A newly established reverse genetic system for a circular RNA virus reveals new requirements for infection and its biocontrol potential

Abstract

Ambiviruses are fungal-infecting circular RNA viruses that uniquely combine viroid-like and viral features, yet the function of their conserved ORF-B protein and their effects on hosts remain unknown, hindered by the lack of a reverse genetics system. Here, we constructed the first infectious cDNA clone of an ambivirus, Fusarium graminearum ambivirus 1 (FgAV1), using a head-to-tail dimer placed downstream of a fungal promoter. FgAV1 was horizontally transmitted via hyphal anastomosis to virus-free Fusarium graminearum strains. Notably, a reverse-oriented dimer construct was also infectious and transmissible. Targeted mutagenesis revealed that both ORF-A- and ORF-B-encoded proteins and the presence of embedded ribozymes are indispensable for ambivirus replication. Our results further demonstrate that FgAV1 infection triggers a fungal RNAi response, extending the antiviral role of host sRNAs to circular RNA viruses. Furthermore, FgAV1 infection suppressed fungal growth and significantly reduced the virulence of F. graminearum on wheat. These findings provide novel insights into ambivirus replication and their potential in fungal pathogen biocontrol.

INTRODUCTION

RNA molecules are widely hypothesized to have preceded DNA and proteins in the early stages of life, functioning both as carriers of genetic information and as catalysts, a scenario known as the “RNA world” hypothesis (1–4). RNA viruses, which are regarded as constituting the closest extant recapitulation and potentially a vestige of the primordial RNA world, infect a wide range of hosts, from animals and plants to fungi and bacteria, spanning the two main life domains (Bacteria and Eukarya), while their presence in Archaea has been only suggested based on metagenomic data from Archaea-enriched environments but never confirmed experimentally (5). RNA viruses are currently organized in two realms: Ribozyviria (a very small group of viruses with circular RNA genomes replicated in trans by host DNA dependent RNA polymerases, DdRP) and Riboviria (encoding for an RNA-directed RNA polymerase, RdRP); RdRP is a possible candidate as the first protein that appeared on earth (6). RdRP-encoding viruses have high negative impact as disease-causing agents but serve as valuable models for studying virus replication, evolution, and host-virus interactions (3). Fungal viruses (mycoviruses) constitute a particularly intriguing group: they are predominantly RNA viruses, with only a few DNA viruses, and display striking genomic diversity, encompassing members with double-stranded RNA, positive- and negative-sense single-stranded RNA genomes, and even capsidless elements with viroid-like properties (7–9). Moreover, recent advances in metatranscriptomics have further uncovered unexpected diversity within this group, including highly unusual ambiviruses (10).

Ambiviruses possess circular, single-stranded RNA genomes of approximately 4.3–5.2 kb, with a compact rod-like conformation, featuring two large, non-overlapping open reading frames (ORFs) arranged in ambisense orientation (ORF-A and ORF-B). ORF-A possibly encodes a structurally conserved putative RdRP, placing ambiviruses within the realm of RdRP-encoding RNA viruses (Riboviria), whereas ORF-B encodes a highly conserved protein of unknown function and with no detectable homology with any protein outside the phylum Ambiviricota (10, 11). Remarkably, their genomes harbor functional ribozymes in both polarities, a hallmark of chloroplastic viroids, some viroid-like satellite RNAs and members of the realm Ribozyviria (10–12). The presence of active ribozymes and the accumulation of circular forms of both polarity strands in Cryphonectria parasitica infected by Cryphonectria parasitica ambivirus 1 was experimentally confirmed by Forgia et al. (10), strongly indicating a symmetric rolling-circle replication mechanism, analogous to that employed by chloroplastic viroids. The unique hybrid nature of ambiviruses, combining features of RNA viruses and viroid-like elements, has led recent taxonomic revisions to assign them to a distinct phylum, Ambiviricota, within the realm Riboviria and the kingdom Orthornavirae (11). The discovery of ambiviruses not only helps to illuminate the evolutionary links between viroid-like RNAs and conventional RNA viruses, but also provides insights into the diverse replication strategies of circular RNA viruses: so far circularity of viral RNA genomes was experimentally proven only for members of the Ribozyviria (Hepatitis delta-like elements, which do not encode an RdRP), while only members of the Ambivircota were experimentally proven to be circular in the Riboviria (11). Other possible circular genomes might exist in Riboviria, as is the case of some members of the family Chuviridae (13), and some mitovirus sequences (12).

Despite their intriguing biology, research on ambiviruses remains in its infancy. To date, most studies have relied heavily on metatranscriptomic data, which have revealed their abundance in diverse fungal lineages (14). Yet, little is known about their molecular biology, replication requirements, or effects on host fungi. No reverse genetics system has been developed for any ambivirus, creating a critical bottleneck for functional investigations. In particular, the role of ORF-B, which is strikingly conserved across ambivirus clades (10, 11), is still enigmatic: is it translated into a functional protein, or does its conservation reflect RNA structural constraints necessary for replication or the self-complementarity required for the conserved rod-shaped secondary structure of the genome? Similarly, while the presence of ribozymes strongly implies a role in processing circular RNA genomes during replication, their functional indispensability has never been formally demonstrated.

Another major gap concerns the ecological and phenotypic consequences of ambivirus infection, together with the molecular mechanisms that drive their replication. Many mycoviruses cause asymptomatic infections, while some attenuate fungal growth, sporulation, pigmentation, or virulence (15). The best-known examples are Cryphonectria hypovirus 1 (CHV1), which causes hypovirulence in C. parasitica, the causal agent of chestnut blight, and Sclerotinia sclerotiorum hypovirulence associated DNA virus 1 (SsHADV1). CHV1 has been successfully deployed as a biological control agent in Europe, serving as a landmark case for mycovirus-mediated disease control (15–17). Beyond CHV1 and SsHADV1 (18), a wide range of mycoviruses from diverse genomic groups have also been reported to induce hypovirulence in fungi, including the dsRNA viruses Stemphylium lycopersici alternavirus 1 (SlAV1) (19), Rosellinia necatrix megabirnavirus 1 (RnMBV1) (20), Heterobasidion partitivirus 13 (HetPV13-an1) (21), the ssRNA virus Fusarium graminearum Virus 1 (FgV1) (22) and the circular DNA virus Diaporthe sojae circular DNA virus 1 (DsCDV1) (23), among many others. These cases highlight the potential of mycoviruses as hypovirulence factors in agricultural biocontrol. However, the impact of ambiviruses on fungal phenotypes remains largely unexplored. It is unclear whether they act primarily as silent passengers or actively remodel host fitness and pathogenicity.

Ambiviruses mostly infect basidiomycetes, but some proven examples of ambiviruses infecting ascomycetes also exist: in this regard, one particularly relevant host is the ascomycetous fungus Fusarium graminearum, a devastating plant pathogen responsible for Fusarium Head Blight (FHB) in cereals worldwide. FHB not only reduces grain yield but also contaminates harvests with mycotoxins such as deoxynivalenol (DON), posing severe risks to food security and safety (24). Recent work identified Fusarium graminearum ambivirus 1 (FgAV1) in the strain Fg266 MV, which harbors several distinct mycoviruses (25). Compared with its ambivirus-free counterpart, Fg266 MV exhibited reduced DON production and increased attractiveness to fungivores (25). Given that these phenotypic alterations arose amid multiple co-infecting viruses, the direct effect of FgAV1 on fungal growth and pathogenicity has yet to be experimentally determined. Nonetheless, these findings raise the intriguing possibility that FgAV1, like certain other mycoviruses, may attenuate fungal virulence or reshape host ecology, highlighting its potential as a biological control factor.

Here, we established the first reverse genetic system for an ambivirus, using FgAV1 as a model. By constructing infectious cDNA clones, we determined that viral infection can be initiated from dimeric genome constructs, and unexpectedly, that reverse-oriented dimers are also competent to initiate replication. Through targeted mutagenesis, we demonstrated that ORF-B translation is indispensable for infectivity. We also confirmed that ribozyme motifs are absolutely essential for viral replication, consistent with their predicted function in processing circular RNA genomes. Importantly, FgAV1 infection alters host biology, suppressing mycelial growth and reducing virulence on wheat. These findings not only provide direct functional evidence for the unique genome architecture of ambiviruses but also prove their potential role in shaping fungal pathogenicity. Collectively, our work establishes a foundation for mechanistic studies of ambivirus replication and highlights their possible applications in biocontrol of plant fungal diseases.

RESULTS

Assembly of an infectious cDNA clone of FgAV1

To establish a reverse genetics system for FgAV1 (Fig. 1A depicts the viral genome structure), we generated a cDNA clone based on the expression vector pHX9-nv, which contains the C. parasitica glyceraldehyde-3-phosphate dehydrogenase (gpd-1) promoter and terminator for high-level constitutive expression in filamentous fungi and a hygromycin phosphotransferase (hph) gene for selection. To validate the expression capability of the pHX9-nv vector and its regulatory elements in our target heterologous host (the virus-free F. graminearum strain Fg1731), we first constructed control plasmids pHX9-GFP and pHX9-mCherry expressing fluorescent reporter genes downstream the promoter region (Fig. 1B). Transformation of Fg1731 protoplasts with these plasmids resulted in robust GFP and mCherry fluorescence throughout the fungal hyphae, as clearly visualized by confocal microscopy (Fig. 1C). This unequivocally demonstrated efficient transcription and translation from the pHX9-nv vector in the strain Fg1731.

Fig. 1. Construction and validation of FgAV1 infectious cDNA clone.

(A) Genomic organization of FgAV1 (sense: 5′ → 3′). ORF, open reading frame; Rz (+), ribozyme contained in the plus strand of FgAV1; Rz (−), ribozyme contained in the minus strand of FgAV1; NB probe, location of the northern blot probes indicated by a blue dashed line. (B) Schematic diagram of the pHX9-GFP, pHX9-mCherry, pHX9-FgAV1-monomer and pHX9-FgAV1-dimer (not to scale). Pgpd, C. parasitica glyceraldehyde-3-phosphate dehydrogenase (gpd-1) promoter; Tgpd, C. parasitica glyceraldehyde-3-phosphate dehydrogenase (gpd-1) terminator; Red arrow, FgAV1 ORF-A; Green arrow, FgAV1 ORF-B; hph, hygromycin phosphotransferase gene for selection. (C) Confocal microscopy observations of GFP or mCherry expression in Fg1731 hyphae transformed with pHX9-GFP or pHX9-mCherry, confirming promoter activity (scale bar: 100 μm). (D) Northern blot analysis of FgAV1 replication in FgAV1 Monomer and Dimer transformants. RNA preparations were separated in agarose electrophoresis under denaturing conditions, transferred to membrane and hybridized with specific DIG-RNA probes for (+) and (−) FgAV1 RNA. (+) polarity and (−) polarity refer to negative strand probe (hybridizing with viral + strand RNA) and positive strand probe (hybridizing with viral − strand polarity); by convention, viral + strand and − strand are those that can directly express ORF-A or its antisense, respectively. Leftmost panels show RNA staining of the membranes with methylene blue. Lanes 1 and 2, total RNA from TBSV- and TMV-infected plants, respectively, are used as RNA markers. Signals corresponding to linear and circular genomic FgAV1 RNA forms (4579 nt) are indicated on the right. A smaller band of the positive sense is also abundantly accumulating (dashed blue arrow).

Leveraging the validated pHX9-nv system, we constructed two FgAV1 expression plasmids (Fig. 1A depicts the viral genome structure). The pHX9-FgAV1-monomer contained a single, full-length cDNA copy of the FgAV1 genome and served as a negative control, as monomeric constructs are predicted not to support the production of replication-competent viral RNAs. The pHX9-FgAV1-dimer contained a head-to-tail tandem dimer of the FgAV1 full-length cDNA (Fig. 1B), a configuration designed to allow the production of full-length monomeric genomic RNA forms through the self-cleavage activity of ribozymes within the dimeric transcript, thereby enabling to launch a productive viral infection. Total RNAs were extracted from Fg1731 transformants “Monomer” (harboring pHX9-FgAV1-monomer) and “Dimer” (harboring pHX9-FgAV1-dimer). PCR amplification from RNase-free DNase I-untreated RNA revealed amplicons in both transformants, indicating that the monomer and dimer cDNAs were successfully introduced into the host (fig. S1). However, after RNase-free DNase I treatment of the RNA samples, no PCR products were detected, confirming complete removal of residual plasmid DNA and excluding transgene-derived artifacts (fig. S1) after DNase treatment.

To specifically detect authentic viral RNA replication (distinct from that resulting from mere transcription from the gpd-1 promoter), we performed northern blot analysis on the DNase I-treated RNA samples using strand-specific riboprobes complementary to the positive-sense (genomic) or negative-sense (replicative intermediate) FgAV1 RNA. Prior to analyzing the transformants, we validated the specificity of these probes using RNA extracted from the naturally FgAV1-infected strain Fg266 MV and the virus-free strain Fg1731 (fig. S2). As expected, probes specific to both polarities detected viral RNA only in the naturally infected Fg266 MV sample, confirming probe specificity and reliability (fig. S2): specifically, a linear and circular form of the genomic fragment can be seen both in the positive sense and negative sense (higher than the 4.8 kb marker of TBSV genomic RNA); furthermore an RNA species positioning between the two fungal ribosomal RNAs is abundantly accumulating in the positive sense orientation, and a lower band (doublet in positive sense) running approximately with the lowest band of the plant ribosomal RNA is also present both in positive and negative sense orientation. Analysis of the transformants revealed a stark contrast. No viral RNA of either polarity was detected in the strain Monomer (Fig. 1D). In contrast, strain Dimer exhibited clear signals of circular and linear forms of both positive-sense (genomic) and negative-sense (replicative intermediate) viral RNAs (Fig. 1D); furthermore, an abundant putative subgenomic RNA (or defective RNA) is specifically and abundantly accumulating as positive sense RNA orientation; the lowest defective RNA is absent in this extract.

The detection of negative-sense RNA is a definitive marker of active viral RNA replication. Therefore, these results provide the first evidence that the dimeric cDNA clone, but not the monomeric control, successfully initiated autonomous FgAV1 replication in the previously virus-free Fg1731 host. This establishes pHX9-FgAV1-dimer as the first reported infectious cDNA clone for an ambivirus.

Reverse-oriented dimer cDNA clone can initiate FgAV1 replication

In the case of ambivirus genomic structure, positive and negative sense orientations were arbitrarily assigned based on the RdRP coding sense (positive sense). To investigate whether the orientation of the FgAV1 cDNA within the expression vector influenced infectivity, a reverse-oriented dimer clone was constructed (pHX9-FgAV1-Rev-dimer), in which the tandem full-length cDNAs of FgAV1 were inserted in the opposite orientation relative to the gpd-1 promoter (Fig. 2A) and that putatively cannot express the ORF-A -the RdRP- from the primary transcript, a step that is supposed to be required to initiate replication. Transformation of the virus-free strain Fg1731 with this construct yielded PCR products when RNA samples were analyzed without DNase I treatment, confirming successful delivery of the Rev-dimer plasmid into fungal cells (Fig. 2B). No products were detected from DNase I-treated RNA, excluding residual plasmid DNA contamination (Fig. 2B) after DNase treatment.

Fig. 2. FgAV1 replication from reverse-oriented dimer clone.

(A) Schematic diagram of the pHX9-FgAV1-Rev-dimer (not to scale). See the legend to Fig. 1B for details. (B) PCR detection of FgAV1 from the Rev-Dimer transformant. M, Marker; Rev-Dimer DNA, DNase I-untreated nucleic acid; Rev-Dimer RNA, DNase I treated nucleic acid; +, cDNA from strain Fg266 MV, the original strain carrying the virus FgAV1; −, cDNA from virus-free strain Fg1731. (C) Northern blot analysis of FgAV1 replication in the Rev-Dimer transformant. Lanes 4 and 5, RNA preparations from Fg266 MV (positive control, +) and from the virus-free strain Fg1731 (negative control, −), respectively. For additional details, refer to the legend of Fig. 1D. A smaller band present in both positive (as a doublet) and negative polarity is indicated by an orange dashed arrow. (+) polarity and (−) polarity refer to negative strand probe (hybridizing with viral + strand RNA) and positive strand probe (hybridizing with viral − strand polarity); by convention, viral + strand and − strand are those that can directly express ORF-A or its antisense, respectively.

Surprisingly, northern blot analysis of DNase I-treated RNA revealed clear hybridization signals with both positive-sense and negative-sense probes in the Rev-Dimer transformant (Fig. 2C). The detection of linear and circular negative-sense RNA forms indicated active replication, and this pattern was indistinguishable from that observed with the forward-oriented dimer (Fig. 1D), demonstrating that the reverse-oriented dimer construct is also capable of initiating FgAV1 replication. Both the putative subgenomic and smaller defective RNA bands are present in this extract. The unexpected finding that negative sense dimer cDNA can initiate viral replication challenges the conventional paradigm that primary transcripts from reverse-oriented viral cDNAs (which express a full length ORF-B, but not the ORF-A) cannot directly launch replication cycles. Given the possible viroid-like aspects of the replication cycle for ambivirus, contributions from host RNA polymerases cannot be excluded. For this reason, we tested if ORF-A is strictly required for FgAV1 replication generating a mutant ORF-A with the catalytic GDD sub motif in the RdRP palm domain deleted (putatively inactivating the RdRP functionality): we could not provide evidence of specific viral replication of the mutant in DNA positive transformants by northern blot analysis (fig. S3). Replication initiated by a reverse-oriented viral cDNAs therefore suggests a remarkable capacity of the host transcript machinery to recognize hidden promoters in the expression plasmid and in specific in the ambiviral sequences themselves to overcome this orientation barrier.

Translation of ORF-B is required for FgAV1 replication

ORF-B-encoded protein is very conserved among ambiviruses with a few subdomains with conserved amino-acid positions in multiple sequence alignments even among the most distantly related ambiviruses (fig. S4). To explore the potential role of ORF-B in the FgAV1 life cycle, we first predicted the three-dimensional structure of the ORF-B-encoded protein using AlphaFold3. The resulting overall 3D-structure (Fig. 3A and fig. S5) was well supported with at least four subdomains displaying a very high plDDT (>90) and an overall pTM of 0.64. The four conserved subdomains spanned region 4–91, 92–180, 267–429, and 430–643 and are here named subdomain X, Y, W and Z, respectively (Fig. 3, B and C and fig. S5, B and C). Structural comparison with the protein data bank (PDB) and big fantastic viral database (BFVD) (26) of the whole molecule using Foldseek Search Server (https://search.foldseek.com/search) revealed a highly conserved RNA recognition motif (RRM)-like fold between residues 92–144 (an example of structural overlap with Caenorhabditis elegans RRM domain contained in a PICS complex is displayed in Fig. 3D). The RRM motif has the typical structure that comprises three anti-parallel β-sheets and two α-helical regions: conserved Y residues are also present in two adjacent antiparallel sheets. Regarding the other three subdomains, positive hits originated when searching the BFVD database using Foldseek server with domain W 280–409 (Fig. 3C): surprisingly the positive hits relate to viral RdRP palm domains. In Fig. 3E are shown the overlapped structures of FgAV1 ORF-B domain W280–409 with the palm domain of Pasivirus A (BFVD accession A0A513Q8Y0). Although structurally conserved with the ABC subdomains of an RdRP palm domain, the catalytic core with the conserved D residues present in A and C subdomains are not present in ORF-B encoded proteins.

Fig. 3. Viral ORF-B-encoded protein has a conserved 3D structure with RRM-like and RdRP-like domains.

(A) Alphafold 3 prediction of FgAV1 ORF-B-encoded protein. The 3D structure of the protein is colored according to the Predicted Local Distance Difference Test (pLDDT) confidence score: dark blue and light blue are for very high and high confidence respectively; yellow and orange are for low and very low confidence respectively. A graph representing the expected position errors is also depicted with the 4 subdomains of high confidence X, Y, W and Z (with the spanning regions labelled to the right end. (B) The 92–180 Y domain (which includes the slightly shorter 92–144 RRM), within which conserved tyrosine residues are highlighted. (C) The 267–429 W domain (which includes the slightly shorter 280–409 palm domain). (D) Overlapping of the predicted 92–144 RRM domain with the Cryo-EM structure of the RRM domain of the PCS complex from Caenorhabditis elegans (PDB 7D1L). (E) Overlapping of the 280–409 domain with the palm domain of Pasivirus A (BFVD A0A513Q8Y0). The GDD and D residues of the catalytic subdomain C and A are highlighted. In (D) and (E), Foldseek searches were performed of the BFVD and PDB databases.

The high structural conservation of ORF-B-encoded proteins (comparable to the ORF-A-encoded RdRP) suggests a basic role in viral replication. To functionally test this hypothesis, we generated mutant constructs disrupting its translation initiation codon. The start codon (AUG) of ORF-B was deleted in both forward (pHX9-FgAV1-ΔAUG-ORFB) and reverse-oriented (pHX9-FgAV1-Rev-ΔAUG-ORFB) dimer configurations within the pHX9 vector (Fig. 4A). This design specifically abrogates ORF-B protein translation while only minimally affecting (and only locally, maintaining the rod-shape overall structure) the nucleotide sequence and RNA secondary structures (fig. S6). To test infectivity, transformation of Fg1731 protoplasts with these constructs was confirmed by PCR (Fig. 4B). Northern blot analysis of DNase I-treated RNAs from both transformants failed to detect viral RNAs of either polarity using strand-specific probes (Fig. 4C). This indicates that the deletion of the ORF-B start codon abolished the ability of FgAV1 to replicate.

Fig. 4. Viral ORF-B is essential for FgAV1 replication.

(A) Schematic diagram of the pHX9-FgAV1-ΔAUG-ORFB and pHX9-FgAV1-Rev-ΔAUG-ORFB (not to scale). See the legend to Fig. 1B for details. (B) PCR detection of FgAV1 from ΔAUG-ORFB and Rev-ΔAUG-ORFB transformants. M, marker; ΔAUG-ORFB DNA and Rev-ΔAUG-ORFB DNA, DNase I-untreated nucleic acid; ΔAUG-ORFB RNA and Rev-ΔAUG-ORFB RNA, DNase I-treated nucleic acid; +, cDNA from strain Fg266 MV, the original strain carrying the virus FgAV1; −, cDNA from virus-free strain Fg1731. (C) Northern blot analysis of FgAV1 replication in ΔAUG-ORFB and Rev-ΔAUG-ORFB transformants. Lanes 3 and 4, RNA preparations from ΔAUG-ORFB and Rev-ΔAUG-ORFB strains. See Figs. 1D and 2C legends for more details. Blue and orange dashed arrows indicate the position of a putative subgenomic and a putative defective RNA. (+) polarity and (−) polarity refer to negative strand probe (hybridizing with viral + strand RNA) and positive strand probe (hybridizing with viral − strand polarity); by convention, viral + strand and − strand are those that can directly express ORF-A or its antisense, respectively.

Consistent with the replication defect, hyphal anastomosis experiments showed no transmission to virus-free Fg1731 recipients from either ΔAUG-ORFB (fig. S7A) or Rev-ΔAUG-ORFB strains (fig. S7B) after 28 days. RT-PCR analysis confirmed absence of viral RNA in recipient mycelia. Taken together these results demonstrate that translation of ORF-B is indispensable for FgAV1 replication and subsequent horizontal transmission. The replication defect persists even in the reverse-orientation configuration that supports wild-type virus replication, underscoring ORF-B’s non-redundant role in the ambivirus lifecycle.

Ribozyme structures are essential for FgAV1 replication

Given the unique dual ribozyme architecture characteristic of ambiviruses, we assessed the functional necessity of these catalytic RNA elements by constructing pHX9-FgAV1-ΔRz, a forward dimer clone lacking both ribozyme sequences (Fig. 5A) present in a dimer of the positive sense RNA. This deletion abrogates the possibility of self-cleavage due to ribozyme presence while preserving all coding sequences. Transformation of Fg1731 protoplasts with pHX9-FgAV1-ΔRz was confirmed by PCR (Fig. 5B). Importantly, northern blot analysis of DNase I–treated RNAs from ΔRz transformants failed to detect viral RNAs of either polarity (Fig. 5C). This absence of viral RNAs demonstrates that the ribozyme structural motif of plus polarity, likely involved in the RNA processing, is essential for initiating FgAV1 replication. Consistent with this replication defect, co-culture experiments showed no horizontal transmission of FgAV1 from ΔRz transformant strain to recipient virus-free Fg1731 via hyphal anastomosis, as confirmed by RT-PCR of recipient mycelia (fig. S7C). We then proceeded to a finer evaluation of the role of the HHrbz that could overcome the limitation of the large deletion we constructed. We derived a mutant cDNA clone with an A residue deletion in the core catalytic domain of the HHrbz that previous work (27, 28) had shown to abolish self-cleaving activity. Our results show that DNA-positive virus transformants do not display evidence of viral replication, with no accumulation of circular and linear forms of viral RNA (fig. S3). Together, these findings demonstrate that the ribozymes encoded within the FgAV1 genome are absolutely required for replication. Given that ribozymes are a defining feature distinguishing ambiviruses from canonical RNA viruses, our results provide direct experimental evidence that these catalytic RNA elements are not merely structural hallmarks but are essential functional components of the ambivirus replication cycle.

Fig. 5. Ribozyme deletion abolishes FgAV1 replication.

(A) Schematic diagram of the pHX9-FgAV1-ΔRz (not to scale). See the legend to Fig. 1B for details. (B) PCR detection of FgAV1 from the ΔRz transformant. M, Marker; ΔRz DNA, DNase I-untreated nucleic acid; ΔRz RNA, DNase I-treated nucleic acid; +, cDNA from strain Fg266 MV, the original strain carrying the virus FgAV1; −, cDNA from virus-free strain Fg1731. (C) Northern blot analysis of FgAV1 replication in the ΔRz transformant. See Figs. 1D and 2C legends for details. Blue and orange dashed arrows indicate a putative subgenomic and a putative defective RNA. (+) polarity and (−) polarity refer to negative strand probe (hybridizing with viral + strand RNA) and positive strand probe (hybridizing with viral − strand polarity); by convention, viral + strand and − strand are those that can directly express ORF-A or its antisense, respectively.

Horizontal transmission of FgAV1 via hyphal anastomosis

Having established that both the forward (pHX9-FgAV1-dimer) and reverse-oriented (pHX9-FgAV1-Rev-dimer) infectious cDNA clones initiate autonomous FgAV1 replication in F. graminearum strain Fg1731, we next investigated the possibility of horizontal transmission of the virus from these infected donor strains to a virus-free recipient strain via hyphal anastomosis. This approach was designed to exclude the possibility that the cDNA integrated in the nucleus could be indeed necessary for maintaining infection in the transgenic system. Virus-infected donor strains (Dimer transformant strain or Rev-Dimer transformant strain) were co-cultured with the original virus-free Fg1731 recipient strain for 28 days, facilitating hyphal contact and potential anastomosis. After this period, mycelia from the recipient Fg1731 sector of the co-culture interface were carefully sampled, avoiding donor hyphae (Fig. 6, A and B - left panels). Total RNA was extracted from the recipient Fg1731 mycelia sampled three successive serial transfers after co-culture. The recipient strain was unable to grow on hygromycin confirming the absence of transgene. RT-PCR analysis using FgAV1-specific primers readily detected viral RNA in recipient strains co-cultured with either the Dimer donor (Fig. 6A - right panel) or the Rev-Dimer donor (Fig. 6B - right panel) but did not detect any DNA corresponding to the virus cDNA. Crucially, no viral signal was detected in control Fg1731 cultured alone (Fig. 6, A and B - right panels). This demonstrated successful horizontal transmission of FgAV1 from both donor strains to the initially virus-free Fg1731 recipient via hyphal fusion.

Fig. 6. Viral horizontal transmission via hyphal anastomosis.

(A) Co-culturing of the Dimer transformant strain with virus-free strain Fg1731 to assess FgAV1 transmission through hyphal fusion. The fungi were photographed at 28 days after culturing. Fungal strains (Fg1731/ FgAV1) were obtained from the recipient side of the fungus and FgAV1 infection was detected by RT-PCR. (B) Co-culturing of the Rev-Dimer transformant strain with virus-free strain Fg1731. FgAV1 was detected by RT-PCR in the recipient fungal side. (C) Northern blot of recipient hyphae after co-culture with Dimer donor, showing both positive- and negative-strand RNAs of FgAV1, indicative of successful horizontal transmission. Lanes 3 and 4, RNA preparations from two Fg1731/FgAV1 strains. For additional details, see the Figs. 1D and 2C legends. (D) Northern blot analysis of recipient hyphae after co-culture with Rev-Dimer donor. Lanes 3 and 4, RNA preparations from two Fg1731/Rev-Dimer strains. See Figs. 1D and 2C legends for more details. Blue and orange dashed arrows indicate a putative subgenomic and a putative defective RNA. (+) polarity and (−) polarity refer to negative strand probe (hybridizing with viral + strand RNA) and positive strand probe (hybridizing with viral − strand polarity); by convention, viral + strand and − strand are those that can directly express ORF-A or its antisense, respectively.

To further validate these findings, northern blot analyses were conducted using strand-specific probes targeting positive-sense (genomic) and negative-sense (replicative intermediate) FgAV1 RNAs. Both probes detected clear viral signals corresponding to linear, circular and possibly a subgenomic RNA or defective RNA forms in recipient isolates obtained from dual cultures with Dimer (Fig. 6C) or Rev-Dimer (Fig. 6D) donors, confirming active viral replication in the newly infected Fg1731 strains. Collectively, these results demonstrate that FgAV1 can be horizontally transmitted via hyphal anastomosis and establish replication in previously confirmed virus-free hosts, regardless of whether the donor originated from a forward- or reverse-oriented infectious cDNA clone, confirming experimentally infectivity (ability to infect new individuals horizontally through anastomosis) of FgAV1.

Since Sanger sequencing on the two plasmids containing the monomer FgAV1 cDNA used to build the infectious dimeric clone highlighted some mutations compared to the published consensus sequence (data S1), RNAseq was performed on the Fg1731/FgAV1 isolate to monitor the emergence of new mutations (compared to the cDNA sequence of the infectious clone) in the replicating virus or to observe if the mutations inserted in the cDNA clones were retained. Using as reference the dimer sequence (data S1) inserted in the infectious clone, four mutations were found by Sanger sequence of the two cDNA monomers (A to C in position 1551, T to C in position 1840, T to G in position 2515 and A to C in position 7841 compared to the consensus sequence in natural infection. Among these, only three mutations can be found in the sequence supposed to originate the replicating virus, as the monomer sequence of FgAV1 initiating the active replication resides between the two self-cleaving sites of the ribozyme in positive polarity (position 773 to position 5351 of the dimer sequence) (data S1). After mapping the Illumina reads generated from the sequencing of the Fg1731/ FgAV1 isolate, only one mutation was observed compared to the original wild type FgAV1, A to C in position 1551 which changes the aa residue 185 of ORF-B from D to A.

These results further demonstrate that our infectious clone system is able to start the autonomous replication of FgAV1. In a transgene free context, the infectious clone is able to revert mutations through active replication that resulted in a restoration of the wild-type consensus sequence in two out of the three mutations present in the infectious clone, but tolerating a substitution in the ORF-B that is stably incorporated in the progeny.

We also addressed the possibility of transmission through co-culture with different species within the Fusarium genus, using a FgAV1 positive strain as a donor, but as expected, and contrary to what we recently showed with Trichoderma species, the results were negative (table S3). We then proceeded to evaluate whether transformation of Trichoderma atrobrunneum, Trichoderma spirale and Fusarium pseudograminearum with the FgAV1 infectious clone would generate infected strains with a replicating virus. Similarly, we could not provide evidence of replication through minus-strand selective qRT-PCR (results indistinguishable from monomer transformation), suggesting that the FgAV1 host range is limited.

FgAV1 infection may trigger a host RNA interference (RNAi) response

Small RNA (sRNA) sequencing was performed on one isolate of Fg1731/FgAV1 to determine if ambiviruses could accumulate sRNAs (a hallmark of RNAi response) as observed for viruses in fungi and in plants, and for canonical viroids in plants. After Illumina sequencing of the sRNAs obtained from Fg1731/FgAV1 isolate, the raw reads were cleaned from the adapters, low quality reads and only reads longer than 18 nt were kept for the analysis. From the 11,860,082 raw reads generated, 10,366,858 clean reads were obtained and mapped on a dimeric version of the viral sequence to detect sRNA mapping on the full circular sequence of the genome. The mapping constraints allowed one mismatch in the alignment. Results show that the number of reads mapping on FgAV1 was 149,432. These reads were almost equally distributed on the negative and positive sense of the viral genome (77,839 reads mapping on the positive sense and 71,593 mapping on the negative sense).

The length distribution of the reads mapping on the ambivirus shows a peak at 21 nt (Fig. 7A) which is compatible with the antiviral sRNA response observed in fungi (29, 30). The peak is obtained by observing the length distribution of the reads mapping on the positive or negative strand, independently (Fig. 7A). The distribution of the reads on the viral genome (Fig. 7B) shows different peaks in position 1–21, 320–340, 425–450, 1850–1880 and 3695–3725 (repeated in the second monomer in Fig. 7B). Interestingly, peaks around position 320 and 425 are enriched in reads mapping on the negative sense of the viral genome, while the other peaks have reads that mostly map on the positive sense. In Fig. 7C, the distribution of the reads on the viral genome is shown separately for each size of the sRNAs, revealing that the peak around position 1 is enriched in reads with lengths of 22, 24, 25, 26 and 31 nt, peaks in positions 320 and 425 are enriched for 23, 24 and 25 nt reads, the peak at position 1850 is rich in reads between 18 and 21 nt and peak 3695 is particularly rich in reads of 23 nt. Lastly, by plotting a heatmap of the base frequencies for each position of the sRNA length, a prevalence of U is seen at the first nucleotide of the whole population of viral sRNA (Fig. 7D). Indeed, 112,304 out of the 149,432 sRNAs mapping on the ambivirus start with a U, confirming a prevalence for this nucleotide in 75% of the cases. Collectively, these data provide compelling evidence that FgAV1 replication triggers a robust and canonical RNA interference response in F. graminearum. Our data provide evidence suggesting that the fungal antiviral machinery based on sRNAs could be involved in the targeting of the infection by circular RNA viruses as already described for linear viruses.

Fig. 7. Small RNA profiling of FgAV1-infected F. graminearum reveals antiviral RNAi activity.

(A) Length distribution of sRNAs mapping to the positive-sense (genomic, in red) and negative-sense (antigenomic, in green) strands of FgAV1. (B) sRNA distribution on a dimer of the genomic sequence of FgAV1, red peaks show sRNAs mapping on the positive sense and green peaks show sRNA mapping on the negative sense. (C) Size-class specific distribution of sRNAs along the FgAV1 genome. sRNAs of specific length (on the right) are plotted to show length-specific peaks of sRNAs mapping on a dimer of the genomic sequence of FgAV1. (D) Nucleotide frequency heatmap of the viral sRNAs mapping on the genome of FgAV1. For each position in the sRNAs length, the frequency of each nucleotide is shown as a scale of intensity from zero (black) to one hundred (red).

FgAV1 infection reduces fungal growth and virulence

To evaluate the biological impact of FgAV1 on its fungal host, we compared the growth and pathogenicity of the virus-free strain Fg1731 with three distinct Fg1731 colonial isolates derivatives that had acquired FgAV1 via hyphal fusion with the donor strain Dimer (Fg1731/FgAV1). Colony morphology of virus-free and virus-infected strains on PDA plates showed no macroscopic differences in appearance (Fig. 8A). However, mycelial expansion of the infected isolates was consistently slower than that of the virus-free control at both 48 h and 72 h (Fig. 8A). Quantitative measurements showed that virus-infected colonies exhibited reduced radial expansion compared to virus-free controls at both 48 h (14% reduction, P < 0.0001) and 72 h (10% reduction, P < 0.0001) (Fig. 8B). This growth suppression was further corroborated by a 9% decrease in hyphal extension rate (P < 0.05) in infected strains (Fig. 8C). In addition to the impairment in vegetative growth, we assessed reproductive capacity by quantifying conidiation. The virus-infected strain Fg1731/FgAV1 produced 32% fewer conidia after 7 days in Synthetic Low Nutrient (SLN) medium compared to the virus-free Fg1731 control (P < 0.0001) (Fig. 8D). Critically, FgAV1 infection also led to a significant reduction of fungal virulence on durum wheat. When conidial suspensions were inoculated into wheat spikelets, the disease severity observed on heads inoculated with Fg1731/FgAV1 was 73% lower than that on heads inoculated with the virus-free Fg1731 strain (P < 0.001) (Fig. 8E). This reduced virulence was reflected in a significant preservation of yield, with kernels from Fg1731/FgAV1-inoculated durum wheat showing a 29% increase in weight compared to those from the Fg1731-inoculated controls (P < 0.05) (Fig. 8F). Collectively, these results revealed that while colony morphology remained largely unaffected by viral infection, FgAV1 significantly impaired hyphal growth kinetics, conidiation, and virulence, key determinants of fungal fitness and pathogenicity. We also evaluated DON production in rice culture comparing three distinct anastomosis-infected isolates to the virus-free recipient strain. Only two of the three virus-infected isolates showed a marked inhibition of DON accumulation (fig. S8), showing some variability that could not be accounted simply by the presumed stability of the viral or host genomic features.

Fig. 8. Effects of FgAV1 on fungal morphology, growth, and virulence.

(A) Colony morphology of Fg1731 and FgAV1 transfectant (Fg1731/FgAV1) cultured on PDA in the dark at 48 h and 72 h. (B) Comparison of the colonial diameter among the strain Fg1731 (n = 19 replicates) and Fg1731/FgAV1 (n = 29 replicates). (C) Growth rate after 3 days on PDA. (D) Conidia production after 7 days in SLN medium (n = 6 replicates). (E) FHB symptoms and disease severity in the wheat heads infected with the strain Fg1731 (n = 6 replicates) and Fg1731/FgAV1 (n = 9 replicates). At 18 days after inoculation, there was a significant reduction in disease severity in the wheat heads infected with the Fg1731/FgAV1 strain. (F) Kernel weight at maturity in the wheat heads infected with the strain Fg1731 (n = 18 replicates) and Fg1731/FgAV1 (n = 27 replicates). Significant increase in kernel weight was observed in the wheat heads infected with the Fg1731/FgAV1 strain. Asterisks indicate levels of statistical significance (t test): *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

DISCUSSION

The goal of this study was to establish a reverse genetics platform for the ambivirus FgAV1 and use it to define the viral elements required for infection, replication, transmission, host interaction, and potential biocontrol relevance. We focus on five major points: (i) the significance of FgAV1 within the expanding diversity of circular RNA agents; (ii) the establishment of the first infectious ambivirus clone; (iii) the essential roles of ORF-B and ribozymes in replication; (iv) the unexpected replication competence of reverse-oriented constructs; and (v) the interaction of FgAV1 with fungal RNAi and its hypovirulence potential.

Circular RNA infectious agents were once considered rare biological exceptions, but metatranscriptomics has revealed a broad and diverse world of circular or viroid-like RNAs across ecosystems, including evidence for replication in bacteria, plants, fungi, and occasional animal hosts (13, 31). This expanding group includes classical plant viroids and Hepatitis Delta Virus (HDV), the historical prototype of circular RNA viruses in humans (32), as well as recently described agents such as ambiviruses, which encode viral RdRPs (10, 11), Obelisks, which encode proteins of unknown function (33), and Zeta viruses, which contain unusual open reading frames (34). Circularity has also been occasionally reported in some Chuviridae, despite their classification as negative-sense riboviruses and their lack of recognizable ribozymes (10, 12, 35, 36). Similarly, some mitoviruses contain ribozymes in both orientations and may possess circular genomes (10, 12), with RNAse R-based evidence available in at least one case (37). These examples suggest that ribozyme-dependent circularity may have evolved multiple times among RNA viruses.

A major obstacle to functional studies of these circular RNA elements has been the absence of infectious clone systems. Here, we established the first infectious cDNA clone for an ambivirus, FgAV1. To our knowledge, this also represents the first reverse genetics system for a circular RNA virus within Riboviria, excluding HDV, which is considered a satellite virus replicated by host DNA-dependent RNA polymerase machinery. This system provides a powerful tool to dissect FgAV1 biology, host interaction, and transmission, and it may guide reverse genetics approaches for other circular RNA replicons. From an applied perspective, this is particularly relevant because some mycoviruses are known hypovirulence factors, with CHV1 serving as a successful biological control precedent against chestnut blight in Europe (15–17). The availability of an infectious FgAV1 clone therefore opens both mechanistic and translational opportunities.

The reverse genetics system allowed us to define three essential components of FgAV1 replication: the catalytic GDD triad in ORF-A, ORF-B and the ribozymes. Ambiviruses may have originated from ancient duplication of an RdRP-encoding segment, producing complementary regions able to fold into rod-shaped RNA structures (38); in this model, ORF-B may represent an exapted derivative of ORF-A, with some ambiviruses later splitting this region into ORF-B1 and ORF-B2 (39, 40). Our mutational data show that ORF-B is not a passive genomic feature: deletion of its start codon abolished replication in both forward and reverse constructs. Structural predictions suggest that ORF-B contains RNA recognition motif-like domains, supporting a possible RNA-binding function (41). These RRM-like motifs are consistent with the structural plasticity of canonical RRM domains (42, 43), and one ORF-B region resembles an RdRP palm-like domain lacking catalytic residues, consistent with hypotheses that RdRP palm domains derive from ancestral RRM modules (44). ORF-B may therefore contribute to FgAV1 replication by mediating RNA binding, protecting the viral genome, promoting helicase-like activity, or assisting replication-complex assembly; more speculatively, it could also participate in RNA circularization. The possibility of RNA protection is supported by the observation that Ceratobasidium ambivirus 1 accumulates in microsomal fractions rather than as soluble naked RNA (45), whereas a helicase-like role is plausible because RRM-containing DEAD-box helicases occur in some RNA viruses, including hypovirids (46). Alternatively, ORF-B might assist RNA ligation, a function performed by chloroplast tRNA ligase during Avsunviroidae replication (47).

The ribozymes were equally indispensable. Ribozymes are defining features of viroid-like replicons and catalyze self-cleavage during rolling-circle replication (48). Previous work showed in vivo ribozyme activity and circular ambiviral RNA accumulation using 5′ RACE and northern blot assays (10), but direct genetic evidence for ribozyme essentiality was lacking. Here, deletion of ribozyme motifs, as well as mutation of a catalytic residue in the hammerhead ribozyme, abolished viral replication. These results demonstrate that ribozymes are functional requirements rather than structural hallmarks. Thus, FgAV1 appears to use a hybrid replication strategy: ORF-A likely synthesizes multimeric RNA intermediates, ribozymes process these intermediates, and ORF-B provides an essential cofactor function. This differs from canonical RNA viruses, which rely primarily on protein enzymes (49), and from viroids, which rely on ribozymes and host enzymes (50).

One unexpected result was that reverse-oriented dimeric constructs were also infectious. Because positive and negative orientations in ambiviruses are assigned according to the RdRP-coding strand, reverse-oriented transcription should not directly produce an ORF-A-translatable RNA. Nevertheless, reverse dimers initiated replication and generated transmissible virus. This suggests that ambiviral sequences or plasmid contexts may contain cis-acting elements recognized by host transcriptional machinery, or that alternative transcriptional events can overcome orientation constraints. Importantly, mutating ORF-A catalytic residues abolished replication, confirming that the viral RdRP remains essential. The reverse-orientation result therefore reveals unusual flexibility in replication initiation from cDNA but does not eliminate the requirement for ORF-A function.

FgAV1 also interacts with host antiviral defenses. Fungal RNAi is a conserved antiviral pathway that processes viral double-stranded RNA into small interfering RNAs (51, 52). FgAV1-derived sRNAs showed a predominant 21-nt peak and near-symmetric mapping to both strands, resembling the response to Fusarium graminearum virus 3, which induces FgDICER-2 and FgAGO-1 (53). The uneven distribution of vsiRNA hotspots also parallels other mycoviruses, likely reflecting viral RNA structures favorable for Dicer processing (53, 54). The strong 5′ U bias suggests recruitment into AGO-containing silencing complexes, a hallmark of antiviral RNAi across kingdoms (55, 56), also reported in mycovirus-infected Botrytis cinerea (57) and Trichoderma species (28). Despite this RNAi pressure, FgAV1 maintained stable replication and horizontal transmission, similar to CHV1, which persists in C. parasitica partly through the p29 silencing suppressor (58–60), and to F. graminearum viruses such as FgV1, which may interfere with FgAGO-1 function (53). Biologically, FgAV1 reduced fungal growth, conidiation, and wheat virulence, supporting its potential as a hypovirulence factor and possible biocontrol agent, analogous in principle to CHV1-based chestnut blight control (15–17).

Several questions remain unresolved. The biochemical function of ORF-B is still unknown, the mechanism allowing reverse-oriented constructs to initiate replication requires clarification, and the apparent host-range restriction of FgAV1 limits immediate application. Future work should define ORF-B interactors and enzymatic or RNA-binding activities, identify host factors involved in replication and circularization, test whether ORF-B modulates RNAi, and evaluate FgAV1 stability, transmission, and disease-control efficacy under field-relevant conditions. Furthermore, FgAV1 infection of RNAi knock out mutants would clarify the specific F. graminearum host components involved in generating the sRNA response we here characterized.

MATERIALS AND METHODS

Fungal strains and culture conditions

Strain F. graminearum 266 MV (Fg266 MV), harboring FgAV1, was isolated in 2018 from a naturally infected maize ear collected in Pocking, Bavaria, Germany (25). The virus-free F. graminearum strain, Fg1731 (isolated from durum wheat from Cadriano, Italy in 2023), served as the wild-type and as recipient strain for infectious clones constructed in this study. All strains, including the transformants generated here, and the strains arising from Fg1731 after dual culturing with strains Dimer, Rev-dimer, ΔAUG, Rev-ΔAUG and ΔRz, respectively, are listed in table S1 and were cultured at 26°C on potato dextrose agar (PDA) medium for 3–5 days for morphological observation or on cellophane-covered PDA medium for nucleic acid extractions. A collection of different Fusarium species used in co-culture experiments for ambivirus transmission is displayed in table S3 and is part of the collection present at the plant pathology laboratory of the University of Sassari (Italy). One isolate of F. pseudograminearum (CAL47b, isolated from durum wheat in Italy), one isolate of Trichoderma spirale (T45) and one of T. atrobrunneum (T99), both from uncultivated soils in Sardinia (61), have been used to test the host range specificity of FgAV1 by transformation.

RNAseq analysis of the fungal isolates

Isolate Fg1731 was pre-selected for absence of dsRNA in a preliminary screen. To confirm the absence of viruses and viroid-like RNAs infecting the recipient isolate, the total RNA of F. graminearum Fg1731 was included in an RNA pool that was subsequently sequenced by RNAseq (Illumina). The RNA of the ten fungal isolates pooled in the sample were equally represented. The ribosomal depletion and cDNA library were performed with the TruSeq Stranded Total RNA Ribo-Zero H/M/R Gold kit (Illumina), and after the library preparation, 50 million reads in paired end were generated on a Novaseq X apparatus (Illumina). The obtained raw reads were cleaned with Bbtools (https://archive.jgi.doe.gov/data-and-tools/software-tools/bbtools/) following a pipeline from the JGI (dx.doi.org/10.17504/protocols.io.gydbxs6). Obtained clean reads were assembled through SPAdes version 3.15.3 (62) and the contigs generated were used to detect the presence of viruses and viroid-like RNAs. For viral and viral ORFan detection, a pipeline based on DIAMOND (63) blastx against the nr NCBI database (version April 2024) was applied as previously published (64). For the detection of viroid-like RNAs, a pipeline based on the identification of ribozyme containing circular contigs was applied as also described in previous work (10). All the viral and viroid-like sequences detected with this approach were then associated with the fungal isolate of origin by RT-PCR to allow the association with the host. The raw reads of the pooled sample including Fg1731 isolate are deposited in the NCBI SRA database (SRR33245417). The same approach was applied to the F. graminearum Fg1731/Dimer isolate to obtain the consensus sequence of FgAV1 after transfection and in a transgene-free context. The raw reads generated from this sample are deposited in the NCBI SRA database (SRR33245416).

Construction of FgAV1 infectious clones

To generate the constructs for the analysis of promoter activity, the coding region of GFP was amplified from the plasmid R3-p21-TripB-GFP (65) using primers GFP-StuI-For and GFP-NotI-Rev (table S2), which introduced StuI and NotI restriction sites. In parallel, the mCherry coding sequence was amplified from the plasmid p-mCherry (a generous gift of Prof. Liying Sun, from Northwestern University, Xian, China) with primers mCherry-EcoRV-For and mCherry-NotI-Rev (table S2) engineered with the same restriction sites. The PCR fragments were first subcloned into the pCR-Blunt vector (Invitrogen, Carlsbad, CA, USA) for sequence verification. After digestion with StuI and NotI, the verified inserts were ligated into the corresponding sites of pHX9-nv, which was generated from the parental plasmid pHX9 by inverse PCR with primers CHV1–25-Rev and CHV1–12700-For to delete the genome of CHV1 (16). This backbone contains unique KpnI, StuI, and NotI sites for insertion of foreign sequences under the control of the C. parasitica gpd-1 promoter and terminator and also carries the Escherichia coli hph gene as a selectable marker driven by the Aspergillus nidulans trpC transcriptional elements, yielding the reporter plasmids pHX9-GFP and pHX9-mCherry.

To construct the infectious clone of FgAV1, the full-length viral genome monomers were amplified from the strain Fg266 MV using primers FgAV1-875F-NotI and FgAV1-870R-KpnI, as well as FgAV1-775F-KpnI and FgAV1-770R-NotI (table S2). These primers introduced KpnI and NotI restriction sites at the ends of each monomer, being KpnI a restriction site present in FgAV1 sequence. The resulting PCR fragments were first ligated into the pCR-Blunt vector, generating pCR-Blunt-FgAV1-Lig1 and pCR-Blunt-FgAV1-Lig2, and sequence-verified by Sanger sequencing. Following verification, pCR-Blunt-FgAV1-Lig1 and pCR-Blunt-FgAV1-Lig2 were digested with KpnI and NotI, and the resulting full-length fragments were ligated to generate a tandem dimer. This dimer was then inserted into NotI-digested pHX9-nv, and the orientation of the insert was confirmed by sequencing, yielding both the forward and reverse dimers, named pHX9-FgAV1-dimer and pHX9-FgAV1-Rev-dimer, respectively. Additionally, a single monomer fragment excised from pCR-Blunt-FgAV1-Lig1 was ligated into KpnI/NotI-digested pHX9-nv to serve as a negative control, designated pHX9-FgAV1-monomer.

Building on the construction of the FgAV1 monomer and dimer clones, several targeted mutants were generated using a unified strategy. To delete the start codon (AUG) of ORF-B or remove the ribozyme sequence, full-length monomer fragments were amplified from both pCR-Blunt-FgAV1-Lig1 and pCR-Blunt-FgAV1-Lig2 using primers DeltaAUG-ORF-B-For/DeltaAUG-ORF-B-Rev (table S2) for deleting AUG of ORF-B and DeltaRibozymeplu8s-For/DeltaRibozymeplu8s-4534Rev (table S2) for deleting the ribozymes. For deleting 9 nucleotides corresponding to the cDNA coding for the GDD residues of the ORF-A, we used primers ΔGDD-ORF-A-For and ΔGDD-ORF-A-Rev (table S2). Finally, to erase 1 nt predicted to be crucial for the ribozyme self-cleavage activity, primers Δ-A-Rz-For and Δ-A-Rz-Rev (table S2) were used. In the case of the ΔGDD mutant, only pCR-Blunt-FgAV1-Lig2 was used as template, since the other monomer is already resulting in a truncated ORF-A monomer in the construct. The PCR products were phosphorylated and circularized using T4 DNA ligase, generating the plasmids pCR-Blunt-FgAV1-ΔAUG-ORFB-Lig1, pCR-Blunt-FgAV1-ΔAUG-ORFB-Lig2, pCR-Blunt-FgAV1-ΔRz-Lig1, pCR-Blunt-FgAV1-ΔRz-Lig2, pCR-Blunt-FgAV1-ΔGDD-ORFA-Lig2, pCR-Blunt-FgAV1-Δ-A-Rz-Lig1 and pCR-Blunt-FgAV1-Δ-A-Rz-Lig2: all the monomers were verified by Sanger sequencing. For each verified monomer mutant, KpnI and NotI digestion was performed, and the resulting corresponding fragments were ligated ends to form dimers, which were subsequently inserted into the NotI site of pHX9-nv. Sequencing confirmed both forward and reverse orientations for the ΔAUG-ORFB dimer mutants, designated pHX9-FgAV1-ΔAUG-ORFB and pHX9-FgAV1-Rev-ΔAUG-ORFB, while the ΔRz dimer was similarly assembled into pHX9-nv, generating pHX9-FgAV1-ΔRz. Finally, the ΔGDD-ORFA dimer and Δ-A-Rz dimers generated pHX9-FgAV1-ΔGDD-ORFA and pHX9-FgAV1-ΔA-Rz, used only in the positive sense orientations for protoplast transformation.

Protoplast preparation and transfection assays

Two plugs of virus-free F. graminearum Fg1731 from a PDA plate were added to 200 mL of SLN medium containing 9.9 mM KNO₃, 7.3 mM KH₂PO₄, 0.3 mM MgSO₄·7H₂O, 6.7 mM KCl, 1.1 mM glucose, and 0.6 mM sucrose, and cultivated at 26°C with shaking at 130 rpm for 5 days. The culture was subsequently filtered and centrifuged to gather macroconidia. One ml of macroconidia suspension (containing approximately 10⁷ macroconidia) was used to product protoplasts following a previous method according to Moradi et al. (66). For F. pseudograminearum, conidia have been harvested from well sporulated PDA and protoplasts have been obtained according to (67), while the subsequent transformation steps were the same for both strains, and were accomplished utilizing the Polyethylene glycol (PEG)-mediated method, as previously described (66) with minor modifications. For the transformation, 500 μL PEG solution containing 10 mM Tris-HCl, 50 mM CaCl₂, 60% (W/V) PEG 4000, and 20 μg (∼50 μL) of recombinant plasmids were mixed with 500 μL of protoplasts and incubated at room temperature for 20 min. Subsequently, 10 mL of PEG solution and 20 mL of STC buffer were added for preparing protoplast mixture. Each 3 mL of protoplast mixture was added to 17 mL of melted (50°C) regeneration medium, spread onto plates, and then incubated at room temperature for 16–18 h. Subsequently, the upper culture medium (10 mL of regeneration medium containing 900 μg/mL hygromycin B was spread on the top of each plate, and the cultures were incubated for 3–5 days at 26°C in the dark for emerging transformants. Transformants were purified by three consecutive transfers from the edges of colonies grown on plates containing 300 μg/mL hygromycin B. The two Trichoderma sp. isolates were transformed following the method described by Formiglia et al. (28).

Nucleic acid extraction and analyses

Total RNA was extracted using the Spectrum Plant Total RNA Kit (Sigma-Aldrich, St. Louis, MO, USA) according to the manufacturer’s instructions. For F. graminearum transformants, 1 μg of total RNA was treated with RNase-free DNase I (Invitrogen, Carlsbad, CA, USA) to remove genomic and transgene DNA. Both untreated and DNase-treated RNAs were analyzed by PCR using Phusion High-Fidelity DNA Polymerase (New England Biolabs, Ipswich, MA, USA). RNA from virus-infected strains obtained through dual culture was further examined by RT-PCR. First-strand cDNA was synthesized using MultiScribe Reverse Transcriptase (Applied Biosystems, Foster City, CA, USA) and subsequently amplified as described above. RNAs from strains Fg266 MV and Fg1731 were reverse-transcribed into cDNA and used as positive and negative controls, respectively. All primers used in this study are listed in table S2.

Northern blot hybridization assays

Northern blot hybridization was performed as previously described (10). DNase-treated RNA from transformants strains or RNA from virus-infected strains was separated on 1% denaturing agarose gels prior to hybridization. Detection was carried out using digoxigenin (DIG)-labeled riboprobes specific for the (+) or (−) strands of FgAV1, which were synthesized by in vitro transcription from linearized plasmids using the Dig-RNA labeling mix (Roche Diagnostics GmbH, Germany). Chemiluminescent signals were visualized with a Chemidoc Touch Imaging System (Bio-Rad, Hercules, CA, USA). RNAs from Fg266 MV and virus-free Fg1731 were included as positive and negative controls, respectively.

Confocal microscopy observation

Fungal isolates were grown on PDA plates without antibiotic and slides were prepared from aerial mycelia diluted in water collected when fungal colonies reached the margin of the plates. Fluorescent signals were acquired using a Zeiss LSM 900 confocal microscope equipped with an achroplan 10X/0.25 objective. GFP was imaged using excitation at 488 nm and emission collected between 500–545 nm, while mCherry was detected with excitation at 561 nm and emission collected between 595–620 nm. Image processing and analysis were carried out with ZEN 2.3 software (Carl Zeiss AG, Oberkochen, Germany).

Horizontal transmission assays

To assess the possibility of horizontal transmission of FgAV1 between strains of F. graminearum, fresh mycelial agar discs from several donor strains (Dimer, Rev-Dimer, ΔAUG-ORFB, Rev-ΔAUG-ORFB, and ΔRz) and the recipient strain Fg1731 were co-cultured on PDA plates. Each donor-recipient pair was cultured together in a 15 cm-diameter plate. After 28 days of co-culture, two mycelial derivative isolates were collected from the margin of each recipient colony and transferred to fresh PDA plates. Viral presence in the derivative isolates was screened by RT-PCR using primers FgAV1-Screen1–1555-For and FgAV1-Screen2–1980-Rev (table S2) and subsequently confirmed by northern blot analysis. Co-cultures were also assembled to test transmission among different Fusarium species, summarized in table S3.

Small RNA sequencing and bioinformatic analysis

sRNA library preparation (Illumina TruSeq Small RNA Library Prep Kit) and sequencing was performed by Novogene Europe (Germany) starting from the total RNA of the isolate of interest; sequencing was performed with Illumina Sequencing SE50 platform. Raw reads were submitted to Fasqc to detect the presence of Illumina adaptors. After identifying the adaptor used for the library, Cutadapt (68) was used to remove the adaptor sequence and to filter for sequences longer than 18 nt and with a quality threshold higher than 20. Clean reads were mapped on a dimeric sequence of the ambiviral genome through bowtie (69) allowing one mismatch with the -v alignment mode. The aligned reads were sorted with the --al flag for further analysis. The barplot of the length distribution of the viral reads was generated by loading the mapped reads file in R (70) through the Biostrings library (71) and using the width information to build the graph through the ggplot2 library (72). The total number of reads mapping on the ambivirus genome was calculated by transforming the SAM output of the bowtie alignment into a sorted BAM file and by indexing it through SAMtools (73). With SAMtools view, the sorting of the positive and negative mapping reads was done using the flag -f 0x10 to sort the negative mapping reads and -F 0x10 to sort the positive mapping reads. The analysis related to the reads distribution on the viral genome, the reads distribution for each strand and the heatmap showing the nucleotide frequencies for each nucleotide in the read sequence were done using the R package viRome (74) which is specifically designed to perform analysis of viral sRNAs. The sRNA reads were deposited in the NCBI with the following ID (SRR36219901).

In silico prediction and analysis of FgAV1 ORF-B encoded protein 3D structure and FgAV1 RNA genomic fold

ORF-B encoded protein sequence was used to retrieve 10 homologues from the NCBI protein database, including the most distantly related. The proteins were aligned with PROMALS3D (75) using standard parameters. A protein 3D structure model was derived with Alphaphold3 (76). ChimeraX was used to display graphically the structures (77). Protein 3D structure database searches were performed using FoldSeek (78) searching the PDB100 20240101 and BFVD (26) 2023_02 and AlphaFold/UniProt50 v6 Alphafold/Swiss-Prot V6 and AlphaFold/Proteome V6 databases using the TM-align mode.

RNAfold was used to derive the secondary structure of the circular genomic RNA (WT and ΔAUG-ORFB) (79).

Biological characterization of Fg1731/FgAV1 strains

We assessed the mycelial morphology, growth, conidiation and virulence on wheat spikes of Fg1731 derivatives carrying the virus after co-culture with strain Dimer (Fg1731/FgAV1) and its virus-free strain Fg1731. After fungal isolates were grown on PDA plates for 3 days at 26°C, agar plugs (5 mm diam) were taken from the actively growing margins and placed at the center of fresh PDA plates. Colony diameters were measured at 48 h and 72 h to determine mycelial growth. The growth rate of fungi was calculated by measuring colonial diameters daily after culturing for 3 days at 26°C. To evaluate conidiation, two mycelial plugs of each isolate (virus-free Fg1731 and Fg1731/FgAV1) were inoculated into 200 mL of SLN liquid medium and cultivated at 26°C with shaking at 130 rpm for 7 days. After incubation, the cultures were centrifuged at 10,000 g for 5 min to pellet the conidia. The conidial pellets were then resuspended in 2 mL of water, and conidial production was quantified using a hemocytometer under a light microscope.

For the virulence assay, 250 μL of the conidial suspension (1x10³ conidia/mL) of each isolate (virus-free Fg1731 and Fg1731/FgAV1) was sprayed onto each single head of durum wheat (cv Iride) at anthesis stage. All inoculated heads were enclosed in plastic bags for 48 hours to promote infection. The experiment included two biological replicates for the virus-free Fg1731 and three for Fg1731/FgAV1, with each biological replicate consisting of three technical replicate heads. Disease severity was assessed at 18 days post-inoculation using a 0–6 symptom severity scale and was calculated using the McKinney index (80) which expresses the percentage of the maximum severity of disease (i.e., 100) according to the formula I = Ʃ[(cf)/nN] 100, where c = disease class, f = frequency, n = number of observations, and N = the highest value of an empirical scale adopted, estimating the severity of symptoms on the stem (i.e., 6). At maturity, kernels from each head were manually collected and weighed. DON production in rice culture was evaluated following methodology previously described (25).

Statistical analysis

Analyses of the data referred to the measures of colony growth, conidia production or diseased size in the virulence assay on the wheat spikelets were performed with Student’s t test analysis of variance using GraphPad Prism version 8.0. Values are shown as means ± standard deviation. Significant differences occurred when P < 0.05. The sample size (n) for each statistical analysis has been reported in the corresponding figure legends.

Supplementary Materials

This PDF file includes:

Figs. S1 to S8

Tables S1 to S3

Data S1