INTRODUCTION

Top Abstract Introduction Materials and methods Results Discussion References

The process of viral entry is a key step in the initiation of human immunodeficiency virus type 1 (HIV-1) infection. Attachment of the HIV-1 virion to the target cell is mediated by binding of the viral envelope glycoprotein complex to the CD4 receptor on the cell surface (reviewed in reference 49). The envelope glycoprotein (Env) is processed from an oligomeric precursor protein, gp160, into two noncovalently linked subunits: the surface subunit, gp120, which is responsible for binding to CD4, and the transmembrane subunit, gp41, which initiates membrane fusion. In addition to CD4, HIV-1 has also been shown to require an interaction between gp120 and a coreceptor, members of a family of seven-transmembrane G protein-coupled chemokine receptors, on the cell surface in order to mediate membrane fusion (18). Many viruses, such as the influenza virus, are endocytosed following binding to their receptor (35). Their viral glycoproteins then undergo a pH-activated conformational change to a fusion-competent form in the acidic environment of the endosome (13, 23). HIV, however, fuses directly with the plasma membrane at the surface of the target cell in a pH-independent manner (37, 38, 53). In the case of HIV, it has been shown that binding to CD4 induces conformational changes in gp120 and increases the exposure of gp41 epitopes independent of gp120 dissociation (48, 50, 51). Thus, it has been proposed that the envelope glycoprotein of HIV undergoes receptor-induced fusion activation analogous to the pH-induced activation of other viral glycoproteins.

The amino-terminal fusion peptide of gp41 was originally identified by its similarity to functional domains of paramyxovirus glycoproteins (22) and has since been extensively characterized through mutagenesis as the primary fusion domain (3, 19, 52). Based on analogy to similar rearrangements in the glycoproteins of influenza virus (13) and Rous sarcoma virus (24, 27), the CD4-induced conformational change in the HIV-1 glycoprotein is believed ultimately to lead to the insertion of the fusion peptide into the target cell membrane. The details of this conformational change and how the gp120-coreceptor interaction contributes to it are unclear. However, recent studies have suggested that domains of gp41 may also be involved in a conformational change leading to fusion. Mutagenesis of a heptad repeat, leucine zipper-like motif in the ectodomain of gp41 demonstrated that the hydrophobicity of position 573 in the center of the repeat dictated the efficiency with which gp41 mediated fusion (9, 10, 16). It has been hypothesized, based on peptide studies (7, 55-60), that this motif is involved in forming a coiled-coil structure in the glycoprotein oligomer, similar to that formed by an acid-activated fragment of hemagglutinin (HA) (5), following receptor binding.

While it has been shown that the cytoplasmic tail of gp41 is not required for fusion (17, 21, 62, 63), we and others have shown that the fusion event does require that gp41 contain a membrane-spanning peptide anchor, since substitution of a covalently linked lipid anchor for the membrane-spanning domain of gp41 did not support fusion (46, 54). The structural requirements of this membrane-spanning domain, however, are less clear. Truncation of the gp41 carboxy terminus beyond the lysine residue at position 683 (HXB2 strain) is sufficient to cause a loss of membrane anchoring and secretion of the glycoprotein in a soluble form (17, 43), whereas truncation to just after the arginine at position 696 results in a stably anchored protein that is expressed on the cell surface (43), suggesting that the minimal membrane-spanning domain could be residues 684 to 695. Similar results have been obtained for the simian immunodeficiency virus (SIV) Env proteins where residues 691 to 707 are sufficient to anchor the protein in the membrane and mediate incorporation into virions (29). However, these truncated glycoproteins were nonfunctional, and residues C terminal to this minimal anchor domain are important for the function of the glycoprotein (43). These results suggest that while the glycoprotein can be forced to use an abbreviated membrane-spanning domain, such a domain is not sufficient for the completion of the fusion reaction.

The studies described here focus on a membrane-proximal region in the ectodomain of gp41 that is unusually rich in tryptophan residues, several of which are conserved in both primate and ungulate lentiviruses (Fig. 1A and B). This region overlaps a biologically important domain of gp41. Two peptides, DP-178 and SJ-2176, derived from this membrane-proximal region of gp41 have been shown to potently inhibit HIV-mediated fusion (28, 61). A neutralizing human monoclonal antibody, 2F5, has also been described whose epitope overlaps the C terminus of the DP-178 peptide (40, 41). Interestingly, this epitope has been shown to become inaccessible following CD4 binding, suggesting that this region may undergo a change in conformation (51).

View larger version (32K): [in this window] [in a new window]   FIG. 1.   Sequence conservation and predicted structure of the tryptophan-rich region. (A) Schematic representation of gp41, showing the location and sequence variability of the tryptophan-rich region and the overlapping sequences of the 2F5 epitope and DP-178 peptide. Shaded regions indicate the N-terminal fusion peptide and the membrane-spanning domain. The amino acids in the HIV-1 sequence database which occur at each position are denoted underneath the respective amino acid in the HXB2 strain. (B) Alignment of the tryptophan-rich sequence of HIV-1 HXB2 gp41 with homologous regions of HIV-2, SIV, and visna virus glycoproteins. Conserved tryptophan residues are boxed. (C) Helical net projection of the tryptophan-rich region. The outlined region indicates the cluster of conserved hydrophobic residues shown in black.

In order to examine the role of the tryptophan-rich membrane-proximal domain of the HIV-1 gp41 ectodomain in the structure and function of the envelope glycoprotein, we have created deletion, substitution, and insertion mutations in this region by site-directed mutagenesis. Our analyses of the resulting mutant glycoproteins indicate that a stretch of 17 hydrophobic and uncharged amino acids immediately adjacent to the border of the membrane-spanning domain is dispensable for the normal maturation, transport, and CD4-binding ability of the protein but is required for Env-mediated membrane fusion. In contrast to the deletions and multiple point mutations required to abrogate fusion, even subtle mutations of the conserved tryptophan residues had a profound effect on incorporation of the mutant glycoproteins into virions and on virus infectivity. Features of the structure and function of this tryptophan-rich motif are discussed in relation to its possible function in the fusion mechanism and glycoprotein incorporation into virus.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and methods Results Discussion References

Cell culture. COS-1 and 293T cells were obtained from the American Type Culture Collection. HeLa-CD4/LTR--gal cell lines were obtained through the AIDS Reference and Reagent Program, Division of AIDS, National Institute of Allergy and Infectious Diseases, and were originally contributed by Michael Emerman. Cells were maintained in Dulbecco's modified Eagle medium (DMEM) containing 10% fetal bovine serum. HeLa-CD4/LTR--gal cells were additionally maintained in medium containing hygromycin and G418 (Geneticin) as recommended by the contributor.

Mutagenesis and construction of plasmids. Construction of the AK1 and AK2 mutants has been described previously (46). To create the AK1-K2 mutation, constructs containing either the AK1 or AK2 mutation were digested with AvaI, and the resulting fragments of the two constructs were ligated to one another in such a way as to delete the intervening sequence between the two mutations. The +FLAG and +DAF mutations were created by site-directed mutagenesis with the Altered Sites system from Promega.

Protein expression and radioimmunoprecipitation. SV40-based env expression plasmids were transfected into COS-1 cells in 60-mm-diameter plates by using DEAE-dextran (1 mg/ml). At 36 to 48 h posttransfection, the cells were starved in methionine- and cysteine-free DMEM for 20 min and pulse-labeled in methionine-cysteine-free DMEM supplemented with [³⁵S]methionine-cysteine (357 µCi/ml; DuPont-NEN) for 30 min. The labeled cells were then chased in complete DMEM for 3 h prior to harvest of the medium and lysis of the cells. Cells were lysed by a 10-min incubation on ice in lysis buffer (1% Nonidet P-40, 0.1% sodium dodecyl sulfate [SDS], and 0.5% sodium deoxycholate, in phosphate-buffered saline [PBS]). Cellular debris was removed by centrifugation in a microcentrifuge for 5 min at 4°C. HIV-1 glycoproteins were immunoprecipitated from the cell lysate (C) and medium (M) by incubation for 1 h at 4°C with AIDS patient serum. The immune complexes were incubated for 30 min at room temperature with fixed Staphylococcus aureus cells (Staph A) and pelleted in a microcentrifuge. The pellets were washed three times in lysis buffer lacking 0.5% sodium deoxycholate and once in 20 mM Tris-HCl (pH 6.8) and were analyzed by SDS-polyacrylamide gel electrophoresis (PAGE).

CD4 binding assay. Following pulse-chase labeling of transfected COS-1 cells as described above, cells were lysed in lysis buffer lacking SDS. The cell lysate was clarified by centrifugation and divided into two tubes. One-half of the lysate was incubated with patient serum as described above. The other half was incubated overnight with 0.5 µg of CD4-immunoglobulin G (IgG) (Genentech; kindly provided by Mark Mulligan [University of Alabama at Birmingham]) at 4°C. Both samples were immunoprecipitated with Staph A and analyzed as described above.

Cell surface biotinylation. Glycoproteins on the cell surface were biotinylated based on the method of Lisanti et al. (34). Following pulse-chase radiolabeling as described above, transfected COS-1 cells were placed on ice in a cold room and washed three times with ice-cold PBS containing calcium and magnesium (PBS-C/M). The cells were then incubated with PBS-C/M containing 0.5 mg of NHS-SS-biotin per ml for 30 min on ice. The reaction was quenched by incubation of the cells in ice-cold serum-free DMEM for 10 min on ice. The cells were then washed twice with ice-cold PBS-C/M containing 20 mM glycine and lysed in lysis buffer containing 20 mM glycine. Following immunoprecipitation with patient serum as described above, the Staph A pellet was resuspended in PBS containing 2% SDS and boiled for 5 min. The sample was then diluted with PBS to a final concentration of 0.05% SDS and incubated with 40 µl of a 50% slurry of streptavidin-agarose beads (Pierce) at 4°C overnight. The beads were then pelleted, washed, and prepared for SDS-PAGE as described above for immunoprecipitates.

Cell-cell fusion assays. Twenty-four hours posttransfection, COS-1 cells in a six-well plate were trypsinized, mixed approximately 1:10 with untransfected HeLa-CD4/LTR--gal indicator cells, and replated in a six-well plate. Twenty-four hours after replating, fusion of the two cell types was detected microscopically following staining in situ with 5-bromo-4-chloro-3-indolyl--D-thiogalactopyranoside (X-Gal) as described previously (31).

Virus expression and pelleting. Virus expression was assayed by labeling COS-1 cells transfected with pNL4-3 proviral clones, followed by lysis, immunoprecipitation, and SDS-PAGE as described above for the analysis of glycoprotein expression.

Virus entry assay. Sixty to seventy-two hours after transfecting 293T cells as described for virus pelleting, culture supernatants were collected and filtered through a 0.45-µm-pore-size filter. Relative levels of reverse transcriptase activity were determined for each sample as previously described (16), and the volumes were normalized by dilution with complete medium. The normalized supernatants were used to infect HeLa-CD4/LTR--gal cells in duplicate as described previously by Kimpton and Emerman (31). Supernatants were diluted as necessary to remain within the quantifiable range of the assay.

	RESULTS

Top Abstract Introduction Materials and methods Results Discussion References

Mutagenesis of the membrane-proximal region of the HIV-1 transmembrane glycoprotein. To examine the functional role of the tryptophan-rich membrane-proximal region of gp41, deletion, substitution, and insertion mutations were created (Fig. 2, left). Two mutations, AK1 and AK2, were designed previously which created AvaI sites at the coding sequences for the lysines at either the N-terminal (AK1) or C-terminal (AK2) border of the 17-amino-acid tryptophan-rich region (46). These mutations substitute an Asn-Ser-Gly peptide sequence at either end of the tryptophan-rich sequence. In order to determine if the tryptophan-rich region of gp41 was necessary for fusion, a preliminary mutant, AK1-K2, was constructed which contains a deletion of 18 residues, including the entire tryptophan-rich sequence.

View larger version (36K): [in this window] [in a new window]   FIG. 2.   Diagram of mutations and their corresponding fusogenicities. The Env mutations within the tryptophan-rich region are diagrammed in the left panel, juxtaposed to the corresponding name of the mutant and the cell-cell fusion data from a representative experiment. Amino acid changes are underlined, and deletions are denoted by periods. COS-1 cells expressing glycoprotein were mixed 1:10 with HeLa-CD4/LTR--gal cells and replated. The cells were stained 24 h later with X-Gal, and blue syncytia were quantitated microscopically. The average number of syncytia per low-power field was determined by counting six nonoverlapping fields per well for each of two wells and averaging the total. The average number of nuclei per syncytium was determined by quantitating 15 syncytia in each of two wells and averaging the total. Wild-Type (A) and Wild-Type (B) indicate results from two separately prepared plasmid clones.

Expression of envelope glycoprotein mutants. Each of the plasmids containing mutated env genes was transiently expressed in transfected COS-1 cells. The proteins were metabolically labeled in a pulse-chase experiment, and the resulting products were immunoprecipitated from both the cell lysates and the cell medium (Fig. 3). All of the mutant glycoprotein products were expressed at levels similar to those of the wild-type protein and were processed normally to gp120 and gp41, with the sole exception of the initial deletion, AK1-K2. This mutation was found to significantly reduce the efficiency with which the glycoprotein precursor was proteolytically processed (Fig. 3, AK1-K2, lanes C and M). Processing of the 665-682 mutant glycoprotein, however, was not significantly affected (Fig. 3, 665-682, lanes C and M). Quantitation of the proportion of gp120 found in the medium relative to cell-associated gp120 revealed no significant effects of the mutations on gp120-gp41 association relative to wild-type Env (data not shown).

View larger version (97K): [in this window] [in a new window]   FIG. 3.   Expression and maturation of envelope glycoprotein mutants. SV40-based env expression plasmids were transfected into COS-1 cells. Cells were labeled with [35S]methionine-cysteine, and HIV-1 glycoproteins were immunoprecipitated as described in Materials and Methods from the cell lysate (C) or culture medium (M) by using AIDS patient serum for analysis by SDS-PAGE (8% polyacrylamide). Mock, transfected with wild-type env plasmid lacking a eukaryotic transcriptional promoter.

Effects of the mutations on cell-cell fusion. In order to assess the fusogenicity of different constructs, we have found that quantitation of the average number of nuclei per syncytium, rather than the number of syncytia, typically provides a more consistent and reproducible measure of this biological activity. For this study, we used a highly sensitive approach to assay cell-cell fusion. COS-1 cells expressing the mutant glycoproteins as demonstrated in the previous experiments were mixed with HeLa-CD4/LTR--gal (MAGI) cells at a ratio of approximately 1:10. Because the env expression plasmids also express the viral Tat transcriptional activator protein, fusion of the two cell types induces the expression of a -galactosidase reporter gene under control of the viral long terminal repeat (LTR) promoter (31). Because the -galactosidase has been modified to contain a nuclear targeting signal, the nuclei of the resulting syncytium stain a dark blue with X-Gal in situ (31). This simplifies the microscopic quantitation of both the average number of syncytia in a low-power field and the average number of nuclei per syncytium. Immunofluorescent staining of selected cultures scoring negative for fusion in this assay demonstrated that individual COS-1 cells were indeed expressing high levels of glycoprotein but were failing to fuse with the MAGI cells (data not shown).

Ability of the mutant glycoproteins to bind CD4. To determine whether mutations in the tryptophan-rich region affected the ability of the glycoprotein to bind CD4, selected mutant glycoproteins which exhibited reduced levels of fusion or a lack of fusion were immunoprecipitated from metabolically radiolabeled COS-1 cells with CD4-IgG in parallel with patient serum (Fig. 4). All of the mutant glycoproteins bound CD4-IgG at levels similar to those of wild-type glycoprotein, indicating that the membrane-proximal region was dispensable for the efficient binding of the glycoprotein to its primary receptor.

View larger version (48K): [in this window] [in a new window]   FIG. 4.   CD4-binding capability of envelope glycoprotein mutants. Radiolabelled HIV-1 glycoproteins were immunoprecipitated from one-half of the cell lysate with CD4-IgG (bottom panel) and from the other half of the lysate with patient serum (top panel). The immunoprecipitates were analyzed by SDS-PAGE (8% polyacrylamide).

Transport of the mutant glycoproteins to the cell surface. Because the ability of the mutant glycoproteins to form syncytia could be affected by reduced cell surface expression, it was necessary to determine the relative levels of each of the fusion-defective mutant glycoproteins on the surface of transfected cells compared to the level of wild-type glycoprotein. Metabolically radiolabeled cells expressing mutant glycoproteins were treated on ice with the membrane-impermeable, thiol-cleavable biotinylation reagent NHS-SS-biotin. Viral glycoproteins were initially immunoprecipitated from the cell lysate of transfected COS-1 cells by using HIV-positive patient serum to isolate the viral protein from cellular proteins. The immune complexes were then denatured by being boiled in SDS, and a portion of this immunoprecipitate was reserved for analysis in parallel with subsequent samples. The remaining sample was incubated a second time with streptavidin-agarose to isolate just the subpopulation of viral glycoproteins which were exposed to biotinylation reagent on the cell surface.

View larger version (31K): [in this window] [in a new window]   FIG. 5.   Biotinylation of envelope glycoprotein mutants expressed on the cell surface. Proteins on the surface of metabolically radiolabeled glycoprotein-expressing cells were biotinylated with NHS-SS-biotin. Viral proteins were isolated by immunoprecipitation with patient serum, and a portion of the immunoprecipitate was analyzed by SDS-PAGE (top panel). The remaining immunoprecipitate was boiled to denature antibodies and incubated with streptavidin-agarose. Following washing, biotinylated proteins were released by reducing agent and analyzed by SDS-PAGE (8% polyacrylamide) (bottom panel). ERRS is a modified HIV-1 glycoprotein containing an endoplasmic reticulum retrieval signal attached to the C terminus of its cytoplasmic tail.

Expression of the envelope glycoprotein mutants in the context of virus. The hydrophobic residues within the 17-amino-acid membrane-proximal region are all highly conserved, including the five tryptophan residues, which are present in every isolate of HIV-1 sequenced to date (Fig. 1A). Thus, it was surprising that only relatively severe mutations affecting these residues abrogated fusion in the cell-cell assay. In order to determine the effects of these mutations in a more biologically relevant functional assay, representative mutations were cloned into the pNL4-3 proviral construct for analysis of their effects on virus entry.

View larger version (85K): [in this window] [in a new window]   FIG. 6.   Expression of envelope glycoprotein mutants in the context of virus. Radiolabeled viral proteins were immunoprecipitated from COS-1 cells expressing pNL4-3 proviral constructs by using patient serum and analyzed by SDS-PAGE (9% polyacrylamide). The positions of the Pr55gag and p24gag products are indicated to the left.

Effects of the mutations on virus entry. To determine the effect of the mutations within the tryptophan-rich region on virus entry, virus was harvested from cells expressing the mutant proviral constructs and filtered to remove cellular debris. The relative levels of reverse transcriptase activity were determined for each of the virus stocks as a measure of the relative virus concentrations. The viral stocks were then normalized by appropriate dilution with media and used to infect MAGI cells. As with the cell-cell fusion assay above, the use of MAGI cells provides a highly sensitive assay for virus entry. Upon infection of the MAGI cell and expression of viral proteins, Tat induces expression of the -galactosidase reporter gene. After staining in situ 48 h after virus adsorption, individual virus entry events can be scored by the presence of either single blue cells or blue syncytia, both containing intensely stained nuclei. Titration of virus stocks was shown to give linear results in this assay over a range of at least 3 logs, and in situ staining and microscopic quantitation have been shown to be much more sensitive than staining of cell lysates for colorimetric quantitation (31).

View larger version (21K): [in this window] [in a new window]   FIG. 7.   Virus entry and cell-cell fusion in the context of virus. (A) Culture medium from cells expressing virus was filtered and normalized for reverse transcriptase activity. The normalized medium was used to infect duplicate wells (12-well plates) of HeLa-CD4/LTR--gal cells. Cells were stained with X-Gal in situ, and the numbers of blue foci (syncytia or single cells) were quantitated microscopically. Virus entry was quantitated as the average number of blue foci per milliliter by first counting the number of blue foci per 16-mm2 field (low power) for six nonoverlapping fields in each of two wells. The average number of blue foci per field was then multiplied by the total number of fields per well, and the result was corrected for the volume of reverse transcriptase-normalized virus supernatant used. (B) Cell-cell fusion was determined as in the legend to Fig. 2, except that proviral constructs were expressed in 293T cells and mixed with HeLa-CD4/LTR--gal cells. Wild-Type (XUN) [A] and Wild-Type (XUN) [B] indicate results from two separately prepared pNL4-3/XUN plasmid clones. pNL4-3env, pNL4-3 containing a frameshift mutation in gp120.

Incorporation of the envelope glycoprotein mutants into virions. In order to determine whether defects in infectivity reflected a difference in the efficiency with which the mutant glycoproteins were incorporated into virions, the composition of released virions was assessed following metabolic labeling. Supernatants were collected from cells expressing the proviral constructs and filtered to remove cellular debris. Virus was pelleted from the supernatants through a sucrose cushion and solubilized for immunoprecipitation and SDS-PAGE. As can be seen in Fig. 8, the mutant glycoproteins were all detected in the viral pellets at levels much lower than those of wild-type glycoprotein. This demonstrates that even subtle changes in the tryptophan-rich region can dramatically affect incorporation of glycoprotein into virions.

View larger version (47K): [in this window] [in a new window]   FIG. 8.   Incorporation of envelope glycoprotein mutants into virions. Supernatants were collected from metabolically radiolabeled 293T cells expressing pNL4-3 proviral constructs and filtered to remove cellular debris. Virus was pelleted from the filtered supernatants through a sucrose cushion and solubilized in lysis buffer. Viral proteins were immunoprecipitated and analyzed by SDS-PAGE (9% polyacrylamide).

	DISCUSSION

Top Abstract Introduction Materials and methods Results Discussion References

The region proximal to the N-terminal border of the membrane-spanning domain of the HIV-1 glycoprotein is unusually rich in tryptophan residues, containing five that are conserved in all sequenced strains of HIV-1, and is predicted by computer algorithms to form an -helix with a partial amphipathic character (Fig. 1C). Three of these tryptophan residues, the first, third, and fourth in this region, are also completely conserved among sequenced strains of HIV-2 and SIV (Fig. 1B). Two of the tryptophan residues align with the analogous region in the visna virus lentiviral glycoprotein, and a third is present within the membrane-spanning domain of this protein. In addition to the tryptophan residues, the other hydrophobic residues in this region are also very highly conserved among different strains of HIV-1 (Fig. 1A). This region of the HIV-1 glycoprotein also overlaps the C-terminal sequence of a peptide, DP-178, which has been shown to potently inhibit fusion and virus entry (61), as well as the epitope for a neutralizing human monoclonal antibody, 2F5 (Fig. 1A). These factors suggest an important role for the membrane-proximal region in the function of lentiviral glycoproteins.

All of the mutated Env proteins were expressed at levels comparable to those of wild-type Env and, except for AK1-K2, were transported and processed normally. This suggests that all of these proteins folded in a manner similar to that of wild-type Env, since multiple studies suggest that misfolded oligomeric membrane glycoproteins tend not to be transported through the normal secretory pathway but are, rather, retained in the endoplasmic reticulum and degraded intracellularly (14). In addition, Env proteins with mutations which resulted in reduced levels of fusion were maintained on the cell surface at steady-state levels comparable to those of wild-type Env. Thus, the decreased levels of fusion seen with these mutants do not appear to be due to lower concentrations of Env on the cell surface. Finally, mutated Env proteins exhibiting reduced levels of fusion were capable of binding CD4 as efficiently as wild-type Env, indicating that the effect on fusion was independent of primary receptor binding.

Previous analyses by successive truncations of gp41 have demonstrated that amino acid residues 666 to 682 of the HIV-1 transmembrane glycoprotein are not sufficient to anchor the protein in the membrane (17, 43). Our results from the deletion of this region (mutant 665-682) demonstrate further that this region is also not necessary for the stable anchoring of the envelope glycoprotein complex in the plasma membrane. This is consistent with lysine 683 forming the amino-terminal border of the membrane-spanning domain. Our results also demonstrate that residues 666 to 682 are not necessary for the transport and processing or CD4-binding capability of the envelope glycoprotein.

A deletion of 11 residues, which included the entire 2F5 epitope and which overlapped the DP-178 and tryptophan-rich regions, has been shown previously to abrogate fusion (44). This deletion overlapped the site of the +FLAG insertion, which we demonstrate also abrogated fusion. However, since our 666-670 mutation did not abrogate fusion, the effect of the previously reported 11-residue deletion appears not to have been due solely to the deletion of residues within the tryptophan-rich region.

The mutant AK2 is similar to a mutant reported by Helseth et al. in which the lysine which forms the amino-terminal border of the membrane-spanning domain was substituted for by an isoleucine (26). However, unlike that mutant, which essentially abrogated fusion, AK2, which substitutes three amino acids including an asparagine for the membrane-bordering lysine, maintained a significant level of fusogenicity (about 30% of the wild-type level). Since substitution of the hydrophobic residues at positions 684 and 685, which were also substituted in the AK2 mutation, had been shown previously not to affect fusion (20), it is likely that substitution of Lys 683 is responsible for the effect of the Helseth mutation. Perhaps the polarity of the asparagine in this position in the AK2 mutant, together with the adjacent serine, preserves the functionality of this region.

The central seven amino acid residues in the tryptophan-rich region, 671 to 677, appear to be critical for the cell-cell fusion activity of the transmembrane glycoprotein. The reduction of the fusogenicity of the glycoprotein by scrambling these central residues (mutant SC7) suggests that the clustering of hydrophobic residues in an -helical secondary structure as modeled in Fig. 1C may indeed be important, but not essential, for the function of this region. The mutation FN was found to have no significant effect on fusion. This result would argue against a topological requirement for a helical structure, but it does not necessarily preclude the existence of such a structure. Indeed, when the mutated sequence was modeled as a helical net, the conserved hydrophobic residues still tended to cluster together on one face of the helix, although in a somewhat distorted arrangement (data not shown). The lack of major effect caused by the proline substitution for the central tryptophan (W672P) further suggests that a rigid helical structure in this region is not required, although again this does not preclude the existence of a relatively refractory helical structure. Indeed, in the context of membranes, proline residues have been found to stabilize -helical secondary structure, while disrupting -sheet structures (33).

Residues 666 to 670 and 678 to 682 also appear to affect fusion, because deletion of these residues reduced, but did not block, cell-cell fusion. Our results suggest that the specific amino acid sequence of residues 665 to 667 is not critical for cell-cell fusion, since the AK1 mutant was fusogenic. Therefore, it is unlikely that the epitope for the neutralizing monoclonal antibody 2F5 which overlaps these residues is involved in a specific molecular interaction during the cell-cell fusion event. Since the epitope for 2F5 is accessible to antibody, it seemed possible that this epitope might accommodate the additional hydrophilic residues added in the +FLAG mutation and perhaps allow the use of the commercially available anti-FLAG M2 monoclonal antibody (Kodak-IBI) to neutralize virus. However, the +FLAG mutation blocked fusion, perhaps by acting in an analogous fashion to the bound antibody in preventing fusion.

Recent results from structural analyses of gp41 suggest that the inhibitory peptide, DP-178, might act by competing for a binding groove (for a C-terminal heptad repeat) on a coiled-coil trimer formed from an amino-terminal heptad repeat (7, 56). Since 666-670 maintained a reduced level of fusogenicity, it seems likely that the C-terminal portion of DP-178, which overlaps residues 665 to 673, is not absolutely required for the formation of a stable six-helix coiled-coil bundle. However, in an analysis of the inhibitory potential of truncated DP-178 peptides, this overlapping region appeared to be important for the inhibitory function of DP-178 (59). One might expect that DP-178 would inhibit fusion of glycoproteins with mutations in this region as well as, if not better than, the wild-type glycoprotein, since it would now be competing with a truncated C-terminal heptad repeat. Mutants AK1, W(2,3)A, and W666A/W672S were therefore tested for their ability to be inhibited by DP-178 in the cell-cell fusion assay, and all were inhibited completely by the peptide, as was the wild-type protein (data not shown). Recent studies with more sensitive fusion assays indicate that such mutants are indeed more susceptible to inhibition by DP-178 (39).

Although the tryptophan-rich region is predicted to be -helical, modeling the region as a -strand also positions all five tryptophans on the same side of the structure (data not shown). However, in this case, the conserved hydrophobic residues are interspersed with the nonconserved polar residues. Interestingly, the structure of the fusion peptide of gp41 from peptide studies appears to be a -strand until it contacts the lipid bilayer, at which time it transitions to a partly -helical structure (36). Thus, the tryptophan-rich region, which has similar characteristics to the fusion peptide and could interact with membranes, may also undergo a conformational change involving the conserved tryptophan residues. In support of this hypothesis, the 2F5 epitope which overlaps the N-terminal end of this region can no longer be recognized by antibody following CD4 binding (51), suggesting that this region may indeed undergo a conformational change or become sequestered in the membrane during fusion. Thus, fusion-inhibitory mutations within the tryptophan-rich domain may interfere with a late step in membrane fusion, perhaps involving an interaction with, and possibly the disruption of, the membrane in which the glycoprotein is anchored.

Substitution of alanines or other amino acids for the individual conserved tryptophan residues had no significant effect on cell-cell fusion. It is important to note, however, that the overnight cell-cell fusion assay employed here may not reveal subtle variations in the kinetics of the fusion reaction among different mutated proteins. Results from mutations combining multiple tryptophan substitutions demonstrate both the unusual resiliency and intricacy of the function of this region during fusion. The fact that W(1-5)A and W(1-3)A both abrogated syncytium formation suggested that the first three tryptophans were critical for the function of this region, but other combinations of mutations have failed to pinpoint any particular tryptophan residue as critical for the function of this domain. Indeed, recent studies suggest that even the W(1-5)A mutant can induce the formation of small fusion pores but that these are unable to be propagated into syncytia (39).

A striking feature of many of the mutations analyzed was the propensity to inhibit virus entry to levels at the limit of detection, while having only a limited effect on syncytium formation. For example, the W672P and W672F mutants exhibited less than a 20% reduction in fusogenicity compared to wild-type Env, yet reduced the number of virus entry events by almost 90% compared to the wild type. The most likely explanation for this dramatic effect on virus entry is that the reduced concentration of Env protein on the surface of the virion is below a critical concentration required for virus-cell fusion. In contrast, the high concentration of Env protein on the cell surface may facilitate fusion, overcoming a block to fusion which occurs in the context of virus, perhaps by enhancing the recruitment of Env oligomers into fusion pores. Studies with influenza virus HA have demonstrated that variations in the cell surface concentration of fusogenic protein oligomers correspond directly with the efficiency of fusion pore formation (11). W672F and W672P, however, were much more efficient at mediating virus entry relative to the other mutants, even though cell-cell fusion efficiencies and the reduced levels of incorporation appeared similar. This discrepancy could indicate that the virus entry assay is able to discriminate more subtle differences in fusogenicity than the cell-cell assay. It does indicate that these mutants are less defective for fusion in the context of the virus than the other mutants tested.

It is clear from the striking effect of a conservative single amino acid substitution (W672F) on incorporation of glycoprotein into virions and virus entry that the function of this region in the virus life cycle is dependent specifically upon the tryptophan residues which lie within it, rather than simply on the hydrophobic moment of the domain or the presence of an aromatic side chain. This result may explain the absolute conservation of these residues among all strains of HIV-1. Tryptophan residues have been suspected to play an important role in a number of biologic processes (2). One possible role for the conserved tryptophan residues in enhancing the incorporation of glycoprotein into virions could involve an interaction with components of the membrane, such as cholesterol (12). Such interactions could facilitate the enrichment of glycoprotein within certain subdomains of the plasma membrane (4) through which the virus might bud. HIV does contain a higher cholesterol/phospholipid ratio in its viral membrane than is found in most cell membranes (1). An interaction with cholesterol in the target cell membrane might also facilitate membrane fusion. Interestingly, the presence and positioning of tryptophan residues in the bee venom peptide melittin have been shown to be critical for its hemolytic activity (2). Thus, the conserved tryptophan residues could interact with cholesterol in either the membrane in which the glycoprotein is anchored, the target cell membrane, or both. The effects of these mutations within the tryptophan-rich region of HIV-1 Env are similar to those found upon mutagenesis of an analogous region of the Newcastle disease virus fusion protein, which was also relatively refractory to single amino acid substitutions but sensitive to multiple, combined substitutions (45). This region of the Newcastle disease virus fusion protein contains a leucine zipper-like motif and was also hypothesized to interact with membranes during fusion because of its proximity to the membrane.

In an earlier mutational analysis of this region, Cao et al. (6) concluded that the single amino acid substitution W672P resulted in Env proteins which were up to twice as fusogenic as wild-type Env in a syncytium assay. Cell-cell fusion was quantitated by counting the number of syncytia per unit of area in a transfected T-cell culture. We were unable to duplicate these results in syncytium assays using adherent cells. Cao et al. also reported that the effect of W672P and other single amino acid substitutions within this region on virus entry was minimal. Virus entry into T cells was determined by using an env transcomplementation system (25), and complemented virus was produced by cotransfection of an env expression plasmid with an env-defective provirus containing a chloramphenicol acetyltransferase (CAT) reporter gene. Relative levels of entry were determined by measuring the CAT activity in the lysates of T cells following incubation with virus. In a complementation assay such as this, it may be possible to "overload" virions with more Env protein than would normally be incorporated into virus expressed from a complete genome. If such were the case, then the complementation assay might be expected to underestimate the detrimental effects of subtle mutations on virus replication in much the same way as a syncytium assay may underestimate a mutation's effect on membrane fusion. This may explain why the same mutation was much more inhibitory in the context of an intact provirus from our analysis than it was in the complementation system from the previous study.

It is unclear at present how this region, and the conserved tryptophan residues in particular, are functioning in the membrane fusion mechanism. It is also unclear how even conservative mutations within this tryptophan-rich motif dramatically reduced the efficiency with which glycoprotein was incorporated into virions. Interestingly, however, mutations within the proteolytic cleavage site of gp160 have also been shown to reduce glycoprotein incorporation (15), suggesting that the conformation of the glycoprotein ectodomain can affect molecular interactions during virus assembly.