Functional Analysis of the Disulfide-Bonded Loop/Chain Reversal Region of Human Immunodeficiency Virus Type 1 gp41 Reveals a Critical Role in gp120-gp41 Association

Abstract

Human immunodeficiency virus type 1 (HIV-1) entry into cells is mediated by the surface-exposed envelope protein (SU) gp120, which binds to cellular CD4 and chemokine receptors, triggering the membrane fusion activity of the transmembrane (TM) protein gp41. The core of gp41 comprises an N-terminal triple-stranded coiled coil and an antiparallel C-terminal helical segment which is packed against the exterior of the coiled coil and is thought to correspond to a fusion-activated conformation. The available gp41 crystal structures lack the conserved disulfide-bonded loop region which, in human T-lymphotropic virus type 1 (HTLV-1) and murine leukemia virus TM proteins, mediates a chain reversal, connecting the antiparallel N- and C-terminal regions. Mutations in the HTLV-1 TM protein gp21 disulfide-bonded loop/chain reversal region adversely affected fusion activity without abolishing SU-TM association (A. L. Maerz, R. J. Center, B. E. Kemp, B. Kobe, and P. Poumbourios, J. Virol. 74:6614–6621, 2000). We now report that in contrast to our findings with HTLV-1, conservative substitutions in the HIV-1 gp41 disulfide-bonded loop/chain reversal region abolished association with gp120. While the mutations affecting gp120-gp41 association also affected cell-cell fusion activity, HIV-1 glycoprotein maturation appeared normal. The mutant glycoproteins were processed, expressed at the cell surface, and efficiently immunoprecipitated by conformation-dependent monoclonal antibodies. The gp120 association site includes aromatic and hydrophobic residues on either side of the gp41 disulfide-bonded loop and a basic residue within the loop. The HIV-1 gp41 disulfide-bonded loop/chain reversal region is a critical gp120 contact site; therefore, it is also likely to play a central role in fusion activation by linking CD4 plus chemokine receptor-induced conformational changes in gp120 to gp41 fusogenicity. These gp120 contact residues are present in diverse primate lentiviruses, suggesting conservation of function.

Retroviruses enter cells via fusion of viral and cellular membranes, a process that is mediated by the viral envelope glycoprotein (Env) complex. The Env complex is a hetero-oligomer comprising a surface-exposed subunit (SU), which mediates viral attachment by binding to cellular receptor(s). The transmembrane (TM) protein anchors the Env complex to the viral envelope and infected cell surface and is responsible for membrane fusion. The functional gp120 (SU)-gp41 (TM) complex of human immunodeficiency virus type 1 (HIV-1) is derived from an inactive precursor (gp160) following cleavage by cellular convertases in the Golgi (26, 39). gp120 first binds to the primary receptor, CD4 (15, 31, 34), resulting in the creation of a high-affinity binding site for one or more coreceptors, usually CXCR4 (21) and/or CCR5 (2, 17, 18, 52, 58). These binding events trigger conformational changes in gp41 that correlate with membrane fusion activity (22, 29).

The linear organization of the gp41 ectodomain includes an N-terminal fusion peptide, a coiled-coil-forming sequence, a disulfide-bonded loop region, and a C-terminal α-helical segment. The ectodomain is anchored in the viral envelope by an ∼20-residue membrane-spanning sequence which precedes an ∼150-residue cytoplasmic domain. The three-dimensional structures of HIV-1 and simian immunodeficiency virus (SIV) gp41 core fragments lacking the fusion peptide, disulfide-bonded loop, and membrane-spanning sequence have been solved by X-ray crystallography and nuclear magnetic resonance. These studies revealed trimeric rods composed of a central coiled coil and an antiparallel C-terminal α-helix packed against the outside of the coiled coil (6, 11, 37, 51, 55, 60). This trimeric helical hairpin structure is conserved in TM protein core fragments from retroviruses, human T-lymphotropic virus type 1 (HTLV-1) (32) and murine leukemia virus (MuLV) (20), and other enveloped viruses, including influenza virus (5, 13), Ebola virus (38, 53), simian virus 5 (3), and respiratory syncytial virus (61). Helical hairpins are similar to the fusion-pH-induced conformation of the influenza virus TM protein, HA2, placing the hydrophobic N-terminal fusion peptide and C-terminal transmembrane sequence at the same end of the rod in a fusion-activated conformation (for reviews see references 12, 50, and 54).

The results of X-ray crystallographic and biochemical studies indicate that the viral envelope glycoproteins described above exist in at least two conformations, a prefusogenic metastable conformation which is converted to a thermostable, fusogenic hairpin conformation following an activation trigger (5, 9, 12, 13, 48, 50, 57). The activation trigger for HA2 is endosomal (low) pH, while cellular receptor-SU interactions induce retroviral fusion (16, 28, 29). Fusion activation of HA2 induces the relocation of the fusion peptide from the glycoprotein core to the tip of the central coiled coil for insertion into the target membrane and refolding of the C-terminal segment for antiparallel packing on the outside of the coiled coil (5, 14). The refolding of HA2 into a hairpin structure would relocate the membrane-spanning sequence to the same end of the rod as the fusion peptide, drawing together the viral and cellular membranes for fusion. The idea that retroviral TM hairpins also correspond to a fusion-activated conformation is consistent with the observations that synthetic peptides corresponding to the C-terminal helical segments of HIV-1, SIV, and HTLV-1 TM hairpins are potent inhibitors of Env-mediated fusion and viral entry (30, 33, 37, 49, 56). The gp41 C-terminal peptides appear to inhibit membrane fusion by binding to a CD4-induced site in the central coiled coil of a gp120-gp41 fusion intermediate (10, 22, 47), blocking formation of the gp41 hairpin.

Formation of the HA2 fusion-activated hairpin involves refolding of the C-terminal portion of the central coiled coil and the creation of a chain reversal region that is stabilized by hydrophobic interactions, connecting the coiled coil and antiparallel C-terminal segment (5). The structure of the HA2 chain reversal region differs significantly from the structures of homologous regions in HTLV-1 (32), MuLV (20), and Ebola virus (38, 53) TM hairpins. In HTLV-1 gp21, this structure comprises a 3₁₀-helix, a conserved Gly-Gly motif, a short α-helix, and the disulfide-bonded loop. Similar chain-reversing structures are observed in p15E and GP2, which are packed perpendicularly to the coiled coil through hydrophobic and polar interactions (20, 32, 38, 53).

By using structure-based site-directed mutagenesis, we found that the HTLV-1 gp21 disulfide-bonded loop/chain reversal region plays a critical role in fusion activity (36). A conserved hydrophobic cluster formed by Leu-385, Leu-386, and Phe-387 at the base of the HTLV-1 gp21 coiled coil, a conserved Gly-Gly hinge motif preceding the disulfide-bonded loop, and a salt bridge formed between Glu-398 in the disulfide-bonded loop and Arg-380 near the base of the coiled coil (Fig. 1A) are all required for wild-type levels of HTLV-1 Env fusion function. Our data also suggested that a solvent-exposed aromatic residue C terminal to the disulfide bonded loop, Phe-402 (Fig. 1A), may originate from the core of the prefusogenic Env complex to occupy a solvent-exposed position following fusion activation. Based on these observations, we proposed a model whereby the chain reversal region transmits a conformational signal from receptor-bound SU to induce the fusion-activated TM protein hairpin conformation.

FIG. 1.

Structural comparison of chain reversal regions from HTLV-1 gp21(354–416) (Protein Data Bank [PDB] accession number 1MG1) (32) (A) and HIV-1 gp41 (559–644) modeled on the SIV_MAC gp41 crystal structure (PDB accession number IQBZ) (60) (B). The structures have been aligned according to the hydrophobic 4-3 repeat sequences of the coiled-coil-forming helices. Coiled-coil-forming helices and C-terminal helices are shown as ribbons colored pink and light blue, respectively, while the rest of the chain is colored grey. The disulfide-bonded loop of gp21 is colored yellow. The region of gp41 that would comprise the disulfide-bonded loop is also shown in yellow. The side chains of residues mutated in this study of gp41 and in a previous study of gp21 (36) are indicated, as are the side chains of some residues interacting with the mutated amino acids. The HIV-1_HXB2R gp160 numbering system is used for gp41. To convert to HIV-1_BH8 numbering, subtract 5; to convert to SIV_MAC numbering, add 13. These panels were prepared using RIBBONS 2.8 software. (C) Structure-based sequence comparison of the chain reversal/disulfide-bonded loop region from HTLV-1 gp21 (residues 374 to 415), MuLV p15E (536 to 581), Ebola virus (Zaire) GP2 (582 to 628), and selected primate lentivirus TM proteins (corresponding to HIV-1 gp41 residues 579 to 644). The lentiviruses used in the alignment include HIV-1_BH8 (group M, GenBank accession number K02011), HIV-1_YBF30 (group N, AJ006022), SIV_CPZGAB (X52154), HIV-1_ANT70 (group O, L20587), SIV_MAC239 (M33262), SIV_MND (S28084), SIV_AGMTAN (U58991), and SIV_SYK (AAA74712). α-helical regions as observed in the crystal structures of HTLV-1 gp21, HIV-1_HXB2R, and SIV_MAC gp41 are shown as cylinders (coiled-coil region, pink; C-terminal helix, light blue). The HIV-1_HXB2R gp160 numbering system is used for the lentiviral gp41 sequences. Residues identical to those of HIV-1_BH8 are highlighted in black; homologous residues are grey. Basic residues in the coiled coil and disulfide-bonded loop are highlighted in blue, and acidic residues are red. The hydrophobic clusters of gp21, p15E, and GP2 are highlighted in green, conserved glycines are dark grey, and the conserved aromatic residue C terminal to the disulfide bridge is light green. Deletions are indicated by dots. The gp41 residues mutated in this study are indicated by asterisks. As a guide, some HTLV-1 gp21 and HIV-1 gp41 residues are numbered.

The HIV-1 gp41 disulfide-bonded loop appears to also play an important role in Env fusion function, since random mutagenesis of the loop sequence can decrease or block Env fusion activity and HIV-1 entry competence (40). However, despite this functional parallel, comparisons of gp21 and gp41 disulfide-bonded loop/chain reversal regions indicate both structural and sequence differences (Fig. 1). These differences prompted a closer examination of the functional role of the HIV-1 gp41 disulfide-bonded loop/chain reversal region. By using the structural and functional information obtained previously for HTLV-1 gp21 (36) to guide a comparative mutational analysis of gp41, we found that the gp41 disulfide-bonded loop/chain reversal region plays a central role in gp120-gp41 association.

MATERIALS AND METHODS

Cells and virus.

293T cells were maintained in Dulbecco's modification of minimal essential medium–10% fetal calf serum (complete medium). Wild-type and mutant Env glycoproteins were expressed in 293T cells (250,000 cells per 4.5-cm² well of Linbro 12-well culture dishes; ICN Biomedicals Inc., Aurora, Ohio) by using the vaccinia virus-T7 RNA polymerase transient-expression system (41). Briefly, 293T cells were transfected with expression constructs by the Fugene procedure (Boerhinger Mannheim GmbH, Mannheim Germany) and at 5 h posttransfection were infected with the recombinant vaccinia virus vTF7.3, which directs expression of bacteriophage T7 RNA polymerase. The recombinant vaccinia virus vTF7.3 was obtained from T. M. Fuerst and B. Moss (41).

Plasmid constructs.

The construction of pTMenv.2, which directs T7 promoter-driven expression of the env open reading frame (BH8 clone of HIV-1_LA1 [46]), is described elsewhere (45). Mutations were introduced into an EcoRI-HindIII fragment (HIV-1_BH8 env nucleotides 1141 to 1902) by PCR mutagenesis. The sequences of Env mutants were confirmed by the ABI Prism BigDye terminator system (Applied Biosystems). For some mutants, the T7 expression cassette of pTM.1 was ligated into the EagI site of wild-type and mutated pTMenv.2 vectors in order to attenuate Env expression levels in transfection experiments. A dual expression vector for cytomegalovirus promoter-driven CD4 expression and bacteriophage T7 promoter-driven firefly luciferase expression. pT4luc, was prepared for use in cell-cell fusion assays. A fragment containing the CD4 open reading frame was excised from pT4B, which was obtained from Richard Axel (35), using BamHI and EcoRI for ligation into the BamHI/EcoRI sites of pUC18. The EcoRI/XbaI fragment containing the CD4 open reading frame was then ligated into the multiple-cloning site of pCDNA3 (Invitrogen) to give pCDNA.T4. The oligonucleotide primers 5′-GGTATCGATGACGGCCGGTTCTTTCC and 5′-CTATTTTTCCTTCGTCGGCCGTACGCTC were used to PCR amplify the T7 promoter-driven firefly luciferase expression cassette from pTMluc (36), which was cloned into the NruI site of pCDNA.T4 to yield pT4luc. pc.FUSIN (17), which directs CXCR4 expression from a cytomegalovirus promoter, was obtained from N. Landau.

Antibodies.

Human antibodies to gp120 were purified from pooled HIV-1-positive human plasma by affinity chromatography on a recombinant gp120 (HIV-1_SF2 strain)-Sepharose CL4B column (human anti-rgp120) as described previously (44). Anti-gp41 monoclonal antibodies (MAbs) obtained through the AIDS Research and Reference Reagent Program, National Institute for Allergy and Infectious Diseases, include 126-6, 98.6, and 50-69 from S. Zolla-Pazner (59), MD-1 from R. Myers (R. Myers, T. Meiller, W. Falkler, Jr., J. Patel, and J. Joseph, Abstr. 93rd Gen. Meet. Am. Soc. Microbiol. 1993, abstr. T-70, 1993), and C8 from G. Lewis (1). Immunoglobulin G (IgG) was purified from the plasma of an HIV-1-positive individual using protein A-Sepharose (Pharmacia Biotech).

Western blotting.

At 16 h postinfection, transfected and infected 293T cells were lysed for 10 min on ice in phosphate-buffered saline (PBS) containing 1% Triton X-100, 0.02% sodium azide, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 5 μg of aprotinin ml⁻¹, 5 μg of leupeptin ml⁻¹, and 1 mM dithiothreitol. Lysates were clarified by centrifugation at 10,000 × g at 4°C prior to polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate (SDS-PAGE) in 7.5-to-15% polyacrylamide gradient gels under reducing conditions. Proteins were transferred to nitrocellulose prior to Western blotting with ¹²⁵I-labeled MAb C8, which was prepared by the chloramine T procedure (25).

Biosynthetic labeling and immunoprecipitation.

At 16 h postinfection, transfected and infected 293T cells were incubated for 30 min in cysteine and methionine-deficient medium (ICN, Costa Mesa, Calif.) and then labeled for 45 min with 150 μCi of Tran³⁵S-label (ICN). The cells were then washed and chased in complete medium for 4.5 h prior to lysis. All incubations were performed at 37°C. Cell lysates and clarified culture supernatants were immunoprecipitated with various human antibodies as described previously (45).

Surface biotinylation.

At 16 h postinfection, infected and transfected cells were chilled on ice, washed twice with ice-cold PBS and then incubated with 1.25 mM sulfo-NHS-long-chain biotin (Pierce) in PBS for 30 min on ice. The biotinylated cells were washed once with ice-cold PBS and quenched with 50 mM glycine–PBS (500 μl) for 30 min on ice. The cells were lysed by the addition of 170 μl of ice-cold 4% Triton X-100–2.4 M KCl–0.2 M Tris HCl (pH 7.4), clarified by centrifugation at 10,000 × g for 10 min at 4°C, and precleared with bovine serum albumin-Sepharose CL4B for 16 h at 4°C. Envelope proteins were immunoprecipitated from the precleared cell lysates with IgG from an HIV-1-positive individual and protein G-Sepharose (Pharmacia Biotech). Biotinylated Env proteins were visualized following SDS-PAGE on 4-to-12% polyacrylamide gels under reducing conditions, transfer to nitrocellulose, blotting with neutravidin-horseradish peroxidase (Pierce Chemical Co.) and chemiluminescence (Boehringer Mannhein GmbH).

Luciferase reporter assay for cell-to-cell fusion.

293T effector cells were transfected with Env expression constructs and then infected with vTF7.3 at 5 h posttransfection. 293T target cells were cotransfected with pT4luc and pc.FUSIN. At 24 h posttransfection, effectors and targets were resuspended in 1 ml of complete medium containing 1 μg of actinomycin D per ml and 40 μg of cytosine arabinoside per ml and cocultured for a further 16 h at 37°C. Cells were then lysed and assayed for luciferase activity using the Promega (Madison, Wis.) luciferase assay system.

RESULTS

Comparison of HTLV-1, HIV-1, and SIV TM protein disulfide-bonded loop/chain reversal regions.

The crystal structures of HTLV-1 gp21 (32) and SIV_MAC gp41 (60) disulfide-bonded loop/chain reversal regions are shown in Fig. 1A and B, respectively. In HTLV-1 gp21, the disulfide-bonded loop/chain reversal region forms a compact substructure that is packed roughly perpendicular to the coiled coil, connecting the coiled coil to the antiparallel C-terminal segment. Very similar structures mediate the chain reversal in MuLV p15E (20) and Ebola virus GP2 (38, 53). However, the structural details of the homologous region in HIV-1 and SIV_MAC gp41 are incomplete (6, 11, 37, 51, 55, 60). The disulfide-bonded loop has been deleted from HIV-1 gp41 constructs used in three-dimensional structure analyses (11, 37, 51, 55), whereas the cysteines that would form the disulfide bridge have been replaced with alanine in SIV_MAC gp41 constructs (Fig. 1B) (6, 60).

To help identify functionally relevant residues of the gp41 chain reversal region, we compared HTLV-1 gp21, MuLV p15E, and Ebola virus GP2 amino acid sequences with those of representative primate lentiviruses (Fig. 1C). (The numbering system of HIV-1_HXB2R is used for gp41 in this paper. To convert to HIV-1_BH8 numbering, subtract 5; to convert to SIV_MAC numbering, add 13.) Whereas conserved gp41 residues appeared to coincide in sequence with amino acids associated with conserved gp21, p15E, and GP2 structural features (Fig. 1C), the three-dimensional context of these gp21 and gp41 residues differed markedly (Fig. 1A and B). (We have used the more complete crystal structure of SIV_MAC gp41 [60] as a homology model in this study. SIV_MAC and HIV-1_HXB2R gp41 structures are very similar, with root mean square deviations in main-chain atoms of less than 1 Å.). Whereas the side chains of HTLV-1 gp21 hydrophobic cluster residues (Leu-384, Leu-385, and Phe-386) are orientated into the center of the coiled coil, capping the C terminus of the coiled coil, the apparently coincident gp41 residues, Leu-593 and Trp-596, are part of a helical extension of the coiled coil, their side chains pointing into solvent. We also examined the conserved charged residues found within the disulfide-bonded loop of most retroviral TM proteins. In HTLV-1, MuLV, and Ebola virus TM proteins, the acidic residue in the disulfide-bonded loop forms a salt bridge with a basic residue near the base of the coiled coil (e.g., Glu-398 and Arg-380 in HTLV-1 gp21) (Fig. 1A). Most lentiviruses (except for SIV_MND) contain one or more basic residues within the disulfide-bonded loop, but ion pairing with conserved acidic residues (Glu-584 and Asp-589) of the coiled coil is unlikely, since the acidic residues take part in intermonomer (Glu-584–Arg-579) and intramonomer (Asp-589–Lys-588) salt bridges (Fig. 1B). The more C-terminally located Phe-402 of gp21 is solvent exposed, but Trp-610 of gp41 forms a hydrophobic cluster with Val-608 and Tyr-586. In order to better understand the functional role of the gp41 chain reversal/disulfide-bonded loop region, we introduced conservative substitution mutations at these and other less conserved sites and tested their effects on HIV-1_BH8 Env protein synthesis, processing, cell-cell fusion activity, gp120-gp41 association, cell surface expression, and antigenic structure.

Synthesis and processing of HIV-1 Env mutants.

The synthesis and processing of HIV-1_BH8 Env mutants in transfected and infected 293T cells was assessed by Western blotting with MAb C8, which is directed to an epitope in the gp41 cytoplasmic domain (1). Similar levels of gp160 and processed gp41 were observed for wild-type and mutant Env glycoproteins (Fig. 2). These results are consistent with the folding of mutants into translocation-competent forms that are cleaved in the Golgi in a similar manner to wild-type Env.

FIG. 2.

Synthesis and processing of HIV-1 Env glycoprotein mutants in 293T cells. Env-expressing 293T cells were lysed at 16 h posttransfection and subjected to reducing SDS-PAGE in 7.5-to-15% polyacrylamide gradient gels. Proteins were transferred to nitrocellulose followed by Western blotting with MAb C8, directed against an epitope in the cytoplasmic domain of gp41. Env proteins were visualized following scanning in a PhosphorImager SF. The asterisk indicates a proteolytic product of gp160. The figure was prepared using Powerpoint software and is representative of three independent experiments. wt, wild type.

Cell-cell fusion activity of gp41 mutants.

We next screened the Env mutants for cell-cell fusion activity using a luciferase reporter assay. Figure 3 shows that alteration of the conserved acidic residues Glu-584 (E584Q, E584D, and E584N) and Asp-589 (D589K) resulted in nonfunctional Env glycoproteins. The interhelical Arg-579–Glu-584 and intrahelical Lys-588–Asp-589 salt bridges observed in the gp41 crystal structures (Fig. 1B) are therefore essential for Env fusion competence. By contrast, substitutions at Lys-601 within the disulfide-bonded loop were tolerated to various degrees, with the K601E, K601Q, and K601A mutants maintaining between 35 and 75% of wild-type fusion activity. However, a basic residue appears to be preferred at this position, since the K601H mutant retained full fusion activity.

FIG. 3.

Cell-cell fusion activities of Env glycoprotein mutants using a luciferase reporter assay. 293T effector cells were transfected with wild-type (wt) and mutated pTMenv.2 expression vectors and then infected with vTF7.3, while target 293T cells were cotransfected with pT4luc and pc.FUSIN. At 16 h postinfection, effectors and targets were mixed and cocultured for 16 h prior to lysis and assay for luciferase activity. The relative fusion activities of Env proteins are expressed as follows: (ratio of luciferase activity induced by mutant Env to luciferase activity induced by wild-type Env) × 100. The means ± standard errors of the means from three independent transfections are shown.

Analysis of the hydrophobic residues, Leu-592, Leu-593, and Trp-596 that are located N terminal to the disulfide bridge revealed that Val and Ala substitutions were tolerated at the variable Leu-592 position but not at the conserved Leu-593 site. Replacement of Trp-596 with a smaller aromatic residue, Phe, led to an approximately 50% reduction in fusion activity, while mutation to smaller residues (His or Leu) abolished fusion activity. Fusion function was also abolished when Leu-593 and Trp-596 were switched in the L593W/W596L double mutant, indicating that simply maintaining hydrophobicity at this site is not sufficient for fusion competence. Introduction of Lys and Ala substitutions at the partially conserved polar site (Gln-591) led to an approximately 50% reduction in fusogenicity.

HTLV-1 gp21 contains an essential Gly-Gly motif N terminal to the disulfide bridge, whereas gp41 contains a single conserved glycine at this position (Fig. 1C). The gp41 glycine (Gly-597) is also essential for fusion function, since substitution with residues that allow various degrees of backbone flexibility, Pro < Ala < Ser, led to the abolition of function. Finally, replacement of the conserved Trp-610 residue, located C terminal to the disulfide-bonded loop, with Phe or His also abolished fusion activity.

gp120-anchoring abilities of selected gp41 mutants.

We selected one or more mutants from each group for further characterization. To facilitate the biochemical analyses, we used the pTMenv.2/EC series vectors for attenuated Env protein expression and more efficient intracellular transport and processing. Infected and transfected cells were pulse-labeled with Tran³⁵S-label for 45 min and then chased for 4.5 h to allow intracellular transport of newly synthesized Env glycoproteins (19). Consistent with previous studies (27), both gp160 and gp120 were immunoprecipitated by the anti-rgp120 antibody from wild-type-Env-transfected cell lysates, whereas only gp120 was obtained from clarified culture supernatants (Fig. 4). Analysis of the mutants revealed three phenotypic groups: E584Q, Q591A, and W596F retained normal levels of cell-associated gp120; D589K, K601E, and W596L retained lower levels of cell-associated gp120; the other mutants in Fig. 4 had shed all of their gp120 into culture supernatants. These results confirmed that all but three gp41 mutants (E584Q, Q591A, and W596F) had defective gp120-anchoring abilities. These results also indicate that the targeted residues (except for Glu-584 and Gln-591) play a pivotal role in gp120-gp41 association and that in most cases, loss of fusion function is due to an unstable gp120-gp41 complex. The results of the fusion assay (Fig. 3) suggested that the basic residue (Lys-601) in the disulfide-bonded loop was required for full fusion activity. Figure 4 indicates that the gp120-anchoring abilities of Lys-601 mutants correlate approximately with their fusion activity, with K601E (35% fusion competent) anchoring low levels of gp120 whereas K601H (100% fusion competent) and K601Q (75% fusion competent) have a wild-type gp120-anchoring phenotype. This result indicates that the basic residue in the disulfide-bonded loop is involved in a stable gp120-gp41 association that is required for full fusion activity.

FIG. 4.

gp120-anchoring ability of gp41 mutants. Env-expressing 293T cells were labeled with Tran³⁵S-label for 45 min and chased in complete medium for 4.5 h before lysis. Cell lysates (C) and clarified culture supernatants (S) were immunoprecipitated with human anti-rgp120 and protein G-Sepharose. gp160 and gp120 bands were visualized following SDS-PAGE in 4-to-12% gradient gels under reducing conditions and scanning in a PhosphorImager SF. The relative fusion activities of wild-type and mutated Env glycoproteins (means ± standard errors of the means, taken from Fig. 3) are shown below the corresponding gel lanes. The figure was prepared using Powerpoint software and is representative of three independent experiments.

Cell surface biotinylation of Env mutants.

We confirmed the cell surface expression of the selected mutants by treating transfected and infected cells with a membrane-impermeant biotinylation reagent, followed by immunoprecipitation with IgG from an HIV-1-positive individual and blotting with neutravidin-peroxidase. Figure 5 indicates that substantial amounts of surface-expressed gp160 and gp120 were detected on wild-type-Env-transfected cells. By contrast, gp160 but not gp120 was detected for the cleavage site mutant, EnvC−/− (K500N/K502I/R503S/R504S/R508S/K510N/R511S), consistent with blocked processing. Analysis of the mutant panel revealed the absence of surface-localized gp120 for all but two mutants. Normal levels of surface-localized gp120 were retained by Q591A, which was also found by radioimmunoprecipitation assay (Fig. 4) to anchor wild-type levels of gp120. In contrast to the radioimmunoprecipitation results, very little (W596F) or no (E584Q and W596L) surface-expressed gp120 was detected by biotinylation for these mutants. These differences may be due to the shedding of gp120 from unstable E584Q, W596F, and W596L mutant gp120-gp41 complexes following the washing, biotinylation, and quenching procedure. These data indicate that the mutants tested are transport competent, since wild-type levels of gp160 were detected at the cell surface; however, all but two mutations had led to undetectable levels of surface-expressed gp120.

FIG. 5.

Cell surface expression of Env glycoprotein mutants. 293T cells were transfected with wild-type (wt) and mutated pTMenv.2/EC vectors followed by infection with vTF7.3. At 16 h postinfection, cells were washed twice with PBS, incubated with 1.25 mM sulfo-NHS-long-chain biotin for 30 min, and then quenched with 50 mM glycine–PBS. Envelope proteins were immunoprecipitated from precleared cell lysates with IgG from an HIV-1-positive individual and protein G-Sepharose. Biotinylated Env proteins were visualized following SDS-PAGE on 4-to-12% polyacrylamide gradient gels under reducing conditions, transfer to nitrocellulose, and blotting with neutravidin-horseradish peroxidase. The figure was prepared using Powerpoint software from a single experiment.

Probing the antigenic structure of selected mutants with anti-gp41 conformation-dependent MAbs.

The conformation of the gp41 domain was further assessed by using the human MAbs MD-1, 126-6, and 98-6, which bind to conformational epitopes in the C-terminal portion of the gp41 ectodomain (23, 59; Myers et al., Abstr. 93rd Gen. Meet. Am. Soc. Microbiol.), and MAb 50-69, which maps to a site that includes the disulfide-bonded loop (59). The MAbs MD-1, 126-6, 98-6, and 50-69 are able to detect structural disturbances in gp41, as mutations in the gp41 coiled-coil sequence that block gp160 oligomerization also block epitope formation (45). The MAbs 126-6, 98-6, and 50-69 can also recognize the putative fusion-activated hairpin form of gp41 (24). Figure 6 shows that all mutants tested were efficiently immunoprecipitated by the MAbs, indicating that the gp41 domain of the mutants acquired a global conformation similar to that of the wild-type gp41 domain. Because the mutants were processed normally, expressed at the cell surface, and recognized by conformation-dependent MAbs, we infer that the loss of gp120-anchoring ability is due to the alteration of a critical contact in the gp120-gp41 interface or due to a localized structural alteration at the mutation site which does not affect the global conformation of the gp41 domain.

FIG. 6.

Recognition of Env mutants by conformation-dependent anti-gp41 MAbs. Env-expressing 293T cells were labeled with Tran³⁵S-label for 45 min and chased in complete medium for 4.5 h before lysis. Cell lysates were immunoprecipitated with the various anti-gp41 MAbs or human anti-rgp120 using protein G-Sepharose. gp160 bands were visualized following SDS-PAGE in 4-to-12% gradient gels under reducing conditions and scanning in a PhosphorImager SF. The figure was prepared using Powerpoint software.

DISCUSSION

Conservative substitutions in the HIV-1 gp41 disulfide-bonded loop/chain reversal region adversely affected gp120-gp41 association and cell-cell fusion activity. These effects were not due to folding defects in the mutated Env proteins, since glycoprotein maturation appeared normal, with the mutants being processed, expressed at the cell surface, and recognized by conformation-dependent MAbs in a manner similar to that of wild-type Env. The gp120 association site includes conserved aromatic and hydrophobic residues, Leu-593, Trp-596, Gly-597, and Trp-610, on either side of the disulfide-bonded loop and a basic residue, Lys-601, within the loop. Because this region has a central role in gp120 contact, it may link CD4 and chemokine receptor-induced conformational changes in gp120 to gp41 fusion activation. These findings contrast those obtained in a similar study of HTLV-1 gp21, where analogous mutations diminished or abolished cell-cell fusion activity without affecting SU-TM protein association (36).

A comparison of retroviral TM proteins reveals sequence and structural differences that may reflect an altered chain reversal region function. For example, primate lentiviruses can be segregated from other retroviral genera, including HTLV/bovine leukemia virus, C-type retroviruses, and D-type retroviruses, based on their chain reversal region sequences (32, 43) (Fig. 1C). Primate lentiviruses contain a disulfide-bonded CX₂(K/R)X₂C loop motif in the gp41 chain reversal region that is linked to the C-terminal antiparallel α-helix by a 10- to 15-residue sequence (based on the gp41 crystal structure shown in Fig. 1B). By contrast, HTLV-1, HTLV-2, bovine leukemia virus, and C-type and D-type retroviruses have a CX₄EXCCF motif connected by a shorter linker (5 residues in HTLV-1 gp21) to C-terminal helical elements. The coiled-coil helices of gp41 continue for three helical turns beyond the last residue of the hydrophobic 4-3 repeat (Leu-587), providing a surface onto which the linker sequence can pack in an extended conformation. By contrast, the gp21 chain reversal region is more compact, with the central helices being terminated at the last heptad repeat residue (Leu-385) by a hydrophobic cluster. The chain then exits the coiled coil at 90° to the three-fold symmetry axis before turning again by 90° to continue in an antiparallel direction.

The disulfide-bonded CX₄EXCC loops of HTLV-1 gp21 and MuLV p15E adopt very similar structures, the acidic residue (Glu-398 in gp21) forming a salt bridge with an arginine of the coiled coil (20, 32). The salt bridge is functionally relevant, since a Glu-398-to-Asn mutation abolished gp21 fusion function and helical hairpin formation (36). The conserved basic residue in the HIV-1 disulfide-bonded loop, Lys-601, is not essential for fusion activity, with K601E, K601A, and K601Q mutants retaining between 35 and 75% of wild-type fusion activity, but a basic residue is nevertheless preferred at this position, since the K601H mutant was fully functional. It is unlikely that Lys-601 of the gp41 CX₂(K/R)X₂C loop makes an analogous salt bridge with conserved acidic residues (Glu-584 and Asp-589) in the coiled coil of the gp41 hairpin, since they form intermonomer (Glu-584–Arg-579) and intramonomer (Asp-589–Lys-588) ion pairs (Fig. 1B), both of which were found to be essential for Env fusion activity. Furthermore, the extended central helices would place the CX₂(K/R)X₂C loop at the tip of the central rod, with the basic residue some distance from gp41 acidic side chains, suggesting that the chain reversal region of gp41 is stabilized via a different mechanism than its HTLV-1 and MuLV counterparts. The basic residue in the gp41 disulfide-bonded loop appears to be a minor determinant of gp120 association that also contributes to Env fusion competence. Whereas K601Q and K601H mutants anchored wild-type levels of gp120, the K601E mutant exhibited enhanced gp120-shedding, perhaps due to the substituted Glu-601 carboxylate group being inappropriately close to a gp120 acidic residue.

The highly conserved Leu-593, Trp-596, and Gly-597 residues, preceding the gp41 disulfide bridge, were found to be important for both fusion function and gp120 association. Conservative substitutions at these positions resulted in a decrease (W596F) or absence of cell-associated gp120, whereas mutations at nearby less-conserved residues, Gln-591 and Leu-592, were tolerated. Leu-593 and Trp-596 form a solvent-exposed hydrophobic patch at the base of the coiled coil adjacent to a cavity associated with Gly-597. These residues may provide a site of association for the conserved, predominantly hydrophobic residues in the gp120 N-terminal (Val-36, Tyr-40) or C-terminal (Ile-491, Gly-495) regions which are important for gp41 association (27). The presence of the Gly-597 cavity may allow accommodation of bulky gp120 side chains such as that of Tyr-40, while Lys-601 could form contacts with the gp120 acidic residues Glu-32, Glu-47, and Glu-492. While we found that Trp-596 is important for gp120 association and fusion activity, an earlier study showed that a W596M substitution inhibits HIV-1-induced cytopathic effects in Jurkat T cells but not viral entry (8). Evidence that the HIV-1 gp41 disulfide-bonded region is in intimate contact with gp120 N- and C-terminal regions has been provided by Binley et al. (4), who produced disulfide-linked gp120-gp41 by introducing cysteines into these gp41 and gp120 regions. Our results extend this observation by identifying conserved amino acids in the gp41 disulfide-bonded region that are critical for gp120 association.

Primate lentiviruses contain a conserved Val-X-Trp motif which is located C terminal to the disulfide bridge (Fig. 1B and C). In the gp41 helical hairpin structure shown in Fig. 1B, these residues are packed against Tyr-586 of the coiled coil, tethering the extended chain (Thr-606–Leu-615) preceding the C-terminal helix onto the coiled coil. A W610F substitution was sufficient to cause complete shedding of gp120 from gp41, suggesting that the Trp-610 side chain is available for a critical contact with gp120 in the gp120-gp41 prefusion complex. That Trp-610 is in close contact with gp120 residues is indicated by the finding that a W610C gp41 mutant can form an intersubunit disulfide bond with gp120 V36C, W45C, or A501C substitution mutants (4). The Tyr-586–Val-608–Trp-610 hydrophobic cluster observed in the gp41 crystal structure (Fig. 1B) is likely to be formed in the gp41 helical hairpin after fusion activation. However, we cannot rule out the possibility that the Tyr-586–Val-608–Trp-610 cluster is formed in the gp120-gp41 prefusion complex and that a cavity in W610F and W610H mutants induces a localized structural disturbance and gp120 dissociation.

In an earlier study, Cao et al. (7) found that substitutions in the C-terminal portion of the gp41 ectodomain (E647L, Q652L, L669P, F673P, I675S, N677R, W678A, and Y681P) led to reduced gp120-gp41 association. Despite the nonconservative nature of the substitutions, gp120-gp41 association was maintained at 42 to 77% of wild-type levels. These C-terminal-region mutants (except for N677R) also exhibited normal or enhanced fusion (syncytium-forming) activity, indicating a minor role in gp120 association that is unrelated to Env fusion function. Furthermore, these C-terminal residues do not appear to be sufficient for a stable gp120-gp41 complex in the context of mutated disulfide-bonded region residues Leu-593, Trp-596, Gly-597, Lys-601, and Trp-610.

Recently, we used HTLV-1 as a prototype CX₂EX₂CCF motif-containing retrovirus to analyze the functional role of the disulfide-bonded loop/chain reversal region. Our data are consistent with the idea that this region plays a conserved role in transmitting a conformational signal from receptor-bound SU to TM to induce the fusion-activated hairpin (36). In MuLV p15E, the unpaired Cys adjacent to the disulfide loop appears to contact a thiol-disulfide exchange motif CXXCL in SU, indicating that the disulfide-bonded loop is in close contact with SU in the prefusion SU-TM complex. This contact is also likely in the other CX₂EX₂CCF-containing retroviruses, since they also have the CXXCL sequence in SU (42). Receptor binding by SU may induce a conformational change or relocation of the CX₂EX₂CCF motif, which then triggers refolding of the TM protein to become fusion active. However, mutations in HTLV-1 gp21 that affected fusion function did not induce shedding of SU, indicating that other SU-TM contact sites can maintain the SU-TM complex despite a mutated chain reversal region. The present study shows that the analogous region in HIV-1 gp41 is a critical SU association site and is therefore likely to also play a role in conformation transduction from receptor-bound gp120 to gp41. The gp41 residues found here to be important for gp120-gp41 association are conserved in primate lentiviruses, suggesting a conserved functional role.