Subunit Stoichiometry of Human Immunodeficiency Virus Type 1 Envelope Glycoprotein Trimers during Virus Entry into Host Cells

Abstract

The envelope glycoproteins of human immunodeficiency virus type 1 (HIV-1) function as a homotrimer of gp120/gp41 heterodimers to support virus entry. During the process of virus entry, an individual HIV-1 envelope glycoprotein trimer binds the cellular receptors CD4 and CCR5/CXCR4 and mediates the fusion of the viral and the target cellular membranes. By studying the function of heterotrimers between wild-type and nonfunctional mutant envelope glycoproteins, we found that two wild-type subunits within an envelope glycoprotein trimer are required to support virus entry. Complementation between HIV-1 envelope glycoprotein mutants defective in different functions to allow virus entry was not evident. These results assist our understanding of the mechanisms whereby the HIV-1 envelope glycoproteins mediate virus entry and membrane fusion and guide attempts to inhibit these processes.

Human immunodeficiency virus type 1 (HIV-1) encodes a gp160 envelope precursor glycoprotein. HIV-1 envelope glycoprotein spikes mature by trimerization, glycosylation, and proteolytic cleavage of the gp160 precursor into gp120 and gp41 glycoproteins. The glycoproteins are then loaded onto virions as a homotrimer of gp120 and gp41 heterodimers (14, 45). Viral entry is initiated by binding of the gp120 glycoprotein subunits to the cellular receptor CD4 (9, 23). The interaction with CD4 leads to structural changes within gp120 that allow binding to the coreceptors, either CCR5 or CXCR4 (1, 4, 8, 10, 11, 15, 30). CD4 binding also induces conformational changes in the gp41 ectodomain that result in the exposure of a trimeric coiled coil (20, 37). Formation of the gp120/CD4/CCR5 complex is thought to result in exposure of the fusion peptide, a highly hydrophobic region at the gp41 N terminus that contacts the target cell membrane (18). Further rearrangements within the gp41 ectodomain result in the formation of a six-helix bundle structure and the fusion of opposing membranes from the virion and the target cell (6, 7, 13, 41, 43). Peptide or small-molecule inhibitors targeting CD4-induced conformational changes in gp120, gp120-coreceptor binding, or six-helix bundle formation in gp41 have been shown to block HIV-1 infection, representing promising new directions for developing the next generation of anti-HIV-1 therapeutics (27, 35, 44, 49). The number of subunits within an HIV-1 envelope glycoprotein trimer that is required for virus entry is unknown. This information specifies the number of subunits that have to be bound and inhibited by candidate drugs to limit HIV-1 infection.

Cooperativity among the three individual subunits in an envelope glycoprotein trimer may allow efficient function. The existence of such cooperativity was suggested by a study of the effects of fusion-inducing pH decreases on conformational changes in heterotrimers composed of hemagglutinin glycoproteins from two different strains of influenza virus (5). Complementation in a cell-cell fusion assay among HIV-1 envelope glycoprotein mutants defective in receptor binding and fusion peptide function has been reported (36). Such complementation implies subunit cooperativity within a trimer and suggests that fewer than three functional subunits can achieve membrane fusion. However, these studies did not provide a precise estimate of the number of functional subunits that are required for a trimer to support virus entry. Interestingly, soluble envelope glycoprotein trimers of simian immunodeficiency virus apparently bind only a single molecule of soluble CD4, whereas three antibody molecules can be bound (22). In contrast, more than one soluble CD4 molecule could bind to a soluble envelope glycoprotein trimer of HIV-1_ADA (48). In this study, we attempted to estimate the subunit stoichiometry within an envelope glycoprotein trimer during virus entry by testing the function of heterotrimers of HIV-1 envelope glycoproteins carrying one or two loss-of-function mutant subunits.

MATERIALS AND METHODS

Theoretical model building.

Recently, by studying viruses carrying heterotrimers between the wild-type HIV-1 envelope glycoproteins and closely related dominant-negative or neutralization escape mutants, we demonstrated that an individual HIV-1 envelope glycoprotein trimer functions independently to mediate virus entry (46, 47). This stoichiometry has been designated T=1, where T denotes the minimum number of trimers that function as a unit to mediate virus entry. The logical arguments, model-building techniques, and experimental approaches applied in the present study are similar to those used previously to investigate the stoichiometry of HIV-1 entry and neutralization.

The HIV-1 gp160 precursor trimerizes soon after its synthesis in the endoplasmic reticulum (12). Subsequently, proteolytic cleavage produces the mature, functional trimer composed of three gp120 and three gp41 subunits (19, 39). When two different species of viral glycoproteins are coexpressed in a cell, heterotrimers can form. Heterotrimers between two species of closely related viral glycoproteins can form with an efficiency close to that predicted by random mixing of two species of monomeric subunits (5, 38). The frequency of different species of envelope glycoprotein trimers formed by random mixing is dependent on the frequency of mutant monomers (f_M) in the pool of available envelope glycoprotein monomers. Then, the frequency of different trimers in a viral stock can be calculated as follows: the frequency of the wild-type homotrimer is (1 − f_M)³; the frequency of the heterotrimer with two wild-type subunits and one mutant subunit is 3(1 − f_M)²f_M; the frequency of the heterotrimer with one wild-type and two mutant subunits is 3(1 − f_M)f_M²; and the frequency of the mutant homotrimer is f_M³.

Let S be the number of wild-type subunits within an envelope glycoprotein trimer that is required for trimer function. For the purpose of model building, we assume that an individual envelope glycoprotein trimer functions in an all-or-none fashion; thus, a trimer with S or more wild-type subunits is assumed to function like a wild-type homotrimer, and a trimer with fewer than S wild-type subunits is considered to be totally nonfunctional. If all three subunits in a trimer are required for function (S=3 model), only the wild-type homotrimers, which are present at a frequency of (1 − f_M)³, are potentially functional in a viral stock. In the S=2 model, in which two wild-type subunits in a trimer are required for function, both the wild-type homotrimers and the heterotrimers with two wild-type subunits and one mutant subunit are potentially functional. In this case, the frequency of functional trimers is (1 − f_M)³ + 3(1 − f_M)²f_M or 1 − 3f_M² + 2f_M³. In the S=1 model, in which only one wild-type subunit in an envelope glycoprotein trimer is required for function, all of the envelope glycoproteins except the mutant homotrimers are potentially functional. Thus, functional envelope glycoproteins exhibit a frequency of 1 − f_M³ in this case.

HIV-1 stocks contain a very high proportion of functionally defective virions (often over 99.9%); thus, the chance that one HIV-1 virion carries more than one infectious unit is statistically negligible (29, 32). Because each envelope glycoprotein trimer functions independently (T=1) (46, 47), the infectivity of an HIV-1 viral stock should directly reflect the frequency of virion-associated trimers that are potentially functional. The experimentally determined infectivity of a virus preparation containing a given proportion of mutant envelope glycoproteins can be expressed as a relative infectivity, RI(%), which is a percentage of the infectivity observed for a viral stock with only the wild-type envelope glycoproteins. The relationships between the f_M and RI(%) values of viral stocks predicted by the three candidate models are shown in Fig. 1. Results consistent with the S=3 model can hypothetically be achieved in one of two ways: either by the requirement for three functional subunits in an envelope glycoprotein trimer or by the presence of a transdominant-negative mutant subunit. The latter scenario has been observed previously with two transdominant-negative mutant envelope glycoproteins (47).

FIG. 1.

Theoretical models for the subunit stoichiometry of the HIV-1 envelope glycoprotein trimer. S is defined as the number of wild-type subunits within one HIV-1 envelope glycoprotein trimer that is required for the trimer to support virus entry. f_M is defined as the frequency of mutant monomers in the total pool of HIV-1 envelope glycoproteins. The frequency of potentially functional trimers was calculated using the theoretical assumptions described in Materials and Methods. The frequency of potentially functional trimers in a viral stock is predictive of the infectivity relative to that of the wild-type viruses [RI(%)].

The overall relationship among S, f_M, and RI(%) may be summarized as RI(%) = {1 − [f_M³ + 3f_M²(1 − f_M)(S − 1) + 3f_M(1 − f_M)²(S − 1)(S − 2)/2 − 3f_M²(1 − f_M)(S − 1)(S − 2)/2]} × 100, assuming that each envelope glycoprotein trimer functions independently to mediate viral infection, i.e., T=1 (46, 47). Reduction of this formula by S=1, S=2, or S=3 would result in the f_M/RI(%) relationships for individual models, as described above.

Plasmids expressing HIV-1 envelope glycoproteins.

The functionality of trimers composed of wild-type and mutant HIV-1 envelope glycoproteins was evaluated using envelope glycoproteins derived from the HIV-1_YU2 isolate. HIV-1_YU2 is a primary isolate, uses CCR5 as a coreceptor, and is very resistant to neutralizing antibodies. The wild-type HIV-1_YU2 gp160 glycoproteins were expressed from the pSVIIIenv vector (40). The glycoprotein mutants were created by the PCR-based QuikChange protocol (Stratagene). The D368R, R315G/L317S, and L520E mutants disrupt CD4 binding, CCR5 binding, and membrane fusion, respectively (17, 34, 42). The residue numbers correspond to those of the prototypic HXBc2 envelope glycoproteins, according to current convention (25). The presence of the desired mutations and the absence of unintended coding changes were confirmed by DNA sequencing of the entire env reading frame. All plasmids expressing the wild-type and mutant envelope glycoproteins were prepared using a QIAFilter kit (QIAGEN), quantified, stored, and used as a set.

Analysis of HIV-1 envelope glycoprotein variants.

To examine the expression, processing, and gp120-gp41 association of the mutant HIV-1 envelope glycoproteins, 9 μg of the pSVIIIenv plasmid was cotransfected with 1 μg of a plasmid expressing the HIV-1 Tat protein into 293T cells by using Lipofectamine reagent (Invitrogen) according to the manufacturer's recommendations. Beginning at approximately 16 to 24 h after transfection, the cells were labeled with 200 μCi each of [³⁵S]methionine and [³⁵S]cysteine in methionine- and cysteine-free medium for 16 to 24 h. The secreted gp120 glycoproteins were harvested in the culture medium after a brief centrifugation to remove cell debris. The labeled cells were washed with 5 ml phosphate-buffered saline (PBS) and stripped from the plates with 10 mM EDTA-PBS. The harvested cells were lysed in 1 ml lysis buffer (20 mM Tris-HCl, pH 7.4, 0.5 M NaCl, 0.5% Nonidet P-40, and 1× protease inhibitor cocktail [Pharmacia]) for 30 min at 4°C. Cell lysates were collected after clearing the cell debris by centrifugation at 14,000 rpm in an Eppendorf microfuge for 30 min at 4°C. For immunoprecipitations, 400 μl of medium or cell lysate was incubated overnight at 4°C with 3 μl of pooled sera from HIV-1-infected individuals and 50 μl of protein A-Sepharose (10% in PBS) (Pharmacia) that had been preincubated with 5% bovine serum albumin in PBS. After three washes with 1 ml lysis buffer, the beads were boiled for 5 min in 1× sodium dodecyl sulfate (SDS) sample buffer with 1.5% β-mercaptoethanol. The protein samples were then analyzed on 8% SDS-polyacrylamide gels (Invitrogen) and visualized by autoradiography.

Reporter virus stocks and single-round infectivity assay.

Recombinant HIV-1 encoding firefly luciferase and pseudotyped with the wild-type or mutant envelope glycoproteins was produced as previously described (21, 47). Briefly, 293T cells in 100-mm tissue culture dishes were cotransfected by the Lipofectamine reagent with 2 μg of the pSVIIIenv plasmid expressing the HIV-1 envelope glycoprotein variants, 2 μg of the pCMVΔP1ΔenvpA plasmid, and 6 μg of the pHIV-1Luc plasmid. The pCMVΔP1ΔenvpA plasmid encodes the packaging components (Gag/Pol proteins) and the Tat protein of HIV-1. The pHIV-1Luc plasmid encodes a packageable HIV-1 vector that is defective in all HIV-1 genes except tat and that expresses the luciferase reporter gene. The viral stocks, which are capable of a single round of infection, were harvested 2 days later, aliquoted, and stored at −80°C. The infectivity of recombinant viruses was measured by incubation of the viruses with Cf2Th-CD4/CCR5 cells. Target cells (6 × 10³ cells per well) were seeded into 96-well tissue culture isoplates (EG&G Wallac) and cultured for 16 to 24 h. Viral stocks and the tissue culture medium used for the dilution were prewarmed to 37°C, and serial dilutions of the viral stocks were made. After thorough removal of the media from the target cells, 100 μl of the diluted virus suspension was added to each well. After 48 h, viral infectivity was quantified by measuring the luciferase activity in the target cells with a luciferase detection kit (Pharmingen) and an automated luminometer (Microlumat Plus; EG&G Berthod). For each sample of diluted virus, multiple (four or six) wells of target cells were infected in parallel and the mean value of luciferase activity obtained. All three mutant envelope glycoproteins described above (D368R, R315G/L317S, and L520E) were defective in the ability to support virus entry by at least 4 orders of magnitude relative to the wild-type HIV-1_YU2 envelope glycoproteins (see Fig. 4).

FIG. 4.

Lack of genetic complementation between envelope glycoprotein mutants with defects in different functional modalities. The D368R, R315G/L317S, and L520E mutant HIV-1_YU2 envelope glycoproteins are defective in CD4 binding, CCR5 binding, and membrane fusion, respectively. Equal amounts of plasmids encoding the indicated mutant HIV-1_YU2 envelope glycoproteins were cotransfected in 293T cells to produce luciferase reporter viruses. The infectivity of the resulting viral stocks was measured on Cf2Th-CD4/CCR5 target cells. The infectivities of viruses with single wild-type or mutant envelope glycoproteins were measured in parallel. The means and the ranges of variation in luciferase counts from four parallel wells of infection are shown on a logarithmic scale.

To produce viral stocks with different f_M values, 293T cells were transfected as described above except that the pSVIIIenv plasmids expressing the wild-type HIV-1_YU2 envelope glycoproteins and one derivative mutant were mixed at a given ratio, keeping the total amount of plasmid DNA at 2 μg. To minimize the potential for pipetting errors in creating viruses with very low or high f_M values, the pSVIIIenv plasmids were first diluted so that all pipetting could be performed in the 4- to 20-μl range. All viral stocks associated with one series of different f_M values were prepared and subsequently tested as a set. Recombinant viral production was measured with a standard reverse transcriptase (RT) assay, and all viral stocks had RT levels within 20% of each other (data not shown). The infectivities of recombinant viruses with different ratios (f_M values) of wild-type and mutant envelope glycoproteins were determined. In pilot experiments, we compared variation in infectivity of viral stocks with the wild-type glycoproteins by normalization either to the volume of input virus or to the RT activity. Due to higher variance of infectivity of different viral stocks normalized by RT activity (data not shown), normalization by viral stock volume was employed in all subsequent experiments. The infectivities of the viral stocks associated with a given series of f_M values were expressed as a percentage of the infectivity of the wild-type virus to generate percentages of relative infectivity [RI(%)]. The relationship between f_M and RI(%) values was then compared with the curves predicted by the candidate theoretical models of subunit stoichiometry. Quantitative data fitting was conducted to fit the observed RI(%) data to the values expected by candidate models based on individual mutants and on aggregation of all five mutants.

Data fitting.

The construction of theoretical models and their predictions were explained in “Theoretical model building” above. Four parallel measurements of RI(%) for viruses with f_M values of 0.2, 0.4, 0.6, and 0.8 were collected for individual envelope glycoprotein mutants, and the means of the four primary data points were generated. To analyze the model fitness for each individual mutant, the means of the RI(%) data were used for analysis due to the ease of data handling; the RI(%) values of f_M = 0 and f_M = 1 were excluded from the analysis because they were predetermined as 0% and 100%, respectively, thus contributing no variance to the assessment of fitness. To examine the deviation of the observed RI(%) data [RI(%)^observed] from the RI(%) values predicted by a given model [RI(%)^expected], total modeling variance of a given mutant was calculated by Σ[RI(%)^observed − RI(%)^expected]²/RI(%)^expected for f_M values of 0.2 to 0.8. For each mutant, the total modeling variances were separately generated for models of S=1, S=2, or S=3; the candidate model with the smallest total modeling variance was considered to best describe the observed data.

Simultaneous estimation of S and T values.

We use T to designate the number of HIV-1 envelope glycoprotein trimers required for virus entry. Two independent approaches indicated that each HIV-1 envelope glycoprotein trimer functions independently to mediate virus entry, i.e., T=1 (46, 47). We attempted to simultaneously estimate the T and S values by using the data generated from the loss-of-function mutants in this study without any preset assumptions of possible S and T values. To estimate the S and T values based on the aggregated data from all five envelope glycoprotein mutants, a mixed nonlinear regression model using all four primary RI(%) data points from each of five mutants was developed, thus including 20 independent data points at each level of f_M from 0 to 1.0 (2). Let RI(%)_ij = {1 − [f_M³ + 3f_M²(1 − f_M)(S − 1) + 3f_M(1 − f_M)²(S − 1)(S − 2)/2 − 3f_M²(1 − f_M)(S − 1)(S − 2)/2]}^T × 100 + e_ij + u_i, where i = 1, 2, … 20 and j = 1, 2, … 6. The S and T values are the fixed-effect parameter to be computed, u_i is the random-effects parameter for the ith set, and e_ij is the residual error for the ith set at the jth f_M level. The i parameter designated the 20 individual RI(%) data points at each level of f_M; the j parameter represented six levels of the f_M values. We assumed that u_i and e_ij were independent and both followed a normal distribution. This nonlinear regression model was solved for S and T by use of an SAS procedure, PROC NLMIXED (2). The Newton-Rhapson method was used as an optimization technique in the computation (36a).

Complementation of HIV-1 envelope glycoprotein mutants for function in virus entry and cell-to-cell fusion.

To test whether functional complementation between envelope glycoprotein mutants with defects in different functional modalities can occur during viral entry, 1 μg each of two of the three plasmids expressing D368R, L317S/R315G, and L520E mutant envelope glycoproteins of HIV-1_YU2 gp160, in pairwise combinations, was cotransfected with 2 μg of the pCMVΔP1ΔenvpA plasmid and 6 μg of the pHIV-1Luc plasmid to make viral stocks, as described above. The infectivities of these heterotrimer viral stocks were measured in parallel with those of viruses with singular wild-type or mutant envelope glycoproteins made under the same conditions. The means and ranges of variation of luciferase signals from four wells of parallel infections were compared.

To test whether three different HIV-1 envelope glycoprotein mutants could complement each other for function in a three-way combination by forming heterotrimers with three different mutant subunits, we made luciferase reporter viruses by cotransfecting 0.66 μg each of the three plasmids expressing the D368R, L317S/R315G, and L520E mutant HIV-1_YU2 gp160s with 2 μg of the pCMVΔP1ΔenvpA plasmid and 6 μg of the pHIV-1Luc plasmid. In one set of experiments, these viruses were compared with viruses carrying the wild-type or D368R, L317S/R315G, or L520E mutant envelope glycoproteins of HIV-1_YU2 gp160 individually and those carrying the pairwise combinations of the three mutant glycoproteins, as described above.

To examine the potential of these mutant envelope glycoproteins to complement each other in inducing cell-cell fusion or syncytium formation, 293T cells (the “effector cells”) in six-well plates were transiently transfected with 0.2 μg of the plasmid expressing HIV-1 Tat and 0.9 μg each of the plasmid DNAs expressing two of the three mutant HIV-1_YU2 envelope glycoproteins by the standard Lipofectamine procedure. For comparison, cells were transfected with the plasmids expressing wild-type and mutant HIV-1 gp160 glycoproteins individually and with the pcDNA3.1 vector as a negative control. Approximately 3 × 10³ Cf2Th-CD4/CCR5 cells were seeded in the 96-well plates to serve as the target cells. After 24 h of cocultivation of the effector cells (0.1% of cells transfected in six-well plates) and the target cells in six parallel wells, the level of syncytium formation was quantified by counting giant cells with five or more nuclei under a light microscope. The average numbers and ranges of variation of syncytia in the six parallel wells were compared.

RESULTS

HIV-1 envelope glycoprotein mutants.

Our strategy for investigating the subunit stoichiometry associated with HIV-1 entry involves the functional analysis of viruses containing mixtures of wild-type and defective envelope glycoproteins. Based on existing information on functional motifs within the HIV-1 envelope glycoproteins, we designed mutant HIV-1_YU2 envelope glycoproteins defective in one of the following functions: CD4 binding, CCR5 binding, or fusion peptide activity.

The D368R mutant is altered in aspartic acid 368, which forms a key salt bridge with CD4, and is markedly defective in CD4 binding (26, 42). The D368R HIV-1_YU2 envelope glycoproteins were expressed and processed comparably to the wild-type HIV-1_YU2 envelope glycoproteins (Fig. 2A). Although similar amounts of the D368R mutant and wild-type gp120 envelope glycoprotein were shed into the medium (Fig. 2A), the precipitation of the shed gp120 by soluble CD4-immunoglobulin (Ig) was undetectable for the D368R mutant (Fig. 2B, left panel). The soluble CD4-Ig molecule contains the four N-terminal domains of CD4 fused to the Fc domain of mouse IgG1. These results are consistent with the poor ability of the D368R mutant to interact with CD4.

FIG. 2.

HIV-1_YU2 envelope glycoproteins use an S=2 subunit stoichiometry to support viral infection. (A) Expression and processing of the HIV-1_YU2 envelope glycoprotein variants. Transfected 293T cells expressing the wild-type and mutant envelope glycoproteins were radiolabeled with [³⁵S]methionine and [³⁵S]cysteine for 24 h. The secreted glycoproteins were harvested from the culture medium, and the cell-associated glycoproteins were harvested from lysed cells. The radiolabeled glycoproteins were precipitated with pooled sera from HIV-1-infected individuals and analyzed on 8% SDS-polyacrylamide gels. The gp160 envelope glycoprotein precursors and the mature gp120 glycoproteins are indicated. (B) The receptor-binding abilities of the wild-type and mutant gp120 glycoproteins were assessed as in our previously published studies (24, 28). To test the glycoproteins' ability to bind CD4 in solution, CD4-Ig was incubated with protein A-agarose and culture medium containing ³⁵S-labeled gp120 at room temperature for 3 h. After being washed with the cell lysis buffer, the precipitated glycoproteins were resolved on 8% SDS-polyacrylamide gels (left panel). To measure CCR5 binding, ³⁵S-labeled culture media were concentrated 10-fold by using a Centriprep YM30 filter (Amicon) and then incubated with 3 × 10⁶ Cf2Th-CCR5 cells and 2 μg of soluble CD4 at 37°C for 1 h. After removal of the unbound envelope glycoproteins by three washes with PBS, the target cells were lysed in the cell lysis buffer. The cell-bound glycoproteins in the lysates were then detected by immunoprecipitation with pooled sera from HIV-1-infected persons and resolved by SDS-polyacrylamide gel electrophoresis (right panel). (C) Relative infectivities of viruses with mixtures of the wild-type and mutant HIV-1_YU2 envelope glycoproteins. Recombinant luciferase-expressing HIV-1 with various ratios (f_M) of wild-type and mutant envelope glycoproteins (D368R, L317S/R315G, or L520E) was incubated with Cf2Th-CD4/CCR5 cells. The luciferase activity in the target cells was measured, and the relative infectivity of a given viral stock was calculated by normalization to the infectivity of the viral stock carrying only the wild-type HIV-1_YU2 envelope glycoproteins. The means and the ranges of variation from four or six parallel wells of infection are shown (solid lines). All experiments were performed at least twice with two independent sets of viral stocks in one series of f_M values, and results from a single typical experiment are shown. Also shown are the curves expected for the theoretical models of S=1, S=2, and S=3 (dashed lines).

The L317S mutant of the HIV-1_YU2 envelope glycoproteins has an alteration in the gp120 V3 loop that decreases the efficiency of CCR5 binding (34). The L317S mutant retained some ability to support HIV-1 entry when pseudotyped onto a recombinant luciferase-expressing HIV-1 vector (data not shown). To eliminate this residual function, a second residue change (R315G) was introduced into the V3 loop; the resulting mutant, R315G/L317S, was efficiently expressed but exhibited decreases in proteolytic processing of the gp160 glycoprotein precursor (Fig. 2A). Nonetheless, substantial amounts of gp120 were present in the supernatants of cells expressing the R315G/L317S envelope glycoproteins. The wild-type and R315G/L317S gp120 glycoproteins in cell supernatants were concentrated 10-fold with a Centriprep YM30 filter (Amicon). The concentrated gp120 proteins were incubated with 20 μg/ml soluble CD4 and 3 × 10⁶ Cf2Th-CCR5 cells, which express CCR5. After extensive washing, the cell-bound gp120 was detected by immunoprecipitation. In contrast to the wild-type HIV-1_YU2 gp120 glycoprotein, the R315G/L317S mutant gp120 did not detectably bind CCR5 on the cell surface (Fig. 2B, right panel).

The L520E mutant contains a charged residue in the normally hydrophobic fusion peptide at the gp41 N terminus (17). The L520E mutant envelope glycoproteins were efficiently expressed and processed and exhibited a level of gp120-gp41 association that was comparable to that of the wild-type envelope glycoproteins (Fig. 2A).

Subunit stoichiometry of the HIV-1 envelope glycoproteins.

The infectivity of recombinant HIV-1 viruses containing different ratios (f_M values) of wild-type and mutant envelope glycoproteins was determined by using Cf2Th-CD4/CCR5 target cells. The relationship between RI(%) and f_M for these viruses is shown in Fig. 2C. For the three mutant envelope glycoproteins, the observed curves were consistent with the S=2 theoretical curve.

The subunit stoichiometry of the HIV-1 envelope glycoproteins was examined for another virus strain, HIV-1_HXBc2. The HIV-1_HXBc2 envelope glycoproteins differ substantially from those of HIV-1_YU2; HIV-1_HXBc2 is a well-characterized T-cell line-adapted isolate, uses CXCR4 as a coreceptor, and is very sensitive to neutralizing antibodies (16). The D368R mutant of the HIV-1_HXBc2 envelope glycoproteins has been shown to be defective for CD4 binding (42); the D368R mutant was efficiently expressed and processed compared to the wild-type HIV-1_HXBc2 envelope glycoproteins (Fig. 3A). We screened several existing candidates for an HIV-1_HXBc2 envelope glycoprotein mutant that is defective for CXCR4 binding and is suitable for our purpose (3). The R308L mutant was incapable of supporting virus entry (data not shown) but was expressed and processed relatively efficiently (Fig. 3A). Several attempts to identify candidate HIV-1_HXBc2 mutants that are defective for fusion peptide function and suitable for our experimental design were not successful, due to a low level of residual function in supporting virus entry (F522Y and F522L), abnormal protein processing (L520E), or significant shedding of the gp120 subunits (F522Y/L523S) (data not shown).

FIG. 3.

S=2 subunit stoichiometry of the HIV-1_HXBc2 envelope glycoproteins in virus entry. (A) Expression and processing of the HIV-1_HXBc2 envelope glycoprotein variants. The wild-type and mutant HIV-1_HXBc2 envelope glycoproteins were detected in the culture media and cell lysates as described in the legend for Fig. 2A. (B) Relative infectivities of viruses with mixtures of wild-type and mutant HIV-1_HXBc2 envelope glycoproteins. The relative infectivity of recombinant luciferase-expressing HIV-1 with various ratios (f_M) of wild-type and mutant HIV-1_HXBc2 glycoproteins (R308L or D368R) was measured using Cf2Th-CD4/CXCR4 cells and reported as described in the legend for Fig. 2C.

Viral stocks pseudotyped with the wild-type, D368R mutant, or R308L mutant HIV-1_HXBc2 envelope glycoproteins were produced as described before. The infectivity of these viral stocks was measured on Cf2Th-CD4/CXCR4 target cells, as described above. The two mutant HIV-1_HXBc2 gp160 envelope glycoproteins exhibited decreases of at least 4 orders of magnitude in the ability to support virus entry (data not shown). Viral stocks with a series of f_M values were produced by coexpressing the wild-type and either the D368R or the R308L mutant HIV-1_HXBc2 envelope glycoproteins in individual sets. The relationships between RI(%) and f_M observed for the D368R and R308L mutants were close to that of the S=2 model curve (Fig. 3B). Thus, the S=2 subunit stoichiometry appears to be applicable to the envelope glycoproteins of two phenotypically distinct HIV-1 isolates.

Quantitative model fitting to estimate S values based on individual HIV-1 envelope glycoprotein mutants.

We applied statistical methods to evaluate the fitness of the model predictions for the relationships between the f_M values and the observed RI(%) values. In the theoretical models, the RI(%) values for individual HIV-1 envelope glycoprotein mutants are predicted by the formula (1 − f_M)³, 1 − 3f_M² + 2f_M³, or 1 − f_M³ for models of S=1, S=2, or S=3, respectively (see Materials and Methods). Using these formulae, the expected RI(%) values [RI(%)^expected] for viral stocks with f_M values of 0.2, 0.4, 0.6, or 0.8 were generated (Table 1). For each HIV-1 envelope glycoprotein mutant, the model fitness was judged based on the cumulative deviations of the observed RI(%) [RI(%)^observed] data from the RI(%)^expected values for the four viral stocks with f_M values of 0.2, 0.4, 0.6, or 0.8. The viruses with f_M values of 0 or 1.0, i.e., the viruses with pure mutant or wild-type envelope glycoproteins, had RI(%) values of 0% or 100% by definition; thus, they did not contribute to the variance for the purpose of modeling. The total modeling variance represents the cumulative deviation of the RI(%)^observed values from the model predictions. The total modeling variance for each envelope glycoprotein mutant was calculated by Σ[RI(%)^observed − RI(%)^expected]²/RI(%)^expected for the f_M values of 0.2, 0.4, 0.6, and 0.8. For all five mutant envelope glycoproteins of HIV-1_YU2 and HIV-1_HXBc2, the total modeling variances were the smallest for the S=2 model by very large margins (Table 2).

TABLE 1.

RI(%)^expected values for models with different S values

Model	RI(%)expected value at fM value of:
	0.2	0.4	0.6	0.8
S=1	99.2	93.6	78.4	48.8
S=2	89.6	64.8	35.2	10.4
S=3	51.2	21.6	6.4	0.8

TABLE 2.

Analysis of variance of RI(%)^observed values^a from RI(%)^expected values

HIV-1 strain	Mutant	RI(%)observed value at fM value of:				Total modeling varianceb
		0.2	0.4	0.6	0.8	S=1	S=2	S=3
YU2	D368R	96.1	77.6	43.4	9.3	50.4	5.0	488.7
	R315G/L317S	89.1	62.2	31.5	9.2	71.8	0.6	290.9
	L520E	70.5	55.6	26.4	10.1	88.8	7.6	232.3
HXBc2	D368R	74.4	51.2	22.4	8.7	98.0	10.2	169.7
	R308L	72.4	47.8	23.2	8.4	102.2	12.3	155.8

^aThe mean RI(%)^observed values were calculated from four independent wells of infections, as described in Materials and Methods. The RI(%)^expected values are given in Table 1.

^bFor each HIV-1 envelope glycoprotein mutant, the total modeling variance was calculated using the formula Σ[RI(%)^observed − RI(%)^expected]²/RI(%)^expected for the data points associated with f_M values of 0.2, 0.4, 0.6, and 0.8. The lowest modeling variance for each mutant is highlighted in bold.

Simultaneous estimation of S and T values by use of aggregated data from all five HIV-1 envelope glycoprotein mutants.

Our previous work indicates that the minimal functional unit of HIV-1 envelope glycoproteins in mediating virus entry is composed of a single trimer, i.e., T=1. This conclusion was supported by studying the target size of anti-HIV-1 Env neutralizing antibodies (46) and by using dominant-negative mutants of HIV-1 envelope glycoproteins (47). Furthermore, a T=1 stoichiometry is observed with the envelope glycoproteins of two other viruses, Friend murine leukemia virus and avian sarcoma/leukosis virus type A. Thus, a significant amount of evidence supports the T=1 model for at least some retroviruses. We also wished to estimate the S and T values by using a parametrical statistical method to analyze the data generated from the above-described study of the five envelope glycoprotein mutants. Unlike the above-described analysis of individual HIV-1 envelope glycoprotein mutants, the multivariate analysis was conducted without any assumption about the S and T values. Because of the limited statistical power when only the five means were used for modeling, four primary RI(%) data points from each of the five mutants with f_M values of 0 to 1.0, i.e., 20 independent RI% values at each of the six levels of f_M, were included to enhance the modeling power. According to the general relationships among RI(%), f_M, and S (see Materials and Methods), let RI(%)_ij = {1 − [f_M³ + 3f_M²(1 − f_M)(S − 1) + 3f_M(1 − f_M)²(S − 1)(S − 2)/2 − 3f_M²(1 − f_M)(S − 1)(S − 2)/2]}^T × 100 + e_ij + u_i, where i = 1, 2, … 20 and j = 1, 2, … 5. A mixed nonlinear regression model was developed to solve for the S and T values by use of an SAS procedure, PROC NLMIXED (2). When all 20 primary RI(%) data points at all six values of f_M were included, the model converged nicely on an estimated S value of 1.97 (95% confidence interval, 1.70 to 2.25) and an estimated T value of 0.78 (95% confidence interval, 0.54 to 1.01). This estimation of T is consistent with our previous studies (46, 47), providing further support for the conclusion that individual HIV-1 envelope glycoprotein trimers mediate virus entry independently. The estimation of S is consistent with that derived by model fitting of the data obtained from the studies of individual envelope glycoprotein mutants. Thus, the results support a model in which two wild-type or functional subunits within an HIV-1 envelope glycoprotein trimer are required for the trimer to mediate virus entry.

Absence of complementation between mutants with defects in different functional modalities.

It has been reported that envelope glycoprotein mutants that are defective in different functional modalities could complement cell-cell fusion activity when coexpressed in the same cells by use of vaccinia virus vectors (36). In this context, two subunits of the trimer cannot possess wild-type function to directly fulfill the S=2 subunit stoichiometry. For example, if a CD4-binding-defective mutant is coexpressed with a CCR5-binding-defective mutant, the two kinds of homotrimers will be nonfunctional, and two species of heterotrimers could be formed; one kind of heterotrimer has one wild-type CD4-binding motif and two wild-type CCR5-binding motifs, and the other kind of heterotrimer has two wild-type CD4-binding motifs but one CCR5-binding motif. Thus, neither of the two species of heterotrimer would be comprised of two subunits functional in each of the two modalities, i.e., CD4 binding and CCR5 binding. To fulfill the requirement of the S=2 subunit stoichiometry, cross talk among subunits of the trimer would be required.

To test whether complementation between envelope glycoprotein mutants with defects in different functional modalities can occur during viral entry, the D368R, R315G/L317S, and L520E mutant HIV_YU2 envelope glycoproteins were coexpressed pairwise at a 1:1 ratio to produce viral stocks containing heterotrimers. The infectivities of these heterotrimer-containing viral stocks were compared with those of viruses with homotrimeric wild-type or mutant envelope glycoproteins. The three viruses with homotrimeric mutant envelope glycoproteins exhibited infectivities that were approximately 4 logs lower than that of the wild-type virus (Fig. 4). Viruses with the three pairwise combinations of mutant envelope glycoproteins did not demonstrate infectivities beyond those observed for viruses with the individual mutant envelope glycoproteins (Fig. 4). Similarly, no functional complementation during virus entry was observed with the D368R and R308L mutants of the HIV-1_HXBc2 envelope glycoproteins (data not shown). Thus, these envelope glycoprotein mutants did not appreciably complement one another to support virus entry.

When the three mutants (D368R, R315G/L317S, and L520E) of the HIV-1_YU2 envelope glycoproteins are all coexpressed in equal quantities in cells, a unique kind of heterotrimer that contains one subunit consisting of each of these three mutant glycoproteins could be formed. Theoretically, this specific heterotrimer could comprise up to 22% of the total envelope glycoprotein trimers (3/3 × 2/3 × 1/3, i.e., 6/27, or 22%). This type of heterotrimer would theoretically retain two subunits that can negotiate each of the three functions of the HIV-1 envelope glycoproteins. Thus, there is a remote possibility that these heterotrimers will satisfy the requirement of the S=2 subunit stoichiometry. Hypothetically, if the heterotrimers with three subunits defective in three different functions are functional, such a viral stock should exhibit a level of infectivity close to 22% of that of the wild-type viruses and more than 3 orders of magnitude over those of the viruses with individual mutant envelope glycoproteins. To explore this, we made luciferase reporter viruses by cotransfecting three plasmids expressing the D368R, L317S/R315G, and L520E mutant HIV-1_YU2 envelope glycoproteins in equal quantities. The infectivity of such viral stocks was not significantly greater than that of the viruses with single mutant envelope glycoproteins (Fig. 4). Thus, the existence of two functional motifs in each of the subunits of an HIV-1 envelope glycoprotein trimer was insufficient to allow viral entry, suggesting the lack of efficient intersubunit communication within a trimer.

We also examined the potential of these mutant envelope glycoproteins to complement each other to induce cell-cell fusion or syncytium formation. In these experiments, 293T cells were transiently transfected with pairwise combinations of plasmid DNAs expressing two of the three mutant HIV-1_YU2 envelope glycoproteins to create effector cells. These effector cells were cocultivated with Cf2Th-CD4/CCR5 target cells in 96-well plates, and syncytia were counted 24 h later. 293T cells expressing single wild-type or mutant envelope glycoproteins were used as controls. In all of the cells expressing pairwise combinations of mutant HIV-1 envelope glycoproteins, syncytium formation was not appreciably higher than in the cells expressing the individual mutants (data not shown). Similarly, cotransfecting three different mutant envelope glycoproteins did not result in significant syncytium formation (data not shown). Thus, functional complementation among the mutant envelope glycoproteins in mediating cell-cell fusion was not detected in our experimental setting.

DISCUSSION

Here we provide evidence that at least two of the subunits of an HIV-1 envelope glycoprotein trimer must function to result in detectable virus entry. We conclude that S=2 from fitting the observed entry data of viruses with mixtures of wild-type and mutant HIV-1 envelope glycoproteins to the curves predicted by theoretical models. The theoretical models for this study were generated using some basic assumptions about HIV-1 envelope glycoprotein and its function. These assumptions underlie the construction of these theoretical models and allow us to reach conclusions using this approach. These assumptions are as follows: (i) the wild-type and mutant envelope glycoproteins are expressed in the same cells after cotransfection, the relative levels of the paired envelope glycoproteins are proportional to the ratios of the transfected plasmids, and the monomers of the paired Env proteins form trimers by random association; (ii) each HIV-1 virion particle contains no more than one infectious unit, or, in other words, the infectious units in HIV-1 viral stocks function independently; and (iii) all heterotrimers containing mutant subunits function in an all-or-none fashion. Detailed justifications for the first two assumptions, i.e., random mixing and independence of infectious units, have been published previously (47) and will not be repeated here. Violation of the third assumption would imply that some heterotrimers exhibit partial function. This would result in our underestimation of S. Thus, we considered the possibility that S is actually 3 and that some heterotrimers retain partial function. For the observed data to be explained by the S=3 theoretical model, the heterotrimers containing mutant subunits would need to exhibit significant levels of partial function. This is particularly the case for the data points where the f_M values are high and the observed RI(%) values are readily measurable. For the five viral stocks with f_M values of 0.8, the observed RI(%) values were 9.3, 9.2, and 10.4 for D368R, L317S/R315G, and L520E of HIV-1_YU2, respectively, and 8.7 and 8.3 for D368R and R308L of HIV-1_HXBc2, respectively; these values are at least 10-fold higher than the RI(%) value of 0.8% predicted by the S=3 model. Thus, very high levels of heterotrimer function, approaching that predicted by the S=2 model, are required to explain the data. Finally, a fourth assumption, i.e., that T=1, was used to fit the data from the study of individual HIV-1 envelope glycoprotein mutants. However, we also analyzed the data derived from the study of all five mutants by a mixed nonlinear regression model without making any assumptions about T. The results from the latter analysis support our conclusion that S equals 2 and are consistent with a T value of 1.

The S=2 subunit stoichiometry of envelope glycoprotein trimers in HIV-1 entry specifies the minimum requirement of subunits in support of trimer function. During the normal HIV-1 entry process, all three subunits in an envelope glycoprotein trimer may uniformly contribute to function. We observed no evidence of functional complementation between HIV-1 envelope glycoprotein mutants defective in different entry-related functions. Thus, minimal cross-communication occurs among the trimeric subunits with respect to transmission of receptor-induced conformational changes triggering membrane fusion.

Several different types of envelope glycoprotein inhibitors have been developed as potential therapeutics against HIV-1 infection (27, 35, 44, 49). Neutralizing antibodies block HIV-1 envelope glycoprotein spikes using a 1:1 stoichiometry (46). Such a neutralizing stoichiometry is not necessarily in conflict with the S=2 subunit stoichiometry reported here, given that neutralizing antibodies can function by steric hindrance through their relative bulkiness (33) or by interference with overall structural changes in the trimer complex (31). It is of interest to define the stoichiometry of peptide blockers, e.g., T20, that interfere with the conformational changes in the trimer required for overall envelope glycoprotein function (7). As an inference from the S=2 subunit stoichiometry, molecular blockers that compete with CD4 or CCR5 for binding to individual gp120 subunits may need to bind two or more subunits of the targeted trimer to effectively inhibit its function. Understanding the requirements for inhibition of HIV-1 entry may assist the development of effective therapies or prophylactic approaches.