Our analysis of related ranaviruses presents a novel approach to genomic comparison that differs from other studies. We analyzed the genomes of closely related isolates of FV3 through 454 GS-FLX technology and STR comparison. The scale (fully sequenced genomes), and the nature of comparison (using viral isolates of FV3), set our investigation apart from past studies. We found that the 3 strains we examined displayed slightly different levels of virulence during in vivo studies. By sequencing the one genome yet to be sequenced, we were able to highlight areas that may be important in generating infectious phenotypes of FV3. This kind of analysis has never been done for FV3 or related ranaviruses: thus, it provides greater insight into the genetic variation among these closely related DNA viruses and the possible genetic basis of ranaviral virulence.
Our results demonstrate that genetic variation is present between closely related FV3 isolates in both coding and non-coding regions. The SSME genome was sequenced and compared to the published wt-FV3 genome, along with related ranaviral genomes RGV, SSTIV, and TFV. Comparisons revealed that SSME was divergent from wt-FV3, and aza-Cr. This variation could be due to the fact that SSME was isolated from a spotted salamander rather than anurans, and so the strain may have evolved in order for it to better adapt to its novel host, as is seen during serial passage . For example, the pathogenicity of Dengue virus was altered by serial passaging Dengue virus 27 times which resulted in 25 nucleotide changes between 2 strains . This finding is interesting given the fact that our results also showed that SSME was the least virulent of our 3 strains during an in vivo study in anurans (Figure 1). Thus, genetic mutations observed in SSME, such as deletions in ORFs 65L, 66L and 49/50L, could have had an effect on strain virulence in anuran hosts. Also, past sequencing results  as well as our own genomic sequencing of wt-FV3 and aza-Cr genomes revealed only 13 nucleotide differences (data not shown) between the strains, supporting our in vivo finding that the two strains cause essentially the same amount of tadpole mortality (97% and 96% respectively).
One of the most significant genetic variations found in SSME was a 757 bp deletion that deleted all of the 65L coding region and most of the 66L coding region (Figure 2). Among other ranaviruses, the functions of these genes have not been determined. Further sequencing of ranaviruses has shown that 65L is present in both RGV and SSTIV as 69L and 68L respectively [23, 28], while this form of 66L appears to be unique to FV3, as other ranaviruses have a 139 nucleotide insertion in their related 66L regions (Figure 2). Another genetic variation specific to SSME was a 13 amino acid deletion in ORF 50L (Figure 3). Interestingly, a second variation was found in this area in the form of single nucleotide deletion. This deletion lead to the merger of ORFs 49L and 50L in all examined ranaviruses other than TFV and wt-FV3 (Figure 3). In terms of function, 49L has multiple SAP motifs, which are DNA/RNA binding domains predicted to be involved in chromosomal organization and DNA replication . Thus, this new 49/50L ORF may function in viral replication.
Single nucleotide deletions were found in multiple ranaviruses within the 43R genes (Figure 4) and 46L (Tables 1 and 2) genes. In 43R, the deletion was present in all ranaviruses analyzed other than wt-FV3, and resulted in a frameshift mutation (Figure 4). In 46L, the original stop codon was lost, leading to the extension of the ORF by 319 bps (Tables 1 and 2). This was found in all genomes other than TFV and wt-FV3. Supposedly, 46L encodes for a neurofilament triplet H1-like protein . However, the extended version of 46L that we discovered has a putative conserved domain known as a microneme/rhoptry antigen in the area previously thought to be non-coding. Micronemes and rhoptries are organelles possessed by Apicomplexa protozoans that secrete proteins involved in parasite entry into a host cell, specifically possessing protein-binding motifs that recognize ligands on the host cell surface . Although usually associated with protozoan parasites, these microneme/rhoptry antigens found in 46L could give further indication as to 46L’s function.
The changes present in 43R, 46L, 49/50L, 65L, and 66L represent the main variable regions amongst the related ranaviral isolates we analyzed. This suggests that these are changeable areas across ranaviruses, and could be used in the future to help explain variable infectious phenotype. Moreover, multiple amino acid deletions present in 65L, 66L, and 49/50L, were limited to the SSME strain, which displayed the lowest level of virulence during tadpole infection. Thus, changes that are unique to SSME may present areas of the genome that are particularly effective in viral attenuation, specifically in an alternative host. As FV3 mortality and morbidity continues to worsen and fluctuate across environmental regions, examination of these genomic areas may prove useful as an initial way to investigate the genetic basis behind infective changes . Further research could be used to explain variations in the virulence of different FV3-like isolates.
As we had already identified variation in the coding regions of closely related FV3 isolates, we decided to further our understanding of variation within highly variable sites by investigating 3 recently identified STR regions . These are known to be variable areas: for instance, although FV3 and SSTIV share 99% genome sequence identity, they share only approximately 50% of repeats in common . Thus, we predicted that these repeat regions would have greater inter-strain variability that would provide useful information when trying to understand overall genetic variability between ranaviral isolates. In order to test this prediction, we analyzed our sequenced SSME genome and reference ranaviral genomes, along with 6 environmental samples isolated from the same waterway. We also sequenced wt-FV3 and aza-Cr to check for repeat number stability across their past viral passages. Analysis revealed that repeat copy number was variable between isolates, even between those from the same geographic location, but that there was some conservation (Figure 5). Specifically, wt-FV3 and aza-Cr were identical, and samples F4, E5, and E3 were also identical at all three regions (Figure 5); this finding was not surprising given that aza-Cr is the result of the wt-FV3 strain treated with azacytidine and does not necessarily represent a strain with a separate evolutionary history . It also implies stability in the repeat regions, as these regions have not changed between the two strains since their initial separation. However, the finding that viral isolates from the same geographical area have variability may limit the use of STRs as a geographic marker. This STR analysis allowed us to better quantify the small scale genetic variation that is present in highly variable genetic sites amongst FV3 isolates, thus furthering our understanding of genetic variation beyond coding regions.
The STR analysis we performed in our study has the potential to contribute to our understanding of FV3 tracking and strain designation. Surveillance and phylogeographical analysis of FV3 are pivotal in understanding how the pathogen varies between different habitat sites and amphibian species, as well as for revealing possible sources of a disease outbreak . It can also have direct effects on conservation by aiding in strategy development to minimize die-offs in high-risk areas, and in creating vaccines through knowledge of the FV3 genome itself [14, 31]. However, exact taxonomic identification of viruses in amphibian populations has been difficult given the lack of detailed molecular data on FV3 and other ranaviruses. The methodology used to classify these viruses in the past has been through comparing the major capsid protein (MCP) of different viral isolates . However, the use of the MCP as a tool to distinguish between different ranaviruses, as well as between different strains of the same virus, has been called under scrutiny [24, 32, 33]. Thus, there is a need to develop new methods of strain tracking for ranaviral isolates.
The use of STRs for ranavirus strain identification has precedence in other virus studies. In one such study, 12 isolates of human cytomegalovirus (HCMV) were isolated from various individuals infected with the virus. The isolates were then tested for variable repeats in 24 polymorphic regions, and based on this analysis, each viral isolate was designated as an individual strain of HCMV . Many of the HCMV repeats used in this study were found in non-coding regions of the genome, similarly to the ones used in our study. The study suggested that these changes in repeats are evolutionarily neutral and so appropriate for strain identification, not only in HCMV, but in other similar, large genome DNA viruses .
Other studies have used coding instead of non-coding repeat regions to identify viral strains [35, 36]. In our study, Region 1 is found in the 19R ORF, unlike the non-coding areas of Regions 2 and 3. There are many examples of functional microsatellites that are known to affect viral characteristics based on copy number, including hepatitis C virus and vesicular stomatitis virus [35, 36]. Therefore, in addition to being potentially useful in viral tracking, STRs from the Region 1 coding region may have functional significance in FV3.