Understanding Resistance to β-Lactams and β-Lactamase Inhibitors in the SHV β-Lactamase

Bacterial resistance to β-lactam/β-lactamase inhibitor combinations by single amino acid mutations in class A β-lactamases threatens our most potent clinical antibiotics. In TEM-1 and SHV-1, the common class A β-lactamases, alterations at Ser-130 confer resistance to inactivation by the β-lactamase inhibitors, clavulanic acid, and tazobactam. By using site-saturation mutagenesis, we sought to determine the amino acid substitutions at Ser-130 in SHV-1 β-lactamase that result in resistance to these inhibitors. Antibiotic susceptibility testing revealed that ampicillin and ampicillin/clavulanic acid resistance was observed only for the S130G β-lactamase expressed in Escherichia coli. Kinetic analysis of the S130G β-lactamase demonstrated a significant elevation in apparent Km and a reduction in kcat/Km for ampicillin. Marked increases in the dissociation constant for the preacylation complex, KI, of clavulanic acid (SHV-1, 0.14 μm; S130G, 46.5 μm) and tazobactam (SHV-1, 0.07 μm; S130G, 4.2 μm) were observed. In contrast, the kinacts of S130G and SHV-1 differed by only 17% for clavulanic acid and 40% for tazobactam. Progressive inactivation studies showed that the inhibitor to enzyme ratios required to inactivate SHV-1 and S130G were similar. Our observations demonstrate that enzymatic activity is preserved despite amino acid substitutions that significantly alter the apparent affinity of the active site for β-lactams and β-lactamase inhibitors. These results underscore the mechanistic versatility of class A β-lactamases and have implications for the design of novel β-lactamase inhibitors.

It is particularly noteworthy that, among the inhibitor-resistant class A ␤-lactamases, the S130G substitution in TEM and SHV involves the conserved motif Ser-130/Asp-131/Asn-132, the "SDN" loop. Ser-130 and Asn-132 are hydrogen-bonded to the catalytically important Lys-73 in the active site. This observation challenged our understanding of the role of Gly at the 130 position in inhibitor-resistant variants of the SHV and TEM ␤-lactamases (e.g. SHV-10 and TEM-59) (5,13). Exploring the consequences of amino acid substitutions on catalytic activity, we show that the apparent affinity of the active site is greatly reduced for substrates and inhibitors. In contrast, S130G and SHV-1 ␤-lactamases are inactivated at nearly the same rate. Our hypothesis, based on a molecular model, is that movement of a catalytic water molecule may potentially serve the same role as the hydroxyl group of Ser-130 in substrate binding and catalysis. These results have implications for the future design of ␤-lactams and ␤-lactamase inhibitors.
Mutagenesis-PCR-based site-saturation and site-directed mutagenesis was performed using a QuikChange™ mutagenesis kit (Stratagene), as reported (9,15). Two complementary degenerate oligonucleotides (Genosys Biotechnologies, The Woodlands, TX) were first constructed for site-saturation mutagenesis at the Ser-130 position. For site-directed mutagenesis, specific oligonucleotides were designed based upon common codon usage (Genosys Biotechnologies).
DNA Sequencing-We performed DNA sequencing with an A.L.F. Express automated DNA sequencer (Amersham Biosciences) using the Thermo Sequenase TM fluorescent-labeled primer cycle sequencing kit in a manner similar to methods published previously (8,9,15).
Antibiotic Susceptibility-E. coli DH10B expressing the mutated bla SHV genes were phenotypically characterized by LB agar dilution minimum inhibitory concentrations (MICs) 2 using a "Steers replicator" that delivered 10 4 colony forming units/spot. Antibiotics used and their suppliers were described previously (8,9,15). Concentrations employed for determining MIC values were in g/ml. MICs were read after 18 -24 h of incubation at 37°C and performed three times for each antibiotic.
Determination of Steady-state Expression Using an Enzyme-linked Immunosorbent Assay-By using a polyclonal anti-SHV antibody, a quantitative enzyme linked immunosorbent assay (ELISA) method was employed to assess steady-state expression levels (16). The ELISA plates were read at ϭ 492 nm and OD values of the samples were compared with an internal standard curve using the SHV-1 ␤-lactamase.
Kinetics-Kinetic constants of the SHV-1, S130G and S130T ␤-lactamases were determined by continuous assays at room temperature, 25°C, using an Agilent™ 8453 diode array spectrophotometer. Each experiment was performed in 20 mM phosphate-buffered saline at pH 7.4. Measurements were obtained using ampicillin (⌬⑀ 235 The kinetic experiments and analyses were patterned after the methods of Imtiaz et al. (17) and will be described briefly below.
The kinetic parameters, V max and K m , were obtained with non-linear least squares fit of the data (Michaelis-Menten equation) using A direct competition assay was performed to determine the dissociation constant for the preacylation complex, K I, of the inhibitors (clavulanic acid and tazobactam). We used a final concentration of 100 M nitrocefin as the indicator substrate and 7 nM SHV-1 ␤-lactamase or 31 nM S130G ␤-lactamase in these determinations. The data were analyzed according to Equation 2 to account for the affinity of nitrocefin for the SHV-1 ␤-lactamase: The first-order rate constant for enzyme and inhibitor complex being inactivated, k inact was measured directly by monitoring the reaction time courses in the presence of inactivators. A fixed concentration of enzyme, nitrocefin, and increasing nM concentrations of inactivator, I, were used in each assay. The k obs for inactivation was determined graphically as the reciprocal of the ordinate of the intersection of the straight lines obtained from the initial, v 0 , and final, v f , steady-state velocities. The k obs was calculated from the relationship k obs ϭ 1/t. Each k obs was plotted versus I and fit to Equation 3 to determine k inact : The second-order rate constant, k inact /K I (the rate constant for reaction of free enzyme with free inhibitor to give inactive enzyme) was determined by the fit of k obs to Equation 3.
The partitioning of the initial enzyme inhibitor complex between hydrolysis and enzyme inactivation, i.e. the turnover number (t n ϭ k cat /k inact ) was obtained in the following manner. First, we incubated increasing amounts of inhibitor (clavulanic acid or tazobactam) with a fixed concentration of SHV-1 or S130G ␤-lactamase in a total volume of 100 l of 20 mM phosphate-buffered saline, pH 7.4, at room temperature. After 24 h, an aliquot (10 l) was removed from the mixture and the steady-state velocity was measured and compared with a control sample with no inhibitor added. The proportion of clavulanic acid or tazobactam relative to SHV-1 or S130G that resulted in 90% inactivation after 24 h was used to determine the ratio of inhibitor to enzyme in with SHV-1, S130G, S130T, and S130X ␤-lactamases Antibiotic SHV-1 S130G S130T S130X a   a 90-min experiment. We reacted 1.6 M SHV-1 and 25 M S130G with 60 M and 1 mM clavulanic acid, respectively, in a final volume of 200 l. Similarly, we reacted 1.6 M SHV-1 and 25 M S130G with 7.75 M and 125 M of tazobactam, respectively. From these mixtures, a set aliquot was removed at designated time points, and steady-state velocities were measured. The contributions of amino acid substitutions to the relative change in Gibbs free energy of the transition state with substrates and inhibitors were determined by using Equations 4 and 5 (18 -20): ⌬⌬G ϭ ϪRT ln(k inact /K I S130G /k inact /K I SHV-1 ) (Eq. 5) Molecular Modeling-Insight II™ software (Accelrys) was used to construct the S130G variant and to perform energy minimizations as described previously (9) based on the SHV-1 crystal structure (Protein Data Bank code 1SHV) (21).

Mutagenesis-Site-saturation and site-directed mutagenesis
to obtain all 19 amino acid substitutions at the 130 position in SHV-1 was successfully performed, and the most common codons used were selected (8). The DNA sequence of each mutant bla SHV was first determined in E. coli™ XL1 Blue cells and confirmed in E. coli DH10B cells.
Antibiotic Susceptibility-The effects of each amino acid substitution at the 130 position in the SHV ␤-lactamase expressed in E. coli DH10B are summarized in Table I (also Fig. 1 and Table I in "Supplemental Material"). Only the S130G and S130T ␤-lactamases expressed in E. coli were ampicillin resistant (MICs Ն 32 g/ml). Resistance to ampicillin/clavulanic acid was demonstrated by the E. coli DH10B strain with the S130G ␤-lactamase. None of the variant ␤-lactamases expressed in E. coli conferred more resistance than SHV-1 against ampicillin/tazobactam or piperacillin/tazobactam. ␤-Lactamase Expression Levels-To investigate whether increased relative amounts of ␤-lactamase produced were responsible for the observed MICs, we assayed ␤-lactamase expression levels in the constructs using a polyclonal antibody in an ELISA format (16) (Fig. 1 ''Supplemental Material''). Except for S130D, all variants produced less ␤-lactamase than the wildtype enzyme, particularly the S130G ␤-lactamase (11% relative to SHV-1 ␤-lactamase). Hence, increased quantities of S130G enzyme were not responsible for the resistance to ampicillin/ clavulanic acid.
Kinetic Parameters of SHV-1 and S130G ␤-Lactamases-The steady-state kinetic parameters are reported in Table II. Large increases in K m were demonstrated by S130G ␤-lactamase for nitrocefin (700 M, 140-fold increase). For the S130G ␤-lacta-mase the k cat values were reduced for ampicillin and nitrocefin. The k cat /K m ratios clearly show that the S130G mutation markedly reduced catalytic efficiency. Compared with SHV-1 ␤-lactamase, S130G is 3.6% as efficient catalyzing ampicillin and 0.14% as efficient hydrolyzing nitrocefin (Table III).
Most strikingly, we observed that the K I for clavulanic acid with the S130G variant was 340-fold greater (0.14 M versus 47 M; Table IV). For tazobactam, the difference in K I was also significantly elevated (K I ϭ 0.07 M for SHV-1 versus 4.2 M for S130G; 60-fold increase). The difference in first-order rate constants for inactivation, k inact s, was within 17% for clavulanic acid and 40% for tazobactam (Table IV). The second-order rate constants, k inact /K I , demonstrate a larger difference between SHV-1 and S130G ␤-lactamase, the increase in K I being responsible for this.
In the time-dependent inactivation experiments, the ratio of I/E required to reduce activity by 90% (t n ) was equal for SHV-1 and S130G at 24 h (clavulanic acid, 40/1; tazobactam, 5/1). At 20 min, the period corresponding to the dividing time of E. coli, the activity of S130G was reduced 55% by clavulanic acid; SHV-1 ␤-lactamase activity was reduced Ͼ 90% (Fig. 2). In contrast, by 15 min, SHV-1 and S130G ␤-lactamases were inactivated Ͼ 80% by tazobactam. The k cat s indicate that both SHV-1 and S130G ␤-lactamases hydrolyze clavulanic acid and tazobactam (Table IV).
Molecular Modeling-Molecular modeling calculations indicate that the S130G ␤-lactamase is not significantly different in overall energy when compared with SHV-1, despite the loss of a key H-bonding residue in the active site. We do not observe other side chains close enough to assume the role of Ser-130 (22). In contrast, the catalytic water moves 2.37 Å nearer to the Lys-73:HZ2 atom. Additionally, there is movement of the important Lys-73 residue (0.5 Å for 73: HZ2) toward the catalytic Ser-70. The changes in active-site relationships as a result of the S130G substitution are shown in Fig. 3.

DISCUSSION
These data show that the S130G amino acid substitution is essential for resistance to ampicillin/clavulanic acid in SHV ␤-lactamase. The major effect of the S130G mutation on apparent K m and K I emphasizes the precise topological role of this amino acid (17,(22)(23)(24)(25)(26)(27). A direct consequence of the Ser 3 Gly substitution is reflected by the amount of S130G versus SHV-1 inhibited in 20 min, the time required for E. coli cell division. The time-dependent inactivation of S130G by tazobactam occurred more rapidly than with clavulanic acid. The magnitude of the K I elevation combined with the rates of inactivation for tazobactam was not enough to result in an elevation of ampicillin/tazobactam (data not shown) or piperacillin/tazobactam MICs. This has been noted previously with laboratory isolates that produce larger amounts of ␤-lactamase than clinical strains.
In addition to these findings, two important observations of the kinetic behavior of SHV-1 and S130G ␤-lactamases with clavulanic acid and tazobactam deserve emphasis: S130G and SHV-1 ␤-lactamases are inactivated by clavulanic acid at sim-    magenta). Key active-site residues are highlighted as well as the catalytic water molecule (CAT H 2 O). These structures were obtained using the AMBER force field. ilar rates, and hydrolysis rates are comparable. Analogous results were obtained with tazobactam. These data indicate that the S130G ␤-lactamase, as a clinically important ␤-lactam/ ␤-lactamase inhibitor-resistant enzyme, maintains the fundamental hydrolytic mechanism common to all ␤-lactamases (12).

FIG. 3. Stereo view of the superposition of the energy-minimized active-site structures of SHV-1 wild type (shown colored by element) and the S130G SHV variant (shown in
From our model (Fig. 3), we hypothesized that an active-site water molecule may relocate to a region nearer to the Lys-73 residue. The water molecule may also function with Lys-73 in a coordinated proton shuttle and may substitute for the Ser-130 in protonating the ␤-lactam nitrogen (26,27). In the x-ray structures of the inhibitor resistant TEM-30 (R244S) and TEM-84 (N276D), the relocation of a conserved water plays a crucial role in the binding affinity of inhibitors (10,12). The molecular modeling studies of the E166N amino acid substitution of TEM-1 ␤-lactamase revealed the movement of a catalytic water molecule was essential in modification of the substrate profile (28). In a similar manner, movement of a water molecule in the active site of S130G may be an important process that preserves catalytic activity.
The mechanism by which S130G is inhibited by clavulanic acid or tazobactam challenges our understanding of the catalytic pathway to inactivation. By using the x-ray structure of SHV-1 inhibited by tazobactam as a guide, we hypothesize that after acylation of S130G by clavulanic acid 1, the reactive imine 2 is formed (Fig. 4). The imine rearranges to form the cis-and trans-enamine 3, 4. This stable intermediate could account for the inhibition observed in our experiments and would occur in both the variant and wild-type enzymes. The absence of the side chain of Ser-130 would deprive this enzyme of the hydroxyl group that acts as the necessary nucleophile required in the final irreversible inactivation step (6,22). Hence, formation of the cross-linking or bridging species (vinyl ether), and the Ser-130 bound carboxylic acid moiety would not be possible in the S130G ␤-lactamase. However, nucleophilic attack by an active-site water molecule could yield the aldehyde 5 or hydrate aldehyde 6. This also may also lead to inhibition. Lastly, the aldehyde 5 or hydrate aldehyde 6 could undergo hydrolysis and regenerate the active enzyme 7 (22). Alternatively, the imine 2 could be directly hydrolyzed to yield the aldehyde 5 and hydrate aldehyde 6. This pathway is entirely consistent with that proposed for the inhibitor-resistant TEM-33 (M69L) and N276D-substituted enzymes (10,11). Our kinetic data for tazobactam and clavulanic acid suggest that the rate that each intermediate is formed and the number of branch points differ for each inhibitor. Experiments are in progress attempting to define the predominate pathway for clavulanic acid and tazobactam in inhibitor-resistant SHV enzymes. The design of future ␤-lactam and ␤-lactamase inhibitors should take into ac-count that alterations in the oxyanion hole and free movement of water molecules in the active site will permit more than one pathway to catalysis and inhibition. A deeper appreciation of these alternative mechanisms may avert the unwelcome emergence of even more resistant ␤-lactam bacteria because of excessive ␤-lactam use.