Testing the feasibility of targeting a conserved region on the S2 domain of the SARS-CoV-2 spike protein

Abstract

The efficacy of vaccines against the SARS-CoV-2 virus significantly declines with the emergence of mutant strains, prompting investigation into the feasibility of targeting highly conserved but often cryptic regions on the S2 domain of spike protein. Using tools from molecular dynamics, we find that exposure of a conserved S2 epitope located in the central helices below the receptor binding domains would require large-scale motion beyond receptor binding domain up-down motion, but, along the reaction coordinates we explored, it is unlikely to be exposed by such large-scale dynamic fluctuations of the S1 domain without any external facilitating factors, despite some previous computational evidence suggesting transient exposure of this region. Furthermore, glycans, particularly those on N165 and N234, hinder S2-exposing opening dynamics, and thus stabilize spike in addition to immunologically shielding the protein surface. Although the S2 epitope region examined here is central to large-scale conformational changes during viral entry, free energy landscape analysis obtained using the path coordinate formalism reveals no inherent "loaded spring" effect, suggesting that a vaccine immunogen would tend to present the epitope in a prefusion-like conformation and may be effective in neutralization. These findings contribute to a deeper understanding of the dynamic origins of the function of the spike protein, as well as further characterizing the feasibility of the S2 epitope as a therapeutic target.

Figure 1

Simulation systems. (A) The head portion of the spike protein with one RBD in the up state. A portion of the S2 central helix region, the erected RBD, and its nearest NTD are shown using ribbons. Glycans are shown in licorice. The reaction coordinate is the angle θ determined by the centers of mass P, Q, and R (marked with gray spheres) of residue groups given in the main text. The NTD that moves with the RBD is shown in dark blue. An S2-exposing open state structure where the system has moved further along the reaction coordinate is also shown on the right. (B) Residues 912–1034 used in S2 helix extension simulation. These reference structures were obtained after equilibration of the prefusion bent state (right) and the postfusion extended state (left). We assumed that the extended membrane spanning intermediate is in the postfusion conformation. (C) A contact map of residue pairs used to define the reaction coordinate between pre- and postfusion states in (B) (Eq. 2). The matrix elements corresponding to the pairs included in the calculation are shown in black. To see this figure in color, go online.

Figure 2

(A) The PMF of the glycosylated (dark pink) and unglycosylated (light blue) spikes along the S2-exposing opening angle. Arbitrarily selected frames from the sampling are shown below, with corresponding angles marked by vertical gray dashed lines. The angle threshold of 22.7° is marked with arrows. (B and C) Boxplots showing the AASA of the S2 region for glycosylated (dark pink) and unglycosylated spike (light blue) along the same coordinate as in (A). Box boundaries correspond to the first and third quartiles, whiskers extend up to the last data point within 1.5 times the interquartile distance, outliers are denoted as points, the medians as horizontal lines, and the means as black dots. (D) The means from (B) and (C) are plotted together for comparison of the two systems. A plot of PMF against AASA is shown in Fig. S14A. To see this figure in color, go online.

Figure 3

(A) A snapshot of the up state spike, taken from the last frame of the umbrella with bias center closest to the minimum of the PMF. The opening RBD, located on chain A, is colored orange. The closest NTD, located on chain C, is colored dark blue. Residues 950–1030 of the S2 domain are colored as in Fig. 1A. Fourteen glycans for which free energy perturbation theory is implemented are shown in black. The strongest contributing glycans, on N165(C) and N234(C), are highlighted in surface representation. (B) A closer look at the neighborhood of glycans N165(C) and N234(C), as seen from above. (C) Each plot shows the PMFs before (pink line) and after (black line) removing the interactions between each of 14 glycans and the rest of the system. The bottom right plot shows the resulting PMF after (nonperturbatively) removing all 14 glycans. To see this figure in color, go online.

Figure 4

PMF calculated from S2 helix extension simulations. The extended structure is at ${\xi }_A=-0.993$ and the prefusion bent “helix-turn-helix” structure is at ${\xi }_B=+0.993$. Arbitrarily chosen structures from the umbrella sampling trajectories are annotated along the PMF to show how the structure changes along the reaction coordinate. The location of reference states A and B are shown as vertical dashed lines. The shaded region near $\xi =0$ samples states with nonzero boundary potential (see Fig. S17). To see this figure in color, go online.

Figure 5

Phase space explored in helix extension simulations. The postfusion state A is located at $s\left(X,A\right)=1$, $s\left(X,B\right)=0.486$ and the prefusion state B is at $s\left(X,A\right)=0.486$, $s\left(X,B\right)=1$. Umbrella centers along the coordinate ξ are shown as black lines. The circular arc with radius 0.69 centered at $\left(1,1\right)$ is the start of the repulsive radial boundary potential. Positions of structures sampled during the second half of the 16 ns umbrella sampling are shown as bands of various colors. The final PMF obtained (see Fig. 4) is shown in the background with a color scale. To see this figure in color, go online.