CD4-induced activation in a soluble HIV-1 Env trimer

Abstract

The HIV envelope glycoprotein (Env) trimer undergoes receptor-induced conformational changes that drive fusion of the viral and cellular membranes. Env conformational changes have been observed using low-resolution electron microscopy, but only large-scale rearrangements have been visible. Here, we use hydrogen-deuterium exchange and oxidative labeling to gain a more precise understanding of the unliganded and CD4-bound forms of soluble Env trimers (SOSIP.664), including their glycan composition. CD4 activation induces the reorganization of bridging sheet elements, V1/V2 and V3, much of the gp120 inner domain, and the gp41 fusion subunit. Two CD4 binding site-targeted inhibitors have substantially different effects: NBD-556 partially mimics CD4-induced destabilization of the V1/V2 and V3 crown, whereas BMS-806 only affects regions around the gp120/gp41 interface. The structural information presented here increases our knowledge of CD4- and small molecule-induced conformational changes in Env and the allosteric pathways that lead to membrane fusion.

Figure 1. HDX profile of SOSIP.664 trimers

(A) The sequence of mature, full-length Env is shown with the SOSIP modifications indicated. To the right, structural elements including variable loops 1–5, N/C termini, and heptad repeats of gp41 (HR1 and HR2) are mapped onto the ribbon diagram for one protomer from the BG505 SOSIP.664 trimer crystal structure (PDB 4NCO) (Julien et al., 2013a). (B,C) The H/D-exchange profiles are shown for unliganded BG505 and KNH1144 SOSIP.664 trimers. Percent exchange is shown after 3 s, 1 min, 30 min and 20 h for all observable peptic fragments at the midpoint of their primary sequence. For example, the exchange profile of fragment 105–111 is plotted at position 108. Individual exchange plots with errors are shown in Figures S2 and S3.

Figure 2. HDX-MS comparisons of 293F vs. 293S GnTI−/− expression system

A) The butterfly plot shows the exchange profiles of BG505 SOSIP.664 trimers from 293S GnTI−/− cells (all high mannose glycoforms, top) vs. 293F cells (contains complex glycosylation, bottom). B) The differences at each time point are plotted in the difference plot underneath, revealing no major changes attributed to differences in glycosylation types.

Figure 3. Changes upon CD4 binding by HDX

Butterfly plots comparing the exchange profiles for SOSIP.664 trimers unliganded (top) and in complex with sCD4 (bottom) for KNH1144 (A) and BG505 (B). The corresponding difference plots below highlight the regions that gain and lose protection upon sCD4 binding, plotted above and below the axis, respectively. (C) Regions that are more protected (blue) or less protected (red) upon CD4 binding are mapped on the BG505 SOSIP.664 trimer crystal structure (PDB 4NCO) (Julien et al., 2013a). KNH1144 sequence data were used for these heat maps due to the greater coverage, although BG505 shows nearly identical trends for the regions covered. Individual exchange plots with errors are shown in Figures S2 and S3. Labeled spheres highlight the specific residues monitored by oxidative labeling.

Figure 4. Solvent accessibility changes upon CD4 binding monitored by oxidative labeling

(A) Modification rates of peptides used as “dosimeters” for ensuring that both the unliganded and CD4-bound data sets were exposed to similar amounts of oxidative labeling. The signal for the unmodified form (normalized to a non-irradiated sample) of Leucine Enkephalin (left) and Substance P (right) are shown as a function of exposure to synchrotron radiation. (B) The degree of oxidation is shown as a function of radiation exposure for various regions of SOSIP.664 (BG505) in the unliganded (gray circles) or CD4-bound (squares) state. Percent modified refers to the signal intensity of the oxidized from (+16 Da) relative to the sum of signals for the modified and unmodified peptide, with the sites of oxidation shown in bold. Error bars represent the standard deviation between duplicate measurements. Examples of mass spectra revealing changes in oxidation rates are shown in Figure S6. The oxidation sites are illustrated on the trimer in Figure 3C. *Data for peptide 656–664 encompass oxidation at residues L660, L661, and L663.

Figure 5. Changes upon NBD-556 and BMS-806 binding by HDX

Butterfly and difference plots showing the effects of NBD-556 (A), and BMS-806 (B) on KNH1144 SOSIP.664 trimers. Differences upon binding NBD-556 (C) and BMS-806 (D) are plotted on the trimer crystal structure (PDB 4NCO), revealing regions that are more protected (blue) or less protected (red) upon CD4 binding. All of the major changes in the highlighted regions are significant as assessed by the experimental error (individual plots shown in Figure 6).

Figure 6. Key CD4-induced changes in trimeric SOSIP.664 monitored by HDX-MS

(A) The SOSIP.664 trimer crystal structure (PDB 4NCO) (Julien et al., 2013a) was modeled into the EM density of unliganded Env on virions (EMD 5019; (Liu et al., 2008): V1/V2 (yellow) and V3 (green), the bridging sheet (red), HR1 (purple) and gp120 inner domain (black). (B) The sCD4-bound gp120 core structure (PDB 3JWD; (Pancera et al., 2010) was modeled into the sCD4-bound electron density map (EMD 5455) (Tran et al., 2012) with CD4 shown in blue. In the CD4-bound state, the bridging sheet is reorganized, positioning V1/V2 away from the trimeric interface and exposing the V3 loop. Individual exchange profiles of the key regions of interest are shown for unliganded trimer (blue), sCD4 bound (red), NBD-556 bound (orange) and BMS-806 bound (cyan). Error bars represent standard deviations from duplicate measurements. EM renderings were made with Chimera (Pettersen et al., 2004).

Figure 7. Allosteric networks in Env trimers

CD4 binding leads to two allosteric effects within Env trimers (red to yellow pathways). Pathways are highlighted on a single protomer of the trimer with neighboring protomers rendered transparent. (A) Repositioning of the elements of the bridging sheet leads to the disruption of the trimeric interactions within V1/V2 and V3, while somehow also affecting the FPPR, whose location and relation to the crown region have yet to be interpreted in high resolution structures. (B) The conformational changes induced within three layers of the inner domain also influence HR1 within gp41.