Formation of Virus Assembly Intermediate Complexes in the Cytoplasm by Wild-Type and Assembly-Defective Mutant Human Immunodeficiency Virus Type 1 and Their Association with Membranes

Abstract

We have previously identified two distinct forms of putative viral assembly intermediate complexes, a detergent-resistant complex (DRC) and a detergent-sensitive complex (DSC), in human immunodeficiency virus type 1 (HIV-1)-infected CD4⁺ T cells (Y. M. Lee and X. F. Yu, Virology 243:78–93, 1998). In the present study, the intracellular localization of these two viral assembly intermediate complexes was investigated by use of a newly developed method of subcellular fractionation. In wild-type HIV-1-infected H9 cells, the DRC fractionated with the soluble cytoplasmic fraction, whereas the DSC was associated with the membrane fraction. The DRC was also detected in the cytoplasmic fraction in H9 cells expressing HIV-1 Myr− mutant Gag. However, little of the unmyristylated Gag and Gag-Pol proteins was found in the membrane fraction. Furthermore, HIV-1 Gag proteins synthesized in vitro in a rabbit reticulocyte lysate system in the absence of exogenous lipid membrane were able to assemble into a viral Gag complex similar to that of the DRC identified in infected H9 cells. The density of the viral Gag complex was not altered by treatment with the nonionic detergent Triton X-100, suggesting a lack of association of this complex with endogenous lipid. Formation of the DRC was not significantly affected by mutations in assembly domains M and L of the Gag protein but was drastically inhibited by a mutation in the assembly I domain. Purified DRC could be disrupted by high-salt treatment, suggesting electrostatic interactions are important for stabilizing the DRC. The Gag precursor proteins in the DRC were more sensitive to trypsin digestion than those in the DSC. These findings suggest that HIV-1 Gag and Gag-Pol precursors assemble into DRC in the cytoplasm, a process which requires the protein-protein interaction domain (I) in NCp7; subsequently, the DRC is transported to the plasma membrane through a process mediated by the M domain of the matrix protein. It appears that during this process, a conformational change might occur in the DRC either before or after its association with the plasma membrane, and this change is followed by the detection of virus budding structure at the plasma membrane.

The assembly of retroviruses primarily follows two distinct pathways (15, 35, 40). In type B and type D retroviruses, which include the mouse mammary tumor virus and Mason-Pfizer monkey virus (M-PMV), respectively, the newly synthesized Gag and Gag-Pol precursors are transported to a specific intracytoplasmic site at which electron-dense immature viral capsids are formed (24–27). The preformed viral capsids are subsequently targeted through an energy-dependent mechanism (37) to the plasma membrane, wrapped by cellular lipid membrane containing viral Env glycoproteins during budding, and released as extracellular viral particles (24–27).

In the case of type C retroviruses and lentiviruses, including HIV-1, crescent-shaped budding intermediates at the plasma membrane are the first viral structures that are visualized by electron microscopy before the appearance of spherical extracellular particles (15, 35, 40). The steps that precede the formation of electron-dense budding structures at the plasma membrane are therefore not well defined. It is possible that all the viral components, such as Gag, Gag-Pol, and the viral RNA genome, are individually targeted from the cytoplasm to the plasma membrane, from which virus assembly and budding occur simultaneously. Alternatively, some of the viral proteins, including the nascent Gag and Gag-Pol precursors, could form an electron-transparent complex in the cytoplasm that is subsequently targeted to the site of virus budding.

It is conceivable that these two distinct retroviral assembly pathways (the type B/D and type C/lentivirus) share many similar features, since mutations in the Gag molecule can change the assembly pathway from one to the other. The MA domain is likely a major determinant that distinguishes between the two assembly pathways. A single amino acid substitution in the MA domain of M-PMV Gag can convert a type D retrovirus to a type C-like morphogenesis (27). A large deletion in the matrix domain of HIV-1 Gag results in the formation of electron-dense immature viral capsid structures in the cytoplasm of infected cells (32), and formation of HIV-1 immature viral capsids in the absence of lipid membranes has been observed in vitro when the MA domain is deleted (5, 13).

As part of our study of HIV-1 morphogenesis, we have recently identified two distinct HIV-1 assembly intermediate complexes in HIV-1-infected CD4⁺ T cells, a detergent-resistant complex (DRC) and a detergent-sensitive complex (DSC) (18). Myristic acid modification of the HIV-1 Gag proteins, a signal required for plasma membrane binding and virus production, was not required for the formation of the DRC but was essential for the formation of the DSC (18). Furthermore, lipid membrane-disrupting detergent destroyed the DSC but not the DRC, suggesting that the formation of DSC requires stable association with the plasma membrane (18). However, the intracellular localization of the DRC and DSC was still undefined.

We have now examined the localization of the HIV-1 DRC and DSC by use of a subcellular fractionation method that has allowed us to separate large pelletable complexes in the cytoplasm from the membrane fractions. We found that the DRC in the wild-type HIV-1-infected CD4⁺ T cells was fractionated into the cytoplasmic fraction, whereas the DSC was pelleted with the membrane fraction. Furthermore, the DRC formed by Myr− mutant Gag molecules was also fractionated into the soluble cytoplasmic fraction, but not in the pelleted membrane fraction. In vitro-synthesized Gag proteins assembled into a DRC-like complex without addition of lipid, and the density of this complex was not affected by treatment with the nonionic detergent Triton X-100. These results suggest that HIV-1 Gag and Gag-Pol precursors can assemble into a DRC in the cytoplasm and that DRC formation does not require interaction with the lipid membrane. Results of experiments involving mutations of the well-characterized assembly domains M, I, and L of HIV-1 Gag demonstrated that although the M and L domains were not essential for DRC formation, mutation of the I domain disrupted the formation of the DRC.

MATERIALS AND METHODS

Cells and sera.

Uninfected H9 cells, wild-type HIV-1-infected H9 cells, H9 cells expressing myristylation-negative HIV-1 Gag and Gag-Pol (Myr−/H9), protease mutant Gag and Gag-Pol (Pr−/H9), and a pol deletion mutant (ΔPol/H9) were established and maintained as previously described (17, 18). H9 cells expressing p6^gag-truncated and p6^gag-plus-NCp7-truncated forms of HIV-1 Gag proteins were also established as previously described (6). An HIV-1-positive human serum was obtained from an HIV-1-infected patient from Baltimore, Md. Sheep polyclonal anti-gp120, anti-gp41, and anti-CD4 antisera were obtained from the AIDS Research and Reference Reagent Program, National Institutes of Health, Bethesda, Md. Rabbit polyclonal heat shock protein 70 (HSP70) antiserum was purchased from Stressgen Corp. The alkaline phosphatase (AP)-conjugated goat anti-rabbit immunoglobulin G (IgG) was purchased from Sigma Immuno Research. AP-conjugated goat anti-mouse IgG and AP-conjugated rabbit anti-goat IgG antibodies were purchased from Jackson Immuno Research Laboratories, Inc.

Osmolysis and subcellular fractionation of CD4⁺ T cells.

H9 cells were harvested by centrifugation at 2,000 rpm in a Sorvall RT6000B centrifuge (Du Pont) for 10 min. Cells were washed twice in RPMI 1640, resuspended in hypotonic buffer (20 mM Tris-HCl [pH 7.8], with 10 mM KCl, 1 mM EDTA, 0.1% 2-mercaptoethanol, and 2 μg of aprotinin per ml), and incubated at 4°C for up to 5 h without agitation. During incubation, the cells were monitored by tryptophan blue exclusion under light microscope to determine the completeness of the lysis. After more than 95% of the cells were lysed, the soluble cytoplasmic fraction of the cells was separated from the pelleted membrane fraction by low-speed centrifugation at 1,500 × g for 30 min. The resulting supernatant (cytoplasmic fraction) was carefully separated from the pellet. The pellet was resuspended in the original volume of hypotonic buffer by tipping the bottom of the tube, poured into a tight-fitting Dounce homogenizer, homogenized on ice, and centrifuged at 1,500 × g for 10 min. This supernatant was then collected as the membrane fraction. Total proteins from equal portions of the cytoplasmic and membrane fractions were precipitated by trichloroacetic acid (TCA), separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), and analyzed by Immunoblotting with the various antibodies. To examine the localization of viral assembly intermediate complexes, the cytoplasmic fraction or the membrane fraction was loaded onto a sucrose equilibrium-density gradient and analyzed as previously described (18).

To compare the efficiency of DRC formation by wild-type, Myr-negative, p6^gag-truncated, and p6^gag-plus-NCp7-truncated HIV-1 Gag, we disrupted H9 cells expressing each of these different forms of Gag molecules by osmolysis as described above. The cytoplasmic fractions were centrifuged at 100,000 × g for 1 h to pellet the DRC complexes. The pellets and the supernatants were adjusted to equal volumes and analyzed by SDS-PAGE and immunoblotting. Cytoplasmic fractions from H9 cells expressing wild-type and p6^gag-plus-NCp7-truncated HIV-1 Gag were also analyzed by sucrose equilibrium-density gradient centrifugation and immunoblotting as previously described (18).

Formation of DRC in vitro by HIV-1 Gag proteins.

The HIV-1 Gag proteins were expressed by using plasmid pGEM11Zgp, which contains the full-length HIV-1 Gag coding region after the SP6 promoter, in an in vitro-coupled transcription-translation system with a rabbit reticulocyte lysate (TnT Translation System; Promega). As suggested by the manufacturer, reactions were carried out with 27.5 μl of rabbit reticulocyte lysate, 2 μl of reaction buffer, 1 μl of SP6 RNA polymerase, 2 μl of a 1 mM amino acid mixture, 1 μl of RNase inhibitor RNasin (40 U/μl), 1 μl of cut plasmid DNA (1 μg/μl), and 15.5 μl of deionized water. The reaction mixture was incubated at 30°C for 2 h. The reaction samples were then incubated in the presence or absence of 0.5% Triton X-100 for 10 min at room temperature and then subjected to discontinuous equilibrium density centrifugation as previously described (18). After the sucrose gradient centrifugation, each fraction was pelleted and analyzed by SDS-PAGE and immunoblotting as described previously (18).

Protease digestion.

Cells were harvested by low-speed centrifugation and washed twice in RPMI 1640 medium and then were resuspended in hypotonic buffer and incubated at 4°C for 20 min. The cells were homogenized on ice in a tight-fitting Dounce homogenizer, and nuclei and unbroken cells were removed by centrifugation at 2,000 rpm for 10 min in a Sorvall RT6000B centrifuge. The resulting postnuclear supernatants were divided into two 5-ml portions: one portion was incubated with 5 mg of trypsin per ml (Boehringer Mannheim) at 37°C for 30 min, and the trypsin was then inhibited by addition of an excess antitrypsin inhibitor (30 mg/ml) (Boehringer Mannheim). As a control, the other 5-ml portion was simultaneously incubated under identical conditions in the absence of trypsin. Both samples were subjected to 16 to 60% discontinuous sucrose equilibrium-density gradient centrifugation, and the resulting fractions were analyzed by SDS-PAGE and immunoblotting as previously described (18).

High-salt treatment.

HIV-1-infected H9 cells were lysed in phosphate-buffered saline (PBS) containing 1% Triton X-100 and subjected to 16 to 60% discontinuous sucrose equilibrium-density gradient centrifugation as previously described (18). Fractions 9 and 10 from a freshly centrifuged 16 to 60% discontinuous sucrose equilibrium-density gradient (3.8 ml total) were diluted in 6.2 ml of hypotonic buffer and divided into two equal portions. The two portions were incubated with or without 1 M NaCl for 10 min at room temperature and then layered onto the top of a 20 to 60% discontinuous sucrose equilibrium-density gradient prepared by layering 2.3 ml of each stock sucrose solution in TNE buffer (0.01 M Tris-HCl [pH 7.2], 0.1 M NaCl, 0.001 M EDTA) (from 60 to 24% in 4% increments) in a 36-ml ultracentrifuge tube and then layering 4.6 ml of 20% sucrose solution on the top of the gradient. The samples were ultracentrifuged, and fractions were collected and pelleted as previously described (18). The pelleted materials were analyzed by SDS-PAGE and immunoblotting.

By using an in vitro-coupled transcription-translation system, HIV-1 Gag proteins were expressed in a rabbit reticulocyte lysate as described above. After incubation, the reaction mixture was divided into two portions. NaCl (1 M) was added into one-half of the reaction mixture and incubated for 10 min at room temperature. As a control, the other portion was simultaneously incubated without 1 M NaCl. Each reaction mixture was subjected to discontinuous equilibrium-density centrifugation and analyzed by SDS-PAGE and immunoblotting.

Immunoblot analysis.

Samples were separated by SDS-PAGE (12% polyacrylamide) under reducing conditions and transferred simultaneously onto two nitrocellulose filters for 24 h as described previously (18). The filters were blocked by incubation in 3% nonfat dry milk in washing solution (0.2% Tween in PBS) at room temperature for 1 h. They were washed three times with washing solution and incubated with the primary antibody at room temperature for 1 h. The filters were then washed three times with washing solution and incubated with the appropriate AP-conjugated secondary antibody at room temperature for 1 h. The filters were washed three times with washing solution and once with PBS. The antibody-labeled proteins were visualized by reaction with 5-bromo-4-chloro-3-indolylphosphate and nitroblue tetrazolium substrate.

RESULTS

Development of a cellular fractionation method that separates large complexes in the cytosol from membranes.

In order to address the question of whether the DRC or DSC was associated with membrane, it was necessary to separate the cytosol from membrane compartments. The conventional method for separation of cytosol from membrane relies on cell lysis by homogenization followed by ultracentrifugation at 100,000 × g, which produces a membrane pellet (P100) and cytosol supernatant (S100). However, this conventional method cannot efficiently separate large cytoplasmic complexes from membranes, because they both are pelleted at 100,000 × g. Our previous studies (18) indicated that the DRC could also be pelleted at 100,000 × g. To solve this problem, we adapted a subcellular fractionation protocol that has been successfully used to separate the cytosol from the membrane fractions of erythrocytes (8, 33).

To separate the soluble cytosol fraction of CD4⁺ T cells from the pelletable membrane fraction, uninfected H9 or Myr−/H9 cells were harvested by low-speed centrifugation and osmotically lysed by incubation in hypotonic buffer without agitation, as described in Materials and Methods. This gentle osmolysis allows cells to burst open and release the cytosolic material. During incubation, the cells were monitored to assess the extent of the cell lysis (data not shown). After more than 95% of the cells were lysed, the soluble cytoplasmic fraction (including any putative protein complexes) was separated from the pelleted membranes, nucleus, and other cell debris by low-speed centrifugation (1,500 × g). The membrane fraction was then generated by homogenization followed by another low-speed centrifugation (1,500 × g) to separate it from the nucleus and other cell debris. Equal amounts of total proteins from the soluble cytoplasmic and membrane fractions were then precipitated by TCA and separated by SDS-PAGE.

To determine whether the membrane fraction was adequately separated from the soluble cytoplasm, we performed immunoblot analysis with antibodies recognizing either membrane proteins, such as the viral Env proteins gp160 and gp120, or soluble proteins, such as HSP70. Immunoblotting with the polyclonal anti-gp41 antiserum demonstrated that the uncleaved gp160 was largely fractionated with the pelletable membranes in the Myr−/H9 cells (Fig. 1A); in addition, gp160 and gp120 were predominantly found in the pelletable membrane fraction (Fig. 1B). As expected, no gp160 or gp120 was detected in the soluble cytoplasmic or pelletable membrane fractions of uninfected H9 cells (Fig. 1A and B). Immunoblotting with a polyclonal anti-HSP70 antiserum demonstrated that the HSP70 proteins were predominantly fractionated with the soluble cytoplasmic fraction of both uninfected H9 and Myr−/H9 cells (Fig. 1C). The separation of cytosol from membrane was also demonstrated by immunoblotting with a polyclonal anti-CD4 antiserum. CD4 was largely detected in the pelletable membrane fraction of the uninfected H9 cells (Fig. 1D). These results demonstrated that soluble cytoplasmic portions were sufficiently separated from the pelletable membrane fraction by our new subcellular fractionation procedure. Since the viral Env glycoprotein precursor gp160, synthesized in the endoplasmic reticulum, has been shown to be processed into its mature form (gp120 and gp41) in the Golgi apparatus and subsequently transported to the plasma membrane (7, 34, 38), it appears that the pelletable membrane fraction we obtained included both intracellular membranes and the plasma membrane.

FIG. 1.

Subcellular fractionation and detection of DRC in Myr−/H9 cells. Uninfected H9 and Myr−/H9 cells were osmolysed by incubation in hypotonic buffer, and the soluble cytoplasmic fraction (S) of the osmolysed cells was separated from the membrane fraction (P) as described in Materials and Methods. The total proteins in an equal volume of each fraction were precipitated by TCA and separated by SDS-PAGE. Immunoblot analysis with a polyclonal anti-gp41 antiserum (A) and a polyclonal anti-gp120 antiserum (B) was performed to determine the location of the viral Env glycoproteins. Likewise, immunoblotting with a polyclonal anti-HSP70 antiserum (C) and a polyclonal anti-CD4 antiserum (D) was performed to determine the location of HSP70 (indicated by arrowheads) or CD4, respectively. (E) Equal volumes of the S and P fractions were then subjected to the discontinuous sucrose equilibrium-density gradient as previously described (18). After centrifugation, each fraction was collected and analyzed as previously described (18). Viral proteins were visualized by immunoblotting with an HIV-1-positive human serum. The positions of the viral proteins Pr55^gag and Pr160^gag-pol are indicated on the right, and molecular mass markers (kilodaltons) are indicated on the left.

The DRC in Myr−/H9 cells was fractionated with the soluble cytoplasmic fraction, not with the pelletable membrane fraction.

Using our new subcellular fractionation method, we first addressed the question of whether the DRC identified in the Myr−/H9 cells is localized to the cytoplasm or associated with the membrane fraction (Fig. 1E). The soluble cytoplasmic fraction of the Myr−/H9 cells was separated from the pelletable membrane fraction, and each fraction was subjected to the discontinuous sucrose equilibrium-density gradient to identify the DRC, as previously described (18). The fractionated materials were then pelleted by high-speed centrifugation and separated by SDS-PAGE, and the presence of viral proteins was analyzed by immunoblotting (Fig. 1E).

Immunoblotting with an HIV-1-positive human serum indicated that the DRC in the Myr−/H9 cells was predominantly fractionated in the soluble cytoplasmic fraction, which consisted of Pr55^gag and Pr160^gag-pol precursors at a density of 1.09 to 1.13 g/ml (Fig. 1E, top panel). In contrast, the pelleted membrane fraction contained little of either Pr55^gag or Pr160^gag-pol precursor (Fig. 1E, bottom panel). These findings suggest that the DRC identified in the Myr−/H9 cells is localized largely to the soluble cytoplasmic fraction and that myristic acid modification is essential for the membrane association of HIV-1 Gag proteins.

While the DRC was localized to the cytoplasmic fraction of the wild-type HIV-1-infected H9 cells, the DSC was associated with the membrane fraction.

Next, we asked where the DRC and DSC were localized in wild-type HIV-1-infected H9 cells. To address this question, we subjected wild-type HIV-1-infected H9 cells to subcellular fractionation and equilibrium density centrifugation as described above. In the soluble cytoplasmic fraction, immunoblot analysis with HIV-1-positive human serum demonstrated that the DRC, consisting of Pr55^gag and Pr160^gag-pol precursors at a density of 1.09 to 1.13 g/ml, was largely localized to the cytoplasmic fraction (Fig. 2, top panel). This result is consistent with the localization of the DRC to the cytoplasmic fraction of Myr−/H9 cells (Fig. 1E). In the pelleted membrane fraction, however, the mature viral proteins, such as CAp24, MAp17, and RTp66, were cofractionated together at a density of 1.15 to 1.17 g/ml (Fig. 2, bottom panel), characteristic of the DSC (18). Thus, in HIV-1-infected H9 cells, the DRC cofractionated with the soluble cytoplasm, whereas the DSC was predominantly associated with the pelletable membranes.

FIG. 2.

Subcellular fractionation of DRC and DSC from HIV-1-infected H9 cells. HIV-1-infected H9 cells expressing wild-type Gag proteins (Myr+) were subjected to subcellular fractionation and discontinuous equilibrium-density centrifugation as described in the legend to Fig. 1. Viral proteins were visualized by immunoblotting with an HIV-1-positive human serum. The positions of the viral proteins as detected in the released virions are indicated on the right, and molecular mass markers (kilodaltons) are indicated on the left. The positions of the DRC and DSC in the sucrose gradients are also indicated. S fraction, soluble cytoplasmic fraction; P fraction, pelleted membrane fraction.

HIV-1 Gag proteins could assemble into the DRC in vitro in the absence of exogenous lipid membrane.

We also made use of a cell-free system in order to explore the idea that HIV-1 Gag proteins could assemble into the DRC in the absence of cellular membranes. HIV-1 Gag proteins were expressed in a rabbit reticulocyte lysate system as described in Materials and Methods, and the reaction mixture was subjected to discontinuous sucrose equilibrium-density centrifugation to detect DRC. After centrifugation, each fraction was collected from the bottom of the gradient and pelleted by high-speed centrifugation as previously described (18). The pelleted materials were then separated by SDS-PAGE and visualized by immunoblotting with the HIV-1-positive human serum (Fig. 3). The HIV-1 Gag proteins synthesized in the rabbit reticulocyte lysate reaction formed a DRC-like complex, which sedimented at a density of 1.10 to 1.12 g/ml (Fig. 3A).

FIG. 3.

Formation of DRC by HIV-1 Gag proteins expressed in vitro. The HIV-1 Gag proteins were expressed in a rabbit reticulocyte lysate system as described in Materials and Methods. Reaction samples were then incubated in the absence (A) or presence (B) of Triton X-100 for 10 min at room temperature and subjected to discontinuous equilibrium-density centrifugation as described in the legend to Fig. 1. Viral proteins were visualized by immunoblotting with an HIV-1-positive human serum. The positions of the viral proteins are indicated on the right, and molecular mass markers (kilodaltons) are indicated on the left. Lane V, protein profile of released mature virions.

It is possible that the rabbit reticulocyte lysate used in our experiments contains some endogenous lipid membranes and that the HIV-1 Gag complex we detected could therefore be the result of association with these endogenous membranes in the reaction mixture. Since the nonionic detergent Triton X-100 has been shown to solubilize the lipid membrane and increase the density of immature viral particles from 1.15 to 1.17 g/ml to over 1.23 g/ml in our gradient system (18), we asked whether this detergent would affect the density of the Gag complex identified in our cell-free system.

When the in vitro transcription-translation reaction mixture was incubated with 1% Triton X-100 and then analyzed as described above, we found that the in vitro-synthesized Gag proteins still sedimented at a density of 1.10 to 1.12 g/ml after Triton X-100 treatment (Fig. 3B), similar to that observed in the absence of Triton X-100 (Fig. 3A). A small quantity of HIV-1 Gag proteins that had been present at the top of the gradient in the absence of Triton X-100 (Fig. 3A) disappeared after Triton X-100 treatment (Fig. 3B). It is possible that the presence of these HIV-1 Gag proteins at the top of the gradient in the absence of Triton X-100 was the result of association with endogenous lipid membrane. Taken together, these results indicate that the density of the Gag complex formed in the cell-free system was not altered by nonionic detergent Triton X-100 treatment, suggesting a lack of association with the lipid membrane.

High-salt treatment disrupted the DRC identified in both HIV-1-infected CD4⁺ T cells and the cell-free system.

Extensive mutagenesis studies have previously demonstrated that myristic acid modification of the HIV-1 Gag proteins is essential for their intracellular transport to the plasma membrane and for virus production (3, 9, 10, 12, 21, 32, 42, 44). Immunofluorescent staining (42) and electron microscopy (10) have indicated that unmyristylated HIV-1 Gag proteins have a defect in plasma membrane targeting. A similar defect has been described for the unmyristylated M-PMV (26) and spleen necrosis virus (36) Gag proteins. Other investigators, using a subcellular fractionation technique, have reported that unmyristylated HIV-1 Gag proteins fractionated into both the P100 (membrane) and S100 (cytosol) fractions under low-salt conditions (3). However, under high-salt conditions (1 M NaCl), unmyristylated HIV-1 Gag proteins are exclusively found in the S100 fraction (3). Therefore, it has been suggested that these differences in sedimentation behavior under the different salt conditions could be attributed to a difference in the type of association of the Gag proteins with membranes, rather than to a difference in their intracellular transport to the plasma membrane. An alternative explanation is that the unmyristylated Gag and Gag-Pol proteins assemble into DRCs in the cytoplasm, which are then copelleted with the membrane microvesicles into the P100 membrane fraction. Under high-salt conditions, on the other hand, the unmyristylated Gag proteins of the DRC are disrupted into nonpelletable Gag proteins, which are subsequently fractionated in the S100 cytosolic fraction. To test the second explanation, we examined the effect of 1 M NaCl on the stability of the DRC.

DRCs from HIV-1-infected H9 cells were purified and divided into two portions. One portion was adjusted to a final concentration of 1 M NaCl, and then reloaded onto a second sucrose equilibrium-density gradient. As a control, the other portion was simultaneously subjected to the second sucrose gradient without 1 M NaCl treatment. After centrifugation, each fraction was pelleted and analyzed by SDS-PAGE and immunoblotting. As expected, the DRC in the absence of 1 M NaCl again sedimented at a density of 1.09 to 1.13 g/ml in the second sucrose gradient (Fig. 4A, top panel). When 1 M NaCl was added, however, the Pr55^gag and Pr160^gag-pol precursors of the DRC could not be detected in the second gradient. A small quantity of the Pr55^gag precursors was detected at a density of about 1.05 g/ml toward the top of the second gradient (Fig. 4A, bottom panel). This result indicates that 1 M NaCl disrupted the DRC into nonpelletable Gag proteins and a small amount of pelletable Gag complex, which is less dense than the DRC. A similar result was observed for DRC formed in the cell-free system (Fig. 4B). Our findings therefore indicate that the HIV-1 Gag protein observed in the pelleted membrane fraction under low-salt conditions is likely the result of the cosedimentation of the DRC with membrane microvesicles at 100,000 × g. Under high-salt conditions, the Gag proteins of the DRC could not be pelleted because of the disruption of the DRC. These findings also suggest that ionic interactions between Gag and Gag and/or Gag and Gag-Pol precursors are important for stabilizing the DRC.

FIG. 4.

Disruption of the DRC into nonpelletable Gag proteins by high-salt (1 M NaCl) treatment. (A) The DRCs in wild-type HIV-1-infected H9 cells were purified by sucrose equilibrium-density gradient centrifugation as described previously (18) and then divided into two equal portions. As a control, one-half was reloaded over a second sucrose equilibrium-density gradient without addition of 1 M NaCl (A, top panel). The other half was treated with 1 M NaCl at room temperature for 10 min prior to centrifugation (A, bottom panel). (B) HIV-1 Gag proteins were expressed in a rabbit reticulocyte lysate system as described in Materials and Methods and divided into two portions. One-half was loaded over a sucrose equilibrium-density gradient without addition of 1 M NaCl (B, top panel). The other half was treated with 1 M NaCl at room temperature for 10 min prior to centrifugation (B, bottom panel). Gradient fractions were pelleted and analyzed by SDS-PAGE and immunoblotting with an HIV-1-positive human serum. Lanes V and M show the protein profile of released mature HIV-1 viral particles (indicated at left) and molecular mass markers in kilodaltons (indicated at right), respectively.

The Gag precursor proteins in the DRC were digested by trypsin, whereas those in the DSC were not.

We have previously observed that the DRC has a much lower density (approximately 1.10 to 1.12 g/ml) than that of the naked immature viral capsid (>1.23 g/ml). It has also been proposed that HIV-1 Gag protein in the cytosol may adapt a different conformation once it has bound to the membrane (31, 45). Since the DRC is apparently not associated with the membrane, whereas the DSC is, it is important to ask whether the DRC can be conformationally distinguishable from the DSC. To address this question, we have used protease digestion analysis to investigate whether the Gag proteins of either the DRC or the DSC are accessible to trypsin. HIV-1-infected H9 cells were lysed by homogenization, and the postnuclear supernatant was prepared as described previously (18). The postnuclear supernatant containing the DRC and DSC was incubated with trypsin, and the reaction was then inhibited by addition of excess trypsin inhibitor. The protease-treated samples were subjected to discontinuous equilibrium-density gradient centrifugation, and each fraction was then pelleted by a high-speed centrifugation as described previously (18). The pelleted materials were analyzed by SDS-PAGE and immunoblotting with the HIV-1-positive human serum.

In the absence of trypsin treatment, both the DRC (fractions 8 to 10), at a density of 1.09 to 1.13 g/ml, and the DSC (fractions 5 to 7), at 1.15 to 1.17 g/ml, were detected (Fig. 5, top panel), as previously described (18). After trypsin treatment, however, the quantity of Pr55^gag precursors in the DRC was significantly decreased (fractions 8 to 10, Fig. 5, bottom panel). A putative digestion product of the Gag protein, approximately 40 kDa (indicated by an asterisk) was detected at the same position as the DRC (fractions 8 to 10, Fig. 5, bottom panel) and at lighter fractions. This result is consistent with our previous observation that the DRC is sensitive to trypsin treatment (18). In contrast to the DRC, the Pr55^gag precursors in DSC were apparently not affected by trypsin treatment under the same conditions (Fig. 5, bottom panel). These results indicate that the Gag precursor proteins in the DRC were more accessible to trypsin than those in the DSC, suggesting that there may be a conformational difference between the DRC and the DSC. Although Pr55^gag precursors in DSC were relatively resistant to trypsin digestion, these proteins are still sensitive to protease K digestion or high-concentration salt treatment (18), suggesting that the DSC is not fully surrounded by a lipid membrane.

FIG. 5.

Sensitivity of the DRC and DSC to trypsin digestion. Homogenates from wild-type HIV-1-infected H9 cells were prepared as described in Materials and Methods and divided into two equal portions. One portion was incubated in the absence of trypsin (top panel), and the other was incubated with 5 μg of trypsin per ml (bottom panel) at 37°C for 30 min. After incubation, excess antitrypsin inhibitor was added to inhibit further proteolytic digestion. Samples were then subjected to sucrose equilibrium-density centrifugation as previously described (18). Viral proteins were analyzed by immunoblotting with an HIV-1-positive human serum. Lanes V and M show the protein profile of released mature HIV-1 viral particles (indicated at left) and molecular mass markers in kilodaltons (indicated at right), respectively.

The I domain of HIV-1 Gag is required for the formation of the DRC.

Three distinct virus assembly domains, M, I, and L, have been proposed for retroviral Gag proteins (35, 40). Our previous observations (18) and data presented here by using the myristylation mutant of HIV-1 Gag suggest that the M domain HIV-1 Gag is not essential for the formation of the DRC. To study the role of the I and L domains of HIV-1 Gag in the formation of the DRC, we established H9 cells expressing mutant Gag molecules containing a truncation of p6^gag (Pr48/H9, L domain mutant) or NC plus p6^gag (Pr41/H9, L-plus-I domain mutant). The efficiency of DRC formation by the full-length Gag, unmyristylated Gag, p6^gag-truncated, and NC-plus-p6^gag-truncated Gag was evaluated by sedimentation experiments as described in Materials and Methods. Approximately 50% of the full-length Gag (Pr− and ΔPol) and the unmyristylated Gag (Myr−) in the cytoplasm existed in the pelletable complex form. It appeared that the presence (Pr−) or absence (ΔPol) of Gag-Pol precursor did not significantly influence the formation of Gag complexes (Fig. 6). The p6^gag-truncated Gag (Pr48) also formed DRC, and it appeared that more of p6^gag-truncated Gag was present in the complex form than as soluble Gag (Fig. 6). In contrast, formation of the DRC from the NC-plus-p6^gag-truncated Gag (Pr41) was significantly reduced; a majority of the NC-plus-p6^gag-truncated Gag molecules were present as soluble Gag (Fig. 6). To address this question more vigorously, we have used sucrose equilibrium-density gradient centrifugation to study DRC formation by wild-type and NC mutant Gag. Although DRC could be readily detected in the cytoplasmic lysate of H9 cells expressing wild-type Gag (Fig. 7A), little DRC was detected in the cytoplasmic lysate of H9 cells expressing the NC-plus-p6^gag-truncated Gag molecules (Fig. 7B), when comparable total amounts of wild-type and mutant Gag molecules were loaded onto the sucrose gradients (Fig. 7, lanes T). These results suggest that the I domain of HIV-1 Gag is critical for the formation of DRC, a finding that is consistent with the idea that the I domain plays an important role in mediating protein-protein interactions (35, 40).

FIG. 6.

Identification of regions in HIV-1 Gag that are important for DRC formation. The efficiency of DRC formation in the cytoplasm of H9 cells expressing the full-length uncleaved Gag and Gag-Pol (Pr−/H9), the full-length uncleaved Gag without Gag-Pol (ΔPol/H9), unmyristylated Gag and Gag-Pol (Myr−/H9), p6^gag-truncated Gag (Pr48/H9), and NC-plus-p6^gag-truncated Gag (Pr41/H9) was evaluated by sedimentation experiments as described in Materials and Methods. The viral proteins present in the pelletable DRC form (lanes C) and as nonpelletable soluble Gag (lanes S) were analyzed by immunoblotting with an HIV-1-positive human serum.

FIG. 7.

Formation of DRC by HIV-1 full-length and NC deletion mutant Gag proteins analyzed by discontinuous equilibrium-density centrifugation. Cytoplasmic lysates from H9 cells expressing the full-length uncleaved Gag (A) or NC-plus-p6^gag-truncated Gag (B) were subjected to 16 to 60% discontinuous sucrose equilibrium-density gradients as described in Materials and Methods. Eighteen fractions were collected from each gradient, each fraction was pelleted by a high-speed centrifugation, and fractions 3 to 14 from each gradient were analyzed by SDS-PAGE and immunoblotting with an HIV-1-positive human serum. As a control, total viral Gag proteins present in the cytoplasmic lysates were also analyzed side by side (lanes T).

DISCUSSION

In the present study, we have made use of a new subcellular fractionation method to investigate the intracellular localization of two putative HIV-1 virus assembly intermediate complexes, DRC and DSC. By adapting a technique used to separate cytosol from the membrane ghosts of erythrocytes (8, 33), we were able to separate the cytosol from the membranes of HIV-1-infected CD4⁺ T cells. By using this method, we demonstrated that the DRC formed by unmyristylated HIV-1 Gag proteins in H9 cells was fractionated with the soluble cytoplasmic fraction. Little of the unmyristylated HIV-1 Gag proteins was sedimented in the pelleted membrane fraction where most of the viral Env proteins gp160 and gp120 were detected. This observation indicates that the DRCs formed in the cytoplasm of the Myr−/H9 cells are defective in their membrane association. This result is consistent with those of previous studies, which have suggested that myristic acid modification of HIV-1 Gag is essential for the intracellular transport of HIV-1 Gag molecules to the plasma membrane (3, 9, 10, 12, 21, 32, 42, 44).

The DRCs formed by wild-type HIV-1 Gag were also largely fractionated with the soluble cytoplasmic fraction, whereas the DSCs were predominantly associated with the pelleted membrane fraction. Formation of a DRC-like structure was also detected when HIV-1 Gag was synthesized in a cell-free system in the absence of exogenous lipid membrane. Furthermore, the density of these in vitro-generated DRCs was not altered by treatment with the nonionic detergent Triton X-100, suggesting a lack of association with endogenous lipid. Collectively, these findings support the idea that the HIV-1 Gag and Gag-Pol proteins assemble into DRC in the cytoplasm and are subsequently transported to and associated with the plasma membrane.

During the morphogenesis of type-C retroviruses, including HIV-1, electron-dense viral structures are not typically visible in the cytoplasm of infected cells. However, the results presented here and in a previous report (18) suggest that interaction among HIV-1 Gag and/or Gag-Pol precursors occurs prior to their association with the plasma membrane. Other lines of evidence are also accumulating to support such a concept. Coimmunoprecipitation experiments with anti-p6 antibody have revealed the presence of Gag and Gag-Pol precursor complexes in T cells expressing wild-type or myristylation-negative HIV-1 Gag and Gag-Pol precursors (17). Also, it has been reported that membrane-targeting-defective mutants of HIV-1 Gag can be rescued into wild-type Gag particles (20, 42) and can even interfere with virus infectivity (42). Furthermore, other studies have demonstrated incorporation of unmyristylated HIV-1 Gag-Pol precursors into wild-type Gag particles that leads to the activation of viral protease (17, 23, 30). In retroviruses such as Rous sarcoma virus, in which the Gag molecule is not modified by myristic acid, MA mutants that block plasma membrane targeting can also be rescued into particles when coexpressed with wild-type Gag molecules (39).

Formation of immature HIV-1 capsids in vitro in the absence of lipid membrane has been observed for MA-deleted HIV-1 and RSV Gag molecules (4, 5, 13). Assembly of viral capsid structures in the cytoplasm has also been observed for a myristylation-negative mutant form of HIV-1 Gag which contains a large MA deletion (32). On the other hand, formation of immature capsid structures in vitro (19) and formation of DSC in HIV-1-infected cells (18) by full-length HIV-1 Gag molecules are dependent on interaction with lipid membranes. This situation is clearly different from that in type D retroviruses, which form viral capsids in the absence of lipid membranes (28, 37), even with full-length Gag molecules. The molecular mechanism that is responsible for this difference is not clear but seems to be determined by the MA domain (27).

Protease digestion analysis with trypsin has demonstrated that the Gag precursors in the DRC are more sensitive than those in the DSC, suggesting there may be a difference in the conformational structures of the DRC and DSC. It is possible that the Gag molecules in the DRC are more loosely packed than those in the DSC, making the former more accessible to trypsin. This explanation would be consistent with the fact that the DRC has a much lower density (about 1.10 to 1.12 g/ml) than that of the naked immature viral capsid (>1.23 g/ml) (18). It may also explain why the DRCs in the cytoplasm are not readily visible by electron microscopy, whereas virus budding structures (presumably some of which are DSC) at the plasma membrane are more visible. It is conceivable that a conformational change occurs after the DRCs become associated with the plasma membrane. In this case, the myristic acid modification of Gag molecule may play an important role in this process.

Although the formation of the HIV-1 DRC does not require a membrane association signal (M domain) or the L domain in p6^gag, which is required for efficient virus release from the cell surface (11, 14, 22, 29, 41), its formation does require the putative protein-protein interaction domain (I domain) located in the NC region. Previous studies have demonstrated an important role for the NC protein in retrovirus assembly (15, 35, 40). In some cases, the function of the HIV-1 NC in virus assembly has been successfully replaced by other known protein-protein interaction modules (43). How the NC protein might stimulate protein-protein interaction remains to be determined. Since positively charged amino acids in HIV-1 NC have been shown to play an important role in stimulating virus assembly (2, 6), it is interesting to note that in our study, the DRC could be disrupted by high-salt treatment, suggesting certain electrostatic interactions may be important in maintaining the DRC structure. It has been suggested that interaction between retroviral NC protein and RNA may be critical for virus assembly (1, 4, 5, 13, 16). Determination of whether electrostatic interactions in DRC involve protein-RNA interaction or protein-protein interaction requires further study.