D614G Substitution of SARS-CoV-2 Spike Protein Increases Syncytium Formation and Virus Titer via Enhanced Furin-Mediated Spike Cleavage

ABSTRACT Since the D614G substitution in the spike (S) protein of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) emerged, the variant strain has undergone a rapid expansion to become the most abundant strain worldwide. Therefore, this substitution may provide an advantage for viral spreading. To explore the mechanism, we analyzed 18 viral isolates containing S proteins with either G614 or D614 (S-G614 and S-D614, respectively). The plaque assay showed a significantly higher virus titer in S-G614 than in S-D614 isolates. We further found increased cleavage of the S protein at the furin substrate site, a key event that promotes syncytium formation, in S-G614 isolates. The enhancement of the D614G substitution in the cleavage of the S protein and in syncytium formation has been validated in cells expressing S protein. The effect on the syncytium was abolished by furin inhibitor treatment and mutation of the furin cleavage site, suggesting its dependence on cleavage by furin. Our study pointed to the impact of the D614G substitution on syncytium formation through enhanced furin-mediated S cleavage, which might increase the transmissibility and infectivity of SARS-CoV-2 strains containing S-G614.

of viral genetic variations usually provide valuable insights into the mechanisms linked to pathogenesis, antiviral drug resistance, and immune responses, which also impact the development of new vaccines, antiviral drugs, and diagnostic tests (1)(2)(3). Therefore, analysis of viral genomes and monitoring of the evolutionary trajectory of SARS-CoV-2 over time have been meticulously conducted to identify any specific genetic variations that contribute to the transmissibility and virulence of SARS-CoV-2 (4)(5)(6).
Among the genetic variations that have evolved during the course of the SARS-CoV-2 outbreak, the D614G substitution in the spike (S) protein, which corresponds to a change of the A nucleotide at genome position 23403 to a G, has been identified as the signature of the A2a clade of SARS-CoV-2, the most prevalent clade (7,8). This substitution emerged at low frequency in early March 2020 but rapidly expanded to become the most abundant clade worldwide by April and May 2020 (7), and it was thus proposed to provide a selective fitness advantage during the outbreak. B. Korber et al. recently reported that the strain containing the spike protein with G at position 614 (S-G614) is associated with higher viral loads in the upper respiratory tracts of infected individuals (though not increased disease severity) than the loads produced by viruses containing S-D614 (7). Several recent reports further demonstrated that pseudotyped viruses or the engineered viruses containing S-G614 exhibit significantly higher infectivity than those containing S-D614 (9)(10)(11). Sera from most convalescent CoV disease 2019 (COVID-19) patients may neutralize both S-D614-and S-G614-pseudotyped viruses with comparable efficiencies (7,12). As this mutation locates outside the receptor binding domain (RBD) of S, similar affinities for binding of S-D614 and S-G614 to the angiotensin-converting enzyme 2 (ACE2) receptor were documented (13)(14)(15). These findings thus raised the possibility that the D614G substitution confers increased infectivity and transmissibility neither through an increased binding to ACE2 nor through increased escape of immune surveillance. Elucidating the molecular basis for the higher infectivity of the D614G virus is urgent to understand its predominance and to design an effective treatment strategy for patients.
The host protease-mediated cleavage of envelope protein is involved in the maturation, pathogenesis, and infectivity of many viruses. S of SARS-CoV-2 also undergoes two proteolytic cleavages critical for membrane fusion and viral entry. One cleavage site is located at the S1 and S2 boundary, a polybasic motif recognized by furin and related proprotein convertases (PCs); the other is located at the S29 site within the S2 domain, which is recognized by the TMPRSS2 protease (16). In our recent study, we found that the cleavage of S by furin/PCs at the S1/S2 boundary is critical for S-mediated syncytium formation, another pathogenic event that contributes to increased viral transmission (17). As noted previously, the D614G substitution is located in the C-terminal region of the S1 domain of the S protein, close to the furin cleavage site (between amino acids [aa] 685 and 686) (Fig. 1A). Meanwhile, recent cryogenic electron microscopy (cryo-EM) studies suggested that the D614G substitution might induce conformational changes in the S protein (13,18,19). It thus raised the possibility that the D614G substitution contributes to the increased accessibility of the S protein for cleavage by furin through a conformational change and thus increases the membrane fusion activity of the virus; this may be the basis for the increased infectivity, transmission capability, and virus titer of S-G614-containing SARS-CoV-2. To test this hypothesis, we first compared the virus titers and rates of cleavage of the S protein of 18 clinical SARS-CoV-2 isolates. The effects of the D614G substitution on enhanced S cleavage and syncytium formation were further quantitatively validated in cells expressing S protein. This cell-based assay system was further used to examine the critical role of furin-mediated S cleavage for the effect of the D614G substitution. We expect that the findings will provide an explanation for the rapid expansion of S-G614containing SARS-CoV-2 and help develop a therapeutic strategy for intervening in its spread in the population.  isolates (NTU01 to NTU18) (see Tables S1 and S2 in the supplemental material) containing either S-D614 or S-G614 from infected Calu-3 and Vero E6 cells. Interestingly, we found that the S-G614-containing viruses (n = 10) had significantly higher virus titers than the S-D614-containing viruses (n = 8), as determined by a plaque assay ( S-G614-containing viral isolates showed increased cleavage of the S protein compared to that of S-D614-containing viral isolates. To examine if the D614G substitution might contribute to increased accessibility of the S protein for cleavage by furin, we compared the patterns of S protein cleavage at this site for viruses containing either the S-G614 or the S-D614 protein. Viruses in the supernatants of cells infected with either group of viruses were harvested for immunoblot analysis. Interestingly, the results showed significantly increased cleavage of the S protein into the S1 and S2 fragments in S-G614containing viruses compared to that in S-D614-containing viruses in Calu-3 cells (Fig. 1D); the cleaved S/total S percentage was shown to be 55.95% 6 9.86% (in S-G614 viruses) versus 27.71% 6 16.05% (in S-D614 viruses) (P , 0.001) (Fig. 1D, right). Moreover, we found a higher ratio of spike to nucleocapsid (N) in the S-G614-than in the S-D614-containing viruses (total S/N, 1.9 6 0.44 versus 1.0 6 0.31; P , 0.001; cleaved S/N, 3.7 6 0.39 versus 1.0 6 0.27; P , 0.001) (Fig. S2).
Consistently, a significant difference in the rates of cleavage of the S protein between the lysates from cells infected with the two groups of viruses was also demonstrated by immunoblotting (Fig. 1E), with the cleaved S/total S percentages being 64.96% 6 5.99% in S-G614 viruses versus 39.13% 6 11.65% in S-D614 viruses (P , 0.001) (Fig. 1E, right). A higher cleavage of S in S-G614-than in S-D614-containing viruses was also supported by the results from Vero E6 cells infected with the two groups of viruses ( Fig. S1C and D). Therefore, these results suggested a putative effect of the D614G substitution in enhancing the cleavage of the S protein.
The syncytial phenotype and S cleavage were increased in S-G614-expressing cells compared to those in S-D614-expressing cells. Genetic heterogeneity beyond the D614G substitution among different SARS-CoV-2 isolates might cause confusion regarding our observation. Therefore, we tested our hypothesis in cultured cells expressing only the S protein. Codon-optimized expression plasmids for wild-type S-D614 and the single-substitution mutant (S-G614) were individually transfected into Vero E6 cells, which were harvested at 24 h posttransfection for analysis. We first examined the syncytial phenotype via the observation of fused cells containing multiple nuclei visible by light microscopy. Compared with control cells transfected with only the vector, cells expressing the S-D614 protein exhibited a moderate syncytial phenotype, which was increased in cells expressing the mutant S-G614 protein ( Fig. 2A). Immunoblot analysis further revealed increased cleavage of the S-G614 proteins relative to the S-D614 protein and a higher cleaved S/total S ratio in the lysate of S-G614-expressing cells than in S-D614-expressing S-mediated syncytium formation was increased in S-G614-expressing cells compared to that in S-D614-expressing cells, dependent on furin-mediated S cleavage. To confirm the effect of the D614G substitution on S and ACE2 bindingmediated cell syncytium formation, we established a luciferase-based reporter assay to quantitatively compare levels of syncytium induction by the S-G614 and S-D614 proteins (schematically illustrated in Fig. 2C). As documented, the interaction between S and ACE2 at the membranes of infected cells and adjacent cells primes the cleavage of S by host proteases to release the fusion peptide essential for syncytium formation (20). Therefore, in our experimental design, we selected ACE2-null 293T cells transfected with an expression plasmid for either S-D614 or S-G614 as S 1 and ACE2effector cells. Two ACE2 1 target cell lines were selected: 293T cells stably transfected with human ACE2 (293T-hACE2) and Vero E6 cells (with endogenous ACE2 expression). Neither effector cells nor target cells alone developed syncytia due to the lack of the binding partner (ACE2 or the S protein). To quantitatively examine the fusion of effector and target cells, we developed a one-hybrid luciferase reporter assay. The pGAL4DBD-hAR-NTD plasmid was cotransfected with an individual S expression plasmid into the effector cells; the pGAL4/UAS-TK-Luc plasmid was transfected into the target cells. The effector and target cells were then cocultured and harvested for reporter activity analysis. The expression of pGAL4/UAS-TK-Luc in the target cells was activated only upon formation of syncytia consisting of effector and target cells (mediated by the interaction of S and ACE2), which is driven by the transcriptional activator encoded by the pGAL4DBD-hAR-NTD plasmid in the effector cells.
As expected, the expression of pGAL4DBD-hAR-NTD or S alone in the effector cells did not activate luciferase expression in the target cells (Fig. 2D, lanes 2 to 5), which was elevated only when pGAL4DBD-hAR-NTD and the S protein were coexpressed in the effector cells. The expression of S-G614 significantly increased the luciferase activity compared to that upon the expression of S-D614 in the effector cells (Fig. 2D, lane 6 versus lane 7). The results from the reporter assay thus provide quantitative evidence supporting increased syncytium induction by the mutant S-G614 protein compared with that induced by the S-D614 protein, consistent with the findings from light microscopy observation.
Since our previous study showed that furin inhibitors can block spike cleavage and spike-mediated syncytium formation both in SARS-CoV-2-infected cells and spike-expressing cells (17), we want to further examine whether the enhanced syncytium formation by S-G614 expression was mediated through cleavage by furin/PC protease(s). We treated Sexpressing effector cells with the furin/PC inhibitor decanoyl-RVKR-chloromethylketone (CMK). Immunoblot analysis demonstrated that cleavage of the S-D614 or S-G614 protein was completely blocked by CMK treatment (Fig. 2E, left). The syncytial phenotype in the S-D614-and S-G614-expressing cells was also significantly decreased by CMK treatment (Fig. 2E, right). The difference in luciferase activity induced by the S-D614 and S-G614 proteins was diminished by CMK treatment (Fig. 2D, lane 8 versus lane 9), suggesting that the  syncytium formation induced by the S-G614 protein is dependent on the presence of active furin/PCs in cells. Consistently with this finding, the difference in the effects of S-D614 and S-G614 on reporter activity was abolished when the furin substrate site was mutated by introduction of the R682A substitution (Fig. 2D, lane 10 versus lane 11); the effects of the substitution on the cleavage of the S protein and syncytium formation are shown in Fig. 2E. All together, these results suggest that the putative function of the D614G mutation in the S protein of SARS-CoV-2 is dependent on enhanced cleavage at the furin substrate motif at the S1/S2 boundary, which contributes to an increased membrane fusion activity.

DISCUSSION
Recently, several studies showed that the D614G variant SARS-CoV-2 replicates more efficiently in primary human proximal airway epithelial cells than the wild-type virus. They also provided in vivo evidence showing that the D614G variant exhibited significantly faster droplet transmission between hamsters than the wild-type virus at early stages after infection (9,11). Our study here provides a possible mechanistic explanation in support of the notion that the D614G substitution can increase virus transmission through enhanced membrane fusion mediated by increased cleavage of the S protein by furin/PC proteases.
In our previous study, we examined the effect of furin cleavage on viral replication by either preinfection treatment or postinfection treatment of the virus-infected cells with the CMK furin inhibitor (17). The results suggested that the furin-mediated cleavage of spike might not contribute to viral RNA synthesis and virion release but had a critical role in membrane fusion for the viral entry step. The current study further revealed that the D614G substitution in S might contribute to increase the cleavage by furin/PCs and thus increase the syncytium formation of S-G614-containing SARS-CoV-2 strains. Therefore, the effect of enhanced furin cleavage of the spike protein caused by the D614G substitution might increase viral entry into the cells and contribute to higher virus titers after multiple infection cycles.
Regarding the mechanism, as revealed by cryo-EM, Wrobel et al. demonstrated that proteolytic cleavage of spike at the S1/S2 boundary could turn the spike trimer from a "closed" form into an "open" form, a receptor-binding-competent structure with high affinity for the ACE2 receptor (21). Park et al. also showed that cleavage of S by furin might prime for subsequent S cleavage by TMPRSS2 and help release the fusion peptide for subsequent membrane fusion activity (22). These studies thus provide possible mechanisms for increased furin-mediated S cleavage in S-G614 viruses to enhance viral entry, which meanwhile may also contribute to higher syncytium formation of cells infected with S-G614 viruses. Both events may contribute to the increase in virus titer of the S-G614 viruses. In fact, the contribution of increased syncytia in the spreading of viruses has been documented for several viruses (22)(23)(24)(25)(26), possibly through direct cellto-cell spreading and escape from the innate immune surveillance (27,28). Based on these observations, the increased syncytium formation induced by the D614G substitution might increase the spreading of SARS-CoV-2. This possibility is worthy of being tested in vivo in proper animal models.
Regarding how the substitution may affect furin cleavage, through cryo-EM analysis, several studies well demonstrated the effects of the D614G substitution on the structure of spike protein. As noted, the RBDs in the D614 spike trimer are presented mainly as the closed form (13,18,19). One structure change caused by the D614G substitution is the loss of the salt bridge between D614 and K854, which reduced the stability of the spike trimer in the closed form. Therefore, the RBDs in the G614 spike trimer are preferentially presented in the open form, which is more competent for ACE2 binding. Meanwhile, disruption of the salt bridge between D614 and K854 might unfold a domain (aa 827 to 855) in the S2 subunit; this unfolding promotes fusion-peptide-containing domain (aa 815 to 825) to undergo membrane fusion (19,21). These structural changes caused by the D614G substitution might contribute to a higher affinity of the spike protein for the furin/PCs, which has been demonstrated by the computational prediction of furin binding to the RRAR motif of S-G614 versus S-D614 and is preliminarily supported by the results of an in vitro proteolytic assay (18,29).
Though the mechanism for the increased cleavage by furin/PCs by the D614G substitution was delineated in assays of cells expressing only the S protein, it has been well supported by virus-infected cells, as most S-G614-containing viruses show more evidence of the cleavage of S in association with higher virus titers. Despite this, the effect of D614G might be confounded by other genetic variations among different virus isolates. For example, the D614G mutation was noted to be associated with mutations in viral nsp3 and an ORF1b protein variant (P314L) (5,30). The coexistence of D614G and P314L may suggest additional mechanisms for the selection advantage of viruses expressing S-G614 in terms of viral infectivity and warrant further investigation.
In addition, our results showed quite interesting differences between the amounts of cleavage of S present in the supernatant (55.95% and 27.71%) and those in the lysates (64.96% and 39.13%). During biogenesis, the spike monomer travels from the endoplasmic reticulum (ER) to the ER-Golgi intermediate compartment (ERGIC), where it forms a trimer and incorporates into viral particles, budding into the lumens of vesicles, which are then trafficking for release through exocytosis (31). The cleavage of spike by active furin occurs primarily in the Golgi compartment and mainly in the trans-Golgi network (TGN) (32). The spike protein not packaged in the viruses remained in the membrane, which is more accessible for cleavage by the active furin in the TGN. Therefore, the cleavage form of spike, as reflected by the cleaved S/total S ratio, might be more predominant in the S from cell lysates than in that from the viruses in the supernatant. Based on the results, the difference in spike cleavage might be truly reflected by the analysis of the purified viruses. However, due to the safety concern, we did not conduct the virus purification experiment in the current study.
Moreover, we found a higher S/N ratio in S-G614 viruses than in S-D614 viruses (Fig. S2). Consistently, Zhang et al. reported higher spike density in S-G614 viruses (15). Through alignment of the closed conformation of the S-G614 trimer and S-D614 trimer by cryo-EM, Zhang et al. found that residues 620 to 640 at C-terminal domain 2 (CTD2) formed an ordered "630 loop" in the S-G614 trimer but a disordered one in the S-D614 trimer. The ordered loop might stabilize the cleaved S1/S2 trimer and prevent S1 shedding from S-G614 viruses (14), which provides a possible mechanism for the higher spike density in S-G614 viruses (15). Since the S1 subunit is responsible for the binding of S with ACE2, a decrease of S1 shedding from the trimer may increase the functional S protein for both viral infectivity and spike-mediated syncytium formation.
Currently, S-G614-containing SARS-CoV-2 has expanded and is the dominant strain worldwide. As the enhancement of cleavage by furin/PCs is critical for its transmission, blockade of this mechanism by targeting furin/PC protease activity for inhibition might become a potential antiviral strategy to block transmission of this strain. As shown in our recent study, two furin/PC inhibitors, CMK and naphthofluorescein, can abolish S cleavage, virus production, and pathogenic syncytium formation of an S-G614-containing SARS-CoV-2 isolate (NTU03) (17). Consistently, both CMK treatment and mutagenesis of the furin cleavage site can decrease syncytium formation in S-D614-and S-G614-expressing cells. The finding that CMK treatment can further decrease the luciferase reporter activity of the R682A mutant, suggesting that CMK might have some other effect on reporter activity, independently of the cleavage by the furin/PCs, needs further investigation. According to our studies, such furin inhibitors thus become leads for further antiviral development for the prevention and treatment of S-G614-containing SARS-CoV-2 infections.

MATERIALS AND METHODS
Viruses. Sputum specimens from SARS-CoV-2-infected patients were kept in viral transport medium. Virus (multiplicity of infection [MOI] = 0.02) isolated from the specimens was propagated in Vero E6 cells in Dulbecco's modified Eagle's medium (DMEM) supplemented with 2 mg/ml tosylsulfonyl phenylalanyl chloromethyl ketone-trypsin, which may enhance infectivity and viral isolation efficacy, as documented previously (33-36). The 18 virus isolates used in the current study were hCoV-19/Taiwan/NTU01/2020 to hCoV-19/Taiwan/NTU18/2020. Plaque assay. The plaque assay was performed as previously described, with minor modifications (37). In brief, the Vero E6 cells (2 Â 10 5 cells/well) were seeded in 24-well tissue culture plates and maintained in DMEM supplemented with 10% fetal bovine serum (FBS) and antibiotics. After 24 h of incubation, SARS-CoV-2 was treated to the cell monolayer for 1 h at 37°C. After removal of the virus, the cell monolayer was washed once with phosphate-buffered saline (PBS), and the cells were maintained with medium containing 1% methylcellulose and incubated for 5 to 7 days. Subsequently, cells were fixed with 10% formaldehyde overnight and stained with 0.5% crystal violet in order to count the plaques. Virus titers are means from three independent experiments.
Plasmid construction. Humanized pUC57-2019-nCoV-S was a kind gift from Che Ma at the Institute of Genomics Research Center, Academia Sinica, Taiwan. The spike sequence was cloned into the pcDNA3.0-HA vector with the addition of a hemagglutinin (HA) tag at the spike protein C terminus via NheI and XbaI sites. A QuikChange II site-directed mutagenesis kit (Agilent) was used to generate the mutant D614G spike or R682A spike construct. The primer set for D614G spike was S-D614G-F (59-CTCGGTACAATTCACGCCCTGATACAGCACGGC-39) and S-D614G-R (59-GCCGTGCTGTATCAGGGCGTGAAT TGTACCGAG-39). The primer set for R682A spike was 59-ACGCTCCGGGCTCTTGCGGGAGAGTTTGTCTG-39 and 59-CAGACAAACTCTCCCGCAAGAGCCCGGAGCGT-39.
Cell culture experiments. Cell-cell fusion assay. A quantitative GAL4-based mammalian one-hybrid assay was established to assess cell-cell fusion activity. This reporter assay contains two plasmids (kindly provided by Hsiu-Ming Shih at the Institute of Biomedical Sciences, Academia Sinica, Taiwan). One is the reporter construct pGAL4/UAS-TK-Luc, which encodes firefly luciferase under the control of the GAL4 response element (upstream activation sequence [UAS]) and a thymidine kinase (TK) promoter. The other is the transcriptional activator construct pGAL4DBD-hAR-NTD, which consists of aa 1 to 560 of the AR transcriptional activation domain with the GAL4 DNA-binding domain (DBD) fused at its N terminus. The protein encoded by pGAL4DBD-hAR-NTD can bind to the UAS GAL4 response element of the pGAL4/UAS-TK-Luc reporter and activate the transcription of the luciferase reporter gene by the transcriptional activator of AR-NTD.
RNA extraction and Northern blot analysis. RNA was extracted with a NucleoSpin RNA kit (Macherey-Nagel) according to the manufacturer's instructions. Northern blotting was performed as previously described (39). In brief, 0.2 mg of RNA was denatured and separated by an 0.8% agarose-formaldehyde gel at 70 V for 5 h. The agarose gel was soaked in 50 mM NaOH for 50 min to break the large RNA fragment. Subsequently, the gel was washed with 100 mM Tris-HCl (pH 7.5) for 30 min and incubated in 20Â SSC buffer (1Â SSC is 0.15 M NaCl plus 0.015 M sodium citrate) for 20 min and then capillary transferred to a positively charged Hybond-N nylon membrane (Amersham Biosciences) overnight. RNA was immobilized by UV crosslinking (1,800 Â 100 mJ/cm) and hybridized at 50°C overnight with digoxigenin (DIG)-labeled probes generated with a PCR DIG probe synthesis kit (Roche Diagnostics). The primers used to synthesize the DIG-labeled nCoV19-N cDNA probe were 59-AAGCTGGACTTCCCTATGGTGC-39 and 59-CCTTGGGTTTGTTCTGGACCACG-39. The probes of porphobilinogen deaminase (PBGD) used as the internal control in Northern blotting were 59-GGTGACCAGCACACTTTGGG-39 and 59-AGCCGGGTGTTGAGGTTTCC-39.

D614G Mutation Enhances Spike Cleavage and Syncytium
® qRT-PCR. The qRT-PCR was conducted by following the protocol described previously (17). In brief, RNA extracted from SARS-CoV-2-infected Vero E6 cells was reverse transcribed using a SuperScript III reverse transcriptase system (Thermo Fisher Scientific). Quantitative PCR targeting the E gene was performed using FastStart DNA SYBR green on a LightCycler 1.5 (Roche Diagnostics), with the primer set 59-ACAGGTACGTTAATAGTTAATAGCGT-39 and 59-ATATTGCAGCAGTACGCACACA-39. The RNA level of the E gene in the cells was determined relative to the internal control, the cellular PBGD gene, with the primer set 59-GCATCGCTGAAAGGGCCTTCC-39 and 59-TCATCCTCAGGGCCATCTTCATGC-39.
Statistical analysis. The virus titers quantified by plaque assays in triplicate are shown as means 6 standard deviations (SD). Results from the reporter assay are shown as data representative of three independent experiments and are presented as means 6 SD. Differences in data from the virus titer, qRT-PCR, and reporter assays between each indicated paired sample group were evaluated by Student's t test. A P value of 0.05 or lower was used to indicate statistical significance (*, P , 0.05; **, P , 0.01; ***, P , 0.001).
Data availability. The sequencing data for the 18 virus isolates used in the current study (hCoV-19/ Taiwan/NTU01/2020 to hCoV-19/Taiwan/NTU18/2020) have been deposited in GISAID (https://www.gisaid.org/ CoV2020/). The accession codes are listed in Table S1 in the supplemental material, and the mutations in these 18 viral isolates are summarized in Table S2.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.   We declare no competing interests.