Bacterial strains and plasmids used in this study are listed in Table 1. Staphylococcus aureus strains were grown at 37°C with aeration in tryptic soy broth (TSB) or tryptic soy agar (TSA) (Difco Laboratories, Detroit, Mich.). Transposition mutant RUSA208 [4] and the conditional mutant strains RN4220pCadmurT-gatD and COLpCadmurT-gatD, the double mutant RUSA208pCadmurT-gatD, the complemented strains COLpCadmurT-gatD+pSKmurT and COLpCadmurT-gatD+pSKgatD and the control strain COLpCadmurT-gatD+pSK were grown in the presence of the respective antibiotics (Table 1). The growth medium was supplemented with 0.2 µM of cadmium chloride (CdCl₂; Sigma, St. Louis, MO), unless otherwise described.

Escherichia coli strains (Table 1) were grown in Luria-Bertani broth (LB; Difco Laboratories) with aeration at 37°C. Erythromycin (10 µg/ml), neomycin sulphate (50 µg/ml), kanamycin (50 µg/ml), chloramphenicol (10 µg/ml) and ampicillin (100 µg/ml) from Sigma were used for the selection and maintenance of S. aureus and E. coli mutants.

The amino acid sequences of ORFs SACOL1951 (MurT) and SACOL1950 (GatD) were retrieved from the UniProtKB database [5], and their domain architecture was checked using the InterProScan tool [6]. The domains were aligned through TCoffee [7]. Given the limited similarity between sequences, secondary structure inference was used as an independent benchmark for the alignment. This inference was accomplished through Psipred [8]. Position specific annotation other than the one present in the InterPro documentation was collected from references [1], [9].

Restriction enzymes from New England Biolabs (Beverly, MA) were used as recommended by the manufacturer. Routine PCR amplification was performed with Tth DNA polymerase (HT Biotechnology, Cambridge, United Kingdom) and PCR amplification for cloning purposes was performed using Pfu DNA polymerase (Stratagene, Heidelberg, Germany).

For plasmid DNA extraction High pure Plasmid Purification Kit (Roche, Basel, Switzerland) was used. PCR and digestion products were purified using High pure PCR Purification Kit (Roche). Ligation reactions were performed using Rapid DNA Ligation kit (Roche).

Reverse transcription (RT)-PCR was performed as described [10] using total RNA from strain COL as the template. The primers used for the reverse transcription reactions are described in Table 2 and the amplification conditions were: 94°C for 2 min; 40 cycles of 94°C for 30 s, 53°C for 30 s, and 72°C for 2 min; and one final extension step of 72°C for 5 min.

A 918- bp DNA fragment of murT gene was amplified by PCR using chromosomal DNA from strain COL as a template and the specific primers PmurT′-R and PmurT′-F (Table 2). The amplification conditions used were as follows: 94°C for 4 min; 30 cycles, each consisting of 94°C for 30 s, 50°C for 30 s, and 72°C for 1 min 30 s; and one final extension step of 72°C for 10 min. The amplified fragment and the integrative plasmid pBCB20, carrying a cadmium chloride inducible promoter (R.G. Sobral and M.G. Pinho, unpublished) were both digested with SmaI and ligated, generating plasmid pMurT′.

Plasmid pMurT′ was electroporated into competent cells of RN4220 with a Gene Pulser apparatus (Bio-Rad, California) under conditions described previously [11]. Selection of the transformants was performed using kanamycin (50 µg/ml), neomycin sulphate (50 µg/ml) and 0.2 µM of Cadmium chloride. The correct insertion of pMurT′ into RN4220 chromosome was confirmed by PCR, using an internal murT primer chosen outside the region cloned and an internal pBCB20 primer (Table 2). The murT-gatD conditional mutation was then transduced, by phage 80α to the background of COL as previously described [12] and mutant COLpCadmurT-gatD was obtained.

A 1673 bp DNA fragment, including the complete murT coding sequence and 300 bp of the immediate upstream region was amplified from COL genome using the primers PmurTSalI and PmurTBamHI (Table 2). The amplified murT fragment and plasmid pSK5632 [13] were digested with SalI and BamHI and ligated, generating the replicative plasmid pSKmurT. The same strategy was used for the construction of the replicative plasmid pSKgatD, in which a 1088 bp DNA fragment including the complete gatD gene sequence and 300 bp of the immediately upstream region.

Plasmids pSKmurT and pSKgatD were separately introduced into RN4220 by electroporation and subsequently transferred to COLpCadmurT-gatD by transduction, generating COLpCadmurT-gatD+pSKmurT and COLpCadmurT-gatD+pSKgatD, respectively. Plasmid pSK5632 was also introduced in the conditional mutant, providing the control strain COLpCadmurT-gatD+pSK.

The murT-gatD conditional mutation was transduced, using phage 80α, to the background of RUSA208. The obtained double mutant RUSA208pCadmurT-gatD, has a transposon insertion in glnRA operon and the murT-gatD operon under the control of pCad promoter.

Cells were grown in TSB at 37°C to mid-exponential phase (OD_{620 nm} of 0.7). Prior to harvesting the cells, the RNA protect reagent (QIAGEN, Hilden, Germany) was added to the cultures. Total RNA was isolated as previously described [14]. PCR amplified internal fragments of the murT, gatD, SACOL1949-SACOL1948, SACOL1952, glnA and pta genes were used as probes for hybridization (the primers used are listed in Table S1). The DNA probes were labeled with [α-³²P]dCTP (Perkin Elmer, MA, USA).

Isolation of cell wall was performed as described [2], [15]. Briefly, cells were harvested by centrifugation, washed twice with cold 0.9% NaCl, resuspended in 0.9% NaCl and boiled for 20 min. After chilling on ice, the suspension was centrifuged and washed twice with 0.9% NaCl. The cells were disrupted using 106 µm glass beads (Sigma) and FastPrep FP120 apparatus (Bio 101, La Jolla, Calif.), purified, washed, and boiled for 30 min in 5% SDS, diluted in 50 mM Tris/HCl pH 7, to remove non-covalently bound proteins. After centrifugation, 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°C to degrade cell-bound proteins. Purified cell walls were washed with double-distilled water and lyophilized.

Lyophylised cell wall was treated with 49% of hydrofluoric acid for 48 hours at 4°C in order to remove teichoic acids. The teichoic acid free peptidoglycan was washed with water several times to remove all traces of hydrofluoric acid and then lyophylised.

Identical amounts of peptidoglycan were digested with mutanolysin (1 mg/ml; Sigma). The resulting muropeptides were reduced with sodium borohydride and separated by reverse-phase-high performance liquid chromatography (RP-HPLC) using a Hypersil ODS (Runcorn Cheshire, UK) column (3 µm particle size, 250×4.6 mm, 120 Å pore size) and a linear gradient from 5% to 30% MeOH in 100 mM sodium phosphate buffer pH 2.5 at a flow rate of 0.5 ml/min as described [16].

Highly purified cell wall was prepared as previously described [2] and resuspended to a final concentration of 10 mg/ml. Cell wall material (500 µg) was digested with lysostaphin (300 µg) in 20 mM amonium acetate, pH 4.8, for 24 h at 37°C with stirring. Subsequently, cellosyl (Höchst AG, Frankfurt, Germany) (15 µg) was added to the reaction mixture which was incubated for 12 h at 37°C. Finally, additional 15 µg of cellosyl was added and the incubation continued for an additional 12 h. The enzymatic reaction was stopped by boiling the samples for 5 min and insoluble contaminants were removed by centrifugation. The digested cell wall was reduced with sodium borohydride and the resulting monomeric muropeptides were separated by RP-HPLC using a Prontosil (Bischoff, Leonberg, Germany) column (3 µm, particle size, 250×4.6 mm, 120 Å pore size), and a linear gradient from 0% to 30% MeOH in 10 mM sodium phosphate buffer pH 6.0 at a flow rate of 0.5 ml/min.

The eluted fractions corresponding to the most predominant peaks of the chromatograms were collected after HPLC separation, concentrated to 10–20 µl, and acidified with 1% trifluoroacetic acid (TFA). The samples were then desalted and further concentrated using ZipTips (C- 18, Millipore, UK) according to the standard protocol recommended by the manufacturer. The material was eluted from the ZipTip with 3 µl of 50% acetronitrile, 0.1% TFA and was sprayed directly into a Finnigan LTQ-FT mass spectrometer (Thermo, Bremen, Germany) operating in positive mode (Pinnacle Proteomics Facility, Newcastle University, UK) [17].

The UDP-linked peptidoglycan precursors from the cytoplasmic pool were isolated using a modified protocol [18]. Briefly, vancomycin (Sigma) was added (at five times the minimal inhibitory concentration) to mid-exponential grown cultures and incubation proceeded for additional 30 minutes. The cultures were then chilled below 10°C, cells were harvested, suspended in cold water and slowly stirred into the same volume of boiling water for 15 minutes. After centrifugation the supernatant was collected, lyophilized, dissolved in water and the pH was adjusted to 4.0 using 20% phosphoric acid. The suspension was again centrifuged and the pH of the supernatant adjusted to 2.0. The suspension was centrifuged at 4°C for 1 h at 200000 g.

The UDP-linked peptidoglycan precursors were separated through the same column used to separate the muropeptides of peptidoglycan – using a linear gradient from 0 to 30% of MeOH in 100 mM sodium phosphate buffer (pH 2.0), with a flow rate of 0.5 ml/min. Compounds to be analyzed by MS were isolated and desalted using the same column as before with a linear gradient from 0 to 30% of MeOH in 10 mM of sodium phosphate (pH 4.3) for 25 min with a flow rate of 0.5 ml/min. Mass spectral data were obtained by MALDI-TOF analysis (Pinnacle Proteomics Facility, Newcastle University, UK).

Overnight grown cultures of strains COL and COLpCadmurT-gatD, COLpCadmurT-gatD+pSKmurT, COLpCadmurT-gatD+pSKgatD and COLpCadmurT-gatD+pSK were diluted 1∶1,000 into fresh TSB supplemented with the respective antibiotics (Table 1). The conditional mutants were grown in media containing CdCl₂ concentrations at 0, 0.01, 0.05 and 0.2 µM. The cultures were incubated at 37°C with agitation and the OD₆₂₀ nm was monitored over time.

Overnight grown cultures of strains COL and COLpCadmurT-gatD, COLpCadmurT-gatD+pSKmurT and COLpCadmurT-gatD+pSKgatD and COLpCadmurT-gatD+pSK were plated on TSA supplemented with increasing concentrations of CdCl₂ (0, 0.01, 0.05 and 0.2 µM) and incubated at 37°C for 24 hours. Oxacillin (Sigma) diffusion disks (1 mg) were used to determine inhibition halos.

To analyze the susceptibility of peptidoglycan to lysozyme hydrolysis, a turbidometric assay was used as described [2], [19]. Briefly, 0.5 mg of purified peptidoglycan from the conditional mutant, grown with and without CdCl₂, were sonicated in 1 ml of 100 mM Sodium-Potassium phosphate buffer pH 6.6. Human lysozyme or hen egg white lysozyme (Sigma) was added to a final concentration of 300 µg/ml and the reaction was incubated at 37°C. The optical density was monitored at 660 nm.

The impact of lysozyme on exponential growth was determined as described [19]. Overnight cultures of the conditional mutant grown with inducer were diluted to an OD_{620 nm} of 0.1 in fresh TSB (with and without inducer). The cultures were incubated at 37°C until an OD_{620 nm} of 1.0. Then, each culture was diluted 1∶10 into fresh TSB medium and lysozyme (300 µg/ml) was added at an OD_{620 nm} of 1.0. The growth was monitored for several hours.

The same procedure was done using 20 µg/ml of Polymyxin B (Sigma), a cationic antimicrobial peptide.

DNA sequence analyses of murT-gatD region suggested that murT and gatD are located in the same operon and might be co-transcribed from a common promoter: the murT stop codon and the gatD methionine codon are separated by 4 bp only; both genes are transcribed in the same direction and no promoter region sequence could be found upstream from gatD (Figure 1A).

Reverse transcription-PCR (RT-PCR) was performed using total cDNA of strain COL with forward primers specifically binding to murT and a reverse primer specifically binding to gatD. The test yielded products of the expected size (Figure 1B, lanes C and D). No PCR product was obtained from the negative control using primers from the SACOL1949-1948 region, which was found by northern blotting not to be transcribed (Figure 1B, lane A). A PCR product of the expected size was obtained for the positive control, using primers internal to pta, a housekeeping gene.

The results of the RT-PCR test indicated that both murT and gatD are co-transcribed from a common promoter.

Analysis of genome sequences available showed that the murT and gatD genes occur, widespread among bacteria, as a syntenic block, although it is not a universal feature. This is in agreement with our RT-PCR results, which identified the two genes as a small operon. The distribution of this syntenic block among the prokaryotes, with emphasis on the Staphylococcaceae, is shown in Figure S1.

In order to explore the functions of these uncharacterized genes we constructed a mutant strain containing a single chromosomal copy of murT-gatD under the control of an inducible promoter (pCad). A DNA fragment of murT gene which includes the first 298 codons and the ribosome binding site but not the promotor region, was cloned into the integrative plasmid pBCB20 (see Table 1). The recombinant plasmid was electroporated into RN4220 and the chromosomal construct was transduced into the background of the MRSA strain COL. The only complete functional copies of murT and gatD genes were located immediately downstream from the pCad, generating mutant COLpCadmurT-gatD (Figure S2). Hence, this strain expresses the murT-gatD genes when grown in the presence of Cd²⁺, and both genes are depleted when Cd²⁺ is absent from the growth medium (see below).

Northern blotting assays were performed in order to confirm the specificity of transcription of the murT-gatD operon controlled by the CdCl₂ concentration in the medium. The transcription of murT, gatD, SACOL1952 and SACOL1948-SACOL1949 genes was analyzed for COL and mutant COLpCadmurT-gatD grown with several concentrations of inducer. The level of murT and gatD transcription was found to increase with the inducer concentration in the medium (data not shown). No alterations were detected under the same conditions in the transcription level of the ORFs located in the immediate vicinity of the murT-gatD operon, SACOL1952 and SACOL1949-SACOL1948, which were found to be not transcribed even for strain COL (data not shown). The housekeeping gene pta was used as control.

For strain COL, a single transcript was visualized for each gene: an mRNA structure of approximately 1780 nt long hybridized with murT probe and an mRNA structure of approximately 2300 nt long was obtained for hybridization with gatD probe. The size of this last transcript matches the size of both genes, consistent with their co-transcription.

Cell walls of parental strain COL and of the conditional mutant, grown with and without Cd²⁺, were purified and digested with cellosyl and lysostaphin. The resulting monomeric muropeptides were reduced and analyzed by RP-HPLC. The muropeptide profiles revealed that, when the transcription of murT-gatD operon was inhibited, two new muropeptide structures appeared in the RP-HPLC profile (Figure 2A – peaks V and VI). These two muropeptide species showed shorter retention times than peak I, which is common to all the profiles. To identify the structural modifications, all peaks annotated in Figure 2A were isolated and analyzed by MS.

The MS results (Table 2) indicate that the two new peaks (V and VI) observed in the profile of the murT/gatD depleted cells corresponded to muropeptide structures with D-iso-glutamate in the stem peptide replacing D-iso-glutamine. Peaks I, II, III and IV correspond to muropeptide structures with D-iso-glutamine (Figure 2B).

Amidated muropeptides (Peak I) were still present when the transcription of murT-gatD operon was inhibited. This could be due to the activity of MurT and GatD expressed by residual transcription from the pCad promoter or to the presence of other enzymes with the same activity. These findings identify the protein products of murT-gatD as essential for the full amidation of the D-glutamic acid residues in the S. aureus peptidoglycan.

The cell walls of the parental strain COL and the conditional mutant COLpCadmurT-gatD grown with different concentrations of inducer were extracted, the peptidoglycan purified, digested with muramidase and the muropeptides analyzed by RP-HPLC (Figure 3A).

The elution profile of the conditional mutant grown in the absence of CdCl₂ showed longer retention times for all peaks, when compared with COL. In addition, the peaks corresponding to muropeptide structures with higher oligomerization level (retention time over 60 min) were split into two or more smaller peaks eluting at very similar retention times. The elution profiles of the mutant grown with 0.01 µM and 0.05 µM of CdCl₂ showed gradual re-establishment of the parental muropeptide pattern. For cells grown in 0.2 µM CdCl₂ supplemented medium, the optimal inducer concentration, the peptidoglycan HPLC profile was indistinguishable from that of strain COL.

The muropeptide elution profile of COLpCadmurT-gatD, grown in the absence of inducer, showed similarities to the elution profile of the previously characterized glnRA transposition mutant RUSA208 [4] (Figure 3B). In RUSA208, the transposon inserted into the glnR gene which codes for the repressor of the glutamine synthetase operon glnRA, resulting in the abolishment of glnA transcription.

The impact of the glnRA mutation on the peptidoglycan of RUSA208 has been described as the substitution of the normal D-iso-glutamine residues by D-iso-glutamic acid at position 2 of the stem peptide [4]. Substitution of iso-glutamine by iso-glutamic acid residues has been observed among muropeptide monomers (Peak 5A in Figure 3B), among dimeric muropeptides (Peaks 11A & 11B in Figure 3B), among the tripeptide structures (Peak 15B in Figure 3B) and among three of the stem peptides represented by peaks 15A, B & C in Figure 3B. All these structures are also present in the conditional mutant COLpCadmurT-gatD grown in the absence or at suboptimal concentrations of the inducer (see Figure 3A).

The glnA gene sequence in COLpCadmurT-gatD was identical to that in strain COL, excluding the possibility that a mutation in glnA causes the deficiency in peptidoglycan amidation as it occurs in RUSA208 strain. Also, the transcription of the glnA gene did not vary with the Cd²⁺ concentration in COLpCadmurT-gatD (data not shown) discarding the hypothesis that murT and/or gatD may indirectly reduce glnA transcription.

The peptidoglycan profiles of RUSA208 and COLpCadmurT-gatD grown with no CdCl₂, showed that amidation of the muropeptides still occured partially. This may be due to a leaky expression of murT-gatD operon through pCad promoter in the absence of CdCl₂. In the case of RUSA208, other sources of amino group, besides glutamine, may be used, although less efficiently.

The peptidoglycan HPLC profile of the double mutant RUSA208pCadmurT-gatD showed a virtually complete lack of amidated muropeptides (Figure 3A), indicating that the gene products of these two operons are together needed for the amidation of the glutamic acid residue of the peptidoglycan.

The transcriptional analysis showed that the expression of both murT and gatD genes is being controlled in COLpCadmurT-gatD mutant, through the concentration of inducer added to the medium. For this reason we constructed two independent complementation mutants, COLpCadmurT-gatD+pSKmurT and COLpCadmurT-gatD+pSKgatD, by separately introducing into the COLpCadmurT-gatD mutant, the replicative plasmid pSK5632 with either the murT or the gatD gene. Cloning of the murT-gatD operon into pSK5632 was also attempted, but this construct did not yield viable E. coli transformants. Strain COLpCadmurT-gatD+pSK harboring pSK5632 with no cloned gene was constructed and used as control.

With the two complementation strains available, we obtained three distinct levels of re-establishment of the normal peptidoglycan: i) the in trans complementation with several copies of the murT gene showed a partially restored peptidoglycan with a small amount of muropeptides containing glutamic acid residues (COLpCadmurT-gatD+pSKmurT – 0 µM CdCl₂, Figure S3A); ii) the in trans complementation with several copies of gatD gene showed no re-establishment of the normal peptidoglycan profile (COLpCadmurT-gatD+pSKgatD – 0 µM CdCl₂, data not shown); iii) the in trans complementation with several copies of the murT gene and sub-optimal expression of the chromosomal copy of murT-gatD operon showed complete restoration of the peptidoglycan profile (COLpCadmurT-gatD+pSKmurT – 0.01 µM CdCl₂, Figure S3B). In the latter case (iii), the 0.01 µM CdCl₂ of added inducer is responsible for providing a sub-optimal number of copies of murT-gatD transcripts, adding to the already available copies of murT transcript provided in trans. The few copies of gatD provided in this condition are enough for a complete re-establishment of the normal peptidoglycan composition. Thus, complementation of the murT-gatD-depletion phenotype requires the expression of murT and at least a basal level of gatD.

In order to identify the biosynthetic stage at which amidation occurred, the cell wall precursor pool was analyzed by RP-HPLC from strains COL and for the murT-gatD conditional mutant grown with and without the inducer. The HPLC profiles were identical for the three conditions analyzed (Figure S4). The major peak, eluting at 38 minutes, was isolated from the cytoplasmic fractions of COL and of the murT-gatD conditional mutant grown with and without the inducer. The corresponding structures were analyzed by MALDI-TOF MS. The results indicated an identical molecular mass of 1149.35 (neutral mass) for each of the three samples, consistent with the structure of the UDP-MurNAc-L-Ala-D-iGlu-L-Lys-D-Ala-D-Ala, the last cytoplasmic precursor. The presence of D-iso-glutamate in these three structures indicated that the conversion of glutamic acid to iso-glutamine residues must occur at a later stage of cell wall precursor biosynthesis – most likely in the lipid phase – confirming an earlier finding [20].

MurT shares approximately 15% identity and 53% similarity with the sequence of the Mur ligases of S. aureus. Interestingly, while MurT shares the characteristic Mur ligase central domain [1], [21] as defined at InterPro (IPR013221), Pfam (PF08245) and Panther (PTHR23135) MurT lacks the flanking N- and C-terminal domains (Figure S5A).

Among the conserved residues were some critical motifs required for ATP and Mg²⁺ binding and other conserved sites that may not be directly involved in catalysis (Figure S5A). In addition, the MurT protein has a C-terminal domain of unknown function (Pfam: DUF1727, InterPro: IPR013564), which is also found at the C-terminus of more than 900 sequences of prokaryotic proteins at UniProt, and in 5 different domain architectures, all of them sharing the same ORF, or in contiguous ORFs, with Mur central domain (PF08353).

GatD shows similarity to one of the two domains of a cobyric acid synthetase protein: a glutamine-dependent amidotransferase (Gn-AT), with glutamine amide transfer (GAT) activity. Its architecture comprises the overlapping domain signatures of CobB/CobQ_GATase (InterPro: IPR017929), and GATase_3 (InterPro: IPR011698) domains. Through multiple sequence alignment of the N-terminal region of three known Gn-ATs, the absence of a large fragment was noted in GatD (Figure S5B). This missing fragment included important residues for the dethiobiotin synthase activity [9] and part of the ATP binding motif. By placing the representation of the secondary structures over the sequence alignment, we can observe considerable agreement between the shared regions, especially near the reactive center of GATase_3 (Figure S5B). This domain harbored the conserved residues directly involved in GAT activity, according to IPR011698. GatD was also found to contain the unusual Triad family glutamine amidotransferase domain with conserved Cys and His residues (Figure S5B), but lacking the Glu residue of the catalytic triad, as the CobB and CobQ proteins [9].

The murT-gatD operon emerged as a syntenic block that seems to be widespread among bacteria. Interestingly, for the distant taxa of Actinobacteria, in some rare cases, the two ORFs are merged into a single one (Figure S1).

The genome co-localization of the two determinants, together with data from sequence analysis, led us to suggest a model for the coordinated function of MurT and GatD proteins in the peptidoglycan glutamate amidation (Figure 6).

Both proteins together harbor all domain functions required for amidation of peptidoglycan precursor: MurT may be responsible for the recognition of the reaction substrates, the lipid-linked peptidoglycan precursor and ATP, while GatD could be the catalytic subunit involved in the transfer of the amino group from free glutamine to the peptidoglycan precursor. The GatD sequence lacks an ATP binding motif which is common to all members of the Gn-AT family suggesting an activity that depends on the MurT protein which exhibits a typical Mur ligase central domain including the ATP binding motif (Figure 6).

Experiments are in progress to better define the roles of MurT and GatD proteins in the mechanism of amidation of S. aureus peptidoglycan precursor. Irrespective of mechanistic details, the results with the conditional mutant of murT/gatD clearly indicate that the amidation of glutamic acid residues in the S. aureus peptidoglycan is catalyzed by the concerted action of these two enzymes. The murT-gatD operon appears to be the last missing genetic determinant to account for the structural variation in the S. aureus peptidoglycan.