Bacteriophage Protein–Protein Interactions

B. Electron microscopy

Another important method for the investigation of physical relationships between phage components is electron microscopy, especially in combination with genetic analysis. Mutants often have an aberrant morphology. If certain structures are missing in the mutant virion, it can be concluded that the mutated gene encodes this structure or is at least involved in its assembly. For instance, many laboratory strains of λ harbor a frameshift mutation in the stf gene, and these mutants lack side tail fibers (Hendrix and Duda, 1992). Many such studies not only showed which genes encode which proteins, but also indicated how the proteins interact (e.g., when particular structures could be “removed” from the virion by defined gene knockouts).

More recently, many bacteriophage capsids and phage protein complexes have been analyzed by cryoelectron microscopy and three-dimensional reconstructions. This technique gives resolutions approaching that of X-ray crystallography, especially for proteins present in symmetric arrays in the virion, and has proven ideal for analyzing protein–protein interactions in large complexes, especially in combination with high-resolution structural data from other techniques, such as X-ray crystallography.

C. Yeast two-hybrid screens

More recently, the yeast two-hybrid system (Y2H) has become an important method to detect pairwise protein interactions. Interaction screens of phage proteins can be carried out easily by cloning all open reading frames in appropriate vectors and then testing these in a systematic pairwise fashion. This has been done for several phages, including coliphages T7 (Bartel et al., 1996) and λ (Rajagopala et al., 2011), as well as Streptococcus pneumoniae phages Cp-1 and Dp-1 (Häuser et al., 2011; Sabri et al., 2010). The whole process can also be automated.

One disadvantage of the Y2H system is that a typical pairwise screen, also called an “array screen” because of the systematic arrangement of clones in microtiter plates, may only detect 20–30% of all interactions. That is, the rate of false negatives is on the order of 70–80%. This can, in theory, be overcome by using multiple vectors that differ in the structure of the fusion proteins, for instance in the production of N- or C-terminal fusion proteins (Stellberger et al., 2010). A combination of multiple vectors can achieve interaction detection rates of up to 80% (i.e., false negative rates as low as 20%) (Chen et al., 2010).

Y2H also suffers from false positives that can only be understood with additional experimentation. Nonetheless, a multivector Y2H strategy can help in the recognition of false positives. For instance, if multiple vector combinations are used, interactions found in all or most cases are more likely to be “true positives,” whereas interactions detected in a lower fraction of cases are less reliable and may be false positives.

D. Protein complex purification and mass spectrometry

Whole phage virions can be purified, typically by CsCl density gradient centrifugation. Proteins from purified phage can then be denatured and separated on polyacrylamide gels or subjected directly to trypsin digestion and subsequent liquid chromatography (LC) and mass spectrometry (MS). The mass of peptides determined by MS can usually identify the corresponding proteins unambiguously, and thus composition of the phage particles. Once all proteins in a virion are identified, at least some can be assigned easily to known structures, such as the icosahedral capsid or tail sheath. Typically, only a fraction of all phage-encoded proteins are represented in the mature virion (e.g., less than 20 of the ≈70 phage λ proteins). Especially for little-studied phages it can be extremely helpful to know which proteins are in the virion. For example, in an analysis of Streptococcus phage Dp-1, only 8 of the 72 predicted proteins were identified by LC/MS and were thus confirmed structural proteins (see later).

E. Biochemistry of protein complexes

Biochemical approaches often involve the purification of protein complexes and thus yield protein interaction information. This approach can be made technically easier if individual proteins are affinity tagged. For instance, phage-encoded proteins can often be fused to affinity column-binding proteins such as glutathione-S-transferase or oligohistidine without affecting their function. Such tagged proteins and proteins bound to them can then be purified using affinity reagents. The untagged proteins in the complex can then be identified by mass spectrometry or immunoblotting (if antibodies against specific proteins are available). A range of experimental techniques is available for the characterization of phage protein complexes and their activities; a comprehensive overview cannot be given here. For instance, chemical cross-linking of interacting proteins is used frequently to spatially locate physically associated proteins, and variable spacer lengths can determine the steric proximity of the binding partners. Other tests include activity assays (e.g., proteases or other modifying enzymes that may be required for phage assembly), replication, or other processes. Some examples of such activities are described here.

B. Coliphage λ

Phage λ may be the best-studied temperate phage and has been under investigation since the early 1950s. There are probably thousands of papers that have been published on phage λ and its closer relatives N15, HK97, and P22. For instance, there are more than 400 citations in a review of lambdoid phages that do not even focus on the genetic switch of λ, the paradigm for gene regulation (Hendrix and Casjens, 2006). The switch has been described in detail by many other papers and in its own book (Court et al., 2007; Ptashne, 2004). This section summarizes current knowledge of the interactions among phage λ proteins. Finally, a summary of interactions with its host, E. coli, is given. For a more detailed description of phage λ biology and regulation, the authors refer the reader to other reviews (Calendar, 2006; Court et al. 2007; Hendrix and Casjens, 2006; Hendrix et al., 1983; Oppenheim et al., 2005; Ptashne, 2004).

1. The λ virion

The phage λ genome and virion are shown in Figure 3. Note that the virion contains tail fibers that are not visible in most λ micrographs. The variant of λ used in most laboratories around the world, λPaPa, has a frameshift mutation in the structural stf gene and lacks the long “side tail fibers” of λ, Ur-λ, that comes directly from the original E. coli K-12 lysogen. It is still not entirely clear how many different proteins are in the λ particle but 12 are known with confidence and 2 more (gpM and gpL) are possibly in the particle in very low numbers. A current model of the particle with its known proteins is shown in Figure 3. Based on this model, we can predict at least 23 PPIs in the virion (Table III), although several are inferred from genetic experiments and remain to be detected by biochemical or biophysical methods. The virion adsorbs to sensitive cells through an interaction between the tail tip protein gpJ and the host LamB outer membrane protein. DNA is then released through the tail into the cytoplasm.

FIGURE 3

Genome and virion of phage λ. (A) Genome of phage λ. Colored ORFs correspond to colored proteins in (B). Main transcripts are shown as arrows. After Hendrix and Casjens in Calendar (2006). (B) Schematic model of λ virion. Numbers indicate the number of protein copies in the particle. It is unclear whether gpM and gpL proteins are in the final particle or only required for assembly. (C) Electron micrograph of phage λ. Modified after Hendrix and Casjens (2006).

TABLE III

Published protein–protein interactions of phage λ

λ 1	λ 2	Notes	References
gpA	gpNu1	gpA (N-terminal) – gpNu1 (C-terminal)	Catalano, 2000; de Beer et al., 2002; Hang et al., 1999
gpA	gpB	gpA (C-terminal) – gpB (= portal)	Catalano, 2000; Duffy and Feiss, 2002
gpW	gpB	gpW (NMR, head–tail joining)	Casjens, 1974; Casjens et al., 1972
gpW	gpFII	gpW required for gpFII binding	Casjens, 1974; Maxwell et al., 2001
gpB	gpB	12-mer (22 amino acids removed from gpB N-terminal)	Kochan, Carrascosa and Murialdo, 1984; Tsui and Hendrix, 1980
gpC	gpE	Covalent PPI (in virion?)	Hendrix and Casjens, 1974;Hendrix and Duda, 1998
gpC′	gpB		Murialdo, 1979
gpC	gpNu3	gpC may degrade gpNu3 (before DNA packaging)	Hendrix and Casjens, 1975; Hohn et al., 1975; Medina et al., 2010; Murialdo and Becker, 1978
gpD	gpD	gpD forms trimers	Mikawa et al., 1996; Sternberg and Hoess, 1995; Yang et al., 2000
gpE	gpE	Major capsid protein	Casjens and Hendrix, 1974; Dokland and Murialdo, 1993; Lander et al., 2008
gpU	gpU	“Probably a hexamer”	Katsura and Tsugita, 1977
gpD	gpE		Casjens and Hendrix, 1974; Dokland and Murialdo, 1993; Lander et al., 2008
gpU	gpV	gpU is head-proximal tail protein	Pell et al., 2009
gpV	gpV		Buchwald et al., 1970a,b; Casjens and Hendrix, 1974; Katsura, 1983
gpO	gpP		Wickner and Zahn, 1986; Zahn and Blattner, 1985, 1987; Zahn and Landy, 1996)
gpO	gpO	gpO–gpO interactions when bound to ori DNA	Alfano and McMacken, 1989
gpB	gpA	Genetic evidence	Frackman et al., 1985; Sippy and Feiss, 1992
gpFI	gpA	Genetic evidence	Catalano and Tomka, 1995
gpFI	gpE	Genetic evidence	Murialdo and Tzamtzis, 1997
gpH	gpG/gpGT	gpG/gpGT holds gpH in an extended fashion	Hendrix and Casjens, 2006
gpV	gpGT	T domain binds soluble gpV	Hendrix and Casjens, 2006
gpH	gpV	gpV probably assembles around gpH, displacing gpG/gpGT	Xu et al., 2004
CI	CI	Forms octamer that links OR to OL	Bell and Lewis, 2001; Dodd et al., 2001
Exo	Bet		Radding et al., 1971
SieB	Esc	Esc is encoded in frame in SieB and inhbits SieB	Ranade and Poteete, 1993a,b
CII	CII	Homotetramers	Ho and Rosenberg, 1982
Xis	Int		Sam et al., 2002
Xis	Xis	Xis-Xis binding mediates cooperative DNA binding	Sam et al., 2004
Int	Int	Dimer	Warren et al., 2003
Stf	Stf	Likely homotrimer by homology with T4 fibers
gpS	gpS	Large ring in inner membrane	Savva et al., 2008
gpS	gpS′	gpS′ inhibits gpS ring formation	Grundling et al., 2000
gpB	gpE	Copurify in procapsid	Hendrix and Casjens, 1975
gpNu3	pNu3	Monomer–dimer equilibrium	C. Catalano, personal communication
CIII	CIII	Dimer	Halder et al., 2007
Cro	Cro	Dimer; X-ray str	Anderson et al., 1979
gpRz	gpRz1	Heteromultimer supposed to span the periplasm	Summer et al., 2007
gpNu3	gpB	gpNu3 required for gpB incorporation into procapsid	Ray and Murialdo, 1975

2. Assembly of virion head (Fig. 4)

FIGURE 4

Assembly of the phage λ head. Head assembly has been subdivided into five steps, although most steps are not very well understood in mechanistic terms. Note that the tail is assembled independently. GpC protease, scaffolding protein gpNu3, and portal protein gpB form an ill-defined initiator structure. Protein gpE joins this complex in a step requiring the chaperonins GroES and GroEL. Proteins gpNu1, gpA, and gpFI are required for DNA packaging. GpD joins and stabilizes the capsid as a structural protein; gpFII and gpW are connecting the head to the tail that joins once the head is completed. Modified after Georgopoulos et al. (1983) and Lander et al. (2008).

λ encodes 10 proteins required for head assembly, of which 6 end up in the mature virion. Head assembly starts with assembly of the portal, which requires the host GroEL/S chaperone for folding. It is thought that the portal serves as the nucleus for assembling the remaining capsid. However, the mechanistic details of assembly are not well understood. GpC is involved as a head maturation protease that becomes attached covalently to E, the main capsid protein. The resulting fusion proteins subsequently get trimmed to make slightly shortened products named X1 and X2 (Hendrix and Casjens, 1974). The authors note, however, that attempts to identify X1 and X2 were unsuccessful and thus X1 and X2 may be artifacts (Medina et al., 2010). GpNu3 acts as putative scaffolding protein that aids coat shell assembly and is removed from the capsid by proteolysis before DNA is packaged. Structures of the procapsid and mature λ capsid have been solved by cryo-EM, showing the T=7 arrangement of capsid protein gpE clustered into hexamers and pentamers (Dokland and Murialdo; Lander et al., 2008). Trimers of the decoration protein gpD bind to the capsid at trivalent interaction points between the capsomers and stabilize the capsid (Sternberg and Weisberg, 1977). Comparison of the procapsid with the mature capsid reveals the considerable structural transitions that occur upon capsid maturation and DNA packaging, but this process is not well understood mechanistically.

3. Assembly of virion tail (Fig. 5)

FIGURE 5

Assembly of the phage λ tail. The λ tail is made of at least six proteins (gpU, V, J, H, tfa, stf) with another seven required for assembly (gpI, M, L, K, G/T, Z). Assembly starts with protein gpJ, which then, in a poorly characterized fashion, requires proteins gpI, gpL, gpK, and gpG/T to add the tape measure protein H. GpM joins and then gpG and gpG/T leave the complex so that the main tail protein gpV can assemble on the gpJ/H scaffold. Finally, gpU is added to the head-proximal end of the tail. GpZ is required to connect the tail to the preassembled head. Protein gpH is cleaved by the action of gpU and gpZ (Tsui and Hendrix, 1983). It remains unclear if proteins gpM and gpL are part of the final particle (Hendrix and Casjens, 2006). From http://www.pitt.edu/~duda/lambdatail.html.

Tail assembly begins with the antireceptor protein gpJ. Proteins gpI, gpK, gpL, and gpM assemble around gpJ in that order, although no mechanistic or structural details are known about this process. In parallel events, gpH, gpG, and gpG-T (the frameshifted form of gpG including the T reading frame) form a complex that binds the tail tip and the major tail protein (gpV). The tape meassure protein gpH determines the length of the phage tail (Katsura, 1990). Once the correct length is achieved, gpV assembly stops and the proximal tail protein gpU is added. Protein gpZ somehow facilitates head–tail joining without being assembled into the final virion.

5. Anti-termination

Escherichia coli RNA polymerase transcribes λ DNA starting from the immediate early promoters P_L and P_R, which leads to expression of the N and cro genes, among others. As soon as the gpN-binding sites (nut sites) are transcribed, gpN modifies the transcription complex by recognizing a stem-loop structure in the mRNA and by binding to host Nus proteins [note that gpN interacts with nut RNA (not DNA)]. This modified complex is capable of overriding the termination signal that normally stops immediate early transcription. Because gpN is continually being degraded by the Lon protease, ongoing anti-termination requires continual transcription of the N gene (Fig. 6). In addition to the anti-terminator for early lytic genes (gpN), a second anti-terminator, gpQ, is required for late gene expression. GpQ anti-termination differs from that of gpN in that gpQ binds to the “Q binding element” (QBE, also known as the qut site) of the P_R′ promotor (i.e., unlike the gpN protein, gpQ binds DNA). When RNA polymerase (RNAP) initiates transcription at the P_R′ promotor, it almost immediately pauses at a pause-inducing sequence. Pausing allows gpQ to contact σ ^{in the RNAP, which causes a conformational change in the enzyme that enables gpQ to bind the β subunit (RpoB). The gpQ-containing RNAP is then able to override the termination signal at t}_R′ (Deighan et al., 2008; Nickels et al., 2002).

FIGURE 6

Interactions of phage λ with its host, E. coli. (A) Overview of λ activities in E. coli: (a) free phage, (b) binding and entry, (c) DNA circularization and supercoiling, (d) transcription, (e), replication, and (f) virion assembly. Details are shown in B–E. (B) Infection and DNA ejection, (C) DNA ligation, (D) gene regulation. An asterisk means weakly bound to the RNA polymerase. (E) DNA replication. Phage proteins are labeled in red, host proteins in black. Modified after Das (1992) and Mason and Greenblatt (1991).

C. Salmonella phage P22

P22 infects Salmonella enterica serovar Typhimurium and was originally isolated by Zinder and Lederberg (1952). It is a temperate member of Podoviridae. Its virion proteins are not particularly closely related to other tailed phage groups, but its early genes and overall lifestyle are very closely related to phage λ. For this reason, it is often placed in the “lambdoid” phage group. Phage P22 is also a very important tool for those who use Salmonella genetics, as it is particularly easy to handle in the laboratory and is a very good generalized transducing phage (i.e., it packages a significant amount of host DNA by accident and can be used to move host alleles from one Salmonella strain to another).

2. P22 virion assembly

The P22 virion is assembled by proteins encoded by 14 genes in its late operon (Fig. 7) (Botstein et al., 1973; Eppler et al., 1991; King et al., 1973; Poteete and King, 1977; Youderian and Susskind, 1980). Some P22-like phages, for example, phages ε34 and L, have a 15th virion assembly gene, dec, at the promoter proximal end of the cluster, whose encoded protein “decorates” the outside of the head and stabilizes it (Gilcrease et al., 2005; Tang et al., 2006; Villafane et al., 2008). The P22 late operon encodes lysis proteins very similar to those of λ; however, the virion assembly genes are not closely related to λ or to any other tailed phage type. P22 virion assembly is best known for the fact that its head-assembling scaffolding protein was the first to be discovered and studied (Cortines et al., 2011; King and Casjens, 1974; Prevelige et al., 1988; Weigele et al., 2005). The P22 procapsid assembles as a complex of 415 coat protein molecules, 12 portal ring subunits, and about 250 scaffolding protein molecules. At about the time when DNA is packaged, all 250 scaffolding proteins are released intact from the procapsid and are reused with newly synthesized coat protein in subsequent rounds of procapsid assembly. Thus, the P22 scaffolding protein acts catalytically in head assembly and individual molecules can participate in at least five rounds of procapsid assembly. A particularly high-resolution asymmetric cryo-EM reconstruction of the virion has been determined (Lander et al., 2006; Tang et al., 2011), and because assembly of the four proteins of its short tail has also been quite well studied [X-ray structures are known for three tail proteins and the portal protein (Olia et al., 2006, 2007; Steinbacher et al., 1997)], the P22 virion is perhaps the best understood of all tailed phage virions. The details of P22 virion assembly have been reviewed in light of its evolution and genome mosaicism (Casjens and Thuman-Commike, 2011), and the authors refer the reader to this review for further details. The published PPIs of phage P22 are summarized in Table V. Interactions between P22 and its host are much less well studied than those of λ and are mostly understood by homology arguments in the case of proteins that are similar in the two phages. Known P22–host interactions are listed in Table VI.

FIGURE 7

Genome and virion of phage P22. (A) Genome of phage P22 with a scale in kbp below. Green open reading frames (ORFs; rectangles) are transcribed left to right; red ORFs are transcribed right to left. Known RNA polymerase promoters and transcripts are shown as black flags and arrows, respectively. (B) The asymmetric (not icosahedrally averaged) P22 virion three-dimensional cryo-EM reconstruction from Tang et al. (2011). Numbers in parentheses indicate molecules/virion, and bold numbers indicate proteins for which X-ray structures are known. (C). Negatively stained electron micrograph of P22 virions.

TABLE V

P22 intraphage protein interactions

P22 1	P22 2	Description	References
gp3	gp3	Small terminase subunit; homononomer in solution	Nemecek et al., 2007, 2008
gp3	gp2	Purify as complex; bind each other in vitro	Nemecek et al., 2007; Poteete and Botstein, 1979
gp2	gp1	gp2 is large terminase subunit; assembles in packaging motor to gp1 by analogy with other phages (gp2 by itself is monomer in solution); evolutionary argument for P22 gp2–gp1 interaction	Casjens, 2011
gp1	gp1	Portal protein; homododecamer; EM of 12-mer; EM of virion; X-ray structure	Bazinet et al., 1988; Lander et al., 2006; Olia et al., 2011
gp1	gp5	Purify as complex (procapsids); three-dimensional (3D) reconstructions of virion and procapsid	Botstein et al., 1973; Chen et al., 2011; Lander et al., 2006
gp1	gp8	gp8 required for assembly of gp1 into procapsid; interaction seen in procapsid 3D reconstruction	Chen et al., 1989; Greene and King, 1996; Weigele et al., 2005
gp8	gp8	Scaffolding protein; forms homodimers and tetramers in solution	Parker et al., 1997
gp8	gp5	Purify as complex (procapsid); bind in vitro	Fuller and King, 1982; Greene and King, 1994; Parker et al., 1998
gp5	gp5	Coat (major capsid) protein; forms T=7 l icosahedral shell of virion	Botstein et al., 1973; Casjens, 1979; Lander et al., 2006
gp5	gp1	Purify as complex (procapsids); interaction seen in 3D reconstruction	Chen et al., 2011; Lander et al., 2006
gp4	gp4	Tail protein; monomer in solution (naturally unfolded?); gp4s barely touching each other in dodecamer ring in virion (i.e., contacts not large); EM, X-ray structure	Olia et al., 2011; Tang et al., 2011
gp4	gp5	Contact in high-resolution 3D reconstruction of virion	Olia et al., 2006, 2011; Tang et al., 2011
gp4	gp1	Order of assembly, EM, 3D reconstruction; X-ray cocrystal structure	Olia et al., 2006, 2011; Tang et al., 2011
gp4	gp10	12 gp4s contact 6 gp10s; order of assembly, EM, 3D reconstruction	Lander et al., 2006; Olia et al., 2007
gp4	gp9	gp4 ring contacts 6 gp9 trimers; 3D reconstruction	Lander et al., 2006
gp10	gp10	Tail protein; probably hexamer in solution; 3D reconstruction	Lander et al., 2006; Olia et al., 2007
gp10	gp9	6 gp4s contact 6 gp9 trimers; 3D reconstruction	Lander et al., 2006
gp10	gp26	6 gp10s contact one gp26 trimer	Landeret al., 2006
gp26	gp26	Tail needle protein; trimer in solution - crystal structure; one trimer in virion	Olia et al., 2007
gp16	gp8	Mutants of gp8 do not incorporate gp16 into virion	Chen et al., 1989; Greene and King, 1996; Weigele et al., 2005
C1	C1	C1 is homologue of λ CII; homotetramer	Ho et al., 1992
Mnt	Mnt	Repressor of antirepressor gene; tetramer	Nooren et al., 1999
Arc	Arc	Repressor of antirepressor gene; dimer in solution, tetramer on DNA	Bowie and Sauer, 1989; Breg et al., 1990; Brown, Bowie and Sauer, 1990
Ant	C2	Antirepressor; binds to and inhibits C2 repressor	Susskind and Botstein, 1975
gp9	gp9	Tailspike protein; trimer in solution, 6 trimers in virion; crystal structure; binds to polysaccharide on cell surface	Lander et al., 2006; Steinbacher et al., 1994, 1997
Int	Xis	Form heteromultimer	Cho et al., 2002
Xis	Xis	Excisionase; homomultimers	Mattis et al., 2008
Erf	Erf	Recombination protein; homomultimeric ring of 10–14 subunits	Poteete et al., 1983
C3	C3	C3 is homologue of λ CIII; λ CIII is dimer by disulfide bond; P22 interaction by homology with λ CIII	Halder et al., 2007
SieB	Esc	Similar to λ but not studied in detail	Ranade and Poteete, 1993a
C2	C2	Prophage repressor homologue of λ CI; homodimer	De Anda et al., 1983
Cro	Cro	Dimer; solution studies and crystal structure	Darling et al., 2000; Newlove et al., 2004
gp18	gp12	Origin binding protein gp18 recruits DnaB-like gp12 to replication origin by analogy with λ (note that gp12 is not a homologue of λ P protein; gp18 is homologous to λ O protein only in N-terminal domain)	Backhaus and Petri, 1984
Rz	Rz1	Rz and Rz1 are supposed to bind to one another and form a bridge that spans the periplasm in phage λ; P22 gp15 and Rz1 should do the same by homology.	Summer et al., 2007
gp13	gp13	Holin forms rings of large diameter in λ; by homology same should be true of P22 holin	Savva et al., 2008
gp13	gp13′	Antiholin–holin interactions delay the formation of holin rings in λ; by homology same should be true in P22	Rennell and Poteete, 1985
Dec	Dec	Trimer; Note that P22 has no Dec, but its very close relative phage L does and L Dec binds to P22 capsids.	Tang et al., 2006
Dec	gp5	Dec occupies a subset of the quasi-equivalent local threefold positions on the shell	Tang et al., 2006

TABLE VI

P22–host protein interactions

P22	Host	Description	References
gp5	GroEL	Solution studies	de Beus et al., 2000
gp24	NusA? etc?	Homologue/analogue of λ gpN; likely binds same host factors as λ gpN	Franklin, 1985; Hendrix and Casjens, 2006
C3	FtsH(HflB)	C3 is homologue of λ CIII; inhibition of FstH by homology with λ CIII	Semerjian et al., 1989
C1	FtsH(HflB)	C1 is homologue of λ CII; CII is degraded by FtsH	Backhaus and Petri, 1984
C2	RecA	P22 repressor autocleavage stimulated by RecA binding like λ	Prell and Harvey, 1983; Tokuno and Gough, 1976
Xis, gp18, gp24, C1	Host protein degradation machinery	These may be degraded analogously to their λ homologues, but direct information is not available	Backhaus and Petri, 1984
gp23	RNA polymerase?	By homology to λ; gp23 is nearly identical to λ gpQ	Pedulla et al., 2003
gp12	Host replication machinery	Possible interaction with host DnaG primase	Wickner, 1984
Abc2	RecC	Solution studies	Murphy, 2000

D. Enterobacteriophage P2 and its satellite phage, P4

Bacteriophage P2 was originally isolated from the Lisbonne & Carrère strain of E. coli by Bertani in 1951 and is a member of the Myoviridae family of viruses. It has an icosahedral head and a contractile tail and is a 33.6-kb double-stranded DNA genome (Bertani, 1951; Bertani and Six, 1988; Nilsson and Haggård-Ljungquist, 2006). 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).

Bacteriophage P2 has attracted much interest as a “helper” for bacteriophage P4, an 11.6-kb replicon that can exist either as a plasmid or integrated into the host genome as a prophage (Briani et al., 2001; Deho and Ghisotti, 2006; Lindqvist et al., 1993). P4 lacks genes encoding major structural proteins, but can utilize the structural gene products encoded by P2 for assembly of its own capsid (Christie and Calendar, 1990; Lindqvist et al., 1993; Six, 1975). An overview of the P2 structural biology and a summary of its interactions are shown in Figure 8 and Table IX.

FIGURE 8

(A) Schematic diagram of the 33.6-kb P2 genome. Open reading frames (ORFs) are indicated by arrows that reflect their direction and size and are color coded as follow: red/orange, capsid-associated proteins (gpQ, O, N, L); blue, terminase proteins (gpP, M); yellow, lysis proteins (gpY, K, lysA, lysB); green, tail-related proteins (gpR, S, FI, FII, T, U, D); blue, base plate and tail fiber-related proteins (gpV, W, J, H, G); pink, transcriptional control proteins (Ogr, C, Cox); brown, integrase (Int); and purple, replication-related proteins (gpA,B). Nonessential genes and ORFs of unknown function are white. Promoters are indicated by arrows above ORFs. (B) Diagram of protein–protein interactions listed in Table IX. Each P2 and P4 protein is shown as a circle, colored as given earlier. Host proteins are shown as white boxes. (C) Schematic diagram of the P2 virion, color coded as given earlier. Copy numbers of proteins, when known, are indicated in parentheses. (D) Electron micrograph of a P2 virion, stained negatively with uranyl acetate.

3. Head structure and assembly

The main structural proteins involved in P2 capsid assembly are gpN, capsid protein; gpO, internal scaffolding protein; and gpQ, connector or portal protein (Chang et al., 2008; Lengyel et al., 1973; Linderoth et al., 1991). P2 capsids are assembled as empty precursor procapsids consisting of 415 copies of the gpN capsid protein arranged with T=7 icosahedral symmetry and clustered in hexamers and pentamers that make trivalent interactions (Dokland et al., 1992) (Fig. 9). GpQ forms a dodecameric ring that is incorporated into the capsid at one fivefold vertex and through which the DNA is subsequently packaged (Doan and Dokland, 2007; Rishovd et al., 1994).

FIGURE 9

P2 assembly. Schematic diagram of the P2/P4 capsid assembly pathway. (A) Capsid protein gpN, scaffolding protein gpO, and connector (portal) protein gpQ are assembled into the T=7 P2 procapsid. (B) GpN, gpO, and gpQ are then processed to their mature forms, N*, O*, and Q*. DNA is packaged into the procapsid by the P2 terminase complex (gpM and gpP), accompanied by expansion into the mature, angular capsid (B). (C) In the presence of the P4-encoded external scaffolding protein Sid, P2 structural proteins are assembled into a small, T=4 P4 procapsid. Loss of Sid (D), protein processing, and DNA packaging lead to the mature, expanded P4 capsid (E). The decoration protein Psu is added to the mature capsid (F).

GpO plays a dual role in the assembly process as both a protease and a scaffolding protein (Chang et al., 2009; Dokland, 2011). The full-length, 284 residue gpO protein is associated with procapsids at early times (Marvik et al., 1994), but is cleaved autoproteolytically between residues 141 and 142, leaving an N-terminal fragment, gpO* (formerly known as h7), that remains inside the mature capsid (Chang et al., 2008; Lengyel et al., 1973). Concomitantly, gpN and gpQ are cleaved—presumably by gpO—to their mature forms, gpN* and gpQ*, by removal of 31 and 26 residues, respectively, from the N terminus (Rishovd and Lindqvist, 1992; Rishovd et al., 1994). The protease activity of gpO resides in its N-terminal domain, which binds tightly to gpN and constitutes a serine protease similar to that of the herpesviruses (Chang et al., 2009; Cheng et al., 2004; Dokland, 2011). The C-terminal 90 amino acids comprise a scaffolding domain, which is necessary and sufficient for promoting capsid assembly (Chang et al., 2008, 2009; Wang et al., 2006). It too presumably binds to gpN during assembly, but any interaction is transient (Chang et al., 2009). Both N- and C-terminal domains form dimers in solution (Chang et al., 2009).

Circular P2 genomic DNA substrates are packaged into empty procapsids by the terminase complex, which consists of the small and large terminase proteins (gpM and gpP, respectively). Terminase recognizes and cleaves P2 DNA at the cos sites (Bowden and Modrich, 1985; Ziermann and Calendar, 1990). DNA packaging is associated with major structural transitions that lead to a larger, more angular and thin-shelled, mature capsid (Chang et al., 2008; Dokland et al., 1992). A head completion protein, gpL, is added (Chang et al., 2008), and tails, which are assembled via a separate pathway, are attached to the unique gpQ-containing portal vertex (Lengyel et al., 1974).

E. Pneumococcus phage Dp-1 and Cp-1

The lytic bacteriophages Dp-1 and Cp-1 infect the Gram-positive bacterium Streptococcus pneumoniae, which colonizes the mucosal surface of the upper respiratory tract and causes serious invasive infections such as pneumonia, meningitis, and sepsis (van der Poll and Opal, 2009). Dp-1 was the first isolated Streptococcus phage and is an unassigned member of the Siphoviridae (Fig. 10) (McDonnell, Lain and Tomasz, 1975). It was reported that the Dp-1 virion surprisingly contains lipids that might be arranged as a bilayer surrounding the capsid (Lopez et al., 1977), a feature not reported for any other phage belonging to the Caudovirales. The lytic enzyme Pal of Dp-1 has been studied intensively because it kills virulent streptococci efficiently and has been used as an enzybiotic to cure pneumococcal infections in animals (Loeffler et al., 2001; Rodriguez-Cerrato et al., 2007). The podovirus Cp-1 lytic enzyme Cpl1 was also demonstrated to be an efficient enzybiotic (Loeffler et al., 2003). The Cp-1 genome is a linear, 20-kbp dsDNA predicted to encode 28 proteins and a packaging RNA (pRNA). Like ϕ29-like viruses, Cp-1 uses protein priming for the initiation of DNA replication (Martin et al., 1996). The biology of Dp-1 and Cp-1 is otherwise only poorly understood.

FIGURE 10

The Dp-1 genome and interactome. (A) Dp-1 genome map. Open reading frames (ORFs) corresponding to their transcription direction are shown as arrows (genome map). Predicted transcripts and functional gene clusters are indicated. All protein–protein interactions (PPIs) identified by systematic Y2H screens among Dp-1 proteins are shown as lines that connect the corresponding ORF symbols. Proteins that bind themselves (homomers) are highlighted by gray ORF symbols. Other ORF colors: refer to Figure C-E. (B–E) A Dp-1 virion model based on identified binary PPIs, mass spectrometry analysis (MS) of mature Dp-1 virions, and homology predictions. (B) Dp-1 virion proteins, including structural proteins identified by MS analysis (in red), or proteins with a homology-based annotation that indicates a virion-related function (in black). (C) Core model of the Dp-1 virion: proteins highlighted in red were identified by MS analysis as structural components. Proteins highlighted in blue represent hypothetical gene products that interact with structural components, thus presumably playing a transient role in virion assembly as they were not identified by MS analysis of Dp-1 virions. Red edges that connect the proteins were shown to interact in Y2H tests. (D) Y2H screens do not identify all expected PPIs. Red lines represent PPIs that were expected to occur among the corresponding proteins known from homologues but were not identified in the systematic Y2H screens. (E) Systematic Y2H screens reveal unexpected PPIs. Proteins are included that were shown to bind with structural proteins and putative morphogenetic factors, e.g., DNA metabolism (DNA polymerase subunits, RecB, DNA ligase, DnaG), queuosine biosynthesis (Que proteins), lysis (Holin), and transcription (sigma factor). Interactions are symbolized by edges that connect the corresponding proteins (in the case of sigma factor gp69, interaction partners are highlighted by an asterisk). (F) Electron micrograph of a mature Dp-1 virion. C–F from Sabri et al. (2010), with permission.

The 56.5-kbp genome of Dp-1 is predicted to encode 72 proteins (Fig. 10A) (Sabri et al., 2011). Approximately half of the gene products could be functionally annotated by homology predictions. However, in this and hundreds of other completely sequenced phage genomes, more than half of phage gene products could not be annotated. Proteome-wide Y2H screens, coupled with mass spectrometry analysis (MS) of mature Dp-1 virions, identified 156 unique PPIs among 57 Dp-1 proteins [a detailed list can be found in Sabri et al. (2011)], and MS revealed that the mature virion is composed of eight structural proteins. A Dp-1 virion model (Figs. 10B and 10C) physically links many proteins with unknown function to the virion. Because they bind structural proteins but are absent from the mature virion, these Y2H PPIs either indicate novel structural components (when found by MS of the virion particles) or virion assembly facilitators. For instance, homology comparisons of gp40, gp41, and gp53 predict that these proteins are structural (Fig. 10B). However, MS analysis did not detect them as components of the virion. Even though functional predictions by homology comparisons are extremely helpful in proteome annotation, a certain fraction of proteins may be wrongly annotated in the absence of other experimental data (i.e., whether a protein is annotated as a structural component or an assembly facilitator can have an important impact on future experiments).

Such a comprehensive analysis also reveals unexpected interactions that normally would not be tested in hypothesis-driven experiments (Fig. 10E). For example, Dp-1 proteins believed to be involved in DNA replication and recombination, transcriptional regulation, and other pathways exhibit interactions with structural components. This indicates that DNA packaging and DNA replication/repair might be coupled (as in phage T4), for example, by resolving intermediate DNA structures in parallel with packaging. Although the identified PPIs were helpful in obtaining the first hints about the function of the proteins, not all expected PPIs between the structural proteins were detected (Fig. 10D). Nevertheless, this study demonstrated that homology-based genome annotation complemented by proteomic approaches is a useful strategy to gain novel insights into the biology of poorly understood bacteriophages.

The same combination of genome annotation and proteomics was repeated for the 28 putative Cp-1 proteins (Häuser et al., 2011). They detected a total of 12 structural proteins by MS analysis. More than half the proteins showed at least one protein–protein interaction with 17 binary interactions in total. Notably, nearly all were detected between proteins encoded in the structural gene cluster, reflecting a highly modular organization of its interactome. In contrast, Dp-1, with its larger proteome/interactome, shows more cross-communication between proteins belonging to different pathways, although a functional enrichment of interactions between proteins belonging to related cellular processes was still detected (Sabri et al., 2011). It is likely that larger phage proteomes/interaction networks routinely exhibit a higher frequency of protein cross-communication at the protein interaction level due to regulation between various pathways (i.e., the larger the proteome, the more promiscuous the interaction network).

1. Growth physiology

T7 grows on rough strains of E. coli (i.e., those without full-length O-antigen polysaccharide on their surface) and some other enteric bacteria, but close relatives have been described that infect smooth and even capsulated strains (Molineux, 2006a). In such phages, tail fibers present on T7 are replaced with tailspikes with enzymatic activity that degrades the O- or K-antigens on the cell surface. T7 has a short latent period of 17 min at 37°C, which has resulted in most physiological studies being conducted at 30°C where infected cells lyse after 30 min. However, adaptation of T7 to high fitness with an excess of an E. coli K-12 strain growing under optimal conditions at 37°C in rich media results in a phage with a latent period of only ≈11 min. This adapted phage can undergo an effective expansion of its population by more than 10 ^{in 1 hr of growth (}Heineman and Bull, 2007). There is no obvious subtlety in the T7 developmental pathway: the early region of its genome is transcribed efficiently by the host RNA polymerase, leading to synthesis of a phage-specific RNA polymerase and the concurrent inhibition of the host enzyme. Thus, T7 rapidly takes control of the transcriptional capacity of the infected cell. As E. coli mRNAs are also intrinsically unstable, T7 therefore also gains control of the translational capacity of the cell. Further, because T7 uses nucleotides derived from the host chromosome for its own replication, more than 80% of the DNA in progeny is derived from the bacterial chromosome and only a small burst is observed after infection of DNA-free minicells. T7 development is not known to be dependent on host chaperones, although it does code for two proteins that have been suggested to have chaperone activity in capsid and tail assembly. Aside from host components used for phage adsorption, plus general biosynthetic processes and the translational apparatus, T7 development is remarkably independent of host proteins: only E. coli RNA polymerase, thioredoxin—the cofactor for T7 DNA polymerase—and (deoxy)cytidine monophosphate kinase are known to be essential.

2. Early steps in infection

T7 is a member of Podoviridae, possessing only a short stubby tail that needs to be extended functionally in order to span the cell envelope, which thus allows its genome to be translocated into the cell. Tail extension is affected through internal head proteins that are ejected into the cell at the initiation of infection prior to or concomitant with transport of the leading end of the genome (Chang et al., 2010; Kemp et al., 2005). A schematic of the T7 virion is shown in Figure 2. In T7, the core consists of several copies each of three proteins (gp14, gp15, and gp16), which must interact, albeit perhaps loosely, with themselves and each other, in the phage capsid. However, during their ejection into the cell they must at least partially dissociate and also partially unfold in order to pass through the narrow channel through the portal and tail (Molineux, 2006b). The N-terminal domain of gp16 refolds spontaneously in the periplasm of the infected cell to form an enzyme with lytic transglycosylase activity (Moak and Molineux, 2000, 2004). This muralytic activity is only conditionally essential for T7 infection (Moak and Molineux, 2000).

FIGURE 2

Phage T7: genome, virion, and interactome. ORFeome and protein interaction map of bacteriophage T7. Cloned open reading frames (ORFs) and random fragments were used to generate a two-dimensional interaction map. ORFeomes can also be used by structural genomics projects to derive three-dimensional structures. A combination of interaction maps and crystal structures often allows reconstruction of protein complexes as in viral particles or their subunits. (Top) T7 genome with each ORF represented as a box. ORFs indicated by integral numbers were identified originally as essential by genetic screens. Other genes have decimal numbers; of these, only 2.5 and 7.3 are essential. Proteins found to interact with other proteins are indicated by colored boxes; colors indicate a simplified assignment to functional classes as shown. Interactions between proteins are shown as lines, with self-interactions shown as dimers. Hatched lines indicate expected interactions that have not been found by two-hybrid analysis. Gray proteins correspond to host proteins (E. coli). Blue proteins are involved in virus assembly and structure, and, where known, their location in the virus particle is shown on the right. Proteins with thick borders have been crystallized and their structure determined. Genetic map modified after Dunn and Studier (1983). Figure modified after Uetz et al. (2004). Interactions based primarily on data in Table II (see references therein).

The C-terminal residues of gp16, likely together with a more central portion of gp15, also refold in the periplasm and then insert spontaneously into the cytoplasmic membrane, completing the channel across the cell envelope for DNA translocation into the cell. Likely the last of the core proteins to exit the phage head, gp14 is found exclusively within the outer membrane of the infected cell. Using energy derived from the membrane potential, these proteins then ratchet the leading 1 kb of the T7 genome into the cell (Chang et al., 2010; Kemp et al., 2005). In both T7 and T3, genome internalization by this process stops, and the remainder of the DNA is normally pulled into the cytoplasm as a consequence of host and then phage RNA polymerase-catalyzed transcription (Garcia and Molineux, 1995, 1996; Kemp et al., 2004). Several strong promoters lie on this leading 1-kb segment of the T7 genome.

3. Early genes

Transcription from T7 early promoters not only internalizes the remainder of the genome but also leads to expression of early T7 genes, some of which interact and inhibit or modulate the activity of host proteins (Molineux, 2006a). Of the early proteins, only T7 RNA polymerase is essential for phage growth. The first T7 protein to be synthesized is gp0.3, an anti-type I restriction protein (Studier, 1975). The structure of gp0.3 reveals that it resembles double-stranded DNA, and the protein binds to the specificity subunit of the heteropentameric EcoKI enzyme, preventing both it and the cognate modification enzyme from recognizing its target sequence (Atanasiu et al., 2001, 2002; Kennaway et al., 2009; Murray, 2000; Stephanou et al., 2009). T7 gp0.3 is active against all type I enzymes and thus T7 grows normally on B, C, and K-12 strains of E. coli regardless of its previous host. However, the protein affords no protection against type II and little, if any, against type III enzymes. Gp0.3 proteins of different T7-like phages fall into two distinct groups with no sequence similarity; both have anti-restriction activity but the group typified by T3 gp0.3 also has SAMase activity (Studier and Movva, 1976). The second major early protein whose function is known is gp0.7, a serine-threonine kinase (Rahmsdorf et al., 1974) that is also responsible for the shut off of host transcription. The two activities are separable (Simon and Studier, 1973). Gp0.7 phosphorylates several host proteins, including RNA polymerase (Severinova and Severinov, 2006; Zillig et al., 1975), the translational apparatus including EF-G and the small ribosomal subunit protein S6 (Robertson and Nicholson, 1990, 1992; Robertson et al., 1994), and both RNase III (Marchand et al., 2001) and RNase E (Mayer and Schweiger, 1983). T7 shuts down translation of host mRNAs but it remains unclear whether EF-G or S6 is involved directly. An unknown T7 late gene also modulates the specificity of RNase E (Savalia et al., 2010). Although it is likely that others do so, only 1 other of the 10 early T7 proteins is currently known to interact with host proteins. Gp1.2 binds to E. coli dGTPase, converting it into a rGTP-binding protein (Huber et al., 1988; Nakai and Richardson, 1990); gp1.2 also binds to the membrane protein FxsA and the F plasmid protein PifA (Cheng et al., 2004; Schmitt and Molineux, 1991; Wang et al., 1999).

4. T7 RNA polymerase

The hallmark of T7 is of course its phage-encoded RNA polymerase, which, together with its regulator T7 lysozyme, is responsible for most T7 gene expression. It is also essential for synthesizing RNA primers at the primary origin of replication (Fujiyama et al., 1981) and for initiation of DNA packaging (Zhang and Studier, 1997, 2004). In addition to several complexes containing a promoter and product RNAs (Cheetham and Steitz, 1999; Cheetham et al., 1999; Kennedy et al., 2007; Tahirov et al., 2002; Yin and Steitz, 2002, 2004), the crystal structure of the RNA polymerase–lysozyme complex has been determined (Jeruzalmi and Steitz, 1998). Many mutants in both RNA polymerase and lysozyme have been characterized that affect their functional interaction, but it is not obvious from the crystal structures how the mutations affect the regulation of transcription. Lysozyme is thought to target the initial transcribing complex in a promoter strength-dependent manner and have no effect on transcription elongation (Huang et al., 1999; Stano and Patel, 2004; Zhang and Studier, 1997, 2004). Lysozyme is, however, important in pausing or terminating transcription at the class II terminator CJ, where it probably interacts directly with the T7 terminase protein gp19 and thus directs the packaging machinery to create the DNA terminus (Lyakhov et al., 1997, 1998; Zhang and Studier, 2004).

After T7 RNA polymerase has been synthesized, there is no further requirement for the host polymerase and it actually becomes inhibitory to phage development. Inactivation is caused by the T7 gp2 protein, which binds to the β′ subunit and inhibits transcription by interfering with strand separation of the open complex at most promoters (Camara et al., 2010; Nechaev and Severinov, 1999). In vivo, the only T7 promoter that must be inhibited by gp2 is A3 (Savalia et al., 2010). Gp2 was found to interact with T7 endonuclease gp3 by a yeast two-hybrid screen (Bartel et al., 1996). An interaction between gp2 and gp3 (albeit not necessarily direct) was suggested earlier (Mooney et al., 1980), but the idea has not yet been critically evaluated experimentally.

5. DNA replication

T7 has been one of the primary model systems for understanding the biochemistry of DNA replication. Central to T7 replication is T7 DNA polymerase, which comprises a heterodimer of gp5 and the host thioredoxin. Binding of gp5 to thioredoxin is strong, with the latter serving as the processivity factor for DNA replication, increasing the length of DNA synthesized per binding event 100-fold. The interface between the two proteins also provides the docking site for binding of the other two proteins, SSB (gp2.5) and the primase–helicase gp4, which are central to T7 replication in vivo. The process of T7 replication in vitro has been reviewed recently (Hamdan and Richardson, 2009).

Two other proteins in the DNA replication cluster are known to interact with host proteins. The gene 5.5 product binds to the abundant nucleoid-associated protein H-NS (Ali et al., 2011) and in vitro relieves its inhibition of both E. coli and T7 RNA polymerase-catalyzed transcription (Studier, 1981). In vivo, T7 5.5 mutants direct reduced rates of DNA replication and exhibit a small plaque phenotype (Lin, 1992; Liu and Richardson, 1993; Studier, 1981). However, the phenotype persists in hns-defective strains, although an hns mutant that renders gene 5.5 essential has been isolated (Liu and Richardson, 1993). Gp5.5 also interacts with the λ proteins of RexAB system; it is not known whether this is a direct protein–protein interaction or whether gp5.5 interacts with a product of RexAB activity. A gene 5.5 missense, but not a null mutant, makes T7 sensitive to Rex-mediated exclusion. The product of gene 5.9 inhibits the E. coli RecBCD nuclease in a comparable manner to λ Gam protein; this interaction is also nonessential as mutants lacking gene 5.9 grow normally in rec(BCD) ^{hosts (}Lin, 1992).

The assembly of T7 virions involves several symmetry mismatches and many protein–protein interactions, some of which may be helped by other T7 proteins that potentially have a chaperone-like activity (Kemp et al., 2005). Note that, GenBank annotations notwithstanding, gp13 is not a structural component of the virion (Kemp et al., 2005). Assembly is thought to initiate with oligomerization of the portal protein gp8 to form a dodecameric ring, although both 12- and 13-mers are observed following overexpression of gene 8 (Cerritelli and Studier, 1996b; Kemp et al., 2005; Valpuesta et al., 1992, 2000). The pathway of prohead formation may then involve assembly of the internal core (gp14, gp15, and gp16) on the inner surface of the portal, perhaps concomitant with the recruitment of capsid protein gp10 hexamers bound to scaffolding protein gp9. Subsequent recruitment of gp10 pentamers and hexamers results in shell formation around a scaffold of gp9 (Cerritelli et al., 2003a). The 12-fold symmetrical portal protein faces a symmetry mismatch with the 5-fold vertex of the icosahedral capsid shell. The internal core of T7 contains 4 copies of gp16, 8 copies of gp15, and either 10 or 12 copies of gp14 (Agirrezabala et al., 2007; Cerritelli et al., 2003b; Kemp et al., 2005). These proteins are ejected into the infected cell in order to transport the phage genome into the cytoplasm, and the protein–protein interactions in the mature virion are likely quite different than those found in the infected cell. Gp14 lies at the base of the internal core and is therefore juxtaposed to the 12-fold symmetrical portal protein. Complicating our understanding of the T7 virion is the identification of ≈18 copies of the small gp6.7 protein, which is ejected into the outer membrane of the infected cell at the initiation of infection (Kemp et al., 2005). It is not known where gp6.7 is located; it likely forms part of the internal core or lies within the portal channel.

6. Tail

The outer face of the portal is the site of assembly of the sixfold symmetrical stubby tail. The tail is composed of three proteins—gp7.3, 11 and 12—to which six trimers of the tail fiber protein gp17 are attached. Each trimer consists of three domains: an N-terminal third that binds to the tail; this domain is conserved in phages related to T7 that have a different host range. This domain is followed by a triple-stranded coiled-coil of ≈120 residues and then a C-terminal half consisting of globular domains that interacts with the cell surface lipopolysaccharide (Steven et al., 1988). The body of the tail consists of 12 copies of gp11, 6 copies of gp12, and ≈30 copies of the small protein gp7.3 (Kemp et al., 2005). How these proteins are arranged is unknown. It has been suggested that gp7.3 functions as a tail assembly protein that remains with the mature virion; the protein is ejected into the outer membrane of the infected cell (Chang et al., 2010; Kemp et al, 2005). Host range mutants of T7 suggest that all three proteins interact directly with the lipopolysaccharide of the host cell.