We collected fecal samples from bats belonging to the R. cornutus and R. ferrumequinum species in Niigata, Chiba, and Shizuoka Prefectures (Appendix Figure 1, panel A). Using real-time reverse transcription PCR, we successfully detected the envelope gene sequence of sarbecovirus in 1 or 2 R. cornutus bat samples in each prefecture (Table 1). In contrast, all R. ferremuquinum bat samples were negative. These data suggested that bat sarbecoviruses are distributed among R. cornutus bats at various locations in Japan.

In our previous study, we showed that a vesicular stomatitis virus‒based pseudotyped virus possessing the S protein of Rc-o319 sarbecovirus from R. cornutus only infected RcACE2-expressing cells, but not hACE2-expressing or other Rhinolophus ACE2–expressing cells (7). Therefore, to isolate bat sarbecoviruses, we established RcACE2-stably expressing cells (Vero-RcACE2) based on Vero/TMPRSS2 cells. Using Vero-RcACE2 cells, we successfully isolated bat sarbecoviruses, which exhibited extensive cytopathic effect with syncytium formation (Appendix Figure 1, panel B) from real-time reverse transcription PCR‒positive fecal samples from each prefecture. We designated the Niigata isolate as Rc-os20, the Chiba isolate as Rc-mk2, and the Shizuoka isolate as Rc-kw8. We further isolated the cultivable Rc-o319 strain by using Vero-RcACE2 cells.

We determined the full-genome sequence of all isolates by using next-generation sequencing and deposited the sequences in GenBank (accession nos. LC663958, LC663959, and LC663793). We found that sequence homologies were high (range 94.8%–96.8%) among all isolates from Japan (Table 2), However, Rc-mk2 and Rc-os20 lacked the entire open reading frame 8 coding region.

Figure 1

Figure 1. Phylogenetic tree of sarbecoviruses from bats in Japan, generated by using the full-genome nucleotide sequences with the maximum-likelihood analysis combined with 500 bootstrap replicates. Red indicates strains isolated in this...

We also performed similarity plot analysis of entire genome sequence by using each isolate as a query, which indicated that similarities among isolates were high throughout the entire genome sequence, except for coding regions of the N-terminal domain (NTD) and receptor-binding domain (RBD) of the S gene, although NTDs of Rc-o319 and Rc-kw8 were conserved (Appendix Figure 2). No clear recombination among the isolates were observed as analyzed by RDP5 software (12). Phylogenetic analysis showed that the isolates from Japan formed a single genetic cluster and positioned in a clade containing SARS-CoV-2‒related sarbecoviruses, which might be designated the Japanese clade of bat sarbecoviruses (Figure 1).

We aligned the receptor-binding motif of the S protein of isolates from Japan with that of other sarbecoviruses (Appendix Figure 3, panel A). We observed that all isolates had a 9-aa deletion in this motif, as previously observed in Rc-o319, and had relatively conserved residues with Rc-o319. In addition, phylogenetic tree analysis of RBD showed that strains from Japan were included in the clade of viruses that use ACE2 orthologs as a strain receptor (Appendix Figure 3, panel B). Therefore, we assumed that these new strains from Japan use RcACE2 as a receptor.

Figure 2

Figure 2. Growth kinetics of sarbecovirus isolates from bats in Japan. Rhinolophus cornutusbat isolates Rc-o319 (A), Rc-os20 (B), Rc-mk2 (C), and Rc-kw8 (D) or SARS-CoV-2 (B.1.1.7) € were inoculated into...

To test this hypothesis, we compared the replication of isolates from Japan with that of a control SARS-CoV-2 (B.1.1.7, Alpha variant) in Vero-RcACE2, Vero-hACE2, Vero-ACE2KO, and Vero/TMPRSS2 cells. Whereas the 4 bat isolates replicated well in Vero-RcACE2 only, they did not replicate in Vero/TMPRSS2, Vero-hACE2, or Vero-ACE2KO cells, suggesting their RcACE2-dependent infectivity. In contrast, we observed that SARS-CoV-2 replicated efficiently in Vero/TMPRSS2, Vero-RcACE2, and Vero-hACE2 cells, but not in Vero-ACE2KO cells (Figure 2), suggesting multiple ACE2-dependent infectivity, including that of R. cornutus bats. These data suggested that at isolates from Japan use only bRcACE2 as a receptor, showing narrow host specificity.

We isolated bat sarbecoviruses from R. cornutus bats in several locations in Japan that were phylogenetically positioned in the same cluster of the SARS-CoV-2–related viruses. These isolates used only bat ACE2 as a receptor and did not replicate in hACE2-expressing cells, forming a unique type, and suggesting a low potential for human infection.

To our knowledge, this type of bat sarbecoviruses has not been previously isolated (13) because African green monkey Vero cells having highly similar ACE2 to hACE2 were used for viral isolation attempts in the previous studies (4,5). Cultivable bat sarbecoviruses provide a useful and powerful tool to determine their characteristics, such as receptor specificity and pathogenicity in animals, leading to elucidation of spillover potential.

Rhinolophus spp. bats are relatively short-distance migrants (14) and lack frequent cross-contact between bat groups, explaining why most genome sequences were highly conserved among strains from Japan. Exceptions were the RBD-coding and NTD-coding regions of the S gene, which show high variation caused by immune pressure (15), suggesting that they diverged relatively recently from the undefined ancestral virus. Because sarbecoviruses might mutate and infect humans by intermediate hosts in wildlife or livestock, epidemiologic studies of sarbecoviruses in wildlife, including bats, need to be conducted on a long-term basis for risk assessment of their zoonotic potential.