Description of a Novel Mycovirus in the Phytopathogen Fusarium culmorum and a Related EVE in the Yeast Lipomyces starkeyi

A new mycovirus was found in the Fusarium culmorum strain A104-1 originally sampled on wheat in Belgium. This novel virus, for which the name Fusarium culmorum virus 1 (FcV1) is suggested, is phylogenetically related to members of the previously proposed family ‘’Unirnaviridae’’. FcV1 has a monopartite dsRNA genome of 2898 bp that harbors two large non-overlapping ORFs. A typical -1 slippery motif is found at the end of ORF1, advocating that ORF2 is translated by programmed ribosomal frameshifting. While ORF2 exhibits a conserved replicase domain, ORF1 encodes for an undetermined protein. Interestingly, a hypothetically transcribed gene similar to unirnaviruses ORF1 was found in the genome of Lipomyces starkeyi, presumably resulting from a viral endogenization in this yeast. Conidial isolation and chemical treatment were unsuccessful to obtain a virus-free isogenic line of the fungal host, highlighting a high retention rate for FcV1 but hindering its biological characterization. In parallel, attempt to horizontally transfer FcV1 to another strain of F. culmorum by dual culture failed. Eventually, a screening of other strains of the same fungal species suggests the presence of FcV1 in two other strains from Europe.


Introduction
Mycoviruses have been described in all major fungal taxa and more extensively in genera with economic or medical importance [1]. Amongst those, the genus Fusarium has received much attention leading to the discovery of many mycoviruses, especially in Fusarium oxysporum and Fusarium graminearum [2,3]. In this study, we characterize for the first time a virus from the phytopathogenic species F. culmorum.
F. culmorum belongs to the fungal complex causing Fusarium head blight and Fusarium foot and root rot on several plant species [4]. In addition to its pathogenicity, F. culmorum is known to produce mycotoxins that contaminate crop products, hampering their consumption by humans and cattle [5,6]. Currently, the control of this pathogen relies on the extensive use of fungicides. However, besides the potentially detrimental effect of such chemicals on the environment, F. culmorum has developed related resistance [7]. Thus, there is a need for alternative control strategies. Mycoviruses conferring hypovirulence have been proposed for this purpose [8], motivating the search for such potential biocontrol agents.
The majority of known mycoviruses are made of dsRNA [1], hence screening for dsRNA in fungal mycelia is a fast method to discover new mycoviruses. Despite the fact that dsRNA molecules have been previously detected in F. culmorum, no viral sequence has been described yet [9,10]. By screening Viruses 2020, 12, 523 3 of 13 (1000 replicates) [23]. The tree was visualized using MEGA-X [24]. Alternatively, a phylogenetic tree was also obtained using MrBayes (with rtREV model) [25] and visualized using MEGA-X.

Evaluation of Vertical Transmission
The F. culmorum strain A104-1 was grown on synthetic nutrient-poor agar (SNA) medium for 10 days at room temperature (RT). Macroconidia were harvested by adding and collecting 1 mL of sterile water containing Tween-20 (one drop per mL) to the cultures. The suspension was then poured on 3% (w/v) agar plates with serial dilutions and incubated O/N at RT. Sixty germinated macrononidia were individually transferred onto potato dextrose agar (PDA) plates and subsequently incubated for 10 days at RT. Total RNA was extracted from the produced mycelia using Trizol reagent (Invitrogen, Waltham, MA, USA) according to the manufacturer's instructions. Then, RNA was subjected to RT-PCR using the specific primers 5 -GGCGTTAAGGCGTTGGGCTACT-3 (positions 1701-1722) and 5 -CCTCGACCCGCAACGCC-3 (positions 2200-2184) with the enzymes already described.

Chemical Treatment Assay
Cycloheximide and rifampicin were used in the aim to obtain a virus-free isogenic line of the fungus. Briefly, 20 mL of carboxymethyl cellulose (CMC) liquid medium supplemented with either cycloheximide (1 mg/L) or rifampicin (1 mg/L) were inoculated with plugs of mycelia and incubated at 120 rpm at 25 • C in darkness. After 5 days of culture, conidia were collected and poured onto PDA supplemented with the same chemicals. Fourteen individual fungal colonies were isolated for each treatment and sub-cultured for at least three serial passages via hyphal-tipping on similar media. Afterwards, mycelia were grown on PDA without chemical and screened for the presence of FcV1 by RT-PCR as already described.

Horizontal Transmission Assay
A mycelia plug of the fungal strain A104-1 was grown on PDA next to a mycelia plug from the F. culmorum strain P1P2 that was obtained from the strain UK99 after repeated passages on tebuconazole supplemented media [7]. Seven days after co-culture, 6 mycelia plugs located on the confrontation zone were transferred onto PDA containing tebuconazole (1 mg/L) for selection. Plugs from fast growing mycelia sectors were then transferred onto PDA and screened for the presence of FcV1 by RT-PCR as already described.

Prevalence of FcV1 in Other F. culmorum Strains
Fifty-three strains of F. culmorum originated from diverse locations (Table S3) were grown on PDA for 10 days at RT. Following this, the produced mycelia were screened for the presence of FcV1 by RT-PCR, as described above. Strains that were positive were further analysed by dsRNA analysis.

Genome Sequence Analysis of FcV1
Following a dsRNA extraction on the F. culmorum strain A104-1, one major band of ca. 3 kb was visible on agarose gel in addition to fungal DNA and RNA ( Figure 1A). The dsRNA nature of this band was proven by resistance to DNase I and S1 nuclease ( Figure S1). An RT-PCR on this dsRNA using a degenerate primer yielded two amplicons ( Figure S2) that allowed to start the sequencing. After completing the latter, the full sequence was determined to be 2898 bp in length. Due to its dsRNA nature, the sequence was assigned to the genome of a novel virus for which the name Fusarium culmorum virus 1 (FcV1) is proposed. The full sequence has been deposited in the Genbank database under the accession number MN187541. A BLASTn search for the total nucleotide sequence resulted in significant hits (E-value ≤ 4 × 10 −10 , 5-80% coverage) corresponding to members of the proposed family "Unirnaviridae". Like other members of this group (see Table 1 for viruses abbreviations and Viruses 2020, 12, 523 4 of 13 characteristics), FcV1 genome harbors two large non-overlapping ORF (ORF1 and 2) on the positive strand, at positions 15-959 and 974-2794, respectively ( Figure 1B). ORF1 encodes for a putative protein of 314 aa with an estimated mass of 34.02 kDa. The typical -1 slippery motif G_GAU_UUU [26] is found at the 3 end of ORF1 (positions 950-956), suggesting that ORF2 is translated by programmed ribosomal frameshifting (PRF). This PRF hypothesis is supported for unirnaviruses by the fact that a protein with homology to both ORF1 and ORF2 was recovered during an investigation on ChNRV1 translation products [27]. In the case of FcV1, the hypothetically produced protein ORF1+2p contains 926 aa, with an estimated mass of 103.53 kDa.
for a putative protein of 314 aa with an estimated mass of 34.02 kDa. The typical -1 slippery motif G_GAU_UUU [26] is found at the 3′ end of ORF1 (positions 950-956), suggesting that ORF2 is translated by programmed ribosomal frameshifting (PRF). This PRF hypothesis is supported for unirnaviruses by the fact that a protein with homology to both ORF1 and ORF2 was recovered during an investigation on ChNRV1 translation products [27]. In the case of FcV1, the hypothetically produced protein ORF1+2p contains 926 aa, with an estimated mass of 103.53 kDa.
The 5′-and 3′-UTR of FcV1 genome are 14 and 104 bp long, respectively. An alignment of unirnaviruses 5′-termini evidenced the conserved motif GGAAA/UA/UUA/U in FcV1, UvURVHNND1, BbNV1, ChNRV1, AlRV1, ClV1 and ThV1 ( Figure 1C). An alignment of unirnaviruses 3′-termini only highlights the presence of series of C residues of variable lengths (3 to 6) in FcV1, UvURVHNND1, BbNV1, ChNRV1, AlRV1 and ThV1 ( Figure 1C). The termini of other unirnaviruses do not contain these motifs, probably due to incomplete sequencing. Interestingly, FcV1 3′-UTR is predicted to fold into stable stem-loop structures ( Figure 1D, with minimum optimal energy ΔG= −37.32 kcal/mol). This prediction was also evidenced for several related viruses [28] and such secondary structures might be of biological importance in terms of replication and/or translation, a common feature of viral UTRs.  The 5 -and 3 -UTR of FcV1 genome are 14 and 104 bp long, respectively. An alignment of unirnaviruses 5 -termini evidenced the conserved motif GGAAA/UA/UUA/U in FcV1, UvURVHNND1, BbNV1, ChNRV1, AlRV1, ClV1 and ThV1 ( Figure 1C). An alignment of unirnaviruses 3 -termini only highlights the presence of series of C residues of variable lengths (3 to 6) in FcV1, UvURVHNND1, BbNV1, ChNRV1, AlRV1 and ThV1 ( Figure 1C). The termini of other unirnaviruses do not contain these motifs, probably due to incomplete sequencing. Interestingly, FcV1 3 -UTR is predicted to fold into stable stem-loop structures ( Figure 1D, with minimum optimal energy ∆G= −37.32 kcal/mol). This prediction was also evidenced for several related viruses [28] and such secondary structures might be of biological importance in terms of replication and/or translation, a common feature of viral UTRs.  1 Apparently not fully sequenced. 2 No -1 PRF motif found.

Analysis of FcV1 ORF1p and Retrieving of ORF1p Homologs
As expected by nucleotide similarities, a BLASTp search for FcV1 ORF1p resulted in significant hits (E-value ≤ 5 × 10 −50 , ≥ 69% coverage) corresponding to the predicted ORF1p of other unirnaviruses. All those proteins are similar in size (314-394 aa, except PmRV1 for which ORF1p might be not fully sequenced given the absence of the conserved 5 -terminus) and show 34-61% pairwise identity ( Figure 2).

Analysis of FcV1 ORF1p and Retrieving of ORF1p Homologs
As expected by nucleotide similarities, a BLASTp search for FcV1 ORF1p resulted in significant hits (E-value ≤ 5 × 10 -50 , ≥ 69% coverage) corresponding to the predicted ORF1p of other unirnaviruses. All those proteins are similar in size (314-394 aa, except PmRV1 for which ORF1p might be not fully sequenced given the absence of the conserved 5′-terminus) and show 34-61% pairwise identity ( Figure 2). The function of those proteins is currently unknown. Campo et al. [27] pointed out a high similarity score between the predicted topology of ChNRV1 ORF1p and a portion of the capsid protein (CP) of Saccharomyces cerevisiae virus L-A. However, no true viral particle has been  The function of those proteins is currently unknown. Campo et al. [27] pointed out a high similarity score between the predicted topology of ChNRV1 ORF1p and a portion of the capsid protein (CP) of Saccharomyces cerevisiae virus L-A. However, no true viral particle has been associated with any unirnavirus by transmission electron microscopy. Instead, it was shown that BbNV1 most likely exhibit linear chain-like structures when analyzed by atomic force microscopy [36]. Those are believed to be a particular association of the genomic dsRNA with ORF1p [36]. The absence of true virion is also observed for two other related groups of monopartite dsRNA viruses, namely ustiviruses and amalgaviruses (see Section 3.3), which also have a ca. 3 kb genome encoding for two ORFs and containing a PRF motif. While ORF1p of ustiviruses are much smaller (ca. 180 aa) than those of unirnaviruses, those of amalgaviruses are more similar in size (ca. 380 aa). ORF1p from these three viral groups show small pairwise identity (Figure 2), yet Nibert et al. [37] have shown that they all exhibit an α-helical coiled coil (CC) in their predicted tertiary structures. A similarity plot of an alignment of FcV1 and other unirnavirus ORF1p (Figure 3, black line) shows that the N-terminal region is highly variable while the C-terminal region containing the predicted CC ( Figure 3, red line) is conserved, suggesting the importance of this structure in unirnaviruses ORF1p function. CC in their respective C-terminal region ( Figure S3). They were predicted from a metagenomic study in marine microorganisms in the north-west Pacific Ocean [39]. No other information is available regarding those sequences besides the fact that their assembling reads were obtained following a dsRNA extraction protocol from cells collected from surface water, suggesting than they represent sequences of dsRNA viruses infecting those organisms. Unexpectedly, a hypothetical protein (ODQ71866.1) from Lipomyces starkeyi strain NRRL Y-11557 [40] was also evidenced by a BLASTp search for FcV1 ORF1p (E-value = 2 × 10 −16 , 55% coverage). A total of 153 out of 187 aa (82%) of this protein shows strong similarity to the conserved C-terminal region of unirnaviruses ORF1p, although its total length is much shorter. Such a result could correspond to an endogenous virus-like element (EVE), resulting from the integration of a viral cDNA into its host genome.
To get more insights on this potential EVE, we used the MycoCosm portal [41]. We retrieved the gene of interest inside the L. starkeyi genome in the scaffold 8 at positions 129932-130495, on the minus strand ( Figure 4, blue arrow). This gene seems to be unique to L. starkeyi since a BLASTp search did not reveal any significant hit in other eukaryotic genomes, advocating for an integration event. A nucleotide deletion in the stop codon of the ORF prolongs the coding region by 192 nt, leading to an additional fragment of 64 aa (Figure 4, dotted blue arrow). Remarkably, this additional fragment also displays similarity with unirnaviruses ORF1p. Overall, the pairwise aa identity of this region ranges from 30.1 to 39.5% with homolog regions of unirnaviruses ORF1p. According to expressed sequence tag (EST) data, it seems that this gene is transcribed (Figure 4, red line). Interestingly, another gene is found on the same transcript (ODQ71867.1; Figure 4, green arrow) This gene contains an intron and produces a hypothetical protein exhibiting a conserved helix-turn-helix (HTH) domain homolog to that of OrfB of IS605 transposase (pfam:12323, E-value = 10 -12 ). However, the remaining sequence does not match to any other characterized protein. Nevertheless, it is tempting to speculate that this protein might have been involved in the horizontal gene transfer event that led to the viral endogenization. Interestingly, it was proposed that amalgaviruses ORF1p originated from a nucleocapsid protein (NC) from a phlebo/tenuivirus during a genomic recombination event, involving a partitivirus as the donor for the replicase [38]. This hypothesis was supported by results obtained from searches using softwares based on pairwise comparison of profile hidden Markov models (HMM). When similar searches were carried out for unirnaviruses ORF1p, no proteins corresponding to known phlebo/tenuivirus NC were listed in the hits. Instead, using an alignment of unirnaviruses ORF1p as a query for a search using HMMERsearch, three significant hits were obtained corresponding to uncharacterized proteins: A0A2V0RJY0, A0A2V0RJC8 and A0A2V0RBA2 (E-value ≤ 5.2 × 10 −9 ). Those proteins are longer (544, 488 and 546 aa, respectively) than unirnaviruses ORF1p but contain a similar CC in their respective C-terminal region ( Figure S3). They were predicted from a metagenomic study in marine microorganisms in the north-west Pacific Ocean [39]. No other information is available regarding those sequences besides the fact that their assembling reads were obtained following a dsRNA extraction protocol from cells collected from surface water, suggesting than they represent sequences of dsRNA viruses infecting those organisms.
Unexpectedly, a hypothetical protein (ODQ71866.1) from Lipomyces starkeyi strain NRRL Y-11557 [40] was also evidenced by a BLASTp search for FcV1 ORF1p (E-value = 2 × 10 −16 , 55% coverage). A total of 153 out of 187 aa (82%) of this protein shows strong similarity to the conserved C-terminal region of unirnaviruses ORF1p, although its total length is much shorter. Such a result Viruses 2020, 12, 523 7 of 13 could correspond to an endogenous virus-like element (EVE), resulting from the integration of a viral cDNA into its host genome.
To get more insights on this potential EVE, we used the MycoCosm portal [41]. We retrieved the gene of interest inside the L. starkeyi genome in the scaffold 8 at positions 129932-130495, on the minus strand (Figure 4, blue arrow). This gene seems to be unique to L. starkeyi since a BLASTp search did not reveal any significant hit in other eukaryotic genomes, advocating for an integration event. A nucleotide deletion in the stop codon of the ORF prolongs the coding region by 192 nt, leading to an additional fragment of 64 aa (Figure 4, dotted blue arrow). Remarkably, this additional fragment also displays similarity with unirnaviruses ORF1p. Overall, the pairwise aa identity of this region ranges from 30.1 to 39.5% with homolog regions of unirnaviruses ORF1p. According to expressed sequence tag (EST) data, it seems that this gene is transcribed (Figure 4, red line). Interestingly, another gene is found on the same transcript (ODQ71867.1; Figure 4, green arrow) This gene contains an intron and produces a hypothetical protein exhibiting a conserved helix-turn-helix (HTH) domain homolog to that of OrfB of IS605 transposase (pfam:12323, E-value = 10 -12 ). However, the remaining sequence does not match to any other characterized protein. Nevertheless, it is tempting to speculate that this protein might have been involved in the horizontal gene transfer event that led to the viral endogenization.

Analysis of FcV1 ORF2 and Phylogenetic Analysis
As mentioned above, FcV1 ORF2 is probably translated by a -1 PRF to produce the fusion protein ORF1+2p. While the function of FcV1 ORF1p in the viral cycle remains elusive, FcV1 ORF1+2p is likely to be associated with viral replication. Indeed, ORF2 sequence contains a predicted RNAdependent RNA-polymerase (RdRp) domain (cd01699, E-value = 2.84 × 10 −16 ) at positions 239-433. Inside this RdRp domain, the palm motifs A, B and C [42] were readily identified in FcV1 as well as in other unirnaviruses ( Figure 5).

Analysis of FcV1 ORF2 and Phylogenetic Analysis
As mentioned above, FcV1 ORF2 is probably translated by a -1 PRF to produce the fusion protein ORF1+2p. While the function of FcV1 ORF1p in the viral cycle remains elusive, FcV1 ORF1+2p is likely to be associated with viral replication. Indeed, ORF2 sequence contains a predicted RNA-dependent RNA-polymerase (RdRp) domain (cd01699, E-value = 2.84 × 10 −16 ) at positions 239-433. Inside this RdRp domain, the palm motifs A, B and C [42] were readily identified in FcV1 as well as in other unirnaviruses ( Figure 5).
A maximum-likelihood phylogenetic analysis based on an alignment of FcV1 ORF2 with related ORF strongly places FcV1 together with other unirnaviruses (Figure 6). A similar result was obtained following a Bayesian method ( Figure S4). The proposed family "Unirnaviridae" is related to several other clusters of monopartite viruses with similar genome architecture belonging to the family Amalgaviridae and another group for which the genus name ''Ustivirus" has been proposed [43]. The family Amalgaviridae accommodates the genus Amalgavirus as well as the proposed genera ''Zybavirus" and ''Anlovirus" [26,28]. While amalgaviruses infect plants, members of "Zybavirus" and "Anlovirus" have been found in the yeast Zygosaccharomyces bailii and in the microsporidian Anlospora locustae, respectively [26,28]. As far as ustiviruses are concerned, they have been detected in fungi [43].
Viruses 2020, 12, 523 8 of 13 As mentioned above, FcV1 ORF2 is probably translated by a -1 PRF to produce the fusion protein ORF1+2p. While the function of FcV1 ORF1p in the viral cycle remains elusive, FcV1 ORF1+2p is likely to be associated with viral replication. Indeed, ORF2 sequence contains a predicted RNAdependent RNA-polymerase (RdRp) domain (cd01699, E-value = 2.84 × 10 −16 ) at positions 239-433. Inside this RdRp domain, the palm motifs A, B and C [42] were readily identified in FcV1 as well as in other unirnaviruses ( Figure 5). Figure 5. Alignment of unirnaviruses motifs A, B and C from the RNA-dependent RNA-polymerase (RdRp) palm domain. Identical residues are highlighted by asterisks, highly conserved residues by double dots and less conserved but related residues by single dots. Numbers of residues separating the different motifs are given in grey.
A maximum-likelihood phylogenetic analysis based on an alignment of FcV1 ORF2 with related ORF strongly places FcV1 together with other unirnaviruses (Figure 6). A similar result was obtained following a Bayesian method ( Figure S4). The proposed family "Unirnaviridae" is related to several other clusters of monopartite viruses with similar genome architecture belonging to the family Amalgaviridae and another group for which the genus name ''Ustivirus'' has been proposed [43]. The family Amalgaviridae accommodates the genus Amalgavirus as well as the proposed genera ''Zybavirus'' and ''Anlovirus'' [26,28]. While amalgaviruses infect plants, members of "Zybavirus" and "Anlovirus" have been found in the yeast Zygosaccharomyces bailii and in the microsporidian Anlospora locustae, respectively [26,28]. As far as ustiviruses are concerned, they have been detected in fungi [43]. Recently, putative new viral species related to amalgaviruses and ustiviruses have been detected in arthropods: Beihai barnacle virus 14 (BbV14), Hubei partiti-like virus 59 (HPLV14) and Leshenault partiti-like virus (LPLV) [44,45]. However, these viral sequences were obtained from deep sequencing of total nucleic acids extracted from arthropods, leaving open the possibility that their actual hosts might be arthropod endosymbionts rather than the arthropods themselves. Two viruses somehow related to ustiviruses have been detected in marine microorganisms, namely Bryopsis mitochondrion-associated virus (BmaV) and Diatom colony-associated dsRNA virus 2 (DcaV2) [44,45]. Notably, those viruses also exhibit a similar genome organization consisting of two ORF and a PRF motif.
Despite exhibiting the genome organization of Totiviridae, all the aforementioned viruses are actually more related to members of the family Partitiviridae and a group of bipartite viruses which were previously referenced as "CTTV-like viruses", named after the famous Curvalaria thermal tolerance virus [28]. Both viral groups are associated with isometric particles. While CTTV-like viruses have been found exclusively in fungi, partitiviruses are multipartite viruses infecting fungi, plants, protists and insects.

Biological Characterization and Prevalence in Other Strains
In order to assess the vertical transmission rate of FcV1, sixty macroconidia produced by the F. culmorum strain A104-1 were isolated and allowed to produce new mycelium. An RT-PCR screening Recently, putative new viral species related to amalgaviruses and ustiviruses have been detected in arthropods: Beihai barnacle virus 14 (BbV14), Hubei partiti-like virus 59 (HPLV14) and Leshenault partiti-like virus (LPLV) [44,45]. However, these viral sequences were obtained from deep sequencing of total nucleic acids extracted from arthropods, leaving open the possibility that their actual hosts might be arthropod endosymbionts rather than the arthropods themselves. Two viruses somehow related to Viruses 2020, 12, 523 9 of 13 ustiviruses have been detected in marine microorganisms, namely Bryopsis mitochondrion-associated virus (BmaV) and Diatom colony-associated dsRNA virus 2 (DcaV2) [44,45]. Notably, those viruses also exhibit a similar genome organization consisting of two ORF and a PRF motif.
Despite exhibiting the genome organization of Totiviridae, all the aforementioned viruses are actually more related to members of the family Partitiviridae and a group of bipartite viruses which were previously referenced as "CTTV-like viruses", named after the famous Curvalaria thermal tolerance virus [28]. Both viral groups are associated with isometric particles. While CTTV-like viruses have been found exclusively in fungi, partitiviruses are multipartite viruses infecting fungi, plants, protists and insects.

Biological Characterization and Prevalence in Other Strains
In order to assess the vertical transmission rate of FcV1, sixty macroconidia produced by the F. culmorum strain A104-1 were isolated and allowed to produce new mycelium. An RT-PCR screening showed that 100 % of those progeny mycelia were infected with the virus, highlighting the ability of FcV1 to be efficiently vertically transmitted in its host.
Since conidial isolation did not result in the loss of FcV1, the fungal host was treated with two different chemicals known for their ability to cure of mycovirus, namely ribavirin [46] and cycloheximide [47]. An RT-PCR screening of mycelia produced by chemically treated conidia revealed the presence of FcV1 in all of them, suggesting the inability of the two chemicals to cure the virus, at least in the tested experimental conditions.
As an alternative to the production of a virus-free isogenic line of the strain A104-1, another way towards the biological characterization of FcV1 would be the infection of a virus-free strain. A dual culture combining the strain A104-1 and the selectable strain P1P2 was achieved to this end. In practical terms, both strains were cultured on PDA and mycelia plugs from the confrontation zone were subcultured on selective media. None of the produced mycelia gave positive signal for FcV1 by RT-PCR. At this point, it is unclear whether this lack of transmission is due to the inability of FcV1 to be transmitted in this fashion or if viral transfer is impeached by vegetative incompatibility between both fungal strains.
The presence of FcV1 in other F. culmorum strains was also investigated. In order to do so, fifty-three strains that were originally sampled in Europe, Russia, Turkey, New Zealand and Canada were tested. Interestingly, two European strains gave the amplicon with the expected size (499 bp) by RT-PCR and exhibited a similar ca. 3 kb band when analyzed for the presence of dsRNA ( Figure S5): PVS-Fu 353 from Pozzo S. Nicola (Sassari in Sardinia, Italy) [48] and Fc30 from Vaihingen (Germany) [49].

Discussion
Kotta-Loizou et al. [30] originally suggested the name "Unirnavirus" when they described BbNV1 that was phylogenetically related to AlRV1, implying the creation of a new group to gather both viruses. Since then, several mycoviruses with related RdRp have been discovered in different fungal species, leading the same authors to propose the viral family name "Unirnaviridae" [36]. In addition to related RdRp, all those viruses share the following features: a monopartite dsRNA genome of ca. counterfeit viral capsid protein derived from an EVE that interferes with the assembly of a similar virus, acting therefore as an antiviral immunity system [54]. Whether the EVE found in L. starkeyi could act in a similar way would be worth testing. But at first, confirming the existence and expression of the detected EVE by (RT)-PCR is a prerequisite.
We were unable to cure the strain A104-1 from FcV1 by isolation of macroconidia. This vertical transmission rate of 100% was already observed for UvNV1 [50]. In parallel, the amalgavirus Southern tomato virus was shown to be seed-transmitted at rates ranging from 70 to 90% [55]. Altogether, it seems that those related viruses have evolved efficient strategies to be vertically retained in their respective host.
Since we were not able neither to cure the virus from the strain A104-1 nor to infect another strain, the biological effect of FcV1 remains undetermined. Previously, it was shown that ChNRV1 induces conidial defects on its host strain mutated for dcl1, whereas it seems to be asymptomatic in the wild-type strain [27]. As far as the related amalgaviruses and ustiviruses are concerned, no direct correlation between viral infection and symptom was ever evidenced to our knowledge. Actually, many of the amalgaviruses have been discovered randomly through transcriptomic data [29,[48][49][50], probably illustrating a persistent nature. It seems that those viral species do not represent relevant candidates for biocontrol purpose given the absence of an impact on the host, even though phenotype alternation under particular conditions cannot be ruled out.
Eventually, we performed a screening of other F. culmorum and we gave evidences for the presence of the virus in two other strains from Europe. In the future, it will be interesting to obtain the sequences of those potential FcV1 isolates as well as to test more strains in order to shed light on the nucleotide diversity of this virus.
In summary, in this study we have described FcV1, a novel monopartite mycovirus infecting F. culmorum. To our knowledge, it is the first full genomic sequence for a virus infecting this fungal species. Our studies also broaden the diversity of mycovirus belonging to "Unirnaviridae" and supports the proposition of this taxon being considered as a new viral family.