INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

The proteins required for the assembly of infectious retroviruses are initially expressed as three separate polyproteins, Gag, Gag-Pol, and Env (reviewed in references 6 and 26). Of these three proteins, the Gag polyprotein, Pr55^Gag in the case of human immunodeficiency virus type 1 (HIV-1), is the major structural protein in the viral particle and provides all of the functions needed for retroviral assembly and budding (reviewed in reference 8). The HIV-1 Gag-Pol polyprotein is produced by a translational frameshift within the C terminus of Gag that occurs infrequently, at about 5% of the level of Pr55^Gag expression. This precursor contains the viral enzymes required for replication: protease, reverse transcriptase, and integrase. The Env protein complex consists of two proteins: the surface glycoprotein gp120^SU, which is noncovalently attached to the transmembrane protein, gp41^TM. This complex initiates infection by binding to the host cell and provides for viral entry into the cell by directly fusing with the plasma membrane.

Electron microscopy studies have shown that assembling lentiviruses form a bar-like structure at the cortex of the plasma membrane, presumably consisting of both Gag and Gag-Pol precursors (reviewed in reference 16). As this structure grows, the nascent particle adopts a spherical shape and finally buds from the cell with a coat of plasma membrane that contains Env. The incorporation of surface glycoproteins into retroviral particles (as well as into other enveloped viruses) appears to lack specificity since Envs from distantly related retroviruses as well as glycoproteins from unrelated enveloped viruses can be incorporated into retroviruses by a process called pseudotyping (63). The exceptions to this pseudotyping phenomenon are the lentiviral Envs, which cannot be incorporated into type C viruses due to the long cytoplasmic tails found in their TM proteins (41). While there are several reports of interactions between HIV-1 Gag and Env during assembly (4, 7, 10, 11, 13, 36, 40, 60, 62), the interactions between budding retroviruses and their Envs are not yet understood.

Lentiviruses bud as immature virions containing Gag and Gag-Pol proteins. These polyproteins are cleaved into the individual proteins by the viral protease that is present in the Gag-Pol precursor. This processing starts during budding and is completed well after particle release (26, 28, 29, 57). Cleavage of Gag and Gag-Pol allows for structural rearrangements and interactions that, in turn, induce a morphological rearrangement and form a condensed, well-ordered mature virion structure (reviewed in reference 16). This maturation event is required for viral infectivity (31). The Pr55^Gag polyprotein is cleaved by the HIV-1 protease into six major mature protein products: p17^MA (matrix), p24^CA (capsid), p2^Gag (sp1), p7^NC (nucleocapsid), p1^Gag (sp2), and p6^Gag (20). Results from a large number of studies have produced a basic understanding of the roles that MA, CA, and NC play in the virus life cycle both as domains in Pr55^Gag and as individual proteins (reviewed in reference 6). In contrast, the function of p6^Gag in the virus life cycle is not as well understood as those of the other Gag proteins.

The two functions that have been demonstrated for p6^Gag have been mapped to the ends of the molecule. The C terminus of this protein has been shown to be required for the incorporation of Vpr, an accessory protein that appears to have important roles in pathogenesis (32, 33, 37). The N-terminal portion of p6^Gag contains a late (L) viral assembly domain that is required for efficient budding of Gag (18). Studies using site-directed mutagenesis and the construction of chimeric viruses have shown that a PTAPP sequence that is highly conserved among the different isolates of HIV-1 is required for the L-domain function of p6^Gag (18, 25, 47, 50).

The p6^Gag protein has an unusual amino acid composition: overall, its sequence has many polar and positively charged residues (44). Interestingly, analysis of the Gag proteins from HIV-1_MN virions has demonstrated that this C-terminal 16-amino-acid (aa) tail is removed from approximately 30% of the p6^Gag molecules found in mature HIV-1_MN (21). This appears to be the result of an HIV-1 protease cleavage between aa 36 and 37 within the p6^Gag sequence KELY³⁶-P³⁷LAS. Comparisons between several HIV-1 strains show that this minor cleavage site as well as the hydrophobic tail of p6^Gag is well conserved (44), raising the possibility that this region is important for HIV-1 replication. To determine the importance of this cleavage site and the general nature of the C-terminal region of p6^Gag, the amino acids at this minor cleavage site and some of the hydrophobic residues in this portion of p6^Gag were altered by site-directed mutagenesis. Mutant viruses were tested for infectivity, and their p6^Gag proteins were analyzed. The results show that maintenance of the cleavage site as well as many of the hydrophobic residues is not required for virus replication. Unexpectedly, the data also revealed that mutations within p6^Gag can block the incorporation of Env into HIV-1 virions.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

DNA mutagenesis. The pNL4-3 infectious molecular clone of HIV-1 was used for these studies (1) and altered by site-directed mutagenesis using PCR-based methods, either by a single amplification with a mutagenic primer method or by two rounds of amplification using the previously described overlap extension procedure (24). Mutations introduced into p6^Gag by this method are as follows: L35P, nucleotide (nt) 2237, TC; Y36C, nt 2240, AG; Y36F, nt 2240, AT; Y36S, nt 2240, AC; P37H, nt 2243, CA; L41P, nt 2255, TC; L44P, nt 2264, TC; and P49L, nt 2278, CT. The TML751X mutation altered nt 8472 from an A to a G, producing a leucine-to-nonsense mutation at codon 751, resulting in the truncation of the gp41^TM cytoplasmic tail. This mutation is similar to those previously described by Freed and Martin (12) and Mammano et al. (40). An Env-defective mutant, Env, was constructed so as to contain a 1 frameshift in the Env precursor leader by the deletion of a T at nt 6349. Mutant DNA fragments were inserted into the pNL4-3 backbone (at the ApaI-BclI sites for the p6^Gag mutants, the EcoRI-KpnI sites for the Env frameshift, and the BamHI-XhoI sites for the L751X mutant) to produce the mutant molecular clones. After construction, the region of the DNA that was PCR amplified was sequenced to confirm the mutation and to rule out the possibility of any additional changes introduced during the mutagenesis process.

Cell culture methods. 293T transformed human kidney and HeLa-CD4-LTR-lacZ (HCLZ; gift of David Waters, AIDS Vaccine Program) cell lines were cultured in Dulbecco's modified Eagle's medium; H9 T-cell leukemia cells and peripheral blood mononuclear cells (PBMC) (provided by the blood donor program of the National Institutes of Health, Bethesda, Md.) were cultured in RPMI 1640. All media were supplemented with 10% (vol/vol) fetal bovine serum, 2 mM L-glutamine, 100 U of penicillin per ml, and 100 µg of streptomycin per ml. All cell culture products were obtained from Life Technologies Inc. (Gaithersburg, Md.). Transfections of 293T cells were carried out by using a calcium phosphate mammalian cell transfection kit (5 Prime-3 Prime, Inc., Boulder, Colo.) according to the manufacturer's recommendation. DEAE-dextran transfections of H9 cells were carried out on 5 × 10⁶ H9 cells for 20 min at room temperature at a DNA concentration of 5 µg/ml and a DEAE-dextran concentration of 200 µg/ml in 0.1 M Tris-HCl-buffered RPMI 1640 medium without supplements. Virion production was measured by reverse transcriptase assay on cell culture supernatants as previously described (17). All infections were done in the presence of polybrene (2 µg/ml), with sheared salmon sperm DNA (sssDNA) as a negative control. The HCLZ assay, a method to measure viral infectivity similar to that used for the MAGI assay (30), was performed as previously described (17). Briefly, a six-well tissue culture cluster (product no. 3506; Costar, Cambridge, Mass.) was seeded with 5 × 10⁵ HCLZ cells the day before infection. The cells were infected with dilutions of virus, and the assay was developed for -galactosidase activity with a 5-bromo-4-chloro-3-indolyl--galactoside stain 48 h postinfection. Positive-staining cells (those colored blue from infection) were observed by light microscopy and counted (as blue cell-forming units [BCFU]) to score infection events. The ability of viruses to replicate was measured by infecting H9 cells as previously described (17). Briefly, H9 cells were exposed to 10-fold dilutions of virus, and the cultures were monitored periodically for infection by the presence of and increase in reverse transcriptase activity in the culture medium. Infections using PBMC were carried out as follows. PBMC were separated from whole blood by centrifugation with Ficoll and stimulated with phytohemagglutinin (PHA) for 48 h before infection. Infections were carried out by plating 10⁶ cells and dilutions of virus into 24-well plates; 24 h later, the PHA and virus were removed and 2 ml of RPMI 1640 medium with 50 U of recombinant interleukin-2 (Life Technologies Inc.) was added. Samples were taken at 4 and 12 days postinfection and assayed for reverse transcriptase activity. Pseudotyping of HIV-1 mutants was accomplished by calcium phosphate cotransfection of 293T cells with a vesicular stomatitis virus G protein (VSV-G) expression construct pCMVHg (5) (a gift of Jane Burns, University of California) or a clone of the Mo10A1 murine leukemia virus (MuLV), pRB161-7 (46) (gift of Robert Gorelick, National Cancer Institute, Frederick Cancer Research and Development Center).

Protein analysis. Virions were isolated by centrifugation through a 20% sucrose pad in an SW28Ti rotor at 25,000 rpm at 4°C for 1 h. Lysates of transfected cells were prepared by washing T150 flasks that contain the adherent cells in phosphate-buffered saline (Life Technologies), followed by lysis with 2 ml of a 20 mM Tris-Cl (pH 7.5)-0.2 M NaCl-1% (vol/vol) Triton X-100 solution. Cell debris were removed from the lysates by centrifugation at 2,500 × g. Immunoblot analysis was performed as previously described, using the enhanced chemiluminescence procedure (Amersham Life Science, Arlington Heights, Ill.) (22). Equal amounts of virions as measured by reverse transcriptase activity (~1.2 × 10⁷ cpm) and 2% of the cell lysates were used for immunoblot analysis. Monoclonal antibody against gp120^SU, rabbit antiserum against gp41^TM and p6^Gag, and goat antiserum against p24^CA were obtained from the AIDS Vaccine Program. Monoclonal antibody against gp41^TM was obtained from New England Nuclear Inc. (Boston, Mass.). For microscale high-pressure liquid chromatography (HPLC), HIV-1 samples were treated with freshly prepared guanidine-HCl (Pierce, Rockford, Ill.) to a final concentration of 4 M, and then the samples were reduced by the addition of 3% (vol/vol) -mercaptoethanol (Sigma, St. Louis, Mo.) for 30 min at 37°C. The samples were separated by using a Shimadzu HPLC system (composed of LC-10AD pumps, SCL-10A system controller, CTO-10AC oven, FRC-10A fraction collector, and SPD-M10AV diodearray detector) with a 2.1- by 100-mm Poros R2/H narrow-bore column (Boehringer GmbH, Mannheim, Germany) at a flow rate of 300 µl/min, using a 3 to 60% (vol/vol) linear gradient of water with 0.1% (vol/vol) trifluoroacetic acid and acetonitrile with 0.0875% (vol/vol) trifluoroacetic acid, respectively. The molecular weights of selected proteins were measured by matrix-assisted laser desorption/ionization-time-of-flight mass spectrometry using a Shimadzu Kompact Maldi-II laser desorption mass spectrometer. One microliter of sample, containing approximately 1 to 100 pmol, was mixed with 0.3 µl of matrix suspension (sinapinic acid [10 mg/ml] in 50% water-50% acetonitrile) directly on the stainless steel target plate. Laser desorption/ionization mass spectra were acquired with a low laser fluence; each spectrum was the sum of at least 100 laser shots. Three molecular mass standards, substance P (1,349.6 Da), bovine insulin (5,734.6 Da), and ubiquitin (8,565.9 Da), were used as standards for the mass measurement of the p1, p7^NC, and p6^Gag proteins.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Mutagenesis of the p6^Gag tail region. The hydropathic character of the p6^Gag sequence was analyzed by the Kyte-Doolittle algorithm (34), and the plot confirmed that this protein is predominantly hydrophilic (Fig. 1). However, the analysis also showed that some of the residues in the C-terminal 16 aa of p6^Gag have a more hydrophobic character than the rest of the protein. It has been shown that a substantial fraction of the p6^Gag proteins inside HIV-1 virions (approximately 30% in the case of HIV-1_MN) are cleaved internally between Y³⁶ and P³⁷ (21) (Fig. 1). Interestingly, this cleavage physically separates the highly hydrophilic region from the more hydrophobic region, producing three species of p6^Gag inside the virion: a full-length 52-aa, an N-terminal 36-aa, and a C-terminal 16-aa species. The functional significance of this cleavage site or of the p6^Gag tail's hydrophobic nature is not understood.

View larger version (19K): [in this window] [in a new window]   FIG. 1.   Mutations in p6Gag. Hydropathic analysis of p6Gag (using a 10-aa window) is presented above its sequence; mutations made to the C-terminus of p6Gag are presented below. Amino acids are represented by the standard single-letter code. The L domain in p6Gag is boxed. The internal protease cleavage site in p6Gag is indicated below the amino acid sequence.

Infectivity and replication of mutants. The mutant DNA constructs were transiently transfected into 293T human kidney fibroblasts to produce virus. All of these mutants appeared to produce near-wild-type levels of reverse transcriptase at the 48- and 72-h harvests (data not shown). The mutant viruses were examined for the ability to carry out a single round of infection, using HCLZ cells in an LTR-lacZ Tat complementation assay. The results (Table 1) showed that all but two of the mutant viruses tested gave titers that were essentially equivalent to those of the wild-type virus. The Y36F mutant had a titer that was approximately 300-fold lower than titers for the other mutant and wild-type samples, while the Y36S-L41P mutant had essentially no titer (2 BCFU/ml) in the HCLZ assay.

5

View larger version (71K): [in this window] [in a new window]   FIG. 2.   Immunoblot analysis of the single-substitution mutants. A single set of blots was exposed to antiserum against p6Gag (A) and then successively stripped and reprobed with antiserum to p24CA (B). The virion samples produced by transfection (corresponding to ~1.2 × 107 cpm of reverse transcriptase activity) are identified above each lane. sssDNA, mock control; WT, sample of the wild-type NL4-3 virions. The apparent molecular masses as determined by relative mobility are presented at the left.

Cleavage of the p6^Gag internal protease site. Due to the hydrophobic nature of the N-terminal p6^Gag cleavage product, immunoblot analysis is unable to reliably detect this protein (19) and cannot be used to determine whether the minor protease site was processed in the mutants. Therefore, these mutant virions were analyzed by microscale HPLC so that the various products of p6^Gag can be isolated, characterized, and quantitated. Approximately 5 µg (total protein) of virus produced from the wild-type DNA and each mutant construct was subjected to microscale HPLC analysis, and the peaks containing mature Gag proteins were identified by immunoblotting, mass spectrometry, sodium dodecyl sulfate-polyacrylamide gel electrophoresis followed by staining of the gel with Coomassie brilliant blue (C-PAGE), and protein sequence analyses. These analyses revealed no significant differences in the HPLC profiles of the other mature Gag proteins between the wild-type and mutant samples, confirming that overall processing occurs normally for all the mutants (Fig. 3 and data not shown). The HPLC chromatograms of the p6^Gag region for all the viruses tested are presented in Fig. 3. Compared to the wild-type sample, which eluted as expected from previous experiments, the mutant p6^Gag proteins were resolved at somewhat different positions within the chromatograms due to the sequence changes in these relatively small proteins. The molecular weights of the proteins within the peaks in this region of the HPLC were determined by mass spectroscopy (Table 2). The data matched the values calculated from the full-length sequence of the wild-type NL4-3 and mutant p6^Gag proteins, confirming the identity of the peaks and demonstrating that these proteins were not internally cleaved. Sequence and mass spectrometry data of the peak that elutes between p1 and p6^Gag, labeled p6^f in Fig. 3, revealed that it contained both of the N- and C-terminal p6^Gag cleavage fragments. This peak, when present, was found in most of the mutant samples except for Y36F, where the N-terminal cleavage product appeared to elute later in the chromatogram, probably due to the hydrophobic Y-to-F exchange. Mass spectrometry of this peak showed that it had the expected size for the Y36F N-terminal fragment, confirming the identity of this peak. The mass spectrometry data for all of the p6^Gag-containing peaks were in good agreement with the values calculated from the sequences (Table 2), showing that the majority of these mutant proteins were full length, correctly processed, and not posttranslationally modified.

View larger version (17K): [in this window] [in a new window]   FIG. 3.   HPLC analysis of p6Gag mutants. The complete HPLC chromatogram for wild-type NL4-3 virions is shown at the top, with the region presented below boxed. The identities of the Gag proteins determined by C-PAGE, immunoblotting, protein sequence, and mass spectrometry are identified above the peaks. The p1Gag peak is shaded; the p6Gag fragment peak, labeled p6f on the wild-type (WT) profile, is identified by a black triangle; full-length mutant p6Gag is highlighted with an arrow. The relative amount of p6Gag cleaved versus the total p6Gag present (all three species) is presented on the right.

Protein analysis of Y36S-L41P. To examine the reason for the striking defect in infectivity and replication exhibited by Y36S-L41P, proteins from the mutant were analyzed. Immunoblot analysis of wild-type and Y36S-L41P mutant virions with p6^Gag antiserum revealed that the mutant p6^Gag band migrated higher than the wild-type protein (Fig. 4A), similar to the result with the other mutants. Stripping and reprobing the blot with p24^CA antiserum showed that the mutant and wild-type samples produced equivalent signals of p24^CA, indicating that similar amounts of virus were present on the blot and the processing to produce this protein occurred normally (Fig. 4A). HPLC and mass spectrometry analysis of the Y36S-L41P mutant showed that the mature p6^Gag of the Y36S-L41P mutant was not cleaved at the internal Y³⁶-P³⁷ site in p6^Gag (Fig. 3; Table 2), yet the other Gag proteins eluted as expected (data not shown), confirming that Gag is processed normally.

View larger version (38K): [in this window] [in a new window]   FIG. 4.   Immunoblot analysis of Y36S-L41P and Y36F virions. (A) Blots of both wild-type NL4-3 (WT) and Y36S-L41P virion samples produced from transfection (corresponding to ~1.2 × 107 cpm of reverse transcriptase activity) as well as samples of lysates isolated from cells transfected with viral constructs (2% of the lysate from a confluent T150 flask). (B) Immunoblots of wild-type and Y36F virions samples produced from transfection (corresponding to ~1.2 × 107 cpm of reverse transcriptase activity). Samples are identified above the lanes, and the antiserum used for each reaction is noted above each blot. Molecular masses as determined by relative mobility are presented on the left.

Rescue of Y36S-L41P by pseudotyping. The results shown above suggest that the Y36S-L41P mutant is primarily defective due to the absence of Env in the particles. However, this mutation may also produce other defects in the viral life cycle that cannot be determined due to the inability of this mutant to efficiently bind and enter host cells. To test this possibility, the mutant was cotransfected with a VSV-G expression plasmid in hopes of pseudotyping the Y36S-L41P mutant and bypassing the Env incorporation defect. As controls, the wild-type construct and an Env mutant that contains a frameshift mutation in the leader sequence of the env gene (thus does not produce gp120^SU or gp41^TM) were also cotransfected with the VSV-G expression construct. The resultant viruses were assayed for infectivity in the HCLZ assay to determine whether these virions could undergo one round of replication. The results, presented in Table 3, showed that VSV-G clearly rescued the Y36S-L41P mutant to near wild-type levels, higher than the level of the pseudotyped Env mutant. The ability of the VSV-G envelope to rescue the infectivity of the Y36S-L41P mutant suggests that the inability to incorporate Env is the only defect in this mutant. However, recent work by Aiken has shown that VSV-G pseudotyping of an Env HIV-1 mutant suppresses the requirement for Nef and viral sensitivity to cyclosporin A compared to HIV-1 that uses its own Env to enter the cell (2). Therefore, even though the Y36S-L41P was rescued by VSV-G pseudotyping, it is still possible that the endocytic mode of entry used by this glycoprotein masks other defects in this mutant. To test this possibility, the Y36S-L41P mutant was cotransfected with an expression plasmid for Mo10A1, an MuLV whose Env, like HIV Env, uses a direct fusion mechanism with the host plasma membrane and can infect human cells (42, 43, 46, 55). Previous studies have shown that MuLV Env can pseudotype HIV-1 and rescue HIV-1 Env-deficient viruses (35, 39, 53). The results showed that the MuLV Env rescued the Y36S-L41P mutant (Table 4). The efficiency of the rescue was considerably lower than the VSV-G results, consistent with data from other studies (2, 38). The reduced levels may be due to differences in incorporation into the virion and/or receptor usage by these two surface glyco-proteins. Since the source of MuLV Env was the Mo10A1 molecular clone, this DNA also produced an MuLV that can infect human cells. However, since MuLV does not produce Tat, it did not score positive in the HCLZ assay (Table 4). Since pseudotyping with either VSV-G or 10A1 Env protein can complement the mutant phenotype, these data confirm that the defect in Y36S-L41P is in the incorporation of the HIV-1 Env into the virion and not in other aspects of assembly or infection.

View larger version (36K): [in this window] [in a new window]   FIG. 5.   Immunoblot analysis of virions with the gp41TM L751X mutation. The same blot of both wild-type NL4-3 and mutant virion samples produced from transfection (corresponding to ~1.2 × 107 cpm of reverse transcriptase activity) was analyzed with gp41TM antibody (A) and then successively stripped and reprobed with gp120SU antibody (B) and p24CA antiserum (C). Lanes: 1, sssDNA (negative control); 2, wild-type NL4-3; 3 and 4, two separate clones of NL4-3/TML751X; 5 and 6, two separate clones of Y36S-L41P/TML751X; 7, Y36S-L41P. Molecular masses as determined by relative mobility are presented on the left.

Y36S-L41P exhibits a dominant negative effect. The Env incorporation defect of the Y36S-L41P mutant could be due to the inactivation of an important domain in Gag or, alternatively, to the introduction of a sequence that blocks the incorporation of Env. Studies have shown that wild-type Gag can complement Gag-Pol precursor mutants in trans (48, 54) and that Gag mutants can dominantly interfere with wild-type Gag assembly in trans (58), suggesting that HIV-1 Gag molecules can form functionally mixed particles. To determine whether wild-type Gag can complement the Y36S-L41P, the mutant and wild-type constructs were cotransfected into 293T cells at a ratio of 9 to 1 to coexpress these two Gag proteins. As controls, transfections with similar amounts of mutant or wild-type plasmids with the appropriate amount of carrier were also tested. The resultant virus stocks were assayed for a single round of infectivity with the HCLZ assay. The data (Table 6) showed that the 9-to-1 cotransfection produced a titer 3,000-fold lower than that produced by the wild-type transfection. It is important to note that while the ratio of mutant Gag to wild-type Gag is different in the cotransfections, the total expression of viral proteins is essentially the same as for either the wild-type or mutant DNA transfections (Fig 6B). These results show that the mutant dominantly interferes with the ability of the wild-type Gag proteins to incorporate Env. A control cotransfection of sssDNA and wild-type DNA at a 9-to-1 ratio produced a titer 10-fold lower than that produced by the wild-type transfections. The large difference between this control and the 9-to-1 mutant-to-wild-type cotransfection shows that the reduction in titer was due to the presence of the mutant phenotype and not simply less wild-type expression. The virus stock produced from a 1-to-1 cotransfection of Y36S-L41P and wild-type yielded a titer 10-fold lower than that found after transfection with the wild type (Table 6). This weaker inhibition demonstrates that the negative effect was dependent on the relative amount of mutant DNA in the transfection. Taken together, these results show that the Y36S-L41P mutant Gag inhibits the infectivity of the wild-type virions in trans, apparently by forming virions that contain both mutant and wild-type Gag proteins during particle formation.

View larger version (36K): [in this window] [in a new window]   FIG. 6.   Immunoblot analysis of Y36S-L41P and wild-type cotransfectants. A blot containing virion samples produced from the cotransfections (corresponding to ~1.2 × 107 cpm of reverse transcriptase activity) was reacted with anti-gp41TM antibody and then stripped and reacted with antiserum against p24CA. Lanes: 1, sssDNA (negative control); 2, wild-type NL4-3; 3, Y36S-L41P and NL4-3 transfection at a 1-to-1 ratio; 4, Y36S-L41P and NL4-3 transfection at a 9-to-1 ratio; 5, Y36S-L41P.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

The results presented here show that all of the single-amino-acid mutations within the internal cleavage site and the hydrophobic tail of p6^Gag inhibited cleavage yet had a minor effect on viral replication and infectivity. Therefore, protease processing at this site is not essential for virus infectivity. In fact, the Y36F mutant had decreased infectivity yet had the second-highest level of internal cleavage of the mutants, showing that internal cleavage of p6^Gag and infectivity are not linked.

The basis for the Y36F mutant defect is currently unknown. The levels of virus released into the medium as determined by reverse transcriptase assay and immunoblotting do not appear to be significantly lower than those for the other mutants. Furthermore, immunoblot analysis for Env showed that this mutant incorporated amounts of Env similar to those incorporated by the wild type.

The observation that mutations distal to the protease site also inhibited the use of this site was unexpected since it is thought that the four amino and three carboxyl residues flanking the scissile bond fit into the protease substrate-binding pocket and most directly affect proteolysis (9, 27, 52, 57, 61). An explanation for these results is that the tertiary structure of p6^Gag strongly influences this cleavage. However, structural studies of p6^Gag produced in bacteria have failed to show any ordered structure, and in vitro studies have found that peptides containing this cleavage site as well as recombinant p6^Gag protein itself are not cleaved by protease (56). Since p6^Gag is first produced as a sequence within the Pr55^Gag precursor, the ability of protease to cleave at this site may be determined more by structures adopted and interactions made during assembly and processing of the Gag polyprotein than by the primary sequence around the cleavage site.

Unlike the single-substitution mutants, the Y36S-L41P mutant was unable to infect cells because it did not incorporate significant amounts of Env. The inability to acquire the Env protein complex appears to be the only significant defect for this mutant since it could be rescued by pseudotyping with either VSV-G or MuLV Env. In some cases, the HCLZ assay did detect a very low amount of infectivity with the double mutant, raising the possibility that some Env was incorporated into the virion. This result could also be explained by rare Env-independent infection events arising from nonspecific virus-cell fusion. The results of the replication assays and the failure to isolate a revertant after attempts at long-term culturing of either transfected or infected H9 cells show that while these viruses might infect cells at a very low level, they fail to replicate at a detectable level.

The cotransfection experiments presented here demonstrated that this mutation acts in a trans-dominant negative fashion. This implies that the failure to incorporate Env is due to the presence of a sequence in p6^Gag that blocks incorporation rather than the alternative, the loss of a p6^Gag sequence that is required for Env incorporation. This interpretation is supported by a finding that truncation of the C-terminal tail of p6^Gag has little effect on the incorporation of gp41^TM into virions (45).

The mechanism for the exclusion of Env by the Y36S-L41P mutant is not clear. Studies using polarized cells have shown that the cytoplasmic tail of Env can influence the site of viral budding (36), suggesting a coordination of Env and Gag localization. Therefore, it is possible that the failure to incorporate Env into Y36S-L41P virions is due to an inability of the mutant Pr55^Gag proteins to assemble in a region of the plasma membrane that contains HIV-1. However, several observations argue against this hypothesis. It has been shown that deletion of the C terminus of the gp41^TM cytoplasmic tail causes HIV-1 to bud from both surfaces of polarized cells (36), indicating that this truncated Env no longer localizes to a specific region of the plasma membrane. Therefore, the inability of the Y36S-L41P mutant to incorporate the truncated Env suggests that a polarized budding phenomenon is not responsible for the Env packaging defect. Furthermore, recent experiments by Mammano et al. have shown that truncation of the cytoplasmic tail of HIV-1 Env allows it to be incorporated into MuLV (41). This result together with the fact that either MuLV Env or VSV-G can pseudotype both MuLV and HIV-1 shows that all of these glycoproteins, even if confined to a limited area, are present in a common region on the surface of the cell. Therefore, the differential ability to incorporate MuLV Env and VSV-G and not the HIV-1 full-length or truncated Env argues that the defect of the mutant is not due to an alteration of the site of budding. Finally, the trans-dominant negative effect of this mutant appears to be the result of an interaction between the mutant and wild-type Gags, most likely producing virions that contain mixtures of the two. This effect suggests that these Gags assemble and bud from similar regions of the cell.

Our data are consistent with a model in which the presence of the mutant p6^Gag protein within the Gag precursor excludes HIV Env but not other Env proteins during assembly and budding. Some p17^MA mutants appear to sterically block the long cytoplasmic tail of gp41^TM from fitting into the immature budding virion at the cortical membrane-Gag interface (3, 11, 12, 23, 40, 51). The failure of Y36S-L41P mutant to package the truncated gp41^TM demonstrates that the mechanism for this defect is different from that of the p17^MA mutants. The location of p6^Gag in assembling and budding virions is not known. However, given the current body of data, it is unlikely that the majority of p6^Gag is present near the surface of the plasma membrane (16, 59). Thus, from our data as well as the established models of immature virion structure, it appears unlikely that this defect is due to p6^Gag structurally blocking the insertion of Env into the virion at the cortical membrane as seems the case for the p17^MA mutants.

While it is not clear how this mutant Gag excludes Env, it is important to appreciate that p6^Gag is required for efficient budding, as it harbors the L domain for HIV-1 (18, 25, 47, 50). The mechanism for L-domain action is unknown, though in the case of avian viruses the L domain appears to interact with a cellular protein (15). A recent study by Garnier et al. (14) found that the entire p6^Gag region within Pr55^Gag is an important determinant of virus particle size. Taken together, these observations implicate p6^Gag in the regulation of assembly and budding, possibly in conjunction with host proteins, by an unknown mechanism. While the mutations presented here do not affect the L domain proper, the Env incorporation defect identified here is likely to be linked to the role of p6^Gag in assembly and budding, probably as a domain in Pr55^Gag. Env could be excluded from the virion either directly by the mutant p6^Gag during its role in assembly or by interfering with host proteins that might assist in this process.

These results identify a novel aspect in the role of p6^Gag in HIV-1 assembly. Since relatively little is known about p6^Gag, this observation may provide an important clue to its function.