Trimer Enhancement Mutation Effects on HIV-1 Matrix Protein Binding Activities

ABSTRACT

The HIV-1 matrix (MA) protein is the amino-terminal domain of the HIV-1 precursor Gag (Pr55Gag) protein. MA binds to membranes and RNAs, helps transport Pr55Gag proteins to virus assembly sites at the plasma membranes of infected cells, and facilitates the incorporation of HIV-1 envelope (Env) proteins into virions by virtue of an interaction with the Env protein cytoplasmic tails (CTs). MA has been shown to crystallize as a trimer and to organize on membranes in hexamer lattices. MA mutations that localize to residues near the ends of trimer spokes have been observed to impair Env protein assembly into virus particles, and several of these are suppressed by the 62QR mutation at the hubs of trimer interfaces. We have examined the binding activities of wild-type (WT) MA and 62QR MA variants and found that the 62QR mutation stabilized MA trimers but did not alter the way MA proteins organized on membranes. Relative to WT MA, the 62QR protein showed small effects on membrane and RNA binding. However, 62QR proteins bound significantly better to Env CTs than their WT counterparts, and CT binding efficiencies correlated with trimerization efficiencies. Our data suggest a model in which multivalent binding of trimeric HIV-1 Env proteins to MA trimers contributes to the process of Env virion incorporation.

IMPORTANCE The HIV-1 Env proteins assemble as trimers, and incorporation of the proteins into virus particles requires an interaction of Env CT domains with the MA domains of the viral precursor Gag proteins. Despite this knowledge, little is known about the mechanisms by which MA facilitates the virion incorporation of Env proteins. To help elucidate this process, we examined the binding activities of an MA mutant that stabilizes MA trimers. We found that the mutant proteins organized similarly to WT proteins on membranes, and that mutant and WT proteins revealed only slight differences in their binding to RNAs or lipids. However, the mutant proteins showed better binding to Env CTs than the WT proteins, and CT binding correlated with MA trimerization. Our results suggest that multivalent binding of trimeric HIV-1 Env proteins to MA trimers contributes to the process of Env virion incorporation.

INTRODUCTION

The matrix (MA) domain of the human immunodeficiency type 1 (HIV-1) precursor Gag (Pr55Gag) protein serves several assembly-related functions. One role is to direct Pr55Gag proteins to assembly sites at cell plasma membranes (PMs) that are enriched for phosphatidylinositol-4,5-bisphosphate (PI[4,5]P2) and cholesterol (1–8). This function is accomplished in part by virtue of an amino-terminal myristate moiety and, in part, through direct binding to PI(4,5)P2 (1). Evidence indicates that the PI(4,5)P2-binding site on MA overlaps a binding site for RNA, which suggests a chaperone model in which MA-RNA binding protects the MA domain of Pr55Gag from associating with inappropriate intracellular membranes prior to delivery to the PI(4,5)P2-rich PM assembly sites (9–16). In some experimental systems, it has been shown that Pr55Gag proteins with MA deletions (ΔMA) or swaps of MA with other membrane binding domains are capable of directing the assembly of conditionally infectious viruses, but in these cases, assembly and replication efficiencies have tended to be compromised (17–21).

A second role of HIV-1 MA is to facilitate the incorporation of envelope (Env) proteins into virions (22–25). How MA accomplishes this is complicated. One confounding issue is that some cell surface proteins can be assembled into HIV-1 particles in what appears to be a passive fashion (26). Indeed, this is how glycoproteins from other viruses can pseudotype with the cores of HIV-1 (17, 27–31). Presumably, PM proteins that localize to HIV-1 assembly sites can be incorporated into viruses as innocent bystanders, and interestingly, several heterologous viral glycoproteins are assembled efficiently into both wild-type (WT) and ΔMA HIV-1 viruses (17, 31). The same is not true for the WT HIV-1 Env protein (17, 18). However, HIV-1 Env proteins that carry deletions of the long Env cytoplasmic tail (CT) can be assembled efficiently into either WT or ΔMA HIV-1 particles (18), at least in cell types that permit the efficient cell surface localization of ΔCT Env proteins (32, 33).

The above-described results suggest that ΔCT Env proteins assemble passively into HIV-1 virions, whereas WT Env proteins require a direct or indirect interaction with MA for virion incorporation (17, 18, 22–25). Because of the unusual length of the HIV-1 Env CT, one can speculate that WT Env trimers are hindered from assembling into lattices organized by PrGag proteins unless they are guided by an interaction with MA. Data supporting such a hypothesis are available. HIV-1 MA has a propensity to form trimers (22–25, 34–36), and mutations of residues at the spokes of those trimers (such as 12LE, 16EK, 30LE, 34VE, and 98EV) (Fig. 1A) have been shown to impair WT HIV-1 Env virion incorporation (22–25, 37). MA also has been shown to assemble in vitro in two forms of hexamer lattices, and in both lattices, residues 12, 16, 30, 34, and 98 point toward hexamer centers (Fig. 1B and C), suggesting that Env CT tails can be coordinated in these locations (38, 39).

FIG 1.

Alternative MA assemblies. (A) Atomic model of the HIV-1 MA trimer (PDB entry 1HIW) is shown with glutamine 62 (Q62) at the trimer interface in red, and residues that affect Env protein incorporation (12L, 16E, 30L, 34V, and 98E) are in purple. (B) Observed (56) organization of MA proteins on PI(4,5)P2-containing membranes as a hexamer of trimers, with a large protein-free hexamer hole in green. (C) Observed (55) organization of MA proteins on membranes with no PI(4,5)P2, where trimer contacts are not evident, and hexamer holes (green) are smaller in diameter. In both panels B and C, Q62 residue locations are colored red, and other residues that affect Env protein incorporation are depicted in dark blue/purple.

Unexpectedly, recent investigations have revealed that another mutation, 62QR, located at the MA trimer interface (Fig. 1A), impacts WT HIV-1 Env incorporation (22–25). Cell culture studies showed that the 62QR mutation suppressed the Env incorporation defects of the 12LE, 16EK, 30LE, 34VE, and 98EV MA mutations and of an Env CT mutation that has the same phenotype (22–25). How is the 62QR suppression mediated? One possibility (22–25) is that 62QR stabilizes trimers such that MA lattices which form large hexamer holes (Fig. 1B) are favored over those that feature small hexamer holes (Fig. 1C). Such a model is parsimonious in that it does not require specific MA-CT interactions, but it does not directly address why MA lattices have not been observed in immature HIV-1 virions, why WT HIV-1 Env fails to incorporate into ΔMA virions, or why mutations at the spokes of MA trimers affect Env incorporation (22–25, 37, 40–44).

Due to the importance of the 62QR mutant, we have examined its activities in vitro in comparison with WT MA. Our results show that 62QR MA proteins had increased capacities to trimerize relative to their WT counterparts, but that the WT and 62QR proteins appeared to organize similarly on membranes. The two variants showed slight differences in binding to PI(4,5)P2-containing membranes and RNA ligands, but 62QR proteins bound significantly better than WT proteins to the HIV-1 Env CT. Our results suggest a model in which multivalent binding of trimeric HIV-1 Env proteins to MA trimers contributes to the process of Env virion incorporation.

MATERIALS AND METHODS

Proteins for analysis.

WT myristoylated and unmyristoylated [designated Myr(−)] MA C-terminally His-tagged proteins were expressed in Escherichia coli strain BL21(DE3)/pLysS (Novagen) from pET-11a-based vectors and purified and analyzed as described previously (9, 11, 13, 38, 39). The WT MAGFP protein was purified as described above. It was constructed from a pET-11a-based MACA expression vector (38, 39) by insertion of the coding region for the green fluorescent protein (GFP) from pEGFP-c1 (Clontech) near the C terminus of MA, as in the previously reported vector HIVgpt-MAGFP (20). The MA-GFP juncture residues are AALALPVATM-GFP-KSGLRSAD, where the first and last residues represent HIV-1 MA residues 119 and 120. The 62QR MA and MAGFP variants and the 12LE and 12LE/62QR variants were transferred from the NL4-3 constructs (22–25, 37, 45) and were purified similarly to their WT counterparts. The Schistosoma japonica glutathione S-transferase (GST) and GST-CT proteins also were expressed and purified from Escherichia coli strain BL21(DE3)/pLysS. GST was expressed from pGEX4T3 (GE Healthcare), and GST-CT was expressed from pGEX4T3-CT, where the cytoplasmic tail coding region of the HXB2 Env protein was amplified with BglII ends by PCR and inserted in frame C-terminally to the GST coding region of pGEX4T3 at the pGEX4T3 BamHI site. The juncture sequence is GGA TCT GTT AAC AGA GTT-ENV CT-ATT TTG CTT TAA AGA TCG, where the underlined codons represent the HXB2 Env codons 705 to 708 and 854 to 856. The GST and GST-CT proteins were purified under nondenaturing conditions from 400-ml cultures (optical density at 600 nm [OD₆₀₀] of 0.6 to 0.9) that were induced for 4 h at 25°C via the addition of 0.5 mM isopropyl β-d-1-thiogalactopyranoside (IPTG). Induced cultures were pelleted and resuspended in 5 ml of GST sonication buffer (50 mM sodium phosphate [pH 7.8], 300 mM NaCl) containing protease inhibitors (15 μg/ml aprotinin, 1.5 mM phenyl methane-sulfonyl fluoride [PMSF], 36 μg/ml chicken trypsin inhibitor, 36 μg/ml soybean trypsin inhibitor, 6 μg/ml benzamidine, 15 μg/ml leupeptin, 7.5 μg/ml pepstatin A) plus 12.5 μg/ml DNase 1 (Roche). Suspended bacteria were incubated on ice for 30 min, subjected to two rounds of French press lysis, clarified by a 10-min centrifugation at 4°C and 7,500 × g, and incubated with rocking for 1 to 3 h at 4°C with 1 to 2 ml of reduced glutathione agarose resin (Sigma, Pierce). Samples and resin then were loaded onto columns, unbound material was collected, columns were washed twice with 10 ml sonication buffer plus inhibitors, and five 1-ml elutions of elution buffer (50 mM Tris, pH 8.0, 20 mM reduced glutathione) were eluted. Elutions were desalted by dialysis in two 3-h steps at 4°C with 4 liters of storage buffer (50 mM sodium phosphate [pH 7.0], 150 mM NaCl, 1 mM β-mercaptoethanol [BME]) and analyzed for purity by SDS-PAGE plus staining and immunoblotting with anti-GST primary antibody (SC-138; at 1:1,000 dilution; Santa Cruz Biotech). GST purifications yielded protein stocks that were >90% pure and were stored frozen (−80°C) in aliquots at 1 to 2 mg/ml. GST-CT purifications yielded samples that corresponded to 50% GST-CT and 50% processed GST and were stored frozen as described above.

RNA binding assays.

Fluorescence anisotropy (FA) binding assays were performed as described previously (11, 13). Briefly, 10 nM fluorescein isothiocyanate (FITC)-labeled Sel25 RNA oligomer (5′ FITC-GGACA GGAAU UAAUA GUAGC UGUCC-3′; Invitrogen) samples were incubated in 25 mM sodium phosphate (pH 6.0), 50 mM NaCl with successive additions of 30 μM protein to achieve final concentrations of 0 to 5,120 nM protein. Incubations were performed in 12- by 75-mm disposable borosilicate glass tubes, and measurements were obtained on a PanVera Beacon 2000 fluorescence polarizer (Invitrogen) at an excitation wavelength of 490 nm. Readings were made in triplicate at room temperature, polarization values were calculated from emitted light intensities according to the ratio of parallel minus perpendicular to parallel plus perpendicular (18–20), and binding isotherms were fitted using Prism.

Electrophoretic mobility shift assays (EMSAs) were performed using the chemically synthesized untagged Sel25 RNA (Invitrogen) as described previously (11, 13). EMSA binding reactions were performed for 30 min at 4°C in 25 mM sodium phosphate (pH 6.0), 50 mM NaCl, and employed 15 μM Sel25 plus 0 to 37.5 μM MA. Bound and free RNAs were separated by electrophoresis on 12% native polyacrylamide gels in 0.5× Tris-borate buffer (44.5 mM Tris base, 44.5 mM boric acid [pH 8.0]) and visualized by staining with Stains-all (Sigma) (11, 13).

Membrane binding.

Liposome bead binding assays were performed as described previously (9, 13) with WT or 62QR MA beads and fluorescently tagged liposomes. MA-coated beads were prepared by following our standard protocol (9, 13), and the estimated bead-bound MA used in each assay was 300 nM. 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-phospho-ethanolamine-N-(7-nitro-2-1,3benzoxadiazol-4-yl) (NBD-DOPE; Avanti), and brain PI(4,5)P2 Avanti) by sonication (9, 11, 13). PC liposomes were composed of 20% cholesterol, 79.8% PC, and 0.2% NBD-DOPE. PIP(4,5)P2-containing liposomes were composed of 20% cholesterol, 69.8% PC, 0.2% NBD-DOPE, and 10% PI(4,5)P2. Binding reactions were performed for 16 to 18 h at 4°C with 60 μl MA beads plus 2 μl of 2 mg/ml liposomes in wash buffer (25 mM sodium phosphate [pH 7.8], 50 mM NaCl, 0.1 mg/ml bovine serum albumin [BSA]). After binding incubations, beads were pelleted, washed twice with 0.3 ml wash buffer, and suspended in 50 μl of wash buffer on ice. For viewing, 10-μl samples were applied to microscope slides, covered with 22-mm by 22-mm coverslips, and imaged on a Zeiss Axioplan fluorescence microscope using a 20× (LDA-Plan) objective with Zeiss filter set 10 (excitation band pass, 450 to 490 nm; beam splitter Fourier transform, 510 nm; emission band pass, 515 to 565 nm). Bead brightness values were calculated and averaged as described previously (9, 13) and were normalized to values obtained with WT MA-coated bead plus PI(4,5)P2-containing liposome incubations.

CT binding assays.

Beads for pulldown assays were prepared by incubation of 100 μl of packed glutathione agarose resin (Sigma, Pierce) in 0.25 ml of assay buffer (25 mM sodium phosphate [pH 7.0], 150 mM NaCl, 10% glycerol, 0.1 mg/ml BSA) plus 5 μg GST or GST-CT at 4°C for 1 h. Beads then were pelleted, washed with assay buffer, and resuspended in assay buffer. Binding reactions included 33 μl resin in a total of 0.1 ml assay buffer containing 1.5 μM MA or CA protein, and they were performed at 4°C for 16 to 18 h on a rotator. After binding reactions, beads were pelleted and unbound (total) samples were collected. Beads were then washed three times for 1 min each with 0.5 ml assay buffer (minus BSA), and bound material was eluted in 15-min incubations with elution buffer (50 mM Tris [pH 8.0], 20 mM glutathione). Total and bound samples were fractionated by SDS-PAGE and immunoblotted (17, 20, 31) with primary antibodies to MA (01848170; 1:2,000; Capricorn), CA (Hy183; 1:30 from hybridoma media), and anti-GST primary antibody (SC-138; 1:1,000 dilution; Santa Cruz Biotech) for protein detection. Total and bound band levels were quantified by densitometric scanning and analysis with Image J software (46) and normalized as percent bound levels. Fluorescent protein GST binding assays were performed with GST and GST-CT beads prepared as described above in incubations with 1.5 μM WT or 62QR MAGFP protein. Binding reactions and washes were performed as described above. After washing, beads with bound MAGFP proteins were imaged and analyzed as described for membrane binding assays.

Cross-link analysis.

Protein cross-linking was performed by UV light exposure, which has been reported to cross-link proteins at tyrosine, cysteine, and histidine residues (47–49). For UV cross-linking, 20-μl drops of 50 μM protein in 50 mM sodium phosphate (pH 7.0), 150 mM NaCl, 1 mM BME were placed in the center rows of 96-well plates (vinyl; Costar) at 25°C. UV cross-linking reactions were performed with a UV Stratalinker (Stratagene) using the auto-cross-link parameter (1,200 μJ [×100]) for 0, 1, 3, or 10 min. At each time point, samples were removed from the plate and processed for electrophoresis and immunoblotting (17, 20, 31). Monomers, dimers, and trimers were determined by comparison of mobilities with known size standards (161-0374; Bio-Rad) and were quantified by densitometric scanning and analysis with Image J software (46). For each 10-min UV reaction, the amounts of monomer and trimer signals were normalized to the monomer signals to obtain estimates of trimer formation.

EM.

Assembly of MA proteins on lipid monolayers, electron microscopy (EM), and image analysis proceeded by following methods described previously (38, 39). WT and 62QR MA proteins (250 μM) were assembled in 25 mM sodium phosphate (pH 7.8), 5 mM sodium acetate, 300 mM NaCl, 20% glycerol, 1.25 mM BME beneath lipid monolayers of 60% egg PC (Avanti), 20% cholesterol (Sigma), and 20% brain PI(4,5)P2 (Avanti). After incubations, samples were lifted onto lacey EM grids (Ted Pella), rinsed for 30 s on drops of distilled water, stained for 45 to 60 s with 1.33% (wt/vol) uranyl acetate, wicked, and dried. Samples were imaged under low-dose conditions at 120 keV on the OHSU FEI Tecnai transmission electron microscope (TEM) equipped with an Eagle 4 megapixel charge-coupled device (CCD) at 4.37 Å/pixel and defocus values of −200 to −1,500 nm to give a contrast transfer function (CTF) first zeroes of beyond 2 nm. Ordered areas in images were boxed, Fourier transformed, indexed, and unbent using the 2dx image processing package (50). Resulting amplitudes and phases (aph) files were subjected to space group analysis (38, 39), yielding best fits based of p6 symmetry for both WT and 62QR crystals, with real space dimensions for both proteins of a = b = 87.6 Å and γ = 120°. Merged WT and 62QR aph files were complete to 25 Å, were filtered to that resolution limit, and were back-transformed assuming p6 symmetry to yield Fourier filtered images in which proteins appear white and protein-free areas are dark.

RESULTS

Matrix protein oligomerization.

Because the HIV-1 MA 62QR mutation at the matrix trimer interface strongly suppressed the defects imposed by mutations located at MA trimer spokes (22–25) (Fig. 1), it was of interest to compare the activities of the 62QR protein relative to the WT protein. Initially, we compared the sedimentation profiles of purified 62QR and WT myristoylated MA (here referred to as MA) proteins. Our results indicated that 50 μM WT MA (in 50 mM sodium phosphate, pH 6.0, 100 mM NaCl, 2 mM Tris-2-carboxyethyl phosphine) sedimented with an apparent molecular mass of 21,606 Da, slightly higher than the predicted 16,812 Da, whereas the 62QR mutant sedimented with an apparent mass of 33,532 Da (data not shown). These results were consistent with the notion that the proteins oligomerize as dimers and/or trimers, but we were unable to fit data clearly into monomer-dimer, monomer-trimer, or monomer-dimer-trimer equilibrium models.

As an alternative approach, we subjected WT and 62QR MA proteins to UV cross-link analysis. To do so, purified proteins were mock or UV treated, separated by electrophoresis, and subjected to immunoblot detection. As shown in Fig. 2 (upper left), while UV treatment increased the proportion of WT MA dimers, levels of trimers were low. In contrast, with 62QR MA, UV treatment yielded both dimers and trimers (Fig. 2, upper right). Profile plots of samples receiving UV treatment for 10 min clearly showed a significant trimer band for 62QR MA, but the band was less clear for WT MA (Fig. 2, lower left). Quantitation of trimers versus monomers after 10 min of UV treatment in 10 independent experiments indicated that WT MA protein trimers were observed at 2% monomer levels, while 62QR trimers were observed at 11% monomer levels (Fig. 2, bottom right). These results indicate that the 62QR trimer interface mutation stabilized MA trimers.

FIG 2.

Matrix protein trimerization. Purified WT or 62QR MA proteins at 50 μM concentrations were UV cross-linked for 0, 1, 3, or 10 min (left to right) as described in Materials and Methods, separated by electrophoresis, and immunoblotted with an anti-MA antibody. Monomer (1), dimer (2), and trimer (3) sizes were determined by the mobilities of size standards run in parallel. The bottom left shows the band intensity profile plots of the 10-min-time-point lanes from the immunoblot images above, with the WT plot as a dotted line and the 62QR plot as a solid line; note the intensity difference of the trimer bands. The bottom right shows the percentage of trimers (relative to monomers) from the 10-min time point averaged from 10 separate experiments, with standard deviations as indicated. At the bottom, the amounts of monomers (light gray), dimers (dark gray), and trimers (black) from the 10-min time point were quantified as a percentage of the MA monomer signals.

As noted in the Introduction, one hypothesis as to how 62QR proteins suppress Env incorporation defects posits that membrane-bound lattices of 62QR proteins better accommodate Env CT tails than do WT lattices as a consequence of MA trimerization (Fig. 1B and C). We examined this possibility via analysis of how WT and 62QR proteins organize on membranes. For this, WT and 62QR MA proteins were assembled on membranes containing cholesterol and PI(4,5)P2 and analyzed by transmission EM. As observed previously (39), the MA proteins assembled into small ordered two-dimensional (2D) crystalline patches that were amenable to image analysis. Consequently, ordered regions were boxed and Fourier transformed, and lattice reflections were indexed to give unit cell parameters. Interestingly, both WT and 62QR MA proteins yielded unit cells with hexagonal (p6) symmetry and identical real space unit cell sizes (a = b = 87.6 Å, γ = 120°). To visualize the MA-membrane lattices, Fourier transforms were filtered and back-transformed, and assemblies are shown in Fig. 3, viewed from the membrane side with protein regions in white. Of particular note is that the assemblies appear nearly identical, with trimer units surrounding equally sized hexamer holes. These results do not lend support to the hypothesis that WT and 62QR proteins organize on membranes with differently sized hexamer holes (22–25), but it is possible that potential differences were masked by the 25-Å resolution of our reconstructions.

FIG 3.

Membrane-bound organization of MA proteins. WT and 62QR MA proteins were assembled onto lipid monolayers containing PI(4,5)P2, lifted onto EM grids, stained, and imaged. Fourier transforms of ordered regions in micrographs were indexed, Fourier filtered, and back transformed as described in Materials and Methods. In each case, assemblies are shown perpendicular to the membrane, protein regions are in white, protein-free regions are dark, and real space unit cell dimensions were a = b = 87.6 Å and γ = 120°. Note that at this resolution (25 Å), matrix proteins appear as interconnected trimers (one trimer indicated in each panel) around hexamer holes, and that the two assemblies are nearly indistinguishable.

RNA and membrane binding.

The WT HIV-1 MA protein binds to RNAs and PI(4,5)P2-containing membranes at overlapping binding sites (11, 13). To examine whether the 62QR mutation affects this binding, we initially tested myristoylated and myristoylation-minus [Myr(−)MA] WT and 62QR proteins for binding to the previously identified (11, 13, 51) Sel25 RNA ligand. Fluorescence anisotropy (FA) assays were performed with fluorescently tagged Sel25 ligand and increasing concentrations of WT and 62QR MA and Myr(−)MA (Fig. 4A). For both the MA and Myr(−)MA proteins, we observed slightly higher affinities of the 62QR than the WT variants to Sel25: for the myristoylated proteins the calculated binding constants were 496 nM (WT) and 180 nM (62QR), while the unmyristoylated protein constants were 334 nM (WT) and 130 nM (62QR). These trends also were observed in gel shift (EMSA) assays, in which bound (B) and free (F) Sel25 RNAs were detected after incubation with protein, and electrophoretic separation (Fig. 4B). Interestingly, we did not see significant differences in bound RNA mobilities between the WT and 62QR proteins, consistent with the previously observed 1:1 binding of WT MA and Sel15 RNA (11) and the low levels of observed oligomers for both WT and 62QR proteins in our cross-link analysis (Fig. 2).

FIG 4.

Matrix binding to 25mer RNA ligands. Binding of WT or 62QR MA (Myr+) or unmyristoylated MA (Myr−) to the Sel25 RNA ligand was tracked by fluorescence anisotropy (A) or electrophoretic mobility shift assay (B). In panel A, 10 nM fluorescent Sel25 RNA samples were incubated with increasing concentrations of protein, as indicated. Fluorescence millipolarization values were measured and plotted as percentages of the maximum polarization values obtained for each sample. The curves represent averages from triplicate measurements, and K_ds (dissociation constants) were calculated from half-maximal polarization values. In panel B, 15 μM samples of Sel25 were incubated with 0, 7.5, 15, 22.5, 30, and 37.5 μM protein, after which free (F) and bound (B) RNAs were separated by electrophoresis and stained. At the bottom of each panel, densitometrically determined percent bound (%B) values are given. Note that M indicates mock lanes containing bromphenol blue dye that was excluded from RNA-containing lanes, and that results are representative of three independent experiments.

As for membrane binding to WT and 62QR MA, we addressed this using liposome bead-binding assays, as we have in the past (9, 11). Here, fluorescent PC-cholesterol liposomes, prepared with or without PI(4,5)P2, were assayed for binding to beads coated with WT or 62QR MA proteins. As shown in Fig. 6 (left), the WT and 62QR beads both bound PI(4,5)P2-containing liposomes at much higher levels than the PC-cholesterol liposomes. Thus, selectivity for binding to PI(4,5)P2 membranes was retained for the 62QR proteins. However, in multiple measurements, we did observe that PI(4,5)P2 liposomes bound 30 to 40% less well to the 62QR than the WT MA beads (Fig. 5, right). Because MA residue 62 maps away from the PI(4,5)P2 binding site (3, 14, 52), this binding difference would appear to be due to an as-yet undetermined effect.

FIG 6.

MA binding to the HIV-1 Env CT. The indicated WT MA, 62QR MA, or CA proteins at 1.5 μM were incubated with beads coated with glutathione S-transferase (GST) (CT−) or GST-CT (CT+), after which total unbound proteins were collected (total) and beads were washed and eluted to collect bound proteins. Total and bound samples were electrophoretically separated and immunoblotted in parallel for protein detection. Immunoblot bands were quantified densitometrically, and binding levels are expressed as percentages of bound versus total protein. Calculated percentages derive from five (62QR MA), four (WT MA), or three (CA) independent experiments.

FIG 5.

Membrane binding of WT and 62QR MA. Liposome binding assays were performed with beads coated with WT or 62QR (62) MA proteins and fluorescently labeled liposomes containing PC plus cholesterol (PC) or PC plus cholesterol plus 10% PI(4,5)P2 (PIP). Examples of liposome-bound beads are shown on the left. Liposome binding levels were determined by measuring bead fluorescence brightness levels from at least 17 beads of each type imaged in two separate experiments. Values were normalized to the overall average of PI(4,5)P2-containing liposomes bound to WT MA beads.

Matrix-Env CT binding.

Since the 62QR mutation was shown to affect Env incorporation into virions in a CT-dependent fashion (22–25), it was of interest to analyze WT and 62QR MA-CT binding in vitro. Although different previous investigations have resulted in some uncertainty concerning the nature of MA-CT interactions (22–25, 30–33, 52, 53), several studies have reported MA-CT binding (52, 53). Previous work indicated that the HIV-1 Env CT could be purified from bacteria as a GST fusion protein (GST-CT) and, when prebound to glutathione resin, could be used successfully in MA binding assays (52). Because of this, we expressed and purified GST-CT (and GST) from bacteria under nondenaturing conditions. In this regard, it is pertinent to note that our efforts to purify the CT as a His-tagged or maltose binding protein-tagged protein under nondenaturing conditions were unsuccessful (data not shown).

For binding reactions, GST- or GST-CT-coated beads were incubated with WT or 62QR MA proteins (or CA as a control), after which unbound (total) and bound fractions were isolated, separated by electrophoresis, and detected by immunoblotting. As shown in Fig. 6 (top), none of the proteins bound well to GST beads, and the CA and WT MA proteins also bound poorly to GST-CT beads. In contrast, the 62QR MA proteins clearly bound to the GST-CT beads. From multiple experiments, we found that approximately 30% of the input 62QR MA protein bound to the GST-CT beads (Fig. 6, bottom). In contrast, our results indicated that GST-CT binding levels were close to background levels for the WT MA and CA proteins. Thus, GST-CT binding levels appeared to parallel trimerization levels observed in cross-linking experiments (Fig. 2) for the MA proteins.

As a second approach to assay MA-CT binding, we bound WT or 62QR MAGFP proteins to GST or GST-CT beads and tracked binding via fluorescent detection of GFP. Our results (Fig. 7) were consistent with results shown in Fig. 6. As shown in Fig. 7 (left), while neither MAGFP protein bound well to GST beads, the WT variant bound slightly to GST-CT beads, and the 62QR variant bound considerably better. Quantitation of fluorescence levels from multiple beads indicated that 62QR binding to GST-CT was higher than WT binding (Fig. 7, right).

FIG 7.

MAGFP binding to the HIV-1 Env CT. WT or 62QR MAGFP proteins (30 μM) were incubated with beads coated with glutathione S-transferase (GST) or GST-CT, after which beads were washed and imaged for fluorescence detection. Examples of WT and 62QR protein binding to GST-CT and GST beads are shown on the left. On the right, MAGFP binding levels were quantified by measuring fluorescence brightness values from 8 to 10 beads of each class. Values were normalized to the bead brightness average of WT MAGFP bound to GST-CT beads.

Because the 62QR mutation was selected as one that suppressed the Env incorporation defects of MA variants such as the 12LE mutation (22–25), it was of interest to examine how 62QR affects the oligomerization and CT binding capabilities of 12LE mutant proteins. To do so, 12LE and 12LE/62QR MA proteins were purified and subjected to UV cross-linking and CT binding analysis. As shown in Fig. 8 (top), while 12LE dimers were detected, trimers were present at only 2% monomer levels, similar to that observed for the WT MA protein (Fig. 2). In contrast, UV treatment clearly yielded 12LE/62QR trimers (top center), and trimers were observed at 12% monomer levels (top right), similar to the 62QR protein (Fig. 2). These results indicate that the 62QR mutation fosters MA trimerization on both WT and 12LE backgrounds. Importantly, the 62QR mutation also conferred an increased capacity for CT binding to 12LE MA proteins. As shown in Fig. 8 (bottom left), 12LE binding to either GST or GST-CT resin appeared minimal. In contrast, while hardly any 12LE/62QR MA bound to GST resin, approximately 20% of the protein bound to GST-CT, two thirds of the level observed for the 62QR MA protein itself (Fig. 6). These results indicate that the 62QR mutation allows 12LE MA proteins to bind to the HIV-1 Env CT.

FIG 8.

Analysis of 12LE and 12LE/62QR MA proteins. The 12LE and 12LE/62QR proteins were analyzed for trimerization (top) as described in the legend to Fig. 2 and for Env CT binding as described in the legend to Fig. 6. In the top left, 50 μM concentrations of the indicated proteins were UV cross-linked for 0, 1, 3, or 10 min (left to right) and detected as monomers (1), dimers (2), or trimers (3) after electrophoresis and immunoblotting. At the top right, the percentages of trimers relative to monomers derive from eight independent experiments. The bottom left shows immunoblots of bound and total unbound 12LE and 12LE/62QR proteins from 1.5 μM MA incubations with GST-CT (CT+) or GST (CT−) beads. At the bottom right, percentages of bound versus total protein were derived from four independent experiments.

DISCUSSION

The 62QR mutation, located at the hub of MA trimer structures (Fig. 1), is a unique mutation in that it suppresses the envelope incorporation defects of an Env CT mutation and of multiple MA mutations (including 12L, 16E, 30L, 34V, and 98E) that locate toward the ends of MA trimer spokes. Our data favor a model (20–22) in which the 62QR change promotes MA trimerization (Fig. 2). The mutation also slightly increased the affinity of MA for the small Sel25 RNA ligand (Fig. 4), but our gel shift data (Fig. 4) are consistent with our previous observation (11) of 1:1 MA-RNA binding for both the WT and 62QR variants. Thus, we hypothesize that the addition of a basic residue near other residues that have been implicated in MA-RNA binding (11) accounts for the observed differences in WT and 62QR Sel25 binding. We also noted that the WT MA protein bound slightly better to PI(4,5)P2-containing membranes than did the 62QR protein (Fig. 5). It will be of interest to extend these observations in the future.

Given that the 62QR mutation impacted CT-dependent Env incorporation into virions in cell culture (22–25), we tested its effects relative to the WT protein in in vitro MA-CT binding assays. To do so, we adapted the previously described MA-CT binding assay (52) and found that 62QR MA proteins bound better to GST-CT beads than their WT counterparts (Fig. 6 and 7). Additionally, we observed that the 62QR mutation in the context of the 12LE Env incorporation mutation conferred the ability to bind GST-CT beads (Fig. 8). Of particular interest was that GST-CT binding of MA proteins correlated with their abilities to trimerize (Fig. 2 and 8). What might explain this behavior? One pertinent consideration is the oligomerization status of the glutathione resin-bound GST-CT. Because the Schistosoma japonicum GST expressed from pGEX plasmids preferentially forms dimers (54), it is reasonable to expect that resin-bound GST-CT packs at least pairs of CTs in close proximity. However, the packing density of glutathione agarose (>40 mg GST per ml of settled resin) is such that GST-CT monomers may be 8 to 9 nm apart from each other on average, assuming that the entire volume of each bead can bind GST-CT ligand. This GST-CT packing density permits at least two possibilities. One possibility is that some percentage of CTs assemble into trimer structures, and these trimers bind to MA trimers in a 1:1 stoichiometry. An alternative is that the spokes of MA trimers (which are 6 to 7 nm apart) bind independently to separate CTs, and MA-CT binding avidity depends on relatively weak MA spoke-CT interactions and MA trimerization. In this case, we envision that MA trimers may align beneath Env trimers at membranes, as shown in Fig. 9, and that separate spokes of an MA trimer may bind to the separate CTs of Env trimers at three different hexameric lattice cage holes (Fig. 1).

FIG 9.

Overlay of Env and MA trimers. The HIV-1 Env SU plus TM (gp120+gp41; left) and observed TM residues (right) from PDB entry 3J5M are overlaid onto MA trimer structures (PDB entry 1HIW) as backbone and side-chain traces or space-filling models. The CT structures are not known and not depicted but are envisioned in this overlay to bind to separate members of the MA. In such a model, CT tails of a single Env trimer bind to separate monomers of the MA trimer at separate hexamer holes of the MA lattices depicted in Fig. 1.

It is worthwhile to note that the modes of MA-CT binding proposed above do not argue against the roles of cellular factors in fostering Env protein transport to assembly sites or in facilitating virion incorporation (32, 33). The trimer binding models also do not contradict the possibility that hexamer hole sizes in MA lattices contribute to the regulation of Env assembly into virus particles (22–25). Although we did not observe significant differences between membrane-bound WT and 62QR MA lattices (Fig. 3), the resolution of our reconstructions was such that potentially important distinctions might be masked. In this regard, whereas MA lattices have not as yet been observed in EM reconstructions of HIV-1 virions (37–41), observations concerning viral pseudotyping, Env fluidity in virus particles, and CT cleavage support models that infer close interactions between Env CT tails and MA lattices (22–25, 52, 53, 55, 56). We believe that further study of 62QR and other MA mutations will help elucidate the complex interplay between the HIV-1 envelope and matrix proteins.