Bacteriophage PRD1 DNA entry uses a viral membrane-associated transglycosylase activity



Amino acid sequence analyses have indicated that the amino-terminal part of bacteriophage PRD1 structural protein P7 carries a conserved transglycosylase domain. We analysed wild-type PRD1 and different mutant particles in zymograms and found a glycolytic activity that was associated with protein P7. This is the first time a putative bacteriophage or plasmid lytic transglycosylase has been shown to have an enzymatic activity. In the absence of protein P7, the phage DNA replication and host cell lysis were delayed. Gene VII of PRD1 is known to encode proteins P7 and P14. In this investigation, the open reading frame coding for P14 was mapped to the 3′ end of gene VII. Proteins P7 and P14 probably form a heteromultimeric complex, which is located at the particle vertices and is involved in the early steps of the PRD1 life cycle


Peptidoglycan consists of glycan strands formed of alternating N-acetylglucosamine and N-acetylmuramic acid residues cross-linked by short oligopeptides. It forms a covalently closed network-like structure around the bacterial cell. The sacculus protects bacteria from harmful environmental influences, provides mechanical strength and determines the typical shape of the cell. Despite the peptidoglycan barrier, macromolecular complexes, such as virus genomes, flagella, pili and virulence factors, are capable of crossing the cell envelope (for reviews of the sacculus structure and its role in transenvelope transport, see Dijkstra and Keck, 1996a; Höltje, 1998).

Peptidoglycan-modifying enzymes are ubiquitous in bacteria as well as in other organisms (Jollès, 1996). Lytic transglycosylases are muramidases that catalyse the cleavage of the β-1,4-glycosidic bond between the N-acetylmuramic acid and N-acetylglucosamine residues. This results in 1,6-anhydromuramic acid carrying products that conserve the energy of the cleaved bond in the form of a ring structure (Höltje et al., 1975; Taylor et al., 1975). Six different lytic transglycosylases have been identified in Escherichia coli, and their enzymatic activities have been characterized (Slt70, MltA-D and EmtA; Höltje et al., 1975; Ursinus and Höltje, 1994; Ehlert et al., 1995; Dijkstra and Keck, 1996b; Kraft et al., 1998; van Asselt et al., 1999a). It has been proposed that they function in the enlargement and division of the sacculus, as well as in the export of bulky compounds (Dijkstra and Keck, 1996a; Höltje, 1998). Koonin and Rudd (1994) defined a conserved sequence motif characteristic of transglycosylases. Sequence analysis has led to the identification of genes encoding putative transglycosylases, not only in Gram-positive and Gram-negative bacteria, but also in bacteriophages and plasmids (Koonin and Rudd, 1994; Bayer et al., 1995; Mushegian et al., 1996). The actual lytic activity has not been verified in the case of putative bacteriophage or plasmid enzymes, although it has been shown that they are involved in the transport of plasmid DNA, phage genomes and virulence factors across the cell envelope (Bayer et al., 1995; García and Molineux, 1996; Mushegian et al., 1996). Moak and Molineux (2000) have recently demonstrated that the putative transglycosylase of bacteriophage T7 facilitates infection of host cells grown to high cell density and that the predicted catalytic glutamate is needed for the beneficial effect.

PRD1 is a broad-host-range double-stranded DNA bacteriophage that infects a variety of Gram-negative hosts harbouring an N-, P- or W-type conjugative plasmid (Olsen et al., 1974). The plasmid encodes the phage receptor, but is not otherwise involved in the virus life cycle (Lyra et al., 1991). The approximately 25 structural proteins encoded by the PRD1 genome can be divided to four categories. These include the terminal protein that is covalently bound to the 5′ ends of the phage genome, capsid proteins, membrane-associated proteins and vertex proteins (Bamford et al., 1983; 1991; Grahn et al., 1999; Rydman et al., 1999). Similarly to adenoviruses, the icosahedral capsid of bacteriophage PRD1 is organized on a pseudo T = 25 lattice with 240 copies of the trimeric major coat protein P3 (Butcher et al., 1995; Benson et al., 1999). The vertices are occupied by an adenovirus-like spike–penton complex formed of proteins P2, P5 and P31 (Mindich et al., 1982a; Grahn et al., 1999; Rydman et al., 1999; J. Caldentey, R. Tuma and D. H. Bamford submitted; J. K. H. Bamford and D. H. Bamford submitted). The capsid surrounds an internal membrane vesicle that encloses the linear virus genome (Bamford and Mindich, 1982; Butcher et al., 1995). The spherical membrane comprises approximately half lipids and half protein (Davis et al., 1982; Bamford et al., 1991). The phage membrane is capable of undergoing a structural transformation from a spherical vesicle into a tubular tail-like structure, and it has been proposed that PRD1 uses this tail-tube to inject its genome through the host cell envelope (Lundström et al., 1979; Bamford and Mindich, 1982).

In this work, we show that bacteriophage PRD1 structural protein P7 has glycolytic activity. Mutant particles lacking protein P7 show delayed asynchronous DNA replication and lysis.


The 3′ end of gene VII encodes protein P14

Earlier studies have shown that both proteins P7 and P14 are encoded by gene VII. This was revealed by the ability of five different monoclonal P7 antibodies to recognize both proteins and by their absence from sus234 mutant particles containing a single amber mutation in gene VII (Hänninen et al., 1997). However, the exact location of the nucleotide sequence encoding protein P14 has not been defined.

In the course of an NTG-induced PRD1 mutant screening, 288 new amber mutants were obtained (Rydman et al., 1999). Complementation analysis mapped the sus471 mutation to gene II that encodes the receptor-binding protein of PRD1 (Table 1). Analysis of sus471 mutant particles in SDS–polyacrylamide gels and by Western blotting revealed that protein P7 was also absent. P14 and other structural proteins studied could be detected at wild-type levels (Fig. 1). Phages grown in suppressor hosts (thus having protein P2 and being infectious) did not have any detectable amounts of P7. We discovered that the specific infectivity of P2P7 mutant particles was decreased because of the absence of the receptor-binding protein P2 (9.1 × 107 pfu µg−1 protein in purified virus preparates, compared with 1 × 1010 pfu µg−1 protein in wild-type virus; Bamford and Bamford, 1990). Instead, P7 particles had a specific activity close to that of wild-type viruses (5 × 109 pfu µg−1 protein). This shows that protein P7 is not essential for PRD1 propagation and is in agreement with earlier observations (Mindich et al., 1982b).

Table 1. . Complementation of P7 mutants with cloned genomic fragments.
PhageStrainPropertiesTitre (pfu ml−1)
PRD1 sus234 S. typhimurium DS88Non-suppressor6.0 × 105
S. typhimurium DB7156(pLM2)Suppressor4.6 × 1011
E. coli HMS174(pSU18)(pLM2)Complementation control1.5 × 105
E. coli HMS174(pALH71)(pLM2)Includes PRD1 gene VII5.4 × 1010
E. coli HMS174(pPR46)(pLM2)Includes nucleotides 12534–12986 from PRD1 genome4.6 × 1010
E. coli HMS174(pPR49)(pLM2)Includes nucleotides 12675–12986 from PRD1 genome2.8 × 108
PRD1 sus471 S. typhimurium DS88Non-suppressor9.0 × 105
S. typhimurium DB7154(pLM2)Suppressor6.8 × 1011
E. coli HMS174(pSU18)(pLM2)Complementation control1.8 × 105
E. coli HMS174(pJB21)(pLM2)Includes PRD1 gene II1.0 × 1011
Figure 1.

Protein composition of PRD1 sus471 mutant particles. SDS–polyacrylamide gel and corresponding Western blot with polyclonal P2 antiserum and monoclonal P7/P14 antibodies. Lanes a and d, purified wild-type PRD1; lanes b and e, P2P7 mutant particles (sus471 propagated in a non-suppressor host); and lanes c and f, P7 mutant particles (sus471 propagated in a suppressor host). In addition to proteins P7 and P14, P7/P14 antibodies also recognize the major membrane-associated phage protein P11, as well as weakly the major coat protein P3.

To investigate whether the receptor binding of the suppressor-corrected protein P2 differs from that of the wild-type P2, we compared the adsorption rates of suppressor-grown sus471 mutant and wild-type particles (Fig. 2). No significant difference was detected. The plating efficiency of sus471 infection was normal on both E. coli and Salmonella typhimurium cells, but the plaque size was reduced.

Figure 2.

The adsorption of wild-type (filled circles) and P7 mutant particles (empty circles) to S. typhimurium DS88 cells. The receptorless SL5676 cells were used to demonstrate the receptor specificity of the binding (triangles, wild-type and P7 particles).

Mutant genes VII of sus234 and sus471 were sequenced, and a G→A change was detected in nucleotides 12721 and 12212 respectively (Fig. 3). In the case of sus234, codon 178 (TGG) is converted to an amber stop codon TAG. In the case of sus471, an opal stop codon TGA replaces codon 8 (TGG). This explains the absence of protein P7 in sus471 particles and suggests that translation of protein P14 is initiated somewhere in the region bounded by these two mutations. A closer inspection of this region revealed three possible ATG initiation codons at positions 12534, 12621 and 12675 with corresponding putative protein products of approximately 15.0, 12.2 and 10.3 kDa respectively. As the region preceding the ATG codon at 12621 seemed to have no homology to the Shine–Dalgarno sequence, we cloned only the two other open reading frames (ORFs) to a low-copy vector pSU18, resulting in plasmids pPR46 and pPR49. In complementation analyses, both plasmids were capable of rescuing the sus234 amber mutant (Table 1). The titres acquired with pPR46-carrying strains were two orders of magnitude higher than those obtained with pPR49 and five orders of magnitude higher than the titres of sus234 infecting a control strain carrying pSU18 without any inserts. The molecular mass of protein P14 approximated from SDS–polyacrylamide gels (≈18 kDa; Hänninen et al., 1997) also indicates the ATG at position 12534 as the most probable starting codon. We conclude that protein P14 is encoded by an ORF comprising nucleotides 12534–12986 at the 3′ end of gene VII in the PRD1 genome (Fig. 3).

Figure 3.

Gene VII of bacteriophage PRD1 and its protein products. The conserved transglycosylase motif characterized by Koonin and Rudd (1994) consists of amino acids 18–97 in the amino-terminal half of the protein. The predicted transmembrane helix (TM) comprises amino acids 219–236 near the carboxyl-terminus of protein P7. Location of the nucleotide changes in P7 (sus471) and P7/P14 (sus234) mutants are indicated by stars. Nucleotide sequences cloned into plasmids pPR46 and pPR49 for complementation analysis are indicated below the gene, whereas proteins P7 and P14 encoded by the gene are shown above it.

Gene VII, as well as the ORF encoding protein P14, was analysed using the phdhtm program for potential membrane-spanning helices. In both cases, a predicted transmembrane helix including amino acids 218–236 (protein P14) or 219–236 (full-length protein P7) was discovered (Fig. 3).

Protein P7 has glycolytic activity

The lytic activity of renatured protein P7 was detected in zymograms. Phage proteins were given an opportunity to obtain their native conformations by incubating gels in the renaturation buffer for 2–3 days. Four different incubation temperatures, 37°C, 22°C, 15°C and 4°C, were tested. P7 was found to be active at all temperatures, but a reduced activity was detected at 37°C. The highest activity was obtained by incubating zymograms at 4°C for 3 days. Two clear zones resulting from the lysis of peptidoglycan substrate added to the gel were detected in wild-type PRD1 particles (Fig. 4). The faster moving zone was shown to be protein P15, the enzyme responsible for the lysis of host cells at the end of the PRD1 life cycle (Caldentey et al., 1994). The structural role of P15 will be discussed elsewhere (P. S. Rydman and D. H. Bamford, to be published). The slower moving zone with lytic activity was associated with all mutant particles except sus234 and sus471, identifying it as protein P7. Protein P14 had no detectable activity. This localizes the muralytic domain of protein P7 to its amino-terminal half carrying the conserved transglycosylase motif (Fig. 3).

Figure 4.

Zymogram analysis of the wild-type and different mutant PRD1 particles. Hen egg white lysozyme (14 kDa) was used as a control. In wild-type PRD1, two clear zones indicating peptidoglycan hydrolysis activities appeared. These correspond to phage proteins P7 (27 kDa) and P15 (17 kDa) (see the text).

The absence of protein P7 leads to delayed asynchronous DNA replication and lysis

The adsorption assay showed that P7 particles bind to host cells as efficiently as wild-type viruses (Fig. 2). Previously, it has been shown that the newly synthesized viral DNA appears ≈ 20 min after infection by monitoring the incorporation of [3H]-thymidine (Davis and Cronan, 1983). We determined the thymidine incorporation during sus471 infection. Figure 5A shows that, in P7 particle infections, replication is delayed, and its initiation is asynchronous. In addition, the appearance of full-length phage DNA in infected cells was probed by two different DNA isolation methods. In both cases, the results were the same. In the wild-type infection, saturated amounts of phage DNA were detected at ≈ 20 min after infection. In sus471 infections, a saturated but somewhat lower amount of full-length phage DNA appeared some 20 min later (Fig. 5B and C).

Figure 5.

A. [3H]-thymidine incorporation into replicating DNA in S. typhimurium DS88 cells treated with nalidixic acid (filled circles) and corresponding cells infected with wild-type PRD1 (open circles) or P7 mutant particles (triangles). Nalidixic acid inhibits host replication by inactivating DNA gyrase. PRD1 DNA replication is not affected.

B and C. Agarose gel analysis of the appearance of full-length viral DNA in DS88 cells infected with wild-type PRD1 (B) and with sus471 mutant phages (C). C represents the non-infected control cells. PRD1 genomic DNA is indicated by an arrow.

Figure 6 shows the one-step growth curves in which non-suppressor (A) and suppressor (B) host cells have been infected with wild-type and different mutant viruses (sus170, sus234 and sus471, amber mutations in genes II, VII and amber mutation in gene II as well as opal mutation in gene VII respectively). Infecting viruses have been produced in suppressor hosts and, in consequence, they have normal protein composition, with the exception of sus471, which is lacking P7 (the opal mutation cannot be corrected by amber suppressors). The progeny viruses generated in a non-suppressor host have the respective mutant phenotypes. Mutant phages with amber mutations in gene II (sus170) and gene VII (sus234) have growth curves resembling the wild-type virus, irrespective of the suppressor capabilities of the host. Instead, sus471-infected suppressor cells lyse with nearly the same efficiency as the cells infected with wild-type viruses, but lysis is delayed ≈ 20 min. sus471-infected non-suppressor cells lyse less effectively about 30 min later than cells infected with wild-type PRD1. As sus170-infected and sus234-infected cells lyse normally, the lack of P2 or P7 alone cannot be responsible for the inefficient lysis of non-suppressor cells caused by the sus471 mutant.

Figure 6.

One-step growth curves of wild-type and different mutant PRD1 phages in a non-suppressor (A) and a suppressor (B) host. Cells were infected with wild-type PRD1 (filled circles), sus170 mutant (amber mutation in gene II, open circles), sus234 mutant (amber mutation in gene VII, filled triangles) and sus471 mutant (amber mutation in gene II, opal mutation in gene VII, open triangles). Cells were infected with the multiplicity of infection of 6. The life cycle of sus471 is delayed 20–30 min, irrespective of the suppressor capabilities of the host.


Gene VII of PRD1 consists of two different regions. The amino-terminal domain carries a glycolytic activity that probably serves the virus by degrading peptidoglycan during the entry of viral DNA into the host cell. The carboxy-terminal domain is identical to the essential phage protein P14 and seems to anchor the transglycosylase domain to the viral membrane (Fig. 3). A similar modular structure has been discovered in those lytic transglycosylases for which detailed structural information is available (Slt70 and Slt35; van Asselt et al., 1999a,b). We envisage that, at some point in PRD1 evolution, the nucleotide sequence encoding the lytic activity of protein P7 was captured from the host cell genome and added in front of a pre-existing structural phage gene creating the modern gene VII.

We currently view the injection of PRD1 DNA into the host cell cytoplasm as a multistage process. It appears that approximately 7–10 proteins are involved in the events associated with the receptor binding and phage DNA delivery. At the first stage, the primary adsorption protein P2 reversibly binds to the receptor (Grahn et al., 1999). This probably triggers the release of the spike–penton complex consisting of proteins P2, P5, P31 and the peripentonal major coat protein trimers. The removal of the vertex complex creates an opening in the capsid shell (Rydman et al., 1999) that enables membrane tube-tail formation. The two lytic activities carried in PRD1 particles, proteins P7 and P15, are probably associated with the emerging tail-tube. The presence of only one lytic enzyme is sufficient for successful infection. However, delayed DNA entry and lysis seem to be associated with P7 particle infection (Figs 5 and 6). In wild-type infections, these enzymes apparently work together and create a localized opening into the peptidoglycan layer, thus enabling the tail-tube to reach the cytoplasmic membrane. The events associated with the crossing of the inner membrane remain unknown, except that a transient K+ ion and ATP leakage from infected cells have been detected by electrochemical measurement (Daugelavicius et al., 1997).

The copy number of protein P7 has been estimated to be approximately 20 copies per virion (Davis et al., 1982). For protein P14, no such estimates are available but, in Western blots in which wild-type particles are probed with monoclonal P7/P14 antibodies, proteins P7 and P14 produce bands with approximately equal intensities. As they both carry the same epitope, this probably indicates similar copy numbers. Thus, some 20 copies of protein P14 would be present in a single PRD1 virion. Earlier studies have suggested that protein P7 forms multimeric complexes (Luo et al., 1993a,b). As protein P14 is included in protein P7 and the extra domain of protein P7 consists almost entirely of the transglycosylase consensus sequence, it is very likely that the putative multimerization signal resides in the P14 region. The primary results from our attempts to purify recombinant proteins P7 and P14 from E. coli cells indicate that they are both released together with membrane components. This is in agreement with the presence of a predicted transmembrane helix at the carboxyl-terminus of the proteins. We envisage that proteins P7 and P14 together form a heteromultimeric complex located at every vertex of the PRD1 particle, but anchored to the viral membrane.

Our results show that the putative transglycosylase of phage PRD1, protein P7, is actually associated with glycolytic activity and that this activity functions in the early stages of virus infection. These findings suggest that corresponding activities also exist for other putative phage and plasmid transglycosylases.

Experimental procedures

Bacteria, plasmids and phages

Bacterial strains, plasmids and phages used in this study are listed in Tables 2 and 3. Plasmids pALH71, pJB21, pPR46 and pPR49 were constructed by amplifying genes VII, II and two fragments at the 3′ end of gene VII (see Table 2) by polymerase chain reaction (PCR) and inserting them into low-copy plasmid pSU18 using standard molecular cloning techniques (Sambrook et al., 1989). Cells were grown in Luria–Bertani medium (LB; Sambrook et al., 1989) and, when appropriate, chloramphenicol (25 µg ml−1), kanamycin (25 µg ml−1) and/or tetracycline (20 µg ml−1) were added. For the production of wild-type and mutant phage particles, DS88 cells were infected at a multiplicity of infection (MOI) of 6. After lysis, the wild-type particles were concentrated and purified in 5–20% sucrose rate zonal gradients, as described previously (Bamford and Bamford, 1990). Light-scattering zones formed by viral particles were collected and harvested by centrifugation resulting in 1× virus.

Table 2. Bacterial strains used in this study.
StrainGenotype or descriptionReference
Salmonella typhimurium LT2 DS88SL5676 ΔH2, H1-i::Tn10 (Tcs) (pLM2) Bamford and Bamford, 1990
 standard non-suppressor host 
S. typhimurium LT2 DB7154(pLM2)DB7100 leuA414(Am), hisC527(Am) supD10 Winston et al. (1979)
 suppressor host for sus170 and sus471 
S. typhimurium LT2 DB7156(pLM2)DB7100 leuA414(Am), hisC527(Am) supF30 Winston et al. (1979)
 suppressor host for sus234 
S. typhimurium LT2 SL5676ΔH2, H1-i::Tn10 (Tcs)
 used in adsorption assay as non-binding
B.A.D. Stocker, Stanford University,
 Palo Alto, CA, USA
Escherichia coli K-12 DH5αUsed in peptidoglycan purification Sambrook et al. (1989)
E. coli K-12 HMS174Non-suppressor host for complementation
Campbell et al. (1978)
Table 3. Phages and plasmids used in this study.
PRD1 wtWild typeWild type Olsen et al., 1974
sus170 Amber mutation in gene IIP2 Mindich et al. (1982b)
sus234 Amber mutation in gene VIIXIVP7P14 Mindich et al. (1982b)
sus471 Amber mutation in gene II,
opal mutation in gene VII
P2P7/P7 when propagated in
non-suppressor/suppressor host
This study

Mutagenesis, isolation and mapping of phage mutants

Techniques for the N-methyl-N′-nitro-N-nitrosoguanidine (NTG) mutagenesis and isolation of amber mutants have been described earlier (Mindich et al., 1982b). Mutations were mapped by complementation analysis as described previously (Mindich and McGraw, 1983; Bamford et al., 1991). The initial screening was performed using large subgenomic clones that included several phage genes. If a positive result was obtained, mutants were mapped to the gene level using clones containing a single phage gene. The nucleotide sequence of gene VII from sus234 and sus471 mutants was determined using an automated sequencer (Perkin-Elmer ABI Prism 377XL) at the DNA Synthesis and Sequencing Laboratory, Institute of Biotechnology, University of Helsinki.

Adsorption and DNA replication assays

Phage adsorption test was accomplished by incubating 300–600 phage particles with 2 × 108Salmonella cells grown to the optimal adsorption phase (2 × 109 cfu ml−1; Kotilainen et al., 1993). The infection mixture was incubated for 0.5, 5, 10 and 15 min at 20°C, and the cells were collected by centrifugation (5000 r.p.m. for 3 min at 20°C). The number of non-adsorbed phage particles in the supernatant fractions was determined by plating them on DB7154(pLM2) lawns.

The labelling of replicating PRD1 DNA with [3H]-thymidine was accomplished according to the method of Mosig and Colowick (1995), except that nalidixic acid (40 µg ml−1) was added to inhibit host replication (Davis and Cronan, 1983). The amount of incorporated 3H was determined by scintillation counting.

Intracellular phage DNA was detected by extracting from infected DS88 cells at different time points (10, 15, 25, 30 and 35 min after infection for wild type and 20, 25, 30, 35, 40 and 45 min after infection for sus471). Two different DNA isolation methods were used, the alkaline lysis method and the isolation of the total cellular DNA of the cell (Sambrook et al., 1989). Isolated DNAs were analysed in 1% agarose gels.

Substrate preparation for zymogram assay

E. coli peptidoglycan sacculi used in zymogram analysis were prepared according to the method of Hoyle and Beveridge (1984) by boiling cells derived from 500 ml of stationary phase culture in 4% SDS. Sacculi were washed with 2 M NaCl as described by Bernadsky et al. (1994) to remove peptidoglycan-associated proteins, resuspended in 5 ml of Milli-Q water and kept at −20°C until used.

Zymogram assay

Lytic activity was detected by zymogram analysis as described by Bernadsky et al. (1994). SDS–polyacrylamide gels (14%; 101 × 82 × 0.5 mm) were casted according to the method of Olkkonen and Bamford (1989), except that E. coli peptidoglycan preparation was added to reach a final concentration of 15% (v/v). Before casting gels, peptidoglycan sacculi were homogenized by sonication. After electrophoresis, gels were rinsed with Milli-Q water and soaked for 30 min in Milli-Q water and for 30 min in renaturation buffer (25 mM potassium phosphate buffer, pH 7.4, 0.1% Triton X-100) with gentle agitation at room temperature. Gels were then transferred to fresh renaturation buffer and incubated for 48–72 h at the indicated temperature. After incubation, zymograms were stained with 0.1% methylene blue in 0.01% KOH at 37°C for 1 h and destained with Milli-Q water.

Analytical methods

SDS–PAGE was performed as described by Olkkonen and Bamford (1989). Western blotting was done by transferring proteins from SDS–PAGE gel onto a polyvinylidene difluoride (PVDF) membrane (Millipore) and visualized with the ECL detection system (Biological Industries) using horseradish peroxidase (HRP)-conjugated swine anti-rabbit Igs (Dako) or peroxidase-conjugated horse anti-mouse Igs (Vector) as secondary antibodies. Polyclonal P2 antiserum (from rabbit; Grahn et al., 1999) and monoclonal P7/P14 antibodies (7N41, from mouse; Hänninen et al., 1997) were used as primary antibodies for protein composition analysis of mutant phage particles. Protein concentrations were determined with Coomassie brilliant blue reagent (Biocell) using BSA as a standard (Bradford, 1976).

The nucleotide sequence encoding protein P14 as well as the entire gene VII was analysed for helical transmembrane regions using the PHDhtm (profile fed neural network systems from Heidelberg) program (Rost et al., 1994; 1995).


We thank Ms M.-L. Perälä for her excellent technical assistance. Drs J. K. H. Bamford and A.-L. Hänninen are warmly thanked for providing plasmids pJB21 and pALH71. M. Moak and I. Molineux are acknowledged for sharing their unpublished information on the putative transglycosylase of phage T7. Helsinki Graduate School in Biotechnology and Molecular Biology is acknowledged for a fellowship to P.S.R. This work was supported by the Finnish Academy of Sciences grants 62993 and 64298 (D.H.B.).