 A and C and SI Appendix, Fig. S4 A and C). Thus, in PBMCs, we observed donor-dependent variability in the capacity of the Env mutants to replicate, recapitulating the phenotypes observed in both Jurkat and CEM12D7 T cell lines. Consistent with our previous results, when we infected PBMCs with 293T-derived, luciferase-encoding Env-mutant viruses, we again observed defects in single-cycle infectivity (Fig. 4E), although the reductions were not statistically significant for the Env-P81S mutant. Thus, in T cells, the Env mutants are capable of robust replication despite their generally low particle infectivity.

Fig. 4.

Replication kinetics of Env mutants in primary cells recapitulate their phenotypes in cell lines. (A–C) 293T-derived, VSV-G–pseudotyped, Env mutants were used to infect PBMCs from three independent donors (donors 1–3) in duplicate (SI Appendix, Fig. S4). Jurkat cells (D) were included for comparison; replication kinetics were monitored by measuring RT activity. (E) PBMCs from three independent donors were infected with 293T-derived Env-pseudotyped pNLuc reporter viruses. Data from donor 1 are from two independent experiments, donor 2 from three independent experiments, and donor 3 from one experiment shown as means ± SD; ns, not significant. **P < 0.01, ***P < 0.001, and ****P < 0.0001.

The Compensatory Env Mutants Do Not Enhance Virus Release Efficiency, Env Expression, or Incorporation of Env or Pol Products into Virions.

To characterize the ability of the compensatory Env mutants to replicate despite exhibiting low cell-free particle infectivity, we investigated the properties of these mutants through biochemical analyses in Jurkat cells. We infected Jurkat cells with 293T-derived, VSV-G pseudotyped Env-mutant viruses and measured the expression of metabolically labeled viral proteins in cellular and viral lysates by radioimmunoprecipitation (SI Appendix, Fig. S5A). As expected, the Env compensatory mutants did not exhibit deficiencies in virus release efficiency compared with WT or Env (−) clones (SI Appendix, Fig. S5 A and B). There was also no significant defect in mutant cellular Env expression or Env processing (gp120/gp160) or virion Env, RT (p66/p51), or IN (p32) incorporation compared with WT (SI Appendix, Fig. S5). These results are in contrast to a recent study in which Env-mediated HIV-1 escape from APOBEC3G restriction was associated with increased incorporation of RT in virions (40). Taken together, these results indicate that the phenotype of the Env mutants cannot be explained by effects on virus assembly and release, viral protein expression, or the incorporation of Env or Pol products into virions.

Mutagenesis of Env Residues Y61, P81, and A556 Reinforces the Phenotypes of the Original Env Mutants.

To understand in more detail the effects of mutations at Env residues Y61, P81, and A556 on HIV-1 replication and infectivity, we introduced both conservative and nonconservative changes at these positions. We observed that nearly all of the mutants replicated efficiently in Jurkat cells (Fig. 5A) yet exhibited severe defects in cell-free particle infectivity (Fig. 5B). One exception is the conservative Env-Y61F mutant, which displays WT levels of infectivity and forms syncytia in Jurkat cultures. These data corroborate our observations with the original three Env mutants that are highly defective for cell-free infectivity yet can replicate efficiently in Jurkat cells, in some cases with kinetics faster than those of the WT.

Fig. 5.

Mutagenesis of Env residues Y61, P81, and A556 confirms the phenotypes of the original Env mutants. (A) Jurkat cells were transfected with the indicated proviral clones and replication kinetics were monitored by measuring RT activity. Data are representative of at least two independent experiments. (B) Single-cycle infectivity of the indicated mutants was measured in TZM-bl cells as in Fig. 3B. Data from at least three independent experiments are shown as means ± SD.

Mutations in Env Can Confer Drug Resistance.

Given the overall robust replication observed with the Env compensatory mutants in Jurkat cells and in some PBMC donors, we next asked if the mutants could rescue a replication defect unrelated to Gag or Env. We transfected Jurkat cells with pNL4-3 proviral clones containing a nonactive-site IN mutation, N155E (41), in the presence or absence of Env compensatory mutations. We observed that all three of the compensatory mutations largely rescue the replication defect imposed by IN-N155E (Fig. 6A), demonstrating that these mutations can broadly rescue replication-deficient viruses.

Fig. 6.

Env mutations partially rescue a replication-delayed IN mutant and provide a replicative advantage in the presence of DTG. (A) Jurkat cells were transfected with the indicated proviral clones and replication kinetics were monitored by measuring RT activity. (B) Jurkat cells were transfected with the indicated proviral clones in the presence of 1.5 or 3 nM DTG and replication kinetics were monitored by measuring RT activity. Cells and virus from the WT escape mutant in the presence of 3 nM DTG were collected at the day of peak replication (gray line, day 28 posttransfection). (C) Schematic of HIV-1 Env indicating the location of the mutations identified by sequencing of DTG-resistant viruses. Mutations in Env are indicated by underlined residues and amino acid position (NL4-3 numbering). Location of the mutations within the genome are indicated within dashed regions. Domains are defined as in Fig. 1A. Jurkat (D) and CEM12D7 (E) cells were transfected with the WT or Env-A539V pNL4-3 proviral clones in the presence or absence of 3 nM DTG and replication kinetics were monitored by measuring RT activity.

Given the ability of the Env mutations to enhance the replication of an IN mutant, we asked whether they could also overcome inhibition mediated by the second-generation INSTI DTG, an ARV that is difficult for HIV-1 to evade both in vitro and in vivo (42). We transfected Jurkat cells with the WT or Env-mutant pNL4-3 proviral clones in the presence of three concentrations of DTG and monitored replication kinetics. At the lowest concentration of DTG tested, 0.3 nM, all viruses replicated like the no-drug control; however, at higher concentrations of inhibitor, 1.5 nM and 3 nM, we observed replication of the three Env compensatory mutants but no, or severely delayed, replication of the WT (Fig. 6B). These results demonstrate that the Env mutations are able to confer escape from DTG. In the presence of 3 nM DTG, WT-transfected cultures showed evidence of virus replication at ∼30 d posttransfection (Fig. 6B, Right), suggesting the acquisition of DTG-resistance mutations. We collected virus-containing supernatants at the day of peak replication, infected new cultures of Jurkat cells, and found that the repassaged virus exhibited partial DTG resistance compared with naïve virus in the presence of 3 nM DTG. We collected cells from this repassaged, partially DTG-resistant virus, extracted the viral DNA, PCR amplified, and sequenced. We did not identify any mutations in IN, but rather identified three mutations in Env: E209K, A539V, and H641Y (Fig. 6C). These results were confirmed by the observation that the Env-A539V mutation arose under the same conditions in two independent selections in Jurkat cells.

To determine whether the Env mutations that arose during virus propagation in DTG were able to confer resistance to DTG, we engineered these mutations into pNL4-3 and evaluated their effects on replication kinetics in Jurkat cells in the presence or absence of DTG. The Env-A539V mutant replicated with a peak on day 15 in the presence of 3 nM DTG, a concentration at which WT did not replicate (Fig. 6D). Additionally, the Env-A539V mutant replicated with WT or faster-than-WT kinetics in the absence of DTG (Fig. 6D), a phenotype similar to that of the Env compensatory mutants described above. Similarly, the Env-E209K and H641Y mutants replicated with WT or faster-than-WT kinetics in the absence of DTG, but conferred only partial resistance to DTG (SI Appendix, Fig. S6A). Combining Env-A539V with either Env-E209K or Env-H641Y resulted in double mutants that replicated in the presence and absence of 3 nM DTG with kinetics similar to those of the WT without DTG (SI Appendix, Fig. S6B). In the presence of 3 nM DTG, the Env-A539V mutant also conferred a replication advantage over WT in CEM12D7 cells (Fig. 6E). Further corroborating these findings, we also identified Env-A539V independently during de novo selection for DTG resistance in CEM12D7 cells. To extend our observations to a clinically relevant HIV-1 strain, we propagated the clade C transmitted/founder virus, K3016 (CH185_TF), in the presence of 3 nM DTG. Consistent with our findings with NL4-3, the virus acquired a mutation in Env, specifically Env-T529I. Remarkably, residue T529 of K3016 corresponds to Env-A539 of NL4-3, the position of the A539V mutation.

Env Mutants Exhibit Decreased Sensitivity to DTG by Enhancing Virus Spread.

To further characterize the mechanism of DTG-resistant Env mutants, we focused on the two gp41 mutants, Env-A556T and Env-A539V. We calculated the DTG IC₅₀ values of Env-A539V and A556T by performing a spreading infection in Jurkat T cells in the presence of serial dilutions of DTG (SI Appendix, Fig. S7). The Env-A556T and A539V mutants displayed a fold-change in resistance compared with WT of ∼4.6 and 5.3, respectively (SI Appendix, Fig. S7); these values are comparable to currently characterized DTG drug-resistance mutations in IN [Stanford HIV Drug Resistance Database (43)]. Interestingly, and in contrast to the Env compensatory mutants obtained during propagation of the p6-mutant viruses, Env-A539V exhibits near-WT levels of cell-free particle infectivity (SI Appendix, Fig. S8). The robust replicative fitness exhibited by Env-A556T in Jurkat cells and in some PBMC donors, despite severe defects in cell-free infectivity, strongly suggests that this mutant is proficient in cell-to-cell infectivity. Indeed, we observed that the Env-A556T mutant exhibits a statistically significant increase in cell-to-cell transmission efficiency compared with WT in the presence and absence of DTG (Fig. 7). The Env-A539V mutant also exhibits a statistically significant increase in cell-to-cell transmission efficiency in the presence and absence of DTG (Fig. 7). Taken together, these results demonstrate that the Env mutants that exhibit reduced sensitivity to DTG at concentrations that inhibit WT are proficient in cell-to-cell transmission.

Fig. 7.

DTG-resistant gp41 mutants Env-A556T and Env-A539V exhibit enhanced cell-to-cell transmission relative to WT. Cell-to-cell transmission of the indicated mutants was measured by infecting Jurkat donor cells with VSV-G–pseudotyped pBR43IeG-Env mutant viruses, normalizing for GFP⁺ cells, and inoculating target Jurkat cells in the presence or absence of DTG. The accumulation of GFP⁺ cells during a 48-h coculture was measured. Data from four independent experiments were normalized to WT in the absence of drug, and plotted as means ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.

The gp41 Env Mutants A556T and A539V Increase the MOI During a Spreading Infection.

To determine the mechanism by which Env-A556T and Env-A539V confer DTG resistance, we again utilized pBR43IeG, which allows us to quantify replication kinetics as a function of viral gene expression. We inoculated Jurkat T cells with virus-producing 293T cells, allowing for cell-to-cell transfer, and measured GFP expression over time (Fig. 8A and SI Appendix, Fig. S9 A and B). In this system, the Env mutants replicate with accelerated kinetics compared with WT. We also observed that at days of peak replication, we measured a higher percentage of GFP⁺ cells with the Env mutants compared with WT (Fig. 8A). To further characterize the replication properties of the mutant viruses, we calculated the geometric mean fluorescence intensity (MFI) of the GFP⁺ cells at days of peak replication and found that cells infected with Env mutant viruses exhibit dramatically brighter GFP fluorescence compared with WT (Fig. 8B and SI Appendix, Fig. S9C). Interestingly, we also observe an increase in MFI with WT and Env-A539V when we infect Jurkat cells by spinoculation at a high MOI (SI Appendix, Fig. S10 A and B). The increase in virally encoded GFP expression, as measured by MFI, in the cells infected with the Env mutants relative to WT (Fig. 8B) suggests that, in the context of a spreading infection, the Env mutants may be overcoming blocks to viral replication by increasing the effective MOI, resulting in an increase in the number of productive infection events per target cell.

Fig. 8.

Env-A539V confers resistance to DTG by accumulating a high percentage of infected cells and by increasing the effective MOI. (A) Jurkat cells were inoculated with virus-producing 293T cells expressing the indicated proviral clones and replication kinetics were monitored by measuring %GFP⁺ cells. (B) The geometric MFI of GFP⁺ cells from A was calculated at days of peak infection. Data from four independent experiments are shown as means ± SD. *P < 0.05, **P < 0.01.

Discussion

In this study, we describe mutations within HIV-1 Env that rescue replication-defective p6–Alix binding site mutants. We further characterize the Env compensatory mutants—Y61H, P81S, and A556T—and find that, despite exhibiting robust replication in the Jurkat T cell line and PBMCs, they display severe defects in fusogenicity and cell-free particle infectivity. This panel of Env substitution mutants not only rescues the replication defects conferred by disruption of p6–Alix binding but also enhances the replication of an IN mutant and confers resistance to the INSTI DTG. We attribute the ability of these Env mutants to rescue replication defects to their efficient transmission via a cell-to-cell route. De novo selections for DTG resistance in both Jurkat and CEM12D7 T cell lines led to the identification of an additional Env mutation, A539V, which confers DTG resistance in both T cell lines. Unlike the Y61H, P81S, and A556T mutations, A539V is highly fit in its ability to infect via both cell-free and cell-to-cell routes. We also provide evidence that the gp41 mutants A556T and A539V exhibit decreased sensitivity to DTG by increasing the MOI by cell-to-cell transmission. Finally, selection for DTG resistance with a subtype C transmitted/founder virus led to the selection of an Env mutation at the same position in gp41 as the A539V mutation, demonstrating that the phenomenon of Env-mediated DTG resistance is not confined to the subtype B clone NL4-3.

The selection of the original three Env escape mutations in this study is influenced by several factors, including cell type, route of virus transmission, and Alix function. Our initial selection experiments (37) were performed in Jurkat cells, which are infected inefficiently by cell-free HIV-1 (28), and in which p6–Alix binding site mutants replicate with a severe delay. In PBMCs, the p6–Alix binding site mutants display variable phenotypes, depending on the donor (37). While it is well established that the interaction between the PT/SAP motif of p6 and the ESCRT-I subunit Tsg101 plays a key role in HIV-1 budding, Alix appears to play a more auxiliary and cell-type–dependent role in HIV-1 replication. The observed reduction in cell-to-cell transmission efficiency that we observed with a p6–Alix-binding site mutant suggests that recruitment of Alix to the VS may promote virus spread.

We demonstrate that the high levels of replicative fitness of the Env mutants described here in Jurkat cells and in some PBMC donors is a result of their competence to mediate cell-to-cell transmission. The three Env mutants that were acquired during propagation of the p6-mutant viruses in Jurkat cells (Env-Y61H, P81S, and A556T) are impaired in their ability to induce syncytia and thus may decrease bystander cell killing and syncytial apoptosis (44). The selection of fusion-defective mutants has been observed previously (45); mutations in Env arose in HIV-1–infected SupT1 cells that improved replication in the absence of syncytia (46, 47), and Env mutations that abrogated syncytium formation were reported to arise during propagation of a Vif-defective virus (40). The lack of cell–cell fusion potentially increases the production of progeny virus before cell death (46, 47). The conservative and nonconservative substitutions that we introduced at positions Y61, P81, and A556 confirm our initial finding that poorly infectious mutants can replicate efficiently in Jurkat cells. The mutants that replicate efficiently in Jurkats and PBMCs despite low cell-free particle infectivity are likely functionally and structurally adapted to promote cell-to-cell transmission. Indeed, this idea is in agreement with the suggestion that cell–cell fusion is inhibited during VS formation perhaps because Env is retained in a prefusogenic state (27).

The positions of the Env compensatory mutations are highly conserved within subtype B viruses. Among the >6,000 subtype B Env amino acid sequences in the Los Alamos National Laboratory (LANL) database, Y61 is 93.7% conserved, P81 is 99.78% conserved, and A556 (A558 in HXB2; gp41 A47) is 99.91% conserved. A similar degree of conservation is found within subtype C viruses. Env-Y61H is present in 708 sequences in the LANL database, 76 of which are subtype B and 34 of which are subtype C. Env-P81S occurs in only two subtype B sequences and Env-A556T is present in only two nonsubtype B sequences (circulating recombinant form strains AE and BF). Env-Y61H was previously identified as an escape mutation that arose during treatment of HIV-1_LAI–infected cells with the HIV-1 attachment inhibitor BMS-378806 (48). To our knowledge, neither Env-P81S nor any other mutation at that position has been previously characterized in the literature. Finally, Env-A556 has been mutagenized extensively in the context of characterizing intra- and interheptad repeat (HR) interactions of the gp41 six-helix bundle (49, 50), which is consistent with our findings with A556T. Recently, Env-A556T (A558T in HXB2) was also selected, along with several substitutions in the C1 domain of gp120, as a resistance mutation to the peptidic inhibitor VIR165 (51). This study suggested that the C1 and HR1 substitutions may alter the kinetics of Env conformational changes, providing resistance by limiting substrate accessibility (51). In our study, the Env compensatory mutations are also located within the C1 domain of gp120 and the HR1 domain of gp41. Although gp120 and gp41 interact noncovalently, C1 and HR1 have been shown through mutagenesis and structural studies to be critical for the stability of gp120–gp41 association in the unliganded state (50, 52–54). These observations suggest that the Env mutations described here may alter the stability of gp120–gp41 interactions.

We mapped the location of the Env compensatory mutations onto a monomeric Env prefusion structure and a trimeric Env CD4-bound structure (SI Appendix, Fig. S11 A and B). We found that the amino acid positions of the three Env compensatory changes are closely clustered in the monomeric structure, with Y61 and A556 only 4.5 Å apart (55) (SI Appendix, Fig. S11A). Interestingly, the trimeric, CD4-bound Env structure highlights a rearrangement of Y61 to a solvent-exposed location central to all three gp120 monomers (56) (SI Appendix, Fig. S11B). Our mutagenesis studies show that conservative mutation Y61F is well tolerated in terms of cell-free particle infectivity, syncytium formation, and replication relative to WT (Fig. 5). P81 is located in a loop region just C-terminal to the ⍺0 helix in the CD4-bound structure; however, the gp41 residues surrounding A556 are not annotated in this structure (56). The Env mutants described here will provide useful tools for further studies of Env structure and function, particularly in understanding how Env regulates cell-free vs. cell-to-cell modes of virus transmission.

In addition to changes in Env, we also found that mutations in Vpu arose in response to the substitutions in the p6–Alix binding site (Fig. 1A and SI Appendix, Fig. S1 C and D). We commonly observe Vpu-inactivating mutations during the propagation of replication-delayed HIV-1 in Jurkat cells. The basis for these loss-of-function mutations in Vpu remains to be explored, but could be associated with a beneficial effect of low-level tetherin expression in cell-to-cell viral transmission (57, 58).

Our characterization of the Env compensatory mutants continued with the observation that they confer a global replication advantage in the presence of a replication-delayed IN mutant. We believe this is not due to bypassing integration, but rather, as discussed above, to the ability of the Env mutants to overcome blocks in the replication cycle by mediating effective cell-to-cell transmission. We also found the Env compensatory mutants can replicate in the presence of DTG at concentrations that inhibit WT virus. De novo selection experiments led to the emergence of a DTG-resistant virus that lacked mutations in IN, but instead contained substitutions in Env. One of these Env mutations, Env-A539V, confers DTG resistance in both Jurkat and CEM12D7 cells, a phenotype that correlates with near-WT cell-free particle infectivity.

The INSTIs—Raltegravir (RAL), Elvitegravir (EVG), DTG, and Bictegravir (BIC)—are the most recently Food and Drug Administration-approved class of anti-HIV inhibitor (59). HIV-1 readily develops resistance to RAL and EVG both in vitro and in vivo as a result of mutations in IN; however, DTG is more difficult for the virus to escape (42, 60, 61). Recently, IN-R263K has emerged in ARV-experienced, INSTI-naïve patients experiencing virological failure on a DTG regimen and in a patient during DTG monotherapy (62–64). IN-R263K confers weak DTG resistance but, interestingly, can spread in culture in the context of cell-to-cell transmission (32, 65). Additional IN mutants in patients experiencing virological failure during DTG monotherapy include Q148H/R, N155H, G118R, and S230R (64). The fold-change in DTG resistance of reported IN mutants is significantly lower than that observed with IN mutants resistant to RAL and EVG; our calculated fold-change for Env-A556T and Env-A539V is comparable to several of these DTG-resistant mutants (43). Interestingly, we observed no differences in DTG IC₅₀ values in the context of a cell-free, single-cycle assay, supporting the hypothesis that reduced DTG sensitivity conferred by the Env mutations is manifested during cell-to-cell transmission.

Several studies have investigated the ability of ARVs to inhibit both cell-free and cell–cell infectivity (29, 33, 35, 66–68). A model proposed by Baltimore and coworkers (29) suggests that a reduction in sensitivity to ARVs due to multiple infections per cell caused by cell-to-cell spread can be a source of ongoing replication in the presence of ARVs. Similarly, Mothes and coworkers (33) reported a correlation between the ability of ARVs to inhibit cell-to-cell transmission and effectiveness against high local MOI at cell–cell contacts. These models and observations would lead to the prediction that mutations that enhance cell-to-cell transmission may lead to drug resistance and that the concentrations of ARVs required to inhibit a cell-free infection event may be insufficient to inhibit cell-to-cell transmission, as we observed in this study (Fig. 9). The hypothesis that cell–cell transfer of the Env mutants results in more proviruses per cell, relative to WT, is supported by the increased MFI observed in mutant vs. WT-infected cultures (Fig. 8B). While it is challenging to define the significance of cell-to-cell transmission in vivo, it is clear that ARVs need to remain effective against the high MOI occurring during cell-to-cell transmission (27).

Fig. 9.

Model for enhanced cell-to-cell transmission with mutant Envs. A model for cell-free and cell-to-cell infection for viruses encoding either WT (Upper) or mutant Env (Lower) proteins in the absence (Left) or presence (Right) of DTG. Producer cells are outlined in blue, target cells in black. Infected cells are indicated by the presence of a provirus (black) within the nucleus (gray). Inhibition with DTG is shown in red solid or dashed lines, indicating complete or partial inhibition, respectively.

Similar to Y61, P81, and A556, A539 (A541 in HXB2; gp41 A30) is 97.99% conserved across clade B viruses, occurring in only 4 subtype B and 14 subtype C sequences of a total of 175. Mutations at this position have been described previously in several different contexts, but largely as fusion inhibitor escape mutations (shown mapped onto a monomeric Env prefusion structure and a trimeric Env CD4-bound structure; see SI Appendix, Fig. S11 C and D, respectively). Env-A539V was also selected during escape from a deleted form of the antiviral factor IFN-induced transmembrane protein 1 (IFITM1) (69) and in response to the MxB restriction factor (70). These studies also observed the appearance of inactivating mutations in Vpu (69, 70). Mutations in Env and Vpu also provided escape from IFITM1 restriction in SupT1 cells; one of the Env mutations, Env-G367R, was described as being defective for cell-free infectivity, but able to spread via cell-to-cell transmission (71–73). These data provide further support for the concept that mutations in Vpu and Env can overcome barriers to virus replication by promoting cell-to-cell transmission.

The identification of ARV-resistance mutations outside of the target gene is rare, although not unprecedented (74). For example, several studies have observed that patients on protease inhibitor (PI)-containing regimens experience virological failure in the absence of drug-resistance mutations in protease (PR) (75–78). Siliciano and colleagues proposed that env sequences from these patients contain mutations, specifically in the Env CT, that confer PI resistance (79). A connection between the gp41 CT and PI resistance may be linked to the role of virus maturation, triggered by PR, in activating Env fusion activity (80, 81). Several clinical reports have observed failure of DTG-containing therapy in the absence of IN mutations, suggesting that mutations outside IN may confer resistance in these patients (82, 83). A recent study observed mutations in the nef gene that conferred INSTI resistance (84, 85). However, to our knowledge, we present a unique instance of de novo selection of Env mutations that confer resistance to DTG in vitro. Ongoing studies should determine whether the Env mutations described here confer resistance to other classes of ARVs, and whether mutations in Env can contribute to escape from ARV therapy in infected individuals (86). The results of this work will provide fundamental insights into mechanisms of drug resistance and viral spread in vivo.

Methods

Cell Culture.

The 293T [obtained from American Type Culture Collection (ATCC)] and TZM-bl [obtained from J. C. Kappes, X. Wu, and Tranzyme, Inc. through the NIH AIDS Reagent Program (ARP), Germantown, MD] cells were maintained in DMEM containing 5% or 10% (vol/vol) FBS, 2 mM glutamine, 100 U/mL penicillin, and 100 µg/mL streptomycin (Gibco) at 37 °C with 5% CO₂. Jurkat (87), CEM12D7 (88), and Jurkat-1G5 (89) (obtained from E. Aguilar-Cordova and J. Belmont through the NIH ARP, Germantown, MD) T cell lines were maintained in RPMI-1640 medium containing 10% FBS, 2 mM glutamine, 100 U/mL penicillin, and 100 µg/mL streptomycin (Gibco) at 37 °C with 5% CO₂. PBMCs were stimulated with 2 µg/mL PHA-P for 3–5 d before infection, then cultured in 50 U/mL IL-2.

Preparation of Virus Stocks.

The 293T cells were transfected with HIV-1 proviral DNA using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s instructions. Virus-containing supernatants were filtered through a 0.45-µm membrane 48 h posttransfection and virus was quantified by measuring RT activity. VSV-G–pseudotyped virus stocks were generated from cells cotransfected with proviral DNA and the VSV-G expression vector pHCMV-G (90), at a DNA ratio of 10:1. Env-pseudotyped reporter virus stocks were generated from cells cotransfected with pNLuc (91) and the Env expression vector pIIINL4-Env (87), at a DNA ratio of 10:1.

Cloning and Plasmids.

The full-length HIV-1 clade B molecular clone pNL4-3 (pNL4-3 WT) was used for this study (92). The pNL4-3/KFS clone, referred to here as pNL4-3 Env (−) was described previously (93). pNL4-3 clones bearing p6 mutations p6-Y36A, Y36S/L44H, Y36S/L44R, L41R, and L41A were described previously (10, 16, 37). pNL4-3 clones bearing Vpu, Env, p6/Vpu, or p6/Env mutations were constructed with the QuikChange Site-Directed Mutagenesis kit (Stratagene) into subclones of pNL4-3 according to the manufacturer’s instructions, and were then recloned into pNL4-3. pIIINL4-Env clones bearing Env mutations were generated as above. The IN mutant pNL4-3/IN-N155E was a kind gift from Alan Engelman, Dana Farber Cancer Institute, Boston, MA (41) and IN/Env double mutants were generated as above. pBR-NL43-IRES-eGFP-nef⁺ (pBR43IeG) is a proviral vector that coexpresses Nef and eGFP from a single bicistronic RNA (obtained from F. Kirchhoff through the NIH ARP, Germantown, MD) (94, 95). pBR43IeG clones containing Env mutations were constructed as above. The full-length HIV-1 clade C molecular clone pK3016 (CH185_TF) was reported previously (96). DNA for transfections was purified in large-scale quantities using MaxiPrep Kits (Qiagen) and mutations were verified by sequencing (Macrogen).

Virus Replication Assays.

Virus replication was assayed in a T cell line model of spreading infection, as previously described (97). Briefly, T cells were transfected with proviral clones (1 µg DNA/1 × 10⁶ cells) in the presence of 700 µg/mL DEAE-dextran or were infected with RT-normalized virus-containing supernatants. Virus replication was quantified by measuring RT activity in collected supernatants over time. Where indicated, the assay was initiated by inoculation of T cells with GFP virus-producing 293T cells at a ratio of 10³ 293T:10⁶ Jurkat, a density that has been optimized to recapitulate WT kinetics; here, virus replication was quantified by measuring GFP⁺ cells by flow cytometry over time. Cells were fixed in 4% PFA and analyzed by flow cytometry using a FACSCalibur (BD); data were collected via CellQuest and processed via FlowJo. When indicated, genomic DNA was extracted from infected cells using the QIAamp Genomic DNA Extraction Kit (Qiagen); viral DNA was amplified by PCR, and sequenced (Macrogen) (SI Appendix, Table S1). Frequency of residues at each mutant position was determined by the AnalyzeAlign tool from LANL (https://www.hiv.lanl.gov/content/sequence/ANALYZEALIGN/analyze_align.html). DTG was a kind gift from S. Hughes, National Cancer Institute/NIH, Bethesda, MD. Data were plotted (transformed and normalized) and IC₅₀ values were calculated using GraphPad PRISM. Curves were fit using nonlinear regression as log(inhibitor) vs. normalized response, variable slope using a least squares (ordinary) fit. Structural modeling were performed in MacPyMOL.

Single-Cycle Infectivity Assays.

TZM-bl is a HeLa-derived reporter cell line that contains a stably integrated HIV–LTR–luciferase construct (38, 39). TZM-bl cells were infected with serial dilutions of RT-normalized virus stocks in the presence of 10 µg/mL DEAE-dextran. Approximately 36 h postinfection, cells were lysed with BriteLite luciferase reagent (Perkin-Elmer) and luciferase was measured in a Wallac BetaMax plate reader. Technical duplicates were normalized to pNL4-3 WT and averaged; data represent the average of independent, normalized experiments. To measure single-cycle infectivity in PBMCs, cells were infected in duplicate with the indicated RT-normalized Env mutant pseudotyped pNLuc viruses. Luciferase was measured 48 h postinfection as above. Technical duplicates were normalized to pNL4-3 WT and averaged; data represent the average of independent, normalized experiments.

Fusion Assay.

The 293T cells were cotransfected with the indicated pIIINL4-Env expression vectors and the HIV-1 Tat expression vector pSV-Tat (98) at a ratio of 10:1. Twenty-four hours posttransfection, 293T cells were removed and overlaid onto TZM-bl or Jurkat-1G5 cells (Jurkat-derived reporter cell line containing a stably integrated HIV-1–LTR–luciferase construct) with serial dilutions in duplicate. Twenty-four hours postoverlay, luciferase was measured as above. Technical duplicates were normalized to pNL4-3 WT and averaged; data represent the average of independent, normalized experiments.

Cell-To-Cell Transmission Assay.

Donor Jurkat cells were infected with 293T-derived VSV-G–pseudotyped pBR43IeG Env mutant viruses. Forty-eight to 72 h postinfection, the percent of GFP⁺ donor cells was measured by flow cytometry. Infected donor Jurkat cells were cocultured with uninfected target Jurkat cells at a ratio that normalized the GFP⁺ input cells to ∼10% per coculture in the presence or absence of 1.5 nM DTG. Forty-eight hours postcoculture, cells were fixed in 4% PFA and analyzed by flow cytometry. Data were collected via CellQuest and processed via FlowJo.

Statistics.

Statistics were calculated using GraphPad PRISM. Unpaired t tests were performed and two-tailed *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001 were considered statistically significant.

Ethics Statement.

PBMCs were obtained from anonymous, deidentified blood donors to the NIH Department of Transfusion Medicine Blood Products Program (NIH CC-DTM).