Identification, Molecular Characterization, and Biology of a Novel Quadrivirus Infecting the Phytopathogenic Fungus Leptosphaeria biglobosa

Here we report the molecular characterisation of a novel dsRNA virus isolated from the filamentous, plant pathogenic fungus Leptosphaeria biglobosa and known to cause significant alterations to fungal pigmentation and growth and to result in hypervirulence, as illustrated by comparisons between virus-infected and -cured isogenic fungal strains. The virus forms isometric particles approximately 40–45 nm in diameter and has a quadripartite dsRNA genome structure with size ranges of 4.9 to 4 kbp, each possessing a single ORF. Sequence analysis of the putative proteins encoded by dsRNAs 1–4, termed P1–P4, respectively, revealed modest similarities to the amino acid sequences of equivalent proteins predicted from the nucleotide sequences of known and suspected members of the family Quadriviridae and for that reason the virus was nominated Leptosphaeria biglobosa quadrivirus-1 (LbQV-1). Sequence and phylogenetic analysis using the P3 sequence, which encodes an RdRP, revealed that LbQV-1 was most closely related to known and suspected quadriviruses and monopartite totiviruses rather than other quadripartite mycoviruses including chrysoviruses and alternaviruses. Of the remaining encoded proteins, LbQV-1 P2 and P4 are structural proteins but the function of P1 is unknown. We propose that LbQV-1 is a novel member of the family Quadriviridae.


Introduction
Fungal viruses or mycoviruses are ubiquitous and have been detected in all major groups of fungi including members of the divisions Ascomycota, Basidiomycota and Glomeromycota [1]. There are at least five established mycovirus families, and one established genus, whose members have multi-segmented dsRNA genomes: families Reoviridae, Partitiviridae, Chrysoviridae, Quadriviridae, and Megabirnaviridae, and the genus Botybirnavirus [1], which possess genome segment numbers of 11-12, 2, 4, 4, 2, and 2, respectively. In addition, there are two proposed families Alternaviridae and Polymycoviridae [2][3][4], which possess 4 and 4-8 genomic components, respectively. All of these viruses are encapsidated in rigid, spherical virus particles apart from polymycoviruses, which are not conventionally encapsidated [4,5]. Although many mycoviruses have no or few obvious effects on their host fungi, some induce phenotypic alterations including hypovirulence (attenuated virulence) or hypervirulence (enhanced virulence).
Phoma stem canker (blackleg) is an internationally important disease of oilseed rape (Brassica napus, canola, rapeseed), causing serious losses in Europe, Australia, and North America. For instance UK losses of ca. £100 million per season have been estimated using national disease

Fungal Strains, Culture Conditions, and Dsrna Extraction
The Chinese L. biglobosa isolate W10 was grown on V8 agar plates and incubated for five days at 20 • C in darkness to produce confluent cultures. The isolate was confirmed as L. biglobosa following sub-culturing on potato dextrose agar (PDA) plates on the basis of morphological phenotype and by PCR amplification of the ribosomal RNA region incorporating the internal transcribed spacer (ITS) regions and the 5.8S rRNA gene [10]. The dsRNA elements known to be present in W10 [9] were purified by LiCl fractionation [11] and gel electrophoretic analysis of dsRNA was done according to standard protocols [4]. A virus-cured isogenic line was obtained using a combination of treatment with cycloheximide and hyphal tipping [4,9].

Virus Purification and Transmission Electron Microscopy (TEM)
Mycelia grown in PD broth with shaking at 25 • C were harvested five days after inoculation using sterile Miracloth (Merck Millipore, Danvers, MA, USA) and rapidly frozen in liquid N 2 and kept at −80 • C until processing. Virus purification was performed as described before [5]. Purified virus was negatively stained with 1% uranyl acetate on carbon-coated 400-mesh copper grids and examined using a FEI Tecnai 20 transmission electron microscope. Isolation of dsRNA from purified virus was performed using phenol/chloroform treatment. DNase I (Promega, Madison, WI, USA) and S1 nuclease (Promega, Madison, WI, USA) treatments of purified dsRNAs were performed according to the manufacturer's instructions.

cDNA Cloning and RNA Ligase-Mediated Rapid Amplification of cDNA Ends (RLM-RACE)
After electrophoretic separation on agarose gels dsRNAs were used, either collectively or individually, as templates for cDNA synthesis and PCR amplification of products using random priming, sequence-specific priming and RLM-RACE, which were subsequently cloned and sequenced [12,13]. At least three different clones were sequenced, covering the same part of each segment of the viral genome.

Sequence and Structure Bioinformatic Analysis
Sequence similarity searches of the GenBank, Swissprot and EMBL databases were conducted using the BLAST programs [14]. Searches for protein motifs were conducted using the Pfam database [15]. For phylogenetic analysis the RdRP protein sequences were aligned with MUSCLE as implemented by MEGA 6 [16], the alignment was improved manually and all positions with less than 30% site coverage were eliminated. Maximum likelihood phylogenetic trees were constructed in MEGA 6 using the LG + G + I substitution model. Structural models of the putative capsid proteins encoded by dsRNA 2 and dsRNA 4 were constructed using Phyre2 [17] and molecular graphics images Following extraction of dsRNA from the Chinese L. biglobosa isolate W10, it was discovered that it contained four dsRNA elements 4.9-4.0 kbp in size designated as dsRNA 1 to dsRNA 4, based on their gel mobility. Similar electrophoretic patterns have been reported for 10 more L. biglobosa isolates originating from China and the UK [9]. A representative agarose electrophoresis of the dsRNA elements isolated from L. biglobosa isolate W10 is shown in Figure 1A. As described previously, isolate W10 was successfully freed of infection with mycovirus dsRNAs using a combination of cycloheximide treatment and hyphal tipping to generate strain W10-VF-1 which was confirmed by electrophoretic isolation of dsRNA, Northern blotting and RT-PCR [9]. Comparison of the colony morphologies of the isogenic L. biglobosa W10 and W10-VF-1 lines revealed significantly different phenotypes associated with virus infection ( Figure S1). A comparison of the mycelial growth rates of the two isogenic lines by examining radial expansion growth curves and biomass production emphatically demonstrated that infection of L. biglobosa with the mycovirus results in increased growth rate, as illustrated in Figure S1, and concomitant hypervirulence of the host fungus, phenomena which have been investigated in detail elsewhere [9].

Virus Particles, Genomic dsRNAs, and Their Sequences
Purified virus particles had an icosahedral structure and particle diameter was estimated to be ca. 40-45 nm ( Figure 1B) and most particles were peripherally penetrated by the stain giving them a doughnut-like appearance. The chemical nature of purified preparations of dsRNA, either prepared directly from mycelia or isolated from virions, was confirmed by its insusceptibility to DNase 1 and S1 nuclease. The agarose gel electrophoretic patterns of the four dsRNA species with size estimates of 4.9, 4.4, 4.4, and 4.0 kbp were reminiscent of those seen for two members of the family Quadriviridae, Rosellinia necatrix quadrivirus strains W1118 and W1075, where the accumulation of dsRNA 1 and dsRNA 4 were also consistently less than those of dsRNA 2 and dsRNA 3 ( Figure 1A). Based on these findings, the virus under investigation was nominated Leptosphaeria biglobosa quadrivirus-1 (LbQV-1).
A random cDNA library prepared from a mixture of LbQV-1 dsRNAs 1 to 4 was constructed. Approximately 50 clones with cDNA inserts of 500-1000 bp were selected, sequenced and assembled into contigs. The contigs covered almost the entire length of each dsRNA segment and gaps were filled by RT-PCR using oligonucleotides based on known sequences to generate amplicons, which were subsequently cloned and sequenced [4]. Terminal sequences of the four dsRNAs were determined by 3 -RLM-RACE and fragments for each end of both strands of the dsRNAs were cloned and sequenced. Because of heterogeneity at the extreme termini at least 12 clones were analysed, together with cloned internal amplicons and amplicons linking RACE clones with the assembled contigs. A random cDNA library prepared from a mixture of LbQV-1 dsRNAs 1 to 4 was constructed. Approximately 50 clones with cDNA inserts of 500-1000 bp were selected, sequenced and assembled into contigs. The contigs covered almost the entire length of each dsRNA segment and gaps were filled by RT-PCR using oligonucleotides based on known sequences to generate amplicons, which were subsequently cloned and sequenced [4]. Terminal sequences of the four dsRNAs were determined by 3'-RLM-RACE and fragments for each end of both strands of the dsRNAs were cloned and sequenced. Because of heterogeneity at the extreme termini at least 12 clones were analysed, together with cloned internal amplicons and amplicons linking RACE clones with the assembled contigs.
The complete nucleotide sequences of the dsRNA 1-4 segments are 4917, 4543, 4490, and 4048 bp in size and each respectively contains an ORF on the plus-strand, potentially encoding proteins 1559 amino acids (aa; 172 kDa), 1383 aa (152 kDa), 1367 aa (153 kDa), and 1111 aa (120 kDa) in size, flanked by 5'-and 3'-untranslated regions (UTRs). Following the nomenclature adopted for other quadriviruses [7,8], these four proteins were nominated P1-P4. The 5'-UTRs of the four segments are respectively 48, 69, 42 and 58 bp long while the 3'-UTRs are 189, 322, 344 and 654 bp long ( Figure 2; Table 1). Irrespective of which LbQV-1 dsRNA was used as template all 3'-RLM-RACE clones corresponding to the 5'-terminus of the plus-sense strand had sequences 5'-N/ACGA-( Figure 3) and are identical to the 5'-terminal sequences of the prototype Rosellinia necatrix quadrivirus strains W1118 and W1075 (RnQV1-W118 and W1075; [7,8]) and four large (L) dsRNAs isolated from plants infected with Amasya cherry disease (ACD-L), which probably represent the incomplete genomes of two closely-related quadriviruses of cherry or fungal origin [19]. However, the 3'-terminal sequences of the plus-sense strand of the LbQV-1 dsRNAs were highly variable and were extremely heterogeneous ( Figure 3). Interestingly, however, there was significant homology between regions upstream of the 3'-termini of the LbQV-1 dsRNAs and these regions also contained characteristic sequence motifs including some small stem loop structures that were also identified in the 3'-UTRs The complete nucleotide sequences of the dsRNA 1-4 segments are 4917, 4543, 4490, and 4048 bp in size and each respectively contains an ORF on the plus-strand, potentially encoding proteins 1559 amino acids (aa; 172 kDa), 1383 aa (152 kDa), 1367 aa (153 kDa), and 1111 aa (120 kDa) in size, flanked by 5 -and 3 -untranslated regions (UTRs). Following the nomenclature adopted for other quadriviruses [7,8], these four proteins were nominated P1-P4. The 5 -UTRs of the four segments are respectively 48, 69, 42 and 58 bp long while the 3 -UTRs are 189, 322, 344 and 654 bp long ( Figure 3; Table 1). Irrespective of which LbQV-1 dsRNA was used as template all 3 -RLM-RACE clones corresponding to the 5 -terminus of the plus-sense strand had sequences 5 -N/ACGA-( Figure 4) and are identical to the 5 -terminal sequences of the prototype Rosellinia necatrix quadrivirus strains W1118 and W1075 (RnQV1-W118 and W1075; [7,8]) and four large (L) dsRNAs isolated from plants infected with Amasya cherry disease (ACD-L), which probably represent the incomplete genomes of two closely-related quadriviruses of cherry or fungal origin [19]. However, the 3 -terminal sequences of the plus-sense strand of the LbQV-1 dsRNAs were highly variable and were extremely heterogeneous ( Figure 4). Interestingly, however, there was significant homology between regions upstream of the 3 -termini of the LbQV-1 dsRNAs and these regions also contained characteristic sequence motifs including some small stem loop structures that were also identified in the 3 -UTRs of two isolates of RnQV1 and ACD-L dsRNAs 1-4. Heterogeneity at the extreme 3'termini is also a feature of the genomes of both RnQV1 and ACD-L dsRNAs 1-4.
Viruses 2019, 11, 9 5 of 10 the presence of these proteins in the purified LbQV-1 preparation. Further PMF analysis of the proteins displayed on SDS-PAGE was not pursued but, by analogy with the full molecular characterization of the prototype quadrivirus RnQV1 [7,8], it can be assumed that the smaller proteins represent proteolytic degradation products of both proteins encoded by LbQV-1 dsRNA 2 and dsRNA 4 plus some host proteins. The known cryo-EM structure at 3.7 Å resolution of RnQV1 (PDBID 5ND1) was used as template for predicting the structure of the LbQV-1 putative capsid proteins and their interactions ( Figure 4B).

Amino Acid Similarities and Phylogenetic Analysis
PSI-BLAST searches of the amino acid sequences of LbQV-1 P1-P4 generally showed modest similarities to the amino acid sequences of equivalent proteins predicted from the nucleotide sequences of known and suspected quadrivirus dsRNAs (Table 1). LbQV-1 P1 was distantly related to RnQV1-W1075 and RnQV1-W1118 (22% and 20% identity), ACD-L dsRNA 1 and ACD-L dsRNA 2 (26% and 27% identity) P1 equivalents and the sequence of one dsRNA component of a grapevine associated totivirus 1 (GaTV-1; 28% identity; [20]). Similarly a PSI-BLAST search with P2 showed only 23% and 22% amino acid sequence identity to equivalent proteins of RnQV1-W1118 and RnQV1-W1075. The results of PSI-BLAST searches with LbQV-1 P3 showed that it shared significant sequence similarity to known and suspected quadriviruses and was more closely related to the putative RdRPs encoded by ACD-L dsRNA 3 and -L dsRNA 4 (44% and 45% identity) and cherry chlorotic rusty spot (CCRS) L dsRNA 3 and -L dsRNA 4 (43% and 44% identity) as compared to RnQV1-W1118 and RnQV1-W1075 (35% and 36% identity). The PSI-BLAST search also showed that

Assignment of Structural Protein Genes
SDS-PAGE analysis of purified LbQV-1 showed the presence of two major proteins corresponding to ca. 130 kDa and 100 kDa as well as minor proteins of ca. 55 kDa, 40 kDa, and 35 kDa (Figure 2A). It is unclear whether all the observed proteins are viral since some of them may be derived from the host. Following PMF, four and six tryptic peptides were found to be derived from the N-terminal amino acid sequence encoded by dsRNAs 2 and 4, respectively (Table S1), verifying the presence of these proteins in the purified LbQV-1 preparation. Further PMF analysis of the proteins displayed on SDS-PAGE was not pursued but, by analogy with the full molecular characterization of the prototype quadrivirus RnQV1 [7,8], it can be assumed that the smaller proteins represent proteolytic degradation products of both proteins encoded by LbQV-1 dsRNA 2 and dsRNA 4 plus some host proteins. The known cryo-EM structure at 3.7 Å resolution of RnQV1 (PDBID 5ND1) was used as template for predicting the structure of the LbQV-1 putative capsid proteins and their interactions ( Figure 2B).

Amino Acid Similarities and Phylogenetic Analysis
PSI-BLAST searches of the amino acid sequences of LbQV-1 P1-P4 generally showed modest similarities to the amino acid sequences of equivalent proteins predicted from the nucleotide sequences of known and suspected quadrivirus dsRNAs (Table 1). LbQV-1 P1 was distantly related to RnQV1-W1075 and RnQV1-W1118 (22% and 20% identity), ACD-L dsRNA 1 and ACD-L dsRNA 2 (26% and 27% identity) P1 equivalents and the sequence of one dsRNA component of a grapevine associated totivirus 1 (GaTV-1; 28% identity; [20]). Similarly a PSI-BLAST search with P2 showed only 23% and 22% amino acid sequence identity to equivalent proteins of RnQV1-W1118 and RnQV1-W1075. The results of PSI-BLAST searches with LbQV-1 P3 showed that it shared significant sequence similarity to known and suspected quadriviruses and was more closely related to the putative RdRPs encoded by ACD-L dsRNA 3 and -L dsRNA 4 (44% and 45% identity) and cherry chlorotic rusty spot (CCRS) L dsRNA 3 and -L dsRNA 4 (43% and 44% identity) as compared to RnQV1-W1118 and RnQV1-W1075 (35% and 36% identity). The PSI-BLAST search also showed that LbQV-1 P3 shares modest amino acid sequence identities of 20% to 30% to RdRPs from other mycoviruses within the families Totiviridae, Chrysoviridae, and Megabirnaviridae and an alignment of representative RdRP domains was generated ( Figure S2; [21]). A Pfam search showed that LbQV-1 P3 belongs to the viral RdRP family (RdRp_4). Sequences of representative RdRPs from fungal dsRNA viruses were used to construct maximum likelihood phylogenetic trees. As shown in Figure 5, the tree placed LbQV-1 P3 together with RdRPs encoded by ACD-L dsRNA 3, -L ds RNA 4, RnQV1-W1118, RnQV1-W1075, and GaTV-1 in a separate clade from other fungal dsRNAs. A PSI-BLAST search with LbQV-1 P4 showed only 24% amino acid sequence identity to equivalent proteins of both RnQV1-W1118 and RnQV1-W1075.

Discussion
The Quadriviridae is a family of non-enveloped spherical viruses with quadripartite dsRNA genomes of 3.5-5.0 kbp, comprising 16.8-17.1 kbp in total [22], which includes one genus with two strains of a single prototype species RnQV1 [7,8]. Here we have characterised a novel quadrivirus LbQV-1 from the filamentous fungus L. biglobosa, which shares some, but not all, biological and molecular features of the Quadriviridae. For instance, unlike RnQV1, LbQV-1 can be transmitted intracellularly via anastomosis and vertically through conidia, and it elicits alterations to the growth and pathogenicity of L. biglobosa resulting in the uncommon occurrence of hypervirulence ( Figure  S1; [9]). As with RnQV1, it is not known whether LbQV-1 has an extracellular phase. Transcription and replication are presumably carried out in a spherical particle by virion-associated RdRP.

Discussion
The Quadriviridae is a family of non-enveloped spherical viruses with quadripartite dsRNA genomes of 3.5-5.0 kbp, comprising 16.8-17.1 kbp in total [22], which includes one genus with two strains of a single prototype species RnQV1 [7,8]. Here we have characterised a novel quadrivirus LbQV-1 from the filamentous fungus L. biglobosa, which shares some, but not all, biological and molecular features of the Quadriviridae. For instance, unlike RnQV1, LbQV-1 can be transmitted intracellularly via anastomosis and vertically through conidia, and it elicits alterations to the growth and pathogenicity of L. biglobosa resulting in the uncommon occurrence of hypervirulence ( Figure S1; [9]). As with RnQV1, it is not known whether LbQV-1 has an extracellular phase. Transcription and replication are presumably carried out in a spherical particle by virion-associated RdRP.
Similar to RnQV1, LbQV-1 particles consist of rigid spherical virions ca. 40-45 nm in diameter ( Figure 1B) which are encapsidated by two major structural proteins P2 and P4 encoded by dsRNA 2 and dsRNA 4, respectively ( Figure 3). In both RnQV1 strains W1075 and W1118, the P2 and P4 proteins, Viruses 2019, 11, 9 8 of 10 which are more than 80% identical, form an asymmetric hetero-dimer subunit, and 60 of these build a T = 1 capsid [23]. We assume that LbQV-1 possesses a similar virion structure, having demonstrated moderate sequence identity between the equivalent proteins involved (Figure 3; Table 1). It is not known whether LbQV-1 P2 and P4 have acquired new functions through the insertion of complex domains at preferential insertion sites on the capsid outer surface as has been demonstrated for RnQV1 [23][24][25]. Both RnQV1 strains and LbQV-1 possess four linear dsRNA genome segments, termed dsRNA 1 to dsRNA 4 in a decreasing order of length from 5.0-3.5 kbp and each contains a single large ORF on the positive-sense strand of each dsRNA genomic segment (Figure 3). It is anticipated that each genomic dsRNA of LbQV-1 and both strains of RnQV1 are encapsidated separately [7,8]. LbQV-1, RnQV1-W1118, and RnQV1-W1075 genomic dsRNAs all had the same 5 -terminal motif sequence of 5 -N/ACGA- (Figure 4). The same motif is also present in the ACD-L dsRNAs, which probably represent the dsRNA 1 and 3 segments of two closely-related quadriviruses of cherry or fungal origin [19]. The lengths of all of the 5 -UTRs of fully sequenced quadrivirus dsRNAs are relatively small ranging in size from 35 to 106 nucleotides and those in both isolates of RnQV1 and LbCV-1 dsRNA 2 and 4 segments contain (CAA) n repeats which are presumably translational enhancers, as reported for chrysoviruses [26]. Examination of the 3 -UTRs of the four dsRNA segments of two strains of RNQV1, LbQV-1 and ACD-L L reveal heterogeneity in the very terminal end sequences ( Figure 4). However, as compared to RNQV1 and ACD-L the sequence heterogeneity found in all four LbQV-1 dsRNAs is extensive and regions of identity between them were only discovered some distance upstream of the very terminal end sequences ( Figure 4). We believe that these observations for LbQV-1 and the discovery of an extraordinarily 645 nt long 3 -UTR for dsRNA 4 are not artefactual as we obtained the same results with a large number of representative clones in several different experiments. The occurrence of a large number of repeat sequences with the motif CA(A) n = 2-8 towards the 5 -terminus of the LbQV-1 dsRNA 4 3 -UTR is another novel feature for a quadrivirus the significance of which is unknown.
Sequence and phylogenetic analysis using the P3 sequence, which encoded the RdRP, revealed that LbQV-1 was most closely related to known and suspected quadriviruses and members of the genus Totivirus in the family Totiviridae [7,8] rather than other quadripartite mycoviruses, including chrysoviruses and alternaviruses ( Figure 5; Figure S2; [7,8]). Of the remaining LbQV-1 encoded proteins, whereas LbQV-1 P2 and P4 are structural proteins the function of P1 is unknown. Based on the findings presented in this report, we propose that LbQV-1 is a novel member of the genus Quadrivirus in the family Quadriviridae.