Probing Active Site Chemistry in SHV β-Lactamase Variants at Ambler Position 244

Inhibitor-resistant class A β-lactamases are an emerging threat to the use of β-lactam/β-lactamase inhibitor combinations (e.g. amoxicillin/clavulanate) in the treatment of serious bacterial infections. In the TEM family of Class A β-lactamases, single amino acid substitutions at Arg-244 confer resistance to clavulanate inactivation. To understand the amino acid sequence requirements in class A β-lactamases that confer resistance to clavulanate, we performed site-saturation mutagenesis of Arg-244 in SHV-1, a related class A β-lactamase found in Klebsiella pneumoniae. Twelve SHV enzymes with amino acid substitutions at Arg-244 resulted in significant increases in minimal inhibitory concentrations to ampicillin/clavulanate when expressed in Escherichia coli. Kinetic analyses of SHV-1, R244S, R244Q, R244L, and R244E β-lactamases revealed that the main determinant of clavulanate resistance was reduced inhibitor affinity. In contrast to studies in the highly similar TEM enzyme, we observed increases in clavulanate kinact for all mutants. Electrospray ionization mass spec-trometry of clavulanate inhibited SHV-1 and R244S showed nearly identical mass adducts, arguing against a difference in the inactivation mechanism. Testing a wide range of substrates with C3-4 carboxylates in different stereochemical orientations, we observed impaired affinity for all substrates among inhibitor resistant variants. Lastly, we synthesized two boronic acid transition state analogs that mimic cephalothin and found substitutions at Arg-244 markedly affect both the affinity and kinetics of binding to the chiral, deacylation transition state inhibitor. These data define a role for Arg-244 in substrate and inhibitor binding in the SHV β-lactamase.

that Arg-244 coordinates a water molecule with the backbone carbonyl of Val-216 (10,15,16). This water molecule is postulated to play a key role in inhibitor affinity and provide the proton source essential for terminal inactivation of the enzyme (Fig. 2) (16). In addition, the guanidinium group of Arg-244 contributes directly to substrate affinity through hydrogen bonding with the conserved ␤-lactam carboxylate (10) (Figs. 1 and 2). Inhibitor-resistant TEM ␤-lactamases with substitutions at Arg-244 exhibit reduced rates of inactivation (k inact ) as well as diminished affinity for clavulanate and substrates (14,16).
We hypothesized that the absence of mutations identified in SHV at this position, despite heavy clinical drug pressure, is an indication of important differences in active site chemistry between the two highly similar enzymes. To test this hypothesis, we investigated the effects of multiple amino acid substitutions by site-saturation mutagenesis and characterized select SHV ␤-lactamases by steady-state kinetics. To compare the nature of the intermediates in the inactivation process, we incubated SHV-1 and SHV R244S, a clavulanate-resistant variant, with clavulanate and resolved the adducts by electrospray ionization mass spectrometry. In addition, we studied the inhibition of several SHV ␤-lactamase variants at position 244 using two boronic acid transition state analogs that are chemically related to the antibiotic cephalothin (Fig. 1). Our results reveal mechanistic differences in clavulanate inactivation between SHV-1 and TEM-1 and suggest a new role for Arg-244 in SHV.

EXPERIMENTAL PROCEDURES
Plasmids and Mutagenesis-bla SHV-1 , subcloned into phagemid vector pBC SK(Ϫ) (Stratagene, La Jolla, CA) and maintained in ElectroMAX TM DH10B TM T1 R cells (Invitrogen) as previously described (17) served as the template for all mutagenesis described herein. Site-saturation mutagenesis was performed on the bla SHV-1 construct using the QuikChange II site-directed mutagenesis kit (Stratagene) and primers degenerate at Ambler position 244. Two microliters of the mutagenesis reaction was transformed into Escherichia coli DH10B electrocompetent cells (Invitrogen) and plated onto Luria-Bertani (LB) agar plates using chloramphenicol (Sigma-Aldrich, 20 g/ ml) for selection. One hundred colonies were screened for mutations at amino acid position 244 by DNA sequencing with an ALF Express TM automated DNA sequencer (GE, Piscataway, NJ). The Thermo Sequenase TM fluorescence-labeled primer cycle sequencing kit was used according to the protocol provided. Special attention was given to selecting mutants with the most common codon usage in SHV-1.
Because the R244K, -Met, and -Phe mutants were not obtained in the initial screen, bla SHV-1(R244K) , bla SHV-1(R244M) , and bla  were constructed by site-directed mutagenesis, using specific mutagenic oligonucleotides as previously described (18). DNA sequencing confirmed the presence of the mutated codons as described above.
Immunoblots-E. coli DH10B cells were grown to A 600 ϭ 0.8 and frozen. Twenty microliters of each culture was lysed by 10-min incubation at 100°C in SDS loading dye buffer. Immunoblots were performed with an anti-SHV polyclonal antibody to confirm expression of full-length protein from all 20 constructs as previously described (18).
Antibiotic Susceptibility-Minimal inhibitory concentrations (MICs) were determined by the agar dilu- tion method, using a Steer's Replicator that delivers 10 4 colony forming units per 10-l spot. Antibiotics tested include ampicillin, cephalothin, and piperacillin (all from Sigma-Aldrich) and lithium clavulanate (GlaxoSmithKline, Surrey, United Kingdom). Clavulanate susceptibility was determined in the presence of 50 g/ml ampicillin. MIC values reported are the most frequent number observed (mode) in at least three separate experiments.
Protein Expression and Purification-E. coli DH10B cells containing the bla SHV-1 and bla SHV R244S , R244Q , R244L , and R244E genes in pBC SK(Ϫ) were grown overnight in SOB medium (per liter: 20 g tryptone, 5 g yeast extract, 0.58 g sodium chloride, 0.19 g potassium chloride (Fisher Scientific)) harvested by centrifugation at 4°C, and frozen. ␤-Lactamase was liberated using stringent periplasmic fractionation with 40 g/ml lysozyme and 1 mM EDTA, pH 7.8. Preparative isoelectric focusing was performed with the lysate in a Sephadex granulated gel using ampholines in the pH range of 3.5-10 (Amersham Biosciences) as previously described (19). The protein was eluted and dialyzed with 20 mM diethanolamine buffer, pH 8.5. Protein concentration was assessed by Bio-Rad protein assay with bovine serum albumin standards. Purity of Ͼ95% was determined using 10% SDS-PAGE.
The dissociation constants of the pre-acylation complex (K d ) were determined by direct competition with the indicator substrate nitrocefin for the ␤-lactams (cephalothin, piperacillin, and meropenem). The K i values for the inhibitors (clavulanic acid and the boronic acid inhibitors, compounds 1 and 2) were also determined by direct competition, because k inact /K i was Ͻ Ͻ1.
Enzyme concentrations for these experiments were adjusted so that an initial rate of nitrocefin hydrolysis between 1 and 1.5 M/s in the absence of inhibitor was achieved. Final concentrations were 7 nM (SHV-1), 10.5 nM (R244Q), 21 nM (R244S and R244L), and 105 nM (R244E). Nitrocefin concentrations equal to the K m ( Table 2) were used for all experiments, except for R244L in which 200 M was used due to absorbance limitations. The data were analyzed according to Equation 2 to determine K i , where i ϭ fraction of enzyme inhibited, [S] ϭ nitrocefin concentration, [I] ϭ inhibitor concentration, and K m refers to the K m value of the enzyme for nitrocefin (20).
The first-order rate constant for enzyme and inhibitor complex inactivation, k inact , was measured directly by monitoring the reaction time courses in the presence of clavulanate. A fixed concentration of enzyme and nitrocefin, and increasing concentrations of clavulanate ([I]), were used in each assay. The k obs for inactivation was determined graphically by Equation 3, where A ϭ absorbance, t ϭ time (seconds), v f ϭ final reaction velocity, v 0 ϭ the initial reaction velocity in the first 5 s, and A 0 ϭ the initial absorbance. Each k obs was plotted versus [I] and fit to Equation 4 to determine k inact .
The partitioning of the initial enzyme inhibitor complex between hydrolysis and enzyme inactivation (k cat /k inact ) was obtained in the following manner. First, we incubated increasing amounts of clavulanate with a fixed concentration of ␤-lactamase in a total volume of 40 l of 20 mM phosphate-buffered saline, pH 7.4, at room temperature. After 24 h, the sample was added to a 1-ml cuvette containing phosphate-buffered saline and nitrocefin (see above), and initial rates of hydrolysis were assessed. The proportion of clavulanic acid relative to enzyme that resulted in Ն90% inactivation after 24 h was the k cat /k inact for the enzyme.
K i values for the boronic acid compounds were determined as described above. However, because of time-dependant inactivation, enzyme and the boronic acid compound 2 were preincubated for 5 min in phosphate-buffered saline before initiating the reaction with the addition of substrate (10,21,22).
Timed inactivation experiments were also performed with clavulanate and the boronic acid compound 2. To control for the difference in affinities, K i concentrations of inhibitor were used in each assay. Enzyme and inhibitor were incubated in phosphate-buffered saline for a range of time points before the reaction was initiated by nitrocefin (see above for enzyme and nitrocefin concentrations). Mass Spectrometry-For intact protein mass spectrometry, 40 M of SHV-1 or R244S were incubated for 15 min with and without the addition of 40 mM lithium clavulanate. Each reaction was terminated by the addition of 1:10 volume of 1% trifluoroacetic acid and immediately desalted and concentrated using a C 4 ZipTip (Millipore, Bedford, MA) according to the manufacturer's protocol. Samples were then placed on ice and analyzed within 2 h.
Spectra of the intact SHV-1 and SHV R244S proteins were generated on an Applied Biosystems (Framingham, MA) Q-STAR XL quadrupole-time-of-flight mass spectrometer equipped with a nanospray source. Experiments were performed by diluting the protein sample with acetonitrile/1% formic acid to a concentration of 10 M. This protein solution was then infused at a rate of 0.5 l/min, and data were collected for 2 min. Spectra were deconvoluted using the Applied Biosystems (Framingham, MA) Analyst program.
Synthesis of Boronic Acid Inhibitors-1 H and 13 C NMR spectra were recorded on a Bruker DPX-200 or Avance 400 spectrometer; chemical shifts (␦) are reported in parts per million (ppm) downfield from tetramethyl silane (TMS) as internal standard (s, singlet; d, doublet; t, triplet; q, quartet; m, multiplet; br, broad signal); coupling constants (J) are given in hertz. Mass fragmentations were determined on a Finnigan MAT SSQ A mass spectrometer (electron impact (EI), 70 eV). Optical rotations were recorded at 20°C on a PerkinElmer Life Sciences polarimeter, and specific rotations are given in 10 Ϫ1 deg cm 2 g Ϫ1 .
All reactions requiring anhydrous conditions were performed under argon using oven-dried glassware. Tetrahydrofuran (THF) was dried according to classic procedures and distilled from sodium/benzophenone before use.

RESULTS
Mutagenesis and Immunoblotting-16 of 19 amino acid substitutions were obtained in the initial sequencing screen. bla SHV-1(R244K) , bla SHV-1(R244M) , and bla SHV-1(R244F) were constructed by site-directed mutagenesis. Full-length expression of all variants was confirmed by immunoblotting (data not shown).
Antibiotic Susceptibility-To test the effects of the single amino acid substitutions at Arg-244 on in vivo ␤-lactam susceptibility, MICs against all 19 variants were determined for ampicillin, ampicillin/clavulanate, piperacillin, and cephalothin in E. coli DH10B cells (Table 1). Twelve amino acid substitutions in SHV-1 increased the MIC values for the inhibitor combination ampicillin/clavulanate by at least two dilutions (from 50/2 to Ͼ50/8 g/ml). MIC values for the penicillins (ampicillin and piperacillin) and the cephalosporin, cephalothin, were universally decreased. Kinetic Behavior of Inhibitor Resistant ␤-Lactamases with Clavulanate-Susceptibility data served as a guide for selection of ␤-lactamases for further kinetic analysis. Four clavulanateresistant enzymes were chosen: SHV R244S, R244Q, R244L, and R244E.
A common feature of the resistant enzymes is reduced affinity for clavulanate, with K i values increasing 60-to 1000-fold (Table 2). Unexpectedly, the clavulanate-resistant enzymes exhibited increased k inact values. This is in sharp contrast to the TEM enzymes, R244Q and R244T, which exhibited a 100-fold decrease in k inact (14). At concentrations equal to the K i for clavulanate, all SHV Arg-244 variants were rapidly inactivated (Fig. 4). This is again much different from the analysis of R244S, R244Q, and R244T TEM variants at Arg-244, which never achieve complete inactivation (14,16).
Despite the increased rates of inactivation, the markedly decreased affinity of these mutant enzymes for clavulanate reduced the inactivation efficiency (k inact /K i ) by 25-to 500-fold (Table 2). Thus, decreased affinity is the primary cause for resistance to inactivation by clavulanic acid.
In 24-h inactivation experiments, SHV-1, R244S, and R244E enzymes required approximately the same amount of clavulanate to achieve a 90% reduction in enzyme activity (k cat /k inact , or partition ratio) ( Table 2). Alternatively, the R244Q and R244L enzymes exhibited a 3-to 9-fold increase in k cat /k inact , respectively. The derived k cat values were increased for all mutants tested, most notably the R244Q and R244L mutants, and this enhanced ability to catalyze the turnover of inhibitor likely contributed to resistance.
Determining the Nature of the Intermediates: Electrospray Ionization MS with SHV-1 and R244S-Clavulanate undergoes a multistep reaction pathway (2). Candidate intermediates were observed previously for SHV-1 and S130G, also an inhibitorresistant variant (Fig. 5) (2). Intact mass spectrometry was performed on SHV-1 and R244S to observe the covalent intermediates in the inactivation pathway. The R244S variant was chosen, because it is the most common mutant found in TEM at position 244.
As seen in Fig. 5, when SHV-1 and R244S were incubated with clavulanate, nearly identical covalent intermediates were formed. This includes, within experimental error, the ⌬ ϩ52 adduct that is postulated to represent the terminally inactivated, cross-linked enzyme species, and the ⌬ ϩ70, ⌬ ϩ88, and ⌬ ϩ155 Da adducts previously observed in TEM-1 and SHV-1 (2,27). The only differences between the two spectra include a ⌬ ϩ175 adduct seen in only SHV-1 and a ⌬ ϩ194 peak visualized in R244S only. The ⌬ ϩ175 peak is very minor and may be hidden within the shoulder of the ⌬ ϩ157 adduct of R244S. Alternatively, the ⌬ ϩ194 peak in R244S could be a protein modification (oxidation), which shifted the ⌬ ϩ175 adduct by ϩ19.
Kinetic Behavior of Inhibitor Resistant Enzymes with ␤-Lactam Substrates-Kinetic parameters for SHV-1 and the inhibitor-resistant mutants R244S, R244Q, R244E, and R244L for ampicillin and nitrocefin are reported in Table 2. Affinity is reduced for all variants, with R244S and R244Q demonstrating the lowest K m values. Interestingly, k cat is universally reduced; the enzymes with the highest k cat /k inact values for clavulanate (R244Q and R244L) display the highest k cat values for substrates among the Arg-244 variants. Catalytic efficiency (k cat /K m ) of all variants was greatly reduced for ampicillin (14-to 700-fold) and nitrocefin (5-to 175-fold).
To assess the importance of the position of the C 3-4 carboxylate in substrate affinity, we next tested piperacillin, cephalothin, and meropenem (Fig. 1). Because accurate hydrolysis of these substrates was difficult to measure for the Arg-244 vari-  ants, affinity was assessed by competition reaction with the indicator substrate nitrocefin (Table 3). Again, all substitutions at position 244 resulted in enzymes with reduced affinities for these substrates (2-to 180-fold increase in K d for piperacillin, 20-to 750-fold for cephalothin, and a striking 400-to Ͼ2700fold for meropenem). As seen above, R244S and R244Q retained the highest affinities.
Probing the Active Site: Cephalothin Boronic Acid Transition State Analogs-Boronic acid analogs have been developed in recent years, both as high affinity inhibitors of ␤-lactamases and as probes to study reaction mechanism. They also serve to explore determinants of binding specificity (10,11,21,22,28). A majority of these compounds have been designed as achiral molecules containing the R1 side chains of penicillins and cephalosporins. Recently, chiral compounds have been developed to take advantage of affinity gains of the C 4 carboxylate (Fig. 1).
To investigate the importance of Arg-244 in SHV for coordination of the ␤-lactam carboxylate, we synthesized an achiral (compound 1) and chiral (compound 2) boronic acid derivative of cephalothin (Fig. 1). In particular, we reasoned that the molecular architecture of 2 would closely resemble the interactions displayed by the antibiotic cephalothin (the carboxylic moiety at C 4 included) with the ␤-lactamase.
As expected, compound 1 inhibits SHV-1 and R244S, R244Q, and R244L with similar affinities (Table 3). In contrast, the R244E enzyme demonstrated an ϳ4-fold reduction in affinity compared with wild type. Because this analog should not interact directly with residue 244, we submit that this difference suggests a rearrangement in the tertiary structure of this variant.
Testing the chiral boronic acid transition state inhibitor, compound 2, there was a 7-fold enhanced affinity for SHV-1 compared with compound 1. R244S and R244Q both showed only modest improvements in K i of binding to the chiral compound (Ͻ2-fold), and 244Leu and R244E had reduced affinity for 2, indicating an unfavorable interaction of the C 4 carboxylate.
As has been described for other ␤-lactamase families (10, 21), we observed time-dependent inhibition of compound 2, which is in contrast to the fast on/fast off time-independent inhibition of other (including 1) boronic acid compounds. To compare our analysis to others, we chose a 5-min preincubation for our studies (10,11,21). To ensure the suitability of the 5-min preincubation for reaching equilibrium, we used the K i concentrations measured at 5 min to further study the time course of inhibition from 0 to 3600 s (60 min) (Fig. 6). Interestingly, although all the mutant enzymes did reach steady state by 5 min, SHV-1 was increasingly inhibited with time, resulting in near complete inhibition by the 3600-s time point. Therefore, the true steady-state K i value for SHV-1 with the chiral inhibitor is likely much lower than the 5-min measurement of 3.8 M.

DISCUSSION
We show that many substitutions at Ambler position 244 in SHV produce the inhibitor-resistant phenotype. However, kinetic analysis of selected variants suggests that the demonstration of resistance to clavulanate in SHV is unique. The principal characteristic of resistance in SHV variants at Arg-244 is a reduction in affinity for the inhibitor. This property is shared by both TEM and SHV. However, unlike TEM, SHV mutants at position 244 do not demonstrate a reduction in k inact . These inhibitor-resistant ␤-lactamases undergo rapid inactivation with clavulanate provided adequate inhibitor concentrations are achieved. Interestingly, both the inhibitor-susceptible and   inhibitor-resistant SHV enzymes follow the same reaction pathway; the products of inactivation of SHV-1 and R244S by clavulanate, as determined by mass spectrometry, are identical.
What accounts for this key difference in behavior between the TEM and SHV ␤-lactamases? We answer this question by examining the atomic structures of TEM-1, SHV-1, and the  SHV S130G variant. Examining a representation created from the PDB coordinates available for the TEM-1 apo-enzyme, a water molecule is clearly positioned between Arg-244 and the backbone carbonyl of Val-216 (Figs. 2 and 7) (10). One explanation for the catalytic deficit in TEM mutants at position 244 is displacement of that water molecule, which is essential for secondary clavulanate ring opening in the inactivation pathway. Comparing the apo-enzyme crystal structures of TEM-1 (26) and SHV-1 (25), the distances between the closest guanidinium nitrogen of Arg-244 and the backbone carbonyl oxygen Val-216 are significantly different (Fig. 7). For TEM, this distance is 5.19 Å, whereas in SHV the distance is 7.85 Å, a large distance for coordination of a water molecule. Therefore it is likely that, in SHV, protonation of the inhibitor is achieved by either a coordinated water molecule elsewhere in the active site, or from bulk water in the medium. Thus, substitutions at position 244 in SHV affect affinity but do not retard inhibitor turnover. This explanation is reminiscent of the observations made in the determination of the S130G apo-enzyme structure. In S130G the catalytic water molecule is only evident as the inhibitor is bound (29). Our data support the general hypothesis that the active site of inhibitor-resistant SHV enzymes is rehydrated. This raises the possibility that, despite different mutations, inhibitor-resistant SHV enzymes follow a common pathway to inactivation.
Studying the dissociation constants of penicillins, cephalosporins, and carbapenems provided us deeper insight into the reliance of Arg-244 binding to the different spatial positions of the C 3-4 carboxylate and its role in catalysis (Fig. 8).
First, affinity for the penicillins is the least affected by substitutions at Arg-244. The carboxylate in this case is hanging from the sp 3 -hybridized and S-configured C 3 carbon. In our model, it is likely that Arg-244 contributes one or two weak hydrogen bonds to the carboxylate in the Michaelis complex, with additional hydrogen bonding interactions coming from other residues in the binding pocket (Fig. 8B). For this reason, most inhibitor-resistant variants at Arg-244 in SHV retain weak penicillinase activity, and their emergence in the context of current ␤-lactam/␤-lactamase inhibitor combinations, which utilize penicillins, is not excluded.
Second, the reductions in affinity are at the very least 20-fold (SHV R244Q) for cephalothin. Different from penicillins, the cephalosporin carboxylate is directly linked to the C 4 sp 2 -hybridized carbon of the six-membered ring and, therefore, coplanar with C 3 -C 4 , with reduced rotational freedom due to conjugation. This likely brings the carboxylate in closer approximation to Arg-244 (Fig. 8C). The dramatic loss of cephalosporinase activity among Arg-244 variants shows that clavulanateresistant variants that arise in the clinic would be susceptible to treatment with cephalosporins. That being said, excessive use of cephalosporins in the clinical setting may mask the emergence of inhibitor-resistant SHV enzymes.
Most revealing is our data for meropenem affinity. Like clavulanate, carbapenems act as very effective covalent inhibitors of class A ␤-lactamases. 8 Interestingly, replacement of Arg-244 results in total loss of affinity for meropenem. All of the mutants bound meropenem with millimolar affinity, and two of them (R244E and R244L) had K d values Ͼ100 mM. This indicates that the carboxylate linked to the C 3 sp 2 -hybridized carbon of the five-membered ring makes several crucial hydrogen bonding interactions with the guanidinium group of Arg-244 (Fig. 8D). Caution should be taken in the development of future inhibitors with planar C 3 carboxylates, because substitutions at residue 244 could seriously compromise their activity.
Last, we studied two boronic acid transition state analogs to further probe the contributions of Arg-244 variants to ␤-lactam carboxylate affinity. The achiral cephalothin boronic acid compound 1 contains just the R1 side chain of cephalothin. In contrast, the chiral cephalothin analog 2 more closely imitates the interactions of the natural substrate by the addition of a C 4 carboxylate on a phenyl ring (Fig. 1). Comparing affinity of both compounds allowed us to determine the extent to which each substitution of the ␤-lactamase affects binding to the carboxylate independently.
Against SHV-1, compound 2 demonstrates significantly greater affinity for the active site (nearly a log fold) compared with compound 1. We observed that the R244S substitution at 244 has a modest, 40% increase in binding to 2 versus 1 (24 M versus 34 M). This supports our claim (see above) that there is another residue in the binding pocket that contributes to binding of the carboxylate, possibly Thr-235 (16). Alternatively, hydrophobic interactions with the phenyl ring of compound 2 could stabilize the inhibitor.
Compared with R244S, a slightly higher affinity is seen with R244Q (60% increase) for compound 2, which may imply that the Gln residue itself is able to weakly interact with the inhibitor carboxylate. On the other hand, increased K i values for 2 with R244L and R244E suggest unfavorable interactions with the inhibitor carboxylate and explain the drastic affinity reductions for ␤-lactams.
The differences in the time-dependent inactivation of compound 2 between the wild-type and mutant enzymes also provided a window into the importance of Arg-244 for turnover of substrate. The chiral compound 2 has been shown in crystal structures with TEM, CTX-M, and AmpC-type ␤-lactamases to assume the deacylation transition state (10,21,22). In TEM, although no structural rearrangements were seen upon binding to the inhibitor, there was movement of catalytic water molecules and a rearrangement of the inhibitor (10). If this reorganization is responsible for the time-dependent binding of compound 2 to SHV-1, the reduction in time dependence among the Arg-244 mutants may indicate a novel substrate deacylation role for Arg-244 (Fig. 6). Ongoing studies to determine the microscopic rate constants k 2 and k 3 of SHV variants at Arg-244 using stopped-flow kinetics will provide evidence for this role. Co-crystallization of SHV-1 and Arg-244 variants with the cephalothin transition state analogs are in progress.
In conclusion, we present evidence that TEM and SHV are different in their intrinsic binding and turnover of inhibitors (12). Despite similarities in sequence and phenotype between TEM and SHV, the kinetic differences revealed herein suggest that these two class A ␤-lactamases follow unique pathways in response to antibiotic pressure (an argument that would have significant implications for the comparative study of penicillininactivating enzymes among prokaryotes). Awareness of the subtle yet mechanistically important differences in inactivation chemistry among class A ␤-lactamases could prove crucial in the future development of ␤-lactamase inhibitors.