Molecular Basis of Resistance to Muramidase and Cationic Antimicrobial Peptide Activity of Lysozyme in Staphylococci

It has been shown recently that modification of peptidoglycan by O-acetylation renders pathogenic staphylococci resistant to the muramidase activity of lysozyme. Here, we show that a Staphylococcus aureus double mutant defective in O-acetyltransferase A (OatA), and the glycopeptide resistance-associated two-component system, GraRS, is much more sensitive to lysozyme than S. aureus with the oatA mutation alone. The graRS single mutant was resistant to the muramidase activity of lysozyme, but was sensitive to cationic antimicrobial peptides (CAMPs) such as the human lysozyme-derived peptide 107R-A-W-V-A-W-R-N-R115 (LP9), polymyxin B, or gallidermin. A comparative transcriptome analysis of wild type and the graRS mutant revealed that GraRS controls 248 genes. It up-regulates global regulators (rot, sarS, or mgrA), various colonization factors, and exotoxin-encoding genes, as well as the ica and dlt operons. A pronounced decrease in the expression of the latter two operons explains why the graRS mutant is also biofilm-negative. The decrease of the dlt transcript in the graRS mutant correlates with a 46.7% decrease in the content of esterified d-alanyl groups in teichoic acids. The oatA/dltA double mutant showed the highest sensitivity to lysozyme; this mutant completely lacks teichoic acid–bound d-alanine esters, which are responsible for the increased susceptibility to CAMPs and peptidoglycan O-acetylation. Our results demonstrate that resistance to lysozyme can be dissected into genes mediating resistance to its muramidase activity (oatA) and genes mediating resistance to CAMPs (graRS and dlt). The two lysozyme activities act synergistically, as the oatA/dltA or oatA/graRS double mutants are much more susceptible to lysozyme than each of the single mutants.


Introduction
In humans, lysozyme is found in a wide variety of fluids, such as tears, breast milk, and respiratory and saliva secretions, as well as in cells of the innate immune system, including neutrophils, monocytes, macrophages, and epithelial cells [1,2]. Lysozyme is an important protein in the innate defense response against invading microorganisms and acts on bacteria by hydrolyzing the ß-1,4 glycosidic bonds between N-acetylmuramic acid (MurNAc) and N-acetylglucosamine (GlucNAc), resulting in degradation of peptidoglycan (PG), and subsequent cell lysis [3,4]. Most bacterial species are sensitive to lysozyme, but some important human pathogens, such as Staphylococcus aureus, Neisseria gonorrhoeae, and Proteus mirabilis, are resistant. The mechanisms behind the high resistance of S. aureus to lysozyme are unknown, although several studies suggest that O-acetylation at position C-6 of the MurNAc residue contributes to lysozyme resistance [5][6][7][8][9]. Recently, we were able to prove that indeed O-acetyltransferase A (OatA) of S. aureus is responsible for O-acetylation of the PG, and this leads to resistance to the muramidase activity of lysozyme [10]. We also showed that the MurNAc was Oacetylated only in pathogenic, lysozyme-resistant staphylococci (e.g., S. aureus, S. epidermidis, S. lugdunensis, and others). All nonpathogenic species (e.g., S. carnosus, S. gallinarum, or S. xylosus) were lysozyme sensitive and lacked PG-specific Oacetylation. Therefore, OatA can be regarded as a general virulence factor [11].
Although the oatA mutant was less resistant to lysozyme than the wild type (WT) S. aureus, it still was more resistant than, for example, Micrococcus luteus, suggesting that other factors, such as a high degree of peptide cross-linking, may also contribute to lysozyme resistance [12]. Recently, we showed that the presence of wall teichoic acid (WTA) increased lysozyme resistance [13]. One also has to consider that lysozyme does not only comprise muramidase activity but also antimicrobial peptide activity, as demonstrated by catalytically inactivate lysozyme or peptides isolated from digested lysozyme, and by synthetic lysozyme-derived peptides [14][15][16][17].
Here, we show that the extremely high resistance of S. aureus to lysozyme can be genetically dissected as a) resistance to muramidase activity and b) resistance to inherent cationic antimicrobial peptide (CAMP) activity. Furthermore, we characterized via transcriptome analysis the two-component system (TCS), GraRS, which, in addition to many virulence genes, also controls the dlt operon to mediate resistance to lysozyme and other CAMPs.

Susceptibility of oatA and graRS Single and Double Mutants to Lysozyme and CAMPs
In our search for highly susceptible lysozyme mutants in S. aureus, we isolated two Tn917 transposon mutants in SA113oatA::kan that revealed higher sensitivity to lysozyme than the oatA mutation alone. Chromosomal sequencing of the flanking Tn917 insertion sites revealed that Tn917 was inserted in SA0615 [18]. SA0615 and the upstream gene SA0614 have the features of a typical TCS and were recently named GraRS (glycopeptide resistance-associated), because overexpression of GraR (response regulator) and GraS (sensor histidine kinase) increased vancomycin resistance [19]. To further study the role of TCS in lysozyme resistance, we constructed a deletion mutant by substituting graRS with an erythromycin B cassette to yield SA113 graRS::erm ( Figure  1). In addition, we also constructed an oatA::kan/graRS::erm double knockout. Sequencing and complementation with pTXgraRS, a vector in which the graRS genes are induced into expression by xylose, confirmed the correct replacement. Whereas the oatA/graRS double mutant was highly susceptible to lysozyme, both single mutants were only marginally affected, but were still more sensitive than the WT, which is completely lysozyme resistant (Figure 2A-2D).
The oatA/graRS double mutant was much more lysozyme sensitive than each of the single mutants. This hypersensitivity of the double mutant can be explained by dual activities of lysozyme that act in a synergistic way. To study this phenotype in more detail, we investigated whether the graRS single mutant is affected by the muramidase activity of lysozyme. Indeed, the isolated PG from the graRS single mutant was completely resistant to lysozyme hydrolysis, in contrast to the oatA mutant. As expected, PG of the oatA/ graRS double mutant was also hydrolysed, although the sensitivity was less pronounced, as in the oatA single mutant ( Figure 3). Therefore, the increased sensitivity of the double mutant likely came from its higher susceptibility to lysozyme's CAMP activity. This was confirmed by the addition of LP9, polymyxin B, or gallidermin to a growing culture, which caused immediate growth arrest in the graRS mutant, whereas the WT was much less affected ( Figure 4A and 4B), and only the lantibiotic gallidermin inhibited the WT. In addition, we demonstrated that heat-inactivated lysozyme exhibits CAMP activity, but no muramidase activity. Heat-inactivated lyso- (B) In the graRS deletion mutant, graRS is substituted by the erythromycin B resistance cassette. Note that ermB gene has a weak transcription terminator, and transcriptional read-through to the following vraFG genes is likely. graR, response regulator; graS, sensor histidine kinase; vraF, ABC transporter ATP-binding protein; vraG, ABC transporter permease; SA0612 and SA0613 are hypothetical proteins (HP

Author Summary
In humans, lysozyme plays an important role in the suppression of bacterial infections. However, some bacterial pathogens, such as Staphylococcus aureus, are completely resistant to lysozyme. Here we demonstrate that lysozyme acts on S. aureus in two ways: as a muramidase (cell wall lytic enzyme) and as a cationic antimicrobial peptide (CAMP). S. aureus has developed resistance mechanisms against both activities by modifying distinct cell wall structures. Modification of the peptidoglycan by O-acetylation (OatA) renders the cells resistant to the muramidase activity. Modification of teichoic acids by D-alanine esterification (Dlt) renders the cells resistant to lysozyme's CAMPs and other CAMPs. Transcriptome analysis of the glycopeptide resistance-associated (GraRS) twocomponent system revealed that this global regulator controls 248 genes such as other global regulators, colonization factors, or exotoxin-encoding genes. Since GraRS also upregulates the dlt operon, it was not surprising that in the graRS mutant teichoic acid D-alanylation is markedly decreased, which explains its increased sensitivity to CAMPs. By comparative analysis of mutants we were able to dissect genes that were responsive to the dual activities of lysozyme. Here we show how efficiently S. aureus is protected from the human defense system, which enables this pathogen to cause persistent infections.
zyme showed no activity (neither lytic nor CAMP activity) to the oatA mutant or to the isolated PG of oatA, but it was able to inhibit the growth of the oatA/graRS double mutant ( Figures 2B, 2D, and 3). This result suggests that GraRS controls genes involved in CAMP resistance. This effect was not only achieved with hen egg-white, but also with human lysozyme.

Comparative Transcriptome Analysis of WT and graRS Mutant
To find out which genes are responsible for the high susceptibility to CAMPs in the graRS mutant, we carried out a comparative transcriptome analysis of the WT strain and the graRS mutant. We detected 115 genes whose mRNAs were upregulated (Table 1) and 133 genes whose mRNAs were downregulated by GraRS ( Table 2). The complete list of up-and down-regulated genes with their National Center for Biotechnology Information PID numbers is presented in Dataset S1. In order to give an impression of which genes are controlled by GraRS, some examples are mentioned below.
In the graRS mutant, genes that are involved in RNA and amino acid synthesis and glycolysis shows highly gene transcription rates. In particular, the urease genes (ureA-G) all 12 pur genes were 2-to 32-fold up-regulated as compared to the WT, whereas purR (repressor) appeared not to be influenced by GraRS. Interestingly, the amount of oatA transcript increased in the graRS mutant, which could explain the slightly higher resistance of the graRS mutant to the muramidase activity of lysozyme ( Figure 3). A number of genes that were down-regulated included global regulators (rot, sarS, mgrA), cell surface protein encoding genes (the Ser-Asp rich fibrinogen-binding proteins SdrC and SdrE), the major autolysin gene (atlA) and an autolysin/adhesin gene (aaa) [20], exoprotein encoding genes (hlb, hlgA,B, lukM,F, and geh), transporter encoding genes (essA/essC, oppB, and norB), capsule encoding genes (capA,H,I,J,K) and PIA encoding genes (icaADBC), genes responsible for D-alanyl esterification of teichoic acids (TAs) (dltA,B,D), and the alanine dehydrogenase gene (ald1). The pronounced decrease of expression of the ica [21][22][23] and dlt operons [24] and atlA [25] explains why the graRS mutant showed a biofilm-negative phenotype on microtiter plates (unpublished data). With a few genes, such as rot, ureC, and dltA, we verified the transcriptome data by reverse transcriptase (RT)-PCR (Table 3).
Next, we asked which of the 115 less expressed genes in the graRS mutant were responsible for the increased susceptibility to CAMPs. A most likely candidate was the dlt operon (encoding enzymes involved in D-alanylation of TAs). Its transcript was decreased 2.1-fold to 2.9-fold as compared to WT, and indeed, the D-alanylation of TAs was decreased 46.7% in the graRS mutant compared to WT (Table 3). It has been previously shown that inactivation of the dlt operon in S. aureus confers sensitivity to defensins, protegrins, and other antimicrobial peptides [26]. The observed decrease of ald1 transcription by a factor of 3.5 is in line with the decreased dlt transcription. Ald1 is the alanine dehydrogenase, which is involved in the synthesis of L-alanine.

Comparison of graRS and dltA Mutants
Because the dlt operon is less expressed in the graRS mutant, we investigated lysozyme susceptibility with a dltA deletion mutant, which is well-known to be sensitive to  CAMPs [26]. Indeed, the dltA mutant was more sensitive to lysozyme ( Figure 2E); however, this sensitivity was not due to its muramidase activity, as the isolated PG of the dltA mutant was not hydrolyzed by lysozyme ( Figure 3). Furthermore, growth of the dltA mutant was inhibited whether active or heat-inactivated lysozyme was applied ( Figure 2E). When the susceptibility of graRS and dltA mutants to LP9, polymyxin B, and gallidermin were compared, both mutants were similarly more susceptible to these CAMPs ( Figure 4B and 4C). However, there were two distinctions: a) the susceptibility of the dltA mutant was more pronounced than that of the graRS mutant, and b) even in the presence of gallidermin or polymyxin B, the graRS mutant started to grow after some time and reached the same optical density (OD) values after 24 h as the control culture lacking CAMPs. In contrast, the dltA mutant remained sensitive to gallidermin and polymyxin B and was unable to resume growth. In the presence of LP9, growth resumed after a similar lag period as in the graRS mutant; this can possibly be explained by its proteolytic degradation. Not only the single but also the double mutants oatA/graRS and oatA/dltA were sensitive to the CAMP activity of LP9, although the susceptibility was less pronounced as with the graRS and dltA single mutants. However, the oatA single mutant was completely resistant to LP9, indicating that oatA is resistant to CAMPs ( Figure 4D-4F). With respect to gallidermin-and polymyxin B-induced cell lysis, it has been observed that CAMPs such as lantibiotics induce autolysis in staphylococci by increasing PG hydrolase activity [27]. We assume that gallidermin and polymyxin B, which are also CAMPs, very likely have a similar effect.
We asked whether the increasing insensitivity of the graRS mutant after prolonged growth is some short lasting CAMPinduced adaptation or whether it is based on selection of resistant mutants. To answer this question, we inoculated from a 24-h graRS culture treated with polymyxin B ( Figure  5B) a new culture and challenged it again with polymyxin B ( Figure 5C). The subculture revealed no growth retardation, which suggests that the graRS phenotype is unstable and that polymyxin B-resistant revertants were quickly selected. Since the dltA revealed a stable phenotype, we assume that in the selected revertants dltA expression was increased to WT levels.
Hypersensitivity of the oatA/dltA and oatA/graRS Double Mutants to Lysozyme The highest susceptibility to lysozyme was observed with the oatA/dltA double mutant, which was more than 66-fold and 333-fold more sensitive to lysozyme than the dltA and oatA single mutants, respectively ( Figure 2B, 2E, and 2F; Table  4). The oatA/graRS mutant is not quite as sensitive as the oatA/ dltA mutant. Another difference is that the oatA/dltA mutant stays lysozyme sensitive even after 24 h of cultivation ( Figure  2D and 2F), indicating that the dltA mutant phenotype cannot easily revert to the WT phenotype. The lower susceptibility of the oatA/graRS double mutant can possibly be explained by the fact that the TA in this mutant still contains 53.3% Dalanyl esters, whereas the dltA mutant completely lacks Dalanylation in its TAs ( Table 3).
The high susceptibility of the double mutants is based on the dual activities of lysozyme: a) the oatA mutant is sensitive to the muramidase activity of lysozyme but is insensitive to CAMPs ( Figures 2B, 3, and 4F), and b) the dltA and graRS mutants are sensitive to CAMPs, but insensitive to the muramidase activity of lysozyme (Figures 3, 4B, and 4C). The extremely high lysozyme susceptibility of the oatA/dltA double mutant can only be explained by a synergistic effect of the two activities.

Increased Lytic Activity of Mutanolysin by Lysozyme and LP9 in the graRS Mutant
Mutanolysin is a muramidase that is able to hydrolyze Oacetylated PG [28] but does not normally cause cell lysis in WT S. aureus or its graRS mutant at a concentration of 100 lg/ ml. However, when the graRS mutant was treated with mutanolysin in combination with lysozyme or LP9, the lytic activity (indicated by decrease in OD) was strongly increased ( Figure 5A). Because the O-acetylated graRS mutant is insensitive to the catalytic activity of lysozyme, we assume that mutanolysin acts through its lytic activity, and LP9 and lysozyme through their CAMP properties. We have not investigated how the stimulating effect of lysozyme and LP9 on cell lysis is accomplished. However, we assume that it is caused by the concerted action of PG hydrolysis by mutanolysin and induced autolysis by lysozyme and LP9, as mentioned above.

Minimal Inhibition Concentration Values of SA113 and Various Mutants
The minimal inhibition concentration (MIC) values for lysozyme, polymyxin B, and gallidermin in WT and various mutants are summarized in Table 4. Both the WT and the   graRS mutant were completely resistant to lysozyme at a concentration of 50 mg/ml. However, the graRS mutant was 17-and 4-fold more susceptible to polymyxin B or gallidermin. The sensitivity to the CAMPs is very likely due to the aforementioned decrease in expression of the dlt operon, which corresponds with decreased D-alanylation of the TAs.
The oatA mutant was more susceptible to lysozyme than the graRS mutant, but, similar to WT, was completely insensitive to heat-inactivated lysozyme or CAMPs, indicating that oatA is only sensitive to the muramidase activity of lysozyme. The oatA/graRS double mutant was almost 17-fold more sensitive to lysozyme than the oatA mutant, which can be explained by the fact that this double mutant is sensitive to both the muramidase and the CAMP activities of lysozyme. The two activities exert a synergistic effect on the double mutant. The dltA single mutant was over 25-fold more sensitive to lysozyme than the WT and 5-fold more sensitive than the oatA single mutant, demonstrating the importance of lysozyme's CAMP activity. Furthermore, the dltA mutant exhibited the highest susceptibility to polymyxin B and gallidermin, but was completely insensitive to lysozyme's muramidase activity (Figure 3). With a MIC of only 30 lg/ml, the oatA/dltA double mutant revealed the highest susceptibility to lysozyme. Indeed, it has a 20-fold greater sensitivity to lysozyme than the oatA/graRS double mutant. The oatA/dltA double mutant is 333-fold and 66-fold more sensitive than the single oatA or dltA mutants, which illustrates the extremely high synergistic effect of lysozyme when it can exert both muramidase and CAMP activities. Overexpression of graRS in the graRS mutant or the WT by pTXgraRS resulted in an approximately 2-fold increase in polymyxin B resistance, indicating that even in WT cells, CAMP resistance can be further increased.

Discussion
One of our research aims was to identify genes involved in staphylococcal lysozyme resistance. We have already elucidated two genes and corresponding enzymes that contribute to resistance against the muramidase activity of lysozyme. Since the target of muramidase is PG, it is not surprising that the mechanism of resistance is masking PG by modification. In S. aureus there are two PG modifications that are involved in resistance to lysozyme's muramidase activity. One modification is O-acetylation catalyzed by the PG-specific Oacetyltransferase A, OatA, and we have shown that the oatA mutant is more susceptible to the muramidase activity of lysozyme than the WT [10]. The other modification is WTA [29] that is covalently linked to the same C 6 position in MurNAc as in the O-acetyl group. TagO is a specific UDP-Nacetylglucoseamine transferase, which is involved in the first step of WTA synthesis. The tagO deletion mutant completely lacks WTA [30]. Although the tagO mutant still shows high lysozyme resistance, a oatA/tagO double mutant, however, is much more susceptible to lysozyme's muramidase activity than the oatA mutation alone [13]. Here, we show that the high lysozyme resistance of S. aureus is not only based on resistance to the muramidase activity of lysozyme, but also to its inherent CAMP resistance.
The described global two-component regulator, GraRS, was identified in an oatA-minus background by increased lysozyme susceptibility in an oatA/graRS double mutant. The graRS mutant was more susceptible to CAMPs than the WT. We assume that the reason for the increased susceptibility of the graRS mutant was a decrease in dlt expression, and consequently, GraRS up-regulates dlt expression. The Dlt enzymes modify TAs by the incorporation of D-alanine esters rendering the cells resistant to CAMPs, very likely by repulsion [26]. We showed that the dltA mutant is even more susceptible to lysozyme-derived LP9 and other CAMPs than the graRS mutant, because in the dltA mutant, D-alanine esters were completely absent in TAs, the mutant was stable, and no revertants were observed. Heat-inactivated lysozyme does not affect either the growth of the oatA or that of the graRS mutant. The latter effect is surprising, as the graRS mutant is sensitive to the other CAMPs (LP9, gallidermin, polymyxin B). However, the oatA/graRS mutant was sensitive to heatinactivated lysozyme, suggesting that the bulky molecule has better access to the cell envelope when the PG is de-Oacetylated. Likewise, sensitivity of the dltA mutant to heatinactivated lysozyme can also be explained by better access to the cell envelope because of the lack of D-alanine esters in TAs.
The next interesting question was, how do CAMPs act in the dltA, oatA/graRS, or oatA/dltA mutants? Killing of Gramnegative bacteria could be demonstrated by lysozyme-derived peptides that were transported through the outer membrane and damaged the inner membrane by pore formation [17]. Several authors assume that lysozyme and CAMPs are not only acting as membrane permeabilization agents, but also activate autolytic wall enzymes of Gram-positive bacteria, thus causing cell lysis [31][32][33]. It has also been shown that lipoteichoic acids can bind and inhibit autolysins, depending on their degree of D-alanylation [34][35][36]. Similar results were also obtained in a dlt mutant of Lactococcus lactis, which showed increased autolysis [37]. In line with these observations, the   graRS and dltA mutants also showed increased autolysis when treated with Triton X-100 (unpublished data), suggesting that in these mutants, too, CAMPs activate autolytic enzymes. We assume that the observed synergistic effect of lysozyme in the oatA/graRS and oatA/dltA double mutants is caused by the simultaneous activation of autolytic enzymes and the muramidase activity of lysozyme. A similar synergistic effect is seen by treatment with mutanolysin in combination with LP9 (inducing autolysis) or lysozyme (cannot exert its muramidase activity as the PG is O-acetylated) as shown in the graRS single mutant ( Figure 5A). For the first time (to our knowledge), we have traced and dissected genes that were responsive to the dual activities of lysozyme.
Until now, little was known about the two-component system GraRS. We became interested in the regulation of GraRS because we wanted to trace the gene(s) that caused the increased CAMP susceptibility in the graRS mutant. Comparative transcriptome analysis of SA113, an 8325-derivative, and its graRS mutant revealed that 115 genes were upregulated and 133 genes were down-regulated by GraRS (Tables 1 and 2). Among the down-regulated genes was the vraFG operon, which immediately follows the graRS operon. However, in studying intermediate level of vancomycin resistance in S. aureus, Ambrose Cheung and colleagues found that vraFG is positively controlled by GraRS [38]. This contradictory result can be explained by the genetic organization of our graRS::ermB deletion mutant (Figure 1). In our mutant, the ermB cassette is in the same orientation as the vraFG genes. Since the ermB transcription terminator is very weak, we assume that there is a transcriptional readthrough into the vraFG genes. This explains why in our graRS deletion mutant, the vraFG genes were up-regulated instead of down-regulated.
GraRS up-regulates transcription of global regulators such as the SarA homologs Rot, SarS, and MgrA. We compared our GraRS transcriptome results with that of the recently published transcriptome studies of Rot [39], MgrA [40], and ArlRS [41] (Tables 1 and 2; Figure 6). Rot is a repressor of exoproteins but positively regulates cell surface proteins, and SarS is a positive activator of protein A. MgrA appears to be  Table 3.   SA1793, which encodes a hypothetical protein with a phagerelated function. Many of the down-regulated genes are involved in RNA and amino acid synthesis or glycolysis. lrgA, which encodes a holin-like protein with murein hydrolase activity, is also down-regulated by GraRS but up-regulated by ArlRS and MgrA. Most of the genes are exclusively regulated by GraRS, such as ica, pur, mgrA, sirA,C, atlA, aaa, dnaJ,K, grpE, and vraF,G. These results illustrate that there is a distinct cross-regulation between GraRS, ArlRS, Rot, MgrA, and probably some other global regulators. GraRS is not only important for resistance to glycopeptides, lysozyme, and other CAMPs. Our data suggest that GraRS also has an intermediate role between other global regulators (Agr, MgrA, Rot, and SarA), as GraRS up-regulates both adhesins as well as exoproteins and toxins (e.g., hlb, hlgA,B, lukM,F, geh). GraRS is possibly involved in the establishment of persistent infections by the up-regulation of colonization factors (e.g., ica, atl, aaa, fib, sirA, sirC, sdrC, sdrE), factors involved in resistance to CAMPs (dlt), factors involved in intermediary vancomycin resistance (vraF,G, as mentioned above), and factors involved in biofilm formation (e.g., dlt, atl, ica). It would be interesting to study the graRS mutant in an animal model for chronic infection.
Transposon mutagenesis. Transposon mutagenesis was carried out in the DoatA::kan mutant using the temperature-sensitive plasmid pTV1ts and was performed as described by Bera et al. [10].
Effects of cationic agents on exponential growth. Overnight cultures were diluted to an OD 578nm of 0.1 in 50 ml of BM and the cultures were incubated with shaking at 37 8C. OD was determined every hour. Ten milliliters of each culture were transferred into a new 100-ml flask when the cultures reached an OD 578nm of nearly 1.0. Then, cationic agents, such as hen egg-white lysozyme and human lysozyme (Sigma-Aldrich, http://www.sigmaaldrich.com/), LP9 (a lysozyme-derived 9-aa peptide, 107 R-A-W-V-A-W-R-N-R 115 -NH 2 ) (EMC, http://www.microcollections.de/), polymyxin B (Sigma-Aldrich), gallidermin (Genmedics, http://www.genmedics.com/), or mutanolysin (Sigma), were added. Lysozyme was inactivated by heating for 1 h at 100 8C and placed on ice. The OD 578nm of all cultures was measured hourly up to 7-8 h and after 24 h.
MIC assay. The overnight cultures were diluted in BM with 0.5% xylose to a concentration of 0.5 3 10 6 CFU per ml and aliquoted in 0.5-ml samples, and cationic agents in different concentrations were added. The cultures were incubated with shaking at 37 8C for 20-24 h and MIC was determined.
Biofilm assay. An overnight culture was diluted 1:200 in fresh TSB with 0.5% glucose, and 200 ll were filled into microtiter plates and incubated for 20-24 h at 37 8C without shaking. The supernatant was removed and the plate was washed two times with PBS (pH 7.4). The plate was dried and the cells were colored with 0.1% safranine.
Isolation of PG. One liter of BM was inoculated with an overnight culture of the WT SA113 or the mutants. Strains were grown for 12 h with shaking at 37 8C. Cells were centrifuged, washed two times with cold 0.9% NaCl, diluted in 0.9% NaCl, and boiled for 20 min. After the cells were chilled on ice, they were again centrifuged and washed twice with 0.9% NaCl. The cells were disrupted in a mechanical grinding device using glass beads 150-212 lm (Sigma-Aldrich) at 4 8C, centrifuged and washed two times with cold H 2 O bidest , boiled for 30 min in 2% SDS to remove noncovalently bound proteins, and washed four times with H 2 O bidest . The cell wall fragments were diluted in 0.1 M Tris/HCl (pH 6.8) and incubated with 0.5 mg/ml trypsin for 16 h at 37 8C to degrade cell-bound proteins. After centrifugation and washing with water, the PG was lyophilized.
Turbidometric assay of PG. For analyzing the susceptibility of PG to lysozyme, we used a modified method turbidometric assay as described by Clarke [42]. The PG of the WT SA113 and the mutants were sonicated and diluted to 0.5 mg in 1 ml of 80 mM PBS (pH 6.4). After the addition of 300 lg lysozyme per ml, the decrease in optical density was monitored at the beginning (0 h) and after 4 h at OD 660nm and calculated as percentages. Quantification of D-alanylation of TA by HPLC. S. aureus strains were grown in BM with 0.25% glucose overnight, centrifuged, washed three times, and resuspended in ammonium acetate buffer (20 mM [pH 6.0]). The OD 600nm was adjusted to 30. Aliquots (1 ml) were heatinactivated by incubation at 99 8C for 10 min and centrifuged, and pellets were dried. After incubation at 37 8C for 1 h with 100 ll of 0.1 N NaOH, 100 ll of 0.1 N HCl were added for neutralization and samples were dried. For derivatization, 100 ll of triethylamine and 100 ll of Marfey's reagent (1-fluoro-2,4-dinitrophenyl-5-L-alanine amide; Sigma) (10 mM) were added. After incubation at 40 8C for 1 h, samples were dried and resuspended in DMSO:H 2 O (1:1). Quantification of D-alanine was performed by HPLC as previously described [43].
RNA isolation and real-time RT-PCR. SA113 and the graRS deletion mutant were cultivated in 50 ml of BM and harvested at mid-exponential growth phase. Before RNA isolation, two volumes of RNAprotect bacteria reagent (Qiagen, http://www.qiagen.com/) were added to 10 ml of culture and centrifuged. The cells were lysed by the addition of 50 lg/ml of lysostaphin (0.5 mg/ml) (Genmedics) in TE buffer and total RNA was isolated using the RNeasy Mini Kit (Qiagen). Contaminating DNA was degraded with the DNase Kit (Ambion, http://www.ambion.com/) according to the manufacturer's instructions. LightCycler RT-PCR was carried out using the Light-Cycler RNA amplification Kit SYBR Green I or with the LightCycler RNA amplification kit for hybridization probes (Roche Biochemicals, http://www.roche.com/). The internal control gyr was quantified using 10-fold serial dilutions (10 4 to 10 8 copies/ll) of a specific RNA standard using oligonucleotides specific for gyr (gyr297F: TTAGTGTGGGAAATTGTCGATAAT and gyr574R: AGTCTTGTGA-CAATGCGTT TACA), dltA (dltA1: TGGCGTTGAAAGACTAGGC and dltA2: TTACGAACTCAGACTGGCG), rot (rot1: TTCAGCGAGATT-GAAAGCG and rot2: GTTGCTCTACTTGCAATGG) or ureC (ureC1: G A T A T C A T T G C C G C T G A A G G a n d u r e C 2 : A A A G C A -GATGGTGTTGCACC) as described [44]. Standard curves for dltA and rot were generated using 5-fold serial dilutions of WT SA113 RNA or for ureC of the graRS mutant RNA. Differences between WT and the graRS mutant were determined by n-fold change and calculated as a percentage of the mRNA product. The specificity of the PCR was verified by size determination of the amplicons by agarose gel electrophoresis. To check for DNA contamination, each sample was subjected to PCR by using the LightCycler DNA amplification kit SYBR Green I (Roche Biochemicals). In none of the cases an amplification product was detectable.
Transcriptome analysis. Transcriptome analysis was carried out as described by the microarray manufacturer Scienion (http://www. scienion.de/) and Resch et al. [45]. cDNA was synthesized from isolated RNA (1 lg) during mid-exponential growth (4 h) derived from WT SA113 (labeled in green with Cy3 [532 nm]) or from the graRS mutant (labeled in red with Cy5 [635 nm]). cDNAs from WT and the graRS mutant were pooled and hybridized on four DNA microarrays. Scienion performed DNA transcriptome analysis by comparing the intensity of each Cy3-labeled gene of the WT with the intensity of each Cy5-labeled gene of the graRS mutant as a ratio of the medians (532/635). The threshold was set at a 2-fold difference in gene expression. Genes whose RNA level was higher in WT (2.0 and more) were categorized as being positively regulated by GraRS. In contrast, genes that had higher RNA levels (2.0 and more) in the graRS mutant were described as being negatively regulated by GraRS. The significance of differences (n-fold) in gene expression was calculated by One-Sample t-Test-Benjamini-Hochberg (Adv); results ,0.051 are significant, and some genes from Tables 1 and 2 are higher than 0.05.