Kinetic and Mechanistic Characterization of Recombinant Lactobacillus viridescens FemX (UDP- N -acetylmuramoyl Pentapeptide-lysine N 6 -Alanyltransferase)*

The FemABX family encodes enzymes that incorporate L -amino acids into the interchain peptide bridge of Gram-positive cell wall peptidoglycan and are novel nonribosomal peptidyl transferases that use aminoacyl-tRNA as the amino acid donor. We previously reported the identification of the femX gene from Lactobacillus viridescens and recombinant expression of active FemX (LvFemX) that catalyzes the transfer of L -Ala from Ala- tRNA Ala to the (cid:1) -amino group of L -lysine of UDP-MurNAc pentapeptide (Hegde, S. S., and Shrader, T. E. (2001) J. Biol. Chem. 276, 6998–7003). Recombinant LvFemX exhibits K m values of 42 and 15 (cid:2) M for UDP-MurNAc pentapeptide and Escherichia coli Ala-tRNA Ala , respectively, and exhibited a k cat value of 660 min (cid:3) 1 . Initial velocity and inhibition kinetic studies support an ordered sequential mechanism for the enzyme, and we propose that catalysis proceeds via a ternary complex. The pH dependence of the activity was bell-shaped, de-pending on the ionization state of two groups exhibiting apparent p K a values of 5.5 and 9.3. Chemical modifica- tion of the enzyme and the kinetics of inactivation, and protection by substrate, indicated the involvement of carboxyl

The ⑀-NH 2 group of diamino acids such as meso-diaminopimelic acid, L-lysine, or L-ornithine of the stem peptide of peptidoglycan is cross-linked to D-Ala 4 of the neighboring peptide either directly or via interchain peptide bridge (Fig. 1). The interchain peptide bridges are synthesized only in a subset of Gram-positive bacteria that include many important human pathogens and in a very few Gram-negative species, including pathogenic spirochetes but excluding non-pathogenic species of the flora (1). The enzymes that incorporate L-amino acids into the interchain cross bridge; unlike the mur-encoded enzymes in the peptidoglycan biosynthetic pathway, are novel nonribosomal peptidyl transferases and use aminoacyl-tRNA as the amino acid donor (2)(3)(4). The interchain peptide is specific for a given species and varies both in sequence and chain length from species to species (Fig. 1). In Staphylococcus aureus, the femX (now renamed as fmhB), femA, and femB genes were shown to be responsible for the incorporation of glycine 1 , gly-cine 2-3 , and glycine [4][5] , respectively, of the pentaglycine interchain peptide (5)(6)(7)(8). Similarly, femAB homologs of Streptococcus pneumoniae designated as murM and murN were reported to be responsible for the addition of L-Ala/L-Ser to L-Lys and subsequent incorporation of a second L-Ala residue, respectively (9 -11). Though the femABX gene products (including murMN and femAB homologs from other species) have been implicated in interchain peptide formation, no catalytic activity has been demonstrated for these proteins. In S. aureus and other bacteria, FemX purifies as a large, cell wall-associated multienzyme-lipid complex (3,4) and neither recombinant FmhB nor Streptococcus pyogenes FemAB homologs showed any catalytic activity using UDP-MurNAc pentapeptide (UDP-MPP) 1 as substrate (8,12), suggesting that these enzymes act after conjugation of UDP-MPP substrate to the undecaprenylphosphate lipid carrier (3). In contrast, Lactobacillus viridescens (also known as Weissella viridescens) has been shown to produce a soluble FemX (EC 2.3.2.10) that catalyzes the transfer of L-Ala from alanyl-tRNA to UDP-MPP before its lipid conjugation (2). We have purified L. viridescens FemX (LvFemX), identified the femX gene using reverse genetics and demonstrated the catalytic activity for the recombinant LvFemX (Ref. 12 and Scheme 1). Recently, two recombinant enzymes, namely BppA1 and BppA2 (equivalent to the S. aureus FemX and FemA) from Enterococcus faecalis have been shown to specifically transfer L-Ala from alanyl-tRNA to the first and second positions of the side chain, respectively (13,14). However, no kinetic or mechanistic studies on this important family of enzymes have been reported so far. In the present study, we report the substrate specificity, kinetic parameters of various tRNA and UDP-MPP substrates, kinetic mechanism, solvent kinetic isotope effects, and the involvement of Asp 109 in the catalytic function of LvFemX.

Expression and Purification of the Proteins
Wild-type and mutant LvFemX were expressed and purified as described previously (12). Protein concentrations were estimated by the Bio-Rad protein assay method using bovine serum albumin as a standard.

Isolation and Purification of UDP-MurNAc Pentapeptides
UDP-MPP Lys and UDP-MPP Dap from S. aureus and Bacillus subtilis, respectively, were isolated and purified to homogeneity as described previously (12), with an additional round of Mono-Q and Biogel-P2 chromatography (FPLC, Amersham Biosciences). The purity of the products was determined by electrospray mass spectrometry and the concentration determined spectrophotometrically using a molar absorption coefficient of 10 4 M Ϫ1 cm Ϫ1 . UDP-MPP Orn was isolated and purified from Lactobacillus cellobiosus (ATCC 11739) as described above. L. cellobiosus was grown in MRS medium (Difco) at 37°C to mid-log phase (OD 600 ϭ 1.2), vancomycin was added to a concentration of 40 g/ml, and the flasks were incubated for an additional 3 h. The UDP-MPP Lys -Ala (product of the FemX) was prepared enzymatically and purified as described above.

In Vitro Transcription of E. coli Ser-tRNA, Ala-tRNA, and Microhelix Ala
All tRNAs used in the study were synthesized in vitro using T7-RNA polymerase. The synthesis of various tRNAs from encoding plasmids using deoxyoligonucleotide templates were performed as described previously (15)(16)(17) with some modifications. Briefly, plasmid-encoded tRNA genes (100 g of plasmid) were digested with BstN1, the restriction enzyme removed by phenol/chloroform extraction and the linearized plasmid recovered by ethanol precipitation. The transcription reactions of 1 ml included: 40 mM Tris (pH 8.2), 25 mM NaCl, 5 mM DTT, 2 mM spermidine, 7 mM each NTP, 250 units of RNase inhibitor, 15 mM MgCl 2 , 100 g of bovine serum albumin, 2 units of inorganic pyrophosphatase, 100 g of purified T7-RNA polymerase and 100 g of linearized plasmid template, and the reactions were incubated for 10 -12 h at 37°C. Proteins from the reaction mixture were removed by phenol/ chloroform extraction followed by ethanol precipitation of the nucleic acids. The precipitate was dissolved in DNase digestion buffer (Promega), and the plasmid template was digested with 20 units of RNasefree DNase. The DNase was removed by phenol/chloroform extraction and the tRNA transcripts were precipitated twice with 2 volumes of ethanol, re-dissolved in 10 mM Tris (pH 7.5) containing 5 mM DTT and 200 units of RNase inhibitor. The transcripts were analyzed by electrophoresis on a denaturing acrylamide gel (12%) in Tris borate/EDTA buffer containing urea. For microhelix Ala (17) two deoxyoligonucleotide templates: micro-p, 5Ј-AATTGCTGCAGTAATACGACTCACTATAG-GGGCTATAGCTCTAGCTCCACC-3Ј (T7 promoter sequence is underlined) and micro-t, 5Ј-TGGTGGAGCTAGAGCTATAGCCCCTAT-AGTGAGTCGTATTACTGCAGCAATT-3Ј (transcribing sequence is underlined) were used instead of plasmid template. Forty micrograms of each of the oligonucleotides annealed by heating to 80°C and cooling to room temperature before adding to the transcription reaction as described above. The final transcripts were also heated to 80°C and cooled to room temperature before storage. The typical yield of tRNA transcript from the above reactions was 8 -9 mg.

Aminoacylation of tRNA
For routine assays and kinetic studies, aminoacylation of tRNA reactions were coupled to the FemX reaction (see below). Pre-charging of tRNA was performed as described previously (18) with minor modifications. Briefly, 500-l reaction mixtures contained 100 M Ala-tRNA, 5 mM ATP, 10 mM MgCl 2 , 150 M [ 14 C]Ala, 5 mM DTT, 100 units of RNase inhibitor, 1 unit inorganic pyrophosphatase, 2-3 M AlaRS-459-6H in 50 mM Tris (pH 7.5) was incubated at room temperature for 1 h. Charged tRNA was recovered and stored at Ϫ80°C as described (18). Aminoacylation of Ser-tRNA was performed using commercially available E. coli aminoacyl-tRNA synthetase.

Measurement of FemX Activity
Reaction rates were measured by monitoring the incorporation of radioactivity [ 14 C]Ala into purified UDP-MPP (2,12). The 40-l assay mixture in 50 mM Tris (pH 7.5) contained 1 mM ATP, 60 -80 M [ 14 C]Ala, 10 mM MgCl 2 , 2-3 M AlaRS (19) or 500 units of E. coli aminoacyl-tRNA synthetase (Sigma) in the case of Ser-tRNA), 0.1 unit of inorganic pyrophosphatase, UDP-MPP, and tRNA. The reaction was initiated by the addition of 2 l of appropriately diluted FemX (40 -100 ng), incubated for 1-2 min at 25°C, and stopped by rapidly transferring the tubes to boiling water bath for 2 min. 2 l of the quenched reaction mixtures were spotted onto precoated cellulose TLC plates, and the incorporation of [ 14 C]Ala into product was quantitated as described (Ref. 12 and Scheme 1).
Initial velocity kinetic data were fitted using Sigmaplot 6.0. Equation 1 was used to fit simple substrate saturation kinetics, where the concentration of one substrate was varied at a fixed, saturating (ϳ5ϫ K m ) concentration of the other substrate. Equation 2 was used to fit initial velocity patterns, where concentrations of both substrates were varied.
In Equations 1 and 2, v is the measured reaction velocity, V is the FIG. 1. Primary structure of cross-linked peptidoglycan. A, primary structure of a fragment of S. aureus peptidoglycan. S. aureus peptidoglycan contains L-Lys in the stem peptide (diamino acid) and a pentaglycine interchain peptide. S. aureus FemX transfers Gly 1 from glycyl-tRNA to the ⑀-NH 2 group of Lys followed by the addition of Gly 2-3 and Gly 4-5 by FemA and FemB, respectively. The interchain peptide formation proceeds in the direction of C termini to N termini and the ␣-NH 2 group of the terminal Gly 5 is cross-linked to the penultimate D-Ala residue (D-Ala 4 ) of a neighboring pentapeptide releasing the terminal D-Ala residue (curved arrows). The cross-linking step is a transpeptidase (TP) reaction, which is inhibited by both ␤-lactams and vancomycin (V, vancomycin binds to the terminal D-Ala-D-Ala). B, sequence of interchain peptide bridges in different species. L-amino acids of the interchain peptide are added to the ⑀-NH 2 group of the diamino acid (either L-Lys or L-Orn) of the stem peptide by Fem ABX family enzymes. In those organisms that lack an interchain peptide, the diamino acid (m-Dap 3 ) of the stem peptide is directly cross-linked to the D-Ala 4 of a neighboring peptide. SCHEME 1. A, the reaction catalyzed by LvFemX. B, FemX assay; shown is an inverted image of autoradiogram (developed for 20 h) of FemX reaction products separated on TLC. Lane 1, reaction mixture lacking FemX; lane 2, with added FemX. maximal velocity, A and B are the concentrations of substrates (aminoacyl-tRNA and UDP-MPP), K a , and K b are the corresponding Michaelis-Menten constants, and K ia is the inhibition constant for substrate A. Equations 3, 4, and 5 were used to fit competitive, noncompetitive, and uncompetitive inhibition, respectively. In Equations 3, 4, and 5, I is the concentration of the inhibitor, and K is and K ii are the slope and intercept inhibition constants for the inhibitor, respectively.
The solvent kinetic isotope effects on V and V/K were determined by measuring the initial velocities using saturating concentrations of UDP-MPP and varying concentrations of Ala-tRNA in H 2 O or 75% D 2 O as solvent. Solvent deuterium isotope effects were calculated from Equation 6, where, I represents the fraction of isotope, and E VK and E V are isotope effects on V/K and V (Ϫ 1), respectively.

pH Profile
The reaction rates of [ 14 C]Ala transfer from alanyl-tRNA to UDP-MPP by LvFemX were measured over the pH range of 5.5-9.1 (in 50 mM MES buffer, pH 5.5-6.8 and 50 mM Tris and/or Hepes buffer, pH 7.1-9.1) using 600 M each substrate. The reaction mixture was preincubated for 20 min before the addition of FemX. Assays were also performed using 300 M precharged tRNA (to ensure the completeness of the aminoacylation reaction over a wide pH range), and the rates were fitted to Equation 7, where C is the pH-independent plateau value, K a is the ionization constant for the acidic group, K b is the ionization constant for the basic group, and H is the hydrogen ion concentration. Reaction with EDC/Nitrotyrosine Ethyl Ester (NTEE)-FemX solutions (100 M, 500 l) in 50 mM MES/Hepes buffer (75:25, v/v), pH 6.0, were incubated with 5 mM EDC and 5 mM NTEE at room temperature in the absence and in the presence of 120 M microhelix Ala for 45 min. Subsequently, the reaction was arrested by the addition of 10% (w/v) trichloroacetic acid, precipitated protein was collected by centrifugation, washed extensively with chilled acetone, air-dried, and dissolved in 0.1 M NaOH. The number of nitrotyrosyl groups incorporated was determined spectrophotometrically, at 430 nm, using a molar absorption coefficient of 4600 M Ϫ1 cm Ϫ1 (20).

Site-directed Mutagenesis
Site-directed mutagenesis was carried out using the QuickChange site-directed mutagenesis kit (Stratagene). Deoxyoligonucleotide synthesis and other procedures were performed according to the manufacturer's instructions. C-terminally His 6 -tagged mutants were purified as described previously (12).

RESULTS
Substrate Specificity of LvFemX-In all kinetic measurements, the rates of aminoacylation of tRNAs were adjusted so that the FemX reaction was rate-limiting. Initial velocities were determined at pH 7.5, at 5-8 different concentrations of each substrate. The data were fitted by nonlinear, least square curve fitting. Kinetic constants for aminoacyl-tRNAs, determined at a saturating concentration of UDP-MPP Lys and UDP-MPP Orn are listed in Table I. The steady-state kinetic parameters for UDP-MPP Lys and UDP-MPP Orn determined at saturating concentrations of different aminoacyl-tRNAs are also summarized in Table I.
Kinetic Mechanism-The initial velocity pattern was determined using UDP-MPP Lys and Ala-tRNA Ala at five different concentrations for each substrate. The resultant double reciprocal plot was intersecting ( Fig. 2A), indicating a sequential kinetic mechanism. Product inhibition experiments were carried out using UDP-MPP Lys -Ala versus either UDP-MPP Lys or Ala-tRNA Ala . UDP-MPP Lys -Ala exhibited linear, competitive inhibition versus UDP-MPP Lys (Fig. 2B)  Solvent Kinetic Isotopic Effects-Solvent kinetic isotope effects on acyl transfer were determined by measuring initial velocities at pH 7.5, in both H 2 O and 75% D 2 O. Reactions were performed at a fixed, saturating concentration of UDP-MPP Lys and at varying Ala-tRNA Ala concentrations. The solvent kinetic isotope effect on V was 1.7 Ϯ 0.2 and on V/K was 1.1 Ϯ 0.1 (Fig. 2D).
pH Profile-The pH activity profile of the Ala transfer reaction by LvFemX was bell-shaped ( Fig. 3) and suggested the involvement of two groups exhibiting pK a values of 5.52 Ϯ 0.03 and 9.32 Ϯ 0.05 whose ionization is critical for catalytic activity.
Modification of Carboxyl Groups and Site-directed Mutagenesis-LvFemX lost more than 95% of its activity when incubated with EDC. No loss of enzymatic activity was observed in control samples. Semilogarithmic plots of residual activity as a function of time of inactivation at various concentrations of the EDC (Fig. 4) were linear up to 5% remaining activity, indicating that the inactivation follows pseudo first-order kinetics. The slope of the plot of the log of inactivation rate versus the log [EDC] was unity (Fig. 4, inset) suggesting that at least one molar equivalent of the reagent binds to one molecule of the enzyme for inactivation. Incubation of the enzyme with 1.2 molar equivalents of either Ala-tRNA or microhelix Ala protected against inactivation by EDC, with about 85% of the initial activity being retained after incubation with EDC (Fig.  4). Modification of the enzyme with EDC in the presence of NTEE resulted in the incorporation of 5.05 nitrotyrosyl groups per molecule of the enzyme, with a loss of more than 95% of the initial activity. However, modification of the enzyme by EDC and NTEE, in the presence of 1.2 equivalents of either Ala-tRNA or microhelix Ala , resulted in the incorporation of 3.1 nitrotyrosyl residues per molecule of the enzyme, with reten-  tion of 85% of the initial activity. This suggests that the loss of catalytic activity was due to modification of two carboxyl groups essential for the catalytic activity. Similarly, modification of LvFemX with the more reactive, carboxylate-specific reagent, Woodward reagent K (2-ethyl-5-phenylisoxazolium-3Ј-sulfonate) inactivated the enzyme completely at a molar ratio of 1:20 as the result of the modification of 5 carboxyl groups, as determined by electrospray mass spectrometry of the inactivated FemX (data not shown). Inclusion of Ala-tRNA again protected against the modification of two carboxyl groups and the enzyme retained 80% of the initial catalytic activity (data not shown). To identify which carboxyl groups are involved in catalytic activity and protected against modification by Ala-tRNA, we mutated five carboxylate containing amino acid residues, which are conserved among FemABX family members to their amido-isosteres. Three of the mutant forms of FemX; E205Q, E215Q, and E316Q retained 100, 85, and 40%, respectively, of the wild-type activity. However, the mutated enzymes, D109N and E320Q, lost more than 99 and 96% of the catalytic activity, respectively, as compared with wild-type enzyme. The steady-state kinetic parameters of UDP-MPP and Ala-tRNA Ala substrates for the D109N and E320Q mutant enzymes are listed in Table II. DISCUSSION The FemABX family of enzymes were identified in the late 1960s and early 1970s and shown to be aminoacyl-tRNA-dependent nonribosomal peptidyl transferases (2-4, 21, 22). However, no mechanistic studies have been reported on this interesting family of peptide bond synthesizing enzymes, perhaps due to the extreme complexity of their substrates and assay procedures. The genes encoding FemABX enzymes were discovered in the 1990s and LvFemX was the first active re- combinant enzyme reported (12). The fact that LvFemX is a soluble protein that acts before lipid conjugation of its UDP-MPP substrate and exhibits good activity using unmodified E. coli Ala-tRNA Ala , make it a good candidate for preliminary mechanistic studies of members of the FemABX enzymes. The method described here (adapted from Ref. 15) to prepare in vitro transcribed tRNAs using T7 RNA polymerase is an inexpensive, and efficient, method to generate tRNAs (or any RNA molecules from a DNA template) on the milligram scale.
From the determined steady-state kinetic values, particularly the relative V/K values, Ala-tRNA Ala was the best donor substrate for LvFemX. Microhelix Ala exhibited an identical k cat value but a 2-fold increase in K m value. However, the increase in K m could be partially attributed to the likely heterogeneity in microhelix preparations (which has been observed before using the T7 RNA polymerase in vitro, see Ref. 16) that could result in the generation of unchargable forms of microhelix, which would be observed in the spectrophotometric method of quantitation used. This suggests that FemX primarily recognizes the stem loop of tRNA and that other regions of the tRNA are not required for the peptidyl transferase activity of FemX. The K m value for Ala-tRNA Ser was identical to Ala-tRNA Ala but its maximal velocity was 3 times slower compared with Ala-tRNA Ala . Serine was transferred at an even slower rate from Ser-tRNA Ser . Though Ser-tRNA Ser is a poor substrate, it exhibited a 2-fold lower K m value compared with Ala-tRNA Ala . Gly-tRNA Gly was a very poor substrate for which the activity could be demonstrated, but for which the kinetic parameters could not be obtained with precision. The K m value of Ala-tRNA Ala (15 M) for LvFemX appears to be high considering the physiological importance of the FemX reaction and the cellular concentrations of tRNAs. However, the affinity of LvFemX for heterologous tRNAs from different sources varies considerably and the L. viridescens Ala-tRNA Ala (natural source) was previously shown to be the best substrate (2).
Among the UDP-MPPs tested, UDP-MPP Lys was the best substrate. UDP-MPP Orn exhibited a comparable k cat value but a 3-fold higher K m value. UDP-MPP Dap was an extremely poor substrate for LvFemX using E. coli Ala-tRNA Ala , with only trace amounts of [ 14 C]Ala being transferred upon prolonged incubation at high concentrations (1 mM) of the pentapeptide. Further, UDP-MPP Dap did not exhibit any significant inhibitory effect versus UDP-MPP Lys or UDP-MPP Orn at concentrations up to 1 mM (data not shown). The intersecting initial velocity plot obtained with UDP-MPP Lys and Ala-tRNA Ala suggests a sequential kinetic mechanism where both substrates must be bound to the enzyme for catalysis to occur. In order to distinguish between the ordered and random addition of substrates we determined product inhibition patterns by UDP-MPP-Ala and Ser-tRNA. The observation of linear, competitive inhibition versus UDP-MPP and linear, noncompetitive inhibition versus Ala-tRNA Ala is consistent with the ordered addition of UDP-MPP followed by Ala-tRNA Ala . The observation of uncompetitive, dead end inhibition by Ser-tRNA versus UDP-MPP confirms the order of substrate addition.
To probe the potential rate-limiting nature of the chemical reaction, solvent kinetic isotope effects were determined at pH 7.5, a region where small changes in pH(D) did not have any effect on kinetic parameters (data not shown). The unitary values of the solvent kinetic isotope effects on V/K suggest that aminoacyl-tRNA is kinetically "sticky." The larger values of the solvent kinetic isotope effects on V, which includes steps from the precatalytic ternary complex through final product release may reflect the effects of solvent isotopic substitution on the chemical step, the release of products, or the conformational changes that allow this to occur. Although small, the inequality of D2O V and D2O V/K argues against a rapid equilibrium random mechanism where the chemical step is rate-limiting.
The pH dependence of the activity of LvFemX suggests that the deprotonation of a group exhibiting a pK value of 9.32 Ϯ 0.05 and the protonation of a group exhibiting a pK value of 5.52 Ϯ 0.03 causes a loss of catalytic activity (Fig. 3). The group that exhibits a pK value of 5.52 is most likely an enzyme group that acts as a general base to promote catalysis via abstraction of a proton from the ⑀-amino group of the lysine residue of UDP-MPP Lys . A possible role for the group exhibiting the pK value of 9.32 is to function as general acid donating a proton to the 3Ј-oxygen of the ribose of tRNA during the collapse of the tetrahedral intermediate. We propose a model for the reaction catalyzed by LvFemX that includes this acid/base assistance (Fig. 5). UDP-MPP binds to the free enzyme followed by Ala-tRNA Ala . In the ternary complex, nucleophilic attack on the carbonyl of the aminoacylated-tRNA generates the zwitterionic, tetrahedral intermediate. Based on our pH studies, we invoke the participation of a general acid to assist in protonating the 3Ј-hydroxyl group of the tRNA ribose moiety and a general base to deprotonate the positively charged amine. A very similar mechanism involving catalysis by promoting the appropriate decomposition of a similar tetrahedral intermediate has been proposed for an aminoglycoside 2Ј-N-acetyltransferase (23).
We could demonstrate that the carbodiimide, EDC, inactivated the enzyme in a time-dependent manner. Though EDC is relatively specific for carboxyl groups, it can react with cysteines (24). However, LvFemX does not contain any cysteine  residues, suggesting that the loss of catalytic activity can be attributed to the modification of catalytically essential carboxyl group(s). Protection against inactivation by substrates suggested that the loss of activity was due to the modification of carboxyl groups located at, or near, the active site and not due to nonspecific modification of the enzyme. A multiple sequence alignment of members of the FemABX family revealed the presence of five highly conserved Asp or Glu residues. Mutation of either Glu 205 or Glu 215 to the corresponding glutamine had no effect on the catalytic activity. Glutamate 316 is totally conserved in FemABX family members; however, the E316Q mutant exhibited only a modest 3-fold decrease in the catalytic activity. Asp 109 and Pro 110 are totally conserved in FemABX family members, and Pro 110 is essential for the structural integrity of the enzyme, as mutation of this residue results in total loss of catalytic activity (12). The V/K value of the D109N mutant decreased more than 230-fold (Table II), and this is primarily due to a 60-fold decrease in k cat , with only a modest increase in K m values of the mutant enzyme. Glutamate 320 is another totally conserved residue and the mutant E320Q enzyme exhibited a 20-fold lower V/K value and a 12-fold lower k cat . These data suggest roles for Asp 109 and Glu 320 in the catalytic function of the enzyme. FemABX homologs are present in all organisms that produce branched peptidoglycan and are absent in species which produce directly cross-linked peptidoglycan (14,25). However, the family members share low degrees of sequence similarity (12,14). The three-dimensional structure of S. aureus FemA (26) reveals it has two domains; a helical arm domain and a globular domain that resembles histone acetyltransferases (Fig.  6A). LvFemX exhibits 38% sequence similarity compared with S. aureus FemA. Alignment of the LvFemX sequence with the S. aureus FemA sequence and structure reveals an amino acid deletion in FemX corresponding to the helical arm domain. There are no highly conserved amino acid residues present in the helical arm domain. The helical arm domain of FemA was proposed to be the tRNA binding site based on similarities to bacterial seryl-tRNA synthetases (26). Despite the truncated 12 amino acid helical arm domain, FemX can efficiently bind and catalyze amino acid transfer from aminoacyl-tRNAs. LvFemX is a cytoplasmic enzyme in vivo, while other known FemABX enzymes are cell wall-associated, and we suggest that the helical arm may be involved in cell wall association.
The globular domain of FemA has a large, L-shaped channel that traverses the surface of the domain (Fig. 6B) that has been proposed to be the pentapeptide substrate binding cavity (26). Asp 108 and Asp 396 (equivalent to LvFemX Asp 109 and Glu 320 ) are located at the bottom of the cleft with their side chain carboxylates facing toward the cleft at a distance of 4.21 Å from each other (Fig. 6, B and C). In the near vicinity are two conserved residues, Tyr 71 and Arg 76 (equivalent to LvFemX Tyr 73 and Lys 76 ) at a distance of 2.46 and 4.69 Å, respectively, from Asp 108 (Fig. 6C). These structural observations, coupled with our kinetic studies, chemical modification and site-directed mutagenesis suggest that Asp 109 is the general base exhibiting a pK value of 5.5 whose protonation abolishes catalytic activity. The function of Glu 320 may be to fix the positions of Tyr 73 and Lys 76 , either of which may serve as the general acid based on the higher pK value of 9.3.
In penicillin-resistant strains of S. aureus, femA and femB could be deleted with resultant restoration of ␤-lactam sensitivity, but fmhB (FemX) is essential for the viability (8). Whereas S. pneumoniae MurM and MurN (equivalent to FemX and FemA) are reported to be nonessential, MurM is essential for ␤-lactam resistance. These data argue that FemX is an excellent target for narrow spectrum antibiotic development against pathogenic bacteria that produce branched peptidoglycan. Inhibitors developed against FemX would be expected to be active in organisms in which FemX is essential (such as S. aureus) and to be effective in restoring ␤-lactam sensitivity where FemX is not essential for viability (e.g. S. pneumoniae). Subray S. Hegde and John S. Blanchard -Alanyltransferase) 6 N -acetylmuramoyl Pentapeptide-lysine N FemX (UDP-