Bacteriophages have played a major role contributing to our knowledge of molecular biology, not only in the role of model viruses but also as tools to investigate mRNA, genes, genetic codes and genomes. The first sequenced genomes were the RNA bacteriophage MS2 [1] and the DNA bacteriophage Ф-X174 [2]. As important as Drosophila was in shaping the field of genetics and Tobacco Mosaic Virus was in advancing the study of virology and biochemistry, the RNA phage MS2 (family Leviviridae) was fundamental in laying the foundation of molecular biology. Thus, it is important to continue adding to the basic understanding of phages. The observations presented in this study were rather serendipitous, in that the focus was not on searching for natural recombinant bacteriophages. Nonetheless, evidence of a recombination event was revealed during a ssRNA bacteriophage sequencing project [3].

Male-specific ssRNA (FRNA) coliphages belong to the family Leviviridae. They are classified into two genera (Levivirus and Allolevivirus) which are subdivided into four genogroups (genogroups I and II in Levivirus and genogroups III and IV in Allolevivirus). Investigating the genetic diversity of FRNA phages Vinjé et al. [4] conducted a phylogenetic analysis of 32 Levivirus field strains using a 189 bp replicase gene fragment. This study revealed three main clusters: genogroup I, genogroup II and a potential novel group, designated JS, which clustered between genogroup I and genogroup II. The putative JS group, represented by phages, WWTP1_50 and 2GI13, had a >40% sequence diversity in the 189 bp replicase gene sequence when compared to strains from genogroups I and II. As these strains were isolated from widely separated geographical regions (Massachusetts and South Carolina) Vinjé et al., [4] proposed that JS may form a stable lineage. This report suggested further genomic sequencing and serological data were needed to confirm whether these strains formed a novel genogroup or whether they were the result of recombination or rearrangement events [5].

In its simplest form, recombination occurs when two disparate DNA or RNA strands exchange or merge stretches of their sequences whereas mutation involves the substitution, deletion or insertion of a nucleotide resulting in the change of the nucleotide sequence of a gene or an amino acid sequence of a protein. In some RNA viruses, RNA recombination events can occur when two or more strains infect the same host. Proposed models for the formation of novel RNA sequences include (i) cleavage and ligation in RNA molecules or RNA secondary structures [6], (ii) replicative template switching whereby the RNA-dependent RNA polymerase (replicase) switches from one template to another RNA template, also known as copy choice [5,7], and (iii) RNA transesterification which occurs when the polymerase adds a separate RNA fragment to the 3' terminus of the original RNA template [5].

Historically, experiments with ssRNA coliphage mutants failed to provide evidence for recombination and the investigators concluded that RNA phages would not undergo recombination [8]. A potential flaw in the conclusion may have been that the study occurred at the time when FRNA phages were thought to possess only three genes, not four. In all likelihood, laboratory-applied selective pressure failed to detect or generate a specific recombinant. This failure may not necessarily reflect the lack of recombination or responsible mechanisms that could occur under actual environmental conditions encountered by ssRNA coliphages.

The first indication of RNA recombination in a male-specific FRNA phage was the report of small, non homologous, recombinant RNA molecules produced from a purified template-free Qβ replicase molecule [9]. The investigators noted similar RNA molecules were present in E. coli cells infected with phage Qβ. Chetverin et al., [7] studied this phenomenon by observing the formation of novel sequences in RNA molecules which suggested that this recombination event occurred as a transesterification reaction catalyzed by a conformation acquired by Qβ replicase during RNA synthesis [5,7]. Nucleotide sequences of recombined RNA molecules non-homologous to the parent RNA were formed in the absence of DNA intermediates, demonstrating an RNA recombination mechanism in the presence of Qβ replicase [5]. Therefore, it was plausible to have recombination in environmental ssRNA male-specific coliphage (Leviviridae) isolates.

In the present study, two JS strains, DL52 and DL54, were isolated during an environmental genotyping study of Leviviridae FRNA phages [10,11]. As in the Vinjé study [4], strains DL52 and DL54 were isolated from separate coastal waters. These phages were placed into the putative JS subgroup using the genotyping methods of Vinjé et al. [4]. The objective of this study was to determine whether the existence of a novel JS-like subgroup representing a third Levivirus cluster as proposed by Vinjé et al., [4] could be verified. The approach taken here was to compare sequences from the JS strains to nucleotide and amino acid sequence data from entire genomes of 10 levivirus genogroup I strains and 5 levivirus genogroup II strains [3]. Analysis of the novel JS strains provided evidence to determine whether these Levivirus strains clustered to genogroup I, II, a combination of groups I and II or a unique genogroup. To further understand the phylogeny of these JS strains, complete genomic sequencing, amino acid composition, phylogenetic, bioinformatic and statistical analyses were performed.

Reported here are whole genome sequence data and bioinformatic analyses supporting the hypothesis that two novel FRNA isolates, DL52 and DL54, were the result of natural recombination. Initially classified as a JS subgroup of genogroup I within the Levivirus genus (family Leviviridae), these strains were isolated from seawater approximately 1130 km apart in the Rachel Carson Reserve, Beaufort, NC, and Narragansett Bay, RI. Findings that JS strains were highly similar to three out of four genes (maturation, capsid and lysis) in genogroup I MS2-like strains, shared the catalytic site location in the RNA-dependent RNA polymerase (replicase) gene, had an identical 3’ signature [3] and then greatly diverged along a stretch of the replicase gene all supported the occurrence of a recombination event.

In this study, two JS strains shared >95% amino acid identity in three (maturation, capsid and lysis) Levivirus genogroup I MS2-like genes but only an 84–85% amino acid identity to the otherwise, highly conserved replicase protein. In comparison, genogroup I strain fr was uniformly different from all other genogroup I strains in all four proteins [3]. Cophenetic correlations and bootstrap analysis strengthen the possibility that JS strains were recombinants as the JS strains were only a subgroup of genogroup I and not a novel genogroup. Throughout the Leviviridae family, subgroups emerge within genogroups, however, subgroup strains differ in all four genes from the parent genogroup [3]. It is therefore plausible to propose natural recombination in these two novel JS-like FRNA coliphages as data presented here suggested a specific genetic rearrangement or recombination event in the replicase gene.

Interestingly, different Leviviridae subgrouped strains originating from across the globe display high amino acid similarity among subgrouped strains [3]. Recombination may explain why Leviviridae strains circulate as discrete subgroups independent of geographical location. Although the two unique JS-like strains were isolated from NC and RI, they shared 96.73% nucleotide similarity across the entire genome. Thus, either a single, natural recombination event occurred as a de novo mutation in each strain or identical, natural recombinations along a hot spot in these genomes formed these strains. In either case discovering that geographically-separated JS-like strains acquired the same recombination event is intriguing.

Largely responsible for the diversity of RNA viruses [12] RNA-RNA recombination was observed in several positive-sense, ssRNA human and animal viral taxa including caliciviruses, coronaviruses, hepatitis, dengue, enteroviruses and astroviruses [13,14,15,16,17,18,19,20,21]. For example, genetic exchange in ssRNA viruses was first demonstrated in polioviruses [22,23].

Recombination events frequently alter the RNA-dependent RNA polymerase region. Human Noroviruses, a positive sense ssRNA virus with a genome length of 7400–8300 nt, are considered to belong to a prototype strain if they share approximately 85% overall nucleotide sequence identity and a high amino acid sequence identity (>95%) in the polymerase gene [20]. A naturally occurring human Norovirus strain shared 95% amino acid sequence identity with the capsid sequences from a Mexico cluster and 95% amino acid identity to the polymerase in a Lordsdale virus cluster. Sequences from the natural strain were obtained from one viral isolate. The combination of sequences in the one strain being complementary to two distinct human Norovirus clusters led to the proposition that this strain was a naturally occurring recombinant [20].

Genetic recombination is known to occur in certain Enteroviruses, a positive ssRNA virus having an approximate 7500 nt genome. Poliovirus recombination occurs in vaccine-derived strains [24] in the human population as a single infected individual excretes a high proportion of recombinants [18]. To determine if other enteroviruses undergo natural recombination, isolates of echoviruses were collected from a meningitis outbreak. Nucleotide sequences were clustered based on a capsid protein (VP1) and RNA-dependent RNA polymerase (3D). Dendrogram relatedness of the echovirus strains grouped the VPI sequences to the prototype strains. However, the RNA polymerase sequences did not cluster to the prototype strains, suggesting genetic recombination among the outbreak strains [18].

Human astroviruses are positive sense, ssRNA with a genome length of approximately 6,800 nucleotides [19] and a polyadenylated 3' tail [21]. Two sets of strains were investigated for recombination; one set was identified from a child care center in Houston, TX, and the two other strains were found in stool samples from two children in Mexico City. The pool of strains shared >97% nucleotide sequence similarity in two out of three genomic regions. The novel strain clustered to one group based on the capsid region. When the RNA-dependent RNA polymerase was analyzed, the novel strain clustered to a separate human astrovirus group. The strains were identified as naturally occurring recombinants on the evidence of high sequence similarity to a few genes of one prototype and similarity to different genes in a second prototype. A total of 64 additional human astroviruses lacked these novel traits [19].

An enteric turkey astrovirus is a non-enveloped, positive sense ssRNA virus with a polyadenylated 3' tailed genome of approximately 7 kb. The most conserved gene in the avian and mammalian astrovirus is the RNA-dependent RNA polymerase or replicase. Genetic analysis of capsid and polymerase sequences from twenty-three turkey astrovirus strains resulted in 8 clusters for the capsid gene and two phylogenetic clusters for the RNA polymerase gene. Computer-generated analyses identified polymerase gene recombination in strains of turkey astrovirus [14].

In this study, four different recombination detection programs along with manual alignment predicted strains DL52 and DL54 as recombinants although the exact amino acid and/or nucleotide breakpoint varied somewhat along the RNA-dependent RNA polymerase gene. As expected the breakpoints did not occur within the catalytic-site domain. The sliding window approach as used with many recombination programs is based on an arbitrarily chosen window length, thus affecting the sensitivity and accuracy when pinpointing the precise breakpoint [25]. Overall, the use of a variety of recombination algorithms provided a stronger, more rigorous scientific case. When comparing the Levivirus strains, manual alignment provided a more accurate picture of where the recombination event occurred along the genome but it did not provide a statistical analysis. Therefore, statistical analysis in combination with manual alignment resulted in a more confident assessment of recombination. Evidence for recombination among positive ssRNA viruses exists within the RNA-dependent RNA polymerase (replicase) gene for numerous viruses as described here and supports the data that natural recombination can occur within the Leviviridae family.

The results of this study provide genetic evidence, bioinformatic and statistical analyses suggesting a natural recombination event in the formation of a genogroup I subgroup JS-like levivirus, represented by two strains, DL52 and DL54. There was high nucleotide and amino acid identity in three genes, the maturation, capsid and lysis genes (≥95%) but a lack of similarity in the replicase gene (84–85%) when JS strains were compared to genogroup I MS2-like strains. Four different recombination programs demonstrated one or two breakpoint regions in the replicase gene, signifying a recombination event. The recombination event occurred downstream of the replicase catalytic site thereby maintaining viral integrity and replication function. Thus, primers for oligonucleotide hybridization probes targeting the replicase beyond the catalytic site would not hybridize to JS strains. In contrast, molecular assays targeting the maturation, capsid or lysis sequences would presumptuously place JS strains as an MS-2 like genogroup.

Phylogenetic tree analysis produced a cophenetic correlation which showed (i) ten genogroup I strains, including strain fr, (ii) the JS subgroup strains, and (iii) the genogroup II strains all formed faithful clusters with correlations of 100, 90 and 98, respectively. The cluster cutoff method, however, revealed only two relevant clusters, (i) genogroup I strains, which included fr and JS, and (ii) genogroup II strains. Therefore, the novel JS strains are not a unique Levivirus genogroup. The proposed classification of JS strains is genogroup I subgroup “JS-like”.

Although both JS strains were prepared for sequencing in the same laboratory, these strains were field-collected by different investigators and shipped to another location where they were plaque-purified and preliminarily classified. Therefore, the possibility that contamination resulted in false recombinants seems unlikely. Likewise, the possibility of cloning and/or PCR errors contributing to the nucleotide and amino acids changes would not have led to both JS strains being almost identical in the non recombinant regions as seven genogroup I strains (DL1, DL2, DL13, DL16, R17, J20 and ST4), three genogroup II strains (DL10, DL20, T72) and two JS strains (DL52 and DL54) were sequenced in this lab in no certain strain or fragment order using the same methods and sequencing company [3]. Finally, to the best of our knowledge, this is the first description of recombinant viruses from natural isolates in ssRNA Leviviridae bacteriophages.