Distinctive Roles of Furin and TMPRSS2 in SARS-CoV-2 Infectivity

Abstract

The spike protein (S) of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) directs infection of the lungs and other tissues following its binding to the angiotensin-converting enzyme 2 (ACE2) receptor. For effective infection, the S protein is cleaved at two sites: S1/S2 and S2'. The "priming" of the surface S protein at S1/S2 (PRRAR₆₈₅↓) [the underlined basic amino acids refer to critical residues needed for the furin recognition] by furin has been shown to be important for SARS-CoV-2 infectivity in cells and small-animal models. In this study, for the first time we unambiguously identified by proteomics the fusion activation site S2' as KPSKR₈₁₅↓ (the underlined basic amino acids refer to critical residues needed for the furin recognition) and demonstrated that this cleavage was strongly enhanced by ACE2 engagement with the S protein. Novel pharmacological furin inhibitors (BOS inhibitors) effectively blocked endogenous S protein processing at both sites in HeLa cells, and SARS-CoV-2 infection of lung-derived Calu-3 cells was completely prevented by combined inhibitors of furin (BOS) and type II transmembrane serine protease 2 (TMPRSS2) (camostat). Quantitative analyses of cell-to-cell fusion and S protein processing revealed that ACE2 shedding by TMPRSS2 was required for TMPRSS2-mediated enhancement of fusion in the absence of S1/S2 priming. We further demonstrated that the collectrin dimerization domain of ACE2 was essential for the effect of TMPRSS2 on cell-to-cell fusion. Overall, our results indicate that furin and TMPRSS2 act synergistically in viral entry and infectivity, supporting the combination of furin and TMPRSS2 inhibitors as potent antivirals against SARS-CoV-2. IMPORTANCE SARS-CoV-2, the etiological agent of COVID-19, has so far resulted in >6.1 million deaths worldwide. The spike protein (S) of the virus directs infection of the lungs and other tissues by binding the angiotensin-converting enzyme 2 (ACE2) receptor. For effective infection, the S protein is cleaved at two sites: S1/S2 and S2'. Cleavage at S1/S2 induces a conformational change favoring the S protein recognition by ACE2. The S2' cleavage is critical for triggering membrane fusion and virus entry into host cells. Our study highlights the complex dynamics of interaction between the S protein, ACE2, and the host proteases furin and TMPRSS2 during SARS-CoV-2 entry and suggests that the combination of a nontoxic furin inhibitor with a TMPRSS2 inhibitor significantly reduces viral entry in lung cells, as evidenced by an average synergistic ∼95% reduction of viral infection. This represents a powerful novel antiviral approach to reduce viral spread in individuals infected by SARS-CoV-2 or future related coronaviruses.

Keywords: Calu-3 cells; HeLa cells; SARS-CoV-2; TMPRSS2; cell-to-cell fusion; furin; spike; viral entry; viral infection.

FIG 1

Processing of S peptides by furin and TMPRSS2. (A) Schematic representation of the primary structure of preproS, including its domains, the predicted furin-like S1/S2 site generating the S1 and S2 subunits, and the S2′ site preceding the fusion peptide (FP). The signal peptide (SP), N-terminal domain (NTD), receptor binding domain (RBD) to ACE2, the two heptad repeats HR1 and HR2, the transmembrane domain (TM), the cytosolic tail (CT), and the C-terminal V5 tag are indicated. (B) In vitro furin activity against peptides mimicking the S1/S2 (and its mutants) and S2′ cleavage site sequences of the spike protein from SARS-CoV-2 and SARS-CoV-1, as described in Table 1. Each substrate was tested at final protease concentrations of 2 and 100 nM. (C) In vitro TMPRSS2 activity (at 50 nM) against peptides mimicking the indicated S1/S2 and S2′ cleavage site sequences as described in Table 1.

FIG 2

Processing of spike glycoprotein in HeLa cells and critical role of ACE2. (A) Western blot analyses of the processing of WT proS into V5-tagged S2 and S2′ by the proprotein convertases furin, PC5A, PACE4, and PC7 following cotransfection of their cDNAs in HeLa cells. The migration positions of immature proS_im, S2, and S2′ as well as the actin loading control are emphasized. V, empty pIRES-EGFP-V5 vector. The estimated percent cleavages into S1/S2 and S2′ are shown and were calculated as the ratio of the V5 immunoreactivity of the cleaved form to the sum of all forms. As estimated by qPCR, the percent mRNA levels of overexpressed PCs relative to furin are also shown. The data are representative of those from at least three independent experiments. (B) Western blot analyses of HeLa cells following cotransfection with cDNAs coding for either WT S protein (WT) or its double Ala mutant (R685A + R682A) (μS1/S2) in the absence or presence of furin cDNA at a ratio of S to protease of 1:2. *, inconsistently observed oligomeric forms of proS. (C) Western blot analyses of HeLa cells cotransfected with V5-tagged spike protein, WT S or its furin-optimized S2′ (KRRKR₈₁₅↓SF) mutant (μS2′), and empty vector (V) or furin. (D) Identification of S2′ cleavage site by tandem mass spectrometry (MS/MS). WT spike glycoprotein was immunoprecipitated from HeLa cells using V5-agarose beads, then resolved by SDS-PAGE, and subjected to silver staining (left side); the positions of the slices are indicated (1 to 10). The MS/MS analysis of peptides generated by a Lys-specific protease (K814↓) is indicated; the data represent the ratio of SFIEDLLFNK₈₂₅ to R₈₁₅SFIEDLLFNK₈₂₅ (right side). (E and F) Western blot showing the impact of ACE2 on the processing of WT and μS1/S2 spike glycoproteins by furin. HeLa cells expressing empty vector, WT proS, or its μS1/S2 mutant with or without furin, ACE2, or both were analyzed by Western blotting using anti-V5 antibody. The ratio of cDNAs used was 1:1:1 for S, ACE2, and furin. The data are representative of those from at least three independent experiments.

FIG 3

Importance of furin in the processing of the spike glycoprotein. (A) HeLa cells were first transfected with control nontargeting siRNA (siCTL) or siRNA furin (siFur) at a final concentration of 20 nM, or mock transfected (N), and 24 h later, transfected with empty vector or with that coding for a V5-tagged spike glycoprotein for an additional 48 h. Following lysis, proteins were resolved by SDS-PAGE followed by WB with anti-V5 or anti-furin antibodies. (B) HeLa cells transfected with empty vector, V5-tagged WT spike protein, or its S1/S2 site mutant (μS1/S2) were treated with endoglycosidase F (endo-F) and endoglycosidase H (endo-H) or mock treated (NT) and analyzed as described for panel A. (C and D) HeLa cells transfected with V5-tagged wild-type spike protein or S1/S2 single mutants (C) or S2′ single or double mutants (D) in the absence (V) or presence of overexpressed furin were lysed and analyzed by WB.

FIG 4

Inhibition of PCs by BOS compounds. (A) Chemical motif of BOS inhibitors and representative structure of BOS-318. (B) In vitro BOS inhibition of the cleavage of the fluorogenic dibasic substrate FAM-QRVRRAVGIDK-TAMRA by each of the proprotein convertases furin, PC5 (PCSK5), PACE4 (PCSK6), and PC7 (PCSK7). All experiments were performed in 10 different wells, and the average pIC₅₀ (nanomolar) was calculated. Shown for comparison is the inhibitory pIC₅₀ negative log of the IC₅₀ value when converted to molar) of the furin-like inhibitor RVKR-cmk; determination was performed >100 times. (C) In vitro inhibition of furin by the BOS compounds. Furin (2 nM) was incubated with increasing concentration of BOS inhibitors, and its enzymatic activity against the synthetic peptides DABSYL/Glu-TNSPRRAR↓SVAS-EDANS (5 μM) was measured at pH 7.5 (n = 3). (D) Golgi assay; the table presents the effects of BOS inhibitors on U2OS cells expressing each of furin, PC5A, PACE4, and PC7 simultaneously transduced with a BacMam-delivered construct containing a Golgi-targeting sequence followed by a 12-amino-acid furin/PCSK cleavage site from bone morphogenic protein 10 (BMP10) and GFP at the C terminus (GalNAc-T2-GGGGS-DSTARIRR↓NAKG-GGGGS-GFP). Dibasic cleavage releases NAKG-GGGGS-GFP, thereby reducing the Golgi-associated fluorescence estimated by imaging. (E) Furin inhibitors (BOS) abrogate endogenous processing of the spike glycoprotein. HeLa-ACE2 cells were transiently transfected with a cDNA encoding an empty vector or with one expressing the V5-tagged spike glycoprotein (spike-V5). At 5 h pretransfection, cells were treated with the vehicle DMSO (NT, duplicate), with the furin inhibitors at the indicated concentrations, or with RVKR-cmk at 50 μM. At 24 h posttransfection, media were replaced with fresh ones lacking (NT) or containing the inhibitors for an additional 24 h. Cell extracts were analyzed by Western blotting using MAb V5. All data are representative of those from at least three independent experiments.

FIG 5

Furin-like inhibitors and camostat treatment decrease SARS-CoV-2 infection in Calu-3 cells. (A) Replication kinetics was studied at 12, 24, and 48 h postinfection by plaque assay to determine PFU of SARS-CoV-2 in the supernatant of infected Calu-3 cells treated or not with 1 μM BOS-318, BOS-857, and BOS-981. A line graph presents results of the triplicate plaque assay results (mean ± SD). (B) Virus titers in the supernatant (24 h postinfection) of infected Calu-3 cells treated with the indicated concentrations of BOS-318 were determined by plaque assay (mean ± SD of triplicates) (left). The selectivity index (SI) of BOS-318 in Calu-3 cells as shown in top right panel was determined by 50% cytotoxic concentration (CC₅₀)/half-maximal inhibitory concentration (IC₅₀). The left y axis indicates the inhibition of virus titer relative to that of the untreated control group (red). The right y axis indicates cell viability relative to that of the untreated control group (green). The CC₅₀, IC₅₀, and SI values for each inhibitor are as shown. Representative analyzed plaque images of infected Calu-3 cells treated with indicated doses of BOS inhibitors are shown at the bottom right. (C) Immunoblots for the infected Calu-3 cells (right) and viral particles secreted in the supernatant (left) with and without treatment with BOS inhibitors indicate reduced viral protein levels. Immunoblots were probed for the full-length (proS_m) and cleaved (S2) fragments of viral S protein and nucleocapsid (N) protein as indicated; β-actin was included as the loading control for the cells. (D) Virus titers in the supernatant (24 h postinfection) of infected Calu-3 cells treated with BOS-318 and/or camostat (Camo) were determined by plaque assay (mean ± SD of duplicates) (top). Representative analyzed plaque wells of infected Calu-3 cells are shown. Imaged plaque plates were processed, and plaques were counted using algorithm-based Matlab software as described in Materials and Methods.

FIG 6

Furin-like inhibitors strongly reduce SARS-CoV-2 infection in Calu-3 cells. Calu-3 cells were treated with indicated concentrations of BOS-857 (A) and BOS-981 (B) and infected with SARS-CoV-2 for 24 h. Virus titers in the supernatant were determined by plaque assay on Vero E6 cells (mean ± SD of triplicates). The selectivity indices of BOS-857 and BOS-981 in Calu-3 cells as shown at the top right were determined by CC₅₀/IC₅₀. The left y axis indicates the inhibition of virus titer relative to that of the untreated control group (red). The right y axis indicates the cell viability relative to that of the untreated control group (green). Representative analyzed plaque wells of infected Calu-3 cells are shown. Imaged plaque plates were processed, and plaques were counted using algorithm-based Matlab software as described in Materials and Methods.

FIG 7

Furin-like inhibitors modestly reduce virus production in SARS-CoV-2-infected Vero E6 cells in a concentration-dependent manner. (A) Vero E6 cells treated or not with 1 μM BOS-318, BOS-857, or BOS-981 were infected with SARS-CoV-2 for up to 45 h. Virus titers in the supernatant obtained at 12, 24, and 48 h postinfection were determined by plaque assay on Vero E6 cells. A line graph presents results of the triplicate plaque assay (mean ± SD). (B to D) Virus released in the supernatant (48 h postinfection) of infected Vero E6 cells treated with indicated concentrations of BOS-318 (B), BOS-857 (C), or BOS-981 (D) were determined by plaque assay (mean ± SD of triplicates).

FIG 8

Processing of SARS-CoV-2 S by furin-like convertases and TMPRSS2 is critical for viral entry in human lung epithelial cells but not in model HEK293 cells stably expressing ACE2. (A) Furin cleavage of proS at the S1/S2 site is required for SARS-CoV-2 pseudoviral entry in Calu-3 cells (left) but not HEK293T-ACE2 cells (right). Cells were inoculated with nanoluciferase-expressing HIV particles pseudotyped with SARS-CoV-2: wild-type spike, double Ala mutant spike (μS1/S2), or furin-optimized spike (mS2′). Inhibition of proS processing at S1/S2 by a novel furin-like inhibitor (BOS-318) during pseudovirion packaging prevents viral entry in Calu-3 cells but not in HEK293T-ACE2 cells. (B) Western blot analyses show inhibition by BOS-318 of proS processing at the S1/S2 site. Purified pseudovirions and cellular extracts of producing HEK293T17 cells treated or not with BOS-318 inhibitor were separated on an SDS-PAGE gel and analyzed for HIV-1 p24 and V5-tagged S protein (proS_m or cleaved, S2) as indicated. (C) Pretreatment of Calu-3 cells with 1 mM BOS-318 (B), 100 mM Camostat (C), or both (B and C) markedly reduces viral entry. In panels A and C, Calu-3 cells were transduced with nanoluciferase-expressing HIV particles pseudotyped with SARS-CoV-2 WT, μS1/S2, or mS2′ S for 72 h and analyzed for nanoluciferase expression. Viral entry was expressed as fold increase over that given by bald particles (pseudovirions made in the absence of S). Each dot represents a different experiment with median luciferase activity calculated from three biological replicates. Two to four experiments were performed for each cell type. Error bars indicate SD.

FIG 9

Spike-induced cell-to-cell fusion relies on Furin cleavage at S1/S2. (A) Cell-to-cell fusion between donor cells (HeLa) expressing the fusogenic SARS-CoV-2 spike protein (S) along with the HIV transactivator Tat and acceptor cells (TZM-bl) that express ACE2. Upon fusion, Tat is transferred from donor to acceptor cells, thereby inducing luciferase expression. (B) HeLa cells transfected with an empty vector (a) or expressing SARS-CoV-2 spike (b, seen at higher magnification in panel c), or μS1/S2 (d) were cocultured with TZM-bl cells for 18 h and the number of syncytia (white arrowheads) was examined using CellMask to probe for the plasma membrane and DAPI to stain the nuclei. Cell-to-cell fusion (white arrowheads) was evaluated using confocal microscopy. Scale bars = 60 μm for panels a, b, and d and 30 μm for panel c. (C) Donor cells were transfected with vectors expressing either no protein (empty vector), μS1/S2, or WT spike in the absence (NT) or presence (DMSO) of a vehicle or with the furin inhibitors BOS-318, BOS-981, BOS-857 (300 nM), and RVKR (10 μM). Acceptor cells were transfected with a vector expressing ACE2. After 48 h, donor and acceptor cells were cocultured for 18 h. Relative luminescence units (RLU) were normalized to the V value arbitrarily set to 1. Data are presented as mean values ± SD (n = 3); one-way analysis of variance (ANOVA) and Dunn-Sidàk multiple-comparison test were used. (D) Donor HeLa cells expressing WT S or the indicated S mutants and variants were cocultured with acceptor TZM-bl cells expressing ACE2. The extent of fusion is represented as a ratio between the RLU measured for each condition and that of donor cells expressing empty vector. The bar graph represents the average of 3 experiments performed in triplicates. Data are presented as mean values ± SEM (n = 3). One-way ANOVA and Bonferroni multiple-comparison test and two-way ANOVA and Dunn-Sidàk multiple-comparison test were used.

FIG 10

Exogenous TMPRSS2-generated shedding of ACE2 differentially regulates S-induced fusion at the plasma membrane of WT S versus μAS1/S2. Donor HeLa cells expressing doubly tagged (N-terminal HA tag and C-terminal V5 tag) WT spike glycoprotein or its S1/S2 mutant (μAS1/S2) were cocultured with acceptor TZM-bl cells expressing (empty vector), ACE2, or ACE2 plus TMPRSS2 (TMP) and treated with DMSO (vehicle control) or camostat (120 μM). Within the same experiment, cell-to-cell fusion (A) was assessed in parallel with spike processing in cells and media by Western blotting (B and C). (A) The extent of fusion is represented as a ratio between the RLU measured for each condition and that of donor cells expressing V. The bar graph represents the average of 2 experiments performed in triplicates. (B and C). Western blot analyses of media and cell extracts of the cocultured cells with donor cells overexpressing doubly tagged (N-terminal HA tag and C-terminal V5 tag) spike glycoprotein, WT (B) or μAS1/S2 (C). Media were subjected to immunoprecipitation with anti-HA agarose for the secreted forms of spike protein (S1 and S1_L), followed by Western blotting with anti HA-HRP. In the cell extracts, spike glycoproteins and ACE2 were immunoblotted with anti-V5 MAb and a polyclonal ACE2 antibody, respectively. The data are representative of those from three independent experiments. The Western blot analyses of TMPRSS2 (using a TMPRSS2 antibody) in the cells are shown separately, emphasizing the migration positions of the zymogen proTMPRSS2 and its autocatalytically generated products: the N-terminal TMPRSS2_Nt and C-terminal catalytic TMPRSS2_Cat.

FIG 11

The C-terminal collectrin-like domain of ACE2 may be critical for the regulation of cell-to-cell fusion of spike glycoprotein when exogenous TMPRSS2 is present: effect on secretion of S1_L. (A) Schematic representation of the primary structure of human ACE2 with emphasis on the C-terminal collectrin-like domain (aa 616 to 768 [light gray]), TMPRSS2 cleavage region (aa 697 to 716 [black]), and the polybasic amino acid segments in which K/R were mutated to A (C0 and C4) (amino acids underlined and in bold). Also shown are the peptidase domain (aa 19 to 615 [white]) containing the regions involved in the interaction with the spike SARS-CoV protein (hatched) and transmembrane domain (TM). (B) HeLa cells were cotransfected with ACE2 (WT ACE2 or the ACE2 C0+C4 mutant) and TMPRSS2 (WT or the S441A active-site mutant [μTMPRSS2]) or empty vector. Media and cell extracts were analyzed by Western blotting for shed ACE2 (sACE2) and ACE2, respectively. The migration positions of the ∼95-kDa and ∼80-kDa sACE2 in the media, as well as the intracellular disulfide bridged catalytic (cat) and N-terminal (Nt) fragments of TMPRSS2, are emphasized. (C) Donor HeLa cells expressing WT S-HA or μAS1/S2-HA were cocultured with acceptor TZM-bl cells expressing ACE2, WT ACE2, or the ACE2 C0+C4 mutant in presence or absence of TMPRSS2. From the same experiment, cell-to-cell fusion was assessed (C), in parallel with WB analyses of cells and media (D). The extent of fusion is represented as a ratio between the RLU measured for each condition and that of donor cells expressing an empty vector. The bar graph presents the average of 3 experiments performed in triplicates. Data are presented as mean values ± SD (n = 3). One-way ANOVA and Tukey’s multiple-comparison test were performed. (D) Coculture media were subjected to immunoprecipitation with anti-HA agarose for the secreted forms of spike protein (S1) followed by Western blotting with anti-HA-HRP. Spike glycoproteins in the cell extracts were immunoblotted with anti-V5 MAb. The Western blot data are representative of those from three independent experiments. (E) HeLa cells were transiently coexpressed with doubly tagged spike protein (N-terminal HA tag and C-terminal V5 tag), WT S or its mutants, μS1/S2 or μAS1/S2, and ACE2 alone or in combination with WT TMPRSS2 (TMPRSS2) or its S441A active-site mutant (μTMPRSS2), at an S/ACE2/TMPRSS2 ratio of 1:0.5:0.5. The immunoblot of the 24-h-conditioned media was first probed for secreted S1 and S1_L (HA-HRP antibody), stripped, and next probed for sACE2.

FIG 12

Proposed model for the processing of S protein and its blockade by Furin and TMPRSS2 inhibitors. (Boxed panel) Schematic representation of the S glycoprotein domains of SARS-CoV-2, including the N-terminal (NTD) and C-terminal (CTD) domains of S1, the furin-S1/S2, and the furin/TMPRSS2-S2′ processing sites, as well as the fusogenic α-helix that follows S2. Binding of the receptor binding domain of S1 to the membrane-associated ACE2 in target cells and the cell surface expression of TMPRSS2 and furin are also schematized. (Right portions) (1) BOS inhibitors (or mS1/S2 mutant) completely prevent fusion of donor HeLa cells expressing S glycoprotein with acceptor HeLa-ACE2 cells, which lack endogenous TMPRSS2. In this context, furin is a major processing enzyme cleaving at S1/S2 and generating S2′. (2) In acceptor HeLa cells expressing TMPRSS2 (+), maximal prevention of cell-to-cell fusion can be achieved by a combination of furin (BOS, phenocopying the mS1/S2 or mAS1/S2 mutants) and TMPRSS2 (camostat) inhibitors, which blocks S2′ production, ACE2 shedding (sACE2), and separation of sACE2-S1L complex from S2. (3) Optimal blockade of SARS-CoV-2 infection of Calu-3 cells, which express endogenously both furin and TMPRSS2, is also achieved by a combination of furin (BOS) and TMPRSS2 (camostat) inhibitors.