Cell Wall Attachment of a Widely Distributed Peptidoglycan Binding Domain Is Hindered by Cell Wall Constituents*

The C-terminal region (cA) of the major autolysin AcmA of Lactococcus lactis contains three highly similar repeated regions of 45 amino acid residues (LysM domains), which are separated by nonhomologous sequences. The cA domain could be deleted without destroying the cell wall-hydrolyzing activity of the enzyme in vitro. This AcmA derivative was capable neither of binding to lactococcal cells nor of lysing these cells while separation of the producer cells was incomplete. The cA domain and a chimeric protein consisting of cA fused to the C terminus of MSA2, a malaria parasite surface antigen, bound to lactococcal cells specifically via cA. The fusion protein also bound to many other Gram-positive bacteria. By chemical treatment of purified cell walls of L. lactis and Bacillus subtilis, peptidoglycan was identified as the cell wall component interacting with cA. Immunofluorescence studies showed that binding is on specific locations on the surface of L. lactis, Enterococcus faecalis, Streptococcus thermophilus, B. subtilis, Lactobacillus sake, and Lactobacillus casei cells. Based on these studies, we propose that LysM-type repeats bind to peptidoglycan and that binding is hindered by other cell wall constituents, resulting in localized binding of AcmA. Lipoteichoic acid is a candidate hindering component. For L. lactis SK110, it is shown that lipoteichoic acids are not uniformly distributed over the cell surface and are mainly present at sites where no MSA2cA binding is observed.

The major autolysin AcmA of Lactococcus lactis subsp. cremoris MG1363 is a peptidoglycan hydrolase that is required for cell separation and is responsible for cell lysis during stationary phase (1,2). The 40.3-kDa secreted mature protein is subject to proteolytic degradation (3,4) resulting in a number of activity bands in a zymogram of the supernatant of a lactococcal culture. Bands as small as that corresponding to a protein of 29 kDa were detected, all representing products of AcmA (2). From experimental data and homology studies, we inferred that AcmA consists of two domains: an active site domain and a C-terminal region containing three highly homologous re-peats of 45 amino acids (cA), which might be involved in cell wall binding. Since the smallest active protein is 29 kDa, it was proposed that the protein undergoes C-terminal proteolytic breakdown (1,2). Nearly all cell wall hydrolases seem to consist of a catalytic domain and usually, but not always, a domain containing a number of specific amino acid repeats (5,6). Mur1 of Streptococcus thermophilus and Mur, the N-acetylmuramidase of Leuconostoc citreum, do not contain such repeats (7,8). Peptidoglycan hydrolyzing activity of Mur could be detected in vitro, but Mur on its own was not able to complement an acmA mutation in L. lactis. However, expression of a Mur-cA fusion protein was able to play the role of AcmA in cell separation after cell division in L. lactis acmA (8).
Cell wall hydrolases of various bacteria and bacteriophages contain repeats similar to those present in AcmA. These repeats are also called LysM (lysin motif) domains, since they were originally identified in bacterial lysins (9). The presence of the LysM domains is not limited to bacterial proteins. They are also present in a number of eukaryotic proteins, whereas they are lacking in archaeal proteins (10).
A cell wall binding function has been postulated for a number of proteins containing LysM domains (5,9,11,12). Partially purified muramidase-2 of Enterococcus hirae, a protein similar to AcmA and containing six LysM domains, binds to peptidoglycan fragments of the same strain (13). The p60 protein of Listeria monocytogenes contains two LysM domains and was shown to be associated with the cell surface (14). The ␥-Dglutamate-meso-diaminopimelate muropeptidases LytE and LytF of Bacillus subtilis have three and five repeats, respectively, in their N termini and are both cell wall-bound (15)(16)(17)(18). However, which particular parts of these enzymes entailed the binding capacity has not been examined in any of these studies.
Some spore-specific proteins in B. subtilis also contain LysM domains. His tag fusions of the spore proteins YdhD, YkuD, and YkvP (containing one LysM domain) were produced during sporulation and could be detected in mature spores (19,20). SafA (YrbA), a protein involved in spore assembly, localized to the outer rim of the cortex and is apparently targeted to the spore by its N-terminal LysM domain (21). When the signal sequence of ␤-lactamase was replaced by the two N-terminal LysM domains of the spore protein YaaH, ␤-lactamase assembled in spores. The authors speculated that the LysM domain functions as a kind of signal sequence involved in assembly on forespores (20).
The structure of one of the two LysM domains of the Escherichia coli lytic transglycosidase MltD has been resolved by NMR studies (10). The domain has a ␤␣␣␤ structure. In the same study also a potential substrate-binding site could be identified. A loop, present between ␤-strand 1 and ␣-helix 1 lies at the end of a shallow groove on the surface of the domain. A conserved aspartate or glutamate in this shallow groove could be involved in the interaction with the ligand (10).
In this paper, we describe that the LysM domains bind directly to peptidoglycan. The binding is not species-specific; the domain binds to many Gram-positive bacteria with different peptidoglycan structures. Using immunofluorescence microscopy, we show that the domain binds to specific loci on the cell surface. Specific chemical treatments of cells and cell walls indicate that a cell wall component, extractable with trichloroacetic acid, is responsible for hindering of AcmA binding, thereby causing this localized binding.
Chemicals and Enzymes-All chemicals used were of analytical grade and, unless indicated otherwise, obtained from Merck. Enzymes for molecular biology were purchased from Roche Applied Science and used according to the supplier's instructions. Purified peptidoglycan from Staphylococcus aureus was obtained from Fluka Chemie (Zwijndrecht, The Netherlands). Peptidoglycan from Micrococcus luteus and Curtobacterium flaccumfaciens were kind gifts of Prof. Dr. S. J. Foster (University of Sheffield).
Enzyme Assays and Optical Density Measurements-AcmA activity was visualized on G 1 ⁄2M17 agar plates containing 0.2% autoclaved lyophilized Micrococcus lysodeikticus ATCC 4698 cells (Sigma) as halos around L. lactis colonies after overnight growth at 30°C. X-prolyl dipeptidyl aminopeptidase (PepX) was measured as described by Buist et al. (3). Absorbance was measured in a Novaspec II spectrophotometer (Amersham Biosciences) at 600 nm. Autolysis of L. lactis was analyzed by following the absorbance at 600 nm of a 100-fold diluted overnight culture in fresh G 1 ⁄2M17 medium for 6 days at 30°C.
DNA Manipulations and Transformations-Molecular cloning techniques were performed essentially as described by Sambrook et al. (23). Electrotransformation of E. coli and L. lactis was performed by using a gene pulser (Bio-Rad) as described by Zabarovsky et al. (24) and Leenhouts et al. (25), respectively. Minipreparations of plasmid DNA from E. coli and L. lactis were obtained by the alkaline lysis method as described by Sambrook et al. (23) and Seegers et al. (26), respectively. PCR products were purified using the High Pure PCR purification kit (Roche Applied Science).
Construction of AcmA Derivatives-A stop codon and EcoRI restriction enzyme site (italics and underlined in oligonucleotide REPDEL, respectively) were introduced in acmA at the end of the sequence specifying the active site domain. This was done by PCR using the primers REPDEL (5Ј-CGCGAATTCCTTATGAAGAAGCTCCGTC) and ALA-4 (5Ј-CTTCAACAGACAAGTCC), annealing within the sequence encoding the signal peptide of AcmA and pGKAL1 as the template. The PCR product was digested with SacI and EcoRI and cloned into the corresponding sites of pBluescriptSKϩ, leading to pDEL. Subsequently, the 1187-bp PflMI/EcoRI fragment of pGKAL1 (2) was replaced by the 76-bp PflMI/EcoRI fragment of pDEL, resulting in pGKAL5, which was obtained in L. lactis MG1363acmA⌬1.
SDS-PAGE, Detection of AcmA Activity, Western Blotting, and Immunodetection-L. lactis cell and supernatant samples were prepared as described before (3). AcmA activity was detected by a zymogram staining technique using SDS-polyacrylamide (12.5 or 17.5%) gels containing 0.15% autoclaved, lyophilized M. lysodeikticus ATCC 4698 cells (Sigma) as described before (2). The standard low range and prestained low and high range SDS-PAGE molecular weight markers of Bio-Rad were used as references. SDS-polyacrylamide gels were stained with Coomassie Brilliant Blue (Bio-Rad).

Laboratory collection
Cell Wall Isolation and Chemical Treatment-Cultures of L. lactis and B. subtilis (50 ml each) were pelleted, and cells resuspended in 5 ml of water were broken by vigorous shaking in the presence of glass beads (29). Nonbroken cells were removed by centrifugation (5 min, 2000 ϫ g). The cell walls were isolated from the resulting supernatant by centrifugation (15 min, 20,800 ϫ g). Cell walls were washed several times with water and lyophilized. Equal aliquots of cell walls (0.1 mg) were resuspended in 1 ml of water, 10% SDS, or 10% trichloroacetic acid and boiled for 15 min. Subsequently, the cell walls were washed, once with 1 ml of phosphate-buffered saline (PBS 1 ; 75 mM NaP i , pH 7.3, 68 mM NaCl) and four times with 1 ml of water. L. lactis and Lactobacillus casei cells were treated with trichloroacetic acid by boiling 25 l of cell culture in 1 ml of 10% trichloroacetic acid. Lactobacillus helveticus was treated with SDS by boiling 25 l in 10% SDS. The cells were washed as described for the chemically treated cell walls.
Purification of PA3-L. lactis PA1001 containing plasmid pPA3 was grown overnight at 30°C in GM17 supplemented with 5 g/ml chloramphenicol. Subsequently, the cells were diluted 100-fold in 1 liter of fresh GM17 and induced with 1 ml of nisin-containing supernatant of an overnight culture of L. lactis NZ9700. The cells were cultivated for 24 h at 30°C. Cells and supernatant were separated by centrifugation for 20 min at 10,000 rpm, after which the latter was filter-sterilized and concentrated with Vivaflow 200 (molecular mass cut-off of 10 kDa) to 20 ml and buffered with 80 ml of 25 mM sodium phosphate buffer, pH 5.8. To lower the salt concentration in the sample, 100 ml of H 2 O were added. The sample was loaded on a 5-ml prepacked AKTA-prime SP-Sepharose column (Amersham Biosciences), which was equilibrated with 25 mM sodium phosphate buffer, pH 5.8. The column was washed with 25 mM sodium phosphate until the absorbance measured at 214 nm did not change anymore. Subsequently, bound proteins were eluted with a 200-ml NaCl gradient from 0 to 0.5 M. Fractions of 4 ml were collected. The fractions containing highly pure PA3 were collected and desalted using a high trap desalting column (Amersham Biosciences). The protein was kept in H 2 O at Ϫ20°C at concentrations ranging from 0.4 to 0.8 mg/ml.
Binding of Proteins to Lactococcal Cells, Cell Walls, and Purified Peptidoglycan-The cells of 2 ml of exponential phase cultures of MG1363acmA⌬1 were collected by centrifugation and gently resuspended in an equal volume of supernatant of similarly grown MG1363acmA⌬1 carrying either plasmid pGK13 (empty vector), pGKAL1 (expressing acmA), or pGKAL5 (expressing truncated acmA) and incubated at 30°C for 20 min. Subsequently, the mixtures were centrifuged. Of the supernatants, 0.4 ml was dialyzed against three changes of 1 liter of demineralized water, after which they were lyophilized and dissolved in 0.2 ml of denaturation buffer (30). The cell pellets were washed with 2 ml of fresh medium, after which cell-free extracts were prepared in 1 ml of denaturation buffer.
Binding of the MSA2cA fusion protein was studied by growing L. lactis NZ9000 containing pNG304 or pNG3041 until an A 600 of 0.4 was reached. The cultures were induced with nisin (by adding one-one thousandth volume of the supernatant of the nisin producer L. lactis NZ9700) and incubated for 2 h. The cells of 1 ml of MG1363acmA⌬1 culture, 0.1 mg of cell walls, or 0.5 mg of peptidoglycan were resuspended in 1 ml of supernatant of either of the two induced cultures. Then the suspensions were spun down, and the pellets were washed with 1 ml of fresh M17 medium, resuspended in 100 l of denaturation buffer (30), boiled for 5 min, and subjected to SDS-10% PAGE followed by Western blot analysis. The supernatant fractions, containing nonbound proteins, were concentrated 10 times using phenol and ether extraction (31). The concentrated supernatants were diluted with 1 volume of SDS sample buffer and subjected to SDS-10% PAGE.
To study the binding efficiency of MSA2cA to different peptidoglycan types, 50-g samples of peptidoglycan isolated from M. luteus and C. flaccumfaciens were mixed with the protein. A dilution range of the protein was prepared by diluting the supernatant containing MSA2cA in supernatant of an L. lactis MG1363acmA⌬1 culture.
To study binding of PA3, 15 g of purified PA3 was added to 70 l of trichloroacetic acid-pretreated L. lactis cells (described above) in a total volume of 1 ml in 20 mM Tris-HCl, pH 8.0 (10 10 cells/ml). The samples were incubated for 20 min on a blood cell suspension mixer at room temperature. Subsequently, the samples were spun down (3 min, 20,000 ϫ g), and unbound proteins in the supernatant were precipitated with 5% trichloroacetic acid for 1 h on ice and subsequently centrifuged for 20 min at 20,000 ϫ g. Precipitated proteins were washed with acetone, air-dried, and resuspended in SDS-sample buffer. The cell pellets were washed with 1 ml of 20 mM Tris-HCl, pH 8.0, and resuspended in SDS-sample buffer. Samples were analyzed by SDS-PAGE and Coomassie staining.
Immunofluorescence Microscopy-Samples (25 l) of bacterial cultures or the chemically treated bacterial cells were pelleted and washed once with 1 ml of water. The pellets were resuspended in L. lactis NZ9000 (pNG3041) culture supernatant containing MSA2cA and incubated for 5 min at room temperature. The cells were washed once with 1 ml of 1 ⁄2M17 and subsequently incubated for 10 min at room temperature in 100 l of PBS containing 4% bovine serum albumin. Subsequently, the cells were resuspended in 100 l of PBS containing anti-MSA2 antibody (diluted 1:400) with 2% bovine serum albumin for 1 h. After three washing steps with 1 ml of PBS containing 0.1% Tween 20 (Sigma) (PBST), the cells were incubated for 1 h in 2% bovine serum albumin in PBS, containing a 1:400 dilution of Oregon Green antirabbit antibody (Molecular Probes, Eugene, OR). Subsequently, the cells were washed three times with 1 ml of PBST and transferred to a poly-L-lysine-coated microscopic slide (Omnilabo, Breda, The Netherlands), which was air-dried at ambient temperature. Vectashield (Vector, Burlingame, CA) was added, a coverslip was mounted, and fluorescence was visualized with a Zeiss microscope (Carl Zeiss, Thornwood, CA) and an Axion Vision camera (Axion Technologies, Houston, TX).
Lectin Binding Studies-Fluorescein-labeled Ricinus communis agglutinin I lectin (RCA120) (Vector) was diluted to 10 g/ml in lectin binding buffer (10 mM HEPES, pH 7.5, 0.15 M NaCl). L. lactis cells from 100 l of overnight culture were spun down, and the pellet was resuspended in 100 l of the diluted lectin. The suspension was incubated for 10 min on ice, and the cells were washed once in lectin binding buffer (100 l) and resuspended in the same volume. Samples (5 l) were spotted onto microscopic slides for fluorescence microscopy as described above.

AcmA Is a Modular Protein Containing an N-terminal Cell
Wall-degrading Domain-We have previously analyzed the amino acid sequence of the L. lactis autolysin AcmA using Blast algorithms and PFAM and reported that mature AcmA probably consists of two domains (2). The N-terminal domain is predicted to contain the active site, whereas the C-terminal domain contains a putative cell wall binding domain. This cell wall binding domain consists of three homologous so-called LysM domains (9), separated by nonhomologous amino acid sequences (Fig. 1A). To investigate whether the N-terminal domain of AcmA indeed carries the active site, a deletion variant lacking the putative cell wall binding domain was created by introducing a stop codon downstream of the codon for Ser 218 . The gene encoding truncated AcmA variant (A 1-218 ; see Fig.  1A) is expressed from the acmA promoter in the vector pGKAL5. The AcmA-negative strain L. lactis MG1363acmA⌬1 was transformed with pGKAL1, encoding wild type AcmA, or pGKAL5 and was plated on plates containing cell walls of M. lysodeikticus. Cells expressing acmA produced a clear halo around their colonies, whereas colonies of cells producing A 1-218 did not. Cells containing pGKAL5 formed long chains, resulting in culture sedimentation, as has also been shown for the AcmA-negative strain L. lactis MG1363acmA⌬1 (2). Autolysis was substantially reduced upon deletion of the C-terminal domain. Whereas a 36.7% reduction of the absorbance at 600 nm was observed for L. lactis MG1363acmA⌬1 (pGKAL1) after 60 h of incubation at 30°C, only 15.6% of A 600 reduction was observed for L. lactis MG1363acmA⌬1 expressing A 1-218 . This reduction was similar to that of L. lactis MG1363acmA⌬1 (15.2%) (Table II) Zymographic analysis of the cell fractions of overnight cultures of these three strains showed that AcmA and A 1-218 are active (Fig. 1B). The truncated AcmA variant A 1-218 is less active than AcmA, but expression of both proteins is equal, as is shown using Western analysis with anti-AcmA antibodies (Fig. 1B). In the supernatant of cultures producing full-length AcmA or A 1-218 , lytic activities could be detected corresponding to the calculated molecular weights of the processed, secreted forms of the respective proteins (40.3 and 17.1 kDa) (results not shown). These data show that the N-terminal domain of AcmA is active in vitro and, thus, contains the active site of the enzyme. This truncated form of AcmA is not able to lyse cells or separate chains in vitro, suggesting that the C-terminal domain of AcmA is required for both of these processes.
The C-terminal Domain of AcmA Is Required for Cell Wall Binding and Can Target a Heterologous Protein to Lactococcal Cells-Proof that the C-terminal domain (cA) of AcmA is required for cell binding was obtained by mixing supernatant fractions of end-exponential phase cultures containing AcmA or A 1-218 with cells of L. lactis MG1363acmA⌬1. After incubation, cells and supernatant were separated and examined for the presence of lytic activity using zymographic analysis. AcmA was mainly present in the cell fraction, whereas A 1-218 was still present in the supernatant, indicating that the latter had lost its cell binding abilities (results not shown). The breakdown products of AcmA were also capable of binding to the lactococcal cells (data not shown). As shown by Poquet et al. (4), these breakdown products of AcmA are generated by the housekeeping surface protease HtrA. To investigate whether cA is capable of targeting heterologous proteins to the cell surface of L. lactis, a fusion protein containing the human malaria parasite Plasmodium falciparum surface antigen MSA2 and cA was constructed. The entire fusion protein was produced by nisin induction using the pNZ8048 (32) derivative pNG3041 (Fig. 1A) in L. lactis NZ9000. The fusion consists of the 184-amino acid preprosequence of the lactococcal proteinase PrtP for proper secretion in L. lactis, followed by MSA2 (223 amino acid residues) and cA (Fig. 1A). MSA2cA is secreted as a 62.5-kDa protein. L. lactis NZ9000 (pNG304) expresses a fusion protein FIG. 1. A, schematic representation of pGKAL1 (expressing AcmA), pDEL, pGKAL5 (expressing A 1-218 ), pNG304 (expressing MSA2*), pNG3041 (expressing MSA2cA), and pPA3 (expressing PA3). SS (black), signal sequence of AcmA; SS (gray), signal sequence of Usp45 of L. lactis. Rx (dark gray), repeats; light gray, Thr-, Ser-, and Asn-rich intervening sequences (2); ppPrtP (black), signal and pro-sequence of the lactococcal proteinase PrtP (44). The AcmA active site domain is shown in white, and the P. falciparum antigen MSA2 is indicated by a striped bar. pDEL is a pBluescriptSKϩ derivative containing the SacI/EcoRI fragment of the PCR products obtained with primers ALA-4 and REPDEL in the corresponding sites of pBluescriptSKϩ. The PflMI/EcoRI fragment of pGKAL1 was replaced by the PflMI/EcoRI fragment of pDEL, resulting in plasmid pGKAL5.  containing the preprosequence of PrtP and the MSA2 protein, without cA (MSA2*).
The supernatants of cells secreting MSA2cA or MSA2* upon nisin induction were mixed with cells of L. lactis MG1363acmA⌬1. After incubation, cells and supernatants were separated and examined for the presence of MSA2 antigen by Western immunoblotting. As shown in Fig. 2, MSA2* was mainly present in the supernatant, whereas MSA2cA fractionated with the lactococcal cells, showing that cA is capable of specifically targeting MSA2 to the cell surface.
AcmA and MSA2cA Bind to Other Gram-positive Bacteria-A whole-cell-binding assay was performed to examine whether AcmA and MSA2cA are capable of binding to other Grampositive bacteria. Washed cells of several bacterial species were mixed with AcmA or MSA2cA. Protein binding was examined using zymographic analysis (AcmA) and Western immunoblotting (MSA2cA). AcmA and MSA2cA were both capable of binding to cells of B. subtilis, E. faecalis, S. thermophilus, L. casei, L. sake, Lactobacillus buchneri, Lactobacillus plantarum, Listeria inocua, and Clostridium beijerinckii (results not shown, but see below). The C-terminally truncated variant of AcmA, A 1-218 , and MSA2* did not bind to cells of these bacteria (data not shown).
Binding of cA to Cells of Gram-positive Bacteria Is Localized-After incubation with cells of L. lactis, E. faecalis, S. thermophilus, L. casei, L. sake, and B. subtilis, MSA2cA could be detected on the cell surface by immunofluorescence microscopy using anti-MSA2 antibodies and a fluorescent secondary antibody. MSA2cA was present at specific locations on the cell surface in all cases (Fig. 3A). MSA2cA clearly binds at the poles of L. casei cells. In the case of the cocci L. lactis, E. faecalis, and S. thermophilus, the protein binds around the septum. Binding to B. subtilis is at spots all over the cell surface, whereas MSA2cA binds to the entire cell surface of L. sake but not to the poles.
The cA Domain Binds to Peptidoglycan-Binding of cA to the cells of several different Gram-positive bacteria implies that the LysM repeats bind a cell wall component present in all of these cells. The localized binding observed above suggests either that this component is only present at specific loci or that it is present all over the surface but shielded at certain sites. To determine to which component cA binds, equal aliquots of lactococcal cell walls were treated with SDS to remove cell wall-associated proteins or with trichloroacetic acid, which is believed to remove peptidoglycan-associated polymers, among which are carbohydrates such as (lipo)teichoic acids (33).
Treated cell walls were subsequently incubated with a supernatant containing MSA2cA, followed by Western analysis. Fig.  4 shows that MSA2cA binds to both SDS-and trichloroacetic acid-treated cell walls, which suggests that it binds to the peptidoglycan of the cell wall. Similar results were obtained with chemically treated cell walls from B. subtilis (Fig. 4).
Next, purified peptidoglycan from S. aureus was incubated with MSA2cA or MSA2* and pelleted by centrifugation. Fig. 5A shows that MSA2cA associates with peptidoglycan, whereas MSA2* is solely present in the supernatant. The bacteria mentioned above all have A-type peptidoglycan. To test whether MSA2cA was able to bind to B-type peptidoglycan, increasing amounts of MSA2cA (from supernatant) were added to 50 g of peptidoglycan isolated from M. luteus (a typical A-type peptidoglycan) or C. flaccumfaciens (B-type), and bound protein was analyzed by Western immunoblotting. M. luteus and C. flaccumfaciens were shown to bind similar amounts of MSA2cA. Breakdown products of the protein are also able to bind, as is visible in Fig. 5B.
MSA2cA binding on trichloroacetic acid-treated cells of L. lactis and L. casei was examined by immunofluorescence microscopy. Fig. 3B shows that MSA2cA is able to bind to the entire cell surface and not only to the poles of the trichloroacetic acid-treated cells. SDS treatment did not alter the localization of the bound protein (results not shown).
Binding of AcmA or MSA2cA to L. helveticus is very inefficient (results not shown). The strain used for these experiments (L. helveticus ATCC 15009) is known to produce an S-layer (34). After boiling of the cells in SDS to remove the S-layer (35), binding of MSA2cA was more efficient and occurred over the whole L. helveticus cell surface (Fig. 3B). This result indicates that S-layer proteins hinder binding of MSA2cA. For all of the above mentioned binding experiments, we used L. lactis supernatant containing the MSA2cA protein. L. lactis PA1001 (pPA3) contains a pNZ8048 derivative that expresses the cA domain (amino acids 220 -437). Its secretion is driven by the signal sequence of the lactococcal USP45 protein (36) (see Fig. 1A for a diagram of PA3 protein). To exclude a role for an unidentified component present in the growth medium in binding to peptidoglycan, pure cA protein (PA3) was used in binding studies. PA3 was shown to fractionate with trichloroacetic acidpretreated L. lactis cells, showing that the cA domain itself binds to peptidoglycan (Fig. 6, lane 2). No PA3 protein was found in the supernatant fraction (Fig. 6, lane 4). The smear in lane 2 of Fig. 6 is caused by the (partly) degraded proteins in the trichloroacetic acid-pretreated cells. 2 When the same amount of PA3 was centrifuged without adding any trichloroacetic acid-pretreated cells, a small amount of PA3 protein was found in the pellet (Fig. 6, lane 1), probably representing aggregates of PA3. However, since similar PA3 samples were used in both experiments, all soluble proteins (Fig. 6, lane 3) specifically bound to trichloroacetic acid-pretreated cells (Fig.  6, compare lanes 3 and 4).
Lipoteichoic Acids Are Localized in L. lactis Cell Walls-L. lactis subsp. cremoris SK110 contains galactosyl residues in its lipoteichoic acids (LTA). Sijtsma et al. (37) have shown that the galactosyl moiety obstructs the attachment of bacteriophage to its binding site in the cell wall. Strain SK112, a derivative of SK110 lacking galactosyl residues in its LTA, is phagesensitive. The lectin RCA120, which is specific for terminal galactosyl units, binds specifically to the LTA isolated from SK110 and not to LTA from SK112 (37). Fluorescein-labeled RCA120 binds to those regions in the cell surface of SK110 that are not bound by MSA2cA (Fig. 7). Since galactosyl residues are only present in LTA (37), this would indicate that LTA in the cell wall of L. lactis SK110 is present only at the nonpolar sites. After boiling of the cells in 10% trichloroacetic acid, no galactosyl-containing polymers could be detected in the cell wall of SK110 using the RCA120 binding assay (results not shown), whereas MSA2cA binds to the entire surface of these trichloroacetic acid-treated SK110 cells. RCA120 did not bind to cells of MG1363. DISCUSSION In this paper, we show that AcmA is a modular enzyme consisting of an N-terminal active site domain and a C-terminal substrate-binding domain, containing three repeated LysM domains. A repeatless AcmA mutant was constructed and shown to be active on M. lysodeikticus cell walls, albeit with severely reduced efficiency. Cells expressing this derivative grew in long chains, sedimented, and did not autolyze. These results indicate that, although the N terminus of AcmA contains the active site, the presence of the C-terminal domain is needed for the enzyme to retain appreciable activity. It is tempting to speculate that this apparent increase in catalytic FIG. 4. Binding of MSA2cA to chemically treated cell walls of L. lactis and B. subtilis. Cell walls were isolated as described under "Experimental Procedures." Equal amounts of cell walls were boiled in water, 10% SDS, or 10% trichloroacetic acid and washed extensively with water. Subsequently, 1 ml of supernatant of an induced culture of L. lactis NZ9000 (pNG3041) containing MSA2cA protein was incubated for 5 min with the cell walls (0.5 mg). Subsequently, cell walls were subjected to centrifugation and were loaded onto an SDS-10% polyacrylamide gel. After electrophoresis, the MSA2 antigen was detected by Western immunoblotting using anti-MSA2 antibodies.

FIG. 5.
A, MSA2cA binds to purified peptidoglycan from S. aureus. Purified peptidoglycan (0.5 mg) isolated from S. aureus (Fluka) was mixed with supernatants from nisin-induced cultures of L. lactis NZ9000 (pNG304), containing MSA2* protein, and L. lactis NZ9000 (pNG3041), containing MSA2cA. After 5 min of incubation, the peptidoglycan was subjected to centrifugation. Peptidoglycan and supernatant fractions were analyzed by Western immunoblotting using anti-MSA2 antibodies. PG, peptidoglycan (bound fraction); sup, supernatant after incubation with peptidoglycan (nonbound fraction). B, MSA2cA binds to A-and B-type peptidoglycan. MSA2cA protein was diluted 100, 20, 10, 4, or 2 times in MG1363acmA⌬1 supernatant. Also, undiluted protein (U) was used. The diluted protein was mixed with 50 g of peptidoglycan isolated from M. luteus or C. flaccumfaciens, and bound protein was analyzed using Western immunoblotting. efficiency of AcmA is caused by the repeat domain by allowing the enzyme to bind to its substrate, the peptidoglycan of the cell wall. As was postulated by Knowles et al. (38) for the cellulase binding domains in cellobiohydrolases, such binding would increase the local concentration of the enzyme. The repeats could be involved in binding alone or could be important for proper positioning of the catalytic domain toward its substrate.
The C-terminal domain cA consists of three LysM domains separated by nonhomologous sequences. The hypothesis that the LysM domains of AcmA are involved in cell binding (2) was corroborated in this study. First of all, we show that AcmA is indeed capable of binding to bacterial cells. To prove that it was the C terminus of AcmA that facilitated binding and not some intrinsic cell wall binding capacity of the N-terminal domain, both the active site domain and the cA were separately produced in L. lactis. The active site domain A 1-218 did not bind to lactococcal cells, whereas the cA domain did bind to these cells when added from the outside. Cell wall binding was also obtained for the fusion protein MSA2cA in which the repeat domain was fused to the human malaria parasite antigen MSA2*, whereas MSA2* did not bind to lactococcal cells. In a previous study, Buist et al. (1) have shown that AcmA can operate intercellularly; AcmA-free lactococcal cells can be lysed when grown together with cells producing AcmA. Combining this observation with the results presented above allows us to conclude that AcmA does not only bind when confronting a cell from the outside but, indeed, is capable of hydrolyzing the cell wall with concomitant lysis of the cell.
LysM domains are present in proteins of many different bacteria and also of eukaryotes (10). The presence of similar repeats in proteins of different bacterial species strongly suggests that they recognize and bind to a cell wall component that is common to these bacteria. We show here that the LysM domains of AcmA bind specifically to peptidoglycan, the major cell wall component in bacteria. Binding is observed to many different bacteria and to different peptidoglycan types, suggesting that the LysM domains recognize a part common in all peptidoglycans. The peptidoglycans used in this paper have different structures; L. lactis, S. aureus, B. subtilis, and M. luteus peptidoglycans belong to the A-type peptidoglycan. Although there are differences in the composition of the peptide part, the A-type peptidoglycans all have L-Ala as the first amino acid, which attaches the peptide to the glycan chain.
Besides that, the cross-link in this type of peptidoglycan is always between the fourth (D-Ala) and the third amino acid (always a diaminoacid, such as meso-diaminopimelic acid (e.g. B. subtilis) or L-Lys (e.g. S. aureus)) (39). C. flaccumfaciens peptidoglycan is a B-type peptidoglycan; the first amino acid residue is not L-Ala but L-Gly. Cross-linking of the peptidoglycan is achieved by formation of a bond between D-Ala in the fourth position and the second amino acid (D-Glu) in the other peptide chain (39). The only moiety that A-type and B-type peptidoglycans have in common is the N-acetylglucosamine-Nacetylmureine polymer (glycan) (39). Furthermore, these structural differences between the peptidoglycans tested in this study, like amino acid composition and mode of cross-linking, did not seem to influence the efficiency of binding (Fig. 5B). Hence, we postulate that the LysM domain binds to glycan. The exact binding site in peptidoglycan is currently under study.
Although peptidoglycan is present on the entire cell surface of Gram-positive bacteria, the MSA2cA protein was not able to bind to the whole cell surface of L. lactis, L. casei, L. sake, and B. subtilis, as was shown by immunofluorescence microscopy studies. Apparently, other cell wall constituents prevent binding of a protein containing LysM domains, allowing binding only to those sites on the cell surface where this component is absent. These notions of localized binding fit with a very early observation that lysis in L. lactis starts at a specific locus, the equatorial ring (40). The sensitivity of the assay does not allow us to exclude the possibility that low amounts of protein can bind to the whole surface of the lactococcal cell, but the acid treatment clearly shows that the removal of cell wall components increases the binding capacity of the cell surface significantly. When AcmA-specific antibodies were used in immunofluorescence microscopy studies, no signal could be detected on the cell surface of L. lactis cells expressing AcmA. Expression of AcmA is probably too low to perform such a study (1).
The S. aureus autolysin Atl is specifically targeted to the cell division site by a repeat domain (41). The three repeats in Atl are not homologous to LysM domains, but they serve the same function in directing the enzyme to its site of action. Baba et al. (41) conclude that site-specific targeting of muralytic enzymes cannot be achieved by enzyme-substrate interactions, because the substrate peptidoglycan is present on the whole cell surface. In their view, localized binding could only be achieved by using a localized cell wall component. Here we show, however, that LysM domains do bind peptidoglycan. Mur of L. citreum and Mur1 of S. thermophilus are devoid of substrate binding domains and only contain an active site domain. Nevertheless, they have been shown to bind to the cell wall of the producing strains. A clear function of the cell wall binding domain of a peptidoglycan hydrolase is to direct the enzyme to its site of action, to the poles of the cell in the case of AcmA. Mur and Mur1 probably bind less specifically to the cell wall, probably even to the whole cell surface, which was already suggested for Mur (8).
HtrA is responsible for breakdown of AcmA (4) and also the fusion protein MSA2cA. 3 The protease cleaves in cA (42), and the breakdown products still retain cell binding ability (Fig.  5B), showing that not all three LysM domains are necessary for binding.
The component(s) that prevent the LysM domains from binding all over the cell surface are removed by boiling cells or cell walls in trichloroacetic acid; after this treatment, the cell walls bind more MSA2cA (Fig. 4), and binding takes place over the entire cell surface (Fig. 3B). Trichloroacetic acid is believed to remove cell wall-associated molecules (e.g. teichoic acids and polysaccharides) from the cell wall (33). The highly localized binding of AcmA also strongly suggests that the component hindering the binding of the enzyme is not present everywhere on the lactococcal cell surface. Chemical treatment of cell walls, especially extraction with trichloroacetic acid, has been used to identify bacteriophage receptors in the lactococcal cell wall. Extraction with trichloroacetic acid of cell walls from L. lactis strains F7/2, Wg2-1, and E8 resulted in a strong reduction of phage binding (33,43). The reduction of phage binding correlated with the release of significant amounts of carbohydrates and phosphorus-containing material. The hypothesis that a carbohydrate cell wall component covalently linked to peptidoglycan was the phage receptor in L. lactis E8 was confirmed by lectin binding studies. When lectins that specifically bind galactose or glucosamine where mixed with cells of this strain, they prevented binding of phages. This revealed that galactose and glucosamine were required for the adsorption of the phages.
L. lactis SK110 contains LTA with galactosyl units and is resistant to phage attack (37). Using the fluorescein-labeled lectin RCA120, we show that galactosyl-decorated LTA is not uniformly distributed over the cell wall of strain SK110 and seems to be present at those sites to which MSA2cA does not bind; whereas MSA2cA mainly binds to the poles and the septum, RCA120 binds to the entire cell except the poles and septum. Assuming that all LTA is decorated with galactosyl residues in this strain, this would suggest that LTA is present at specific sites on L. lactis cells and that LTA is involved in the hindering of binding of MSA2cA to the cell surface. The galactosyl moiety of LTA is not involved in this phenomenon, since no galactosyl residues could be detected on the cell surface of MG1363, whereas MSA2cA binds to MG1363 cells at the same specific loci as seen in SK110. LTA, a decoration other than galactosyl residues, or a cell wall carbohydrate associated with LTA present in both MG1363 and SK110 is probably involved in hindering of LysM domain binding to peptidoglycan. The exact nature of this component is currently under study.
We have shown in this paper that L. lactis regulates the binding of AcmA by allowing the autolysin to bind only there where it is needed. In doing so, L. lactis protects itself against the potentially lethal activity of its autolysin.