Conformational trajectory of the HIV-1 fusion peptide during CD4-induced envelope opening

Categories: Disease & Virus

May 17, 2025

FP remains accessible for antibody binding after CD4-induced Env opening

To assess CD4-induced changes in FP accessibility, we measured binding to the FP-targeted antibody VRC34.01 at different time points following incubation of BG505.SOSIP Env with either CD4 alone, or together with the fragment antigen binding (Fab) of the coreceptor-mimicking antibody 17b that recognizes an epitope presented upon CD4-induced Env opening (Fig. 1C and Supplementary Fig. S1). VRC34.01 binding decreased after CD4-induction, and the decrease was more profound in the presence of 17b Fab. A control experiment without the addition of CD4 or 17b showed no change in VRC34.01 binding to BG505.SOSIP Env. We next assessed simultaneous changes in FP exposure measured by binding to VRC34.01, and Env opening measured by binding to 17b (Fig. 1D and Supplementary Fig. S1). 17b binding increased post CD4 addition, indicating Env opening and exposure of the bridging sheet, while VRC34.01 binding decreased.

To visualize the impact of CD4-bound Env conformations on FP positioning, we incubated BG505 SOSIP Env with CD4 and 17b Fab at 25 °C, and performed single particle cryo-EM on the Env complexes at selected time-points, 1.3 h (hr), 20 h, and 3 days, post CD4/17b addition, with VRC34.01 Fab added 30 min before the samples were vitrified for cryo-EM analysis (Figs. 1E, 2, and Supplementary figs. S2–S7, Table S1). We identified three particle populations across the three cryo-EM datasets that differed in their stoichiometries of bound VRC34.01 Fab (Fig. 1E, Supplementary Fig. S7, and Supplementary Table S2). Population 1 dominated at all three time points and consisted of a partially open Env in which each of the three gp120-gp41 protomers were bound to CD4, 17b Fab and VRC34.01 Fab. Another population, named Population 2, was detected at all three time points, albeit in smaller proportions relative to Population 1 (Supplementary Fig. S7). In Population 2, each of the three gp120 subunits were bound to CD4 and 17b Fab, while only two protomers were bound to VRC34.01 Fab. The proportion of Population 2 relative to Population 1 increased with longer incubation times. At the 3-day time point, a third population, named Population 3, was detected that resembled Populations 1 and 2 in their bound CD4 and 17b stoichiometries but only had a single VRC34.01 Fab bound, leaving two protomers not bound to VRC34.01.

**Fig. 2: A partially open intermediate on the HIV-1 entry pathway retains FP accessibility to antibody binding.**

While the experiments described above were performed by incubating Env with the ligands at 25 °C, similar trends were observed in SPR binding assays when the incubations were performed at 37 °C (Supplementary Figs. S1 and S9). A cryo-EM dataset obtained by incubating BG505 SOSIP with CD4 and 17b Fab at 37 °C for 2 h, followed by a 30 min incubation with VRC34.01 Fab at 37 °C prior to plunge freezing yielded structures representing Populations 1, 2 and 3. The appearance of the Population 3 structure earlier than could be detected in the 25 °C cryo-EM datasets suggested that the CD4-induced Env conformational changes are more rapid at 37 °C than at 25 °C. The recurrence of these structures across independent experiments highlights the reproducibility of these structural states.

In summary, we identified three populations of CD4-induced Env in our cryo-EM datasets with differing stoichiometries of bound VRC34.01. These results confirmed that FP remained accessible to VRC34.01 binding despite substantial Env opening and suggested FP accessibility to antibody binding is hindered at the sites in Populations 2 and 3 that were not bound by VRC34.01.

A functional CD4-triggered partially open Env intermediate has an antibody-accessible FP

Two distinct structural configurations of the FP have thus far been defined in the literature, one that is antibody accessible in the pre-fusion closed Env⁷ and a second that is sequestered within a gp41-gp120 pocket in a partially or fully open CD4-induced Env^17,21. In this study, we have discovered new CD4-induced Env intermediates that are open enough to bind the bridging sheet-directed antibody 17b and yet retain the ability to bind a FP-targeting antibody. To understand Env-structural changes that enable CD4-induced opening, while the FP remains in an antibody-accessible configuration, we first examined Population 1, which was the dominant population in all the cryo-EM datasets (Figs. 1E, 2, and Supplementary Fig. S7, Tables S1 and S2). We selected the Population 1 reconstruction from the 1.3-hr time-point for our analysis as it contained the largest number of particles and the highest resolution among the Population 1 structures from the three datasets.

In Population 1, the gp120 subunits exhibited known structural markers of the CD4-induced conformation^2,4,17, including the bridging sheet at the 17b-binding interface and residues 63–73 assembled into the α0 helix (Fig. 2A and Table S2). The gp41 subunit appeared conformationally less perturbed and was bound to VRC34.01 with a similar interaction interface dominated by the FP and the gp120 N88 glycan, as previously observed in the structure of VRC34.01 in complex with the closed BG505 SOSIP (Fig. S10)⁷. Although no symmetry had been applied during the cryo-EM data processing, the three protomers were highly similar in the symmetrically open Population 1 intermediate (Fig. S10). Our Population 1 structure revealed a similar gp120 opening geometry as the cryo-ET structure of membrane-associated HIV-1_ADA.CM Env bound to three membrane-associated CD4 molecules (Fig. 2B)⁶, suggesting that Population 1 represents a physiologically relevant entry intermediate.

We next studied the Population 1 structure using a previously defined set of vectors that report on structural rearrangements associated with rigid body movements in gp120 relative to gp41 (Fig. 2C and Table S3)²⁴. These vectors describe the orientation of gp120 relative to the gp41 three-helix bundle, capturing rotation of gp120 away from the trimer central axis and rotation orthogonal to the trimer central axis. These measures effectively capture differences between closed, open, and intermediate state Envs. All three Population 1 protomers clustered together in all measures examined and were similar to previously published structures of BG505 (PDB: 6CM3) or B41 (PDB: 6EDU) SOSIP bound to CD4, 17b and 8ANC195 Fab²¹. The Population 1 structures were distinct from previously published open and open occluded state structures (PDBs: 5VN3 and 5VN8, respectively) in gp120 rotations described by a dihedral angle (φ) that defines orthogonal rotation and angles (θ₁ and θ₂) describing rotations relative to the trimer central axis (Fig. 2C). However, the distance between the gp120 core and W571 was similar between the open and intermediate structures. Contrasting each with the closed state structure clusters indicates the open and open occluded structures occupy distinct angles in the dihedral and the angle between the gp41 three-helix bundle and gp120 termini, while the Population 1 and CD4, 17b, 8ANC195-bound Envs differ in the angle describing gp120 rotation away from the central trimer axis. In summary, our vector analysis indicates that the partially open Population 1 intermediate described here shifts the gp120 domains away from the central axis, while the open and open-occluded structures shift the gp120 domains orthogonal to the trimer central axis.

We compared the configuration of the FP in the previously published partially open CD4, 17b, 8ANC195-bound structure (PDB: 6CM3) and the partially open CD4,17b, VRC34.01-bound Population 1 structure resolved in this study (PDB: 9D90) (Fig. 2D). In the Population 1 structure (VRC34.01-bound), the FP was extended out of the Env core to bind the VRC34.01 antibody, whereas, in the 8ANC195-bound structures, the FP was buried in an intra-protomer gp120/gp41 hydrophobic pocket. The formation of the pocket for FP sequestration in the CD4, 17b, 8ANC195-bound structure was facilitated by a shift in the position of the FP proximal region (FPPR) primarily involving straightening of the FPPR helix creating space for FP burial. The distance between the Cα atoms of FPPR residue Gln540 and the gp120 residue Phe233 measured at 13 Å for the Population 1 structure and at 19 Å for the CD4,17b, 8ANC195 bound BG505 SOSIP structure (Fig. 2D). A similar FP configuration was observed in the B41-complex with CD4, 17b and 8ANC195, suggesting that this is an isolate-independent conformational state (Supplementary Fig. S11). The difference in FP accessibility while maintaining overall similar protomer geometry suggested that in this intermediate state the FP can either be antibody-accessible or it can be occluded. While VRC34.01 binds FP and stabilizes its accessible configuration, 8ANC195 stabilizes the FP occluded configuration.

In summary, we identified a CD4-triggered partially open Env intermediate on the HIV-1 entry pathway, with a protomer geometry that accommodates an antibody-accessible or a buried FP, with a conformational change in the FPPR being the major facilitator for this conformational switch of FP.

Mechanism of downstream FP sequestering and antibody inaccessibility

In addition to the near symmetric, partially open Population 1 state where VRC34.01 was bound at each of the three FP sites, we also identified populations that were bound to either one or two VRC34.01 Fabs, leaving two and one FP sites unbound, respectively, despite saturating amounts of VRC34.01 Fab being used for sample preparation (Figs. 1E, and 3A, B and Supplementary Fig. S7 and Tables S1, S2). As expected, based on binding to antibody 17b, the bridging sheet and the α0 helix were formed in all gp120 protomers in the Population 2 and Population 3 structures (Fig. 3A, B, and Supplementary Table S2). Examining the unbound sites revealed FP sequestered within a gp120/gp41 pocket in an antibody-inaccessible configuration (Fig. 3C, D), thus providing a structural explanation for the lack of antibody binding to these FP sites.

**Fig. 3: Burial of FP upon gp120 opening.**

Unlike the near-symmetric Population 1 structure, Populations 2 and 3 displayed marked asymmetries. To quantify Env opening, we measured interprotomer distances between the CD4-binding site gp120-residue Asp368 and gp120-residue Pro124 at the Env trimer apex (Fig. 3E, F). As previously recognized, the closed and open Env structures showed substantial differences in these distances²³ (Fig. 3E). In the closed Env trimer (PDB: 5ACO)²⁵, the distance between the Asp368 residues and Pro124 residues measured 14.7 Å and 56 Å, respectively. These distances were much larger in the CD4,17b-bound open Env trimer (PDB: 5VN3) at 76.7 Å and 84.3 Å, respectively. By contrast, in the CD4,17b,8ANC195-bound partially open Env (PDB: 6CM3), these distances were intermediate between the open and closed, at 65 Å and 79.4 Å, respectively. Since all three structures were reconstructed by imposing C3 symmetry during cryo-EM map refinement, each of these interprotomer distances was identical within each structure. These distances measured in the Population 1 structure were similar to the distances in the CD4,17b,8ANC195-bound partially open structure (PDB: 6CM3) (Fig. 3F). Since no symmetry was applied during the reconstruction of the Population 1 map, three distances were noted for each measure: 63 Å, 66 Å, and 66.9 Å for the interprotomer distances between residue Asp368, and 75 Å, 77.6 Å and 78.2 Å for the interprotomer distances between residue Pro124. In Population 2, the two protomers that were bound to VRC34.01 showed similar separation as observed in Population 1, 61.4 Å and 74.6 Å between residues Asp368 and Pro124, respectively. The protomer that was not bound to VRC34.01 showed greater gp120 displacement with these distances approaching closer to those observed in the CD4,17b-bound open Env trimer (PDB: 5VN3). In Population 3, the two protomers that were not bound to VRC34.01 had buried FPs and showed gp120 geometries closer to the fully open Env conformation.

Taken together, our results demonstrate that FP burial rendering it inaccessible to an FP-directed antibody requires further Env opening and transition of Env geometry past a symmetrically open intermediate that occurs earlier during CD4-induced Env opening. This CD4,17b,VRC34.01-bound BG505 SOSIP Population 1 (PDB: 9D90, this study) intermediate resembles the cryo-ET structure of the CD4-bound HIV-1_ADA.CM Env⁶, and the single particle cryo-EM structures of CD4,17b,8ANC195-bound BG505 SOSIP (PDB: 6CM3)²³ and CD4,17b,8ANC195-bound B41 SOSIP (PDB: 6EDU)²³. In this early intermediate, FP can adopt a buried (antibody-inaccessible) or an antibody-accessible conformation. This intermediate is characterized by gp120 protomers opening like the petals of a tulip where the Env trimer apex separates and gp120 is displaced from the trimer central axis, as a rigid body hinging about the gp120 N/C termini at the trimer base. For stable sequestration of FP that renders it unavailable for antibody binding, further displacement of the gp120 protomers is needed, in the form of a lateral rotation in a plane roughly parallel to the viral membrane and about an axis orthogonal to the central trimer axis.

Conformational changes in gp41 required for stable FP sequestering

To elucidate the gp41 structural features involved in stable FP sequestration, we examined differences in the vicinity of the FP between the partially open Population 1 conformation (PDB: 9D90; this study) and the previously described fully open Env (PDB 5VN3)¹⁷. The fully open structure showed a greater displacement of the gp120 subunits visualized by the clear separation of signature α0 helix from gp41, while in the Population 1 structure this region was helical but remained associated with the gp41 subunit (Fig. 4A).

**Fig. 4: FP trajectory upon CD4-induced Env opening.**

At the FP site, the most striking difference was observed in the gp120/gp41 pocket where FP was buried in the fully open structure versus this region in the partially open intermediate (Fig. 4B, C). In the CD4,17b-bound fully open structure (PDB; 5VN3), this pocket was much larger and measured at ~26 Å between FPPR residue Gln540 and gp120 residue Phe233, with the buried FP adopting an extended loop conformation to fill the larger space of the pocket. By contrast, in the partially open intermediate, this distance measured 13 Å in Population 1 (CD4,17b, VRC34.01-bound structure) where the FP was exposed and 19 Å in the CD4,17b, 8ANC195-bound structure where the FP was buried and assumed a helical conformation. Progressive straightening of FPPR along with changes in both HR1 and HR2 regions of gp41 orchestrated the enlargement of this pocket, which in the fully open structure assumes an interprotomer character with one of its walls lined with the HR1 helix of the adjacent protomer. Thus, the concerted gp120/gp41 re-organizations that resulted in the formation of a larger FP-binding pocket may be responsible for the stable sequestration of the FP in the fully open structure.

We observed that the Population 2 gp41 with buried FP had a conformation similar to that of the fully open Env (PDB: 5VN3), with HR1 helices showing close overlap and the FPPR straightened out further compared to the partially open Population 1 and the 6CM3 structures, albeit not to the extent of the fully open structure (Fig. 4D). In both Population 2 and fully open 5VN3 structures, the movement of the HR2 region around residues 638–662 (indicated by red arrow in Fig. 4D) creates room for the FPPR unbending. The gp120/gp41 pocket in this Population 2 protomer measured 21 Å, with the cavity size approaching that of the cavity measured in the fully open structure.

In summary, our data show that the FP configuration undergoes stepwise changes as a consequence of CD4-induced movements in gp120 and gp41. From a closed, pre-receptor state where FP is accessible to antibodies, Env proceeds to partially open states where FP remains available to the FP-directed antibody VRC34.01. Only upon more extensive rotation of gp120 and concerted changes in gp41 does FP become fully buried within a gp120/gp41 pocket and, as a result, no longer accessible to antibody binding.

smFRET analysis of Env opening on the virion surface

smFRET analysis of Env on the surface of intact virions has revealed conformational shifts of virus Env from a pre-triggered (PT) state through a pre-receptor closed (PC) state to a fully open CD4-bound conformational state (CO) in response to CD4 activation^26,27. The pre-fusion, pre-receptor closed Env on virions resembles the FP-accessible Env structure complexed with three VRC34.01 (PDB: 5I8H), while the fully open Env was associated with the symmetric Env structure bound with three CD4 and three 17b (PDB: 5VN3), and a pre-triggered state was suggested that is undefined in currently available structures^26,27. We asked whether the structural differences between partially open VRC34.01-bound Env structures characterized in this study and the fully open FP-sequestered Env would be reflected at the global population level of Env conformations presented on virions. We performed smFRET experiments at room temperature (~25 °C) of two different fluorescently click-labeled Env_BG505 on intact HIV-1_Q23 virions²⁸, in which donor/acceptor fluorescent probes were placed between gp120 V1 and V4 or between gp120 V4 and gp41 α6, respectively (Supplementary Fig. S12A). Placing FRET probes at different paired structural elements of Env allowed us to visualize global conformational changes of Env from two different structural perspectives, gp120 V1-V4 and gp120-gp41 (Fig. 4E–G, and Supplementary Fig. S12, and Table S4). Using these two imaging systems, we observed distinct FRET histograms of virus Env and similar trends of histogram shifts across different experimental conditions: ligand-free, in the presence of ligands VRC34.01, VRC34.01 + sCD4 + 17b, and sCD4 + 17b (Figs. 4E–G and S12, Table S4). The distinct FRET histograms observed from the gp120 V1-V4 (Supplementary Fig. S12B, D) and gp120-gp41 (Supplementary Figs. S12F and 4F) imaging systems are expected, as they capture Env dynamics from two different structural perspectives (Supplementary Figs. S12C and 4E). Due to differences in viewing angles, the FRET efficiencies associated with each primary state vary between systems. Of note, the similarity in shift directions under ligand-free and ligand-present conditions (Figs. 4E–G and S12) suggests that Env undergoes global conformational changes at the population level, independent of the observation angle. In this analysis, we applied the previously well-defined three-state (PT, PC, CO) Gaussian distributions²⁸ to describe the FRET histograms, which reflect the overall conformational landscape of Env on virions. As expected, ligand-free Env exhibited predominance of the pre-triggered conformation, and Env, in the presence of sCD4 and 17b, prevailed in the fully open CD4-bound state. In the presence of VRC34.01, a decrease was observed in the PT population with an increase in the PC population, consistent with previously published results with the JR-FL Env⁷. For the VRC34.01 + sCD4 + 17b sample, the smFRET histograms suggested that the Env conformational distributions resided between the PC and CO conformations (Fig. 4F, and Supplementary Figs. S12B, D, F). Quantifying and comparing the propensity of each primary conformational state occupied by virus Env under ligand-free and different ligand-bound conditions, we observed distinct conformational effect on Env by VRC34.01 in the presence of sCD4 + 17b, positioned on the Env activation pathway between the effect of VRC34.01 alone and the sCD4 + 17b CO state (Fig. 4G and Supplementary Fig. S12E). These results were consistent between the observations from the gp120-gp41 (Fig. 4E–G and Supplementary Fig. S12F) and the gp120 V1-V4 structural perspectives (Supplementary Figs. S12B–E). Thus, smFRET analysis of the impact of VRC34.01 on the CD4,17b-bound Env was consistent with VRC34.01 stabilizing an intermediate state on the path of CD4-induced Env opening.

Visualizing conformations preceding CD4-induced transitions of HIV-1 Env

We next sought to visualize CD4-induced Env conformational transitions that occur upstream to Population 1 by performing single particle cryo-EM analysis on a sample of BG505 SOSIP that was incubated with sCD4 for 2 h, followed by the addition of VRC34.01 Fab 30 min before sample vitrification. As antibody 17b works synergistically with CD4 to open Env (Fig. 1C), we rationalized that excluding 17b may allow us to capture earlier stages of CD4-induced Env conformational changes. Two distinct particle populations were revealed in the cryo-EM dataset, which yielded reconstructions of 4.08 Å (Population 4) and 4.14 Å (Population 5) global resolutions. For both populations, all three protomers were bound to one each of CD4 and VRC34.01 Fab (Fig. 5A, B and Supplementary Fig. S13, and Table S1, S3). The two populations differed in the extent of rotation of their gp120 subunits. In Population 4, one of the three gp120 protomers was rotated roughly to the extent observed in the Population 1 structure, while the other two protomers were minimally rotated from their pre-receptor conformation. The distances between the CD4 binding site residue Asp 368 in the two minimally rotated protomers measured 60.2 Å and was thus closer to the distance observed in the closed, pre-receptor Env (~56 Å) (Fig. 3E) than to the distances measured in the partially open Population 1 intermediate (~75–78.2 Å) (Fig. 3F). The third protomer that had rotated was separated in this measure from the two other protomers by 61.7 Å and 66.9 Å. The bridging sheet and the α0 helix, which are the structural components associated with CD4-induced Env opening, were only seen in the rotated gp120 protomers. Although two protomers were minimally rotated, their V1V2 regions were unstructured, thus suggesting that CD4-induced disruption of the V1V2 cap of the closed, pre-receptor Env precedes the rotation of the protomers. This is consistent with time-resolved, temperature-jump small-angle x-ray scattering studies that suggest an order-to-disorder transition in the trimer apex precedes Env transitions involving protomer rotation²⁹. In Population 5, two gp120 protomers were rotated and showed formation of the bridging sheet and α0 helix, while the third protomer remained minimally rotated. Both Populations 4 and 5 were bound to VRC34.01 Fab at all the three FP sites, as expected, since the gp120 subunits had not rotated far enough to allow the gp41 conformational changes required for FP burial and steric inaccessibility to antibodies.

**Fig. 5: Partially open early intermediates and a stepwise mechanism for CD4-induced Env opening.**

In summary, Population 4 and 5 structures represent conformational states preceding the partially opened Population 1 conformation. Taken together, our results demonstrate sequential CD4-induced opening of the gp120 protomers and are consistent with previous studies that show initiation of CD4 binding to the closed Env begins with a single CD4, which induces opening of gp120 protomers needed for binding of additional CD4 molecules^16,30.

Source link