Insights into class D beta-lactamases are revealed by the crystal structure of the OXA10 enzyme from Pseudomonas aeruginosa.

BACKGROUND: beta-lactam antibiotic therapies are commonly challenged by the hydrolytic activities of beta-lactamases in bacteria. These enzymes have been grouped into four classes: A, B, C, and D. Class B beta-lactamases are zinc dependent, and enzymes of classes A, C, and D are transiently acylated on a serine residue in the course of the turnover chemistry. While class A and C beta-lactamases have been extensively characterized by biochemical and structural methods, class D enzymes remain the least studied despite their increasing importance in the clinic. RESULTS: The crystal structure of the OXA10 class D beta-lactamase has been solved to 1.66 A resolution from a gold derivative and MAD phasing. This structure reveals that beta-lactamases from classes D and A, despite very poor sequence similarity, share a similar overall fold. An additional beta strand in OXA10 mediates the association into dimers characterized by analytical ultracentrifugation. Major differences are found when comparing the molecular details of the active site of this class D enzyme to the corresponding regions in class A and C beta-lactamases. In the native structure of the OXA10 enzyme solved to 1.8 A, Lys-70 is carbamylated. CONCLUSIONS: Several features were revealed by this study: the dimeric structure of the OXA10 beta-lactamase, an extension of the substrate binding site which suggests that class D enzymes may bind other substrates beside beta-lactams, and carbamylation of the active site Lys-70 residue. The CO2-dependent activity of the OXA10 enzyme and the kinetic properties of the natural OXA17 mutant protein suggest possible relationships between carbamylation, inhibition of the enzyme by anions, and biphasic behavior of the enzyme.

pH, inhibitor, and substrate specificity studies on Escherichia coli penicillin-binding protein 5.

The recent structural determination of Escherichia coli penicillin-binding protein 5 (PBP 5) provides the opportunity for detailed structure-function studies of this enzyme. PBP 5 was investigated in terms of its stability, linear reaction kinetics, acyl-donor substrate specificity, inhibition by a number of active site-directed reagents, and pH profile. PBP 5 demonstrated linear reaction kinetics for up to several hours. Dilution of PBP 5 generally resulted in substantial loss of activity, unless BSA or a BSA derivative was added to the diluting buffer. PBP 5 did not demonstrate a significant preference against a simple set of five alpha- and epsilon-substituted L-Lys-D-Ala-D-Ala derivatives, suggesting that PBP 5 lacks specificity for the cross-linked state of cell wall substrates. Among a number of active site-directed reagents, only some thiol-directed reagents gave substantial inhibition. Notably, serine-directed reagents, organic phosphates, and simple boronic acids were ineffective as inhibitors. PBP 5 was stable over the pH range 4.6-12.3, and the k(cat)/K(m) vs. pH profile for activity against Ac(2)-L-Lys-D-Ala-D-Ala was bell-shaped, with pK(a)s at 8.2 and 11.1. This is the first complete pH profile, including both acidic and basic limbs, for a PBP-catalyzed DD-carboxypeptidase (CPase) reaction. Based on its structure, similarity to Class A beta-lactamases, and results from mutagenesis studies, the acidic and basic limbs of the pH profile of PBP 5 are assigned to Lys-47 and Lys-213, respectively. This assignment supports a role for Lys-47 as the general base for acylation and deacylation reactions.

Inhibition of transglycosylation involved in bacterial peptidoglycan synthesis.

The continuing spectre of resistance to antimicrobial agents has driven a sustained search for new agents that possess activity on drug resistant bacteria. Although several paths are available to reach this goal, the most generalized would be the discovery and clinical development of an agent that acts on a new target which has not yet experienced selective pressure in the clinical setting. Such a target should be essential to the growth and survival of bacteria, and sufficiently different from, or better still non-existent in, the human host. The transglycosylation reaction that polymerizes biochemical intermediates into peptidoglycan qualifies as such a target. This biochemical system accepts the basic unit N-acetylglucosamine-beta-1, 4-N-acetyl-muramyl-pentapeptide-pyrophosphoryl-undecaprenol (lipid II), and leads to polymerization of the N-acetylglucosamine -beta-1, 4-N-acetyl-muramyl-pentapeptide segment into peptidoglycan. Approaches to targeting this reaction include modification of known glycolipid and glycopeptide natural product antibiotics. The synthesis and antibacterial activity of synthetic analogs of moenomycin having novel antibacterial activities not present in the parent structure will be presented, together with the combinatorial chemistry and assay systems leading to their discovery. Likewise, we will discuss chemical modifications to specific glycopeptide antibiotics that have extended their spectrum to include vancomycin resistant enterococci that substitute D-alanyl-D-lactate for D-alanyl-D-alanine in their peptidoglycan. Two differing theories, one positing the generation of high affinity, specific binding to D-alanyl-D-lactate via glycopeptide dimerization and/or membrane anchoring, and the other supporting direct targeting of the modified glycopeptide to the transglycosylation complex, seek to explain the mechanism of action on vancomycin resistant enterococci. Biochemical evidence in support of these two theories will be discussed.

Structure-based design approaches to cell wall biosynthesis inhibitors.

This review summarizes some of the published attempts to incorporate protein and NMR structures in the design of new antibiotics that specifically target Cell Wall biosynthesis. Most of the steps involved in peptidglycan synthesis have been investigated as potential strategies against cell wall inhibition. Structural information has been most useful in the design of molecules in the Mur enzyme pathway, penicillin binding proteins and lactamases, as well as proteins that are part of the final steps of transglycosylation - in particular, d-Ala-d-Ala ligase. Several unique issues exist in the design of effective antibacterials, such as the significant differences in protein structure between organisms, such as the case of MurB in which a large amino acid loop that occupies the active site of the E. Coli is gone in the Staph aureus enzyme. Additionally, bacterial resistance is an important issue, and in some cases, structural information can be used to understand the source of this resistance. For example, mutations within the d-Ala-d-Ala ligases lead to the inability of Vancomycin antibiotics to bind.

Automated analysis of enzyme inactivation phenomena. Application to beta-lactamases and DD-peptidases.

In the presence of a reporter substrate, the progressive inactivation of an enzyme was easily studied by directly transmitting absorbance readings to a microcomputer. Pseudo-first order rate constants as high as 0.3 sec-1 were rapidly and accurately measured. When utilization of the reporter substrate did not exceed 10%, the rate of the reaction (vt) could be considered as proportional to the active enzyme concentration at any time during the analysis and the decrease of vt was first order with time. This simple method was used to follow the inactivation of beta-lactamases (EC 3.5.2.6) by various physical and chemical agents. When a large proportion (30-80%) of reporter substrate was destroyed, a correction was introduced to account for the corresponding decrease of its rate of utilization. This enabled experiments to be performed with a DD-peptidase and a substrate exhibiting a low delta epsilon upon hydrolysis. For the first time, the inactivation of a penicillin-sensitive enzyme by a beta-lactam could be continuously and directly observed. Finally, the method was extended to the study of hysteresis phenomena.

Recent developments in carbapenems.

Carbapenems are beta-lactam antibiotics characterised by the presence of a beta-lactam ring with a carbon instead of sulfone in the 4-position of the thyazolidinic moiety. The first carbapenem to be utilised in therapy was imipenem, the N-formimidoyl derivative of thienamycin. Imipenem is coadministered with cilastatin, an inhibitor of human renal dehydropeptidase I, as imipenem is hydrolysed by this enzyme. Meropenem was the first carbapenem with a 1-beta-methyl group and 2-thio pyrrolidinyl moiety, which renders this antibiotic stable to renal dehydropeptidase I. Other carbapenems for parenteral administration later discovered include biapenem, panipenem, ertapenem, lenapenem, E-1010, S-4661 and BMS-181139. Carbapenems which are orally administered include sanfetrinem, DZ-2640, CS-834 and GV-129606. Carbapenems have an ultra-broad spectrum of antibacterial activity and stability to almost all clinically relevant beta-lactamases. This differentiates them from all other currently available classes of beta-lactam antibiotics. However, Class B beta-lactamases, along with some rare Class A and D enzymes, are able to hydrolyse these antibiotics. Although Class B enzymes are generally chromosomally-encoded (isolated from Stenotrophomonas maltophilia, Aeromonas spp., Bacillus cereus, Bacteroides fragilis, Flavobacterium spp. and Legionella gormanii), plasmid-metallo-beta-lactamases now are appearing in B. fragilis, Pseudomonas aeruginosa, Acinetobacter baumannii and members of Enterobacteriaceae such as Serratia marcescens and Klebsiella pneumoniae. The number of these enzymes compared to the number of other beta-lactamase types is still low, however, it is likely that they will spread due to the increased selective pressure of carbapenem use. The very broad spectrum of antimicrobial activity associated with a good clinical efficacy and a favourable safety profile makes the carbapenems valuable as 'first-line' antibiotics in initial empirical therapy for the treatment of severe infections.

Antibacterial and pharmacokinetic properties of M14659, a new injectable semisynthetic cephalosporin.

In vitro and in vivo antibacterial activities and pharmacokinetics of M14659 were investigated. In vitro activity of M14659 was superior to that of ceftazidime against Staphylococcus aureus. Against Gram-negative bacteria except Pseudomonas aeruginosa, its activity was almost equal to that of ceftazidime. M14659 was more active against P. aeruginosa including multi-drug resistant strains than cefsulodin, cefoperazone or ceftazidime. Affinities of M14659 for penicillin-binding proteins (PBPs) of Escherichia coli and P. aeruginosa were 2 to 14 times higher for PBP-1A and PBP-1B than found for ceftazidime, and almost the same for PBP-3. In vivo activity of M14659 against S. aureus was superior to that of cefamandole, cefotaxime or ceftazidime. Against Gram-negative bacteria including P. aeruginosa, M14659 was 2 to 220 times more active than cefotaxime or ceftazidime. Plasma half-life of M14659 in mice was about 3 times longer than that of ceftazidime. M14659 administered intravenously to mice was mainly excreted in urine without metabolism, and its recovery rate was almost equal to that of ceftazidime.

Tertiary structural similarity between a class A beta-lactamase and a penicillin-sensitive D-alanyl carboxypeptidase-transpeptidase.

beta-Lactam antibiotics--the penicillins, cephalosporins and related compounds--act by inhibiting enzymes that catalyse the final stages of the synthesis of bacterial cell walls. Recent crystallographic studies of representative enzymes are beginning to reveal the structural bases of antibiotic specificity and mechanism of action, while intensive efforts are being made to understand the beta-lactamase enzymes that are largely responsible for bacterial resistance to these antibiotics. It has been suggested that the beta-lactamases and beta-lactam target enzymes may be evolutionarily related and some similarity of amino-acid sequence around a common active-site serine residue supports this idea. We present here the first evidence from a comparison of three-dimensional structures in support of this hypothesis: the structure of beta-lactamase I from Bacillus cereus is similar to that of the penicillin-sensitive D-alanyl-D-alanine carboxypeptidase-transpeptidase from Streptomyces R61.

Effects of thiol reagents on Streptomyces K15 DD-peptidase-catalysed reactions.

The 26,000-Mr DD-peptidase of Streptomyces K15 binds one equivalent of thiol reagents as 5,5'-dithiobis-(2-nitrobenzoate) or p-chloromercuribenzoate (pCMB). Derivatization of the DD-peptidase by pCMB decreases the efficacy of the initial binding of the ester carbonyl donor Ac2-L-Lys-D-Ala-D-lactate to the enzyme (K), the rate of enzyme acylation by the donor (K+2) and the rate of enzyme deacylation (k+3). However, the value of the k+2/k+3 ratio, and therefore the percentage of total enzyme which, at saturating concentrations of the donor, is present as acyl-enzyme at the steady state of the reaction, are not modified. The enzyme's binding sites for pCMB and benzylpenicillin are not mutually exclusive. But, when compared with the native enzyme, the pCMB-derivatized enzyme undergoes acylation by benzylpenicillin with a decreased second-order-rate constant (k+2/K) value and gives rise to a penicilloyl adduct of increased stability. Since the acyl-enzyme mechanism is not annihilated by pCMB derivatization, it is proposed that basically, and like all the other DD-peptidases/penicillin-binding proteins so far characterized, the Streptomyces K15 DD-peptidase is an active-site-serine enzyme.

Purification of two DD-carboxypeptidases/transpeptidases with different penicillin sensitivities from Proteus mirabilis.

Two membrane-bound enzymes of Proteus mirabilis with the dual functions of peptidoglycan DD-carboxypeptidase and transpeptidase (named DD-carboxypeptidase/transpeptidase H and L) were isolated and purified by selective solubilization with the nonionic detergent Genapol X-100, affinity chromatography on matrix-bound ampicillin, and preparative isoelectric focusing in the presence of detergent. Purified enzymes H and L were, respectively, penicillin-binding proteins 4 and 5 among seven major penicillin-binding proteins present in P. mirabilis. The enzymes differed in the following properties. Enzyme H had an Mr of 49,000; isoelectric point at pH 8.2; high sensitivity to benzylpenicillin and permanent inactivation because of high stability of the enzyme-antibiotic complex EI* (half-life 300 min); fragmentation of benzylpenicillin with formation of phenylacetylglycine during the slow decay of EI*; it functioned as an endopeptidase on peptide-crosslinked side chains of peptidoglycan. Enzyme L had an Mr of 43 000; isoelectric point at pH 5.9; low sensitivity to benzylpenicillin and low stability of EI* (half-life 7.2 min) with rapid recovery of enzyme activity; no function as an endopeptidase. The properties of enzyme L were identical with those of the single active DD-carboxypeptidase found previously in the spheroplast L-form of P. mirabilis grown in the presence of benzylpenicillin. We conclude that the partial penicillin resistance of P. mirabilis, with growth as L-form and synthesis of peptide-crosslinked peptidoglycan, depends on the continuing fuction of enzyme L as a DD-carboxypeptidase and transpeptidase in the presence of the antibiotic.

Primary structure of the COOH-terminal membranous segment of a penicillin-sensitive enzyme purified from two Bacilli.

D-Alanine carboxypeptidase is a penicillin-sensitive intrinsic membrane enzyme which is composed of a hydrophilic NH2-terminal catalytic domain (Mr congruent to 45,000 to 47,000) and a COOH-terminal membranous segment (approximately 20 to 30 amino acids in length) (Waxman, D. J., and Strominger, J. L. (1979) J. Biol. Chem. 254, 4863-4875; Waxman, D. J., and Strominger, J. L. (1981) J. Biol. Chem. 256, 2059-2066). The primary structures of the COOH-terminal 30 amino acids of two D-alanine carboxypeptidase purified from bacterial membranes were determined (residues numbered from the COOH terminus): Bacillus stearothermophilus: (formula see text) Water-soluble fragments of the B. stearothermophilus D-alanine carboxypeptidase were shown to be formed by cleavage after Phe27 or after Leu25 as indicated by carboxypeptidase A and B analysis and by the release of the four COOH-terminal chymotryptic peptides (Val26-Leu25, Ser24-Phe16, Val15-Trp12, and Thr11-Leu1) upon formation of water-soluble chymotrypsin D-alanine carboxypeptidase. This indicates that the membranous fragment is largely contained within the COOH-terminal 24 residues. Thus, this bacterial membrane protein probably does not contain the significant cytoplasmic domain characteristic of transmembrane proteins such as glycophorin. The absence of an uninterrupted stretch of 20 to 25 uncharged residues suggests that the membrane anchoring of D-alanine carboxypeptidase may differ from that of simple transmembrane proteins. Possible structures for the membranous segment of D-alanine carboxypeptidase are discussed.

Effect of piperacillin on D-alanine carboxypeptidase activities from Pseudomonas aeruginosa.

Membrane-bound D-alanine carboxypeptidase activity from Pseudomonas aeruginosa is very sensitive to inhibition by piperacillin.

Chemical modifications of the active site of Streptomyces R61 DD-carboxypeptidase.

The DD-carboxypeptidase of Streptomyces R61 is an exocellular enzyme related to the bacterial peptidoglycan cross-linking enzymes, and, like them, is inhibited by penicillin. The active-site reagents methanesulfonyl fluoride and diisopropylfluorophosphate inhibit catalytic activity and binding of penicillin G indicating the involvement of a serine residue in both processes. For methanesulfonyl fluoride the second-order rate constant (0.7 M-1 min-1) is comparable to that of classical serine proteases. For diisopropylfluorophosphate, which binds to the enzyme stoichiometrically, the second-order rate constant (1.5 M-1 min-1) is at least two orders of magnitude smaller. The arginine-specific reagents methylglyoxal, 2,3-butanedione and phenylglyoxal inactive DD-carboxypeptidase in borate buffer with second-order rate constants of 70, 70 and 120 M-1 min-1, respectively. Inactivation correlates with stoichiometric binding to the enzyme. Peptidase and esterase activities are similarly affected, suggesting that substrate binding in both cases requires an arginine-carboxyl group interaction. Penicillin binding is also inhibited, but the degree of inhibition depends on the alpha-dicarbonyl side chain. Binding of alpha-dicarbonyls to DD-carboxypeptidase facilitates subsequent binding of diisopropylfluorophosphate suggesting that interaction of these compounds with the active site might induce a conformational change on the enzyme making the serine residue more accessible to the modifying reagent.

Primary structure of the Streptomyces R61 extracellular DD-peptidase. 2. Amino acid sequence data.

In order to confirm the Streptomyces codon usage, the Streptomyces R61 DD-peptidase was fragmented by cyanogen bromide cleavage of the carboxymethylated protein, trypsin digestion of the carboxymethylated protein and trypsin digestion of the protein treated with beta-iodopenicillinate and endoxo-delta 4-tetrahydrophthalic acid. The isolated peptides, which altogether represented more than 50% of the polypeptide chain, were sequenced. The data thus obtained were in excellent agreement with the primary structure of the protein as deduced from the nucleotide sequence of the cloned gene. Though a weak acylating agent, beta-iodopenicillanate reacted selectively with the active site of the DD-peptidase and formed an adduct which mas much more stable than that formed with benzylpenicillin, thus facilitating the isolation and characterization of the active-site peptide.

Mode of interaction between beta-lactam antibiotics and the exocellular DD-carboxypeptidase--transpeptidase from Streptomyces R39.

The exocellular DD-carboxypeptidase-transpeptidase of Streptomyces R39 is inhibited by beta-lactam antibiotics according to the same general scheme of reaction as the exocellular DD-carboxypeptidase-transpeptidase of Streptomyces R61. However, the values for the kinetic constants involved in the reaction are very different for the two enzymes and provide an explanation for the observation that the R39 enzyme is more sensitive to beta-lactam antibiotics than the R61 enzyme. Further, particular beta-lactams influence the kinetic constants to different extents depending on the source of the enzyme, so that a physical basis for the spectrum of antibiotic activity against particular enzyme systems is provided.

Mode of action of a structurally novel beta-lactam.

CP-35,587 {6-[d-2-amino-2-(p-hydroxyphenyl)acetamido]-2,2-dimethyl-3- (5-tetrazolyl)-penam} is a member of a new family of beta-lactam antibacterial agents. Because the biospectrum of CP-35,587 has features similar to both penicillins and cephalosporins, experiments were carried out to explore its mode of action. CP-35,587 did not inhibit peptidoglycan transpeptidase from Escherichia coli, but it did inhibit dd-carboxypeptidase. Similar results were obtained with cephalexin. When these antibiotics were compared for effects on growth and morphology of E. coli, it was observed that filaments formed after exposure to either antibacterial agent for 30 to 60 min. The filaments enlarged, and fragmentation occurred until very few viable cells remained after exposure to CP-35,587. After 180 min in the presence of cephalexin, the cells began to divide, and the filaments formed cross-walls reaching control values in 24 h. CP-35,587 and cephalexin had similar effects on the morphology of the Klebsiella cell: the cells became enlarged within 30 min; with increasing exposure, the filaments became longer, with evidence of cytoplasmic emptying and ghost cell formation. These ghostlike tubules eventually broke apart, leaving fragments. These data indicate differences in the mode of action of CP-35,587 from those of most other beta-lactam antibiotics.

Effects of sulfhydryl reagents on the binding and release of penicillin G by D-alanine carboxypeptidase IA of Escherichia coli.

Purified D-alanine carboxypeptidase IA of Escherichia coli is inhibited by penicillin G and binds penicillin G reversibly. The binding of penicillin to the enzyme is relatively insensitive to sulfhydryl reagents, while release of penicillin from the enzyme is severely inhibited by these reagents. The inhibition of release parallels the inhibition of carboxypeptidase activity by the sulfhydryl reagents. In the presence of the sulfhydryl reagent p-chloromercuribenzoate, an acyl-enzyme intermediate, produced by the reaction of carboxypeptidase IA with diacetyl-L-lysyl-D-alanyl-D-alanine, accumulates and can be isolated. These results indicate that binding of penicillin to carboxypeptidase IA occurs by an acylation step of the carboxypeptidase reaction, while penicillin release occurs by a deacylation step of the reaction. Only the latter is inhibited by sulfhydryl reagents.