The cell wall binding domains (CBD) of bacteriophage endolysins target the enzymes to their substrate in the bacterial peptidoglycan with extraordinary specificity. Despite strong interest in these enzymes as novel antimicrobials, little is known regarding their interaction with the bacterial wall and their binding ligands. We investigated the interaction of Listeria phage endolysin PlyP35 with carbohydrate residues present in the teichoic acid polymers on the peptidoglycan. Biochemical and genetic analyses revealed that CBD of PlyP35 specifically recognizes the N-acetylglucosamine (GlcNAc) residue at position C4 of the polyribitol-phosphate subunits. Binding of CBDP35 could be prevented by removal of wall teichoic acid (WTA) polymers from cell walls, and inhibited by addition of purified WTAs or acetylated saccharides. We show that Listeria monocytogenes genes lmo2549 and lmo2550 are required for decoration of WTAs with GlcNAc. Inactivation of either gene resulted in a lack of GlcNAc glycosylation, and the mutants failed to bind CBDP35. We also report that the GlcNAc-deficient phenotype of L. monocytogenes strain WSLC 1442 is due to a small deletion in lmo2550, resulting in synthesis of a truncated gene product responsible for the glycosylation defect. Complementation with lmo2550 completely restored display of characteristic serovar 1/2 specific WTA and the wild-type phenotype.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Bacteriophage endolysins are cell wall hydrolysing enzymes synthesized in phage-infected cells at the end of the lytic multiplication cycle. They digest the bacterial murein sacculus from within and induce lysis of the host cell, enabling progeny virions to be released (Loessner, 2005; Borysowski et al., 2006). Because most endolysins do not feature signal peptides, they gain access to the bacterial peptidoglycan through the action of holins, which oligomerize in the membrane, thereby producing small pores allowing the endolysins to reach their substrate (Young, 1992; Wang et al., 2000). Endolysins of phages for Gram-positive bacteria commonly show a modular organization of at least two distinct functional domains: an N-terminal enzymatically active domain and a C-terminal cell wall binding domain (CBD). The enzymatically active domain determines the enzymatic mechanism of the protein, whereas the CBD is responsible for high cell wall specificity of the enzyme, targeting the protein to the ligands present in or on the bacterial cell wall (Loessner et al., 2002; Loessner, 2005). Both their unique catalytic mechanisms and their specificity in substrate recognition and cell wall binding contribute to the well-defined and specific spectrum of lytic activity of endolysins.
When exogenously applied to Gram-positive cells, endolysins can digest the peptidoglycan and cause lysis from without due to the absence of an outer membrane in these bacteria (Loessner, 2005). Because of the highly specific lysis and killing, endolysins represent interesting antibacterial agents (Fischetti, 2005; Loessner, 2005; Borysowski et al., 2006; Hermoso et al., 2007). Recombinant CBD polypeptides can also be used for immobilization of bacterial target cells on solid surfaces for subsequent magnetic separation and detection, and, when tagged with fluorescent proteins, for differential staining and identification (Kretzer et al., 2007; Schmelcher et al., 2010).
Several different Listeria phage endolysins have been characterized and analysed for their binding spectra and catalytic mechanism (Loessner et al., 1995; 1996; 2002; Zimmer et al., 2003; Korndörfer et al., 2006; Schmelcher et al., 2010) and the individual functional domains can be recombineered to further increase lytic activity and binding affinity (Schmelcher et al., 2011). However, next to nothing was known regarding the underlying molecular principles and genes responsible for those processes, and specific ligands responsible for binding of the endolysin CBDs had not been identified. Previous work indicated that binding of CBDs of Listeria phage lysins Ply118 and Ply500 relies on serovar- and strain-specific carbohydrate ligands, which most likely are covalently anchored to the Listeria peptidoglycan (Loessner et al., 2002).
Listeria peptidoglycan belongs to the A1γ chemotype, featuring meso-diaminopimelic acid directly cross-linked with the fourth amino acid (d-Ala) of a neighbouring stem peptide (Schleifer and Kandler, 1972). Based on surface markers such as somatic (O) and flagellar (H) antigens, the genus Listeria features 16 different serovars, and 13 serovars among the groups 1/2, 3, 4 and 7 are associated with the pathogen Listeria monocytogenes (Seeliger and Höhne, 1979; Gorski, 2008). Variation among the different serovars is mostly determined by differences in carbohydrate substitution of the polyribitol-phosphate (RboP) subunits of wall teichoic acids (WTA) (Fiedler et al., 1984; Fujii et al., 1985; Uchikawa et al., 1986). Serovar groups 1/2, 3 and 7 strains carry poly-RboP WTAs and bear N-acetylglucosamine (GlcNAc) and/or rhamnose (Rha) as substituents on the C2 and C4 position of the ribitol backbone, whereas in serovar 4, 5 and 6 cells, the GlcNAc moiety is incorporated into the WTA chain and by itself decorated with galactose (Gal) and/or glucose (Glc) residues (Uchikawa et al., 1986; Fiedler, 1988).
It was previously demonstrated that WTAs are required for binding of phage endolysin PlyP35 (Schmelcher et al., 2010). However, the precise nature of the ligand remained unclear. Here, we show that CBDP35 specifically binds to WTA-associated and specifically oriented GlcNAc moieties. We employ in silico analysis and mutant Listeria to identify two genes required for WTA glycosylation, and show that the naturally GlcNAc-deficient strain WSLC 1442 features a deletion in one of these genes.
CBDP35 binds to WTAs
Previous studies suggested that the WTAs on Listeria cell walls may be required for binding of endolysin CBDs (Loessner et al., 2002; Schmelcher et al., 2010). The basic structures of the WTA molecules featured by the strains used in this study are depicted in Fig. 1. To determine the interaction of the WTA with CBDP35, cell walls of L. monocytogenes strain EGDe were first subjected to a mild extraction of attached carbohydrate polymers, including WTAs, followed by fluorescence assays using green fluorescent protein (GFP)-tagged CBDP35.
Figure 2 shows that exposure of the Listeria wall to pH 6.0, 5.0 and 4.0 had no significant effect on the binding of CBDP35, indicating that the peptidoglycan and associated carbohydrate polymers are not affected by these mildly acidic conditions. When cell walls were treated with lower pH buffers (pH 3.0 and 2.5), however, CBD binding was strongly reduced; exposure to glycine/HCl at pH 2.5 resulted in a sevenfold decrease of GFP fluorescence (Fig. 2). The decrease in cell wall-associated phosphorus at low pH reflects the removal of WTAs from cell walls. It also correlates well with the decrease in CBDP35 binding, indicating the function of WTA as ligand.
To more directly demonstrate the role of WTA in CBDP35 binding, purified and soluble WTA from L. monocytogenes strains EGDe (serovar 1/2a), WSLC 1442 (serovar 1/2a, GlcNAc-deficient), WSLC 1020 (serovar 4a) and WSLC 1042 (serovar 4b) were used to competitively inhibit the binding of CBDP35 to purified walls of L. monocytogenes EGDe. WTA from EGDe and WSLC 1020 had a strong effect on binding of fluorescent CBDP35 (inhibition approximately 60% and 80%, respectively), whereas the WTA polymers from WSLC 1442 and WSLC 1042 did not hinder binding (Fig. 3). These results are consistent with the findings that CBDP35 strongly bound to EGDe and WSLC 1020 cells, whereas strains WSLC 1442 and 1042 were not labelled (Table 1). Taken together, these data strongly indicate that WTA and attached carbohydrates serve as binding ligands for CBDP35. Intriguingly, the use of purified WTAs revealed that WTAs of strain EGDe were able to inhibit the binding, while those from strain WSLC 1442 did not affect the interaction.
Table 1. Binding of fluorescently labelled CBDP35 and WGA to L. monocytogenes cells.
Previous work showed that strain WSLC 1442 features GlcNAc-deficient WTA (Wendlinger et al., 1996), which prompted us to consider GlcNAc as a likely CBDP35 ligand. In order to provide more details on the requirements for CBDP35 recognition, we tested the effect of single carbohydrate molecules known to occur as substituents in Listeria serovar 1/2 and 4 WTA, such as GlcNAc, Rha, Glc and Gal. GFP-tagged CBDP35 was pre-incubated with these carbohydrates, and binding to intact L. monocytogenes cells was measured by quantitative fluorescence analysis (Fig. 4). We found that GlcNAc reduced the binding of CBDP35 to serovar 1/2a cells of strain EGDe by almost 80%, while Rha, Glc and Gal had no effect, again indicating that the binding of CBDP35 to the Listeria cell walls requires GlcNAc decoration of the WTAs.
We then tested CBDP35 binding inhibition by a variety of N-acetylated sugars [GlcNAc, N-acetylgalactosamine (GalNAc), N-acetylmuramic acid (MurNAc) and N-acetylmannosamine (ManNAc)] on purified cell walls of strain EGDe. Here, GlcNAc again significantly inhibited CBDP35 binding by approximately 35% (Fig. 5). Surprisingly, GalNAc and MurNAc were almost as effective as GlcNAc and reduced the binding of CBDP35 by 28% to 32% (Fig. 5). The fact that non-N-acetylated Glc or Gal were ineffective (Fig. 4) demonstrates the importance of N-acetylation for CBDP35 recognition. However, ManNAc showed no inhibitory activity. It features an axial orientation of the N-acetyl group, whereas GlcNAc, GalNAc and MurNAc all have N-acetyl groups in the equatorial orientation (Fig. 5). This small but significant structural difference suggests that configuration of the N-acetyl group is critical for CBDP35 recognition and binding. Taken together, these data indicate that terminal GlcNAc residues in WTAs of L. monocytogenes serovar 1/2 and 3 strains serve as binding sites for CBDP35, with an equatorial N-acetyl group as a possible key feature.
Inactivation of L. monocytogenes genes lmo2549 or lmo2550 results in WTA GlcNAc deficiency
The lmo2549 gene in L. monocytogenes serovar 1/2a strain EGDe (Glaser et al., 2001) encodes a putative WTA glycosylation protein of 145 amino acids, which shares 82% identity with GtcA of L. monocytogenes serovar 4b strains (Promadej et al., 1999; Autret et al., 2001). It is flanked by another serovar 1/2 specific gene (lmo2550), specifying a protein of 315 residues of unknown function. Gene lmo2550 was previously designated as csbB due to sequence similarity of its encoded protein with CsbB of Bacillus subtilis (Autret et al., 2001). Based on sequence homologies, both products of lmo2549 and lmo2550 are possibly involved in WTA glycosylation. To test the role of these two genes on WTA carbohydrate decoration, we generated Δlmo2549 and Δlmo2550 null mutants in the L. monocytogenes EGDe wild-type background. Both EGDeΔlmo2549 and EGDeΔlmo2550 showed no visible morphological alteration or growth defects, indicating that cell division or viability was not affected by the in-frame deletions.
Compositional analyses of the WTA repeating units of wild-type EGDe and its deletion mutants EGDeΔlmo2549 and EGDeΔlmo2550 were performed by electrospray ionization tandem mass spectrometry (ESI-MS/MS). Representative spectra of the WTA monomers are shown in Fig. 6. For EGDe, the major peak at m/z 502.22 characterizes the [M + H]+ ion of the monomeric WTA unit. Additional ions are ascribed to the Li+ (m/z +7) and the Na+ adduct (m/z +23) of the WTA monomer. ESI-MS/MS analysis yielded two additional peaks at m/z 356.15 and 204.09, indicating the presence of both Rha and GlcNAc. The WTA monomers of the mutant strains EGDeΔlmo2549 and EGDeΔlmo2550 revealed essentially identical ion spectra, which are in agreement with their predicted WTA composition. Both spectra are dominated by a strong m/z 305.16 signal, corresponding to the lithiated [M + Li]+ monomeric WTA unit. The additional ions were identified as the sodium adducts [M + Na]+ of the WTA monomers. Subsequent MS/MS analyses of the [M + Li]+ ions at m/z 305.16 of both WTAs generated fragmentation peaks at m/z 159.10, which are consistent with a ribitol backbone resulting from elimination of the Rha (−146). In conclusion, the data confirm that disruption of genes lmo2549 or lmo2550 in serovar 1/2 strains leads to absence of GlcNAc substitution in EGDe WTA.
In addition, the presence or absence of GlcNAc was monitored using GFP-labelled CBDP35 and a wheat germ agglutinin (WGA)-Alexa Fluor 594 conjugate (Fig. 7 and Table 1). GlcNAc residues were recognized on the parental strain EGDe, whereas EGDeΔlmo2549 and EGDeΔlmo2550 were negative. Transcomplementation of the mutants in EGDeΔlmo2549::pPL2(lmo2549) and EGDeΔlmo2550::pPL2(lmo2550) completely restored CBDP35 binding to normal levels (Fig. 7). We conclude that lmo2549 and lmo2550 are required for glycosylation of WTAs with GlcNAc, required for binding of CBDP35.
GlcNAc deficiency in L. monocytogenes strain WSLC 1442 is due to lmo2550 mutation
It was previously reported that the serovar 1/2a L. monocytogenes strain WSLC 1442 is a mutant lacking GlcNAc in its WTA (Wendlinger et al., 1996). However, the reason for this unusual phenotype was not known. To address the question whether the GlcNAc deficiency in this strain could be due to mutations in lmo2549 or lmo2550, we determined the sequences of the respective WSLC 1442 genes and compared them with L. monocytogenes EGDe. While lmo2549 was identical, lmo2550 from WSLC 1442 features a two nucleotide (nt) deletion (nt 749–750) (Fig. 8). The mutation leads to a frameshift and introduction of a premature translation stop at position 259, resulting in a C-terminally modified protein lacking the last 57 amino acids. The truncated product is probably non-functional and unable to mediate the GlcNAc decoration of WTAs.
To provide direct evidence for this assumption, strain WSLC 1442 was complemented with an intact copy of the gene on pPL2(lmo2550). The results of CBDP35 binding assay (Fig. 7) confirmed that the two base pair deletion in lmo2550 is in fact responsible for the lack of GlcNAc, and that complementation with the gene can restore a GlcNAc proficient phenotype in the WSLC 1442 cells.
Phage endolysins display a modular architecture composed of two functional domains, comprising a catalytic and a substrate recognition site (Loessner, 2005). Until now, only a few ligands for bacteriophage endolysins have been identified. Endolysin Cpl-1 from pneumococcal phage Cp-1 was shown to possess choline-binding repeats for anchoring the protein to choline moieties in the pneumococcal WTA (Garcia et al., 1988; Hermoso et al., 2003; 2007), and the endolysin Lyb5 from the Lactobacillus fermentum bacteriophage ΦPYB5 contains LysM domains, which are thought to bind to peptidoglycan (Hu et al., 2010).
Although several Listeria phage endolysins have been examined (Loessner et al., 2002; Korndörfer et al., 2006; Schmelcher et al., 2010), identity of the ligands recognized by their CBD domains remained unknown. Our data presented here demonstrate that CBDP35 specifically interacts with the GlcNAc carbohydrate residues in the poly-RboP WTAs of L. monocytogenes, which is the first report of a direct interaction between a phage endolysin and WTA-associated sugar decoration.
GFP-tagged CBDP35 attaches to the cell surface of all L. monocytogenes serovar 1/2 and 3 strains (except some unusual isolates such as strain WSLC 1442), and to many serovar 4, 5 and 6 strains (Schmelcher et al., 2010). In the WTA molecules found in serovars 1/2, 3, and the rare serovar 7, the ribitol monomers are directly connected via phosphodiester bonds between C1 and C5 of adjacent units and may be substituted with GlcNAc and/or Rha residues at C2 and C4, depending on the serovar (Fiedler and Ruhland, 1987; Fiedler, 1988). Interestingly, CBDP35 fails to bind to strains WSLC 1442 and the serovar 7 strain WSLC 1034, both of which lack GlcNAc residues in their WTAs (Seeliger and Höhne, 1979; Wendlinger et al., 1996).
This prompted us to investigate the role of peptidoglycan-associated WTAs on the binding of CBDP35. We found that stepwise depletion or complete removal of WTAs on cell walls resulted in reduction or prevention of CBDP35 binding, indicating that the appropriate type of WTA is required and sufficient as a ligand. The latter was further confirmed by competitive inhibition using purified WTA molecules featuring terminal GlcNAc residues on the ribitol backbone. In contrast, Rha appears to have no influence on binding of CBDP35, as serovar 3 strains lacking this sugar exhibit the same uniform decoration of the cell surface as the serovar 1/2 strains (Schmelcher et al., 2010). Thus, the failure of WSLC 1442 (only Rha, no GlcNAc) and WSLC 1034 (no Rha, no GlcNAc) to bind CBDP35 is solely due to the lack of terminal GlcNAc substitution in the WTA molecules.
CBDP35 was able to decorate various strains of serovar 4, 5 and 6. Although the WTAs of serovar 4a (e.g. WSLC 1020) and 4b strains (e.g. WSLC 1042) both contain GlcNAc, only WTAs of serovar 4a were able to act as ligand for CBDP35. This difference is explained by structural WTA variation. In WTAs of serovars 4, 5 and 6, GlcNAc is integrated into the poly-RboP chain (Uchikawa et al., 1986; Fiedler and Ruhland, 1987; Fiedler, 1988) and may be decorated with glucosyl and/or galactosyl residues (Uchikawa et al., 1986). Serovar 4b strains are unique in that GlcNAc is substituted with both Gal and Glc, while serovar 4a strains feature no additional carbohydrates linked to the integrated GlcNAc (Uchikawa et al., 1986). Therefore, it seems reasonable to assume that serovar 4a WTA features fully accessible GlcNAc molecules, whereas the two sugar substituents in 4b WTA prevent access of CBDP35 to its recognition site. Similarly, the binding of CBDP35 to strains of serovar 4c and 4d (Schmelcher et al., 2010) may be explained by their less complex WTA structure, in which the integrated GlcNAc is substituted by only a single carbohydrate (Fujii et al., 1985; Fiedler, 1988) and may still be accessible to CBDP35. The molecular details on how exactly CBDP35 recognizes GlcNAc are not yet clear. Besides GlcNAc as important structural motif, other parts of the poly-RboP chain seem to be required for recognition and binding of CBDP35. More detailed analysis of the molecular interaction between WTA and CBDP35 is currently being attempted, using nuclear magnetic resonance and X-ray crystallography.
It was interesting to note that the binding of CBDP35 to cell walls could not only be inhibited by GlcNAc, but also by other (structurally related) N-acetylated monosaccharides such as GalNAc and MurNAc, while ManNAc had no effect. This finding can be explained by the assumption that CBDP35 requires the N-acetyl group in an equatorial configuration, in contrast to the axial orientation offered by ManNAc. To further address the question whether selective binding of CBDP35 to GlcNAc in WTAs depends on a binding pocket for N-acetylated carbohydrates on poly-RboP, structural analysis of the protein–ligand complex is required, which will help to understand the specific interaction on the molecular scale.
Aside from its specificity for Listeria cells, CBDP35 was also reported to be able to bind to some Staphylococcus aureus strains (Schmelcher et al., 2010). This unusual finding can now be explained by the fact that most S. aureus strains also feature poly-RboP WTA on their surfaces (Endl et al., 1983), which can be further substituted with d-alanine at the C2 position, or GlcNAc at the C4 position (Sanderson et al., 1962; Vinogradov et al., 2006). Because the latter conformation precisely reflects the corresponding structure in L. monocytogenes, it can be assumed that CBDP35 also recognizes and binds to the GlcNAc residues in S. aureus poly-RboP WTA. In contrast, CBDP35 is unable to bind to either group A or group C streptococci (data not shown). Their surface carbohydrates consist of a polyrhamnose backbone, featuring either GlcNAc (group A) or GalNAc (group C) as substitution (Coligan et al., 1978). This finding again indicates that binding of CBDP35 to GlcNAc is dependent on its structural context, namely the poly-RboP backbone.
The reasons and constraints for species- and serovar-specific differences in WTA structure and glycosylation pattern and the underlying genetic variation in Listeria strains of different serovar groups remain to be elucidated. In L. monocytogenes, several genes have been reported to affect carbohydrate decoration of WTAs, e.g. gtcA (Promadej et al., 1999; Autret et al., 2001), gltA and gltB (Lei et al., 2001), and glcV (Spears et al., 2008). In serovar 4b, GtcA was reported to be important for Gal and Glc substitution of the integrated GlcNAc molecule (Promadej et al., 1999).
We present convincing biochemical and genetic evidence for the role of lmo2549 and lmo2550 in WTA glycosylation with GlcNAc in serovar 1/2 and 3. The role of GlcNAc for CBDP35 binding was demonstrated by using GlcNAc-deficient mutants, and subsequent complementation. Genes lmo2549 and lmo2550 were previously designated as gtcA and csbB (Autret et al., 2001), based upon their sequence homology to gtcA from a serovar 4b strain (Promadej et al., 1999), and to csbB of B. subtilis (Huang and Helmann, 1998). Based on Tn1545-based insertional inactivation of gtcA (lmo2549) in strain EGD (serovar 1/2a), the authors suggested that the gene might be involved in glycosylation of WTAs with Rha, as its inactivation did not affect adsorption of phage A118 (Autret et al., 2001). However, while binding of phage A118 is associated with Rha in WTA molecules (Wendlinger et al., 1996), GlcNAc is not required. It is now clear that the previously found inhibition of A118 binding by high concentrations of glucosamine hydrochloride (Wendlinger et al., 1996) was due to the high concentrations of chloride ions in the assay, which reflects the sensitivity of A118 binding to ionic strength (M.R. Eugster et al., unpublished). Contrary to its N-acetylated derivative, glucosamine itself is not present in Listeria WTA, but represents a by-product of chemical WTA sample preparation and derivatization of GlcNAc (Kamisango et al., 1983).
Mass spectrometry analyses and binding assays clearly indicate that lmo2549 and lmo2550 directly affect GlcNAc substitution in WTAs of L. monocytogenes serovar 1/2 strains, whereas there is no effect on the Rha substitution. Mass spectra for WTAs of mutants Δlmo2549 and Δlmo2550 are in agreement with corresponding mass spectrometry results for the GlcNAc-deficient strain WSLC 1442 (Eugster and Loessner, 2011). However, it remains unknown how lmo2549 and lmo2550 and their products contribute to the linkage of GlcNAc residues to the WTA backbone. Interestingly, lmo2549 (gtcA) is present in both serovars 4b (Promadej et al., 1999) and 1/2 (Autret et al., 2001), although both authors have assigned different putative functions to the gene product. Based on sequence homologies and in silico analyses, lmo2549 (gtcA) in L. monocytogenes EGDe encodes a GtrA-like protein, predicted to be an integral membrane protein involved in synthesis of cell surface polysaccharides, and lmo2550 is predicted to encode a putative glycosyltransferase (CAZy family glycosyltransferase 2). These enzymes catalyse transfer of sugar moieties from carriers such as UDP-glucose, UDP-N-acetylgalactosamine, GDP-mannose or TDP-rhamnose to various substrates including cellulose, dolichol phosphate and teichoic acid. Lmo2550 shares high identity scores with CsbB of B. subtilis (Autret et al., 2001), which is suggested to be a putative membrane-anchored glycosyl transferase participating in peptidoglycan biogenesis (Akbar and Price, 1996; Huang and Helmann, 1998), and also shows similarity to a bactoprenol glucosyltransferase (Mavris et al., 1997). Because lmo2549 and lmo2550 are organized in tandem, it is likely that both proteins act in concert to transfer GlcNAc residues to the WTA backbone. Also, the observation that lmo2549 and lmo2550 mutants do not suffer any obvious growth defects indicates that no major cellular biosynthesis pathways are affected by their inactivation.
L. monocytogenes WSLC 1442 lacks GlcNAc in its WTAs (Wendlinger et al., 1996; Eugster and Loessner, 2011). The genetic approach described here revealed that this deficiency is due to a small deletion in lmo2550, resulting in a frameshift and a C-terminally truncated gene product. Because this gene is required for GlcNAc attachment, this mutation is responsible for lack of the sugar in the WSLC 1442 WTA. Together with the genetic complementation analysis using lmo2550, we conclude that this particular isolate represents a mutant of an otherwise normal serovar 1/2 strain. In fact, this phenotype seems to be not infrequent (M.R. Eugster, unpubl. obs.), and may actually receive positive selection by the fact that the lack of GlcNAc residues in WTAs can confer phage resistance as a result of decreased adsorption (Wendlinger et al., 1996). Recent data from our laboratory indicate that this GlcNAc-negative phenotype can efficiently be selected for by challenge with phage that requires the molecule as a binding ligand, either for its tail associated receptor proteins or the endolysin CBDs, both of which recognize WTA as their primary binding ligands.
Bacterial strains, plasmids and culture conditions
Strains and plasmids used in this study are listed in Table 2. Escherichia coli were grown aerobically at 37°C in Luria–Bertani (LB) medium, and used for amplification of plasmids and recombinant protein synthesis. L. monocytogenes were cultured in brain heart infusion (BHI) medium or tryptose broth, at 30°C with shaking. Streptococcus pyogenes and Streptococcus equi were cultivated anaerobically at 37°C, in BHI media. Ampicillin (100 µg ml−1 for E. coli), tetracycline (30 µg ml−1 for E. coli), and chloramphenicol (10 µg ml−1 for E. coli or L. monocytogenes) were added to the media as required (Table 2).
pPL2 harbouring lmo2549 under the control of Piap; Camr
pPL2 harbouring lmo2550 under the control of Piap; Camr
DNA techniques and biochemical analysis
General molecular biology techniques were performed by using standard protocols (Sambrook and Russel, 2001). T4 DNA ligase and restriction enzymes were purchased from Roche Diagnostics (Rotkreuz, Switzerland) and Fermentas (Le Mont-sur-Lausanne, Switzerland), and used according to the manufacturer's instructions. DNA fragments and PCR products used in cloning steps were created with Phusion High-Fidelity DNA polymerase (Finnzymes, Espoo, Finland) and purified with GenElute PCR Clean-Up Kit (Sigma-Aldrich, Buchs SG, Switzerland) when needed. Plasmids were purified using the GenElute Plasmid Miniprep Kit (Sigma-Aldrich). Oligonucleotide primers for PCRs were designed according to the sequence of L. monocytogenes serovar 1/2a strain EGDe (GeneBank/EMBL accession No. AL591824). All plasmid constructs were transformed into E. coli strain XL1-Blue MRF' for propagation. All PCR products, plasmids, and mutant bacteria were verified by DNA sequencing. Sugars and other carbohydrates used for competitive binding inhibition were purchased from Sigma-Aldrich.
Protein production and purification
Production of a His-tagged CBD of phage lysin PlyP35 fused to GFP (HGFP-CBDP35) was carried out as described before (Loessner et al., 2002; Schmelcher et al., 2010). Briefly, cultures of E. coli strain XL1-Blue MRF' harbouring plasmid pHGFP-CBDP35 (Schmelcher et al., 2010) were grown at 30°C in modified LB medium (15 g l−1 tryptose, 8 g l−1 yeast extract, 5 g l−1 NaCl) supplemented with tetracycline (30 µg ml−1) and ampicillin (100 µg ml−1). At an OD600 of 0.5, gene expression was induced by adding 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG). After incubation at 30°C for 4 h, cultures were stored overnight at 4°C to allow complete maturation of the GFP chromophore, harvested by centrifugation, and resuspended in buffer A (50 mM NaH2PO4, 500 mM NaCl, 5 mM imidazole, 0.1% Tween 20, pH 8.0). Cells were broken using a French Press (SLM Aminco) at 100 MPa, and the lysate was centrifuged (20 000 g, 45 min), filter-sterilized (0.2 µm PES membrane), and the raw extracts stored at −20°C. After thawing, recombinant HGFP-CBDP35 was purified by Immobilized Metal Affinity Chromatography (IMAC) using an ÄKTA Purifier and an XK 16 column (GE Healthcare, Glattbrugg, Switzerland) packed with nickel-nitrilotriacetic acid (Ni-NTA) resin (Qiagen, Basel, Switzerland). After extensive washing with buffer A, buffer B (buffer A plus 250 mM imidazole) was used for elution of His-tagged protein. Fractions were concentrated (Vivaspin columns, cut-off 10 kDa; Sartorius, Dietikon, Switzerland), and dialysed against two changes of dialysis buffer (50 mM NaH2PO4, 100 mM NaCl, 0.005% Tween 20, pH 8.0) before storage at −20°C. Protein purity was assayed by SDS-PAGE, and concentrations were determined using a NanoDrop ND-1000 spectrophotometer. Adsorption coefficients were calculated as described earlier (Gill and von Hippel, 1989).
Preparation of Listeria cell walls
Crude cell walls from L. monocytogenes strain EGDe used for binding assays and extraction of WTAs were isolated from exponentially growing bacteria and purified as described previously (Fiedler et al., 1984; Valyasevi et al., 1990; Wendlinger et al., 1996; Navarre et al., 1999; Eugster and Loessner, 2011). Briefly, bacteria were grown in tryptose broth medium to an OD600 of 0.7, heat-killed by steaming the culture for 30 min, harvested by centrifugation (7000 g, 10 min, 4°C), and resuspended in SM buffer (100 mM NaCl, 8 mM MgSO4, 50 mM Tris, pH 7.5). Cells were lysed by two passages through a One Shot Cell Disrupter (Constant Cell Disruption System, Northants, UK), at a pressure of 270 MPa. Cell debris and unbroken cells were removed by centrifugation (1400 g, 5 min), and cell walls were recovered by centrifugation of the supernatant (20 000 g, 30 min, 4°C). The pellet containing the cell envelopes was washed three times with water and resuspended in SM buffer. Cell walls were then treated with DNase and RNase (25°C, 4 h), and subsequently incubated with proteinase K (25°C, 2 h) with gentle mixing (each enzyme with a final concentration of 100 µg g−1 wet cell walls). Then, cell walls were harvested by centrifugation, resuspended in 4% SDS (w/v, in water) and incubated for 30 min at 100°C. All SDS-insoluble material was collected by centrifugation (20 000 g, 30 min, 20°C), and the detergent removed by extensive washing (5 times) with deionized water. Finally, cell walls were resuspended in water, lyophilized, and stored at −20°C.
Extraction and purification of WTAs
Lyophilized peptidoglycan was resuspended in 25 mM glycine/HCl extraction solution (pH 2.5), and boiled at 100°C for 10 min to release the WTAs from the peptidoglycan backbone (Kojima et al., 1983; Fujii et al., 1985; Kaya et al., 1985). Insoluble material was sedimented by centrifugation (30 000 g, 30 min, 4°C), and the extraction repeated. Finally, supernatants containing the soluble WTA polymers were pooled and dialysed against purified water at 4°C (Spectra/Por, MWCO 1000). The crude WTA preparations were further purified by anion exchange chromatography (IEX) on a HiTrap DEAE-Sepharose Fast Flow column (GE Healthcare), as described earlier (Fiedler et al., 1984; Wendlinger et al., 1996; Eugster and Loessner, 2011). The material was loaded onto the column equilibrated with starting buffer (10 mM Tris/HCl, pH 7.5), and eluted with a linear salt gradient of 0 to 1 M NaCl in approximately 20 column volumes, at a flow rate of 1 ml min−1. Elution was monitored at 205 nm, and samples were assayed for total phosphorus content to identify WTA-containing fractions. Phosphate content was determined photometrically as described elsewhere (Eugster and Loessner, 2011). Finally, phosphorus-containing fractions containing purified WTA polymers were pooled, dialysed against water (Spectra/Por, MWCO 1000 Da), lyophilized, and stored at −20°C until use.
Analysis of WTA carbohydrates by mass spectrometry
Mass spectrometry analysis of WTA carbohydrates was performed according to a novel procedure (Eugster and Loessner, 2011). In brief, purified WTA polymers were selectively cleaved to monomeric units by hydrolysis of phosphodiester bonds using 48% hydrogen fluoride for 16 h at 0°C (Glaser and Burger, 1964; Lipkin et al., 1969; Fiedler et al., 1981). After degradation, lyophilized WTA monomers were dissolved in a mixture of 50% acetonitrile/50% water (v/v) containing 3 mM lithium trifluoroacetate (Li-TFA). Samples were subjected to ESI-MS/MS analysis at the Functional Genomics Center Zurich (FGCZ) of ETH Zurich and the University of Zurich (Switzerland). Positive ion mass spectra were acquired on a Q-TOF Ultima API mass spectrometer (Micromass, UK).
Cell wall treatments
Purified lyophilized cell walls of L. monocytogenes strain EGDe were treated with buffers of various pH values and tested for CBDP35 binding to examine the influence of pH on ligand extraction. The chemical treatments were as follows: 100 mM sodium citrate buffer at pH 6.0, 5.0, and 4.0; 25 mM glycine/HCl buffer at pH 3.0 and 2.5. Treatments of cells walls with these buffers were carried out as described above for extraction of WTA molecules, i.e., heating to 100°C for 10 min. After the treatments, wall material was washed twice with H2O, lyophilized and resuspended in PBST buffer. Then, these cell walls were used for CBDP35 binding tests by fluorescence plate assays as described below. In addition, supernatant fractions of treated cell wall suspensions were assayed for total phosphate content to determine the degree of WTA removal.
Cell wall decoration of Listeria cells and cell walls
The ability of GFP-labelled CBDP35 or WGA-Alexa Fluor 594 conjugate (Invitrogen, Basel, Switzerland) to bind to Listeria cells, other bacteria, and isolated walls was tested by fluorescence binding assays as reported earlier (Loessner et al., 2002; Eugster and Loessner, 2011). Bacteria cells from late log phase were harvested by centrifugation (7000 g, 1 min, 4°C) and resuspended in 1/10 volume of PBST (50 mM NaH2PO4, 120 mM NaCl, 0.1% Tween 20, pH 8.0). Purified cell walls were diluted to a final concentration of 10 µg ml−1 in PBST.
For qualitative binding assays, 100 µl of cells was mixed with 50 µl of a stock solution of 0.1 mg ml−1 HGFP-CBDP35 and incubated for 5 min at room temperature. Bacteria or cell walls with bound protein were recovered by centrifugation at 16 000 g for 1 min, washed twice with PBST buffer, and resuspended in 50 µl of PBST buffer for fluorescence microscopy (Leica TCS SPE; Leica, Heerbrugg, Switzerland). Binding assays with WGA were performed in the same fashion, using a WGA stock solution of 0.1 mg ml−1.
For quantitative fluorescence measurements, 100 µl of cells or cell wall suspension (prepared as described above) was incubated with 6 µl of a HGFP-CBDP35 stock solution (0.1 mg ml−1). Supernatants of the centrifuged binding assays containing unbound HGFP-CBDP35 were dispensed into black 96-well microplates, and fluorescence readings were performed using a Victor3 multilabel counter (PerkinElmer, Schwerzenbach, Switzerland). The fraction of protein bound to cells or cell walls could then be calculated by subtracting the fluorescence of unbound protein in the supernatant from total fluorescence of the controls (tubes with no cells added).
Competitive inhibition of CBDP35 cell wall association
A series of experiments were performed to examine the binding specificity of CBDP35 and the effect of competitive inhibition. Towards this end, purified WTA polymers of Listeria strains EGDe, WSLC 1442, WSLC 1020 and WSLC 1042, as well as various WTA-associated single sugars and acetylated carbohydrates (Glc, Rha, Gal, GlcNAc, GalNAc, MurNAc and ManNAc) were tested for their effectiveness in suppressing binding of HGFP-CBDP35 to bacterial cells and cell walls. Carbohydrates or purified WTAs were pre-incubated with HGFP-CBDP35 for 10 min at room temperature, and then added to cells or cell walls. Carbohydrates were dissolved in water, sterile-filtered (0.2 µm PES membrane), and added to a final concentration of 200 mM. Purified WTAs were used for competitive inhibition at a final concentration of approximately 1 µg µl−1. Binding of the protein was then quantitatively determined as described above. All experiments were performed in triplicate. Results are stated as mean values ± 1 standard deviation (SD).
L. monocytogenes genes lmo2549 and lmo2550
Based on in silico sequence alignments and analysis using CLC Main Workbench (Aarhus, Denmark), we identified two putative genes predicted to be involved in WTA glycosylation. The open reading frames (ORFs) corresponding to lmo2549 and lmo2550 were amplified from chromosomal DNA of L. monocytogenes strain EGDe, using primer pair 5′-TGT TTC GCT TGA GCT CTT AGT AGA ACC TGA C-3′ and 5′-TGC TGG TTT CGC TAT CTC ATT AGG CAC-3′ for lmo2549, resulting in a PCR fragment of 884 bp. Primer pair 5′-GTT GTA AAA CCA CCC ATG ATT AAA TAC ATC AAG ATA C-3′ and 5′-CAA TGA AGC TTT TTA TGA ATT ACT ACA AGA GGA AAT G-3′ was used for gene lmo2550, yielding a sequence of 1168 bp. The same primer pairs were used to verify the amplification products by DNA sequencing.
Generation of L. monocytogenes deletion mutants
In-frame deletion and generation of L. monocytogenes mutants Δlmo2549 and Δlmo2550 was performed using splicing-by-overlap-extension (SOE) PCR (Horton et al., 1989), followed by allelic exchange mutagenesis. In order to allow homologous recombination of chromosomal DNA and a plasmid, SOE primers were designed to amplify ∼ 500 bp sequences flanking the ORFs of interest. L. monocytogenes EGDe chromosomal DNA was utilized as template for PCR. The upstream region of ORF lmo2549 was amplified using SOE primer pair lmo2549_f_del_A (5′-ATC AGG ATC CGC AAC TTC ACA GGG AAA AG-3′) and lmo2549_r_del_B (5′-GAA AAT GGT TAG ACG TGT TTA GTA AAT GGA TCA TTT TC-3′), and the downstream region was generated with SOE primers lmo2549_f_del_C (5′-GAT CCA TTT ACT AAA CAC GTC TAA CCA TTT TCT TAT TTT GTT C-3′) and lmo2549_r_del_D (5′-ATC AAA GCT TCC AGA TAC AGG TGA TTT TAG ATT GC-3′). A subsequent PCR using the external primer pair lmo2549_f_del_A and lmo2549_r_del_D and the two flanking fragments as template generated a ∼1000 bp fragment of lmo2549 with an in-frame deletion. The SOE-PCR products also feature restriction sites (underlined in primer sequences) introduced with the external primers. The resulting PCR products were digested with appropriate restriction enzymes, ligated into the corresponding sites in the temperature-sensitive shuttle vector pKSV7 (Smith and Youngman, 1992), and transformed into E. coli, resulting in plasmid pKSV7(Δlmo2549). All plasmids were initially constructed in E. coli XL1-Blue MRF'.
To generate the Δlmo2550 in-frame deletion, plasmid pKSV7(Δlmo2550) was constructed as described above, using primer pair 5′-ATC AGA GCT CAG AGT AAT TAA GTG CTT TTA CAC ACA GTA TAG-3′ and 5′-CCT GCA TAT AAT GAG GTA GAA GAA TAT AAC GGA GAG AAA G-3′ (primers A and B) for the upstream sequence, and primer pair 5′-CTC CGT TAT ATT CTT CTA CCT CAT TAT ATG CAG GAA CAG ATA TTG-3′ and 5′-AAA AGG ATC CGA AAT ATA GAA GAG GGC GGT AG-3′ (primers C and D) for the downstream sequence of ORF lmo2550.
Plasmids pKSV7(Δlmo2549) and pKSV7(Δlmo2550) were then electroporated (Park and Stewart, 1990) into L. monocytogenes EGDe. Allelic replacement was realized by homologous recombination (double crossover) between L. monocytogenes DNA sequences on the plasmid and their corresponding chromosomal alleles. For this purpose, integration mutants were selected for the first homologous recombination by growth for several generations at the non-permissive temperature (42°C) in presence of chloramphenicol and subsequently grown at the permissive temperature (30°C) to select for the second homologous recombination event, resulting in the excision of the sequence and loss of resistance (Smith and Youngman, 1992).
Antibiotic-sensitive clones were checked by PCR and DNA sequencing to verify deletions in EGDeΔlmo2549 and EGDeΔlmo2550.
Sequence analysis of glycosylation genes in L. monocytogenes WSLC 1442
In order to identify putative mutations in WTA glycosylation genes lmo2549 and lmo2550 of the GlcNAc-deficient strain WSLC 1442, the corresponding ORFs were amplified by PCR from isolated chromosomal DNA using the same primer pairs mentioned above for the analysis of strain EGDe. The obtained sequences were aligned to the reference sequences of wild-type strain EGDe and analysed using CLC Main Workbench (Aarhus, Denmark).
Functional complementation of glycosylation-defective mutants
Plasmid pPL2, a site-specific phage integration vector (Lauer et al., 2002), was used for complementation of mutants EGDeΔlmo2549 and EGDeΔlmo2550, and strain WSLC 1442 featuring a 2 bp deletion in lmo2550. The lmo2549 gene was amplified from L. monocytogenes EGDe chromosomal DNA using primer pair 5′-ATC AGG ATC CTT ATT TTG TAG AAG AAT ATA ACG GAG AG-3′ and 5′-ATC AGT CGA CTT ATT TTT TCA CTT TGA AAA TGA TC-3′ (flanking restriction sites are underlined), generating a 500 bp fragment encompassing lmo2550 and 41 bp upstream sequence. To create plasmid pPL2(lmo2550), a 1010 bp fragment containing lmo2550 and 42 bp upstream region was amplified from EGDe DNA, using forward primer 5′-ATA AGG ATC CGT TTT TAG AGT AAT GAG TGG AAA AAA G-3′ and reverse primer 5′-ATC AGT CGA CTC ATA CTA TGT CTT CTT TCT CTC C-3′. The PCR fragments were ligated into the pPL2 BamHI and SalI sites downstream of a Piap promoter, introduced into E. coli strain XL1-Blue MRF' by electroporation, followed by identification of correct transformants. Integration vectors pPL2(lmo2549) and pPL2(lmo2550) were then transformed into the L. monocytogenes mutants EGDeΔlmo2549 and EGDeΔlmo2550. Selection of recombinants was performed on BHI plates supplemented with 10 µg ml−1 chloramphenicol. For confirmation of pPL2 integration at the single copy tRNAArg locus, a colony PCR was performed with primer pair NC16 (5′-GTC AAA ACA TAC GCT CTT ATC-3′) and PL95 (5′-ACA TAA TCA GTC CAA AGT AGA TGC-3′) (Lauer et al., 2002) specifically amplifying a 499 bp product in Listeria strains that contain an integration vector at tRNAArg-attBB' (Lauer et al., 2002). The resulting strains were designated as EGDeΔlmo2549::pPL2(lmo2549) and EGDeΔlmo2550::pPL2(lmo2550). The same procedures were used for transformation of L. monocytogenes WSLC 1442, yielding 1442::pPL2(lmo2549) and 1442::pPL2(lmo2550). The null-mutants and complemented strains were then used in further experiments regarding CBD binding and WTA compositional analysis, as described above.
We are grateful to Serge Chesnov (FGCZ, University of Zurich, Switzerland) for mass spectrometry measurements, and to Regula Bielmann, Yves Briers, Fritz Eichenseher, Lars Fieseler and Mathias Schmelcher (ETH Zurich, Switzerland) for helpful discussions. We thank Christoph Jans (ETH Zurich, Switzerland) and Roger Stephan (University of Zurich, Switzerland) for providing strains. Karin Hotz (ETH Zurich, Switzerland) is acknowledged for critical reading of the manuscript.