|Home | About | Journals | Submit | Contact Us | Français|
The human immunodeficiency virus type 1 (HIV-1) matrix (MA) protein targets HIV-1 precursor Gag (PrGag) proteins to assembly sites at plasma membrane (PM) sites that are enriched in cholesterol and phosphatidylinositol-(4,5)-bisphosphate [PI(4,5)P2]. MA is myristoylated, which enhances membrane binding, and specifically binds PI(4,5)P2 through headgroup and 2′ acyl chain contacts. MA also binds nucleic acids, although the significance of this association with regard to the viral life cycle is unclear. We have devised a novel MA binding assay and used it to examine MA interactions with membranes and nucleic acids. Our results indicate that cholesterol increases the selectivity of MA for PI(4,5)P2-containing membranes, that PI(4,5)P2 binding tolerates 2′ acyl chain variation, and that the MA myristate enhances membrane binding efficiency but not selectivity. We also observed that soluble PI(4,5)P2 analogues do not compete effectively with PI(4,5)P2-containing liposomes for MA binding but surprisingly do increase nonspecific binding to liposomes. Finally, we have demonstrated that PI(4,5)P2-containing liposomes successfully outcompete nucleic acids for MA binding, whereas other liposomes do not. These results support a model in which RNA binding protects MA from associating with inappropriate cellular membranes prior to PrGag delivery to PM assembly sites.
The matrix (MA) domain of the human immunodeficiency virus type 1 (HIV-1) precursor Gag (PrGag) protein serves several functions in the viral replication cycle. One essential function is to target PrGag proteins to their assembly sites at the plasma membranes (PMs) of infected cells (4, 5, 11, 16, 25, 29, 30, 33, 35, 39, 43-45, 47, 50, 54, 56, 57). A second function is the recruitment of the viral surface/transmembrane (SU/TM; also referred to as gp120/gp41) envelope (Env) protein complex into virions (14, 15, 18, 19, 27, 51-53). In addition to these activities, numerous reports have attributed nucleic acid binding properties to retroviral MAs (24, 38, 47), and with some viruses MA appears to serve in an encapsidation capacity (24). While no encapsidation role has been assigned for HIV-1 MA, experiments have shown that MA can substitute for the HIV-1 nucleocapsid (NC) protein assembly function (38) under some circumstances, presumably by virtue of its facility to concentrate PrGag proteins by binding them to RNAs (38).
A number of structural studies have been conducted on HIV-1 MA (1, 22, 41, 42, 49). The protein is N terminally myristoylated and composed of six α helices, capped by a three-strand β sheet (7, 22, 41, 42, 49). The protein trimerizes in solution and in crystals (22, 28, 49) and recently has been shown to organize as hexamers of trimers on lipid membranes (1). The membrane binding face of HIV-1 MA is basic, fostering its ability to associate with negatively charged phospholipid headgroups (1, 22, 30, 41, 42, 49). The importance of such an interaction has been underscored in molecular genetic experiments which demonstrated that depletion of PM phosphatidylinositol-(4,5)-bisphosphate [PI(4,5)P2] reduced the assembly efficiency of HIV-1 (9, 36). Consistent with these observations, HIV-1 MA preferentially binds to soluble PI(4,5)P2 mimics through contacts with the headgroup and 2′ acyl chain, and binding promotes exposure of the MA myristate group and protein oligomerization (17, 21, 40-43, 46). However, PI(4,5)P2 is not the only lipid to demonstrate an association with HIV-1. In particular, HIV-1 appears to assemble at cholesterol-rich PM sites, cholesterol is highly enriched in HIV-1 virions, and cholesterol depletion reduces viral infectivity (2, 6, 8, 20, 23, 26, 31, 34, 37). The HIV-1 lipidome shows additional differences from the PM lipids of infected cells (2, 5, 8), suggesting that other lipids could affect PrGag-membrane binding or virus assembly site selection.
To gain a better understanding of the functions and interactions of HIV-1 MA, we have examined the liposome and nucleic acid binding properties of purified myristoylated MA. Using liposome flotation assays and a novel liposome bead binding assay, we have demonstrated that the PI(4,5)P2 binding specificity of MA is enhanced by cholesterol, that protein myristoylation increases membrane binding efficiency but not specificity, and that 2′ acyl chain variation is compatible with PI(4,5)P2 binding. We also examined whether soluble PI(4,5)P2 mimics could compete with liposomes for MA binding. Surprisingly, we found that soluble mimics not only failed to compete with PI(4,5)P2 liposomes but also increased MA binding to membranes that do not contain acidic phospholipids. Finally, we have observed that while MA does bind nucleic acids, nucleic acid binding is outcompeted by PI(4,5)P2-containing liposomes. Our results suggest models for PrGag-membrane and RNA association and the HIV-1 assembly pathway.
Myristoylated HIV-1 MA and matrix plus capsid (MACA) proteins, as well as unmyristoylated MA proteins, were expressed in Escherichia coli strain BL21(DE3)/pLysS (Novagen) along with Saccharomyces cerevisiae N-methyltransferase from pET-11a-based vectors kindly provided by Michael F. Summers (University of Maryland Baltimore County) (41, 42, 49). Myristoylated proteins were purified under conditions that yield only low levels (<10%) of unmyristoylated species as verified by mass spectrometry with one of our preparations and as described previously (1, 41, 42, 49). The unmyristoylated MA protein was purified similarly, except that myristic acid was excluded from the bacterial growth media during the induction phase. The proteins were desalted by buffer exchange in Sephadex G25 spin columns in 10 mM sodium phosphate (pH 7.8), supplemented with β-mercaptoethanol (1 mM final concentration), aliquoted, and stored at −80°C under nitrogen gas. Protein purities were evaluated after fractionation by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (1, 3, 50, 51, 58) by Coomassie blue staining (1, 3, 58) and immunoblotting using a primary anti-HIV CA monoclonal antibody (Hy183; kindly provided by Bruce Chesebro, Rocky Mountain Laboratory) or sheep antiserum (no. 286; lot IDV-010) to MA obtained from Michael Phelan via the NIAID AIDS Research and Reference Program.
Liposomes were prepared from stock solutions in chloroform of cholesterol (Sigma), 1,2-dioleoyl-sn-glycerol-3-phosphocholine (PC; Avanti), 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine-N-(7-nitro-2-1,3benzoxadiazol-4-yl) (NBD-DOPE; Avanti), 1,2-diacyl-sn-glycero-3-phospho-l-serine (PS; Avanti), soy phosphatidylinositol (PI; Avanti), brain PI(4,5)P2 (Avanti), 1,2-dioleoyl-sn-glycero-3-phospho-(1′-myoinositol-4′5′)-bisphosphate [DOPI(4,5)P2; Avanti], and 1,2-dipalmitoyl-sn-glycero-3-phospho-(1′-myoinositol-4′,5′)-bisphosphate [DPPI(4,5)P2; Echelon]. Lipids were dried in glass vials with a stream of nitrogen gas, supplemented with liposome buffer (10 mM HEPES [pH 7.4], 50 mM NaCl, 0.002% sodium azide), and suspended by 20 30-s rounds of sonication in a Branson 1210 bath sonicator with incubations on ice between each sonication round. Final liposome lipid concentrations were 2 mg/ml, and liposomes were stored for up to 1 month under nitrogen at −20°C. By weight, cholesterol-negative PC liposomes were composed of 99.8% PC and 0.2% NBD-DOPE, while cholesterol-containing PC liposomes were composed of 20% cholesterol, 79.8% PC, and 0.2% NBD-DOPE. PS, PI, brain PI(4,5)P2, DOPI(4,5)P2, and DPPI(4,5)P2 liposomes were prepared at either 10% (wt/wt) or concentrations indicated in the text by correspondingly reducing PC concentrations.
Liposome flotation assays followed previously described protocols (9, 10). Briefly, proteins at 20 μg/ml and liposomes at 200 μg/ml were incubated at 25°C for 30 min in 0.2 ml of float buffer (20 mM HEPES, pH 7.4). After incubations, samples were supplemented with 1 ml of 85.5% sucrose in float buffer, layered with 2.8 ml of 65% sucrose in buffer, and topped with 1 ml of 10% sucrose in float buffer. The step gradients, prepared in Beckman TLA 100.1 centrifuge tubes, were centrifuged at 128,000 × g for 4 h at 4°C. After centrifugation, five 1-ml fractions were collected from gradient tops to bottoms and subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis. MACA proteins from gradient fractions were detected by immunoblotting as described above, and protein levels in each fraction were quantitated using NIH Image software from immunoblot images generated by an Epson Perfection 1240U scanner. For graphing purposes, the percentages of protein detected in the top two (liposome) fractions of each gradient were plotted.
For bead binding assays, 0.2 ml of packed nickel-nitrilotriacetic acid (Qiagen) beads was washed with 0.5 ml of wash buffer (25 mM sodium phosphate [pH 7.8], 50 mM NaCl, 0.1 mg/ml bovine serum albumin [Sigma A4503]), suspended in 0.5 ml of wash buffer supplemented with 20 μl of 10-mg/ml bovine serum albumin, incubated for 5 min at 4°C, and washed three times with 0.3 ml wash buffer. After the washes, packed beads were supplemented with 0.2 ml wash buffer minus or plus 2 or 3 μg of His-tagged MA or CA. Proteins and beads were incubated for 2 h at 4°C, after which beads were pelleted, washed twice with 0.3 ml wash buffer, and suspended in a total volume of 360 μl wash buffer on ice; the estimated bead-bound MA or CA concentration used in each assay was 300 nM.
For liposome binding assays, 60 μl of beads in wash buffer was supplemented with 2 μl of 2-mg/ml NBD-DOPE-tagged liposomes (prepared as described above), and incubated for 16 to 18 h at 4°C, after which beads were pelleted (1 min; 13,700 × g), quickly washed twice with 300 μl wash buffer, and resuspended in 50 ml of wash buffer on ice. For viewing of fluorescently tagged beads, samples were mixed by pipetting up and down and 10-μl samples were applied to microscope slides and covered with a 22-mm by 22-mm coverslip. Fluorescent green beads were imaged on a Zeiss AxioPlan fluorescence microscope using a 20× (LDA-Plan) objective and Zeiss filter set 10 (excitation band-pass, 450 to 490; beam splitter Fourier transform, 510; emission band-pass, 515 to 565). For every individual experiment, exposure and gain settings were identical, and care was taken to make sure that brightness values (on a scale of 0 to 4,095) were below the maximum. After collection of multiple bead images per sample as gray-scale tagged-image-file-format files using Improvision OpenLab software, images were ported to NIH Image J for analysis (3). For each bead, normalized bead brightness values were determined. To do so, areas and average brightness values were collected from circled beads. Average background brightness values were calculated by dividing total image average brightness values (with bead area brightness values zeroed out) by total image areas minus bead areas. Normalized bead brightness values (average bead brightness minus background brightness) were averaged for all beads of a given incubation in an experiment, and results of experiments were normalized to results with liposomes containing 10% PI(4,5)P2 and without competitors.
For PI(4,5)P2 analogue addition experiments, the di-C4-PI(4,5)P2 (Echelon) or di-C8-PI(4,5)P2 (Avanti) analogues were added to incubations from stocks in water to achieve final concentrations of 20 or 80 μM, respectively. For nucleic acid binding studies, two fluorescently tagged probes were used. One was a 30-mer DNA oligonucleotide (5′-Alexa 594-TGTGTGCCCTGTGAAAATGTGCCCTGTGTG-3′; Invitrogen) that was available in our lab: gel shift assays (55) indicated that its Kd (dissociation constant) for MA was approximately 15 μM. The second nucleic acid corresponded to HIV-1 Mal isolate RNA nucleotides (nt) 1 to 4001. This was prepared by T7 RNA polymerase transcription from a SalI-linearized SP64HIV1-4005 plasmid (kindly provided by J.-C. Paillart, CNRS-Université de Strasbourg) as described previously (58) but with the use of 0.5 mM unlabeled nucleoside triphosphates plus 0.1 mM Cy3-CTP (GE Healthcare) in the transcription buffers. For nucleic acid binding studies, fluorescently tagged nucleic acids at the indicated concentrations replaced liposomes in binding assays and were imaged using Zeiss filter set 15 (excitation band-pass, 546/12; beam splitter Fourier transform, 580; emission long pass, 590). For nucleic acid-liposome competition assays, liposomes and nucleic acids were mixed at the concentrations indicated in the legends. Colorized versions of images were prepared in Adobe Photoshop by converting them to 8-bit red-green-blue tagged-image-file images and adjusting selective whites and neutrals to green (100-0-100-0) or red (0-100-100-0) settings (3).
A number of studies concerning the phospholipid binding properties of HIV-1 MA have demonstrated its affinity for negatively charged phospholipids (1, 11, 30, 56). Based on observations that cellular reduction of PI(4,5)P2 also reduced HIV-1 release from cells (36), recent investigations have shown how MA binds to soluble PI(4,5)P2 analogues (41) and have indicated that in vitro-translated HIV-1 PrGag proteins preferentially bind to membranes containing PI(4,5)P2 (9). Because experiments with soluble headgroup mimics exclude possible membrane contributions and because the components of in vitro translation systems are not completely characterized, we have undertaken related membrane binding investigations using purified myristoylated MA proteins (unless otherwise indicated), defined liposomes, and a novel fluorescent liposome bead binding assay. As illustrated in Fig. Fig.1,1, for this assay, fluorescently tagged liposomes are incubated with MA-coated beads, and binding efficiency is monitored conveniently by fluorescence microscopy of the beads. As a complementary assay, we employed the previously described liposome flotation assay (9, 10). For these experiments, myristoylated MACA proteins were used, as they allowed us to detect proteins with a high-titer CA antibody and because CA did not demonstrate membrane binding properties by itself (see below).
To examine the phospholipid headgroup influences on MA-membrane binding, we initially tested neutral (zwitterionic) PC liposomes or PC liposomes doped with 10% of the negatively charged PS or PI(4,5)P2 phospholipids. Under these conditions, we found that proteins bound poorly to PC liposomes in flotation assays (Fig. (Fig.2,2, top), as indicated by the low signals in the top (liposome) fractions. Consistent with the notion that basic MA residues mediate binding to acidic phospholipids, proteins bound well to both PS- and PI(4,5)P2-containing liposomes, but we observed little distinction between the two under these conditions. Similar results were obtained in bead binding assays (Fig. (Fig.2,2, bottom). As illustrated, both fluorescently tagged liposomes containing PS and those containing PI(4,5)P2 lit up MA-coated beads (A, B, and G), whereas brightness values were low for MA beads incubated with PC liposomes (C and G). Control incubations with CA-coated beads yielded low binding levels with all of the liposome compositions (D to G).
Because HIV-1 appears to assemble at cholesterol-rich PM sites (2, 6, 8, 20, 23, 26, 31, 34, 37), we tested cholesterol effects on liposome binding. Thus, we altered the compositions of our PC, 10% PS, and 10% PI(4,5)P2 liposomes by adding cholesterol to a final concentration of 20%, within the observed range for HIV-1 (5, 8), and correspondingly reducing the liposome PC concentrations. In liposome flotation assays, this modification had the effect of slightly increasing protein binding to PC liposomes, but it also increased protein specificity for PI(4,5)P2- versus PS-containing liposomes (compare Fig. Fig.2,2, top, and Fig. Fig.3,3, top). A similar effect was observed in bead binding assays (Fig. (Fig.3,3, bottom). As shown, while PC-cholesterol liposome binding to MA beads remained low (C and D), the presence of cholesterol resulted in a relative enhancement of MA binding to liposomes containing PI(4,5)P2 versus those containing PS (compare Fig. 3A, B, and D with Fig. Fig.2G).2G). Control experiments indicated that the presence of cholesterol did not improve the background binding of any of our liposomes to CA beads (Fig. (Fig.33 legend). Given this observed effect of cholesterol on the membrane binding specificity of MA, we included the sterol in all the studies described below.
Using liposomes containing 20% cholesterol and 60 to 75% PC, we next tested the effects of different concentrations of PI(4,5)P2 and PS in bead binding assays. For these experiments, the concentrations of PI(4,5)P2 or PS were varied from 5 to 20% of the total liposome weight. With PI(4,5)P2, we observed that MA-liposome binding increased between 5 and 10%, and higher concentrations showed little additional effect (Fig. (Fig.4A).4A). This result is comparable to those observed for the membrane binding of in vitro-translated HIV-1 PrGag proteins (9). Relative to PI(4,5)P2, various PS liposome concentrations between 5 and 20% gave a more direct effect, with MA binding increasing at each jump in the PS concentration (Fig. (Fig.4A).4A). We did not test PS concentrations of higher than 20% to avoid complications due to potential changes in lipid fluidity and packing (1, 32, 48).
In addition to the above studies, it also was of interest to examine how MA myristoylation influenced membrane binding. Consequently, we compared binding levels of MA and myristoylation-minus MA proteins to cholesterol-PC liposomes or to liposomes containing 10% PS or PI(4,5)P2. We observed (Fig. (Fig.4B)4B) that loss of the MA myristate moiety reduced liposome binding but not to the extent observed for in vitro-translated PrGag proteins (9): potential reasons for these differences are discussed below. Significantly, while the efficiency of MA binding to membranes decreased with myristoylation-minus proteins, selectivity for acidic phospholipids and PI(4,5)P2 in particular appeared to be maintained (Fig. (Fig.4B).4B). These results support a model in which phospholipid headgroup-directed MA-membrane binding triggers myristate exposure (17, 21, 40-43, 46), which then nonspecifically enhances MA affinity for membranes.
Nuclear magnetic resonance analysis of MA binding to soluble PI(4,5)P2 analogues surprisingly showed that MA associates with the compounds via headgroup and 2′ acyl chain contacts, suggesting that MA “flips out” 2′ acyl chains from bilayers during membrane binding (17, 41). To probe potential contributions of 2′ chains for MA binding, we performed assays using several commercially available PI(4,5)P2 phospholipid preparations. One of the preparations was the brain PI(4,5)P2 (PIP2) preparation used in the experiments described above: this was a naturally occurring mixture in which the main species carries 1′ stearoyl (18:0) and 2′ arachidonyl (20:4) chains. We also employed dioleoyl-PI(4,5)P2 (DOPIP2) with two unsaturated 18:1 acyl chains and dipalmityl-PI(4,5)P2 (DPPIP2), with two saturated 16:0 acyl chains. Soy PI (main species 1′ palmitoyl [16:0], 2′ linoleoyl [18:2]) was used as a control for these studies.
When cholesterol-PC liposomes with 10% PIP2, DOPIP2, or DPPIP2 were utilized in liposome flotation assays, we did not detect significant effects on protein-membrane binding. As shown in Fig. Fig.55 (top), 30% of the protein samples were found in the top (liposome-bound) fractions for all three variants. In contrast, very little protein bound to liposomes in which PI replaced the 4,5-bisphosphate species (Fig. (Fig.5,5, top). Similar results were observed with our bead binding assay (Fig. (Fig.5,5, bottom). As illustrated, fluorescent brain PIP2- (A), DOPIP2- (B), and DPPIP2-containing (C) liposomes bound well to MA beads, in contrast to the PI (D) liposomes (E). Together, these experiments suggest that the nature of 2′ acyl chains does not alter MA-PI(4,5)P2 binding preferences dramatically, at least for the limited sampling of variants tested here.
Another potentially interesting line of inquiry was testing whether soluble PI(4,5)P2 analogues could compete with liposomes for MA binding. Such studies could be relevant to the development of compounds that might inhibit virus assembly by impairing membrane association. For these experiments, we tested whether PI(4,5)P2 analogues with saturated 1′ and 2′ four (di-C4) or eight (di-C8) carbon chains could inhibit MA binding to cholesterol-PC or cholesterol-PC-10% PI(4,5)P2 liposomes. As shown in Fig. Fig.6,6, 20 μM and 80 μM di-C4 and di-C8 PI(4,5)P2 analogues failed to compete with PI(4,5)P2-containing liposomes for MA binding. It is plausible that this failure of competition could be due to some saturation of our binding assay, but it is noteworthy that the analogue concentrations employed were approximately 3- to 12-fold greater than the total PI(4,5)P2 concentrations present in the PI(4,5)P2-containing liposomes. Remarkably, we also discovered that addition of the soluble analogues to cholesterol-PC incubations actually increased liposome binding to MA, with the 80 μM di-C8-PI(4,5)P2 addition causing the greatest change (Fig. (Fig.6A).6A). Interestingly, 80 μM di-C8-PI(4,5)P2 addition also increased cholesterol-PC liposome binding to myristate-minus beads, but the effect was less than that with the myristoylated protein (Fig. (Fig.6B).6B). Possible explanations for these observations are outlined in the Discussion.
As with other retrovirus MA proteins, a number of reports have documented the nucleic acid binding abilities of HIV-1 MA (24, 38, 47), although the relevance of these observations to retrovirus replication is uncertain. We modified our bead assay (Fig. (Fig.1)1) to examine MA-nucleic acid binding by substituting fluorescent nucleic acids for fluorescent liposomes. As expected, our experiments indicated that fluorescently tagged, T7-transcribed HIV-1 RNA (nt 1 to 4001) and a randomly chosen tagged 30-mer DNA oligonucleotide both bound well to HIV-1 MA-coated beads but not to uncoated (mock) or CA-coated beads (Fig. (Fig.7A).7A). Interestingly, when MA bead-plus-oligonucleotide incubation mixtures were supplemented with PI(4,5)P2-containing liposomes, nucleic acid binding levels were diminished, whereas cholesterol-PC and cholesterol-PS liposomes showed only minor reductions (Fig. (Fig.7B).7B). These results suggest that membranes containing PI(4,5)P2 may compete with nucleic acids for MA.
To confirm the above results, MA bead binding experiments were performed with rhodamine-tagged (red) 30-mer oligonucleotides and NBD-tagged (green) PI(4,5)P2-containing liposomes. When either of these ligands was incubated separately with MA beads, high levels of binding were observed (Fig. 8A and B). In contrast, when the two fluorescent ligands were mixed, nucleic acid binding levels dropped dramatically (Fig. 8C and E), while liposome binding levels fell only slightly (Fig. 8D and E). These results support the notion that nucleic acids and PI(4,5)P2-containing membranes may compete for MA binding, with potentially important consequences for the HIV-1 life cycle.
Using liposome flotation assays and a novel liposome bead binding assay, we have characterized the binding of HIV-1 MA to membranes containing different phospholipids, and our results (Fig. (Fig.22 to to5)5) support the notion (9, 17, 36, 41) that MA binds preferentially to PI(4,5)P2 versus PC, PS, or PI. We did not observe the level of PI(4,5)P2 selectivity shown with soluble headgroup analogues (41), but this may be due to the presence of membranes and/or the possibilities of multivalent binding in our studies. We also did not achieve the PI(4,5)P2 selectivity found in some studies with in vitro-translated HIV-1 PrGag proteins (9). This could be a consequence of higher protein-to-lipid ratios or the absence of the missing Gag domains in our system. However, it is worth noting that RNase treatment of in vitro translation extracts reduces PrGag selectivity for PI(4,5)P2-containing liposomes (A. Ono, University of Michigan, personal communication), and this effect of nucleic acids is discussed in more detail below. Another condition which reduced selectivity in our hands was the removal of cholesterol (Fig. (Fig.22 versus Fig. Fig.3).3). This may arise from alterations of membrane fluidity (33, 48) in our assays, but it also correlates with the observation that HIV-1 assembles at cholesterol-rich membranes in cells (2, 6, 8, 20, 23, 26, 31, 34, 37), and it is possible that cholesterol modifies the way that PI(4,5)P2 is presented to MA (33).
From our results alone, we cannot distinguish whether the MA binding preference for PI(4,5)P2 versus other phospholipids is simply a reflection of its number of charges, rather than a specificity for lipid headgroups. However, nuclear magnetic resonance analysis showed that MA binds both the PI(4,5)P2 headgroup and elements of the 2′ acyl chain (17, 41). Because of this, we tested several PI(4,5)P2 acyl chain variants but did not observe marked binding distinctions (Fig. (Fig.5).5). Thus, it is possible that MA is not particularly picky as to the type of 2′ chain that is offered or that greater differences may be seen with a wider variety of 2′ acyl chain variations. MA myristoylation also failed to yield a major effect on headgroup selectivity, but it did increase membrane binding levels for all liposomes tested (Fig. (Fig.4B).4B). This observation is entirely compatible with a scenario (17, 21, 40-43, 46) in which MA-phospholipid headgroup binding triggers myristate exposure, which nonselectively enhances membrane binding. Our soluble PI(4,5)P2 analogue experiments (Fig. (Fig.6)6) are also consistent with this scenario. We initially thought that di-C4-PI(4,5)P2 and di-C8-PI(4,5)P2 might compete effectively with PI(4,5)P2-containing membranes as shown in Fig. Fig.9A,9A, but this prediction was not borne out (Fig. (Fig.6).6). Instead, we observed that di-C4-PI(4,5)P2 and especially 80 μM di-C8-PI(4,5)P2 enhanced MA binding to PC liposomes (Fig. (Fig.6A).6A). The addition of 80 μM di-C8-PI(4,5)P2 also increased PC liposome binding to myristate-minus MA beads, but the extent of the increase was less than that with the wild-type protein. Because soluble PI(4,5)P2 exerted a greater effect on the wild type than did myristylation-minus MA binding, our results suggest that soluble PI(4,5)P2 analogues can trigger myristate exposure, which then fosters liposome binding (Fig. (Fig.9B).9B). However, the observation that PI(4,5)P2 analogue addition does affect the myristate-minus protein suggests that the water-soluble analogues also partition somewhat into the PC membranes, offering binding targets for MA (Fig. (Fig.9C9C).
In addition to membrane binding, for a number of years, researchers have reported that retrovirus MA proteins possess nucleic acid binding properties (24, 38, 47). Interestingly, HIV-1 Gag proteins that are defective with regard to their NC-RNA binding properties assemble much more poorly if MA also is absent (38), possibly suggesting that MA has an RNA binding capacity in vivo. Nevertheless, the significance of MA-nucleic acid binding for HIV-1 replication remains to be elucidated. Our results clearly indicate the ability of MA to bind nucleic acids (Fig. (Fig.77 and and8).8). Interestingly, we also have demonstrated that PI(4,5)P2-containing liposomes can outcompete nucleic acids for MA binding, whereas PC and PS liposomes do not (Fig. (Fig.77 and and8).8). While our experiments were performed at concentrations lower than physiological salt conditions (50 mM versus 150 mM NaCl), we speculate that our results are physiologically relevant. Recent observations suggest that RNase treatment of Gag in vitro translation extracts reduces the selectivity of Gag binding for PI(4,5)P2 (A. Ono, University of Michigan, personal communication). When these results are taken into consideration, it implies that MA-RNA binding may increase the ability of the protein to discriminate between phospholipid headgroups. We envision that this could occur if the MA affinity for RNA were between its affinity for PI(4,5)P2 and that for other phospholipids, so that RNA binding could protect MA from binding to inappropriate membranes. Such a model is depicted in Fig. Fig.10,10, which illustrates the potential binding of PrGag MA and NC domains to viral RNA, followed by an MA switch to membrane binding at PM assembly sites. It is noteworthy that this HIV-1 Gag protein conformation change is consistent with previous observations and modeling (12, 13), and further examination of HIV-1 MA interactions with membranes and RNAs will be of interest.
We are grateful to Michael F. Summers for bacterial expression vectors and to Akira Ono for discussing results prior to publication. We also appreciate the help and support of Robin Barklis, Hans Barklis, Michael Jarvis, Ben Kukull, Claudia Lopez, Greg McMurray, Hank McNett, Mary Turnock, Mike Webb, Mark Woolman, Robin Woolman, and Suraj Yalamuri.
This research was supported by National Institutes of Health grants GM060170 and AI071798 to E.B.
Published ahead of print on 23 September 2009.