INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Among the tailed phages, newly synthesized virus particles are usually released from bacterial host cells following the synergistic action of a two-component lysis system: a hydrophobic membrane protein, termed holin, forms nonspecific pores or lesions in the cell membrane to promote access of a cell wall-hydrolyzing enzyme (endolysin) to the peptidoglycan substrate (44, 45). Such dual-component lysis systems, of which the phage  S and R gene products represent the best studied prototypes, have recently been discovered in several phages of both gram-negative and gram-positive hosts. The system appears to be extremely heterogeneous, since at least 11 apparently unrelated holin gene families have been identified (8) which are associated with one or more of at least five endolysin enzyme functions, such as the hydrolysis of glycosidic linkages (by muramidases and glucosaminidases) and the hydrolysis of amide bonds (by amidases and peptidases) (24, 44).

Putative holin proteins may be identified by a number of characteristic properties: (i) the presence of encoding genes, usually located immediately upstream of the endolysin genes; (ii) the presence of at least two hydrophobic transmembrane domains with no net charge, separated by a short beta-turn linker; (iii) a highly charged, hydrophilic, C-terminal domain; and (iv) a dual-start motif [5'-ATG-(NNN)_{1 or 2}-ATG-...-3'], not always present, permitting the synthesis of two products of different length, which are thought to regulate the pore-forming process (8, 45). Holin proteins range in size between 60 and 185 amino acids (22, 45).

At least for the phages of gram-positive hosts, however, the dual-component lysis system may not be universal. Although the presence of holins has been shown or suggested for several phages (2, 6, 10, 17, 22, 24, 27, 37, 41, 43), no genes encoding putative holins have yet been found to be associated with the endolysin genes of Listeria phage A511 (24) and Bacillus cereus phages (23). It is interesting that the N terminus of the endolysin of one of the Bacillus phages (TP21) shows extensive sequence homology to a signal sequence (including a cleavage site) from related Bacillus cell wall autolysins.

Staphylococcus aureus bacteriophage 187 is of special interest because it is the only member of staphylococcal bacteriophage species 187 and the sole representative of serogroup L phages (1). It differs from all other S. aureus phages by its host range, DNA restriction enzyme profiles (12), and distinctive set of virion proteins (21). Strains lysed by this phage have never been found to be lysed by any other phage (3), probably due to specific teichoic acids which are required for phage reception (29). We describe here the cloning and functional analysis of the unusual lysis system of this virus, which is comprised of a very small, putative holin (Hol187), whose coding sequence is fully embedded into the genetic module containing the enzymatically active domain of the large endolysin gene. We also provide evidence that the peptidoglycan hydrolase is related to major staphylococcal autolytic enzymes, and confirm the nonspecific membrane lesion-forming, holin-like nature of Hol187 by the complementation of a defective  S allele.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Organisms, plasmids, and culture conditions. All bacterial strains, phages, and plasmids used throughout this study are listed in Table 1. Escherichia coli JM109(DE3) was used in initial detection assays and for the overexpression of the ply187 endolysin gene cloned in pSP72. JM109 and W3110 were used for the expression of hol187 and its derivatives, and LE392 and HB101 were hosts for the propagation and assay of  phages. S. aureus 187 and Micrococcus luteus were grown in brain heart infusion broth or tryptose media at 37°C. Phage 187 was propagated on double-layer agar plates. Virus particles were concentrated from the lysates by polyethylene glycol 8000 precipitation and purified by ultracentrifugation on stepped CsCl gradients as previously described (33, 46). E. coli was grown in standard Luria-Bertani media (33) at 37°C. For the selection of plasmid-bearing cells, ampicillin was added at 100 µg/ml.

Cloning and identification of the lysis genes. Phage DNA was extracted and purified according to standard methods (33). For the construction of an expression library in E. coli, DNA was partially digested with AluI. Following electrophoresis in low-melting-point agarose, fragments in the range of 1,500 to 3,000 bp were recovered by -Agarase digestion (Boehringer) and concentrated by ultrafiltration (Microcon 100; Amicon). DNA fragments were then ligated into pSP72, which had been linearized with SmaI.

DNA sequencing and computer analysis. The nucleotide sequence of the plasmid insert was determined on both strands by primer walking with synthetic oligonucleotides. The sequence downstream of the central AluI site in ply187 (see Fig. 1) was determined by direct sequencing of phage 187 DNA, by using primers as sequences became available. Sequenase, version 2.0 (U.S. Biochemicals), and -³⁵S-dATP (Amersham) was used in all sequencing reactions. The program DNAsis for Windows, version 2.01 (Hitachi), was used for the analysis of nucleotide and amino acid sequences.

Cloning and expression of truncated Ply187 in E. coli. To determine the enzymatically active domain(s) of Ply187, specific fragments of ply187 (see Fig. 3) were derived by PCR. For optimal expression, a consensus ribosomal binding site (RBS) and spacer to the ATG start codon was provided on the forward (Fwd) primers. The following forward and reverse (Rev) primers were used (RBSs are underlined and start codons and stop codons are shown in boldface): 1-Fwd, 5'-ACTTGGATCCGAGGAGAAATTACTATGGCACTGCCTAAAACGGG TAAACC-3'; 471-Fwd, 5'-ACTTGGATCCGAGGAGAAATTACTATGGCACAAAACAATCCTGCACCTAAAGAC-3'; 471-Rev, 5'-ACTTGGATCCTTATGGTGGTGTAGGTTTCGGTTCTGC-3'; 892-Rev, 5'-ACTTGGATCCTTAAGCTAATGACAAACATTCGATTTCATT-3'; and 1887-Rev, 5'-ACTTGGATCCTTATTTTTTATATTGATCGTATATAAAAT-3'. Purified phage 187 DNA (20 ng) was used as a template in the amplification reaction with Taq polymerase (Qiagen). PCR products were digested with BamHI and ligated into the BamHI site of pSP72. Transformation of plasmids, screening for clones expressing a lytic activity, and confirmation of correct insertion were done as described above. Finally, the lytic activity of the individual recombinant proteins was scored by comparing the sizes of the clearing zones, which appeared 0.5 to 1 h after overlaying with the indicator cells. For the possible differentiation of amidase from glucosaminidase activity, M. luteus cells were also tested here (38).

Cloning and expression of hol187 in E. coli. To investigate the effect of heterologous expression of hol187 in E. coli and to study specific alterations in Hol187, PCR was used for the construction of mutated hol187 genes by modified forward primers (Hol187, 5'-ATCAGAATTCGAGGAGAAATTAATATGTTGATGGTTATTATGGTCGGCAATGTTGGGATT-3'; Hol187-S, 5'-ATCAGAATTCGAGGAGAAATTAATATGGTTATTATGGTCGGCAATGTTGGG ATT-3'; Hol187-L, 5'-ATCAGAATTCGAGGAGAAATTAATATGTTGCTGGTTATTATGGTCGGCAATGTTGGGATT-3'; and Hol187-Ch, 5'-ATCAGAATTCGAGGAGAAATTAATATGTTGATGGTTATTATGGACGACAATG TTGGGATT-3'). The reverse primer was 5'-ATCAGAATTCTTATATCACCTGGTTTAGGGACAAAAT-3' (see Table 1 for the individual products). The DNA fragments resulting from the amplification of phage 187 DNA were digested with EcoRI (Boehringer) and ligated into pBluescript II, followed by electroporation into JM109. This strain was selected because it features the laqI^q mutation, which more efficiently controls the expression of genes cloned under control of the lac promoter. The nucleotide sequences and orientations of the cloned genes were verified by using primers complementary to the T7 and T3 promoters flanking the multiple cloning site in this vector. To investigate the effects of the four gene products on growth (i.e., cell viability) of plasmid-bearing E. coli, log-phase cultures (50-ml volume; optical density at 600 nm [OD₆₀₀], approximately 0.1) were induced for gene expression by the addition of 1 mM IPTG, and growth (measured as optical density) was monitored over time.

Determination of total and released ATP during hol187 expression. To determine the effect of Hol187 on the integrity (i.e., leakiness) of the E. coli cytoplasmic membrane, the release of ATP into the medium was determined following the overproduction of Hol187 in JM109(pHOL187). Cells were diluted (in a volume of 50 ml) to approximately 10⁷ cells/ml, and hol187 expression was induced with 1 mM IPTG. Samples (1 ml) were taken at the indicated time points (see Fig. 6). The ATP assay was the highly sensitive firefly luciferase type and was carried out according to the instructions of the manufacturer (bioluminescence assay HS II; Boehringer). The emitted light was quantified in a photon-counting tube luminometer (Lumat 9501/16; Berthold), with a delay of 0.5 s and integration of the signal over the following 10 s.

12

16

Release of -galactosidase. The four plasmids specifying native Hol187 and mutated proteins (pHOL187, pHOL187-S, pHOL187-L, and pHOL187-Ch) were electroporated into E. coli W3110. This strain was used here because it produces native -galactosidase. Transformants were plated directly onto agar plates containing the chromogenic substrate X-Gal (5-bromo-4-chloro-3-indolyl--D-galactopyranoside; Sigma), and incubated for 24 h. W3110 control cells were plated on medium without antibiotic.

Immunological detection of Hol187. To verify that Hol187 is actually made in phage-infected S. aureus cells, antibodies were raised in rabbits against a synthetic 18-mer peptide derived from the Hol187 amino acid sequence DTGTLRHQATQEIWHGID. A cysteine residue was added to the N-terminal end, and the peptide was subsequently conjugated to keyhole limpet hemocyanin. The conjugate was used to immunize rabbits, and the specific antibodies from the final bleeding were purified from the serum by affinity chromatography to the immunizing peptide, which had been coupled to an activated Sepharose column (Genosys Biotechnologies, Cambridge, United Kingdom). The resulting purified antibody fraction reacted with the immunizing peptide but showed no cross-reaction with S. aureus proteins, as determined in preliminary control blots. The preimmune serum obtained from the rabbits was tested in a control assay and yielded no signal (results not shown).

Complementation of the S gene lysis defect in gt11. E. coli host cells LE392 and HB101 were grown in Luria-Bertani media supplemented with 0.2% maltose and 10 mM MgSO₄ (33). The PCR-generated hol187 genes were ligated (out of frame with lacZ) into the EcoRI site of gt11 (5, 16, 22) under the control of P_lac. Phage genomes were packaged according to the instructions of the manufacturer (Packagene system; Promega), and aliquots containing the recombinant virus particles were plated on soft-agar plates with LE392 or HB101 as hosts. A negative control (gt11) and a positive control ( wild type) were also plated. Following incubation for 16 h at 40°C, 10 individual plaques of each of the recombinant phages from the HB101 lawns were picked. The phages were eluted in 200 µl of SM buffer (33) and treated with a drop of chloroform, and the supernatant was replated on HB101. Again, single plaques were picked and the phages were eluted. The presence of the desired hol187 genes and mutations in the phage clones was verified by PCR amplification of the insert and subsequent nucleotide sequencing (results not shown).

Nucleotide sequence accession number. The DNA sequence reported here appears in the EMBL, GenBank, and DDBJ databases under accession no. Y07740.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Cloning and sequencing of phage 187 lysis genes. DNA purified from S. aureus phage 187 was shotgun-cloned into the inducible expression vector pSP72. Colonies of E. coli expressing a lytic activity could be readily identified by the clearing zones when overlaid with a lawn of S. aureus cells. Nucleotide sequencing of the plasmid insert and subsequent computer analysis allowed the identification of a putative open reading frame. However, we found that the sequence was incomplete, since no stop codon was present on the cloned fragment. Therefore, the remaining nucleotide sequence was determined directly from phage DNA (Fig. 1A). The ply187 gene (1,887 nucleotides) is preceded by an RBS (5'-GAAGTGAT-3') with high sequence similarity to that of the 3' end of the S. aureus 16S rRNA gene (5'-GAGGTGAT-3' [26]). An inverted sequence repeat (stem-loop structure) is present at the immediate 3' end of the gene (G = 35 kJ/mol), which includes the TAA stop codon and probably functions as a transcriptional terminator.

View larger version (23K): [in this window] [in a new window]   FIG. 1.   (A) Genetic map of the S. aureus phage 187 lysis gene region. The indicated AluI fragment was initially cloned and sequenced on plasmid pPL187, and the remaining nucleotide sequence was determined directly from phage DNA. (B) The holin gene hol187 is fully embedded in the 5'-terminal part of the endolysin gene ply187, in a different reading frame (+1). Deduced amino acid sequences of both gene products (Hol and Ply [partial]) are shown below the nucleotide sequence.

Ply187 is related to major staphylococcal autolysins. The complete ply gene product consists of 628 amino acids, has a calculated molecular mass of 71.6 kDa, and has a predicted pI of 9.9. Computer analysis showed no potential signal peptide in Ply187. Comparison of its amino acid sequence to those of other proteins included in some of the current databases (EMBL and GenBank) is presented in Fig. 2. The amino terminus of Ply187 shows 31% identity and 73% similarity over 113 amino acids (aa) to the N terminus of LytA from prophage 11 of the lysogenic S. aureus strain NCTC 8325 (42), which is an N-acetylmuramoyl-L-alanine amidase (EC 3.5.1.28). Extensive homology was observed between the central-to-C-terminal domain of Ply187 and the C termini of the major autolysins Atl-A of S. aureus (11, 30) and Atl-E of Staphylococcus epidermidis (15). Both proteins show significant sequence identity (46 and 51%, respectively) and similarity (80 and 78%, respectively) to Ply187. Both Atl enzymes are first synthesized as single polypeptides equipped with signal peptides and propeptides of various sizes. However, the Atl enzymes are then processed to yield two functional lytic enzymes: a 60-kDa protein with amidase activity and a 51-kDa portion representing the C-terminal portion of the initial gene product with probable glucosaminidase activity (11, 15, 30). By deducing the sequence homologies to LytA and especially to the Atl enzymes, two domains were tentatively assigned to Ply187: the N terminus represents a probable amidase (17.8 kDa), and the C-terminal portion (53.7 kDa), whose function is unknown, might bear glucosaminidase activity (see Fig. 3). However, as reported below, no staphylolytic activity could be observed when the latter Ply187 domain was separately expressed and tested in the overlay assay.

View larger version (46K): [in this window] [in a new window]   FIG. 2.   (A) Alignment of the N-terminal amino acid sequences of Ply187 and LytA, an amidase endolysin from the S. aureus phage 11. Identical amino acid residues and conservative replacements are indicated by vertical lines and colons, respectively. Homologous regions are shown in boldface. (B) Homology of the C-terminal domains of Ply187 and the major autolysins of S. aureus (Atl-A) and S. epidermidis (Atl-E). A few gaps have been introduced to optimize alignment.

Enzymatic activity of Ply187 is located in the N terminus. The initially observed lytic activity resulted from the expression of only a fragment of ply187 on plasmid pPL187. The cloned fragment consists of 891 bp and enabled the synthesis of a polypeptide corresponding to the N-terminal 297 aa (33.9 kDa) of Ply187, which represents only 47% of the actual enzyme. This result strongly suggested that the catalytic site is located in the N terminus of Ply187. To further support this hypothesis, individual lytic activities of different fragments of Ply187 were determined by overproduction in E. coli (Fig. 3). Expression of the complete enzyme from plasmid pPL187-F3 yielded only relatively weak activity (small lysis zones), compared with the larger lytic zones observed around colonies expressing a fragment corresponding to the initially cloned 891-bp fragment (pPL187-F2). Surprisingly, the strongest activity resulted from the expression of the smallest fragment, i.e., the N-terminal 157 aa of Ply187 (pPL187-F1). In contrast, the polypeptides corresponding to amino acids 158 to 297 of the native enzyme (pPL187-F4) and amino acids 158 to 628 (pPL187-F5) produced no visible lysis in the overlay assays with S. aureus or M. luteus. These results indicated that the N terminus contains the only peptidoglycan-hydrolyzing activity of Ply187.

View larger version (18K): [in this window] [in a new window]   FIG. 3.   Determination of the enzymatically active domain of Ply187. PCR cloning was used to express the native ply187 gene and four fragments derived therefrom in E. coli. The solid bars represent the enzymatically active amidase domains, and the dotted bars indicate the C-terminal domains. The insert in pPL187-F2 corresponds to the DNA fragment initially cloned on plasmid pPL187.

A holin-like gene is fully embedded in ply187. The small reading frame found within the part of the endolysin gene corresponding to the N terminus translates into a putative protein of 6.4 kDa (Fig. 1B and 4), with a predicted pI of 6.1. Its sequence reveals several properties characteristic of the holin protein family (45): (i) two stretches of generally hydrophobic amino acids with a neutral net charge (putative transmembrane helices); (ii) a hydrophilic beta-turn linker separating the membrane-spanning domains; and (iii) a dual-start motif (5'-ATG-TTG-ATG-...-3'), permitting the possible expression of two polypeptides of different lengths (57 and 55 amino acids, respectively). The distinct domains evident from the hydrophobicity plot (Fig. 4A) correlate well with the proposed secondary structure (Fig. 4B). Interestingly, the hol187 gene product lacks a highly charged C-terminal domain, present in most other holins identified to date (45). Moreover, its coding sequence is fully embedded (on the same strand) in a different reading frame (+1) within the endolysin gene, which has not previously been found for any of the known or suspected holin genes.

View larger version (27K): [in this window] [in a new window]   FIG. 4.   (A) Hydrophobicity analysis graph of Hol187, according to the algorithm of Kyte and Doolittle (20). Hydrophobic regions are in the area above zero, and the central hydrophilic domain is indicated in the area below zero. The mean hydrophobicity index is 0.97, which strongly indicates a membrane location. (B) Simplified schematic representation of the predicted secondary structure of Hol187 (antiparallel helices). Charged amino acids are indicated with plus and minus signs, respectively.

Hol187 is found in the cytoplasmic membrane of phage-infected cells. The lysis of S. aureus host cells infected by phage 187 is relatively slow (Fig. 5); cultures start to decrease in optical density at approximately 75 min postinfection and continue to clear during the following hour. Immunoblotting with specific peptide antibodies directed against Hol187 indicated that the protein is synthesized in S. aureus cells upon infection with phage 187 and enabled the detection of the protein in membrane fractions of these cells, whereas uninfected control cells yielded no signal. Since relatively large amounts of membrane extracts were necessary to produce clear signals, it appears that the protein is present in the cells in very small amounts.

View larger version (36K): [in this window] [in a new window]   FIG. 5.   (Upper panel) Time course of lysis of S. aureus cells infected with phage 187 at time zero, presented as optical density of the cultures. (Lower panel) Spatial appearance of Hol187 in membrane fractions of phage infected cells as revealed by immuno-dot blotting in control cells (A) and phage-infected cells (B). Numbers indicate the time in hours at which the individual samples were taken.

Synthesis of Hol187 in E. coli causes growth retardation and membrane damage. The expression of hol187 from pHOL187 in E. coli JM109 following induction of the gene caused severe growth impairment, i.e., no further increase in the optical density of the cultures compared to the control. The same effect was observed with cells carrying a mutated hol187 gene (pHOL187-S), where the first two amino acids were deleted. Expression of only the longer, 57-aa product from plasmid pHOL187-L had a somewhat less detrimental effect. The introduction of two negatively charged amino acids into the first putative transmembrane domain of the protein (pHOL187-Ch) completely abolished the protein function; the culture afterward grew normally (results not shown).

View larger version (73K): [in this window] [in a new window]   FIG. 6.   (Upper panels) Effect of hol187 expression on total cellular ATP content (left), and ATP release into growth medium (right), by E. coli cells harboring either pBluescript (control) or pHOL187. Gene expression was induced at time zero, and samples were taken and analyzed at the indicated time points. (Lower panels) Effect of the synthesis of Hol187 and mutated proteins on the release of -galactosidase from E. coli W3110 cells into the medium (indicated by the blue zones around colonies) during growth on agar plates containing the chromogenic substrate X-Gal. (A) Control (W3110 without plasmid); (B) W3110(pHOL187); (C) W3110(pHOL187-S); (D) W3110(pHOL187-L); (E) W3110(pHOL187-Ch).

hol187 can substitute for  S. Propagation (i.e., plaque formation) of gt11 requires a host featuring supF, to compensate for the Sam100 mutation. To determine whether Hol187 could cause release of the R gene product into the periplasm, we tested the ability of recombinant gt11 to form plaques on the nonpermissive host HB101. Results are presented in Table 2 and clearly show that complementation of the defective S allele with functional Hol187 (native, M1L2, and M3L) allowed plaque formation. Again, gt11::hol187-S produced somewhat larger plaques than gt11::hol187-L, whereas the V7D G8D mutation in gt11::hol187-Ch did not support R-mediated cell lysis. These data strongly suggest that Hol187 does, in fact, function as a holin.

View this table: [in this window] [in a new window]   TABLE 2.   Ability of Hol187 to complement the lysis defect (Sam100) of gt11

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Many peptidoglycan-hydrolyzing (lytic) enzymes are composites of specifically adapted modules, encoding substrate recognition and hydrolysis domains. Such a modular organization was shown or proposed for several investigated lysins from phages or bacteria, where the catalytic activity is (almost always) located in the N-terminal region, while the C-terminal part contains the target-specific binding domains (4, 18, 22, 23, 24, 25, 35, 40, 41). It is shown here that the N-terminal domain (25%) of Ply187 contains the enzymatically active site, which, as deduced from its convincing homology to LytA, may act as an N-acetylmuramoyl-L-alanine amidase. Although the large C-terminal domain of Ply187 shows homology to the proposed glucosaminidase domains of the Atl proteins, the respective Ply187 fragments revealed no lytic activity. Moreover, C-terminal deletion of the enzyme (up to 75%) strongly increased lytic activity in the overlay assays. Thus, our results suggest that the C terminus does not represent an essential substrate recognition and/or binding domain, since it seems to be dispensable for staphylolytic activity. Therefore, the actual in vivo function of this large domain of Ply187 remains to be investigated. In this context, it would be interesting to clarify whether Ply187 exerts its in vivo activity as a full-length polypeptide or, as do the Atl enzymes, might also be posttranslationally processed to yield two polypeptides. A site similar to the one which is proteolytically cleaved in AtlA (Ala⁷⁷⁶ [30]) can also be tentatively identified in Ply187 (Ala¹⁵⁸) by amino acid sequence alignment (results not shown). This would separate the proposed amidase domain from the large C-terminal part but would still leave open the hypothetical function of the latter.

In several cases, lytic phage enzymes are thought to be closely related to the lytic enzymes harbored by their individual host bacteria (autolysins and others), since extensive sequence similarities were found in specific domains (functional modules) of the respective enzymes (13, 22, 23, 35). With Ply187, we found striking homology to major staphylococcal cell wall hydrolases, i.e., the Atl autolysins. Therefore, it is tempting to speculate that the different domains (i.e., genetic modules) of Ply187 could have been acquired or exchanged through horizontal gene transfer between phage DNA and the bacterial genome.

We identified the small hol187 coding sequence, translated from a different reading frame, fully embedded within ply187. Holin genes are generally found to be closely associated with the endolysin sequences on the phage genomes (i.e., directly upstream of the peptidoglycan hydrolase gene), and they may overlap the latter by a few base pairs to allow translational coupling to occur (22, 44). The observation that hol187 is fully embedded out of frame within the endolysin gene on the same strand is an unusual finding. In fact, coding sequences entirely encompassed within other genes seem to be very rare among life forms having double-stranded DNA as genetic material. A relevant example is  Rz, which contains another small gene (Rz1) translatable from a different reading frame (14). Although Rz1, a small lipoprotein localized in the outer membrane of phage infected E. coli cells, may somehow be involved in the lysis event (19), it is clearly not a holin.

We have shown that Hol187 is synthesized in phage-infected S. aureus cells prior to cell lysis as a late gene product and that it is present in the cytoplasmic membrane. The best indication, however, that Hol187 functions as a holin is that it is able to complement a defective  S allele and allows propagation and plaque formation by recombinant phages in the nonsuppressive HB101 background. Moreover, the gene product shows all properties characteristic of the group of small, hydrophobic proteins termed holins (44), except that it is the first example described that lacks the highly charged C terminus common to other holins (45). It was previously thought that this region is crucial for correct orientation and/or oligomerization in the cytoplasmic membrane. However, the results of a recent study on C-terminally truncated  S holin indicated that the hydrophilic C terminus is actually nonessential (32). Hol187 belongs to the class II holins (with only two possible membrane-spanning domains) and is the smallest holin protein described to date. The introduction of a negative charge into the first proposed transmembrane domain of Hol187 resulted in loss of activity. This supports the present model, which suggests that the overall net charge of the membrane-spanning domains must be neutral.

The hol187 gene reveals two possible translational start codons (ATG), separated by a TTG encoding a leucine residue. It is interesting that the latter codon itself may also serve as a translational start, which is not uncommon in staphylococcal genes (16, 28, 31, 39) and has recently been found for the (entirely different) holin gene of S. aureus phage Twort (22). However, it seems unlikely that the dual ATG codons of hol187 are coincidental, because this motif is a frequent and characteristic feature of many holin genes (8). When the mutated hol187 variants were expressed in an E. coli background, we observed that the full-length polypeptide (Hol187-L; 57 aa) may be somewhat less active than the shorter product translated from the second possible start codon (Hol187-S; 55 aa). In the best studied model of holin function (phage ), this difference seems to be more pronounced: the longer S-107 polypeptide inhibits membrane hole formation and was reported not to cause lysis by itself (7, 9). However, its independent overexpression from a high-copy-number plasmid under the control of the lac promoter also resulted in cell lysis (36). While the opposing nature of the two S variants is explained by different N-terminal net charges, the two possible hol187 products are separated by an uncharged leucine residue. At present, it is entirely unclear how Hol187 function may be regulated in its natural host.

Compared to  S, Hol187 is relatively slow acting. Low-level background expression from plasmid vectors is slightly inhibitory but not strictly lethal for the cells, whereas promoter induction results in membrane lesions and cell starvation. In phage-infected S. aureus cells, lysis begins at approximately 75 min and continues up to 150 min. Compared to -infected E. coli cells, this is more than twice the time needed to complete an infective cycle. This may simply reflect the different timing of lysis, which should not terminate the infective cycle before particle morphogenesis is complete. Nevertheless, we have shown here that Hol187 functions as a holin: it causes release of intracellular molecules into the periplasm and, most notably, is able to substitute for  S, i.e., it promotes access of the heterologous R-gene product to the E. coli murein.

In conclusion, we found compelling evidence that the embedded hol187 gene encodes a holin-like protein, whose probable function is to permeabilize the staphylococcal cytoplasmic membrane to permit access of the Ply187 amidase endolysin to the cell wall. Hol187 is novel in more than one respect and joins the 11 apparently unrelated holin families currently known (45). These findings indicate that there is an even greater heterogeneity of holin genes (i.e., in gene families) than previously established. Even though the primary amino acid sequences of holins are totally dissimilar, their functional characteristics are well conserved.