Selection and Characterization of Amino Acid Substitutions at Residues 237–240 of TEM-1 (cid:98) -Lactamase with Altered Substrate Specificity for Aztreonam and Ceftazidime*

Recently, natural variants of TEM-1 (cid:98) -lactamase with amino acid substitutions at residues 237–240 have been identified that have increased hydrolytic activity for extended-spectrum antibiotics such as ceftazidime. To identify the sequence requirements in this region for a given antibiotic, a random library was constructed that contained all possible amino acid combinations for the 3-residue region 237–240 (ABL numbering system) of TEM-1 (cid:98) -lactamase. An antibiotic disc diffusion method was used to select mutants with wild-type level activity or greater for the extended-spectrum cephalosporin ceftazidime and the monobactam aztreonam. Mutants that were selected for optimal ceftazidime hydrolysis contained a conserved Ala at position 237, a Ser for Gly substitution at position 238, and a Lys for Glu at position 240. Mutants selected for aztreonam hydrolysis exhib- ited a Gly for Ala substitution at position 237, a Ser for Gly substitution at position 238, and a Lys/Arg for Glu at position 240. The role of the A237G substitution in dif-ferentiating between ceftazidime and aztreonam was further investigated by kinetic analysis of the A237G, E240K, G238S:E240K, and A237G:G238S:E240K enzymes. The A237G single mutant and the G238S:E240K double mutant exhibited increases in catalytic efficiency for both ceftazidime and aztreonam. However, the triple mutant A237G:G238S:E240K, 8 chosen to bracket

The most common mechanism of bacterial resistance to ␤-lactam antibiotics is the production of ␤-lactamase, a bacterial enzyme that catalyzes the hydrolysis of ␤-lactams. ␤-Lactamases are grouped into 4 classes (A, B, C, and D) based on primary sequence. TEM-1 ␤-lactamase, a class A serine hydrolase, is the most prevalent ␤-lactamase found in Gram-negative bacteria. TEM-1 ␤-lactamase is capable of hydrolyzing both penicillins and cephalosporins. However, it cannot hydrolyze the recently developed extended-spectrum antibiotics, such as ceftazidime and aztreonam. Extended-spectrum antibiotics were developed in part to combat Gram-negative bacteria that had developed resistance to existing penicillins and first and second generation cephalosporins through the expression of ␤-lactamases such as TEM-1. However, within a few years after the introduction of the extended-spectrum antibiotics, clinical isolates were discovered that were capable of hydrolyzing these new antibiotics (1). These enzymes, termed extendedspectrum ␤-lactamases, contain one to four amino acid substitutions near the active site and are derived either from the TEM or SHV ␤-lactamases (2). These substitutions occur at residues 104, 164, and 237-240, both as individual and as combinatorial mutations (numbering of amino acids according to Ambler et al. (26)). Specifically, the substitutions of a lysine for glutamate at position 104 (E104K), a serine for arginine at position 164 (R164S), a serine for a glycine at position 238 (G238S), and a lysine for a glutamate at position 240 (E240K), either alone or in combination, have been shown to provide increased catalytic activity toward the extended-spectrum cephalosporin ceftazidime (3)(4)(5)(6). Understanding how amino acid substitutions alter the substrate specificity of ␤-lactamase for a given antibiotic may aid in the design of new antibiotics that avoid inactivation by extended spectrum ␤-lactamases. For example, by knowing if amino acid substitutions in extended spectrum ␤-lactamases act through enhanced binding of a specific part of an antibiotic side chain it may be possible to modify that part of the antibiotic to avoid inactivation. In addition, knowledge of how mutations alter the substrate specificity of ␤-lactamase may be of direct empiric value for combination therapy. If the set of substitutions in ␤-lactamase that result in increased activity are different for two antibiotics then the combination of those antibiotics in therapy may avoid the development of resistance.
Previously, random replacement mutagenesis has been used to randomize the nucleotide sequence of three contiguous codons in the bla TEM-1 gene to create libraries that encode all possible amino acid combinations for the target region (6 -10). In the random library covering residues 238 -241, ␤-lactamase mutants were selected and characterized having 100-fold higher activity than the wild-type level toward ceftazidime (6). That study determined that the G238S and E240K substitutions were the primary cause for ceftazidime resistance while substitutions at position 241 provided only a minor improvement.
While residue 241 is on the outer edge of the active site, crystallographic studies of TEM ␤-lactamase have shown residue 237 to be part of the hydrogen-bond network that stabilizes initial binding of the substrate in the active site (11,12). Since residues 237-240 are the only contiguous residues in which substitutions have been shown to greatly increase the hydrol-ysis spectrum of TEM ␤-lactamase (4 -6, 8, 13), and residue 237 seems to have a greater importance than residue 241 in hydrolysis, the region 237-240 is more relevant for the study of enzyme-substrate interactions in extended-spectrum ␤-lactamases. In this report, random replacement mutagenesis has been used to create the random library L237-240, 1 which contains all possible amino acid combinations for residues 237-240. This library was used to select mutants with high levels of activity toward the cephalosporin ceftazidime and the monobactam aztreonam to study the importance of this region in determining substrate specificity. Ceftazidime and aztreonam contain an identical aminothiazole-oxime side chain and so amino acid substitutions that affect ceftazidime hydrolysis may exhibit a similar effect on aztreonam hydrolysis. In fact, analysis of kinetic parameters of the E104K, R164S, G238S, and E240K enzymes indicates that the catalytic efficiency of each of these enzymes is increased for both ceftazidime and aztreonam (4,5). In this study, the majority of mutants had a serine for glycine substitution at position 238 and a lysine or arginine for glutamate substitution at position 240, which reinforces the importance of the G238S and E240K/R substitutions in the hydrolysis of antibiotics with an aminothiazoleoxime side chain (3)(4)(5)(6). All of the mutants selected for ceftazidime hydrolysis had the wild-type alanine conserved at position 237, but 17 out of 18 mutants selected for aztreonam hydrolysis had an A237G substitution. The role of the individual substitutions was determined by introducing the substitutions individually and in combination. Both the A237G single and G238S:E240K double substitution improve catalytic efficiency for both ceftazidime and aztreonam. However, when the A237G substitution is added to the double mutant, catalytic efficiency is decreased 12-fold for ceftazidime but is increased 3-fold for aztreonam. Thus, the effect of the A237G substitution on ceftazidime hydrolysis is strongly dependent on whether it resides in the wild-type or G238S:E240K enzyme.

EXPERIMENTAL PROCEDURES
Materials-All enzymes were purchased from New England Biolabs, except T7 DNA polymerase, which was purchased from U. S. Biochemical Corp. Ceftazidime was provided by Glaxo (Greenford, United Kingdom) and aztreonam was provided by Bristol-Meyers Squibb, Inc. Nitrocefin and antibiotic paper discs containing either ceftazidime or aztreonam were purchased from Becton Dickinson (Cockeysville, MD). G-75 Sephadex was obtained from Pharmacia Biotech Inc. (Piscataway, NJ).
Plasmid pBG66 is the parent plasmid of all random library constructions (9). The pBG66 plasmid is a 4.8-kilobase derivative of pBR322 and pBR325 that contains the wild-type bla TEM-1 gene and the cat gene, which encodes for chloramphenicol acetyltransferase. The plasmid also contains the ColE1 and f1 origins of DNA replication.
Oligonucleotides and Random Replacement Mutagenesis-Oligonucleotide primers used for mutagenesis and DNA sequencing were synthesized by Genentech, Inc., at the PAN facility at Stanford University Medical School, and at Genosys Biotechnologies, Inc. Random replacement mutagenesis was done as described previously (17).
Briefly, the strategy behind the modified mutagenesis was to first insert a unique XhoI restriction site into a location within the bla gene, which has been targeted for mutagenesis. The XhoI recognition sequence is flanked by two 12 base arms which are complementary to the sequence adjacent to the site targeted for mutagenesis. The restriction site was positioned at or near the middle of the three codons to be randomized, and the second base of the middle codon was deleted to create a frameshift mutation. A XhoI restriction site was previously inserted at codon 238 (9). The frameshift mutation, resulting from this insertion, renders the bla gene non-functional. Subsequent randomization is achieved by replacing the unique restriction site with a 9-base randomized DNA sequence. An oligonucleotide was designed to replace the 9-base window (including the XhoI site) with random sequence, 5Ј-NNS NNS NNS-3Ј (where N indicates an equal probability of any base, and S indicates an equal probability of either C or G). This insured all amino acids would be sampled in the window. Two 14-base complementary arms flank the random sequence. The calculations involved in the determination of the percent of randomization have been described previously (9). Library DNA was electroporated into E. coli XL1-B cells for further screening.
Plasmid Isolation and DNA Sequencing-Single-stranded plasmid DNA was prepared for sequencing as described (18). DNA sequencing was performed using the dideoxy chain termination method (19). Oligonucleotides were designed to prime synthesis from specific sites within the bla TEM-1 gene.
Selection of Active Mutants-Selections were done using the disc diffusion method. The TEM-1 ␤-lactamase or the L237-240 was introduced into E. coli XL1-B and was grown in 2 ϫ YT media plus 12.5 g/ml chloramphenicol to stationary phase with aeration at 37°C. A 1:20 dilution of this culture was made with Mueller-Hinton media, and a sterile cotton swab was used to plate the dilution on Mueller-Hinton agar plates. Antibiotic paper discs containing either 30 g of ceftazidime or aztreonam were applied to the surface of the plates which were then incubated overnight at 37°C. The zone of inhibition for wild-type ␤-lactamase was used as a control. Any colonies from L237-240 that grew near or within the wild-type zone of inhibition were considered to have greater than or equal to TEM-1 activity.
Antibiotic Susceptibility-Minimum inhibitory concentrations (MICs) were determined by broth microdilution. 1 ϫ 10 4 E. coli XL1-B cells containing the selected mutant ␤-lactamase were inoculated into microtiter wells containing 100 l of LB media having 2-fold dilutions of the antibiotic being tested. The ranges of antibiotic concentrations tested for ceftazidime and aztreonam were 0.03-64 g/ml. The microtiter plates were incubated at 37°C for 18 -24 h. The plates were examined visually, and the lowest concentration of antibiotic that inhibited visual growth was recorded as the MIC. A 2-fold difference in MIC values was determined to be insignificant.
Site-directed Mutagenesis-The E240K and G238S:E240K mutants were previously constructed by cassette mutagenesis (6). The A237G mutant was originally isolated from the L235-237 random library in a selection for ampicillin resistance (9).
␤-Lactamase Purification-TEM-1 ␤-lactamase and the mutant ␤-lactamases were purified to Ͼ90% homogeneity. The wild-type TEM-1, E240K, and G238S:E240K enzymes were previously cloned into an expression vector under the control of the tac promoter (6). E. coli XL1-Blue cells containing the expression vector were grown in LB media (18) containing 25 g/ml kanamycin to early log phase. At an OD 600 of ϳ0.5, 0.1 mM isopropyl-1-thio-D-galactopyranoside was added and the culture was incubated an additional 3 h until it reached late log phase. In contrast, the A237G and A237G:G238S:E240K enzymes were purified from E. coli XL1-B cells containing the pBG66 plasmid and were expressed from the natural ␤-lactamase promoter. These strains were grown in culture for 9 h to reach late log phase. From this point, all enzymes were purified by the method that follows. ␤-Lactamase and other periplasmic proteins were first isolated by an osmotic shock procedure (20). The solution obtained by osmotic shock was adjusted to a final concentrate of 100 mM NaAc, pH 7.5, 800 mM NaCl (buffer A). The protein solution was concentrated to a 5-ml volume with an Amicon Centriprep-10 concentrator. The concentrated protein solution was then applied to a 1-ml HiTrap zinc chelating column (Pharmacia). ␤-Lactamase bound strongly to the column while other periplasmic proteins eluted out with buffer A. ␤-Lactamase was eluted using a linear gradient of buffer B (100 mM NaAc, pH 4.0, 800 mM NaCl). Fractions containing ␤-lactamase activity were identified by nitrocefin hydrolysis and SDS-polyacrylamide gel electrophoresis. The ␤-lactamase fractions were concentrated with a Centriprep-10 concentrator. The ␤-lactamase was further purified by Sephadex G-75 gel filtration chromatography using a 25 mM sodium phosphate buffer, pH 7.0. The purity of the enzymes was verified by SDS-polyacrylamide gel electrophoresis. All preparations were Ͼ90% pure. The purified enzymes were stored at 4°C until the determination of kinetic parameters.
Enzyme Kinetics-The kinetics of TEM-1 ␤-lactamase and the se-lected mutants were determined with ampicillin, cephaloridine, ceftazidime, and aztreonam. The extinction coefficients used were: ampicillin, 235 nm, ⌬⑀ ϭ 900 M Ϫ1 cm Ϫ1 (6); cephaloridine, 260 nm, ⌬⑀ ϭ 10,200 (22). The hydrolysis of the substrates was monitored spectrophotometrically at 30°C in 0.05 M phosphate buffer, pH 7.0, on a Beckman model DU 640 spectrophotometer using a 0.1-or 1.0-cm pathlength cuvette as needed. Kinetic parameters V max and K m were determined by: 1) initial velocity kinetic analysis by fitting to the equation of a rectangular hyperbola using unweighted nonlinear least squares (23), or 2) using complete progress curves with starting concentrations at least 3 ϫ K m . The progress curves were evaluated by determining the instantaneous rate at a minimum of 50 points in the reaction by taking the derivative of the substrate concentration at those points in the reaction course. This data was fitted to the Michaelis-Menton equation using unweighted nonlinear least squares (23). To insure that substrate inhibition was not occurring, initial rates were determined at 6 to 8 substrate concentrations chosen to bracket the K m value determined by the progress curve. In the case of aztreonam, the K m was measured as the K i with cephaloridine as the reporter substrate. The enzyme and inhibitor were preincubated for 5 min at 30°C before the substrate was added. At least five concentrations of reporter substrate bracketing the K m were tested with at least three concentrations of inhibitor. Enzyme concentration was at least 10-fold less than inhibitor concentrations and at least 100-fold less than reporter substrate concentrations. K m /V max values were then plotted versus inhibitor concentration to obtain K i . These experiments were done at least in triplicate and analyzed using the EnzymeKinetics v1.42 program from Trinity Software. The k cat was obtained directly from the linear de-  (2). Note that residues 238 and 240 are adjacent to each other in the TEM sequences. The absence of residue 239 is due to the ABL numbering system for Class A ␤-lactamases (26). In previous experiments with the random library 238 -241, ␤-lactamase mutants were selected that had 100-fold greater resistance than wild-type toward ceftazidime (6). It was found that the multiple substitution of G238S:E240K was responsible for the large increase in catalytic efficiency for ceftazidime.
In order to better study the role of substitutions at positions important for extended-spectrum ␤-lactamase activity, the random library 237-240 was constructed. Selection experiments were done using the extended-spectrum cephalosporin ceftazidime and the monobactam aztreonam (Fig. 1). The similarity of these substrates should provide further information about the ability of substitutions in this region to not only increase cat- Altered Substrate Specificity of ␤-Lactamase alytic efficiency but also to differentiate between similar substrates. The L237-240 library was transformed into E. coli XL1-B cells to isolate mutants with high level activity for ceftazidime or aztreonam. The transformants were selected using a disc diffusion method with discs containing either 30 g of ceftazidime or 30 g of aztreonam. E. coli cells containing wild-type TEM-1 ␤-lactamase or the L237-240 library were spread on Mueller-Hinton agar plates and the antibiotic discs were applied to the surface of the plates. The antibiotic diffuses outward from the disc forming a circular concentration gradient on the agar plate. After an overnight incubation at 37°C, a zone of inhibited bacterial growth will be visible around each antibiotic disc. The radius of the zone of inhibition of wild-type enzyme for each antibiotic is a relative indicator of the wildtype level of activity and was used as a control for these experiments. Only mutant transformants with greatly enhanced activity toward the antibiotic being tested could survive at a much closer radius to the antibiotic disc than the cells containing the wild-type enzyme. The transformants growing within the wild-type zone of inhibition radius on the random library plates were selected for further study. The disc diffusion selection was performed twice for each antibiotic.
The DNA sequences of 20 mutants selected for ceftazidime hydrolytic activity (10 from each selection) and 18 mutants selected for aztreonam hydrolytic activity (10 from the first selection and 8 from the second) were determined to identify the substitutions that resulted in increased ␤-lactamase activity toward these antibiotics. For the ceftazidime selection ( Fig.  2A), alanine was conserved at position 237 for all the mutants, suggesting that alanine is the optimal amino acid at this position for ceftazidime hydrolysis. At position 238, 15 of 20 mutants had a serine while 5 of 20 had an asparagine residue. At position 240, all 20 mutants had either a lysine or arginine. The predominance of serine at 238 and lysine at 240 is in agreement with previous studies showing the importance of these substitutions for ceftazidime hydrolysis (3,4,6). For the aztreonam selection (Fig. 2B), glycine was observed at position 237 in 17 of the 18 mutants sequenced, suggesting that glycine is the optimal amino acid for aztreonam hydrolysis. At position 238, 12 of 18 mutants had G238S, 5 of 18 had G238T, and 1 of 18 had G238N. At position 240, lysine or arginine was observed in 15 of 18 mutants selected for aztreonam hydrolysis while the remaining mutants contained either glutamine or threonine. The predominance of serine at 238 and of lysine or arginine at 240 in the aztreonam selection matches the profile of the ceftazidime selection. This result suggests the substitutions at position 238 and 240 act via the aminothiazole-oxime side group that is common to both antibiotics. In contrast, the sequence requirements at position 237 for maximal aztreonam hydrolysis appear to be different from those required for maximal ceftazidime hydrolysis.
Antibiotic Susceptibility-To better understand the extent to which the mutations conferred increased hydrolytic activity, the MICs were determined for aztreonam and ceftazidime for all the selected mutants (Fig. 2). The MICs for ceftazidime were 60-to 100-fold greater than wild-type for mutants with the G238S and E240K/R substitutions. However, the G238S: E240K/R mutants that contained the additional substitution of A237G had MICs only 30-fold greater than wild-type. Thus, the addition of the A237G substitution to the G238S:E240K double mutant decreased the activity of the enzyme toward ceftazidime.
The MIC's for aztreonam were 250-fold higher than wildtype for the G238S:E240K double mutant. However, in contrast to the ceftazidime MIC's, the addition of the A237G substitution to the double mutant resulted in a further increase in the aztreonam MIC. These results suggest that the A237G substitution allows the enzyme to discriminate between aztreonam and ceftazidime.
The kinetics of ampicillin and cephaloridine hydrolysis (Tables I and II) were determined to compare the effect of the substitutions on the hydrolysis of a penicillin and a non-extended-spectrum cephalosporin with that of ceftazidime and aztreonam. The effect of the A237G substitution on the hydrolysis of ampicillin and cephaloridine followed a similar pattern in that for both substrates the K m value was increased relative to the wild-type enzyme. The double mutant G238S:E240K exhibits a large decrease in both k cat and K m values. This has been reported previously and it was shown that the majority of this effect is due to G238S substitution (6). The K m values determined for the G238S:E240K enzyme in this study are somewhat lower than has been described previously (6). However, all data reported here was done with the same batches of enzyme and antibiotic substrate to facilitate a direct comparison of values. The addition of the A237G substitution to the G238S:E240K enzyme increases the K m value of the triple mutant for both ampicillin and cephaloridine relative to the G238S:E240K double mutant. Therefore, both alone and in combination with the G238S:E240K substitutions, the A237G substitution increases the K m for ampicillin and cephaloridine hydrolysis.
As seen in Table III, all of the mutant enzymes had increased catalytic efficiency for ceftazidime. With ceftazidime as substrate, a high K m value (Ͼ1000 M) prevented the individual values of k cat and K m from being determined for all but the G238S:E240K enzyme. The k cat /K m values were estimated by examining the reaction rate at [S] Ͻ Ͻ K m (24). As described previously (6), the effects of the G238S and E240K substitutions on ceftazidime hydrolysis were that G238S causes a significant decrease in K m , allowing the k cat and K m of G238S: E240K to be measured, and E240K causes an increase in k cat and a small decrease in K m (6). Accordingly, the double mutant G238S:E240K provides a 1294-fold increased catalytic efficiency (k cat /K m ) over the wild-type enzyme. Interestingly, the A237G substitution alone provides a 13-fold increase in catalytic efficiency for ceftazidime, but the triple mutant A237G: G238S:E240K provides only a 109-fold increase, which is 12fold less than the double mutant. The results suggest that the A237G substitution, while beneficial on its own, detracts from the contributions of the G238S:E240K double mutant. Because the K m of the G238S:E240K double mutant is 178 M and the K m of the A237G:G238S:E240K triple mutant is too high to measure (Ͼ1000 M), it can be concluded that the addition of the A237G substitution to the double mutant results in at least a 5-fold increase in K m . Note that an increase in K m is also observed with ampicillin and cephaloridine as substrates (Tables I and II). However, because of the inability to measure K m in the A237G single mutant, it is unclear whether the increase in catalytic efficiency of this enzyme is due to decreased K m or increased k cat or both.
With aztreonam as the substrate (Table IV), each substitution lowered the K m value such that k cat and K m could be determined for each enzyme except wild-type. The effect of the A237G and E240K substitutions was to lower the K m to a level that we could measure. An even larger reduction, relative to that of A237G and E240K, was observed for the K m values of G238S:E240K and A237G:G238S:E240K, with reductions of an additional 200 -600-fold. The addition of the G238S substitution to the E240K enzyme had the effect of further reducing both K m and k cat . In contrast to the kinetic patterns observed with ceftazidime, the triple mutant A237G:G238S:E240K had the highest catalytic efficiency for aztreonam, with a 3437-fold increase over the wild-type enzyme. The reason for this is that the addition of the A237G substitution to the G238S:E240K enzyme has the effect of lowering K m and thus providing a 3-fold increase in catalytic efficiency over the G238S:E240K double mutant. It is interesting that the A237G substitution resulted in a lower K m value when aztreonam was the substrate versus a higher K m value for ampicillin, cephaloridine, and ceftazidime.
A comparison of the substrate specificity, as measured by the catalytic efficiency, of wild-type TEM-1 ␤-lactamase and each of the mutant enzymes is shown in Fig. 3. It is apparent from the data that the wild-type enzyme has very high catalytic efficiency for ampicillin and cephaloridine and poor catalytic efficiency for ceftazidime and aztreonam. The double mutant G238S:E240K, although, has moderately high catalytic efficiency for all substrates. With the addition of A237G, however, the triple mutant exhibits a further increase in catalytic efficiency for aztreonam hydrolysis while showing reduced efficiency for ampicillin, cephaloridine, and ceftazidime hydrolysis. Thus, the double substitution of G238S:E240K provides the wild-type enzyme with the ability to hydrolyze antibiotics with an aminothiazole-oxime side group, and the further addition of A237G allows the enzyme to differentiate between a monobactam and a cephalosporin with the aminothiazole-oxime side group.

DISCUSSION
The goal of this study was to identify substitutions in the 237-240 region of TEM-1 ␤-lactamase that alter specificity and enhance catalytic efficiency toward different antibiotic substrates, such as the cephalosporin ceftazidime and the monobactam aztreonam. In order to accomplish this goal, random replacement mutagenesis was used to completely randomize residues 237-240 to form a random library that contains all the possible amino acid combinations for that region. Then, mutants were selected that had greater than the wild-type   level of activity toward the given antibiotic. It was found that the preferred sequence for ceftazidime hydrolysis was a conserved alanine at position 237, a serine or asparagine at position 238, and a lysine or arginine at position 240. The preferred sequence for aztreonam hydrolysis involved more variability and was found to have a glycine at position 237, a serine, threonine, or asparagine at position 238, and a lysine, arginine, or threonine at position 240. Previous studies with TEM-1 and SHV-type ␤-lactamases have shown the importance of the G238S:E240K double mutant for ceftazidime hydrolysis (3,6). For TEM-1 ␤-lactamase, the G238S:E240K double substitution has been shown to increase k cat while decreasing K m (6). Characterization of these enzymes for aztreonam hydrolysis indicate that the increases in k cat /K m are mostly due to very large reductions in K m . To better understand the role of the A237G substitution, we determined the contributions of the A237G, E240K, G238S: E240K, and A237G:G238S:E240K substitutions in ampicillin, cephaloridine, ceftazidime, and aztreonam hydrolysis. Characterization of these enzymes for ampicillin and cephaloridine hydrolysis indicate that the A237G substitution results in an increased K m value which causes a slight decrease in the overall catalytic efficiency for these substrates (Tables I and II). However, characterization of these enzymes for ceftazidime hydrolysis indicate that, although A237G alone provides an increase in catalytic efficiency, when it is added to the G238S: E240K double mutant a reduction in catalytic efficiency occurs and therefore the A237G substitution detracts from the contri-butions of the other substitutions.
In contrast to the hydrolysis of ceftazidime, the A237G substitution is beneficial for aztreonam hydrolysis both individually and in combination with the G238S:E240K double mutant. The k cat /K m of the A237G mutant is increased 19-fold more than the wild-type enzyme (Table IV). In addition, the k cat /K m of the G238S:E240K double mutant for aztreonam hydrolysis is increased 3-fold when the A237G substitution is added (Table  IV). Whereas optimal ceftazidime hydrolysis is achieved by the double mutant G238S:E240K, optimal aztreonam hydrolysis is achieved by the triple mutant A237G:G238S:E240K.
The contributions of the substitutions to the overall catalytic efficiency can be calculated for each substrate from the data in Tables I through IV. This was done using the following equation (24) Tables I-IV. Each substrate is listed on the horizontal axis.
of the functionally equivalent carboxylate group (Fig. 1). The sulfonic acid is therefore in a different position relative to the carbonyl-carbon of the lactam ring than is the carboxylate. This may result in weaker interactions between the sulfonic acid oxygens and the side chains of Ser-235 and Arg-244. A model of aztreonam in the active site of the wild-type TEM-1 ␤-lactamase is shown in Fig. 4. The A237G substitution may act by altering the conformational flexibility of the S3 ␤-strand on which both Ser-235 and Ala-237 are found. This could reposition the side chain of Ser-235 so that it could form a stronger hydrogen bond with the sulfonic acid group. Alternatively, the removal of the methyl group at position 237 in the A237G mutant may allow the side chain of Arg-244 to rotate to a position closer to the sulfonic acid group and form a stronger interaction. Either of these possibilities might improve the binding of the substrate and account for the reduced K m value. A change in enzyme mechanism also must be considered. Acylation has been determined as the rate-limiting step of TEM-1 ␤-lactamase for the extended-spectrum cephalosporin cefotaxime, and the G238S substitution was shown to improve the acylation rate, although not enough to change the rate-limiting step (31). The large reduction in K m for aztreonam could be explained by a possible change in rate-limiting step from acylation to deacylation. However, a detailed interpretation of the effect of the A237G substitution awaits further structural and kinetic data on the enzyme.