The 24,886,212 raw reads obtained by HTS were processed and reassembled de novo into transcripts using the paired-end splicing method in Trinity to obtain 1,224 contigs. BLASTx analysis revealed that a contig consisting of 2,751 nt had the highest amino acid sequence identity (44.41%) with Cannabis sativa mitovirus 1 (Accession no. BK010438/ DAB41756). The presence ParMV1 in P. polyphylla var. yunnanensis symptomatic leaves was confirmed by RT-PCR (Fig. 1C).
The complete genome sequence of ParMV1 (Accession no. MT269666) consist of 2,751 nt with an A + U content of 56.05%. ParMV1 contains a single ORF (nt:358-2,637) which is predicted to encode an RNA-dependent RNA polymerase (RdRp) with a molecular mass of about 86.42 kDa, and 5′ and 3′ untranslated regions (UTRs) of 357 bp and 114 bp respectively (Fig. 1A). The complete genome sequence and the size of the 5’- and 3’-UTR fragments are consistent with previously reported plant mitoviruses. In addition, the BLASTx analysis showed that ParMV1 has high sequence similarity with plant mitoviruses (Score 507–599, Query Cover 76–79% and E-value of 0.0). These results indicate that ParMV1 has a typical mitovirus genome structure.
Six highly conserved motifs (Ι to VΙ) of mitovirus were identified in ParMV1 by multiple sequence alignments of RdRp amino acid sequences of ParMV1 and other mitovirus, based on the criteria of Bruenn et al [15,16,27−30]. These motifs include motif-Ι (307FGKLACAIEGGGKRRIFAIGNYVKQRLLRPYHDWFSTVLGRIPNDGTYNQL357), Motif-ΙΙ (368LLYSFGLKSAIDRWP382), Motif-ΙΙΙ (427FVTGQPLGYHCSWPLFALSHHWVVWMAA454), Motif-ΙV (464FQNYAVLGDDIVLADSSVAAQYQSIL489), Motif-V (492LGVEISYQKSLVS504), Motif-VΙ (510EFAKR514) (Fig. 2A). Two potential stem-loop structures with initial ∆G values of -27.4 kcal/mol and − 34.5 kcal/mol respectively were predicted in the 5' and 3' UTRs of ParMV1 (Fig. 1B). A potential panhandle structure was also predicted with a ∆G value of -13.2 kcal/mol because of the inverted complementarity of the 5' and 3' UTRs. These motifs and terminal sequence features (stem-loop and panhandle structures) typical of mitoviruses confirm that ParMV1 is typical of mitovirus. Comparative sequence alignment analysis showed that the RdRp of ParMV1 shared 23.1–40.6% amino acid and 32.3–45.7% nucleotide sequence identities with the species of Mitovirus (Table S2, S3). Cannabis sativa mitovirus 1 isolate (Accession no. BK010438/ DAB41756) had the highest sequence identities (45.7% aa and 40.6% nt) with ParMV1, inferring that ParMV1 is a plant mitovirus.
Mitoviruses were previously thought to infect only fungi, but the detection of complete genome sequences (2.7 to 3.0 kb) of nearly 20 plant mitoviruses in beet, hemp and petunia in the past years indicates that mtoviruses can also infect plants. A number of mitovirus sequences have been identified in the transcriptomes of a large collection of invertebrates [31], but the origin of these viruses still remains unclear. It has been proposed that the mitoviruses either arose from plant genetic elements or by horizontal transfer from fungal mitochondria to plant mitochondria [32–35]. Subsequent studies have suggested that plant mitoviruses are formed by the integration of fungal mitovirus cDNA fragments in the mitochondrial DNA of vascular plants [36]. Systematic analysis of mitoviruses in 10 plant species revealed that plant mitoviruses did not originate from pathogenic fungi [26].
In the diseased P. polyphylla var. yunnanensis leaves from which ParMV1 was detected and isolated, no fungal infection was detected. These diseased plants only exhibited viral-like symptoms. Besides, if ParMV1 had evolved from endophytic fungi, ParMV1 should have clustered with fungal mitoviruses in the clade containing fungal mitoviruses, but not with plant mitoviruses as seen in the RdRp-inferred phylogenetic tree (Fig. 2B). These results therefore indicate that ParMV1 is a plant mitovirus rather than a fungal mitovirus.
The presence of this novel mitovirus was confirmed by HTS and RT-PCR, combined with complete genome structure analysis, molecular phylogeny and comparative genome sequence analysis. Although studies have shown that fungal mitoviruses can affect the growth and virulence of fungi, leading to morphological abnormalities in mitochondria and growth defects, it is uncertain whether or not mitoviruses have a direct impact on plant health [34,37−38]. As a new plant mitovirus, it is uncertain whether ParMV1 directly caused the observed viral disease symptoms in the host plant. Since plant mitoviruses rely on hosts mitochondrial for replication, it is plausible that these viruses may have potential effects on plant hosts mitochondria that may be detrimental to the plant. Further studies are therefore required to ascertain the effects of mitoviruses on their plant host.