Reassessment of the complete genome sequence of BoDV-2. The complete genome sequence of BoDV-2 isolate No/98 is deposited in NCBI GenBank; however, it is suspected that this sequence might contain several errors in the L gene because three blocks of amino acid differences are unnaturally accumulated at the C-terminal domain (CTD) of L compared to those of representative strains of BoDV-1, such as He/80 and V (Fig. 1B) 6. Therefore, we reassessed the whole-genome sequence of BoDV-2 isolate No/98 via RNA sequencing (RNA-seq) using total RNA extracted from persistently infected Vero cells available at the Friedrich-Loeffler-Institut. Compared to the published sequence (accession no. AJ311524), the reassessed sequence contains one and five nucleotide differences in the P and L genes, respectively (Fig. 1A). Two of them in the L gene, uracil at positions 2762 and 3334, which are located in the polyribonucleotidyltransferase (PRNTase) domain (Fig. 1B), were completely substituted with guanine, resulting in amino acid changes of leucine at position 921 with arginine and cysteine at position 1112 with glycine (Fig. 1A). In addition, three SNPs were detected in the L gene, two of which, positions 3743 and 4250, induce nonsynonymous amino acid changes in the CTD of L, though they do not overlap with the blocks of amino acid differences described in a previous paper (Fig. 1A and B) 6. A nonsynonymous SNP was also detected at position 456 of the P gene (Fig. 1A). Furthermore, we performed rapid amplification of cDNA ends (RACE) analysis to determine the exact 3’ and 5’ terminal sequences, which could not be reconstituted by RNA-seq analysis. Similar to BoDV-1 18,22, BoDV-2 possesses four nucleotides overhung at the 3’ end of both the genome and antigenome without a complementary sequence at the 5’ end of the opposite strand (Fig. 1C). Compared to the published sequence, the reassessed sequence lacks uracil at the 5’ end of the genome (Fig. 1D).
The polymerase activity of BoDV-2 L is restored by substituting a nucleotide. In the reassessed sequence, nucleotides at positions 2762 and 3334 of the L gene are completely substituted compared to the reference sequence (Fig. 1A). To determine whether these differences affect BoDV-2 L gene expression, we constructed L cDNA expression plasmids possessing each nucleotide substitution independently (BoDV-2 LR or LG) or in combination (BoDV-2 LRG) and performed western blotting (Fig. 2A). In a previous study, we could not detect expression of FLAG-tagged BoDV-2 L with the reference sequence, though that of FLAG-tagged BoDV-1 L was clearly demonstrated, suggesting the low expression potential of the FLAG-BoDV2 L construct 23. However, in the present study, we could detect the expression of BoDV-2 L with the reference sequence when a linker sequence was inserted between FLAG and the L gene to enhance the accessibility of the antibody to the tag (Fig. 2B). In addition, we confirmed expressions of the reconstructed BoDV-2 L variants in transfected cells (Fig. 2B).
To determine the polymerase activities of the reconstructed BoDV-2 L variants, we performed a BoDV-2 minireplicon assay. As shown in Fig. 2C, reporter activity was not detected for BoDV-2 L with the reference sequence, indicating that it lacks any polymerase activity. Interestingly, BoDV-2 L possessing a nucleotide substitution at position 2762, namely, BoDV-2 LR, restored its polymerase activity, whereas a substitution at position 3334, namely, BoDV-2 LG, did not (Fig. 2C). Accordingly, BoDV-2 LRG, which possesses both nucleotide substitutions simultaneously, exhibited the same level of polymerase activity as BoDV-2 LR (Fig. 2C), indicating that a difference in the nucleotide at position 2762 is detrimental to the polymerase activity of BoDV-2 L.
To examine whether nucleotide differences at positions 2762 and 3334 of the L gene affect recovery of rBoDV-2 via reverse genetics, we constructed BoDV-2 cDNA plasmids expressing the full-length BoDV-2 antigenomes possessing each or both nucleotide substitutions. HEK293T cells were transfected with each BoDV-2 cDNA plasmid and five helper plasmids and cocultured with uninfected Vero cells at 3 days post transfection. After two weeks from coculture, rBoDV-2s that contain a nucleotide substitution at position 2762 of the L gene, rBoDV-2 LR and LRG, were successfully rescued (Fig. 2D). Interestingly, although nucleotide substitution at position 3334 did not affect the polymerase activity of BoDV-2 L in the minireplicon assay (Fig. 2C), rBoDV-2 LRG, which possesses both substitutions at positions 2762 and 3334, seemed to be recovered more efficiently than rBoDV-2 LR, which possesses a substitution at position 2762 alone (Fig. 2D), suggesting that the nucleotide at position 3334 of the L gene also plays a role in BoDV-2 replication. To examine the growth ability of rBoDV-2 LRG in comparison with that of the nonrecombinant BoDV-2 isolate No/98, Vero cells infected with rBoDV-2 LRG or isolate No/98 were cocultured with uninfected Vero cells at a ratio of 1 to 19 (Fig. 2E), and viral propagation and the copy number of viral RNA were evaluated every 3 or 4 days by immunofluorescence assay (IFA) and RT-qPCR, respectively. While the nonrecombinant isolate No/98 spread to almost all Vero cells within 2 weeks of coculture, infection of rBoDV-2 LRG exhibited limited spread within this period (Fig. 2F). Similarly, the copy number of viral RNA increased 20-fold in isolate No/98-infected cells, whereas it barely increased in rBoDV-2 LRG-infected cells (Fig. 2G). These results suggest that the growth ability of rBoDV-2 is not only determined by the L2762 and L3334 substitutions, even though both substitutions induce high polymerase activity in the minireplicon assay.
Nonsynonymous SNPs in the L gene facilitate the growth of BoDV-2. Our RNA-seq analysis identified two nonsynonymous SNPs in the CTD of the L gene (Fig. 1A), indicating that the Vero cells persistently infected with BoDV-2 isolate No/98 contain a viral L quasispecies. Therefore, we constructed L expression plasmids harboring each or both nucleotide substitutions at positions 3743 and 4250 based on BoDV-2 LRG and performed a minireplicon assay (Fig. 3A). While the substitution of uracil at position 3743 with cytosine, LRGT, increased polymerase activity by 1.3-fold, that of adenine at position 4250 with guanine, LRGR, decreased it by 0.7-fold. Substitution of both nucleotides, LRGTR, slightly decreased polymerase activity (Fig. 3B).
To further investigate the impact of these SNPs on the growth ability of BoDV-2, we generated rBoDV-2 possessing each or both substitutions in the L gene and evaluated their growth kinetics. Although the substitution at position 4250 of the L gene had a detrimental effect on polymerase activity in the minireplicon assay (Fig. 3B), L proteins possessing either SNP substitution remarkably facilitated viral propagation and RNA synthesis of rBoDV-2 (Fig. 3C and D). These findings suggest that these SNPs influence viral replication through mechanisms other than enhancing polymerase activity in cells.
A nonsynonymous SNP in the P gene expedites the growth of BoDV-2. We also identified another nonsynonymous SNP at position 456 of the P gene (Fig. 1A). The P gene encodes the viral polymerase cofactor, which directly interacts with the L protein and plays a crucial role in viral polymerase activity 24,25. Therefore, we examined the effects of this SNP on viral polymerase activity and viral growth ability using both a minireplicon assay and rBoDV-2 variants. Substitution of guanine at position 456 of the P gene with adenine, PI, resulted in a slight increase in the polymerase activity of LRG but not that of LRGTR, as observed in the minireplicon assay (Fig. 4A and B). Furthermore, the P gene with the introduced SNP significantly facilitated viral propagation (Fig. 4C and D) and RNA synthesis of rBoDV-2 (Fig. 4E and F). These observations indicate that the nonsynonymous SNP detected in the P gene also contributed to the increase in rBoDV-2 growth ability.
Lack of superinfection exclusion maintains the low-fitness polymorphism in cells persistently infected with BoDV-2. Since the isolation of BoDV-2 isolate No/98 from a pony brain in 1999, it has been maintained for a long time in persistently infected Vero cells. Notably, although BoDV-2-infected Vero cells have undergone repeated passages, L- and P-sequence variants with low viral replication ability appear to be the major quasispecies within persistently infected cells. The coexistence of viral quasispecies in infected cells has been shown to affect overall viral replication and pathogenicity and therefore plays a role in the adaptation and evolution of viral populations 26.
To determine the significance of the coexistence of BoDV-2 quasispecies in persistently infected cells, we performed infection experiments using cells infected with different sequences of rBoDV-2. First, we generated rBoDV-2 with different fluorescent marker genes, mCherry and GFP, inserted into the artificial intergenic region between the P and M genes (Fig. 5A) based on rBoDV-2 PILRGTR with high-growth ability. As shown in Fig. 5B, both viruses exhibited similar growth kinetics, confirming that the effects of these two fluorescence marker proteins on the growth ability of BoDV-2 are not different. We therefore examined the propagation of rBoDV-2 variants with different growth abilities, namely, low-growth rBoDV-2 LRG-mCherry (LRG-mCherry) and high-growth rBoDV-2 PILRGTR-GFP (PILRGTR-GFP), by a competition assay using Vero cells infected with each recombinant virus. When Vero cells infected with LRG-mCherry or PILRGTR-GFP were cocultured with uninfected Vero cells at a ratio of 1:1:23, PILRGTR-GFP rapidly spread throughout the culture, and the number of mCherry-expressing cells gradually decreased (Fig. 5C). Similarly, after 14 days of coculture, the genomic RNA of PILRGTR-GFP accounted for almost all the BoDV-2 genomic RNA in the population (Fig. 5D). However, at 14 days after coculture, almost 1.0% of the cells coexpressed mCherry and GFP, and the proportion of cells expressing mCherry alone or coexpressing mCherry and GFP was approximately 3.3% of the total (Fig. 5C), indicating that high-growth PILRGTR-GFP superinfected LRG-mCherry-infected cells and that low-growth rBoDV-2 was maintained without elimination from the cells superinfected with the PILRGTR-GFP.
To confirm the absence of superinfection exclusion between the rBoDV-2 variants, we next cocultured Vero cells infected with LRG-mCherry and PILRGTR-GFP at a 1:1 ratio. As a result, the percentage of cells coexpressing mCherry and GFP gradually increased; after 14 days of coculture, the proportions of cells expressing GFP alone, mCherry alone, and coexpressing mCherry and GFP were approximately 46.3%, 18.3%, and 32.4%, respectively (Fig. 5E). This finding suggests that the high-growth PILRGTR-GFP predominantly superinfected to the LRG-mCherry-infected cells. On the other hand, the proportion of viral genomic RNA of LRG-mCherry in the culture increased to some extent (Fig. 5F), indicating that the resident viruses were not eliminated even among the cells superinfected with the high-growth virus; rather, the replication of the low-growth viral genome was upregulated by support from the polymerase derived from high-growth virus. In addition, we cloned Vero cells coinfected with both LRG-mCherry and PILRGTR-GFP (Fig. 5G) and cocultured them with uninfected Vero cells to investigate whether superinfection affects the propagation abilities of coinfected variants. As shown in Fig. 5H, the proportion of cells infected with only the PILRGTR-GFP virus increased; the LRG-mCherry virus could also be transmitted to uninfected Vero cells, albeit only slightly. These observations suggest that BoDV-2 does not exclude superinfection, allowing low-fitness variants to persist within infected cells while maintaining their characteristics and thereby maintaining population diversity.