The multicomponent antirestriction system of phage P1 is linked to capsid morphogenesis

Summary

Bacterial Type I restriction-modification (R-M) systems present a major barrier to foreign DNA entering the bacterial cell. The temperate phage P1 packages several proteins into the virion that protect the phage DNA from host restriction. Isogenic P1 deletion mutants were used to reconstitute the previously described restriction phenotypes associated with darA and darB. While P1ΔdarA and P1ΔdarB produced the expected phenotypes, deletions of adjacent genes hdf and ddrA also produced darA-like phenotypes and deletion of ulx produced a darB-like phenotype, implicating several new proteins of previously unknown function in the P1 dar antirestriction system. Interestingly, disruption of ddrB decreased P1’s sensitivity to EcoB and EcoK restriction. Proteomic analysis of purified virions suggests that packaging of antirestriction components into P1 virions follows a distinct pathway that begins with the incorporation of DarA and Hdf and concludes with DarB and Ulx. Electron microscopy analysis showed that hdf and darA mutants also produce abnormally high proportions of virions with aberrant small heads, which suggests Hdf and DarA play a role in capsid morphogenesis. The P1 antirestriction system is more complex than previously realized and is comprised of multiple proteins including DdrA, DdrB, Hdf, and Ulx in addition to DarA and DarB.

Introduction

Bacteriophage P1 was discovered by G. Bertani in 1951 as a temperate phage residing in Escherichia coli strain ‘Li’ (Bertani, 1951). P1 is a myophage of the order Caudovirales, with a 94.8 kilobase (kb) unit genome of linear dsDNA (Lobocka et al., 2004). Unlike most other known temperate phages, P1 lysogenizes E. coli as a circular plasmid maintained at one copy per host chromosome (Ikeda and Tomizawa, 1968; Rosner, 1972). P1 has also played a major role in molecular genetics as the premier generalized transducing phage of E. coli (Lennox, 1955; Calendar, 1988).

In Caudovirales phages such as P1, the infection cycle begins with the adsorption of the phage to the host surface and the ejection of phage DNA into the host cell. Bacteria have several mechanisms to defend themselves against phage infection, including restriction and modification (R-M) systems that recognize and cleave foreign DNA in a site-specific manner (Fig. 1A). These systems are broadly divided into three categories, Type I, II and III, which are distinguished by their subunit makeup, DNA cleavage mechanisms and the nature of their DNA recognition sequences (Kruger and Bickle, 1983; Loenen et al., 2014).

Fig. 1. The antirestriction system of phage P1 protects DNA from Type I Restriction and Modification (R-M) mediated cleavage.

A. Phage infection begins with the translocation of the linear 95 kb P1 genome from the phage head into the host cytoplasm. Depending on the bacterial host where phages are propagated, specific bases in the phage DNA are methylated (denoted as stars). If the methylation pattern in the DNA matches the specificity of the host Type I R-M system, the DNA is protected from cleavage (left), resulting in normal plating efficiency. If the phage DNA is unmethylated (or improperly methylated), the DNA is cleaved by the host Type I R-M system (middle), resulting in reduced plating efficiency. Iida et al. (1987) demonstrated that in P1, virion-associated proteins (triangles) can protect P1 DNA from Type I restriction, even if it is unmodified (right). It is believed that at least some of these virion-associated proteins must translocate into the cell to function.

B. Partial genomic map of P1 shows genes associated with the antirestriction system. Rectangular blocks represent genes, and those above the black line are transcribed rightwards, and genes below the line are transcribed leftwards. Besides the previously described darA and darB, several other genes are associated with the P1 antirestriction system. The dar genes (gray) are organized in two separate predicted transcriptional units. One transcript contains darB and ulx (darB operon, driven by the PdarB promoter) and the other contains hdf, darA, ddrA and ddrB (darA operon, driven by Pdar). The ruler at the bottom of the figure corresponds to the nucleotide position as it appears in the published P1 genome (NC_005856.1).

Type I R-M systems are composed of three subunits, HsdR (R), HsdM (M) and HsdS (S) (Hsd for host specificity of DNA). These subunits can form two different oligomeric complexes; R₂M₂S₁, which can catalyze both restriction and modification, and M₂S₁, which can only catalyze modification (Suri et al., 1984; Murray, 2000). The HsdS subunit recognizes specific, bipartite DNA sequences of 13–15 bp, the HsdM subunit recognizes the modification status in the recognition DNA sequences and the HsdR subunit catalyzes DNA cleavage at a non-specific location up to a few kb away from the recognition sequence, if DNA is unmethylated or is methylated inappropriately (Murray, 2000). Unlike Type I R-M systems, Type II R-M systems have independent restriction and modification enzymes. These restriction enzymes cleave DNA within the recognition sites if the DNA is not appropriately modified (Bickle and Kruger, 1993). Similar to Type II R-M systems, Type III systems are comprised of separate modification and restriction subunits. The modification subunit can function by itself as a modification methylase. However, a complex of modification and restriction subunits is required for endonuclease function, in which DNA cleavage occurs 25–27 bp away from the recognition site (Tock and Dryden, 2005).

P1 encodes the virion-associated proteins DarA and DarB (dar standing for defense against restriction), which are required to protect the P1 genomic DNA (gDNA) from restriction by host Type I R-M systems (Fig. 1A) (Iida et al., 1987). DarA, with a predicted molecular mass of 69 kDa, is expressed late in the lytic cycle and is processed into a smaller 60 kDa form when incorporated into the virion (Streiff et al., 1987). DarB has a predicted molecular mass of 251 kDa and appears to be incorporated into the virion intact (Iida et al., 1987). Iida et al. demonstrated by plaque assay that DarA is required for protection of DNA against restriction by the EcoA Type I R-M system and DarB is required for protection against the EcoB and EcoK systems. The efficiency of plating (EOP) of P1 darA mutants was reduced in EcoA, EcoB and EcoK strains whereas the EOP of P1 darB mutants was reduced only in EcoB and EcoK strains. Because complementation studies indicated that the Dar proteins function only in cis, it has been proposed that these are ejected into the host cell along with phage DNA at the initiation of infection, where they exert their antirestriction activity (Fig. 1A) (Iida et al., 1987). The antirestriction proteins synthesized during the phage lytic cycle only act in the following infection cycle, and the antirestriction activity is not specific to P1 gDNA, but any DNA that is packaged into the P1 virion (Iida et al., 1987). Since darA⁻ phages also exhibit the darB⁻ phenotype and darA⁻ virions lack both DarA and DarB proteins, it appears that DarA is required for the incorporation of DarB into the P1 capsid (Iida et al., 1987). The 2,255-residue DarB contains an identifiable N-terminal methyltransferase domain and a central DExD-like helicase domain, enzymatic functions that could be imagined to play a direct role in the protection of DNA from restriction (Lobocka et al., 2004). Bioinformatic analysis of DarB revealed that it is widely distributed among bacteria, primarily associated with mobile DNA elements including phages, plasmids and conjugative transposons (Gill et al., 2011). It has been proposed that DarB could represent a larger class of proteins that facilitate DNA mobility by protecting mobile DNA elements from Type I R-M system-mediated cleavage during their transfer between hosts.

To further explore the antirestriction system of phage P1, we used a recombineering approach to create isogenic deletion mutants of darA, darB, and other P1 genes with previously unknown function including hdf, ddrA, ulx and ddrB. The results are discussed in terms of a network of proteins involved in the assembly of P1 antirestriction system and their effects on capsid morphogenesis.

Results and discussion

Genomic context of the P1 antirestriction system

The genome of phage P1 was completed in 2004 (NC_005856.1), showing the positions of the darA and darB structural genes within the P1 chromosome (Lobocka et al., 2004). As shown in the partial P1 genetic map (Fig. 1B) the 1,920 bp darA gene is located in a polycistronic operon with its expression driven by LPdar (LP for late promoter) (Guidolin et al., 1989), and darB (6,768 bp) is located ~10 kb downstream of darA in a separate operon driven by PdarB (Lobocka et al., 2004). The darA gene is located immediately downstream of hdf, and upstream of ddrA, ddrB and hxr. The genes lydA and lydB, upstream of hdf, are known to function as the phage holin and antiholin respectively (Schmidt et al., 1996). Two operons are located between the dar operons: one containing the transcript of the phage’s SAR endolysin lyz (Xu et al., 2004) and another on the opposite strand encoding the predicted phage portal, portal protease and single-stranded DNA binding protein (Lobocka et al., 2004). The darB gene is upstream of ulx and the lysogeny maintenance gene lxc (Lobocka et al., 2004).

The genes shown in gray in Fig. 1B were deleted and replaced with kan markers by a lambda Red recombineering approach (Datsenko and Wanner, 2000). Initially darA and darB were deleted, followed by hdf, ddrA, ddrB, hxr and ulx. Because lydAB, lyz, lxc and the phage structural proteins Prt and Pro already had assigned functions, they were not considered to be relevant to this study and were not studied further. Deletion of hxr produced no detectable antirestriction phenotype against EcoA, EcoB or EcoK (data not shown) and genes downstream of hxr were not examined further.

darB and ulx provide protection against EcoB and EcoK

The antirestriction phenotypes associated with P1 darA and darB were first reported in 1987 (Iida et al., 1987): P1, when propagated on a modification-deficient host, exhibited a slight (less than 10-fold) reduced efficiency of plating (EOP) on EcoA and EcoK lawns and an EOP of ~10⁻² on EcoB. Under the same conditions, P1darA⁻ plated at an EOP ~10⁻⁴ to 10⁻⁵ on all three restrictive hosts, whereas P1darB⁻ was restricted (EOP ~10⁻⁴) only on EcoB and EcoK.

The previously reported darB⁻ phenotype, defined as severely reduced plating efficiency on EcoB and EcoK hosts, was recapitulated in an isogenic deletion of darB (Fig. 2A). As previously reported, P1ΔdarB did not exhibit an antirestriction defect in the EcoA host. In addition to reproducing the darB phenotype, the antirestriction defect in P1ΔdarB could be efficiently complemented in trans by inducing the mutant phage from a lysogen of the non-modifying host strain WA921 carrying the plasmid pdarB, expressing DarB from an arabinose-inducible promoter (Fig. 2A). In this complementation system, the progeny phage could incorporate the DarB protein expressed from the plasmid in the propagating host and produce the antirestriction phenotype on infection of a restricting strain. Phages complemented in this way could only express the antirestriction phenotype for their first infection cycle. Complemented mutant phage that were re-propagated on WA921 exhibited the same EOP as the original P1ΔdarB when plated to all three of the restricting strains (data not shown), indicating the absence of recombination between the phage and complementing vector during phage propagation. The reduced EOP of P1ΔdarB in EcoB and EcoK cells indicates that in the absence of DarB, P1 DNA, after being ejected into host cells, is susceptible to EcoB and EcoK restriction, but is protected from EcoA restriction.

Fig. 2. darB and ulx are required for protection of P1 against EcoB and EcoK restriction. Parental P1 or isogenic deletion mutant phages were thermally induced from lysogenized modification-deficient WA921 containing either empty pBAD24 or the respective complementing plasmid. Phages were then plated on E. coli hosts containing the Type I EcoA, EcoB, or EcoK R-M systems and WA921; efficiency of plating (EOP) for each phage was calculated as the ratio of plaques formed on lawns of the R-M-containing strains to the plaques formed on WA921. The EOP data shown have been normalized to the EOP of parental P1 induced from WA921 containing pBAD24.

A. The relative EOP of P1ΔdarB was reduced by ~10⁻³ and 10⁻⁴ in EcoB and EcoK strains respectively (gray bars).

B. The relative EOP of P1Δulx was reduced by ~10⁻² and 10⁻³ in EcoB and EcoK strains respectively (gray bars). The restriction phenotype associated with both darB and ulx could be complemented in trans (white bars). No notable phenotype was observed in the EcoA strain. The data shown are averages of three biological replicates and error bars represent standard deviation.

Other genes near darB were examined for their possible role in antirestriction. The gene ulx (146 codons; ulx for upstream of lxc (Lobocka et al., 2004)) is located between darB and lxc and was previously of unknown function. P1Δulx showed a partial darB-like restriction phenotype, with EOP attenuated approximately 10-fold less than for P1ΔdarB (Fig. 2B). The ulx-associated restriction phenotype could also be complemented in trans (Fig. 2B). In both darB and ulx deletions, complementation produced slightly higher EOP’s than the parent P1 phage, suggesting complementation could provide a greater copy number of protein available for incorporation into the virion and subsequent greater protection against restriction. The plaque sizes for both P1ΔdarB and P1Δulx were comparable to P1. Moreover, a double deletion mutant of darB and ulx showed plating deficiency similar to P1ΔdarB (data not shown). Thus, both darB and ulx are required for full protection of P1 DNA against EcoB and EcoK restriction.

darA, hdf and ddrA contribute to protection against EcoA, EcoB and EcoK

In agreement with previous work, the isogenic deletion P1ΔdarA had reduced EOP when plated on strains expressing EcoA, EcoB or EcoK (Fig. 3A). Relative to P1, the EOP of P1ΔdarA was decreased by ~10⁻³, 10⁻² or 10⁻⁴ in cells expressing EcoA, EcoB or EcoK respectively. The EcoA-associated darA restriction phenotype could be only partially complemented in trans by darA alone, and the EcoB and EcoK phenotypes were not affected. It was previously reported that darA mutants could be complemented to nearly wt levels in trans by a P1 DNA fragment containing darA, along with the upstream gene hdf and the downstream gene ddrA of the darA operon (Iida et al., 1998). In our experiments, P1ΔdarA was more efficiently complemented in trans by a construct containing darA, hdf and ddrA (data not shown). Better complementation in presence of hdf and ddrA raised a possibility of the polar effects of darA deletion, in which the restriction phenotype observed could either be associated with hdf or ddrA, the other genes present in the complementing plasmid. To rule out the roles of Hdf and DdrA in complementation, nonsense mutations were introduced in both hdf and ddrA to prevent their expression. Same degree of complementation was observed with phdf_am_darA_ddrA_am (Fig. 3A). The low efficiency of complementation exhibited by the pdarA vector may be due to instability of the mRNA transcript containing darA alone, as the longer DNA fragments included on phdf_am_darA_ddrA_am and the previously reported P1 DNA fragment provided better complementation.

Fig. 3. hdf, darA and ddrA are required for protection of P1 against EcoA, EcoB and EcoK restriction. Phages were induced and EOP assays were performed as described previously. The EOP data shown have been normalized to EOP of parental P1 induced from modification deficient WA921 containing pBAD24.

A. The relative EOP of P1ΔdarA was reduced by ~10⁻³, 10⁻² and 10⁻⁴ in EcoA, EcoB and EcoK strains respectively (dark gray bars). The darA-associated restriction phenotype could be complemented partially by a plasmid bearing only darA (light gray); stronger complementation was observed with a plasmid expressing a longer transcript containing darA, along with the upstream gene hdf_Am, and downstream ddrA_Am (white bars).

B and C. Both P1Δhdf and P1ΔddrA exhibited a darA-like restriction phenotype (gray bars). The restriction phenotypes of both hdf and ddrA could be complemented in trans (white bars). The data shown are averages of three biological repeats and the error bars represent standard deviation.

Previous studies of darA and darB function used P1 mutants that contained either IS insertions or deletions that often spanned hdf and ddrA (Iida et al., 1987). To determine if these genes could also play a role in the protection of P1 DNA from host restriction, the isogenic mutants P1Δhdf and P1ΔddrA were created and evaluated for their antirestriction phenotypes. Both P1Δhdf and P1ΔddrA showed darA-like phenotypes, with sensitivity to EcoA, EcoB and EcoK restriction similar to that exhibited by P1ΔdarA (Fig. 3B and 3C). Unlike darA, the hdf and ddrA phenotypes could both be fully complemented in trans by vectors containing hdf or ddrA alone (Fig. 3B and 3C). The plaque sizes for both P1Δhdf and P1ΔdarA were smaller than that of P1, whereas the plaques of P1ΔddrA appeared to be of normal size.

ddrB modulates the EcoB and EcoK antirestriction phenotype

Having determined the restriction phenotypes in ulx, hdf and ddrA mutants, we tested other genes located within the darA and darB operons (Fig. 1B). Surprisingly an isogenic deletion of ddrB was found to have a ~10-fold higher EOP on cells expressing EcoB or EcoK than that of the parental P1 (Fig. 4). This phenotype could again be complemented in trans by a vector containing ddrB alone, with the EOP returning to ~3 in the complemented phage. Deletion and complementation of ddrB produced no notable phenotype on EcoA strains. Both ulx and darB are similar to ddrB in this regard, in that their deletion affects susceptibility to EcoB and EcoK restriction, with no apparent effect on EcoA strains (Fig. 2). It seems likely that ddrB exerts its phenotype via action on Ulx and/or DarB, either by regulating the copy number of these proteins incorporated into the virion, or by modulating their activity after the antirestriction system is deployed in the host cell.

Fig. 4.

ddrB negatively affects the darB phenotype. Phages were induced and EOP assays were performed as described previously. The EOP data shown have been normalized to EOP of parental P1 induced from modification-deficient WA921 containing pBAD24. Unlike the other restriction phenotypes described, disruption of ddrB increased the relative EOP of the phage in EcoB and EcoK strains by approximately 10-fold compared to wild-type P1 (gray bars). This effect did not extend to EcoA strains, suggesting that DdrB negatively impacts the activity of DarB. The ddrB phenotype could also be complemented in trans (white bars).

Virion proteomics of P1 and its mutants

It was shown previously that P1darA⁻ phages fail to package DarB (Iida et al., 1987), leading to the conclusion that, in addition to being required for protection of DNA against EcoA, DarA is also required for incorporation of DarB into the virion. In this study, multiple genes in addition to darA and darB have been identified with roles in the antirestriction phenotype of P1. These phenotypes can be placed into two broad categories: darA-like, which affects phage EOP on EcoA, EcoB and EcoK strains, and darB-like, in which EOP is affected only on EcoB and EcoK strains. To determine how these multiple P1 genes could produce similar restriction phenotypes, proteomic analysis of purified virions was conducted (Fig. 5). Protein bands corresponding to DarB, Hdf, DarA and DdrB could be assigned by mass spectrometry analysis (underlined in Fig. 5A). The protein band corresponding to DdrA is assigned based on missing band density corresponding to its predicted protein molecular mass in lanes of P1ΔddrA virions compared to the parental P1. The protein band corresponding to Ulx was assigned based on western blotting of purified P1Δulx virions complemented with a Ulx-FLAG fusion (Inclan et al., 2016) (Supporting Information Fig. S1).

Fig. 5. Comparative SDS-PAGE of parental P1 and isogenic mutants suggests cascading dependencies for protein incorporation into the P1 virion.

A. SDS-PAGE of parental P1 and antirestriction mutants. Molecular masses of the size standard are provided in kDa on the left. Relevant protein bands are labeled on the left; proteins that have been identified by mass spectrometry are underlined. DdrA is assigned based on the missing band density in the gel image and predicted molecular mass, and Ulx is assigned based on Western blotting of P1Δulx complemented with Ulx-FLAG with anti-FLAG antibody. At the bottom of the figure, the restriction sensitivity phenotypes of the P1 mutants are indicated: B, K (sensitive to EcoB and EcoK restriction), or A, B, K (sensitive to EcoA, EcoB and EcoK restriction).

B. The table summarizes protein presence or absence in P1 mutants. P1Δulx shows a faint DarB band, which suggests that DarB can be incorporated in the absence of Ulx, albeit inefficiently. When either hdf or darA is disrupted, P1 fails to package both Hdf and DarA, suggesting that Hdf and DarA are co-dependent for virion incorporation.

Disruption of darB resulted in the loss of DarB and Ulx (Fig. 5), indicating only Ulx is dependent on DarB for incorporation. Coupled with the conserved domains present in DarB, the EcoB and EcoK phenotypes observed for P1ΔdarB (Fig. 2A) suggests that DarB is directly responsible for EcoB and EcoK antirestriction activity. Band densitometry indicates P1Δulx packages only ~15% of the DarB as the parental P1 (data not shown). Thus, DarB can be packaged into the virion in the absence of Ulx but packaging efficiency is greatly reduced. This reduced packaging of DarB is consistent with the ~10-fold weakened antirestriction phenotype exhibited by P1Δulx compared to P1ΔdarB (Fig. 2).

Compared to the parental P1, a single ~50 kDa band is missing in the P1ΔddrB virion; this species was confirmed as DdrB by mass spectroscopy analysis. Aside from this 50 kDa species, all bands including those corresponding to DarB, Ulx, DarA, Hdf, and DdrA are present, indicating that no other virion proteins are dependent on DdrB for their incorporation. Since ddrB encodes a protein with a predicted molecular mass of ~109 kDa, the appearance of DdrB at a position corresponding to 50 kDa suggests that the protein is proteolytically processed for incorporation into the virion; no ~109 kDa band corresponding to DdrB was observed in SDS-PAGE. Based on MS-MS analysis of the gel slice containing this predicted DdrB band from the parental P1, high peptide coverage (5–27-fold, covering 88% of residues) was observed for the N-terminal portion of the protein sequence, with coverage dropping abruptly after residue 421 (1 to 3-fold, covering ~42% of residues), suggesting that it is primarily the N-terminal portion of the DdrB protein that is present in the P1 virion. Cleavage of the protein at a position at or shortly after 421 would result in a ~46 kDa product, which matches closely the observed 50 kDa DdrB band. Presumably some small amount of the C-terminal fragment somehow remains with the virions, and this fragment would also be ~50 kDa and co-migrate with the C-terminal portion. The phenotype associated with ddrB deletion is an increase in protection against EcoB and EcoK restriction. One possibility is that the absence of DdrB allows for incorporation of more DarB into the virion, providing greater antirestriction activity. However, band densitometry does not convincingly show greater DarB band intensity in ddrB mutants (data not shown). An alternative hypothesis is that DdrB actively suppresses DarB-associated antirestriction activity. In either case, the role of DdrB in the P1 antirestriction system is not clear, as its presence appears to counterintuitively hinder the fitness of P1 as measured by its ability to overcome host restriction.

As shown in Fig. 5, P1Δhdf, P1ΔdarA and P1ΔddrA virions all fail to incorporate DarB, DdrB and Ulx. EcoB and EcoK restriction sensitivity of these mutants can be explained by the absence of DarB because DarB is most likely responsible for EcoB and EcoK antirestriction activity as discussed above. However, hdf, darA and ddrA mutants are also all sensitive to EcoA restriction, which suggests that one of these three genes is responsible for the EcoA antirestriction activity of P1. Deletion of either hdf or darA results in virions missing both Hdf and DarA, which indicates that these proteins are co-dependent for virion incorporation. Both the P1Δhdf and P1ΔdarA virions are also missing DdrA in addition to DarB, DdrB and Ulx mentioned above. It was suggested that DarA is responsible for EcoA protection (Iida et al., 1987), however, DarA was found to be present in P1ΔddrA virions, which are sensitive to EcoA restriction (Figs. 3C and 5). This suggests that DarA is not solely responsible for protection from EcoA restriction as previously proposed. Aside from DarB, DdrB and Ulx, the only protein observed to be absent in P1ΔddrA is DdrA itself. Thus, DdrA may be the protein that protects P1 DNA from EcoA-mediated endonuclease activity, although it is not clear if DarA and Hdf are also required for DdrA activity or are simply required for incorporation of DdrA into the virion.

Band densitometry was performed on SyproRuby-stained SDS-PAGE of P1 using ImageJ (Schneider et al., 2012) to estimate the copy numbers of P1 antirestriction system components associated with the virion. These calculations suggest that there are ~45 DarB, ~100 DarA, ~40 DdrB and ~400 Hdf molecules in the average P1 virion (Supporting Information Table S2). The copy number of Ulx and DdrA could not be estimated due to their low mass and band intensity. These values indicate that P1 capsids contain ~35 MDa of major capsid protein and a combined mass of ~28 MDa of Dar proteins (DarB, DarA, DdrB and Hdf) per virion, indicating the Dar proteins contribute to a significant proportion of the mass of P1 heads.

Morphological defects are linked to hdf and darA

Under normal conditions, P1 is known to produce three morphological head size variants termed ‘big’ (P1B, ~86 nm), ‘small’ (P1S, ~65 nm) and ‘minute’ (P1M, ~47 nm) (Walker and Anderson, 1970). The P1B variant (hereafter referred to as ‘normal’ head size or P1N) packages the complete genome and is infectious, while P1S and P1M are defective as they can package only ~40% and ~10% of the P1 genome respectively (Walker and Anderson, 1970). The proportion of P1S virions has been reported to vary depending on the host, with P1 lysates induced from E. coli K-12 containing 20–37% P1S heads; P1M heads have been reported as less than 1% of most lysates (Walker and Anderson, 1970).

In isopycnic gradient purification of P1 and its antirestriction mutants, variations in banding patterns were observed in some of the mutant phages, particularly P1Δhdf and P1ΔdarA (Fig. S2). These mutants exhibited more intense bands at regions in the gradients corresponding to lighter buoyant densities than the parental P1, which suggested that the disruption of these genes also affected viral morphogenesis. The phage bands were extracted whenever possible and the head size of purified phages were measured from negative-stained TEM images and their distribution is described in Supporting Information Fig. S2.

To study the effects of antirestriction gene deletions on the overproduction of P1S virions, the parental and mutant phages were induced from lysogens of E. coli MG1655 and a head-size survey was conducted for each phage by measuring the diameters of several hundred phage heads photographed from negative-stain TEM images prepared from the phage lysates. Observed phage head diameters ranged from 52 to 95 nm (Fig. 6). In accordance with previous reports, head diameters >67 nm were classified as ‘normal’ (i.e., P1N) and head diameters ~67 nm were classified as small (P1S) variants. Virions with ‘minute’ (P1M) head size were rarely observed and were excluded from classification. All phages produced heads of a bimodal size distribution, with a smaller size class of ~60 nm and a normal size class centered around ~80 nm (Fig. 6, Supporting Information Table S3). As shown in Fig. 6, the parental P1 produced virions with ~82% normal-sized heads and ~18% small heads, in agreement with previous observations of P1 (Walker and Anderson, 1970). In lysates of P1Δhdf and P1ΔdarA this relationship was essentially inverted, with P1Δhdf producing ~85% small-headed virions and P1ΔdarA producing ~82% small-headed virions. Other phage mutants produced small heads at frequencies similar to the parental P1, ranging from 14.1% for P1ΔddrB to 28.1% for P1ΔdarB, however this variation is within the range previously reported for P1 (Walker and Anderson, 1970).

Fig. 6. Head size variation in parental P1 and mutants. Lysates of phage P1 contain virions of two major head size classes. The virions of >67 nm head diameter have been classified as ‘normal’ and those of ≤67 nm as ‘small’.

A. Transmission electron micrograph (TEM) showing typical normal head (left) and small head (right) virions.

B. The diameters of parental P1 and antirestriction mutant virions were measured from TEM images. The X-axis in each histogram denotes the head size diameter in nm and the Y-axis denotes the percentage of virions of each head size present. The phage genotype (parental P1 or antirestriction mutant) and the number of virions measured are denoted in the upper right of each histogram. On the right, the percentages of virions falling into the small and normal head size classes for each phage mutant are summarized. The darB, ulx, ddrB and ddrA mutants have similar head size distributions as parental P1, while this distribution is reversed in hdf and darA mutants, which are dominated by small-headed virions.

Previous work has shown that deletion of the entire darA operon, or transposon insertions within lydA and darA, resulted in overproduction of P1S virions in a phenotype denoted Vad⁻ (for viral architecture determinant) (Iida et al., 1998). Based on the data presented here, the Vad⁻ phenotype can be assigned to two genes within the darA operon, hdf and darA. These genes are directly adjacent to each other (Fig. 1B) and are the earliest non-lysis genes present in the Pdar transcript. Deletion of either gene results in the failure to incorporate any of the other virion-associated antirestriction components (Fig. 5), and the loss of antirestriction activity against EcoA, EcoB and EcoK (Fig. 3). The presence of the Vad⁻ phenotype in both P1Δhdf and P1ΔdarA indicates that both Hdf and DarA play important roles in determining head size during P1 morphogenesis.

In many viruses (bacteriophages such as P22, T4, λ and eukaryotic viruses such as HSV-1 and other herpesviruses), assembly of the major capsid protein into phage heads of the correct size and shape is guided by a scaffolding protein, resulting in a spherical immature procapsid (Mateu, 2013). Studies in phages T4, P22 and λ have shown that procapsids are assembled initially without any DNA (King et al., 1980). During pro-capsid maturation and concomitant DNA packaging, the scaffolding protein is lost from the capsid (Dokland, 1999). Absence of functional scaffolding proteins results in aberrant capsid assembly or no assembly at all (Dokland, 1999; Aksyuk and Rossmann, 2011). No major scaffold protein has been identified for P1 (Lobocka et al., 2004). Disruption of hdf and darA results in an increased proportion of small P1 heads (Fig. 6), suggesting that Hdf and DarA play roles in directing proper capsid assembly, most likely at the time of procapsid formation. However, Hdf and DarA are still present in mature virions (Fig. 5) and hence do not fit the classical definition of scaffolding proteins which are present in viral proheads but absent in mature virions (Dokland, 1999). Moreover, the presence or absence of these proteins do not result in strict phenotypes: the deletion of hdf or darA still produces ~20% normal-sized phage heads, and the parental P1 containing intact hdf and darA produces ~20% aberrant capsids. Hdf and DarA may not play a direct role in the observed antirestriction activity of P1, but rather their absence prevents the incorporation of the other antirestriction system components.

Since the absence of either Hdf or DarA was found to induce greater production of small head-sized virions, we wished to determine if the small head-size progeny of parental P1 are capable of incorporating Hdf and DarA. Comparative SDS-PAGE band intensities of P1N and P1S virions suggest that small-head phages incorporate ~65% of the major capsid and ~40% of the DarB proteins relative to normal P1N virions (Supporting Information Fig. S3). These numbers are comparable to what would be expected from a reduction in head diameter from 80 to 60 nm, with concomitant reductions of capsid surface area by ~56% and capsid volume by ~42%. While the P1S virions are capable of incorporating both DarA and Hdf, band densitometry suggests that P1S heads incorporate only ~14% and ~12% of DarA and Hdf proteins respectively, relative to P1N (Supporting Information Fig. S3). At this point it is difficult to determine if P1S heads are produced as a result of reduced incorporation of Hdf and DarA, or if lower amounts of Hdf and DarA are incorporated into pro-heads that have already committed to the smaller size.

The incorporation of the P1 antirestriction proteins into the capsid appears to be linked to the early stages of capsid morphogenesis. All of the virions run on SDS-PAGE in Fig. 5 were taken from CsCl bands corresponding to normal phage buoyancy (1.46–1.48 g cc⁻¹, Supporting Information Fig. S2) and electron microscopy surveys of these fractions confirmed that they contained >95% normal head size P1N virions (data not shown). Even though the P1Δhdf and P1ΔdarA virion proteomes shown in Fig. 5 are the minority fraction of the phage lysate containing normal-sized heads, the downstream antirestriction components DdrA, DdrB, DarB and Ulx are still absent from the virions, demonstrating that the formation of normal-sized heads is in itself not sufficient for incorporation of these antirestriction proteins. This suggests that it is Hdf and DarA themselves, and not their effects on head morphogenesis, that determine the incorporation of the other antirestriction proteins.

Conclusions

The P1 Dar system is composed of multiple proteins: Hdf, DarA, DdrA, DdrB, DarB and Ulx. The role of Hdf and DarA in antirestriction system is unique because they are also involved in the production of normal head-size virions. In the presence of Hdf and DarA, ~80% of P1 progeny are of normal head-size and ~20% are of small head-size (Fig. 7A). The distribution of normal and small head-size is reversed in the absence of either Hdf or DarA (Fig. 6). Because the absence of either Hdf or DarA results in a high abundance of aberrant small capsids, it is probable that these proteins play a role in the early stages of prohead formation. The proteins Hdf, DarA and DdrA are required for the protection of phage DNA from restriction by the EcoA Type I R-M system, and DarB is required for protection against EcoB and EcoK. Ulx enhances the antirestriction phenotype of DarB, and DdrB appears to negatively affect this activity by an unknown mechanism (Fig. 7B). All of these phenotypes could be complemented in trans. Proteomic analysis of isogenic P1 mutants shows a clear order of incorporation for each component into the mature virion, as illustrated in Fig. 7B. In this proposed model, Hdf and DarA are incorporated first in a co-dependent manner, shortly after prohead initiation. The absence of either Hdf or DarA results in failure to incorporate any of the other antirestriction proteins. Following Hdf and DarA, DdrA is incorporated, followed by DarB, DdrB and Ulx. The incorporation of these last three proteins are not strictly co-dependent: the absence of DarB results in no apparent Ulx incorporation, but absence of Ulx results only in reduced DarB incorporation. This behavior suggests that Ulx acts as a chaperone or packaging factor to increase the amount of DarB incorporated into the virion. The presence of DdrB results in reduced antirestriction activity, presumably by negatively affecting DarB but the mechanism is not known.

Fig. 7. Models of P1 head-size determination and antirestriction component incorporation.

A. The parental P1 with intact darA and hdf produces ~80% normal head-size (P1N) and ~20% small head-size (P1S) progeny. In the absence of either Hdf or DarA, this ratio is inverted, producing progeny with ~80% aberrant small heads.

B. Incorporation of P1 antirestriction system components follows a distinct pathway. The pathway presumably begins with the initiation of prohead formation by the phage portal and scaffold proteins, followed by co-incorporation of Hdf and DarA into the prohead. In addition to exacerbating the head-size defect as shown in Panel A, the absence of Hdf or DarA prevents the incorporation of all downstream Dar components. The incorporation of Hdf and DarA does not confer protection against EcoA, EcoB or EcoK restriction. Incorporation of DdrA appears to be the next step, as ddrA mutants incorporate only DarA and Hdf. Incorporation of DdrA provides protection against EcoA restriction. The incorporation of DarB, DdrB and Ulx is difficult to separate into separate steps, but packaging of DarB into virions provides protection against EcoB and EcoK restriction. Ulx appears to be an enhancer of DarB packaging as ulx mutants incorporate less DarB than normal, while DdrB acts as a negative regulator of DarB activity.

Given the incorporation of these proteins into the P1 virion and the cis-acting nature of the antirestriction phenotype, it is evident that the P1 antirestriction system must exert its activity on infection of the host cell (Iida et al., 1987). Since it is difficult to imagine how these proteins could act while still in the confines of the capsid, they must be introduced into the host cytoplasm via the phage tail on infection. It is not a requirement that all six of the antirestriction proteins described here be translocated into the cytoplasm to perform some direct protective action. Some proteins, particularly the more upstream components such as DarA and Hdf, likely only serve to ensure the incorporation of the active components into the virion. Strictly speaking, a minimum of only two proteins are required to account for the observed phenotypes: one that provides protection against EcoA, and one that protects against EcoB and EcoK. Based on bioinformatic evidence, DarB is the likely candidate to provide EcoB and EcoK protection as the predicted methyltransferase and helicase domains of DarB suggest a possible enzymatic mechanism to provide protection from host restriction. The protein or proteins responsible for the observed anti-EcoA activity are more difficult to assign. DdrA is one candidate, as phage lacking DdrA contain DarA and Hdf but do not express any antirestriction activity.

The translocation of phage proteins from the head into the host cytoplasm is a known feature in several Caudovirales phages. The capsid of the temperate Salmonella Typhimurium podophage P22 contains three internal proteins, gp16, gp20 and gp7, that leave the confines of the capsid to perform their function of assisting in translocation of DNA across host membranes (Jin et al., 2015). Coliphage N4 is known to translocate its ~380 kDa DNA-dependent RNA polymerase from the phage head into the host cytoplasm early in the infection process (Choi et al., 2008). Perhaps the closest analog to the P1 system shown in this work is the paradigm coli-phage T4, whose heads contain three non-essential internal proteins, IPI, IPII and IPIII (Black and Ahmad-Zadeh, 1971). The ~10 kDa T4 IPI is ejected into the host cytoplasm during phage infection and protects the incoming phage DNA by inhibiting a glucosyl-hmC DNA-specific restriction endonuclease (Bair et al., 2007). The proteins of the Dar system are assembled into the P1 capsid and at least some of them are ejected into the host cytoplasm, where they protect P1 DNA from cleavage by Type I R-M systems. In our study of the P1 antirestriction system, we determined that the incorporation of the P1 antirestriction proteins follow a distinct pathway that involves at least six proteins. We also found that proteins involved in the assembly pathway influence head morphogenesis. Phage P1 is the first reported instance, to our knowledge, where the antirestriction system has also been implicated in capsid morphogenesis.

Experimental procedures

Bacterial strains and phages

The bacterial strains and the parent phage P1CMclr100 used in this study were obtained from the Coli Genetic Stock Center, Yale University, and are listed in Table 1. P1CM is a derivative of P1kc that acquired chloramphenicol resistance from the R-factor R₁₄ (Kondo and Mitsuhashi, 1964). P1CMclr100, hereafter referred to simply as P1, is a thermoinducible mutant of P1CM (Rosner, 1972). All phage mutants used in this study are single-gene deletions of P1. Unless otherwise noted, E. coli strains were cultured on LB broth [10 g L⁻¹ Bacto Tryptone (BD), 5 g L⁻¹ Bacto yeast extract (BD), 10 g L⁻¹ NaCl (Avantor)] or LB agar (LB broth amended with 15 g L⁻¹ Bacto agar) at 37°C. P1 lysogens were cultured and maintained at 30°C on LB amended with 10 μg mL⁻¹ chloramphenicol (LB cm) or 10 μg mL⁻¹ chloramphenicol plus 30 μg mL⁻¹ kanamycin (LB cm + kan).

Table 1.

Bacterial strains, phages and plasmids.

Strains, phages or plasmids	Genotype or relevant characteristic	Reference/source
E. coli strains
WA2379	leu− met− lac− ${\mathrm{r}}_{\mathrm{A}}^{-}{\mathrm{m}}_{\mathrm{A}}^{-}$	(Arber and Wauters-Willems, 1970)/The Coli Genetic Stock Center, Yale University (CGSC)
W3110	F− l− rpoS(Am) rph-1 Inv(rrnD-rrnE) ${\mathrm{r}}_{\mathrm{K}}^{+}{\mathrm{m}}_{\mathrm{K}}^{+}$	(Iida et al., 1987; Hayashi et al., 2006)/CGSC
WA921	thr− leu− met− lac− ${\mathrm{r}}_{\mathrm{K}}^{-}{\mathrm{m}}_{\mathrm{K}}^{-}$	(Wood, 1966; Arber and Wauters-Willems, 1970)/CGSC
WA960	${\mathit{thr}}_B^{+}$ leu− met− lac− ${\mathrm{r}}_{\mathrm{B}}^{+}{\mathrm{m}}_{\mathrm{B}}^{+}$	(Wood, 1966; Arber and Wauters-Willems, 1970)/CGSC
BW25113(pKD46)	F− l− rpoS(Am) rph-1 rrnB3 ΔlacZ4787 hsdR514 Δ(araBAD)567 Δ(rhaBAD)568	(Baba et al., 2006)/CGSC
BW25141(pKD4)	lacIq rrnBT14 ΔlacZWJ16 ΔphoBR580 hsdR514 ΔaraBADAH33 ΔrhaBADLD78 galU95 endABT333 uidA(ΔMluI)::pir+ recA1	(Datsenko and Wanner, 2000)/CGSC
MG1655	lacIq1tonA::Tn10	(Park et al., 2006)/R. Young
Phages
P1CMclr100	clr100 ts; Chloramphenicol resistant	(Rosner, 1972)/CGSC
P1ΔdarB	in-frame deletion of darB in P1CMclr100	This study
P1Δulx	in-frame deletion of ulx in P1CMclr100	This study
P1ΔdarA	in-frame deletion of darA in P1CMclr100	This study
P1Δhdf	in-frame deletion of hdf in P1CMclr100	This study
P1ΔddrA	in-frame deletion of ddrA in P1CMclr100	This study
P1ΔddrB	in-frame deletion of ddrB in P1CMclr100	This study
P1Δhxr	in-frame deletion of hxr in P1CMclr100	This study
Plasmids
pBAD24	empty cloning vector	(Guzman et al., 1995)/R. Young
pdarB	P1 darB cloned into pBAD24	This study
pulx	P1 ulx cloned into pBAD24	This study
pdarA	P1 darA cloned into pBAD24	This study
phdfam_darA_ddrAam	P1 hdf_darA_ddrAgene fragment cloned into pBAD24 and amber mutations introduced to 9th codon in hdf and 28th codon in ddrA	This study
phdf	P1 hdf cloned into pBAD24	This study
pddrA	P1 ddrA cloned into pBAD24	This study
pddrB	P1 ddrB cloned into pBAD24	This study
pulx-CFLAG	P1 ulx with C-terminal FLAG-tag fusion cloned into pBAD24	This study

Production of phage lysates

Phage lysates were produced by thermal induction of P1 lysogens. Lysogenic strains were grown at 30°C in LB cm to OD₅₅₀ 0.5–0.6. P1 was thermally induced by shifting the culture to 42°C in a shaking water bath for 60–75 min (Iida and Arber, 1977). The crude lysate was harvested when the OD₅₅₀ fell to ~0.2, by centrifugation of the culture at 10,000 × g, 10 min, 4°C and sterilized by passage through a 0.22 μm filter (Millipore).

Lysogenization

Lysogens of P1 and its mutants were produced as previously described (Rosner, 1972) with minor modifications. A fresh overnight culture of the desired E. coli lysogenization host was supplemented with 5 mM CaCl₂, and 100 μL of the culture was mixed with 100 μL of an undiluted phage lysate (~10⁸–10⁹ PFU mL⁻¹) and incubated at RT for ~20 min. The phage-host mixture was then plated to LB cm and a CM^R colony was selected after overnight incubation at 30°C and purified by restreaking. The same procedure was used for producing lysogens of P1 deletion mutants, with the exception that plating was conducted on LB cm +kan.

Generation of PCR fragments for single-gene deletions

All primers used in PCR were ordered from Integrated DNA Technologies. The plasmid pKD4 has an FRT-flanked kan gene which was used at the source of the kanamycin resistance cassette used for gene deletions (Datsenko and Wanner, 2000). To avoid restriction by the Type III R-M system resident in the P1 chromosome, an EcoP1 site normally present in the kan gene was removed by following protocol from QuikChange site-directed mutagenesis kit (Agilent Technologies) changing G608 to A in a silent mutation. To generate FRT-flanked kan insert for gene deletion, the 3’-ends of the forward and reverse primers were designed to match the priming sites gtgtaggctggagctgcttc and atgggaattagccatggtcc of pKD4 (Datsenko and Wanner, 2000). The 5’ ends of both forward and reverse primers were synthesized to include 50 nt homology to the P1 genome up- and downstream of the targeted gene. The flanking regions were chosen to retain the first and last several codons of the targeted genes to avoid polar effects (Datsenko and Wanner, 2000). PCR reactions were performed using Phusion Hi-Fidelity PCR Master Mix (New England Biolabs), following the manufacturer’s recommended protocol. The PCR products were gel purified using the QIAquick Gel Extraction Kit (Qiagen).

Generation of single-gene knockout mutants

The phage lambda Red recombinase mediated homologous recombination method was used to generate isogenic single-gene knock-out P1 mutants (Datsenko and Wanner, 2000). P1 was lysogenized into BW25113(pKD46) and a colony resistant to both chloramphenicol and ampicillin was selected. BW25113(pKD46) lysogenized with P1 was grown in LB amended with 100 μg mL⁻¹ ampicillin and 10 μg mL⁻¹ chloramphenicol to OD₅₅₀ 0.1, 1 mM L-arabinose was added to induce Red proteins, and the culture was grown to OD₅₅₀ 0.6. The cells were harvested and made electrocompetent as described previously (Datsenko and Wanner, 2000). Each electroporation reaction contained 300 ng of gel-purified DNA and 100 uL competent cells in a 0.1 cm cuvette, and was transformed in a Bio-Rad Micro-Pulser^™ electroporator following the manufacturer’s protocol. Cells were recovered in 1 mL SOC for 2 h at 30°C and plated to LB cm +kan at 30°C overnight. CM^R and Kan^R colonies were selected and mutations were verified by PCR amplifying a DNA region spanning the kan insertions and sequencing across the insertion junctions (Baba et al., 2006). These lysogens were then induced as above and the mutant P1 phages were lysogenized into E. coli strain MG1655 as described above for maintenance of the strains.

Complementation of phage mutants

Phage knockouts were complemented in trans by induction of lysogens containing both the mutant prophages and vectors expressing the deleted genes. Complementing genes were amplified by PCR from a P1 DNA template (Supporting Information Table S1) and cloned into pBAD24 at its XbaI and HindIII sites using standard molecular biology techniques (Guzman et al., 1995). Ligation products were transformed into competent E. coli 5-alpha cells (New England Biolabs) and selected by plating on LB agar amended with 100 μg mL⁻¹ ampicillin. The plasmids were extracted as described above and were verified by sequencing before transformation into the r⁻m⁻ strain WA921. Complemented phage were prepared by thermal induction of mutant phage from WA921 lysogens containing the corresponding complementing vector expressing the deleted gene in trans (Table 1). 1 mM L-arabinose (Sigma-Aldrich) was added to the culture at the time of temperature shift to 42°C to induce protein expression from the complementing plasmids. The phage lysate was used to determine efficiency of plating (EOP) on restricting and non-restricting hosts as described below.

Efficiency of plating (EOP) assay

All phages used in EOP assays were induced from modification-deficient WA921 lysogens. Assays were conducted as previously described with few modifications (Arber and Dussoix, 1962; Mise and Arber, 1976; Iida et al., 1987). Host cells, grown to OD₅₅₀ 0.4–0.5 in tryptone broth (10 g L⁻¹ Bacto tryptone, 5 g L⁻¹ NaCl), were incubated with 10 mM CaCl₂ for 30 min at RT. Phage were adsorbed to 300 μL of host cells for 20 minutes at RT. The cells and phage were then plated using the soft agar overlay method using LB (Lennox) plates (10 g L⁻¹ Bacto tryptone, 5 g L⁻¹ NaCl, 5 g L⁻¹ Bacto yeast extract, g L⁻¹ Bacto agar) containing 2.5 mM CaCl₂ as the bottom plates with 4 mL lawns of tryptone top agar (10 g L⁻¹ Bacto tryptone, 5 g L⁻¹ NaCl, 7 g L⁻¹ Bacto agar). Plaques were counted after overnight incubation at 42°C. EOP was calculated as the ratio of plaques appearing on the lawn of the restricting strain to the number of plaques on WA921 lawns. The EOP of each phage mutant was normalized to the EOP of parental P1 on the same plating strain. In complementation assays, EOPs were normalized to P1 induced from WA921(pBAD24). All experiments were replicated three times.

Purification of virions by CsCl isopycnic centrifugation

P1 or P1 mutant was induced in 1L LB as described above. The crude lysate was centrifuged in JA10 rotor for 15 min at 17,000 × g and the supernatant was filter sterilized. The lysate was concentrated by 24 h centrifugation in JA10 rotor at 14,000 × g at 4°C. The pellet was soaked in SM buffer (50 mM Tris-HCl pH 7.5, 100 mM NaCl, 8 mM MgSO₄) at 4°C and extracted after 48 h. 10 μg mL⁻¹ DNAse (Sigma) was added to the concentrated phage and left at RT for 30 min. Phages were then purified by equilibrium centrifugation in cesium chloride as previously described (Boulanger, 2009). The phage bands were extracted with 18 gauge needles and dialyzed against SM buffer in Slide-A-Lyzer 3500 MWCO dialysis cassettes (Thermo Scientific). Small volumes of pre-dialysis samples were saved and used to measure refractive index in an Abbe Refractometer. Refractive indices were converted to density (g cc⁻¹) by comparing to standard CsCl density-refractive index correlation table.

SDS-PAGE analysis

For SDS-PAGE analysis of phage proteins, samples were prepared as described previously, with slight modification (Boulanger, 2009). Approximately 2 × 10¹⁰ PFU of CsCl purified P1 were loaded per lane. For antirestriction P1 mutants, protein loading was normalized to equal amounts of tail sheath protein (57 kDa). Phages were heated in boiling water for 10 min to release DNA from the capsid and samples were then treated with DNase I at 37°C for 2 h. The samples were denatured by heating in boiling water for 5 min in Laemmli sample buffer and loaded on a 4–20% Tris-glycine SDS-PAGE gel (Life Technologies) (Laemmli, 1970). PageRuler Unstained Broad Range Protein Ladder (Thermo Scientific) was used as molecular mass standard. The gel was stained with SYPRO Ruby (Thermo Scientific), following the manufacturer’s recommended protocol for maximum sensitivity. The gel was imaged with Fotodyne gel imager.

Proteomic analysis

Proteins associated with the virion were identified by mass spectrometry (MS) of the protein bands excised from SDS-PAGE gel as described previously (Gill et al., 2011). Briefly, ~10¹¹ PFU of phage P1 was prepared and loaded into 4–20% Tris-glycine SDS-PAGE as described above. Coomassie-stained bands were excised, subjected to reduction with dithiothreitol and alkylation with iodoacetamide. The samples were treated with ~0.4 μg trypsin (Thermo Scientific) and digested peptides were concentrated and desalted using C₁₈ ZipTips (Millipore). The samples were spotted manually onto a matrix-assisted desorption ionization (MALDI) target (Genomic Solutions) using α-cyano 4-hydroxycinnamic acid. MALDI-MS and MS/MS analysis was conducted against the NCBI-nr database as described previously (Gill et al., 2011). P1 DarB, DarA and Hdf were identified by this method.

The virion-associated DdrB was identified by liquid chromatography-tandem mass spectrometry (LC-MS/MS) in collaboration with the Protein Chemistry Lab at Texas A&M University. Four lanes each of ~2 × 10¹⁰ PFU P1 and P1ΔddrB were loaded into a 4–20% Tris-glycine SDS-PAGE and stained with SyproRuby as described above. The gel region missing protein band in P1ΔddrB lanes and corresponding protein band from P1 lanes were excised. The samples were processed for LC-MS/MS as described before (Shevchenko et al., 2006). LC-MS/MS was performed on a LTQ Orbitrap Velos/ETD mass spectrometer (Thermo Fisher) equipped with a NanoLC 2-D HPLC system (Eksigent). The results were analyzed using Mascot (Matrix Science) and X!Tandem (The GPM). Scaffold (Proteome Software) was used to validate MS/MS based peptide and protein identifications. The SDS-PAGE band was annotated based on its presence in P1 and absence in P1ΔddrB lanes.

Transmission electron microscopy

Phages were stained with 2% uranyl acetate and imaged in a JEOL 1200 EX transmission microscope under 100 kV accelerating voltage as previously described (Valentine et al., 1968; Gill et al., 2011). Side-to-side head diameters perpendicular to the axis of the tail were measured electronically using ImageJ (Schneider et al., 2012) and converted to nm against images of a carbon grating replica of known dimensions (Ted Pella, cat# 607).