Discussion

Ranaviruses have been recognized as agents of emerging infectious disease that are now globally distributed (Duffus et al., 2015). Establishing a rapid and effective taxonomic method plays an important role in virus identification and epidemiological investigation. Multiple molecular biology methods, such as RFLP (Restriction endonuclease fragment length polymorphism) profiles and cross-hybridization, are used for virus classification (King et al., 2011). But, with the development of DNA sequencing technology, genomic DNA sequencing and analysis as demarcation criteria are becoming more common. An increasing number of ranavirus genome sequences have been published in the public databases, which can contribute to the investigation of the evolutionary history and taxonomy study. In this study, we employed genomic phylogenic analysis and dot plot comparison to gain a better understanding of classification within ranavirus subspecies. Dot plot analysis can provide insight into genome architecture changes (containing deletions, inversions and duplications) among ranaviruses (Jancovich, Steckler, et al., 2015). These genome architecture changes can be visualized in a picture using JDotter software (Brodie et al., 2004). In previous research, dot plot analysis was widely used in the investigation of the evolutionary history by comparison of genome architecture changes (Chen et al., 2013; Jancovich et al., 2003). In this study, we found that the same group of ranavirus subspecies approximately showed similar linearity pattern (Figure 4) according to dot plot analysis maps of the FV3-AY548484 genome versus other ranavirues genomes. Neither NJ-Tree nor CV-tree can clearly determine taxonomic position of ToRV (Figure 2 and Figure 3), while ToRV should belong to CMTV-like group by Dot plot analysis (Figure 4 and Figure S1). Therefore, dot plot analysis as an important supplementary method can assist phylogenic analysis to confirm taxonomic position of ranaviruses (Table S4).
Presently, the most commonly used classification of newly isolated ranaviruses is phylogenetic analysis based on a single gene (e.g, MCP gene, NF-H1 gene and DNA polymerase gene (Allender et al., 2013; George et al., 2015; Jancovich, Steckler, et al., 2015; Zhou et al., 2013). While phylogenic analysis of a single gene is convenient, it is unlikely to be as robust as whole genome analysis and even causes mistakes. For example, MCP gene is a highly conserved gene and is most widely applied in phylogenic analysis (Jancovich, Steckler, et al., 2015). The Neighbor-Joining phylogenic tree using MCP gene showed that ToRV and CMTV-E was closely related to the FV3-like group, which is not consistent with taxonomic identification based on genome analysis (Figure S5, Cluster10). As a structural protein, MCP gene is too highly conserved sequence to distinguish among various virus strains (Duffus et al., 2013). In addition, Anke C. Stöhr et (Stöhr et al., 2015) constructed a phylogenic tree using concatenated sequences (3223 bp) of MCP, DNApol, RNR-α and RNR-β genes. The tree showed that ToRV was most closely to the FV3-like group, but otherwise shows CMTV-like characteristics in regard to the global arrangement of its genome (Figure S1) (Stöhr et al., 2015). The reason for the difference observed is that the genes used in phylogenic analysis experienced substantial substitution saturation or contained recombinant fragments. For example, the RNR-β gene (Cluster 28) contained recombinant fragments (Table S5). In this study, recombination analysis and substitution saturation analysis were performed to select qualified genes, and the phylogenetic positions of the concatenated 4 selected sequences were also consistent with phylogenetic analysis based on core genes. These results can further improve the accuracy of single-gene or multiple-genes phylogenetic trees in ranavirus taxonomy.
In recent years, there has been a suggestion that GIV/SGIV should be considered as a new genus (Jancovich, Steckler, et al., 2015). We tend to agree with the view based on our study. Firstly, GIV/SGIV share little collinearity with FV3 (Figure 4) and other ranaviruses, which indicated that GIV/SGIV genome architecture were significant different with those of ranaviruses. Secondly, branch length of GIV/SGIV in phylogenetic tree based on ranavirus core genes are 0.27 (Figure 3A). In contrast, branch length among FV3- like, CMTV-like and EHNV-like group are about 0.01~0.03. The branch lengths represent the evolutionary distances. The phylogenetic tree revealed that the evolutionary distances between GIV/SGIV and other ranaviruses (FV3- like, CMTV-like and EHNV-like gourp) are far greater than the evolutionary distances among FV3- like, CMTV-like and EHNV-like group. In addition, in the process of core pan analysis, it is interesting that only 2 genes within GIV and SGIV genomes can be strictly classified into the same cluster when compared without “-nsl” parameter. Meanwhile, 30 ranaviruses (remove GIV and SGIV) share 50 strictly core genes (Figure S6). The default program of PanX is without “-nsl” parameter, which means that long branches gene will be removed from cluster. The branch lengths reflect evolutionary distances among genes within one cluster (Ding et al., 2017). Therefore, the results of core pan analysis indicated that significant portions of strictly core genes within GIV/SGIV were far from the others of 30 ranaviruses in evolutionary distances. In conclusion, genomic characterizations of GIV/SGIV are significantly different from other ranaviruses, and need to be considered as a new genus.