Identification and molecular characterization of an N-acetylmuramyl-l-alanine amidase Sle1 involved in cell separation of Staphylococcus aureus


  • Junko Kajimura,

    1. Department of Bacteriology, Hiroshima University Graduate School of Biomedical Sciences, 1-2-3 Kasumi Minami-ku, Hiroshima City, Hiroshima 734-8553, Japan.
    Search for more papers by this author
  • Tamaki Fujiwara,

    1. Department of Bacteriology, Hiroshima University Graduate School of Biomedical Sciences, 1-2-3 Kasumi Minami-ku, Hiroshima City, Hiroshima 734-8553, Japan.
    Search for more papers by this author
  • Sakuo Yamada,

    1. Department of Microbiology, Kawasaki Medical School, Matsushima 577, Kurashiki, Okayama 701-0192, Japan.
    Search for more papers by this author
  • Yoshika Suzawa,

    1. Department of Bacteriology, Hiroshima University Graduate School of Biomedical Sciences, 1-2-3 Kasumi Minami-ku, Hiroshima City, Hiroshima 734-8553, Japan.
    Search for more papers by this author
  • Tetsuya Nishida,

    1. Department of Bacteriology, Hiroshima University Graduate School of Biomedical Sciences, 1-2-3 Kasumi Minami-ku, Hiroshima City, Hiroshima 734-8553, Japan.
    Search for more papers by this author
  • Yoshihiro Oyamada,

    1. Microbiology Group, Pharmacology and Microbiology Research Laboratories, Dainippon Pharmaceutical, Enoki-Cho 33-94, Suita City, Osaka 564-0053, Japan.
    Search for more papers by this author
  • Ikue Hayashi,

    1. Research Facility, Hiroshima University Faculty of Dentistry, 1-2-3 Kasumi Minami-ku, Hiroshima City, Hiroshima 734-8553, Japan.
    Search for more papers by this author
  • Jun–ichi Yamagishi,

    1. Microbiology Group, Pharmacology and Microbiology Research Laboratories, Dainippon Pharmaceutical, Enoki-Cho 33-94, Suita City, Osaka 564-0053, Japan.
    Search for more papers by this author
  • Hitoshi Komatsuzawa,

    1. Department of Bacteriology, Hiroshima University Graduate School of Biomedical Sciences, 1-2-3 Kasumi Minami-ku, Hiroshima City, Hiroshima 734-8553, Japan.
    Search for more papers by this author
  • Motoyuki Sugai

    Corresponding author
    1. Department of Bacteriology, Hiroshima University Graduate School of Biomedical Sciences, 1-2-3 Kasumi Minami-ku, Hiroshima City, Hiroshima 734-8553, Japan.
      E-mail; Tel. (+81) 82 257 5635; Fax (+81) 82 257 5639.
    Search for more papers by this author

E-mail; Tel. (+81) 82 257 5635; Fax (+81) 82 257 5639.


We purified a peptidoglycan hydrolase involved in cell separation from a Staphylococcus aureus atl null mutant and identified its gene. Characterization of the gene product shows a 32 kDa N-acetylmuramyl-l-alanine amidase that we designated Sle1. Analysis of peptidoglycan digests showed Sle1 preferentially cleaved N-acetylmuramyl-l-Ala bonds in dimeric cross-bridges that interlink the two murein strands in the peptidoglycan. An insertion mutation of sle1 impaired cell separation and induced S. aureus to form clusters suggesting Sle1 is involved in cell separation of S. aureus. The Sle1 mutant revealed a significant decrease in pathogenesis using an acute infection mouse model. Atl is the major autolysin of S. aureus, which has been implicated in cell separation of S. aureus. Generation of an atl/sle1 double mutant revealed that the mutant cell separation was heavily impaired suggesting that S. aureus uses two peptidoglycan hydrolases, Atl and Sle1, for cell separation. Unlike Atl, Sle1 is not directly involved in autolysis of S. aureus.


The major component of the Gram-positive bacterial cell wall is peptidoglycan composed of glycan strands cross-linked by oligopeptides forming a meshwork sacculus that encloses the cell giving it a rigid framework against internal pressure. Peptidoglycan is a dynamic structure maintaining a balance between peptidoglycan biosynthesis and degradation. Staphylococcus aureus is an important pathogen in suppurative infections and a target for chemotherapy. As such, it is a model microorganism for Gram-positive bacteria to study the structure and function of the cell wall. The S. aureus peptidoglycan consists of a peptide l-alanyl-d-isoglutamyl-l-lysyl-d-alanine (l-Ala-d-iGln-l-Lys-d-Ala) in which most of the epsilon amino groups of lysine contain pentaglycine substituents, where L-Ala is linked via an amide bond to a d-lactyl group of muramic acid, a repeating disaccharide, N-acetylmuramic acid-(β1,4)-N-acetylglucosamine (MurNAc-GlcNAc). The peptide is unusual consisting of both D- and L- amino acids forming unique L-D peptide bonds in the peptidoglycan. Cross-linking the glycan strands is achieved by linking the carboxyl of D-Ala at position four in the peptide to the free amino group of a penta-glycine cross-bridge extending from L-Lys to another peptide protruding from the neighbouring glycan strand.

Bacteria produce several peptidoglycan hydrolases (PGHs) that are involved in the degradation of the peptidoglycan. The possible physiological functions include nicking the peptidoglycan for insertion of new monomers involved in remodelling and turnover of the peptidoglycan during cell growth, division and separation (for reviews see ReferencesWard and Williamson, 1984; Shockman and Höltje, 1994; Sugai, 1997). Some of these maintenance enzymes are autolysins disintegrating the peptidoglycan when cells are placed in unfavourable conditions that lead to autolysis (Shockman and Barret, 1983; Ward and Williamson, 1984; Shockman and Höltje, 1994; Sugai, 1997). The principal PGH of S. aureus is identified as a bifunctional autolysin called Atl (Oshida et al., 1995). Atl is initially produced as a 138 kDa protein having an amidase domain and a glucosaminidase domain. It undergoes proteolytic processing to generate two major PGHs: a 62 kDa N-acetylmuramyl-l-alanine amidase (AM) and a 51 kDa N-acetylglucosaminidase (GL) (Oshida et al., 1995; Komatsuzawa et al., 1997). Atl may be involved in cell separation of daughter cells after cell division based on the following observations: Atl and its processed proteins were shown to localize on the cell surface at the septal region of an upcoming cell division site (Yamada et al., 1996); AM and GL are able to disperse cell clusters of S. aureus into single cells without affecting cell viability (Sugai et al., 1995); and a null mutant of atl, the gene for Atl, grew in clusters (Takahashi et al., 2002). Atl is also shown to act as an autolysin in penicillin-induced autolysis of S. aureus (Sugai et al., 1997a). Mutants (Lyt) defective in the production of Atl have been isolated (Mani et al., 1994). The Lyt mutant reveals only one bacteriolytic band at ∼32 kDa observed using S. aureus zymograph gels. The authors suggest the 32 kDa lytic enzyme is vital for cell morphogenesis. We also confirmed that the atl null mutant exhibited a single lytic band at ∼32 kDa in S. aureus zymograph gels (Takahashi et al., 2002). Using the Lyt mutant, a gene encoding a 31.4 kDa PGH has been cloned and is designated as lytM (Ramadurai and Jayaswal, 1997). However, whether the LytM represents the PGH showing a single 32 kDa lytic band in the Lyt mutant remained elusive as the lytM knockout mutant is not presently available.

In this paper we purified a PGH showing a single 32 kDa lytic band from the atl null mutant and identified the gene. Characterization of the gene product revealed that it is not the LytM gene product but an N-acetylmuramyl-l-alanine amidase we designated as Sle1. Mutation of sle1 inhibited the cell separation and induced S. aureus to form clusters suggesting that Sle1 is also involved in cell separation of S. aureus. Our results suggest that S. aureus uses two peptidoglycan hydrolases, Sle1 and Atl for cell separation after cell division.


Purification of the 32 kDa staphylolytic enzyme

Zymographic analysis of cell surface-associated proteins extracted with 4% SDS using S. aureus cell gels revealed multiple bacteriolytic bands at 138, 115, 85, 62 and 32 kDa. Previous studies show that most of these bands except for the 32 kDa band are derivatives of the gene products from atl, a major autolysin gene of S. aureus. The atl deletion mutant possessed a 32 kDa bacteriolytic protein indicating the gene for the 32 kDa bacteriolytic protein is distinct from the atl gene products. Therefore, we attempted to purify this staphylolytic 32 kDa protein and characterize it. Preliminary studies revealed that a small amount of the activity was secreted into the culture supernatant of an overnight culture; therefore, we used the culture supernatant of the atl null mutant JT1413 as the purification source. A 3 ml slurry of dye-ligand affinity gel (TSKgel Blue-Toyopearl 650 ML, Tosoh, Tokyo, Japan) was added to the 1 l culture filtrate of S. aureus JT1413 and the suspension was gently stirred for 12 h at 4°C. The staphylolytic activity does not bind to the Blue-Toyopearl. However, this procedure is important to remove contaminating protein that will prevent purification of the target protein in a later step. The unbound protein was recovered by centrifugation at 6000 g and the supernatant was adjusted to approximately 2 M NaCl. The salted supernatant was passed through a TSKgel Butyl Toyopearl 650 ML (15 × 95 mm) column equilibrated with 0.1 M phosphate buffer containing 2 M NaCl (pH 6.0)(buffer 1). The column was washed with buffer until most of the unbound proteins passed through and the staphylolytic activity was bound to the column. The column was first eluted with 0.1 M phosphate buffer (pH 6.0) without salt, then with 4% SDS. The stapholytic activity was not eluted with the buffer and eluted with 4% SDS. This active fraction was electrophoresed in an acrylamide tube gel using a disc preparative SDS-PAGE apparatus overnight. Zymographic analysis of the eluted fractions revealed fractions 25–28 were positive for the 32 kDa bacteriolytic activity (Fig. 1B). Protein analysis of the corresponding fractions using SDS-PAGE recognized a single protein band in fraction 25 with an estimated molecular mass of 32 kDa by Coomassie brilliant blue staining (Fig. 1A). From a 6 l culture having an initial 33.9 mg of protein and 8483 U of activity, we purified 10.7 µg of protein with 823 U of activity as the final product. The purification yield was 9.7% and the increased purification factor was 307.7. The amino-terminal sequence of the purified protein identified 14 residues: ATTHTVKPGESVVA. A computer-assisted analysis of the deduced amino acid sequence with the GenBank using the blast program identified an S. aureus open reading frame (ORF) with 1002 bp, coding for a hypothetical protein with possible amidase function of 334 amino acid residues with an apparent molecular mass of 35.8 kDa and a pI of 9.93. The deduced amino acid sequence exactly matched the sequence starting at the Ala residue at position 26 of the ORF and we designated it Sle1. Sle1 was ubiquitously found in all reported genome sequences of S. aureus including strains N315, Mu50 and MW2.

Figure 1.

Purification of Sle1 from the atl null mutant (JT1413).
A. Coomassie blue stained 12% SDS-PAGE gel of fractions from disc preparative SDS-PAGE from the concentrated culture supernatants of the atl null mutant (JT1413).
B. Zymographic analysis of fractions using S. aureus-gels.

Sequence analysis

The first 25-amino-acid residues of Sle1 appears to correspond to a signal peptide (SP) where the cleavage site is probably located between Ala and Ala using previous reports (von Heijne, 1986). This assumption is in complete agreement with the deduced N-terminal sequence of the 32 kDa protein demonstrating Sle1 possesses an SP characteristic of a secreted protein and consistent with the 32 kDa protein being secreted into the growth medium. The processed protein has a calculated molecular mass of 33.4 kDa and a pI of 9.84. A schematic representation of the primary structure of the Sle1 is shown in Fig. 2. Self-alignment of the deduced sequence of Sle1 revealed the presence of three repeat sequences (R1, R2 and R3) in the N-terminal region showing a significant similarity to the peptidoglycan binding motif of LysM (Bateman and Bycroft, 2000). LysM was initially identified as a module in a bacteriolytic enzyme in prokaryotes and is now recognized as a widespread protein module present in both prokaryotes and eukaryotes. Similarity searches for the C-terminal end with the predicted coding sequences present in the public databases using the psi-blast, blast and fasta algorithms revealed the presence of ORFs possessing similar sequences with unknown functions in S. aureus and in other Gram-positive bacteria including Staphylococcus epidermidis, Staphylococcus carnosus, Streptococcus pyogenes, Streptococcus thermophilus, Streptococcus pneumoniae, Streptococcus cremoris, Lactococcus lactis and Enterococcus faecalis. Furthermore, several other potential ORFs were identified when the search was extended to the available ‘unfinished’ bacterial genome sequence database in TIGR. Among them, several have a similar size and conform to the modular structure possessing repetitive sequences of the LysM domain in their N-terminals suggesting that homologues are present in other Gram-positive bacterial species. For instance, Heilmann et al. (Heilmann et al., 2003) reported a 35 kDa protein Aae with bacteriolytic activity from S. epidermidis. Aae shows 65% identity and 72% homology to Sle1.

Figure 2.

Domain organization of Sle1.
A. The diagrammatic representation of Sle1 protein is illustrated. The C-terminal of Sle1 has bacteriolytic activity, and three repeats (shaded, R1, R2 and R3) in the N-terminal show homology to the LysM domain.
B. The sequences of R1, R2 and R3 are compared with LysM motifs of the E. coli membrane-bound lytic murein transglycosylase D (Bateman and Bycroft, 2000).

The sequence comparisons discussed above suggested that the C-terminal portion is the catalytic domain. To test this hypothesis, His × 6-tagged 128 C-terminal amino acid residues and 180 N-terminal amino acid residues were generated using the Escherichia coli XL-II blue plasmid. These fusion proteins were purified using Ni-resin and assayed for bacteriolytic activity using zymography. As shown in Fig 3C and D, only the His × 6-tagged C-terminal portion revealed activity and was as potent as His × 6-tagged full-length Sle1. Whereas, the His × 6-tagged N-terminal portion appeared as a whitish precipitate without activity. This suggested the C-terminal part has the catalytically active domain and the N-terminal LysM domain is not essential for the bacteriolytic activity in zymography. Moreover, the N-terminal truncated protein did not show any lytic activity in a Micrococcus luteus-gel suggesting that the truncated protein may have substrate specificity (data not shown).

Figure 3.

Identification of the catalytic domain of Sle1.
A. The diagrammatic representation of truncated Sle1 proteins are illustrated. His-Sle1 is the signal peptide-processed form of Sle1 with a N-terminal His-tag. His-Sle1-N is His-Sle1 with a C-terminal 128-amino-acid deletion. His-Sle1-C is His-tagged C-terminal 128 amino acid.
B. Coomassie blue stained 10% SDS-PAGE of His-Sle1 (lane 1), His-Sle1-N (lane 2) and His-Sle1-C (lane 3).
C. Bacteriolytic activity of His-Sle1 (lane 1), His-Sle1-N (lane 2) and His-Sle1-C (lane 3) in the zymogram S. aureus-gel.
D. Dose dependency of bacteriolytic activity of His-Sle1 (square), His-Sle1-N (lozenge) and His-Sle1-C (triangle) in the zymogram S. aureus-gel. Activity was expressed as arbitrary unit. Determination of staphylolytic activity was carried out as described previously (Sugai et al., 1991).

Identification of the 32 kDa protein as N-acetylmuramyl-l-alanine amidase

The purified peptidoglycan from S. aureus was incubated with the purified His × 6-tagged Sle1 (full length), and the appearance of new free amino groups and reducing sugars during enzymatic hydrolysis was monitored (Fig. 4A). An increase in the concentration of free amino groups together with the decrease in turbidity suggested that the Sle1 is an amidase or an endo-peptidase (Fig. 4A). To determine the N-terminal amino acids generated during the lytic reaction, the supernatant of the reaction mixture was incubated with DNFB, followed by hydrolysis with 4 N HCl. A high-performance liquid chromatography (HPLC) analysis of the DNP-labelled and hydrolysed amino acids identified 2,4-dinitrophenyl-alanine in the Sle1-treated sample (Fig. 4B). To determine the cleavage site, we first compared the muropepetides generated by double digestions with mutanolysin and Sle1 and those with mutanolysin and the 62 kDa amidase (AM of Atl) (Fig. 5). Muropeptides obtained by mutanolysin digestion were separated by reverse-phased HPLC on ODS-Hypersil as described previously (Stranden et al., 1997; Roos et al., 1998) and each peak was subjected to matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF-MS) analysis where the first peak generated an ion with m/z 1276, a measurement that is in close agreement with the structure of a peptidoglycan cleavage fragment GlcNAc-(β1,4)-MurNAc-(l-Ala-d-iGln-l-Lys-(NH2-Gly5-d-Ala)-d-Ala-COOH) corresponding to M1 as described previously (Stranden et al., 1997) and the second peak generated an ion with m/z 2440, which corresponds to M2. In the same way, each fragment was assigned an oligomer as shown in Fig. 5A. The chromatogram of the doubly digested peptidoglycan with mutanolysin and the Sle1 shows a peak corresponding to dimers decreased significantly and a new peak (*) with shorter retention time appeared (Fig. 5B). Otherwise, there was not much difference in the chromatographic pattern between mutanolysin-digested peptidoglycan and that of the doubly digested one by mutanolysin and the Sle1. This observation was reproducible even if the amount of the Sle1 was increased in the reaction mixture. Whereas, the chromatogram of peptidoglycan digested with mutanolysin and 62 kDa amidase revealed a significant decrease in the peaks corresponding to peaks appearing in the mutanolysin-digested peptidoglycan and appearance of several new peaks (Fig. 5C). These results strongly suggest that the mode of action of Sle1 was distinct from that of the 62 kDa amidase. The new peak (*) appeared in the chromatogram of doubly digested peptidoglycan with mutanolysin and Sle1 was analysed by MALDI-TOF-MS. The new peak generated ions with m/z 1250, 1307, 1364 and 1421 (Fig. 6A). Muramyl peptides were generated readily from alkali metal adducts during MALDI where the [M + Na+] ion is the predominant species. Therefore, the molecular mass of each peptide can be deduced from the mass of the [M + Na+] ion. The calculated masses of these could be explained as peptides with the structures: (l-Ala-d-iGln-l-Lys-(Gly5)-d-Ala)-(l-Ala-d-iGln-l-Lys-(Gly)-d-Ala) -d-Ala: (L-Ala-d-iGln-l-Lys-(Gly5)-d-Ala)-(l-Ala-d-iGln-l-Lys-(Gly2)-d-Ala)-d-Ala; (l-Ala-d-iGln-l-Lys-(Gly5)-d-Ala)-(l-Ala-d-iGln-l-Lys-(Gly3)-d-Ala)-d-Ala; and (l-Ala-d-iGln-l-Lys-(Gly5)-d-Ala)-(l-Ala-d-iGln-l-Lys-(Gly4)-d-Ala)-d-Ala to the respective m/z (Fig. 6B). The MALDI-TOF-MS data strongly suggests that Sle1 may cleave mutanolysin-released peptidoglycan between the glycan chain and peptide although the digestion products are apparently different from those of mutanolysin and the 62 kDa amidase as shown in Fig. 5B and C. This is consistent with the increase of 2,4-dinitrophenyl-alanine by Sle1 digestion of the peptidoglycan (Fig. 4B). Taken together these results suggest Sle1 acts as an N-acetylmuramyl-l-alanine amidase.

Figure 4.

Time course digestion of S. aureus peptidoglycan with His-Sle1 and the measurement of the DNP-labelled amino acid.
A. Measurement of turbidity, amount of free amino groups and reducing sugars. Control (filled circle); His-Sle1 (open square).
B. Determination of N-terminal amino acids. The products resulting from the enzyme reaction was incubated with 1-fluoro-2,4-dinitrobenzene and the labelled compounds were hydrolysed with a high concentration of HCl followed by separation using reverse phase chromatography as described in the Experimental procedures. Only DNP-l-alanine increased during the enzyme reaction. Control (white); His-Sle1 (black).

Figure 5.

Comparison of muropeptides generated by digestion of purified S. aureus peptidoglycan using mutanolysin, mutanolysin-Sle1 and mutanolysin-62 kDa amidase. S. aureus peptidoglycan was incubated with (A) mutanolysin, (B) mutanolysin and Sle1, or (C) mutanolysin and 62 kDa amidase. The solubilized peptidoglycan was separated on a Hypersil C18 reverse-phase HPLC column (see Experimental procedures).

Figure 6.

MALDI-TOF-MS analysis of the peak (Fig. 5B*).
A. The compound corresponding to the purified peak (Fig. 5B*) of reverse phase HPLC was subjected to MALDI TOF-MS analysis.
B. Diagram of the peptidoglycan subunits with masses of 1227, 1284, 1341 and 1398.

Construction and the phenotype of the sle1 mutant

To analyse the function of Sle1, we attempted to disrupt the sle1 in the genome of RN450 using insertion inactivation with a 900-base internal sle1 fragment cloned into the thermo-sensitive S. aureusE. coli shuttle vector pYT1. The pYT1 was cut using BamHI and HindIII at the multicloning site and the 900-base internal sle1 fragment was introduced to generate pLyt36. S. aureus RN450 was transformed with pLyt36 at 30°C and the transformants were shifted to 42°C in the presence of tetracycline to induce a single cross-over. The colonies were screened for the loss of the 32 kDa bacteriolytic band using S. aureus peptidoglycan-containing zymography. One of the mutants was designated as Lyt36 and the successful inactivation of sle1 was confirmed using Southern blot analysis. To assess if the mutant phenotype was attributed to the inactivation of sle1, we performed complementation of the sle1 mutation with a plasmid carrying a DNA fragment containing the sle1 locus. Amplification by polymerase chain reaction (PCR) generated an 1.1 kb DNA fragment containing the sle1 locus, which was cloned into the pGEM-T easy plasmid vector, and a BamHI-EcoRI digestion DNA fragment was subcloned into the S. aureusE. coli shuttle vector pCL15 to generate pJT1131. Lyt36 was transformed with pJT1131 by electroporation. One of the transformants designated JT1482 was further characterized. We analysed zymograms using a 4% SDS extract of RN450, the sle1 mutant Lyt36 and sle1-complemented Lyt36, JT1482. In polyacrylamide gels containing S. aureus cells, the 32 kDa bacteriolytic enzyme activity was not detected in the Lyt36 extract while a clear 32 kDa bacteriolytic band was observed using the S. aureus wild-type strain RN450 (data not shown). The 32 kDa bacteriolytic band was restored using the extract of the complemented mutant JT1482. Western blot analysis of the extracts with anti-Sle1 serum further supported these observations (Fig. 7A). Anti-Sle1 detected the 32 kDa proteins in extracts of the wild-type strain RN450 and of the complemented mutant JT1482, but not from the extract of the mutant Lyt36 (data not shown). Further, when the Sle1 mutant was grown in stationary culture, the cells grew at the bottom of the test tube while the wild-type RN450 and the complemented mutant JT1482 grew normally resulting in uniform turbid medium with few sediment cells (Fig. 7B). Phase contrast light microscopic observation showed the mutant formed clusters but the appearance was somewhat different from that of the atl null mutant. Individual cells in clusters were arranged in a much more regular manner than those of the atl null mutant and the cells appeared to be tightly bound together as shown in Fig. 7C. Scanning electron microscopy (SEM) observation showed irregular morphology with unseparated daughter cells suggesting the retardation of cell separation in the mutant cells (Fig. 7D). In contrast, RN450 and JT1482 appeared as cocci or diplococci with far fewer clusters (Fig. 7C). Transmission electron microscopy (TEM) observations further supported the inhibition of cell separation in the mutant cells (Fig. 7E, a–d). Some cells showed several fold in-growths of septa in different planes without subsequent cell separation resulting in formation of giant cells. Furthermore, the mutant cells possessed a rough outer surface compared with the wild type; whereas, the complemented mutant JT1471 had a restored smoothness to the outer surface that was observed in the wild-type RN450 (Fig. 7E, e).

Figure 7.

Comparison of the wild type, sle1 mutant and sle1-complemented mutant.
A. Western blotting analysis. Western blotting analysis was performed using antisera to Sle1 (a) or non-immune serum (b) as primary antibody. Lane 1, RN450; lane 2, Lyt36; lane 3, JT1482.
B. Turbidity of a culture grown overnight without shaking.
C. Light microscopic observation of the cultures (× 345).
D. Scanning electron microscopic observation of the cells. Bar = 0.2 µm. Lane 1, RN450; lane 2, Lyt36; lane 3, JT1482.
E. Transmission electron microscopic observation of the cells. Bar = 0.2 µm. Lyt36 (a–d); RN450 (e).

The phenotype of the Sle1 mutant suggested the involvement of Sle1 in the separation of daughter cells during cell division. In the absence of Sle1 activity, cell clusters were formed. Changes in the structure of the peptidoglycan are likely to occur in the Sle1 mutant compared with the wild-type cells. Therefore, we prepared peptidoglycan from the wild-type RN450 and the mutant, and the enzymatic hydrolysates obtained by digestion with mutanolysin were analysed using HPLC. The peptidoglycan of the Sle1 mutant showed increased relative amounts of monomer (+ 22.1%), dimers (+ 14.9%) and trimers (+ 26.9%) (Table 1). Conversely, the relative amount of the oligomers (n > 4) was reduced in the mutant. These changes might possibly reflect a defect in the splitting of the septum.

Table 1.  The per cent muropeptides and overall degree of cross-linking in the parent and mutant.
MuropeptideaRelative amounts (%)
RN450JT1462 (Δ%)
  • a

    . Sum of the peak areas in the indicated fractions.

  • b

    . The cross-linking (CL) value was calculated as follows: 0.5 × dimer (%) + 0.67 × trimer (%) + 0.9 × oligimer (%).

Monomer 6.2 7.51 (22.1)
Dimer 7.2 8.32 (14.9)
Trimer 7.4 9.4 (26.9)
Oligomer79.274.77 (−5.6)
CL(%)b79.977.75 (−2.6)

As peptidoglycan hydrolases can cleave covalent bonds in murein saccules that maintain the internal pressure of the cells, they could possibly work as autolytic enzymes. We have previously demonstrated that atl gene products are autolytic enzymes that are involved in penicillin-induced cell lysis. Therefore, we investigated the effect of the sle1 mutation on penicillin-induced lysis and on lysis of non-growing cells suspended in buffer. There was no difference between the wild type and the mutant in the rate of lysis in both cases suggesting that Sle1 is not directly involved in penicillin-induced lysis of S. aureus (data not shown). This is further evidence that the sle1 product enzyme is different from the atl gene product.

Phenotype of sle1 and atl double mutant

We have previously demonstrated that atl gene products are involved in cell separation. The atl null mutant grew in clusters, but those clusters were composed of well-separated cocci whose individual cells were connected to each other by a marginal peptidoglycan bridge. This led us to predict the presence of another peptidoglycan hydrolase(s) involved in cell separation (Takahashi et al., 2002) and we have thus identified Sle1. To obtain more insight into the biological function of Sle1 and Atl, we constructed a strain lacking both Sle1 and Atl. This double mutant was created using phage transduction of the mutation from the atl knockout strain JT1457 into JT1462. The double mutant was viable but its growth was significantly impaired. The double mutant revealed no bacteriolytic bands in zymogram analysis, and formed irregular clusters in which cell separation was inhibited (Fig. 8). These results clearly indicated that S. aureus uses two peptidoglycan hydrolases, Atl and Sle1 for cell separation after division.

Figure 8.

Electron microscopic observation of atl/sle1 double mutant. Transmission electron microscopic observation of the cells. Bar = 0.5 µm.

Sle1 and pathogenesis of staphylococcal infection

Alteration of the cell surface in the mutant suggested that the Sle1 mutation might affect physicochemical properties of the cell surface, which is primarily involved in the initial attachment of the bacteria. Therefore, we assessed whether the Sle1 mutation affected pathogenesis of S. aureus using an acute infection model. Staphylococci were injected intraperitoneally using eight mice per trial group. The wild type caused lethal infection in half of all infected animals [50% lethal dose (LD50)] when 9.3 × 106 cfu were injected whereas the Sle1 mutant required 1.08 × 108 cfu.


We have identified an N-acetylmuramyl-l-alanine amidase with a molecular mass of 32 kDa. The protein designated as Sle1 possesses a modular structure composed of an N-terminal cell-wall binding domain (LysM domain) and a C-terminal catalytic domain. Presence of several ORFs showing a similar modular structure in Gram-positive bacteria suggests that they conform to a bacteriolytic enzyme family using N-acetylmuramyl-l-alanine amidase activity. The phenotype of the sle1 mutant clearly indicates that sle1 is important for splitting the septum during cell division. We have previously demonstrated that atl gene products are involved in cell separation (Takahashi et al., 2002). The phenotype of the atl/sle1 double mutant demonstrated that both Atl and Sle1 are possibly the only peptidoglycan hydrolases involved in cell separation of S. aureus. Involvement of multiple peptidoglycan hydrolases in cell separation has been reported in a number of other bacteria. For instance, three amidases, AmiA, B and C, are involved in cell separation of E. coli (Heidrich et al., 2001). In S. pneumoniae, LytA and LytB are shown to be involved in cell separation (Severin et al., 1997; Yothrer et al., 1998; De Las Rivas et al., 2002) and further, in Bacillus subtilis LytF and CwlF are involved in cell separation (Ohnishi et al., 1999). In these cases a single mutation has little effect on cell separation and the double or triple mutants showed a phenotypic impaired cell separation with extraordinarily long chain formation (Ohnishi et al., 1999; Heidrich et al., 2001). Similarly the atl/sle1 double mutant had a phenotype where cell separation was severely impaired. Isolation of the Sle1 mutant indicated that Sle1 is not an essential enzyme for staphylococcal cell growth at least under laboratory conditions. However, cluster formation does inhibit the dissemination of daughter cells and thus could affect distribution of the cells during an infection. Our results of lowered pathogenesis in the Sle1 mutant suggest this is a fundamental function of Sle1 in infection with S. aureus. Recently, 35 kDa bacteriolytic protein Aae was identified in S. epidermidis (Heilmann et al., 2003). Aae is highly homologous to Sle1 (65% identity and 72% homology in amino acid sequences) and shows similar modular structure, the repetitive sequences in the N-terminal half and the catalytic domain in the C-terminal half. Aae showed adhesive properties to fibrinogen, fibronectin and vitronectin suggesting its role as an adhesin in colonization of the bacteria to host-factor coated material as well as host tissue. Whether or not Sle1 possesses such adhesive properties besides its bacteriolytic activity remains to be investigated.

Reverse-phase chromatography of the peptidoglycan digests and subsequent MALDI-TOF-MS analysis of the digested fractions clearly indicated that the mode of action of Sle1 and the 62 kDa amidase are different. Sle1 selectively targets a small number of N-acetylmuramyl-l-Ala bonds in the peptidoglycan saccules as double digestion of the peptidoglycan with Sle1 and mutanolysin decreased only mutanolysin-generated dimers and other oligomers were virtually unaffected. Furthermore MALDI-TOF-MS analysis shows a dimeric peptide with an incompletely free Gly chain. These results strongly suggest that Sle1 predominantly cleaves N-acetylmuramyl-l-Ala bonds in dimeric cross-bridges that interlink two murein strands in the peptidoglycan. In E. coli, trimeric cross-bridges that interlink three murein strands are assumed to accumulate at the sites of uncleaved septa according to a hypothetical growth model called the ‘three-for-one’ mechanism (Höltje, 1998). AmiC, one of the amidases involved in cell separation of E. coli specifically degrades the thick murein rings resulting in blocked septum formation. But workers could not demonstrate the specific cleavage of the oligomeric cross-bridges by AmiC (Heidrich et al., 2001). The fact that an amidase, which specifically recognizes dimeric cross-bridges that interlink two murein strands, is involved in cell separation may suggest the presence of a unique peptidoglycan structure in the septal region. Further study on the cell surface localization of Sle1 and the investigation of the peptidoglycan architecture in detail will contribute to the understanding of the molecular mechanism of staphylococcal cell separation by these peptidoglycan hydrolases.

Experimental procedures

Bacterial strains and culture conditions

Bacterial strains and plasmids used in this study are listed in Table 2. Staphylococcus was grown in either Trypticase Soy Broth (TSB, Becton and Dickinson Microbiology Systems, Cockeysville, MD) or brain–heart infusion (BHI) broth (Becton and Dickinson); and Escherichia was grown in Luria–Bertani (LB) broth. For enzyme purification, bacteria growing exponentially in TSB were inoculated into 3 l of fresh medium and incubated with agitation by a rotary shaker for 6 h at 37°C until the cells reached the mid-log phase. The culture was then centrifuged at 6000 g for 15 min at 4°C, and the culture supernatant was passed through a membrane filter (pore size, 0.2 µm; Advantec, Tokyo, Japan). When necessary, chloramphenicol (10 µg ml−1), tetracycline (3 µg ml−1) or ampicillin (100 µg ml−1) was added to the medium.

Table 2.  Bacterial strains, plasmids and primers.
Strain or plasmidVectorRelevant characteristicsSource or reference
S. aureus
 RN450 NCTC8325-4R. Novick
 RN4220 NCTC8325-4 rR. Novick
 Lyt36 RN450 (sle1::pLyt36)This study
 JT1413 RN450 (atl::cat)J. Takahashi
 JT1471pCL15RN450(pJT1131)This study
 JT1482pCL15Lyt36(pJT1131)This study
 JT1462 RN450(sle1::pLyt36)This study
 JT1483pCL15JT1462 (pJT1131)This study
 JT1457 RN450(atl::cat)This study
 JT1517 RN450(atl::cat, sle1::pLyt36)This study
E. coli
 XL-II blue  recA1 endA1 gyrA96 thi-1 hsdR17 supE44 relA1 lec[F′ proAB lacIqZDM15 Tn10(Tetr) Amy Camr] 
 pGEM T-easy PCR cloning vector; Ampr in E. coliPromega
 pYT1 Shuttle vector (Ts); Ampr in E. coli, Tcr in S. aureusK. Hiramatsu
 pCL15 Shuttle vector; Ampr in E. coli, Camr in S. aureusC. Lee
 pQE-30 Expression vectorQIAGEN
 pJT1086pGEM T-easy1.0 kb PCR fragment containing sle1 (RN450)This study
 pJT1090pQE-301.0 kb BamHI-HindIII fragment of pJT1086A. Tomasz
 pJT1092pGEM T-easy0.5 kb PCR fragment containing sle1 (RN450)This study
 pJT1093pGEM T-easy0.5 kb PCR fragment containing sle1 (RN450)This study
 pJT1095pQE-300.5 kb BamHI-HindIII fragment of pJT1092This study
 pJT1096pQE-300.5 kb BamHI-HindIII fragment of pJT1093This study
 pLyt36pYT10.9 kb PCR fragment containing sle1 (RN4220)This study
 pJT1128pGEM T-easy1.1 kb PCR fragment containing sle1 (RN450)This study
 pJT1131pCL151.1 kb BamHI-EcoRI fragment of pJT1128This study
 pJT1084pGEM T-easy1.8 kb PCR fragment containing amidase domain in atl (RN450)This study
 pJT1088pQE-301.8 kb BamHI-HindIII fragment of pJT1084This study
GenePrimerOligonucleotide sequence (5′-3′)Size of amplified products (kb)


The bacteriolytic enzyme was analysed by zymography using polyacrylamide gels containing heat-killed cells of either S. aureus FDA209P or M. luteus ATCC4698 as described elsewhere (Sugai et al., 1990). For quantification of bacteriolytic activity, the minimal dose that exhibited a visible bacteriolytic band in the zymogram was defined as one unit (Ohta et al., 1998). In some experiments, each bacteriolytic band was scanned and the activity was evaluated using densitometry (Sugai et al., 1991).

Protein purification

Chromatography media, TSKgel Blue-Toyopearl 650 ML (Tosoh, Tokyo, Japan) and TSKgel Butyl-Toyopearl 650 ML (Tosoh) were used. In the last step of purification we used a disc preparative gel electrophoresis system (Nihon Eido, Tokyo, Japan). The fraction of interest was electrophoresed in a 12% acrylamide gel (36 by 95 mm) according to the manufacture's specifications and the proteins were separated. The fractions were manually collected in 0.5 ml aliquots and the bacteriolytic activity was determined by zymography and the proteins were detected by SDS-PAGE as described below.

N-terminal amino acid sequencing

Purified protein separated by SDS-PAGE was electrotransferred using a Trans-Blot polyvinyliden difluoride membrane (Bio-Rad Laboratories, Hercules, CA). The protein band of interest was stained with 0.1% Coomassie brilliant blue R-250 and sequenced with an automated Edman degradation using a Model 49X Procise protein sequencer (PE Applied Biosystems) with standard blot sequencing cycles.

DNA manipulations

Routine DNA manipulations were performed by standard procedures (Sambrook et al., 1989) where transformation of S. aureus by electroporation was performed. Southern blotting of DNA and hybridization were performed as described previously (Sugai et al., 1997b). The primers used in this study are listed in Table 2.

Purification of recombinant His × 6-tagged protein

Escherichia coli XL-II blue (the plasmid vector carrier) was grown at 37°C with vigorous shaking to A660 of 0.7 and then 1 mM isopropyl-β-d-thiogalactopyranoside was added leading to induction of the His × 6-tagged protein. After 3 h incubation, the bacteria were centrifuged, resuspended in 100 mM phosphate buffer (pH 6.8) and disrupted with an Ultrasonic disruptor (Tomy Seiko). The preparation was centrifugated at 9000 g for 30 min and the pellet was solubilized with 8 M urea, 0.1 M Tris-Cl, 0.1 M phosphate buffer (pH 8.0) and the His × 6-tagged protein was purified using Ni-resin (Qiagen) according to the manufacturer's instructions. The purified sample was extensively dialysed against 0.1 M phosphate buffer (pH 8.0) to renature the enzyme activity before use.

Preparation of peptidoglycan

Peptidoglycan was prepared essentially as described previously (Roos et al., 1998). Briefly, bacteria were grown in 500 ml BHI to A660 of 0.7. The cells were washed twice by centrifugation with PBS and resuspended in 10 ml of 1 M NaCl. The cells in a Duran flask with 20 g of glass beads (0.1–0.11 mm) were broken using elliptical shaking at 2000 rev per min for 3 min using CO2 cooling. Gram staining revealed complete lysis. To remove the glass beads and unbroken cells, the suspension was centrifuged at 3000 g for 2 min. The supernatant was further centrifuged at 9000 g for 15 min and the pellet was resuspended in 0.5% SDS and incubated at 60°C for 30 min. The crude cell wall was prepared by centrifugation at 9000 g for 15 min, and the pellet was washed six times with water. The crude cell wall was incubated with 200 µg ml−1 trypsin in 1 M Tris-Cl (pH 7.0) for 18 h at 37°C. The preparation was centrifuged (9000 g for 15 min), washed with 1 M Tris-Cl (pH 7.0), then with 1 M Tris-Cl, 1 M NaCl (pH 7.0) and again with 1 M Tris-Cl (pH 7.0). Finally, the cell wall was washed three times with water. After centrifugation at 9000 g for 15 min, the pellet was resuspended in 1 ml 40% (w/v) aqueous hydrofluoric acid and incubated for 18 h at 4°C to remove the teichoic acids. The purified peptidoglycan was isolated by centrifugation, washed in water, and lyophilized.

Assay of lytic activity and enzyme specificity

The purified peptidoglycan was used as the substrate for the lytic enzyme activity assay. Lytic activity was determined by the rate of the decrease in the turbidity of the peptidoglycan suspended in 0.1 M Tris-Cl (pH 7.5) [1 mg (dry weight) ml−1]. The test enzyme (25 µg) was mixed with 1 ml of the peptidoglycan suspension. The turbidity at A660 was monitored over time and the concentration of free amino groups and reducing sugars were analysed. The reducing sugars or free amino groups of enzymatically hydrolysed peptidoglycan were determined using a modified Park–Johnson procedure (Thompson and Shockman, 1969) or a Ghuysen procedure using 1-fluoro-2,4-dinitrobenzene (FDNB) (Ghuysen et al., 1966) respectively. To identify the N-terminal amino acid at the cleavage site, hydrolysed samples were dried under vacuum and resuspended in a 500 µl mixture (4:1, v/v) of buffer A (10% acetonitrile and 0.02 N acetic acid) and buffer B (90% acetonitrile and 0.02 N acetic acid). The hydrolysed DNP compounds were analysed using HPLC with a reverse-phase column (Wakosil-II5 C18, 4.0 × 250 mm, Wako, Kyoto, Japan) (Nugroho et al., 1999).

HPLC analysis of enzyme-treated muropeptides

Muropeptides generated using digestion with the His-Sle1 and purified peptidoglycan was solubilized with mutanolysin in the presence or absence of the His-Sle1. Purified peptidoglycan (1.3 mg dry weight) was suspended in 0.1 M phosphate buffer containing 4 mM MgCl2 (pH 6.8) and was incubated with mutanolysin (50 µg ml−1) alone; His-Sle1 (25 µg ml−1) and mutanolysin (50 µg ml−1); or 62 kDa amidase (25 µg ml−1) (Sugai et al., 1995) and mutanolysin (50 µg ml−1). The mixtures were incubated for 16 h at 37°C, followed by addition of 60 µl 20% phosphoric acid adjusting the pH to 4.0 and incubated at 95°C for 5 min to stop the digestion. The mixture was then added to 175 µl 1.5 M sodium borate buffer (pH 9.0) and 10 mg solid sodium borohydride and incubated at room temperature for 15 min to reduce the soluble peptidoglycan to its muramisitol derivative. The excess borohydride was quenched using 120 µl 20% phosphoric acid. To hydrolyse the remaining O-acetyl groups present on the muramic acid residues, the mixture was adjusted to pH 12.0 with 4 N NaOH and incubated for 1.5 h at 37°C. The muropeptide solution was then adjusted to pH 2.5 using 4 N HCl prior to centrifugation at 10 000 g for 15 min. The supernatant was passed through a membrane filter (0.22 µm) to remove the insoluble contaminants. Separation of the muropeptides was carried out using a Hypersil ODS (5 Å, 250 × 4.6 mm) column. A linear gradient was used from 5% (v/v) methanol in 50 mM NaH2PO4 (pH 2.5) to 30% (v/v) methanol in 50 mM NaH2PO4 (pH 2.8) where at 210 min the muropeptides were detected at A206.

Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry analysis

The dried HPLC fractions containing the muropeptides of interest were suspended in 0.1% trifluoroacetic acid after desalting with a C-18 ZipTip (Millipore). The samples (1 µl) were cospotted with 2 µl of the matrix, α-cyano-4-hydroxycinnamic acid at 10 mg ml−1 in CH3CN : water : TFA (70:30:0.1). The MALDI-MS spectra were observed using a reflection time-of-flight instrument (BIFLEX III, Bruker Daltonics) in the positive mode.

Electron microscopy

Bacterial cells were harvested, washed twice with phosphate buffered saline and collected by centrifugation. For TEM, the cells were doubly fixed with 2.5% glutaraldehyde and 1% OsO4. The samples were then dehydrated using an ethanol series and embedded in Spurr's Epon. Ultrathin sections were cut with an ultra-microtome and examined with a JEOL JEM-2000 EXII electron microscope at 80 or 100 kV. For SEM, a drop of the bacterial cell suspension was mounted on a glass cover slip and doubly fixed with 2.5% glutaraldehyde and 1% OsO4. For surface conductivity, the samples were stained with 1% tannic acid and fixed with 1% OsO4. After dehydration using an ethanol series and t-butylalcohol, the samples were dried and coated with an osmium plasma coater PMC-5000 (MEIWA SHOJI) and examined with a JEOL JSM-6340F field emission scanning electron microscope at 15 kV.


The penicillin G (PCG) induced autolysis of growing cells and autolysis of cells suspended in buffer solution were measured essentially as previously described (Komatsuzawa et al., 1994). To measure PCG-induced autolysis, cells were grown to A660 of 0.2 at 37°C and aliquots of TSB alone or TSB containing PCG were added to the cell suspension. The cells were then re-incubated and the A660 was monitored at 1 h intervals. For the autolysis assay, cells were grown to A660 of 0.7 at 37°C and then the cells were collected by centrifugation at 10 000 g for 10 min. The pellets were washed with cold phosphate buffered saline twice and suspended to A660 of 0.7 in 50 mM Tris-Cl (pH 8.0). Autolysis was measured as a decrease in A660 after incubating at 37°C.

Animal experimentation

Staphylococci were grown overnight in TSB, diluted into fresh medium, and grown for 6 h at 37°C until A660 of 1.0. Bacteria were centrifuged, washed once and diluted with saline. The virulence of staphylococcal strains was estimated by determining LD50. Groups of eight Std-ddY mice weighing approximately 20 g each were injected intraperitoneally with graduated doses of bacteria, and mortality was monitored.

Antiserum.  The recombinant protein with an equal volume of Freund's complete adjuvant (Difco Laboratories, Detroit, MI) (50 µg of protein per millilitre) was injected subcutaneously into rabbits (2 kg). At 2 and 4 weeks, rabbits were injected intravenously with 100 µg of the recombinant protein. Antiserum was obtained 5 weeks after the first injection. The antiserum was diluted 1000-fold for Immunoblot detection.

Other procedures

SDS-polyacrylamide gel electrophoresis and Western blot were carried out as previously described (Sugai et al., 1995). Immuno-detection of protein was performed using a ECL Western blot analysis system (Amersham Pharmacia). Protein concentrations were determined using the method of Bradford (Bradford, 1976) with bovine serum albumin as the standard.

Nucleotide sequence accession number

The nucleotide sequence of sle1 in S. aureus NCTC8325-4 is in DDBJ, EMBL and the GenBank under accession number AB113206.


We acknowledge K. Uehira and T. Suda of the Electron Microscope Center, and S. Ohmori of the Department of Microbiology, Kawasaki Medical School for use of their electron microscope facilities; T. Fukushima, T. Yamaguchi and J. Sekiguchi of the Department of Applied Biology, Faculty of Textile Science and Technology, Shinshu University for their technical support in HPLC analysis; H. Murakami of the Department of Bacteriology, School of Medicine, Juntendo University and T. Mizuno, a dental student at Hiroshima University for their technical help; and I. Hayashi for MS analysis. We thank N. Ledger for editorial assistance. Identification of the gene for Sle-1 could not have been achieved without the generous data release policies of the genome sequencing centre at TIGR and the Sanger Center. Thanks also go to the Research Center for Molecular Medicine, the Research Facility for Laboratory Animal Science and Research Facility, Hiroshima University Faculty of Dentistry for use of those facilities. This work was supported in part by Research Project grants (No. 9-503, 10-504 and 11-505) from Kawasaki Medical School.

Note added in proof

Heilman, C. et al., identified the same enzyme as a potential adhesion to fibrinogen and fibronectin, and designated it as Aaa. Infect Immun73: 4793–4802, 2005.