The Peptidoglycan Pattern of Staphylococcus carnosus TM300—Detailed Analysis and Variations Due to Genetic and Metabolic Influences

The Gram-positive bacterium Staphylococcus carnosus (S. carnosus) TM300 is an apathogenic staphylococcal species commonly used in meat starter cultures. As with all Gram-positive bacteria, its cytoplasmic membrane is surrounded by a thick peptidoglycan (PGN) or murein sacculus consisting of several layers of glycan strands cross-linked by peptides. In contrast to pathogenic staphylococci, mainly Staphylococcus aureus (S. aureus), the chemical composition of S. carnosus PGN is not well studied so far. UPLC/MS analysis of enzymatically digested S. carnosus TM300 PGN revealed substantial differences in its composition compared to the known pattern of S. aureus. While in S. aureus the uncross-linked stem peptide consists of a pentapeptide, in S. carnosus, this part of the PGN is shortened to tripeptides. Furthermore, we found the PGN composition to vary when cells were incubated under certain conditions. The collective overproduction of HlyD, FtsE and FtsX—a putative protein complex interacting with penicillin-binding protein 2 (PBP2)—caused the reappearance of classical penta stem peptides. In addition, under high sugar conditions, tetra stem peptides occur due to overflow metabolism. This indicates that S. carnosus TM300 cells adapt to various conditions by modification of their PGN.


Introduction
Staphylococcus carnosus (S. carnosus, S. c.) TM300 is an apathogenic, coagulase-negative, "food grade" staphylococcal species [1] commonly used as meat starter culture for raw sausages [2]. Its 2.56 Mbp genome has the highest GC content of all staphylococcal species sequenced so far. While virulence and toxicity factors are almost completely missing from the genome, most of the metabolic pathways are present. As typical for starter cultures, it harbors various sugar degradation pathways [3,4]. The general features of the peptidoglycan (PGN) of S. carnosus TM300 are already known. Like Staphylococcus aureus (S. aureus, S. a.) it belongs to the A3α-type with a penta glycine interpeptide bridge [1,5]. N-acetylmuramic acid (MurNAc) residues in the glycan backbone are similarity. The role of FtsE and FtsX during synthesis of the PGN sacculus has not been investigated in staphylococci so far.
The PGN sacculus can be isolated and digested by a muraminidase such as mutanolysin into fragments called muropeptides. Muropeptides are disaccharide units with peptide moieties that can be cross-linked to other stem peptides and disaccharides thereby resulting in oligomeric structures [36]. During the analyses of S. carnosus TM300 muropeptides, we observed degradation of the first acceptor stem peptides from five (pentapeptide) to three (tripeptide) amino acids, which is in contrast to the known muropeptide pattern of S. aureus [37,38]. This gave us reason to perform a detailed Ultra Performance Liquid Chromatography (UPLC) and UPLC mass spectrometry (UPLC/MS) analysis of the S. carnosus TM300 muropeptide pattern. Our results show that almost all stem peptides that were not part of a cross-link contain only three amino acids (L-Ala-D-iGln-L-Lys) under standard conditions. However, overexpression of HlyD-FtsE-FtsX, a putative protein complex interacting with PBP2, partly inhibited this degradation process. A similar effect was observed for high glucose and fructose concentrations.

Peptidoglycan Analysis
We analyzed the PGN of S. carnosus TM300 by UPLC ( Figure 1) and observed that it differed substantially from the known pattern of S. aureus [37,38]. The whole muropeptide pattern of S. carnosus was shifted to shorter retention times compared to S. aureus. We performed UPLC-MS analysis to determine the chemical composition of the muropeptide peaks (Table 1). All main peaks in the monomeric as well as in the cross-linked fractions (dimer to pentamer fraction) contained muropeptides with stem peptides only consisting of three amino acids: L-Ala-D-iGln-L-Lys, explaining the shift in retention time. In S. aureus, the non-cross-linked stem peptides of monomeric muropeptides as well as the acceptor stem peptide of multimeric muropeptides harbor the complete penta stem peptide. These penta muropeptides are also present in S. carnosus, but in small amounts compared to the respective tri muropeptide. The cross-linked stem peptides are still tetra peptides and do not get degraded into tri peptides, leading to Tri-Tetra n muropeptides (Table 1). During the course of our experiments, we realized that the amount of different muropeptides varied and was influenced by genetic factors as well as composition of the medium. Table 1. Muropeptides of S. carnosus TM300 analyzed by UPLC-MS. Each muropeptide contains one to five stem-peptides and each consists of the three to five amino acids L-Ala-D-iGln-L-Lys(-D-Ala)(-D-Ala). The interpeptide bridges are built of Gly molecules. The sum of all Gly residues present in each muropeptide is given. Muropeptide peaks 4, 7 and 10 (highlighted in bold) are strongly elevated by hlyD-ftsE-ftsX overexpression. Muropeptide peaks 3, 6 and 9 (highlighted in italics) are strongly elevated by high glucose and fructose concentrations.   Table 1 including molecule composition and proposed molecular formula. Retention time of peaks in the UPLC (quaternary pumping system) was about 1 min longer than in the binary UPLC/MS system. Therefore, in Table 1 only TIC retention times are given (TIC: Total Ion current, recorded in positive mode).   Table 1 including molecule composition and proposed molecular formula. Retention time of peaks in the UPLC (quaternary pumping system) was about 1 min longer than in the binary UPLC/MS system. Therefore, in Table 1 only TIC retention times are given (TIC: Total Ion current, recorded in positive mode).

Search for New Proteins Involved in PGN Biosynthesis
We performed a Bacterial-Two-Hybrid (BTH) screen using a genomic library of S. carnosus TM300 to search for new interaction partners of PBP2. This bifunctional transglycosylase/transpeptidase is considered to be one of the main players of PGN biosynthesis in S. aureus [16][17][18][19]39,40]. Among others, we found the so far uncharacterized protein SCA_1997, a conserved hypothetical protein that belongs to the HlyD family of secretion proteins (K02005). Therefore, we refer to this gene as hlyD. Bioinformatical analyses using the Kyoto Encyclopedia of Genes and Genomes (KEGG) [24] showed that if hlyD is present in Bacillus spec. and Staphylococcus spec. it is always located in an operon with two other genes. Each of them possesses its own start codon but they share one Shine-Dalgarno sequence and one promotor region. The first gene (SCA_1996) encodes for a putative ABC transport system ATP-binding protein (K02003) with structural similarity to FtsE of E. coli. The second gene (SCA_1995) is predicted to produce an ABC transport system permease protein (K02004) with structural similarity to E. coli FtsX [30,41]. So far, there is no function described for the staphylococcal HlyD, FtsE and FtsX proteins.
We cloned the genes of each of these proteins into the two vectors of the BTH system and performed interaction studies with all possible pairs ( Table 2). PBP2 interacted with itself as well as with HlyD and FtsX. HlyD also interacted with FtsX. In addition, FtsX interacted with itself and with FtsE. There was no self-interaction of FtsE detected, and FtsE did not directly interact with HlyD ( Figure 2). The interactions of the full length proteins confirmed our genetic screen results, and suggest that HlyD, FtsX and FtsE are involved in the same or in closely related biosynthetic pathways, most likely in cell wall biosynthesis and/or cell division.

Search for New Proteins Involved in PGN Biosynthesis
We performed a Bacterial-Two-Hybrid (BTH) screen using a genomic library of S. carnosus TM300 to search for new interaction partners of PBP2. This bifunctional transglycosylase/transpeptidase is considered to be one of the main players of PGN biosynthesis in S. aureus [16][17][18][19]39,40]. Among others, we found the so far uncharacterized protein SCA_1997, a conserved hypothetical protein that belongs to the HlyD family of secretion proteins (K02005). Therefore, we refer to this gene as hlyD. Bioinformatical analyses using the Kyoto Encyclopedia of Genes and Genomes (KEGG) [24] showed that if hlyD is present in Bacillus spec. and Staphylococcus spec. it is always located in an operon with two other genes. Each of them possesses its own start codon but they share one Shine-Dalgarno sequence and one promotor region. The first gene (SCA_1996) encodes for a putative ABC transport system ATP-binding protein (K02003) with structural similarity to FtsE of E. coli. The second gene (SCA_1995) is predicted to produce an ABC transport system permease protein (K02004) with structural similarity to E. coli FtsX [30,41]. So far, there is no function described for the staphylococcal HlyD, FtsE and FtsX proteins.
We cloned the genes of each of these proteins into the two vectors of the BTH system and performed interaction studies with all possible pairs ( Table 2). PBP2 interacted with itself as well as with HlyD and FtsX. HlyD also interacted with FtsX. In addition, FtsX interacted with itself and with FtsE. There was no self-interaction of FtsE detected, and FtsE did not directly interact with HlyD ( Figure 2). The interactions of the full length proteins confirmed our genetic screen results, and suggest that HlyD, FtsX and FtsE are involved in the same or in closely related biosynthetic pathways, most likely in cell wall biosynthesis and/or cell division. Therefore, we deleted the whole operon and examined the resulting mutant strain S. carnosus TM300 ΔhlyD-ftsEX. We did not observe any difference in growth behavior under standard BM conditions (Supplementary Materials Figure S2A), in minimal M9 medium (Supplementary Materials Figure S2B), or when the amount of NaCl in the medium was varied ( Figure S2C). In addition, deletion of this operon did not have any effect on the muropeptide composition ( Figure S3A). Therefore, we deleted the whole operon and examined the resulting mutant strain S. carnosus TM300 ∆hlyD-ftsEX. We did not observe any difference in growth behavior under standard BM conditions (Supplementary Materials Figure S2A), in minimal M9 medium (Supplementary Materials Figure S2B), or when the amount of NaCl in the medium was varied ( Figure S2C). In addition, deletion of this operon did not have any effect on the muropeptide composition ( Figure S3A). However, when the three genes were overexpressed from a plasmid harboring a xylose inducible promoter, all peaks with muropeptides containing a penta stem peptide were highly elevated while the tri muropeptides were still present in comparable amounts. Muropeptides with tetra stem peptides were slightly elevated as well. As 25 mM xylose itself had no effect on the wild type strain, the observed effect is due to overexpression of the hlyD-ftsEX operon ( Figure 3) while overexpression of hlyD or ftsEX had no effect on muropeptide composition (Supplementary Materials Figure S3B). Astonishingly, we also observed elevated tetra and-to a lower extent-penta muropeptides when we repressed plasmid gene expression with 25 mM glucose, indicating that the PGN of S. carnosus is influenced by genetic as well as metabolic factors.
Antibiotics 2016, 5, 33 6 of 17 However, when the three genes were overexpressed from a plasmid harboring a xylose inducible promoter, all peaks with muropeptides containing a penta stem peptide were highly elevated while the tri muropeptides were still present in comparable amounts. Muropeptides with tetra stem peptides were slightly elevated as well. As 25 mM xylose itself had no effect on the wild type strain, the observed effect is due to overexpression of the hlyD-ftsEX operon ( Figure 3) while overexpression of hlyD or ftsEX had no effect on muropeptide composition (Supplementary Materials Figure S3B). Astonishingly, we also observed elevated tetra and-to a lower extent-penta muropeptides when we repressed plasmid gene expression with 25 mM glucose, indicating that the PGN of S. carnosus is influenced by genetic as well as metabolic factors. (IV) S. carnosus TM300 grown in the presence of 25 mM glucose. Overexpression of hlyD-ftsEX led to an accumulation of penta muropeptides (peak 4, 7 and 10 in panel II) which was not observed when the wild type strain was grown in the presence of 25 mM xylose. Addition of 25 mM glucose in the medium lead to an increase in tetra muropeptides independent of the plasmid pPTX-hlyD-ftsEX (peaks 3, 6 and 9 in panel III and IV, respectively).

Precursor Analysis
To examine if S. carnosus TM300 synthesizes already shortened variants of the classical PGN precursor UDP-MurNAc-pentapeptide, we accumulated and analyzed the PGN precursor of the wild type strain S. carnosus TM300 and of the overexpression strain S. carnosus TM300 pPTX-hlyD-ftsEX ( Figure 4). In S. aureus, the PGN precursor is the UDP-MurNAc-pentapeptide with a mass of 1149.37 g/mol. In both tested S. carnosus strains, this mass was only found in a broad peak group at about 11 min in the TIC of the UPLC-MS. None of the other masses obtained correlated with the UDP-MurNAc-tripeptide (1007.2774 g/mol) or UDP-MurNAc-tetrapeptide (1078.3145 g/mol). This shows that S. carnosus TM300 only synthesizes the classical precursor UDP-MurNAc-pentapeptide and truncation to the tetra and tri stem peptide must occur at a later time point. (IV) S. carnosus TM300 grown in the presence of 25 mM glucose. Overexpression of hlyD-ftsEX led to an accumulation of penta muropeptides (peak 4, 7 and 10 in panel II) which was not observed when the wild type strain was grown in the presence of 25 mM xylose. Addition of 25 mM glucose in the medium lead to an increase in tetra muropeptides independent of the plasmid pPTX-hlyD-ftsEX (peaks 3, 6 and 9 in panel III and IV, respectively).

Precursor Analysis
To examine if S. carnosus TM300 synthesizes already shortened variants of the classical PGN precursor UDP-MurNAc-pentapeptide, we accumulated and analyzed the PGN precursor of the wild type strain S. carnosus TM300 and of the overexpression strain S. carnosus TM300 pPTX-hlyD-ftsEX ( Figure 4). In S. aureus, the PGN precursor is the UDP-MurNAc-pentapeptide with a mass of 1149.37 g/mol. In both tested S. carnosus strains, this mass was only found in a broad peak group at about 11 min in the TIC of the UPLC-MS. None of the other masses obtained correlated with the UDP-MurNAc-tripeptide (1007.2774 g/mol) or UDP-MurNAc-tetrapeptide (1078.3145 g/mol). This shows that S. carnosus TM300 only synthesizes the classical precursor UDP-MurNAc-pentapeptide and truncation to the tetra and tri stem peptide must occur at a later time point.

Investigation of a Putative L,D-Carboxypeptidase
A logical explanation for the observed truncation of the penta stem peptide to a tri stem peptide would be an L,D-carboxypeptidase (LD-CP) activity. Even though such an activity has never been described for staphylococci, we found one potential candidate in the genome of S. carnosus TM300: SCA_0214 is annotated as a cytoplasmic muramoyl-tetrapeptide carboxypeptidase that hydrolyses the bond between a di-basic amino acid and the C-terminal D-alanine in the tetra peptide moiety of PGN. An orthologous gene was not found in S. aureus and no paralog was annotated in S. carnosus TM300. Therefore, a deletion mutant of SCA_0214 was created in S. carnosus TM300. However, deletion of this gene did not alter the PGN pattern compared to the wild type strain (Supplementary Materials Figure S4A). When we overexpressed this gene, we found the resulting protein to accumulate in the cytoplasm and not be transported to the supernatant (Supplementary Materials Figure S4B). As truncation of the stem peptide most likely occurs outside the cell, we exclude SCA_0214 as the sought-after LD-CP.

Influence of Sugars on PGN Composition
As we had observed that high concentrations of glucose (25 mM) led to an increase in muropeptides with tetra stem peptides, we tested other C-sources in this concentration as well. While fructose had the same effect, the muropeptide pattern in the presence of glycerol or ribose was unchanged. PGN of cells grown in B0 medium (B medium without an additional C-source) was also unaffected (Supplementary Materials Figure S5A). Therefore, we examined whether this observation was due to overflow metabolism and deleted the gene for the catabolite control protein A (ccpA). CcpA mainly functions as a gene repressor, but it is also involved in the transcription activation of genes involved in fermentation and overflow metabolism [42,43]. It is itself activated by glucose intermediates like glucose-6-phosphate and fructose-1,6-bisphosphate [44]. Indeed, the effect of high amounts of glucose on PGN composition was lost in the ΔccpA mutant ( Figure 5). In the complemented mutant, the ccpA gene was constitutively expressed from a plasmid, and addition of 25 mM glucose to the medium again caused the appearance of muropeptides with tetra stem peptides in addition to tri stem peptides. Alterations in the PGN pattern of S. carnosus TM300 were not solely reduced to high glucose concentrations and were lost in a ccpA deletion mutant. Therefore, this effect can be explained by an overflow metabolism when an abundance of nutrients is present in the medium rather than solely by a glucose effect. In S. aureus, SA113 an increased glucose concentration showed no effect on the muropeptide pattern (Supplementary Materials Figure S6).

Investigation of a Putative L,D-Carboxypeptidase
A logical explanation for the observed truncation of the penta stem peptide to a tri stem peptide would be an L,D-carboxypeptidase (LD-CP) activity. Even though such an activity has never been described for staphylococci, we found one potential candidate in the genome of S. carnosus TM300: SCA_0214 is annotated as a cytoplasmic muramoyl-tetrapeptide carboxypeptidase that hydrolyses the bond between a di-basic amino acid and the C-terminal D-alanine in the tetra peptide moiety of PGN. An orthologous gene was not found in S. aureus and no paralog was annotated in S. carnosus TM300. Therefore, a deletion mutant of SCA_0214 was created in S. carnosus TM300. However, deletion of this gene did not alter the PGN pattern compared to the wild type strain (Supplementary Materials Figure S4A). When we overexpressed this gene, we found the resulting protein to accumulate in the cytoplasm and not be transported to the supernatant (Supplementary Materials Figure S4B). As truncation of the stem peptide most likely occurs outside the cell, we exclude SCA_0214 as the sought-after LD-CP.

Influence of Sugars on PGN Composition
As we had observed that high concentrations of glucose (25 mM) led to an increase in muropeptides with tetra stem peptides, we tested other C-sources in this concentration as well. While fructose had the same effect, the muropeptide pattern in the presence of glycerol or ribose was unchanged. PGN of cells grown in B0 medium (B medium without an additional C-source) was also unaffected (Supplementary Materials Figure S5A). Therefore, we examined whether this observation was due to overflow metabolism and deleted the gene for the catabolite control protein A (ccpA). CcpA mainly functions as a gene repressor, but it is also involved in the transcription activation of genes involved in fermentation and overflow metabolism [42,43]. It is itself activated by glucose intermediates like glucose-6-phosphate and fructose-1,6-bisphosphate [44]. Indeed, the effect of high amounts of glucose on PGN composition was lost in the ∆ccpA mutant ( Figure 5). In the complemented mutant, the ccpA gene was constitutively expressed from a plasmid, and addition of 25 mM glucose to the medium again caused the appearance of muropeptides with tetra stem peptides in addition to tri stem peptides. Alterations in the PGN pattern of S. carnosus TM300 were not solely reduced to high glucose concentrations and were lost in a ccpA deletion mutant. Therefore, this effect can be explained by an overflow metabolism when an abundance of nutrients is present in the medium rather than solely by a glucose effect. In S. aureus, SA113 an increased glucose concentration showed no effect on the muropeptide pattern (Supplementary Materials Figure S6).

Deletion of ccpA Causes a Decrease in Colony Size and Growth Rate
While the observed variation in PGN composition was lost in the ccpA deletion mutant, this strain was influenced in its growth behavior. There was a decrease in colony size ( Figure 6 and Table  3) of the S. carnosus TM300 ΔccpA mutant of ~60% compared to the wild type strain. Complementation restored the mutant phenotype completely. In liquid culture, the S. carnosus TM300 ΔccpA mutant showed a growth defect during log phase compared to the wild type strain that was restored again by complementation ( Figure 6). Growth rate of the mutant was reduced to 60% in low glucose concentration and to 68% in high glucose concentrations compared to the wild type. This effect was restored by complementation (Table 4). Therefore, in these experiments, the glucose concentration itself had only minor influence, indicating observed effects on the PGN are specific.

Deletion of ccpA Causes a Decrease in Colony Size and Growth Rate
While the observed variation in PGN composition was lost in the ccpA deletion mutant, this strain was influenced in its growth behavior. There was a decrease in colony size ( Figure 6 and Table 3) of the S. carnosus TM300 ∆ccpA mutant of~60% compared to the wild type strain. Complementation restored the mutant phenotype completely. In liquid culture, the S. carnosus TM300 ∆ccpA mutant showed a growth defect during log phase compared to the wild type strain that was restored again by complementation ( Figure 6). Growth rate of the mutant was reduced to 60% in low glucose concentration and to 68% in high glucose concentrations compared to the wild type. This effect was restored by complementation (Table 4). Therefore, in these experiments, the glucose concentration itself had only minor influence, indicating observed effects on the PGN are specific.

Deletion of ccpA Causes a Decrease in Colony Size and Growth Rate
While the observed variation in PGN composition was lost in the ccpA deletion mutant, this strain was influenced in its growth behavior. There was a decrease in colony size ( Figure 6 and Table  3) of the S. carnosus TM300 ΔccpA mutant of ~60% compared to the wild type strain. Complementation restored the mutant phenotype completely. In liquid culture, the S. carnosus TM300 ΔccpA mutant showed a growth defect during log phase compared to the wild type strain that was restored again by complementation ( Figure 6). Growth rate of the mutant was reduced to 60% in low glucose concentration and to 68% in high glucose concentrations compared to the wild type. This effect was restored by complementation (Table 4). Therefore, in these experiments, the glucose concentration itself had only minor influence, indicating observed effects on the PGN are specific.  Table 3. Determination of colony size. The size of ten individual grown colonies was measured digitally. The ccpA mutant showed a 57% decrease in cell size that could be complemented again. The influence of glucose was only minor. The mean values of three independent experiments are given.

The Peptidoglycan Composition of S. carnosus TM300
UPLC/MS analysis of the PGN of S. carnosus TM300 revealed a muropeptide pattern different from the one known of S. aureus. Instead of stem peptides containing the classical five amino acids L-Ala-D-iGln-L-Lys-D-Ala-D-Ala) typical for S. aureus, most stem peptides were degraded to tri peptides missing the two D-Ala. This is the case for monomeric muropeptides (Figure 1a, peak 1) as well as for all cross-linked muropeptides, in which the acceptor stem peptide was also degraded, while the former donor stem peptides are still classical tetra peptides (Figure 1a, peaks 5, 8, 11 and 12). We were able to identify the cross-linked muropeptides up to the pentamer (Tri-Tetra 4 ). Acceptor stem peptides with penta peptides were also found ( Figure 1a, peaks 4, 7 and 10), but in small amounts only. We also found tetra muropeptides containing five (Figure 1a, peak 3) or six to nine glycine residues in the interpeptide bridge (Figure 1a, peak 2). The latter indicates that these muropeptides had been released again from mature PGN. This peak was not influenced by the genetic and metabolic factors tested.
In S. aureus PGN-the best studied PGN in staphylococci-this degradation was never observed [37,38,45]. Tri stem peptides are a classical feature of E. coli [36] but also of other firmicutes like Bacillus subtilis, where the PGN is modified after cross-linking has occurred. Interestingly, we observed the length of the S. carnosus TM300 stem peptide not only to be influenced by genetic factors (in our study hlyD-ftsE-ftsX) but also by the composition of the growth medium (see below). Especially 25 mM glucose, a concentration frequently used for plasmid repression, caused an increase in muropeptides with tetra stem peptides. Changes in stem peptide composition due to medium ingredients have been reported for example for Caulobacter crescentus. It incorporates glycine at position five, even if only traces of free glycine are present in the growth medium [46]. When glycine is depleted in the medium, S. aureus PGN is directly cross-linked and lacks the interpeptide bridge [47]. Cells then enter the stationary phase, indicating this to be a protection mechanism.

The hlyD-ftsE-ftsX Operon
A gene for HlyD (hemolysin D) is present in e.g., Escherichia coli (E. coli) but the orthologue could not be found in staphylococci. The database KEGG and a Blast search for orthologous genes did not lead to any positive results. However, SCA_1997, which was identified in the BTH screen with a genomic library of S. carnosus TM300 as a putative interaction partner for PBP2, was annotated as a conserved hypothetical protein belonging to the HlyD family of secretion proteins. The nomenclature was maintained as SCA_1997 and orthologous genes possess a short HlyD motif that is eponymous for these proteins.
Like the E. coli HlyD, the staphylococcal HlyD is highly similar to the membrane fusion component of the RND family of transporters (RND: Resistance, Nodulation, Cell Division). As the proposed function is to facilitate and enable transport (e.g., resistance through efflux) by bringing the inner and outer membrane together, most members of this family are found in Gram-negative bacteria [48]. Nevertheless, also in Gram-positive bacteria, there are a few representatives of this family. But hlyD is not omnipresent in Staphylococcus. Only about 47% (34 out of 73) of all annotated staphylococci harbor a copy of hlyD. In contrast, in Bacillus, 98% of strains (112 out of 114 strains sequenced) possess this gene. There is also no correlation between hlyD and pathogenicity, as it can be found throughout the whole genus of staphylococci.
While the S. carnosus TM300 FtsEX complex has no sequence similarity to FtsEX of E. coli, the predicted structures of the respective proteins are very similar. An ftsEX depletion mutant in E. coli forms filaments, and growth in LB medium was shown to be salt dependent as the septal ring does not properly assemble in the absence of FtsEX and salt [30]. However, growth of the S. carnosus TM300 hlyD-ftsEX mutant was not affected by the absence or presence of NaCl nor by the use of minimal medium M9 (Supplementary Materials Figure S2) and we did not observe any obvious changes in cell appearance. This indicates the septal ring in staphylococci does not necessarily need ftsEX for stabilization.
Our BTH results are in accordance with known results for PBP2 of S. aureus which was also shown to self-interact [49]. In our experiments, FtsX also interacted with HlyD, with itself and with FtsE ( Figure 2 and Table 2). We could not detect homodimerization of FtsE as was published for the Streptococcus pneumoniae orthologue. In this organism, both FtsE and FtsX formed homodimers and interacted with each other [31]. In various species, FtsE and FtsX have been shown to interact with each other and thereby activate murein hydrolases. This includes PcsB of Streptocccus pneumoniae [31][32][33], the endopeptidase CwlO of Bacillus subtilis [35] and an amidase of E. coli that was activated by FtsEX via EnvC [34]. Our results from S. carnosus point to a role for HlyD-FtsEX in the regulation of a carboxypeptidase, as the overexpression of the hlyD-ftsEX operon led to an increase in muropeptides containing penta stem peptides. To a minor extent, tetra stem peptides were also observed ( Figure 3). Penta stem peptides of S. carnosus are normally degraded into tri stem peptides by a so far unknown enzyme(s). We propose that HlyD-FtsEX must have an inhibitory effect on this/these enzyme/s, as deletion of the whole operon did not alter the muropeptide pattern and therefore did not influence degradation activity. In addition, only overproduction of all three proteins caused an increase in penta and tetra muropeptides, indicating a necessity for complex formation for activity. As tri muropeptides are present in similar amounts to pentas, inhibition of the degradation enzymes is not complete and other factors might contribute to their regulation. This is corroborated by the notion that the hlyD-ftsEX operon is not essential and its deletion had no effect on bacterial growth under various conditions tested.

Investigation of a Putative L,D-Carboxypeptidase (SCA_0214)
The PGN of S. carnosus TM300 mainly consists of tri and tetra peptide structures (Table 1) suggesting L,D-carboxypeptidase (LD-CP) activity. Bioinformatical analysis led to only one possible candidate in S. carnosus TM300: SCA_0214, which is located between SCA_0213 (an NAD dependent epimerase/dehydratase family protein) and SCA_0215 (ilvE, a branched chain amino acid aminotransferase). In contrast to S. carnosus TM300, a gene for an LD-CP is missing from the genome of S. aureus. For instance, the genome of S. a. USA300 does not contain a homologous gene for SCA_0214. Instead, the two genes SCA_0213 and SCA_0215 are located directly next to each other. The absence of a putative LD-CP is in accordance with the presence of only penta and tetra stem peptides in S. aureus PGN [37]. However, deletion of SCA_0214 in S. carnosus TM300 had no effect at all on the muropeptide pattern. It still harbored mainly tri stem peptides ( Figure S4A). Even if SCA_0214 was only able to cleave off D-Ala from a tetrapeptide but not from a pentapeptide, as could be expected from a carboxypeptidase, its deletion should have resulted in an accumulation of tetrapeptides. As this is not the case, we exclude SCA_0214 as being responsible for the occurrence of tripeptides in S. carnosus TM300.
But what could be the role of SCA_0214? Overproduction resulted in protein accumulation in the cytoplasm. This is in accordance with the LD-CP sequence analysis that revealed neither a TAT (Twin Arginine Translocation) nor a Sec (Secretion) sequence. This could point to a role in the cytoplasm, but could also be a mere side effect of protein overproduction.
But which enzyme(s) could cause the reduction from penta to tri stem peptides? As trimming of the stem-peptides during PGN maturation is widely seen in eubacteria, S. carnosus might be "normal", while actually S. aureus is an exception by keeping its pentapeptide intact. Another bioinformatical search in KEGG revealed three annotated DD-CPs in S. carnosus: (1) SCA_0291, the homologue of S. aureus PBP4, which is actually a PGN synthase responsible for secondary cross-linking of PGN [22]; (2) SCA_2445, which is also assigned as an S. aureus PBP4 homologue, but with less homology; and (3) SCA_1643 that is orthologous to LdcB of Streptococcus pneumoniae and B. subtilis but has no orthologue in S. aureus. Both enzymes had originally been annotated as D,D-CP (dacA) but were recently shown to degrade tetra peptides from isolated PGN into tri peptides in vitro and were therefore renamed [50]. SCA_2445 could be the enzyme that creates tetra peptides, and tri peptides are then produced by SCA_1643. We tried to construct deletion mutants of SCA_1643 and SCA_2445 but were unsuccessful.

Influence of Sugars on the Muropeptide Composition
As we had observed an altered PGN pattern when we repressed plasmid gene expression with 25 mM glucose, we tested the influence of various carbon sources on the PGN pattern of S. carnosus TM300. We found an increase of tetra muropeptides when the cells of S. carnosus TM300 were grown with either 25 mM glucose ( Figure 5, chromatogram 2) or fructose ( Figure S5A). Again, tri muropeptides were still present in comparable amounts to these tetra muropeptides. No increase in tetra muropeptides was observed when supplemented with 25 mM xylose ( Figure S5A), the standard gene induction conditions for plasmid pPTX [51,52] that served as a control.
In Gram-positive bacteria, the catabolite control protein A (CcpA), a global regulator of carbon source metabolism, regulates cellular processes that are influenced by the presence of glucose [53]. Therefore, we deleted the ccpA gene and analyzed the PGN composition again under low and high glucose concentrations. As expected, no elevated tetra peaks were observed in the ccpA deletion mutant under high glucose conditions. Instead, the muropeptide pattern was identical to the one obtained from cells grown in 5 mM glucose ( Figure 5, chromatograms 3 and 4). Complementation with a plasmid constitutively expressing ccpA restored the original phenotype, showing that the appearance of tetra muropeptides can be traced back to the influence of glucose and fructose intermediate products on CcpA. Moreover, deletion of ccpA also affected cell growth. The deletion strain show a reduced growth rate in exponential phase and a severe decrease in colony size compared to the wild type strain, indicating a major role for CcpA in the metabolism of S. carnosus TM300.
Glucose and fructose are both metabolized to fructose-1,6-bisphosphate (FBP), an intermediate product of glycolysis. As alterations on the PGN pattern of S. carnosus TM300 did not only occur under high glucose concentrations, this effect could be rather explained by a general overflow metabolism instead of a glucose effect. However, glucose not only serves as a nutrient but also the expression of metabolic genes [54,55] and virulence factors [56] are influenced by glucose.
But how can glucose concentration influence the PGN composition? The increase in tetra muropeptides speaks again for a partial inhibition of a so far unknown LD-CP, although an increase in penta peptide structures was also detected. Taken together, this points to an influence of ccpA and hlyD-ftsEX on the stem peptide degrading enzyme(s) which is albeit not a complete inhibitory effect as tri stem peptides are also still present.
So far we do not understand why S. carnosus TM300 trims its stem peptides in a way that is also seen for other bacteria, but S. aureus does not. It was just shown that in E. coli the eight DD-CPs are active under different conditions. For example, PBP5 (dacA) is mainly active at neutral pH, while PBP6b is only expressed and active at acidic pH. Loss of either enzyme results in morphological defects when the cells are incubated at the respective pH. We, however, did not observe any obvious changes in the morphology of our strains during overexpression of the hlyD-ftsE-ftsX operon or under high glucose conditions. But the notion that even under these conditions tripeptides are present in similar amounts as the newly appearing penta or tetrapeptides can be interpreted as a hint for an important role of the tri peptides in the PGN of S. carnosus TM300.

Materials and Methods
For strains, plasmids and oligonucleotides used please refer to Supplementary Materials Table S1: Bacterial strains, Table S2: Plasmids and Table S3: Oligonucleotides.

Reagents
Unless otherwise stated, all reagents were bought from Sigma-Aldrich.

Gene Deletion
All knockout procedures were performed according to the respective protocols of the vectors. While the pMAD system does not imply the usage of a resistance cassette, the other two knockout vectors were constructed with an ermB (erythromycin B) resistance cassette. Flanking of ermB with loxP sites enabled us to flox out the resistance cassette by the Cre recombinase constitutively produced by the vector pRAB1 [64]. This generated a marker-free mutant minimizing downstream effects caused by the resistance cassette.

Genomic Library
Chromosomal DNA from S. carnosus was isolated [57] using lysostaphin (0.5 mg/mL) for cell lysis. Genomic DNA was sheared using the nebulizer from the TOPO Shotgun Subcloning Kit (Invitrogen, Carlsbad, CA, USA) and separated on a preparative agarose gel. Five pools of DNA from 1.7 to 3 kbp were isolated, cleaned and treated with Klenow fragment to generate blunt ends. DNA from each pool was used for ligation into pKT25 [65], which had been cut with SmaI and dephosphorylated by rAPid Alkaline Phosphatase (Roche, Mannheim, Germany). The resulting plasmids were transformed into NEB 5-alpha electrocompetent E. coli cells (New England Biolabs, Ipswich, MA, USA) according to the manufacturer's protocol. Ligation procedure was performed twice, resulting in ten transformations. Each transformation was plated on three LB/Kanamycin (50 µg/mL) plates (15 cm in diameter) and incubated overnight at 37 • C. All colonies from the same pool were combined and stored in 50% glycerol at −80 • C. From each pool, an aliquot was used to inoculate LB media for plasmid isolation ("prey" vectors). Over 80% of the plasmids contained genomic DNA inserts, and cell wall related genes like pbpA, pbpF, pbp4, rodA and murJ could each be amplified in at least one subpool. SCA_1084 (pbp2) was cloned into pUT18C [65] as "bait" vector.

Bacterial-Two-Hybrid Assays
The method of Karimova et al. was used [65]. The plasmids were co-transformed in E. coli BTH101, plated on LB agar plates containing 100 µg/mL ampicillin, 50 µg/mL kanamycin and 100 µg/mL streptomycin and incubated for 16 h at 37 • C. Ten colonies per each interaction tandem were transferred onto MacConkey agar plates (100 µg/mL ampicillin, 30 µg/mL kanamycin and 100 µg/mL streptomycin) and incubated for 24-48 h at 30 • C. To generate triplicates, the first three colonies per each interaction tandem were inoculated overnight in LB media containing 100 µg/mL ampicillin, 30 µg/mL kanamycin, 100 µg/mL streptomycin and 0.5 mM IPTG. Overnight cultures were diluted in H 2 O bidest (1:4) and cell density was measured at OD 600 . Of each cell suspension, 100 µL was mixed with 20 µL 0.1% SDS, 40 µL chloroform and 1 mL Z buffer (70 mM Na 2 HPO 4 × 12 H 2 O, 30 mM NaH 2 PO 4 H 2 O, 1 mM MgSO 4 , 0.2 mM MnSO 4 , pH 7, 100 mM β-mercaptoethanol). The cells were solubilized by pipetting 10-15 times. Of each supernatant, 100 µL was incubated with 20 µL ONPG solution (ortho-Nitrophenyl-β-galactoside; 4 mg/mL ONPG in Z-buffer without β-mercaptoethanol) for 10 min. ONPG served as substrate for the detection of β-galactosidase activity. The colorless substrate is cleaved to galactose and the yellow colored product ortho-nitrophenol. The reaction was stopped by adding 50 µL of 1 M Na 2 CO 3 and the optical density was determined at 420 nm. The enzymatic activity of the β-galactosidase is given in units/mg dry weight bacteria and was calculated as follows: Calculation of enzyme activity: A = 200 × (OD 420 − OD 420 in control tube)/minutes of incubation × dilution factor Calculation of bacteria dry weight: 1 mL of culture at OD 600 = 1 corresponds to 300 µg dry weight bacteria A negative control was performed by co-transforming the empty plasmids pKT25 and pUT18C into the host strain E. coli BTH101. The cut-off value for positive interactions was four times the negative control.

Peptidoglycan Analysis
The cells were grown in 20 mL BM for 8 h and peptidoglycan was isolated and analyzed by UPLC and UPLC/MS as previously described [37]. Precursors were isolated as published [66] and analyzed by UPLC-MS using the PGN method.

Overproduction of SCA_0214
Main cultures (20 mL B0 media supplemented with either 5 mM glucose for repression or 25 mM xylose for induction) were inoculated with overnight cultures to OD 578 = 0.05. After 8 h culture, supernatant was used to isolate extracellular proteins using StrataClean beads (Agilent Technologies, St. Clara, CA, USA). Intracellular proteins were extracted from pelleted cells. Cells were disrupted using a FastPrep ® -24 (MP Biomedicals, Santa Ana, CA, USA) and 500 µL acid washed glass beads (0.22 mm in diameter, Sigma-Aldrich). Cells were treated 4 times at 6500 rpm for 30 s. After two cycles, the samples were placed on ice for 5 min to reduce potential heat that develops during FastPrep ® -24 treatment and that could lead to protein denaturation. Afterwards, the tubes were centrifuged at 12,000 rpm for 5 min at RT. The supernatant contained the intracellular proteins. Proteins were analyzed by SDS-PAGE stained with Coomassie.

Conclusions
We could show that the PGN of the apathogenic bacterium S. carnosus TM300 differs from the well-studied PGN of the pathogen S. aureus. In TM300 free stem peptides are shortened to contain tri peptides only in contrast to the unmodified penta peptides of S. aureus. This suggests the activities of an LD-and a DD-CP. We could show, that the proposed DD-CP might be regulated by three enzymes that interact with the PGN synthase PBP2: HlyD, FtsE and FtsX. The overexpression of the respective operon results in the reappearance of penta stem peptides. The LD-CP seems to be regulated by the catabolite control protein A (ccpA) as high sugar concentrations result in the appearance of tetra stem peptides. Taken together, our results show that the PGN of S. carnosus can adapt to various conditions. Supplementary Materials: The following are available online at www.mdpi.com/2079-6382/5/4/33/s1. Figure S1: Muropeptide structures of S. aureus and S. carnosus TM300; Figure S2: Growth behavior of S. carnosus TM300 ∆hlyD-ftsEX; Figure S3: Deletion of hlyD-ftsEX and overexpression of single genes does not influence the peptidoglycan pattern; Figure S4: Analysis of the S. carnosus TM300 ∆0214::ermB mutant and localization of the encoded protein; Figure S5: Influence of different sugars on the peptidoglycan of S. carnosus TM3; Figure S6: Influence of glucose on the peptidoglycan of S. aureus SA113 ∆spa;