Cholesterol and Ceramide Facilitate Membrane Fusion Mediated by the Fusion Peptide of the SARS-CoV-2 Spike Protein

Abstract

SARS-CoV-2 entry into host cells is mediated by the Spike (S) protein of the viral envelope. The S protein is composed of two subunits: S1 that induces binding to the host cell via its interaction with the ACE2 receptor of the cell surface and S2 that triggers fusion between viral and cellular membranes. Fusion by S2 depends on its heptad repeat domains that bring membranes close together and its fusion peptide (FP) that interacts with and perturbs the membrane structure to trigger fusion. Recent studies have suggested that cholesterol and ceramide lipids from the cell surface may facilitate SARS-CoV-2 entry into host cells, but their exact mode of action remains unknown. We have used a combination of in vitro liposome-liposome and in situ cell-cell fusion assays to study the lipid determinants of S-mediated membrane fusion. Our findings reveal that both cholesterol and ceramide lipids facilitate fusion, suggesting that targeting these lipids could be effective against SARS-CoV-2. As a proof of concept, we examined the effect of chlorpromazine (CPZ), an antipsychotic drug known to perturb membrane structure. Our results show that CPZ effectively inhibits S-mediated membrane fusion, thereby potentially impeding SARS-CoV-2 entry into the host cell.

Figure 1

(A) Sequence alignment of Spike proteins from all human coronaviruses in the region immediately following the S2′ cleavage site known to be critical for SARS-CoV-2 fusion. The alignment was performed with the Clustal Omega program using the following sequences obtained from UniProt: SARS-CoV2 (P0DTC2), SARS-CoV (P59594), MERS-CoV (K9N5Q8), HCoV-NL63 (Q6Q1S2), HCoV-229E (P15423), HCoV-OC43 (P36334), and HCoV-HKU1 (Q5MQD0). The fusion peptides FP1 (red) and FP2 (blue) are highly conserved notably among MERS-CoV, SARS-CoV, and SARS-CoV-2. (B) Experimental setup used to study the capacity of SARS-CoV-2 fusion peptides to mediate membrane fusion in vitro. Fusion peptides with a C-terminal His₆ tag were reconstituted at t = 0 of the assay into fluorescently labeled liposomes mimicking the viral envelope (v-liposomes) by binding to lipids having an NTA-Ni headgroup. Fusion was monitored between v-liposomes and unlabeled peptide-free liposomes mimicking the cellular membrane (c-liposomes) using a FRET-based lipid mixing assay.

Figure 2

(A) Representative FRET-based lipid mixing experiments between v-liposomes (composed of 92 mol % PC, 5 mol % NTA-Ni, 1.5 mol % NBD, and 1.5 mol % Rho) and c-liposomes of various lipid compositions (including or not 10 mol % PE and 30 mol % CHOL or 10 mol % PE, 30 mol % CHOL, and 20 mol % CER at the expense of PC ) in the absence (yellow) or presence of FP1 (red) or FP2 (blue) added at t = 0 (500 μM lipids and 25 μM peptides). FP1 induced efficient lipid mixing between v- and c-liposomes exclusively composed of PC lipid. Fusion was strongly activated when the c-liposome membrane contained PE and CHOL or PE, CHOL, and CER. No significant lipid mixing was measured under the same conditions with the FP2 peptide, or when the fusion peptides were not lipid-anchored (v-liposomes devoid of NTA-Ni lipids; see Figure S1). (B) Average extent of lipid mixing after 90 min (n = 4–9 independent experiments; error bars are standard errors).

Figure 3

(A) Representative FRET-based lipid mixing experiments between v-liposomes (composed of 92 mol % PC, 5 mol % NTA-Ni, 1.5 mol % NBD, and 1.5 mol % Rho) and c-liposomes of various lipid compositions (including or not 10 mol % PE and 30 mol % CHOL or 10 mol % PE, 30 mol % CHOL, and 20 mol % CER at the expense of PC) in the presence of FP1 and in the presence/absence of CPZ, both added at t = 0 (500 μM lipids, 25 μM peptides, and 50 μM CPZ). All fusion experiments (with or without CPZ) were performed with a final DMSO concentration of 1% (v/v) in buffer H. CPZ strongly inhibited FP1-mediated fusion between v- and c-liposomes regardless of the c-liposome membrane lipid composition. (B) Average extent of lipid mixing after 90 min (n = 4–6 independent experiments; error bars are standard errors). (C) Percentage of fusion inhibition by CPZ after 90 min.

Figure 4

Effect of drugs on in situ cell–cell fusion. (A) Assay principle: HeLa cells stably expressing HiBit-Hsp70 and either co-expressing C-terminally Flag-tagged wild-type Spike (S-Flag) or not were co-cultured with HeLa cells stably expressing LgBit and either co-expressing ACE2 or not. Cell–cell fusion triggers content mixing and nanoluciferase complementation. (B) Donor cells (HiBit-Hsp70 positive cells) either co-expressing S-Flag or not (mock) were co-cultured with acceptor cells (LgBit positive cells) either co-expressing ACE2 or not (mock). After 24 h, Fumonisin B1 (20 μM) or MβCD (2 mM) and/or CPZ (10 μM) were added for 24 h prior to reading nanoluciferase activity. The graph represents the percentage of nanoluciferase activity for each condition. Nanoluciferase activity resulting from content mixing between donor cells co-expressing S-Flag and acceptor cells co-expressing ACE2 was set to 100%. The inset shows the background nanoluciferase activity measured when the donor cells do not co-express S-Flag (n = 2 independent experiments in triplicates; error bars represent standard deviations of all replicates).

Figure 5

(A) Structure of CHOL and CPZ. Both molecules are amphipathic and display a ring-like planar structure. (B) Proposed mechanism of action for FP1 on the cellular membrane. FP1 may interact with the boundaries of membrane domains (depicted in blue) that contain CHOL or CER molecules (not shown here for clarity). CPZ may inhibit FP1–membrane interaction and FP1-induced fusion by altering lipid order, specifically increasing the bilayer thickness outside of membrane domains while decreasing it within these domains.