In this study, we analyzed RNA-seq data originally collected from four tissue samples (flower, bud, leaf, and root) of a three-year-old A. carmichaelii plant with the aim of identifying key genes involved in the biosynthesis of diterpene alkaloids (Rai et al., 2017). The plant, which was cultivated under natural conditions at a research center in Tsukuba, Japan, did not show any symptoms of viral disease. The RNA-seq data are available in the Sequence Read Archive (SRA) of the National Center for Biotechnology Information (NCBI) under the accession numbers SRR6225420–SRR6225423. High-quality A. carmichaelii RNA-seq data were obtained by filtering raw data using sickle (version 1.33 ² ) with the parameter “-q 30 -l 55.” Transcriptome contigs were generated by assembling high-quality reads using the SPAdes Genome Assembler (version 3.15.4 ³ ) with the parameter “--rnaviral” (Bushmanova et al., 2019).

Putative viral genome contigs were initially identified by comparing A. carmichaelii transcriptome contigs with known viral RdRp domain sequences extracted from the Pfam database (release 35.0 ⁴ ) using DIAMOND (version v2.0.14.152 ⁵ ) (Buchfink et al., 2021). Accession numbers of the Pfam RdRp families used for preparation of known viral RdRp domains are PF00602, PF00603, PF00604, PF00680, PF00946, PF00972, PF00978, PF00998, PF02123, PF03431, PF04196, PF04197, PF05273, PF05788, PF05919, PF06317, PF06478, PF07925, PF08467, and PF12426. A polyprotein ORF of a putative potyvirus genome contig was predicted using the NCBI ORFfinder ⁶ . Protease cleavage sites of a polyprotein were identified by a sequence comparison with polyproteins of closely related viruses. Pairwise sequence identities were calculated using FASTA (version 36.3.8h ⁷ ). Sequence logo representations were prepared using WebLogo (version 3 ⁸ ) (Schneider and Stephens, 1990; Crooks et al., 2004).

High-quality reads were mapped to potyviral genome contig sequences using BWA (version 0.7.17-r1194-dirty ⁹ ) with the BWA-MEM algorithm (Li and Durbin, 2009). RNA-seq reads that span a putative RNA polymerase slippage site in the P3 coding region were extracted, and the frequency of insertions or deletions of A residues was determined.

MAFFT (version 7.475 ¹⁰ ), with the parameter “--auto,” was utilized for creating a multiple alignment of potyviral polyprotein sequences (Nakamura et al., 2018). Information-rich segments from the multiple sequence alignment were extracted using trimAl (version 1.4.rev22 ¹¹ ) with the parameter “-automated1” (Capella-Gutierrez et al., 2009). The maximum likelihood phylogenetic tree was constructed using IQ-TREE (version 2.2.0 ¹² ) with the “-B 1000” parameter (Minh et al., 2020). The resulting phylogenetic tree was visualized using MEGA (version 11.0.13 ¹³ ) (Kumar et al., 2018).

RNA-seq reads previously generated from four tissue types (flower, bud, leaf, and root) of a three-year-old A. carmichaelii plant were assembled into contigs (Rai et al., 2017). Two of these contigs, with lengths of 9,492 and 9,441 bp respectively, exhibited significant sequence similarities to RdRp sequences of known potyviruses. They contained a full-length coding region for potyviral polyproteins, although the completeness of the genomes should be verified through a molecular method such as rapid amplification of cDNA ends. The predicted polyprotein ORFs had lengths of 9,210 and 9,165 bp, encoding 3,069 and 3,054 aa polyproteins, respectively.

BLAST searches of the NCBI protein database revealed that the two putative potyvirus genome contigs identified in the A. carmichaelii transcriptome were most similar to tulip breaking virus (TBV), a member of the genus Potyvirus, with approximately 56% amino acid sequence identity of polyprotein sequences (Wylie et al., 2019). The species demarcation criteria for potyviruses are generally accepted as <76% nucleotide identity of the large ORF and <82% amino acid identity of the polyprotein (Adams et al., 2005b; Inoue-Nagata et al., 2022). Therefore, the two newly identified viruses can be considered novel and distinct from previously known potyviruses. The second most closely related known potyvirus was lily mottle virus (LMoV), with around 53% amino acid sequence identity of polyprotein sequences (Zheng et al., 2003).

The polyprotein ORFs of the two newly identified potyvirus genome contigs exhibited a nucleotide sequence identity of 66%, while the amino acid sequence identity of polyprotein sequences was 69%. According to the aforementioned demarcation criteria, the two genome sequences were considered to originate from distinct potyviruses and were tentatively named Aconitum potyvirus 1 (AcoPV1, 9,492 bp) and Aconitum potyvirus 2 (AcoPV2, 9,441 bp). The AcoPV1 and AcoPV2 sequences were deposited in the NCBI GenBank under accession numbers OP271473 and OP271474, respectively.

The large polyprotein of a potyvirus is processed into ten mature peptides (P1, HC-Pro, P3, 6K1, CI, 6K2, VPg, NIa-Pro, NIb, and CP) by the three potyvirus proteases (P1, HC-Pro, and NIa-Pro) (Adams et al., 2005a; Goh and Hahn, 2021; Pasin et al., 2022). Cleavage sites and boundaries of mature peptides were predicted in the AcoPV1 and AcoPV2 polyproteins based on sequence similarities with mature proteins of closely related potyviruses, including TBV and LMoV (Figure 1). The polyprotein sequences of AcoPV1 and AcoPV2 surrounding the cleavage sites matched previously deduced consensus sequences for the potyvirus protease recognition sites (Adams et al., 2005a; Goh and Hahn, 2021).

All members of the family Potyviridae are known to have a secondary ORF, termed PIPO, which is situated in the P3 coding region, and is in frame −1 relative to the P3 (Yang et al., 2021; Inoue-Nagata et al., 2022). The PIPO ORF is expressed into a trans-frame fusion protein, known as P3N-PIPO, due to the insertion of an A residue at the slippery GA₆ motif during replication (Chung et al., 2008). In the genomes of AcoPV1 and AcoPV2, the GA₆ motifs were located at positions 2,802–2,808 and 2,782–2,788, respectively. The PIPO ORFs of AcoPV1 and AcoPV2 encode 59 and 82 amino acids, respectively. No additional GA₆ motif other than that in the P3 coding region was found in AcoPV1 and AcoPV2 sequences.

The GA₆ motif within the P3 coding region of potyviruses induces rare RNA polymerase slippages during replication, resulting in the crucial insertion of an A residue for P3N-PIPO protein production (Olspert et al., 2015; Rodamilans et al., 2015; Hagiwara-Komoda et al., 2016; Olspert et al., 2016; Untiveros et al., 2016). The RNA polymerase slippage rate can be estimated by counting RNA-seq reads originating from the slippage products in the GA₆ motif. To detect and quantify the slippage products of the AcoPV1 and AcoPV2, RNA-seq reads spanning the GA₆ motifs were extracted and examined (Figure 2).

A total of 509 and 1,084 RNA-seq reads were mapped to the GA₆ motif in AcoPV1 and AcoPV2, respectively. Among these, 505 (99.21%) for AcoPV1 and 1,069 (98.62%) for AcoPV2 matched the original unmodified (wild type) sequence. There was one read for AcoPV1 (0.20%) and seven reads for AcoPV2 (0.65%) that potentially originated from the single-A insertion or +1A product, which is essential for P3N-PIPO protein production.

Interestingly, we also observed RNA-seq reads corresponding to other insertion or deletion products. One of these slippage products, the single-A deletion or –1A product, might lead to the production of another trans-frame fusion protein, known as P3N-ALT, as previously reported in clover yellow vein virus (ClYVV) (Hagiwara-Komoda et al., 2016). In AcoPV1, one read (0.20%) was identified as the –1A product, potentially encoding P3N-ALT, if present. The –1A product in AcoPV1 results in the addition of 12 amino acids after the GA₆ motif. In the case of AcoPV2, four reads (0.37%) were identified as the –1A product, which could add five amino acids after the GA₆ motif. Additionally, we also observed other slippage products, including double-A deletion (–2A) or double-A insertion (+2A) products. These slippage products may be translated to produce P3N-ALT with a single amino acid deletion or P3N-PIPO with a single amino acid insertion, respectively.

As a result, we estimated the rate of RNA polymerase slippage, potentially leading to the production of P3N-PIPO or P3N-ALT, to be 0.79% in AcoPV1 and 1.38% in AcoPV2. It is worth noting that the number of RNA-seq reads originating from slippage products is too low to accurately measure their RNA polymerase slippage rates.

To determine the phylogenetic positions of AcoPV1 and AcoPV2 among known potyviruses, we retrieved polyprotein sequences of known potyviruses from the NCBI protein database. The pairwise identities of polyprotein sequences of AcoPV1, AcoPV2, and other potyviruses are presented in Supplementary Table S1. A maximum likelihood phylogenetic tree was inferred from a multiple sequence alignment that included AcoPV1, AcoPV2, and known polyprotein sequences (Figure 3).

Both AcoPV1 and AcoPV2 cluster together into a subclade, supported by a 100% bootstrap value, distinct from the known potyviruses. The clade that includes TBV and LMoV is the most closely related known potyvirus group to AcoPV1 and AcoPV2. Together, the AcoPV1/AcoPV2 and TBV/LMoV subclades form a single clade, supported by a 99% bootstrap value (green box in Figure 3).

The next most closely related clade comprises nine potyviruses: leek yellow stripe virus, Platycodon mild mottle virus, bean yellow mosaic virus, ClYVV, Mediterranean ruda virus, iris potyvirus A, lily yellow mosaic virus, Paris virus 1, and Thunberg fritillary mosaic virus. AcoPV1, AcoPV2, TBV, LMoV, and these nine potyviruses form a robust clade within the genus Potyvirus, supported by a 100% bootstrap value (blue box in Figure 3).

Potyviruses are primarily transmitted by aphids, which feed on sap from leaves and other easily accessible parts of host plants (Gadhave et al., 2020; Xia et al., 2023). To assess the distribution and abundance of AcoPV1 and AcoPV2 within A. carmichaelii tissue samples, we counted all mapped RNA-seq reads to the viral genome sequences (Table 1).

In flower and leaf tissues, the viral RNA-seq reads of AcoPV1 and AcoPV2 were notably higher compared to bud and root tissues. In the flower tissue, AcoPV1 and AcoPV2 viral reads accounted for 0.22% and 0.88%, respectively, of the total RNA-seq reads. In leaf tissue, viral reads corresponding to the AcoPV1 and AcoPV2 genomes comprised 0.70% and 0.36%, respectively. For bud tissue, we observed lower quantities, with 0.02% for AcoPV1 and 0.18% for AcoPV2. In root tissue, both viruses showed the lowest counts among tissues studied (less than 0.01%).

It is notable that the A. carmichaelii plant analyzed in this study did not display any viral disease symptoms. This suggests that viral infection may be at an initial stage, and their distribution may be limited to the original infection site or nearby tissues. Consequently, the higher abundance of AcoPV1 and AcoPV2 RNA-seq reads in leaf and flower tissues may reflect the feeding preferences of aphids, which are likely their primary vectors.