The results presented in this report are in two major subsections. The first contains an analysis of the resemblance between EfCIV583 and a typical SaPI. The second presents an extensive analysis of lactococcal and streptococcal genomes for SaPI-like elements.
E. faecalis EfCIV583
We tested the following key life cycle genes in this element: int/xis (integration/excision), rep/ori (replication initiator, replication origin) and rpr (repressor) for functionality, and noted, incidentally, that the element lacked a recognizable homolog of terS, a gene present in most SaPIs and, in those SaPIs, essential for SaPI-specific transfer (Ubeda et al., 2007b). We also identified the p1 gene that is responsible for EfCIV583 induction.
To confirm the functionality of the int/xis system of EfCIV583, we performed PCR analysis with inward- and outward-directed primers as shown in Figures 2a and b. In both cases, amplicons were obtained, suggesting int functionality. To confirm this, we analyzed int activity ectopically in Escherichia coli. We prepared derivatives of the thermosensitive plasmid pMAK700 (Hamilton et al., 1989) containing the chromosomal attachment site (attC) for EfCIV583 and derivatives of pCN51 (Charpentier et al., 2004), containing the cognate PICI attachment site (attPI) plus int (Figure 2c). Plasmid pairs were tested for cointegrate formation by overnight growth (at 30 °C) followed by plating on doubly selective medium at 43 °C, the restrictive temperature for pMAK700. Colonies were obtained only with plasmids containing the cognate att sites and int gene; no colonies were obtained when the att sites and int genes were from different elements. Cointegrate formation was confirmed by restriction analysis (Figure 2d), and by plasmid sequencing.
The Pri-Rep-ori segment of SaPIbov1, SaPI1 or SaPIn1 can drive replication of a suicide plasmid in S. aureus (Ubeda et al., 2007a, 2008). As the EfCIV583 replicon seemed to have the same overall organization as the SaPIs (see Figure 1 and Supplementary Figure S1A), we tested for functionality of its replicon by cloning the corresponding segment of EfCIV583 (plasmid pJP782) into an erythromycin resistance (Emr) plasmid containing the E. coli ColE1 replicon (incapable of replication in S. aureus or other Gram-positive bacteria). We also constructed plasmids carrying mutations in the rep gene or in the ori site (plasmids pJP1306 and pJP781) of EfCIV583. As controls, we used similar plasmids carrying the replication module from SaPI1 or SaPIbov1 (plasmids pRN9217 and pRN9211, respectively). Plasmids were transferred to S. aureus RN4220, E. faecalis JH2-2 and Bacillus subtilis RL-3 by transformation with selection for the EmR marker of the plasmid. Additionally, as staphylococcal phages can transfer SaPIs and plasmids to Listeria monocytogenes (Chen and Novick, 2009; Chen et al., 2015a), we transduced the functional plasmids from S. aureus to this organism, also with erythromycin selection. Plasmids carrying the pri-rep-ori segment, but not those with mutations in these loci, produced colonies in the recipient strains (Table 1). DNA extraction from the different hosts confirmed the presence of the plasmids containing the replication module (Supplementary Figure S1B). These results confirm that EfCIV583, SaPI1 and SaPIbov1 can replicate in different Gram-positive bacteria, and that this replication depends on the presence of a functional replicon.
Functionality of EfCV583 master repressor
SaPI gene expression is controlled by a master repressor, Stl, analogous to λ-c1 (Ubeda et al., 2008). To test for a comparable repressor in EfCIV583, we cloned into a β-lactamase reporter vector, pCN41 (Charpentier et al., 2004), the region of EfCIV583 corresponding to the SaPI regulatory region, which contains stl, the promoter that it represses and the site of transcriptional divergence (Ubeda et al., 2008; Tormo-Más et al., 2010). Clones were constructed in E. coli either with or without the stl-like gene, which is here designated rpr (repressor; see Figure 3a); the reporter constructs were introduced by transformation in S. aureus RN4220, and β-lactamase activities were measured. The clone containing rpr showed sharply lower β-lactamase activities than that lacking it, consistent with repressor function for Rpr (Figure 3b).
Identification of the phage p1-coded EfCIV583 inducer
Since p1 did not seem to be required for EfCIV583 induction, we wondered whether EfCIV583 was, in fact, fully comparable to a SaPI or, perhaps, was a new type of phage-related element. Accordingly, we subjected the element to the two tests by which the original SaPIs had been defined. First, we SOS-induced strain VE14089 with mitomycin C (MC), removing aliquots of the culture at different times during induction. VE14089 is a plasmid-cured derivative of strain V583 (Supplementary Table S1), which facilitates the interpretation of agarose gels used for the analysis of EfCIV583. The cells were lysed with lysozyme and the lysates analyzed by agarose gel electrophoresis. In the SaPI system, in an experiment of this type, late in the lytic cycle, a new band appears in the electropherogram, migrating more rapidly than the sheared chromosomal (bulk) DNA (Lindsay et al., 1998; Ubeda et al., 2005). This band has the mobility of the SaPI monomeric DNA and represents SaPI DNA released from intracellular mature small particles or filled small proheads during preparation of the DNA sample (Ubeda et al., 2007b, 2008). As shown in Figure 4a, the gel pattern observed with VE14089 following phage induction was indistinguishable from that seen with the classical SaPIs (Lindsay et al., 1998; Ubeda et al., 2005), and the identity of the SaPI-like band was confirmed as EfCIV583 by Southern blotting (Figure 4b).
The second test involved the pelleting of the particles in an MC-induced VE14089 lysate and isolation of the particle DNA followed by agarose gel electrophoresis. With a SaPI-helper phage combination, two bands are seen with such DNA—one of phage monomer size and the other of SaPI monomer size (Ubeda et al., 2007b; Tormo et al., 2008). Again, as shown in Figure 4c, the same gel pattern was seen with EfCIV583 and confirmed by Southern blotting. These results imply that EfCIV583 is packaged in small capsids as are many of the SaPIs (Ruzin et al., 2001; Ubeda et al., 2005); indeed, such small particles have been observed by Matos et al. (2013). The sequence of EfCIV583, however, does not contain any identifiable homologs of the SaPI cpm or ccm genes, which are responsible for small particle formation (Ubeda et al., 2007b; Carpena et al., 2016), and we are presently attempting to identify such genes by mutational analysis.
To test further the implication that EfCIV583 and p1 represent a SaPI-like element and its helper phage, we introduced a selectable marker (tetM) into a putatively non-essential region of EfCIV583 (Supplementary Figure S2), then prepared an MC-induced lysate of JP11028, containing this derivative and phage p1, and tested for transfer of the tetM marker to two different non-lysogenic E. faecalis strains. Transfer of the tetM marker was observed at a frequency of up to ~105/ml (Table 2), three orders of magnitude greater than would be expected for generalized transduction. Incidentally, sequencing of the DNA obtained from phage particles in these experiments revealed that the individual phage or PICI elements are packaged independently, confirming the absence of any fused p1/EfCIV583 genome.
To confirm that p1 is the helper phage for EfCIV583, using pMAD derivative plasmids (Supplementary Table S3), we cured VE14089 of prophages 1–6. We found that the elimination of p1 but not any of the others eliminated EfCIV583 transfer (in accordance with Matos et al., 2013), as well as the appearance of the PICI band following MC induction (Figure 4b and Table 2). Moreover, MC treatment of a non-lysogenic EfCIV583-positive strain did not show either induction or any extra chromosomal species of the EfCIV583 element (Supplementary Figure S3A). Although these results clearly confirm that p1 is required for the induction and packaging of EfCIV583, they raise a further question of the packaging mechanism, since, as noted above, EfCIV583 does not encode a recognizable terS gene. We therefore considered the possibility that, like SaPIbov5 (Quiles-Puchalt et al., 2014), EfCIV583 contains a packaging site that is recognized by the p1 terminase. Since the terminase packaging (pac) site for most phages is embedded in the terS coding sequence, we performed a blast search with the p1 terS coding sequence looking for a matching subsequence within EfCIV583. As shown in Figure 4d, we found such a sequence and suggest that it represents the requisite packaging site for both phage and PICI, and therefore that EfCIV583 uses the p1 terminase for its packaging. This sequence, as occurs with the SaPIs (Quiles-Puchalt et al., 2014), is located in an intergenic region of EfCIV583 (Figure 4d). A BLAST search showed that this sequence was present in the terS gene of two additional E. faecalis prophages and, remarkably, in a pneumococcal PICI, SpnCI670-6B-1.
The existence of a functional Stl-like repressor of EfCIV583 replication, plus the failure of MC to induce the appearance of replicative forms of EfCIV583 in a non-lysogen suggested that, like the SaPI Stl, the EfCIV583 repressor Rpr is SOS-insensitive. We confirmed the SOS insensitivity of the EfCIV583 repressor by means of the β-lactamase reporter fusion, as shown in Figure 3. As the most likely source of the putative derepressor protein was the helper phage p1, we subcloned several segments of the phage, tested them for induction of EfCIV583 with the β-lactamase reporter fusion (Supplementary Figure S3B), and found that a 4.5 kb phage segment could relieve Rpr-mediated repression of the β-lactamase reporter fusion (Supplementary Figures S3C and D). We were then able to demonstrate that a single gene in this segment (EF0309; Supplementary Figure S3E) was responsible, and that a p1 derivative with a mutation in this gene did not induce transfer of the element (Table 2). We then used affinity chromatography with a histidine-tagged derivative of Rpr to test for the formation of a complex between the phage-encoded inducer and the EfCIV583-encoded repressor (Figure 5a), as previously reported for the SaPIs (Tormo-Más et al., 2010). These results suggested that the product of EF0309 acts directly on the repressor, as previously reported for the well-characterized Dut-specific induction of the SaPIs (Tormo-Más et al., 2010, 2013). Interestingly, an orthology analysis of EF0309 revealed that it is highly likely to be the phage’s xis gene, and it is referred to as xis hereafter.
Finally, the role of Xis in EfCIV583 induction was confirmed by the introduction of an in-frame xis deletion in p1, which also eliminated EfCIV583 induction, mobilization and EfCIV583-mediated phage interference (Figures 5b and c and Table 2).
In summary, EfCIV583 appears to closely resemble a typical SaPI, with p1 serving as its helper phage. Our results however, are inconsistent with those of Matos et al. (2013), who reported that, unlike the SaPIs, EfCIV583 can be SOS-induced and that EfCIV583 and p1 are each packaged exclusively in small and large particles, respectively. In both cases, however, our data do not agree with those of Matos et al. (2013). First, we could not demonstrate SOS inducibility and we show, rather, that EfCIV583 is induced by the phage-coded Xis protein (see Figures 3 and 5). Second, since in the Southern blot analyses the p1 and EfCIV583 probes hybridized both with the EfCIV583-sized (small) and the phage-sized (large) DNA bands (Figure 4c), our results suggest that phage DNA can be packaged in the EfCIV583-sized particles, and conversely the island DNA can also be packaged in full-sized phage particles. As with the SaPIs, these results indicate that packaging is not dependent on prohead size (Maiques et al., 2007; Ubeda et al., 2007b). Packaging of a significant proportion of phage DNA in the small particles, which would generate defective phages, could be responsible for the observed phage interference (Matos et al., 2013; Frígols et al., 2015).
Extension to other Gram-positive cocci
A key feature of the present study is that we considered it important to focus on species in which cohesive families of SaPI-like elements could be identified and in which these were the predominant form of phage-related elements in the genus. This would be, first of all, parallel to the S. aureus situation and secondly would reinforce the concept that the biological success of such elements would be underlined by the existence of large—possibly exclusive—intrageneric families. Although the first PICI element was identified in E. faecalis, it subsequently became clear that E. faecalis does not contain a significant family of such elements. Thus, only four others have been identified and all are very closely related (Supplementary Figure S4 and Supplementary Table S4), three of them being in the same site as EfCIV583. Moreover, the orthology analysis of the EfCIV583 ORFs does not reveal membership in any family (see Supplementary Table S5). Nevertheless, EfCIV583 shares not only most of the functional features of a typical SaPI but also the typical genome organization, indicating that that it is clearly a member of the overall PICI family, as described above, despite the atypical pattern of its orthologs.
By contrast, in the lactococci and the streptococci (especially S treptococcus pneumoniae), there does appear to be a series of elements that fits the genomic pattern described for the SaPIs. These elements, which form cohesive families on the basis of ortholog analysis (see below), could readily be separated from other types of inserted elements, occupying specific and unique chromosomal sites. In this report, we characterize these cohesive families of SaPI-like elements, the lactococcal and streptococcal PICIs.
Identification and genomic characterization
For the lactococci, we started with two reports describing six ‘prophage’ genomes in L. lactis strain IL1403 (Bolotin et al., 2001; Chopin et al., 2001). Three of these, bIL309, bIL285 and bIL286, are typical prophage genomes, 35–44 kb in length and three, bIL310, bIL311 and bIL312, are much smaller, 11–15 kb, and lack virion protein genes. Two of the three, bIL310 and bIL312 (but not bIL311), are apparently inducible as their DNA could be detected in lysates after MC induction of the resident prophages present in strain IL1403 (Bolotin et al., 2001). The genomic patterns of these three were highly similar and were also highly similar to the genomic patterns of the SaPIs, as noted above (see Figure 1).
To identify similar elements in other lactococci, we used the KEGG orthology tool (Kanehisa and Goto, 2000). Here, we started with the phage-related element present in the strain CV56 (LlCIV56-1), which is similar to bIL310 in IL1403 (henceforth, LlCI-lL1403(I)), and generated orthology lists for all 22 ORFs in the island (Supplementary Table S6). The orthologs are listed in decreasing order of similarity to the index gene, and the length of the list depends on how well the gene is conserved. Each gene in the list is linked to a KEGG map of the corresponding region of the organism’s genome, and the KEGG map patterns are often highly informative with respect to the local genetic context. Indeed, visual inspection of the KEGG genome map usually enables the identification of such inserted elements. For example, as shown in Supplementary Figure S5A and described above for the SaPIs, a PICI consists of a short set of genes transcribed in one direction and a longer set transcribed in the opposite direction, with the overall size being ~12–16 kb. The int gene is at or near the end of the shorter set, the terS gene near the end of the longer set, within which are one or two large genes corresponding to rep and pri, and the divergence is flanked by regulatory genes corresponding to SaPI stl and str. A typical example is illustrated in Supplementary Figure S5A. Visual scanning of the entire genome often reveals one or more other elements with this pattern or with the prophage pattern (Supplementary Figure S5B); in the staphylococci, streptoocci and lactococci, other types of inserted elements, aside from transposons and insertion sequences, are very rare (although sets of genes with this overall pattern can be found, BLAST searches readily determine whether these are inserted elements such as PICIs or not). PICIs and prophages have small numbers of specific att sites and once these are identified, BLAST searches with the flanking genes reveal the unoccupied sites, as shown in Supplementary Figure S5A.
In the ortholog Tables, generally at least the first 10 orthologs are listed. Occasionally, orthologs are found in the absence of other PICI- or prophage-related genes. These are listed as ‘no insert’. Where there are fewer than 10, either all the matches are listed or prophage genes have begun to appear (and are listed), at which point the list is terminated. The left-hand column in the chart records the locus tag of the gene that has been identified by the KEGG orthology search. In the next three columns are the length of the hypothetical protein (HP), its % similarity with the index protein, and the length of the overlap region between the two that was used for the similarity calculation. For each of the orthologs listed, the corresponding genome region was inspected to determine the type of insert, if any, in which the gene was located. This result is listed in the next column. If it resembles the genomic pattern of a typical PICI or prophage (see below), this is so indicated. ‘NI’ (no insert) means that the flanking regions do not have the pattern of genes typical of PICIs, prophages or other mobile elements, nor do the flanking genes resemble genes of mobile elements. The next two columns indicate the coordinates of the ends of the putative inserted element, and the next, its size. Any relevant comments are in the next column. Several problematical elements, listed as ‘hybrid’, probably represent hybrids between PICIs and prophages. They represent additional examples of possible recombinants and are included for completeness.
By this means, we identified a set of 26 SaPI-like elements in the lactococci, and observed that they occupy seven different att sites (Supplementary Table S7 and Supplementary Figure S6). These elements, like those in IL1403, have a number of features in common with the SaPIs: (i) unique attachment (att) sites that are not also occupied by prophages; (ii) the above-mentioned major point of transcriptional divergence flanked by regulatory genes (also a feature of temperate phages); (iii) absence of bacteriophage morphogenetic and lytic genes; (iv) size around 15 kb; and (v) presence of primase (pri) and initiator (rep) protein genes, plus location and organization of the replication origin. Although the Pri-Rep proteins are often annotated, remarkably, as virulence-related proteins (VapE) or as phage resistance proteins, the genes encoding these proteins are always homologous to the SaPI replication initiation genes (Ubeda et al., 2007a) and are indicated as such in our maps. Of note is the fact that, as with the SaPIs, distinct families of PICI Pri-Rep proteins are encoded in these elements. Thus, in some cases, the pri–rep genes are fused as is the case with some SaPIs. A subset of these newly identified lactococcal PICIs is illustrated in Figure 6a. Their ORFs are color coded to indicate putative functions, and, as can be seen, their organization corresponds closely to that seen in Figure 1 and Supplementary Figure S5A.
A problem with the assembly and characterization of these element families is that the overall recombination frequency in lactoccoci is high, and consequently, genomic rearrangements are common; to pinpoint possible rearrangements, it is especially important to have a family of closely related elements for comparison. An example of genotypic modifications is provided by the elements at the mtlD site in L. lactis (site I). Among the currently available sequenced genomes, four have PICIs at this site, NZ9000, MG1363, IL1403 and CV56. The first two (LlCINZ9000-1 and LlCIMG1363-1) are virtually identical in sequence as are the latter two (bIL312 and LlCICV56-1), and a comparison of the first (LlCINZ9000-1) and third (bIL310) elements reveals several areas of virtual identity separated by three major unshared segments, which are most likely to be insertions. At the extreme 5′ end is a 4 kb unshared region that is also present at the insertion site in strain KW2 and, in a modified form including a transposon, in strain SK11, and is absent in strain KLDS.
As can be seen in the Supplementary Table S6, for most of the 22 ORFs examined, the top 10 orthologs belong to other putative PICIs. Prophages appear toward the end of the list, but mostly for those ORFs that are obviously phage-related (int, pri, rep or terS), and in all cases, the phage ORFs have low similarity with the PICI gene. The other types of ORFs in the list include: (i) accessory genes that were presumably inserted by some unknown type of recombination event and (ii) ORFs encoding HP.
As with most genetic units, the newly identified PICIs always contain many ORFs with unknown functions, whose putative products are annotated as HPs. There are several orthology patterns among the HPs: some HPs are highly conserved and present in many different PICIs, but never elsewhere, whereas others are specific to one or two PICIs only. Prophage orthologs are rarely found, usually very far down when there is a long list. In these cases, the gene may have originated in a prophage or have been acquired by one. We suggest that HPs matching only conspecific PICIs have either been acquired since the divergence of PICIs from their ancestral element or have evolved de novo. Either underscores the long evolutionary independence of these elements. Although the HP ORFs found only in PICIs may be free to diverge unrestrictedly, we have not encountered any that have recently been fragmented by mutation and they nevertheless retain their exclusive membership in the PICI family, supporting the concept that the family is coherent and that its members have evolved independently and in concert.
Packaging of the lactococal elements
We have recently reported that some SaPIs have cos sites. These variant SaPIs, of which SaPIbov5 is the prototype (Viana et al., 2010), are induced by cos phages, which share the same cos site, and are efficiently packaged by these phages, leading to high-frequency intra- and intergeneric transfer (Quiles-Puchalt et al., 2014; Chen et al., 2015a). Since all three phages present in L. lactis IL1403 are cos phages, and since the bIL310 and bIL312 elements present in this strain can be packaged after induction of the resident prophages (Bolotin et al., 2001), we hypothesized that these elements use for packaging the same strategy used by SaPIbov5, namely carrying the cos sequence present in one of the prophages. To test this, we analyzed the sequence contained between the hnh and the terS phage genes. In many phages from Gram-positive bacteria, this region contains the phage cos site (Quiles-Puchalt et al., 2014). Examination of the bIL310 sequence revealed a putative cos site identical to the putative phage bIL286 cos site (Supplementary Figure S7). Related with the other two phages present in the L. lactis IL1403 strain, bIL285 and bIL309, other PICIs also share the putative cos sites present in these phages (Supplementary Figure S7). To test these phage cos sites for function, we cloned the bIL286 and bIL310 putative cos sites to a plasmid, pAGEnt (Supplementary Table S3), which was not transferrable after induction of the resident prophages present in strain IL1403, and found that the cloned cos sites enabled transfer of the plasmids by the bIL286 phages (Supplementary Table S8). This result confirmed the identity of these sequences as cos sites.
Pneumococcal and other streptococcal PICIs
A similar search in eight of the now very large number of pneumococcal genomes revealed essentially the same pattern of SaPI-like elements as in the lactococci. Diagrams of the genomes of 10 of these are presented in Figure 6b and Supplementary Table S7. We have also carried out an orthology analysis for all of the ORFS in one of these, in strain 670-6B, at 0.01M b (ychF site). The results of this analysis, shown in Supplementary Table S9, are similar to those obtained in the analysis of the lactococcal elements (Supplementary Table S6), confirming the idea that the PICI genes belong exclusively to the PICI elements. The difference with the lactococcal elements is that in the pneumococcal PICIs some orthologs are found frequently among other streptococci and orthologs that do not belong to any inserted element are sometimes found in a wide variety of other genera—which may be a result of the high level of transformation competence in pneumococci and related streptococci (Straume et al., 2015). As with the lactococci, with one exception, prophages do not appear or only appear far down the list. The exception is gene SP670_0020, an ORF encoding a 55 amino-acid HP, for which the first three orthologs are three different prophages. These prophages are at different sites in three different host strains, and none of these sites contains a PICI. This ORF presumably represents a very rare episode of gene exchange between PICIs and prophages.
The appearance of other streptococci in this list suggests either that the PICI lineage split from an ancestral element before the differentiation of streptococcal species, or reflects horizontal transfer of these elements. Transfer, however, need not have been mediated by a helper phage, as these streptococci are transformation competent, and pneumococci have a habit of extruding their DNA under certain conditions (Claverys et al., 2007). It is especially notable that a site 3′ to the enolase gene is occupied by closely related PICIs in S treptococcus suis and S treptococcus oligofermentans, as well as in S. pneumoniae (see Figure 6b and Supplementary Table S7).
Examination of KEGG genome maps obtained with orthologs of the rep gene of LlCI1403(I) revealed similar elements in many species. In the pneumococci, only elements of the PICI type were identified in this screen. We thus considered it likely that these elements would belong to a pneumococcus-specific family analogous to those in S. aureus and in the lactococci. We have identified 11 different att sites for the pneumococcal PICIs. As our analysis of the pneumococcal genomes has not been as extensive as that of the lactococcal genomes, there may well be other sites. Again, however, we have not found any intact or defective prophages at any of these sites. Other streptococcal species contain PICIs that are closely related to those in pneumococci (Figure 6b and Supplementary Table S7).
The streptococcal PICIs have been studied in some detail by McShan and co-workers (2012) who have demonstrated that one frequently occupied insertion site in S. pyogenes is between the DNA repair genes, mutS and mutL at ~1.75 Mb in the S. pyogenes genome (Scott et al., 2008). PICI-like elements at this site block transcription of the downstream mutL but excise during exponential growth, restoring transcription and reinsert when stationary phase is reached (Scott et al., 2008). Unlike the typical S. pyogenes PICIs, the PICI-like elements at the mutS/L site appear to have SOS-sensitive repressors and to lack a terS homolog, and therefore, their primary function may be gene regulation rather than transfer. Nguyen and McShan (2014) have suggested that other streptococcal PICIs may also have regulatory roles.
From an evolutionary perspective, as can be seen from the ortholog analysis of SpnCI670-6B (Supplementary Table S9), genes from S treptococcus pyogenes PICIs located between mutS and mutL are in the same list as genes from S treptococcus pyogenes PICIs located elsewhere, suggesting that the streptococcal PICIs belong to a coancestral family that has branched into at least two subfamilies—one involved in regulation, encoding a SOS-inducible repressor, lacking a TerS homolog, and integrating between the DNA repair genes mutS and mutL, and the other in gene transfer, encoding an SOS-insensitive repressor, a TerS homolog, and integrating elsewhere.
In summary, the PICIs within each of these two genera are closely related, suggesting that they are a coherent family that does not contain genetic units of any other type. Very importantly, elements containing additional modules or other complicating segments have not been found in these species—which is why the families are referred to as cohesive.
Many of the PICIs carry identifiable genes or other elements that do not appear to be involved in the PICI lifecycle (accessory genes—see Supplementary Table S7), of which there seem to be at least four types—(i) transposons, IS sequences and other obviously inserted elements, even including, in at least one case, SpnCI-TCH8431-2, a possibly intact plasmid; (ii) genes involved in helper phage interference—well characterized in S. aureus but as yet defined in only one PICI, EfCIV583; (iii) genes that contribute in possibly important ways to the host organism. In the SaPIs, most of these encode superantigens or other virulence genes (Lindsay et al., 1998; Ubeda et al., 2003; Viana et al., 2010), many of which are carried exclusively by the SaPIs. In the lactococci and streptococci, this class includes genes for diverse metabolic activities and resistances to antibiotics, bacteriocins and bacteriophages; superantigen genes have not been observed; and (iv) phage-related genes that are not standard and seem accidental—including occasional capsid, phage protease and regulatory genes. One phage-related regulatory gene, an ltrC homolog, is of interest since ltrC is an essential phage gene that controls late-phage gene transcription and always occurs just 5′ to terS in prophage genomes (Ferrer et al., 2011; Quiles-Puchalt et al., 2013). However, among the PICIs, the ltrC homolog is always 3′ to terS, is sometimes duplicated, and sometimes occurs without any terS homolog. In the SaPIs, terS is adjacent to the phage interference module, one of whose targets is ltrC (Ram et al., 2014). Perhaps, the PICI-carried ltrC homologs, which are <50% similar their prophage counterparts, are involved in phage interference.
In summary, the PICIs, like other non-essential genomic elements, suffer diverse types of rearrangements, including plasmids, insertions of transposons, IS sequences and so on. Such adventitious insertions occur, of course, in any non-essential region, and are not unique to the PICIs.