2.1 Spike protein

Structural analyses of S from SARS-CoV-2 as well as other members of coronavirus family showed that S forms homotrimer on the viral surface and protrudes from the viral membrane to form a distinctive crown-like appearance.^{3, 11, 12} These structures also revealed that the RBD in the S1 subunit is located at the outermost part of the trimer (Figure 1A), allowing S to easily access the receptor.

To bind to a host cell receptor, the RBDs of S1 undergo hinge-like conformational movements that hide or expose the determinants of receptor binding, which are referred to as two distinct conformations, respectively: the “down” represents a receptor-binding inactive state, and the “up” represents a receptor-binding active state.^{3, 10} Interactions between S and the receptor destabilize the prefusion trimer, causing shedding of the S1 subunit and conformational changes in the S2 subunit, resulting in a stable postfusion conformation.^{3, 10}

Unlike SARS-CoV, the S of SARS-CoV-2 has been shown to have a polybasic cleavage site (PRRAR) at the S1/S2 site that can be cleaved by furin during biosynthesis^12-15 (Figure 1B). This cleavage site is required for efficient proteolytic cleavage of S, and well-regulated cleavage at this site is essential for cell-cell fusion and viral entry into human lung cells.¹⁴ The SARS-CoV-2 Alpha variant (B.1.1.7) and Delta variant (B.1.617.2), which are both highly transmissible, exhibit mutations in position 681 that is adjacent to the furin cleavage site (HRRAR in Alpha, RRRAR in Delta). These mutations are predicted to alter furin cleavage and viral infection.

In addition, S is extensively decorated with host-derived glycans. these glycosylations create sugary barriers called glycan shielding that is likely to be important for immune evasion. On the other hand, some glycans could be recognized as epitopes and elicit neutralizing antibodies.¹⁶ Detailed information on S glycosylation and its role in virus infection has been discussed in a review.¹⁷

2.2 ACE2 as SARS-CoV-2 receptors

Several molecules have been reported to be involved in the entry process of SARS-CoV-2 and act as either attachment factors or entry receptors, among which ACE2 is believed to be the most important functional receptor. ACE2 is a type I membrane protein with a single zinc-binding catalytic motif, whose N-terminal domain shares significant sequence homology with angiotensin-converting enzyme (ACE). ACE2 is an important regulator in the renin-angiotensin system, which is cardinal in renal and cardiovascular physiology and pathophysiology.¹⁸

ACE2 also acts as a functional receptor for three coronaviruses: human CoV-NL63, SARS-CoV, and SARS-CoV-2. Among them, human CoV-NL63 infection only causes mild symptoms, whereas SARS-CoV and SARS-CoV-2 cause severe lower respiratory tract diseases. However, all three viruses can recognize and interact with common “virus-binding hotspot” regions in ACE2.^19-22 It has been reported that SARS-CoV-2 RBD has higher ACE2 binding affinity than SARS-CoV RBD, however, the full-length S of SARS-CoV-2 has comparable ACE2 binding affinity to that of SARS-CoV,²³ suggesting the SARS-CoV-2 RBD might be less exposed than SARS-CoV RBD.

While the catalytic site and intracellular domain of ACE2 are not required for the infection, the interaction between peptidase domain (PD) of host ACE2 and RBD of S is shown to mediate the viral entry of SARS-CoV-2.^{21, 22, 24} The structures of SARS-CoV-2 RBD/ACE2 complex have been determined by X-ray crystallography and cryo-electron microscopy.^{3, 21}

ACE2 distribution in human tissues may indicate their susceptibility to SARS-CoV-2 infection.²⁵ Several studies have analyzed the expression of ACE2 in different tissues and reported that ACE2 protein is mainly expressed in the small intestine, kidney, gallbladder, testis, and to a lesser extent in the lung.^25-27 Although the respiratory route does not express the highest ACE2, it is the dominant route of infection, suggesting some in-correlation between ACE2 expression and tissue susceptibility to SARS-CoV-2 infection.

The high expression of ACE2 in intestine and the successful detection of viral RNA in the stool and gastrointestinal tracts of patients with COVID-19, indicate that fecal-oral transmission is a potential transmission pathway.²⁸ However, one major obstacle to this transmission route is that SARS-CoV-2 could not tolerate gastric acid and survive passage into the gut.^{28, 29} Studies have shown that SARS-CoV-2 lost infectivity at low pH.³⁰ Consistent with this, patients taking proton pump inhibitors are more likely to be associated with higher infection rates or more severe clinical outcomes.³¹

In the respiratory tract, higher ACE2 expression was detected in the nose, while lower expression throughout the lower respiratory tract was revealed by high-sensitivity RNA in situ technology,³² and this distribution is in parallel with the susceptibility to SARS-CoV-2 infection. In addition, although lung tissue has relatively low ACE2 expression, circulating soluble ACE2 (sACE2) could contribute to its high susceptibility to viral infection.³³ sACE2 is a form of protease-cleaved ACE2 that is released into the plasma, which retains its ability to bind to SARS-CoV-2 and mediates viral endocytosis entry via AT1 or ACPR1B receptors.³³

2.3 Other co-receptors and receptor-independent virus infection

Besides ACE2, several other molecules have been reported to be involved in the SARS-CoV-2 entry process, which include neuropilin 1 (NRP1), phosphatidylserine receptors, the C-type lectins, ASGR1, KREMEN1, and CD147. Detailed functions of these co-receptors or cofactors have been summarized in reviews.^{10, 34}

In addition, SARS-CoV-2 could also infect cells via receptor-independent pathways. Viral spreads through cell-cell contact have been proved in cell cultures, which is reported to be mediated by S.³⁵ Cell-to-cell transmission has been well documented for other viruses, like HIV and HCV, which could help virus to evade antibody neutralization, resulting in an efficient way of virus transferring to neighboring cells. ACE2 could enhance SARS-CoV-2 cell-to-cell transmission, but it is not absolutely required.³⁵

Another form of receptor-independent transmission is mediated by exosomes or extracellular vesicles (EVs). EVs could transfer viruses or viral components from infected cells to healthy cells and thus facilitate infection. SARS-CoV-2 assembles in single membrane vesicles near ER-Golgi intermediate compartment and egresses through exocytosis.³⁶ Large viral-containing vesicles have been documented in SARS-CoV-2 infected cells and these multivesicular bodies are responsible for the generation of EVs.^{36, 37} EVs could transfer viruses or viral components to target cells by internalization, direct-fusion or receptor-ligand interaction.³⁷ Recent studies have indicated that EVs were involved in SARS-CoV-2 viral spreading and pathogenesis.^{38, 39} On the other hand, EVs could also transfer receptors, like ACE2 to receipt cells,⁴⁰ thus further contributing to the in-correlation of ACE2 expression and SARS-CoV-2 tropism.

3.1 S1/S2 cleavage by furin

As described above, a polybasic cleavage site (PRRAR) at the S1/S2 boundary of SARS-CoV-2 S can be cleaved by furin in the infected cells, resulting in the formation of two noncovalently linked subunits with different functions-S1 for receptor binding and S2 for membrane fusion. The S1 subunit in this non-covalently linked structure is prone to premature dissociation, while the D614G mutation features new interactions that prevent premature loss of S1 subunit and enhances infectivity.^{43, 44} The cleavage of S1/S2 is essential for entry of SARS-CoV-2 into human lung cells as well as cell-cell fusion.¹⁴ In addition, this cleavage site has also been indicated to affect viral fitness and viral transmission,^{45, 46} and thus plays an important role in the infectivity of this virus. Notably, NRP1 has been shown to interact with the carboxyl-terminal of cleaved S1 and enhanced viral entry into host cells.^{47, 48} As NRP1 is abundantly expressed in the olfactory epithelium of the nasal cavity, it has been suggested to be involved in the spread and infection of the virus in the central nervous system.⁴⁹

Furthermore, S1/S2 cleavage is prerequisite for the cleavage at the S2′ site.⁵⁰ Although cleavage at S1/S2 is essential for SARS-CoV-2 infection, this cleavage does not produce conformational changes necessary to induce membrane fusion. A further cleavage at the S2′ site mediated either by TMPRSS2 on cell surface or by cathepsins in endosome is required to expose the fusion peptide (FP) and initiate the fusion of membranes.^{10, 51}

3.2 Direct membrane fusion and TMPRSS2

TMPRSS2 was first identified as a virus-activating protease that cleaves the HA glycoprotein of influenza A virus.⁵² TMPRSS2 belongs to type II transmembrane serine protease (TTSP) that is characterized by an N-terminal transmembrane domain, a stem region, and a C-terminal extracellular serine protease domain that contains a catalytic triad essential for proteolytic activity, comprising histidine, aspartic acid and serine.⁵³ TMPRSS2 is widely expressed in epithelial cells of respiratory tract, heart, prostate and gastrointestinal tracts.^{42, 54, 55} It has been associated with a wide range of physiological and pathological processes, such as digestion, tissue remodeling, blood coagulation, fertility, inflammatory responses, and tumor cell invasion.⁵⁶

TMPRSS2 has been reported to be involved in cell entry of influenza virus H1N1, H7N9, and human coronaviruses HCoV-229E, MERS-CoV, and SARS-CoV.^{56, 57} In vivo studies demonstrated that TMPRSS2-deficient mice showed less severe lung pathology when infected with certain influenza A virus strains, SARS-CoV or MERS-CoV,^58-60 suggesting TMPRSS2 is an essential host factor for these viruses. Therefore, inhibition of TMPRSS2 has been proposed as a promising approach for antiviral development.⁵⁷ It was then shown that TMPRSS2 inhibitor, camostat mesylate reduced the entry of SARS-CoV-2 into TMPRSS2-expressing cells, such as the human airway cells, suggesting TMPRSS2 was also crucial for SARS-CoV-2 infection of respiratory tract.⁶¹ In addition to TMPRSS2, other TTSPs or metalloproteases have also been reported to activate S.^{62, 63} In vitro studies confirmed that furin cleaved at the S1/S2 boundary while TMPRSS2 cleaved at the S2′ site, TMPRSS2 and furin could not compensate for each other for S activation.⁴² Therefore, S activation involves a cleavage at S1/S2 during synthesis and then a second cleavage at S2′ during viral entry into target cells. In particular, TMPRSS2 cleaves S at single arginine or lysine residues located immediately upstream of FP,⁴² allowing FP to insert into the host cell membrane, and thus mediate viral entry at cell surface.^{10, 56}

3.3 Endocytosis and cathepsin

On the other hand, many in vitro studies using various cell lines have demonstrated that lysosomotropic agents could block viral entry,^{13, 24, 61} suggesting an endocytic entry pathway for SARS-CoV-2. Further studies revealed that if cell surface proteases are absent or insufficient, SARS-CoV-2 prefers to enter cells via endocytosis.^{10, 64} Endocytosis is the process through which cells take in external substances by engulfing them with the cell membrane. Clathrin-mediated endocytosis (CME) is the best-characterized pathway, during which clathrin-coated vesicles are formed and extracellular substances are engulfed within.⁶⁵ SARS-CoV has been reported to enter cells via CME as well as other pathways.^{66, 67} Similarly, SARS-CoV-2 could also enter cells via CME,^{24, 68} during which the cytoplasmic tail and enzymatic activity of ACE2 are not required.^{24, 69}

During endocytosis, the S protein conformation changes in an acidic environment. It is generally believed that the cleavage of S at the S2′ site during endocytosis is mediated by the host protease cathepsin, followed by FP exposure and membrane fusion.^{10, 70} However, some studies have proposed that the cleavage sites for cathepsins are upstream of the S2′ site,^{71, 72} therefore, the exact cleavage sites for cathepsin or whether there are other proteases involved need further investigation. Cathepsins are nonspecific proteases with endopeptidase and exopeptidase activity, contributing to protein degradation in late endosomes and lysosomes.⁷³ Cathepsins are highly expressed in most human tissues and there are three catalytic classes of cathepsins: aspartic (D and E), serine (G), and cysteine (B, C, K, L, S, and V) proteases, of which cathepsin B is required for Ebola virus entry, while cathepsin L is more important for SARS-CoV and SARS-CoV-2 virus entry.^{74, 75}

As SARS-CoV-2 could exploit two entry pathways for viral entry, combinational use of inhibitors targeting TMPRSS2 and Cathepsin displayed strong synergy in blocking virus entry.⁷⁶

3.4 TMPRSS2 dictates the entry route used by SARS-CoV-2 to infect host cells

SARS-CoV-2 could infect cells via membrane fusion at cell surface or endocytosis, and the specific pathway used in a given cell type mainly depends on the expression of proteases, in particular TMPRSS2. The route of membrane fusion is preferred when TMPRSS2 is expressed, while in the absence of this protease, the virus will employ endocytosis for viral entry.⁷⁷ A study showed that cell surface membrane fusion occurs within 10 min after infection in a pH-independent manner, while the pH-dependent endocytosis takes about 40–60 min postinfection.⁴¹ Since both ACE2 and TMPRSS2 are expressed in the respiratory tract, especially the upper respiratory tract,^{78, 79} TMPRSS2-mediated membrane fusion is believed to play an important role in the infection of respiratory epithelium, while endocytosis might act as an alternative pathway during the amplification of SARS-CoV-2 in extrapulmonary organs.⁵⁰ In addition, overexpression of TMPRSS2 in non-TMPRSS2 expressing cells eliminated the dependence of infection on the cathepsin,⁴¹ highlighting the importance of TMPRSS2 expression in determining whether the viruses employ either membrane fusion or endocytosis pathways.

3.5 Membrane fusion

Enveloped viruses have evolved to employ a largely conserved mechanism to promote the fusion of viral and cellular membranes.^{80, 81} For most viruses with class I fusion proteins, glycoproteins are all homotrimers that contain coiled-coil structures that must be primed to become fusion-competent.⁸⁰ After the FP inserts into the target membrane, conformation changes cause the α-helical heptad repeats to fold back to form a six-helix bundle (6HB), which will pull membranes to form fusion pores.⁸¹

In the case of SARS-CoV-2, the S2 subunit responsible for membrane fusion forms a central helical bundle with the heptanucleotide repeat 1 (HR1) and HR2 that are essential for membrane fusion.¹⁰ After S binding to ACE2, the S2′ cleavage site is exposed and later processed by cathepsin or TMPRSS2. S2′ cleavage causes conformational changes in S2 and probably the dissociation of S1, followed by FP inserting into cell membrane. This fusion intermediate features a long central three-stranded coiled coil and the 6HB is formed by the folding back of HR2 to HR1.¹⁰ 6HB pulls the viral and cell membranes to apposition (two membranes brought close together) and facilitates the formation of the fusion pore, through which the viral genome is released into the cytosol (Figure 1C).¹⁰

Membrane fusion is not a spontaneous process and several energy barriers must be overcome before fusion occurs. For example, energies are required for membrane apposition, membrane curvature deformation and fusion-pore formation and expansion.⁸² Like other viral protein-mediated membrane fusion, free energy generated during S conformation changes from a metastable state to a thermodynamically more favorable state will be used to overcome the energy barrier for lipid bilayer fusion.^83-85

4.1 Methods for syncytia detection

The traditional method of syncytia assessment is to count multinucleated giant cells under a light microscope. While this method is straightforward, quantification is time-consuming and difficult: small syncytia may be easily overlooked while some unfused cell clusters may be counted by mistake. To overcome these issues, different methods have been developed to quantify syncytia (Figure 2). Slack et al.⁸⁷ established a fusion-dependent promoter activation and green fluorescent protein (GFP) expression system to quantify syncytia, in which a LacR-IE1 transcriptional activator is expressed in one group of cells, while a second group of cells is transfected with a plasmid containing the GFP gene under the control of a LacR-IE1 regulated promoter. Fusion of these two cells leads to promoter activation and GFP expression.⁸⁷ On the other hand, Buchrieser et al.⁸⁸ measured SARS-CoV-2 induced syncytia with split GFP system. Fusion of two groups of cells expressing GFP1-10 and GFP11 respectively, leads to the complementation of GFP, which can be detected with a fluorescent microscope or the high-content imaging system for screening purposes. Sanders et al.⁸⁹ employed dual-fluorescence labeling to quantify SARS-CoV-2 induced syncytia, in which they used RFP to label ACE2-expressing cells and GFP to label spike-expressing cells. Cell fusion of these two types of cells leads to multinucleated cells with both colors.

In addition, syncytia could also be quantified with luminescence. Our group established a coculture system to measure spike-induced syncytia based on split NanoLuc.⁹⁰ Fusion of cells expressing NanoLuc large subunit (LgBiT) and cells expressing small subunit (HiBiT) leads to complementation of NanoLuc luciferase, which can then be measured with live-cell NanoLuc substrate. Compared to the split GFP system, this system is more rapid and sensitive and is more suitable for high-throughput screening. With this system, we also showed that syncytia formation is much more dependent on ACE2 expression while moderate spike expression is able to induce syncytia,⁹⁰ partially explaining why syncytia is a commonly observed feature in SARS-CoV-2 infected tissue.

4.2 Syncytia formation as a pathological feature of severe COVID-19

The presence of multinucleated pneumocytes has been well documented in severe COVID-19 patients,^{91, 92} and patients who died as a direct consequence of SARS-CoV-2 infection had a higher rate of syncytia formation than those who died of other causes,⁸⁹ suggesting that syncytia contributed to the pathology of SARS-CoV-2 infections. Several COVID-19 autopsies also confirmed the presence of multinucleated giant cells in lung tissues.^93-95

Syncytia could facilitate viral transmission as these large multinucleated syncytia could be released from the primary infection sites and then enable multiple viral genomes to be collectively transferred into single cells. Syncytia-mediated viral spread has been proved during SARS-CoV-2 infection.^{35, 96} By using a reconstituted primary bronchial epithelia model, Beucher et al.⁹⁶ showed that syncytia released from the epithelium could transmit SARS-CoV-2 infection more efficiently than cell-free supernatant, suggesting syncytia enable en bloc viral transmission. Another advantage of this syncytia-mediated viral transmission is that it also enables viruses to spread without the need to assemble and/or enter the extracellular environment, thus protecting the virus from neutralizing antibodies as well as restriction factors that target viral assembly and release.⁹⁷

Although syncytia facilitate viral transmission, they might cause direct cytopathic effects to host cells. Syncytia-related cell death has been well documented in viruses such as HIV and RSV,^{98, 99} and the syncytial death via apoptosis or pyroptosis could release viruses into surrounding cells and thus further spread viral infection.⁹⁷ Syncytia induced by SARS-CoV-2 S also led to cell death, which has been confirmed with in vitro cell culture models.^{88, 90} Ma et al.¹⁰⁰ further proved that the death of syncytia induced by SARS-CoV-2 infection was mediated by GSDME-dependent pyroptosis.

In addition, syncytia have also been shown to target lymphocytes for cell death. Zhang et al.⁹⁴ reported that lymphocytes were detected within syncytia as a cell-in-cell structure among COVID-19 patients. This phenomenon was further confirmed with a coculture cell model showing that syncytia could internalize different lymphocytes to form cell-in-cell structure and internalized cells readily underwent quick cell death.⁹⁴ These results suggested that syncytia might contribute to peripheral blood lymphocyte loss in patients.

4.3 Spike protein induces syncytia formation

Pathological studies of COVID-19 patients or nonhuman primate models with syncytia suggest that most syncytia are of lung epithelial origin, but were generally negative for SARS-CoV-2 nucleocapsid protein.^{89, 101} Although many of the detailed mechanism of syncytia formation is unclear, it is generally believed that S protein expressed on infected cell surface interacts with the ACE2 receptor on neighboring cells and causes the syncytia formation, a process resembling the S-mediated viral-cell membrane fusion. Several studies have confirmed that syncytia could be recapitulated with in vitro cell culture systems by coculturing S-expressing and ACE2-expressing cells,^{88, 89, 92, 102} suggesting that S protein is the major fusogen protein involved. However, Buchrieser et al.⁸⁸ reported that no clear syncytia were formed in SARS-CoV-2 infected Caco2 cells, although these cells are permissive to infection, suggesting that syncytia formation is dependent on both viral and host factors (Figure 3).

4.4 Viral factors affecting syncytia formation

The three highly pathogenic coronaviruses, MERS-CoV, SARS-CoV, and SARS-CoV-2 all could cause serious lung injury and fatal infections. However, their abilities to induce syncytia differ. First, S protein from SARS-CoV has poor ability to induce syncytia,^{14, 94} while S from both MERS-CoV and SARS-CoV-2 has been proven to cause syncytia.¹⁴ Coincidentally, spikes from the latter two coronaviruses have the furin cleavage site, which is missing in SARS-CoV S, suggesting that the furin cleavage site might play a role during syncytia formation. Consistently, inhibition of furin activity or mutations destroying the furin cleavage site of SARS-CoV-2 S have been proven to abolish syncytia,^{14, 102} and Zhang et al.⁹⁴ further proved that a bi-arginine motif adjacent to the furin cleavage site containing R682 and R685 dictates syncytium formation. Second, compared to MERS-CoV spike, a crucial third arginine at the furin cleavage site of SARS-CoV-2 conferred complete S1/S2 cleavage, while MERS-CoV S was processed by furin at very low efficiency.¹⁰³ Consistently, nonhuman primates infected with MERS-CoV showed fewer syncytia than those infected with SARS-CoV-2,¹⁰¹ further arguing the role of furin cleavage in syncytia formation. In addition to the furin cleavage site, cellular proteases and ACE2 expression are also important factors in the formation of syncytia. TMPRSS2 moderately increased SARS-CoV2 S-driven fusion, while ACE2 strongly increased it. In contrast, TMPRSS2 strongly increased SARS-CoV S-driven fusion, while ACE2 moderately increased it.¹⁰⁴

As mentioned, S alone is sufficient to trigger syncytia in cells expressing ACE2 receptor.^{88, 94, 102} Sequence analysis showed that the ER-Golgi retrieval motif is not required for syncytia, however, a membrane-proximal region of high cysteine content that can be palmitoylated is essential for membrane fusion.⁸⁹ Palmitoylation of this region has been reported to be required for murine coronavirus (mouse hepatitis virus, MHV), and SARS-CoV spike-mediated syncytia formation,^105-107 suggesting a conserved role of cysteine palmitoylation during coronavirus infection. In addition, other post-translation modifications including phosphorylation could also affect cell-cell fusion. For example, two serines at the edges of the furin cleavage site have been reported to be efficiently phosphorylated by proline-directed and basophilic protein kinases,¹⁰³ which blocked furin cleavage and thus would compromise cell-cell fusion.

4.5 S of VOC and syncytia formation

As an RNA virus, coronaviruses would accumulate mutations due to selection pressures during circulation. SARS-CoV-2 has evolved into different variants, among which, Alpha (B.1.1.7), Beta (B.1.351), Gamma (P.1), Delta (B.1.617.2) and Omicron (B.1.1.529) have been listed as VOCs by the WHO.⁷ Interestingly, some of the variants’ S proteins contain mutations that alter their ability to induce syncytia. Notably, the Delta S has been shown to promote syncytia formation with relatively larger and faster membrane fusion than the other variants.^{24, 108-110} In contrast, although the Omicron variant has increased transmissibility, its S protein has been shown to have a reduced ability to induce the formation of syncytia.^{111, 112} On the other hand, contradictory results have been reported for Alpha and Beta. Rajah et al.¹⁰⁸ reported S proteins of these variants are more syncytiogenic compared to Wuhan ancestral while Hoffmann et al.¹¹³ showed that Alpha and Beta S did not increase syncytia formation. The discrepancy is possibly due to different assay systems employed by different groups.

Of all the mutations associated with the S protein, mutation at P681 draws a lot of attention due to its location at the S1/S2 cleavage site. Two kinds of naturally-occurred mutations have been reported to be associated with variants, which are P681H (Alpha, Omicron) and P681R (Delta).¹¹⁴ Both mutations have been suggested to enhance S cleavage efficiency and alter the ability of syncytia induction.^{24, 97, 110, 113, 114} However, whether there is a direct correlation between S cleavage and syncytia formation still needs further investigation.

In addition to the S1/S2 cleavage efficiency, S-ACE2 affinity has also been suggested to affect syncytia formation and high S-ACE2 affinity correlates to high fusogenicity.¹⁰⁸ Mutations within the RBD, such as N501Y, K417N, and E484K, which are found in VOC, have been reported to increase the affinity of S protein to ACE2 and were associated with higher syncytia formation.¹⁰⁸

4.6 Host factors affecting syncytia formation

In addition to the fusogenic ability of S protein, many host factors can also modulate syncytia formation. Interferon-induced transmembrane (IFITM) proteins are small transmembrane proteins that are products of interferon-stimulated genes (ISGs).¹¹⁵ IFITM can inhibit a broad spectrum of viral infections by blocking viral membrane fusion.¹¹⁵ It has been reported that IFITMs could inhibit SARS-CoV-2 infection, although opposing results have also been reported.^116-118 SARS-CoV-2-induced syncytia could be effectively inhibited by the expression of IFITMs, with IFITM1 being the most potent molecule.^{88, 108} But this inhibitory effect could be reversed by the expression of the serine protease, TMPRSS2,⁸⁸ which increased and accelerated S-induced syncytia with or without IFITMs expression. In vitro cell culture model indicated that both IFITMs and TMPRSS2 modulate the efficiency of fusion when they are present in the same cell with ACE2, rather than in the S-expressing cell.⁸⁸ Another ISG product that inhibits syncytia formation is lymphocyte antigen 6 complex, locus E (LY6E),¹¹⁹ but the specific mechanism is not well understood.

An antisyncytial screening study with clinically approved drugs revealed that drugs regulating intracellular Ca²⁺ levels are top candidates.⁹² Mechanism analysis revealed that the Ca²⁺-activated TMEM16 family is essential for syncytia formation. In particular, overexpression of TMEM16F significantly stimulated S-induced syncytia, while TMEM16F inhibition blunted syncytia.⁹² TMEM16F, an ion channel protein, also functions as a lipid scramblase responsible for phosphatidylserine externalization onto the outer leaflet of the plasma membrane. Phosphatidylserine externalization is believed to serve as a conserved “fuse me” signal¹²⁰ and thus an important event for plasma membrane fusion.

In addition to phosphatidylserine, other lipid molecules are also essential for cell membrane fusion. Sanders et al.⁸⁹ identified several drugs that could perturb membrane lipid composition and inhibit syncytia through high-throughput screening. In particular, cholesterol is required for syncytia formation but via a raft-independent mechanism. Drugs such as 25-hydroxycholesterol or methyl-beta-cyclodextrin that interfere with cell membrane cholesterol could inhibit syncytia in a dose-dependent manner.^{89, 121, 122}

To summarize, successful infection of SARS-CoV-2 in human airway epithelial cells leads to viral replication and the expression of viral-encoded proteins, including S. While most of the newly synthesized S accumulates in ER-Golgi intermediate compartment for virion assembly, some of the S travels past the Golgi and expresses on cell surface.^123-125 The cell surface-expressed S could then interact with ACE2 on neighboring cells and initiate cell–cell fusion. The expression of S on infected cell surface as well as ACE2 expression on target cells are both required for syncytia formation.⁹⁰ Similar to the process of viral-cell membrane fusion, S1/S2 cleavage is also involved in syncytia formation.^{14, 46, 90} Additionally, an efficient cell-cell membrane fusion also requires cleavage at S2′ site,¹²⁶ which is further supported by the fact that TMPRSS2 enhances syncytia formation.^{88, 90} The exposure of the FP will then initiate the membrane fusion process, during which lipid bilayers rearrange with lots of host protein factors and lipid molecules mentioned above. SARS-CoV-2 induced multinucleated pneumocyte is a feature of severe COVID-19 patients. Most of the syncytial cells end up dying together with internalized lymphocytes, with possible concomitant dissemination of viruses to initiate new infections.