Characterization of spike glycoprotein of 2019-nCoV on virus entry and its immune cross-reactivity with spike glycoprotein of SARS-CoV

Since beginning of this century, there have already been three zoonotic outbreaks caused by beta coronaviruses (CoV), SARS-CoV in 2002-2003, MERS-CoV in 2012, and the newly identified 2019-nCoV in late 2019, Wuhan, China. As to Feb 10 th , 2020, there are over 40,000 confirmed cases and over 900 deaths. However, little is known about the biology of this newly emerged virus. Here we developed a lentiviral based pseudovirus system for S protein of 2019-nCoV to study virus entry in BSL2 settings. First, we confirmed that human angiotensin converting enzyme 2 (hACE2) is the main entry receptor for 2019-nCoV. Second, we found that 2019-nCoV S protein mediated entry on 293/hACE2 cells was mainly through endocytosis, and PIKfyve, TPC2, and cathepsin L are critical for virus entry. Third, 2019-nCoV S protein is less stable than SARS-CoV, and it could trigger protease-independent and receptor dependent cell-cell fusion, which might help virus rapidly spread from cell to cell. Finally and more importantly, polyclonal anti-SARS S1 antibodies T62 effectively inhibited entry of SARS-CoV S pseudovirions, but almost had no effect on entry of 2019-nCoV S pseudovirions. Further studies using sera from one recovered SARS-CoV patient and five 2019-nCoV patients showed that there was only limited cross-neutralization activities between SARS-CoV and 2019-nCoV sera, suggesting that recovery from one infection might not protect against the other. Our results present potential targets for development of drugs and vaccines for 2019-nCoV.


Introduction
Coronaviruses (CoVs) infect human and animals and cause varieties of diseases, including respiratory, enteric, renal, and neurological diseases 1 . They are classified into four genera, alpha-CoV, beta-CoV, gamma-CoV, and delta-CoV 2 . Since beginning of this century, there have already been three zoonotic outbreaks of beta-CoVs. In 2002-2003, severe acute respiratory syndrome coronavirus (SARS-CoV) 3,4 , a lineage B beta-CoV, emerged from bat and palm civet 5,6 , and infected over 8000 people and caused about 800 deaths 7 . In 2012, Middle East respiratory syndrome coronavirus (MERS-CoV), a lineage C beta-CoV, was discovered as the causative agent of a severe respiratory syndrome in Saudi Arabia 8 , currently with 2494 confirmed cases and 858 deaths 9  CoV uses its spike glycoprotein (S), a main target for neutralization antibody, to bind its receptor, and mediate membrane fusion and virus entry. Each monomer of trimeric S protein is about 180 kDa, and contains two subunits, S1 and S2, mediating attachment and membrane fusion, respectively. In the structure, N-and C-terminal portions of S1 fold as two independent domains, N-terminal domain (NTD) and C-terminal domain (C-domain) ( Fig 1A). Depending on the virus, either NTD or C-domain can serve as the receptor binding domain (RBD). While RBD of mouse hepatitis virus (MHV) is located at the NTD 14 , most of other CoVs including SARS-CoV and MERS-CoV use C-domain to bind their receptors [15][16][17][18][19] . MHV uses mouse carcinoembryonic antigen related cell adhesion molecule 1a (mCEACAM1a) as the receptor 20 , and the receptors for SARS-CoV and MERS-CoV are human angiotensin-converting enzyme 2 (hACE2) 21 and human dipeptidyl peptidase 4 (hDPP4) 22 , respectively. While S proteins of 2019-nCoV share overall higher percentage of amino acid identities with SL-CoV ZC45 23 than SARS-CoV, the amino acid sequence of potential RBD of 2019-nCoV is more homologous to that of SARS-CoV (74%) than SL-CoV ZC45 (72%). Recently, Zhou et al reported that 2019-nCoV uses hACE2 as the receptor 13 .
However, whether any of these proteases could promote virus entry of 2019-nCoV remain elusive.
In this study, using a lentiviral pseudotype system, we determined cell type susceptibility, virus receptor, entry pathway, and protease priming for 2019-nCoV, and identified several potential drug targets for 2019-nCoV. Surprisingly, we also found that limited crossneutralization between convalescent sera from SARS-CoV and 2019-nCoV patients.

Results
To enhance expression of the S protein of 2019-nCoV in mammalian cells, a codonoptimized cDNA encoding the S protein and 3xFLAG tag was synthesized, and to facilitate incorporation of S protein into lentiviral pseudovirons, the last 19 amino acids containing an ER-retention signal from the cytoplasmic tail of the S protein was removed ( Fig 1A).
The new construct was named 2019-nCoV S here. HEK 293T cells were transfected with 2019-nCoV S plasmid and expression of 2019-nCoV S protein was determined by western blot. There were two major bands, 180 kDa, and 90 kDa, detected by mouse anti-FLAG M2 antibody ( Fig 1B, lane 2), reflecting the full length and cleaved S proteins, respectively.
The band above 250 kDa likely results from dimeric or trimeric S proteins. Consistent with our previous report 27 , MERS-CoV S protein was detected by polyclonal goat anti-MHV S antibodies AO4 (Fig 1C). Interestingly, both 2019-nCoV and SARS S proteins were also detected by polyclonal goat anti-MHV S antibodies AO4, suggesting the presence of a conserved immunogenic epitope among all four different CoVs, likely in S2. Surprisingly, although S1 subunits of 2019-nCoV and SARS-CoV share almost 64% in amino acid identities, 2019-nCoV S protein was barely detected by rabbit polyclonal anti-SARS S1 antibodies T62 (Fig 1D), suggesting that the major epitope(s) for T62 antibodies should lie in the region of S1 where the sequence differs between SARS-CoV and 2019-nCoV. The 2019-nCoV S protein was not detected by either the monoclonal anti-SARS S1 antibody or anti-MERS S2 antibody.
The efficiency of 2019-nCoV S protein incorporation into pseudovirions was also evaluated using monoclonal mouse anti-FLAG M2 antibody and polyclonal goat anti-MHV S antibody AO4. While majority of SARS-CoV S proteins incorporated into pseudovirons were full length, at 180 kDa ( Fig 1H and 1I), most of 2019-nCoV S proteins on lentiviral pseudovirions were cleaved, about 90 kDa (Fig 1G and 1H), likely reflecting presence of extra furin site (R682-R683-A684-R685, Fig 1A) between S1 and S2 in 2019-nCoV S protein. Consistent with the results in cell lysate (Fig 1D), SARS-CoV S proteins, but not 2019-nCoV S proteins, in pseudovirions was readily detected by using polyclonal rabbit anti-SARS S1 antibodies T62 ( Fig 1I).
Next, we determined whether 2019-nCoV S pseudovirions were able to transduce human, monkey, and bat cells. VSV-G pseudovirons were used as a positive control, whereas bald particles with no spike proteins (mock) were served as a negative control. As expected, all cell types were effectively transduced by VSV-G pseudovirons (Fig 2A). Compared to mock control, 2019-nCoV S pseudovirions showed an over 500-fold increase in luciferase activities in Calu3 cells, at a level similar to SARS-CoV S pseudovirions (Fig 2A) prevented by pre-incubation of soluble hACE2 at both 10 g/ml and 50 g/ml (Fig 2D), further supporting the notion that hACE2 is the receptor.
Since majority of S proteins on 2019-nCoV S pseudovirions are cleaved, we then determined whether 2019-nCoV S pseudovirons entered cells through endocytosis or cell surface. HEK 293/hACE2 cells were treated with lysosomotropic agents, ammonia chloride and bafilomycin A, and their effect on virus entry was evaluated. Consistent with previous reports, 20 mM NH 4 Cl and 100 nM bafilomycin A decreased entry of SARS-CoV S and VSV-G pseudovirions by over 99%, compared to no treatment control. More than 98% reduction in transduction on 293/hACE2 cells by 2019-nCoV S pseudovirions was also shown when the cells were incubated with either NH 4 Cl or bafilomycin A (Fig 3A), indicating that 2019-nCoV S pseudovirions enter 293/hACE2 cells mainly though endocytosis, despite that its spike proteins were cleaved.
Phosphoinositides play many essential roles in endocytosis. Among them, one is phosphatidylinositol-3,5-bisphosphate (PI(3,5)P 2 ), which regulates early endosome to late endosome dynamics 32,33 . Phosphatidylinositol-3-phosphate 5-kinase (PIKfyve) is the main enzyme synthesizing PI(3,5)P 2 in early endosome. HEK 293/hACE2 cells were treated with apilimod, a potent inhibitor for PIKfyve 34 . Inhibition of PIKfyve by apilimod significantly reduced entry of SARS-CoV S pseudovirions on 293/hACE2 cells in a dose dependent manner (Fig 3B), whereas it had no effect on entry of VSV-G pseudovirions, which occurred in early endosome. Similar inhibitory effects were observed when HeLa/hDPP4 cells and HeLa cells stably expressing mouse carcinoembryonic antigen related cell adhesion molecule 1a (mCEACAM1a) (HeLa/mCEACAM1a) were treated with apilimod and transduced with MERS-CoV and MHV S pseudovirions (Fig 3B), respectively. Moreover, infection of live MHV on 17Cl.1 cells was also strongly inhibited by apilmod treatment (Fig 3C), no significant cell toxicity was observed on apilimod at any concentration tested (Fig 3C). We then determined whether apilimod could inhibit entry of 2019-nCoV S pseudovirions on 293/hACE2. As expected, apilimod treatment significantly inhibited entry of 2019-nCoV S pseudovirions in a dose dependent manner ( Fig 3D). Similar effects were shown when 293/hACE2 cells were treated with YM201636, another PIKfyve inhibitor ( Fig 3E). These results suggested that PIKfyve might be a potential general drug target for viruses that enter cells through endocytosis. Two pore channel subtype 2 (TPC2) and TRPML1 in lysosome are two major downstream effectors of PI(3,5)P 2 35 . While blocking TPC2 activity by tetrandrine, a inhibitor for TPC2 36 , significantly decreased entry of 2019-nCoV S pseudovirions ( Fig 3F), treatment of cells with 130, a TRPML1 inhibitor, had no effect ( Fig   S1), indicating that TPC2, not TRPML1, is important for 2019-nCoV entry.
Protease "priming" on S protein is an important step for coronavirus entry, and cathepsins shorter time and lower temperature to be inactivated ( Fig 4D and 4E), suggesting that 2019-nCoV S protein might be easier to be triggered. We also found that expressing type II membrane serine protease (TMPRSS) 2, 4, 11A, 11D, and 11E on 293/hACE2 cells significantly enhance 2019-nCoV S protein mediated cell-cell fusion (Fig S2).
Since 2019-nCoV S reacted weakly with polyclonal rabbit anti-SARS S1 antibodies T62 in western blot, we then investigated whether 2019-nCoV S protein in native conformation could be recognized by anti-SARS S1 antibodies T62. HEK 293T cells transiently expressing 2019-nCoV S proteins were incubated with polyclonal anti-SARS S1 antibodies T62 and analyzed by flow cytometry. SARS-CoV S was used as a positive control. As expected, expression of SARS-CoV S proteins on 293T cell surface were readily detected by polyclonal anti-SARS S1 antibodies T62 (Fig 5A). On the contrast, only low level of binding of 2019-nCoV S protein to polyclonal rabbit anti-SARS S1 antibodies T62 was detected substitution had major effect on SARS-CoV S protein expression ( Fig S3) and binding of SARS-CoV S to polyclonal rabbit anti-SARS S1 antibodies T62 (Fig 5D), suggesting that any of these residues might not be any direct antigenic sites for polyclonal antibodies T62.
Next, we tested whether polyclonal rabbit anti-SARS S1 antibodies T62 could inhibit entry of 2019-nCoV S pseudovirions. CoV S protein pseudovirions were incubated with polyclonal antibodies T62 on ice for 1 hr, and their transduction was measured according to luciferase activities. As expected, polyclonal antibodies T62 neutralized SARS-CoV S pseudovirion effectively in a dose dependent manner (Fig 5E). In contrast, even at a concentration of 50 g/ml, polyclonal antibodies T62 did not have marked effect on transduction by 2019-nCoV S proteins (Fig 5E).
Because rabbit polyclonal anti-SARS S1 antibodies did not show significant neutralization  Both S proteins use hACE2 as the receptor for binding and entry, which we further confirmed by flow cytometry analysis and competitive inhibition experiment using soluble hACE2 in this study (Fig 3B and 3C Recently studies showed that early to late endosome maturation is regulated by PI(3,5)P 2 , and inhibition of PIKfyve, the key enzyme synthesizing PI(3,5)P 2 , and TPC2, a downstream effector in lysosome, significantly reduced virus entry of MERS-CoV 49 . We also confirmed this in this study, and further showed that blocking PIKfyve and TPC2 also strongly inhibited entry mediated by 2019-nCoV S protein, indicating that PI(3,5)P 2 pathway might be considered as potential general drug targets for CoV infection.
CoV S protein is one of key component determining virus virulent, tissue tropism, host range and is also one of main targets for neutralization antibodies and vaccine design.
Although the S proteins of 2019-nCoV and SARS-CoV are highly homologous, to our surprise, polyclonal rabbit anti-SARS S1 antibodies T62 did not bind to 2019-nCoV S protein well, and poorly neutralized 2019-nCoV S protein mediated virus entry. Further analysis reveals that major immune-epitopes for T62 antibodies likely lie in the region of RBD (Fig 4A) After washing, cells were then analyzed by flow cytometry.

Soluble hACE2 inhibition assay
Briefly, lentiviruses pseudotyped with 2019-nCoV S, SARS-CoV S or VSV-G were preincubated with serially diluted soluble hACE2 for 1 hr on ice. The mixture were then added on 293/hACE2 cells, followed by centrifugation inoculation for 1 hr at room temperature.
Cells were fed with fresh medium 6 hrs later and lysed at 48 hrs post-inoculation.
Pseudoviral transduction was measured according to luciferase activities.
Pseudovirus neutralization assay SARS-CoV S, 2019-nCoV S, and VSV-G pseudovirions were pre-incubated with serially diluted either polyclonal rabbit anti-SARS S1 antibodies T62 or patient sera for 1 hr on ice, then virus-antibody mixture was added onto 293/hACE2 cells in a 96-well plate. After 6 hrs incubation, the inoculum was replaced with fresh medium.    Experiments were done triplicate and repeated at least three times. One