Lancemaside A inhibits SARS-CoV-2 entry by blocking S protein-mediated viral membrane fusion
Having observed that LA inhibits two main SARS-CoV-2 entry routes to similar extent, we reasoned that a possible target for LA could be a common step of both viral entry pathways. Therefore, we first examined the effect of LA on the interaction between SARS-CoV-2 spike (S) protein and ACE2. To do so, a recombinant protein containing receptor-binding domain (RBD) of the SARS-CoV-2 S protein fused with GFP (S-RBD-GFP) was produced and added into ACE2+ H1299 cells (Figure 3a). Flow cytometry showed that more than 97% of the cells were bound with S-RBD-GFP (Figure 3b), reflecting SARS-CoV-2 S attachment to cellular receptor for viral cell entry. Importantly, pretreatment of ACE2+ H1299 cells with CL root extracts or LA did not affect the binding of S-RBD-GFP to ACE2 on the cells within the effective concentration range we obtained in the anti-SARS-CoV-2 assays (Figure 1a and 2), despite somewhat inhibitory effects were seen at higher concentrations (Figure. 3b). Moreover, LB, LD, and triterpenoid aglycone such as echinocystic acid (EA) did not display any inhibitory effects until their concentration reached 20 µM (Figure S1). Overall, these results suggest that S-ACE2 protein-protein interaction is not a main target for LA to inhibit SARS-CoV-2 infection. Thus, as a next step, we tested another possibility that LA may prevent fusion between the viral and host membranes, a process that commonly occurs in both entry pathways. For the assay, we established two stable cell lines: one that overexpresses S protein with EGFP from a single bicistronic mRNA in HEK293 cells (Spike-HEK293) and the other overexpressing mRuby in ACE2/TMPRSS2+ H1299 cells. Then, dual imaging time-lapse microscopy was performed to monitor cell fusion between these two cells as indicative of the S-mediated membrane fusion during viral entry (Figure S2). Addition of Spike-HEK293 cells to a monolayer of ACE2/TMPRSS2+ H1299 rapidly induced the cell-to-cell fusion, which allows them to fuse continuously with neighboring ACE2/TMPRSS2+ H1299 due to the S protein displayed on the surface of fused hybrid cells, eventually giving rise to enlarged multinucleated cells (Figure 3c). The heterologous cell fusion is mediated by the interaction between S protein and ACE2, because this process was not observed in co-culture of HEK293 expressing GFP only with ACE2/TMPRSS2+ H1299 cells (Figure 3c). We next examined the effects of LA on the cell-to-cell fusion event. Strikingly, LA treatment almost completely blocked the fusion of Spike-HEK293 cells with ACE2/TRMPRSS2+ H1299 cells, implicating impairment of SARS-CoV-2 entry by LA at the membrane fusion step (Figure 3c). Furthermore, we quantified the cell-to-cell fusion by flow cytometry analysis after co-culturing the two cell types in a ratio 1 to 3. The results showed that approximately 30% of total cells were double-positive for mRuby and GFP, indicating that most of Spike-HEK293 were participated in the fusion event (Figure 3d). Meanwhile, the double-positive cells were barely detected when control GFP-HEK293 cells (no S protein) were co-cultured with ACE2/TMPRSS2+cells, again demonstrating the requirement of S protein during the cell-to-cell fusion (Figure 3d). Notably, LA pretreatment strongly blocked the fusion event, with less than 4 % of cells generating cell fusion hybrids (Figure 3d). On the other hand, LB and LD showed limited or no anti-fusion activity (Figure 3d), further supporting our results obtained in pSARS-CoV-2 entry assay (Figure 1d). Next, we investigated how LA may affect membrane fusion. We previously reported that PD, a triterpenoid saponin derived from PG, redistributes membrane cholesterol and inhibits SARS-CoV-2 infection (Kim et al., 2021). Membrane cholesterol is essential component for SARS-CoV-2 to gain entry into the host cells through membrane fusion (Sanders et al., 2021). As LA has a similar chemical structure to PD, we reasoned that LA might also alter membrane cholesterol content and undertook confocal microscopy analysis of filipin staining to investigate membrane cholesterol distribution after LA treatment. We found that 1 h treatment of ACE2+ H1299 cells with 10 μM LA significantly led to an increase in cholesterol content at the plasma membrane and in endosomes (Figure 3e and 3f). Importantly, these redistributions of cholesterol were also observed upon treatment of parental H1299 cells with LA (Figure 3e and 3f), indicating that LA affects cell membrane directly, not by affecting ACE2. Moreover, we treated ACE2+ H1299 cells with LB, LD, and EA, and compared their effects on membrane cholesterol with that of LA. The cells showed that LB also increased the amount of cholesterol in plasma membrane by about 1.7-fold above DMSO-treated control, as revealed by filipin staining, but it was lower than that of LA (2.3-fold increase) (Figure S3). In addition, LD and EA did not show any effects on membrane cholesterol content (Figure S3). Intriguingly, the increased levels of membrane cholesterol caused by triterpenoid saponins from CL is well correlated with their inhibitory ability against S-mediated membrane fusion and viral infection (Figure 3d and Figure 1d), supporting the idea that membrane cholesterol might be a direct target for LA to exert its anti-SARS-CoV-2 activity. Taken together, our data suggest that LA interferes with the S-mediated membrane fusion, possibly by altering distribution of cholesterol on the host cell membrane, leading to inhibition of main SARS-CoV-2 infection routes.
Lancemaside A blocks syncytia formation Given that LA has an inhibitory activity for membrane fusion, we next sought to investigate the effects of LA on the formation of multinucleated giant cells called syncytia, which is a result of the continuous fusion of SARS-CoV-2-infected cells with neighboring cells and often detected in lung tissue from COVID-19 patients (Buchrieser et al., 2020; Bussani et al., 2020). For the assay, we employed a spit-GFP complementation technology, in which half of GFP is expressed separately in different cells and a functional GFP protein can be reconstituted upon cell-to-cell fusion. ACE2/TMPRSS2+ H1299 cells stably expressing each of the two fragments of the reporter protein were generated and plated in equal number, and then S protein was ectopically expressed in these cell mixtures to induce syncytia formation, which can be detected by GFP fluorescence (Figure 4a and 4b). In DMSO-treated control cells, multiple enlarged GFP+ cells with clustered nuclei were detected, indicating syncytia were formed (Figure 4c, upper panel). Using automated image analysis, we quantified the efficiency of syncytia formation by dividing the number of nuclei in GFP+multinucleated cells by the total number of nuclei in the field of view (Figure 4c, lower panel) and found that approximately 30% of the total number of cells plated formed GFP+ syncytia (Figure 4d). Importantly, the number of GFP+ cell clusters was dose-dependently reduced by LA treatment, with IC50values of 3.94 μM, indicating that LA effectively inhibits the S-mediated syncytia formation (Figure 4c and 4d). As virus-induced syncytia formation generally promotes programmed cell death (Hooper, Zaki, Daniels & Middleton, 2001; Nardacci, Perfettini, Grieco, Thieffry, Kroemer & Piacentini, 2015), we next asked if LA is able to prevent the viral activation of apoptotic pathways. Consistent with previous studies, the S-mediated syncytia promoted apoptosis as evidenced by western blot analysis of the proteolytic cleavage of caspase 3 and 9 (Figure 4e). Notably, these apoptotic markers were not detected under no S expression and LA pretreatment conditions (Figure 4e). Together, these results indicate that LA blocks the S-induced syncytia formation and consequent apoptotic cell death.