Identification and characterization of an endogenous pararetrovirus in Gentiana triflora plants Authors: Masamichi Isogai, Shunsuke Tezuka, Akira Abe, Yutaro Takamura, Hajime Yaegashi Published: 2026 ABSTRACT AbstractDeep sequencing and de novo assembly of small RNAs (sRNA analysis) provide a powerful approach to detect plant RNA and DNA viruses. Here, using sRNA analysis of Gentiana triflora (gentian) plants affected with kobu-sho disease, a serious disease of gentian, we found caulimovirus-like sequences, provisionally designated as gentian caulimovirus-like sequence (GCVS). Analysis of the complete nucleotide sequence indicated that GCVS is a novel virus with a genomic structure and organization typical of members of genus Caulimovirus (family Caulimoviridae). Based on Southern blot hybridization and rolling circle amplification analyses, GCVS seems to be integrated into the genomes of a wild gentian plant and three F₁ gentian cultivars and their respective parents and without any episomal viral forms in plants. Consistently, no episomal virus was detected in gentian plants with kobu-sho disease. To confirm GCVS integration in the gentian genome, we used a BLASTn analysis of the gentian draft genome (4,374 contigs; total: 3,657,985,820 nucleotides [nt]) using the GCVS sequence as a query and identified 1,929 GCVS-derived fragments (total: 3,116,680 nt) across 280 contigs. All fragments were shorter than the full-length GCVS genome (8,125 nt), and more than 90% were less than 4,000 nt long. Open reading frames in these fragments encoded proteins containing numerous amino acid substitutions and were frequently disrupted by premature stop codons. Collectively, these features indicate that GCVS represents a non-infectious endogenous pararetrovirus, defined as a pararetroviral sequence that has been integrated into the host genome and is no longer capable of autonomous replication or infection.Supplementary InformationThe online version contains supplementary material available at 10.1007/s00705-026-06654-w. Introduction Kobu-sho, a severe disease of Gentiana triflora (gentian) plants characterized by shortened internodes with gall formation on the stem and chlorosis at the shoot apex, markedly reduce the market value of the plants [23]. In 2013, gentian kobu-sho-associated virus (GKaV) was detected in gentian plants affected by kobu-sho disease (hereafter referred to as kobu-sho gentian plants) during a sequence analysis of high-molecular-weight double-stranded RNAs (dsRNAs) extracted from the diseased plants [25]. Although GKaV has not yet been recognized by the International Committee on Taxonomy of Viruses, phylogenetic analyses of the putative RNA-dependent RNA polymerase and RNA helicase encoded by the GKaV genome indicate similarity to members of the Flaviviridae family. In addition, a genomic region of GKaV has been suggested to be involved in gall formation in gentian plants [5]. However, GKaV has not yet been confirmed as the causal agent of kobu-sho disease because back-inoculation of the virus into gentian plants has not been successful. Moreover, viruses such as members of the Caulimoviridae family (plant pararetroviruses) that do not produce high-molecular-weight dsRNAs in infected cells cannot be detected by sequence analysis of dsRNA. Therefore, a comprehensive virome analysis of kobu-sho gentian plants using additional approaches is required. As an alternative to analyzing the virome in kobu-sho gentian plants, small RNAs (sRNAs) can be used to identify plant DNA and RNA viruses, viroids, and previously unknown viral agents [27, 42]. Plants use sRNAs to induce sequence-specific gene silencing, which is central to antiviral defense in plants [11]. dsRNAs or hairpin RNAs are processed by Dicer-like proteins into 21–24 nucleotide (nt) sRNAs, which are subsequently loaded onto Argonaute proteins to form the RNA-induced silencing complex (RISC). RISCs use these sRNAs as guides to degrade viral RNAs derived from both RNA and DNA viruses through post-transcriptional gene silencing (PTGS) [16, 53] and also to target DNA viruses via RNA-directed DNA methylation, resulting in transcriptional gene silencing (TGS) [31, 56]. Consequently, high-throughput sequencing and bioinformatic analyses of plant sRNA populations enable the identification of plant-infecting viruses, a strategy generally referred to as sRNA analysis. When pararetroviruses are detected in plants by sRNA analysis, it is necessary to determine whether they exist as episomal forms and/or as endogenous pararetroviruses (EPRVs) integrated into the plant genome because, although pararetroviruses lack an integrase gene and do not require integration into the host genome as part of their replication cycle, they have been reported to be integrated into the genomes of numerous monocotyledonous and dicotyledonous vascular plants, and ferns [8, 10, 12, 18, 19, 37, 46, 55]. Most EPRVs are replication-defective and occur as fragmented sequences of various sizes, ranging from short segments to nearly complete viral genomes [2]. In some interspecific hybrids of plants belonging to the genera Musa, Nicotiana, and Petunia, however, some EPRVs become reactivated, giving rise to episomal viruses—such as banana streak virus (genus Badnavirus), tobacco vein clearing virus (genus Solendovirus), and petunia vein clearing virus (genus Petuvirus)—and cause disease in their hosts [32, 38, 44]. Plant pararetroviruses are currently classified into 11 genera in the family Caulimoviridae. Members of the genus Caulimovirus (caulimoviruses) have a circular dsDNA genome of approximately 7.1–9.8 kilobase pairs (kbp) containing six or seven open reading frames (ORFs) [17]. ORFs 1–6, which are conserved among all caulimoviruses, respectively encode the movement protein (MP), aphid transmission factor (ATF), DNA-binding protein, coat protein (CP), polymerase polyprotein (comprising protease, reverse transcriptase, and ribonuclease H), and transactivator/viroplasmin (Tav). ORF7 has not been detected in infected plants, and its function remains unknown. In this study, sRNA analysis of kobu-sho gentian plants revealed the presence of caulimovirus sequences, tentatively designated as the gentian caulimovirus sequence (GCVS). To date, neither caulimoviruses infecting gentian plants nor EPRVs integrated into the gentian genome have been reported. Therefore, we determined the taxonomic classification of GCVS and investigated whether GCVS is derived from an episomal virus and/or from an EPRV and its possible relevance to kobu-sho disease. Materials and methods Sample materials In June 2022, plants of gentian cultivar (cv.) Light-Pink-Ashiro with kobu-sho symptoms were collected from a commercial field in Hachimantai City, Iwate Prefecture, Japan. The plants were in their third year after transplantation to the field in 2020, and kobu-sho symptoms had been observed in the field since 2021. In addition, one plant each of wild gentian and three F₁ cultivars of Light-Pink-Ashiro, Ashiro-no-Natsu, and Ashiro-no-Aki —with their respective maternal and paternal parents (Ei-P and INP; AZ1-Early-1 and NE2N-No. 1; and N27-No. 29 and Y8-No. 2, respectively), were obtained from the Hachimantai City Floricultural R&D Center. Nicotiana tabacum plants were grown in pots in a growth chamber at 25 °C in continuous light and used as control plants. Blueberry red ringspot virus (BRRV), a member of the genus Soymovirus in the Caulimoviridae family (soymovirus), was used as an episomal virus. BRRV-infected and -uninfected highbush blueberry (Vaccinium corymbosum) trees grown in an orchard at the Iwate University Research Farm were used in this study [22]. sRNA sequencing Total RNA was extracted from each gentian plant using TRI Reagent solution (Merck KGaA, Darmstadt, Germany) according to the instructions. RNA quantity and quality were assessed using an Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA, USA), a NanoDrop spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA), and 1% agarose gel electrophoresis. A total of 2 µg of RNA, with an RNA integrity number of 7.1 was used to prepare a next-generation sequencing library. An sRNA library was constructed using the NEBNext Multiplex Small RNA Library Prep Set for Illumina (New England Biolabs, Ipswich, MA USA) and the manufacturer’s protocol. For removing technical sequences, including adapter and PCR primer sequences and low-quality bases (Phred score < 20), pass-filter FASTQ files were processed using Cutadapt version 1.9.1 [34], yielding high-quality clean reads. Data analysis Reads of 18–30 nucleotides (nt) obtained from deep sequencing of sRNAs were assembled into contiguous sequences (contigs) using Velvet version 1.2.10 [57] with a k-mer size of 15. The resulting contigs were used for BLASTx and BLASTn searches of the NCBI GenBank database for viruses (as of January 2019) to identify plant viruses present in kobu-sho gentian plants. Cloning and sequencing of GCVS GCVS amplicons were cloned using a TOPO TA Cloning Kit (Invitrogen, Carlsbad, CA, USA). The inserts were sequenced using the BigDye Terminator v1.1 Cycle Sequencing Kit (Applied Biosystems, Foster City, CA, USA) with either M13 forward primer (5′-TGTAAAACGACGGCCAGT-3′) and M13 reverse primer (5′-GAGCGGATAACAATTTCACACAGG-3′) or sequence-specific primers designed based on newly obtained sequence data. Phylogenetic analysis Proteins encoded by GCVS ORFs 1 and 3–6 were phylogenetically analyzed using MEGA12 [28]. Amino acid sequences were aligned using ClustalW, and a phylogenetic tree was constructed using the maximum likelihood method with the best-fit protein substitution model implemented in MEGA12. Polymerase chain reaction (PCR) and reverse transcription-PCR (RT-PCR) tests for GCVS Total RNA and total DNA were extracted from gentian leaf samples using TRI Reagent solution (Thermo Fisher Scientific) and the NucleoSpin Plant II kit (Macherey-Nagel, Düren, Germany), respectively. Total RNA was reverse-transcribed using random hexamer primers and ReverTraAce reverse transcriptase (Toyobo, Osaka, Japan) [22]. The reverse transcribed product and total DNA extracted from gentian leaf samples were amplified by PCR for RT-PCR and PCR tests, respectively, using Takara Ex Taq DNA polymerase (Takara, Shiga, Japan) with a sense primer (5′-GGACAATTAAAATTTTCCGGATATAAAC-3′) and an antisense primer (5′-ATTTAATTGGATCTATTGGATTTTCACTAC-3′) specific for GCVS [22]. The PCR was performed using a Takara PCR thermal cycler Dice model TP600 (Takara) with 30 cycles of denaturation for 30 s at 94 °C, annealing for 30 s at 55 °C, and extension for 45 s at 72 °C. The amplicons were then separated using 1.2% agarose electrophoresis and visualized with ethidium bromide. Preparation of a digoxigenin (DIG)-labeled antisense RNA probe to detect GCVS The DNA fragment spanning the ORF5 region of GCVS was amplified using Takara Ex Taq DNA polymerase (Takara) with a sense primer (5′-GGACAATTAAAATTTTCCGGATATAAAC-3′) and an antisense primer (5′-GGGTAATACGACTCACTATAGGAGACATATGAGGTGATTTACTAGG-3′, underlined: T7 RNA polymerase promoter site). The PCR was run for 30 cycles of denaturation at 94 °C for 30 s, annealing at 55 °C for 30 s, and extension at 72 °C for 45 s. The amplified DNA fragment was separated in an agarose gel and purified using the MonoFas DNA Purification Kit I (GL Sciences Inc., Tokyo, Japan), then used as a template for in vitro transcription with T7 RNA polymerase in the presence of DIG-11-UTP (Roche Diagnostics, Mannheim, Germany) to generate the DIG-labeled antisense RNA probe. Southern blot hybridization One hundred nanograms of total DNA from a gentian leaf was separated by electrophoresis in a 0.8% agarose gel prepared in 1× TBE buffer (89 mM Tris, 89 mM boric acid, 2 mM EDTA) and stained with ethidium bromide. The DNA was transferred onto a positively charged nylon membrane (Hybond-N+, GE Healthcare, Buckinghamshire, UK) by capillary blotting using the protocol of Sambrook and Russell [45]. Prehybridization was performed at 50 °C for 30 min. Hybridization was carried out for 16 h at 50 °C using the DIG-labeled antisense RNA probe corresponding to the ORF5 region of GCVS prepared above. After hybridization, the membrane was washed twice for 5 min each at room temperature in 2× SSC containing 0.1% SDS, then twice for 15 min each at 65 °C in 0.1× SSC containing 0.1% SDS. For chemiluminescence detection, anti-DIG alkaline phosphatase-conjugated Fab fragments and the CDP-Star chemiluminescent substrate were used as instructed (Roche Diagnostics, Mannheim, Germany). Chemiluminescence was quantified using an ImageQuant LAS 4010 system (GE Healthcare). Rolling circle amplification (RCA) Thirty nanograms of total DNA from a gentian plant were used as a template for RCA analysis using the Illustra TempliPhi 500 Amplification Kit (GE Healthcare) as instructed. Amplified products were digested with restriction enzyme EcoRI, separated by agarose gel electrophoresis and visualized with ethidium bromide. RT-PCR test for RNAs corresponding to all six ORF regions of GCVS in kobu-sho and asymptomatic gentian plants Total RNA was extracted from gentian leaves using TRI Reagent solution (Merck KGaA). To remove potential DNA contamination, the RNA samples were treated with recombinant DNase I (Takara) according to the manufacturer’s protocol. RT reaction was performed using a random hexamer and ReverTraAce reverse transcriptase (Toyobo), and the reverse transcribed product was used a template for PCR amplification. PCR was carried out as described above using primer pairs 1 F/1R, 2 F/2R, 3 F/3R, 4 F/4R, 5 F/5R, and 6 F/6R corresponding to ORF1 to ORF6, respectively (Supplementary Table 1). To confirm that the amplification products were derived from RNA rather than contaminating DNA, control reactions without reverse transcriptase were also performed. Analysis of GCVS integration into gentian genome The GCVS sequence was used as a query for a BLASTn analysis against a genome assembly of G. triflora (gentian draft genome) [1], using an E-value cutoff of 1 × 10− 3 and a minimum sequence identity threshold of > 70%. Results sRNA analysis Deep sequencing of sRNAs from kobu-sho gentian plants yielded 9,187,505 reads (the DNA Data Bank of Japan Sequence Read Archive under accession number DRR932625), which were assembled into 5,325 contigs (Supplementary Data 1). BLASTx analysis of these contigs revealed that seven contigs shared high amino acid sequence identities (78.3–100%) with six caulimoviruses, including carnation etched ring virus (CERV), cauliflower mosaic virus (CaMV), horseradish latent virus (HLV), Lamium leaf distortion virus (LLDV), Mirabilis mosaic virus (MMV), and soybean putnam virus (SbPV) (Table 1). In contrast, no caulimovirus-derived sequences were detected in the BLASTn analysis, likely because amino acid sequences conserved among caulimoviruses make BLASTx analysis more sensitive than nucleotide sequence-based BLASTn analysis for detecting these viruses [17]. Neither BLASTx nor BLASTn analysis detected contigs showing high sequence identity to viruses belonging to genera of the family Caulimoviridae other than Caulimovirus. Table 1Caulimoviruses with high amino acid sequence identity to contigs obtained from sRNA analysis of kobu-sho-affected gentian plants, their corresponding virus genomic regions and accession numbers, and the nucleotide positions of the contigs in the reconstructed GCVS genomeContigNumber of amino acids in contigAmino acid sequence identity (%)CaulimovirusVirus genomic regionAccession numberNucleotide positions in the GCVS genomeNODE_267_length_39_cov_40.02564217100cauliflower mosaic virusORF4NP_056727.13,576-3,628NODE_4254_length_48_cov_30.87500020100horseradish latent virusORF5YP_006907834.14,771-4,83220100cauliflower mosaic virusORF5NP_056728.1NODE_4828_length_54_cov_36.9444432295.5soybean putnam virusORF5YP_006607892.14,924–4992NODE_10099_length_50_cov_28.7000012095Lamium leaf distortion virusORF5YP_001931961.14,834-4,897NODE_6804_length_47_cov_58.8297881888.9cauliflower mosaic virusORF1NP_056724.1522–578NODE_5044_length_45_cov_9.1333331687.5cauliflower mosaic virusORF6NP_056729.16,263-6,318NODE_11834_length_59_cov_24.7118642378.3carnation etched ring virusORF5NP_612577.14,503-4,5752378.3Mirabilis mosaic virusORF5NP_659397.1 Caulimoviruses with high amino acid sequence identity to contigs obtained from sRNA analysis of kobu-sho-affected gentian plants, their corresponding virus genomic regions and accession numbers, and the nucleotide positions of the contigs in the reconstructed GCVS genome In addition to identifying the six caulimoviruses, BLASTx analysis identified 232 contigs showing homology to a polyprotein of GKaV (YP_007438864.1) (Supplementary Data 2). Consistent with this result, BLASTn analysis detected 164 contigs with > 85% nucleotide sequence identity to GKaV (Supplementary Data 3). Mapping of virus-derived sRNA reads to the GKaV RNA genome (NC_020252.1) revealed high genome coverage (86.9%) and a high mean read depth of 278.4. To date, nine viruses—broad bean wilt virus 2, clover yellow vein virus, impatiens necrotic spot virus, cucumber mosaic virus, GKaV, gentian mosaic virus, gentian ovary ringspot virus, gentian virus A, and Sikte waterborne virus—have been reported to naturally infect gentian plants [4, 14, 15, 24, 25, 39, 54]. Among these nine, only GKaV was detected by BLASTn and BLASTx in kobu-sho gentian plants. Nucleotide sequence analysis of caulimovirus-like sequences detected by BLASTx Based on four of the seven contigs with high amino acid sequence identity to caulimoviruses, PCR primers MP-F, MP-R, CP-F, CP-R, POL-F, POL-R, TAV-F, and TAV-R were designed (Table 2) and used for PCR with total DNA from kobu-sho gentian leaves. The extension time was adjusted according to the expected amplicon size, using the protocol for the PCR to detect GCVS. Amplicons of approximately 3.0 kbp (fragment a), 1.3 kbp (fragment b), 1.5 kbp (fragment c), and 2.3 kbp (fragment d) were obtained using primer pairs MP-F/CP-R, CP-F/POL-R, POL-F/TAV-R, and TAV-F/MP-R, respectively (Fig. 1). Table 2Primer sequences for PCR designed from four contigs with high amino acid sequence identity to caulimovirusesContigPrimer namePrimer sequence (5’ to 3’)NODE_6804_length_47_cov_58.829788MP-FACCTAGCCTACAATAAATTTATGTTTGMP-RCAAACATAAATTTATTGTAGGCTAGGNODE_267_length_39_cov_40.025642CP-FTGTTGGATCTGCAATATCGCP-RCTCGATATTGCAGATCCAACNODE_10099_length_50_cov_28.700001POL-FGATTCTGGCAGGTATTGCTAGPOL-RCTAGCAATACCTGCCAGAATCNODE_5044_length_45_cov_9.133333TAV-FTGCAGGTATCTACACAACATGTAV-RTGTTGTGTAGATACCTGCATG Primer sequences for PCR designed from four contigs with high amino acid sequence identity to caulimoviruses Reconstructed genome organization of gentian caulimovirus sequence (GCVS). (a) Linearized GCVS genome. ORFs 1–6 encode proteins with features typical of viruses in the genus Caulimovirus (family Caulimoviridae): ORF1 encodes a viral movement protein; ORF2, an aphid transmission factor; ORF3, a caulimovirus DNA-binding protein; ORF4, a zinc finger–containing protein related to the caulimovirus coat protein; ORF5 contains reverse transcriptase, caulimovirus peptidase A3, and RNase H domains; and ORF6 encodes a caulimovirus viroplasmin. (b) Fragments a–d correspond to PCR-amplified products generated using respective primer pairs MP-F/CP-R, CP-F/POL-R, POL-F/TAV-R, and TAV-F/MP-R After cloning and sequencing of fragments a–d (the DNA Data Bank of Japan databases under accession numbers LC925024-LC925027, respectively), the regions between adjacent fragments were further amplified using additional primer sets designed based on the newly obtained nucleotide sequences, then the resulting products were also cloned and sequenced. No additional sequences were detected between fragments a–b, b–c, c–d, or d–a (Fig. 1), indicating that fragments a–d originated from a theoretical single circular DNA molecule. Moreover, the remaining three contigs for which primers were not designed were also derived from the same molecule consisting of fragments a–d (Table 1). Together, these results suggest that the seven contigs obtained by our sRNA analysis were derived from a single caulimovirus genome (= GCVS). However, comparison of nucleotide sequences among multiple clones of each amplified fragment revealed substitutions, insertions, and deletions as described below. Therefore, the full-length nucleotide sequence of GCVS was constructed by generating a consensus sequence through alignment of sequences from at least six independent clones, using a publish strategy for analysis of endogenous viral elements [3, 29]. Complete nucleotide sequence of GCVS The complete nucleotide sequence of GCVS was 8,125 nt long (Fig. 1, supplemental data 4). Nucleotide numbering was initiated at the 5′ end of the negative-sense strand primer-binding site (5′-TGGTATCGGAGC-3′), which is complementary to the consensus tRNAMET sequence [35, 43, 48]. Comparative sequence analysis of the six ORFs using the NCBI Conserved Domain Database [33] revealed the presence of conserved domains characteristic of caulimoviruses: a viral MP domain (amino acids 52–236 of the ORF1 protein; pfam01107), an ATF domain (amino acids 2–148 of the ORF2 protein; pfam03233), a caulimovirus DNA-binding protein domain (amino acids 15–107 of the ORF3 protein; pfam03310), a zinc finger-containing domain related to the caulimovirus CP (amino acids 391–407 of the ORF4 protein; smart00343), a reverse transcriptase domain (amino acids 302–481 of the ORF5 protein; cd01647), a caulimovirus peptidase A3 domain (amino acids 51–253 of the ORF5 protein; pfam02160), an RNase H domain (amino acids 576–697 of the ORF5 protein; cd09274), and a caulimovirus viroplasmin domain (amino acids 122–163 of the ORF6 protein; pfam01693). The genome size and organization of GCVS elucidated above are typical of caulimoviruses. Among caulimoviruses and caulimovirus-related, unclassified viruses, GCVS shared the highest overall nucleotide sequence identity with LLDV (61% identity across the full-length genome; Table 3). Comparison of individual ORFs showed that the highest nucleotide sequence identity was 64% for ORF1 with CaMV and figwort mosaic virus (FMV), 61% for ORF2 with Angelonia bushy stunt virus (ABSV), 58% for ORF3 with FMV and SbPV, 61% for ORF4 with CaMV, 69% for ORF5 with LLDV and SbPV, and 58% for ORF6 with CERV, Dahlia mosaic virus (DMV), and HLV (Table 3). At the amino acid level, the highest sequence identity for each ORFs was 51% for ORF1 with FMV and SbPV, 43% for ORF2 with CaMV and HLV, 39% for ORF3 with DMV and MMV, 46% for ORF4 with CERV, 66% for ORF5 with ABSV and SbPV, and 39% for ORF6 with Atractylodes mild mottle virus and SbPV (Table 3). Phylogenetic analyses based on the deduced amino acid sequences of GCVS ORF1 and ORFs 3–6 further supported the close relationship among GCVS and caulimoviruses (Fig. 2). A phylogenetic tree based on the deduced amino acid sequence of GCVS ORF2 was not constructed because soymoviruses lack a gene corresponding to GCVS ORF2, precluding the definition of an appropriate outgroup. Table 3Percentage nucleotide (nt) and amino acid (aa) sequence identities for sequences of open reading frames (ORFs) shared between the reconstructed GCVS genome and other caulimoviruses and two caulimovirus-related, unclassified virusesGenusNt identityPercentage nt sequence identity, and aa sequence identityVirus nameAccessionFull genomeORF1ORF2ORF3ORF4ORF5ORF6 Caulimovirus Lamium leaf distortion virusEU5544236162, 3757, 3957, 3860, 4269, 6557, 37cauliflower mosaic virusV001416064, 5059, 4255, 3561, 4566, 6457, 38figwort mosaic virusX061666064, 5160, 3858, 3860, 4267, 6357, 35Atractylodes mild mottle virusKR0803276063, 5059, 4256, 3758, 4067, 6557, 39carnation etched ring virusX046586062, 4859, 3855, 3760, 4667, 6558, 37Angeloca bushy stunt virusKU5088006063, 4661, 3856, 3558, 3869, 6656, 35Dahlia mosaic virusJX2723205962, 4859, 3656, 3957, 3961, 5558, 34Mirabilis mosaic virusAF4546355962, 4658, 3756, 3959, 4065, 6156, 35horseradish latent virusJX4299235962, 4859, 4257, 3758, 4466, 6458, 37soybean putnam virusJQ9269835963, 5159, 4158, 3760, 4169, 6657, 39strawberry vein banding virusX973045857, 3656, 3057, 3058, 3764, 5256, 33Metaplexis yellow mottle-associated virusMW6562145758, 3956, 3256, 2957, 3762, 5356, 32Pueraria virus AMZ8261385757, 3957, 3251, 2858, 3764, 5355, 34caulimovirus-related, unclassified virusesDahlia common mosaic virusJN03227365963, 4860, 3956, 3759, 3966, 6257, 34Eupatorium vein clearing virusEU5698315861, 4857, 3555, 3558, 3663, 5357, 36 Percentage nucleotide (nt) and amino acid (aa) sequence identities for sequences of open reading frames (ORFs) shared between the reconstructed GCVS genome and other caulimoviruses and two caulimovirus-related, unclassified viruses Phylogenetic trees based on amino acid sequence alignments of proteins encoded by GCVS and members of the genera Caulimovirus (caulimoviruses) and Soymovirus (soymoviruses). Phylogenetic trees were constructed using the maximum likelihood method with the best-fit protein substitution model implemented in MEGA12. The newly identified GCVS is in bold. Accessions for amino acid sequences of caulimoviruses included in the analyses are ABSV (Angelica bushy stunt virus, KU508800), AMMV (Atractylodes mild mottle virus, KR080327), CaMV (cauliflower mosaic virus, V00141), CERV (carnation etched ring virus, X04658), DMV (Dahlia mosaic virus, JX272320), FMV (figwort mosaic virus, X06166), HLV (horseradish latent virus, JX429923), LLDV (Lamium leaf distortion virus, EU554423), MMV (Mirabilis mosaic virus, AF454635), MYMaV (Metaplexis yellow mottle-associated virus, MW656214), PVA (Pueraria virus A, MZ826138), SbPV (soybean Putnam virus, JQ926983), and SVBV (strawberry vein banding virus, NC_001973). Soymoviruses BRRV (blueberry red ringspot virus, MZ004932), and SbCMV (soybean chlorotic mottle virus, X15828) were used as outgroups. Bootstrap values more than 30% are indicated at branching points in the phylogram as a percentage of 1,000 iterations Detection and seed transmission of GCVS in a commercial field To determine whether GCVS is specifically associated with kobu-sho disease, total DNA and total RNA were extracted from leaves of 30 symptomatic and 20 asymptomatic gentian plants (cv. Light-Pink-Ashiro) collected from a commercial field and tested them using PCR and RT-PCR, respectively. GCVS was detected in all samples in both assays, regardless of the presence or absence of kobu-sho symptoms. These results suggest that GCVS DNA is present in all samples and that GCVS-derived RNA is transcribed in plant cells. In addition, seeds produced after natural pollination in the commercial field were collected from kobu-sho plants and apparently healthy gentian plants (cv. Light-Pink-Ashiro). PCRs for GCVS using total DNA from 20 seedlings from each seed source detected GCVS in all samples, whether the maternal plants had kobu-sho symptoms or not. Southern blot hybridization and RCA of total DNA from wild gentian and three F₁ G. triflora cultivars and their respective parents GCVS was detected in all gentian plants with and without kobu-sho symptoms. Although episomal caulimoviruses have not been reported to be transmitted through seeds [17], GCVS was detected in all seedlings examined, indicating 100% seed transmission of GCVS. Therefore, GCVS might be present as an endogenous pararetrovirus (EPRV) in gentian plants. To examine this possibility, we used a GCVS-specific probe for Southern blot hybridization with total DNA from wild gentian and three F₁ gentian cultivars (cv. Light-Pink-Ashiro, Ashiro-no-Natsu, and Ashiro-no-Aki) and their respective maternal and paternal parents (with N. tabacum as the control). No hybridization signals were detected in total DNA from N. tabacum, whereas all gentian samples yielded clear hybridization signals exclusively at positions corresponding to high-molecular-mass genomic DNA (Fig. 3a). If GCVS were present as an episomal virus, then we would expect to detect circular double-stranded DNA genomes—characteristic of active pararetroviruses—as a supercoiled form migrating faster than 8,125 bp, a linear form at 8,125 bp, and a relaxed circular form migrating slower than 8,125 bp during agarose gel electrophoresis. However, no hybridization signals corresponding to any of these episomal viral DNA forms were observed. Southern blot hybridization analysis using GCVS-specific probe and total DNA from apparently healthy Gentiana triflora (gentian) plants and gentian plants collected before and after symptoms appeared. (a) Total DNA from apparently healthy plants of wild gentian and F₁ cultivars Light-Pink-Ashiro, Ashiro-no-Natsu, and Ashiro-no-Aki and their respective maternal and paternal parents (Ei-P and INP; AZ1-Early-1 and NE2N-No. 1; and N27-No. 29 and Y8-No. 2, respectively). (b) Total DNA from kobu-sho-affected cv. Light-Pink-Ashiro collected May 15, June 15, July 15, and August 15, 2023; kobu-sho symptoms were present from June on. Total DNA was separated by agarose gel electrophoresis (top of each panel), then subjected to Southern blot hybridization (bottom of each panel). Total DNA from Nicotiana tabacum plants was used as a negative control To further assess the presence of episomal GCVS in these gentian plants, we used RCA, which selectively amplifies circular DNA molecules using random hexamer primers and Phi29 DNA polymerase. No amplicons were obtained from total DNA from the wild gentian, three F₁ cultivars, or the respective parents (Fig. 4a); however, amplicons were detected from the positive controls: bacterial plasmid DNA (pUC19) and total DNA from blueberry leaves infected with BRRV (soymovirus). These results demonstrate that the RCA reliably detected circular dsDNA molecules. Collectively, these findings indicate that GCVS does not exist as an episomal virus but is instead integrated into the gentian genome as an endogenous element. Rolling circle amplification (RCA) of total DNA from apparently healthy gentian plants and plants collected before and after symptoms appeared to detect episomal forms of GCVS. (a) Total DNA from wild gentian and F₁ cultivars Light-Pink-Ashiro, Ashiro-no-Natsu, and Ashiro-no-Aki and their respective maternal and paternal parents (Ei-P and INP; AZ1-Early-1 and NE2N-No. 1; and N27-No. 29 and Y8-No. 2, respectively). (b) Total DNA from cv. Light-Pink-Ashiro collected on May 15, June 15, July 15, and August 15, 2023; kobu-sho symptoms were present from June on. Bacterial plasmid DNA (pUC19) and total DNA from blueberry plants infected with BRRV were used as positive controls; total DNA from BRRV-uninfected blueberry plants was used as negative controls. Asterisks (*) indicate RCA products Southern blot hybridization and RCA of total DNA from kobu-sho gentian plants Some endogenous pararetroviruses (EPRVs) are known to be reactivated, producing episomal viruses and occasionally causing disease symptoms in plants [32, 38, 44]. We therefore investigated whether GCVS can be reactivated to form an episomal virus in kobu-sho gentian plants. Gentian plants are perennial crops, and once affected by kobu-sho disease, they do not recover; symptoms reappear in June on new shoots from overwintered buds. Accordingly, leaf samples were collected monthly from May to August 2023 from gentian plants that had kobu-sho symptoms in 2022. In the southern blot hybridization using total DNA from these samples and the GCVS-specific probe, all gentian samples yielded clear hybridization signals exclusively at positions corresponding to high-molecular-mass genomic DNA, but no hybridization signals were detected from genomic DNA from N. tabacum (Fig. 3b). Importantly, no hybridization signals corresponding to episomal viral DNA—including complete circular dsDNA, linear dsDNA, and relaxed circular dsDNA forms—were detected at any sampling time, either before the appearance of kobu-sho symptoms (samples collected in May) or after (samples collected from June to August). To further assess the presence of episomal GCVS in kobu-sho gentian plants, we used RCA. No amplicons indicative of episomal virus formation were detected in any samples collected during this period (Fig. 4b). In contrast, DNA from the positive controls (bacterial plasmid DNA [pUC19] and total DNA from blueberry leaves infected with BRRV) were amplified, but not from the negative controls (DNA from BRRV-negative blueberry plants). Collectively, neither Southern blot hybridization nor RCA provided evidence of the presence of episomal GCVS before or after the appearance of kobu-sho symptoms. Detection of GCVS-derived RNA in gentian plants by RT-PCR DNA spanning the ORF5 region of GCVS was amplified by RT-PCR test. To further investigate whether RNAs corresponding to each of the six ORF regions of GCVS are present in gentian plants, we subjected RNA from both kobu-sho-affected and apparently healthy gentian plants to RT-PCR. RNAs corresponding to all six ORF regions were detected (Fig. 5). These results suggest that GCVS-derived RNAs are transcribed across the entire GCVS genome in plant cells, regardless of the presence or absence of kobu-sho symptoms. RT-PCR detection of GCVS RNA in kobu-sho-affected (lanes 2 and 3) and apparently healthy (lanes 4 and 5) gentian plants. Fragments corresponding to GCVS ORFs 1–6 were amplified using respective primer pairs 1 F/1R (a), 2 F/2R (b), 3 F/3R (c), 4 F/4R (d), 5 F/5R (e), and 6 F/6R (f). Arrows indicate positions of expected bands for each ORF. PCR was also performed without reverse transcription to confirm DNA was not present in the RNA samples; no amplicons were detected (data not shown) GCVS sequences in the gentian genome To determine whether GCVS is integrated into the gentian genome, we used the GCVS sequence as the query for a BLASTn analysis of a gentian draft genome assembly comprising 4,374 contigs with a total length of 3,657,985,820 nt, (Fig. 1). This analysis identified 1,929 GCVS-derived fragments distributed across 280 contigs of the gentian draft genome. The cumulative length of these fragments was 3,116,680 nt, corresponding to approximately 0.085% of the draft genome. However, because the correspondence between these contigs and gentian chromosomes is unknown [1], it is not possible to determine whether the GCVS fragments are widely distributed across all chromosomes or confined to specific chromosomes. Analysis of the insertion boundaries revealed that the most frequent insertion start site corresponded to nucleotide position 1 of the reconstructed full-length GCVS sequence (247 fragments; Supplemental Fig. 1a), whereas the most frequent insertion end site was nucleotide position 8072 (76 fragments; Supplemental Fig. 1b). Notably, no fragments spanned the entire region from nucleotide positions 1 to 8072. Moreover, more than 90% of the GCVS-derived fragments were shorter than 4,000 nt (Fig. 6). The top 10 GCVS-derived fragments ranked by query coverage (Fragments 1–10) are shown in Fig. 7. Nucleotide substitutions, insertions, and deletions were observed in the ORFs present in these fragments, leading to amino acid substitutions and premature stop codons in the encoded proteins. (Table 4; Fig. 7). Frequency distribution of nucleotide sequence lengths of GCVS-derived fragments within the gentian genome. Bars represent 100-nt intervals. Numbers at bottom right in parentheses: nucleotide length / corresponding frequency GCVS-derived fragments (Fragments 1–10) ranked by coverage of the reconstructed GCVS genome. Boxes represent ORFs; black dots indicate the positions of premature stop codons; “qcov” denotes genome coverage Table 4Percentage nucleotide sequence identity and number (no.) of inserted and deleted bases in nucleotides in each ORF between the reconstructed GCVS genome and GCVS-derived fragments (Fig. 7, Fragments 1–10)FragmentORF1ORF2ORF3ORF4ORF5ORF6Identity(%)No. insertionsNo.deletionsIdentity(%)No.insertionsNo.deletionsIdentity(%)No.insertionsNo.deletionsIdentity(%)No.insertionsNo.deletionsIdentity(%)No.insertionsNo.deletionsIdentity(%)No.insertionsNo.deletions197.7 a0098.70294.21497.93897.82095.655294.60094.20091.41192.334951482.8 a88394.86097.51089.32295.90294.512281.1 a728493.5001000092.90090.63094.112183.7 a67592.77595.10091.71393.13396.50381.4 a118694.53294.41192.611930494.49683.1 a98792.65595.11196.30092.901395.291179.8 a919894.40091.321492.92391.76793.4161880 a911994.20095.500880090.471894.22581.4 a4810941094.903910793.26995.21484.6 a69a When ORF1 and ORF6 did not encompass the entire region, only the regions present were compared Percentage nucleotide sequence identity and number (no.) of inserted and deleted bases in nucleotides in each ORF between the reconstructed GCVS genome and GCVS-derived fragments (Fig. 7, Fragments 1–10) a When ORF1 and ORF6 did not encompass the entire region, only the regions present were compared Examination of the genomic distribution of GCVS-derived fragments revealed pronounced clustering in specific contigs. Contig_1326 harbored the largest number of fragments (52), whereas scaffold_3803 contained 33 fragments, including Fragment 1, which showed the highest query coverage (Fig. 8). As exemplified by these contigs, even regions containing numerous GCVS-derived insertions showed extensive fragmentation of GCVS sequences, with individual fragments distributed across different genomic regions and both plus and minus strands present within each contig. Distribution of GCVS-derived fragments in contigs of the gentian draft genome. (a) Contig_1326, which harbors the largest number of GCVS-derived fragments (52 fragments). (b) Scaffold_3803, which contains 33 GCVS-derived fragments, including Fragment 1, which shows the highest coverage of the reconstructed GCVS genome (96.1%; Fig. 7). Red and blue arrows represent GCVS-derived fragments on the plus and minus strands of the GCVS nucleotide sequence, respectively. Numbers above the arrows indicate the start and end nucleotide positions of each fragment, based on the plus strand of the reconstructed full-length GCVS genome. Black numbers indicate the nucleotide lengths between adjacent GCVS-derived fragments Discussion In this study, we reconstructed the complete nucleotide sequence of GCVS by assembling GCVS-derived fragments that were integrated into the gentian genome (Figs. 1 and 7, and 8). Although the reconstructed sequence exhibits a typical caulimovirus genome organization and retains conserved amino acid motifs in ORFs 1–6, no episomal form of GCVS was detected (Figs. 3 and 4); therefore, it remains unclear whether the reconstructed sequence represents a fully functional viral genome. According to the species demarcation criterion for the genus Caulimovirus established by the International Committee on Taxonomy of Viruses, a nucleotide sequence divergence of more than 20% in the polymerase region (RT + RNase H) is required to define a distinct species [17]. Comparative analyses revealed that GCVS had greater than 20% nucleotide divergence from all recognized caulimoviruses and caulimovirus-related unclassified viruses in this region, supporting its classification as a novel caulimovirus species (Table 3). This conclusion is further supported by phylogenetic analyses based on the deduced amino acid sequences of ORF1 and ORFs 3–6, in which GCVS clustered with caulimoviruses rather than with soymoviruses (Fig. 2). Collectively, these findings suggest that a previously undiscovered caulimovirus species is integrated into the gentian genome, and GCVS was therefore designated endogenous gentian pararetrovirus. Although some latent EPRVs can be reactivated to generate episomal viruses, occasionally accompanied by virion formation and disease symptoms [21, 32, 38, 44], in the present study, we found no evidence for the presence of episomal GCVS in one wild gentian plant, three F₁ gentian cultivars, or their respective parental lines (Figs. 3 and 4). Furthermore, episomal GCVS was not detected in kobu-sho-affected gentian plants either before or after symptom development (Figs. 3 and 4). A total of 1,929 GCVS-derived fragments were identified in the draft genome, all of which were shorter than the full-length GCVS genome (Figs. 6 and 8). In addition, the ORFs within Fragments 1–10, which contain nearly complete viral genomes, frequently harbored premature stop codons and amino acid substitutions (Fig. 7; Table 4). The accumulation of such mutations under relaxed selective constraints following viral integration into the host genome is likely to impair the function of viral proteins required for autonomous episomal replication [13]. In endogenous pararetroviruses such as banana streak virus, activation of endogenous viral sequences has been reported to occur through recombination [38]. However, even if a full-length genome was reconstructed from GCVS-derived fragments by recombination, the presence of multiple deleterious mutations in coding regions would likely prevent the production of a functional episomal virus. Together, these results suggest that although numerous GCVS-derived sequences are present in the gentian genome, GCVS most likely persists as a replication-defective endogenous pararetroviral element [20, 29]. This interpretation is further supported by previous observations that caulimovirus-like particles were not detected by electron microscopy using negative staining of homogenized material from kobu-sho-affected gentian plants [25]. Overall, our findings provide new insights into the structure, diversity, and evolutionary fate of endogenous pararetroviral sequences in the gentian genome. Most EPRVs are thought to accumulate at high copy numbers through vertical transmission after integration of viral DNA into host chromosomes via illegitimate (nonhomologous) recombination [20]. Although large-scale gentian domestication in Japan began approximately 60 years ago, the integration of GCVS into the genome of wild gentian indicates that the integration predates gentian domestication. Replication-defective EPRVs are generally regarded as molecular fossils of ancient viral infections that became fixed in host germline genomes over thousands to millions of years [9, 19, 51]. Accordingly, surveys of GCVS across diverse Gentiana species from different geographic regions may help identify the original host species in which GCVS was first integrated and provide insights into the geographic origin of GCVS and the evolutionary history of the genus Gentiana. Although nine plant viruses are known to infect gentian plants in Japanese commercial fields, in the present study, we only detected GKaV in the sRNA analysis of the kobu-sho gentian plants. This result was corroborated by the RT-PCR of total RNA from the plants which showed only GKaV among the nine reported viruses. GKaV has been detected in more than 99% of kobu-sho gentian plants [25] and in poenies with leaf distortion disease, which is characterized by root gall formation, in the Netherlands and the United States [47]. Although GCVS-derived sequences were detected by sRNA analysis, no episomal form of GCVS was identified by either Southern blot hybridization or RCA. In addition, GCVS seems to be present as a replication-defective EPRV. On the basis of these results, episomal GCVS is unlikely to be involved in the development of kobu-sho disease. The CaMV ORF6 protein (Tav) induces chlorosis through host-mediated responses [6, 52]. In caulimoviruses, ORF6 is translated from a monocistronic RNA transcribed from the 19 S promoter; thus, any involvement of the GCVS ORF6 in causing kobu-sho symptoms would presuppose transcriptional activation of the 19 S promoter from integrated GCVS fragments. Although DNA methylation is dynamically regulated and can be influenced by environmental stress or aging [30, 40, 44], no evidence yet supports that the GCVS 19 S promoter is activated or demethylated in gentian plants. More importantly, the GCVS ORF6 regions harbor multiple deleterious mutations. Such disruptions would effectively abolish the production of a functional Tav protein even if transcription were hypothetically initiated. These observations indicate that the ORF6 sequences derived from integrated GCVS fragments are unlikely to contribute to the development of chlorosis in the apical region of kobu-sho gentian shoots. As mentioned earlier, sRNAs play central roles in TGS and PTGS, which constitute major antiviral defense mechanisms in plants [26]. In petunia vein clearing virus (PVCV), a full-length viral genome is integrated into the host genome; nevertheless, viral transcription is suppressed by TGS mediated through methylation of the viral promoter [30]. Methylation signals targeting EPRVs are also thought to act on episomal viruses, leading to methylation of viral promoters [36, 41]. In addition, sRNAs are known to induce PTGS by targeting viral RNAs [7, 49, 50]. In this study, GCVS-derived sequences were detected in the sRNA analysis. GCVS-derived RNAs were also transcribed across the entire GCVS genome in plant cells (Fig. 5). These findings indicate that GCVS sequences integrated into the gentian genome are transcriptionally active in gentian nuclei and that their transcripts are subsequently processed into sRNAs in the cytoplasm. These observations raise the possibility that GCVS-derived sRNAs contribute to antiviral defense via both TGS and PTGS, potentially conferring resistance to episomal caulimoviruses with high sequence identity to GCVS. Such defense mechanisms may have led to the elimination of episomal GCVS in gentian plants. Future studies will clarify whether sRNAs derived from GCVS have any antiviral activity against viruses that share high nucleotide sequence identity with GCVS. Electronic Supplementary Material Below is the link to the electronic supplementary material. Supplemental Data 1 (FA 433 KB) Supplemental Data 2 (XLSX 31.9 KB) Supplemental Data 3 (XLSX 26.6 KB) Supplemental Data 4 (DOCX 17.4 KB) Supplemental Fig 1 (DOCX 16.2 KB) Supplemental Table 1 (PPTX 67.1 KB) Acknowledgements This work was supported by JSPS KAKENHI Grant Number 24K08905 to M.I. Author contributions Conceptualization: M. Isogai; Methodology: M. Isogai, S. Tezuka, and A. Abe; Investigation: M. Isogai, S. Tezuka, A. Abe; Sampling of diseased gentian plants: M. Isogai, S. Tezuka, and Y. Takamura; Cultivation and supply of apparently healthy gentian cultivars: S. Tezuka, and Y. Takamura; Monthly discussion: M. Isogai, S. Tezuka, and H. Yaegashi; Writing – original draft and visualization: M. Isogai; Review and editing: all authors; Funding acquisition: M. Isogai. Funding Open Access funding provided by Iwate University Data availability No datasets were generated or analysed during the current study. Declarations Ethical approval This article does not contain any studies with human participants or animals performed by any of the authors. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplemental Data 1 (FA 433 KB) Supplemental Data 2 (XLSX 31.9 KB) Supplemental Data 3 (XLSX 26.6 KB) Supplemental Data 4 (DOCX 17.4 KB) Supplemental Fig 1 (DOCX 16.2 KB) Supplemental Table 1 (PPTX 67.1 KB) Data Availability Statement No datasets were generated or analysed during the current study.