The fungal strain L3 (D. seriata) was collected and identified during a mycoviral screening survey conducted in 2018 from a diseased leaf of the Lokath plant (Khan et al., 2021b). Some mycoviruses identified as a result of this screening survey have already been published (Khan et al., 2021a; Khan et al., 2021b; Khan et al., 2022). For isolating purified cultures of L3, surface sterilization of leaves was performed with 1% sodium hypochlorite solution and further washing was done using autoclaved sterile distilled water. Samples were air-dried in a safety cabinet and infected parts were inoculated on a fresh potato dextrose agar (PDA, Becton, Dickinson and Co.) plate to obtain fungal cultures. Incubation of fungal cultures was performed at 25°C for 4 to 8 days for growth purposes. Glycerol stocks (40% v/v) of the pure colonies were prepared and stored at -80°C for future use. Fungal identification was performed based on the morphology ( Figure 1 ) and internal transcribed spacer (ITS) region sequencing ( Supplementary Figure 1 ) (White et al., 1990). Total RNA fractions were obtained as described earlier (Eusebio-Cope and Suzuki, 2015). The classical dsRNA extraction method using cellulose chromatography column was employed to extract dsRNA fractions from the fungal samples (Bhatti et al., 2012; Eusebio-Cope and Suzuki, 2015). The obtained dsRNA fractions were then treated with RQ1 DNase (Promega Corp.) and S1 Nuclease (Thermo Fisher Scientific Inc.) to eliminate genomic DNA and ribosomal RNA contaminations.

Mycelial plugs taken from one month old mycelia were cultured on 2% water agar plates at 25°C and were allowed to grow under dark conditions. Once growing hyphal tips were confirmed under a dissecting microscope in a clean bench, single hyphal tips from single hyphae were excised with a scalpel and were allowed to grow on new water agar plates. Single sub-isolates were obtained after four rounds of hyphal tipping. All regenerated isolates were cultured on cellophane overlaid PDA to examine for the presence of mycoviruses via dsRNA extraction or RT-PCR assay. A total of 30 independent hyphal tip isolates were obtained. Phenotype of the parental strain (L3) and a partially cured hyphal tip subculture (L3ht1) were compared.

Total RNA fractions were separately obtained from three dsRNA-positive fungal isolates, L3 (D. seriata, a phytopathogen), SP1 (F. mangiferae, a phytopathogen) and HK (Geotrichum candidum, a soil-inhabitant) and were pooled together as already described by Khan et al. (Khan et al., 2021a; Khan et al., 2021b). Ribosomal RNA depletion, cDNA library construction and RNA sequencing were performed by Macrogen Inc. (Tokyo, Japan) as already described by Khan et al. (Khan et al., 2021b). The assembly of RNA sequencing reads was performed using the Read Mapping algorithm with a default parameter in the CLC Genomics Workbench (version 11, CLC Bio-Qiagen) assembled 42,770 contig sequences were subjected to local BLAST searches to check sequence similarities with deposited virus reference sequences (RefSeq) in the National Center for Biotechnology Information (NCBI) database (data not shown). The presence of mycoviruses corresponding to virus-like contigs was confirmed via RT-PCR amplification from total RNA fractions of the L3 strain with virus specific primers ( Supplementary Table S1 ). The terminal sequences of viral genomes or segments were achieved through the 3′ RNA ligase mediated rapid amplification of complementary cDNA ends (RLM-RACE) procedure as described by Suzuki et al. (Suzuki et al., 2004). RT-PCR and RACE amplicons were subjected to Sanger sequencing.

Viral sequence analysis was achieved by performing alignments through GENETYX DNA-processing software. Furthermore, open reading frames (ORFs) were predicted using the online version of the ORF finder program (http://www.ncbi.nlm.nih.gov/gorf/gorf.html). Standard genetic codon usage was selected for the prediction of ORFs for all mycovirus candidates identified in strain L3. Sequence similarity searches were conducted by NCBI BLAST (BLASTn, BLASTx and BLASTp programs) (https://blast.ncbi.nlm.nih.gov/Blast.cgi). NCBI conserved domain database (CDD) (http://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi) and motif finder (https://www.genome.jp/tools/motif/) tools were utilized for detecting conserved domains (Marchler-Bauer et al., 2017). Conservation at 5′- and 3′-terminal sequences of the genomes or segments and the motif prediction (I-VIII) for RNA-dependent RNA polymerase (RdRP) core domains were achieved by conducting nucleotide and amino acid sequence alignments, respectively, through the UGENE program (http://ugene.net/) (Okonechnikov et al., 2012). The viral sequences were deposited in the GenBank/ENA/DDBJ database. GC content (%) was calculated through the online server of ENDMEMO program (http://www.endmemo.com/bio/gc.php). The predicted molecular mass of the putative protein encoded by viral genomes or their segments was calculated by an online protein molecular weight estimation program (https://www.bioinformatics.org/sms/prot_mw.html). Transmembrane domains were predicted by TMHMM – 2.0 (https://services.healthtech.dtu.dk/service.php?TMHMM-2.0).

Phylogenetic analyses were performed using RdRP deduced amino acid sequences. Multiple sequence alignments were constructed using the online server of MAFFT (version 7) (https://mafft.cbrc.jp/alignment/server/) (Katoh and Toh, 2008). Unreliable regions of alignments were trimmed using Gblocks ver. 0.91b (Talavera and Castresana, 2007). Maximum likelihood (ML) trees were constructed using the online version (3.0) of PhyML with automatic model selection by SMS (Smart Model Selection) (Lefort et al., 2017) (http://www.atgc-montpellier.fr/phyml-sms/). Bootstrap values with 1000 replicates were used for the analysis and values less than 500 were masked. Trees were rooted at the mid-point for defining the out-group. Final ML trees were visualized using the iTOLL; Interactive Tree Of Life online server (https://itol.embl.de/upload.cgi).

Four contig sequences (contig numbers 33, 70, 348, and 151) ( Table 1 ) appeared to represent the four dsRNA segments of a single chrysovirus which we tentatively named Diplodia seriata chrysovirus 1 (DsCV1) ( Figure 2 ). DsCV1 dsRNA1 to 4 were numbered with their decreasing sizes (3594 bp, 3265 bp, 3058 bp and 3202 bp, respectively) ( Figure 2 ) ( Table 1 ). Multiple alignments of terminal nucleotide sequences showed conservation at both termini (except for the dsRNA3 5’ terminal), as reported for other known chrysoviruses ( Supplementary Figure S2A ). CAA rich repeats were also observed at 5′ termini of all dsRNAs, which is commonly observed in chrysoviruses ( Supplementary Figure S2A ). CAA repeat sequences are considered to be translation enhancer elements in tobamoviruses (Jiang and Ghabrial, 2004; Kotta-Loizou et al., 2020).

DsCV1 dsRNA1 encodes for an RdRP (designated P1) of 1100 amino acids (aa) with a predicted molecular weight of 128 kDa. DsCV1 P1 was found to have an RdRP domain (pfam02123, E-value 2.7e-56) predicted from location 362 aa to 815 aa in NCBI CDD search ( Figure 2 ) and eight conserved motifs (I–VIII) observed in RdRPs of typical dsRNA viruses ( Supplementary Figure S2B ). A BLASTp search showed that DsCV1 RdRP exhibited a high sequence identity to Macrophomina phaseolina chrysovirus 1 (MpCV1) RdRP (63.4% identity; Table 1 ) (Marzano et al., 2016). Similarly, the other proteins P3 and P4 encoded by DsCV1 dsRNA3 and dsRNA4, respectively, showed the highest aa sequence identity to the corresponding proteins of MpCV1: 33.4% identity for P3 (911 aa, 101 kDa) and 60.6% identity (848 aa, 95.2 kDa) ( Table 1 ) (Marzano et al., 2016). P3 and P4 encode hypothetical proteins with unknown functions. The presence of small ORFs upstream of the major ORFs was observed in each genomic segment (except for dsRNA1) of an alphachrysovirus, Cryphonectria nitschkei chrysovirus 1 (CnCV1) (Shahi et al., 2021). No such small ORF was detected in DsCV1 genome segments.

Phylogenetic analysis, based on the complete aa sequence of the RdRP of DsCV1 and selected members of the Chrysoviridae family indicated that DsCV1 was phylogenetically close to MpCV1 and clustered with alphachrysoviruses ( Figure 2 ). These results are in accordance with the BLASTp search result ( Table 1 ). According to ICTV species demarcation criteria for chrysoviruses (≤70 % and ≤53 % identity in RdRP and CP) (Kotta-Loizou et al., 2020), DsCV1 appears to represent a new species of the genus Alphachrysovirus, the first chrysovirus infecting the phytopathogenic fungus D. seriata. Since CnCV1 is likely to have a very narrow host range (Shahi et al., 2021), it is worth interesting whether DsCV1 and other chrysoviruses have narrow or broad host ranges.

The four contig sequences (contig numbers 980, 965, 245, and 706) were assumed to represent the four dsRNA segments (dsRNA1 to dsRNA4) of a polymycovirus designated as Diplodia seriata polymycovirus 1 (DsPmV1) ( Table 1 and Figure 3 ). DsPmV1 dsRNA 1-4 were 2439 bp, 2189 bp, 2014 bp, and 1133 bp in length, respectively ( Figure 3 ), with their predicted GC content ranging of 61–63%. The multiple nucleotide sequence alignment of all four DsPmV1 dsRNAs showed a high degree of conservation at the 5’ and 3’ termini ( Supplementary Figure S3A ). The 5’ termini of all dsRNAs shared 5’-CGAUUAAAACUU…-3’ sequence and the 3’ termini shared conserved 5’-GGGG…-3’ tetranucleotide. DsPmV1 dsRNA1 has a single ORF (designated P1) encoding an RdRP of 764 aa with a predicted molecular weight of 83.6 kDa ( Figure 3 ). A motif search of the NCBI CDD revealed P1 has an RdRP domain (fam00680, E-value 7.4e-⁰⁸) spanning amino acid positions 409–590 aa (Koonin and Dolja, 1993) ( Figure 3 ) along with eight conserved motifs (I-VIII) of typical RdRPs ( Supplementary Figure S3B ). A conserved GDNQ motif was observed in DsPmV1 P1, which is a peculiar feature of members of the family Polymycoviridae. This tetrad is commonly observed in (-)ssRNA viruses in the order Mononegavirales ( Supplementary Figure S3B ) (Kanhayuwa et al., 2015; Zhai et al., 2016; Kotta-Loizou and Coutts, 2017; Sato et al., 2020a). A BLASTp search of the NCBI protein database found that DsPmV1 P1 shares 75.8% aa sequence identity with the RdRP of Botryosphaeria dothidea RNA virus 1 (BdRV1) (Hu and Kurgan, 2019; Mahillon et al., 2019) ( Table 1 ). DsPmV1 dsRNA2 encodes for the 694 aa protein (designated P2, 74.7 kDa). P2 shows 81.6% aa sequence identity to BdRV1 P2 ( Table 1 ), but no conserved motifs were observed in this protein. The dsRNA3-encoded protein (613 aa, designated as P3) has a predicted molecular mass of 66.4 kDa ( Figure 3 ) and a motif suggesting putative class I S-adenosylmethionine-dependent methyltransferase (AdoMet-MTase; cd02440) as in the case for other polymycovirus P3 proteins (Kanhayuwa et al., 2015; Zhai et al., 2016; Jia et al., 2017; Sato et al., 2020a). DsPmV1 P3 shares 83.5% aa sequence identity with BdRV1 P3 ( Table 1 ). Although the experimental data are not available, the presence of this domain suggests that P3 is involved in capping genomic dsRNA or messenger RNAs as proposed earlier (Kanhayuwa et al., 2015). DsPmV1 dsRNA4 encodes a protein of 275 aa (designated P4, 28.6 kDa) ( Figure 3 ). This segment was predicted to contain a proline-alanine-serine rich protein (PASrp) [P(21), A(41), and S(21) residues] and showed 77.0% similarity with P4 of BdRV1 ( Table 1 ).

Phylogenetic analysis, based on the complete aa sequence of the RdRPs, suggested that DsPmV1 shows close phylogenetic affinity to BdRV1 in the family Polymycoviridae ( Figure 3 ). According to polymycovirus species demarcation criteria (≤70 % aa sequence identity in the RdRP) (https://talk.ictvonline.org/files/ictv_official_taxonomy_updates_since_the_8th_report/m/fungal-official/9407), DsPmV1 should be classified as the same species as BdRV1 based on the high level of aa sequence identity. It will be an interesting question whether these two viruses might have co-evolved with their host fungi or whether they might have been horizontally transferred between the two phylogenetically closely related fungi.

Two contig sequences (contig numbers 73 and 225) showed similarities with the dsRNA segments encoding RdRP and CP of already reported partitiviruses, respectively ( Table 1 ), and corresponded to the genomic segments of a partitivirus termed Diplodia seriata partitivirus 1 (DsPV1). The genome organization of DsPV1 is shown in Figure 4 . No interrupted poly(A) stretches or poly(A) tail was observed at the 3′ terminal of DsPV1 segments ( Supplementary Figure S4A ), which is commonly observed at the 3′ terminus of many partitiviruses (Vainio et al., 2018). DsPV1 dsRNAs have 76 nt and 137 nt at the 5′ terminus and 97 nt and 116 nt at the 3′ terminus, which constitute the untranslated regions (UTRs) of dsRNA1 and dsRNA2, respectively ( Figure 4 ). DsPV1 shared the conserved terminal sequences, i.e., 5′-CCCAA…-3′ at the 5′ terminus and 5′-…CCCCUUCGGG-3′ at the 3′ terminus with Metarhizium brunneum partitivirus 1 (MbPV1) ( Supplementary Figure S4A ) (accession no. QHB49873) (Wang et al., 2020) ( Table 1 ). DsPV1 dsRNA1 encodes RdRP (573 aa), while DsPV1 dsRNA2 encodes CP ( Figure 4 ). DsPV1 dsRNA1 (1851 bp) encodes RdRP (573 aa, 66.0 kDa) ( Figure 4 ). BLASTp search showed that DsPV1 RdRP exhibits high aa sequence identity to that of Colletotrichum eremochloae partitivirus 1 (CePV1) (66.8%) and to MbPV1 (63.5%) ( Table 1 ). We identified six conserved motifs (III-VIII) in the DsPV1 RdRP ( Supplementary Figure S4B ). DsPV1 dsRNA2 comprises of 1755 nts and was predicted to encode for CP. DsPV1 CP showed 50.7% aa sequence identity with Penicillium aurantiogriseum partitivirus 1 (PaPV1).

An RdRP-based phylogenetic tree showed that DsPV1 belongs to the newly proposed genus “Epsilonpartitivirus” in the family Partitiviridae together with MbPV1, CePV1, PaPV1 and many others (Nerva et al., 2017) ( Figure 4 ). The ICTV-set species threshold for partitiviruses is ≤ 90% aa sequence identity in RdRP and ≤ 80% aa sequence identity in CP (Vainio et al., 2018). Based on this criterion, DsPV1 is considered to represent a new species in this proposed genus.

A contig sequence (contig number 995) showed similarities with members of the genus Victorivirus under family Totivirdae ( Table 1 and Figure 1 ) and corresponded to an undivided dsRNA genome of a victorivirus named Diplodia seriata victorivirus 1 (DsVV1). The complete genome sequence of DsVV1 comprises of 4989 bp in length and possessed two large ORFs (ORF1 and ORF2), which encode for putative CP (708 aa, 74 kDa) and RdRP (826 aa, 91 kDa) ( Figure 5 ). DsVV1 CP and RdRP shared 85.2% and 81.6% aa sequence identity with those of Sphaeropsis sapinea RNA virus 2 (SsRV2) from a pine pathogen Diplodia pinea (Preisig et al., 1998) ( Table 1 ). A multiple nucleotide sequence alignment showed that DsVV1 shared conservation at both termini with reported vivtoriviruses, Magnaporthe oryzae virus 2 (MoV2; accession no. LC573906) and Sphaeropsis sapinea RNA virus 2 (SsRV2; accession no. AF039080) ( Supplementary Figure S5A ). NCBI CDD search showed that ORF1 (CP) and ORF2 (RdRP) were predicted to have a conserved domain of totivirus CP (pfam05518, spanning from 15 aa to 707 aa) and RdRP domains (pfam02123, spanning from 101 to 574 aa), respectively ( Supplementary Figure S5B ). At the junction of ORF1 and ORF2, a conserved tetranucleotide sequence 5′- AUG A-3′ (a potential re-initiation start codon is underlined) was observed ( Figure 5 ). This tetranucleotide could be involved in translational termination and re-initiation of the two ORFs, which can express two proteins CP and RdRP separately, as for those of Helminthosporium victoriae 190S virus (Hv190SV) and some other victroiviruses (Li et al., 2015). A pseudoknot structure important for stop/re-initiation translation is predicted to form directly upstream of the tetranucleotide sequence 5′- AUG A-3′ at a position similar to that in Hv190SV (Li et al., 2015). As a putative stop/restart facilitator, some victoriviruses and hypoviruses have a pentanucleotide sequence at their junction (5’-UAAUG-3’) (Guo et al., 2009; Chiba et al., 2013; Jamal et al., 2019).

Phylogenetic analysis, based on the complete aa sequence of the RdRp (ORF2) of DsVV1 and selected members of the Totiviridae family indicated that DsVV1 made clade with reported victoriviruses, though with a low supporting value 50.2% for victoriviruses ( Figure 5 ). The ICTV species threshold for victoriviruses is set low: <60%. Thus, it is reasonable to classify DsVV1 in the same species as SsRV2, both of which were isolated from Diplodia fungi.

Two contig sequences (contig numbers 218 and 310) showed resemblance with narna-like viruses [(+)ssRNA genome], particularly members of the newly proposed family “Splipalmiviridae” (Sutela et al., 2020; Sato et al., 2022) ( Table 1 ). Splipalmiviruses have been coined for viruses with the split palm (italicized letters are used in the term “splipalmivirus”) domains of RdRP that are encoded separately by different genomic segments (Sutela et al., 2020). The obtained viral sequences likely corresponded to the genomic segments of a splipalmivirus tentatively named Diplodia seriata splipalmivirus 1 (DsSpV1). The complete sequences of DsSpV1 RNA1 and RNA2 shared 12 conserved nucleotides (5′-UUUUGCUUGCGA—3′) and four nucleotides (5′—GUUU-3′) at the 5′ and 3′ termini, respectively ( Supplementary Figure S6A ), which suggest that both contigs represent the genome of a single virus. During data analyses, a third contig sequence (contig number 175) has been detected as a candidate of the third RNA segment of DsSpV1( Table 1 ), however their terminal sequences were still not determined.

DsSpV1 RNA1 (2144 nucleotides, nt) and RNA2 (2168 nt) harbored single ORFs encoding the N-terminal (designated P1, 648 aa and 73.6 kDa) and C-terminal parts (designated P2, 673 aa and 77.2 kDa) of the RdRP, as observed in other splipalmiviruses ( Figure 6 ) (Sato et al., 2022). DsSpV1 RNA1-encoded P1 shared aa sequence identity of 58.3% and 56.5% with those of Erysiphe necator associated narnavirus 11 and Cryphonectria naterciae splipalmivirus 1 (CnSpV1), respectively (Ruiz-Padilla et al., 2021; Sato et al., 2022). DsSpV1 RNA2-encoded P2 shared 45.3% and 44.3% aa sequence identity with those of CnSpV1 and Aspergillus flavus narnavirus 1 (AfuNV2), respectively (Zoll et al., 2018; Sato et al., 2022). DsSpV1-P1 contained RdRP motifs G, F, A, and B, while that encoded by DsSpV1-P2 had RdRP motifs C and D ( Supplementary Figure S6B ). The length of RNA1 was smaller than RNA2, as has been observed in the case of CnSpV1 (Sato et al., 2022). The third DsSpV1 RNA (RNA3 candidate) has a size of ~900 nt, which encodes for a hypothetical protein (designated P3, 173 aa and 18.9 kDa). DsSpV1 P3 showed 41.1% aa sequence identity to the CnSpV1 counterpart (P3). Known splipalmiviruses have different numbers of RNA segments ranging from two to seven (Sutela et al., 2020; Chiba et al., 2021a; Jia et al., 2021; Ruiz-Padilla et al., 2021). AfuNV2 and CnSpV1, closely related to DsSpV1, have three and four RNA segments, respectively. We could not find a fourth or additional segment(s) of DsSpV1 in the NGS data set (data not shown).

To estimate the phylogenetic relationships, the split RdRP sequences encoded by RNA1 and RNA2 of DsSpV1 and other splipalmiviruses, and the representative members of the family Narnaviridae were used to construct phylogenetic trees. The results showed that among the three proposed splipalmiviral genera (Sato et al., 2022), DsSpV1 was placed in a clade with high branch support values of 75.9% (for P1) and 99.9% (for P2) with duasplipalmiviruses ( Figure 6C ). Currently, there is no established species demarcation criteria set by ICTV for splipalmiviruses. DsSpV1 is divergent enough from previously reported duasplipalmiviruses (~58.3% or ~45.3% for the split RdRP aa sequence identity) to classify in a new species in the proposed genus “Duasplipalmivirus”. DsSpV1 will be among rare splipalmivirids that are characterized biologically (see below); its horizontal transmission was confirmed.

Two contig sequences (contig numbers 217 and 38) showed similarities with members of family Botourmiaviridae ( Table 1 ). These two putative mycoviruses were tentatively named as Diplodia seriata botourmiavirus 1 and 2 (DsBOV1 and 2). DsBOV1 and DsBOV2 possess single ORFs that would encode RdRPs of 616 aa (69.7 kDa) and 677 aa (75.7 kDa), respectively ( Figure 8 ). DsBOV2 RNA showed conservation at the termini with already reported Neofusicoccum parvum ourmia-like virus 1 (NpOLV1) (accession no. MK584837) ( Supplementary Figure S7A ) (Nerva et al., 2019b). The DsBOV1 RNA encoded-protein showed aa sequence identity of 45.3% to Sclerotinia sclerotiorum ourmia-like virus 16 RdRP (Jia et al., 2021), while DsBOV2 RNA encoded-protein showed 97.2% aa sequence identity to NpOLV1 RdRP ( Table 1 ). The DsBOV1 and 2 proteins shared eight conserved motifs of RdRP with other reported botourmiaviruses ( Supplementary Figure S7B ).

Phylogenetically, DsBOV1 made clade with scleroliviruses (genus Sclerolivirus) and DsBOV2 made clade with penoliviruses (genus Penolivirus) within family Botourmiaviridae ( Figure 8 ). Based on the ICTV species demarcation criteria (<90% RdRP aa sequence identity) (Ayllon et al., 2020), DsBOV1 should represent a novel species within the genus Sclerolivirus in the family Botourmiaviridae, while DsBOV2 should be classified as a strain belonging to the same species as NpOLV1 within the genus Penolivirus.

A contig sequence (contig number 24) showed sequence similarities with members of the proposed “Ambiguiviridae” and family Tombusviridae (Gilbert et al., 2019). The complete genome of the virus, tentatively named Diplodia seriata ambiguivirus 1 (DsAV1), comprised 3949 nt and harbored two noncontiguous ORFs (ORF1 and ORF2) encoding 372 aa (P1, 41.1 kDa) and 502 aa (P2, 55.7 kDa), which were separated by 160 nt ( Figure 8A ). A database search with BLASTp showed that the putative RdRP aa sequence (ORF2) of DsAV1 shares 97.8% sequence identity to Erysiphe necator associated ambiguivirus 1 (EnAAV1) (accession no. QKN22641, a partial genome sequence with 2856 nt length) (Ruiz-Padilla et al., 2021). DsAV1 ORF1-encoded hypothetical protein shared 51.0% identity to that of Phoma matteucciicola RNA virus 1 (accession no. QNC69246) (Zhou et al., 2021). DsAV1 P2 contained five conserved motifs (IV-VIII) of RdRP domain (pfam00998 spawning from map positions 2 to 456) ( Figure 8A ) ( Supplementary Figure S8A ). An interesting observation was that RdRPs of DsAV1 and other known ambiguiviruses have a GDNA tetrad in motif IV in comparison to the GDD motif found in most (+)ssRNA and dsRNA viruses (REF). Generally, the GDD triplet has a detrimental effect on viability in (+)ssRNA viruses (Vazquez et al., 2000; Das et al., 2021), while dsRNA polymycovirus or ssRNA hadaka virus RdRPs have GDNQ in place of GDD, as mentioned before (Sato et al., 2020a; Sato et al., 2020b Kanhayuwa et al., 2015) ( Supplementary Figure S8A ). A total of four transmembrane domains were observed at the N-terminus of the ORF1 (P1) protein ( Supplementary Figure S8A ), as has been observed in other related viruses (Ai et al., 2016; Canizares et al., 2018). This suggests these domains are important for (+)ssRNA viruses and are reported to play a role in anchoring viral proteins to host membrane, as proposed for Magnaporthe oryzae virus A (Ai et al., 2016) and Verticillium dahliae RNA virus 1 (Canizares et al., 2018).

Phylogenetic analysis with RdRP sequences showed that DsAV1 is grouped together with other reported fungal RNA viruses (ambiguiviruses) and well-separated from other related viruses from plants (family Tombusviridae) and insects (unassigned viruses) ( Figure 8B ). Classification of possible members of the recently proposed family “Ambugiviridae” has not been established. Based on the high degree of sequence similarity (~60.0% RdRP aa sequence identity), both DsAV1 and EnAAV1 should belong to a same novel species in the proposed family. Biological information is unavailable for all reported ambiguiviruses. In this regard, it is of importance that DsAV1 could be eliminated from and re-introduced into the host fungus (see below).

As shown above, D. seriata strain L3 was found to harbor at least eight mycoviruses. To check the possible effect of these viruses on the host fungal strain, we attempted to cure it of the viruses by a few methods. While single spore isolation is often used for this purpose, L3 turned out not to produce spores on PDA media. We then performed hyphal tipping and over 30 subcultures were tested for virus infection. One subculture termed L3ht1, showed a colony morphology different from L3 ( Figure 1 ), and was predicted to have lost a few mycoviruses originally carried in the L3 strain. This could be seen from a dsRNA profile, showing the absence of expected dsRNAs of DsCV1 (3594 bp, 3265 bp, 3058 bp and 3202 bp) and DsPmV1(2439 bp, 2189 bp, 2014 bp, and 1133 bp) in L3ht1 that were present in L3 ( Figure 1 ). Note that some (+)ssRNA viruses, such as splipalmiviruses do not accumulate replicative dsRNA forms in infected fungal host cells (Sato et al., 2022). To confirm which mycoviruses were eliminated in L3ht1, RT-PCR was carried out using the nine primer sets for eight mycoviruses, including split RdRP encoded segments, as listed in Supplementary Table S1 . Consequently, the L3ht1 was shown to harbor five mycoviruses, DsVV1, DsCV1, DsPV1, DsBOV1 and DsBOV2, and to have lost three other mycoviruses, i.e., DsPmV1, DsSpV1 and DsAV1 ( Figure 1 ). For multisegmented RNA viruses such as DsCV1, DsPV1, DsPmV1 and DsSpV1, consistent RT-PCR results were obtained in which different primer sets targeting different genomic segments of a virus of interest were used ( Figure 1 ). Gel electrophoretic analyses of dsRNA and total nucleic acids suggested that whereas L3ht1 accumulated DsVV1 dsRNA more than L3; an opposite trend was observed for DsCV1 ( Figure 1 ). This interesting phenomenon warrants further investigation but suggests antagonistic and synergistic interactions between the eliminated viruses and the viruses whose accumulation was enhanced and suppressed in L3ht1, respectively.

Although we could not isolate virus-free subcultures, phenotypic comparison of the partially cured strain L3ht1 showed milder symptoms, increased colony diameter and regular colony margin, as compared to the original strain L3 (data not shown). Thus, this suggests that one or more of the three mycoviruses is involved in the symptom induction. However, it remains unknown how the respective eliminated mycoviruses contribute to the phenotypic alterations.