Strikingly Different Roles of SARS-CoV-2 Fusion Peptides Uncovered by Neutron Scattering

Abstract

Coronavirus disease-2019 (COVID-19), a potentially lethal respiratory illness caused by the coronavirus SARS-CoV-2, emerged in the end of 2019 and has since spread aggressively across the globe. A thorough understanding of the molecular mechanisms of cellular infection by coronaviruses is therefore of utmost importance. A critical stage in infection is the fusion between viral and host membranes. Here, we present a detailed investigation of the role of selected SARS-CoV-2 Spike fusion peptides, and the influence of calcium and cholesterol, in this fusion process. Structural information from specular neutron reflectometry and small angle neutron scattering, complemented by dynamics information from quasi-elastic and spin-echo neutron spectroscopy, revealed strikingly different functions encoded in the Spike fusion domain. Calcium drives the N-terminal of the Spike fusion domain to fully cross the host plasma membrane. Removing calcium, however, reorients the peptide back to the lipid leaflet closest to the virus, leading to significant changes in lipid fluidity and rigidity. In conjunction with other regions of the fusion domain, which are also positioned to bridge and dehydrate viral and host membranes, the molecular events leading to cell entry by SARS-CoV-2 are proposed.

Figure 1

Interaction of fusion peptides with biomimetic lipid monolayers. (A, B, C) BAM images of the PM monolayer before (at a surface pressure of 22 ± 1 mN·m^–1) and after injection of either FP1 or of FP4. Scale bars are 100 μm. (D, E) Tensiometry binding analysis for FP binding to the PM determined from lateral pressure changes in the lipid monolayer. (D) Uppermost surface pressure increment due to the interaction of FP1, FP2, and FP4 with a PM monolayer, at 3 μM peptide concentration and Π₀ = 22 ± 1 mN·m^–1. The increment in the pressure, ΔΠ, is proportional to the amount of peptide partitioning with the monolayer at the air/water interface, either through binding and/or insertion. (E) Equilibrium analyses for FP1 (green), FP2 (orange), and FP4 (blue) binding to PM. Lines are fits to the data obtained through the Hill-Langmuir equation, as described in SI Methods, eq S1. The dashed lines indicate the fits of the experimental data with the Hill coefficient set to 1. (F) CD molar ellipticity profiles of FP1 (green) and FP4 (blue), in the absence (dotted line) and presence of liposomes (continuous line). The decrease in signal in the presence of lipids suggests a related increase in protein secondary structure.

Figure 2

Neutron reflectivity data of PM monolayers and bilayers with FP1, FP2, and FP4. (A, B, C) Volume fraction profiles normal to the interface of PM monolayers in the absence of Ca²⁺ (derived from data plotted in Figure S3, as described in SI Methods) highlight the distribution of tails (black), heads (magenta), water (cyan), and (A) FP1 (green), (B) FP2 (orange), and (C) FP4 (blue). (D) Bar diagram plot summarizing the volume fraction occupied by the solvent in the lipid headgroup layer of the PM monolayers without any peptide and with FP1, FP2, and FP4. (E, F, G) Volume fraction profiles of solid supported lipid bilayers in the absence of Ca²⁺ (derived from data plotted in Figure S4) highlight the distribution of Si/SO₂ (dark yellow), tails (black), heads (magenta), water (cyan), and (E) FP1 (green), (F) FP2 (orange), (G) FP4 (blue). (H) Volume fraction profiles of solid supported lipid bilayer with FP1 in the presence of 2 mM Ca²⁺ (dark cyan) (derived from data plotted in Figure S5). (I, J) Volume fraction profiles of solid supported lipid bilayer with FP1 in the presence (I, 10 mM Ca²⁺, light green) and then after removal of Ca²⁺(J, after overnight incubation with EDTA, purple). (K) Volume fraction profiles of solid supported lipid bilayer with FP2 in the presence of 2 mM Ca²⁺ (maroon). Data from FIGARO at ILL.

Figure 3

(A, B, C) SANS curves of extruded liposomes (100 nm) with 35% cholesterol content by mass, mixed with (A) FP1, (B) FP2, and (C) FP4. Data for FP3 (no visible effect by SANS) are displayed in Figure S6, as well as data with two other membrane compositions (“O” and “E”). The peptide:lipid mole ratios are given in the legend of Figure 3C. The highest peptide concentrations were measured first, and more diluted compositions were only explored as long as effects were seen in the data, which is why the lowest peptide concentrations are only measured for FP4 (most effective peptide). Overall, the effects seen by SANS are relatively small (indicating preservation of membrane content) compared to macroscopic observation (large aggregates formation). FP2 has almost no effect. FP1 and mostly FP4 show the largest effects. For FP1, the change is in line with disk formation (constant amount of membrane but decreasing number of large vesicles leading to a reduced forward scattering), suggesting membrane perforation and, at the highest peptide concentration, stabilization of disks. For FP4, the increase of size observed at low q suggests binding and fusion between vesicles (aggregation without fusion would retain the shallow form factor oscillation due to the vesicles diameter, while it vanishes upon FP4 addition). The various hypotheses tested and their consequences on SANS data are illustrated in Figure S6. (D, E, F) SANS and NSE results for concentrated “P” and “O” liposomes in the presence of FP1 (at 1:60 and 1:200 peptide:lipid mole ratio). (D) SANS on liposomes “O” (without cholesterol). (E) SANS on liposomes “P” (35% cholesterol by weight). (F) NSE-derived bending rigidities of liposomes “O” and “P” in the presence of FP1. Addition of small amounts of FP1 leads to a slight softening of the membrane. Addition of larger amounts of FP1 leads to an apparent increase of rigidity for liposomes “O” lacking cholesterol, which is probably due to the formation of multilamellar structures. Data from D22 and IN15 at ILL.

Figure 4

QENS-data derived half width at the half-maximum (HWHM), of the Lorentzian function plotted versus q² for the pure lipids, and lipids-peptide complex at high (1:20) and low concentration (1:200) FP1 and FP2 peptides. The straight line represents the fit to the data performed using the jump diffusion model as described in the SI. The observed scattering signal arises mainly from the lipid membrane and represents an ensemble-average over the diffusive motions of the prevailing hydrogens atoms. The observed differences in the measured averaged signal are significant as per the diagonal elements of the covariance matrix of the fits (i.e., neglecting correlated errors). Data from IN5 at ILL.

Figure 5

Proposed fusion mechanism between SARS-CoV-2 and eukaryotic host membrane. The viral membrane bilayer is colored green, the eukaryotic host membrane in blue, and the S2′ protein is in red. The direction of the protein is drawn with red arrows, while the direction of the lipids is drawn with black arrows. FP1and FP4 are represented as ovals, and the structured S2′ protein as a circle (attached to the viral membrane). (A) FP1 forms a fusion initiation point on binding the host membrane. (B) The initiation point enlarges provoking lipids mixing between the viral and host membrane, leading to the growth of a hemifusion diaphragm. (C) FP4 bridges the two membranes together thereby facilitating the fusion of the two membranes into a single bilayer. Moreover, the two membranes coming together exclude the folded S2′ from the growing synapse. (D, E, F) A hemifusion diaphragm is formed and in the endosome, lower free calcium concentrations lead to FP1 orienting itself like FP4, thereby providing further contact between the two membranes. (G) It is also possible that FP1 initiation points may form on both the viral and the plasma membrane. (H) The two membranes form a pore, and as in (F), the pore expands as the Spike protein is excluded by the two membranes coming together due to the bridge function encoded in the Spike fusion peptides.