A dsRNA virus with filamentous viral particles
A complex pattern of dsRNAs in C. Camelliae strain LT-3-1
Nucleic acid preparations enriched in dsRNA were obtained from mycelia of C. camelliae strain LT-3-1 and analyzed by agarose gel electrophoresis. A complex pattern of eight bands was detected in LT-3-1 preparations before and after digestion with DNase I or S1 nuclease (Fig. 1a). Assuming that these bands were generated by dsRNAs, their corresponding sizes were between 2500 and 900 bp as estimated by agarose gel electrophoresis using both dsDNA and dsRNA markers (Fig. 1a). These RNAs were not detected in a typical strain-like DP-3-1 (Fig. 1a).
The dsRNA nature of the eight observed bands was assessed by treatments with RNase III, S1 nuclease, or RNase A (in 2× and 0.1× SSC), together with an ssRNA control (in vitro dimeric transcripts of citrus exocortis viroid (CEVd)) and dsRNAs from a dsRNA mycovirus (Botryosphaeria dothidea chrysovirus 1, BdCV 1). The RNAs extracted from strain LT-3-1 together with BdCV 1 dsRNAs were digested into ~20 bp-sized fragments by RNase III and degraded by RNase A in 0.1× SSC, but they resisted digestion by S1 nuclease and RNase A in 2× SSC. In sharp contrast, CEVd transcripts were completely degraded by S1 nuclease and by RNase A under both ionic conditions, but resisted digestion by RNase III (Fig. 1b). These data strongly support that the RNAs extracted from strain LT-3-1 were indeed dsRNAs (hereafter termed dsRNAs 1–8 according to their decreasing size).
Complete sequence and genomic organization of dsRNAs 1–8
The sequences of the full-length complimentary DNAs (cDNA) of dsRNAs 1–8 were determined by assembling partial cDNAs amplified from each individually purified dsRNAs using real-time (RT) PCR with tagged random primers and RACE13, 14. Regarding the shortest dsRNAs 7 and 8, following ligation of a 3′-closed adaptor (46 nt) to the 3′ end of each RNA strand, they reverse transcribed using a primer complementary to the adaptor (positions 46–26), PCR-amplified with a second primer complementary to the adaptor (positions 34–17), and reamplified by nested PCR with a third primer complementary to the adaptor (positions 17–1). Cloning and sequencing the amplified products confirmed their full-length size. The sizes obtained for dsRNAs 1–8 were 2444, 2253, 2012, 1299, 1122, 1085, 1053 and 990 bp, respectively; the corresponding sequences have been deposited in GenBank under accession numbers KX778766-KX778773. Each dsRNA contains one (dsRNAs 1–4 and dsRNAs 6 and 7) or two (dsRNAs 5 and 8) putative open reading frames (ORFs) on one of the strands; the ORFs were named according to the dsRNA on which they are located, e.g., ORF 5–1 is the first ORF on dsRNA 5 (Fig. 1c). BLASTn searches revealed that dsRNA1 sequence shared low similarities (3.0E−44–2.0E−10; 25–41% coverage; 68–82% identity) to the dsRNA1 of four mycoviruses including two uncharacterized ones: Alternaria tenuissima virus (ATV, accession nos. KP067914), Cladosporium cladosporioides virus 1 (CcV-1, KJ787686), AfuTmV-1, HG975302, and Botryosphaeria dothidea virus 1 (BdRV1, KP245734). However, the remaining RNAs (2–8) shared no detectable similarities with viral RNA sequences.
The 5′-untranslated regions (5′-UTRs) of the coding strand of dsRNAs 1–8 ranged from 34 to 184 nucleotides (nt) and shared 19.1–76.5% identity with each other; the corresponding 3′-UTRs ranged from 63 to 339 nt and shared 19.4–45.3% identity (Fig. 1c; Supplementary Table 1). A conserved sequence (G/CGAUAAUAAN(12)G/CCUA/UGN(5)C), characteristic of dsRNA segmented viruses1, was observed in all the 5′-termini of the coding strand of dsRNAs 1–8. In contrast, the sequence ACCUUAA was conserved only in dsRNAs 1–5 (Fig. 2a), and other sequences were also conserved among some of the dsRNAs, e.g., UCCU between dsRNAs 2 and 5, and GAGAAU/ACAGAGU/CG between dsRNAs 7 and 8 (Fig. 2a). No strict sequence conservation was observed in the 3′-termini of the coding strand of the eight dsRNAs. The tetranucleotide CCCC is shared only by dsRNAs 1, 2, 3, and 8, and the sequence GGN(6)U only by dsRNAs 4, 6, and 7 (Fig. 2b). It is worth noting that dsRNA 5 shares no sequence at the 3′-terminus conserved with the other dsRNAs (Fig. 2b) (confirmed by eight independent 3′-RACE experiments). In addition to the predominant ones, other nucleotides at the termini of each dsRNA were also detected at low frequencies in the RACE experiments, as indicated in Fig. 2a and b together with their frequencies. The sequences conserved in both termini of dsRNAs 1–8 are distinct from those observed in the known related viruses including ATV, CcV-1, AfuTmV-1, BdRV1, BbPmV-1, and BbPmV-2 (e.g., the conserved nucleotides at the 3′-termini for AfuTmV-1, BdRV1, BbPmV-1 and BbPmV-2 are GGGG, UGGGGU, UAGUUUU, UUUU, respectively9, 12, 15). Collectively, these data strongly suggest that the dsRNAs 1–8 associated with C. camelliae strain LT-3-1 are the genomic components of a novel dsRNA virus, which we propose to name “Colletotrichum camelliae filamentous virus 1” (CcFV-1) considering the morphology of their virions (see below).
Putative proteins encoded by the CcFV-1 genome
BLASTp searches revealed that the ORFs of dsRNAs 1–4 of CcFV-1 code for proteins that share amino-acid sequence identities of 47–26% (6.0E−171–1.0E−27, 99–74% coverage) with those encoded by dsRNAs 1–4 of ATV (accession no. KP067914), CcV-1 (KJ787686), AfuTmV-1 (HG975302), and BdRV1 (KP245734) (Supplementary Table 2). The predicted proteins (P1, P2, P3, and P4) encoded by ORFs 1, 2, 3, and 4 of CcFV-1 share amino-acid sequence identities of 47%, 32%, 26%, and 41% with the RNA-dependent RNA polymerase (RdRp), a hypothetical protein, S-adenosyl methionine-dependent methyltransferase capping enzyme, and a PAS-rich protein (PASrp, containing a high proportion of proline (P), alanine (A), and serine (S) residues) of AfuTmV-1, respectively (Supplementary Table 2). Additionally, protein P2 of CcFV-1 displays a sequence identity of 31% (E-value 1.0E−55, coverage 92%) with a hypothetical protein of unknown function from Alternaria sp. (ACL80752) (Supplementary Table 2). However, the remaining ORFs of CcFV-1 dsRNAs 5–8 showed no detectable similarity with sequences encoded by viral RNAs (Supplementary Table 2). It is worth noting that CcFV-1 ORF 5 neither shares detectable identity with the ORFs of dsRNA 5 of BdRV1 and CcV-1, nor with those of dsRNA 6 and −7 of BbPmV-2.
Putative protein functions for the CcFV-1 ORFs were also inferred by a homology search using the HHpred server, with the results obtained agreeing with those from BLASTp searches (Supplementary Table 3). CcFV-1 P1 was highly similar to the RdRp of (+)ssRNA viruses belonging to Caliciviridae (E-values ≤ 1.3E−47) and Picornaviridae (E-values ≤ 4.0E−40), which infect humans or animals (Supplementary Table 3). Sequence alignment also allowed the identification of three amino-acid motifs (IV, V, and VI) observed previously in AfuTmV-1, BdRV1, BbPmV-1, BbPmV-2, and (+)ssRNA viruses, which are characteristic of supergroup 1 of the viral RdRp family, further supporting that P1 is an RdRp (Supplementary Fig. 1). Interestingly, motif VI also contains the GDN triplet followed by Q rather than G/ADD (which is normally invariant for (+)ssRNA viruses). Moreover, a phylogenetic reconstruction of P1 with RdRps of representative members of dsRNA virus families, including unclassified dsRNA viruses, and of three (+)ssRNA virus families (the closest (+)ssRNA viruses according to BLASTp analysis), clustered CcFV-1 together with AFuTmV-1, BdRV1, BbPmV-1, and BbPmV-2. Such assignment separates these five viruses from other dsRNA viruses, placing them closer to (+)ssRNA viruses belonging to the Caliciviridae, as predicted by HHpred analysis, in an intermediate position between dsRNA and (+)ssRNA viruses (Supplementary Fig. 2; Supplementary Table 4). These data suggest that CcFV-1 may be an intermediate link between dsRNA and (+)ssRNA viruses, as proposed for AfuTmV-19. On the other hand, CcFV-1 P3 is highly similar to methyltransferase of diverse bacteria (e.g., Bacillus subtilis, Methanococcus maripaludis, and Escherichia coli) (probabilities of 99.8–99.7%, E-values 1.4E−17–1.3E−15), suggesting that P3 is most likely a S-adenosyl methionine-dependent methyltransferase capping enzyme (Supplementary Table 3). No function could be tentatively ascribed to the remaining CcFV-1 putative proteins due to a lack of reliable conserved motifs. Nonetheless, CcFV-1 P4 has a high proportion of P (7.6%), A (13.4%), and S (10.0%) residues, resembling the PASrps encoded by AfuTmV-1 (P 6.8%, A 8.0%, S 9.7%), BdRV1 (P 3.2%, A 9.7%, S 7.9%), BbPmV-1 (P 6.2%, A 13.5%, S 7.6%) and BbPmV-2 (P 11.2%, A 12.2%, S 8.8%)9, 12, 15.
Virus-like particles associated with CcFV-1 dsRNAs
To determine whether the dsRNAs associated with the strain LT-3-1 were encapsidated, mycelial crude preparations were examined by transmission electron microscopy (TEM), which revealed filamentous virus-like particles longer than 127.1 nm (Supplementary Fig. 3a). The presumed viral particles were purified by centrifugation in stepwise sucrose gradients (60–10% with 10% sucrose increments, or 40–15% with 5% sucrose increments), being predominantly found in the gradient zone from 20 to 30% sucrose (Fig. 3). Assuming that the observed short particles were most likely fragments of the longer ones, only those longer than 1000 nm were considered. Measuring 51 filamentous virus-like particles above that limit revealed widths ranging from 11.9 (virus particles number 3, 34, and 35 in Fig. 3b) to 17.7 nm (number 36 and 37) and lengths from 1000.5 (number 25) to 4426.8 nm (number 32) (Supplementary Fig. 3b; Fig. 3b). Some broken particles that appeared empty, most likely resulting from the loss of their dsRNA components, had inner widths ranging from 4.8 to 7.2 nm (Fig. 3a, indicated by arrows). The presumed viral particles were also examined following purification by centrifugation through a CsCl cushion, with the fraction containing the viral dsRNAs and capsid protein (CP) being ultracentrifuged to collect viral particles. TEM examination revealed similar filamentous particles with the longest one having a width of 16.3 nm and a length of 3266.6 nm (Supplementary Fig. 3c). Importantly, removal of the CcFV-1 dsRNA components (dsRNAs 1–8) from the mycelia of strain LT-3-1 when generating a new subisolate (named LT-3-1D2, see below), was accompanied by removal of the filamentous virus-like particles in the crude extract and all sucrose gradient fractions of subisolate LT-3-1D2, thus supporting a close association between them.
The dsRNAs and proteins composing the virus-like particles
Agarose gel electrophoresis of the nucleic acids extracted from the 10–60% sucrose gradient fractions (at 10% sucrose increments) showed that the typical pattern of CcFV-1 dsRNAs 1–8 was mostly recovered from the 20 and 30% fractions (Fig. 4a, right panel). More stringent sucrose gradient centrifugation, with 15–40% sucrose fractions and 5% sucrose increments, confirmed migration of the dsRNAs 1–8 into the 25% fraction; dsRNAs 1–4 were also found in the 20% fraction and dsRNAs 4–8 in the 30 and 35% fractions (Fig. 4a, left panel). These data suggest that dsRNAs 1–8 are separately encapsidated in the virus particles of CcFV-116. No dsRNA bands were recovered from any of the gradient fractions from mycelial extracts of strain LT-3-1D2 performed in parallel.
SDS-PAGE analysis of the proteins from the 40–15% sucrose fractions of strain LT-3-1 revealed the presence of three dominant bands with estimated molecular masses of 80, 70 kDa (p80 and p70, in the 15–20% fractions), and 31 kDa (p31, in the 20–30% fractions). However, while p31 was not observed by SDS-PAGE of the dsRNA-lacking subisolate LT-3-1D2 derived from strain LT-3-1, p80 and p70 were detected in the 15–20% sucrose fractions, thus suggesting that they are host-encoded proteins (Fig. 4b, left panel). To further examine the nature of p80, p70, and p31 (Fig. 4b, left panel, lanes 15–30%), they were eluted from the gel and subjected to peptide mass fingerprinting (PMF). The sequencing data showed no match between the peptide fragments from p80 and p70 (15 and 22, respectively) and the putative proteins encoded by dsRNAs 1–8. Instead, the sequence of these fragments matched proteins from Colletotrichum spp., confirming that p80 and p70 are fungal proteins (Supplementary Table 5). In contrast, all peptide fragments (12, 12, and 13 fragments) from p31 preparations separately eluted from 20%, 25%, and 30% sucrose fractions, respectively, matched (scores 692, 723, and 774, respectively) amino-acid regions (60–69%) of the putative P4 protein (Supplementary Table 6). This finding suggested that p31 is the structural protein P4 encoded by CcFV-1 ORF4 (see below).
P4 forms the capsid of the CcFV-1 virus particles
P4 was extracted from strain LT-3-1 mycelia, purified from SDS-PAGE, and injected into mice to trigger production of a polyclonal antibody (PAb) against this protein (PAb-P4). The antibody strongly and specifically recognized P4 from 20, 25, and 30% sucrose fractions of strain LT-3-1, while no reactivity was observed with protein extracts from strain LT-3-1D2 (Fig. 5a, left panel). These results indicate that the purified P4 protein used as antigen was indeed uncontaminated with host proteins. Western blotting and indirect enzyme-linked immunosorbent assay (ELISA) analysis revealed a titer of ~128,000-fold dilution for PAb-P4, with an optimum at 2000-fold dilution (Fig. 5a, right panel).
To further confirm that the CcFV-1 dsRNAs were encapsidated, the filamentous virus-like particles prepared by sucrose gradient centrifugation were subjected to immunosorbent electron microscopy (ISEM). The virus-like particles from the 20, 25, and 30% sucrose fractions of strain LT-3-1 were all clearly decorated by PAb-P4 (Fig. 5b, middle and right panels) compared with direct TEM observation (Fig. 5b, left panel). When crude extracts of strain LT-3-1 mycelia were subjected to ISEM analysis the filamentous virus-like particles were also clearly decorated by PAb-P4 (Fig. 5d; Supplementary Fig. 4a). However, similar particles were not observed in either crude extracts or in the sucrose gradient fractions from strain LT-3-1D2, supporting that P4 encapsidates the dsRNAs of CcFV-1. Measurements of 28 decorated virus-like particles longer than 1000 nm revealed widths ranging from 12.0 to 17.7 nm (Supplementary Fig. 4b), matching the sizes observed by TEM (Fig. 3b). It is worth noting that among the decorated virus particles in these preparations, the longest had a length of 4661.6 nm (and a width of 17.7 nm, virus particle number 28 in Supplementary Fig. 4b) (Fig. 5d) and the second longest length of 4590.9 nm (and a width of 16.5 nm, number 17) (Supplementary Fig. 4a), both of which were observed in a crude extract. To the best of our knowledge, these are the longest virus-like particles ever reported (Supplementary Table 7).
CcFV-1 induces phenotypic changes on its fungal host
Forty-seven subisolates generated from conidial colonies of strain LT-3-1 cultured on potato dextrose agar (PDA) were analyzed by dsRNA electrophoresis and RT-PCR amplification of a 492-bp fragment of CcFV-1 dsRNA 2. CcFV-1 was absent in the subisolates, suggesting no vertical transmission via conidia. The phenotypes of three CcFV-1-free subisolates (LT-3-1D1, LT-3-1D2, and LT-3-1D3) chosen at random, together with their growth rate and virulence, were observed on PDA and measured over a 5-day period. The three CcFV-1-free subisolates exhibited morphologies distinct from the parental strain LT-3-1, including presence of hyphal rings, reduction in black pigments, and normalization of pigment distribution at the bottom of the PDA plate (Fig. 6a). Moreover, the three CcFV-1-free subisolates displayed enhanced growth rates (6.3–6.5 vs. 5.4 mm/day) and virulence (6.1–14.9 vs. 1.0 mm lesion lengths) compared with strain LT-3-1 (Fig. 6b).
Ultra-thin hyphal sections of strains LT-3-1 and LT-3-1D2 were examined by TEM. Compared with strain LT-3-1D2, the cells of hyphal compartments of strain LT-3-1 showed abnormalities, including formation of large membranous vesicles that might correspond to viroplasms proposed for virus replication and assembly17, and reductions in Woronin bodies and vacuoles (Supplementary Figs. 5a and b). Virus-like particles with a width similar to those of the CcFV-1 were observed in the vesicles (Supplementary Figs. 5c and d). In addition, ill-defined structures were also observed in the cytoplasm of the hyphal cells of strain LT-3-1 (Supplementary Figs. 5e and f).
CcFV-1 dsRNAs are infectious and produce virus particles
To further corroborate that the CcFV-1 dsRNAs were encapsidated in filamentous virus particles, naked dsRNA components were transfected into C. camelliae cells to assess whether they could induce synthesis of new filamentous virus-like particles in vivo. Protoplasts prepared from the CcFV-1-free strain LT-3-1D2 were transfected with S1- and DNase I-treated CcFV-1 dsRNAs (1–8) purified from mycelia. A total of 129 subisolates regenerated from C. camelliae-transfected protoplasts were randomly subjected to dsRNA electrophoretic analysis, resulting in the presence of all eight transfected dsRNA bands in the mycelial extracts of only three subisolates (named LT-3-1T1, LT-3-1T2, LT-3-1T3), supporting that they were infected by CcFV-1 dsRNAs. When subisolate LT-3-1T2 was subjected to TEM and ISEM analyses, the filamentous virus-like particles were visible in fractions (20–30%) after sucrose gradient centrifugation of mycelial extracts (Fig. 5c, left panel), and were decorated by PAb-P4 (Fig. 5c, middle and right panels). Biological analysis showed different morphologies, compared with strain LT-3-1D2, for the CcFV-1-transfected subisolates (LT-3-1T1, LT-3-1T2, LT-3-1T3), including disappearance of hyphal rings and increase of black pigments, resembling the colony morphology of strain LT-3-1 (Fig. 6a). The abnormal distribution of black pigments and their high concentration were completely recovered in strain LT-3-1T2 in comparison with strain LT-3-1 (Fig. 6a, middle panel). For the three CcFV-1-infected subisolates, prolonged culture for more than 15 days resulted in the spreading of the black pigments over the bottom of the PDA plate, similarly to strain LT-3-1 but in contrast with all CcFV-1-free isolates including LT-3-1D1 to -D3 (Fig. 6a). The CcFV-1-infected subisolates also exhibited decreased growth rates (5.7 vs. 6.5 mm per day) and virulence (1.0–4.3 vs. 13.1 mm lesion lengths) compared with strain LT-3-1D2, but analogous to strain LT-3-1 (Fig. 6b).
To check whether the CcFV-1 dsRNAs could be transfected into another strain (with a genetic background different from LT-3-1) and promote accumulation of the filamentous particles in C. camelliae cells, protoplasts of strain DP-3-1 were also transfected with purified CcFV-1 dsRNAs 1–8. Ten of 94 subisolates gained the tranfected dsRNAs, and one positive subisolate (DP-3-1T28) together with strain DP-3-1, were examined by sucrose gradient centrifugation: the filamentous virus particles were observed in 20–30% sucrose fractions from mycelial extracts of subisolate DP-3-1T28, but not of strain DP-3-1. Parallel experiments in which protoplasts of strains DP-3-1 and LT-3-1D2 were transfected with purified particles, instead of with dsRNAs, failed to provide any positive result.
Comments