Pirates of the Caudovirales

Abstract

Molecular piracy is a biological phenomenon in which one replicon (the pirate) uses the structural proteins encoded by another replicon (the helper) to package its own genome and thus allow its propagation and spread. Such piracy is dependent on a complex web of interactions between helper and pirate that occur at several levels, from transcriptional control to macromolecular assembly. The best characterized examples of molecular piracy are from the E. coli P2/P4 system and the S. aureus SaPI pathogenicity island/helper system. In both of these cases, the pirate element is mobilized and packaged into phage-like transducing particles assembled from proteins supplied by a helper phage that belongs to the Caudovirales order of viruses (tailed, dsDNA bacteriophages). In this review we will summarize and compare the processes that are involved in molecular piracy in these two systems.

Introduction

Inspired by stories of piracy in the South China Sea (Rosenberg, 2009), the term “molecular piracy” was coined by Bjørn Lindqvist while he was on sabbatical in the Dokland lab in Singapore, to describe the P2/P4 system of bacteriophages, where the “pirate” phage/plasmid replicon P4 usurps the structural gene products of an unrelated “helper” bacteriophage for its own propagation (Christie and Calendar, 1990, Lindqvist et al., 1993). In fact, the term “molecular piracy” was not entirely new, but has been used previously to refer, variously, to the way by which viruses take control of their hosts’ biosynthetic machinery (Flaitz and Hicks, 1998, Fujimuro et al., 2007), and to the acquisition of host genes during viral evolution (Ahuja and Murphy, 1993, Choi et al., 2001, Sinkovics et al., 1998). Of course, by the first definition, all viruses are pirates, since all viruses require functions supplied by the host cell for their own propagation, and exchange of genetic material is a fundamental mechanism in the evolution of viruses and other organisms. Some viruses, such as HIV, vaccinia or herpesviruses even incorporate host proteins into their virions (Maxwell and Frappier, 2007, Ott, 2008), but such incorporation tends to be incidental or play an auxiliary role, rather than serving as an integral part of the viral structure.

In our definition, molecular piracy refers specifically to the case in which one infectious genetic element (the “pirate”) uses the structural proteins encoded by a viral replicon (the “helper”) for assembly of its own virion. This characteristic distinguishes the pirate/helper systems from the satellite viruses commonly found in eukaryotes (Hu et al., 2009), or the recently described “virophage”, which depends on and interferes with the replication of mimivirus (La Scola et al., 2008). Although these satellites depend upon the helpers for their propagation, they encode their own capsid proteins. Even hepatitis delta virus, which packages its genome containing nucleocapsids within a viral envelope formed by glycoproteins encoded by a Hepatitis B virus helper, encodes its own nucleocapsid protein (Sureau, 2006).

In the P2/P4 system, not only does the pirate depend on the helper for structural proteins, but has the ability to redirect the capsid assembly process to suit its own needs. As it turns out, the P2/P4 system is not the only example of such a phenomenon. More recently, a similar system was discovered in Staphylococcus aureus, where genetic elements called pathogenicity islands (SaPIs) are mobilized by specific helper phages (Lindsay et al., 1998, Novick et al., 2010) and are packaged into phage-like transducing particles using structural proteins supplied by the helper phage (Poliakov et al., 2008, Tallent et al., 2007, Tormo et al., 2008). These two molecular pirates are not degenerate versions of their helpers, but rather independent replicons that have evolved a highly specialized machinery to exploit helper bacteriophages for their own benefit. An additional example of molecular piracy has been described in Sulfolobus, where two nonconjugative plasmids have been shown to exploit archaeal fuselloviruses for packaging and spread. However, little is known about the underlying molecular mechanisms in this system (Arnold et al., 1999, Wang et al., 2007).

The focus of this review will be on the mechanisms used by the P4-related elements and the SaPIs to manipulate their respective helper phages, which are members of the order Caudovirales - tailed, double-stranded DNA (dsDNA) bacteriophages. The biology of the P2/P4 system has been described in great detail in the decades since its discovery (Christie and Calendar, 1990, Deho, G., Ghisotti, D., 2005, Lindqvist et al., 1993). Reports elucidating SaPI biology have a much briefer history, but there have been significant recent advances in our understanding of the interactions between these pathogenicity islands and their helpers in S. aureus (Novick et al., 2010). The molecular piracy that takes place in these systems involves several steps, typically including transcriptional activation, excision and replication of the pirate DNA, and finally assembly and packaging of pirate DNA into virus-like particles made from helper proteins. A variety of interactions between the pirates and their helpers modulate these processes, ranging from gene regulation to morphogenetic control (Table 1). In the following sections, we will discuss each of these interactions separately and also outline where the systems differ.

Table 1.

Comparison of the steps involved in molecular piracy by SaPI1 and P4

	SaPI/80α	P4/P2
Derepression	80α Sri, Dut, gp15 bind and inactivate different SaPI Stl repressors	P2 Cox activates immunity-insensitive transcription from P4 PLL promoter
Reciprocal Derepression	No known mechanism	P4 Epsilon binds and inactivates the P2 master repressor C
Mutual Transactivation	No known mechanism	P2 Ogr and P4 Delta each activate both P2 and P4 late promoters
Excision	SaPI encoded. Derepression by helper required	Independent of helper; P4 encoded. Derepression by helper enhances
Replication	SaPI encoded Derepression by helper required	Independent of helper, P4 encoded Derepression by helper enhances
Capsid size redirection	Internal scaffold; SaPI CpmA and CpmB	External scaffold; P4 Sid P4 Psu provides additional stability
DNA packaging	Headful packaging. SaPI-encoded TerS redirects specificity SaPI Ppi blocks phage DNA packaging	Cos-site packaging. No known P4-encoded functions

Overview of the P2/P4 system

Bacteriophage P2 was originally isolated from the Lisbonne & Carrère strain of Escherichia coli by Bertani in 1951 and is a member of the Myoviridae family of viruses, having an icosahedral head (capsid) and a contractile tail, and a 33.6 kb double-stranded DNA genome (Bertani, 1951, Bertani and Six, 1988, Nilsson and Haggard Ljungquist, 2005). P2 is a so-called “non-inducible” phage; unlike λ and many other prophages P2 is not mobilized by UV light. Several other P2-related phages have also been shown to function as helpers for P4, including PK (the original helper in the strain from which P4 was isolated) (Six, 1963, Six and Klug, 1973), P3 (Lin, 1984) and coliphage 186 (Sauer et al., 1982). P2-like prophages are common in the environment (Breitbart et al., 2002) and are present in about 30% of strains in the E. coli reference collection (Nilsson et al., 2004), in enterohemorrhagic E. coli, and in a variety of other γ-proteobacteria, including strains of Salmonella, Pseudomonas, Serratia, Haemophilus, Vibrio, Yersinia and the Burkholderia cepacia complex (Garcia et al., 2008, Lynch et al., 2010, Nilsson and Haggard Ljungquist, 2005). Most of the characterization of helper exploitation by P4 has been carried out using P2, however, which will be the focus of the discussion here.

P4 is an 11. 6 kb replicon that can exist either as a plasmid or integrated into the host genome like a prophage (Briani et al., 2001, Deho, G., Ghisotti, D., 2005, Lindqvist et al., 1993). P4 is genetically unrelated to P2, and while it has been described as a satellite phage it is probably more appropriate to consider it as an integrative plasmid that has acquired functions for helper phage piracy. P4 lacks genes encoding major structural proteins and requires all of the morphogenetic genes of its helper phage (Six, 1975). A second P4-like element found in E. coli, retronphage φR73, can also exploit P2 as a helper (Inouye et al., 1991). The exploitation of P2 by P4 can take place under a variety of circumstances, including P2 infection of a strain carrying P4 in either the immune-integrated or multicopy plasmid state, P4 infection of a P2 lysogen, and coinfection by both phages. In each of these scenarios, P4 responds to the presence of the helper phage by interacting with certain phage-encoded functions and by activating P4 functions that allow it to manipulate the helper phage appropriately. The nature and timing of the regulatory crosstalk between P4 and its P2 helper depends on the infection conditions, and appears to be designed to optimize P4 reproduction.

Mobilization of S. aureus pathogenicity islands (SaPIs)

SaPIs are a family of 14–27 kb genetic elements that are integrated into the S. aureus host genome and contain phage-like repressor, integrase and terminase genes. Different SaPIs also express a variety of superantigen toxins and other virulence and antibiotic resistance factors. Ten of the 17 identified to date in staphylococcal genomes have been shown to be inducible by either known or endogenous prophages (Novick et al., 2010). The two best characterized SaPIs are SaPI1 and SaPIbov1, found in S. aureus strains RN4282 and RF122, respectively (Novick et al., 2010). SaPI1 carries genes for the toxic shock syndrome toxin 1 (tst) and enterotoxins K (sek) and Q (seq) (Ruzin et al., 2001). SaPIbov1 carries tst as well as genes encoding enterotoxins C (sec) and L (sel), and is associated with bovine pathogenic S. aureus (Fitzgerald et al., 2001). Some SaPIs (SaPI2, SaPI5) are found in clinically important MRSA strains, including USA200 and USA300 (Highlander et al., 2007). Related elements have also been identified in other staphylococcal species (Kuroda et al., 2005, Takeuchi et al., 2005) and in streptococci (Scott et al., 2012).

While normally repressed and stably integrated in the host genome, SaPIs become activated and mobilized when a compatible helper bacteriophage enters the lytic cycle, whether through infection or by induction of an endogenous prophage. Upon mobilization, the SaPI genomes are packaged into transducing particles formed by structural proteins encoded by the helper phage (Tallent et al., 2007, Tormo et al., 2008). Helper phages for SaPI mobilization belong to a large family of temperate transducing phages found in S. aureus that are members of the family Siphoviridae, with ds DNA genomes ranging from 39.6 – 45.9 kb (Kwan et al., 2005). Several helper phages for different SaPIs have been described, including φ11, 53, 80, 80α, and φNM1 (Christie et al., 2010, Dearborn and Dokland, 2012, Lindsay et al., 1998). The phage-induced mobilization of SaPIs is specific: thus, SaPI1 can be mobilized by phage 80α but not by the closely related φ11, nor the more distantly related phage 80; SaPI2 can be mobilized both by phage 80 and by 80α. SaPIbov1 can be mobilized by φ11 and 80α, but not by 80 (Christie et al., 2010). φ13 is able to induce SaPI1 excision and replication, but not to package the genome into phage particles (Lindsay et al., 1998). The helper/SaPI specificity is manifested at several levels, which will be discussed in greater detail in the following sections.

Derepression

SaPIs are normally integrated stably in their host genome in a repressed state, as is the case with P4 in the immune-integrated state. For both of these elements, an essential first step in mobilization is relief of repression by a helper phage-encoded function. Each has evolved a mechanism that exploits helper phage genes which also perform other roles in the phage life cycle. In addition, P4 has the ability to derepress a resident helper prophage (mutual derepression).

P4 immunity and derepression of P4 by P2

P4 immunity involves a unique mechanism in which a short, stable RNA (CI RNA) regulates transcription termination through sequence-specific binding. Leftward transcription of the P4 α operon (Fig 1), which encodes functions required for both plasmid and lytic growth, initiates from two different promoters (Deho et al., 1988). One of these is constitutive and the other is subject to complex regulatory control. In a P4 lysogen, transcription from the constitutive early leftward promoter, P_LE, yields a transcript of ~ 300 nucleotides that terminates upstream of the P4 replication genes (Briani et al., 2000, Deho et al., 1992) and is subsequently processed to generate the CI RNA (Forti et al., 2002). Termination depends upon RNA-RNA interactions between the CI RNA and two specific target sequences in the untranslated leader region of the nascent transcript. (Sabbattini et al.,1995). In order to overcome P4 immunity, this RNA-mediated termination must be circumvented. This is accomplished by initiation of transcription about 400 bp upstream of P_LE from a second promoter, P_LL. Translation of two nested genes in this longer transcript leads to translational suppression of the CI RNA-mediated transcription termination and therefore expression of the P4 replication functions (Forti et al., 1999).

Fig 1.

Derepression of immune-integrated P4 by a P2 helper phage requires activation of transcription from P_LL, which bypasses the P4 immunity system and leads to transcription of the P4 replication genes (Saha et al., 1989). This requires the product of the P2 cox gene (Saha et al., 1989). Activation of P_LL by Cox also leads to the induction of P4 prophage excision that is observed upon P2 infection of a P4 lysogenic strain (Six and Lindqvist, 1978). Excision requires the P4 Vis protein, which is encoded by the first reading frame in the P_LL transcript (Cali et al., 2004). Derepression of P4 following P2 infection appears to be primarily a mechanism for allowing P4 to survive P2- mediated host cell death rather than a mode for efficient horizontal transfer, since the yield of P4 when P2 infects a P4 lysogen is normally quite low, less than 1 P4 per cell (Six and Lindqvist, 1978). However, if P2 replication is blocked (by infection of a bacterial host that lacks the rep gene, or by mutation in a phage replication function) the yield of P4 increases to a level comparable to that seen during P4 infection of a P2 lysogen (Six and Lindqvist, 1978).

P2 immunity and reciprocal derepression of P2 by P4

When the P2 helper is present as a prophage, P4 is able to derepress it to activate expression of required helper functions. P4 infection of P2 lysogens gives rise to about 100 P4 and about 10⁻³ P2 per infected cell (Six and Klug, 1973). As in other temperate phages, P2 early transcription initiates from a pair of divergent promoters encoding competing repressors that regulate the lysogeny functions (Fig 1). The leftward transcript encodes the P2 immunity repressor, C, and the phage integrase, while the rightward transcript includes genes encoding the repressor of the lysogenic promoter (Cox), as well as the replication functions. C regulates its own promoter and blocks expression of Cox, while Cox blocks expression of C (Saha et al., 1987). P2 Cox is a remarkable protein with multiple roles; it functions not only as the repressor of the lysogenic promoter but also as the recombination directionality factor for prophage excision (Yu and Haggard-Ljungquist, 1993) and, as discussed above, positively regulates the P4 P_LL promoter to derepress P4.

The derepression of P2 by P4 requires the P4 ε gene product (Geisselsoder et al., 1981, Liu et al., 1997). Epsilon binds to the P2 immunity repressor and interferes directly with binding of the repressor to its operator (Liu et al., 1998). This leads to expression of the helper early genes and to in situ replication of the P2 prophage, which does not excise (Six and Lindqvist, 1978). The ε gene is essential for P4 growth in a P2 lysogen, but not during a P2 + P4 co-infection of a nonlysogenic cell. However, Epsilon does appear to contribute to interference with growth of the helper phage during a coinfection (Diana et al., 1978). The interaction between Epsilon and the phage repressor determines whether P4 can use a lysogenic helper phage. The P2-related phage 186, which has morphogenetic genes similar to those of P2 but an unrelated repressor (Kalionis et al., 1986), cannot be derepressed by P4 and can only serve as a P4 helper if it is growing lytically (Sauer et al., 1982).

Derepression of SaPIs

In the absence of helper phage, SaPIs are maintained in a stable repressed state by a master repressor, Stl. Like prophage repressors, Stl binds to a region between two divergent promoters where it inhibits most SaPI gene expression. Inactivation of stl by mutation leads to SaPI excision and replication (Ubeda et al., 2008). Thus, derepression by the helper phage is a key regulatory step in SaPI mobilization. Remarkably, the Stl proteins of different SaPIs are widely divergent, and the ability of a particular helper phage to derepress a given SaPI appears to be a primary determinant of helper phage-SaPI specificity (Tormo-Mas et al., 2010). Three different derepression proteins encoded by phage 80α have been identified, and each targets a different SaPI (Fig 2). All of these proteins share a common mechanism; they act as antirepressors by direct binding and inhibition of their respective Stl proteins (Harwich, 2009, Tormo-Mas et al., 2010). SaPI1 is derepressed by Sri, the product of 80αORF22, SaPIbov1 is derepressed by Dut, the product of 80α ORF32, and SaPIbov2 is derepressed by the product of 80α ORF15. Each of these genes is nonessential for phage growth but required for mobilization of the respective SaPI (Tormo-Mas et al., 2010). Two of these phage-encoded antirepressors have other known functions. Sri was previously identified in the related phage 77 as a protein that inhibited host DNA replication by binding to DnaI (Liu et al., 2004), while Dut is a dUTPase (Tormo-Mas et al., 2010). Like P4, SaPIs have apparently evolved to sense the presence of a helper phage by exploiting genes that play another role in the biology of the phage.

Fig 2.

Transactivation

Both P4 and the SaPIs depend upon their helper phages for gene products needed for virion assembly, DNA packaging, and cell lysis. During lytic growth of the helper phages, these functions are expressed late in infection as part of the normal temporal regulation of the phage morphogenetic genes. In the P2/P4 system, at least, there is a second set of reciprocal interactions that regulate late gene transcription, allowing P4 to optimize exploitation of the helper phage under the different conditions it might encounter. This is accomplished by a pair of related transcriptional activators encoded by P2 and P4 that recognize the same promoters on both genomes but differ in the efficiencies with which they activate gene expression.

The P2 morphogenetic genes, encoding the head, tail, packaging and lysis functions, lie in four operons expressed late in infection (Fig 3). P2 late gene transcription requires the product of the phage ogr gene, a transcriptional activator that binds to a site about 55 bp upstream of the initiation sites for the four P2 late promoters (Christie and Calendar, 1985, Christie et al., 2003) and interacts with the C-terminal domain of the α subunit(s) of E. coli RNA polymerase (Ayers et al., 1994, Sunshine and Sauer, 1975, Wood et al., 1997). Ogr belongs to a family of zinc-binding transcription factors found almost exclusively among P2-related phages and their satellites, with a C₂C₂ motif essential for metal binding and activity (Julien et al., 1998, Pountney et al., 1997).

Fig 3.

P4 has two operons that are expressed during lytic growth (Fig 3), and the two P4 late promoters have the same conserved sequence element found upstream of the P2 late promoters. The leftward P_LL promoter is the same promoter that is derepressed by P2 Cox to initiate P4 excision and replication from the prophage state. The second late promoter, P_sid, regulates rightward transcription of three genes involved in helper exploitation: sid, δ and psu. Sid and Psu play roles in P4 capsid assembly (see below). The third gene product, Delta, is an Ogr homologue that activates transcription from the two P4 late promoters and the four P2 late promoters. Likewise, Ogr activates transcription from the two P4 late promoters as well as the four P2 late promoters (Dale et al., 1986, Deho et al., 1988, Halling and Calendar, 1990). Although there is extensive similarity among proteins in the P2 Ogr family, they fall into two functionally discrete classes. Members of the “helper” class, exemplified by Ogr, activate the P4 late promoters better than the P2 late promoters. Members of the “satellite class,” exemplified by the Delta proteins of P4 and φR73, activate the P2 late promoters better than the P4 late promoters and are able to cause transcription in the absence of replicating P2 DNA ((Julien and Calendar, 1996, McAlister et al., 2003). These differences contribute to earlier expression of P4 late genes in the presence of a P2 helper and maximize expression of the P2 late genes in the presence of P4. They also allow P4 to activate directly the transcription of the P2 morphogenetic genes required for packaging and lysis, bypassing their normal requirement for P2 DNA replication.

There is at this point no evidence to suggest a similar set of complex, reciprocal interactions as a general mechanism regulating helper phage exploitation by SaPIs. In contrast to the P2/P4 system, infection of a helper phage lysogen by a SaPI-containing particle has not been reported to lead to a burst of progeny SaPI virions. Helper phage late transcription, studied in most detail for 80α and φ11, appears to initiate from a single late promoter that is activated by the RinA transcription factor, which is encoded by a gene that lies immediately upstream of the late operon. Deletion of rinA eliminates phage production and essentially eliminates SaPI1 transduction by 80α (Ferrer et al., 2011). This argues that SaPI1 does not encode a function that can replace rinA in helper phage late gene transcription. Consistent with this, no increase in 80α late transcription was detected following prophage induction in the presence of SaPI1 (Harwich, 2009). However, there is still significant residual transduction of SaPIbov1 by both 80αΔrinA and φ11ΔrinA (Ferrer et al., 2011), suggesting that unlike SaPI1, SaPIbov1 may encode a function that can activate helper phage late transcription to some extent.

RinA does not appear to have a reciprocal influence on SaPI1 transcription. The SaPI genes involved in capsid size determination and packaging specificity (discussed below) lie in a six-gene operon designated as operon 1, which is preceded by a LexA-regulated promoter (Ubeda et al., 2007). During 80α infection, transcription of these genes in SaPI1 requires derepression of the SaPI and initiates farther upstream, at a promoter that has not yet been identified (Harwich, 2009). Transcription from the Lex-A regulated promoter would lead to a burst of operon 1 expression during SOS induction of a resident helper prophage, which might improve SaPI yield but is not essential for mobilization. A φ11ΔrinA mutant did not show any impairment in transcription of SaPIbov1 operon 1, even under conditions where transcription from the LexA-regulated promoter was blocked by mutation (Ferrer et al., 2011). This indicates that the helper phage RinA transcription factor does not play a direct role in controlling SaPI operon 1 expression.

Assembly and capsid size determination

Tailed, dsDNA bacteriophages of the Caudovirales assemble their capsids (or heads) as empty precursors—procapsids—from the major capsid protein (CP), typically requiring a scaffolding protein (SP) that acts as a chaperone for the assembly process (Fig 4) (Dokland, 1999, Fane and Prevelige, 2003). The main exception to the requirement for SP is the HK97-like phages, in which an N-terminal sequence in CP appears to serve this role (Conway et al., 1995, Duda et al., 1995). During DNA packaging, the capsid undergoes expansion accompanied by major conformational changes in CP (Johnson, 2010). Tail structures (and sometimes “decoration” proteins) are added to the finished capsid. Capsids are either icosahedral or elongated with icosahedral caps, and - in spite of weak or undetectable sequence homology - all members of the Caudovirales studied to date share a characteristic and unique CP fold, called the HK97-like fold (Johnson and Chiu, 2007, Wikoff et al., 2000).

Fig 4.

One of the most striking features about the piracy both in the P2/P4 system and in the mobilization of SaPIs is the redirection of the helper phage assembly pathway to form capsids that are about 1/3 the size (45 nm, T=4) of those normally made by the phage itself (60 nm, T=7), commensurate with the difference in size of the genomes (Figs 4 and 5A) (Dearborn et al., 2011, Dearborn et al., 2012, Dokland et al., 1992, Ruzin et al., 2001, Spilman et al., 2011). The small capsids are unable to package complete phage P2 genomes, thus this redirection of the assembly pathway strongly interferes with P2 multiplication.

Fig 5.

How do the pirate elements carry out this size change? In P4, the size redirection depends on a P4 size determination gene, sid (Barrett et al., 1976), which encodes an external scaffolding protein that forms an external dodecahedral cage around the procapsid (Fig 4A) (Marvik et al., 1995). Sid is an elongated protein made up of bundles of α-helices (Fig 5B) (Dearborn et al., 2012). Trimers of Sid connect hexamers of the gpN capsid protein across the threefold axes, forcing the shell into a T=4 architecture (Fig 6A). P4 sid mutants fail to form small capsids, and while P4 DNA can still get packaged as dimers or trimers into large capsids, the efficiency is low (Shore et al., 1978). Mutants in gpN, called sir (sid responsiveness) render the capsid protein resistant to the Sid-induced size redirection and thus do not form small capsids (Six et al., 1991). These mutations are clustered in an external loop in the gpN CP, where they presumably interfere with gpN–Sid interactions (Fig 5B) (Dearborn et al., 2012). Conversely, mutations in Sid, named super-sid or nms (N mutation sensitive) (Kim et al., 2001), which are clustered in a C-terminal α-helix, recover the ability of Sid to form small capsids even on a P2 Nsir background (Dearborn et al., 2012).

Fig 6.

Expression of gpN and Sid alone is sufficient to efficiently form small procapsids (Dokland et al., 2002), although the gpO SP is incorporated when both proteins are present (Fig 6A) (Wang et al., 2006). However, gpO is required for the formation of viable P4 phage (Six, 1975), presumably due to other functions of gpO, in particular the protease activity that resides in its N-terminal domain, O*, which remains inside the mature capsids (Fig 4A) (Chang et al., 2009, Dokland, 2013). Indeed, the mutant Oam279, which lacks the C-terminal 47 amino acids and is defective in scaffolding activity, retains protease activity and is viable in the presence of Sid (Agarwal et al., 1990).

The P4-encoded psu (polarity suppression) gene product, which acts as a suppressor of rho-dependent transcription termination (Pani et al., 2009, Sauer et al., 1981), also serves a role as a decoration protein that is added to the outside of the completed capsid (Fig 4A) (Dokland et al., 1993). Psu apparently stabilizes the inherently less stable P4 capsids against environmental stress (Isaksen et al., 1993).

Size determination by SaPIs works differently. In the most well described system—SaPI1 mobilized by phage 80α—two SaPI1 proteins, gp6 and gp7, are both required for efficient small capsid formation (fig 4B) (Damle et al., 2012, Poliakov et al., 2008). Homologous proteins are found in most, but not all, SaPIs, and the corresponding capsid morphogenesis genes have been named cpmA (gp7) and cpmB (gp6) (Damle et al., 2012, Dearborn and Dokland, 2012, Ram et al., 2012). These two proteins, CpmA and CpmB, are sufficient to induce small capsid formation when expressed during phage infection or upon co-expression with just CP and SP in a S. aureus co-expression system (Damle et al., 2012, Ram et al., 2012, Spilman et al., 2012).

Unlike the P2/P4 system, there is no Sid-like external scaffolding protein. Instead, SaPI1 procapsids contain internal fingerlike projections absent from the helper phage procapsids (Fig 5A). CpmB, which has a structure similar to that of the SP of bacteriophage φ29, acts as an internal scaffolding protein (Dearborn et al., 2011, Morais et al., 2003). CpmB binds as a dimer to the inside of the SaPI1 shell (Fig 5C) and most likely competes with the cognate 80αSP for the same binding site on the 80α CP (Fig 6B).

The role of CpmA in size determination is less clear. CpmA is only present in procapsids in small amounts, suggesting that its action is transient in nature (Poliakov et al., 2008). Deletion of cpmB in SaPI1 led to the formation of a large number of non-isometric “monsters” (Damle et al., 2012, Dearborn et al., 2011), and while CpmB alone could promote small capsid formation in the absence of SP, CpmA had an inhibitory effect on capsid assembly (Spilman et al., 2012). Small procapsids lack the internal scaffolding core that can be seen in reconstructions of large procapsids (Spilman et al., 2011). The role of CpmA may be to reorganize the scaffolding core that would otherwise prevent small capsid formation, or to bind SP to allow access to binding sites on CP by CpmB (Fig 6B).

Size redirection depends on compatibility between CpmA/CpmB and the helper capsid proteins. SaPI2, for example, forms small capsids when mobilized by phage 80α, but not by phage 80, presumably due to incompatibility with the phage 80 CP, which shares only 16% sequence identity with that of 80α (Christie et al., 2010). Size determination of SaPIbov1 by 80α also appears to be less efficient than for SaPI1 even though the CpmA and CpmB proteins are almost identical (Dearborn and Dokland, 2012). Other factors, including relative protein expression levels, may also play a role in this process.

It should be pointed out that capsid size redirection is not absolutely essential in either of these systems. P4 sid mutants are viable, although reduced in burst size (Diana et al., 1978, Shore et al., 1978). SaPI1 cpmAB mutants are also viable, and appear to be transduced at normal frequency (Damle et al., 2012, Ram et al., 2012). Furthermore, in some phage/SaPI systems, size redirection does not occur. For example, SaPIbov2 (27 kb) and SaPIbov5 do not contain cpmAB homologs, and do not produce small capsids (Novick et al., 2010, Ram et al., 2012). However, the fact that the cpmAB genes are highly conserved when present and always come together (Novick et al., 2010) suggest that they do confer an evolutionary advantage – presumably by interfering with helper phage growth. Both sid and cpmAB mutants have lost the ability to interfere with their helper phages (Damle et al., 2012, Diana et al., 1978, Ram et al., 2012), and SaPIs that lack size redirection have other interference mechanisms, as discussed below.

DNA packaging

In the Caudovirales, DNA is packaged into the procapsids through a ring-shaped portal at one fivefold vertex in an ATP-dependent process that requires a terminase complex, which consists of small (TerS) and large (TerL) subunits (Black, 1989, Feiss and Rao, 2012, Fujisawa and Morita, 1997). The large terminase subunit is responsible for prohead binding, DNA translocation and DNA cleavage, while the small subunit is involved in DNA recognition and binding (Catalano, 2005, Feiss and Rao, 2012, Roy et al., 2012, Teschke, 2012). P4 and SaPIs have evolved different strategies to exploit the DNA packaging machinery of their helper phages. P4 has simply co-opted the P2 packaging machinery by incorporating the same packaging signals into its own genome. P2 and P4 contain identical 55 bp cos site sequences that include the 19 bp cohesive ends found in virion DNA (Ziermann and Calendar, 1990). DNA packaging and cos site-specific cleavage requires the small (gpM) and large (gpP) terminase subunits as well as procapsids (Pruss et al., 1975, Bowden and Modrich, 1985). For both genomes, covalently closed circular DNA molecules are the preferred packaging substrate, unlike the linear concatemers preferred by most bacteriophages (Black, 1989, Bowden and Modrich, 1985, Fujisawa and Morita, 1997, Pruss and Calendar, 1978).

SaPIs, in contrast, redirect the specificity of the DNA packaging machinery of their helpers (Fig 7). Like their helper phages, SaPIs replicate as linear concatemers, and are packaged as headfuls, resulting in virion DNA that is terminally redundant and partially circularly permuted (Ruzin et al., 2001). The phage TerS protein recognizes a pac site on the phage genome that lies within the terS coding sequence (KD Lane, EK Read, GEC; unpublished), as is the case for the pac site of several other phages that use headful packaging, including P22 (Wu et al., 2002) and PY100 (Schwudke et al., 2008). An initial cut is then followed by several rounds of headful packaging.

Fig 7.

In the presence of the SaPI, a SaPI-encoded TerS subunit together with the phage-encoded TerL directs the specific cleavage and packaging of SaPI DNA by binding to a SaPI-specific pac sequence that lies in an intergenic region upstream of the operon that encodes SaPI terS (JC Bento, KD Lane, EK Read, GEC; unpublished). The SaPI-encoded TerS is required for high frequency transduction for both SaPI1 and SaPIbov1, while the phage-encoded TerS is required for packaging of helper phage DNA (Ubeda et al., 2009).

The compatibility of the SaPI-encoded TerS with the helper phage TerL likely accounts for some of the observed SaPI-helper specificity. For example, phage φ13, a cos site phage, can induce SaPI1 excision and replication but fails to produce SaPI1 transducing particles (Ruzin et al., 2001). This is presumably due to an inability to form a functional hybrid between the cos-site based DNA packaging machinery of the phage and the pac site-based TerS subunit of SaPI1.

Some SaPIs also influence DNA packaging at another level, by interfering directly with the packaging of helper phage DNA. This novel mechanism requires the SaPI ppi (phage packaging interference) gene (originally called pif; (Tormo-Mas et al., 2010)), which encodes a protein that binds directly to the phage TerS protein and blocks packaging of phage DNA (Ram et al., 2012). The known Ppi proteins fall into two conserved subsets, each of which appears to target a different phage small terminase superfamily (Ram et al., 2012).

Interference

Both P4 and SaPIs interfere with the multiplication of their helper phages. In the case of P4, capsid size determination appears to be the primary interference mechanism. Interference with P2 by P4 can range from 5- to 10- fold in a simultaneous co-infection to greater than 500-fold if P4 is given a ten minute head start or is present as a multicopy plasmid (Diana et al., 1978 Deho and Ghisotti, 2005). Although there is some evidence that a still unidentified P4 function may augment P4 Sid for full interference with P2 growth, the degree of interference seen when both phages are growing lytically correlates with the percentage of small capsids formed (Nilssen et al., 1996). Furthermore, P2 sir mutants, do not form small capsids, are also resistant to interference and exhibit normal phage growth (Six et al., 1991).

In the case of SaPIs, the situation is considerably more complex. There are at least three strategies for interference, not all of which are used in the interactions between a particular SaPI and a specific helper phage. While small capsid formation certainly prevents packaging of a complete helper phage genome and thereby interferes with phage growth, the loss of the ability to form small capsids by mutation of either SaPI1 cpmA or cpmB alone does not relieve SaPI1 interference with 80α (Damle et al., 2012) The interference retained by cpmA or cpmB mutants does not appear to depend on any SaPI1 functions other than cpmA or cpmB, suggesting a second direct role for these gene products in helper interference. The effect of the size determination genes also differs for different helper phages.

A second level at which interference has been documented is the inhibition of phage DNA packaging by the SaPI-encoded ppi genes (Ram et al., 2012). These genes fall into two different but related families, each of which appears to target different helper phages depending which family the phage small terminase subunit belongs to. For example, the SaPI1 ppi gene does not interfere with the growth of 80α, but does block φ12, while the SaPIbov2 ppi gene strongly interferes with 80α growth (Ram et al., 2012). Different allelic variants of both cpmAB and ppi confer differing levels of interference, which in some cases are additive and in others redundant. An additional SaPI gene involved in interference has also recently been identified. This gene, ORF17 in SaPI2, blocks growth of phage 80 (which is not affected by the SaPI2 ppi or cpmAB genes) but not 80α and has homologs in other SaPIs as well (Ram et al., 2012). The mechanism for this third interference function remains to be elucidated.

Conclusion and perspectives

Horizontal gene transfer (HGT) is now commonly accepted to play a major role in prokaryotic evolution (Koonin and Wolf, 2008, Toussaint and Chandler, 2012). The vehicles that drive this ongoing exchange of genetic material, the so-called mobilome, includes viruses, plasmids, transposons, and a variety of other selfish elements. Bacteriophages play multiple roles as agents of HGT. They mediate the exchange of fragments of chromosomal DNA via generalized and specialized transduction. Temperate phage integration and excision contributes to the remodeling of bacterial chromosomes, and can interrupt genes or bring in new phage-encoded functions via lysogenic conversion. The pirate elements we have described add a new dimension to phage-mediated HGT. They differ from other phage-like elements in that they do not encode their own capsids. They differ from other types of mobile DNA in that they have found a way to directly manipulate bacteriophages, through changes in gene expression and morphogenesis, as vehicles for their own specific high frequency transduction. P4 and the SaPIs both exhibit specialized adaptations to the lifestyles of their helper phages that allow them to exploit these phages to their advantage.

How did these elements arise? One possibility is that the pirates evolved from temperate phages, retaining just those phage-like functions required for integration/excision, replication and helper exploitation. Alternatively, they may have been independent extrachromosomal replicons that have acquired genes conferring the ability to manipulate phage gene expression and utilize phage proteins for their own purpose. The answer may depend on the specific element, since the lifestyle of P4 differs greatly from the SaPIs. P4 can exist and replicate as a plasmid independently of P2, and it has been proposed that P4 evolved from an ancestral plasmid replicon by acquisition of independent modules for site-specific integration and for helper exploitation (Deho and Ghisotti, 2005). The complex web of mutual interactions between P2 and P4 suggests that this is a finally tuned and highly evolved relationship. The SaPI lifestyle more closely resembles that of a prophage; it has a phage-like repressor and integration functions and it does not exist as an independent extrachromosomal replicon. Accordingly, it has been suggested that SaPIs may have evolved from prophages (Novick et al., 2010). However, the absence of genes encoding any virion structural proteins and the acquisition of functions allowing exploitation of helper phages indicates that SaPIs are not merely some kind of defective prophage, but like P4 have co-evolved with their helpers in a highly specific manner.

Despite differences in lifestyle and regulatory circuitry, P4 and the SaPIs share certain common features (Table 1). Both encode integration/excision and replication functions and do not depend on helper functions for these processes. Both have the ability to sense lytic multiplication of their respective helper phages and respond by excising and escaping from the bacterial host. This provides a clear evolutionary advantage, since lytic infection by a phage would mean the death of the host cell and the loss of the pirate element. Remodeling of the helper phage capsid is another conserved feature, and it is striking that these two pirates have evolved quite different mechanisms to accomplish this outcome. While not obligatory for transduction of the pirate genome, capsid size redirection leads to the packaging of sub-genomic fragments of the helper phage DNA and thereby interferes with phage propagation. This is likely of evolutionary benefit to the host, since fewer cells in the surrounding population would be lysed – and would also benefit the pirate, since it would increase the likelihood that bacteria infected by the transducing particles carrying the pirate element would not also be infected by a phage. The importance of interference in the pirate-helper relationship is underscored by the fact that the SaPIs have evolved at least three independent mechanisms for helper phage interference. One remaining unresolved question is what the helper phages get out of the three way relationship between the bacterial host, the helper phage, and the pirate. Why have the helper phages not evolved resistance to this interference by losing the functions required to derepress the pirates or altering the genes targeted by the interference functions? The finely tuned relationship between these pirates and their helpers suggests that these elements are highly co-evolved in a way that must be of mutual benefit.

Molecular piracy, once thought to be a curiosity seen only in the P2/P4 system, has turned out to be considerably more widespread. Many S. aureus genomes contain one or more SaPIs, and the presence of similar elements in other Gram positive genera has led to their general designation as “phage-related chromosomal islands” (Novick et al., 2010). A BLAST search reveals P4-like elements in the genomes of a number of enterobacteria, including members of the genera Escherichia, Shigella and Salmonella. Identification of the helper phages for these related elements, and further study of the interactions between them, is likely to reveal additional mechanisms by which these pirates can exploit their helpers. With the explosion of available genome sequences, we anticipate the discovery of similar elements in other systems which will provide fertile ground for further study.

• Molecular pirates are replicons that hijack helper phage virions for their spread

• Molecular piracy involves highly specific helper phage/pirate interactions

• Pirates can modify the capsid assembly pathway of their helper phages

• Some pirates redirect DNA packaging specificity

• Pirate replicons interfere with the lytic growth of their helpers