The Lactococcal Phages Tuc2009 and TP901-1 Incorporate Two Alternate Forms of Their Tail Fiber into Their Virions for Infection Specialization*

Background: Siphoviridae virions often possess lytic domains facilitating host-penetration. Results: Tuc2009 and TP901-1 virions may contain full-length or truncated tail fibers, possessing or lacking a lytic domain, respectively. Conclusion: Phages with a lytic domain infect stationary phase cells better, whereas truncated derivatives have higher adsorption efficiencies. Significance: The heterogeneous phage population serves to most effectively infect bacteria where levels of cell wall cross-linkage differ. Lactococcal phages Tuc2009 and TP901-1 possess a conserved tail fiber called a tail-associated lysin (referred to as Tal2009 for Tuc2009, and Tal901-1 for TP901-1), suspended from their tail tips that projects a peptidoglycan hydrolase domain toward a potential host bacterium. Tal2009 and Tal901-1 can undergo proteolytic processing mid-protein at the glycine-rich sequence GG(S/N)SGGG, removing their C-terminal structural lysin. In this study, we show that the peptidoglycan hydrolase of these Tal proteins is an M23 peptidase that exhibits d-Ala-d-Asp endopeptidase activity and that this activity is required for efficient infection of stationary phase cells. Interestingly, the observed proteolytic processing of Tal2009 and Tal901-1 facilitates increased host adsorption efficiencies of the resulting phages. This represents, to the best of our knowledge, the first example of tail fiber proteolytic processing that results in a heterogeneous population of two phage types. Phages that possess a full-length tail fiber, or a truncated derivative, are better adapted to efficiently infect cells with an extensively cross-linked cell wall or infect with increased host-adsorption efficiencies, respectively.

It is predicted that 60% of the estimated 10 31 phages in the biosphere belong to the Siphoviridae (1,2). Yet despite their prevalence, the molecular details surrounding the initiation of infection by such siphophages are not fully resolved. The Sipho-viridae tail tip is a multiproteinaceous structure responsible for host recognition and the initiation of phage infection. Siphoviridae often possess a central tail fiber suspending from their tail tip known to be essential for various aspects of phage infection such as tail morphogenesis, bacterial adsorption, cell envelope penetration, and directing DNA ejection (3)(4)(5)(6)(7)(8)(9)(10)(11).
Coliphages and T5 and Bacillus phage SPP1 are arguably the best studied models for understanding the various roles that central tail fiber proteins fulfill during Siphoviridae infection. possesses a central tail fiber, composed of three copies of gpJ protein, which forms a hub around which the tail organelle assembles (3). The C terminus of gpJ functions as a receptorbinding protein, attaching to LamB on the surface of Escherichia coli (5,12). The tail tube tips of reversibly bound particles are ϳ17 nm from the cell surface, whereas the tail tube tips are in direct contact with the bacterial envelope when this phage is irreversibly bound to its host (13). The transition from reversibly to irreversibly bound phage coincides with conformational changes in gpJ (detected through proteinase sensitivity) and the appearance of a transmembrane channel proposed to be formed by the tape measure protein (TMP) 2 gpH (10,14).
The tail protein Pb2 of Siphoviridae phage T5 serves both as the TMP and straight tail fiber. Insertion of purified Pb2 into black lipid membranes results in the formation of pores of ϳ2-nm diameter (15). The C terminus of Pb2 possesses a PGH domain, which promotes the penetration of the E. coli cell envelope (9). Located between the TMP and tail fiber region domains of Pb2 are two transmembrane domains. These have been shown to insert the tail fiber of T5 into a lipid layer containing the proteinaceous receptor, FhuA, of T5. Insertion of Pb2 into FhuA-containing liposomes causes conformational changes in Pb2, shortening its length while simultaneously doubling its width from 2 to 4 nm (9,17).
The central tail fiber of Bacillus subtilis phage SPP1 is composed of three copies of gp21 (11). The N terminus of gp21 forms a cap at the distal end of the tail tube, which has been shown to open at the onset of infection (18). The C terminus of gp21 acts as a receptor binding protein attaching to the membrane-spanning protein YueB (19 -21). Irreversible adsorption of SPP1 to YueB initiates infection and is accompanied by conformational changes transmitted across the full-length of the tail tube to the portal protein by means of a cascading rotational movement of the major tail protein of the siphophage (11).
Siphoviridae phages infecting lactic acid bacteria, and in particular Lactococcus lactis, are the single largest cause of industrial milk fermentation problems and result in significant economic losses (22). The P335 phage species is one of the three major lactococcal infecting phage species that cause regular problems for the dairy industry (the other species being c2 and 936) (23,24). Scientific efforts to understand and control this particular problem in the dairy industry have allowed lactococcal phages to become one of the most intensely studied phage groups and one of the model systems for understanding how Siphoviridae phages infect their host (25).
The tail fiber of lactococcal phages Tuc2009 and TP901-1 is composed of a trimer of the so-called Tal protein, which between them share 95% amino acid identity (26). Tal 2009 and Tal 901-1 are predicted to possess a C-terminally located M23 peptidase PGH domain, which distally protrudes from the large host-recognizing baseplate structure of each of these phages (27)(28)(29). The N terminus of Tal 2009 and that of Tal 901-1 both share 26% amino acid identity with the N terminus of SPP1 tail fiber protein gp21 (30). The N-terminal portion of Tal 2009 and Tal 901-1 is therefore expected to fulfill an analogous role at the initiation of infection, opening to facilitate ejection of the TMP and DNA. Located between the predicted C-terminal M23 peptidase domain and the N-terminal tail tube cap of Tal 2009 and Tal 901-1 is a glycine-rich sequence, GG(S/N)SGG. Interestingly, GGSSG*GG has been shown to undergo proteolysis in vitro and in vivo (cleavage location indicated by an asterisk), which results in mature Tuc2009 virions that have incorporated both full-length and truncated Tal proteins (31).
Here, we present the biochemical and mutational analysis of Tal to further understand the role of tail fiber proteins during infection. The importance of the cell wall-degrading activity and proteolytic processing of Tal 2009 and Tal 901-1 for host penetration and adsorption is discussed and placed within the context of current knowledge on host penetration by Siphoviridae phages.

MATERIALS AND METHODS
Bioinformatic Analysis-Relevant DNA sequences were downloaded from NCBI and manipulated using the DNAstar software package (DNASTAR, Inc., Madison, WI). Protein domains were identified using InterProScan from EMBL-EBI. Multiple sequence alignments were generated online using the ClustalW program (EMBL). The amino acid percentage identity between protein sequences was estimated by BLAST2p (http://blast.ncbi.nlm.nih.gov/Blast.cgi).
Bacterial Strains, Culture Conditions, Phage Preparations, and Plasmids-Bacteria, phages, and plasmids used in this study are listed in Table 1. L. lactis strains were grown at 30°C in M17 (Oxoid) broth or agar supplemented with 0.5% (w/v) glucose (GM17). Erythromycin and chloramphenicol were used as necessary at concentrations of 5 and 10 g/ml, respectively. L. lactis strain NZ9000 carrying the TP901-1erm prophage was isolated by simultaneously inoculating 2% (v/v) NZ9000 from a fresh overnight and 2% (v/v) TP901-1erm from a 10 10 PEG-precipitated lysate, followed by a 6-h incubation, before selecting for erythromycin-resistant colonies. Largescale phage preparations of TP901-1erm and mutant derivatives were performed as described previously (4) by inducing 1.6 L of the lysogenized NZ9000 host with mitomycin C. TBT buffer (100 mM NaCl, 50 mM CaCl 2 , 10 mM MgCl 2 , 20 mM Tris-HCl (pH 7.0)) was used as the phage resuspension buffer.
Recombineering and Oligonucleotides-All oligonucleotides used in this study are shown in supplemental Table 1. Recombineering was performed as described previously (32). However, specific modifications optimized for L. lactis such as phosphorothioate linkages and transforming 500 g of oligonucleotides were also employed (33). Recombineering oligonucleotides were ordered from Integrated DNA Technologies (Leuven, Belgium), whereas all other oligonucleotides were ordered from Eurofins MWG (Ebersberg, Germany).
Protein Expression and Purification-The terminal 501 bp of the 3Ј-end of tal 2009 , was overproduced in E. coli using the pQE60 vector construct named pTal 2009 -5⌬N as described previously (31). The 501-bp 3Ј-end of tal 2009 was also used to fuse it in-frame with the DNA encoding two LysM cell wall-binding domains of the endolysin of Tuc2009 by Splicing by Overlap Extension (SOEing) PCR (34,35). The resulting amplicon, which specified a Tal 2009 M23peptidase-LysM fusion protein and designated TalM23-LysM, was cloned into pNZ8048 using standard techniques, to generate plasmid pTalM23-LysM. The TalM23-LysM protein was overproduced in NZ9000 using the NICE expression system (36) and required purification using a denaturing renaturation procedure as described in the QIAexpressionist handbook (Qiagen).
Assay for Lytic Activity-L. lactis UC509.9 cells were autoclaved, and the insoluble material was washed twice with water, once with 10 mM sodium phosphate buffer (pH 6.5), and finally resuspended in the latter buffer to produce a suspension with an A 600 of 0.450. All lysis experiments were performed in triplicate at 30°C. Although only water was added to the negative control, mutanolysin (Sigma), at a concentration of 0.1 units/ ml, was used in all samples to hydrolyze the glycan strand linkages of the lactococcal cell wall. Tal 2009 -5⌬N and TalM23-LysM were added at 1.0 and 1.7 g/ml (equimolar concentrations), respectively, to achieve further degradation of the cell wall. This combined degradation of the cell wall results in lysis and in a reduction in A 600 . Cell wall hydrolysis using Tal 901-1 -5⌬N and Tal 901-1 -5⌬N H892A was performed in a similar manner, except that autoclaved cells were prepared from the TP901-1 host L. lactis 3107. Tal 901-1 -5⌬N and Tal 901-1 -5⌬N H892A were assayed at equal concentrations of 1.75 g/ml.
Determination of Hydrolytic Specificity on Pure Peptidoglycan (PG)-L. lactis MG1363 PG was prepared and hydrolyzed by mutanolysin as described previously (37). The resulting soluble muropeptides were reduced with sodium borohydride. Following this predigestion step and prior to the addition of TalM23-LysM, the samples were adjusted to pH 6.5 by the addition of NaOH. TalM23-LysM-mediated digestion of MG1363 PG-derived muropeptides was carried out for 16 h at 30°C with gentle agitation, using 0.04 mg/ml of enzyme. Samples were then analyzed by reverse phase HPLC with an Agilent UHPLC1290 system using ammonium phosphate buffer and methanol linear gradient as described previously (34). All of the peaks of the mutanolysin digest and most of the peaks (peaks 4 to 12) of the mutanolysin plus TalM23-LysM digest were analyzed without desalting by MALDI-TOF MS using a Voyager-DE STR mass spectrometer (Applied Biosystems) as reported previously (37). Structural identification of the muropeptides was deduced from the previously published reference profile (37).
Phage Assays-Phages used in this study were induced from their lysogenic host by the addition of 3 g/ml mitomycin C (Sigma) to a liquid culture of the relevant strain when it had reached an A 600 of 0.1. Phage lysates were sterilized by filtration (0.45-m pore filter) and stored at 4°C. Phage titers were determined in triplicate by plaque assay employing three independent lysates using the standard double agar method (38) against the TP901-1 sensitive host L. lactis 3107. All GM17 media contained 5 g/liter glycine and 10 mM CaCl 2 .
Adsorption assays were performed against host 3107 at an A 600 of 0.3 with a phage host multiplicity of infection of 0.01. Phages were allowed to adsorb for 10 min at 30°C in the presence of 10 mM CaCl 2 . The adsorption assay control without bacterial cells (i.e. medium only) allowed enumeration of the initial phage titer. The percentage of phage that adsorbed was calculated as ((initial phage titer Ϫ final phage titer)/initial phage titer) ϫ 100 (39).
To compare the ability of TP901-1 and its Tal protein mutant derivatives to infect exponential phase or stationary phase cultures, center of infection (COI) assays were performed. Phage infection of L. lactis 3107 was performed using a multiplicity of infection of 0.01 at 30°C in the presence of 10 mM CaCl 2 . Phages were added to exponentially growing 3107 strain at A 600 ϳ 0.3 (host 1) or to stationary phase cells of 3107 following culture dilution to A 600 ϳ 0.3 using cell-free "spent" GM17 medium (host 2). After 10 min, cells were washed twice in GM17 to remove free phage, and subsequent dilutions were plaque assayed against an uninfected 3107 culture to determine the COI values of host 1 and host 2, after which the efficiency of COI (ECOI) formation was calculated as ((COI host 1 )/(COI host 2 )) (40).
Competition assays between Tal 901-1 mutants TP901-1erm GlyϾArg , which possesses a full-length tail fiber, and TP901-1erm Gly603 *, which has a truncated Tal protein, were performed to directly compare the biological significance of the tail fiber and associated lytic domain. An equal volume of freshly prepared phage lysates of TP901-1erm GlyϾArg and TP901-1erm Gly603 * were mixed together, and the phage mixture was diluted 1:10 into both an exponential and stationary phase culture of L. lactis 3107 as described above, except the phages were allowed only 5 min to establish infection.

JOURNAL OF BIOLOGICAL CHEMISTRY 5583
nitrocellulose membrane (Hybond-C Extra, Amersham Biosciences) in an electrotransfer unit (Bio-Rad) for 2 h at 100 V, in a 10% methanol, 10 mM 3-(cyclohexylamino)-1-propanesulfuronic acid (CAPS; pH 11.0) transfer buffer, which was kept cool by placing the unit on ice. The two primary antibodies for Western blots, recognizing the N terminus of the Tal protein or the baseplate protein BppU, had been described previously (41) and shown to interact with the relevant protein of both Tuc2009 and TP901-1 (4). Western blot secondary antibodies, IRDye 680RD goat anti-rabbit antibodies (Li-Cor Biosciences), and prestained protein molecular weight ladder (11-250 kDa; Li-Cor Biosciences) were detected in the 700-nm channel using an Odyssey CLx infrared imager (Li-Cor Biosciences). (31) previously demonstrated that the P335 species phage Tuc2009 possesses a tail-associated lytic activity. Analysis of the genomes of two c2-type and forty 936-type phages did not reveal PGH domains associated with tail proteins (data not shown). The genomes of the eleven publicly available P335 species phages (23) all contain a gene that encodes an easily recognizable (due to its large size) TMP, followed by two genes that specify the Dit (distal tail protein) and tail fiber protein (see Fig.  1) (25). ClustalW alignments of the presumed TMP, Dit, and tail fiber protein sequences resulted in four closely related P335 species phage with a lysozyme-like domain associated with their TMP, five closely related P335 species phages with M23 peptidase domains associated with their tail fiber proteins, and two P335 species phage with no recognizable PGH domain (Fig. 1).

Analyses of P335 Species Structural Proteins to Identify Virion-associated PGH Domains-Kenny and colleagues
Tal 2009 PGH Activity against Lactococcal Cells-The constructs Tal 2009 -5⌬N and TalM23-LysM both generate proteins with the same the C-terminal M23 peptidase of Tal 2009 tail fiber; however, the latter contains two additional LysM cell wall binding domains ( Fig. 2A). To verify the cell wall-degrading activity of Tal 2009 -5⌬N and TalM23-LysM proteins, the lytic activity of these purified recombinant enzymes was assayed on autoclaved bacterial cells as outlined under "Material and Methods." Neither the addition of water (negative control) nor the addition of mutanolysin at low concentrations (0.1 units/ ml) resulted in significant lysis of autoclaved cells, as measured through A 600 reduction of the PG-containing suspension (Fig.  2B). In contrast, a combination of mutanolysin plus Tal 2009 -5⌬N or TalM23-LysM resulted in a clear reduction in A 600 . TalM23-LysM was significantly more effective at hydrolyzing PG relative to Tal 2009 -5⌬N Ͼ20 h (Fig. 2B), which is consistent with other reports where the addition of cell wall binding motifs were used to increase the lytic activity of PGH catalytic domains (42)(43)(44). TalM23-LysM was therefore used in subsequent analyses.
Characterization of Tal 2009 PGH Hydrolytic Specificity-The PGH catalytic domain of Tal 2009 is located at its C-terminal extremity, as reported previously and confirmed here, and contains an M23 peptidase domain (31). The M23 peptidase family includes endopeptidases that target the interpeptide bridge of PG. To confirm this prediction, it was necessary to identify the specific chemical bond targeted by the purified TalM23-LysM. PG was extracted from L. lactis MG1363 and first degraded by mutanolysin, a muramidase, as described previously by Courtin et al. (37). The resulting soluble muropeptides were further incubated with the TalM23-LysM recombinant protein. Muro- peptides obtained after mutanolysin digestion (Fig. 3A) and after mutanolysin plus TalM23-LysM digestion (Fig. 3B) were separated by reverse phase HPLC. Comparison of the two obtained profiles revealed that the higher molecular weight muropeptides obtained by mutanolysin digestion of PG such as dimers (peaks 16 -30), trimers (peaks 31-36), and tetramers (peaks 37-39) were degraded by TalM23-LysM to its constituent monomers (peaks 1-15; see supplemental Table 2). This   Table 2). FEBRUARY 22, 2013 • VOLUME 288 • NUMBER 8 pattern of degradation shows that the PGH catalytic domain of Tal 2009 acts as an endopeptidase that hydrolyzes the lactococcal peptidoglycan within the interpeptide bridge at the D-Ala-D-Asp/Asn bond. Thus, similar to other M23 peptidases (45)(46)(47), Tal 2009 is a D,D-endopeptidase able to hydrolyze PG cross-bridges.

Tail-associated Lytic Enzymes of Siphoviridae Phages
Construction of NZ9000-TP901-1erm for Tal 901-1 Mutagenesis-To investigate the precise role of lactococcal Tal proteins during infection, mutagenesis was attempted on the gene encoding the Tal 2009 protein. However, no mutagenesis system was found to be compatible with the host strain UC509.9 of Tuc2009 (results not shown). As Tal 2009 and Tal 901-1 share 95% amino acid identity, and because mutagenesis of the TP901-1 genome has been reported previously (4, 48 -50), mutagenesis was instead performed on the Tal 901-1 genome. However, rather than using the rather laborious pGhost mutagenesis method (51), we decided to employ recombineering, a mutagenesis system that has recently been adopted for L. lactis (32,33).
For the purpose of achieving TP901-1 mutagenesis by recombineering, we constructed a derivative of L. lactis strain NZ9000 that harbored the TP901-1erm prophage (a derivative of TP901-1 carrying an erythromycin marker; (50)), which was designated NZ9000-TP901-1erm (see "Materials and Methods"). The integration of TP901-1erm via attB-attP recombination (52) was confirmed by (a) the lack of a PCR product with primers, which flanked NZ9000s attB sequence, and (b) by generating PCR products from the primers flanking the NZ9000 attB sequence to TP901-1 chromosomal sequences (results not shown). TP901-1erm phages induced from NZ9000 with mitomycin C were still capable of infecting L. lactis 3107 at levels of ϳ10 7 pfu/ml.
Characterization of Tal 901-1 Mutants-Specific mutations were made in TP901-1 Tal protein to investigate the role(s) of the various conserved features observed in Tal 2009 and Tal 901-1 during infection. Mutant TP901-1erm GlyϾArg was based on previous work (31), which had shown that altering the DNA sequence of three glycine codons, corresponding to the proteolytic processing site GGSSGGG of Tal 2009 , to arginine codons, GRSSRRG, prevented in vitro proteolysis. The desired mutation was introduced into the TP901-1erm genome, and one such mutant prophage was selected, and its genome was sequenced to verify the genotype of the generated mutant phage. Western blots detecting the N terminus of Tal 901-1 protein verified that the mutated Tal protein of mutant TP901-1erm GlyϾArg was unable to undergo proteolysis as only a single band corresponding to the full-length Tal 901-1 protein (102.1 kDa) was detected, in contrast to the Tal protein of TP901-1erm (Fig. 4).
The second mutation in tal 901-1 , generating mutant prophage TP901-1erm H892A , was designed to inactivate the M23 peptidase domain, as demonstrated for the Bacillus subtilis prophage SP-␤ (45). The histidine at position 892 (His-892) of Tal 901-1 , underlined in the M23 peptidase sequence PHLHF of TP901-1erm, was targeted for mutation to an alanine residue as this is the first histidine in the HxH motif characteristic of zinc metallopeptidases (53). To verify that the M23 peptidase of mutant TP901-1erm H892A was inactive, recombinant proteins of the C-terminal M23 peptidase domain of TP901-1 were created. Constructs Tal 901-1 -5⌬N and Tal 901-1 -5⌬N H892A , representing the same C-terminal regions of their respective tail fibers as construct Tal 2009 -5⌬N, were assayed against autoclaved cells of L. lactis 3107 (see Fig. 2C). As expected, recombinant enzyme Tal 901-1 -5⌬N derived from TP901-1erm showed lytic activity, whereas Tal 901-1 -5⌬N H892A , constructed from mutant TP901-1erm H892A , showed no activity (see Fig.  2D). Western blots determined that the Tal protein of mutant TP901-1 H892A still undergoes proteolytic processing. This is evident as the full-length and truncated Tal 901-1 proteins (102.1 and 67.4 kDa, respectively) are detected (Fig. 4). These in vivo results show that the M23 peptidase domain Tal 901-1 is not involved in the proteolytic separation of the Tal protein N and C terminus.
The third mutant TP901-1 derivative, TP901-1erm Gly603 *, was designed to introduce a stop codon into the coding sequence of tal 901-1 at the position where it is known to undergo proteolysis. This mutation would be expected to result in a homogeneous population of phages, the reverse of TP901-1erm GlyϾArg , with all of the Tal 901-1 proteins of the TP901-1erm phage population truncated at the sequence GGNSG*GG (position of stop codon highlighted by an asterisk). TP901-1erm Gly603 * also allows a comparison with mutant TP901-1erm H892A , whose Tal protein is also expected to be incapable of degrading peptidoglycan. Western blots verified the truncated TP901-1erm Gly603 * protein (67.4 kDa), which runs slightly below its expected position on the SDS-PAGE gel (Fig.  4). As expected, no full-length Tal 901-1 protein (102.1 kDa) was detected in the Western blots of TP901-1erm Gly603 * in contrast TP901-1erm. Western blots using primary antibodies directed against the upper baseplate protein, BppU, which immediately succeeds tal 901-1 , were performed on all mutants to verify that there were no downstream effects caused by mutating tal 901-1 (results not shown).
Efficiency of Plaque Formation of TP901-1erm and Its Tal 901-1 Mutant Derivatives-To determine the effect of the various mutations of tal 901-1 on the efficiency of plaque formation of its corresponding TP901-1erm-derived phages, the ability of these mutant phages to infect lactococcal host 3107 was tested by plaque assay analysis. Plaque assays of TP901-1erm and its mutant derivatives were performed in triplicate on three independently produced lysates. The efficiencies of plaque formation of TP901-1erm GlyϾArg , TP901-1erm H892A and TP901-1erm Gly603 *, relative to TP901-1erm phage, were thus determined to be 0.408, 0.050, and 0.062, respectively (Fig.  5A). The plaque morphologies of the Tal 901-1 mutant phages were also noticeably different from those produced by wild type TP901-1erm: TP901-1erm GlyϾArg produced plaques that were slightly smaller and hazier compared with TP901-1erm, whereas TP901-1erm H892A and TP901-1erm Gly603 * both formed pinprick-sized plaques (results not shown).
Efficiency of Formation of Centers of Infection-The M23 peptidase domain associated with Tal 2009 and Tal 901-1 is expected to facilitate cell wall penetration and genome delivery, particularly during conditions where the cell wall is extensively cross-linked, as would be expected for stationary phase cells (8,54). To compare the ability of TP901-1erm and mutant derivatives to infect stationary phase cells compared with exponentially growing cells, the phage ECOIs were calculated (see "Materials and Methods"). The ECOI for TP901-1erm was cal- culated as 0.41 Ϯ 0.02, TP901-1erm GlyϾArg as 0.56 Ϯ 0.05, TP901-1erm H892A as 0.22 Ϯ 0.005 and TP901-1erm Gly603 * as 0.27 Ϯ 0.03. The difference in the abilities of the mutant phage to infect stationary phase cells, relative to TP901-1erm, is considered to be statistically significant (p Ͻ 0.05; Fig. 5C).
Competition Assays between TP901-1 Mutants with and without Tail Fiber-To compare and contrast the biological significance of TP901-1 proteolytic processing of Tal 901-1 , the TP901-1 mutants permanently possessing or lacking the tail fiber were mixed together and simultaneously allowed to infect. The number of COIs established could be calculated due to distinguishing plaque morphologies formed by these phages. TP901-1erm GlyϾArg , which possesses the C terminus of Tal 901-1 and thus the M23 peptidase domain, forms more COIs in stationary phase cells than TP901-1erm Gly603 *. In contrast, TP901-1erm Gly603 *, which lacks the C terminus of the tail fiber and the PGH motif, is more efficient at establishing COIs in exponentially growing cells (Fig. 5D).

DISCUSSION
The proteolysis of Tuc2009 and TP901-1 Tal protein represents a molecular "hedge betting" mechanism to produce a naturally mixed population of phages, which are simultaneously adapted for efficient host adsorption and host penetration. Many phages have evolved sophisticated strategies to efficiently infect their hosts; however, to our knowledge, the proteolytic processing step of Tal 2009 and Tal 901-1 represents a novel phage mechanism for infection optimization under particular conditions. Not all phages appear to require PGH domains to facilitate host penetration and infection, as many phages lack any recognizable PGH domain associated with their structural proteins. However, there is growing evidence to support that virion-associated PGH domains facilitate infection under conditions where the cell wall is extensively cross-linked (for a recent review, see Ref. 55). In the current study, we show that the M23 peptidase of Tal 2009 possesses a D-Ala-D-Asp endopeptidase specificity. This tail-associated lytic activity facilitates phage infection by undoing the interpeptide cross-linkage of peptidoglycan glycan strands. As the majority of the sequenced P335 species lactococcal phages possess PGH domains associated with their tail proteins, the localized degradation of the bacterial cell wall is predicted to be an important step in the initial stages of infection by this species of phage.
In this study, we clearly show that TP901-1erm-derived mutants defective in their tail-associated PGH activity are not completely deficient in phage infection but display decreased infection efficiencies. This is consistent with the study of Moak and colleagues (56), who demonstrated that mutagenesis of the virion-associated PGH domain of T7 did not inhibit but merely delayed infection. Furthermore, the reduced ability of TP901-1erm PGH mutants to infect stationary phase cells is in agreement with the study of Puiri et al. (8), where the lytic domain of the tail fiber of Mycobacterium phage TM4 was necessary for efficient infection of stationary phase cells.
It was previously reported by Scholl and colleagues (57) that phage K1-5 encodes and incorporates two different tail fiber proteins into its virion, adapting the phage to penetrating hosts with a different cell envelope composition. Phages Tuc2009 and TP901-1 (and perhaps many other phages) have similarly evolved a novel strategy to generate a heterogeneous population. Through Tal proteolysis, the lactococcal phages Tuc2009 and TP901-1 generate two phage types: incorporating either a truncated Tal protein or incorporating a full-length tail fiber with an associated lytic domain. Phages that feature a PGH domain in their virion are better adapted to infecting bacteria with a higher degree of cross-linkage in their cell wall. However, this evolutionary advantage appears to be a double-edged sword, as the fitness cost is at the expense of phage adsorption efficiency.
The study of lactococcal phages pertains not only to their industrial significance but their broader importance as model systems. The clear differences in the tail fibers and infection mechanisms of lactococcal phages show the multifaceted approach by phages to achieve infection. Studies characterizing these underlying molecular processes serve to better understand Siphoviridae, and their host interactions, in the biosphere.