Substitution of Lysine 213 with Arginine in Penicillin-binding Protein 5 of Escherichia coli Abolishes D-Alanine Carboxypeptidase Activity without Affecting Penicillin Binding*

All penicillin-binding proteins (PBPs) contain a con- served box of homology in the carboxyl-terminal half of their primary sequence that can be Lys-Thr-Gly, Lys-Ser-Gly, or His-Thr-Gly. Site-saturation mutagenesis was used to address the role of the lysine residue at this position (Lys2”) in Escherichia coli PBP 5, a D-alanine carboxypeptidase enzyme. A soluble form of PBP 5 was used to replace Lys213 with 18 other amino acids, and the ability of these mutant proteins to bind [SH]penicillin G was assessed. Only the substi- tution of lysine with arginine resulted in a protein that was capable of forming a stable covalent complex with antibiotic. The affinity of [’4C]penicillin G for the arginine mutant was 1.2-fold higher than for wild-type PBP 6 (4.4 versus 5.1 pg/ml for 20 min at 30 “C), and both proteins showed identical rates of hydrolysis of the [’4C]penicilloyl-bound complex (tl12 = 9.1 min). Surprisingly, the arginine-substituted protein was unable to catalyze D-alanine carboxypeptidase activity in vitro, which suggests that there is a substantial difference in the geometries of the peptide substrate and penicillin G within the active site of PBP 5. and the of radioactivity remaining in the sPBP proteins was determined by SDS-polyacrylamide gel electrophoresis, fluorography, and densitometry as described pentapeptide L-Ala-D-r-Glu-L-Lys-D-Ala-D-Ala (%Fa). The D-alanine pH 8.5, determined by the assay utilizing D- amino-acid oxidase as by Fr6re et al. The kinetic parameters and K,,, determined from Lineweaver-Burk plots of two independent

Another group of enzymes, called p-lactamases, also interacts with penicillin and other p-lactam antibiotics. The production of these enzymes is the major clinical cause of resistance to p-lactam antibiotics (11). P-Lactamases are known to catalyze the hydrolysis of 8-lactam antibiotics through an acyl-enzyme mechanism that is very similar to the mechanism of B-lactam antibiotic interaction with PBPs. When the primary structures of several PBPs and p-lactamases became available, however, the sequence homology between the two groups of enzymes was found to be quite low. The greatest homologies are found within the region of the active-site serine residues. Within this small region, up to 60% homology can be observed in some cases, but it quickly decreases as larger and larger regions are compared. The lack of significant sequence homology between PBPs and p-lactamases brought into question the evolutionary relatedness of these two groups of enzymes. The relatedness was strongly suggested, however, when the crystal structures of either Bacillus licheniformis 749/C #3-lactamase (12) or Bacillus cereus p-lactamase I (13) were compared to the crystal structure of the low molecular weight DD-peptidase from Streptomyces R61 (14). The similarities of the secondary structures of these enzymes were quite striking, even though the two enzymes differed by 8 kDa and showed very little sequence homology.
The evolutionary relationship between these two groups of penicillin-interactive enzymes was further investigated using homology searches and amino acid alignments, using the Streptomyces R61 D-alanyl-D-alanine peptidase as a reference, of all of the known sequences of PBPs and 8-lactamases (15). This method identified several regions, called boxes, that consist of strict identities or homologous amino acids. These boxes, whose importance is highlighted by the known structures of several members of the penicillin-interactive family, are situated in or near the active site of these enzymes and catalyze the acyl-exchange reactions that occur during turnover. One of these conserved regions, called Box 11, is the well-characterized Ser-X-X-Lys tetrad that contains the active-site serine nucleophile. Another interesting region of homology, called Box VII, is a conserved triad Lys-Thr-GIy, Lys-Ser-Gly, or His-Thr-Gly that occurs within the carboxylterminal half of the primary sequence of these enzymes. Kelly et al. (16), from the crystallographic analysis of @-lactam antibiotics bound to the Streptomyces R61 DD-peptidase, have suggested that the binding of substrate is facilitated by the conserved triad of Box VI1 and that the histidine residue contributes to the initial binding. Because these hypotheses have not been tested at the biochemical level, however, the role of the basic residue within the Box VI1 region of PBPs has not been confirmed.
We chose to investigate the role of the conserved lysine (Lys2I3) of Box VI1 in the catalytic mechanism of E. coli PBP 5 to gain a better understanding of the role of this amino acid in the molecular interactions of a PBP with both p-lactam antibiotics and peptide substrate. PBP 5 is a low molecular weight PBP with a M, of 41,300 that is known to catalyze the major D-alanine carboxypeptidase activity in vivo (17,18). It is initially synthesized with a signal sequence that is cleaved off during translocation to the periplasm (19). The carboxylterminal 15 amino acids have been shown to mediate the association of the protein with the membrane as removal of these amino acids results in the synthesis of a soluble periplasmic protein (20). PBP 5 is unique in that it displays the highest p-lactamase activity of all of the E. coli PBPs; at pH 7 and 30 "C, the half-life of the acyl-enzyme complex is -10 min (21, 22). The D-alanine carboxypeptidase activity that it catalyzes can also be easily assayed in vitro. We chose the method of site-saturation mutagenesis to investigate the role of the basic residue in the Box VI1 region of a soluble form of PBP 5 in which Lys213 was mutated to 18 other amino acids and the properties of these mutants were investigated.

EXPERIMENTAL PROCEDURES
Materials-All restriction enzymes and DNA-modifying enzymes were from Pharmacia LKB Biotechnology Inc. ["CIPenicillin G (155 pCi/mg) was from Amersham Corp. [3H]Penicillin G (57.8 mCi/mg) was the generous gift of Dr. Patrick Casey (Merck Sharp and Dohme). Oligonucleotides were synthesized on a Milligen/Biosearch Cyclone Plus DNA synthesizer. The strains of E. coli used were JM103 with M13 vectors, CJ236 (dut-, ung-) for the production of uracil-containing M13 phage, and NK5830 (F' lacIqpro/arg Alac proxrrrNalA rif' recA-56 suo ara thy) for the overexpression of cloned genes of PBP 5 mutants. Sequencing was performed with the Sequenase I1 kit from United States Biochemical Corp.
Construction of Soluble P B P 5 Expression Vector-A soluble form of PBP 5 (sPBP 5) was constructed as described previously (22). The truncated dacA gene encoding sPBP 5 was reconstructed in the expression vector pTTQ18 (Amersham International) by standard cloning techniques (23). Since pTTQ18 uses ampicillin resistance as its selectable marker, which is detrimental to the detection and purification of PBPs, the vector was modified by digesting with DraI, which removes a 700-bp fragment from within the ampicillin resistance gene, and ligating the kanamycin resistance gene from pUC4K (digested with HindII; Pharmacia) back in its place. The final construct, which contains the dacA gene encoding sPBP 5, the kanamycin resistance gene, and the lacF gene, was designated pWT3K. A derivative of pWT3K, called pKM1325, was also constructed to be a recipient of the mutagenized fragments generated below. In this derivative, the 293-bp SalI-BglII fragment within the dacA gene was replaced with the 1325-bp SalI-BglII fragment from pMSG (Pharmacia), which allowed the unambiguous detection of reconstructed dacA mutants by restriction screening (see below).
Site-saturation Mutagenesis-The 1230-bp EcoRI-Hind111 fragment from pWT3K was cloned into M13mp19, and the recombinant M13 phage was used to make uracil-containing single-stranded DNA as described by Kunkel (24). Site-saturation mutagenesis was performed by first using an oligonucleotide to loop-out the AAA lysine codon, which was then followed by another oligonucleotide to loopin random codons in the former position of the AAA codon. This dual procedure was employed to prevent bias for the wild-type codon and presented a way to identify codon insertions by hybridization GTC GAC GGC ATC ACC GGA CAC ACT-3', which consists of the (see below). The loop-out oligonucleotide synthesized was 5'codons for residues 209-217 that surround the KTG triad of sPBP 5, with the exception of Lys213. Uracil-containing single-stranded DNA template was mutagenized with this loop-out oligonucleotide, and positive phage were identified by sequencing. Uracil-containing template was made from one of the positive phage and was used to perform loop-in oligonucleotide-directed mutagenesis. This loop-in oligonucleotide, 5'-GTC GAC GGC ATC NNN AAC CGG ACA CAC T-3', contained all four bases in equimolar amounts in each of the three positions of the former AAA codon. The resultant singlestranded phage were screened with 32P-labeled loop-out oligonucleotide as described previously (22), and single-stranded DNAs were prepared from nonhybridizing phage. Positive phage containing codons for 16 amino acids at codon 213 were identified by dideoxy chain termination sequencing (25) using Sequenase I1 (United States Biochemical Corp.) and a 17-mer oligonucleotide (5"TAACGGCCTGT-TATGGG-3') that primed 26 nucleotides upstream from the KTG triad. The yield of positive phage containing an inserted codon was 50%.
Due to the repeated isolation of the same amino acids with different codons, only 16 mutants were found by this method. This in fact appeared to be due to a bias of the oligonucleotide, such that the ratio of G at each position was much lower than expected. For the remaining three mutants, specific loop-in oligonucleotides were synthesized using the same sequence as above, except that codons specifying for glycine (GGT), glutamic acid (GAG), and tryptophan (TGG) were used in place of the former AAA codon. Mutagenesis was repeated as described above, and the final three mutations were obtained. In cases where degenerate codons were isolated, the mutations containing the most highly used codon based on the codon preference from E. coli as shown by Sharp and Li (26) were selected. The single-stranded DNA from each selected mutant was sequenced through the SalI-BglII fragment to ensure that no other mutation had occurred during mutagenesis. The replicative forms of each mutation were prepared, and the 293-bp SalI-BglII fragment was isolated and used to replace the exogenous 1325-bp SalI-BglII fragment from pKM1325 (constructed as described above).
Expression of sPBP 5 Mutants-Cells harboring the sPBP 5 mutant genes in pKM3K were grown at 37 "C in 50 ml of 2 X YT medium (23) in the presence of 50 pg/ml kanamycin. When the absorbance at 600 nm reached 0.5, protein synthesis was induced by adding isopropyl-1-thio-P-D-galactopyranoside to a final concentration of 0.5 mM and incubating with shaking for an additional 2 h. The cultures were subjected to osmotic shock as described by Kustu and Ames (27), and the sterile shock fluids were concentrated 10-fold on an Amicon ultrafiltration device fitted with a PM-10 filter. The final protein concentrations of shock fluids were 2-5 mg/ml as estimated by the protein assay of Lowry et al. (28).
PHIPenicillin G Binding and Immunoblotting of sPBP 5 Mutants-Aliquots of each shock fluid (20 pg) were incubated with [3H]penicillin G at a final concentration of 40 pg/ml for 20 min at 30 "C, and the labeled proteins were visualized by SDS-PAGE and fluorography. The gel was treated with EN3HANCE (Du Pont-New England Nuclear), dried, and exposed to Kodak XAR-5 film. The mutant sPBP 5-K213Y (sPBP 5 containing a Lys to Tyr mutation at position 213) was not analyzed for binding activity due to its loss during cloning and was not pursued. To determine that absence of binding was not due to the absence of the protein in the shock fluid, the amount of expression of each mutant was assessed by immunoblotting. Aliquots (20 pg) of osmotic shock fluids were separated on a 12% SDSpolyacrylamide gel and transferred to nitrocellulose using a semidry blotting apparatus. After blocking with nonfat dry milk, the nitrocellulose was incubated first with antiserum raised against sPBP 5' and then with alkaline phosphatase-conjugated goat anti-rabbit Fab fragment (Sigma). The sPBP 5 proteins were visualized with 5-bromo-4chloro-3-indolyl phosphate and nitro blue tetrazolium (29). The antiserum was produced by injecting an SDS-polyacrylamide gel suspension of sPBP 5' subcutaneously into New Zealand white rabbits as described by Harlow and Lane (30).
Determination of Affinities of sPBP 5 and sPBP 5-K213R for p4C] Penicillin G-The affinities of sPBP 5 and sPBP 5-K213R for ["C] penicillin G were measured essentially as described by Ghuysen et al. (31). Aliquots (15 pl) of either the purifiedproteins (1.5 pg) or osmotic shock fluids (20 pg) were incubated with 5 p1 of serial dilutions of ['4C]penicillin G (80-1.25 pg/ml final concentration) for 20 min at 30 "C. Following incubation, SDS-PAGE sample buffer was added, and the samples were separated on a 12% SDS-polyacrylamide gel. The gel was treated with EN3HANCE, dried, and exposed to film. The amount of radioactivity present in the proteins was estimated by densitometric scanning of the film with a Hoefer Scientific Instruments GS 300 densitometer.
Determination of Rate of Hydrolysis of Bound p'C]Penicillin G by sPBP 5 and sPBP 5-K213R"Hydrolysis assays were performed essentially as described previously (22). Aliquots of purified proteins (6 pg, 1 pgltime point) or osmotic shock fluids (120 pg, 20 pgltime Site-saturation Mutagenesis point) in 25 mM Tris. HCl, pH 7.3, were incubated with ["Clpenicillin G at a final concentration of 40 pg/ml at 30 'C. After 20 min, nonradioactive penicillin G was added to a final concentration of 4 mg/ml, and the incubation was continued. Aliquots were removed at various times, and the amount of radioactivity remaining in the sPBP 5 proteins was determined by SDS-polyacrylamide gel electrophoresis, fluorography, and densitometry as described above.
D-Alanine Carboxypeptidase Assays-D-Alanine carboxypeptidase activity was measured exactly as described previously (22) using the pentapeptide L-Ala-D-r-Glu-L-Lys-D-Ala-D-Ala ( % F a ) . The amount of D-alanine released after 30 min in 50 mM Trls.HC1, pH 8.5, was determined by the spectrophotometric assay utilizing Damino-acid oxidase as described by Fr6re et al. (32). The kinetic parameters VmaX and K,,, were determined from Lineweaver-Burk plots of two independent experiments.

RESULTS
Box VII Sequences of PBPs and Site-saturation Mutagenesis-Alignment of the sequences of E. coli PBPs, the Streptomyces R61 DD-peptidase, and several p-lactamases in the region of Box VI1 shows the conserved character of the triad ( Fig. 1). This triad can in fact be KTG, KSG, or HTG. We chose to investigate the apparent role of the lysine residue (Lys2I3) of the Box VI1 region in the catalytic mechanism of sPBP 5. A soluble form of PBP 5 was chosen for mutational analysis since it can be separated from endogenous PBP 5 by osmotic shock, and it is known that removal of the carboxylterminal membrane anchor of PBP 5 does not effect its activity (22). The entire sPBP 5 coding sequence was cloned into M13mp19 to generate a template for mutagenesis. Sitesaturation mutagenesis was performed as described under "Experimental Procedures," and mutant genes containing random codon insertions specifying all 20 amino acids at the former position of Lys213 were identified by sequencing.
Subcloning of sPBP 5 Mutants into Expression Vector-For the subcloning of sPBP 5 mutant genes into an expression vector, we employed a strategy that allowed us to clone a small fragment of the PBP 5 mutant gene into an altered plasmid that recreated the entire gene including the mutation. The expression vector pKM1325 contained the wild-type sPBP 5 gene, except that the 293-bp Sun-BglII fragment that encompasses the Box VI1 region (Fig. 2) was replaced with a 1325-bp SaZI-BgZII fragment from an unrelated plasmid (pMSG). This strategy allowed us to subclone a small sequenced fragment from the mutagenesis (all 20 mutants were sequenced through the SalI-BgZII fragment to ensure that

140)
A K K v G P . . . Expression vector for overproduction of sPBP 6 mutants. pKM3K is a derivative of pTTQ18 in which the ampicillin resistance gene was replaced with the kanamycin resistance gene and the sPBP 5 gene was cloned into the EcoRI-Hind111 sites of the polylinker. The DNA encoding sPBP 5 is shown expanded above the map along with a schematic of the protein sequence. The protein is initially translated with a signal sequence that is removed during its translocation to the periplasmic space. The last 15 amino acids of PBP 5 have been deleted, which results in the production of a soluble enzyme. The Box VI1 triad is indicated by X x x Thr Gly, where X x x refers to the site-saturation position that was substituted with 18 other amino acids. Serine 44 is the active-site nucleophile that forms the covalent bond with piactam antibiotics. Kb, kilobase pairs; aa, amino acids. only the intended mutation was present) without the problem of contamination of the original fragment. The final constructs are shown in Fig. 2. Nineteen of these plasmids (the tyrosine mutant was lost during the subcloning process and was not pursued) were transformed into NK5830, and the overproduction of sPBP 5 mutants in the periplasm was induced with 0.5 mM isopropyl-1-thio-/I-D-galactopyranoside.
Expression and Activity of sPBP 5 Mutants-Following the induction of the sPBP 5 variants, the periplasmic fraction was isolated from each culture using osmotic shock and was used to assess both the activities and the levels of expression of each mutant. To assess the [3H]penicillin G binding activities of the sPBP 5 variants, osmotic shock fluids were preincubated with [3H]penicillin G for 10 min at 30 "C and analyzed by SDS-PAGE and fluorography. [3H]Penicillin G was chosen for this experiment since its 50-fold higher specific activity compared with ['4C]penicillin G should have allowed us to observe even low level binding. As shown in Fig. 3A, however, only the Arg mutant and the "revertant" wild-type sPBP 5 proteins were detected under these conditions. To test whether adequate levels of sPBP 5 mutant proteins were present in the samples that did not show binding, immunoblot analysis was performed on each sample (Fig. 3B). The immunoblot revealed that 18 (out of 19 total) variants were stable and expressed at levels equal to or exceeding those of sPBP 5. To ensure that the gene encoding sPBP 5-K213R had not reverted back to the wild-type sequence, the plasmid was isolated and submitted to double-stranded sequencing, which confirmed the presence of an arginine codon (CGT) at position 213. On the basis of this binding activity, we selected sPBP 5-K213R to further characterize its catalytic activity.
Activities of sPBP 5-K213R"To determine whether any subtle effects had occurred within the active site in sPBP 5-K213R, the affinities of sPBP 5 and sPBP 5-K213R for ["C] penicillin G were determined. Because of the nature of the reaction of penicillin G with PBPs, it should be noted that the term affinity is defined as the concentration of antibiotic that results in the covalent labeling of 50% of the particular PBP under defined conditions (in this case, 20 min at 30 "C; see Ref. 31 for further details). The saturation curves shown in Fig. 4 were obtained after incubating the sPBP 5 proteins with [14C]penicillin G at various concentrations for 20 min at 30 "C. From these data, it was determined that sPBP 5-K213R (affinity = 4.4 pg/ml) has a 1.2-fold higher affinity for ["C] penicillin G compared with sPBP 5 (affinity = 5.2 pg/ml).
The effect of the arginine mutation on the interactions of sPBP 5 with penicillin G was further assessed by determining the rate of deacylation of the [14C]penicilloyl-sPBP complex. The purified protein was incubated with a saturating amount of [14C]penicillin G for 20 min at 30 "C. A 100-fold excess of nonradioactive penicillin G was added at t = 0, and at various times, aliquots of the proteins were removed, denatured in SDS, and subjected to SDS-PAGE and fluorography. The amount of radioactivity remaining covalently attached to protein was measured by densitometry. The first-order rate constants and half-lives for the enzymatically catalyzed hydrolysis of penicillin were calculated from a semilog plot of percent activity remaining uersus time of incubation ( Fig. 5 and Table I). The identical half-lives of the penicilloyl-sPBP complexes for both sPBP 5 (tIl2 = 9.2 min) and sPBP 5-K213R (t112 = 9.1 min) clearly indicate that the arginine mutation has no effect upon the rate of hydrolysis of the acylenzyme complex. As a comparison, the hydrolysis rate of sPBP 5' is also shown. sPBP 5' contains a point mutation (G105D) that results in a drastically reduced deacylation rate of the penicilloyl-sPBP 5' complex while still allowing nearnormal acylation with penicillin G (33,34). As reported previously (22) G by sPBP 5, sPBP 5 ' , and sPBP 5-K213R. Purified sPBP 5 proteins (6 pg) were incubated with 40 pg/ml ["C]penicillin G in 25 mM Tris.HC1, pH 7.3, for 20 min a t 30 "C. A 100-fold excess of nonradioactive penicillin G was then added ( t = 0); and aliquots were removed at the indicated times, denatured, and submitted to electrophoresis on a 12% SDS-polyacrylamide gel. Following fluorography, the radioactivity remaining covalently attached to protein was determined by densitometric scanning.

ND
Values were calculated from the data in Fig. 4. *Values were calculated from the data in Fig. 5. 'Values were determined from double-reciprocal plots of initial rates.
ND, not detected.
was extremely long (244 min) when compared with that of sPBP 5-K213R or sPBP 5 (9 min). The half-life determined for sPBP 5 differed slightly from the value determined previously (10.5 min) and may be a result of the higher pH (7.3 versus 7.0) used in these experiments. Because sPBP 5-K213R showed essentially no difference in its interaction with penicillin G, we reasoned that it would also catalyze D-alanine carboxypeptidase activity. To conduct D-alanine carboxypeptidase assays, it was necessary to purify both sPBP 5 and sPBP 5-K213R to homogeneity in large yields. Utilizing covalent ampicillin affinity chromatography, sPBP 5 (1.0 mg) and sPBP 5-K213R (2.6 mg) were purified from osmotic shock fluids and were used to assess D-alanine carboxypeptidase activity (22). The kinetic parameters of the release of D-alanine were determined from the purified proteins and the substrate L-Ala-D-r-Glu-L-Lys-D-Ala-D-Ala (Table I). Surprisingly, whereas sPBP 5 displayed D-alanine carboxypeptidase activity, no activity could be detected with sPBP 5-K213R. The purified protein was reassayed for penicillin binding at the same time to determine whether the protein had lost its binding activity, but it showed the identical activities that were determined in osmotic shock fluid. A competition experiment in which L-Ala-D-y-Glu-L-Lys-D-Ala-D-Ala was preincubated at pH 8.5 with sPBP 5-K213R for various times before the addition of [14C]penicillin G showed that sPBP 5-K213R did not appear to accumulate an acyl-enzyme complex (data not shown). Our sPBP 5 showed a reproducibly higher K, (8.0 f 0.5 mM) than the sPBP 5 (5.0 f 0.8 mM) reported previously (22), but it is not clear why these values are different from one another. DISCUSSION The proposed role of the histidine residue from the Box VI1 region in the catalytic mechanism of the Streptomyces R61 DD-peptidase was based upon the molecular coordinates of the histidine residue within the active-site cavity determined from the high-resolution crystal structure. The histidine side chain appears to be in position to contribute to the initial binding of substrate and p-lactam antibiotics, presumably by providing a positive charge that can interact with the carboxylate moiety of these ligands (16). We were interested in this hypothesis and sought to confirm the role of this amino acid in another low molecular weight PBP (PBP 5). The role of Lys213 from the Box VI1 region of PBP 5 in the catalytic mechanism of this D-alanine carboxypeptidase enzyme was assessed by the technique of site-saturation mutagenesis. We reasoned that if the protonated amino group on Lys is responsible only for forming an electrostatic interaction with the penicillin carboxylate, then one might expect that either arginine or histidine could substitute for lysine in the active site of sPBP 5. If L y P 3 directly participates in the acid-base reactions that occur during the acylation and deacylation of penicillin G with PBP 5 (as might the histidine residue), then only the protein with a histidine at this position should be able to bind [3H]penicillin G. Finally, if Lys213 is not important in the binding of penicillin G, then at least some of the mutations should show significant binding activity.
Using site-saturation mutagenesis, we substituted lysine 213 of the Box VI1 region of sPBP 5 with 18 other amino acids and tested the ability of these mutant proteins to bind [3H]penicillin G. Only one of these proteins (sPBP 5-K213R) was capable of forming a stable covalent complex with penicillin. It is very illuminating that only arginine was capable of substituting for lysine at this position; since the high pK, of the guanidinium group of arginine makes it unsuitable for participating in acid-base catalysis, the results strongly sug-gest that it is simply the positive charge contributed by these 2 residues that is required for penicillin binding. The inability of histidine to substitute for lysine was unexpected since histidine is found at the same position in the Box VI1 region of Streptomyces R61 DD-peptidase. Since histidine is capable of contributing its positive charge in the Streptomyces enzyme, there are apparently other subtle conformational differences within the active site of sPBP 5 that prevent the productive interaction of the histidinium ion with the penicillin carboxylate moiety.
The activity of sPBP 5-K213R was further investigated, and the mutant was found to exhibit wild-type values of binding affinity and hydrolysis of [l*C]penicillin G. The equivalence of both of these properties in sPBP 5 and sPBP 5-K213R suggests that the arginine residue has virtually no effect on the structure of the active-site cavity in respect to penicillin binding. Interestingly, when D-alanine carboxypeptidase activity was assayed, sPBP 5-K213R consistently showed no activity, even though its interaction with penicillin was normal. This suggests that the arginine residue, although situated in a suitable geometry for interacting with penicillin, is unable to form a productive electrostatic interaction with the substrate carboxylate. This in part may be due to a different positioning of the substrate carboxylate in the active site as compared with the penicillin carboxylate, which is supported by modeling studies of L-lysyl-D-Ala-D-Ala and cephalosporin C in the active-site cavity of the Streptomyces R61 DD-peptidase (16). Another reason for the loss of Dalanine carboxypeptidase activity in sPBP 5-K213R may be the different geometry of the arginine side chain as compared with the lysine side chain. This difference includes both a 1-A increase in the length of the arginine side chain as compared with the lysine side chain and the planar geometry of the guanidinium moiety. These structural differences may influence the binding of substrate such that arginine is not capable of interacting with the substrate carboxylate. Although we cannot be sure of the structural perturbations that have occurred in the active site when Lys213 is replaced by Arg, it does appear that the positioning of the substrate carboxylate in the initial Michaelis complex is different from the penicillin carboxylate and that the arginine residue is capable of discriminating between the two compounds.
Other reports have detailed experiments aimed at the role of lysine in the Box VI1 region of 8-lactamases. The role of Lys234 (from the KTG triad) in the catalytic mechanism of B. licheniformis 6-lactamase was addressed by site-directed mutagenesis, in which it was mutated to either an alanine or glutamate residue (35). These changes resulted in alterations (by 1-3 orders of magnitude) of both K, and kcat, and this led to the conclusion that Lys234 is involved in both ground-state and transition-state binding. The primary sequence of the class C p-lactamase PSE-4 has been recently shown to exhibit a notable variation in the conserved triad, in which the Lys residue is replaced with Arg (36). This was the first report of any native penicillin-interactive protein that contains Arg at this position and supports the view that this residue provides a positively charged electrostatic environment for the binding of the substrate carboxylate. Recently, a report appeared in which Lys234 from the KSG triad of TEM p-lactamase was mutated to an arginine residue (37). The arginine-substituted @-lactamase displayed near wild-type levels of activity with cephalosporins, but showed a 10-fold decrease in K, with no effect on kcat for several penicillins, including penicillin G.
These results also suggest that arginine is capable of replacing lysine in the KSG triad and maintaining near normal activity (at least for some substrates). Our results indicate that the mutation of Lys213 to Arg in sPBP 5 does not affect any of the parameters of penicillin G binding.
In contrast to its interaction with penicillin G, sPBP 5-K213R does not catalyze any detectable D-alanine carboxypeptidase activity. We do not know which of the three parameters in the kinetic scheme of D-alanine carboxypeptidation (binding, acylation, or deacylation) were altered in sPBP 5-K213R. Of these three, we favor the alteration of the binding affinity and/or acylation rate as the parameter(s) most likely altered in the mutant. We base this choice on two factors: 1) the deacylation rate displayed by sPBP 5-K213R for ["C] penicillin G is identical to wild-type values, which indicates that the functioning of the catalytic groups responsible for deacylation are not altered; and 2) no accumulation of an acyl-enzyme complex with the peptide substrate could be observed. The explanation behind the lack of D-alanine carboxypeptidase activity in sPBP 5-K213R is therefore likely to be fundamentally different from that in sPBP 5', which also lacks D-alanine carboxypeptidase activity yet binds penicillin. sPBP 5' lacks penicillin deacylation activity, and its ability to accumulate the acyl-enzyme complex with the depsipeptide substrate (Ac)2-L-Lys-D-Ala-D-lactate can also be correlated with an alteration in k3 (deacylation rate constant) (34). Thus, sPBP 5' has an altered but near-normal acylation rate with both substrate and penicillin G, but is unable to hydrolyze the acyl-enzyme complex. sPBP 5-K213R shows no such defect in the rate of hydrolysis of penicillin G.
It is clear that we shall not truly understand the structural changes that have occurred in the mutation of Lys-213 to Arg until the crystal structure of sPBP 5 has been elucidated. Work on solving the three-dimensional structure of sPBP 5' is in progress'; and once this work comes to fruition, it will be very informative to determine what structural perturbations have occurred within the active site of sPBP 5-K213R.
We have shown that this mutant can be isolated in high yield and purity, and it should be well suited for growing crystals. Comparison of the crystal structures of sPBP 5, sPBP 5', and sPBP 5-K213R should allow us to further define the roles of the active-site residues that participate in binding, acylation, and deacylation of substrate and penicillin.