Subdomain Folding and Biological Activity of the Core Structure from Human Immunodeficiency Virus Type 1 gp41: Implications for Viral Membrane Fusion

Abstract

The envelope glycoprotein of human immunodeficiency virus type 1 (HIV-1) consists of two subunits, gp120 and gp41. The extraviral portion (ectodomain) of gp41 contains an α-helical domain that likely represents the core of the fusion-active conformation of the molecule. Here we report the identification and characterization of a minimal, autonomous folding subdomain that retains key determinants in specifying the overall fold of the gp41 ectodomain core. This subdomain, designated N34(L6)C28, is formed by covalent attachment of peptides N-34 and C-28 by a short flexible linker in place of the normal disulfide-bonded loop sequence. N34(L6)C28 forms a highly thermostable, α-helical trimer. Point mutations within the envelope protein complex that abolish membrane fusion and HIV-1 infectivity also impede the formation of the N34(L6)C28 core. Moreover, N34(L6)C28 is capable of inhibiting HIV-1 envelope-mediated membrane fusion. Taken together, these results indicate that the N34(L6)C28 core plays a direct role in the membrane fusion step of HIV-1 infection and thus provides a molecular target for the development of antiviral pharmaceutical agents.

Viral envelope glycoprotein-mediated membrane fusion is an essential step in the infectious processes of all enveloped animal viruses, including human pathogens such as influenza virus and human immunodeficiency virus type 1 (HIV-1). The envelope glycoprotein of HIV-1 is first synthesized as a polypeptide precursor, gp160, which is then posttranslationally processed to generate two noncovalently associated subunits, gp120 and gp41 (19, 28, 29). gp120 recognizes the target cell by binding to the CD4 glycoprotein and a chemokine receptor (13, 24, 32, 42). gp41 then undergoes conformational changes to become active in promoting virus-cell membrane fusion, a process that leads to viral entry and infection of cells. These conformational changes are thought to be involved in the transition in gp41 from a native (nonfusogenic) to a fusion-active (fusogenic) state (reviewed in reference 7).

The extraviral portion (ectodomain) of the gp41 molecule is directly involved in the membrane fusion process (29). The amino terminus of gp41 contains a hydrophobic, glycine-rich sequence referred to as the fusion peptide that is essential for membrane fusion. There are two 4-3 hydrophobic (heptad) repeat sequences within the gp41 ectodomain which are predicted to form coiled coils, as seen in most viral fusion proteins (6, 12, 16). The N (amino)-terminal heptad repeat is located adjacent to the fusion peptide, while the C (carboxyl)-terminal heptad repeat precedes the transmembrane segment. Limited proteolysis of a fragment corresponding to the gp41 ectodomain led to the identification of a soluble, α-helical complex consisting of a trimer of antiparallel dimers (26, 27). Biophysical studies suggest that three N-terminal helices form an interior, parallel-coiled-coil trimer, while three C-terminal helices pack in the reverse direction into three hydrophobic grooves on the surface of this coiled coil (27). Crystallographic analysis of the gp41 ectodomain core confirmed that it folds into a six-helix bundle (Fig. 1B) (8, 34, 35). A number of studies support the notion that this six-helix structure represents the fusion-active conformation of the gp41 ectodomain core (8, 15, 20, 23, 27, 30, 34, 35).

FIG. 1.

An α-helical core subdomain within the ectodomain of HIV-1 gp41. (A) Schematic representation of gp41. The important functional features of the gp41 ectodomain, the locations of the N-36 and C-34 peptides, and the amino acid sequences of the N-34 and C-28 peptides are shown. N34(L6)C28 consists of N-34 and C-28 plus a linker of six hydrophilic residues. The disulfide bond and four potential N-glycosylation sites are depicted. The residues are numbered according to their position in gp160. (B) Ribbon diagram of the N34(L6)C28 core. The N-terminal helices are depicted in yellow, and the C-terminal helices are in purple. The N-34 and C-28 termini are joined by a linker. The left panel shows an end-on view of N34(L6)C28 looking down the three-fold axis of the trimer. The right panel shows a side view of the N34(L6)C28 trimer.

We report here the identification and characterization of a small subdomain that dictates the folding and stability of the gp41 core from the HIV-1 envelope protein. This subdomain, designated N34(L6)C28, consists of two highly truncated peptides, N-34 and C-28, which are connected by a six-residue hydrophilic linker (Fig. 1A). N34(L6)C28 forms an α-helical, discrete trimer. Moreover, our data on the in vitro folding of the trimeric N34(L6)C28 complex correlate well with the severity of the in vivo phenotypes observed in cells expressing the full-length HIV-1 envelope protein (14). Finally, N34(L6)C28 can inhibit HIV-1-mediated membrane fusion at micromolar concentrations. These results provide evidence that the core structure formed by N34(L6)C28 plays a direct role in the HIV-1 fusion events.

The N34(L6)C28 subdomain.

Previous studies identified an α-helical complex within the gp41 ectodomain consisting of the peptides N-36 and C-34 and showed that these two peptides associate to form a stable, α-helical trimer of heterodimers (Fig. 1A) (26). To facilitate further studies, we designed and constructed a unimolecular (i.e., single-chain) analog for the N-36 and C-34 complex, in which the C terminus of N-36 is connected to the N terminus of C-34 by the hydrophilic linker Ser-Gly-Gly-Arg-Gly-Gly. Plasmid pN36/C34(L6), encoding this unimolecular model designated N36(L6)C34, was constructed by single-strand mutagenesis of pN41/C34(L6) (26). N36(L6)C34 and variants thereof were expressed in Escherichia coli BL21(DE3)/pLysS by using the T7 expression system as previously described (26). Proteins were purified to homogeneity by reverse-phase high-performance liquid chromatography as previously described (26). Protein identity was confirmed by mass spectrometry.

Circular dichroism (CD) spectra were measured with an AVIV 62DS CD spectrometer as previously described (26). Thermal stability was determined by monitoring the change in CD signal at 222 nm as a function of temperature. The midpoint of the thermal unfolding transition (apparent T_m) was determined from the maximum of the first derivative, with respect to the reciprocal of the temperature, of the [θ]₂₂₂ values. The error in estimation of T_m was ±1°C. Apparent molecular weights were determined by sedimentation equilibrium with a Beckman XL-A analytical ultracentrifuge as previously described (26). Data sets (six per peptide) were fitted simultaneously to a single-species model with the program NONLIN (22) to yield an apparent ς value. Specific volumes and solvent densities were calculated as described by Laue et al. (25).

The CD spectrum of N36(L6)C34 is typical of an α-helix, displaying the characteristic minima at 208 and 222 nm. From the magnitude of the CD signal at 222 nm (Table 1), we estimate that 60 residues (∼85% helix content) are in α-helical conformation. This helical structure is remarkably stable; at a neutral pH and a 10 μM protein concentration, the apparent T_m of N36(L6)C34 is 79°C (Table 1). Sedimentation equilibrium experiments showed that the apparent molecular weight of N36(L6)C34 changes with peptide concentration, indicating that N36(L6)C34 tends to aggregate in solution (data not shown).

TABLE 1.

Summary of CD and sedimentation equilibrium data for the gp41 model peptides

Peptide	[θ]222 value (10−3)a	Tm (°C)b	Molecular mass (kDa)c
N36(L6)C34	−26	79	NAd
N34(L6)C28	−31	70	24.4 (23.9)
N34(L5)C28	−31	68	24.9 (23.8)
N34(L10)C28	−29	67	26.6 (25.9)
N30(L6)C28	−20	39	NA

^a[θ]₂₂₂ was measured at 0°C at monomer peptide concentrations of 10 μM in 50 mM sodium phosphate (pH 7.0) and 150 mM NaCl (PBS). 

^bT_m was determined at monomer peptide concentrations of 10 μM. 

^cApparent molecular masses were determined with initial peptide concentrations of 10, 30, and 100 μM at 20°C. The expected value for a trimer is enclosed in parentheses. 

^dNA, not applicable. Data were fit to a single ideal species model plot of ln(absorbance) versus radial distance squared. Nonrandom residuals, indicative of aggregation or derivations from ideality, were observed. 

To trim unfolded regions that potentially contribute to the aggregation, N36(L6)C34 was subjected to limited proteolysis as previously described (26). The peptide fragments were separated and purified by high-performance liquid chromatography, and each peptide was then identified by N-terminal sequencing and mass spectrometry. Five residues at the N terminus of each identified proteolytic fragment were sequenced. Digestion with proteinase K yielded, in addition to N-36 (residues 546 to 581) (observed mass, 4,123 Da; expected, 4,123 Da) and C-34 (residues 628 to 661) with the N-terminal sequence Ser-Gly-Gly-Arg-Gly-Gly (observed mass, 4,721 Da; expected, 4,720 Da), two shorter peptide fragments: C-25, spanning residues 628 to 652 with the N-terminal sequence Ser-Gly-Gly-Arg-Gly-Gly (observed mass, 3,609 Da; expected, 3,608 Da), and N36(L6)C25, spanning residues 546 to 581 (N-36) and 628 to 652 (C25) connected by the linker Ser-Gly-Gly-Arg-Gly-Gly (observed mass, 7,713 Da; expected, 7,713 Da). Digestion of N36(L6)C34 with trypsin generates, in addition to N-36 (residues 546 to 581) with the C-terminal sequence Ser-Gly-Gly-Arg (observed mass, 4,481 Da; expected, 4,480 Da) and C-34 (residues 628 to 661) with the N-terminal sequence Gly-Gly (observed mass, 4,364 Da; expected, 4,363 Da), two shorter peptide fragments: N-34, spanning residues 546 to 579 (observed mass, 3,897 Da; expected, 3,897 Da), and C-28, spanning 628 to 655 with the N-terminal sequence Gly-Gly (observed mass, 3,634 Da; expected, 3,636 Da). The C terminus of N36(L6)C34 is trimmed to Gln 652 with proteinase K or to Lys 655 with trypsin. Trypsin also cleaves N36(L6)C34 at Arg 579. These results indicate that the C-terminal regions of both the helical segments in N36(L6)C34 are not well folded. Accordingly, we adopted the peptides N-34 and C-28 for further studies.

We produced a bacterially expressed single-chain model, designated N34(L6)C28, for the N-34 and C-28 complex. In this molecule, the C terminus of N-34 is connected to the N terminus of C-28 by the linker Ser-Gly-Gly-Arg-Gly-Gly. CD spectra show that N34(L6)C28 is fully helical (Table 1) and highly stable against thermal denaturation, with an apparent T_m of 70°C (10 μM and pH 7.0) under physiological conditions (Table 1). Over a 10-fold range of protein concentrations, the apparent molecular mass of N34(L6)C28 is 24.4 kDa, as determined by sedimentation equilibrium (Table 1). This value, compared to the expected molecular mass of 23.6 kDa for a trimer, indicates that N34(L6)C28 is trimeric in solution.

To examine if the sequence and/or length of the peptide linker per se affects the N-34 and C-28 complex formation, we produced two additional single-chain models with linkers that vary in sequence and length. The two helical segments are connected by the linker Gly-Pro-Arg-Arg-Gly in N34(L5)C28 or by (Gly-Pro-Arg-Arg-Gly)₂ in N34(L10)C28. CD spectra indicate that both the model peptides exist in fully helical conformations which sedimentation equilibrium indicates is a clean trimer (Table 1). Therefore, the linker in N34(L6)C28 appears to stabilize the folded gp41 core for entropic reasons, without perturbing its structure.

To address whether N34(L6)C28 is the smallest cooperatively folded subdomain, we truncated four residues (Leu 576, Gln 577, Ala 578, and Arg 579) from the C-terminal sequence of the N-34 segment in N34(L6)C28 (Fig. 2A). This truncation was chosen for the investigation of how the formation of the central, α-helical coiled-coil interactions gives rise to the cooperative folding and stability of N34(L6)C28. N30(L6)C28 corresponds to a deletion of residues 576 to 579 and was produced in bacteria. Its CD spectrum indicates that N30(L6)C28 is only approximately two-thirds helical, with an apparent T_m of 39°C for a 10 μM solution (Fig. 2B). Moreover, equilibrium sedimentation shows that the radial distribution of N30(L6)C28 is nonlinear and becomes progressively more curved as the concentration is increased, indicating that the peptide in solution forms a mixture of oligomeric species (data not shown). Taken together, these results indicate that N30(L6)C28 is unable to impart the structural specificity conferred by N34(L6)C28. It is likely that the loss of favorable homotrimeric interactions around Leu 576 results in the destabilization of the central, trimeric coiled coil, and therefore contributes to the “misfolding” of N30(L6)C28. Indeed, the CD spectrum indicates that the isolated N-34 peptide contains an ∼90% α-helical structure, while the isolated N-30 peptide is largely unfolded (Fig. 2A). Thus, the N-34 peptide seems to be the shortest N-terminal heptad-repeat sequence required to specify the overall fold of the stable core structure of the gp41 ectodomain. By extension, N34(L6)C28 may represent the smallest stable, cooperatively folded gp41 core.

FIG. 2.

(A) Sequences of the N-34 and N-30 peptides, with the heptad positions marked above the sequence. (B) CD spectra of N30(L6)C28 (10 μM) (filled circles), N-34 (50 μM) (open triangles), and N-30 (50 μM) (open circles) in PBS at 0°C. (C) Temperature dependence of the CD signal at 222 nm for N30(L6)C28 (10 μM) (filled circles), N-34 (50 μM) (open triangles), and N-30 (50 μM) (open circles) in PBS.

A folding defect in N34(L6)C28 arises from fusion-defective mutations of gp120-gp41.

Ile 573, located at an a heptad position in the coiled coil, forms a homotrimeric packing at the center of the coiled coil and interacts with Trp 631 of the C-terminal helix in the crystal structure of the gp41 core (Fig. 3A) (8, 34, 35). Mutagenesis studies show that the hydrophobicity of the side chain at the conserved Ile 573 site is coupled to the fusion activity of the HIV-1 envelope protein complex (14). To understand how these mutations affect gp41 core structure formation, we substituted Ile 573 in N34(L6)C28 with Leu, Val, Ala, Ser, and Pro to generate five mutants. Peptides with conservative mutations (I573L and I573V) were expressed in E. coli at high levels (∼50 mg of protein per liter of culture), while peptide expression was limited to ∼6 mg per liter for I573A and 1 mg per liter for I573S and I573P.

FIG. 3.

Helical structure and thermal stability of the N34(L6)C28 mutant peptides. (A) Helix packing in the hydrophobic layer between Gly 572 and Ile 573 in the interior coiled-coil trimer and Trp 631 and Asp 632 in the outside C-terminal helices. (B) CD spectra of the N34(L6)C28 mutant peptides (10 μM) (I573L, filled triangles; I573V, open circles; I573A, filled circles; I573S, open rhombs; I573P, open triangles) in PBS at 0°C. (C) Temperature dependence of the CD signal at 222 nm for the N34(L6)C28 mutant peptides (10 μM) (I573L, filled triangles; I573V, open circles; I573A, filled circles; I573S, open rhombs; I573P, open triangles) in PBS.

CD spectra of the I573L and I573V mutants indicate that the folded mutant peptides exhibit a >95% α-helical structure (Fig. 3A and Table 2). At a neutral pH and a 10 μM peptide concentration, the apparent T_ms of I573L and I573V are 67 and 65°C, respectively (Fig. 3B and Table 2). Sedimentation equilibrium experiments indicate that both I573L and I573V form clean trimers in solution (Table 2). In contrast, the I573S and I573P mutants form insoluble aggregates at concentrations above ∼10 μM in phosphate-buffered saline (PBS). These mutant peptides are much less helical than the wild-type peptide (approximately 100% helix content for wild type, 42% for I573S, and 48% for I573P) (Fig. 3B and Table 2). The structures of I573S and I573P also unfold at much lower apparent T_ms (70°C for wild type, 25°C for I573S, and 24°C for I573P) (Fig. 3C and Table 2). An intermediate effect is seen in the I573A peptide, which contains an ∼77% α-helical structure, with an apparent T_m of 37°C (Fig. 3B and C and Table 2). Equilibrium sedimentation of the I573A mutant yielded average molecular masses consistent with a clean trimer (Table 2). These results indicate that both the Leu and Val substitutions for Ile 573 can confer the six-helix bundle fold. In contrast, the Pro and Ser mutations each essentially disrupt the trimeric complex formation. Moreover, Ala 573 maintains trimerization specificity at the expense of stability. Thus, the folding and stability of the N34(L6)C28 mutant peptides correlate well with severity of the in vivo phenotypes observed in cells expressing the full-length HIV-1 envelope protein complex. These results strongly suggest that the core structure formed by N34(L6)C28 plays a direct role in the membrane fusion step of HIV-1 infection.

TABLE 2.

Summary of CD and sedimentation equilibrium data for the I573 mutants

Peptide	[θ]222 values (10−3)a	Tm (°C)b	Molecular mass (kDa)c
N34(L6)C28	−31	70	24.4
I573L	−31	67	25.1
I573V	−31	65	24.9
I573A	−24	37	24.1
I573S	−13	25	NAd
I573P	−15	24	NA

^a[θ]₂₂₂ was measured at 0°C at monomer peptide concentrations of 10 μM. 

^bT_m was determined at monomer peptide concentrations of 10 μM. 

^cApparent molecular masses were determined with initial peptide concentrations of 10, 30, and 100 μM at 20°C. 

^dNA, not applicable. I573S and I573P each form an insoluble aggregate at concentrations above ∼10 μM in PBS. 

Many viral membrane fusion proteins can adopt two different tertiary folds. This structural polymorphism is the basis for conformational changes in response to environmental signals and ligand binding. The influenza virus hemagglutinin (HA) protein, for example, irreversibly switches from the native structure to the fusogenic conformation when exposed to a low pH (17, 18, 33, 41). Recent studies suggest that HA is most stable in its fusogenic state, while HA in its native state is metastable and thus has the potential to transform to a more stable, fusogenic state (1, 4, 5, 10). According to this hypothesis, membrane fusion is regulated by the conformational state of the HA protein.

Mutations within the N heptad repeat region of gp41 abolish membrane fusion activity without preventing formation of the native HIV-1 envelope protein complex (3, 11, 14, 38). These results can be reconciled by the hypothesis that these mutations do not disrupt the native structure of gp41 but do inhibit its conformational change to the fusion-active state (9, 38). This view is supported by our finding that fusion-defective mutations lead to great destabilization of the fusion-active core structure of gp41. In principle, introducing mutations into the N heptad repeat region of the gp41 ectodomain could shift the conformational equilibrium between the native and fusogenic folds, thereby allowing the native structure to be trapped in a metastable state for biophysical and structural studies.

N34(L6)C28 inhibits HIV-1 fusion.

Peptides corresponding to the N- and C-terminal heptad repeat regions of the gp41 ectodomain exhibit antiviral activity and block membrane fusion (21, 37, 40). Although the mechanism of action of these peptide inhibitors is not known, considerable evidence suggests that they work, in a dominant-negative manner, by associating with gp41 during the fusion process (9, 15, 23, 27, 30, 31, 39). Since even the smaller N-36 and C-34 complex is too stable to be disrupted by peptide binding, one anticipates that only during the gp41 conformational change to the fusion-active state are the targets for the peptides available (reviewed in reference 7). A recent study of the structure of the ectodomain of simian immunodeficiency virus gp41 challenges this assumption (2).

The inhibitory activities of N34(L6)C28 and variants thereof were determined by an HIV-1 envelope glycoprotein-mediated cell-cell fusion assay as previously described (27). Cells expressing HIV-1 envelope glycoprotein (CHO[HIVe]) (clone 7d2) cells were a generous gift from M. Krieger, and MT-2 cells were obtained through the National Institutes of Health AIDS Research and Reference Reagent Program. CHO[HIVe] cells were plated at 2 × 10⁴ cells/well in a 48-well dish and were grown for 24 h. Then, 10⁵ MT-2 cells were added in the presence of various peptide concentrations. After 14 h of incubation at 37°C, syncytia were counted by microscopic examination at a magnification of 40×.

The N-51 and C-43 complex is also capable of inhibiting fusion, with a 90% inhibitory concentration (IC₉₀) of 1.5 μM (27). Due to difficulties in preparing stoichiometric amounts of the two peptides, this activity was originally believed to be caused by a small fraction of the dissociated C-43 peptide, because the isolated C-43 peptide is an effective inhibitor, with an IC₉₀ of 0.14 μM (27). The single-chain N34(L6)C28 model folds into an extremely stable, six-helix bundle (Fig. 2B). This molecule thus offers an excellent model for investigating the inhibition mechanism of the gp41 core. We examined the relative abilities of N34(L6)C28 and the isolated N-34 and C-28 peptides to block syncytium formation between CHO[HIVe] cells and CD4⁺ target cells (MT-2). Figure 4 shows the inhibition of syncytium formation by N34(L6)C28, N-34, and C-28. N34(L6)C28 and C-28 have similar inhibitory activities, with IC₉₀s of 2.0 and 1.3 μM, respectively, whereas the inhibitory activity of the N-34 peptide cannot be detected below concentrations of 3 μM.

FIG. 4.

Inhibition of syncytium formation. Inhibition of syncytium formation between CHO[HIVe] cells and CD4⁺ MT-2 cells by the isolated N-34 (open triangles) and C-28 (open squares) peptides and the N34(L6)C28 model peptide (open circles). Standard deviations of the means for triplicate samples are indicated by vertical bars.

Several lines of evidence suggest that the dissociated C-28 region is unlikely to account for the inhibitory activity of N34(L6)C28. First, the trimeric N34(L6)C28 complex is highly thermostable under physiological conditions, with a melting temperature of 63°C for a 2 μM solution in PBS. Second, the N34(L6)C28 core is highly resistant to trypsin digestion. Following the incubation of 1 mg of N34(L6)C28 with 0.01 mg of l-(tosylamido-2-phenyl)ethyl chloromethyl ketone-treated bovine trypsin at 37°C for 24 h in PBS, more than 90% of the molecules are still in an α-helical conformation, as judged by CD spectra. Trypsin cleaves N34(L6)C28 at two Arg residues in the six-residue linker region, as identified by mass spectrometry. This resistance was expected because the N-34 and C-28 complex was originally identified by limited proteolysis. Third, the inhibitory activity of N34(L6)C28 was not affected even in the presence of 10 μM of the isolated N-34 peptide. If the inhibitory activity of N34(L6)C28 is due to a small fraction of the dissociated C-28, the addition of N-34 should decrease the core’s activity by associating with the C-28 region. Taken together, our results suggest that in the N34(L6)C28 model of the gp41 core, membrane fusion is inhibited via a mechanism different from that in the dominant-negative model proposed for the isolated N and C peptides. For example, the fusion-active six-helix bundle may act as an inhibitor by interfering with the formation of a fusion pore (36). Understanding the inhibition mechanism of the gp41 core will provide insights into the HIV-1 entry process and could offer new perspectives in the search for effective antiviral therapies.