SARS-CoV-2 Spike Protein Is Capable of Inducing Cell-Cell Fusions Independent from Its Receptor ACE2 and This Activity Can Be Impaired by Furin Inhibitors or a Subset of Monoclonal Antibodies

Abstract

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), which was responsible for the COVID-19 pandemic, efficiently spreads cell-to-cell through mechanisms facilitated by its membrane glycoprotein spike. We established a dual split protein (DSP) assay based on the complementation of GFP and luciferase to quantify the fusogenic activity of the SARS-CoV-2 spike protein. We provide several lines of evidence that the spike protein of SARS-CoV-2, but not SARS-CoV-1, induced cell-cell fusion even in the absence of its receptor, angiotensin-converting enzyme 2 (ACE2). This poorly described ACE2-independent cell fusion activity of the spike protein was strictly dependent on the proteasomal cleavage of the spike by furin while TMPRSS2 was dispensable. Previous and current variants of concern (VOCs) differed significantly in their fusogenicity. The Delta spike was extremely potent compared to Alpha, Beta, Gamma and Kappa, while the Omicron spike was almost devoid of receptor-independent fusion activity. Nonetheless, for all analyzed variants, cell fusion was dependent on furin cleavage and could be pharmacologically inhibited with CMK. Mapping studies revealed that amino acids 652-1273 conferred the ACE2-independent fusion activity of the spike. Unexpectedly, residues proximal to the furin cleavage site were not of major relevance, whereas residue 655 critically regulated fusion. Finally, we found that the spike's fusion activity in the absence of ACE2 could be inhibited by antibodies directed against its N-terminal domain (NTD) but not by antibodies targeting its receptor-binding domain (RBD). In conclusion, our BSL-1-compatible DSP assay allowed us to screen for inhibitors or antibodies that interfere with the spike's fusogenic activity and may therefore contribute to both rational vaccine design and development of novel treatment options against SARS-CoV-2.

Keywords: ACE2; antibodies; cell–cell fusion; furin; glycoprotein; inhibitors; neutralization; receptor-independent; screening; severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2); spike; therapy; variants of concern (VOCs).

Figure 1

Dual split protein (DSP) assay for the automated quantification of cell–cell fusion. (a) Schematic representation of the DSP assay. (i) HEK293T cells were transduced with lentiviruses coding for split chimeric reporter proteins composed of GFP and Renilla Luciferase (RLuc), resulting in DSP1-7 and DSP8-11 cells, which were mixed in a 1:1 ratio. (ii) Co-cultured DSP cells were transfected with plasmids encoding diverse viral fusion proteins either alone or in combination with a plasmid coding for their corresponding receptor. Membrane fusion was quantified through detection of GFP-fluorescence and/or bioluminescence upon RLuc-mediated cleavage of its cell-permeable substrate EnduRen. (iii) The DSP assay was used to identify monoclonal antibodies and inhibitors that block viral glycoprotein-mediated cell–cell fusions. This figure was generated using templates from SMART Servier Medical Art By Servier, used under CC BY 3.0, https://smart.servier.com/, accessed on 1 June 2022. (b) Immunofluorescence analyses of lentivirally transduced HEK293T cells 293T-DSP1-7, 293T-DSP8-11 and 293T-DSP-mix. The individual DSP cell cultures were transfected with pcDNA3-based vectors encoding DSP1-7 (panels a and b), DSP8-11 (panels c and d) or a combination of both (DSP1-7 + DSP8-11; panels e and f). Two days later, cells were fixed, cell nuclei visualized through DAPI and the GFP-fluorescence of the reconstituted DSP reporter protein examined through confocal laser scanning microscopy. (c) Automated quantification of the GFP counts obtained after transfection of the respective DSP cells (293T-DSP1-7, -DSP8-11 and -DSP-mix) with the indicated DSP expression constructs using a CTL-Fluorospot reader. The numbers in the left corner of each image represent the GFP counts of each well as calculated using the software ImmunoSpot (version 6.0.0.2). (d) Validation of the assay via transfection of 293T-DSP-mix cells with gB of human cytomegalovirus (HCMV) (panel b) or the intrinsically fusion-active derivative gB/VSV-G (panel c). The GFP counts measured using the Fluorospot reader (upper left corner of the immunofluorescence images) are diagramed as a column chart.

Figure 2

Cell fusion activity of SARS-CoV-2 spike protein is not dependent on ACE2 expression. (a,b) The 293T-DSP-mix cells were transfected with vectors encoding the indicated viral fusion proteins or an empty vector (mock). Cell–cell fusions were quantified (a) via GFP using the CTL-Fluorospot reader or (b) via bioluminescence, given in relative light units (RLU) upon RLuc-mediated cleavage of EnduRen. (c) Immunofluorescence analyses of 293T-DSP-mix cells transfected with SARS-CoV-2 spike alone (panels a–d) or in combination with its receptor ACE2 (panels e–h). Three days later, cells were fixed, cell nuclei visualized using DAPI and the GFP-signal of the reconstituted DSP-reporter protein (panels b and f) examined through confocal laser scanning microscopy. As a control, spike expression was confirmed by using mAb TRES567 (panels c and g). (d) Pseudovirus entry assay using SARS-CoV-2 spike (CoV-2 spike) or mock (pcDNA3) pseudotyped lentiviral particles for infection of 293T-DSP-mix cells, which were either transfected with (+ACE2) or without ACE2 (-ACE2). Two days post-infection, luciferase reporter gene activity was measured, and this is given in RLU.

Figure 3

Proteolytic cleavage of the SARS-CoV-2 spike protein by furin but not TMPRSS2 is required for ACE2-indpendent cell–cell fusions. (a) Top: Schematic of SARS-CoV-2 spike, which is proteolytically processed by the cellular proteases furin, Cathepsin L and/or TMPRSS2 into its S1, S2 and S2′ subunits, and this cleavage can be blocked by CMK or camostad/nafamostat, as indicated. The magnification shows an amino acid alignment of the respective protease cleavage sites of spike proteins encoded by SARS-CoV-1 [YP_009825051.1] and SARS-CoV-2 [YP_009724390.1]. Numbers indicate the position of the bordering amino acids; basic residues potentially involved in recognition by cellular proteases are highlighted in bold font. Mutants that harbor alanine substitutions in the protease cleavage sites of SARS-CoV-2 spike are indicated at the bottom. (b) DSP fusion assay using 293T-DSP-mix cells transfected with plasmids encoding CoV-2 spike or the respective proteolytic cleavage site mutants as indicated. (c) Western blot analysis using lysates of HEK293T cells transfected with plasmids encoding the indicated spike proteins. Beta-actin detection served as a loading control. (d–f) DSP fusion assay with 293T-DSP-mix cells co-expressing ACE2 and CoV-2 spike (d) or expressing CoV-2 spike alone (e,f) were treated at 4h post-transfection with the furin inhibitor CMK (10 µM), serial dilutions thereof (f) or the TMPRSS2 protease inhibitors camostat (100 µM) and nafamostat (20 µM), respectively (d,e). Nontreated cells (w/o) served as internal controls. Cell–cell fusions were quantified through bioluminescence detection following addition of the substrate EnduRen. (f) The data from (e) were used to calculate the CMK-mediated inhibition of fusion in percent based on the RLU in relation to the nontreated control.

Figure 4

Spike proteins from SARS-CoV-2 variants of concern greatly differ in their capacity to induce cell–cell fusions. (a–f) 293T-DSP-mix cells were transfected with empty vector (mock) or plasmids encoding CoV-2 spike wildtype (WT), mutants (FURmut and pre-lock) or the indicated VOC either alone (a–e) or upon co-transfection with a plasmid coding for ACE2 (d–f). Cell–cell fusions were quantified using the DSP assay for GFP counts as determined with the CTL-fluorospot reader (a,d) or luciferase reporter activity (b,e). (c,f) Immunofluorescence analyses of cell–cell fusions in the absence (c) or presence (f) of exogenous ACE2 expression. GFP signals of the reconstituted DSP reporter proteins were imaged using confocal laser scanning microscopy. The spike variants were stained using the human anti-spike antibody 4C12 and cell nuclei were visualized using DAPI staining.

Figure 5

All tested SARS-CoV-2 spike proteins from variants of concern are dependent on furin cleavage for induction of ACE2-independent cell–cell fusions. (a–g) DSP assay with 293T-DSP-mix cells transfected with empty vector (mock), CoV-2 spike wildtype (WT), mutants thereof or variants of concern as indicated. (a–d) The spike-encoding plasmids were co-transfected together with wildtype furin (a,b, +furin) or its catalytically inactive derivative D153N, thus serving as a dominant negative mutant termed DN-furin (c,d, +DN-furin). (e,f) For pharmacological inhibition of endogenous furin, the transfection mix was replaced at four hours post-transfection with medium containing 10 µM of the furin inhibitor CMK (+CMK). Cell–cell fusions were quantified via GFP (a,c,e) or luciferase activity (b,d,f). (g) The data obtained in (f) were used to calculate the CMK-mediated inhibition of fusion in percent based on the RLU in relation to the respective variant not treated with the inhibitor.

Figure 6

Characterization of SARS-CoV-2 spike’s fusion activity in the absence of ACE2. (a) Schematic of SARS-CoV-2 spike wildtype (WT, Wuhan-Hu-1) and the VOCs Delta and Omicron, as well as chimeric proteins and point mutants thereof. Upper part: Acquired mutations in Delta and Omicron spike are indicated. Lower part: The magnification depicts the corresponding amino acid sequences in proximity to the respective protease cleavage sites. The number in the name of the mutant refers to the amino acid position in WT spike. (b–f) DSP assays of 293T-DSP-mix cells expressing the respective spike variant, chimera or mutant as indicated. Cell–cell fusions were quantified via GFP using the CTL-fluorospot reader (b,d) or bioluminescence upon Renilla-luciferase mediated cleavage of EnduRen (c,e,f). (f) Fusion activity of the individual spike protein in the presence (+ACE2) or absence of ACE2 (-ACE2) as measured by luciferase activity. (g) Data from (f) were used to calculate the n-fold change in fusion activity of the respective spike derivative following ACE2 co-expression relative to their fusion activity when expressed alone.

Figure 7

Blocking SARS-CoV-2 spike’s fusion activity with monoclonal antibodies. (a) Schematic representation of SARS-CoV-2 spike with its S1 and S2 subunits and the target sites of the spike-specific monoclonal antibodies TRES328 (NTD) and TRES567 (RBD). (b–e) DSP assays with 293T-DSP-mix cells transfected with spike in combination with ACE2 (b) or spike alone (c–e). At 4h post-transfection, 25 µg/mL (b,c,e) or serial dilutions (d) of the indicated neutralizing anti-spike antibodies directed against its NTD or RBD were added, and non-neutralizing (nnt) anti-spike antibodies served as internal controls (e). Cell–cell fusions were quantified through bioluminescence detection following addition of the substrate EnduRen. (d) The obtained data were used to calculate the antibody-mediated inhibition of fusion in percent based on the RLU in relation to the nontreated control.