Structure and function of the Mur enzymes: development of novel inhibitors

Authors


Summary

One of the biggest challenges for recent medical research is the continuous development of new antibiotics interacting with bacterial essential mechanisms. The machinery for peptidoglycan biosynthesis is a rich source of crucial targets for antibacterial chemotherapy. The cytoplasmic steps of the biosynthesis of peptidoglycan precursor, catalysed by a series of Mur enzymes, are excellent candidates for drug development. There has been growing interest in these bacterial enzymes over the last decade. Many studies attempted to understand the detailed mechanisms and structural features of the key enzymes MurA to MurF. Only MurA is inhibited by a known antibiotic, fosfomycin. Several attempts made to develop novel inhibitors of this pathway are discussed in this review. Three novel inhibitors of MurA were identified recently. 4-Thiazolidinone compounds were designed as MurB inhibitors. Many phosphinic acid derivatives and substrate analogues were identified as inhibitors of the MurC to MurF amino acid ligases.

Introduction

Increased emergence of pathogenic bacterial strains with high resistance to antibiotic therapy constitutes a serious public health threat. This situation has created an urgent need for the development of new antibacterial agents directed towards novel targets. One of the best known and most validated targets for antibacterial therapy is the machinery for peptidoglycan biosynthesis. Peptidoglycan is an essential component of the bacterial cell wall, conferring mechanical resistance to the higher internal osmotic pressure and maintaining a defined cell shape (Rogers et al., 1980). Therefore, drugs interfering with the proper biosynthesis and assembly of peptidoglycan are effective antimicrobial agents. In addition, peptidoglycan is a structure unique to prokaryotic cells and, thus, optimal for the selective targeting of microbial vital pathways. Peptidoglycan consists of linear glycan chains interlinked by short peptides. The glycan chains are composed of alternating units of N-acetylglucosamine and N-acetylmuramic acid. Muramyl residues bear short pentapeptides, a proportion of which are cross-linked either directly or through a second short peptide. It is this cross-linkage that joins the glycan chains into a macromolecular network of high tensile strength and rigidity. A recent review extensively discussed the genetic, biochemical and physiological data concerning the biosynthesis of peptidoglycan monomer unit (van Heijenoort, 2001). The aim of this review is to focus on the reactions catalysed by the cytoplasmic enzymes MurA to MurF (Fig. 1) and to discuss their use as targets for antibacterial therapy. We report the recent advances made in the development of novel Mur inhibitors, focusing on the enzymes’ structural features and interactions with inhibitors.

Figure 1.

Conversion of UDP-N-glucosamine to UDP-N-acetylmuramyl pentapeptide by the sequential action of MurA to MurF enzymes. UDP-Glc-NAc, UDP-N-acetylglucosamine; UDP-Glc-NAc-EP, UDP-N-acetylglucosamine enolpyruvate; UDP-Mur-NAc, UDP-N-acetylmuramic acid; mDAP, meso-diaminopimelic acid; G+, Gram-positive bacteria; G–, Gram-negative bacteria.

Role of Mur enzymes in peptidoglycan biosynthesis

The biosynthetic pathway of peptidoglycan is a complex two-stage process. The first stage, which occurs in the cytoplasm, is the formation of the monomeric building block N-acetylglucosamine–N-acetylmuramyl pentapeptide. The first committed step in the pathway is the transfer of an enolpyruvate residue from phosphoenolpyruvate (PEP) to position 3 of UDP-N-acetylglucosamine. This reaction is catalysed by MurA. This is followed by a MurB-catalysed reduction of the enolpyruvate moiety to d-lactate, yielding UDP-N-acetylmuramate. A series of ATP-dependent amino acid ligases (MurC, MurD, MurE and MurF) catalyse the stepwise addition of the pentapeptide side-chain on the newly reduced d-lactyl group, resulting in the formation of UDP-N-acetylmuramyl pentapeptide.

Mur enzymes as potential antibacterial targets

A large number of antibiotics in clinical use, mostly β-lactams and glycopeptides, act by inhibiting the later steps in peptidoglycan biosynthesis. However, the earlier steps of the biosynthesis of cytoplasmic peptidoglycan precursor are poorly exploited as antibacterial targets: none of the enzymes involved in these steps is inhibited by known antibiotics or synthetic chemicals of therapeutic usefulness, except for MurA, which is inhibited by fosfomycin. The murA to murF genes are all essential in bacteria. In addition, Mur proteins are highly conserved among various bacterial species, and common structural motifs can be identified. For this reason, a potential Mur inhibitor would be expected to be bactericidal and to have a wide spectrum, which validates the choice of these important bacterial enzymes as targets for the development of new inhibitors.

Structure, catalysis and inhibitors of MurA

Structure and catalytic mechanism

The crystal structures of Escherichia coli MurA com-plexed with UDP-N-acetylglucosamine and fosfomycin (Skarzynski et al., 1996) and of unliganded Enterobacter cloacae MurA (Eschenburg and Schonbrunn, 2000) are both known. MurA has two globular domains connected by a double-stranded linker (Fig. 2A). According to E. coli MurA numbering, the first domain containing the catalytic site Cys-115 comprises residues 22–229, and the second domain comprises residues 1–21 and 230–419. The main-chain fold of each domain is very similar, with three parallel internal helices surrounded by three helices and three four-stranded β-sheets. The overall protein fold is very similar to that of the enzyme 5-enolpyruvylshikimate-3-phosphate (EPSP) synthase, the only other known enolpyruvyl transferase. The catalytic site of MurA is situated in a deep cavity between the two globular domains. The uridinyl ring of UDP-N-acetylglucosamine is sandwiched between two hydrophobic surfaces created by Arg-120 and Pro-121 on one face and by Leu-124 on the other. These interactions, combined with the hydrogen bonds involving uridinyl base atoms N1 and O1, provide good binding specificity for this base. The kinetic mechanism of MurA transferase has been shown to involve the generation of a non-covalently bound tetrahedral phospholactoyl-UDP-N-acetylglucosamine intermediate (Marquardt et al., 1993). The structure and chirality of this adduct were determined from the structure of the complex between the fluoromethyl tetrahedral ketal analogue and the C115A mutant of MurA (Skarzynski et al., 1998). This addition–elimination mechanism for MurA catalysis is outlined in Fig. 3. Kinetic data from another study suggested a mechanism involving the formation of a phospholactoyl–enzyme tetrahedral intermediate preceding the formation of the phospholactoyl–substrate intermediate (Brown et al., 1994). In this covalent enzyme intermediate, Cys-115 is attached to the C-2 of PEP to form an O-phosphothioketal intermediate. However, it was proved that the covalent intermediate is dispensable for catalysis and appears to be off the main catalytic pathway (Kim et al., 1996). The formation of the covalent O-phosphothioketal enzyme intermediate from MurA of E. cloacae was confirmed by solution-state and time-resolved solid-state nuclear magnetic resonance (NMR) spectroscopy (Ramilo et al., 1994). Small-angle X-ray scattering (SAXS) and fluorescence spectroscopy were used to study conformational changes in MurA upon binding to its substrates. Binding of the sugar nucleotide to the free enzyme results in substantial conformational changes, which can be interpreted as the transition from an open to a closed form. PEP also appears to induce a structural change upon its addition to the free enzyme, but this change was less pronounced than that observed upon binding with the sugar nucleotide substrate (Schonbrunn et al., 1998). Another site-directed mutagenesis study performed on E. cloacae MurA demonstrated that Lys-22 is crucial for enzymatic activity and for the formation of the phospholactoyl–enzyme tetrahedral intermediate (Samland et al., 1999). Recent phylogenetic analysis revealed two distinct classes of MurA transferases (Du et al., 2000). The first class of transferases occurs throughout all bacteria except Gram-positive Mycoplasma spp. The second type exists as a duplicate gene copy only in the low-G+C Gram-positive bacteria. Structurally and functionally, both types are highly similar, and one enzyme can substitute for the other.

Figure 2.

Structure of the MurA, -B,-D, -E and -F enzymes shown in the oval ribbon model.

A. Structure of MurA resolved to 1.8 Å (Skarzynski et al., 1996); domain 1 is shown in orange, domain 2 is shown in violet, UDP-N-acetylglucosamine is shown in white, and fosfomycin is shown in blue.

B. Structure of MurB resolved to 1.8 Å (Benson et al., 1997); domain 1 is shown in pink, domain 2 is shown in blue, domain 3 is shown in brown, UDP-N-acetylglucosamine enolpyruvate is shown in white, and FAD (flavin adenine dinucleotide) is shown in yellow.

C. Structure of MurD resolved to 1.7 Å (Bertrand et al., 1999); domain 1 is shown in yellow, domain 2 is shown in red, domain 3 is shown in green, UDP-N-acetylmuramyl-l-alanine is shown in white, and the amino acid substrate (d-glutamic acid) is shown in rose.

D. Structure of MurE resolved to 2 Å (Gordon et al., 2001); domain 1 is shown in yellow, domain 2 is shown in red, domain 3 is shown in green, UDP-N-acetylmuramyl dipeptide is shown in white, and the amino acid substrate (meso-diaminopimelic acid) is shown in rose.

E. Structure of MurF resolved to 2.3 Å (Yan et al., 2000); domain 1 is shown in yellow, domain 2 is shown in red, and domain 3 is shown in green. Domains 1, 2 and 3 of the amino acid ligases (MurD, -E and -F) are shown in the same colours. The image was created using insightII version 2000.1 (Accelrys) on a Silicon Graphics Fuel workstation. Crystal structure co-ordinates were retrieved from the Protein Data Bank (http:www.rcsb.orgpdb): MurA, 1UAE; MurB, 2MBR; MurD, 2UAG; MurE, 1E8C; MurF, 1GG4.

Figure 3.

Addition–elimination mechanism for MurA involving the formation of a tetrahedral phospholactoyl-UDP-N-acetylglucosamine adduct.

Inhibitors

MurA is the only cytoplasmic step inhibited by a clinically used antibacterial agent. Fosfomycin (Christensen et al., 1969; Hendlin et al., 1969), a naturally occurring broad-spectrum antibiotic, is the best known inhibitor of MurA. Fosfomycin has been the drug of choice for the treatment of paediatric gastrointestinal infections resulting from Shiga-like toxin-producing Escherichia coli (STEC) in Japan, and the early administration of this antibiotic is critical for the effective treatment of STEC infections (Horii et al., 1999). Fosfomycin is also among the first-line agents for the treatment of bacterial infections of the urinary tract, which is a common health problem, particularly in women (Nicolle, 2002). Inhibition of MurA enzyme by fosfomycin is competitive, the antibiotic acting as an analogue of PEP and forming a covalent bond with the active cysteine residue of the enzyme (Fig. 4). Fosfomycin is tightly packed between the enzyme and UDP-N-acetylglucosamine, making hydrogen bonds with several different segments of the polypeptide chain. There are also hydrogen bonds between the fosfomycin hydroxyl group and the C-3 hydroxyl of the UDP-N-acetylglucosamine sugar ring and between one of its phosphonate oxygen atoms and the amide nitrogen of UDP-N-acetylglucosamine (Skarzynski et al., 1996). Time-dependent inactivation of MurA by fosfomycin was found to be greatly accelerated by the presence of UDP-N-acetylglucosamine, which suggests that active site conformational changes induced by UDP-N-acetylglucosamine are essential for inactivation (Marquardt et al., 1994). SAXS and fluorescence spectroscopy showed that fosfomycin did not affect the structure of the free enzyme or the sugar nucleotide-bound enzyme (Schonbrunn et al., 1998). However, more recent fluorescence spectroscopy studies conducted by the same researchers showed evidence of fosfomycin-induced structural changes in the case of the UDP-N-acetylglucosamine-liganded enzyme with a time dependence similar to that observed for the inactivation process (Schonbrunn et al., 2000).

Figure 4.

Inactivation of MurA by fosfomycin as a result of the covalent linkage between Cys-115 of MurA and fosfomycin.

There is a high frequency of development of fosfomycin resistance. Fosfomycin enters bacterial cells by active transport through the l-α-glycerophosphate (α-GP) uptake system and the glucose-6-phosphate (G6P) uptake system (Kahan et al., 1974). Chromosomally encoded fosfomycin-resistant strains have an impairment in fosfomycin uptake (Kadner and Winkler, 1973), a low-affinity transferase enzyme (Venkateswaran and Wu, 1972) or overproduction of the enzyme (Marquardt et al., 1992). Fosfomycin resistance encoded by plasmids results from enzymatic modification of the antibiotic in some clinical isolates of Serratia marcescens, Klebsiella pneumoniae, E. cloacae and Staphylococcus epidermidis (Mendoza et al., 1980; Etienne et al., 1989; Suarez and Mendoza, 1991; O’Hara, 1993; Arca et al., 1997). Owing to the importance of fosfomycin in the treatment of STEC infections, the emergence of fosfomycin-resistant STEC isolates has become a significant problem in antibiotic therapy for these infections. A recent report suggests that the fosfomycin resistance in these STEC isolates results from concurrent effects of alteration of the α-GP uptake system and/or the G6P uptake system and of the enhanced transcription of the murA gene (Horii et al., 1999).

It has been suggested that the resistance of Mycobacterium tuberculosis to fosfomycin results from the presence of an aspartate residue instead of the cysteine residue at the active site of the enzyme. Replacement of this aspartate residue by a cysteine reversed the resistance of the enzyme to inhibition by fosfomycin (De Smet et al., 1999). In addition, replacement of the active Cys-115 in the E. coli enzyme counterpart by aspartate and, to a much lesser degree, by glutamate maintained enzymatic activity but conferred complete resistance to the time-dependent inactivation by fosfomycin (Kim et al., 1996).

Three novel inhibitors of the E. coli MurA enzyme, a cyclic disulphide, a purine analogue and a pyrazolopyrimidine, were identified recently (Baum et al., 2001). When preincubated with MurA, these compounds showed IC50s lower than that of fosfomycin (Table 1). The presence of UDP-N-acetylglucosamine during preincubation lowered the IC50 at least fivefold, suggesting that, like fosfomycin, the three compounds may interact with the enzyme in a specific fashion that is enhanced by UDP-N-acetylglucosamine. The three compounds have no apparent structural similarity to fosfomycin. The three compounds exhibited antibacterial activity, but not caused specifically by MurA inhibition, as inhibition of DNA, RNA and protein synthesis was observed in addition to the inhibition of peptidoglycan biosynthesis. The minimum inhibitory concentrations (MICs) of the compounds against test strains were of the same order as fosfomycin. The compounds are proposed to be tightly, but not covalently, associated with MurA. They seem to bind to the MurA enzyme at or near the site at which fosfomycin binds.

Table 1.  Summary of the inhibitors of MurA to MurF enzymes.
EnzymeInhibitorExtent of inhibitionReference
  • a

    . A2pm, meso-diaminopimelic acid.

MurAFosfomycinIC50 = 8.8 µM Baum et al. (2001)
Cyclic disulphideIC50 = 0.2 µM Baum et al. (2001)
Purine analogueIC50 = 0.9 µM Baum et al. (2001)
PyrazolopyrimidineIC50 = 0.3 µM Baum et al. (2001)
MurB4-ThiazolidinonesIC50s 7.7–28.4 µM Andres et al. (2000)
MurCPhosphinateIC50 = 49 µM Reck et al. (2001)
l-Alanine analogues:  
 β-Alanine K is = 110 mM Emanuele et al. (1996)
 β-CN-l-Alanine K is = 3.3 mM Emanuele et al. (1996)
 l-Vinylglycine K is = 5.8 mM Emanuele et al. (1996)
 β-Chloro-l-alanine32.8% Gubler et al. (1996)
 l-Cysteine19% Liger et al. (1995)
 β-Chloro-l-alanine76% Liger et al. (1995)
 β-Cyano-l-alanine88% Liger et al. (1995)
 β-Fluoro-l-alanine94% Liger et al. (1995)
MurDPhosphinateIC50 = 680 nM Tanner et al. (1996)
PhosphinateIC50 < 1 nM Gegnas et al. (1998)
d-Glutamic acid analogues:  
 dl-Homocysteic acid58% Pratviel-Sosa et al. (1994)
 d-erythro-3-Methylglutamic acid47% Pratviel-Sosa et al. (1994)
 d-erythro-4-Methylglutamic acid26% Pratviel-Sosa et al. (1994)
MurEPhosphinateIC50 = 1.1 µM Zeng et al. (1998)
A2pma analogues:  
 (2S,3R,6S)-3-Fluoro-A2pmIC50 = 2.3 mM Auger et al. (1996)
 N-Hydroxy-A2pmIC50 = 0.56 mM Auger et al. (1996)
MurFAminoalkyl phosphinates:  
 N-Acyl phosphinate K i = 700 µM Miller et al. (1998)
 N-Glutaryl phosphinate K i = 200 µM Miller et al. (1998)
 Pseudo-tetrapeptide phosphinate K i = 200 µM Miller et al. (1998)
ATP analogue K is = 33.6 µM Anderson et al. (1996)

Structure, catalysis and inhibitors of MurB

Structure and catalytic mechanism

The reduction of UDP-N-acetylglucosamine enolpyruvate to UDP-N-acetylmuramic acid by the MurB flavoprotein involves a sequence of two half-reactions. First, bound FAD (flavin adenine dinucleotide) is reduced by the two-electron transfer from NADPH. Then, these two electrons are transferred to the C-3 of the enolpyruvyl group (Benson et al., 1993). The crystal structures of the E. coli substrate-free MurB and of MurB complexed with its substrate UDP-N-acetylglucosamine enolpyruvate were both determined (Benson et al., 1995; 1996). MurB is a mixed α+β protein composed of three domains (Fig. 2B). Domain 1 includes residues 3–67 and residues 327–342 and contains six β-strands and α-helices. Domain 2 includes residues 68–201 and contains nine β-strands and two α-helices. Domain 3 includes residues 219–326 and contains six β-strands and three α-helices. Comparisons between the structures of free enzyme and the substrate-bound enzyme show that the absence of UDP-N-acetylglucosamine enolpyruvate causes domain 3 (residues 219–319) to undergo a rigid-body rotation towards domains 1 and 2. Thus, binding of the sugar substrate to MurB results in the closure of the substrate-binding channel over the substrate (Benson et al., 1996). Steady-state kinetic studies demonstrated weak and strong substrate inhibition by NADPH and UDP-N-acetylglucosamine enolpyruvate respectively. Initial velocity measurements as a function of both substrates produced patterns consistent with a ping pong bi bi double competitive substrate inhibition mechanism. These data suggest that NADPH and UDP-N-acetylglucosamine enolpyruvate share a subsite that prevents both molecules from binding at once (Dhalla et al., 1995). NMR studies with perdeuterated, uniformly 13C/15N-labelled MurB confirmed that both substrates use the same binding pocket of the enzyme (Constantine et al., 1997). This result is consistent with the role of the bound FAD cofactor in mediating hydride transfer from the NADPH to the UDP-sugar substrate. The role of Ser-229, acting as a general acid catalyst, at the active site of the enzyme was confirmed by a single point mutation replacing it by alanine and eliminating the hydroxyl group that acts as a proton donor in the second half-reaction (Benson et al., 1997).

Inhibitors

A novel class of 4-thiazolidinone inhibitors of MurB was identified (Andres et al., 2000). Based on the X-ray crystallographic data of MurB, and using the UDP-sugar substrate as a guide, surrogates of the diphosphate moiety were developed. The template was intended to mimic key interactions of the diphosphate with the enzyme and to be able to orient the resultant side-chains in such a way that they would occupy space similar to the glucosamine and uridine moieties of the substrate. In the above study, there was no mention of a microbiological activity for the 4-thiazolidinone compounds, presumably because of the lack of such activity.

Structure and catalysis of amino acid ligases

Structure

The four amino acid ligases, also called Mur synthetases, were crystallized. The three-dimensional structures of MurD, -E and -F were determined, but the three-dimensional structure of crystallized MurC remains to be resolved. Crystallization experiments with MurC produced two crystal forms (Emanuele et al., 1996). Through sedimentation equilibrium analysis, the quaternary structure of MurC was characterized as a monomer–dimer equilibrium (Jin et al., 1996). The enzyme is active as both dimers and monomers, and the specific activity is independent of the oligomerization state of the enzyme.

The first structure described for a Mur synthetase was that of MurD from E. coli (Bertrand et al., 1997). The crystal structure of MurD in the presence of its UDP-sugar substrate (Fig. 2C) was shown to comprise three globular domains, the topology of which is reminiscent of nucleotide-binding folds. The N- and C-terminal domains are consistent with the known dinucleotide-binding fold of dehydrogenases, and the central domain is consistent with the classic mononucleotide-binding fold found in many ATP- and GTP-binding proteins. Domain 1 comprises residues 1–93 and consists of a five-stranded parallel β-sheet surrounded by four α-helices. This domain accounts for the fixation of the UDP moiety of the UDP-sugar substrate. Domain 2 comprises residues 94–298 and consists of a central six-stranded parallel β-sheet surrounded by seven α-helices with a small flanking three-stranded antiparallel β-sheet. Domain 3 comprises residues 299–437 and contains a six-stranded β-sheet with five parallel strands and one antiparallel strand, surrounded by five α-helices. The crystal structure of three complexes of the E. coli MurD with a variety of substrates and products also supported the reaction mechanism proceeding via an acyl-phosphate intermediate (Bertrand et al., 1999). The three-dimensional structure of MurD was found to resemble closely that of folylpolyglutamate synthetase (FPGS), despite a low level of sequence identity, suggesting the presence of a new ADP-forming amide-bond ligase superfamily (Sheng et al., 2000).

The crystal structure of the E. coli MurE enzyme in the presence of its product, UDP-N-acetylmuramyl-tripeptide (Gordon et al., 2001), revealed three globular domains, two of which have a topology reminiscent of the equivalent domain found in the three-dimensional structure of MurD (Fig. 2D). Domain 1 comprises residues 1–88 and consists of a five-stranded β-sheet surrounded by two α-helices. Domain 2 comprises residues 90–338 and consists of a central six-stranded parallel β-sheet surrounded by seven α-helices, closely resembling the central domain of MurD. Domain 3 comprises residues 340–497 and contains a six-stranded β-sheet with five parallel strands and one antiparallel strand, surrounded by five α-helices. It also contains the dinucleotide-binding fold like its MurD counterpart. In addition, the structural determinant responsible for deciding whether the amino acid to be added to the nucleotide precursor is meso-diaminopimelic acid or l-lysine was identified.

MurF enzyme from E. coli (Fig. 2E) was recently shown to consist of three open α/β-sheet domains (Yan et al., 1999; 2000). The N-terminal domain comprises residues 1–81 and consists of a five-stranded β-sheet surrounded by three helices, two in α-helix and one in 310-helix conformation. In addition, there is a very short two-stranded antiparallel β-sheet perpendicular to the main sheet. No known protein structure homologues to this domain were found in databases. The central domain comprises residues 82–310 and consists of a central six-stranded parallel β-sheet surrounded by eight α-helices. In addition, there is a small three-stranded antiparallel β-sheet. The fold of the central domain is similar to the classic mononucleotide-binding fold as found in MurD and -E. The C-terminal domain comprises residues 311–447 and consists of a six-stranded β-sheet surrounded by five α-helices. This domain contains the dinucleotide-binding fold also observed in MurD and -E structure. The three domains of unliganded MurF juxtapose in a crescent-like arrangement, creating a wide-open space where substrates are expected to bind. As such, catalysis is not feasible, and significant domain closure is expected upon substrate binding.

Catalytic mechanism

MurC to -F are believed to function in a similar manner to the ATP-dependent amide-forming enzymes, glutamine synthetase, glutathione synthetase and d-alanine-d-alanine ligase. They all catalyse the formation of an amide or peptide bond, the reaction being driven by the concomitant hydrolysis of ATP into ADP and inorganic phosphate. They operate via a similar chemical mechanism involving carboxyl activation of the nucleotide substrate to an acylphosphate intermediate followed by nucleophilic attack by the amino group of the condensing amino acid or dipeptide, with the formation of a peptide bond and the elimination of a phosphate group (Anderson et al., 1996; Falk et al., 1996; Liger et al., 1996; Tanner et al., 1996; Vaganay et al., 1996; Bertrand et al., 1999). The role of amino acid invariants present in the active site of the Mur superfamily (paralogue invariants) and among different MurD sequences (orthologue invariants) was investigated by submitting these residues in the E. coli MurD enzyme to site-directed mutagenesis (Bouhss et al., 1999a). All mutants had a reduced enzymatic activity, which indicates the importance of these residues for catalysis. MurD, -E and -F possess an invariant lysine residue (Lys-198 in the E. coli MurD numbering). Crystallographic analysis of MurD and MurE has recently shown that this residue is present as a carbamate derivative, a modification presumably essential for Mg2+ binding and acyl phosphate formation. Functional requirement for this residue was recently established by site-directed mutagenesis and chemical rescue experiments (Dementin et al., 2001). It was found that MurD, but not MurC, catalyses the synthesis of adenosine 5′-tetraphosphate from the acyl phosphate intermediate, thereby pointing out a difference between the two enzymes (Bouhss et al., 1999b). This information about Mur synthetase catalysis is useful in designing potential inhibitors of the Mur enzymes.

Inhibitors of amino acid ligases

The most potent and best known inhibitors of amino acid ligases are phosphinate inhibitors, which are derivatives of phosphinic acid. Phosphinic acid is a dipeptide analogue linked to uridine diphosphate by a hydrophobic spacer. It has been found that appropriately substituted phosphinic acids act as slow-binding inactivators of the ATP-dependent amide-forming enzymes. The remarkable feature of this inhibition is that the enzyme promotes the transfer of the γ-phosphate of ATP onto the phosphinate anion to produce ADP and a phosphorylated inhibitor. This process has been confirmed by X-ray diffraction analysis of the phosphorylated inhibitor–enzyme complex in the cases of d-Ala-d-Ala ligase and glutathione synthetase (Fan et al., 1994). The resulting phosphoryl phosphinate moiety closely mimics the tetrahedral intermediate that is proved to form in the normal pathway of the reactions catalysed by the MurC to MurF ligases (Fig. 5). Thus, this tetrahedral intermediate-like phosphoryl phosphinate moiety is tightly bound by the enzymes. No antibacterial activity was reported for the phosphinate compounds synthesized to date. This fact could result from the inefficient transport of these molecules into the cytoplasm or their failure to accumulate up to the required inhibitory concentrations. In addition to phosphinate inhibitors, several substrate analogues (either amino acid analogues or analogues of the UDP-sugar substrates) were found to inhibit Mur synthetases (Table 1). Generally, phosphinate inhibitors are more promising than substrate analogues.

Figure 5.

The proposed reaction mechanism of the ATP-dependent amide-forming enzymes (MurC to F) and the general structure of a phosphorylated phosphinate inhibitor.

One of the key features to be considered in the design of future Mur synthetase inhibitors is the conserved binding motifs and the common kinetic mech-anism among MurC, -D, -E and -F. An inhibitor that recognizes homologous binding motifs will probably bind to and inhibit more than one enzyme. This would lead to a more efficient inhibition of the pathway. In addition, the frequency of emergence of resistance to such an inhibitor would be much lower, as mutations conferring resistance would need to occur in more than one target gene simultaneously.

Inhibitors of MurC

A series of phosphinate transition-state analogues were synthesized and tested on the E. coli MurC enzyme (Reck et al., 2001). A potent inhibitor with an IC50 of 49 nM was identified. This inhibitor was then characterized biochemically (Marmor et al., 2001). The compound exhibits mixed-type inhibition with respect to all three enzyme substrates, suggesting that it forms dead-end complexes with multiple enzyme states. These findings were supported by isothermal titration calorimetry (ITC). Analogues of l-alanine (Table 1) were tested as inhibitors of the E. coli MurC and were shown to be competitive versus l-alanine (Emanuele et al., 1996). Inhibition by other l-alanine analogues (Table 1) was also demonstrated with the E. coli enzyme but without specification of the type of inhibition (Liger et al., 1995; Falk et al., 1996; Gubler et al., 1996). Inhibition of MurC enzyme by l-alanine analogues was first observed in the early 1970s for the enzymes from Bacillus subtilis and Bacillus cereus (Hishinuma et al., 1970; 1971).

Inhibitors of MurD

The d-glutamic acid-adding enzyme (MurD) is the first amino acid ligase for which phosphinate inhibitors have been designed (Tanner et al., 1996). These are the first examples of effective mechanism-based inhibitors of one of the steps of the biosynthesis of the UDP-N-acetylmuramic acid pentapeptide. The most potent designed inhibitor contains a phosphinic acid that has a tetrahedral geometry at the dipeptide centre, which could be enzymatically phosphorylated to give a close analogue of the normal reaction intermediate (Fig. 5). It also retains the charged UDP moiety, which is likely to be important for binding. Based on the assumption that the N-acetyl muramic acid structure could be an important contributor to the potency of the amide-forming enzyme inhibitors, another research group synthesized a generation of inhibitors including this feature (Gegnas et al., 1998). The incorporation of muramic acid and the control of the stereochemical configuration of the α-amino phosphinate increased potency by more than three orders of magnitude compared with Tanner's inhibitor, as indicated by an IC50 value of < 1 nM compared with an IC50 value of 680 nM for Tanner's inhibitor. A series of N-(5-phthalimidopentanoyl)-, N-[2-(2-ethoxy)acetyl]- and N-(7-oxooctanoyl)-phosphono and phosphinoalanine derivatives has been also synthesized and evaluated for inhibition of MurD (Gobec et al., 2001). As for substrate analogues, the effect of various analogues of d-glutamic acid on the E. coli enzyme was investigated, and a few displayed weak inhibition (Pratviel-Sosa et al., 1994). In addition, N-acetylmuramic acid derivatives were also synthesized and tested as potential inhibitors of MurD, but no promising inhibitors resulted from that approach (Auger et al., 1995).

Inhibitors of MurE and MurF

The first phosphinate inhibitor of MurE was designed with structural features based on Tanner's previously reported MurD phosphinate inhibitor (Zeng et al., 1998). The compound inhibited the MurE-catalysed reaction with an IC50 value of 1.1 µM. Several analogues of diaminopimelic acid were also tested as substrates or inhibitors of MurE (Michaud et al., 1990; Auger et al., 1996). Generally, the greatest extents of inhibition were observed with the best substrates, reflecting the fact that these analogues behave as competitive inhibitors.

The only reported phosphinate inhibitors of MurF to date are aminoalkylphosphinate compounds that have been synthesized as transition-state analogues (Miller et al., 1998). They act as reversible competitive inhibitors of the E. coli MurF enzyme, with Ki values ranging from 200 to 700 µM. These compounds showed no antibacterial activity on standard strains. The non-hydrolysable ATP analogue AMP-PCP was tested on the MurF-catalysed reaction and proved to be a potent inhibitor of ATP hydrolysis and was competitive with ATP (Anderson et al., 1996).

Concluding remarks

The extensive use of antibiotics in hospitals and community since their introduction into medical practice has created major evolutionary pressures in bacteria to develop various resistance mechanisms. This phenomenon has led to increased morbidity, mortality and health care costs. The search for new antibacterial agents directed towards novel targets became highly imperative. The biosynthetic pathway of cytoplasmic peptidoglycan precursor is currently gaining much interest as a target site for antibacterial therapy. Since the mid-1990s, many inhibitors of the Mur cytoplasmic enzymes have been reported, but none has yet led to the development of a clinically utilized therapeutic agent. This review offers an overview of the structural features and the catalytic functions of MurA to MurF essential bacterial enzymes, as well as the progress made to date in the conception of Mur inhibitors. A better understanding of the structural characteristics and the elucidation of the mechanisms of Mur enzymes would provide valuable insight into the search and rational design of a new generation of specific bacterial cell wall inhibitors.

Acknowledgements

Research in R.C.L.'s laboratory was supported by the Canadian Bacterial Diseases Network via the Canadian Centers of Excellence and by a FCAR team grant to Roger C. Levesque. R.C.L. is a scholar of exceptional merit from Le Fonds de Recherche en Santé du Québec, A.E.Z. obtained a studentship from Université Laval and from La Fondation Marc Bourgie.

Ancillary