Figure 7. Characteristics of significantly shifted sites.

(A) One monomer of the closed pre-fusion Env trimer (Stewart-Jones et al., 2016) is colored from white to red according to the magnitude of the mutational shift at each site (red indicates large shift). Sites that are significantly shifted according to Figure 6B are in spheres, and all other sites are in cartoon representation. (B) There is no significant difference in the relative solvent accessibility of sites that have and have not undergone significant shifts. This observation holds for both the closed trimer conformation in (A) and the CD4-bound trimer conformation (Ozorowski et al., 2017). The absolute solvent accessibility of each site was calculated using DSSP (Kabsch and Sander, 1983) and normalized to a relative solvent accessibility using the absolute accessibilities from Tien et al. (2013). (C) Sites of significant shifts are clustered in the structures of both the closed and open Env trimers. The left two plots show the distance of each significantly shifted and not-shifted site to the closest other shifted site in the indicated structure. The right-most plot shows the minimum distance across both conformations. The trend for shifts to cluster becomes stronger when considering the minimum distance, suggesting multiple conformations contribute to this trend. (D) Large mutational shifts are not strongly enriched at sites that have substituted between BG505 and BF520, or at sites that contact sites that have substituted. The plots show the magnitudes of the shifts among structurally resolved sites that have substituted between BG505 and BF520, the non-substituted sites that physically contact a substitution in the indicated structure(s) (any non-hydrogen atom within 3.5 angstroms), and all other sites. Figure 7— source data 1 shows that there is a borderline-significant tendency of significantly shifted sites to have substituted. All plots only show sites that are structurally resolved in the indicated structure(s). Structural distances and solvent accessibilities were calculated using all monomers in the trimer. P-values were calculated using the Mann-Whitney $U$ test. Figure 7—figure supplement 1 and Figure 7—figure supplement 2 zoom in on some relevant clusters of sites.

Figure 7—figure supplement 1. Cluster of shifted sites in the post-fusion six-helix bundle of Env.

The left-most panel shows the six-helix bundle of Env (Weissenhorn et al., 1997), focusing on the central three helices, each of which is derived from a different Env monomer and which pack together to form the bundle’s core. The other three helices that make up this bundle are located in another part of the structure. There are $\sim$80 residues per monomer that are resolved in this structure, including four significantly shifted sites and five substituted sites. Of these, three of the shifted sites (582, 583, and 587) and one substituted site (588) form a cluster at one end of the bundle. We show the side chains for each of these sites colored according to whether the site has shifted (red) or substituted (yellow). We wondered whether this cluster was unique to the six-helix bundle, but found that it is also present in roughly the same structural arrangement in both the closed pre-fusion conformation (center panel) and the open CD4-bound conformation (right panel; note that some of the surrounding residues in both the closed and open structures have been hidden for the sake of clarity). As shown in Figure 6C, each of the shifted sites are more tolerant of mutations in BG505 than in BF520. The two shifted sites that point inward towards the bundle’s core (583 and 587) may be important for packing. Based on these structures, it is difficult to discern if or how the substituted site may impact the preferences at the shifted sites nearby.

Figure 7—figure supplement 2. Clusters of shifted sites in highly dynamic regions of Env.

A fine-grained view of two clusters of shifted sites, both of which are in conformationally dynamic regions of Env. In each structure, we only show side chains for shifted sites or other sites of interest, which include substituted sites and sites that are proposed to take part in a hydrophobic network that helps control Env’s dynamics (Ozorowski et al., 2017). These side chains are colored according to the key in the upper right. (A) A cluster of four shifted sites at the apex of the closed pre-fusion conformation of Env (Stewart-Jones et al., 2016), a region that is not resolved in the open CD4-bound conformation. These sites are in Env’s first/second variable loop (164 and 165) and its third variable loop (307 and 309), which pack against one another and against an adjacent Env protomer (the inter-protomer boundary is indicated by a dashed line). Each of the shifted sites has substituted. And these sites are in the immediate vicinity of other non-shifted, but substituted sites. Thus, it seems likely that these shifts at least partially arise from short-range epistatic interactions within this cluster. However, longer range epistatic interactions with more distant sites also seem plausible given the highly dynamic nature of this region. (B) A cluster of shifted sites that are adjacent to a network of hydrophobic residues (blue/purple) that help mediate the conformational change between the closed pre-fusion state to the open CD4-bound state (Ozorowski et al., 2017). One of the shifted sites (purple) is in this network. These sites are shown in context of both structural states, the latter of which is from Ozorowski et al., 2017). Nearly one third of the shifted residues cluster within this structural region (9/30). In contrast to panel (A), only a few of the shifted sites are adjacent to substitutions or have substituted themselves. This trend is found in both the closed and open conformations, suggesting that the shifts may be primarily due to long-range epistatic interactions.