An amino acid substitution that blocks the deacylation step in the enzyme mechanism of penicillin-binding protein 5 of Escherichia coli

A mutant of Escherichia coli has been described that produces an altered form of penicillin-binding protein 5 which still binds penicillin but is unable to catalyse the release of the bound penicilloyl moiety. We show that the mutation is caused by a single nucleotide transition that results in a change from glycine at residue 105 of the wild-type sequence of penicillin-binding protein 5 to aspartate in the mutant.     

Contribution of membrane-binding and enzymatic domains of penicillin binding protein 5 to maintenance of uniform cellular morphology of Escherichia coli

Four low-molecular-weight penicillin binding proteins (LMW PBPs) of Escherichia coli are closely related and have similar DD-carboxypeptidase activities (PBPs 4, 5, and 6 and DacD). However, only one, PBP 5, has a demonstrated physiological function. In its absence, certain mutants of E. coli have altered diameters and lose their uniform outer contour, resulting in morphologically aberrant cells. To determine what differentiates the activities of these LMW PBPs, we constructed fusion proteins combining portions of PBP 5 with fragments of other DD-carboxypeptidases to see which hybrids restored normal morphology to a strain lacking PBP 5. Functional complementation occurred when truncated PBP 5 was combined with the terminal membrane anchor sequences of PBP 6 or DacD. However, complementation was not restored by the putative carboxy-terminal anchor of PBP 4 or by a transmembrane region of the osmosensor protein ProW, even though these hybrids were membrane bound. Site-directed mutagenesis of the carboxy terminus of PBP 5 indicated that complementation required a generalized amphipathic membrane anchor but that no specific residues in this region seemed to be required. A functional fusion protein was produced by combining the N-terminal enzymatic domain of PBP 5 with the C-terminal beta-sheet domain of PBP 6. In contrast, the opposite hybrid of PBP 6 to PBP 5 was not functional. The results suggest that the mode of PBP 5 membrane anchoring is important, that the mechanism entails more than a simple mechanical tethering of the enzyme to the outer face of the inner membrane, and that the physiological differences among the LMW PBPs arise from structural differences in the DD-carboxypeptidase enzymatic core.     

Effects of nucleophiles on the breakdown of the benzylpenicilloyl-enzyme complex EI formed between benzylpenicillin and the exocellular DD-carboxypeptidase--transpeptiase of Streptomyces strain R61.

Serine is one of the enzyme residues with which benzylpenicillin collides as a result of its binding to the Streptomyces strain-R61 DD-carboxypeptidase-transpeptidase enzyme. Nucleophilic attack occurs on C(7) of the bound antibiotic molecule with formation of a benzylpenicilloyl-serine ester linkage, i.e. formation of the benzylpenicilloyl-enzyme EI complex. To reject the bound penicilloyl moiety and consequently to recover its initial activities, the strain-R61 enzyme has developed two possible mechanisms. Pathway A is a direct attack of the serine ester linkage by an exogenous nucleophile, resulting in the transfer of the benzylpenicilloyl moiety to this nucleophile. In pathway B, the benzylpenicilloyl moiety is first fragmented by C(5)-C(6) cleavage and the enzyme-bound phenylacetylglycyl residue thus produced is in turn transferred to the nucleophile. Pathway B occurs with water, glycylglycine and other amino compounds. Both pathways A and B occur with glycerol, other ROH nucleophiles and neutral hydroxylamine. The nucleophilic attacks are enzyme-catalysed.     

Contributions of PBP 5 and DD-carboxypeptidase penicillin binding proteins to maintenance of cell shape in Escherichia coli

Escherichia coli has 12 recognized penicillin binding proteins (PBPs), four of which (PBPs 4, 5, and 6 and DacD) have DD-carboxypeptidase activity. Although the enzymology of the DD-carboxypeptidases has been studied extensively, the in vivo functions of these proteins are poorly understood. To explain why E. coli maintains four independent loci encoding enzymes of considerable sequence identity and comparable in vitro activity, it has been proposed that the DD-carboxypeptidases may substitute for one another in vivo. We tested the validity of this equivalent substitution hypothesis by investigating the effects of these proteins on the aberrant morphology of DeltadacA mutants, which produce no PBP 5. Although cloned PBP 5 complemented the morphological phenotype of a DeltadacA mutant lacking a total of seven PBPs, controlled expression of PBP 4, PBP 6, or DacD did not. Also, a truncated PBP 5 protein lacking its amphipathic C-terminal membrane binding sequence did not reverse the morphological defects and was lethal at low levels of expression, implying that membrane anchoring is essential for the proper functioning of PBP 5. By examining a set of mutants from which multiple PBP genes were deleted, we found that significant morphological aberrations required the absence of at least three different PBPs. The greatest defects were observed in cells lacking, at minimum, PBPs 5 and 6 and one of the endopeptidases (either PBP 4 or PBP 7). The results further differentiate the roles of the low-molecular-weight PBPs, suggest a functional significance for the amphipathic membrane anchor of PBP 5 and, when combined with the recently determined crystal structure of PBP 5, suggest possible mechanisms by which these PBPs may contribute to maintenance of a uniform cell shape in E. coli.     

Penicillin-binding proteins of Thiobacillus versutus

The cytoplasmic membrane of Thiobacillus versutus was found to contain at least nine penicillin-binding proteins (PBPs) with apparent molecular weights as judged by sodium dodecyl sulphate polyacrylamide slab gel electrophoresis of 87000 (PBP1), 81000 (PBP2), 68000 (PBP3), 63000 (PBP4), 57000 (PBP5), 40000 (PBP6), 37000 (PBP70, 33000 (PBP8) and 31000 (PBP9). The PBP pattern of T. versutus was thus quite different from that of the Enterobacteria and the Pseudomonads. Also the properties of the PBPs of T. versutus such as affinity for various beta-lactam antibiotics, heat stability and release of bound penicillin were different from similar properties of Escherichia coli, Pseudomonas aeruginosa and other gram-negative bacteria.     

Hydroxylaminolysis of penicillin binding componenets is enzymatically catalyzed.

The hydroxylaminolysis of the penicilloyl moiety from [14C]penicillin G binding component (PBC) complexes of the Bacillus subtilis D-alanine carboxypeptidase and of the mixture of PBC's of Staphylococcus aureus was inhibited by denaturation of the complexes by heat (55 degrees), detergent (1% sodium dodecyl sulfate), or trichloroacetic acid. The kinetics of inhibition by denaturation were comparable to those of the inhibition of [14C]penicillin G binding to the PBC's and of carboxypeptidase activity of the B. subtilis enzyme under identical denaturing conditions. These data establish that the hydroxylaminolysis is an enzymatically catalyzed process suggesting that penicillin G is bound to an enzymatically active site. Treatment of the denatured [14C]penicillin G-carboxypeptidase complex with sodium borohydride or at pH 12 resulted in the release of the penicilloyl moiety. These results are consistent with a carboxylic ester bond for the penicilloyl-PBC instead of a thiolester linkage as was initially presumed.     

Penicillin-binding proteins of listeria monocytogenes--a re-evaluation

Intact Listeria monocytogenes cells or membranes isolated from them were treated with [3H]penicillin to allow identification of the penicillin binding proteins (PBPs) located in the cytoplasmic membrane. In the former case the PBPs were released from the cells following disruption of the cell wall murein with Listeria monocytogenes bacteriophage lysin. The procedure described by Dougherty et al. (1996) for Escherichia coli, with some modifications, was used to evaluate the M(r)s of the individual PBPs and allowed direct quantitation of their copy number.     

Cephalosporin-sensitive penicillin-binding proteins of Staphylococcus aureus and Bacillus subtilis active in the conversion of [14C]penicillin G to [14C]phenylacetylglycine.

Breakdown of the covalent complex formed between [14C]penicillin G and higher molecular weight, cephalosporin-sensitive penicillin-binding proteins was studied using mixtures of the purified proteins isolated from membranes of Staphylococcus aureus and Bacillus subtilis. These penicillin-binding proteins were found to release the bound 14C label in a first order process characterized by half-lives of 10 to 300 min at 37 degrees C. Denaturation of the penicilloyl.penicillin-binding proctein complex prevented this release, indicating that the process is enzyme-catalyzed. [14C]Phenylacetylglycine was identified as the major labeled fragmentation product, indicating that these cephalosporin-sensitive penicillin-binding proteins, for which no in vitro transpeptidase or carboxypeptidase activity has been found, catalyze the same fragmentation of the bound penicilloyl moiety previously described for several penicillin-sensitive D-alanine carboxypeptidases.     

Studies of the high molecular weight penicillin-binding proteins of Bacillus subtilis.

Seven or eight penicillin-binding proteins (PBPs) were detected in Bacillus subtilis membranes. By introducing covalent affinity chromatography employing cephalosporins as ligands, milligram amounts of three high molecular weight PBPs (PBP 1 ab, Mr = 120,000; PBP 2b, Mr = 94,000; and PBP 4, Mr = 78,000) were obtained without any contamination of the major PBP 5, the D-alanine carboxypeptidase. Small amounts of pure PBP 2b could be isolated by manipulation of the affinity chromatography conditions. Structural and physical properties of these proteins as well as the generation of one major penicilloyl peptide from each PBP by digestion with pepsin suggest that each PBP is the product of a separate gene. No enzymatic activity could be found in mixtures of these high molecular weight PBPs employing substrates used for the transpeptidase and D-alanine carboxypeptidase assays in particulate membrane fractions.     

Release and recycling of eukaryotic initiation factor 2 in the formation of an 80 S ribosomal polypeptide chain initiation complex.

The eukaryotic initiation factor (eIF)-5 mediates hydrolysis of GTP bound to the 40 S initiation complex in the absence of 60 S ribosomal subunits. The eIF-2.GDP formed under these conditions is released from the 40 S ribosomal subunit while initiator Met-tRNA(f) remains bound. The released eIF-2.GDP can participate in an eIF-2B-catalyzed GDP/GTP exchange reaction to reform the Met-tRNA(f).eIF-2.GTP ternary complex. In contrast, when 60 S ribosomal subunits were also present in an eIF-5-catalyzed reaction, the eIF-2.GDP produced remained bound to the 60 S ribosomal subunit of the 80 S initiation complex. When such an 80 S initiation complex, containing bound eIF-2.GDP, was incubated with GTP and eIF-2B, GDP was released. However, eIF-2 still remained bound to the ribosomes and was unable to form a Met-tRNA(f)l.eIF-2.GTP ternary complex. In contrast, when 60 S ribosomal subunits were preincubated with either free eIF-2 or with eIF-2.eIF-2B complex and then added to a reaction containing both the 40 S initiation complex and eIF-5, the eIF-2.GDP produced did not bind to the 60 S ribosomal subunits but was released from the ribosomes. Thus, the 80 S initiation complex formed under these conditions did not contain bound eIF-2.GDP. Under similar experimental conditions, preincubation of 60 S ribosomal subunits with purified eIF-2B (free of eIF-2) failed to cause release of eIF-2.GDP from the ribosomal initiation complex. These results suggest that 60 S ribosome-bound eIF-2.GDP does not act as a direct substrate for eIF-2B-mediated release of eIF-2 from ribosomes. Rather, the affinity of 60 S ribosomal subunits for either eIF-2, or the eIF-2 moiety of the eIF-2.eIF-2B complex, prevents association of 60 S ribosomal subunits with eIF-2.GDP formed in the initiation reaction. This ensures release of eIF-2 from ribosomes following hydrolysis of GTP bound to the 40 S initiation complex.     

Separation of Escherichia coli penicillin-binding proteins into different membrane vesicles by agarose electrophoresis and sizing chromatography.

Membrane vesicles from the envelope of Escherichia coli were separated by electrophoresis through dilute agarose and by sizing chromatography through Sephacryl S-1000. These techniques revealed that proteins were associated with different subsets of vesicles. In particular, dilute agarose electrophoresis clearly separated the inner membrane penicillin-binding proteins (PBPs) into different vesicle groups. Vesicles containing PBPs 4, 6, 7, and 8 migrated rapidly through agarose; vesicles with PBPs 1a, 1b, 2, 3, and 5 eluted later. With the exception of PBP 4, which migrated with PBPs 1 through 5, chromatography through Sephacryl S-1000 was able to distinguish the same two vesicle sets, though the extent of separation was poorer than with agarose. The existence of these associations among vesicles and proteins suggests that there is an organization to the inner membrane of E. coli which is not observed when membrane vesicles are separated solely on the basis of density in sucrose gradients.     

Penicillin-binding site on the Escherichia coli cell envelope.

The binding of 35S-labeled penicillin to distinct penicillin-binding proteins (PBPs) of the "cell envelope" obtained from the sonication of Escherichia coli was studied at different pHs ranging from 4 to 11. At low pH, PBPs 1b, 1c, 2, and 3 demonstrated the greatest amount of binding. At high pH, these PBPs bound the least amount of penicillin. PBPs 1a and 5/6 exhibited the greatest amount of binding at pH 10 and the least amount at pH 4. With the exception of PBP 5/6, the effect of pH on the binding of penicillin was direct. Experiments distinguishing the effect of pH on penicillin binding by PBP 5/6 from its effect on beta-lactamase activity indicated that although substantial binding occurred at the lowest pH, the amount of binding increased with pH, reaching a maximum at pH 10. Based on earlier studies, it is proposed that the binding at high pH involves the formation of a covalent bond between the C-7 of penicillin and free epsilon amino groups of the PBPs. At pHs ranging from 4 to 8, position 1 of penicillin, occupied by sulfur, is considered to be the site that establishes a covalent bond with the sulfhydryl groups of PBP 5. The use of specific blockers of free epsilon amino groups or sulfhydryl groups indicated that wherever the presence of each had little or no effect on the binding of penicillin by PBP 5, the presence of both completely prevented binding. The specific blocker of the hydroxyl group of serine did not affect the binding of penicillin. These observations suggest that a molecule of penicillin forms simultaneous bonds between its S at position 1 and sulfhydryl groups of PBP 5 and between its C-7 and free epsilon amino groups of PBP 5.     

Changes in peptidoglycan composition and penicillin-binding proteins in slowly growing Escherichia coli.

The composition of peptidoglycan of chemostat-grown cultures of Escherichia coli was investigated as a function of growth rate. As the generation time was lengthened from 0.8 to 13.8 h, there was a decrease in the major monomer (disaccharide tetrapeptide) and dimer (bis-disaccharide tetrapeptide), while disaccharide tripeptide moieties increased to greater than 50% of the total wall. The average chain length became much shorter; lipoprotein density tripled, and the number of unusual diaminopimelyl-diaminopimelic acid crossbridges increased fivefold. As cells grew more slowly, amounts of penicillin-binding proteins (PBPs) 1a-1b complex and 4 decreased, while amounts of PBPs 3 and the 5-6 complex increased. We propose that the chemical composition of E. coli cell walls changes with growth rate in a manner consistent with alterations in the activities of PBPs and cell shape.     

The dnaB-dnaC replication protein complex of Escherichia coli. I. Formation and properties

The complex formed between the dnaB and dnaC replication proteins of Escherichia coli is stabilized by ATP binding to dnaC. The dnaB6-dnaC6-ATP6 complex can be maintained without ATP hydrolysis at a concentration as low as 5 x 10(-10) M. The complex is also formed with adenosine 5'-(gamma-thio)triphosphate but generates little or no dnaB activity, suggesting a requirement for ATP hydrolysis in the subsequent stage of binding of the complex to DNA. In this step, dnaC is released, leaving dnaB to function on the associated DNA.     

Sequence of active site peptides from the penicillin-sensitive D-alanine carboxypeptidase of Bacillus subtilis. Mechanism of penicillin action and sequence homology to beta-lactamases

It has been proposed that penicillin and other beta-lactam antibiotics are substrate analogs which inactivate certain essential enzymes of bacterial cell wall biosynthesis by acylating a catalytic site amino acid residue (Tipper, D.J., and Strominger, J.L. (1965) Proc. Natl. Acad. Sci. U.S.A. 54, 1133-1141). A key prediction of this hypothesis, that the penicilloyl moiety and an acyl moiety derived from substrate both bind to the same active site residue, has been examined. D-Alanine carboxypeptidase, a penicillin-sensitive membrane enzyme, was purified from Bacillus subtilis and labeled covalently at the antibiotic binding site with [14C]penicillin G or with the cephalosporin [14C]cefoxitin. Alternatively, an acyl moiety derived from the depsipeptide substrate [14C]diacetyl L-Lys-D-Ala-D-lactate was trapped at the catalytic site in near-stoichiometric amounts by rapid denaturation of an acyl-enzyme intermediate. Radiolabeled peptides were purified from a pepsin digest of each of the 14C-labeled D-alanine carboxypeptidases and their amino acid sequences determined. Antibiotic- and substrate-labeled peptic peptides had the same sequence: Tyr-Ser-Lys-Asn-Ala-Asp-Lys-Arg-Leu-Pro-Ile-Ala-Ser-Met. Acyl moieties derived from antibiotic and from substrate were shown to be bound covalently in ester linkage to the identical amino acid residue, a serine at the penultimate position of the peptic peptide. These studies establish that beta-lactam antibiotics are indeed active site-directed acylating agents. Additional amino acid sequence data were obtained by isolating and sequencing [14C]penicilloyl peptides after digestion of [14C]penicilloyl D-alanine carboxypeptidase with either trypsin or cyanogen bromide and by NH2-terminal sequencing of the uncleaved protein. The sequence of the NH2-terminal 64 amino acids was thus determined and the active site serine then identified as residue 36. A computer search for homologous proteins indicated significant sequence homology between the active site of D-alanine carboxypeptidase and the NH2-terminal portion of beta-lactamases. Maximum homology was obtained when the active site serine of D-alanine carboxypeptidase was aligned correctly with a serine likely to be involved in beta-lactamase catalysis. These findings provide strong evidence that penicillin-sensitive D-alanine carboxypeptidases and penicillin-inactivating beta-lactamases are related evolutionarily.     

Expression and characterization of the ponA (ORF I) gene of Haemophilus influenzae: functional complementation in a heterologous system.

The coding sequence of the Haemophilus influenzae ORF I gene was amplified by PCR and cloned into different Escherichia coli expression vectors. The ORF I-encoded protein was approximately 90 kDa and bound 3H-benzyl-penicillin and 125I-cephradine. This high-molecular-weight penicillin-binding protein (PBP) was also shown to possess transglycosylase activity, indicating that the ORF I product is a bifunctional PBP. The ORF I protein was capable of maintaining the viability of E. coli delta ponA ponB::spcr cells in transcomplementation experiments, establishing the functional relevance of the significant amino acid homology seen between E. coli PBP 1A and 1B and the H. influenzae ORF I product. In addition, the physiological functioning of the H. influenzae ORF I (PBP 1A) product in a heterologous species established the ability of the enzyme not only to recognize the E. coli substrate but also to interact with heterologous cell division proteins. The affinity of the ORF I product for 3H-benzylpenicillin and 125I-cephradine, the MIC of beta-lactams for E. coli delta ponA ponB::spcr expressing the ORF I gene, and the amino acid alignment of the PBP 1 family of high-molecular-weight PBPs group the ORF I protein into the PBP 1A family of high-molecular-weight PBPs.     

Inhibition of bacterial DD-peptidases (penicillin-binding proteins) in membranes and in vivo by peptidoglycan-mimetic boronic acids

The DD-peptidases or penicillin-binding proteins (PBPs) catalyze the final steps of bacterial peptidoglycan biosynthesis and are inhibited by the β-lactam antibiotics. There is at present a question of whether the active site structure and activity of these enzymes is the same in the solubilized (truncated) DD-peptidase constructs employed in crystallographic and kinetics studies as in membrane-bound holoenzymes. Recent experiments with peptidoglycan-mimetic boronic acids have suggested that these transition state analogue-generating inhibitors may be able to induce reactive conformations of these enzymes and thus inhibit strongly. We have now, therefore, measured the dissociation constants of peptidoglycan-mimetic boronic acids from Escherichia coli and Bacillus subtilis PBPs in membrane preparations and, in the former case, in vivo, by means of competition experiments with the fluorescent penicillin Bocillin Fl. The experiments showed that the boronic acids bound measurably (K(i) < 1 mM) to the low-molecular mass PBPs but not to the high-molecular mass enzymes, both in membrane preparations and in whole cells. In two cases, E. coli PBP2 and PBP5, the dissociation constants obtained were very similar to those obtained with the pure enzymes in homogeneous solution. The boronic acids, therefore, are unable to induce tightly binding conformations of these enzymes in vivo. There is no evidence from these experiments that DD-peptidase inhibitors are more or less effective in vivo than in homogeneous solution.     

Hydrolysis of human high-molecular-mass kininogen by human plasma kallikrein. Proposal of a new model concept for the course of reaction in presence and absence of C1(-)-inhibitor

Hydrolysis of high-molecular-mass kininogen was studied by following the changes in the amounts of substrate, intermediates and products as a function of time using quantitative polyacrylamide-gel electrophoresis (silver staining). The experimental data was analysed on the basis of the concept that the overall reaction is composed of three hydrolysis reactions, two positional-change processes of intermediates at the active site, and two product-substrate exchange processes. It is proposed C1(-)-inhibitor to form two types of complexes with kallikrein, one with non-covalent and one with covalent bonds. With an adequately chosen set of reaction-partner concentrations and four different kinds of experimental conditions with respect to kininogen and inhibitor addition to kallikrein, the following results were obtained: 1) Non-covalently bound inhibitor has no effect on the first and the second hydrolysis reaction, but efficiently interferes with the third hydrolysis reaction; 2) Nicked kininogen (first intermediate; one of the two bradykinin bonds split) for the second bond to be hydrolysed undergoes a positional change during which it remains strongly bound to the enzyme, never exchanges with kininogen, and is not displaced by non-covalently bound inhibitor; 3) Intermediate kinin-free kininogen (second intermediate; both bradykinin bonds split and bradykinin released) prior to turning over into stable kinin-free kininogen (final product; histidine-rich fragment split off and released) undergoes a positional change involving dissociation and reassociation so that non-covalently bound inhibitor finds access to the active site; 4) Intermediate kinin-free kininogen to sustain multiple turnovers exchanges with kininogen via a stable complex of such structure that during this process non-covalently bound inhibitor cannot or can only slightly interfere; 5) Stable kinin-free kininogen to sustain multiple turnovers exchanges with intermediate kinin-free kininogen via free enzyme with the effect that non-covalently bound inhibitor efficiently interferes; 6) As hydrolysis proceeds more and more inhibitor becomes covalently bound, gradually leading to complete inactivation of the enzyme.     

Penicillin binding protein 5 affects cell diameter, contour, and morphology of Escherichia coli

Although general physiological functions have been ascribed to the high-molecular-weight penicillin binding proteins (PBPs) of Escherichia coli, the low-molecular-weight PBPs have no well-defined biological roles. When we examined the morphology of a set of E. coli mutants lacking multiple PBPs, we observed that strains expressing active PBP 5 produced cells of normal shape, while mutants lacking PBP 5 produced cells with altered diameters, contours, and topological features. These morphological effects were visible in untreated cells, but the defects were exacerbated in cells forced to filament by inactivation of PBP 3 or FtsZ. After filamentation, cellular diameter varied erratically along the length of individual filaments and many filaments exhibited extensive branching. Also, in general, the mean diameter of cells lacking PBP 5 was significantly increased compared to that of cells from isogenic strains expressing active PBP 5. Expression of cloned PBP 5 reversed the effects observed in DeltadacA mutants. Although deletion of PBP 5 was required for these phenotypes, the absence of additional PBPs magnified the effects. The greatest morphological alterations required that at least three PBPs in addition to PBP 5 be deleted from a single strain. In the extreme cases in which six or seven PBPs were deleted from a single mutant, cells and cell filaments expressing PBP 5 retained a normal morphology but cells and filaments lacking PBP 5 were aberrant. In no case did mutation of another PBP produce the same drastic morphological effects. We conclude that among the low-molecular-weight PBPs, PBP 5 plays a principle role in determining cell diameter, surface uniformity, and overall topology of the peptidoglycan sacculus.     

Nucleotide sequence of the pbpA gene and characteristics of the deduced amino acid sequence of penicillin-binding protein 2 of Escherichia coli K12

We have determined the nucleotide sequence of the pbpA gene encoding penicillin-binding protein (PBP) 2 of Escherichia coli. The coding region for PBP 2 was 1899 base pairs in length and was preceded by a possible promoter sequence and two open reading frames. The primary structure of PBP 2, deduced from the nucleotide sequence, comprised 633 amino acid residues. The relative molecular mass was calculated to be 70867. The deduced sequence agreed with the NH2-terminal sequence of PBP 2 purified from membranes, suggesting that PBP 2 has no signal peptide. The hydropathy profile suggested that the NH2-terminal hydrophobic region (a stretch of 25 non-ionic amino acids) may anchor PBP 2 in the cytoplasmic membrane as an ectoprotein. There were nine homologous segments in the amino acid sequence of PBP 2 when compared with PBP 3 of E. coli. The active-site serine residue of PBP 2 was predicted to be Ser-330. Around this putative active-site serine residue was found the conserved sequence of Ser-Xaa-Xaa-Lys, which has been identified in all of the other E. coli PBPs so far studied (PBPs 1A, 1B, 3, 5 and 6) and class A and class C beta-lactamases. In the higher-molecular-mass PBPs 1A, 1B, 2 and 3, Ser-Xaa-Xaa-Lys-Pro was conserved. In the putative peptidoglycan transpeptidase domain there were six amino acid residues, which are common only in the PBPs of higher molecular mass.