Four functionally distinct groups of genes constitute the PM2 genome. These are genes encoding proteins responsible for genome replication (replication initiation protein P12), transcription regulation (transcription factors P13–P16), structural components of the virion (proteins P1–P10), and proteins involved in cell lysis (P17, P18). The core viral "self" determinants, genes II (protein P2) and IX (protein P9), are separated by the late promoter region and a short gene VII coding for structural protein P7. An open reading frame (ORF) corresponding to gene VII as well as an intergenic region matching the PM2 late promoter (coloured yellow in Fig. 1) were also recognized in all thirteen putative prophages. The region coding for gene II/IX homologs was always found as a block, exactly as in the PM2 genome, except in Photobacterium profundum species SS9 and 3TCK, where the putative prophages contained an additional ORF (Fig 1). Genes coding for structural proteins P5, P8, and P10 were found in 54%, 62%, and 77% of putative corticoviral elements, respectively (Fig. 1). Notably, Vibrio parahaemolyticus RIMD2210633 contained two copies of the PM2-like element, one in each of its two chromosomes, while Methylobacillus flagellatus KT had two absolutely identical (at the nucleotide level) elements in its single chromosome.

As mentioned above, the PM2 genome is organised into three operons – two early and one late 22. The maintenance region of Pseudoalteromonas plasmid pAS28 26 was shown to share similar genomic organisation with the two early transcriptional units of PM2 22. Significant sequence similarity was found only between the leftwards-transcribing operons and the promoter regions (see Fig. 1). Replication protein coding genes of the pAS28 plasmid (ORF 1) and phage PM2 (gene XII) did not share any sequence similarity, although they were located at the equivalent positions (see Fig. 1). In light of those observations, it was proposed that the entire leftwards-transcribing early operon together with the promoter region was obtained from the Pseudoalteromonas plasmid pAS28 via horizontal gene transfer 22. Both genes, XV and XVI, were detected in putative prophages of Vibrio parahaemolyticus, Vibrio vulnificus CMCP6, and Photobacterium profundum SS9, where they were flanking the rest of the viral genes. The remaining corticoviral elements (four out of nine) contained either gene XV or XVI. Interestingly, it seems that in all cases the presumably circular genomes of the corticoviral elements were linearized between the gene XV and XVI. Gene XV was shown to be dispensable in laboratory conditions 27, however, its presence in the genomes of the putative prophages suggests that it might play a beneficial role for the phage in the natural environment.

Surprisingly, gene XII-type replication initiation protein genes were identified only in six of the eleven bacterial species (PM2-type replication system), and ORF 1-type (as in plasmid pAS28) replication protein genes were identified in three of the corticoviral elements (pAS28-type replication system). Putative prophage of Vibrio sp. MED222 encoded an ORF, the product of which shared sequence similarity (23% identity, 59% similarity) with a unique replication protein RepA (accession no. YP_025331) from Pseudomonas alcaligenes plasmid pRA2 (pRA2-type replication system) 28. Furthermore, an identical possible DnaA protein binding site, 5'-TTATCCACA-3', essential for pRA2 plasmid replication, was also found in the putative prophage sequence at the equivalent position, downstream of the repA gene. No putative replication protein-coding genes were identified in the Ralstonia eutropha H16 corticoviral element. However, we identified an ORF coding for a potential DNA binding protein containing the so-called AT-hook motif 29. The AT-hook motif has been shown to be required for the replication and partitioning of Epstein-Barr virus oriP plasmids 30, as well as for the origin recognition complex binding to the replication origin in fission yeast 31. Therefore, it is possible that the AT-hook protein might be involved in the replication of the corticoviral element in Ralstonia eutropha H16, thus comprising the forth replication system (AT-hook-type) found in PM2-like elements. Furthermore, the AT-hook motif-containing protein in Ralstonia eutropha H16 shares sequence similarity with AT-hook motif proteins from Ralstonia pickettii filamentous bacteriophage p12j (accession no. AAQ90254; 35% identity; 54% similarity) and Ralstonia solanacearum plasmid pJTPS1 (accession no. BAA32222; 43% identity, 67% similarity), suggesting possible genetic exchanges between those three Ralstonia elements. Vibrio vulnificus CMCP6 has two chromosomes. The putative corticoviral prophage was found on chromosome 2 only. This prophage has the ORF 1-type replication protein-coding gene. On the other hand, chromosome 1 contains a gene, encoding a putative homolog of the replication initiation protein P12 (NP_761302, E = 1e-65), but no other PM2-related genes. It is possible that a genetic exchange event between the two chromosomes of Vibrio vulnificus CMCP6 gave rise to viruses with the gene XII-type replication initiation protein-coding genes.

We have previously noticed that in the same viral lineage (viruses having a similar mechanism to assemble the virion) replication systems do not need to be related, i.e. the "self" and the replication systems are uncoupled. The final replication product just has to be specifically recognized for genome packaging 12. This observation is supported here by the finding that putative PM2-like prophages encode several types of replication systems. Consequently, genes for genome replication cannot be considered as being part of the viral "self" category, although such genes are among the most conserved ones in viral genomes. This is also true for the tailed dsDNA bacteriophages (order Caudovirales). It has been observed that members of the Podoviridae family use two different strategies to replicate their genomes 32. T7-like phages replicate their genomes by a primase/DNA polymerase mechanism, while φ29-like podoviruses use a covalently linked 5' terminal protein primer 33. Both, T7 and φ29, have the canonical HK97 MCP fold 343536. Interestingly, icosahedral tailless lipid-containing phage PRD1, a type member of the Tectiviridae family, also uses the latter strategy to replicate its linear dsDNA genome 37, but has a completely different MCP topology 1.

In order to escape from the infected cell, bacteriophages have to overcome the cell envelope barrier. All dsDNA bacteriophages characterized so far use the holin-endolysin system to accomplish this step 38. Holins are small phage-encoded membrane proteins that, in a precisely scheduled manner, form nonspecific lesions in the plasma membrane of the infected cell. Endolysins are phage-encoded peptidoglycan-degrading enzymes that may possess endo-β-N-acetylglucosaminidase, N-acetylmuramidase, N-acetylmuramoyl-L-alanine amidase, and/or endopeptidase activities. The endolysins are translocated to the periplasmic space or are activated with the aid of holin proteins. However, no putative endolysins were recognized in the PM2 proteome. Instead, we identified genes (XVII and XVIII) that encode novel lysis-associated proteins P17 and P18 2739. We also discovered that P18 homologues are encoded in the genomes of several enteric (PaP3, Felix 01, and RB49) and marine (P-SSM4, S-PM2, and P-SSM2) tailed bacteriophages 39. This illustrates that lysis genes may be exchanged between bacteriophages belonging to different lineages. Only the putative prophage of Vibrio vulnificus CMCP6 contained a homolog of PM2 gene XVII (32% identity, 82% similarity at the amino acid level), while no homologues of gene XVIII could be identified. Interestingly, putative endolysin genes, similar to those of other dsDNA bacteriophages, were found in corticoviral elements of Vibrio splendidus 12B01 and Vibrionales bacterium SWAT-3 (hits to putative endolysin protein [NP_891673] of phage RB49; E = 4e-31 and E = 1e-30, respectively), Methylobacillus flagellatus KT (lysozyme [NP_690649] of phage B103; E = 4e-27), Methylibium petroleiphilum PM1 (putative lysozyme [NP_958065] of phage PsP3; E = 1e-25), and Ralstonia eutropha H16 (lysin protein [NP_599082] of phage SfV; E = 1e-28) (Fig. 1). Lysis genes have not been shown to participate in virion assembly, e.g. interfere with the "self" function. As shown here, the same family of viruses may have a very different set of genes responsible for progeny release. Consequently, similarly to the replication systems, the lysis-related genes belong to the viral "nonself" category.

In order to establish a stable relationship with the host organism, temperate phages replicate in synchrony with the bacterial chromosome. One mechanism to do so is to integrate the phage genome as part of the host chromosome. This event is catalysed by recombinases (integrase, transposase, invertase) that are often found adjacent to the integration site of the phage genome, thus marking one of the prophage ends 4041. However, not all temperate phages encode recombinases. For example, filamentous vibriophage CTXφ exploits host-encoded recombinases XerC/D to integrate its genome into the bacterial chromosome 42. We identified putative integrases in corticoviral elements of Vibrio splendidus 12B01 (hit to integrase [YP_655517] of phage φMhaA1-PHL101, E = 9e-18), Vibrionales bacterium SWAT-3 (hit to integrase [AAO64735] of phage P2; E = 2e-74), and Oceanobacter sp. RED65 (hit to integrase [YP_654712] of phage F108; E = 3e-64). While putative prophages of Vibrio sp. MED222 and Photobacterium profundum SS9 contained fragments of putative transposase-coding genes (Fig. 1). In addition to an integrase, putative prophages of Vibrio splendidus 12B01 and Vibrionales bacterium SWAT-3 encoded a CI/Cro-type repressor 43, similar to that of a λ-like bacteriophage DMS3 (accession no. YP_950425; E = 1e-13 and E = 7e-12, respectively). We do not currently know whether any of the PM2-like elements is inducible. However, the presence of recombinase-coding genes points to the mobile character of these elements and suggests that at least some of them are functional prophages.

In order to determine in what ecological niches the putative PM2-like prophages could be traced, we checked the origins of the eleven bacteria harboring the corticoviral elements. All of them turned out to be residents of the aquatic ecosystem and were classified into the phylum Proteobacteria in the NCBI Taxonomy database 44.

Marine bacteriophages are the most abundant biological entities in the oceans. Since the oceans are the world's largest biosphere, marine phages are probably the most abundant biological entities on Earth, playing a vital role in the carbon cycling through marine food webs, gene transfer, and conversion of hosts by lysogeny (4546 and references therein). Over 5100 bacterial viruses have been examined in the electron microscope from 1959 to 2000. Of these, about 4950 phages (96.4%) were tailed (order Caudovirales) and only 186 phages (3.6%) were cubic, filamentous, or pleomorphic 47. Tailed phage genomes were also observed to be in great abundance maintained inside the cell, constituting as much as 10 – 20% of some bacterial genomes (reviewed in 41).

In order to compare the relative abundance of PM2-like elements maintained within aquatic bacteria with that of the tailed bacteriophages, 269 genomes of aquatic bacteria available in GenBank (01.05.2007; see Additional file 1) were searched for the presence of putative prophages of tailed viruses. For results to be comparable, the same identification approach as used here for the corticoviral elements was applied. We used the viral "self" determinants of the tailed phages (marine myoviruses [five species], podoviruses [four species], and siphoviruses [two species], as well as from well-studied enterobacterial phages P2 [Myoviridae], λ [Siphoviridae], and T7 [Podoviridae]), i.e. MCP and the larger subunit of the terminase protein, although other virion proteins were suggested to suite better for prophage detection 41. If the bacterial genomes contained genes coding for both selected proteins, we assumed that a prophage is present on the chromosome (Table 1). As the tailed phages are not the object of this manuscript, no further detailed analysis was carried out. However, using this approach the abundance of the putative "tailed" prophages represents the maximal chance to be detected. The highest number of matches, twenty-two, was obtained with MCP and terminase protein sequences of enterophage P2. Among marine tailed phages analysed, the most abundant seems to be bacteriophage VHML-type, with close relatives identified in ten bacterial species. Enteric siphovirus λ and marine podovirus VP16C each had three relatives among 269 aquatic bacterial genomes analysed. Only one of these bacterial species (Vibrio splendidus 12B01) contained both, the "self" determinants of the tailed phage and the putative PM2-like prophage (see Additional file 1). Notably, the two most abundant tailed phages, P2 and VHML, are temperate. Therefore, it seems that their abundance in the genomes of aquatic bacteria reflects the lifestyle they utilize. Interestingly, PM2 and VHML had almost the same number of matches (11 versus 10, respectively) (Table 1). However, it should be noted that PM2 is a strictly lytic phage, which has never been shown to enter the lysogenic pathway. There are no grounds to believe that integration frequency of lytic PM2-like phages should differ from that of the lytic tailed phages. Furthermore, we assume that the quantity of prophage-like elements is related to the quantity of free-living viruses.

PM2 is the only icoshedral tailless marine bacteriophage characterized in detail. However, a number of transmission electron microscopy (TEM) studies have showed that tailless or short-tailed phages are the predominant morphotype in marine samples 4849505152. Since 96% of phages were proposed to be tailed 47, this observation was explained as a result of the dissociation of tails from the capsids during the sample preparation for TEM 46. Our data presented here suggests that the abundance of PM2-like viruses versus tailed phages in the aquatic ecosystem (Table 1) might be vastly underestimated (using a prophage frequency criterion and assuming that mutational rate of the tailed and the PM2-like prophages is the same). This is also supported by the recent sampling of bacteriophages in alkaline hot springs at different geographical locations. Out of one-hundred-fifteen bacteriophage strains isolated during a trial, 50% were icosahedral tailless dsDNA phages, while tailed ones constituted only 44% of isolates 53, suggesting that tailed phages might not be dominant species in all ecological niches. On the other hand, the recent metagenomic study on free-living marine viruses suggested that the most dominant phages are those related to tailed cyanophage P-SSM2 54. However, the abundance of P-SSM2-like phages is questionable as more than 91% of the metagenomic sequences were not significantly similar to any of those in the database 54, therefore leaving space for alternative interpretations when more virus sequence data will be available. Interestingly, the same metagenomic study revealed the previously overlooked high abundance of small icosahedral tailless ssDNA phages (Microviridae virus family) in the samples obtained from the Sargasso Sea 54. It is clear that the relative abundances of any type of viruses cannot be accurately accessed until much more data is available. It is also obviously that phage ecology is only gradually releasing its secrets and further surprises are anticipated.