Characterization of tRNA-dependent Peptide Bond Formation by MurM in the Synthesis of Streptococcus pneumoniae Peptidoglycan*

MurM is an aminoacyl ligase that adds l-serine or l-alanine as the first amino acid of a dipeptide branch to the stem peptide lysine of the pneumococcal peptidoglycan. MurM activity is essential for clinical pneumococcal penicillin resistance. Analysis of peptidoglycan from the highly penicillin-resistant Streptococcus pneumoniae strain 159 revealed that in vivo and in vitro, in the presence of the appropriate acyl-tRNA, MurM159 alanylated the peptidoglycan ϵ-amino group of the stem peptide lysine in preference to its serylation. However, in contrast, identical analyses of the penicillin-susceptible strain Pn16 revealed that MurMPn16 activity supported serylation more than alanylation both in vivo and in vitro. Interestingly, both MurMPn16 acylation activities were far lower than the alanylation activity of MurM159. The resulting differing stem peptide structures of 159 and Pn16 were caused by the profoundly greater catalytic efficiency of MurM159 compared with MurMPn16 bought about by sequence variation between these enzymes and, to a lesser extent, differences in the in vivo tRNAAla:tRNASer ratio in 159 and Pn16. Kinetic analysis revealed that MurM159 acted during the lipid-linked stages of peptidoglycan synthesis, that the d-alanyl-d-alanine of the stem peptide and the lipid II N-acetylglucosaminyl group were not essential for substrate recognition, that ϵ-carboxylation of the lysine of the stem peptide was not tolerated, and that lipid II-alanine was a substrate, suggesting an evolutionary link to staphylococcal homologues of MurM such as FemA. Kinetic analysis also revealed that MurM recognized the acceptor stem and/or the TΨC loop stem of the tRNAAla. It is anticipated that definition of the minimal structural features of MurM substrates will allow development of novel resistance inhibitors that will restore the efficacy of β-lactams for treatment of pneumococcal infection.

MurM is an aminoacyl ligase that adds L-serine or L-alanine as the first amino acid of a dipeptide branch to the stem peptide lysine of the pneumococcal peptidoglycan. MurM activity is essential for clinical pneumococcal penicillin resistance. Analysis of peptidoglycan from the highly penicillin-resistant Streptococcus pneumoniae strain 159 revealed that in vivo and in vitro, in the presence of the appropriate acyl-tRNA, MurM 159 alanylated the peptidoglycan ⑀-amino group of the stem peptide lysine in preference to its serylation. However, in contrast, identical analyses of the penicillin-susceptible strain Pn16 revealed that MurM Pn16 activity supported serylation more than alanylation both in vivo and in vitro. Interestingly, both MurM Pn16 acylation activities were far lower than the alanylation activity of MurM 159 . The resulting differing stem peptide structures of 159 and Pn16 were caused by the profoundly greater catalytic efficiency of MurM 159 compared with MurM Pn16 bought about by sequence variation between these enzymes and, to a lesser extent, differences in the in vivo tRNA Ala :tRNA Ser ratio in 159 and Pn16. Kinetic analysis revealed that MurM 159 acted during the lipid-linked stages of peptidoglycan synthesis, that the D-alanyl-D-alanine of the stem peptide and the lipid II N-acetylglucosaminyl group were not essential for substrate recognition, that ⑀-carboxylation of the lysine of the stem peptide was not tolerated, and that lipid II-alanine was a substrate, suggesting an evolutionary link to staphylococcal homologues of MurM such as FemA. Kinetic analysis also revealed that MurM recognized the acceptor stem and/or the T⌿C loop stem of the tRNA Ala . It is anticipated that definition of the minimal structural features of MurM substrates will allow development of novel resistance inhibitors that will restore the efficacy of ␤-lactams for treatment of pneumococcal infection.
The peptidoglycan in Streptococcus pneumoniae and other Gram-positive pathogens is composed of a carbohydrate polymer consisting of alternating residues of N-acetylglucosamine and N-acetylmuramic acid. Appended to the N-acetylmuramic acid residue is the "stem peptide" composed of up to five amino acids, L-alanyl-␥-D-glutamyl-L-lysyl-D-alanyl-D-alanine. The stem peptides are themselves cross-linked between the ⑀-amino group of the lysine of a pentapeptide and the carbonyl group of the fourth position D-alanine of an adjacent stem (1). The pneumococcal stem peptide is further modified in S. pneumoniae where the lysyl residue ⑀-amino group is substituted by a dipeptide branch consisting of L-alanine or L-serine followed invariably by L-alanine (2)(3)(4).
The pneumococcal genes encoding the enzymes that construct the dipeptide branch, MurM and MurN, add the first and second amino acids to the stem peptide lysine, respectively (10,11). S. pneumoniae has acquired related MurM sequences, within its genome by homologous recombination to create a family of mosaics of related murM genes (12,13). This has endowed the resulting family of MurM variants with vastly differing levels of activity in vivo and differing amino acid specificities for incorporation of alanine and serine (2,12,14).
Although not absolutely essential for high level penicillin resistance in laboratory strains of S. pneumoniae, clinical strains of this organism depend upon the activity of MurM for high level penicillin resistance (14 -16). Despite the medical importance of this protein, knowledge of the enzyme biochemistry of MurM is sketchy and inferred. Unavailability of the peptidoglycan precursor substrates of MurM has previously restricted analysis of this protein to what can be deduced from molecular genetics, bioinformatics, and analysis of peptidoglycan precursor pools within S. pneumoniae. Recent successes in the in vitro synthesis of these precursors in our laboratory and elsewhere have, however, made the analysis of MurM enzymology a realistic proposition (17)(18)(19).
We have therefore carried out a characterization of the enzymatic properties of MurM from two clinical isolates of S. pneumoniae, one highly penicillin-resistant (159) and the other penicillin-sensitive (Pn16), that has allowed us in this paper to 1) confirm the type of enzymatic reaction carried out by MurM; 2) correlate the enzyme biochemistry of MurM with the final composition of the peptidoglycan of these two strains; 3) deduce the specificity of MurM for its peptidoglycan precursor substrates, allowing delineation of what is required for substrate binding by this enzyme; and 4) define those regions of the tRNA Ala substrate of MurM that are required for binding and catalysis.

Chemicals
RNA species corresponding to the sequence of the fulllength pneumococcal tRNA Ala UGC isoacceptor GGG GCC UUA GCU CAG CUG GGA GAG CGC CUG CUU UGC ACG CAG GAG GUC AGC GGU UCG AUC CCG CUA GGC UCC ACC A and corresponding to the 3Ј-aminoacylation site, acceptor stem, and TC loop and stem of the full-length pneumococcal tRNA Ala UGC , GGG GCC UAG CGG UUC GAU CCC GCU AGG CUC CAC CA (RNA minihelix), were synthesized, purified, and supplied with a 5Ј-phosphorylation by Dharmacon Inc. S. pneumoniae Pn16 MurE (MurE Pn16 ) and Pseudomonas   FIGURE 1. The peptidoglycan synthetic pathway. Intracellular cytoplasmic and cell membrane-bound steps are shown with the polymerization reactions that generate the nascent peptidoglycan on the extracellular face of the cell membrane. L-I and L-II denote lipid I and lipid II, respectively. The C-terminal D-alanine that is lost in transpeptidation is shown in yellow on a green background, and the remaining stem peptide amino acids are in black. The GlcNAc residues of the carbohydrate backbone of the peptidoglycan are violet; the serine or alanine added by MurM is in light green, and the alanine added by MurN is brown. The undecaprenyl-phosphate carrier that is cycled between the extracellular and intracellular faces of the cytoplasmic membrane is in red. For purposes of simplification, the figure is drafted assuming that MurM and MurN act exclusively after the formation of lipid II, where formally, they could utilize any intermediate after the MurE step.

Escherichia coli Strains and Plasmids
Details of E. coli strains and plasmids used in this study are indicated in supplemental Materials and Methods.

S. pneumoniae Strains and Isolation of Pneumococcal DNA
Pn16 (110K/70) serotype 42, was isolated in Papua, New Guinea and was penicillin-sensitive (minimum inhibitory concentration Ͻ0.016 g⅐ml Ϫ1 ) (23). 159 serotype 19A was isolated in Hungary (15) and was penicillin-resistant (minimum inhibitory concentration Ͼ16 g⅐ml Ϫ1 ). Strains were propagated on brain heart infusion agar containing 5% (v/v) sheep blood at 37°C in 5% (v/v) CO 2 or in liquid medium in brain heart infusion broth at 37°C in 5% (v/v) CO 2 . DNA was extracted from lawns of pneumococci on brain heart infusion blood agar as described (23).

Micrococcus flavus Membranes
Details of preparation of M. flavus membranes are recorded in supplemental Materials and Methods.

Peptidoglycan Analysis
Peptidoglycan was extracted from late exponential phase S. pneumoniae, purified, and digested with muramidase, and the resulting stem peptides were extracted and fractionated by reverse phase HPLC 2 on a Vydac 218TP54 column (4,24). Peptidoglycan fragment structural assignments were made according to Refs. 2-4, 11.

Protein Analytical Methods
SDS-PAGE, N-terminal protein sequencing, protein assays, and Western blotting for histidine tags were performed according to Ref. 22 and references therein.

Identification and Sequencing of the murM Alleles from Pn16 and 159
To sequence the Pn16 and 159 murM genes, primers were designed using the S. pneumoniae R6 genome sequence (25) at the J. Craig Ventner Institute (formally The Institute for Genome Research) web site to amplify the region between 232 nucleotides 5Ј to the initiator ATG (primer 1, supplemental Table 1) of the murM gene to 162 nucleotides 3Ј of the murM TAA stop codon (primer 2, supplemental Table 1) by PCR. DNA sequence between these primers was amplified by Pwo DNA polymerase according to the manufacturer's instructions. and a product of the correct size (1.6 kb) was purified using a Qiagen spin column and sequenced in both directions.

Cloning, Overexpression, and Purification of MurM from Pn16 and 159
To construct an expression vector carrying a murM allele with a 3Ј sequence encoding a hexahistidine (His 6 ) peptide, 1.3-kb fragments containing the murM allele from Pn16 and 159 were amplified by PCR from the appropriate pneumococcal DNA. Because of sequence divergence at the 5Ј end of the open reading frame between the murM alleles, a 5Ј primer for each gene was designed incorporating an NdeI restriction site for amplification of murM Pn16 and murM 159 , respectively (primers 3 and 4, supplemental Table 1). A single 3Ј primer for the amplification of both alleles was designed to incorporate a 3Ј XhoI site and eliminate the 3Ј stop codon (primer 5, supplemental Table 1). On PCR with PWO polymerase, products of the correct mass for murM Pn16 and murM 159 were obtained, purified, restricted with NdeI and XhoI, and ligated into similarly restricted pET21b as described in Ref. 22. Clones carrying the recombinant murM Pn16 and murM 159 genes were verified by sequencing, and one correct clone was retained for expression of each protein (pET21b::murM Pn16 and pET21b::murM 159 ).
Analysis of whole cells and subcellular fractions thereof by SDS-PAGE and Western blotting suggested that MurM Pn16 and MurM 159 were mostly insoluble but could be solubilized by 1 M NaCl. 3 To purify MurM Pn16 or MurM 159 , all steps were performed at Ͻ4°C. Cell pellets suspended/g in 3 ml of 50 mM HEPES, 1 mM MgCl 2 , 0.5 mM EGTA, 2 mM ␤-mercaptoethanol, 0.2 mM phenylmethanesulfonyl fluoride, 1 M leupeptin, 1 M pepstatin, pH 7.5, ϩ 2.5 mg ml Ϫ1 chicken egg white lysozyme (lysis buffer) were shaken for 30 min, disrupted by sonication, and centrifuged at 10,000 ϫ g for 30 min. The 10,000 ϫ g pellet was extracted in 50 mM sodium phosphate, 1 M NaCl, 0.5 mM EGTA, 2 mM ␤-mercaptoethanol, 1 M leupeptin, 1 M pepstatin, and 0.2 mM phenylmethanesulfonyl fluoride, pH 7.0, for 30 min and then centrifuged for 30 min at 100,000 ϫ g. The supernatant was retained, and the 100,000 ϫ g pellet was re-extracted as above, and the supernatants were combined. The supernatant was sequentially fractionated between 25 and 50% saturation ammonium sulfate and by gel exclusion chromatography on a 500-ml Sephacryl S-200 column in 50 mM NaH 2 PO 4 , 0.5 M NaCl, 0.2 mM phenylmethanesulfonyl fluoride, 1 M leupeptin, 1 M pepstatin, pH 7.0 (phosphate buffer). Fractions containing MurM by Western blot were further purified by immobilized metal affinity chromatography on a 25-ml column of cobalttalon resin (Clontech) in phosphate buffer. Once unbound proteins were eluted, MurM was eluted by a 0 -0.2 M imidazole gradient. The purity and identity of the final products of these purifications were assessed by SDS-PAGE, Western blotting, and N-terminal sequencing.

Cloning, Overexpression, and Purification of Alanyl-tRNA Ala Synthetase (AlaRS) and Seryl-tRNA Ser Synthetase (SerRS) from Pn16 and MurF from 159
Using the pneumococcal sequences in Ref. 26, the above genes were cloned and overexpressed, and their products were purified by immobilized metal affinity chromatography and anion exchange chromatography. Details of these procedures are given in supplemental Materials and Methods.

UDP-MurNAcAEKAA
Syntheses were conducted as for UDP-MurNAcAEK, except that the phosphoenolpyruvate concentration was 99.8 mM and the incubations also contained 15 mM D-alanyl-D-alanine and 9.09 M MurF 159 .
In all cases, the UDP-MurNAc peptide product was freed from protein by centrifugation through a M r 10,000 cutoff membrane, and the filtrate was fractionated on a 50-ml column of Source 30 Q anion exchange resin from which it was eluted using a 0 -1 M ammonium acetate gradient at pH 7.5. Fractions containing the UDP-MurNAcAEK product were identified enzymatically utilizing MurF 159 and by negative ion electrospray-mass spectrometry (ES-MS) for this and all other UDP-MurNAc peptides. All products were lyophilized three times versus water and stored in solution at Ϫ20°C.

Synthesis of Lipid-linked Peptidoglycan Precursors
All syntheses were conducted using Micrococcus flavus membranes essentially as described in Ref. 19. The lipid I or lipid II products were purified as described previously (19).

Analysis of Lipid-linked Peptidoglycan Precursors
To assay lipid-linked precursors, 50 l of lipid I or II species suspended in 50 mM HEPES, 10 mM MgCl 2 , 30 mM KCl, and 1.5% (w/v) CHAPS, pH 7.6, were added to 50 l of 1 M HCl. Samples were boiled for 30 min and neutralized with 2 M NaOH. The phosphate released was assayed according to Refs. 21, 27. Synthesis of lipid precursors was confirmed by TLC on silica and by negative ion ES-MS as in Ref. 19.
No conditions could be found for the amplification of 159 tRNA Ser (3) . 159 tRNA Ser(1) was amplified with platinum Pfx DNA polymerase; however, most unexpectedly, a clean 2-kb product was obtained, the 3Ј termini of which were tagged with ATP and TA cloned into a linearized pCR2.1 vector, according to the manufacturer's instructions (Invitrogen). The 2-kb insert was then sequenced using the vector-specific m13 primer sequences (supplemental Table 1) either side of the insert.

Total tRNA Preparation, Aminoacyl-tRNA Preparation, and tRNA Ala or tRNA Ser Determination
Techniques employed to isolate, preparatively acylate, and assay tRNA Ala and tRNA Ser are described in supplemental Materials and Methods.

Radiochemical
These assays were designed to follow the transfer of label from [ 3 H]alanyl-tRNA Ala and [ 3 H]seryl-tRNA Ser to the peptidoglycan precursor. Initial experiments examining the stability of the aminoacyl linkage to the tRNA suggested that at 37°C at the pH employed in the spectrophotometric method (7.6) the half-life of M. flavus [ 3 H]alanyl-tRNA Ala was 9.8 min. However, this could be extended to 46 min by dropping the pH of the assay to 6.8. 3 Therefore, to avoid interference by depletion of acyl-tRNA substrate in MurM assays, initial rate data were usually obtained at pH 6.8 within the first 10 min of reaction, where loss of alanyl-tRNA Ala through chemical deacylation was Ͻ5%. H]alanyl-tRNA Ala (800 -1000 cpm⅐pmol Ϫ1 unless stated otherwise) and were incubated at 37°C for times specified in the text (although initial rate data were usually taken from the first 2 min of reaction). Reactions were terminated at the appropriate time by the addition of 35 l of ice-cold 6 M pyridinium acetate, pH 4.5, and 70 l of ice-cold butan-1-ol. The incubations were rapidly mixed and centrifuged for 5 min at 1°C at 13,000 ϫ g, after which the butan-1-ol phase was washed with 70 l of water and then assayed for 3 H by liquid scintillation counting. To follow generation of 3 H-serylated lipid precursors, exactly the same procedure was followed, except that the 1 mM L-alanine and 0.45 M [ 3 H]alanyl-tRNA Ala were replaced with 1 mM L-serine and 0.45 M [ 3 H]seryl-tRNA Ser .

Kinetic Data Analysis
Nonlinear regression analyses of dependences of MurM initial velocity on substrate concentration were performed using GraphPad Prism 4 software.

Peptidoglycan Analysis Reveals the Relative Activity and Amino Acid Specificity of MurM in Vivo
To gain insight into the activity of MurM in vivo, we examined the peptide structure of muramidase digests of Pn16 and 159 peptidoglycan by reverse-phase HPLC (Fig. 2). This revealed that the major cross-linked species in Pn16 peptidoglycan is dimeric, composed of two adjacent peptides, without any dipeptide branch between them (peak 4). Unbranched single peptide species (monomers, peaks 1 and 2) also predominated in the first 15 min of the Pn16 chromatogram. Beyond 25 min, multimers of cross-linked peptidoglycan precursors (peaks 5-9) eluted (Fig. 2). These formed the minority of the Pn16 material fractionated and are composed of cross-linked peptide dimers (peaks 5-7) and trimers (peaks 8 and 9), which FIGURE 2. Fingerprints of stem peptides fractionated by reverse phase HPLC from muramidase digests of peptidoglycans from Pn16 and 159. Stem peptides were isolated from pneumococcal peptidoglycans as under "Materials and Methods" and fractionated by reverse phase HPLC with a C-18 column eluted by an 80-min gradient between 0 and 15% (v/v) acetonitrile in 0.1% (v/v) trifluoroacetic acid (4). Peptide elution was monitored at 215 nm, and peptides were identified by comparison with elution of known peptidoglycan fragments (2,4,11,12,14). Stem peptide sequences assigned to each species are in black type; amino acids added by MurM are in red type; amino acids added by MurN are in purple type. Labels 1-9 on the Pn16 HPLC trace and 1 and 3 and I-IX on the 159 HPLC trace correspond to stem peptides categorized in Refs. 2, 4, 11, 12, 14. are linked by dipeptide branches which mostly had serine attached to the lysine of the peptidoglycan peptide (Fig. 2).
This suggested that MurM activity in Pn16 is predominantly serine-specific. In contrast, after 20 min the majority of material eluted on chromatography of muramidase digests of 159 peptidoglycan was composed of monomers (peaks I-III), dimers (peaks (IV-VI), and trimers (peaks VII-IX), all of which were substituted by dipeptide branches (Fig. 2). It was clear that in vivo MurM activity was considerably greater in 159 than in Pn16. Furthermore, in vivo MurM 159 specificity was heavily biased in favor of addition of alanine onto the stem peptide in preference to serine, where the converse was true for MurM Pn16 (Fig. 2).

murM 159 and murM Pn16 Genes Display Marked Sequence Divergence That Underpins the Variation in Peptidoglycan Structure between 159 and Pn16
murM genes display mosaic sequences resulting from homologous recombination among natural populations of streptococci (12,13). To determine whether this might underlie the marked differences between the stem peptide branching in Pn16 and 159, the murM Pn16 and murM 159 alleles were sequenced. Both genes encoded proteins of 406 amino acids. Comparison of the inferred amino acid sequences of MurM Pn16 and MurM 159 in supplemental Fig. 1a revealed considerable amino acid sequence divergence (18%) between the two enzymes overall, with the N-terminal 61 amino acids, residues 115-149 and residues 229 -298, differing in sequence by as much as 31, 42, and 44%, respectively.
Clearly, the sequence divergence and differing genetic lineages of murM Pn16 and murM 159 could contribute to the very differing activities of MurM Pn16 and MurM 159 in vivo. MurM species with a threonine or lysine in position 260 insert predominantly serine or alanine (14). Consistent with this observation and the in vivo preferences of MurM Pn16 and MurM 159 , these proteins possessed a threonine and lysine, respectively, at position 260 (supplemental Fig. S1a).

Overexpression and Purification of MurM 159 and MurM Pn16
To characterize the enzymology of MurM 159 and MurM Pn16 and relate their in vitro behavior to their activity in vivo, we cloned, overexpressed, and purified MurM 159 and MurM Pn16 as their C-terminal (C-(His 6 )) fusions. Pilot experiments in E. coli C41(DE3)/pRIL demonstrated that MurM Pn16 overexpression was in large excess over that of MurM 159 as evidenced by SDS-PAGE and Western blot analysis of whole cells. Subcellular fractionation suggested that both MurM species were almost completely expressed as inclusion bodies. Attempts to obtain soluble MurM, including varying growth temperature and co-expression with chaperones, failed. 4 However, both MurM species could be solubulized by 1 M NaCl. Therefore, we developed a high ionic strength-tolerant purification protocol as follows: high salt solubilization, followed by ammonium sulfate precipitation, followed by gel filtration, followed by immobilized cobalt affinity chromatography to purify MurM.
The protocol yielded, per liter of culture, ϳ3 and 0.5 mg of MurM Pn16 and MurM 159 , respectively, at a purity in excess of 95% (supplemental Fig. S2a). Confirmation of the identity of the purified products involved Western blot identification of the C-(His) 6  The UDP-MurNAc peptides were converted to their lipid I or lipid II derivatives using the MraY and MurG activity associated with M. flavus membranes. Both UDP-MurNAc-AEKAA and UDP-MurNAcAE(DAP)AA were completely converted to their lipid II derivatives in the presence of UDP-GlcNAc as judged by TLC. Likewise, UDP-MurNAcAEKAA was completely converted to its lipid I derivative in the absence of UDP-GlcNAc. However, even under regimes employing extended incubation times and doubling the quantity of M. flavus membranes and UDP-GlcNAc, it proved impossible to completely convert UDP-MurNAc-AEK to its corresponding lipid II derivative as determined by TLC. Therefore, UDP-MurNAcAEK was only converted to its lipid I derivative.
After purification of the lipid I and II species as in Ref. 19

MurM Depends upon tRNA for the Transfer of Alanine or Serine to Lipid II to Generate Lipid II-Ala or Lipid II-Ser in Vitro
To determine whether MurM 159 depended on tRNA for the acylation of lipid II, we carried out a series of incubations designed to generate alanyl-tRNA Ala or seryl-tRNA Ser in situ to determine whether MurM 159 could then transfer the aminoacyl group to lipid II.
Transfer of Alanine-On TLC analysis of the butan-1-ol-soluble components of incubations involving alanyl-tRNA Ala , a product was observed (R f ϭ 0.27) that was absent if ATP, tRNA 159 , AlaRS, lipid II, MurM, or alanine was omitted (Fig.  3a). This product was purified by anion exchange chromatog-raphy (19) (Fig. 3b). Negative ion ES-MS analysis (Fig. 3c) showed the mass spectrum of the product was dominated by a doubly charged species with an (observed/expected for lipid II-Ala) m/z of 971.9925/972.0410, associated with triply and singly charged species with m/z values of 647.9958/647.6914 and 1945.6179/1945.0899, respectively, suggesting the prod-uct was lipid II-Ala. This was further confirmed by ES-MS fragmentation analysis of this species by collision-induced dissociation (supplemental Fig. S3).
Transfer of Serine-On TLC analysis of the butan-1-ol-soluble components from the incubations involving serine, tRNA Ser , SerRS, ATP, MurM 159 , and lipid II, no unique product was observed (Fig.  3d). Purification of the components of this incubation (19) (Fig. 3e) yielded a product that on negative ion ES-MS (Fig. 3f)

MurM Regenerates Unacylated tRNA on Acylation of Lipid II
To further confirm the coupling of SerRS activity to that of MurM 159 implied above, and to demonstrate that unacylated tRNA was the other product of MurM 159 , a continuous spectrophotometric method for monitoring this process was devised. Where SerRS was not limiting for MurM 159 activity, with catalytic concentrations of Ser-tRNA Ser , synthesis of lipid II-Ser from lipid II by MurM 159 generated equivalent amounts of tRNA Ser . This could be reacylated by SerRS, generating an equivalent amount of adenosine 5Ј-monophosphate, which could be coupled to yield a continuous spectrophotometric signal ("Materials and Methods," see "Spectrophotometric"; supplemental Fig. S4a). This assay system generated a SerRS-, tRNA Ser -, ATP-, serine-, lipid II-, and MurM 159 -dependent decrease in NADH absorbance (supplemental Fig. S4b). This confirmed that where MurM 159 accumulated lipid II-Ser (above), tRNA Ser was generated by this enzyme and was recycled with SerRS. The NADH oxidation kinetics suggested consumption of 83% of the lipid II over the duration of the assay. This and the kinetics of the process in supplemental Fig. S4b leads us to tentatively suggest that there is molar equivalence between lipid II-Ser and tRNA Ser generation.

The True Amino Acid Substrates of MurM Are Aminoacyl-tRNAs
To establish that the actual MurM tRNA substrate was an acyl-tRNA, using M. flavus total tRNA, we precharged [ 3 (Table 1).
These assays were conducted for 60 min. However, under initial rate conditions, accumulation of radiolabeled lipid-II-Ala product was linear for about 3 min, and within this time frame the relationship between the measured rate and protein concentration was linear.  (Table 2; p Ͻ 0.0005), which is entirely consistent with the preference of serine over alanine displayed by the MurM Pn16 in vivo. The intrinsic alanylation activity of MurM 159 was also 11.3-and 5.2-fold greater than the lipid II-alanylation (p Ͻ 0.0005) and serylation (p Ͻ 0.0005) activities of MurM Pn16 (Table 2). These data are consistent with the far higher levels of peptidoglycan branching in the 159 peptidoglycan compared with that of Pn16.

The Competence of tRNA Ala and tRNA Ser from 159 and Pn16 as MurM Substrates Does Not Influence Branch Composition of the 159 and Pn16 Peptidoglycan
MurM 159 supported lipid II alanylation to the greatest extent; however, it was not clear whether the tRNA pool in 159 was better suited to supporting MurM activity than the tRNA pool in Pn16. To address this issue, the sequencing of tRNA genes from 159 and Pn16 was undertaken to determine whether there were sequence differences in the tRNA genes that could explain the enhanced levels of MurM activity in 159.
It was only possible to sequence tRNA Ala(1) and tRNA Ala(4) from Pn16 and 159 (see "Materials and Methods") which showed that  tRNA Ala(1) and tRNA Ala(4) were identical in sequence and identical from either organism. All four tRNA Ser genes from Pn16 were sequenced and were found to be identical in sequence and arrangement to the tRNA Ser genes found in penicillin-sensitive pneumococci such as TIGR4 and R6 (25,26). In the case of 159, tRNA Ser(4) and tRNA Ser (2) genes were identical to those in Pn16. The gene encoding 159 tRNA Ser (1) , although identical in sequence to that of Pn16, had, as indicated by BLAST analysis (28), been subject to an IS1167 transposon insertion immediately 5Ј to its start increasing the size of the insert from 573 to 2043 bp. The remaining tRNA Ser ( (Table 2; p Ͻ 0.0005). This suggested that enhanced insertion of serine over alanine in Pn16 peptidoglycan branches was a function of the acyl-tRNA specificity of MurM Pn16 but not caused by a particularly competent seryl-tRNA Ser substrate of this MurM in Pn16.

Changes in the Ratio of tRNA Ala to tRNA Ser in 159 and Pn16 Could Influence the Stem Peptide Branch Composition of 159 and Pn16 Peptidoglycans
Although the competence of alanyl-tRNA Ala and seryl-tRNA Ser from 159 and Pn16 to perform as MurM 159 and MurM Pn16 substrates was equivalent in vitro, in vivo, perturbations in the tRNA Ala :tRNA Ser ratio possibly influenced by the transposon insertion 5Ј to 159 tRNA Ser GCU could affect the substrate availability for MurM and so could impact on the peptidoglycan composition of 159 and Pn16. Therefore, pneumococcal AlaRS and SerRS were used to assay the relative concentration of tRNA Ala and tRNA Ser in Pn16 and 159 total tRNA. This revealed that the 159 tRNA Ala pool (1.363 Ϯ 0.102 pmol of tRNA Ala /g of total tRNA; n ϭ 4 cultures) was 2.2-fold larger than the Pn16 tRNA Ala pool (0.5973 Ϯ 0.107 pmol of tRNA Ala /g of total tRNA; n ϭ 3 cultures; p Ͻ 0.001), whereas the tRNA Ser pool was the same size in both 159 and Pn16 (0.549 Ϯ 0.096 pmol of tRNA Ser /g of total tRNA (n ϭ 5 cultures) and 0.519 Ϯ 0.099 pmol of tRNA Ser /g of total tRNA (n ϭ 3 cultures) respectively; p Ͻ 0.3). The tRNA Ala :tRNA Ser ratio in 159 (2.3) relative to Pn16 (1.2) was raised 2-fold, which could contribute to the enhanced levels of stem peptide branch alanylation in 159 relative to that in Pn16.

Selection of [ 3 H]Acyl-tRNA Substrates for Kinetic Analysis of the Peptidoglycan Precursor Substrate Specificity of MurM
To establish the position of MurM within peptidoglycan synthesis and to determine what portions of the peptidoglycan precursor substrate were crucial for catalysis required a considerable quantity of acyl-tRNA.  Fig. 4, a,  i, and b, i).
The  (Fig. 4a, i, yielded identical values for the k cat app and K m app of MurM 159 for lipid II (Table 3).

Kinetic Analysis of MurM 159 Suggests That Stem Peptide Branches Are Added Exclusively during the Lipid-linked Stages of Cell Wall Synthesis
To determine whether there was any preference displayed by MurM 159 toward lipid I or lipid II, both peptidoglycan precursors were tested as MurM 159 substrates in the butan-1-ol extraction assay. MurM 159 displayed a hyperbolic dependence on lipid I (Fig. 4a, ii). The similarity in K m app and the modest diminution in the lipid I k cat app when compared with lipid II (

Characterization of the Response of MurM 159 to Peptidoglycan Precursors and Precursor Analogues
A Complete Five-amino Acid Stem Peptide Is Not Required for MurM 159 Activity-To determine the importance of the stem peptide D-alanyl-D-alanine to the binding of peptidoglycan precursors to MurM 159 , we tested lipid I-AEK as a MurM 159 substrate. MurM 159 was able to alanylate lipid I-AEK in a MurM 159 -dependent manner. The relationship between MurM 159 velocity and lipid I AEK was characterized by substrate inhibition at Ͼ40 M lipid I-AEK (Fig. 4a, iii). Analysis of the data suggested that it could be fitted by nonlinear regression to Equation 2 ([S] ϭ [lipid I-AEK] (29)), where K s is the binding constant for the lipid I-AEK in a catalytically unproductive mode. The similarity of the k cat app :K m app ratio for both lipid I-AEK and lipid I, 0.46 and 0.21 s Ϫ1 ⅐mM Ϫ1 , respectively ( Table 3    boxyl would have on the activity of MurM 159 , we synthesized lipid II(DAP), where DAP is a derivative of lysine that has been carboxylated on the ⑀-carbon atom to which the side chain ⑀-amino group is attached. Attempts to use lipid II(DAP) as a substrate for MurM 159 failed completely ( Table 3), suggesting that the active site of MurM 159 could not tolerate a negative charge proximal to the lysine ⑀-amino group that undergoes alanylation by MurM 159 .
MurM 159 Can Alanylate Its Product, Lipid II-Ala-Lipid II-Ala is the product of MurM 159 . We therefore initially tested lipid II-Ala for MurM 159 product inhibition. However, to our surprise, MurM 159 transferred [ 3 H]alanine from M. flavus [ 3 H]alanyl-tRNA Ala to lipid II-Ala. The relationship between V o and lipid II-Ala was hyperbolic (Fig. 4a, iv) and was fitted by nonlinear regression to Equation 1. Comparing k cat app /K m app for lipid II-Ala and lipid II (Table 3) suggested that the extra alanylation of the lipid II substrate reduced the catalytic efficiency of MurM 159 72-fold relative to lipid II.

Characterization of the Dependence of MurM 159 on [ 3 H]Alanyl-tRNA Ala
tRNA Secondary Modifications Are Not Required for Recognition by MurM-To study the interactions of the alanyl-tRNA Ala with MurM 159 , we used pure pneumococcal tRNA Ala (UGC isoacceptor; tRNA Ala UGC ), synthesized with a 5Ј-phosphate (Fig. 5a, ii) but otherwise devoid of any secondary modifications to its bases. Analysis of the data in Tables 2 and 4 (Fig. 5b, ii) (Table 4). Both relationships were hyperbolic (Fig. 5, a, i, and b, i). However, MurM 159 concentrations in these assays were significant compared with those of the alanyl-RNAs (Table 4).
To obtain the kinetic constants that characterized the dependence of MurM 159 on its alanylated RNA substrates, the data were fitted by nonlinear regression to Equation 3 below (29,30), which accounts for [MurM 159 ] ([E]).
The  (Table 4) were identical (0.101 and 0.099 s Ϫ1 ⅐M Ϫ1 , respectively), indicating that both RNA substrates were catalytically equivalent. Thus, MurM 159 :tRNA substrate recognition was likely to be limited to the alanyl-acceptor stem, the T⌿C loop, and the T⌿C stem and did not involve the anticodon, D, or variable loops and stems.

DISCUSSION
Peptide bond formation is crucial to the synthesis of the peptidoglycan, which involves three of the four commonly used

Kinetics of utilization of peptidoglycan precursor and peptidoglycan precursor derivatives by MurM 159
All assays were performed for 1 min at 0.45 M ͓ 3 H͔alanyl-tRNA Ala . All assays with lipid derivatives followed the butan-1-ol extraction method (see under "Materials and Methods," "Transfer of 3 H-Amino-acid between ͓ 3 H͔Acyl-tRNA and Lipid-linked Peptidoglycan Precursors"), all assays with UDP-MurNAc substrates used the trichloroacetic acid precipitation procedure (see under "Materials and Methods," "Transfer of 3 H-Amino-acid between ͓ 3 H͔Acyl-tRNA and UDP-linked Peptidoglycan Precursors"). Lipid I-AEK kinetic constants were determined according to Equation 2; all other kinetic constants were determined according to Equation 1 (see text). Fitting data to both equations was performed by nonlinear regression. NA means not applicable. biochemical mechanisms of amino acid carboxyl activation. The stem peptide is constructed from amino acids that are activated by esterification to phosphate (31); the stem peptides are cross-linked after activation by esterification to a highly reactive serine within a penicillin-binding protein transpeptidase (32), or inter-stem peptide branches are built up by condensa-tion of L-amino acids activated by esterification to their cognate tRNA (5)(6)(7)(8)(9)(33)(34)(35)(36). It is with this latter class of reactions that this paper is concerned. S. pneumoniae, probably because of the plasticity of its genome is unique because of the strain-dependent and highly non-uniform chemical nature and abundance of dipeptide in the text. a, ii, secondary structure of the pneumococcal tRNA Ala UGC isoacceptor. Sequence was determined from the gene sequence determined under "Results" and was synthesized as an oligonucleotide for use as a MurM substrate. The anticodon and all stems and loops are labeled, and the G3:U70 base pair in the acceptor stem that is crucial for recognition by AlaRS is highlighted in white against gray (*). a, iii, tertiary structure of the pneumococcal tRNA Ala UGC isoacceptor. A model of tRNA Ala based on that of tRNA Phe is shown, with the major structural elements labeled. b, i-iii, alanyl-minihelix.  Equation 3 in the text. b, ii, secondary structure of the minihelix derivative of the pneumococcal tRNA Ala UGC isoacceptor. Sequence of the acceptor stem and TC loop included the contiguous sequence of nucleotides 1-7 and 49 -76 of the full-length tRNA Ala UGC . The G3:U70 base pair in the acceptor stem crucial for recognition by AlaRS is highlighted in white against gray (*). b, iii, tertiary structure of the minihelix derivative of the pneumococcal tRNA Ala UGC isoacceptor. A model of the tertiary structure of the portion of tRNA Ala recapitulated by the minihelix based on the structure of tRNA Phe is shown, with the major structural elements labeled accordingly.

TABLE 4 Comparison of the substrate dependencies of MurM Pn16 and MurM 159
All assays were performed for 1 min. All assays followed the butan-1-ol extraction method. Lipid II kinetic constants were determined according to Equation 1, and all other kinetic constants were determined according to Equation 3 to account for the relatively high ͓MurM͔:͓͓ 3 H͔alanyl-(t)RNA substrate͔ ratio (see text). Fitting data to both equations was performed by nonlinear regression. branches in its peptidoglycan (12,14), as exemplified here with strains Pn16 and 159. Sequence similarity to S. aureus fmhb (femX) led Filipe and Tomasz (10) to insertionally disrupt the pneumococcal murM sequence. The resulting peptidoglycan phenotype identified the murM gene as being responsible for the addition of the first amino acid of the pneumococcal stem peptide branch (10). We therefore overexpressed and purified MurM Pn16 and MurM 159 to correlate their enzymological properties with their contribution to the structure of pneumococcal peptidoglycan for the first time. The purification of both enzymes was complicated because their solubility depended upon maintenance of a high ionic strength. This feature is also shared with the MurM homologue Enterococcus faecalis BppA1 (33).

MurM
A further hurdle in the study of MurM was the generation of peptidoglycan precursors. Although the synthesis of these molecules in vitro had been described previously, NADPH oxidase activity of MurB, coupled with its sensitivity to substrate inhibition by NADPH, required use of very high concentrations of NADPH and MurB for synthesis of UDP-MurNAc peptides by these methods (17,18,(37)(38)(39)(40). Here, however, by recycling the NADP ϩ co-factor with isocitrate dehydrogenase, we maintained sub-inhibitory NADPH concentrations that were ϳ1% of the final yield of UDP-MurNAc peptide. Similarly, by recycling ADP produced by MurC to -F (Fig. 1), with pyruvate kinase to reform ATP, UDP-MurNAc peptide purification was simplified because, although ADP co-purified with many UDP-MurNAc peptides, ATP eluted after all of them. 3 Our results conclusively showed for the first time that MurM 159 and, by inference, MurM Pn16 supported dipeptide branch synthesis by transferring either alanyl or seryl residues from alanyl-tRNA Ala or seryl-tRNA Ser to the ⑀-amino group of the lipid II stem peptide lysine (Figs. 3, S3, and S4). This is consistent with inter-stem branch synthesis in other Grampositive organisms where L-amino acids or glycine are activated by esterification to tRNA prior to insertion into the stem peptide (5)(6)(7)(8)(9)(33)(34)(35)(36). The work here is an advance on previous analyses of the MurM homologues E. faecalis BppA1 and S. aureus FemX, -A, or -B. In these cases, no demonstration of the transfer of aminoacyl groups from a pre-acylated tRNA to a peptidoglycan precursor was achieved, and so no kinetic analyses of acyl-tRNA usage could be performed (8,33), as was done here. Furthermore, the BppA1 results were based upon utilization of UDP-MurNAc pentapeptide as the stem peptide substrate, where, as the authors of this work concede, it is more likely that the true BppA1 substrate was lipid I or lipid II (33). Unlike the studies of the E. faecalis and Staphylococcus aureus enzymes, our reconstruction of peptidoglycan intermediate synthesis in vitro allowed us to probe in detail the interaction of MurM with its lipid substrates ( Fig. 4; Table 3) (8,33). Finally, the ES-MS fragmentation results presented here are the first direct confirmation that the site of aminoacylation of a lipid peptidoglycan precursor by this class of enzymes is the ⑀-amino group of the lipid II lysine (Fig. S3). This contrasts with previous studies of the S. aureus FemXAB (8), where the site of glycylation of lipid II was not identified.
The markedly different branching phenotype displayed by Pn16 and 159, corresponding to the in vivo activity of MurM ( Fig. 2), was reflected in vitro by MurM 159 activity with pneumococcal alanyl-tRNA Ala , which was considerably greater than with seryl-tRNA Ser and greater than any acyl-tRNA-supported MurM Pn16 activity (Table 2). This was consistent with the almost entire alanylation of stem peptides in 159 (Fig. 2). For MurM Pn16 , seryl-tRNA Ser was a better substrate than alanyl-tRNA Ala , consistent with the residual amounts of serylation and alanylation (where the serylation was greater than alanylation) of stem peptides in Pn16 ( Fig. 2; Table 2). The markedly differing properties of MurM Pn16 and MurM 159 were also reflected in their kinetics of substrate dependence, where consistent with their in vivo activity, both lipid II and alanyl-tRNA Ala UGC were used considerably less efficiently by the Pn16 enzyme compared with its 159 counterpart ( Table 4).
Involvement of both MurM and pneumococcal tRNA in the synthesis of dipeptide branches suggested that the tRNA or the MurM protein had been modified to reflect the differing nature of the peptidoglycan in Pn16 and 159. Analysis of pneumococcal genomes coupled with our own sequencing of tRNA genes of 159 and Pn16 failed to reveal any sequence differences that could account for the elevated levels of peptidoglycan branching in 159. Furthermore, equal amounts of [ 3 H]seryl-tRNA Ser and [ 3 H]alanyl-tRNA Ala from either 159 or Pn16 were equally competent at supporting MurM 159 and MurM Pn16 activity, suggesting that tRNA Ala or tRNA Ser from 159 was as capable of supporting enhanced peptidoglycan branching as tRNA Ala or tRNA Ser from Pn16.
Nevertheless, one of the 159 tRNA Ser GCU genes was directly adjacent to the 3Ј end of an IS1167 transposon insertion that was absent in Pn16. tRNA gene rearrangement by transposon insertion is well known, and it is conceivable that this interfered with transcription of this 159 tRNA Ser GCU gene, elevating the tRNA Ala :tRNA Ser ratio in 159 relative to Pn16 (41,42). This suggests that alterations in tRNA expression as opposed to the tRNA itself may contribute to the high levels of alanylation of peptidoglycan of 159 relative to those in the Pn16 peptidoglycan.
Homologous recombination of streptococcal genes generates a myriad of gene products of variable sequence that has allowed the pneumococcus and other streptococci to develop resistance to major classes of antibiotic as evidenced by sequence differences that have generated ␤-lactam-resistant penicillin-binding proteins and sulfonamide-resistant dihydropteroate synthase variants (43,44). This type of genetic rearrangement has led to a series of families of murM alleles, of which murM Pn16 is 99% identical to the murMA class of murM alleles typified by S. pneumoniae R6 murM and of which murM 159 is 100% identical to the murMB1 allele typified by Hungarian isolate Hun663 (supplemental Fig. S1b) (12,25). Such strain-dependent sequence variation of MurM is believed to underpin variations in MurM activity and peptidoglycan composition (12,14) similar to those observed here.
What does our data reveal about tRNA recognition by MurM? MurM Pn16 and MurM 159 can utilize alanyl-tRNA Ala and seryl-tRNA Ser (Table 2). Therefore, anticodon recognition is not important, which is also so for AlaRS and SerRS (45,46). This correlates with the [ 3 H]alanyl-minihelix data, which suggested that no more than the acceptor stem and the T⌿C loop of tRNA Ala were required for interaction with MurM 159 (Table  4; Fig. 5b, i-iii). Similar conclusions have been drawn regarding the Lactobacillus FemX (47).
AlaRS recognition of tRNA Ala depends on the highly conserved mispaired G:U at position 3:70 of the acceptor stem of the tRNA (Fig. 5a, ii) (45). Although it is possible that this recognition element is targeted by pneumococcal MurM, very recent work by Villet et al. (48) showed that Lactobacillus FemX does not depend upon the G3:U70 base pair for alanyl-tRNA Ala UGC substrate recognition, but instead it depends principally upon the first 2 bp in the acceptor stem of tRNA Ala UGC (G2:C71 and G1:C72; Fig. 5a, ii). All of the S. pneumoniae tRNA Ser and tRNA Ala UGC species have acceptor stems whose first 2 bp are G2:C71 and G1:C72 (25, 26 and data reported herein). Additionally, this acceptor stem motif is shared by pneumococcal tRNA isoacceptors for (amino acid/anticodon) Y/GUA, R/CCG, I/GAU, F/GAA, M/CAU, D/GUC, and V/UAC (25,26,49). Thus, if MurM shares the Lactobacillus FemX mode of tRNA recognition (48), then discrimination between seryl and alanyl addition and limitation of MurM activity to just alanylation and serylation of the stem peptide largely results from recognition of the aminoacyl moiety of the acyl-tRNA substrate by this enzyme.
Comparison of the rates of MurM 159 supported by crude 159 [ 3 H]alanyl-tRNA Ala and synthetic [ 3 H]alanyl-tRNA Ala UGC suggested that the latter was an 8-fold better substrate. This result highlighted the redundancy of tRNA secondary modifications in MurM catalysis. Additionally, these data suggested that the crude pneumococcal tRNA pool contains components that inhibit utilization of alanyl-tRNA Ala by MurM 159 . These components could well be unacylated tRNA species with the G2:C71 and G1:C72 motif (above), including the tRNA Ser isoacceptors which implies a subtlety in the interaction of MurM with the pneumococcal tRNA pool, which might impact upon the final composition of the cell wall.
What do the MurM alanyl-tRNA UGC kinetics reveal about the relationship between peptidoglycan and protein synthesis? K m app values of MurM 159 and MurM Pn16 for [ 3 H]alanyl-tRNA Ala UGC were considerably higher (Table 4) than the 6.2 nM K D of EF-TU for alanyl-tRNA Ala (50). Thus there is potential for considerable competition between ribosomal protein and cell wall synthesis for alanyl-tRNA Ala particularly where amino acids and therefore aminoacyl-tRNA species are in short supply and where, necessarily, alanyl-tRNA Ala will be drawn into protein synthesis. In the staphylococci, this issue is resolved by mutations in the T⌿C and D-loops of two specialized UCC isoacceptor tRNA Gly species that bar involvement of these tRNA Gly UCC species in protein synthesis, thus reserving a pool of glycyl-tRNA Gly for peptidoglycan synthesis (51)(52)(53)(54). In the case of the lactobacilli, maintenance of cell wall branching despite the demands of protein synthesis for alanyl-tRNA Ala probably depends upon the high k cat of FemX (47) relative to that of MurM.
S. pneumoniae is distinguishable from Lactobacillus viridescens and S. aureus because it is not essential for pneumococcal peptidoglycan to be entirely composed of branched stem peptides, nor even to contain any at all (2,4,10,55). It is therefore likely that the pneumococcus is more tolerant to loss of acyl-tRNA from peptidoglycan to protein synthesis than L. viride-scens and S. aureus. Nevertheless, it should be noted that such diminution of pneumococcal stem peptide branching may well adversely impact upon the attachment of important cell surface protein adhesins and virulence determinants ligated to branched stem peptides via sortases prior to peptidoglycan cross-linking (56).
What can be concluded from the in vitro peptidoglycan intermediate substrate specificity of MurM? MurM probably utilizes lipid I and/or II in vivo as suggested by Filipe et al. (14). This is inconsistent with exclusive specificity of S. aureus FemX for lipid II (8). Usage of UDP-MurNAcAEKAA by MurM 159 was insignificant (supplemental Fig. S5; Table 3) in comparison with that of L. viridescens FemX, which utilizes only this substrate (5).
The presentation of the preferred peptidoglycan precursor substrates to MurM and L. viridescens FemX is likely to be radically different. In the case of MurM, its lipid II substrate is tethered to a phospholipid bilayer by an undecaprenyl tail, which is therefore unlikely to interact with the enzyme, whereas the rest of the molecule, including the stem peptide (pyrophosphoryl-(GlcNAc)-MurNAcAEKAA), extends from the membrane surface (57) into the active site of MurM. In contrast, the stem peptide of the soluble peptidoglycan substrate of the L. viridescens FemX is tethered to UDP, which is available for and is indeed essential for interaction with this enzyme (58,59). The vastly differing properties and availabilities of the groups appended to the MurNAc-stem peptide of UDP-MurNAc-AEKAA and lipid I or II probably have precluded the evolution of a FemX or MurM active site that can bind both substrates and might dictate whether peptidoglycan stem peptide branching is a cytoplasmic or a membrane-bound process.
Usage by MurM of either lipid I or lipid II in vivo suggests the N-acetylglucosaminyl group of lipid II was not essential for recognition of the lipid precursor by MurM ( Fig. 4; Table 3). It also indicated that in vivo MurG probably tolerates lipid I species with an acylation of the stem peptide lysine ⑀-NH 2 group to avoid wasteful accumulation of lipid I precursors. This is likely because M. flavus MurG can even process lipid I modified by the addition of bulky pyrene-based fluorophores to the stem peptide lysine (19).
To extend our substrate specificity studies, we examined the consequences of carboxylation of the ⑀-carbon atom of the stem peptide lysine on MurM 159 catalysis. MurM 159 was entirely intolerant of this modification as judged by the observation that lipid II(DAP) was not a substrate. This has also been observed for S. aureus FemA and L. viridescens FemX (8,9). The lysine-containing stem peptide-specific members of the FemXAB family contain a catalytic aspartate (47), which is similarly conserved in MurM (Asp-107; Fig. S1a). It is possible that a substrate (DAP) carboxyl adjacent to the ⑀-NH 2 undergoing acylation, would perturb the environment of the catalytic carboxyl of Asp-107 of MurM sufficiently to prevent catalysis. In this context, it is interesting to note that Streptomyces coelicolor FemX and VanK, which append glycine onto an LL-DAP-containing stem peptide, do not possess conserved active site aspartates analogous to L. viridescens FemX Asp-109 or S. pneumoniae MurM Asp-107 (60,61).
Our results indicated that the stem peptide C-terminal D-alanyl-D-alanine was not essential for MurM 159 catalysis (Fig. 4a, iii). These results contrasted starkly with those of Maillard et al. (58) who found a stem peptide D-alanyl-D-alanine C terminus was required in L. viridescens FemX substrates. In this context, the lipid I-AEK substrate inhibition of MurM 159 might originate in a set of interactions such as those evident in the L. viridescens FemX:UDP-MurNAcAEKAA crystal structure (58), which is not a bound conformation of the substrate that can be catalytically active.
Surprisingly, we found that MurM 159 could append an alanyl group from [ 3 H]alanyl-tRNA Ala onto lipid II-Ala (Fig. 4a, iv). Although this was not evident from the ES-MS analysis of the products of MurM (Figs. 3 and supplemental Fig. S3), the kinetics of alanylation of lipid II-Ala were so poor in the absence of lipid II (Fig. 4a, iv; Table 3) than in its presence, it is unlikely that lipid II-Ala-Ala formation would have been detected by the experiments represented by supplemental Fig. S3 and Fig. 3.
The alanylation of lipid II-Ala by MurM is the activity of MurN in vivo (11). This showed that MurM is an enzymatic homologue of S. aureus FemA (8). A similar flexibility in the active site of S. aureus FemX also allows it under forcing conditions to add two glycine residues instead of one (8). Although clearly of mechanistic and evolutionary relevance, in vivo, mutational inactivation of MurN caused the branches of the stem peptides to be substituted by a single alanine or serine in S. pneumoniae, and consequently, the alanylation by MurM 159 of lipid II-Ala cannot be physiologically significant (11).
What relevance have our results for pneumococcal penicillin resistance? Clinically, MurM is essential for high level penicillin resistance (14 -16). Our data demonstrate the central role played by tRNA in streptococcal cell wall synthesis, and link tRNA metabolism with ␤-lactam resistance. Interestingly, competition of peptidoglycan branching and ribosomal protein synthesis for acyl tRNA might influence the response of pneumococcal infection to ␤-lactam therapy, which in turn might be modified by ribosomally directed antibiotics such as erythromycin and chloramphenicol. Our results have revealed many of the important features of the substrates of MurM involved in binding and catalysis. Pursuit of the structural basis of these interactions and methods for their disruption to furnish novel antibiotic therapies is currently underway in our laboratory.