Role of the Gp16 lytic transglycosylase motif in bacteriophage T7 virions at the initiation of infection



The predicted catalytic glutamate residue for transglycosylase activity of bacteriophage T7 gp16 is not essential for phage growth, but is shown to be beneficial during infection of Escherichia coli cells grown to high cell density, conditions in which murein is more highly cross-linked. In the absence of the putative transglycosylase, internalization of the phage genome is significantly delayed during infection. The lytic transglycosylase motif of gp16 is essential for phage growth at temperatures below 20°C, indicating that these growth conditions also lead to increased cross-linking of peptidoglycan. Overexpression of sltY, E. coli soluble lytic transglycosylase, partially complements the defect in infection of mutant phage particles, allowing them to infect at higher efficiencies. Conversely, an sltY deletion increases the latent period of wild-type phage.


The shape of the eubacterial cell is maintained by the murein sacculus. A neutron small-angle scattering study on the sacculus isolated from exponentially growing Escherichia coli cells showed that 75–80% of the sacculus is a single layer, the remainder being triple-layered (Labischinski et al., 1991). The sacculus completely surrounds the bacterial cell and is the most massive single molecule of the bacterium. Nevertheless, the sacculus is a dynamic structure in which covalent bonds are broken and created during cell growth; about 50% of the murein is broken down and recycled per cell generation (Goodell, 1985). In E. coli, turnover of murein is catalysed by a variety of enzymes, including several distinct lytic transglycosylases (Höltje, 1998). Although most of the E. coli lytic transglycosylases are membrane bound, SltY is a soluble, periplasmic enzyme that is tightly bound to the murein layer in vivo (Walderich and Höltje, 1991).

Peptidoglycan is a barrier to the transport of macromolecules (for a review, see Dijkstra and Keck, 1996a). Specialized pores may be made to accommodate the export of larger molecules by the cell. For example, it has been shown recently that a peptidoglycan-hydrolysing activity is essential for flagellar rod formation (Nambu et al., 1999). In a single-layer peptidoglycan network containing many glycan strands, N-acetyl muramic acid is cross-linked to a neighbouring strand in an alternating fashion (Koch and Woeste, 1992). The network of strands and cross-links prevents the simple diffusion of globular proteins of > 50 kDa (Demchick and Koch, 1996). However, as E. coli undergoes the transition from exponential growth into stationary phase, more cross-linking occurs (Pisabarro et al., 1985). Thus, one would expect that, as cells enter stationary phase, diffusion of even smaller macromolecules across the murein layer could become difficult.

The T7 particle contains only a short, stubby tail, one that is not long enough to span the outer envelope of E. coli (Steven and Trus, 1986). The particle also contains a morphologically distinct structure, the internal core, within its head (Serwer, 1976; Serwer et al., 1997). After infection, the T7 virion ejects the three internal core proteins into the cell (Molineux, 1999). In part, these proteins can be considered an extensible tail, forming a channel distal to the mature virion tail across the cell envelope, to allow translocation of DNA into the cell cytoplasm. An electron micrograph, perhaps showing the core proteins as an extension of the T7 tail, was presented by Serwer (1978). Two of the ejected proteins are large: each of the 12 copies of gp15 in the virion has a mass of 84 kDa, and the three copies of gp16 are each 144 kDa. In principle, these proteins may have difficulty in passing the murein layer, especially in stationary phase cells.

The N-terminal sequences of T7 gp16 exhibit homology with C-terminal sequences of the 70 kDa E. coli soluble lytic transglycosylase SltY (Engel et al., 1991), and it has been suggested that T7 gp16 may have a role in cell lysis at the end of the lytic cycle. However, a role in cell lysis for a putative transglycosylase activity of gp16 would be redundant, because T7 also codes for an N-acetylmuramyl-l-alanine amidase (gp3.5, often called T7 lysozyme) (Inouye et al., 1973; Kleppe et al., 1977). One of the many functions of this protein is known to be in cell lysis at the end of the phage developmental cycle (Silberstein et al., 1975; Cheng et al., 1994). Furthermore, 16am mutants lyse normally after infection of Su hosts (Studier, 1969), and the fact that gp16 is ejected into the cell at the initiation of infection suggests that, if an SltY-like activity is associated with the protein, it is more likely manifest at the beginning of the infection.

The crystal structure of E. coli SltY has been determined (Thunnissen et al., 1994; 1995a; van Asselt et al., 1999). The active-site amino acid is a glutamate at position 478; changing this residue to glutamine abolishes enzyme activity. There is little overall sequence similarity between SltY and the lysozyme family of proteins, but amino acids important in forming the active site are conserved. Conserved motifs in several prokaryotic transglycosylases have been predicted (Koonin and Rudd, 1994; Thunnissen et al., 1995b; Dijkstra and Keck, 1996a). These motifs, including the active-site glutamate, are also found in T7 gp16 (Fig. 1).

Figure 1.

Alignment of residues 2–143 of T7 gp16 with residues 443–592 of E. coli SltY. Identical residues are shaded; conserved motifs important for the architecture of the active site (Dijkstra and Keck, 1996a) are boxed; the catalytic glutamate of SltY, and its counterpart in gp16, are in bold face. (●) Hydrophobic residues of SltY that surround the catalytic glutamate; X denotes residues that, in SltY, bind the peptide moiety of peptidoglycan (van Asselt et al., 1999).

In the work presented here, the glutamate at position 37 of T7 gp16, equivalent to glutamate 478 of SltY, has been substituted with a variety of amino acids, and the course of infection by mutant T7 particles determined. We show that this glutamate is not essential when the cells infected are growing exponentially at 30°C. However, when mutant particles infect cells at a higher density at 30°C or infect exponentially growing cells at 20°C, their latent periods, relative to that of wild type, are extended. Furthermore, we show that the additional time required by mutant virions to complete their infection under these suboptimal conditions is totally accounted for by the increase in time necessary to initiate transcription-mediated entry of the T7 genome into the cell.


Lysates containing mutant T7 particles were prepared by infecting T7 16E37am into isogenic E. coli strains containing 13 different amber suppressor tRNAs. Chromosomal alleles of supD (serine), supE (glutamine), supF (tyrosine) and supP (leucine) were used; other suppressors used are synthetic and are carried on pBR322 derivatives (McClain and Foss, 1988; Kleina et al., 1990; Normanly et al., 1990). For clarity, products of the suppressed gene 16am will be described using the single letter code for amino acids: the mutant protein found in supD strains is thus gp16–37S, that in supE strains is gp16–37Q, etc. In principle, suppressing 16E37am with a glutamate tRNA suppressor should yield gp16–37E, i.e. wild type. However, the suppressor tRNA inserts glutamate only 60–80% of the time, glutamine being the major alternative residue (Kleina et al., 1990; Normanly et al., 1990), and particles of 16amE37 containing gp16–37E are thus referred to as pseudowild.

The catalytic residue of the putative transglycosylase activity is important but not essential for phage growth

Except for virions containing gp16–37C and gp16–37P, latent periods and titres of lysates made at 30°C were normal, and specific infectivities of mutant virions (plaque-forming units, determined using the glutamate suppressor-containing strain, per A260) of purified preparations were all comparable. Growth of 16E37am on strains containing the cysteine or proline amber suppressors consistently resulted in lysates with titres and specific infectivities about 1% of those obtained using the glutamate suppressor. Because both lysate titres and specific infectivities of virions containing gp16–37C and gp16–37P are affected similarly, we conclude that morphogenesis is not grossly affected by the missense substitutions. However, at least 99% of the mutant particles are unable to initiate an infection. Virions containing gp16–37C and gp16–37P were not studied further.

No effects on the latent period were detected when growing stocks at 30°C of phages containing various missense substitutions in gp16. However, because plaque formation requires several rounds of phage growth, any small effects on the latent period as a result of defective phage particles are magnified. T7 16E37am was plated at various temperatures on isogenic strains containing the different amber suppressor tRNAs, and the efficiencies of plating were determined relative to that of the pseudowild phage at 30°C (Fig. 2). At 17°C or below, only the pseudowild phage made plaques at near-normal efficiency. Between 17°C and 22°C, the efficiency of plating of phages containing gp16–37Q, gp16–37S, gp16–37L, gp16–37Y, gp16–37H and gp16–37G rose from < 0.02 to about 0.4; insertion of A, F, K or R at the 16E37am codon was somewhat less effective. The expected catalytic residue for the putative transglycosylase activity of gp16 is therefore not absolutely essential, but is important during infection at low temperature.

Figure 2.

Efficiency of plating of T7 16E37am using different amber suppressors at various temperatures. Efficiencies are relative to 30°C on the glutamate (E) suppressor-containing strain.

The above observations suggested that we could distinguish between phages containing different substitutions at the catalytic site of the putative transglycosylase activity of gp16 by studying the course of infection at low temperature. In addition, it seemed possible that infecting cells at high cell density would prove informative. As an initial characterization for a single round of phage growth, patterns of cell lysis were determined after infection of cells at high multiplicity with stocks of the different mutant T7 particles. A clear separation in lysis times was obtained after infection of cells grown at 20°C to a density of 2.2 × 108 ml−1. Virions containing gp16–37A, gp16–37K and gp16–37R gave a 30–40% increase in latent period, whereas that for virions containing gp16–37F was more than doubled, increasing from about 100 min to 220 min (Fig. 3A). At 30°C, infecting cells at a density of 7.5 × 108 ml−1, i.e. late in the exponential phase of growth, also provided a phenotype associated with mutant virions. Substitutions of G, H, L, Q, S and Y at gp16-E37 provided virions that were somewhat less effective than wild type in causing cell lysis, but virions containing A, F, K and R were severely affected. Latent periods of these mutant particles were about threefold longer than that of wild type (Fig. 3B). These patterns of cell lysis were obtained using cells that lack an amber suppressor, in which the mutant gp16 is not synthesized. We therefore conclude that virions containing different missense substitutions at E37 of gp16 infect cells with varying kinetic efficiencies.

Figure 3.

Lysis of IJ1142 by mutant particles of T7 16E37am. The multiplicity of infection was about 5.

A. 20°C, initial cell density = 2.2 × 108 ml−1.

B. 30°C, initial cell density = 7.5 × 108 ml−1.

Symbols: gp16–37E (●), gp16–37A (Δ), gp16–37F (◂), gp16–37G (▴), gp16–37H (◊), gp16–37K (▸), gp16–37L (□), gp16–37Q (○), gp16–37R (▾), gp16–37S (▪), gp16–37Y (◆).

E. coli soluble lytic transglycosylase partially complements a defective T7 gp16

The putative transglycosylase of T7 gp16 is not essential for phage development in cells growing exponentially at 30°C, but it is possible that the lack of transglycosylase activity in a T7 particle may be complemented by a cellular function. Several lytic transglycosylases have been described in E. coli, and a triply mutant strain deleted for two of the membrane-bound lytic transglycosylases, mltA and mltB, and for soluble lytic transglycosylase sltY has no significant growth defect (Lommatzsch et al., 1997). We tested deletion mutants of sltY, mltA and mltB, both alone and in combination, as hosts for T7. At 20°C, the latent period of pseudowild virions in exponentially growing ΔsltY or sltY+ cells was indistinguishable, whereas latent periods of phages containing A, F, K and R substitutions at gp16-E37 were extended by about 80 min in the deletion mutant (Fig. 4A). At 30°C, infecting pseudowild phage (or T7+, not shown) into ΔsltY cells in the late exponential phase of growth resulted in a 5–10 min delay in lysis of cells, and a 30–40 min delay after infection by virions containing the A, F, K and R substitutions at gp16-E37 (Fig. 4B). Deletions of mltA or mltB, either alone or in combination, had no effect on lysis times; lysis curves, using the triple deletion mutant as host, were indistinguishable (data not shown) from those obtained using ΔsltY alone.

Figure 4.

Lysis of bacterial strains altered in expression of various lytic transglycosylases by mutant particles of T7 16E37am. The multiplicity of infection was about 5.

A and B. IJ1142 and the Δslt strain IJ1735 at (A) 20°C, initial cell density = 2.2 × 108 ml−1; (B) 30°C, initial cell density = 7.5 × 108 ml−1.

C. IJ1142 and IJ1142 containing the sltY plasmid pJL-JS. Symbols: gp16–37E (●, ○), gp16–37A (▪, □), gp16–37F (▴, ▵), gp16–37K (◆, ◊), gp16–37R (▾, ▿). Closed symbols represent infections of IJ1142, open symbols represent infections of the deletion mutant or plasmid-containing strain.

We also tested for effects on phage growth by constitutive overexpression of sltY. At 20°C, there was no effect on lysis times as a result of overexpression of sltY (not shown). Even though overexpressing sltY had no effect on the kinetics of lysis by pseudowild phage, when cells were infected in the late exponential phase of growth at 30°C, the lysis defect of virions containing the A, F, K and R substitutions at gp16-E37 was partially compensated (Fig. 4D). Overexpression of mltA (overexpression of mltB causes cell lysis; Ehlert et al., 1995) had no effect on the kinetics of lysis by pseudowild or mutant phage particles (not shown).

Together, these results lend strong support to the idea that T7 gp16 contains a transglycosylase-like activity that can be important for phage infection under certain conditions. This activity is related by both sequence homology and function to the E. coli soluble lytic transglycosylase SltY, whose function is in peptidoglycan turnover.

The putative transglycosylase activity of T7 gp16 facilitates genome entry

Gene 16 has no known role in mediating cell lysis at the end of an infection. The extended latent periods after infection by mutant phage particles therefore probably reflect the initial stages of the infection. In the T7 life cycle, several sequential steps have been defined that are necessary for internalization of the genome (for a review, see Molineux, 1999). After adsorption, the three internal core proteins of the phage head are ejected into the cell, and a channel for DNA transport is formed across the cell envelope. A transcription-independent process then translocates ≈ 850 bp of T7 DNA into the cell (García and Molineux, 1996), allowing E. coli RNA polymerase to bind to early promoters and initiate the transcription-dependent phases of genome internalization (García and Molineux, 1995).

To determine whether substitutions at T7 gp16-E37 affected genome internalization, we measured the kinetics of phage DNA translocation catalysed by E. coli RNA polymerase (García and Molineux, 1995) after infections by mutant virions. Representative Southern blots from genome entry experiments are shown in Fig. 5. The rates of genome entry, estimated from all blots, are presented in Fig. 6 and Table 1. Under the same conditions of infection, the rates of transcription-catalysed genome entry are not significantly different whether the infecting phage particle is pseudowild or mutant. Once E. coli RNA polymerase has begun to internalize the genome, a mutant gp16 has no effect.

Figure 5.

Time course of methylation of Dam12,16E37am DNA as it enters the infected cell from mutant virions. A schematic DpnI–Sau3AI map of the genome is shown. The 40 kb band represents phage genomes that have not begun to enter the cell. Numbers above each lane represent the time in min after infection when DNA was isolated. DNA in lanes labelled S was digested with Sau3AI as a control for the amount of T7 DNA in each lane. At 30°C and at a cell density of 7.5 × 108 ml−1, IJ1133(pTP166) was infected at a multiplicity of 0.4 by virions containing gp16–37E (A), gp16–37H (B), gp16–37F (C) or gp16–37R (D).

Figure 6.

Rates of genome entry at 30°C estimated from Fig. 5 and from comparable experiments not shown. (●) gp16–37E; (▾) gp16–37R; (▪) gp16–37F; (◊) gp16–37H; (▵) gp16–37A; (□) gp16–37K.

Table 1. Kinetic data derived from genome entry experiments.
Rate of RNA polymerase-
catalysed genome
entry (bp s−1)
Estimated time for
elongation complex
formation (min)
30°C; 2 × 108 cells ml−116E37E343
30°C; 7.5 × 108 cells ml−116E37E214
20°C; 2 × 108 cells ml−116E37E69

Extrapolating the kinetic data of Fig. 6 to y = 0.5 (0.5 kb of the genome inside the cell) yields the times when the RNA polymerase elongation complex is predicted to form at the T7 A1 promoter, the first of the three E. coli promoters on the phage genome to enter the cell by the transcription-independent process (García and Molineux, 1996). Table 1 provides values for all the genome entry experiments we performed for this study using mutant particles of T7 16E37am. At 30°C, when cells are at low cell density and are growing exponentially, there is no significant difference in the time of elongation complex formation at the A1 promoter after infection with different mutant particles. However, when cells are at higher cell density, or are growing at low temperature, the four mutant virions that exhibited the longest latent periods show a substantial delay in the initiation of transcription-dependent genome entry. Relative to pseudowild particles, the increased time necessary before transcription-dependent genome entry is initiated with phage particles containing gp16–37A, gp16–37F, gp16–37K or gp16–37R completely accounts for their increased latent periods. The defects of the various capsid-carried mutant gene 16 products are therefore manifest before E. coli RNA polymerase forms an elongation complex on T7 DNA. The mutant gp16 most probably delays formation of the channel for DNA transport across the cell envelope or the transcription-independent translocation of the initial 850 bp of the phage genome.


We have shown that mutant T7 particles, altered in the expected catalytic residue of a putative transglycosylase associated with the internal core protein gp16, are defective in genome internalization under some conditions of infection. The defect is expressed before the formation of an E. coli RNA polymerase elongation complex at one or more of the three A promoters that lie on the leading 850 bp of the genome. In order to internalize the A promoters, a channel must be created across the cell envelope, and the leading 850 bp of the genome translocated into the cell by a transcription-independent mechanism. The defect in genome internalization is manifest only when the cells to be infected are growing under suboptimal conditions for phage development, including the late exponential phase of cell growth, in which the murein layer is known to be more highly cross-linked (Pisabarro et al., 1985), suggesting that the putative transglycosylase function of wild-type gp16 is indeed enzymatically active. Our observation that the expected catalytic glutamate for any transglycosylase activity of gp16 is also important for an efficient infection of cells growing exponentially at 20°C suggests that the latter also represent conditions in which peptidoglycan is more highly cross-linked than at higher temperatures.

In the phage particle, three copies of the 144 kDa gp16 are found inside the capsid as part of an internal structure called the core, which also contains 12 copies of the 84 kDa gp15 and 18 copies of the 21 kDa gp14 (Steven and Trus, 1986). After adsorption, these three proteins are ejected from the infecting particle into the cell (Molineux, 1999). The structures of these proteins are unknown, but the sizes of gp15 and gp16, in particular, suggest that, in addition to disaggregating from the core structure, these proteins must unfold in order to exit the capsid and enter the phage tail. The diameter of the central hole through the head–tail connector has an average diameter of only 3.7 nm, with a constriction only 2.2 nm wide (Donate et al., 1988; Valpuesta et al., 1992; Kocsis et al., 1995). In the electron microscope, the inner diameter of the tail appears collinear with the hole through the head–tail connector (Steven and Trus, 1986), but nothing is known about the dimensions of the channel across the outer membrane. This is probably formed by gp14, which should therefore be the first of the core proteins to exit the phage head (C.-Y. Chang, P. Kemp, L. R. García and I. J. Molineux, unpublished data). We suggest that gp16 is the second core protein to exit the capsid and that its N-terminal region interacts with the murein layer. At 30°C and when cells are growing exponentially, the predominately single-layer peptidoglycan may be sufficiently porous to allow the unfolded forms of gp16 and gp15 to pass through the mesh of glycan strands and peptide cross-links (Fig. 7A). However, if the peptidoglycan becomes multilayered as cross-links between layers are incorporated, the mesh size of the murein is effectively decreased. Under these conditions, the porosity of the murein would be reduced, and a transglycosylase or related activity of gp16 may become important for a kinetically efficient infection (Fig. 7B).

Figure 7.

Schematic indicating how the three molecules of gp16 ejected into the cell from the virion may insert between glycan strands.

A. Insertion of gp16 into a single peptidoglycan layer.

B. Insertion of gp16 into a multilayered (or more highly cross-linked) cell wall requires transglycosylase or related activity. Thin-line glycan helices with dashes for peptide cross-links are in a different plane from the thick lines. An anhydrosugar formed by the putative transglycosylase activity of gp16 is indicated by the black triangle; for clarity, only one anhydrosugar is shown, although all three gp16 molecules are likely to act similarly.

About 50% of the murein sacculus is degraded, and its constituents are recycled in each division cycle (Goodell, 1985). Several lytic transglycosylases responsible for this turnover are located in the outer membrane and are lipoproteins, which, however, limits the range of their activity. As its name implies, SltY is a soluble, and thus diffusible, periplasmic enzyme that is perhaps one more likely to be subverted by phages to help them infect a cell. An sltY deletion increases the latent period of wild-type T7 and also of pseudowild virions containing gp16-E37, suggesting that a diffusible lytic transglycosylase activity facilitates T7 infection. Although SltY is not essential for infection, even for mutant phage particles containing a defective transglycosylase motif, overexpression of sltY partially complements the defect of mutant virions. These data strongly suggest that the transglycosylase motif of wild-type gp16 is an indicator of enzyme activity. However, it should be noted that, although the hydrophobic residues surrounding the catalytic glutamate of SltY (van Asselt et al., 1999) are well conserved in gp16, most of the muropeptide-binding residues are not (Fig. 1). Peptide cross-links are essential for both SltY and lysozyme activities (Kuroki et al., 1993; Romeis et al., 1993). Proof that gp16 is a transglycosylase will require characterization of the purified protein in vitro, but the lack of conservation of muropeptide-binding residues suggests that, even if T7 gp16 does possess hydrolytic activity against peptidoglycan, its mechanism of enzyme action may differ from both bacterial lytic transglycosylases and lysozymes.

T-even phage particles have long been known to contain a lysozyme activity in the baseplate protein gp5 (Kao and McClain, 1980; Nakagawa et al., 1985; Mosig et al., 1989). The lysozyme activity of gp5 is totally distinct from the better known T4 lysozyme, the product of gene e, which is necessary for cell lysis at the end of the infection. Baseplate formation, and thus complete particle assembly, requires gp5, and it has been suggested that its lysozyme activity locally digested peptidoglycan, allowing the T4 tail tube to penetrate the periplasmic space (Kao and McClain, 1980). Virions of two heat-sensitive gene 5 mutants have been shown to have heat-sensitive lysozyme activity, but the mutations lie outside the region homologous to the gpe lysozyme (Takeda et al., 1998), and formal proof that lysozyme activity per se is required for infection by T4 is still lacking. Nevertheless, the outer diameter of the T4 tail tube is about 9 nm (Moody and Makowski, 1981), greater than the estimated size of the holes through the murein sacculus (Dijkstra and Keck, 1996a), and some means of facilitating tail tube penetration through this potential barrier to infection seems probable.

A high multiplicity of infection (MOI) by T4 (and other T-even phages) effects cell lysis without a requirement for phage gene expression. Lysis-from-without requires gp5 (Nakagawa et al., 1985), suggesting that gp5 may be able to diffuse from its initial site of insertion in the cell and hydrolyse many glycan strands. However, gp16 also helps to catalyse the entry of the T7 genome into the cell (García and Molineux, 1996), and some part of the protein probably inserts into the cytoplasmic membrane (C.-Y. Chang, P. Kemp, L. R. García and I. J. Molineux, unpublished data). T7 gp16 is thus unlikely to be able to diffuse within the infected cell, a conclusion supported by the fact that high MOIs by T7 do not cause lysis-from-without.

The particles of several other phages have also been shown to contain lytic activities. The particles of the lipid-containing phages PM2 (Tsukagoshi et al., 1977) and ø6 (Mindich and Lehman, 1979) both contain lytic activities; in the case of ø6, it is known that enzyme activity is required both early and late in infection. The ø6 P5 protein is necessary for cell penetration by the phage; it digests the peptidoglycan in a step that is a prerequisite for nucleocapsid transport across the cytoplasmic membrane. The PRD1 phage particle actually contains two murein hydrolytic enzymes. Using a zymogram assay on electophoretically separated virion proteins, it has been shown that both PRD1 P7 and P15 are enzymatically active (P. Rydman and D. H. Bamford, 2000). The activity of P15 (a β-1,4-N-acetylmuramidase) is primarily to mediate lysis at the end of the infection cycle, but P7, which contains a lytic transglycosylase motif, facilitates entry of PRD1 DNA into the host cell. Like the putative activity of T7 gp16, muramidase activity of PRD1 P7 is not essential for infection.

The sequence homology between E. coli SltY and T7 gp16 (Engel et al., 1991) was part of our motivation for initiating the work described here. A more recent database search for lytic transglycosylase motifs has revealed that other phage virions, including coliphage P1, may harbour lytic activities (Lehnherr et al., 1998). Interestingly, the genomes of several phages that infect Gram-positive hosts contain putative transglycosylase activities. In principle, the multilayered Gram-positive cell wall should present a more formidable barrier to phage genome penetration of the cell. Localized hydrolysis of peptidoglycan by a phage particle could help to overcome that barrier and facilitate DNA entry into the cell. Murein hydrolytic activities associated with Gram-positive phage particles as they initiate infection do not appear to have been demonstrated, although the genome locations of the coding sequences for YomI of Bacillus subtilis phage SPβ, Rorf1608 of Lactobacillus phage øg1e, orf360 of L. delbrueckii phage LL-H and orf1626 and orf1560 of Streptococcus thermophilus phages sfi19 and sfi21, respectively, suggest that the proteins may be part of a virion. These and other phage proteins, including the defective B. subtilis prophages, PBSX XkdO and the skin element YqbO, all contain a prokaryotic lytic transglycosylase motif. The widespread occurrence of the motif in such a variety of phages will make it unsurprising if many virions are shown to contain murein hydrolytic activity, but whether such activity will prove essential for infection is less obvious.

Homologues of E. coli soluble lytic transglycosylase in several bacterial genera and conjugative or virulence plasmids have also been found by sequence analyses (Koonin and Rudd, 1994; Bayer et al., 1995; Dijkstra and Keck, 1996a,b; Mushegian et al., 1996; van Asselt et al., 1999). Some bacterial homologues have been shown to have enzymatic activity, and their role during cell growth may be similar to that of E. coli SltY. It has also been suggested that these hydrolytic enzymes may facilitate assembly of extracellular organelles and achieving competence for genetic transformation (Dijkstra and Keck, 1996a). Both processes involve transport of macromolecular assemblies across cell walls. In addition, conjugation involves transfer of DNA across the cell envelope, and the conjugal bridge must span the murein layers of both donor and recipient. The R1 gene 19 product and the Ti plasmid VirB1 protein are both needed for efficient plasmid transfer, and both contain sequences conserved in lytic transglycosylases (Bayer et al., 1995; Mushegian et al., 1996). Transglycosylase activity of these proteins has not yet been formally demonstrated and, like the putative activity of T7 gp16, the plasmid proteins are not absolutely required for conjugation. These plasmid data, together with our studies with T7, suggest that the murein sacculus is sufficiently dynamic to allow macromolecular assemblies to cross in either direction. Nevertheless, parasites whose dissemination requires traversing the cell wall may be more successful if, rather than relying on natural turnover of peptidoglycan, they carry their own lytic functions and remove this potential barrier to macromolecular transport.

Experimental procedures

Bacteria, plasmids and phages

Bacterial strains used are listed in Table 2. The ΔsltY, ΔmltA, ΔmltB strain MUF61 (Lommatzsch et al., 1997) and the lytic transglycosylase-expressing plasmids pJL-JS (sltY), pMAT (mltA) and pMK360 (mltB) were gifts from J. V. Höltje. The supP strain MDA2657 (a gift from E. J. Murgola, University of Texas M. D. Anderson Cancer Center, Houston, TX, USA), CR63 (supD) and C600 (supE) were used for transduction of the amber suppressors into IJ1142. The source of dam13-Tn9 was GM2163 (Marinus et al., 1984). IJ1142 was the host for the synthetic tRNA amber suppressor plasmids pFTOR1Δ26 (sup Arg) (McClain and Foss, 1988) and the pGFIB derivatives: sup Ala, sup Cys, sup Glu, sup Gly, sup His, sup Lys, sup Phe and sup Pro (Normanly et al., 1986; 1990; Kleina et al., 1990).

Table 2. E. coli K-12 strains.
NameGenotypeSource or reference
IJ1133ΔlacX74 thi-1Δ(mcrC-mrr)102::Tn10 García and Molineux (1995)
IJ1142ΔlacX74 galK150(Am) trp-49(Am) cysI53(Am) thi-1 Wang et al. (1999)
IJ1198IJ1142 supF Wang et al. (1999)
IJ1219IJ1142 supDx P1(CR63)
IJ1220IJ1142 supEx P1(C600)
IJ1221IJ1142 supPx P1(MDA2657)
IJ1734IJ1142 ΔmltA::Cmrx P1(MUF61)
IJ1735IJ1142 ΔsltY::Knrx P1(MUF61)
IJ1736IJ1142 Δ(srlR-mltB-recA)305::Tn10x P1(MUF61)
IJ1737IJ1142 ΔmltA::CmrΔsltY::Knrx P1(MUF61)
IJ1738IJ1142 Δ(srlR-mltB-recA)305::Tn10 ΔmltA::CmrΔsltY::Knrx P1(MUF61)
IJ1739IJ1142 ΔmltA::CmrΔ(srlR-mltB-recA)305::Tn10x P1(MUF61)
IJ1740IJ1142 Δ(srlR-mltB-recA)305::Tn10ΔsltY::Knrx P1(MUF61)
IJ1741IJ1198 dam-13::Tn9x P1(GM2163)
IJ1742IJ1220 dam-13::Tn9x P1(GM2163)

The T7 16E37am mutation was introduced by site-specific mutagenesis by the ung-dut procedure (Kunkel, 1985) into a subclone of the gene 16 plasmid pAR3496 (a gift from A. R. Rosenberg and F. W. Studier, Brookhaven National Laboratories, New York, USA). The mutation was recombined onto phage by growing T7 Δ16 and Dam12,Δ16 through IJ1142, which carried both the mutant gene 16 and the glutamate suppressor tRNA plasmid. Recombinants were selected on IJ1142 containing only the glutamate suppressor tRNA plasmid. DNA sequencing confirmed that the expected GAA → TAG was the only mutation present on the DNA subjected to mutagenesis. Titres of all T7 16E37am lysates, regardless of which suppressor strain was used to obtain them, were measured at 30°C using IJ1142 containing the glutamate suppressor tRNA plasmid.

Assay of T7 genome entry into bacteria

DNA entry experiments were performed essentially as described by García and Molineux (1995). Briefly, IJ1133 cells carrying a plasmid (pTP166) overexpressing dam (Marinus et al., 1984) were infected at an MOI of 0.4 using CsCl-purified stocks of unmethylated Dam12,16E37am, in which the amino acid inserted at the amber codon varied in different experiments. Cells were grown at the temperature used for infection and, 10 min before infection, 100 µg ml−1 chloramphenicol was added. At various times, portions of the culture were treated with a phenol–ethanol killing solution. Cells were harvested, proteins were digested with proteinase K, and DNA was isolated by successive phenol and phenol–chloroform extractions. DNA was cut with either Sau3AI or DpnI and subjected to Southern hybridizations using randomly primed T7 DNA as a labelled probe. Using imagequant (Molecular Dynamics) software, the intensity of each DpnI band was divided by the intensity of the corresponding Sau3AI band; the ratio corresponds to the fraction of that particular DNA fragment that, at a given time, has been methylated by Dam. Times of fragment entry are defined as the time when the ratio of DpnI to Sau3AI band intensities reaches about 50% of the maximum for that experiment. The maximum value varies in different experiments performed at different times because the fraction of phage particles that rapidly initiate infection in liquid varies. For unknown reasons, even CsCl-purified stocks of unmethylated Dam12,16E37am are not stable during storage at 4°C and rapidly lose titre.


We thank J. V. Höltje for several bacterial strains and plasmids, Promega Corporation for their gift of the glutamate suppressor tRNA plasmid, A. Rosenberg and F. W. Studier for pAR3496, and E. J. Murgola for the supP mutant strain. We gratefully acknowledge the expertise of P. Kemp in constructing the T7 16E37am phage. We thank G. Mosig and F. Arisaka for helping to explain T4 lore. This work was supported by a grant from the National Institutes of Health GM32095. M.M. was also supported in part by National Institutes of Health T32 GM087474.