Evidence in Support of RNA-Mediated Inhibition of Phosphatidylserine-Dependent HIV-1 Gag Membrane Binding in Cells

Abstract

The matrix domain promotes plasma-membrane-specific binding of HIV-1 Gag through interaction with an acidic lipid phosphatidylinositol-(4,5)-bisphosphate. In in vitro systems, matrix-bound RNA suppresses Gag interactions with phosphatidylserine, an acidic lipid prevalent in various cytoplasmic membranes, thereby enhancing the lipid specificity of the matrix domain. Here we provide in vitro and cell-based evidence supporting the idea that this RNA-mediated suppression occurs in cells and hence is a physiologically relevant mechanism that prevents Gag from binding promiscuously to phosphatidylserine-containing membranes.

TEXT

Membrane binding of HIV-1 Gag is one of the essential steps in virus assembly, which takes place primarily at the plasma membrane (PM) (1). The matrix (MA) domain of Gag is essential for directing virus assembly specifically to the PM. MA has an N-terminal myristoyl moiety that facilitates hydrophobic interaction of Gag with membranes (2, 3). The second signal required for efficient association of Gag with membranes is the highly basic region (HBR) in MA, which spans residues 17 to 31 of MA. The HBR mediates the electrostatic interaction with cellular acidic lipids (4–10), in particular, a PM-specific phospholipid, phosphatidylinositol-(4,5)-bisphosphate [PI(4,5)P₂] (9, 11–13). Notably, in vitro studies have shown that MA or its HBR binds RNA as well (13–17). Furthermore, we and others have shown that RNA bound to HBR also regulates membrane binding of Gag in vitro. RNA bound to MA abolishes Gag binding to a negatively charged membrane composed of a neutral lipid phosphatidylcholine (PC) and an acidic lipid phosphatidylserine (PS) at a 2:1 ratio; however, addition of PI(4,5)P₂ allows Gag to alleviate the block imposed by RNA (9, 16, 18). These results collectively support an attractive model in which RNA binding to MA HBR prevents Gag from binding non-PI(4,5)P₂ acidic lipids, of which PS is the most abundant in the cytoplasmic leaflet of cellular membranes. This RNA-mediated inhibition can thus enhance the specificity of Gag binding to membranes that contain PI(4,5)P₂, which in cells is the PM. However, because the available data supporting this model were all obtained by in vitro studies, it remains unknown whether RNA-dependent regulation of membrane binding actually takes place in cells. In this study, we showed that the cellular level of RNA is sufficient for blocking Gag binding to PS. In addition, we observed that Gag present in the cytosol is bound to RNA partially via the MA HBR and that MA-HBR-dependent PS binding of cytosolic Gag is indeed suppressed by RNA.

Inhibition of Gag binding to PC- and PS-containing liposomes by RNA takes place at RNA concentrations lower than that in cells.

Myristoylated full-length Gag synthesized in reticulocyte lysates does not bind negatively charged liposomes containing PS at the physiological concentration of this acidic lipid (∼30%) (12, 19, 20). Even at higher concentrations (∼50%), palmitoyl-oleoyl-PS (POPS), the most abundant form of PS in viral and plasma membranes (21), does not support efficient liposome binding although di-oleoyl-PS does (20). In previous studies (9, 22), we observed that RNase A treatment drastically increases binding of Gag to liposomes with a 2:1 ratio of POPC and POPS (here termed PC+PS liposome). To determine the minimal RNA concentration that is sufficient to inhibit membrane binding of Gag, we added back different amounts of RNA to RNase-treated Gag. Gag synthesized using rabbit reticulocyte lysates as described previously (9, 12) was treated with 0.028 U (400 ng) of RNase A (Qiagen) and incubated in a 30-μl reaction mixture at 37°C for 20 min. RNase A was then inactivated by 10 μl of RNasin (Promega) (40 U/μl). The reaction mixture containing RNase-treated Gag was then incubated with different concentrations of Yeast tRNA (Ambion) for 30 min at 30°C. After another 15-min incubation of Gag with PC+PS liposomes (final volume, 50 μl), the mixture was subjected to equilibrium flotation centrifugation using a sucrose gradient as described previously (12). We found that tRNA in the range of 0.01 to 0.1 μg/μl was sufficient in inhibiting Gag binding to PC+PS liposomes (Fig. 1A). Considering that a typical HeLa cell volume is at most 5 × 10⁻⁶ μl (ranging from 0.5 to 5 × 10⁻⁶ μl) (23) and that a cell has around 10 to 30 pg of RNA (based on the typical RNA yield from a known number of cells) (24, 25), there is at least 2 to 6 μg/μl of RNA in a HeLa cell (data not shown). Consistent with this, the total RNA concentration measured for a mammalian cell line is ∼4 μg/μl (26). Under normal physiological conditions, 80% of cellular RNA is rRNA, 15% to 20% is tRNA or other small RNAs, and the rest is mRNA (27, 28). Thus, the concentration of tRNA that inhibited Gag binding to PC+PS liposomes is lower than the concentrations of both total RNA and tRNA in HeLa and other cell lines (26), suggesting that RNA-mediated inhibition of Gag binding to PS-containing membranes can occur in cells. Notably, Gag binding to PC+PS liposomes containing 7.25 mol% PI(4,5)P₂ was less sensitive to tRNA (Fig. 1B), a finding consistent with the model in which PI(4,5)P₂ is capable of interacting with Gag even in the presence of MA-bound RNA.

Fig 1.

tRNA below intracellular levels inhibits Gag binding to liposomes containing PS but not PI(4,5)P₂. (A) [³⁵S]-labeled Gag synthesized using rabbit reticulocyte lysates was treated with 400 ng of RNase A at 37°C for 20 min. RNase A was blocked using RNasin, and the mixtures were further incubated with the indicated concentrations of tRNA at 30°C for 30 min. PC+PS liposomes were then added and incubated further for 15 min before performing equilibrium flotation centrifugation. Five 1-ml fractions were collected, and 25 μl of each fraction was loaded and analyzed on SDS-PAGE (inset). The liposome binding efficiency was calculated as the amount of membrane-bound Gag (M in inset) as a fraction of total Gag. Data from 3 different experiments are shown as means +/− standard deviations. P values were determined by using Student's t test. **, P < 0.01; ***, P < 0.001. (B) [³⁵S]-labeled Gag incubated sequentially with RNase A, RNasin, and 0.1 μg/μl tRNA as described for panel A was examined for binding to PC+PS liposomes that contained 7.25 mol% PI(4,5)P₂ [PC:PS:PI(4,5)P₂ = 62:31:7.25] in comparison with binding to PC+PS liposomes. The liposome binding efficiency was calculated as described for panel A, and the relative liposome binding efficiency was calculated in comparison with the binding efficiency of RNase-treated Gag not incubated with tRNA for each experiment. The averages of data from 2 different experiments are shown with error bars representing the range of the data. The average liposome binding efficiencies of RNase-treated Gag not incubated with tRNA were 33% for PC+PS liposomes and 59% for PI(4,5)P₂-containing liposomes in these experiments.

RNA binding in cells is mediated partly by MA.

While both total RNA and tRNA concentrations in cells would be sufficient for the inhibition of Gag binding to PS-containing but not PI(4,5)P₂-containing membranes, it is likely that the majority of cellular RNA species are not accessible to the MA HBR in cells due to their localization or interactions with cellular proteins. To determine whether MA HBR is able to interact with RNA in cells, we quantified the amount of RNA bound to Gag immunoprecipitated from cells. In these experiments, we examined nonmyristoylated Gag derivatives encoded in molecular clones that also lack active protease. We used nonmyristoylated Gag constructs to eliminate the variation in immunoprecipitation efficiencies caused by differences in Gag epitope exposure between non-membrane-bound Gag and membrane-bound Gag that forms higher-order multimers (29, 30). We assessed the effect of the 6A2T substitutions in the MA HBR, in which all the basic amino acid residues were replaced with neutral ones (9). As a control, we also examined the RNA binding when NC, the main RNA binding domain of Gag, was replaced with a dimerization motif, leucine zipper (LZ), of a transcription factor, GCN4 (31). This Gag derivative (GagLZ) lacks NC-mediated RNA binding and yet is capable of forming virus-like particles in the context of intact MA (31–33). HeLa cells were transfected using Lipofectamine 2000 according to the manufacturer's instructions with pNL4-3/1GA/PR⁻ (34) and derivatives containing LZ, 6A2T, and 6A2T/LZ changes (constructed using previously reported molecular clones [34–36] by standard molecular cloning techniques). Six hours posttransfection, cells were labeled with [³⁵S]methionine-cysteine. Twenty-four hours posttransfection, cells were lysed using the polysome lysis buffer (23), and Gag was immunoprecipitated using HIV-1 immunoglobulin as described previously (9). A fraction of the immunoprecipitated materials was treated with proteinase K, and coimmunoprecipitated RNA was isolated using Qiazol and a Qiagen miRNAeasy kit as recommended by the manufacturer. In parallel, the remaining immunoprecipitated materials were analyzed on SDS-PAGE, and the amount of Gag was quantified using phosphorimager analysis. The RNA was then quantified using Quant-iT Ribogreen reagent and normalized to the amount of immunoprecipitated Gag. As expected, there was significantly less RNA coimmunoprecipitated with GagLZ than with full-length Gag (Fig. 2). This result is consistent with previous reports showing diminished RNA contents in extracellularly released NC-deficient particles (32, 37, 38). Importantly, the amount of RNA bound to Gag was also reduced when MA HBR was mutated in the full-length Gag context. While introduction of the same MA mutations in the GagLZ context (1GA/6A2T GagLZ) did not cause a statistically significant reduction in RNA binding compared with GagLZ, the amount of RNA bound to 1GA/6A2T GagLZ was most severely reduced relative to full-length Gag among the Gag derivatives. Altogether, these results suggest that not only NC but also MA HBR participates in RNA binding of full-length cytosolic Gag. Consistent with this observation, in vitro studies have shown that MA has stronger affinities for some RNA aptamers than NC, suggesting a potential specific interaction between MA and RNA (13–15, 18, 39). However, these data do not rule out the possibility that the RNA bound to NC is the same RNA that binds MA in cells. If this is the case, due to its inherent high affinity for RNA, NC may facilitate MA-RNA interaction by bringing the RNA into close proximity to the MA HBR.

Fig 2.

MA HBR interacts with RNA in cells. HeLa cells were transfected with pNL4-3/1GA/PR⁻, pNL4-3/1GA/6A2T/PR⁻, pNL4-3/1GA/LZ/PR⁻, or pNL4-3/1GA/6A2T/LZ/PR⁻ and metabolically labeled with [³⁵S] Met/Cys overnight. Cells were lysed and immunoprecipitated with anti-HIV Ig antibody. A fraction of immunoprecipitated materials was examined for the amounts of bound RNA. The amount of coimmunoprecipitated RNA was quantified by fluorometry using Ribogreen reagent. The amount of RNA was normalized to the Gag levels, and the RNA bound to Gag in cells transfected with pNL4-3/1GA/PR⁻ was set to 100%. Data from 3 different experiments are shown as means +/− standard deviations. *, P < 0.05; **, P < 0.01. The amounts of RNAs bound to 1GA/6A2T Gag and 1GA GagLZ were not significantly higher than that of 1GA/6A2T GagLZ.

Gag in the cytosol binds negatively charged liposomes upon removal of RNA.

PS-dependent liposome membrane binding of nonmyristylated Gag synthesized in in vitro translation reaction mixtures is inhibited by RNA present in these reactions (9). To address whether the RNA bound to Gag in cells inhibits PS-dependent membrane binding, HeLa cells (2 × 10⁶ cells) were transfected with HIV-1 molecular clones encoding nonmyristylated Gag as described for Fig. 2. Twenty hours posttransfection, the cells were treated with 100 μl of 0.04% digitonin (in 20 mM HEPES with 150 mM NaCl and Complete protease inhibitor) for 10 min on ice in a microcentrifuge tube. The permeabilized cells were then centrifuged at 13,200 rpm for 10 min in a microcentrifuge to separate the cytosol from perforated cells. The supernatant was then treated or not treated with RNase A (0.7 U or 10 μg) at 37°C for 20 min and further incubated with PC+PS liposomes at 37°C for 15 min. The liposome-bound proteins were then separated on equilibrium flotation centrifugation as previously described (9, 12). Gag present in each fraction was detected using a Typhoon Trio imager (GE) after immunoblotting using HIV-Ig as the primary antibody and Alexa Fluor 488-conjugated anti-human IgG as the secondary antibody. As observed previously for in vitro-synthesized Gag (9), RNase treatment significantly enhanced binding of cytosolic Gag to PC+PS liposomes (Fig. 3). These results indicate that RNAs in cells can block binding of cytosolic Gag to negatively charged membranes that do not contain PI(4,5)P₂. This RNA-mediated inhibition was also observed when NC was replaced with LZ (Fig. 3A and B) or deleted (data not shown), suggesting that neither RNA binding to NC nor NC-driven Gag multimerization is essential for RNA-mediated inhibition of MA-dependent Gag membrane binding. When MA HBR was mutated in the GagLZ context (Fig. 3A and B), there was negligible membrane binding regardless of RNase treatment. This observation indicates that the MA HBR mediated the PC+PS liposome binding of RNase-treated cytosolic Gag examined in these experiments.

Fig 3.

RNA inhibits membrane binding of Gag present in the cytosol. (A and B) HeLa cells (2 × 10⁶) were transfected with pNL4-3/1GA/PR⁻, pNL4-3/1GA/LZ/PR⁻, or pNL4-3/1GA/6A2T/LZ/PR⁻. Twenty hours posttransfection, cells were treated with digitonin, and supernatants containing cytosolic Gag (100 μl) were separated from perforated cells by centrifugation. The supernatants were divided in two aliquots and treated with RNase A or left untreated. Five microliters of PC+PS liposomes (14.6 μg lipid/μl) was then added, and the reaction mixture was further incubated for 15 min before performing sucrose gradient centrifugation. Five fractions were collected and analyzed by Western blotting using anti-HIV Ig as the primary antibody and Alexa Fluor 488-conjugated anti-human IgG as the secondary antibody. M, membrane-bound Gag. (C and D) HeLa cells (6 × 10⁶) were transfected with pNL4-3/PR-, which encodes WT Gag, or with pNL4-3/1GA/PR⁻. Eight hours posttransfection, cells were treated with digitonin, and supernatants containing cytosolic Gag (200 μl) were separated from perforated cells by centrifugation. The supernatants were divided in two aliquots and treated with RNase A or left untreated. Ten microliters of PC+PS liposomes (14.6 μg lipid/μl) was then added, and the reaction mixture was analyzed as described for panels A and B. The amount of membrane-bound Gag versus the total amount of Gag was calculated and is shown as liposome-binding efficiency in panels B and D. Data from at least three independent experiments are shown as means ± standard deviations. ns, not significant; *, P < 0.05; ***, P < 0.001.

The experiments described above were performed using cells expressing nonmyristoylated Gag derivatives to maximize the yield of cytosolic Gag. To determine whether RNA regulates wild-type (WT) Gag in the cytosol in a similar manner, the cytosol of HeLa cells transfected with a molecular clone encoding WT Gag was harvested 8 h posttransfection. At this time point, fluorescently tagged WT Gag shows a cytosolic localization pattern in a larger fraction of cells than at 24 h posttransfection (data not shown). We found that PC+PS liposome binding of WT Gag derived from the cytosol of transfected HeLa cells was enhanced upon RNase treatment (Fig. 3C and D), as observed for nonmyristoylated Gag. These results support the conclusion that RNA prevents cytosolic Gag from binding to prevalent acidic lipids and further suggest that a substantial population of non-membrane-bound Gag that has been observed in cells is maintained due to the block mediated by RNA.

Conclusions.

The data presented in this study show that RNA-mediated inhibition of Gag binding to PS-containing membranes can occur in cells and thus is a physiologically relevant mechanism for the regulation of Gag membrane binding and HIV-1 assembly. While the observed MA-HBR-dependent RNA binding (Fig. 2) supports a direct mechanism whereby RNA outcompetes PS or other non-PI(4,5)P₂ acidic lipids for electrostatic interactions with HBR, we do not rule out a possible indirect mechanism in which a cellular RNA-dependent factor is involved in regulation of MA-acidic lipid interactions. Interestingly, however, a recent report showed that nucleotides facilitate selective binding of synaptotagmin C2 domains to PI(4,5)P₂-containing membranes by blocking electrostatic interactions with PS in untargeted membranes (40). Therefore, the competition between nucleotides and acidic lipids may be a widely used mechanism to promote specific targeting of proteins to PI(4,5)P₂-containing membranes.

In addition to the regulation of Gag membrane binding, previous in vitro studies suggested that the competition between RNA and acidic lipids explains the structural change of full-length HIV-1 Gag between bent and extended shapes (41) and regulation of the NC-dependent RNA chaperone activity (42). These findings collectively highlight the switch function of MA HBR during the late phase of the HIV-1 life cycle. Nevertheless, molecular details of the balance between MA-PI(4,5)P₂ and MA-RNA interactions remain to be determined. In particular, the potential roles for simultaneous binding of MA and NC domains to the same RNA molecule in coordinating Gag membrane binding with Gag multimerization or other steps in the assembly process warrant further investigation. RNA aptamers have been emerging as a new class of potential therapeutics (43). With the current evidence supporting the idea that the MA-RNA interaction regulates membrane binding of Gag in cells, RNA aptamers that specifically interact with MA with a higher affinity than PI(4,5)P₂ might be useful to block HIV-1 assembly in HIV-1-infected individuals.