Peptidoglycan synthetic activities in membranes of Escherichia coli caused by overproduction of penicillin-binding protein 2 and rodA protein.

Penicillin-binding protein (PBP)-2 and the RodA protein are known to function in determining the rod shape of Escherichia coli cells. Peptidoglycan biosynthetic reactions that required these two proteins were demonstrated in the membrane fraction prepared from an E. coli strain that overproduced both of these two proteins and which lacked PBP-1B activity (the major peptidoglycan synthetase activity in the normal E. coli membranes). The cross-linked peptidoglycan was synthesized from UDP-N-acetylmuramylpentapeptide and UDP-N-acetylglucosamine in the presence of a high concentration of cefmetazole that inhibited all of PBPs except PBP-2. The peptidoglycan was synthesized via a lipid intermediate and showed up to 30% cross-linking. The cross-linking reaction was strongly inhibited by the amidinopenicillin, mecillinam, and by other beta-lactam antibiotics that have a high affinity for PBP-2, but not by beta-lactams that had very low affinity for PBP-2. The formation of peptidoglycan required the presence of high levels of both PBP-2 and the RodA protein in the membranes, but it is unclear which of the two proteins was primarily responsible for the extension of the glycan chains (transglycosylation). However, the sensitivity of the cross-linking reaction to specific beta-lactam antibiotics strongly suggested that it was catalyzed by PBP-2. The transglycosylase activity of the membranes was sensitive to enramycin and vancomycin and was unusual in being stimulated greatly by a high concentration of a chelating agent.

II Present address: Fuiisawa Pharmaceutical Co. Research Institute, Tsukuba, Japan. " cell wall, for the formation of the septum, and for the determination of the bacterial cell shape. /?-Lactam antibiotics interfere with peptidoglycan synthesis and have been shown to produce specific morphological effects in Escherichia coli and other related Gram-negative bacteria by binding to the high molecular weight penicillin-binding proteins (PBPsl) (1)(2)(3)(4)(5) . Four of the high molecular weight PBPs of E. coli have been proposed to be primarily involved in peptidoglycan synthesis during the cell cycle. PBP-1A and -lB (Mr approximately 90,000) are involved in cell elongation, PBP-2 ( M , 70,000) in the determination of cell shape, and PBP-3 ( M , 60,000) in septum formation. Recent work suggests that each of these PBPs is a peptidoglycan synthetase and that they act together within the cytoplasmic membrane to catalyze the duplication of the bag-shaped peptidoglycan network (6)(7)(8)(9)(10)(11)(12)(13)(14)(15).
In 1966 Izaki et al. (16) demonstrated a transglycosylaseand a penicillin-sensitive DD-transpeptidase activity in a crude cell membrane fraction of E. coli that together synthesized cross-linked peptidoglycan from lipid-linked precursors. Subsequently it was found (3) that membranes prepared from a mutant that lacked PBP-1B activity failed to catalyze either the transglycosylase or transpeptidase reactions, and this unexpected result was subsequently explained by the discovery that PBP-1B catalyzed both of these reactions (6)(7)(8). The very low level of peptidoglycan synthetic activity in cell membranes prepared from a mutant of E. coli that lacks PBP-1B has allowed us to search for the activities catalyzed by the other high molecular weight PBPs. In this report we describe our investigation on the enzymatic activity of PBP-2. We show that membranes prepared from a strain of E. coli that lacks PBP-1B and which greatly overproduces PBP-2 had very low levels of peptidoglycan synthetic activity. However, if these membranes also contained high levels of the RodA protein (the product of the cell shape gene, mrdB (or rodA)), they catalyzed the following reactions of peptidoglycan synthesis.
[ Escherichia coli Strains and Construction of Plasmids-Strain JST975 lacking PBP-1B (mrcB) was derived from strain JElOll (Fthr leu trp his thy thi ara lac gal xyl mtl rpsL tomi) as described (3). Strain JST9753 (F-lip-9 mrcB proA purB his metB lac gal rpsL) was constructed by mating AB1325 (F-lip-9 proA purB his thi lacy galK xyl mtl rpsL) with CD49751 (HfrC mrcB proA metB lac muk4 tsx) (3) selecting for mtl+ rpsL. Strain JST9753 was transduced to lip+ at 30 "C using Xdlip5cI857Qam73, and lysogens were identified by their immunity to XcI and their thermosensitivity at 43 "C. Xdlip5 is a defective transducing phage carrying the E. coli chromosomal genes from lip to leuS including dacA, mrdE(rodA), mrdA(pbpA), and several other genes (17). The plasmid pHS202, which carries the intact mrdB and mrdA genes and two open reading frames coding for 17-and 7.7-kDa proteins, was constructed by subcloning the 7.2kilobase SalI fragment from XMAdlip24 (18), a defective transducing phage similar to Xdlip5, into the SalI site of pACYC184. pHs503 and pHs504 are derivatives of pHs202 that have the mrdB and mrdA genes, respectively, inactivated by the insertion of Tn5 (18). pHs506 is a derivative of pHs202 that has an insertion of Tn5 which does not inactivate either the mrdB or mrdA genes (18). pTP51, which expresses wild type RodA protein and thermosensitive PBP-2, and pTP71, which expresses thermosensitive RodA protein and wild type PBP-2, were constructed as described below. First, strain TMM3 (lip mrdA3) (19), which produces a thermolabile form of PBP-2, and strain TMM4 (lip mrdB4) (19), which presumably produces a thermolabile form of the RodA protein, were transduced to lip+ using P1 phage to produce strains TMM23 and TMM24, respectively, and were then made A(att-bio). Specialized transducing phage carrying the lip-kuS region were then isolated from the latter strains essentially as described previously (17). The 7.2-kilobase SalI fragment carrying the wild type or mutant mrdA(pbpAf and mrdB(rodA) genes was cloned from each of the resulting transducing phage into the Sa11 site of pACYC184 to produce pTPBl(mrdA3 mrdB+) and pTP7l(mrdA+ mrdB4) as shown in Fig. 1. The strains and plasmids are summarized in Table I. Growth of Cells and Preparation of Membranes-Cells were grown in a modified Lennox broth (20) containing 1% polypeptone, 0.5%   yeast extract, 0.1% glucose, 0.5% NaC1, and 20 pg/ml thymine, adjusted to pH 7.0 with NaOH (L' broth). Overproduction of proteins encoded by Xdlip5cI857Qam73 was achieved by heat induction of strain JST9753 lysogenized with the phage as described previously (12). High levels of plasmid-encoded proteins were achieved by growing the plasmid-containing strains at 30 "C to an A550 of 0.3, adding spectinomycin (170 pg/ml), and continuing growth overnight. The cells were washed with 0.2 volume of saline and were resuspended in the original volume of fresh medium (lacking spectinomycin) and grown for 2 h at 30 "C to allow expression of the cell shape genes on the amplified plasmids. Cells were harvested by centrifugation, washed once with 0.05 M Tris-HC1 buffer, pH 7.5, containing 0.1 mM MgCl, (buffer A), and stored frozen at -80 "C. Bacterial membranes (the particulate membrane fraction) were prepared by disrupting the cells by sonication in buffer A (100 mg, wet weight, of cells/ml) at 0 "C centrifuging at 5,000 X g for 10 min to remove unbroken cells, and pelleting the membrane fraction at 100,000 X g for 30 min. The membranes were washed once in buffer A and resuspended in the same buffer. Detection of PEPS-The levels of PBPs in cell membranes were measured by labeling with ["CC]-or [3H]benzylpenicillin, extracting the cytoplasmic membrane proteins with Sarkosyl (sodium dodecyl sarcosinate, Ciba Geigy), and separating the proteins on a 7.5% sodium dodecyl sulfate-polyacrylamide gel as described (1). A fluorogram was prepared using the 2,5-diphenyloxazole method and prefogged Fuji RX x-ray film as described (21).
Enzyme Assays-The assays of transglycosylase and transpeptidase activities were normally performed in 0.5-ml Pyrex test tubes. The standard reaction mixture contained (in a final volume of 37 pl) 50-60 mM Tris-HC1 buffer, pH 8.5, 1-27 mM MgC12, 13.5 mM potassium EDTA (pH 8.5), 100-400 pg (as protein) of membranes, and, as substrates, 0.36 nmol of UDP-MurNAc-pentapeptide (-L-Ala-D-Glumeso-["C]A2pm-~-Ala-D-Ala) and 10 nmol of UDP-GlcNAc. The reaction mixture was incubated a t 37 "C for the time indicated, and then the reaction was stopped by boiling for 1 min. The mixture was subjected to paper chromatography using Whatman No. 3 MM filter paper and isobutyric acid, 1 M ammonia (1:0.6) as solvent. The radioactivity on the chromatogram was detected by autoradiography and was quantitated using a liquid scintillator (toluene/2,5-diphenyloxazole-1,4-bis[2-(5-phenyloxazolyl)]benzene, 500 ml:2 g:50 mg; counting efficiency of 75%). The determination of the extent of crosslinking was carried out by digestion of the radioactive peptidoglycan product with lysozyme, followed by separation of the products, and measuring the radioactivity in the uncross-linked monomer(s) and cross-linked dimer(s) of the repeating units of the peptidoglycan as described previously (8). The percentage of cross-linking was defined as 50 times the ratio of (radioactivity of cross-linked dimer(s)) to (radioactivity of uncross-linked monomer(s) plus cross-linked dimer(s)). Thus, if everything in the lysozyme digest was cross-linked dimer(s), the ratio was unity and the extent of cross-linking was defined to be 50%.

RodA Protein System in E. coli Membranes and Mecillinam
Sensitivity of the Transpeptidase Activity-The enzymatic activity of PBP-2 has been the subject of much speculation for many years because of the unique function attributed to this protein in the determination of the rod shape of the cell (1, 2). It has finally been established that PBP-2 is a peptidoglycan synthetase, but at least the RodA protein, the product of another cell shape gene, mrdB (or rodA), is required for the expression of its activity. Ishino et al. (12) demonstrated a sequence of reactions for the formation of crosslinked peptidoglycan from the UDP-linked precursors (UDP-MurNAc-pentapeptide (~-alanyl-~-glutamyl-meso-['~C]diaminopimelyl-D-alanyl-D-alanine) and UDP-GlcNAc) in membranes from E. coli JST9753 (mrcB-) cells that had been induced for a defective X prophage (Xdlip5cI857Qam73) carrying the chromosomal region covering leuS to lip (17). As a result of the thermoinduction of the phage, the cells produced a large amount of PBP-2 (the product of the mrdA(pbpA) gene) and of those proteins encoded by the other genes of the transducing phage, including the RodA protein, PBP-5 (the dacA product) and a 54-kDa protein (22) encoded by a gene located between mrdB(rodA) and dacA. The membranes thus contained, in addition to the high levels of PBP-2 and normal levels of PBPs lA, 3,4, and 6, a very large amount of PBP-5 which, because of its D-alanine carboxypeptidase activity, made it difficult to assay the enzymatic reactions catalyzed by PBP-2. Curtis and Strominger (23) have previously purified a small amount of PBP-2 on an affinity column of 6aminopenicillanic acid-CM-Sepharose from a crude mixture of E. coli PBPs, which had been pretreated with cefoxitin, a 7a-methoxycephalosporin. Cefoxitin bound to all of the PBPs, except PBP-2, which alone was adsorbed to the affinity column and could thus be purified. We extended this idea to demonstrate the enzymatic activities of PBP-2 in E. coli membranes by eliminating the enzymatic activity of all other PBPs by pretreatment with a 7a-methoxycephalosporin (5, 12) (cf. Fig. 2). Peptidoglycan was synthesized by the membranes of cells that overproduced PBP-2 and the other proteins expressed from the X transducing phage, in a reaction mixture containing 20 pg/ml 7a-methoxycephalosporin, cefmetazole (24) ( Table 11, Experiment 1). The peptidoglycan formed under these conditions was cross-linked to about 20%, and the only PBP that should be capable of catalyzing the cross-linking reaction under the conditions used in these experiments is PBP-2. The cross-linking was almost completely inhibited by 6.4 pg/ml benzylpenicillin (50% inhibition by 1 pg/ml) and was completely inhibited by 1 pg/ml mecillinam (50% inhibition by 0.2 pg/ml), an amidinopenicillin that binds only to PBP-2 and results in the growth of E. coli as spherical cells (1). The sensitivity of the cross-linking reaction to low concentrations of mecillinam is strong evidence for the involvement of PBP-2 in the reaction since this protein is the only known target of the antibiotic. Other compounds that have high affinity for PBP-2, for example, N-formimidoylthienamycin, also strongly inhibited this transpeptidation reaction (data not shown). However, @-lactams that fail to bind to PBP-2 had no effect, or only a slightly inhibitory effect, on the transpeptidation reaction

by PBP-2 and RodA
Protein 7027  the presence of 20 pg/ml cefmetazole. The peptidoglycan that was formed was 25-30% cross-linked. The absence of Dalanine carboxypeptidase activity in these membranes allowed the transpeptidase reaction to be followed by the release of D-alanine. If UDP-MurNAc-pentapeptide labeled in D-["Cc] Ala-~-['~C]Ala was used, release of D-[14C]Ala could be observed and was inhibited by appropriate amounts of mecillinam and other p-lactams (Table 11, Experiment 2). The transglycosylase activity was inhibited significantly by 5 pg/ ml enramycin (Table 11, Experiment 3) and by 50 pg/ml vancomycin (50% inhibition by 5 pg/ml), but moenomycin (30 pg/ml) had almost no effect. Tunicamycin (2.8 pg/ml) was significantly inhibitory, indicating that the formation of the lipid intermediate was involved in the synthesis of the peptidoglycan (Table 11, Experiment 4).
Both the cross-linked peptidoglycan product and the uncross-linked product synthesized in the presence of mecillinam formed condensed spots at the origin of the paper chromatogram where the heat-inactivated reaction mixture was applied, indicating that the products were insoluble in water in both cases (Fig. 3). The insolubility may have been due to very long glycan chains in the peptidoglycan that was synthesized.
Effect of a High Concentration of Magnesium Zon and a Chelating Agent-As described under "Experimental Procedures," the standard reaction mixture contained 13.5 mM potassium EDTA (either at pH 6.0 or 8.5) and 28 mM magnesium chloride. It is interesting that, for unknown reasons, a high concentration of chelating agent enhanced the synthesis of peptidoglycan in the membrane system. As shown in Table 111, the formation of peptidoglycan was strongly de-

Lipid-linked precursor
Origin - for 90 min at 37 "C in the presence of 50 pg/ml cefmetazole; the reaction mixtures were then boiled, treated with trypsin, and subjected to paper chromatography. Similar results were obtained using membranes from the heat-induced X lysogen. An autoradiogram of a paper chromatogram is shown. A, control; B, plus 1 pg/ml mecillinam; C, plus 50 pg/ml enramycin. pendent on the addition of magnesium ion, probably because of a requirement for this ion in the formation of lipid intermediates. However, more surprisingly, a stimulatory effect of potassium EDTA was observed when it was added in the presence of magnesium. The stimulation of peptidoglycan synthetic activity by the chelating agent was apparently due partly to the chelation of magnesium ions but partly to an unknown mechanism, as an increase in the effective concentration of magnesium ion from 4.6 to 27.8 mM only slightly effected the peptidoglycan synthetic activity, while the enhancement of the activity by increasing the concentration of the chelating agent was higher. The extent of the cross-linkage (transpeptidase activity) also decreased slightly by increasing in the effective concentrations of magnesium ion above 4.6 mM.
Lipid-linked Precursors-The synthesis of peptidoglycan from UDP-linked precursors was inhibited by tunicamycin (the specific inhibitor of the formation of lipid-linked MurNAc-pentapeptide (26)) suggesting that it occurred through the formation of lipid-linked intermediates ( Table 11, Experiment 4). This was further shown by a chasing experiment (Fig. 4). Incubation of membranes for 16 min with labeled UDP-MurNAc-pentapeptide (me~o-[~~C]diaminopimelic acid) and unlabeled UDP-GlcNAc resulted in the synthesis of labeled lipid-linked intermediates, a part of which were incorporated into peptidoglycan over a 60-min period following the addition of an excess of unlabeled UDP-MurNAc-pentapeptide. There was also an appreciable increase in the radioactivity at the position of UDP-MurNAcpentapeptide on the paper chromatogram during the chasing procedure which may have been due to the formation of the nucleotide from the monosaccharide-lipid intermediate by the reverse reaction. Alternatively, it could be a decomposition product of lipid intermediates with a similar chromatographic mobility of UDP-MurNAc-pentapeptide. In any case, only a minor part of the lipid intermediates formed in the first 16 min could be converted into peptidoglycan. If purified preparations of the lipid intermediates (labeled in amino acid or GlcNAc) were used as substrate, only very poor incorporation was obtained. It is probable that the conditions for peptidoglycan synthesis from lipid intermediates in this membrane system are far from optimal, but it is also possible that the PBP-2-RodA protein system only utilizes a special fraction of the lipid intermediates that was not present in the purified lipid preparation. Participation of Both PBP-2 and RodA Protein in the Peptidoglycan Synthetic Activities-The formation of peptidoglycan was carried out using membranes that overproduced both PBP-2 and the RodA proteins (Table IV, plasmid pHS506). Table IV shows the results obtained both in the presence and absence of excess cefmetazole (50 pg/ml). The transglycosylase reaction could be observed in both the presence and absence of cefmetazole. Mecillinam (10 pg/ml) inhibited the cross-linking reaction completely in the presence of cefmetazole but partially in its absence. The stimulation of the transglycosylase reaction by high concentrations of mecillinam has been reported previously (12). On the contrary, only a small amount of peptidoglycan synthesis was observed if membranes were used from cells that carried a plasmid pHs503 that overproduced PBP-2 but which did not express the RodA protein. The small amount of peptidoglycan synthesis in the latter membranes may result from the presence of low levels of RodA protein expressed from the chromosomal copy of the mrdB(rodA) gene. In the absence of cefmetazole the cross-linking reaction was insensitive to mecillinam, indicating that the cross-linking was not catalyzed by PBP-2 when the RodA protein was absent. Membranes prepared from E. coli JST9753/pHS504 which should have high levels of the RodA protein (but not of PBP-2) also showed a lower level of peptidoglycan synthesis. Part of this activity may again be due to PBP-2 formed by the chromosomal copy of the mrdA gene, as the cross-linking of the product peptidoglycan was inhibited by mecillinam in the presence of cefmetazole. However, in its absence, the cross-linking activity was again insensitive to mecillinam. Membranes from the cells that did not involve plasmids also showed low activity of a similar nature.
The overproduction of PBP-2 in the membranes used in these experiments was readily confirmed by the use of the PBP assay with ['*CC]-or [3H]benzylpenicillin, but there is no convenient method for measuring the overproduction of the RodA protein. We, therefore, attempted t o obtain more conclusive evidence of the individual roles of PBP-2 and the RodA protein in peptidoglycan synthesis by using membranes prepared from strains that carry plasmids expressing thermosensitive forms of PBP-2 and the RodA protein (Fig. 1). Fig. 5A shows that the synthesis of peptidoglycan (transglycosylase) and the level of cross-linking (transpeptidase) were similar at 30 and 37 "C in membranes which contained elevated levels of wild type PBP-2 and RodA protein. However, when membranes were used that had elevated levels of wild type RodA protein, but thermosensitive PBP-2, there was a slight decrease in the transglycosylase activity at 37 compared to 30 "C, but a much more significant thermosensitivity of the extent of cross-linking (Fig. 5B). When cell membranes were used that had elevated levels of wild type PBP-2, but thermosensitive RodA protein, thermosensitivity of the transglycosylase activity was more significant than in the other membranes. There was also some effect on the level of cross- formed in the presence of cefmetazole, using membranes prepared from cells expressing greatly elevated levels of PBP-2 and the RodA protein, was digested with lysozyme and the products were separated by two-dimensional thin layer chromatography. The autoradiogram shown in Fig. 6A shows that the main product of lysozyme digestion was the compound C6 or C6', the repeating unit of peptidoglycan GlcNAc-MurNActetrapeptide or GlcNAc-MurNAc-pentapeptide respectively (27,28), and the cross-linked dimer muropeptides C3 or C3', bis(G1cNAc-MurNAc-tetrapeptide) or bis(G1cNAc-MurNActetrapeptide)-D-alanine, respectively (27,28). A few minor products were also obtained but were not characterized. The peptidoglycan synthesized in the membrane system in the presence of 10 pg/ml mecillinam produced mostly C6 or C6' after digestion with lysozyme (Fig. 6B, see also Ref. 8).
The products of lysozyme digestion were dinitrophenylated, and after acid hydrolysis, the dinitrophenyl compounds were . The C3 and C3' products gave free A2pm and mono-dinitrophenyl-A'pm, and the C6 and C6' products, derived from peptidoglycan synthesized both in the presence and absence of mecillinam, gave monodinitrophenyl-Azpm but no free A'pm. More precise characterization of the muropeptides is underway in collaboration with B. Glauner and U. Schwarz in Tubingen.

DISCUSSION
The work of Spratt and Pardee (1) and Spratt (2) in 1975 established that PBP-2 functions in the determination of the rod shape of E. coli cells and that the amidinopenicillin, mecillinam (30), results in the conversion of E. coli into spherical cells by binding to PBP-2. Mutants that produce thermosensitive forms of PBP-2 grow as rod-shaped cells at the permissive temperature and as spherical cells at the restrictive temperature (2). Prior to the above studies Matsuzawa et al. (31) had identified a mutant that formed spherical cells due to a mutation that mapped a t 15 min on the E. coli chromosome. The latter mutant (rodA) (31) was mecillinam resistant but produced normal PBP-2 (32). The cell shape genes identified by the above two classes of mutants were shown to map in the separate but contiguous genes pbpA(mrdA) and rodA(nrdB) (18, 19,22). The mrdB gene product has been identified (33) as a minor component of the cytoplasmic membrane (M, of 31,000). A X transducing phage (22,33) carrying the mrdA and mrdB genes was isolated and shown to also contain the closely linked gene dacA, encoding PBP-5 (34), and the gene for a 54-kDa protein (22) which appears to be a lipoprotein the function of which is ~n k n o w n .~ Overproduction of PBP-5 has been found to result in the production of spherical bacteria (35), but the mrdA and mrdB genes only cause spherical cell shape when their products are defective.
The biosynthetic study reported here shows that PBP-2 has an enzymatic role in peptidoglycan synthesis. The protein is probably a bifunctional enzyme catalzying both the transglycosylase and transpeptidase reactions and, therefore, appears to be similar to PBP-1A, PBP-lB, and PBP-3 which have also been shown to be bifunctional (5,10,11,(13)(14)(15). The results suggest that both PBP-2 and the RodA protein are required for peptidoglycan synthesis, but it is difficult to establish which of the two proteins is responsible for the transglycosylase activity. The transpeptidase activity that we detected was presumably catalyzed by PBP-2 since the reaction was strongly inhibited by low concentrations of mecillinam, which binds exclusively to PBP-2, and by other Plactams that have high affinity for PBP-2 (e.g. benzylpenicillin and N-formimidoylthienamycin (36)), but was only poorly inhibited by those P-lactams that have very low affinity for PBP-2 (e.g. cefmetazole, latamoxef, and MT-141 cefminox (25)). The thermosensitivity of the transpeptidation reaction in membranes that contained high levels of thermosensitive PBP-2 also supports the view that this activity was catalyzed by PBP-2. The RodA protein may regulate the activity of PBP-2 or the two proteins may form a complex which together functions as a peptidoglycan synthetase with a special role in the determination of bacterial cell shape. The proteins may act together to form an initiation piece (or ring) of peptidoglycan at the center of the cell (5,14) to ensure the formation of the correct rod shape of the cell. At present, however, we cannot completely eliminate the possibility that PBP-2 acts as the transpeptidase and that the RodA protein acts as the transglycosylase, but we believe this to be unlikely. Moreover, our membranes also contained two other smaller proteins (7.7-and 17-kDa proteins) presumably encoded by two open reading frames in the chromosomal insert of the plasmids that we used.' The possibility remains that these proteins play some role in the biosynthetic process described above. The clone that contains only genes of PBP-2 and the RodA protein has to be isolated in order to exclude the activity of the smaller proteins. We are also trying to reconstitute a peptidoglycan synthetic system from purified protein components, including PBP-2, RodA, and other proteins such as the 7.7-and 17-kDa proteins (see above), or 54-kDa proteins (22) the function of which were so far not required for the peptidoglycan synthesis in vitro but could be important in the in uiuo process. However, even attempts to reconstitute an active mecillinam-sensitive peptidoglycan synthetic system by sonication of a mixture of membrane vesicles containing high levels of PBP-2 (and smaller proteins but not RodA) and those containing high levels of the RodA protein have been unsuccessful. Furthermore, we have not been able to achieve peptidoglycan synthesis from purified lipid-linked intermediates using either membranes or purified PBP-2.
On the other hand, the amino acid sequence of PBP-2 has recently been obtained from the nucleotide sequence of the mrdA(pbpA) gene.' The amino acid sequence shows several regions of substantial similarity to the sequences of other high molecular weight PBPs determined by others (37,38). A putative penicillin-binding site containing the sequence Ser-Xaa-Xaa-Lys was found which has been found at the active site of all PBPs. The existence of this sequence and the overall similarity to other high molecular weight PBPs add support to the view that PBP-2 acts, like other high molecular weight PBPs (6, 10, l l ) , as a bifunctional enzyme of peptidoglycan synthesis. PBP-2 and the RodA protein are not the only gene products that were proved to have a role in the determination of the shape of E. coli. The gene envB (39) which is located at 71 min on the E. coli chromosome, and the closely linked and possibly identical mreB gene (5), as well as the mreC ( 5 ) gene have also been implicated in the determination of cell shape and the sensitivity of E. coli cells to mecillinam. At present we know little of the function of these genes in the bacterial cell cycle.