Knowledge of the distribution and diversity of RNA viruses is still limited in spite of their possible environmental and epidemiological impacts because RNA virus-specific metagenomic methods have not yet been developed. We herein constructed an effective metagenomic method for RNA viruses by targeting long double-stranded (ds)RNA in cellular organisms, which is a hallmark of infection, or the replication of dsRNA and single-stranded (ss)RNA viruses, except for retroviruses. This novel dsRNA targeting metagenomic method is characterized by an extremely high recovery rate of viral RNA sequences, the retrieval of terminal sequences, and uniform read coverage, which has not previously been reported in other metagenomic methods targeting RNA viruses. This method revealed a previously unidentified viral RNA diversity of more than 20 complete RNA viral genomes including dsRNA and ssRNA viruses associated with an environmental diatom colony. Our approach will be a powerful tool for cataloging RNA viruses associated with organisms of interest.

The study of extracellular DNA viral particles in the ocean is currently one of the most advanced fields of research in viral metagenomic analysis. However, even though the intracellular viruses of marine microorganisms might be the major source of extracellular virus particles in the ocean, the diversity of these intracellular viruses is not well understood. Here, our newly developed method, referred to herein as fragmented and primer ligated dsRNA sequencing (flds) version 2, identified considerable genetic diversity of marine RNA viruses in cell fractions obtained from surface seawater. The RNA virus community appears to cover genome sequences related to more than half of the established positive-sense ssRNA and dsRNA virus families, in addition to a number of unidentified viral lineages, and such diversity had not been previously observed in floating viral particles. In this study, more dsRNA viral contigs were detected in host cells than in extracellular viral particles. This illustrates the magnitude of the previously unknown marine RNA virus population in cell fractions, which has only been partially assessed by cellular metatranscriptomics and not by contemporary viral metagenomic studies. These results reveal the importance of studying cell fractions to illuminate the full spectrum of viral diversity on Earth.

Mycoviruses causing impaired growth and abnormal pigmentation of the host were found in the rice blast fungus, Magnaporthe oryzae. Four dsRNAs, dsRNA 1 (3554 bp), dsRNA 2 (3250 bp), dsRNA 3 (3074 bp) and dsRNA 4 (3043 bp), were detected in isolate S-0412-II 1a of M. oryzae. By picking up single conidia of S-0412-II 1a, cured strains of the fungus were isolated that had completely lost the mycovirus. The cured strains had normal mycelial growth and pigmentation, suggesting that this mycovirus modulates host traits. The buoyant densities of isometric virus particles (~35 nm diameter) containing these dsRNAs in CsCl ranged from 1.37 to 1.40 g cm "3 . The single ORF (3384 nt) of dsRNA 1 encoded a gene product highly homologous to the viral RNA-dependent RNA polymerase of members of the family Chrysoviridae. It is noteworthy that mycovirus S-0412-II 1a was detected not only in host cells but also in culture supernatant. Furthermore, abnormal aggregation of mycelia was observed after adding the mycoviruscontaining culture supernatant to an uninfected strain of M. oryzae and mycoviral dsRNAs were detectable from the aggregated mycelia. This novel dsRNA mycovirus was named Magnaporthe oryzae chrysovirus 1.

A double-stranded RNA (dsRNA) mycovirus was detected in a strain of Alternaria alternata showing impaired growth phenotypes. The A. alternata strain is the Japanese pear pathotype, which produces a host-specific AK-toxin. Sequence analysis of the viral genome dsRNAs revealed that this mycovirus consists of five dsRNAs and is evolutionarily related to members of the family Chrysoviridae; the virus was named Alternaria alternata chrysovirus 1 (AaCV1). AaCV1-ORF2 protein accumulated in dsRNA-high-titer sub-isolates with severely impaired phenotypes; heterologous AaCV1-ORF2 overexpression in Saccharomyces cerevisiae caused growth inhibition. In contrast to this yeast growth inhibition phenomenon, the dsRNA-high-titer isolates displayed enhanced pathogenicity against Japanese pear plants, in accordance with a 13-fold increase in AK-toxin level in one such isolate. These findings indicated that AaCV1 is a novel mycovirus that exhibits two contrasting effects, impairing growth of the host fungus while rendering the host 'hypervirulent' to the plant.

A double-stranded RNA (dsRNA) mycovirus was found in isolate S-0412-II 2a of the rice blast fungus Magnaporthe oryzae. Sequence analysis of the five dsRNA segments (dsRNA1 through dsRNA5) revealed that this mycovirus is closely related to Magnaporthe oryzae chrysovirus 1-A (MoCV1-A), tentatively classified as a member of the Chrysoviridae; therefore, it was named Magnaporthe oryzae chrysovirus 1-B (MoCV1-B). Virus particles were spherical and composed of the ORF1, ORF3 and ORF4 proteins. MoCV1-B-infected isolate S-0412-II 2a showed a more severe impaired phenotype than the MoCV1-A-infected isolate. In a virus-cured isolate, normal growth was restored, implied that MoCV1-B could be involved in this observed phenotype. An unanticipated result was the occurrence of a fungal isolate lacking dsRNA5. The nonessential dsRNA5 had higher sequence identity (96%) with dsRNA5 of MoCV1-A than with the other dsRNA segments (71-79%), indicating that dsRNA5 could be a portable genomic element between MoCV1-A and MoCV1-B.

Magnaporthe oryzae chrysovirus 1 (MoCV1), which is associated with an impaired growth phenotype of its host fungus, harbors four major proteins: P130 (130 kDa), P70 (70 kDa), P65 (65 kDa), and P58 (58 kDa). N-terminal sequence analysis of each protein revealed that P130 was encoded by double-stranded RNA1 (dsRNA1) (open reading frame 1 [ORF1] 1,127 amino acids [aa]), P70 by dsRNA4 (ORF4; 812 aa), and P58 by dsRNA3 (ORF3; 799 aa), although the molecular masses of P58 and P70 were significantly smaller than those deduced for ORF3 and ORF4, respectively. P65 was a degraded form of P70. Full-size proteins of ORF3 (84 kDa) and ORF4 (85 kDa) were produced in Escherichia coli. Antisera against these recombinant proteins detected full-size proteins encoded by ORF3 and ORF4 in mycelia cultured for 9, 15, and 28 days, and the antisera also detected smaller degraded proteins, namely, P58, P70, and P65, in mycelia cultured for 28 days. These full-size proteins and P58 and P70 were also components of viral particles, indicating that MoCV1 particles might have at least two forms during vegetative growth of the host fungus. Expression of the ORF4 protein in Saccharomyces cerevisiae resulted in cytological changes, with a large central vacuole associated with these growth defects. MoCV1 has five dsRNA segments, as do two Fusarium graminearum viruses (FgV-ch9 and FgV2), and forms a separate clade with FgV-ch9, FgV2, Aspergillus mycovirus 1816 (AsV1816), and Agaricus bisporus virus 1 (AbV1) in the Chrysoviridae family on the basis of their RdRp protein sequences.

By identifying variations in viral RNA genomes, cutting-edge metagenome technology has potential to reshape current concepts about the evolution of RNA viruses. This technology, however, cannot process low-homology genomic regions properly, leaving the true diversity of RNA viruses unappreciated. To overcome this technological limitation, we applied an advanced method, Fragmented and Primer-Ligated Double-stranded (ds) RNA Sequencing (FLDS), to screen RNA viruses from 155 fungal isolates, which allowed us to obtain complete viral genomes in a homology-independent manner. We created a high-quality catalog of 19 RNA viruses (12 viral species) that infect Aspergillus isolates. Among them, nine viruses were not detectable by the conventional methodology involving agarose gel electrophoresis of dsRNA, a hallmark of RNA virus infections. Segmented genome structures were determined in 42 per cent of the viruses. Some RNA viruses had novel genome architectures; one contained a dual methyltransferase domain and another had a separated RNA-dependent RNA polymerase (RdRp) gene. A virus from a different fungal taxon (Pyricularia) had an RdRp sequence that was separated on different segments, suggesting that a divided RdRp is widely present among fungal viruses, despite the belief that all RNA viruses encode RdRp as a single gene. These findings illustrate the previously hidden diversity and evolution of RNA viruses, and prompt reconsideration of the structural plasticity of RdRp.

A novel dsRNA virus was identified from the arboreal ant Camponotus yamaokai. The complete nucleotide sequence analysis of the virus revealed that the virus consisted of 5704 bp with two ORFs. ORF1 (3084 nt) encoded a putative capsid protein. ORF2 (1977 nt) encoded a viral RNA-dependent RNA polymerase (RdRp). ORF2 could be translated as a fusion with the ORF1 product by a 21 frameshift in the overlapping ORF1. Phylogenetic analyses based on the RdRp revealed that the virus from C. yamaokai was most likely a novel totivirus, but it was not closely related to the previously known totiviruses in arthropods. Transmission electron microscopy revealed isometric virus particles of ,30 nm diameter in the cytoplasm, which was consistent with the characteristics of the family Totiviridae. The virus was detected by reverse transcription-PCR in all caste members and developmental stages of ants, including eggs, larvae, pupae, adult workers, alates (male and female) and queens. To our knowledge, this is the first report of a member of the family Totiviridae in a hymenopteran; the virus was designated Camponotus yamaokai virus.