Analysis of Minimal Human Immunodeficiency Virus Type 1 gag Coding Sequences Capable of Virus-Like Particle Assembly and Release

Abstract

We have constructed a series of human immunodeficiency virus (HIV) gag mutants by progressive truncation of the gag coding sequence from the C terminus and have combined these mutants with an assembly-competent matrix domain deletion mutation (ΔMA). By using several methods, the particle-producing capabilities of each mutant were examined. Our analysis indicated that truncated Gag precursors lacking most of C-terminal gag gene products assembled and were released from 293T cells. Additionally, a mutant with a combined deletion of the MA (ΔMA) and p6 domains even produced particles at levels comparable to that of the wild-type (wt) virus. However, most mutants derived from combination of the ΔMA and the C-terminal truncation mutations did not release particles as well as the wt. Our smallest HIV gag gene product capable of virus-like particle formation was a 28-kDa protein which consists of a few MA amino acids and the CA-p2 domain. Sucrose density gradient fractionation analysis indicated that most mutants exhibited a wt retrovirus particle density. Exceptions to this rule were mutants with an intact MA domain but deleted downstream of the p2 domains. These C-terminal truncation mutants possessed particle densities of 1.13 to 1.15 g/ml, lower than that of the wt. The N-terminal portions of the CA domain, which have been shown to be dispensable for core assembly, became critical when most of the MA domain was deleted, suggesting a requirement for an intact CA domain to assemble and release particles.

The human immunodeficiency virus (HIV) gag gene encodes a primary core structural protein that is synthesized initially as a polyprotein, Pr55^gag (34, 41). During translation, a myristic acid is cotranslationally attached to the N terminus of Pr55^gag (50, 56), which is required for membrane association and particle assembly (3, 36). At the plasma membrane, the myristylated Pr55^gag molecules self-assemble into virus-like particles and bud out from the cell membrane (5, 48). When virus particles are budding (25), the Pr55^gag is cleaved by the virus-encoded protease (PR) into p17 (matrix [MA]), p24 (capsid [CA]), p2, p7 (nucleocapsid [NC]), p1, and p6 (19, 30, 34). The PR-mediated maturation process of virus particles is essential for virus infectivity (16, 27, 38). In addition to PR, enzymes encoded by pol include reverse transcriptase (RT), RNase H, and integrase, which are required for virus replication (41). The pol gene products are translated as a fusion protein, Pr160^gag-pol, by a ribosomal frameshifting mechanism that occurs at a frequency of 5 to 10% during translation of Pr55^gag (23). The Gag-Pol protein is thought to be assembled into virions via interaction with Pr55^gag (22, 37, 45, 47). Subsequent dimerization of the Gag-Pol molecules induces activation of the embedded PR to cleave Pr55^gag and Pr160^gag-pol (25, 29).

It is clear that the retroviral gag gene contains sufficient information for particle formation (13, 15, 26, 43). The MA protein lies immediately underneath the membrane and forms the viral matrix (14, 39, 40). It is responsible for membrane association and targeting of the Gag precursors to the plasma membrane (2, 11, 46, 63). Mutations within the MA protein sequences have been shown to severely affect stable membrane binding, mutant precursor transport, and particle assembly (12, 14, 46, 54, 62). Incorporation of Env into virus particles is also dependent on the integrity of the MA domain (9, 60). Although subtle mutations in the HIV MA domain may severely disrupt particle assembly (6, 12, 14, 42, 54), a mutant (ΔMA) with a deletion of about 80% of the MA domain has been shown to assemble and process virus particles with wild-type (wt) retrovirus particle densities and to possess wt RT activity (53). Furthermore, replacement of the entire HIV MA with a myristylation signal did not affect particle formation (28). One possible explanation for this discrepancy is that deleterious effects of the smaller mutations on Gag particle assembly have been removed in the MA deletion mutants.

The CA domain is the major core protein of virus particles. Deletion or insertion mutations of the murine leukemia virus CA domain can impair particle assembly (44). However, most regions of the Rous sarcoma virus (RSV) CA can be deleted without significantly affecting particle assembly and release (57–59). Comparative analysis of retroviral Gag proteins identifies a highly conserved sequence, termed the major homology region (MHR), in the C-terminal regions of the CA domains (32). The MHR has been shown to be important for virion assembly in HIV (8, 20, 32, 33, 51) and Mason-Pfizer monkey virus (49). In contrast, a 56-amino-acid deletion mutation in the N-terminal region of HIV CA has been demonstrated to have no major effects on particle assembly and release (7, 54). Concerning the functions of the HIV Gag C-terminal domains, the p7 NC domain contains two Cys-His motifs which are essential for packaging viral RNA into virus particles (1), while the C-terminal p6 domain has been proposed to be involved in the process of virus budding (15, 21). Accumulating data have indicated that the p7NC and p6 domains may not be absolutely required for particle assembly and release (24, 43, 46, 61), while functions of the p2 and p1 peptides are still unclear.

As described above, some regions within HIV gag appear to be dispensable for particle assembly and release. It has been demonstrated in vitro that recombinant HIV CA proteins (10, 18) or purified HIV or RSV CA-NC proteins (4) can assemble into rod-like structures. In addition, a small RSV Gag protein (25 kDa) has been shown to be competent for particle release (55). However, minimum HIV gag sequences required for particle assembly and release from mammalian cells have not been defined. In this study, we constructed a series of C-terminally truncated HIV gag mutants and mutants derived from combination of the C-terminal truncation mutants and an MA deletion (ΔMA) mutant (53). The abilities of these mutants to assemble and release virus particles were assessed by Western immunoblotting and sucrose density gradient fractionation experiments. Localization of the mutant Gag proteins in expressing cells was revealed by indirect immunofluorescence experiments, and mutant particle-associated RT activities were tested by in vitro RT assays. Our results show that mutants with total deletions of about 30 to 50% of HIV type 1 (HIV-1) gag codons still assembled and released particles, which possessed wt retrovirus particle densities. Through these studies, we have identified a minimal HIV gag sequence encoding a 28-kDa recombinant protein capable of particle assembly and release from 293T cells.

MATERIALS AND METHODS

Plasmid construction.

The parental HIV-1 proviral plasmid DNA in this study is HXB₂C (41). Two sets of HIV gag mutations were engineered: one consists of a series of progressive C-terminal truncations (Fig. 1A), and the other one was obtained by combination of an MA deletion mutation (ΔMA) and the C-terminal truncation mutations (Fig. 1B). To construct the p6 Gag deletion mutation, two consecutive stop codons were introduced into the N-terminal-coding region of the p6 gene (nucleotide [nt] 2133). The resultant clone, with the gag reading frame terminated at codon 449, was referred to as T449. The sequence of T449 from nt 2121 is 5′ CCA GGG ATC CTT TAA TAG AGC 3′, which contains a BamHI site (boldface) 5′ to the adjacent stop codons (underlined). To make additional C-terminal truncation mutations, constructs carrying BamHI linker insertions at nt 2071, 1939, 1918, and 1876 were cut with ClaI and BamHI, and the ClaI (nt 831)-to-BamHI fragments of each mutant construct were used to replace the corresponding fragment of T449. These steps resulted in a deletion of gag sequences from linker insertion sites to the T449 BamHI site (nt 2133), with concomitant introduction of the terminator codons, to yield the constructs T431, T387, T380, and T366. The number of each designated construct indicates the position of the gag codon replaced by the stop codons. The juncture sequences for the resultant C-terminal truncation Gag mutants are as follows: T431, nt 2067-5′ ACT GGG ATC CTT TAA TAG AGC 3′; T387, nt 1935-5′ TTT AGG ATC CTT TAA TAG AGC 3′; T380, nt 1914-5′ ATA AGG ATC CTT TAA TAG AGC 3′; and T366, nt 1872-5′ GTT TGG ATC CTT TAA TAG AGC 3′. The ΔMA (53) and ΔNC (52) mutants were as described previously. Briefly, the ΔMA mutation was generated by deletion of the gag coding sequence from the ClaI site at nt 831 to the PvuII site at nt 1147 and insertion of a SalI linker in the deleted region. The resultant construct contained a replacement of 105 deleted codons by sequence encoding four amino acid residues. For construction of the ΔNC mutant, the HIV-1 gag sequence from the ApoI site at nt 1905 to the RsaI site at nt 2066 was removed and replaced by a polylinker, 5′-TCCTGCAGCCCGGGGGATCCGCGGGGT-3′. The other mutants, as illustrated in Fig. 1B, were derived from recombinations of mutant constructs shown in Fig. 1A. Combination of the ΔMA and ΔNC mutants generated the MN construct, and introduction of the ΔMA mutation into T449, T431, T387, T380, and T366 yielded constructs MT449, MT431, MT387, MT380, and MT366, respectively. Each mutant construct was confirmed either by restriction enzyme digestion or by sequencing. All gag mutations were subcloned into HIV gpt (35).

FIG. 1.

HIV gag mutations. Mature processed Gag protein domains of the wt and deletion (dashed lines) and truncation mutants are indicated. All HIV mutants were expressed in the HIV gpt backbone and are described in detail in Materials and Methods. The abilities of the mutants to direct particle release are summarized on the right: +++, release efficiency comparable to wt (≥80% of wt); ++, efficiency about 30% of wt; +, efficiency about 2 to 10% of wt; −, no detectable medium Gag antigens. (A) The ΔMA mutant contains a deletion of 105 codons and a replacement of four amino acid residues in the MA protein. For the ΔNC mutation, 53 codons, including most of NC and a few codons corresponding to the p2 C terminus, were deleted and replaced by 8 codons. The numbers of the C-terminal truncation constructs indicate positions of the HIV gag codons which were replaced by the termination codon. As described in Materials and Methods, T449 was generated by insertion of stop codons in the C terminus of p1. For T431, T387, T380, and T366, the gag coding sequences downstream of the designated codons were deleted. Changed or added codons which resulted from deletion or truncation mutations are underlined. Note that the stop codon insertion in T449 caused an amino acid change in p6 of Gag-Pol from Glu-Phe-Ser-Ser to Asp-Pro-Leu-Ile. The gag-pol frameshift signals were deleted in mutants T431, T387, T380, and T366. (B) Mutant constructs were derived from recombination of the mutants shown in panel A. The combination of ΔMA and ΔNC generated construct MN, and introduction of the ΔMA mutation into T449, T431, T387, and T366 yielded constructs MT449, MT431, MT387, and MT366, respectively.

Cell culture and transfection.

293T cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal calf serum. Confluent 293T cells were split 1:10 onto 10-cm-diameter dishes 24 h before transfection. Twenty micrograms of plasmid DNA of wt or mutant HIV gpt was transfected onto 293T cells by the calcium precipitation method (17), with addition of 50 μM chloroquine to enhance transfection efficiency. At 2 to 3 days posttransfection, culture media and cells were harvested for protein analysis.

Protein analysis.

At 48 to 72 h posttransfection, culture supernatants of transfected 293T cells were collected and filtered through 0.45-μm-pore-size filters, followed by centrifugation through 2 ml of 20% sucrose in TSE (10 mM Tris-HCl, 100 mM NaCl, 1 mM EDTA) plus 0.1 mM phenylmethylsulfonyl fluoride (PMSF) at 4°C for 40 min at 274,000 × g (SW41 rotor at 40,000 rpm). Viral pellets then were suspended in IPB (20 mM Tris-HCl [pH 7.5], 150 mM NaCl, 1 mM EDTA, 0.1% sodium dodecyl sulfate [SDS], 0.5% sodium deoxycholate, 1% Triton X-100, 0.02% sodium azide) plus 0.1 mM PMSF. The cells were rinsed with ice-cold phosphate-buffered saline (PBS), collected in IPB plus 0.1 mM PMSF, and then subjected to microcentrifugation at 4°C for 15 min at 13,700 × g (14,000 rpm.) to remove cell debris. Supernatant and cell samples were mixed with equal volumes of 2× sample buffer (12.5 mM Tris-HCl [pH 6.8], 2% SDS, 20% glycerol, 0.25% bromophenol blue) and β-mercaptoethanol to 5% and boiled for 4 to 5 min. Samples were resolved by electrophoresis on SDS–10% polyacrylamide gels and electroblotted onto nitrocellulose membranes. Membrane-bound HIV Gag proteins were immunodetected by an enhanced chemiluminescence detection system or by a colorimetric method, using as a primary antibody an anti-p24^gag monoclonal antibody (mouse hybridoma clone 183-H12-5C, obtained through the AIDS Research and References Reagent Program, National Institute of Allergy and Infectious Disease and used at a 1:5,000 dilution from purified ascites fluid). For colorimetric immunodetection, the secondary antibody was a sheep anti-mouse immunoglobulin G–alkaline phosphatase conjugate at a 1:2,000 dilution (Vector Laboratories). For enhanced chemiluminescence immunodetection, the secondary antibody was a sheep antimouse horseradish peroxidase-conjugated antibody at a 1:4,000 dilution, and horseradish peroxidase activity detection was by the protocol of the manufacturer (Amersham). Immunodetected bands on films were quantitated with a Personal Densitometer (Molecular Dynamics).

In vitro RT assay.

Culture supernatants of transfected 293T cells were harvested, filtered, and pelleted as described above. Viral pellets were resuspended in 30 μl of TSE buffer. A 10-μl aliquot of each sample was mixed with 40 μl of a reaction cocktail containing 0.1% Triton X-100, 5 mM dithiothreitol, 10 mM MgCl₂, 50 mM Tris-HCl (pH 8.0), 1.2 mM poly(rA)-(dT)₁₅ (Boehringer Mannheim), and 25 μCi of [³H]TTP per ml (38). Reactions were allowed to proceeded at 37°C for 2 h, followed by addition of 5 μl of tRNA (10 mg/ml). The reaction mixtures then were precipitated with ice-cold 10% trichloroacetic acid and filtered with GF/C filters. After the filters were washed and dried, their radioactivities were counted with a Beckman scintillation counter to determine RT activity. To assess particle-associated RT activity for each gag mutant, 10-μl aliquots were analyzed by Western immunoblotting and densitometric quantitation.

Sucrose density gradient fractionation.

Culture supernatants of transfected 293T cells were collected, filtered, and centrifuged through 2-ml 20% sucrose cushions as described above. Viral pellets were suspended in TSE buffer and overlaid on top of premade 20 to 60% sucrose gradients consisting of 1-ml layers of 20, 30, 40, 50, and 60% sucrose in TSE which had been allowed to mix by sitting for 2 h. Gradients were centrifuged at 274,000 × g (SW50.1 rotor; 40,000 rpm) for 16 to 18 h at 4°C, and 500-μl fractions were collected from top to bottom. Each fraction was measured for density and analyzed for Gag proteins by Western immunoblotting.

Indirect immunofluorescence.

The protocol for immunofluorescence was as previously described (54). Briefly, confluent 293T cells were split 1:80 onto coverslips at 24 h before transfection. Two days after transfections, cells were fixed at 4°C for 20 min with ice-cold PBS containing 3.7% formaldehyde. The cells then were washed once with PBS and once with DMEM plus 10% heat-inactivated calf serum (DMEM-calf serum) and permeabilized at room temperature for 10 min in PBS plus 0.2% Triton X-100. Samples were incubated with primary antibodies for 1 h and with secondary antibodies for 30 min. Following each incubation, samples were subjected to three 5- to 10-min washes with DMEM-calf serum. The primary antibody was a mouse anti-p24^gag monoclonal antibody at a 1:500 dilution, and the secondary antibody was a rabbit anti-mouse rhodamine-conjugated antibody at a 1:100 dilution (Cappel). After the last DMEM-calf serum wash, coverslips were washed with PBS three times and mounted in 50% glycerol in PBS for viewing.

RESULTS

Expression and assembly of HIV gag mutants.

In order to define the boundaries of HIV gag sequences necessary for particle assembly and release, a series of truncation and deletion mutations in the HIV gag coding sequences (Fig. 1) were constructed and introduced into a replication-defective HIV vector, HIV gpt (35). As described in Materials and Methods, mutants T449, T431, T387, T380, and T366 were truncated downstream of codons 449, 431, 387, 380, and 366, respectively. Two previously constructed HIV-1 gag mutants, ΔMA (53) and ΔNC (52), also were included in this study.

To test the effects of these mutations on HIV particle assembly and release, mutant and wt HIV gpt constructs were transiently expressed in 293T cells. Culture medium supernatant and cell lysate samples were prepared and subjected to SDS-polyacrylamide gel electrophoresis (SDS-PAGE) followed by electroblotting onto a nitrocellulose filter as described in Materials and Methods. HIV Gag proteins then were immunodetected with an anti-p24^gag monoclonal antibody. Pr55, p41, and the mature Gag product p24 (CA) were observed in the wt cell and medium samples (Fig. 2, lanes 2 and 12). An incompletely processed Gag product, p25, was also visible in the wt cell sample, in agreement with previous reports (34, 42). T449 expressed and released a predicted Gag precursor, Pr50 (corresponding to wt Pr55 with the truncation of p6), as well as p41 and p24/25 (Fig. 2, lanes 3 and 13). Some minor p24-associated bands (p50 [Fig. 2, lane 2] and p55 [lanes 3 and 4]) may result from either partial degradation or incomplete denaturation of the pelleted Gag proteins, as similar phenomena were observed in medium samples from COS7 cells expressing HIV Gag proteins (7, 54).

FIG. 2.

Expression and release of wt and mutant Gag proteins. 293T cells were transfected with the designated constructs. At 48 to 72 h posttransfection, supernatants and cells were collected and prepared for protein analysis as described in Materials and Methods. Supernatant samples (lanes 1 to 9), corresponding to 30% of the total samples, and cell samples (lanes 11 to 19), corresponding to 4% of the total samples, were fractionated by SDS–10% PAGE and electroblotted onto a nitrocellulose filter. HIV Gag proteins were detected with a mouse anti-p24^gag monoclonal antibody at a 1:5,000 dilution, followed by a secondary horseradish peroxidase-conjugated sheep antimouse antibody at a 1:4,000 dilution, and peroxidase activity was determined. Positions of molecular size markers (Std.) (lanes 10 and 20) are indicated on the right, and those of HIV Gag proteins Pr55, p41, and p24 are shown on the left.

Truncation of p6 and p1 (T431) also had no major effects on particle assembly and release (Fig. 2, lanes 4 and 14), as the levels of released mutant Gag precursors were not greatly reduced relative to wt levels. Similarly, mutants T387, T380, and T366 expressed and released mutant Gag precursors Pr42 to Pr39 (Fig. 2, lanes 5 to 7 and 15 to 17, respectively), corresponding to the wt Pr55 with truncations of p7NC-p1-p6 (T387 and T380) or p2-p7NC-p1-p6 (T366). A Gag precursor, Pr42, for ΔNC also was detected in medium supernatants at unreduced levels, relative to wt levels (lanes 8 and 18). Consistent with previous results (53), the ΔMA mutant was assembled and processed, and its Gag products were detected as bands of 42, 28, and 24 kDa (corresponding to wt Pr55, p41, and p24 respectively), with p24 (CA) representing the major species (lanes 9 and 19).

The results shown in Fig. 2 indicate that our HIV gag mutants were still capable of particle assembly, although some mutant proteins were not released as well as wt proteins. However, in order to test the possible effects of multiple mutations on virus particle assembly, we introduced the ΔMA mutation into constructs containing the ΔNC and the C-terminal gag truncation mutations and tested the capabilities for particle assembly and release of each recombinant. As illustrated in Fig. 1B, combination of the ΔMA mutation with the mutations T449, T431, T387, T380, T336, and ΔNC yielded constructs MT449, MT431, MT387, MT380, MT366, and MN, respectively. An addition construct in which ΔMA was combined with a deletion of 56 amino acids (HIV-1 proviral gag sequences from the NsiI site at nt 1251 to the PstI site at nt 1418) yielded the construct designated MNΔNP. Expression and assembly of the recombinant Gag proteins were tested in 293T cells as described above. As shown in Fig. 3, lanes 13 to 19, all mutant constructs expressed Gag proteins with molecular masses corresponding to their predicted gag coding sequences: Gag proteins of MT449 and MT431 were detected as bands of 38 to 37 kDa (Fig. 3, lanes 13 and 14, respectively); the MT387, MT380, and MT366 Gag proteins were detected as bands of about 29 to 28 kDa (lanes 15 to 17); and the MN and MNΔNP Gag proteins were observed as bands of 35 and 30 kDa, respectively (lanes 18 and 19). Interestingly, most of these mutant constructs still could direct the assembly and release of Gag particles into the culture media (Fig. 3, lanes 3 to 9). The two exceptions were mutants MT366 and MNΔNP, which appeared to be blocked in particle release (lanes 7 and 9 respectively). Although the results in Fig. 3 indicate reduced particle release for most of the ΔMA double mutants, we nevertheless have identified a minimum HIV gag coding sequence (MT380) capable of particle assembly. MT380 encoded a small HIV recombinant protein (28 kDa) which consisted mainly of p24-p2, the MA myristylation signal, and a few MA C-terminal residues just before the MA-CA cleavage site.

FIG. 3.

Assembly and release of HIV Gag proteins. 293T cells were transfected with wt HIV gpt and mutant plasmids. Forty-eight hours later, cell and supernatants were collected for protein analysis as described in Materials and Methods. Samples were fractionated by SDS–10% PAGE and then subjected to immunoblot analysis with anti-p24^gag antibody as described in the legend to Fig. 2. Std., standards. Positions of molecular size markers are indicated on the right, and those of HIV Gag proteins Pr55, p41, and p24 are shown on the left.

While we observed that intracellular amounts of wt and mutant Gag proteins were roughly comparable, the levels of some mutant Gag proteins in the medium were remarkably reduced in comparison to wt levels (Fig. 2 and 3). To quantitate these differences, we adopted a previously described methodology to evaluate the effects of gag mutations on HIV particle release (54). Total levels of each Gag protein in cells and medium were quantitated by scanning densitometry, and the extracellular/intracellular Gag protein ratios were determined. For normalization, the ratios obtained with each mutant were divided by wt ratios in parallel experiments. Our results, shown in Fig. 4, indicate that mutants T449, T431, and MT449, possessed medium/cell Gag ratios that were comparable to wt ratios (≥80% of wt). In contrast, the particle release efficiencies were about 30% for T387; 10% for T380, ΔNC MT387, and MN; and about 2 to 5% for T366, MT431, and MT380. Neither MT366 nor MNΔNP Gag antigens were detected in the medium. These results indicate that except for T449, T431, and MT449, the C-terminal truncation or ΔMA double mutation mutants were significantly impaired (T387, T380, T366, ΔNC MT431, MT387, and MN) or completely blocked (MT366 and MNΔNP) in particle release. Our observed low values of mutant protein release could result from inefficient particle release or from protein instability in virus particles. To test the stability of particle-associated Gag proteins, culture media containing wt or mutant virus-like particles were incubated at 37°C for 4 h, pelleted through 20% sucrose cushions, and then subjected to Western immunoblotting analysis. Since we have not observed major differences in the loss of Gag signals after a 4-h incubation (data not shown), we favor the hypothesis that these mutants are inhibited in particle release and that most of their Gag mutant proteins may be trapped intracellularly.

FIG. 4.

Release of HIV Gag proteins from cells. Supernatant and cell samples of wt and mutant constructs were analyzed by Western immunoblotting as described in Materials and Methods. Gag proteins from medium or cell samples were quantitated by scanning mutant and wt Pr55, p41, and p24/25 band densities from immunoblots. Ratios of total Gag protein levels in the media to those in cells were determined for each construct and compared with release levels of wt virus by dividing the release ratio for each mutant by the ratio for the wt in parallel experiments and multiplying by 100. Values for mutants T449, T387, T380, and T366 were derived from three experiments each, and all others were from two experiments each. Error bars indicate standard deviations. dl., Δ.

Indirect immunofluorescence detection of HIV Gag proteins.

To investigate the intracellular locations of mutant Gag proteins, indirect immunofluorescence studies were performed with an anti-p24^gag first antibody and a rhodamine-conjugated rabbit anti-mouse second antibody as described in Materials and Methods. As illustrated in Fig. 5A, wt Gag proteins were detected throughout the cytoplasm of transfected cells with a heterogeneous cytoplasmic staining pattern and a slight perinuclear ring. Similar patterns were seen in the cases of mutants T449 and T431 (Fig. 5D and E, respectively). In contrast, most of the ΔMA Gag proteins appeared to be localized to perinuclear areas (Fig. 5B), a pattern similar to that of ΔMA-transfected COS7 cells (53). The staining patterns of MT449 and MT431 (Fig. 5J and K, respectively) appeared to be roughly similar to the ΔMA pattern. However, these patterns did not correlate with the levels of particle release, as ΔMA and MT449 proteins were efficiently released from cells but MT431 proteins were not. Interestingly, cells expressing Gag proteins with intact MA domains but with deletions or truncations in their NC domains (ΔNC, T387, T380, and T366 [Fig. 5C, F, G, and H, respectively]) showed flat homogeneous staining patterns with no clear perinuclear ring. These patterns were similar to that of COS7 cells expressing HIV Gag–β-galactosidase fusion proteins with intact MA but deleted NC domains (52). About 80 to 90% of ΔNC-transfected cells looked like this, while the others appeared similar to the wt. The proportions of transfected 293T cells that exhibited such a staining pattern were approximately 50 to 60%, 60 to 70%, and 70 to 90% for T387, T380, and T366 respectively, with increased percentages somewhat correlating with the extent of C-terminal truncations. In contrast, MA deletion counterparts MN, MT387, MT380, and MT366 (Fig. 5I, L, M, and N, respectively) and MNΔNP (Fig. 5O) all stained in a heterogeneous punctate pattern with fluorescence extending to cell periphery regions but slightly enriched around perinuclear areas. Such results suggest that the MA deletion double mutant proteins may be trapped intracellularly, perhaps reflecting their impairments in particle release.

FIG. 5.

Indirect immunofluorescence detection of HIV Gag proteins in 293T cells. 293T cells grown on coverslips were transfected with wt HIV gpt (A) and mutants ΔMA (B), ΔNC (C), T449 (D), T431 (E), T387 (F), T380 (G), T366 (H), MN (I), MT449 (J), MT431 (K), MT387 (L), MT380 (M), MT366 (N), and MNΔNP (O). At 48 h posttransfection, cells were fixed and permeabilized for immunofluorescence assays as described in Materials and Methods. The primary antibody was 1:500 dilution of a mouse anti-p24^gag, and the secondary antibody was a 1:100 dilution of rhodamine-conjugated rabbit antimouse antibody. Mock-transfected 293T cells and cells not exposed to the primary anti-Gag antibody yielded no signals (data not shown). Bar in panel O, 10 μm.

Sucrose density gradient fractionation of HIV gag mutants.

Since culture supernatants of transfected 293T cells were centrifuged through 20% sucrose cushions for 40 min, we believed that the recovered Gag proteins in the pelleted media should be virus associated (54). However, because some of these mutants had major deletions, we performed sucrose density gradient fractionation experiments to test whether such large deletion mutations had any effects on virus particle densities. To do so, virus-containing medium samples that had been centrifuged through 2-ml 20% sucrose cushions were resuspended in TSE buffer, layered over premade 20 to 60% sucrose gradients, and centrifuged at 274,000 × g for 16 h, after which fractions were collected and analyzed for sucrose density and Gag proteins. All of our particle-producing mutants were analyzed by this protocol except ΔMA, which previously has been shown to have a wt retrovirus particle density (53). For comparison with wt HIV particle densities in parallel, viral pellets of some mutants also were spun with the wt pellets through the same sucrose density gradients. As predicted, wt HIVgpt, T449, T431, and ΔNC particles all showed peak fractions with densities of 1.16 to 1.18 g/ml (data not shown), consistent with wt retrovirus particle densities. In contrast, the densities of T387 (Fig. 6A) and T380 (Fig. 6B) were about 1.143 to 1.149 g/ml, slightly lower than those of wt HIV particles. The T366 particles, which contain only MA and CA domains, produced particles which exhibited densities of 1.129 g/ml (Fig. 6C), much lower than the wt HIV particle density, suggesting an altered morphology. Cocentrifugation of MN and wt viral pellets showed that both wt and MN Gag proteins had peaks in fraction 5 with a density of 1.177 g/ml (Fig. 6F). Similarly, Gag proteins in particles produced by the constructs MT449, MT431, MT387, and MT380 cosedimented with the wt Gag proteins and banded in fractions with densities between 1.160 and 1.179 g/ml (Fig. 6D and E). Since the assembly of very small HIV Gag proteins into particles was unusual, we further tested whether these mutant particles were membrane enveloped. To do so, medium supernatants from wt-, T366-, MT387-, and MT380-transfected 293T cells were trypsin treated as described previously (28, 58). The results showed that mutant and wt Gag proteins were pelletable following trypsin treatment, suggesting that the mutant Gag proteins were contained within particles (data not shown). Taken together, these results suggest that released Gag proteins were particle associated and that the levels of pelletable medium Gag proteins reflected the levels of virus particles.

FIG. 6.

Sucrose density gradient fractionation of HIV Gag particles. 293T cells were transfected with wt HIV gpt and mutant constructs. At 48 to 72 h posttransfection, supernatants were collected and pelleted through 20% sucrose cushions. Viral pellets were resuspended in TSE buffer. Resuspended viral pellets of each mutant (T387, T380, and T366 [A, B, and C, respectively]) were subjected to sucrose density gradient fractionation (20 to 60%) as described in Materials and Methods. For direct comparison with wt HIV particle densities, wt viral pellets were spun through the same sucrose density gradient with mutant pellets of MT380 (E) or MN (F) or with the pooled pellets of MT449, MT431, and MT387 (D). Each fraction was measured for density and analyzed for Gag protein levels by immunoblot detection. Densities of designated fractions are indicated at the top, while mutant HIV Gag proteins and wt HIV Gag proteins Pr55, p41, and p24 (CA) are shown on the left.

In vitro RT activity assays.

The absence of mature CA proteins indicated that our gag mutants, except T449 and ΔMA, were blocked in particle processing (Fig. 2 and 3). Since the gag sequences covering the Gag-Pol frameshift region had been deleted in the truncation mutants T431, T387, T380, and T366, no functional PR was expected for these mutants or their MA deletion versions (MT431, MT387, MT380, and MT366). Thus, it is not surprising that there were no mature CA proteins in the mutant samples. Nevertheless, we tested the particle-associated RT activities of all assembly-competent mutants. Particles from wt or mutant samples were assayed by using exogenous templates as described in Materials and Methods. As expected, no significant RT activity was observed for truncation mutants T431, T387, T380, and T366 and their MA deletion counterparts (MT431, MT387, and MT380), as the counts per minute of incorporated nucleotide for these mutants were around background levels (Table 1). In contrast, the counts per minute for the wt and the other mutants were at least threefold higher than background levels. To obtain specific activities for each mutant, the ratios of normalized counts per minute versus densitometer-determined virus-associated Gag protein levels were compared with wt levels in parallel experiments. The ΔMA mutant exhibited an RT activity level approaching that of the wt, consistent with the previous results (53). RT activity levels of the p6 deletion mutant T449 also were comparable to wt levels. Interestingly, mutant MT449, derived from combination of the MA and p6 deletion mutations, possessed an RT activity of 0.5% of the wt level, although it efficiently released particles as well as the wt (Fig. 4). Both ΔNC and its MA deletion version MN also had very low RT activity levels (1 to 15%). Low RT levels could be due to reduced stability, processing, or incorporation of Gag-Pol proteins into virions. In addition, stability of the unprocessed immature core and inefficient detergent release of Gag-Pol fusion proteins during RT assays (54) may also account in part for the low RT activities.

TABLE 1.

RT activities of HIV gag mutants^a

Construct	Expt	cpm incorporated	Relative activity (% of wt)b
HIVgpt	1	61,672	100
	2	204,416	100
	3	158,965	100
ΔMA	1	42,627	≥100
	3	196,172	≥100
ΔNC	3	6,598	3
T449	2	101,330	≥100
	3	433,460	≥100
T431	3	1,616
T387	2	3,484
	3	1,664
T380	2	1,250
	3	1,206
T366	2	1,032
	3	1,377
MN	1	5,906	15
	3	4,891	1
MT449	3	13,864	0.5
MT431	3	2,160
MT387	3	2,896
MT380	3	1,855
Mock	1	749
	2	1,335
	3	1,471

^aPreparation of supernatants and RT assays were performed as described in Materials and Methods. For each sample, virus-associated Gag protein levels were determined as described in the legend to Fig. 5. Results of three separate transfection experiments are given. 

^bRelative activities were determined as percentages of wt activities by the equation 100 × [(mutant cpm − background)/mutant Gag protein × wt Gag protein/(wt cpm − background)]. Mutant RT activities with counts per minute at least threefold over the background level were considered positive and were compared with the wt level. 

DISCUSSION

In this study, we constructed a series of HIV gag mutants and tested their abilities to direct the assembly and release of virus particles from 293T cells. We found that the ΔMA mutant, the C-terminal truncation gag mutants T449 and T431, and the ΔMA version of T449 (MT449) could assemble and release virus particles efficiently (Fig. 2 and 4). The NC deletion mutant (ΔNC); the other C-terminal truncation gag mutants, T387, T380, and T366; and the ΔMA versions MN, MT387, and MT380 still assembled and released virus particles, although they demonstrated defective particle release (Fig. 2, 3, and 4). In contrast, MT366 and MNΔNP virus particles were poorly released, as no Gag proteins were detected in the media (Fig. 3 and 4). Our results showing that truncations of the p6 (T449) or p1-p6 (T431) domains had no major effects on particle assembly and release are consistent with previous reports (15, 24, 43, 46, 61). The evidence (Fig. 4) that the p6 deletion mutant virions (T449 and MT449) were released efficiently while the levels of released NC deletion mutant virions (ΔNC and MN) were noticeably reduced indicates that the NC domain is more important than the p6 domain for particle assembly and release.

While almost all of our mutants directed the release of Gag proteins from cells at some level, all of the mutants except T449, T431, ΔMA, and MT449 showed a level of particle release below 50% of wt levels (Fig. 4). Immunofluorescence studies suggested that most of these mutants appeared to have accumulated intracellularly. However, particle release levels did not correlate strictly with the immunofluorescence staining patterns, as ΔMA, MT449, and MT431 all appeared to be enriched at perinuclear membranes, but the MT431 protein was impaired in release, while the ΔMA and MAT449 proteins were not. Immunofluorescence staining of mutants with intact MA but deleted NC domains (ΔNC, T387, T380, and T366) showed an enhanced surface staining without a clear perinuclear ring (Fig. 5). These results suggest that the NC domain may be involved in association with the intracellular membranes or other cellular structural components (52). While the results of immunofluorescence studies are informative about where the proteins have accumulated, we do not know whether the accumulation of mutant proteins in the cells is due to mislocalization or to the impaired transport of mutant proteins.

Deletions downstream of gag codon 387 or 380 (mutants T387 and 380, respectively) did not prevent particle production (Fig. 2), and most of the T387 and T380 mutant particles exhibited a density slightly lower than that of the wt (Fig. 6A and B), in agreement with previous work (24). However, in contrast with observations for the baculovirus system (24, 43), which showed that truncations involving the p2 domain failed to assemble, our mutant T366, with a deletion of the gag sequence downstream of codon 366, still assembled and was released. This discrepancy may be due either to different expression systems employed or to the effects of changed amino acids in the C terminus of our mutant T366 (amino acid residues Leu-Ala-Glu were changed to Trp-Ile-Leu [Fig. 1A]). In the context of MA domain deletions, the MN, MT449, MT431, MT387, and MT380 constructs were still able to direct particle production, although at remarkably reduced levels. The particles produced by these mutants exhibited wild-type HIV densities (Fig. 6D, E, and F), while mutants T387 and T380, which retain MA domains, showed relatively lower particle densities. Possibly the deletion of MA from constructs T387 and T380 permitted mutant Gag cores to adapt tighter, denser conformations, but this hypothesis has yet to be tested.

With regard to the HIV-1 CA domain, although a deletion of 56 amino acid residues in the CA N-terminal portion by itself does not affect particle production (7, 54), the combination of the 56-amino-acid deletion and the ΔMA mutation eliminated particle release (Fig. 3). In contrast, double mutants MN, MT387, and MT380, derived from the combination of the truncation mutations and the ΔMA mutation, clearly could still assemble virus-like particles, although they did not release particles as well as the wt. These results suggest that the CA domain is most important for HIV core assembly. While most of our HIV gag double mutants were impaired in particle release, an extensive genetic analysis of the RSV Gag protein indicates that deletions of over 50% of the RSV gag codons, covering most of the CA residues, showed no major effects on particle assembly and release (59). The fact that the RSV CA does not possess a critical assembly domain (57) or an MHR as the HIV CA does can account in part for the ability of the RSV large-deletion gag mutants to efficiently assemble and release virions.

Three assembly-competent mutants, ΔNC, MN, and MT449, showed complete blocks of particle processing and very low RT activities. This result agrees with previous observations that HIV gag mutants impaired in particle processing have low RT activity levels (54), which may be a consequence of insufficient Gag-Pol incorporation into virions, incomplete processing, or defective Gag-Pol dimerization (29). The fact that ΔNC, MN, and MT449 mutant particles possessed RT activities but showed low PR-mediated particle processing levels indicates that mutations may lead to conformational changes in the Gag-Pol precursor and subsequently interfere with Gag-Pol dimerization. In support of this concept, recent studies have suggested that domains upstream of the PR in HIV Gag-Pol can influence PR dimerization (64). However, a lack of Gag processing also may result from conformational changes of the Gag precursor which interfere with the exposure of the cleavage sites to protease action (31, 42, 54). The finding that our p6 deletion mutant (T449) particles possessed wt RT activity levels but showed incomplete cleavage at p24/25 also may be attributable to one of the above-mentioned possibilities.