To identify a virus latently infecting a hemp (Cannabis sativa) plant, we analyzed transcriptome data originally collected to study the major molecular processes underlying secondary growth and bast fiber (Behr et al., 2016; Guerriero et al., 2017; Behr et al., 2019). The plants were grown in laboratory-controlled conditions and showed no visible viral disease symptoms. The transcriptome data are available in the Sequence Read Archive (SRA) of the National Center for Biotechnology Information (NCBI) under BioProject accession number PRJNA435671.

To construct the extended genome sequence of Melampyrum roseum virus 2 (MelRoV2), we analyzed transcriptome data from Melampyrum roseum (SRA accession numbers DRR082664 and DRR082665) (Kado and Innan, 2018; Bejerman and Debat, 2022).

To collect plant transcriptome data potentially harboring novel tepovirus-like virus genome sequences, we utilized the Serratus Explorer ¹ (Edgar et al., 2022). We selected five known tepoviruses, including potato virus T (PVT), prunus virus T (PrVT), Zostera virus T (ZoVT), Trichosanthes virus A (TrVA), and Ficus tepovirus A (FiVT), as target GenBank records. We chose matches with alignment identities ranging from 60% to 90% and scores between 50 and 100. The resulting plant transcriptome datasets were further filtered to include data with an average sequence length of 100 nt or longer, paired-end layout, and Illumina platform. This process resulted in 192 plant transcriptome datasets being selected.

Raw plant transcriptome sequences were subjected to quality trimming using Sickle (version 1.33) ² with parameters “-q 30 -l 55.” The high-quality reads were then assembled into contigs using SPAdes (version 3.15.5) ³ with the “rnaviral” mode (Prjibelski et al., 2020). BLASTX was used to compare contigs with known viral proteins.

Open reading frames (ORFs) in a putative virus genome contig were predicted using ORFfinder. ⁴ Genome contigs containing complete or nearly complete Rep, MP, and CP ORFs, and showing a minimum amino acid identity of 40% with a known tepovirus Rep protein, were retained. When two or more contigs shared 90% or greater amino acid identity in the Rep proteins, only one contig was retained, and the others were discarded.

Homologous viral proteins were retrieved from the NCBI protein database via the BLAST web server. We used the MAFFT online service ⁵ to generate multiple sequence alignments under default conditions (Katoh et al., 2019). Neighbor-joining phylogenetic tree construction and bootstrap value calculation were also performed using the MAFFT online server. The resulting phylogenetic tree was visualized using MEGA (version 11.0.13) ⁶ (Kumar et al., 2018). Visualization of multiple sequence alignments was prepared using ESPript (version 3.0) ⁷ (Robert and Gouet, 2014).

Transcriptome data generated from hypocotyl tissues of young hemp plants and bast fibers isolated from adult plants were assembled into contigs (Behr et al., 2016; Guerriero et al., 2017; Behr et al., 2019). Sequence comparisons of the transcript contigs with RdRp sequences from known RNA viruses revealed contigs potentially originating from viral genomes (see Supplementary Data S1 for more detailed information). We identified a contig containing ORFs that showed significant sequence similarity to those of known members of the genus Tepovirus (subfamily Trivirinae, family Betaflexiviridae). We tentatively named this virus hemp virus T (HemVT). Its genome sequence has been deposited in the NCBI GenBank under accession number OR346818.

The HemVT genome contains four ORFs, three of which encode Rep, MP, and CP, as is typical of members of the genus Tepovirus. Notably, the fourth ORF did not exhibit amino acid sequence similarity to any known tepovirus ORFs. Instead, it showed approximately 48% sequence identity to NABP-like proteins of GymRhV1 and MelRoV2, which are members of the genus Divavirus (family Betaflexiviridae). The HemVT NABP also displayed marginal sequence identities (15%–20%) to NABP proteins of other Betaflexiviridae viruses, including those from the genera Capillovirus, Carlavirus, Citrivirus, Prunevirus, Trichovirus, and Vitivirus. Therefore, we concluded that the fourth ORF encodes a NABP-like protein, making HemVT the first member of the genus Tepovirus with a NABP homolog.

Following the discovery of the HemVT genome sequence, we hypothesized that additional tepovirus genome sequences with a NABP homolog might exist in plant transcriptome data available in the NCBI SRA. To identify potential tepovirus-like genome sequences, we filtered the SRA datasets using the Serratus Explorer (Edgar et al., 2022). A total of 192 SRA datasets containing reads matching known tepovirus RdRp sequences were downloaded and assembled into contigs. We then selected contigs that contained complete or nearly complete ORFs with significant sequence identities to previously known tepovirus proteins. As a result, we identified 20 additional distinct tepovirus-like genome sequences (see Supplementary Data S2 for more detailed information). These were named according to their host plants: Acanthus hungaricus virus 1 (AcaHuV1), Allium listera virus 1 (AllLiV1), Balanophora indica virus 1 (BalInV1), Capparis spinosa virus 1 (CapSpV1), Cistanche deserticola virus 1 (CisDeV1), Crocus sativus virus 1 (CroSaV1), Davidia involucrata virus 3 (DavInV3), Ferula gummosa virus 1 (FerGuV1), Hylocereus undatus virus 1 (HylUnV1), Lilium lancifolium virus 1 (LilLaV1), Lilium pumilum virus 1 (LilPuV1), Maihueniopsis conoidea virus 1 (MaiCoV1), Panicum virgatum virus 1 (PanViV1), Pogostemon cablin virus 1 (PogCaV1), Pogostemon cablin virus 2 (PogCaV2), Solanum melongena virus 1 (SolMeV1), Vallisneria spiralis virus 1 (ValSpV1), Vallisneria spiralis virus 2 (ValSpV2), Vallisneria spiralis virus 3 (ValSpV3), and Vanilla shenzhenica virus 1 (VanShV1). The genome sequences have been deposited in the NCBI GenBank under accession numbers BK068543–BK068562 (see Supplementary Data S3, S4 for genome and protein sequences, respectively). A summary of all viruses newly identified in this study is presented in Table 1, and their genomic structures are depicted in Figure 1.

Upon examining the genome organization of the newly identified viruses, we found that, in addition to HemVT, four more viruses—CisDeV1, FerGuV1, MaiCoV1, and SolMeV1—contained a NABP-like ORF in their genomes. The other 16 viruses (AcaHuV1, AllLiV1, BalInV1, CapSpV1, CroSaV1, DavInV3, HylUnV1, LilLaV1, LilPuV1, PogCaV1, PogCaV2, VanShV1, PanViV1, ValSpV1, ValSpV2, and ValSpV3) did not contain a NABP ORF. In the case of AcaHuV1, the presence of the fourth ORF could not be determined because its genome sequence was truncated in the middle of the CP ORF.

To establish the phylogenetic relationships of the newly identified viruses and known Tepovirus species, a phylogenetic tree was constructed (Figure 2). We collected the Rep protein sequences of all nine known tepoviruses and representative viruses from other genera in the family Betaflexiviridae. A multiple sequence alignment was generated and a phylogenetic tree was inferred using the MAFFT online service (Katoh et al., 2019). Among the 21 newly identified viruses, 17 (AcaHuV1, AllLiV1, BalInV1, CapSpV1, CisDeV1, CroSaV1, DavInV3, FerGuV1, HemVT, HylUnV1, LilLaV1, LilPuV1, MaiCoV1, PogCaV1, PogCaV2, SolMeV1, and VanShV1) formed a clade with the known tepoviruses, suggesting that they belong to the genus Tepovirus. However, this clade was weakly supported, with a bootstrap value of 36, consistent with previous findings (Goh et al., 2021).

All five viruses containing a NABP homolog (CisDeV1, FerGuV1, HemVT, MaiCoV1, and SolMeV1) were positioned within the Tepovirus clade, confirming the discovery of tepoviruses with NABP-like ORFs. However, the other 12 newly identified tepoviruses lacked the fourth ORF, indicating that most tepoviruses do not possess a NABP-like ORF.

The remaining four newly identified viruses (PanViV1, ValSpV1, ValSpV2, and ValSpV3) formed a distinct subclade with strong bootstrap support (99). This clade is closely related to the genus Vitivirus but does not contain a NABP-like ORF, whereas members of Vitivirus are known to possess NABPs. The absence of a NABP-like ORF and the formation of a distinct clade suggest that these four viruses may represent the founding members of a novel genus closely related to Vitivirus.

Next, we performed sequence comparison and phylogenetic analysis of the five newly discovered tepovirus NABP homologs (Figure 3). First, we searched the NCBI protein database using the five tepovirus NABP homologs as queries. The BLASTP search used an E-value threshold of 1e−5, and 28 known NABP proteins showing sequence similarities to the tepovirus NABP-like proteins were retrieved.

A multiple sequence alignment of the five tepovirus NABP-like proteins and the 28 known NABPs revealed two distinct groups (Figure 3A). The first group included HemVT, GymRhV1, and MelRoV2. The HemVT NABP-like protein shared approximately 48% identity with the NABPs of GymRhV1 and MelRoV2, both members of the genus Divavirus. The second group consisted of the remaining viruses, including four tepoviruses (CisDeV1, FerGuV1, MaiCoV1, and SolMeV1), and 26 known viruses. The four tepovirus NABP-like proteins exhibited 12%–36% identity with previously known NABP proteins. Among the 26 known viruses, 18 belonged to six genera (Capillovirus, Carlavirus, Citrivirus, Prunevirus, Trichovirus, and Vitivirus) within Betaflexiviridae. Notably, eight of the 26 viruses were from two genera (Allexivirus and Potexvirus) in the family Alphaflexiviridae.

The phylogenetic tree inferred from the multiple alignment of NABP homolog sequences confirmed that the HemVT NABP homolog shares ancestry with those of GymRhV1 and MelRoV2 (Figure 3B). The strong bootstrap support (100) for this clade, along with their high sequence similarity, suggests that HemVT, GymRhV1, and MelRoV2 recently obtained their NABP-like ORFs from closely related sources. In the case of GymRhV1 and MelRoV2, it is more plausible that the NABP-like ORF was acquired in their common ancestor before their divergence, as both their Rep and NABP-like proteins show high sequence similarity.

The phylogenetic tree also showed that the NABP-like proteins of CisDeV1, FerGuV1, MaiCoV1, and SolMeV1 share ancestry with those from other genera in Betaflexiviridae and Alphaflexiviridae. However, the exact phylogenetic relationships remain unclear due to low bootstrap support values for the subclades containing them. This ambiguous relationship suggests that these viruses may have acquired their NABP-like ORFs from unrelated sources. This explanation is further supported by the discordance between the Rep and NABP phylogenetic trees. For example, in the Rep tree, MaiCoV1, SolMeV1, and HemVT form a strongly supported subclade (bootstrap value of 100), with HemVT being the closest relative to SolMeV1. However, in the NABP tree, MaiCoV1 and SolMeV1 are distantly placed, and HemVT possesses an NABP-like protein that is distinct from those found in other related viruses. Therefore, it is highly likely that these viruses independently obtained their NABP-like ORFs from unrelated sources.