Inhibition of OXA-1 β-Lactamase by Penems

Abstract

The partnering of a β-lactam with a β-lactamase inhibitor is a highly effective strategy that can be used to combat bacterial resistance to β-lactam antibiotics mediated by serine β-lactamases (EC 3.2.5.6). To this end, we tested two novel penem inhibitors against OXA-1, a class D β-lactamase that is resistant to inactivation by tazobactam. The K_i of each penem inhibitor for OXA-1 was in the nM range (K_i of penem 1, 45 ± 8 nM; K_i of penem 2, 12 ± 2 nM). The first-order rate constant for enzyme and inhibitor complex inactivation of penems 1 and 2 for OXA-1 β-lactamase were 0.13 ± 0.01 s⁻¹ and 0.11 ± 0.01 s⁻¹, respectively. By using an inhibitor-to-enzyme ratio of 1:1, 100% inactivation was achieved in ≤900 s and the recovery of OXA-1 β-lactamase activity was not detected at 24 h. Covalent adducts of penems 1 and 2 (changes in molecular masses, +306 ± 3 and +321 ± 3 Da, respectively) were identified by electrospray ionization mass spectrometry (ESI-MS). After tryptic digestion of OXA-1 inactivated by penems 1 and 2, ESI-MS and matrix-assisted laser desorption ionization-time-of-flight MS identified the adducts of 306 ± 3 and 321 ± 3 Da attached to the peptide containing the active-site Ser67. The base hydrolysis of penem 2, monitored by serial ¹H nuclear magnetic resonance analysis, suggested that penem 2 formed a linear imine species that underwent 7-endo-trig cyclization to ultimately form a cyclic enamine, the 1,4-thiazepine derivative. Susceptibility testing demonstrated that the penem inhibitors at 4 mg/liter effectively restored susceptibility to piperacillin. Penem β-lactamase inhibitors which demonstrate high affinities and which form long-lived acyl intermediates may prove to be extremely useful against the broad range of inhibitor-resistant serine β-lactamases present in gram-negative bacteria.

Ambler class D OXA β-lactamases (EC 3.2.5.6) are among the most rapidly growing and diverse group of β-lactam-inactivating enzymes present in gram-negative bacteria (4, 10, 20, 45, 57). Found primarily in multidrug-resistant Acinetobacter baumannii and Pseudomonas aeruginosa strains, the OXA β-lactamases are able to confer resistance to extended-spectrum cephalosporins (e.g., OXA-10, OXA-14, and OXA-17) as well as carbapenems (e.g., OXA-23, OXA-40, and OXA-58) (10, 16, 57). In contrast to class A TEM-type and SHV-type β-lactamases, the OXA enzymes are also relatively resistant to inactivation by the medically important β-lactam-β-lactamase inhibitor combinations (8, 12). When they are found in clinical isolates, the OXA β-lactamases confer resistance to ampicillin-sulbactam (Unasyn), amoxicillin-clavulanate (Augmentin), and piperacillin-tazobactam (Zosyn) (2, 24, 50). Additionally, many of the bla genes encoding the OXA β-lactamases (bla_OXAs) are carried on plasmids and integrons, making their potential for widespread dissemination a significant clinical concern (3, 14, 17, 37, 38, 44, 45). Thus, the need to develop inhibitors with activity against the OXA β-lactamases is an urgent priority.

To date, acyl phosphates, phosphonates, and penicillinates are the only compounds reported to act as class D inhibitors (1, 31, 46). By acylating the active serine residue, these compounds serve as inhibitors of the OXA enzymes. Unfortunately, the poor stability of phosphonates in aqueous media and the susceptibility of phosphonates to phosophodiesterases limit their clinical use. Penicillinates form acyl enzymes with the OXA-10 β-lactamase, but their affinity is similar to the affinities of clavulanate and the sulfones (K_i range, 100 to 300 μM) (34).

Methylidene penems are novel and highly potent mechanism-based inhibitors of serine-reactive class A and C β-lactamases (41, 55, 56). These compounds contain mono-, bi-, or tricyclic heterocycles that adopt the Z configuration at the C-6 position, which enhances their affinity for the TEM-1 (class A) and AmpC (class C) β-lactamases (41). Since effective inhibitors of class D β-lactamases are still being sought, we reasoned that the methylidene penems would be attractive candidates as inactivators of the OXA β-lactamases. To this end, novel penem inhibitors 1 and 2 (Fig. 1) were synthesized and assayed for their activities against the OXA-1 β-lactamase. OXA-1 was chosen because it is a highly active class D penicillinase and a monomeric enzyme whose atomic structure is known (52). The compounds selected for this study, penems 1 and 2, differ by having a dihydropyrazolo thiazole and a dihydropyrazolo thiazine moiety, respectively. We show that these novel penem inhibitors inactivate the OXA-1 β-lactamase by forming a long-lived acyl enzyme. The antimicrobial activities of penems 1 and 2 combined with piperacillin reveals their potential to protect β-lactam antibiotics from hydrolysis by the OXA-1 β-lactamase.

FIG. 1.

Chemical structures of penem inhibitors (penem 1 and penem 2) and tazobactam (compound 3). Compound 4 is BRL 42715.

MATERIALS AND METHODS

Genetic constructs and host strains.

The bla_OXA-1 gene was cloned from plasmid R_GN238 into pET 12a(+)-KM (kanamycin resistant), as described previously (52). Plasmid R_GN238 bla_OXA-1 was maintained in Escherichia coli DH10B cells (Invitrogen, Carlsbad, CA). This host strain was used for MIC determinations (see below). For protein purification, bla_OXA-1 cloned in the modified vector pET 12a(+)-KM was expressed in Escherichia coli BL21(DE3) cells (Stratagene, La Jolla, CA).

β-Lactamase purification.

The OXA-1 β-lactamase was prepared from E. coli BL21(DE3) cells after induction with isopropyl-β-d-thiogalactopyranoside (IPTG; Sigma Chemical Company, St. Louis, MO). Five hundred-milliliter cultures were induced at an optical density at 600 nm of 0.5 to 0.8 (final IPTG concentration, 0.2 mM). The culture was harvested 3 h after induction and centrifuged, and the pellets were frozen at −20°C and on the next day were resuspended in 50 mM Tris buffer, pH 7.4. β-Lactamase was released from the periplasmic space by using stringent periplasmic fractionation, as described previously (7). This was done with lysozyme and EDTA (7, 22) (all manipulations for the purification of the OXA-1 β-lactamases were done in 50 mM sodium phosphate [monobasic and dibasic] buffer, pH 7.2).

The OXA-1 β-lactamase was initially purified by preparative isoelectric focusing (22). In certain instances, additional purification was performed by high-pressure liquid chromatography with a Sephadex Hi Load 26/60 column (Pharmacia, Uppsala, Sweden). Proteins were eluted with 50 mM sodium phosphate buffer, pH 7.2 (21). The purity of each preparation was assessed by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis. Protein samples were resolved on a 5% stacking, 12% separating SDS-polyacrylamide gel and stained with Coomassie brilliant blue R250 (Fisher, Pittsburgh, PA). Protein concentrations were determined by the Bio-Rad protein assay with bovine serum albumin (Sigma) as the standard.

Kinetic parameters.

Steady-state kinetic parameters were determined with an Agilent 8453 diode array spectrophotometer (19). The kinetic determinations were performed at room temperature (23°C) in 50 mM sodium phosphate supplemented with a saturating concentration of sodium bicarbonate (20 mM) (18). We established that 20 mM bicarbonate is optimal for saturation of the OXA-1 β-lactamase (27).

First, the kinetic parameters V_max and K_m were obtained from the nonlinear least-squares fit of the data (Henri-Michaelis equation) by using the Enzfitter program (Biosoft Corporation, Ferguson, MO). The K_is of the inhibitors were determined as follows (13). In these measurements we used a final concentration of 75 μM nitrocefin (NCF; Becton Dickinson, Cockeysville, MD) as the indicator substrate (change in ɛ₄₈₂ = 17,400 M⁻¹ cm⁻¹). 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 tazobactam and penems 1 and 2. A fixed concentration of enzyme, NCF, and increasing concentrations of tazobactam and penems 1 and 2 (inactivators) were used in each assay. The observed value of k (k_obs) was determined by using a nonlinear least-squares fit of the data to equation 1 by using the Origin (version 7.5) program:

(1)

where A is the absorbance, v₀ (expressed as the variation in the absorbance per unit of time) is the initial velocity, v_f is the final velocity, and t is time. Each k_obs value was plotted against the inhibitor concentration and fit to determine k_inact. In these experiments, the enzyme concentration was 18 nM, the penem inhibitor concentrations ranged from 50 to 1,100 nM, and the tazobactam concentrations ranged from 50 to 3,000 μM. The value of k_inact was used to determine K_i, according to the following equation:

(2)

The turnover number (t_n) and the partitioning of the initial enzyme-inhibitor complex between hydrolysis and enzyme inactivation, k_cat/k_inact (partition ratio), was determined as reported previously (19, 53).

The data were corrected according to the following equation to account for the affinity of NCF for the OXA-1 β-lactamase:

(3)

Chemical syntheses.

Penems 1 and 2 (Fig. 1) were prepared by the method previously described in the literature (56).

¹H NMR.

The proton nuclear magnetic resonance (¹H NMR) spectra of penems 1 and 2 were determined with a Varian 600-MHz NMR instrument. Chemical shifts are reported in parts per million relative to that of residual chloroform (7.26 ppm), tetramethylsilane (0 ppm), or dimethyl sulfoxide (2.49 ppm) as an internal reference, with coupling constants (J) reported in hertz. The peak shapes are denoted as follows: s, singlet; d, doublet; t, triplet; q, quartet; m, multiplet; br, broad. The ¹H NMR spectra of penems 1 and 2 were determined by using the ¹H NMR instrument available at the core facility at Case Western Reserve University. Buffers were prepared in H₂O and were replaced with an equivalent volume of D₂O after dehydration in a Speed Vac apparatus.

ESI-MS.

Electrospray ionization mass spectrometry (ESI-MS) of the intact OXA-1 β-lactamase was performed on an Applied Biosystems (Framingham, MA) Q-Star XL quadrupole time-of-flight mass spectrometer equipped with a nanospray source, as described previously (53). Experiments were performed by diluting the protein sample with acetonitrile-1% formic acid to a concentration of 10 μM. This protein solution was then infused at a rate of 0.5 μl/min, and data were collected for 2 min. Spectra were deconvoluted by using the Applied Biosystems Analyst program.

Tryptic digestion.

Proteolytic digestions were performed by adding trypsin to a solution of OXA-1 β-lactamase at a ratio of 1:25 (trypsin weight to protein weight). Tryptic digestions were carried out for 3 h; terminated by the addition of a 1:10 volume of 1% trifluoroacetic acid; immediately desalted; concentrated by using a C₁₈ ZipTip apparatus (Millipore, Bedford, MA), according to the manufacturer's protocol; and analyzed by ESI-MS on a Q-Star XL mass spectrometer.

Molecular representations.

We used the crystal structure coordinates of the OXA-1 β-lactamase (1M6K; www.rcsb.org) to generate a representation of the OXA-1 β-lactamase as a Henri-Michaelis complex with penem 2 and as an acyl enzyme inactivated by penem 2. To simplify the analysis, water molecules were retained in the initial energy minimization, the hydrogen atoms were added (pH 7.5), and the structure was minimized at a constant dielectric of 1 with the conjugate gradient method and a constant valence force field (CVFF; the Discover_3 module of the Insight II program was used). The model of penem 2 and the seven-member 1,4-thiazepine derivative ring were constructed by using the Builder module of the Insight II program. The structure was optimized by using 1,000 iterations, 0.01 derivatives, and energy minimized to final minimum energies of 133 and 30.81 kcal/mol for penem 2 and the seven-member 1,4-thiazepine derivative ring, respectively. By using the Affinity and the Docking modules, penem 2 and the seven-member thiazepine were docked into the binding site of OXA-1. The water molecules were tethered, and the energy of the enzyme-inhibitor complex was minimized. The model was also examined by using the Affinity module (automatic docking). Twenty conformers (Monte Carlo method) were further studied by using the Simulated_Annealing (SA) module. The SA module refined the 20 structures and lowered the interaction energies. The interaction energy between OXA-1 and penem 2 was calculated to be −40.9 kcal. The trajectory was loaded, and the conformation with the minimum energy was selected (Van der Waals energy, −34.37 kcal/mol; electrostatic energy, −73.87 kcal/mol; and total energy, −108.252 kcal/mol).

Theoretical calculations.

To better understand the spectra obtained from base hydrolysis, ab initio quantum mechanical calculations were performed to predict the ¹H NMR spectral shifts of intermediate compounds by using the Gaussian '03 program. Calculations were performed at the Hartree-Fock level as described previously (54).

Antibiotic susceptibility (MICs).

The MICs of E. coli DH10B cells harboring plasmid R_GN238 bla_OXA-1 were determined in Luria Bertani agar with a Steers replicator that delivered 10 μl of 10⁴ CFU/spot. The β-lactamase inhibitor tazobactam (Fig. 1, compound 3) was supplied by Wyeth Pharmaceuticals (Pearl River, NY). Piperacillin was obtained from Sigma. The concentrations used to determine the MICs were in μg/ml. Breakpoints for susceptibility and resistance were defined by the Clinical Laboratory Standards Institute (CLSI; formerly National Committee for Clinical Laboratory Standards, NCCLS) and interpreted with criteria published in 2005 (CLSI standard M100-S15) (39, 40). For the testing of piperacillin and tazobactam (or the β-lactamase inhibitors penems 1 and 2), a concentration of 4 μg/ml was chosen. The concentration of piperacillin was increased by doubling dilutions.

RESULTS AND DISCUSSION

Kinetics of inactivation of OXA-1 β-lactamase by penems 1 and 2.

We used NCF as the indicator substrate to characterize the inactivation of the OXA-1 β-lactamase and compared the K_is of penems 1 and 2 to the K_i of tazobactam. The k_inact values of penems 1 and 2 for the OXA-1 β-lactamase were 0.13 ± 0.01 s⁻¹ and 0.11 ± 0.01 s⁻¹, respectively; for tazobactam, k_inact was 0.12 ± 0.01 s⁻¹ (Fig. 2). Thus, the novel penem inhibitors demonstrated very low K_is for OXA-1 (penem 1, 45 ± 8 nM; penem 2, 12 ± 2 nM). In contrast, the K_i of tazobactam for OXA-1 was 1,700 to 6,700 times greater (80 ± 14 μM) (Table 1). As expected, both penem 1 and penem 2 showed competitive inhibition. The inactivation rate constants and the large decrease in the K_i values for penems 1 and 2 resulted in a significant improvement in inactivation efficiency (k_inact/K_i).

FIG. 2.

Determination of k_inact and K_i of tazobactam (a), penem 1 (b), and penem 2 (c) against OXA-1. Activity curves were obtained with an Agilent 8452 spectrophotometer and were analyzed by nonlinear least-squares fit of the data with the Origin (version 7.5) program.

TABLE 1.

Kinetic parameters

Compound	Km (or Ki) μM	kinact (s−1)	kinact/Ki (μM−1 s−1)	kcat/kinact
NCF	8.3 ± 1.0
Tazobactam	80 ± 14	0.12 ± 0.01	15 × 10−4 ± 3 × 10−4	349
Penem 1	0.045 ± 0.008	0.13 ± 0.01	2.9 ± 0.6	0
Penem 2	0.012 ± 0.002	0.11 ± 0.01	9.2 ± 1.7	0

By using an inhibitor-to-enzyme ratio of 1:1 (40 μM OXA-1 and 40 μM penem 1 or 2), 100% inactivation was rapidly achieved (≤900 s) and enzyme recovery was not observed up to 24 h. Hence, the partition ratio, or k_cat/k_inact, was 0 (0/1). This stands as a unique finding, since our studies of SHV-1 and clavulanate, sulbactam, and tazobactam showed partial recovery of β-lactamase activity within 1 h after inactivation (42, 51).

ESI-MS and the nature of the inactivation products.

To establish the nature of the inactivation products, ESI-MS was performed with a Q-Star quadrupole time-of-flight mass spectrometer equipped with a nanospray source. We inactivated 40 μM OXA-1 β-lactamase with 1.2 mM penems 1 and 2 in 20 mM diethanolamine, pH 8.6. The ratio of inhibitor to β-lactamase was 30:1, and the incubation time was 15 min (900 s).

The deconvoluted spectra for apo-OXA-1 and penems 1 and 2 for the inactivation of the OXA-1 β-lactamases are presented in Fig. 3, and our results are summarized in Table 2. The ESI-MS measurements were in agreement with the theoretical mass of the OXA-1 β-lactamase, which is 28,132 Da. The preparative method did not permit us to identify a mass increase consistent with the addition of a CO₂ group (carboxylation at Lys70) to the β-lactamase (18, 34, 35, 52).

FIG. 3.

Deconvoluted mass spectra were obtained on a Q-Star quadrupole time-of-flight mass spectrometer equipped with a nanospray source. (a) Deconvoluted spectrum of the OXA-1 β-lactamase; (b), deconvoluted spectrum of the OXA-1 β-lactamase inactivated with penem 1; (c) deconvoluted spectrum of the OXA-1 β-lactamase inhibited with penem 2. (d) Region of the deconvoluted mass spectrum of the OXA-1 β-lactamase tryptic digest. The sequence of the peptide at 5,038 amu is shown with the disulfide bond highlighted. This peptide spans from the N terminus at Ser18 to Lys56, with the disulfide bond being between Cys37 and Cys59. The active-site serine is located at Ser67 and is designated by an asterisk. (e) Region of the deconvoluted mass spectrum of the tryptic digest of OXA-1 β-lactamase inactivated with penem 1. The peak at 5,344 amu represents the covalent attachment of penem 1. The mass shift is indicated. (f) Region of the deconvoluted mass spectrum of tryptic digest of OXA-1 β-lactamase inactivated with penem 2. The peak at 5,359 amu is due to the covalent attachment of penem 2. The mass shift is indicated.

TABLE 2.

MS analysis

Compound(s)	MS units (amu)
Penem 1 (free acid)	306
OXA-1 β-lactamase	28,128 ± 3
OXA-1 β-lactamase + penem 1	28,434 ± 3
Difference	306
Penem 2 (free acid)	321
OXA-1 β-lactamase	28,127 ± 3
OXA-1 β-lactamase + penem 2	28,448 ± 3
Difference	321

Covalent attachment of penems 1 and 2 to the OXA-1 β-lactamase was demonstrated (m/z of 28,128 ± 3 with a change in molecular mass of +306 Da to 28,434 ± 3 Da for penem 1 and m/z of 28,127 ± 3 with a change in molecular mass of +321 to 28,448 ± 3 Da for penem 2). Over the time period studied, we did not find evidence of the fragmentation of penems 1 and 2, as was seen in the inactivation of TEM-1, CMY-2, SHV-1, the Arg244Ser variant of SHV-1, and the Ser130Gly variant of SHV-1 with tazobactam and clavulanate (7, 43, 51, 53). Interestingly, we observed two species in the deconvoluted spectra of each reaction (Fig. 3b and c). The masses of 28,434 ± 3 and the 28,448 ± 3 Da are assigned to the species representing the OXA-1 β-lactamase inactivated by penems 1 and 2, respectively, while the other species of 28,128 ± 3 and 28,127 ± 3 Da are assigned to the OXA-1 β-lactamase without the inhibitor and without CO₂. Assay of β-lactamase activity after inhibitor inactivation and before MS determination did not indicate that the uninhibited β-lactamase was present at the ratio suggested by the spectra (23, 42).

A theoretical enzymatic digestion of the OXA-1 β-lactamase with trypsin (http://prospector.ucsf.edu/prospector/4.0.7/htmL/msdigest.htm) served as a guide for interpretation of the tryptic digestion of the OXA-1 β-lactamase before and after inactivation with penems 1 and 2. The unmatched fragments were compared to the matched fragments, and a series of candidate peptides was identified.

Using the Q-Star apparatus, we detected a peptide with 5,038 atomic mass units (amu). This peptide was assigned to the amino acid sequence shown in Fig. 3d, with the active-site Ser67 and a disulfide bond shown. The N terminus starts at Ser18, and a disulfide bond is present between Cys37 and Cys59. Compared to the tryptic digest of OXA-1, the deconvoluted mass spectrum of the tryptic digest of the OXA-1 β-lactamase inactivated with penem 1 revealed a new peak at 5,344 amu (Fig. 3e). This represents the covalent attachment of penem 1 to the peptide with a mass of 5,038 amu. Similarly, the deconvoluted mass spectrum of the tryptic digest of the OXA-1 β-lactamase inactivated with penem 2 showed a new peak at 5,359 amu, which is consistent with the attachment of penem 2 to the same peptide with a mass of 5,038 amu (Fig. 3f). These data confirm the covalent attachment of each inhibitor to a peptide fragment containing the active-site Ser67.

Base hydrolysis.

In order to determine the nature of the intermediates that are generated from the hydrolysis of penems 1 and 2, we performed base hydrolysis and serial ¹H NMR spectrum determinations (9). Due to the isotope effect on the ionization of weak acids, the pHs in D₂O (pDs) of the resulting buffers are approximately 0.5 unit higher than the pHs of the corresponding buffers in H₂O (6, 48, 49). The ¹H NMR spectra of penems 1 and 2 and the proton assignments are shown in Fig. S1 and S2 in the supplemental material. The first ¹H NMR spectra was that of 5 mM penem 2 at pD 7.2 in 50 mM sodium phosphate buffer (see Fig. S2a in the supplemental material). The pD was raised to 11 to 12 by using Na₃PO₄ and then serial ¹H NMR spectroscopy was performed (see Fig. S2b to f in the supplemental material). We observed the formation of a new species with a half-life of 30 min (see Fig. S2d in the supplemental material). We assigned the spectrum of the initial hydrolysis product to the compound shown in Fig. S2g in the supplemental material on the basis of the chemical shifts predicted by quantum chemical gauge independent atomic orbital calculations done with the Gaussian '03 program.

Analysis of the reaction of BRL 42715 (Fig. 1, compound 4), a previously studied compound, showed that the Z-penem nitrobenzyl ester reacted with the base (sodium methoxide in methanol) results in the formation of a seven-member thiazepine with λ_max values of 253 and 370 nm (9). It is known from crystal structure, kinetic, and computational studies that a novel seven-member 1,4-thiazepine ring is formed from the reaction of methyldiene penems and BRL 42715 with class A enzymes (e.g., TEM-1 and SHV-1) and class C enzymes (11, 15, 36, 41, 55, 56). The chemical similarity of penems 1 and 2 to these compounds, the ¹H NMR spectra, and the results of MS and peptide analysis argue that a similar intermediate, the 1,4-thiazepine derivative, is formed (9, 56). This assertion serves as the experimental foundation for our representation of penem 2 in the active site of OXA-1.

Molecular representations.

The model presented here builds upon the analysis of BRL 42715 (compound 4) reacting with TEM-1 and the Arg244Ser variant (11). The mechanistic analysis for the TEM-1-BRL 42715 interaction compellingly argues for the formation of the S isomer (11). In addition, recent enzymatic studies performed with a series of (5,5)-fused bicyclic 6-methylidene penem inhibitors also favors the formation of the S enantiomer (32). Our representations of the binding of the inhibitor in the active site of OXA-1 were performed with tethered water molecules. Penem 2 was energy minimized by using CVFF (energy, 133 kcal) and was manually docked in the active site of the OXA-1 enzyme (Fig. 4a). Similarly, the 1,4-thiazepine derivative of penem 2 was also energy minimized by using CVFF and was docked manually in the active site in the S configuration (Fig. 4b).

FIG. 4.

Penem 2 and the 1,4-thiazepine derivative of penem 2 were modeled in the active site of OXA-1 β-lactamase by using the Affinity and Docking modules of the Insight II program. (a) Connolly surface diagram of OXA-1 with penem 2 as a Henri-Michaelis complex; (b) Connolly surface diagram of OXA-1 with the 1,4-thiazepine derivative trapped in the binding pocket; (c) important H bonds from OXA-1 to the 1,4-thiazepine derivative.

By comparing the Henri-Michaelis complex (Fig. 4a) to the acyl enzyme (Fig. 4b), we see that Trp102 is rotated ∼30° from the initial position to establish a coplanar hydrophobic π stacking with the bicyclic ring of the inhibitor. In previous crystallographic and computational analyses performed with class A and class C β-lactamases, hydrophobic π-stacking interactions are prominent features (41, 55, 56). In fact, many OXA β-lactamases (e.g., OXA-10, OXA-14, OXA-17, OXA-23, 24, OXA-27, OXA-40, OXA-51, OXA-58, OXA-59, and OXA-69) have Trp102.

Equally notable, the formation of H bonds between the 1,4-thiazepine derivative of penem 2 (as the S isomer) and key residues (Ser115, Phe114, and Thr213) in the OXA-1 β-lactamase are present (Fig. 4c). It is believed that these H bonds contribute to the stability of the acylated species.

Both molecular models indicate conformational changes in the binding site. The residues with the most notable changes are Ser115 and Lys212. Their side chain movements contribute to an increase in the binding site size by several Å (3 to 4 Å). Unfortunately, preliminary molecular dynamic analyses were not able to determine if displacement of the catalytic water molecule results in an acyl ester bond that might also be resistant to hydrolysis (41).

Susceptibility testing.

β-Lactamase inhibitors protect β-lactam antibiotics from inactivation by bacterial β-lactamases (16). A potent β-lactamase inhibitor must lower the MICs substantially. Many factors determine the clinical efficacy of a β-lactam-β-lactamase inhibitor combination against gram-negative bacterial β-lactamases: serum concentrations and the rates of elimination of the β-lactam and β-lactamase inhibitor, the rates of penetration through bacterial porin channels, efflux, and the amount of β-lactamase produced in the periplasmic space. Other important considerations are the affinity for the β-lactamase, the stability of the intermediates formed in the inactivation process, and the dividing time of the organism (19, 26, 28, 29, 51). The complex interplay of these factors is assessed by in vitro susceptibility testing.

To test the efficacies of penems 1 and 2 against the OXA-1 β-lactamase, we chose as our host organism E. coli DH10B. When clinical isolates of E. coli are tested, the breakpoint for susceptibility to piperacillin-tazobactam is ≤16/4 mg/liter; resistance is defined by an MIC of ≥128/4 mg/liter (39, 40).

We first compared the efficacies of penems 1 and 2 added at 4 mg/liter to increasing concentrations of piperacillin. A parallel experiment was performed with the same concentration of tazobactam (4 mg/liter) added to piperacillin. As seen in Table 3, E. coli DH10B cells harboring plasmid R_GN238 with bla_OXA-1 were highly resistant to piperacillin (resistance was defined by an MIC of ≥32 mg/liter; the MIC of E. coli DH10B with bla_OXA-1 was 1,024 mg/liter). Addition of the penem inhibitors 1 and 2 (both at 4 mg/liter) to increasing concentrations of piperacillin markedly reduced the MICs to the susceptible range (4 mg/liter of piperacillin and 4 mg/liter of penem 1 or 2). In comparison to the effect of piperacillin-tazobactam, this effect was very notable (Table 3).

TABLE 3.

MICs

Organism	MIC (mg/liter)a
	Piperacillin	Piperacillin- tazobactam	Piperacillin- penem 1	Piperacillin- penem 2
E. coli DH10B	2	2	2	2
E. coli DH10B blaOXA-1	1,024	512	4	4

^aTazobactam and penems 1 and 2 were each tested at 4 mg/liter.

Conclusion.

In addition to their established activities against class A (TEM-1, SHV-1, IMI-1) and class C (AmpC) β-lactamases, we show that penems 1 and 2 are also extremely effective inhibitors of the OXA-1 β-lactamase, a class D enzyme. Thus, a general and common mechanism of serine β-lactamase inactivation is now reported for methylidene penems. Our data support the thesis that 6-methylidene penem inhibitors inactivate class A, C, and D serine β-lactamases by a common mechanism that leads to the formation of 1,4-thiazepine intermediates. Specifically, the active-site Ser attacks penems 1 and 2 to form the acyl enzyme (compound 5) (Fig. 5). The acylated species leads to the departure of the thiolate from C-5, opening of the second ring, and formation of the imine (compound 6). Finally, this intermediate undergoes 7-endo-trig cyclization to form the cyclic enamines (compounds 7a and 7b) and the 1,4-thiazepine derivative (compound 7c) (53). These investigations also significantly advance the previous work performed with BRL 42715 and OXA-2 by showing that despite a >50% divergence in amino acid sequence and important differences in the active sites of these two OXA enzymes (Asp102 in OXA-2 versus Trp102 in OXA-1), penems 1 and 2 maintain high affinities and high reactivities (33). On the basis of our consideration of the results of MS analysis (two species were present after 15 min if inactivation) and serial ¹H NMR analyses (the formation of a second species with a half-life of 30 min), we postulate that the highly reactive imine (compound 6) is deacylated upon addition of formic acid or acetonitrile, while the cyclic enamine (compounds 7a to 7c) is the stable product.

(4)

In equation 4, E-I*¹ represents the acyl enzyme as the linear imine and E-I*² represents the long-lived cyclized enamine that we suspect is not hydrolyzed by MS. This would explain why there is a complete inhibition of NCF hydrolysis by kinetic analysis, yet a mixture of two species (the apo and inhibited enzymes) is seen by MS.

FIG. 5.

Mechanism of inactivation of OXA-1 β-lactamases by penem 1.

It is possible that the 1,4-thiazepine derivative is not the sole end point of inactivation, nor is 7-endo-trig cyclization to form the S isomer of the cyclic enamine the only path that can be followed. Conceivably, the early steps in catalysis may favor an alternate rotation around the C-7—C-6 bond to result in an R isomer. It should be mentioned that although the thiazepine (compound 7) may exist in several tautomeric enamine forms, the C-7 carbon remains stereogenic in these tautomers. As was shown with SHV-1 and tazobactam, one may envision a pathway in which Ser115 in OXA-1 (the Ser130 equivalent) may serve as a second nucleophile and attack the acyl enzyme (25). Evidence obtained to date does not yet support this alternative pathway.

An explanation for the difference in susceptibility to inactivation by inhibitors such as tazobactam compared to the susceptibility to inactivation by penems 1 and 2 for OXA-1 remains a subject of intense interest. Our analysis provides important early insights into a possible reason for this difference. In class A β-lactamases, such as SHV-1 and TEM-1, the positively charged guanidinium group of Arg244 makes critical H bonds with the C-3 carboxylate of clavulanate either directly or through a strategically located water molecule (53). This water molecule is postulated to play a central role in inhibitor interactions and provides the proton source essential for the terminal inactivation of class A β-lactamases like TEM-1 (53). Inspection of the OXA-1 structure reveals that an Arg244 equivalent is not present (52). The Lys at 212 is 3.5 Å from the C-3 carboxylate. Hence, this important difference (the lack of an Arg244 equivalent and the strategic position of the C-3 carboxylate of the inhibitor) is essential for understanding inhibitor susceptibility or resistance (16). It is also worth noting that class C β-lactamases also lack an Arg244 equivalent and demonstrate resistance to clavulanate, sulbactam, and tazobactam (30). Inspection of the class C β-lactamase structure of Beadle et al. shows that “similar” H-bonding interactions to the C-4 of cephalothin are made by residues Tyr150, Thr319, and W403 (5). Nukaga et al. showed that interactions with these residues were not seen in their study of the inactivation of the GC1 β-lactamase, an extended-spectrum AmpC enzyme with a tripepetide insertion inhibited by a related penem (41).

Finally, we conclude that the mechanism of inactivation by penems is emerging as an extremely important strategy in future β-lactamase inhibitor design. A wide range of OXA β-lactamases as well as inhibitor-resistant class A and C enzymes must be examined to assess the full potencies of these compounds. Nevertheless, it is eminently clear that the design of β-lactamase inhibitors that make stabilizing protein-ligand interactions (π-π interactions) and deacylate slowly can serve as an extremely important synthetic approach to novel β-lactamase inhibition (47). In this regard, penem inhibitors are proving to be key lead compounds with activities against all classes of serine β-lactamases (55).