HIV-1 Lethality and Loss of Env Protein Expression Induced by Single Synonymous Substitutions in the Virus Genome Intronic-Splicing Silencer

Synonymous substitutions can influence virus phenotype, replication capacity, and virulence. In this study, we explored how synonymous codon mutations impacted HIV-1 Env protein expression and virus replication capacity. We changed a single codon, AGG to CGU, which was located in the gp41 coding region (env nucleotide residues 2125 to 2127) and was included in the HIV-1 intronic splicing silencer. This change completely abolished virus replication and Env expression. We also found that changing codon usage in the gp120 region by including an increased number of CpG dinucleotides did not significantly affect Env expression or virus viability. Our findings showed that synonymous recoding was useful for altering viral phenotype and exploring virus biology.

ABSTRACT

Synonymous genome recoding has been widely used to study different aspects of virus biology. Codon usage affects the temporal regulation of viral gene expression. In this study, we performed synonymous codon mutagenesis to investigate whether codon usage affected HIV-1 Env protein expression and virus viability. We replaced the codons AGG, GAG, CCU, ACU, CUC, and GGG of the HIV-1 env gene with the synonymous codons CGU, GAA, CCG, ACG, UUA, and GGA, respectively. We found that recoding the Env protein gp120 coding region (excluding the Rev response element [RRE]) did not significantly affect virus replication capacity, even though we introduced 15 new CpG dinucleotides. In contrast, changing a single codon (AGG to CGU) located in the gp41 coding region (HXB2 env position 2125 to 2127), which was included in the intronic splicing silencer (ISS), completely abolished virus replication and Env expression. Computational analyses of this mutant revealed a severe disruption in the ISS RNA secondary structure. A variant that restored ISS secondary RNA structure also reestablished Env production and virus viability. Interestingly, this codon variant prevented both virus replication and Env translation in a eukaryotic expression system. These findings suggested that disrupting mRNA splicing was not the only means of inhibiting translation. Our findings indicated that synonymous gp120 recoding was not always deleterious to HIV-1 replication. Importantly¸ we found that disrupting an external ISS loop strongly affected HIV-1 replication and Env translation.

IMPORTANCE Synonymous substitutions can influence virus phenotype, replication capacity, and virulence. In this study, we explored how synonymous codon mutations impacted HIV-1 Env protein expression and virus replication capacity. We changed a single codon, AGG to CGU, which was located in the gp41 coding region (env nucleotide residues 2125 to 2127) and was included in the HIV-1 intronic splicing silencer. This change completely abolished virus replication and Env expression. We also found that changing codon usage in the gp120 region by including an increased number of CpG dinucleotides did not significantly affect Env expression or virus viability. Our findings showed that synonymous recoding was useful for altering viral phenotype and exploring virus biology.

INTRODUCTION

Synonymous virus genome recoding is increasing our knowledge of viral biology (1, 2). Synonymous codon mutations do not change protein sequences; nevertheless, they can impact many cellular processes. Indeed, synonymous codon mutations have been shown to affect transcription modifications, translation initiation, translation elongation, translation accuracy, RNA stability, RNA folding and structure, RNA splicing, RNA toxicity, and cotranslational folding (2). Codon usage can also be highly relevant in the temporal regulation of viral gene expression (3). For instance, the human immunodeficiency virus type 1 (HIV-1) Env protein is produced late in the virus replication cycle, and its expression is induced by a viral transinducer, Rev, which is produced earlier in the viral replication cycle. Moreover, HIV-1 gene expression is tightly regulated at the levels of transcription, splicing, mRNA nuclear export, and translation (4).

In simian immunodeficiency virus (SIV), expression of the Env protein depends on the nature of the codons used (3). It has been suggested that in different families of persistent viruses, codon usage is skewed in a distinctive way to enable temporal regulation of late expressing structural gene products, as observed for HIV-1 Env. Similarly, the temporal regulation of the lentiviral Env protein ensures that it is produced late in the lytic cycle of these persisting viruses. Studies have shown that mutations that implemented synonymous codon usage could regulate HIV-1 splicing and replication (5). The regulation of HIV-1 RNA splicing enables balanced splicing and viral replication. However, it remains unclear whether and how codon usage might impact HIV-1 RNA transcription, splicing, transport, or translation.

Synonymous genome recoding can alter both codon usage and dinucleotide frequencies (e.g., CpG or UpA) (1). The HIV-1 genome contains an extremely high frequency of adenines (between 31.7% and 38%) and low GC content. Consequently, HIV-1 protein synthesis can be suppressed by human Schlafen 11 (SLFN11), a novel host restriction factor that utilizes the unique viral codon bias toward AU nucleotides. SLFN11 suppresses HIV-1 protein synthesis in a codon usage-dependent manner (6). Similarly, a recent study showed, with synonymous mutagenesis, that CpG suppression was essential for HIV-1 replication (7). The effect of CpG dinucleotides was exerted posttranscriptionally and independently of translation efficiency (7). Thus, infecting cells with a CpG-enriched HIV-1 Env gp120 region inhibited virion production. Moreover, a zinc finger antiviral protein (ZAP) was identified that could bind to CG sequences in RNA to inhibit virion production. CpG dinucleotides are less frequent than expected in both humans and most mammalian RNA viruses (8). Nevertheless, ZAP may not be the unique factor responsible for the low GC content found in HIV-1.

The present study aimed to explore whether synonymous recoding that altered HIV-1 Env codon usage might affect Env expression, virus replication capacity, and the virus phenotype. To that end, we recoded codons in a way that reduced the usage of codons that are present at high frequencies in the HIV-1 gp160 region. We describe how codon usage recoding impacted the HIV-1 phenotype by altering critical virus mRNA secondary structures implicated in RNA splicing and protein expression.

RESULTS

Previous work showed that codon usage affected the temporal regulation of SIV Env expression (3). In this study, to determine the impact of codon usage on protein expression and virus viability, we recoded the HIV-1 HXB2 Env coding region by replacing six codons, AGG, GAG, CCU, ACU, CUC, and GGG, with the synonymous codons CGU, GAA, CCG, ACG, UUA, and GGA, respectively (Fig. 1; see also Fig. S1 in the supplemental material). These six codons were selected based on their bias of usage. We did not modify regions that overlapped other reading frames to avoid altering other viral genes. Similarly, we did not mutate the Rev response element (RRE). Overall, we introduced a total of 39 mutations in the env gene (R_env [Fig. 1]). These 39 substitutions generated 16 additional CpG dinucleotides. After transfection of the R_env construct into MT-4 cells, no p24 antigen was detected, and no virus infectivity was recovered after five blind cell passages. This result demonstrated the lethality of these 39 synonymous substitutions.

FIG 1.

Schematic representation of the HIV-1 genome and env variants. Wild-type sequences are depicted in black. Altered sequences are depicted in gray. Red sticks represent mutations. Blue sticks represent the mutations in R_env_CpG_2 that were different from the corresponding bases in R_env. LTR, long terminal repeat.

In addition to the 16 new CpG dinucleotides introduced into R_env, we mutated a codon (AGG to CGU) located downstream of the external loop of the HIV-1 intronic splicing silencer (ISS) sequence (HXB2 env positions 2125 to 127 [Fig. S1]). To explore the individual effects of the different introduced substitutions, we generated new variants with only some of the same mutations (Fig. 1; Table 1; Fig. S1). In two of the new variants, we reduced the number of CpGs, either by reverting the newly generated CpGs to wild-type (WT) sequences (R_env_CpG) or by introducing a different mutation (R_env_CpG_2) (Fig. S1). The replication kinetics of R_env_CpG were indistinguishable from those of the WT, but R_env_CpG_2 showed a significantly lower replication capacity (P = 0.003) (Fig. 2A). Of note, the env CGU codon from positions 2125 to 2127 was also reverted to that of the WT (AGG) in R_env_CpG and mutated to AGA in R_env_CpG_2. Thus, synonymous substitutions in this codon might affect virus viability. In the R_env_gp120 construct, all the mutations in glycoprotein 41 (gp41) codons (including the env CGU codon from positions 2125 to 2127) were reverted to WT sequences. This genotype displayed WT replication kinetics (Fig. 2A). Conversely, the R_env_gp41 construct had all the mutations in glycoprotein 120 (gp120) codons reverted to the WT. This genotype was able to replicate (Fig. 2B). However, at the time that p24 antigen was detected, the env codon from positions 2125 to 2127 (referred to here as the env 2125–2127 codon) had completely reverted to the WT. The R_env_gp41 construct had only three mutations, the two env 2125–2127 mutations (AGG to CGU) and an additional G-to-A mutation at env position 1890. Reversion of the env 1890 (R_env_gp120_1) mutation did not revert lethality. Importantly, reversion of the env 2125–2127 codon to the WT (R_env_gp120_2) restored WT replication kinetics (Fig. 2A). Finally, the R_env_2 construct had half of the original 39 substitutions randomly reverted to WT and carried the env 2125–2127 CGU codon. This genotype was able to replicate (Fig. 2B). However, upon virus p24 detection, an additional A-to-G substitution was present at position env 2129. These results demonstrated the relevance of the env 2125–2127 codon in HIV-1 viability.

TABLE 1.

Number of mutations, number of CpG dinucleotides, and codon adaptation index for the HIV-1 env variants included in this study

Env	No. of:		CAIa
	Mutations	CpGs
WT		26	0.1368950
R_env	39	42	0.1315866
R_env_2	21	35	0.1340826
R_env_CpG	20	26	0.1295865
R_env_CpG_2	36	26	0.1291150
R_env_gp41	3	27	0.1360499
R_env_gp120	36	41	0.1320015
R_env_gp120_1	38	42	0.1319750
R_env_gp120_2	37	41	0.1316135

^aCodon adaptation index (CAI) values were calculated from www.jcat.de.

FIG 2.

Replication kinetics of recoded HIV-1 env variants in MT-4 cells. Production of the HIV-1 antigen p24 in culture supernatants was determined on days 0 to 6 to monitor viral replication. (A) Replication kinetics for virus variants that did not generate a mutation or reversion after transfection in cell culture. R_envCpG_2 is significantly less fit than wild-type virus (P = 0.003). (B) Replication kinetics for virus variants that generated or reverted mutations after transfection. For each virus, the slope of the plot of p24 production provided an estimate of the viral replication capacity. The bars indicate the slope of the p24 antigen production graph for each virus after infecting MT-4 cells. The wild-type (HXB2) and mutant env recoded viruses are compared. The significance of the difference between the p24 antigen production slopes was calculated with GraphPrism 7.0.4. Values represent the means ± standard deviations (SD) from at least three independent experiments. When not indicated, differences were not statistically different.

In addition to demonstrating the relevance of the env 2125–2127 codon to virus fitness, we also found that the introduction of 15 additional CpG dinucleotides in gp120 (mutant R_env_gp120) did not have any effect on virus replication capacity (Fig. 2A). To understand the mechanism by which the env 2125–2127 codon impacted viral replication, we cloned all generated env mutants into a eukaryotic expression vector and cotransfected this vector and an HIV-1 Rev expression vector into 293T cells. Immunoblots of the different variants showed that expression levels were drastically reduced in mutant proteins that carried a mutated env 2125–2127 codon (Fig. 3A). Overall, these results were consistent with the results we obtained with infectious viruses; that is, we observed the nearly absent expression of Env in cells transfected with R_env and R_env_gp120_1. Nevertheless, mutants R_env_2 and R_env_CpG, with and without the mutated env 2125–2127 codon, respectively, did not follow this trend (Fig. 3A). We also quantified mRNA expression. We observed no significant differences in the amounts of total virus RNA among cells transfected with the different variants (Fig. 3B). Importantly, we also observed no differences in cytoplasmic or nuclear Env mRNA levels (Fig. 3C and D) among cells transfected with the different variants. This finding suggested that nuclear RNA transport to the cytoplasm was not affected. Taken together, these experiments indicated that changes in the ISS env 2125–2127 codon may have a substantial impact on protein expression, but they did not affect mRNA synthesis or transport.

FIG 3.

Env protein and mRNA production in HEK 293T cells. Full-length WT and env variants were cloned into the pcDNA3.1D/V5-His-TOPO expression vector. Cells were harvested 48 h after transfection, and proteins were subjected to SDS/PAGE and immunoblotting. (A) Immunoblot analyses of protein expression show WT (lane 1), R_env (lane 2), R_env_2 (lane 3), R_env_CpG (lane 4), R_env_CpG_2 (lane 5), R_env_gp41 (lane 6), R_env_gp120 (lane 7), R_env_gp120_1 (lane 8), and R_env_gp120_2 (lane 9). The two bands in Env correspond to gp160 (upper) and gp120 (lower). The two bands in Rev correspond to the two isoforms of the viral protein. Relative to wild-type protein, expression levels of the different mutants were as follows: R_env, 3%; R_env_2, 20%; R_env_CpG, 9%; R_env_CpG_2, 24%,; R_env_gp41, 21%; R_env_gp120, 47%; R_env_gp120_1, 5%; and R_env_gp120_2, 39%. These Env protein expression values were normalized to Rev protein expression. Protein expression quantification was performed with Image Lab 6.0.1 software (Bio-Rad). (B to D) Quantitative PCR results show total levels of Env mRNA (B), cytoplasmic levels of Env mRNA for the WT (1) and R_env (2) (C), and nuclear levels of Env mRNA for the WT (1) and R_env (2) (D). Values represent the means ± SDs from at least three independent experiments.

The HIV-1 Env ISS is a highly phylogenetically conserved RNA stem-loop structure (9). The env 2125–2127 codon is included in the ISS; therefore, we analyzed how the two synonymous mutations introduced in this codon affected the secondary RNA structure of the ISS stem loop. We used RNAfold (version 2.4.10) to analyze a fragment of 250 nucleotides that contained the ISS, the downstream exonic splicing enhancer (ESE3), and the exonic splicing silencer (ESS3a/b) elements (Fig. 4A). We found that the ISS RNA stem-loop secondary structure was disrupted in R_env (Fig. 4B); this structure had lost an external loop (5′-AGUGA-3′) that binds cellular proteins implicated in the virus splicing process (10, 11). We also observed that both the ESE3 and ESS3 structures remained unaffected. On the other hand, R_env_CpG_2, which had the env 2125–2127 codon mutated to AGA instead of CGT, showed a slightly modified ISS RNA secondary structure but retained the external ISS loop (Fig. 4C). When WT (Fig. 4A) and R_env_CpG_2 (Fig. 4C) sequence structures were compared with a consensus structure extracted from Los Alamos HIV database sequences, we observed that the ISS RNA secondary structure is highly conserved (Fig. 4D). The external loop (5′-AGUGA-3′) sequence is always conserved, although some nucleotide variability in the surrounding regions is allowed (Fig. 4D). Therefore, disrupting the ISS stem-loop explained the lethality of R_env and R_env_gp120_1, and the ISS modification observed with R_env_CpG_2 accounted for the reduced replication capacity and Env expression observed with this variant. Taken together, our results confirmed that the stability of the HIV-1 Env ISS secondary RNA structure was essential for Env protein expression and virus replication. In addition, we showed that the env 2125–2127 codon was essential for HIV-1 viability.

FIG 4.

VARNA visualization of RNAfold structures. A fragment of 250 nucleotides of the env gene (env HXB2 positions 2050 to 2300) that contained the ISS (red), the ESE3 (green), and the ESS3 (red) sequences was subjected to RNAfold analysis. The env 2125–2127 codon is highlighted in a blue square. The ISS external loop is highlighted in a gray square. (A) WT env RNA secondary structure. (B) R_env and R_env_gp120_1 RNA secondary structures. The ISS stem-loop is disrupted in these variants. (C) R_env_CpG_2 RNA secondary structure. The ISS stem-loop is slightly modified in this variant. (D) Consensus structure of ISS RNA region calculated by LocARNA. Fragments corresponding to env HXB2 positions 2050 to 2300 were extracted from 153 different HIV-1 subtype B isolates obtained from the Los Alamos HIV database (www.hiv.lanl.gov/). MEF, minimum free energy.

DISCUSSION

Prior studies have documented the relevance of codon usage on virus phenotypes and life cycles (2). Synonymous substitutions were shown to have strong impacts on HIV-1 gag, pol, and env expression and the virus phenotype (5, 7, 12, 13). Synonymous codon changes can produce unexpected outcomes by changing the structure of the mRNA or the protein. A recent report showed that induction of Env protein expression by the SIV Rev protein depended on the nature of the codon usage (3). However, that study only explored Env protein expression and did not explore the impact of coding usage recoding on virus viability and fitness. In this study, we tested the extent to which codon usage affected HIV-1 replication capacity and protein expression.

We found that a single codon change, AGG to CGU, located in the gp41 coding region (env residues 2125 to 2127) and included in the HIV-1 ISS, completely abolished virus replication and Env expression. This result highlighted the utility of synonymous recoding in altering viral phenotypes and exploring virus biology. In addition, we found that codon usage changes introduced in gp120 did not significantly affect Env expression, and consequently, they did not affect virus replication capacity.

The env 2125–2127 codon is located within the phylogenetically conserved HIV-1 ISS sequence (14). Previous studies have described the relevance of the HIV-1 ISS in the control and inhibition of mRNA splicing. The HIV-1 ISS can bind group A cellular heterogeneous nuclear ribonucleoproteins (hnRNPs) (10, 11, 14, 15). HIV-1 ISS also regulates the activity of the 3′ splicing acceptor site, A7, the bipartite exonic splicing silencer, ESS3a/b, and the exonic splicing enhancer ESE3 (14, 16–19). These three regulatory elements, ISS, ESE3, and ESS3a/b, consist of three stem-loops that form the external ISS stem-loop that binds to hnRNPs (20). We demonstrated that synonymous changes in the env 2125–2127 codon disrupted the ISS RNA secondary structure. Interestingly, this codon is located downstream of the external ISS loop sequence 5′-AGUGA-3′. Of note, we also observed that the ESS3 and ESE3a/b stem-loops remained intact after the external ISS loop was disrupted. Our results confirmed that disrupting the RNA secondary structure of the HIV-1 ISS, independently of ESS3 and ESE3, was lethal to HIV-1. Most probably, the external loop sequence 5′-AGUGA-3′ lost its accessibility to other RNA regions and/or proteins. The env 2125–2127 (AGG) codon is extremely conserved throughout all HIV-1 subtypes. However, as shown by the R_env_CpG_2 variant, minor modifications in the ISS stem-loop could be tolerated, as long as the external loop remained conserved; nevertheless, these minor modifications caused some loss of virus fitness.

To explore how codon usage recoding affected protein production, our different Env variants were individually expressed in a eukaryotic expression system. Our results strongly suggest that env variants with a lethal phenotype caused a dramatic decrease in protein production. In contrast to a previous report (3), we found that this reduction was not accompanied by an imbalance in mRNA production or faulty mRNA transport to the cytoplasm. Previous studies have demonstrated that a 5′ splicing donor site was necessary to express HIV-1 Env in eukaryotic systems (21). However, in the expression systems we employed, Env mRNA was expressed without any splicing. The reductions in translation displayed by our variants were not initiated by splicing perturbations. These findings strongly suggested that only translation was affected in our system. Several previous studies have shown that proteins from the hnRNP family were involved in the translation of viral Rev-dependent genes. It has been shown that during HIV-1 infections, hnRNP A1 was retained in the cytoplasm (22). This retention was promoted by the virus, and it allowed hnRNP A1 to act as an internal ribosomal entry site trans-acting factor, which ensured the abundant expression of viral structural proteins. Similarly, hnRNP Q was shown to interact with Rev and increase the levels of Rev-dependent proteins, without altering the levels or transport of the corresponding mRNAs. Thus, hnRNP Q boosted gene expression at the translation step (23). We speculated that by disrupting the ISS secondary structure, R_env lost its capacity to bind the cellular proteins implicated in viral mRNA translation. Accordingly, our results suggested that the ISS stem-loop was implicated in both splicing and protein translation.

HIV-1 has a relatively low number of CpG sequences in its genome (8). This phenomenon has also been observed in many other RNA viruses. The rationale for this low CpG content remains highly controversial. Recently, ZAP, a previously described host antiviral protein (24), was identified as the possible cause for the CpG suppression in HIV-1 and, by extension, other viruses (7). In that study, different synonymously recoded HIV-1 env variants were generated with increasing frequencies of CpGs. Subsequently, they found a correlation between the CpG dinucleotide content and virus viability in cells that expressed ZAP. That study also showed that ZAP could specifically bind CpG dinucleotides and degrade viral RNA (7). In contrast, our results showed that an increased number of CpGs in the gp120 region did not compromise HIV-1 viability or replication capacity in cells that expressed ZAP (7). The R_env_gp120 variant, with 15 additional CpGs in the gp120 region, displayed a replication capacity indistinguishable from that of the WT virus. However, the CpGs introduced in the gp120 region by Takata and colleagues were in different positions than the CpGs introduced in our variants. Therefore, we could not rule out the possibility that only some CpGs are accessible to ZAP, due to RNA secondary structure or interference from the binding of cellular or viral proteins. Intriguingly, Kmiec and colleagues recently showed that the magnitude of CpG suppression did not correlate with the susceptibility to ZAP; instead, only the inhibitory CpGs located in the first 700 bases of the env sequence were targeted by ZAP (25). Nonetheless, it remains unclear why CpGs have been depleted throughout the HIV-1 genome, when only some are targeted by the innate host response (i.e., ZAP).

In conclusion, our findings provided support for the notion that the env 2125–2127 codon and the ISS external loop conformation are implicated in HIV-1 Env expression and virus replication capacity. Our results suggested that translation, but not mRNA transcription or transport, was affected by this RNA secondary-structure disruption. In contrast to other studies, we did not find a clear correlation between the CpG env content and virus viability. However, our study was limited by the fact that we analyzed only a few variants. Further work should clarify the mechanism by which CpG content impacts the expression of different (i.e., Gag and Pol) HIV-1 proteins.

MATERIALS AND METHODS

Cell lines.

MT-4 cells were obtained through the NIH AIDS Reagent Program, Division of AIDS, NIAID, NIH. The cells were propagated in RPMI 1640 medium (Gibco) supplemented with 10% heat-inactivated fetal bovine serum (FBS; Gibco). HEK 293T cells were obtained from the American Type Culture Collection (ATCC). These cells were grown in Dulbecco’s modified Eagle’s medium (DMEM; Gibco) supplemented with 10% heat-inactivated FBS.

Generation of synthetic infectious HIV-1 Env variants.

All designed env sequences were based on the HXB2 strain (www.hiv.lanl.gov). The designed sequences were produced by Invitrogen GeneArt Gene Synthesis. Site-directed mutagenesis was performed to revert mutations in gp41. The oligonucleotides used to revert the env 1890 mutation had the following primer sequences: 5′-CGACCTGGATGGAGTGGGACAGAG-3′ (forward) and 5′-CTCTGTCCCACTCCATCCAGGTCG-3′ (reverse). The oligonucleotides used to revert the env 2125–2127 codon had the following primers: 5′-GAGTTAGGCAGGGATATTCACC-3′ (forward) and 5′-GGTGAATATCCCTGCCTAACTC-3′ (reverse). To obtain infectious viral particles, MT-4 cells were electroporated with three overlapping PCR-amplified fragments, as described previously (26). Briefly, the full-length env was PCR amplified with the following primers: 5′-GAGAGTGAAGGAGAAGTATCAGC-3′ (forward) and 5′-GCAAAATCCTTTCCAAGCCCTTGTC-3′ (reverse). We also PCR amplified an HIV-1 genome fragment that extended from positions 1 to 6306 with primers 5′-TGGAAGGGCTAATTTGGTCC-3′ (forward) and 5′-CTACAGATCATCAATATCCC-3′(reverse) and an HIV-1 genome fragment that extended from positions 8369 to 9709, with primers 5′-ACCCACCTCCCAACCCCGAG-3′ (forward) and 5′-TGCTAGAGATTTTCCACACT-3′ (reverse). Then, the purified PCR products (600 ng each) were cotransfected into MT-4 cells. At 4 and 7 days after transfection, the HIV-1 p24 antigen was quantified (Genscreen HIV-1 antigen [Ag] assay; Bio-Rad) to monitor viral production. Viral particles were collected when the HIV-1 p24 antigen concentration surpassed 500 ng/ml. When p24 was not detected after 7 days of culture, the cells were blind passaged with fresh medium to recover the virus. The designed mutant was considered lethal when p24 was not detected by 30 days posttransfection. Virus titration was performed in MT-4 cells, and values were expressed in terms of the tissue culture dose for 50% infectivity (TCID₅₀), as previously described (13).

Replication capacity assays.

Viral replication capacity was determined by infecting MT-4 cells. Briefly, 0.8 × 10⁶ MT-4 cells were infected at a multiplicity of infection (MOI) of 0.001 (800 TCID₅₀s) and incubated for 3 h at 37°C and 5% CO₂. After incubation, the cells were washed twice with phosphate-buffered saline (PBS) and resuspended in RPMI 1640 supplemented with 10% heat-inactivated FBS in a six-well plate. Every 24 h for 7 days, we collected 300 μl of culture supernatant. Viral p24 antigen was quantified at each time point. To determine growth kinetics, we fit a linear model to the log-transformed p24 data during the exponential growth phase with maximum likelihood methods, as previously described (27).

Plasmids and HIV-1 Env expression.

pcDNA3.1D/V5-His-TOPO (Invitrogen) was used to clone the open reading frames of different env variants and the WT rev. The oligonucleotides used to clone the WT and variant env sequences had the following sequences: 5′-CACCAAACAAGTAAGTATGAGAGTGAAGGAGAAATAT-3′ (forward) and 5′-TCATAGCAAAATCCTTTCCAAGCCC-3′ (reverse). A splicing donor sequence was included upstream of the forward primer, as previously described (21). The oligonucleotides used to clone rev had the following sequences: 5′-CACCATGGCAGGAAGAAGCGGA-3′ (forward) and 5′-TTCTTTAGCTCCTGACTCC-3′ (reverse). For plasmid transfections, we seeded 0.6 × 10⁶ HEK 293T cells into six-well plates, 24 h before transfection. Cells were transfected with 1.5 μg of WT or mutant Env expression plasmids and 0.5 μg of the WT Rev expression plasmid. Transfections were performed with Lipofectamine 3000 reagent (Invitrogen), and cells were collected 48 h after transfection.

Immunoblot analyses and antibodies.

We lysed 10⁶ HEK 293T transfected cells with 100 μl of cell extraction buffer (Thermo Fisher Scientific) supplemented with phenylmethylsulfonyl fluoride and protease inhibitors (cOmplete protease inhibitor cocktail; Sigma-Aldrich). The lysates (10 μg of protein) were separated with electrophoresis on NuPage 4% to 12% bis-Tris gels (Thermo Fisher Scientific) and blotted onto nitrocellulose membranes (Thermo Fisher Scientific). Membranes were probed with antibodies against Hsp90 (C45G5; Cell Signaling Technology), against gp120 (anti-gp120 HIV-1 polyclonal antibody; American Research Products), and against V5 (V5 tag monoclonal antibody, for detecting Rev; Invitrogen).

Real-time quantitative PCR.

To quantify mRNA levels, we collected two vials of 10⁶ HEK 293T cells from each culture that had been transfected with an expression plasmid. Total RNA was extracted from one vial with the High Pure RNA isolation kit (Roche LifeScience). Cytoplasmic and nuclear RNAs were extracted from the other vial with a cytoplasmic and nuclear RNA purification kit (Norgen Biotek Corporation). cDNA was synthesized from 500 ng of RNA with PrimeScript RT master mix (Perfect Real Time; TaKaRa). The transcripts obtained were used to conduct quantitative PCR (qPCR) with TaqMan universal master mix (Applied Biosystems). Oligonucleotides that targeted the RRE region were used to amplify Env mRNAs, with the following sequences: 5′-CAGTGGAATAGGAGCTTTGT-3′ (forward), 5′-TGTACCGTCAGCGTCATTG-3′ (reverse), and 6-carboxyfluorescein (FAM)-5′-CTTGGGAGCAGCAGGAAGCACTAT-3′ (reporter). Rev mRNA was also quantified with the following primer sequences: 5′-GCCCGAAGGAATAGAAGAAGAA-3′ (forward), 5′-GATCGTCCCAGATAAGTGCTAAG-3′ (reverse), and FAM-5′-TGGAGAGAGAGACAGAGACAGATCCA-3′ (reporter). A glyceraldehyde-3-phosphate dehydrogenase (GAPDH) TaqMan gene expression assay (Hs99999905_m1) from Applied Biosystems was used as an internal or endogenous control. The number of RNA copies per microgram of total cytoplasmic and nuclear purified RNAs was calculated with the corresponding plasmid DNA as standards.

RNA sequence computational analyses.

We used RNAfold version 2.4.13 to predict RNA secondary structures (28). The parameters -p and -noLP were set to calculate base-pairing probabilities. We further used LocARNA version 2.0.0RC8 (29) to calculate a structure-guided multiple-sequence alignment of the ISS regions from different HIV genomes. RNA secondary structures were visualized with VARNA version 3.93. Base-pairing probabilities derived from dot plots were color-coded in the RNAfold program.

Statistical analysis.

We evaluated the significance of the differences between replication kinetic slopes with an unpaired t test and Welch’s correction, as implemented in GraphPad Prism 7.04.