Synthesis of the L-alanyl-L-alanine cross-bridge of Enterococcus faecalis peptidoglycan.

The enzymatic synthesis of the complete l-alanyl(1)-l-alanine(2) side chain of the peptidoglycan precursors of Enterococcus faecalis was obtained in vitro using purified enzymes. The pathway involved alanyl-tRNA synthetase and two ligases, BppA1 and BppA2, that specifically transfer alanine from Ala-tRNA to the first and second positions of the side chain, respectively. The structure of the UDP-N-acetylmuramoyl-l-Ala-gamma-d-Glu-l-Lys(N(epsilon)-l-Ala(1)-l-Ala(2))-d-Ala-d-Ala product of BppA1 and BppA2 was confirmed by mass spectrometry (MS) and MS/MS analyses. The peptidoglycan structure of the wild-type E. faecalis strain JH2-2 was determined by tandem reverse-phase high-pressure liquid chromatography-MS revealing that most muropeptides contained two l-alanyl residues in the cross-bridges and in the free N-terminal ends. Deletion of the bppA2 gene was associated with production of muropeptides containing a single alanyl residue at these positions. The relative abundance of monomers, dimers, trimers, and tetramers in the peptidoglycan of the bppA2 mutant indicated that precursors containing an incomplete side chain were efficiently used by the dd-transpeptidases in the cross-linking reaction. However, the bppA2 deletion impaired expression of intrinsic beta-lactam resistance suggesting that the low affinity penicillin-binding protein 5 did not function optimally with precursors substituted by a single alanine.


EXPERIMENTAL PROCEDURES
Protein purification-We previously reported (7) detailed procedures for purification of the UDP-MurNAc-pentapeptide:L-alanine ligase of E. faecalis from extracts of E. coli JM83/pDA29(bppA1) by anion exchange, hydrophobic interaction, affinity (heparin), and exclusion chromatography. The same procedures were used to purify BppA2 from extracts of E. coli JM83 harboring plasmid pDA28(bppA2) (7) except that hydrophobic interaction chromatography was omitted since the enzyme was not soluble in high concentrations of ammonium sulfate. Briefly, E. coli JM83/pDA28(bppA2) was grown to an OD 600 of 0.7 in 1 liter of brain heart infusion (BHI) broth containing 100 µg/ml ampicillin, and induction was performed with 1 mM IPTG for 2 hours at 37 °C. Bacteria were disrupted by sonication, centrifuged (27,000 g for 30 min at 4 °C), and the supernatant was loaded at a flow rate of 1 ml/min onto a 10 ml Bio-Scale Q anion exchange column (Bio-Rad, Ivry sur Seine, France) equilibrated in buffer A (50 mM Tris-HCl, pH 7.2; 2 mM 2-mercaptoethanol). Elution was performed with a 0-1 M NaCl gradient in buffer A. A 5 ml fraction eluting at ca. 360 mM NaCl was diluted by addition of 20 ml of buffer B (50 mM potassium phosphate, pH 7.0; 2 mM 2-mercaptoethanol) and loaded onto a HiTrap heparin affinity column (1 ml, Amersham Biotech Pharmacia, Orsay, France) equilibrated in buffer B at a flow rate of 0.25 ml/min. Elution was performed with a 0-2 M NaCl gradient in buffer B providing a 0.25 ml fraction eluting at ca. 850 mM NaCl. Gel filtration was performed with a Superdex 75 HR10/30 column (Amersham Biotech Pharmacia) equilibrated with buffer A containing 200 mM NaCl at a flow rate 0.5 ml/min. The bppA2 gene product eluted between 8.5 to 9.0 ml (300 µg of protein) and was estimated to be >95 % pure by SDS-PAGE. Proteins were determined by the Bio-Rad assay with bovine serum albumin as a standard. 7 The E. faecalis alanyl-tRNA synthetase containing a C-terminal 6 His tag was purified in two steps based on affinity chromatography on a nickel column and exclusion chromatography as previously described (7).

Analysis of UDP-MurNAc-peptides by mass spectrometry-Samples of the UDP-
MurNAc-peptide products were isolated by rpHPLC, lyophilized, and dissolved in Peptidoglycan structure analysis-Bacteria were grown at 37 °C to an optical density of 0.8 in BHI broth, containing 50 µg/ml of vancomycin for JH2-2(vanA+) and JH2-2∆bppA2(vanA+). Peptidoglycan was extracted with 4% sodium dodecyl sulfate at 100° C and treated with pronase (200 µg/ml) and trypsin (200 µg/ml), as described (13). (formerly designated ORF2 and ORF1, respectively), that are related to the Fem proteins of S. aureus (7). The UDP-MurNAc-pentapeptide:L-alanine ligase activity of BppA1 was detected based on addition of L-[ 14 C]Ala to UDP-MurNAc-pentapeptide followed by detection of radioactive UDP-MurNAc-hexapeptide by rpHPLC coupled to liquid scintillation (7). The assay contained tRNA, Mg 2+ , ATP, and purified E. faecalis alanyl tRNA synthetase to generate the Ala-tRNA substrate of the ligase (7). In this report, we have purified the bppA2 gene product (Experimental Procedures) and shown that addition of the purified protein to the assay resulted in the appearance of a novel radioactive product (peak B in Fig. 2A), in addition to the UDP-MurNAc-hexapeptide product of BppA1 (peak A in Fig. 2A). Kinetics (data not shown) revealed that peak B increased slowly between 60 and 240 min after appearance of peak A. Analysis of peptidoglycan structure-The peptidoglycan of E. faecalis JH2-2∆bppA2 and of the parent strain JH2-2 was analyzed by liquid chromatography coupled to mass spectrometry to evaluate the impact of the deletion of the bppA2 gene on the structure of the cross-bridge. Since isomers containing the same number of alanyl residues cannot be distinguished based on mass determination, the peptidoglycan of derivatives of E. faecalis JH2-2 and JH2-2∆bppA2 harboring the vanA gene cluster was also analyzed to identify muropeptides containing a tetrapeptide stem (Fig. 1). Alone or in combination, introduction of the vanA gene cluster and the ∆bppA2 deletion into E. faecalis JH2-2 generated four peptidoglycan types differing by the number of alanyl residues in the C-terminus of the stem peptides, in the cross-bridges, and in the free N-terminal side chains (Fig. 3). Based on previous analyses (2,13), the majority of the muropeptides from the wild-type strain was expected to contain two D-alanyl residues at the free C-terminus of the peptide stems (a = 2 in Fig. 3A) and two L-alanyl residues both in the cross-bridges (b = 2) and in the free N-terminal side chains (c=2). Depending upon the degree of oligomerization (n), the number of alanyl residues (k), defined as k = a + n*b + c, should be equal to 4, 6, 8, and 10 for n = 0 (monomer), n=1 (dimer), n=2 (trimer), and n=3 (tetramer) (Fig. 3B). The calculated masses of these structures (Fig. 3B) matched the observed masses ( Fig. 3C and data not shown) of the most abundant monomer, dimer, trimer, and tetramer of JH2-2 (Table I).
Muropeptides containing a tripeptide stem may be generated by an L,D-carboxypeptidase cleaving the L-Lys 3 -D-Ala 4 peptide bond (14) or originate from incorporation into the cell wall of incomplete peptidoglycan precursors lacking the C-terminal D-Ala-D-Ala extremity.
Additional muropeptides of lower abundance were accounted for by the following modifications of the main structures described above (Fig. 3 and data not shown). In agreement with previous analyses (2,13,15) were shown to transfer an alanyl residue from L-alanyl-tRNA to UDP-MurNAc-pentapeptide (7,8). In this study, we have identified the ligase for incorporation of the second residue of the N ε -L-Ala 1 -L-Ala 2 side chain of the E. faecalis peptidoglycan precursors based on purification of the bppA2 gene product, demonstration of its UDP-MurNAc-hexapeptide:L-alanine ligase activity, and analysis of the structure of the heptapeptide stem by MS/MS (Fig. 2). The BppA1 and BppA2 ligases were able to function independently from each other and specifically added the first and second residue of the side chain, respectively. This is the first report of the enzymatic synthesis of a complete branched peptidoglycan precursor in vitro.
In wild-type E. faecalis JH2-2, the substantial majority of the muropeptides contained two L-alanyl residues in the cross-bridge and in the free N-terminal side chain ( Fig. 3 and Table I). This observation implies that the BppA1 and BppA2 ligases efficiently synthesized the L-alanyl-L-alanine side chain prior to the translocation of the peptidoglycan precursors to the cell surface. Deletion of the bppA2 gene was associated with production of muropeptides containing a single alanyl residue both in the cross-bridge and in the free N-terminal side chain. Thus, E. faecalis JH2-2 did not produce any enzyme that could substitute for BppA2.
In agreement, BppA1 was unable to add in vitro the second alanyl residue of the side chain and no additional bpp homologue was detected in the genome of E. faecalis V583.
Analysis of the structure of mature peptidoglycan in mutants of S. pneumoniae have established that the murM and murN gene products are required for incorporation of the 1 st (L-Ala or L-Ser) and 2 nd (L-Ala) amino acid of the side chain (16,17). Amino acid sequence identity between BppA1 and MurM (39%) and between BppA2 and MurN (38%) indicates that the proteins necessary for incorporation of the first and second residues of the side chain in E. faecalis and S. pneumoniae may be considered as orthologues (24-25 % if paralogues are compared, ref. 7). The relationships between the Bpp ligases and more distantly related homologues from S. aureus are less obvious (7). In the latter bacterium, inactivation or impaired expression of the fmhB, femA, and femB genes leads to an increase in the relative abundance of muropeptides containing 0, 1, and 3 glycyl residues, respectively (4,18). These observations indicate that the corresponding gene products are required for incorporation of glycyl residues at the 1 st (FmhB), 2 nd and 3 rd (FemA), and 4 th and 5 th positions (FemB) of the pentaglycine side chain. Thus, FemA and FemB may each be responsible for incorporation of two residues. If this is the case, elongation of the side chain should proceed by sequential addition of glycyl residues from glycyl-tRNA to peptidoglycan precursors, since the dipeptide glycyl-glycine is not incorporated into peptidoglycan by particulate enzyme preparations from S. aureus and synthesis of glycyl-glycyl-tRNA was not detected (19). Alternatively, FemA and FemB may only add the 2 nd and 4 th residues. This would imply that the 3 rd and 5 th residues are added to the side chain by as yet uncharacterized enzymes. Candidate genes for these functions do exist in S. aureus since the genome contains a total of five bpp homologues (20), a number of genes that matches the number of glycyl residues in the crossbridge.
Structural variations at the C-terminus of the stem peptide and in the side chain are expected to affect interaction of the D,D-transpeptidases with their donor and acceptor substrates, respectively (Fig. 1). However, expression of the vanA gene cluster and deletion of the bppA2 gene, alone or in combination, had little impact on peptidoglycan cross-linking (Table I). Thus, both modifications of the structure of the peptidoglycan precursors were tolerated by the D,D-transpeptidases of E. faecalis.
In S. aureus, inactivation of femA or femB abolishes methicillin resistance mediated by the low affinity PBP2a (4), whereas fmhB is essential for viability (18). In S. pneumoniae, murM and murN are both unessential genes (16,17). Inactivation of murM prevents expression of β-lactam resistance mediated by mosaic PBPs whereas disruption of murN only produces a modest decrease in the level of resistance to these antibiotics (16,21).
Enterococcus faecalis is intrinsically resistant to third generation cephalosporins, such as ceftriaxone, due to production of a low-affinity D,D-transpeptidase (PBP5) which is produced by nearly all members of the species (22). The bppA2 deletion led to a moderate decrease (8-fold) in the minimal inhibitory concentration of ceftriaxone. This observation suggests that PBP5 did not function optimally with incomplete peptidoglycan precursors as proposed for the low-affinity penicillin-binding proteins responsible for acquired resistance to β-lactam antibiotics in S. aureus and S. pneumoniae (4,16,21).
Screening of bacterial genomes (data not shown) revealed the presence of bpp homologues in all bacteria for which production of branched peptidoglycan precursors containing L-amino acid or glycyl residues can be inferred from existing data on the structure of the cross-bridges (3). Conversely, bpp homologues were not detected if the cross-bridges contained D-amino acids (e.g. E. faecium and Lactococcus lactis) or if the peptidoglycan was directly cross-linked (e.g E. coli and Bacillus subtilis). Thus, the ligases for incorporation of L-amino-acids into the side chain of peptidoglycan precursors appear to form a unique family of non-ribosomal peptide bond synthesizing enzymes that use aminoacyl-tRNAs as substrates. However, the proteins do not possess a unique fold as the recently solved X-ray crystal structure of FemA revealed striking similarities with a histone acetyltransferase (23). Together, these observations suggest that the ligases for synthesis of branched peptidoglycan precursors could be useful targets for the development of narrow-spectrum antibacterial agents active against certain β-lactam resistant Gram-positive bacteria. Such drugs would be expected to restore the activity of β-lactam antibiotics against strains producing low-affinity PBPs and to be active alone if one of the ligases is essential, as this is the case for FmhB in S. aureus.