key: cord-0805250-a4oejky0
authors: Sasaki, Michihito; Uemura, Kentaro; Sato, Akihiko; Toba, Shinsuke; Sanaki, Takao; Maenaka, Katsumi; Hall, William W.; Orba, Yasuko; Sawa, Hirofumi
title: SARS-CoV-2 variants with mutations at the S1/S2 cleavage site are generated in vitro during propagation in TMPRSS2-deficient cells
date: 2020-08-28
journal: bioRxiv
DOI: 10.1101/2020.08.28.271163
sha: 615651d20541074a45ae35e39e1e4b7741f0c47b
doc_id: 805250
cord_uid: a4oejky0

The spike (S) protein of Severe Acute Respiratory Syndrome-Coronavirus-2 (SARS-CoV-2) binds to a host cell receptor which facilitates viral entry. A polybasic motif detected at the cleavage site of the S protein has been shown to broaden the cell tropism and transmissibility of the virus. Here we examine the properties of SARS-CoV-2 variants with mutations at the S protein cleavage site that undergo inefficient proteolytic cleavage. Virus variants with S gene mutations generated smaller plaques and exhibited a more limited range of cell tropism compared to the wild-type strain. These alterations were shown to result from their inability to utilize the entry pathway involving direct fusion mediated by the host type II transmembrane serine protease, TMPRSS2. Notably, viruses with S gene mutations emerged rapidly and became the dominant SARS-CoV-2 variants in TMPRSS2-deficient cells including Vero cells. Our study demonstrated that the S protein polybasic cleavage motif is a critical factor underlying SARS-CoV-2 entry and cell tropism. As such, researchers should be alert to the possibility of de novo S gene mutations emerging in tissue-culture propagated virus strains.

was the dominant pathway employed by SARS-CoV-2 in TMPRSS2-expressing cells [6] .

In contrast, camostat had no impact on entry of the S gene mutant, del2, into 1 4 0 Vero-TMPRSS2 cells but E-64d treatment resulted in a dose-dependent decrease in del2 1 4 1 entry (Fig. 4a) . These results suggested that S gene mutant, del2, can enter Vero-TMPRSS2 1 4 2 cells via cathepsin-dependent endocytosis but not the TMPRSS2-mediated fusion pathway.

Parental Vero cells that do not express TMPRSS2 were inoculated with S gene mutant 1 4 4 viruses in the presence of camostat and/or E-64d. Addition of E-64d inhibited the entry of 1 4 5 all S gene mutants into both Vero-TMPRSS2 and parent Vero cells; by contrast, camostat 1 4 6 had no impact on S gene mutant entry into these target cells (Fig. 4b ). These results 1 4 7 8 suggested that, in contrast to WT virus, S gene mutants enter into cells via 1 4 8 cathepsin-dependent endocytosis only, regardless of the presence or absence of TMPRSS2.

Because WT virus and S gene mutants showed different sensitivities to the treatment 1 5 0 with camostat, an agent currently under exploration as a candidate antiviral for clinical use 1 5 1

[17], we also examined the impact of other antiviral agents including nafamostat (a 1 5 2 TMPRSS2 inhibitor) [18, 19] and remdesivir (a nucleotide analog) [20, 21] . Antiviral 1 5 3 effects in Vero-TMPRSS2 cells were estimated by a cell viability assay based on the 1 5 4 generation of cytopathic effects [22] . Consistent with previous studies [18, 19] , nafamostat 1 5 5 showed higher antiviral efficacy against WT virus than was observed in response to 1 5 6 camostat; however, nafamostat had no antiviral activity against the S gene mutants ( Table   1 5 7 1). In contrast, remdesivir inhibited infection of both WT and S gene mutants with similar 1 5 8 EC 50 values (Table 1 ). These results indicated that S gene mutants are resistant to the 1 5 9 treatment with TMPRRSS2 inhibitors, but are sensitive to antivirals that target post entry In an effort to understand the selection mechanisms underlying the generation of these 1 6 4 mutant variants, we estimated the frequency of S gene mutants in virus population of 1 6 5 SARS-CoV-2 that had undergone serial passage in cultured cells. SARS-CoV-2 from an 1 6 6 original virus stock was underwent passage once (P1) to four times (P4) in Vero or up to 1 6 7 eight times (P8) in Vero-TMPRSS2. Nucleotide sequence heterogeneity at the S1/S2 1 6 8 cleavage site was determined by deep sequencing and variant call analysis. More than one 1 6 9 9 million sequence reads from each passaged sample were mapped onto the S1/S2 cleavage 1 7 0 site and analyzed for sequence variation. No sequence variants were observed in virus 1 7 1 populations until P8 in Vero-TMPRSS2 (Fig. 5a) . In contrast, nucleotide sequence 1 7 2 deletions around the S1/S2 cleavage site corresponding to del1 and del2 mutants were 1 7 3 observed in all three biological replicates of SARS-CoV-2 populations passaged in Vero 1 7 4 cells (Fig. 5a) . Notably, WT nucleotide sequences were detected in fewer than 20% of the 1 7 5 isolates evaluated at P2 and the WT was completely replaced with S gene mutants at P4 in 1 7 6

Vero cells. These results indicated that SARS-CoV-2 propagation in Vero cells results in a 1 7 7 profound selection favoring the S gene mutants. S gene mutants del3 and R685H were not 1 7 8 identified in the virus populations from P1 to P4. An additional variant del4 with a deletion 1 7 9 of 5 amino acids at a point immediately upstream of the RRAR motif (Figs. S2a and S2b), 1 8 0 was detected as a minor variant in sample #1 at P2. These results suggest that these specific 1 8 1 mutations occur only at low frequency.

We then determined the frequency of S gene mutants in virus populations passaged in 1 8 3

Calu-3 and Caco-2, which are cells that endogenously express TMPRSS2 [13, 14] , and also 1 8 4 in 293T-ACE2 that do not express TMPRSS2 [6] . No S gene mutants were identified in 1 8 5 SARS-CoV-2 passaged in Calu-3 and Caco-2 until P4; by contrast, S gene mutants 1 8 6 emerged at P2 in 293T-ACE2 cells (Fig. 5b) . We also identified an additional variant 1 8 7 R682P carrying a single amino acid substitution at the RRAR motif (Figs. S2a and S2b) at 1 8 8 P3 and P4 in 293T-ACE2 cells (Fig. 5b) . Taken together, these results suggest a strong 1 8 9 association between TMPRSS2 deficiency and the emergence of S gene mutants. Trypsin is a serine protease that is typically added to culture medium to induce 1 9 1 cleavage and activation of viral proteins, including the hemagglutinin (HA) protein of 1 9 2 influenza virus and the fusion (F) protein of paramyxovirus to promote growth in 1 9 3 TMPRSS2-deficient cells [23] . Recent studies report that trypsin treatment activates 1 9 4 SARS-CoV-2 S protein and induces syncytia formation in cells that transiently express the 1 9 5 virus S protein [7, 24] . As such, we examined whether exogenously added trypsin could 1 9 6 compensate for TMPRSS2 deficiency and thus inhibit the emergence of S gene mutants 1 9 7 during SARS-CoV-2 propagation in Vero cells. Deep sequencing analysis revealed that S 1 9 8 gene mutants did emerge and accounted for the majority of the virus population after P2 in 1 9 9

Vero cells cultured in serum-free medium with added trypsin ( In this study, we isolated S gene mutants from SARS-CoV-2 WK-521, a strain isolated difficult to impossible to maintain SARS-CoV-2 with the S1/S2 cleavage site in its intact 2 1 6 form.

The present study characterized S gene mutants as SARS-CoV-2 variants that generate 2 1 8 small plaques and that have a narrow range of cell tropism. The phenotypic alterations of S 2 1 9 gene mutants might be explained by noting that the S gene mutants were unable enter target 2 2 0 cells via direct fusion mediated by TMPRSS2. Indeed, a previous study demonstrated that 2 2 1 the polybasic cleavage motif at the S1/S2 cleavage site was indispensable for the entry of 2 2 2 VSV-pseudotyped viruses into Calu-3 cells that expressed TMPRSS2 [7] . Further studies 2 2 3 using infectious S gene mutants will provide new insights into the role of the polybasic 2 2 4 amino acid motif at the S1/S2 cleavage site with respect to both SARS-CoV-2 infection and 2 2 5 its pathogenicity.

At this time, many studies are conducted using SARS-CoV-2 propagated in Vero cells.

Considering the very real possibility that these virus stocks will accumulate S gene 2 2 8 mutations, researchers must pay careful attention to the passage history of any working 2 2 9 stocks of SARS-CoV-2. Moreover, we must be very objective when interpreting the results 2 3 0 from studies using Vero-passaged virus, especially those focused on S protein cleavage, Cells were infected with either WT or S mutants of SARS-CoV-2 at an MOI of 1.

After 24 h, cells were fixed with 3.7% buffered formaldehyde, permeabilized with ice-cold 2 8 3 methanol, and incubated with anti-SARS-CoV-2 S antibody (GTX632604, GeneTex). Cells were inoculated with either WT or S mutants of SARS-CoV-2 at an MOI of 0.1.

After 1 h of incubation, cells were washed twice with phosphate-buffered saline (PBS) and 2 9 2 cultured in fresh medium with 2% FBS. The culture supernatants were harvested at 24, 48, 2 9 3 and 72 h after inoculation. Virus titers were evaluated by plaque assay. 

Vero-TMPRSS2 were infected with either WT or S mutants of SARS-CoV-2 at 4-10 3 1 5 TCID 50 and added to the plates. Plates were incubated at for 3 days, and CPE was 3 1 6 determined for calculation of 50% endpoints using MTT assay. The concentration 3 1 7

achieving 50% inhibition of cell viability (effective concentration; EC 50 ) was calculated. The original stock of SARS-CoV-2 strain WK-521 was serially passaged in Vero,

Vero-TMPRSS2, Calu-3, Caco-2, and 293T-ACE2 cells in complete culture medium or 3 2 2 (for Vero) in serum free DMEM supplemented with 0.5 μ g/ml trypsin (Gibco); three Full-length S protein is cleaved into S1 and S2 proteins at the S1/S2 cleavage site. Functional domains (RBD, receptor binding domain; RBM, receptor binding motif) are highlighted. (b) Multiple amino acid sequence alignments focused on the S1/S2 cleavage site of wild type (WT) and isolated mutant viruses (del1, del2, del3 and R685H). Amino acid substitutions and deletions are shown as gray boxes, and the polybasic cleavage motif (RARR) at the S1/S2 cleavage site is highlighted in red. A red arrowhead indicates S1/S2 cleavage site. 

(a) Vero-TMPRSS2 cells were infected with SARS-CoV-2 WT or del2 mutant in the presence of varying concentrations of the TMPRSS2 inhibitor, camostat, or the cathepsin B/L inhibitor, E-64d, for 1h. At 6 h post-inoculation, the relative levels of viral N protein RNA were evaluated quantitatively by qRT-PCR. (b) Vero-TMPRSS2 and Vero cells were infected with SARS-CoV-2 WT or S gene mutants in the presence of 50 μM camostat and/or 25 μM E-64d for 1 h. At 6 h post-inoculation, the relative levels of viral N protein RNA were quantified by qRT-PCR. Cellular β-actin mRNA levels were used as reference controls. The values shown are mean SD of triplicate samples. One-way analysis of variance with Dunnett's test was used to determine the statistical significance between the responses to treatment with inhibitors and the no-treatment controls; *p < 0.05, **p < 0.01, ***p < 0.001. Multiple (a) nucleotide and (b) amino acid sequence alignments were constructed based on the sequence of WT and SARS-CoV-2 variants identified by deep-sequencing (related to Fig. 4) . Infectious viruses of del4 and R682P were not isolated in this study. Nucleotide substitutions and deletions are shown as gray boxes. Sequence encoding the polybasic cleavage motif (RARR) at the S1/S2 cleavage site is highlighted in red. 

SARS-CoV-2 was serially passaged in Vero cells in serum free DMEM containing trypsin with three biological replicates. Nucleotide sequence diversity at viral S1/S2 cleavage site was determined by deepsequencing.

SARS-CoV-2 (R685H) ATCAGACTCAGACTAATTCTCCTCGGCGGGCACATAGTGTAGCTAGTCAATCCATCATTGC

Multiple nucleotide sequence alignment of S1/S2 cleavage site of wild type and isolated SARS-CoV-2 mutants

Nucleotide substitutions and deletions are shown as gray boxes. Sequence encoding the polybasic cleavage motif (RARR) at the S1/S2 cleavage site is highlighted in red

SARS-CoV-2 (R682P) ATCAGACTCAGACTAATTCTCCTCCGCGGGCACGTAGTGTAGCTAGTCAATCCATCATTGC

SARS-CoV-2 (R682P) ECDIPIGAGICASYQTQTNSPPRARSVASQSIIAYTMSLGAENSVAYSNNS

ECDIPIGAGICASYQTQT----------SQSIIAYTMSLGAENSVAYSNNS SARS-CoV-2 (del2) ECDIPIGAGICASYQTQTNSPR-------QSIIAYTMSLGAENSVAYSNNS SARS-CoV-2 (del3) ECDIPIGAGICASYQTQTNSPRRARSVA---IIAYTMSLGAENSVAYSNNS SARS-CoV-2 (R685H) ECDIPIGAGICASYQTQTNSPRRAHSVASQSIIAYTMSLGAENSVAYSNNS