Purification to homogeneity and properties of two D-alanine carboxypeptidases I From Escherichia coli.

Three homogeneous preparations of D-alanine carboxypeptidases I have been obtained from Escherichia coli strain H2143, termed enzymes IA, IB, and IC. Enzyme IA purified from the membrane after extraction with Triton X-100 appeared on sodium dodecyl sulfate gel electrophoresis to be a polypeptide doublet whose monomer molecular weights were about 32,000 and 34,000. In addition to D-alanine carboxypeptidase activity, it catalyzed a transpeptidase reaction with several substrates, bound [14C]penicillin G, had a weak penicillinase activity, but was devoid of endopeptidase activity. Enzyme IB obtained from the membrane after LiCl extraction and enzyme IC obtained from the supernatant solution were either identical or extremely similar. They were composed of a single polypeptide whose monomer molecular weight was about 41,000. In addition to carboxypeptidase activity, they catalyzed an endopeptidase reaction, had weak penicillinase activity, and had very poor transpeptidase activity, but did not bind [14C]penicillin G. Some data relating to the mechanism of catalysis by these enzymes are described. Their possible physiological role is discussed.

Three homogeneous preparations of n-alanine carboxypeptidases I have been obtained from Escherichiu coli strain H2143, termed enzymes IA, IB, and IC. Enzyme IA purified from the membrane after extraction with Triton X-100 appeared on sodium dodecyl sulfate gel electrophoresis to be a polypeptide doublet whose monomer molecular weights were about 32,000 and 34,000. In addition to n-alanine carboxypeptidase activity, it catalyzed a transpeptidase reaction with several substrates, bound ["Clpenicillin G, had a weak penicillinase activity, but was devoid of endopeptidase activity. Enzyme IB obtained from the membrane after LiCl extraction and enzyme IC obtained from the supernatant solution were either identical or extremely similar. They were composed of a single polypeptide whose monomer molecular weight was about 41,000. In addition to carboxypeptidase activity, they catalyzed an endopeptidase reaction, had weak penicillinase activity, and had very poor transpeptidase activity, but did not bind ['"Clpenicillin G. Some data relating to the mechanism of catalysis by these enzymes are described. Their possible physiological role is discussed.
A number of penicillin-sensitive enzymatic activities have been identified in various bacterial species. These include transpeptidases, n-alanine carboxypeptidases, and endopeptidases (for a recent review, see Ref. 1). The relationship of the various activities to one another (i.e. whether they are catalyzed by one or more than one protein) and the physiological functions of some of these activities require much further clarification, as does the precise mechanism(s) by which they are inhibited by p-lactam antibiotics. o-Alanine carboxypeptidases I and II which remove, respectively, the terminal and penultimate n-alanine residues of the uridine nucleotide substrate, UDP-acetylmuramyl-LAla-nGlu-mesoDap-nAla-nAla, were first described in Escherichia coli and partially purified (2). n-Alanine carboxypeptidase I was competitively inhibited by penicillins, and n-alanine carboxypeptidase II was insensitive to fi-lactam antibiotics. Subsequently, partially purified preparations of n-alanine carboxypeptidase I were shown to possess both endopeptidase (3) and transpeptidase (4) activity (the latter assayed using both natural and synthetic substrates). However, since none of the preparations studied had been purified to homogeneity, it was difficult to state whether or not the various activities observed were functions of the same enzyme protein. Penicillin-sensitive endopeptidase activities in E. coli had also been studied independently of n-alanine carboxypeptidase and transpeptidase (5). The purpose of the present paper is to report the purification to homogeneity of two distinct proteins from Escherichia coli * Supported by Research Grants AI-09152 from the National Institutes of Health, and GB-29747 from the National Science Foundation.
Endopeptidase-The reaction mixture contained in a total volume of 30 ~1, 2 ~1 of 2 M Tris-HCl buffer, pH 7.5, 3 ~1 of the E. coli dimer (C-3, 8000 cpm), and enzyme. After incubation at 37" for 1 hour, 20 ~1 of isobutyric acid/l N NH,OH (5/Z) were added, and the reaction mixture was subjected to paper chromatography overnight on Whatman 3MM filter paper in the solvent above. The chromatogram was subjected to radioautography, and the area corresponding to the E. coli monomer, C-6 (RF = 0.4, R, of the substrate C-3 = 0.2) was cut out and counted.
Alternatively the reaction mixture could be subjected to paper electrophoresis at pH 1.8 (7W formic acid) at 1700 volts for 5 hours, which also separates C-3 and C-6.  10).
In addition, n-alanine carboxypeptidase IA catalyzed the release of the terminal n-alanine residue from UDP-MurNAc-LAla-nGlu-Lys-nAla-nAla (Fig. 12A) and from the synthetic substrate diacetyl-L-lysyl-nalanine-n-alanine (Fig. 12B). Neither of the latter compounds was a very good substrate, and under the conditions of the experiment in Fig. 12, release of the terminal n-alanine residue was incomplete. Data on the K, and V,,,,, of these enzymes are reported below.
Endopeptidase Actiuity-The ability of the three enzymes to  catalyze the hydrolysis of the dimer isolated from the cell wall of Escherichia coli (C-3, bis(disaccharide-tetrapeptide)) was examined. This hydrolysis was catalyzed by o-alanine carboxypeptidases IB and IC, but n-alanine carboxypeptidase IA was totally inactive on this substrate (Fig. 13). It should be noted that the hydro!ysis did not go to completion. This could be due to impurities in the substrate, inactivation of the enzyme during the reaction, or an equilibrium phenomenon. It has not been examined further.
Transpeptidase Activities-Several reactions which can be termed transpeptidation have been examined. The substrates employed were either UDP-MurNAc-LAla-nGlu-mesoDap-nAla-nAla or diacetyl-rkys-nAla-nAla, and the acceptors examined in each case were glycine and hydroxylamine.
n-Alanine carboxypeptidases IA, IB, and IC all catalyzed transpeptidation reactions under some conditions. A comparison of the time course of transpeptidation with UDP-MurNAc-LAla-DGlu-mesoDap-uAla-nAla as substrate and glycine as transpeptidation acceptor at a substrate/acceptor ratio of l/100 is shown in Fig. 14. The amount of enzyme added in each case was such as to catalyze equivalent carboxypeptidase activity on the substrate. It is evident that the extent of transpeptidation under these conditions was greater with o-alanine carboxypeptidase IA than with either n-alanine carboxypeptidase IB or IC. A detailed examination of transpeptidation reactions catalyzed by n-alanine carboxypeptidase IA was carried out with UDP-MurNAc-LAla-nGlu-mesoDap-DAla-nAla as substrate and glycine as acceptor (Fig. 15A). Under these conditions total "hydrolysis" of the substrate (release of n-alanine) was constant up to a concentration of about 120 mM glycine, and the products were partitioned between the hydrolysis product, UDP-MurNAc-LAla-nGlu-mesoDap-DAla, and the transpepti- At concentrations of glycine above 120 mM, there was rate acceleration; at 1.2 M glycine, where the hydrolysis product was totally eliminated, the extent of utilization of substrate for formation of transpeptidation product was 140% of its utilization in the absence of glycine.
A similar result was obtained with diacetyl-LLys-DAla-DAla as substrate. However, in this case the rate acceleration with 1.2 M glycine amounted to 220% of the rate of substrate utilization in the absence of glycine (Fig. 15B). The data obtained with hydroxylamine as transpeptidation acceptor were less satisfactory (Fig. 15, C and D) because hydroxylamine (whether containing salt or salt-free) was a marked inhibitor of the total reaction. The salt (NaCl, equimolar to NH,OH) itself produced some inhibition but not nearly as marked as that found with hydroxylamine.
From these experiments it was possible to demonstrate only that in the presence of the appropriate substrate small amounts of either UDP-MurNAc-LAla-DGlu-mesoDap-DAla-NHOH or diacetyl-Lays-DAla-NHOH were formed (Fig. 15, C and 0). A similar result was obtained with n-alanine carboxypeptidase IB using UDP-MurNAc-r&a-DGlu-mesoDap-DAla-nAla as substrate and glycine as acceptor, except that much higher concentrations of acceptor were required to form transpeptidation products (Fig. 16). For example, the concentration of glycine required to effect formation of equivalent amounts of hydrolysis and transpeptidation products was 600 to 700 mM with enzyme IB and about 90 mM with enzyme IA. Enzyme IB did not catalyze any hydrolysis of diacetyl-L-lysyl-n-alanyl-nalanine, and similarly no transpeptidation reaction occurred. The reactions of enzyme IB with hydroxylamine as transpeptidation acceptor, if it occurred at all, would require extremely high concentrations of the acceptor. Although enzyme IB can catalyze transpeptidation reactions, it does so less efficiently than enzyme IA.

Sensitivity of Enzymatic Activities to fi-Lactam Antibiotics
Sensitivity-All three enzymes were sensitive to inhibition by various B-lactam antibiotics. The concentrations of several of these antibiotics required to produce 50% inhibition of hydrolysis of UDP-MurNAc-r&a-nGlu-mesoDap-nAla-nAla by n-alanine carboxypeptidases IA and IB are summarized in Table IV. It should be noted that enzyme IB is markedly more sensitive than enzyme IA to inhibition by these substances.

Binding of ["C]Penicillin
G-As shown in the purification procedure, particularly in the elution profiles of the various columns, n-alanine carboxypeptidase IA activity coincided with a peak of ["Clpenicillin G binding activity (Fig. 2). On the other hand a second penicillin-binding protein(s) was seen in the intital purification steps in the same region as n-alanine carboxypeptidases IB and IC (Fig. 4); this penicillin-binding profile did not coincide with enzyme activity (Figs. 5 and 7), and during the preparation of enzyme IB and IC it was removed.
The amount of ['*C]penicillin G bound by purified enzyme IA was in the range of 50 to 100 cpm/pg of protein, and for enzymes IB and IC it was less than 7 cpm/wg. Sodium dodecyl sulfate gel electrophoresis of enzyme IA was carried out after prebinding of ["Clpenicillin G. The gel profile of bound ["Clpenicillin G coincided exactly with the Coomassie blue staining material (Fig. 17).
Release of Bound ["C]Penicillin G-Binding of penicillin to E. coli carboxypeptidases had not been observed previously. Moreover, the inhibition of this enzyme activity by penicillins was reported to be reversible on addition of penicillinase (2). A similar situation had been observed earlier in the study of the n-alanine carboxypeptidase from Bacillus stearothermophilus (12). The explanation in the latter case was that this enzyme bound radioactive penicillin but released it with a half-time which was short relative to the incubation time ordinarily used in the assay. Because of this prior experience with Bacillus stearothermophilus, the rate of release of bound ["Clpenicillin G from E. coli carboxypeptidase IA was examined. It was found to be remarkably rapid at 37" with a half-time of about 5 min. At 4" the half-time was about 1 hour (Fig. 18) jetting the mixture to thin layer chromatography, using the same technique as had been employed earlier in a study of the product released by the bacillus enzymes. In the present case it was found that ["Clpenicilloic acid was apparently the product (Fig. 19)   diacetyl-LLys-nAla-nAla or UDP-MurNAc-Ala-nGlu-LLys-nAla-oAla as substrate much lower concentrations of pCMB Under these conditions the activity due to n-alanine carboxyresulted in 50% inhibition, i.e. approximately 0.005 mM. Neither n-alanine carboxypeptidase IB nor IC was inhibited by peptidase IA was entirely inhibited, and the residual activity was exactly equivalent to the activity of the amount of pCMB; at the highest concentration tested (2 mM) approxin-alanine carboxypeptidase IB added in the absence of pCMB. mately 15% activation of both enzyme activities was observed. The question of whether the failure of pCMB to inhibit these In addition, pCMB inhibited the transpeptidase activity of enzyme IA. It had no effect on either the transpeptidase or the enzymes was due to the presence of an inhibitor of pCMB in the enzyme preparation was examined by mixing enzymes IA p endopeptidase activity of enzymes IB and IC (Table V). enicillinase activity of enzyme IA, like its other activities was and IB in the presence of 1 mM pCMB, an amount which also inhibited by pCMB (Table VI). Penicillinase activity of inactivates n-alanine carboxypeptidase IA entirely while actienzymes IB and IC was, however, resistant to pCMB. The E. vating n-alanine carboxypeptidases IB and IC by about 10%. coli penicillinase (free of all carboxypeptidase activity, ob-8The abbreviation used is: pCMB, p-chloromercuribenzoate. tained from the hydroxylapatite column flow through in Fig. 2) Purification of E. coli D-Alanine Carboxypeptidases 1   it seems difficult to equate the physiological transpeptidase activity with the in vitro "transpeptidation" reactions catalyzed by the E. coli carboxypeptidases which have been described in this paper and by other workers (4). Another interesting phenomenon was observed first in 1940 (18). At low penicillin concentrations, cells of E. coli are not killed but form long filaments, i.e. presumably septation is specifically inhibited. The marked difference in penicillin sensitivities of E. coli n-alanine carboxypeptidase IA on the one hand and IB and IC on the other could suggest some relation to the process of filamentation.
But again, the multiplicity of other proteins with which penicillin interacts in E. coli suggests that the data which are presently available should be interpreted with great caution.
An especially interesting feature of the E. coli carboxypeptidases described here is their weak penicillinase activity. Evidence has been presented that this penicillinase activity is an intrinsic property of these enzymes and not due to contamination with the major E. coli penicillinase. If further investigation confirms this finding, then these data suggest that the cleavage of the p-lactam ring of penicillin is analogous to the cleavage of the terminal n-alanyl-n-alanine bond of the various substrates for the carboxypeptidases, i.e. the data support the idea that penicillins are substrate analogs, and also incidentally the idea that n-alanine carboxypeptidases and penicillinases had a common evolutionary origin (19).
The data obtained may be compared to other published data on this enzyme (4). Early studies from this laboratory on n-alanine carboxypeptidase I (2) were carried out with another strain, E. coli B, and thus some of the differences (e.g. precise levels of penicillin sensitivity) may be ascribed to strain differences. The two n-alanine carboxypeptidases reported here (IA and IB; IC) correspond to the two major n-alanine carboxypeptidases reported in a preliminary communication from this laboratory (16). If there are additional minor p-alanine carboxypeptidases and/or endopeptidases (see Fig. 6 in Ref. 16), they will have been discarded in the present purification.
Some of the properties of these enzymes correspond to properties reported in a study of the same enzymes from another laboratory (4), for example, their separation by DEAE-cellulose, the differences in their sensitivities to P-lactam antibiotics, and the transpeptidation reaction. However, enzyme IA was devoid of endopeptidase activity in the present study although the analogous Fraction A (4) had considerable activity. It is conceivable that this difference is due to the higher degree of purification obtained in the present study.
Finally, the present paper contains some preliminary information concerning the mechanism of catalysis by these carboxypeptidases. The partitioning of products between hydrolysis and transpeptidation products in the presence of glycine and the marked rate acceleration seen with high levels of glycine strongly suggest the occurrence of an acyl enzyme intermediate in the reaction catalyzed by carboxypeptidase IA. The hydrolysis of this intermediate would be rate-limiting and its decomposition would be accelerated by the nucleophile, glytine. It should be compared to more extensive data obtained with the B. subtilis and B. stearothermophilus carboxypeptidases.' Since pCMB inactivates the hydrolysis of both UDPacetylmuramyl-pentapeptide and diacetyl-L-lysyl-D-alanyl-Dalanine by carboxypeptidase IA (as well as the transpeptidase and penicillinase activities of carboxypeptidase IA), it seems likely that an SH group participates in some manner in the reactions catalyzed by this enzyme. pCMB is more effective as an inhibitor with the synthetic substrate, diacetyl-L-lysyl-n-alanyl-n-alanine than with the uridine nucleotide substrate, i.e. the natural substrate is more effective in protecting the -SH group from reaction. On the other hand penicillin G binding to o-alanine carboxypeptidase IA is not inhibited at all bypCMB, and in fact the level of binding in the presence of pCMB is 3 times that in its absence. These data could suggest that the presumed acyl enzyme intermediate (of which penicilloyl enzyme may be an analog) is not located on a sulfhydryl group but that a sulfhydryl group is involved in the deacylation reaction. Thus, the penicilloyl enzyme would be stabilized by a substitution of this --SH group. o-Alanine carboxypeptidases IB and IC obviously have a very different catalytic mechanism since they do not bind ["Clpenicillin G and moreover are not inhibited by pCMB. Extension of these preliminary findings to obtain further information about the catalytic mechanism of these enzymes is in progress.