Role of Arginine 214 in the Substrate Specificity of OXA-48

Increasing numbers of variants of the carbapenem-hydrolyzing class D β-lactamase OXA-48 are identified in Enterobacterales worldwide. Among them, OXA-181 and OXA-232 are of particular interest, as they differ from each other by a single amino acid substitution at position 214 (R in OXA-181 and S in OXA-232) that results in reduced carbapenem-hydrolyzing activity for OXA-232. To investigate the role of amino acid position 214 (AA214), the X-ray structure of OXA-232 was determined and AA214 of OXA-48 and of OXA-232 was replaced by G, L, D, E, S, R, and K using site-directed mutagenesis.

ABSTRACT

Increasing numbers of variants of the carbapenem-hydrolyzing class D β-lactamase OXA-48 are identified in Enterobacterales worldwide. Among them, OXA-181 and OXA-232 are of particular interest, as they differ from each other by a single amino acid substitution at position 214 (R in OXA-181 and S in OXA-232) that results in reduced carbapenem-hydrolyzing activity for OXA-232. To investigate the role of amino acid position 214 (AA214), the X-ray structure of OXA-232 was determined and AA214 of OXA-48 and of OXA-232 was replaced by G, L, D, E, S, R, and K using site-directed mutagenesis. These mutants were phenotypically characterized, and three mutants of OXA-232 were purified to study their steady-state kinetic properties. The X-ray structure of OXA-232 along with molecular modeling studies showed that the interaction via a salt bridge between R214 and D159 in OXA-48 is not possible with the G214 or S214 mutation. In contrast, with K214, which is also positively charged, the interaction with D159 is maintained. With the E214 mutant, an alternative binding conformation of imipenem that is not compatible with a nucleophilic attack by S70 was evidenced. Thus, imipenem has a very poor apparent affinity for the E214 mutant because of its nonproductive binding mode. Similarly, we could explain the lack of temocillin hydrolysis by the OXA-232-S214E mutant, which is due to the unfavorable interaction between the negatively charged R1 substituent of temocillin with the E214 residue. Overall, we demonstrate that AA214 in OXA-48-like β-lactamases is critical for the carbapenemase activity.

INTRODUCTION

The intensive use of antibiotics to treat infections led to the emergence of multidrug-resistant pathogens, especially in Gram-negative bacteria (GNB). Among the β-lactams, the carbapenems are considered last-resort antibiotics to treat severe infections caused by GNB (1). The major mechanisms of carbapenem resistance in Enterobacterales are (i) the association of a β-lactamase that very weakly hydrolyzes carbapenems (an extended-spectrum β-lactamase [ESBL] or a cephalosporinase) with decreased outer membrane permeability or (ii) inactivation by specific carbapenem-hydrolyzing β-lactamases (carbapenemases) (1). Genes encoding these enzymes are mostly harbored by plasmids, explaining their rapid spread. β-Lactamases are classified into 4 groups (Amber classes A to D), based upon sequence homology (1, 2). The clinically relevant carbapenemases in Enterobacterales belong to classes A (KPC type), B (NDM, VIM, IMP), and D (OXA-48-like). OXA-48, initially identified in Turkey (3), has since rapidly spread in the Mediterranean area, the Middle East, Europe, and India and is now turning into a major global threat (4). OXA-48 hydrolyzes penicillins, including temocillin, narrow-spectrum cephalosporins, and also carbapenems at a low rate, but it spares expanded-spectrum cephalosporins (ESC), e.g., ceftazidime and cefepime (5). Along with the spread of OXA-48, several variants that differ from OXA-48 by amino acid substitutions or deletions have been reported (http://bldb.eu/BLDB.php?class=D#OXA), and these are mostly located in the β5-β6 loop (6). OXA-181, which differs from OXA-48 by 4 amino acid substitutions (T103A, N110D, E169Q, and S171A), is the second most prevalent OXA-48 variant. OXA-232, which differs from OXA-181 by an additional substitution (R214S) in the β5-β6 loop, is particularly interesting, as its carbapenem-hydrolyzing activity was significantly impaired compared to that of OXA-181 or OXA-48 (7). These results suggested a pivotal role of R214 in the hydrolysis of carbapenems. To further investigate the role of R214 in the hydrolytic profile of OXA-48-like carbapenemases, the X-ray structure of OXA-232 and the steady-state kinetic parameters of in vitro-generated OXA-232 mutants with mutations at amino acid position 214 (AA214) were determined.

RESULTS

Susceptibility testing.

To determine the effect of the substitutions at amino acid position 214 on the hydrolysis of imipenem and temocillin, the MIC values of Escherichia coli expressing OXA-48 and OXA-232 and their respective point mutant derivatives were determined.

The R214S substitution in OXA-48 led to a phenotype similar to that of OXA-232, e.g., reduced MICs for temocillin and imipenem (Table 1). Conversely, OXA-232-S214R restored the MICs for temocillin and imipenem similarly to OXA-48/181. When AA214 was substituted with an uncharged amino acid, such as G and L, the MIC values were similar to those of OXA-232. The most interesting results were obtained by replacements with a negatively charged amino acid at pH 7.0, such as aspartic acid and glutamic acid. Indeed, the MIC values for imipenem and temocillin were remarkably affected, similar to the findings for the E. coli TOP10 control isolate.

TABLE 1.

MICs of β-lactams for E. coli TOP10(pTOPO-OXA-48) and variants, E. coli TOP10(pTOPO-OXA-232) and variants, and E. coli TOP10

β-Lactam	MIC (mg/liter)
	E. coli TOP10(pTOPO-OXA-48)							E. coli TOP10(pTOPO-OXA-232)							E. coli TOP10
	wt	R214S	R214G	R214L	R214K	R214D	R214E	wt	S214R	S214G	S214L	S214K	S214D	S214E
Temocillin	>256	32	32	64	128	8	8	32	256	32	64	128	8	8	8
Imipenem	0.75	0.25	0.25	0.25	0.38	0.25	0.25	0.38	0.75	0.25	0.25	0.5	0.25	0.25	0.25

In a second step, we focused our analysis on the OXA-232-S214K, OXA-232-S214E, and OXA-232-S214G mutants of OXA-232, which have amino acids representative of each amino acid group: a polar positively charged amino acid (K), a polar negatively charged amino acid (E), and an uncharged amino acid (G). The MIC values for other β-lactams were determined and compared to those for the strains with OXA-232, OXA-181 (OXA-232-S214R), and OXA-48 (Table 2). Overall, the MIC values of benzylpenicillin and cephalothin were not affected and the MIC values of the other carbapenems (meropenem and ertapenem) for the mutants varied in the same way that those of imipenem did.

TABLE 2.

MICs of β-lactams for E. coli TOP10(pTOPO-OXA-48), E. coli TOP10(pTOPO-OXA-181), E. coli TOP10(pTOPO-OXA-232) and variants, and E. coli TOP10

β-Lactam	MIC (mg/liter)
	E. coli TOP10(pTOPO-)						E. coli TOP10
	OXA-232-S214G	OXA-232-S214E	OXA-232-S214K	OXA-232	OXA-181	OXA-48
Benzylpenicillin	>256	>256	>256	>256	>256	>256	64
Temocillin	32	8	128	32	>256	>256	8
Cephalothin	16	16	16	16	16	16	4
Imipenem	0.25	0.25	0.5	0.38	0.75	0.75	0.25
Meropenem	0.047	0.032	0.094	0.047	0.125	0.094	0.016
Ertapenem	0.012	0.012	0.094	0.012	0.19	0.094	0.004

Biochemical property determination.

To further characterize the impact of the nature of the residue at position 214 on the hydrolytic profile, the steady-state kinetic parameters of the three OXA-232 mutants (OXA-232-S214G, OXA-232-S214E, and OXA-232-S214K) were determined and compared to those of OXA-48, OXA-181, and OXA-232 (5, 7). Overall, the kinetic studies revealed three patterns (Table 3). OXA-232-S214G exhibited hydrolytic activity toward all tested β-lactams, similar to that of OXA-232. The mutant OXA-232-S214K possessed a catalytic efficiency (k_cat/K_m = 83 mM⁻¹ · s⁻¹) for imipenem ∼4-fold higher than that of OXA-232 (k_cat/K_m = 20 mM⁻¹ · s⁻¹) (7), due to a weak increase in the turnover of the enzyme (k_cat). The most interesting result was observed with the OXA-232-S214E mutant. Indeed, replacement of the amino acid at position 214 by a negatively charged amino acid led to a drastic increase in K_m (>2,000 μM) of at least ∼200-fold compared to that of OXA-232 (K_m = 9 μM) (7), thus decreasing the catalytic efficiency. Moreover, this mutant totally lost its hydrolytic properties for temocillin. All these results were concordant with the observed MIC values. Taken together, our results confirm that the hydrolysis of imipenem depends on the nature of the residue at position 214. In light of these results, it appears that positively charged amino acids favor the hydrolysis of imipenem.

TABLE 3.

Kinetic parameters of OXA-232 mutants^a

Substrate	Km (μM)			kcat (s−1)			kcat/Km (mM−1 · s−1)
	OXA-232-S214E	OXA-232-S214G	OXA-232-S214K	OXA-232-S214E	OXA-232-S214G	OXA-232-S214K	OXA-232-S214E	OXA-232-S214G	OXA-232-S214K
Benzylpenicillin	140 ± 5	95 ± 6.5	95 ± 5	235 ± 20.6	130 ± 11.6	115 ± 9	1678 ± 91.6	1369 ± 166.5	1189 ± 72.2
Temocillin	NH	63 ± 6.7	28 ± 3.6	NH	0.01 ± 0.007	0.03 ± 0.007	NH	0.15 ± 0.07	1 ± 0.29
Cephalothin	100 ± 7	133 ± 12.1	160 ± 13.8	8 ± 0.6	4 ± 0.6	5 ± 0.6	81 ± 10.6	30 ± 3.8	31 ±1.7
Imipenem	>2,000	5 ± 2.6	6 ± 0.7	ND	0.1 ± 0.07	0.5 ± 0.014	ND	20 ± 6.9	83 ± 8.9

^aStandard deviation values are indicated after the kinetic parameter values. NH, no detectable hydrolysis was observed with 1 μM purified enzyme or 500 μM substrate; ND, not determined.

Crystal structure of OXA-232.

The crystals of OXA-232 were obtained at basic pH in 30% (wt/vol) polyethylene glycol (PEG) 3000 and in the presence of 0.2 M Li₂SO₄ (Fig. 1A). Glycerol was added to the mother liquor to cryoprotect the crystals before flash-annealing for data collection. They belonged to hexagonal space group P6₂ with 43.8% solvent, as calculated from the Matthews coefficient of 2.19 ų Da⁻¹ (5, 8). At the time of data acquisition (March 2013), the best available OXA-48 model in the Protein Data Bank (PDB) was 3HBR (1.9 Å) (5), which was obtained from crystals grown at pH 7.5 (lower than the pH 8.5 used in the present study) in PEG 4000 and then dipped into the cryoprotectant ethyleneglycol and which belonged to the monoclinic space group P2₁ with 48.3% solvent (Matthews coefficient, 2.38 ų Da⁻¹). After deletion of 22 N-terminal amino acids, one half of the R214S coordinates in the structure with PDB accession number 3HBR were used as a quasihomologous model for phasing the hexagonal crystal to a 2.2-Å resolution. The refined electron density appeared neat throughout the backbone and most of the side chains of the biological dimer per asymmetric unit of the crystal. Alternate positions for Q41 and Q53 were observed as well for D230, whose one position permitted a salt bridge with R107 at the dimer interface. The average real-space correlation factor, calculated by the use of SFCHECK software (9, 29), was 0.932. Compared to the structure with PDB accession number 3HBR, up to three additional C-terminal amino acids could be displayed in chain B of OXA-232 (only one could be displayed in chain A), but the 8-His tag chain, which was disordered in the surrounding solvent, could not. Additionally, some minor discontinuities observed in some parts of the main chains of the structure with PDB accession number 3HBR could be fixed in the structure with PDB accession number 5HFO (N50D/K51D, T99C, S150A H-bonding D148A). The root mean square deviation (RMSD) values with the two dimers from the structure with PDB accession number 3HBR ranged from 0.3452 to 0.4197 Å (Superpose) (10). Obviously, the OXA-232 tertiary structure of each monomer features the usual class D β-lactamase fold with an α-helical region (α3, residues 73 to 82; α4, residues 110 to 115; α5, residues 120 to 130; α6, residues 132 to 142; α9, residues 185 to 194) and a mixed α-helix/β-sheet region (β1, residues 26 to 28; α1, residues 31 to 35; β2, residues 42 to 48; β3, residues 53 to 56; α2, residues 59 to 62; β4, residues 196 to 199; β5, residues 204 to 212; β6, residues 219 to 227; β7, residues 232 to 240). A total of 98.2% of all residues were inside the favored regions of the Ramachandran plot and 1.8% were in the allowed regions. The quaternary structure of OXA-232 is dimeric, as observed for OXA-48 around a non-crystallographic symmetry 2-fold axis, in compliance with our data obtained by size exclusion chromatography. It was already shown that less monomer surface (200 Ų) was buried in that dimer formation than in the dimer formation with OXA-10 and that it was built upon an intermolecular β-sheet involving β4 from each subunit linked by reciprocal H bonds between the A199 NH and the A199 carbonyl O of one chain with the corresponding one related by the non-crystallograhic symmetry 2-fold axis. Several other H bonds and salt bridges also participated in the dimeric interface stabilization, and among these interactions, an anionic binding site lying on the non-crystallograhic symmetry 2-fold axis was tweezed by two facing R206 residues from non-crystallograhic symmetry-related β5 sheets. This contrasts with the cation-binding site described at the interface of OXA-10, involving other residues replaced in OXA-48. A water molecule was placed in the original structure with PDB accession number 3HBR, but on the basis of a spherical electron density shape, low B factor, and the recurrent presence of halogens in other class D β-lactamase structures (but not upon anomalous signal), it was replaced by a residual chloride coming from the previous purification steps. The same misinterpretation was also done in the structure with PDB accession number 4S2K (11), where chlorine was also reported in the crystallizing medium composition. Sulfate ions have also been seen to make that salt bridge in class D structures grown in the presence of a large amount of sulfate salts (e.g., the structures with PDB accession numbers 6NLW and 5FDH), but here six such anions have been spotted essentially at the molecule surface, as were five opportune glycerol molecules. Carbamylated lysines (KCX 73) can be described in a similar environment in both structures. Regarding the mutation of R214S (the R214 residue was shown in a water-mediated interaction with the drug avibactam in the active site in the structure with PDB accession number 4S2K [11]), this has not structurally disturbed the β5-β6 hairpin and its direct environment, just adding a few solvent molecules in the created void (two or three water molecules and one glycerol molecule in chain A) and modifying the L158B and I215B side chain conformers. Deeper into both active sites, respective carbamylated lysine (KCX 73) can be described in a similar environment in the structures of both OXA-232 and OXA-48. We note, however, the presence of a sulfate ion in the interaction with the same residues that recognize the sulfate moiety of the avibactam (Fig. 1B). More recently, the structure of another variant, OXA-181, which differs from OXA-48 by five substitutions (T103A, N110D, E169Q, S171A, and R214S) has been released (PDB accession number 5OE0) (12) at a similar resolution (2.05 Å). The RMSD with OXA-232 was even lower than that for the former overlay (0.224 and 0.252 Å), in relation to the fact that the structure crystallized in the same unit cell dimensions and space group P6₂ as OXA-232. However, the above-mentioned five substitutions are positioned on the protein surface, so they can potentially modulate the contacts leading to crystal growth. Additionally, the crystallization conditions used in the previous study were slightly different from those used in the present study, with a neutral pH and polyethylene glycol monomethyl ether (PEG MME) 5000 instead of PEG 3000 or PEG 4000 being used for the structure with PDB accession number 3HBR. Like OXA-232, sulfate has been used at a small quantity in the crystallization drop, and in the final model for the structure with PDB accession number 5HFO, one anion was reported to be bound to the amide N of R186A, as for OXA-232. An explanation or even prediction of the variability in β-lactamase crystal forms remains tenuous, since crystals of OXA-48 in complex with inhibitors (see the Beta-Lactamase DataBase at http://bldb.eu/S-BLDB.php for a complete list of structures and references) grew in the same unit cell dimensions (the choice of the P3₂ space group instead of the P6₂ space group was made to refine two more monomers and active sites independently) in different crystallizing solution conditions (2-methyl-2,4-pentanediol [MPD] was the major component) and in the absence of sulfate.

FIG 1.

(A) Superposition of crystal structures of OXA-48 (yellow; PDB accession number 3HBR) and OXA-232 (gray; PDB accession number 5HFO). The β5-β6 loop is delimited by the circle. (B) Partial active-site close-up of sulfate 302A-bound OXA-232. Atoms are colored by type (white, C; blue, N; red, O; yellow, S). Hydrogen bonding and electrostatic interactions are depicted as blue dashes. (C) Alternative docking conformations of imipenem (colored in magenta [A] and yellow [B], closed form of the β-lactam ring) as noncovalent (Michaelis) complexes with the OXA-48-R214E mutant (colored in cyan and generated from the structure with PDB accession number 6P97). Hydrogen bonds and favorable ionic interactions are shown as blue springs. Protein hydrogens are hidden for clarity. (D) Docking conformation of temocillin (colored in green, closed form of the β-lactam ring) as a noncovalent (Michaelis) complex with OXA-48 (colored in pink, PDB accession number 6P97) superposed with the OXA-48-R214E mutant (colored in purple and generated from the structure with PDB accession number 6P97; only the Gln214 residue is shown). Hydrogen bonds and favorable ionic interactions are represented as blue springs, and unfavorable ionic interactions are represented as yellow thick lines. Protein hydrogens are hidden for clarity.

Molecular modeling.

An in silico study was performed to identify the structural determinants that could explain the experimentally determined differences between the hydrolytic profiles of OXA-232 and its variants and those of OXA-48. The OXA-232 and OXA-48 structures were used as starting points, and the mutations S214G, S214K, and S214E were modeled based on predicted low-energy conformations (13). The resulting models showed that lateral chains of serine and lysine mutants are positioned in the same axis and direction as the lateral chain of arginine, without any clashes. The glutamate mutant would be positioned differently and would be oriented toward the active site, establishing hydrogen bonds with the backbone nitrogen and with the side chain hydroxyl of T213. This is the only conformation that avoids clashes of this glutamate mutant with the neighboring residues, I215 and D159, the latter of which is involved in OXA-48 in a salt bridge with R214 that maintains a closed conformation of the active site, which would stabilize the substrates (including carbapenems) in a conformation compatible with an efficient hydrolysis, thus influencing carbapenem turnover (5). For OXA-232 and its variants, the R214-D159 salt bridge cannot be formed, and this may explain the decrease in k_cat values compared to those for OXA-48.

The molecular docking calculations for imipenem and temocillin obtained with the R214E mutant of OXA-48 provided explanations for the different hydrolytic parameters that were observed experimentally. Imipenem showed two alternative binding modes: one that was known, with a single ionic interaction between the imipenem carboxylate and the side chain of R250, compatible with a nucleophilic attack by S70 (Fig. 1C), and a second one, presumably more stable, with two ionic interactions, consisting of one between the imipenem carboxylate and the side chain of R250, as in the previous binding mode, and a second one between the positively charged R2 substituent of imipenem and the mutated residue E214 (Fig. 1C). The latter binding mode is not productive, as it is not compatible with a nucleophilic attack by S70, thus increasing the apparent K_m value. To confirm this hypothesis, we determined the 50% inhibitory concentration (IC₅₀) for imipenem with OXA-232 and with the mutant OXA-232-S214E, the only variant with an increased K_m value. Interestingly, the values obtained (2 μM and 0.059 μM, respectively) showed that the mutant OXA-232-S214E had an IC₅₀ ∼34-fold lower than that of OXA-232. On the other hand, temocillin was the only β-lactam tested possessing a negative charge on the R1 substituent. Our docking calculations showed that this negative charge would establish strong unfavorable ionic interactions with the mutated residue E214, thus precluding binding in a conformation compatible with a nucleophilic attack by S70 (Fig. 1D).

DISCUSSION

OXA-48-producing Enterobacterales are now endemic in many countries, such as Turkey, countries in the Middle East and North Africa, and India, and have widely spread across Europe. Since the first description of OXA-48, several variants have been described (1, 6). These variants can be classified into 3 groups according to their hydrolysis profiles. Most of them, including OXA-181 or OXA-162 (7), have enzymatic activity similar to that of OXA-48 (5). The members of the second group, represented by OXA-163 (7), OXA-247 (14), and OXA-405 (15), have no carbapenemase activity but instead have a marked hydrolytic activity against ESC, similar to the findings for the OXA ESBLs (16). Finally, the members of the third group, represented by OXA-244 (17) and OXA-232 (7), exhibit an overall reduced activity toward all β-lactams, including carbapenems, compared to that of OXA-48. Comparison of the amino acid sequences of the OXA-48 variants suggested a link between the primary structure and the function of these enzymes. Indeed, all OXA-48 variants with an OXA ESBL phenotype (the loss of carbapenem hydrolysis and the gain of activity toward ESC) have amino acid deletions in the β5-β6 loop (Fig. 2). This observation suggests that this loop plays a role in substrate specificity. The phenotypic study of OXA-244 (which differs from OXA-48 by only one substitution, R214G) (17) and the enzymatic study of OXA-232 (which differs from OXA-181 by a single substitution, R214S) underline that residue 214 is crucial for carbapenem and temocillin hydrolysis. To confirm this hypothesis, we generated mutants of OXA-48 and OXA-232 with mutations at position 214, including OXA-232-S214R and OXA-48-R214S, and analyzed their hydrolytic profiles. Thus, the S214R substitution in OXA-232 restored the hydrolytic properties toward temocillin and imipenem. Similarly, the R214S substitution in OXA-48 led to a drastic decrease in imipenem and temocillin hydrolysis. In order to better understand this phenomenon, five mutants with substitutions at position 214 of OXA-48 and OXA-232 were generated. These substitutions corresponded to amino acids that were polar and positively charged (214K), polar and negatively charged (214E and 214D), nonpolar (214L), and glycine (214G). Overall, the substitutions with D, E, L, and G led to decreased MICs of imipenem and temocillin. The S214K substitution in OXA-232 led to increased MIC values, but they were still lower than those for OXA-48. The biochemical study of OXA-232-S214G and OXA-232-S214K showed that these 2 substitutions did not affect the apparent affinity toward β-lactams (K_m values were similar to those of OXA-232 and OXA-48) but did have an effect on the acylation or deacylation steps of substrate catalysis (k_cat values were similar to those of OXA-232 but smaller than those of OXA-48). The catalytic efficiencies (k_cat/K_m) of these two mutants were in agreement with the MIC values. The most significant differences were observed with OXA-232-S214E, which showed a drastic decrease in the apparent affinity for imipenem (the K_m of OXA-232-S214E was at least ∼200-fold higher than that of OXA-232 and OXA-48) and a loss of temocillin hydrolysis. The glutamate substitution seems to have a direct effect on the apparent affinity of imipenem. Analysis of the three-dimensional structure of OXA-48 showed that R214 interacts with D159 via a salt bridge (5), which maintains the shape and the network of water molecules within the binding site. Our molecular modeling study revealed that with the G214 or S214 mutation, this interaction with D159 was lacking, which presumably increases the flexibility of this part of the binding site. In contrast, K is a positively charged polar amino acid which appears to maintain the interaction with D159, even with a lateral chain shorter than that of R. In the E214 mutant, we evidenced an alternative binding conformation of imipenem that was not compatible with a nucleophilic attack by S70. In this case, imipenem acts as an inhibitor with an IC₅₀ value of 0.059 μM, which explains the significantly higher K_m values determined experimentally for this substrate. We also showed that, in the same E214 mutant, the unfavorable interaction between the negatively charged R1 substituent of temocillin with the E214 residue precludes the binding of this antibiotic, thus explaining the structural basis responsible for the lack of temocillin hydrolysis observed experimentally.

FIG 2.

Amino acid sequence alignment of OXA-48 variants. Asterisks indicate identical residues in all three sequences, the colons indicate a replacement by another amino acid but one with the same proprieties. Amino acid motifs that are well conserved among class D lactamases are indicated by gray boxes, and the black-outlined box corresponds to the β5-β6 loop. Positions D159 and R/S214 are indicated by thin and large arrows, respectively. The multiple-sequence alignment was performed with the CLUSTAL O (version 1.2.4) program.

Overall, we demonstrated that amino acid position 214 in OXA-48-like β-lactamases is critical for carbapenemase activity but also for temocillin hydrolysis. This point is of utmost clinical importance and explains why detection of OXA-244 or OXA-232 producers remains a challenge for clinical microbiology laboratories (17–19). Indeed, these isolates do not grow on ChromID Carba Smart medium (bioMérieux, Marcy l’Etoile, France), one of the media most commonly used for the screening of carbapenemase-producing Enterobacterales (18). The ChromID Carba Smart medium is present in a biplate containing on one side a carbapenem and on the other temocillin, two substrates that are only weakly hydrolyzed by OXA-244 and OXA-232 (18).

MATERIALS AND METHODS

Bacterial strains.

The clinical strain Escherichia coli LIEU (19), expressing the OXA-232 β-lactamase, was used to clone the bla_OXA-232 gene. E. coli TOP10 (Invitrogen, Saint-Aubin, France) was used for cloning and mutagenesis experiments, and E. coli BL21(DE3) was used for overexpression experiments (Novagen, Fontenay-sous-Bois, France).

Antimicrobial agents, susceptibility testing, and microbiological techniques.

Antimicrobial susceptibilities were determined by the disc diffusion technique on Mueller-Hinton agar (Bio-Rad, Marnes-la-Coquette, France) and interpreted according to the EUCAST breakpoints, updated in May 2018 (http://www.eucast.org). MICs were determined using the Etest technique (bioMérieux). All antibiotics except temocillin were purchased from Sigma (Saint-Quentin-Fallavier, France); temocillin was from Eumedica (Brussels, Belgium).

PCR, cloning, site-directed mutagenesis, and DNA sequencing.

The recombinant plasmids pTOPO-bla_OXA-232 and pTOPO-bla_OXA-48, obtained from a previous study (7), were used as the templates for the site-directed mutagenesis assays, and specific primers were designed for the different mutations, using the program QuikChange Primer Design (Agilent Technologies). A QuikChange II site-directed mutagenesis kit (Agilent Technologies) was used, following the manufacturer’s recommendations, in order to replace AA214 with a glycine (G), a lysine (K), a leucine (L), an aspartic acid (D), a glutamic acid (E), a serine (S), and an arginine (R). Mutagenesis reaction products were transformed into E. coli TOP10 cells (Invitrogen, Saint-Aubin, France), and selection was performed on a tryptic soy agar plate containing kanamycin (50 μg/ml). For the production of OXA-232, genes were amplified by PCR using the forward primer OXA_23-256NdeI (5′-AAAAACATATGAAGGAATGGCAAGAAAACAAA-3′) and the reverse primer OXA_XhoI-Δstop (5′-AAAAACTCGAGGGGGAATAATTTTTCCTGTTTGAG-3′), to increase the purification yield. For the mutants OXA-232-S214E, OXA-232-S214G, and OXA-232-S214K, the forward primer OXA_NdeI (5′-AAAAACATATGTTGGTGGCATCGATTATCGG-3′) and the reverse primer OXA_BamHI (5-AAAAACTCGAGGAGCACTTCTTTTGTGATGGC-3′) were used. Recombinant plasmids were cloned into pET41b (for the OXA-232 wild type [wt]), allowing the expression of the enzyme with a His tag and pET9a (for OXA-232 mutants) vector (Invitrogen, Life Technologies, Cergy-Pontoise, France), and then transformed into E. coli BL21(DE3) cells (Novagen). All the recombinant plasmids were extracted using a Qiagen miniprep kit and sequenced, using a T7 promoter and M13 reverse primers or T7 terminator (depending on the plasmid), with an automated sequencer (ABI Prism 3100; Applied Biosystems). The nucleotide sequences were analyzed using software available at the National Center for Biotechnology Information website (https://www.ncbi.nlm.nih.gov).

Protein purification.

Overnight cultures of E. coli BL21(DE3) harboring recombinant plasmid pET41b-OXA-232 or pET9a-OXA-232-mut were used to inoculate 2 liters of LB medium broth containing 50 mg/liter kanamycin. The bacteria were cultured at 37°C until an optical density of 0.6 at 600 nm was reached. The β-lactamase was induced overnight with 0.2 mM IPTG (isopropyl-β-d-thiogalactopyranoside) as the inducer, and the cultures were centrifuged at 6,000 × g for 15 min. The pellets were resuspended with the binding buffer 25 mM phosphate sodium, pH 7.4, 300 mM K₂SO₄, 10 mM imidazole for the OXA-232 wt and with the buffer 20 mM bis-Tris H₂SO₄ (pH 7.2) for mutants of OXA-232. Bacterial cells were disrupted by sonication, and the bacterial pellet was removed by two consecutive centrifugation steps at 10,000 × g for 1 h at 4°C; the supernatant was then centrifuged at 96,000 × g for 1 h at 4°C. The OXA-232 wt was purified with a nitrilotriacetic acid-nickel column (GE Healthcare, Freiburg, Germany) by using the elution buffer 25 mM phosphate sodium, pH 7.4, 300 mM K₂SO₄, 500 mM imidazole. Mutants of OXA-232 were purified by using 2 anion-exchange chromatographies on a HiTrap QHP GE Healthcare chromatograph (with 20 mM bis-Tris H₂SO₄, pH 7.2, and then 20 mM piperazine H₂SO₄, pH 9.5) (7). Finally, a gel filtration step was performed for the purification of all β-lactamases with 100 mM sodium phosphate buffer, pH 7, and 150 mM NaCl with a Superdex 75 column (GE Healthcare, Freiburg, Germany). The protein purity was estimated by SDS-PAGE, and the pooled fractions were dialyzed against 10 mM Tris-HCl (pH 7) for the mutants and against 0.1 M HEPES (pH 7.5) for the OXA-232 wt and concentrated using Vivaspin 20 columns (10,000-molecular-weight-cutoff, polyethersulfone [PES] membrane; Sartorius). The protein concentration was determined according to the Bradford method using a Bio-Rad protein assay standard II kit (Bio-Rad, Marnes-la-Coquette, France) with bovine serum albumin (BSA) as a standard.

Kinetics assays.

Steady-state kinetic parameters were determined using a spectrophotometer (Ultrospec 2000; Amersham Pharmacia Biotech), and the assays were performed at 30°C in 100 mM phosphate buffer (pH 7) for 10 min (7). The disappearance of substrate (β-lactam antibiotics) was monitored at the specific wavelengths and converted to initial velocities using the specific extinction coefficients. The k_cat and K_m values were determined by analyzing β-lactam hydrolysis under initial-rate conditions using the Eadie-Hoffstee linearization of the Michaelis-Menten equation with SFWIFT II software (7). Fifty percent inhibitory concentrations (IC₅₀) for imipenem were determined in 100 mM sodium phosphate buffer (pH 7) and 100 μM benzylpenicillin as a reporter substrate.

Crystallization and data collection.

Crystal conditions were screened with 10 mg/ml of concentrated protein using a range of commercially available screens. OXA-232 crystals grew in 0.2 M lithium sulfate, 0.1 M Tris, pH 8.5, and 30% (wt/vol) PEG 3000. Crystals were briefly dipped in cryoprotectant (20% [vol/vol] glycerol in the reservoir solution) and flash-frozen in liquid nitrogen prior to being transferred in a stream of nitrogen at 100 K (delivered by a Rigaku X-tream cryosystem). A 2.21-Å-resolution data set was collected on a Rigaku MicroMax-007-HF rotating-anode generator with Cu K-α radiation, Varimax HF mirror focusing optics (Rigaku), and a MAR345 image-plate detector (MAR Research).

Structure determination and refinement.

X-ray diffraction data sets were processed using the XDS (20) and AIMLESS (21) programs; the crystals diffracted to a 2.21-Å resolution (Table 4). Initial phases were obtained by molecular replacement with the MOLREP program (22) using OXA-48 (PDB accession number 3HBR) as a search probe. Refinement was performed by successive and alternate rounds of refinement with the BUSTER program (23), and model improvement was performed using the Coot program (24). The final model was evaluated using the MolProbity program (25). Data collection and refinement statistics are provided in Table 4.

TABLE 4.

X-ray crystallography and refinement statistics

Parametera	Valueb
Data collection statistics
Diffractometer	Rigaku MicroMax-007-HF-MAR345
Wavelength (Å)	1.5418
Temp (K)	100
Crystal-to-detector distance (mm)	200
Rotation range/image (°)	1
Total rotation range (°)	205
Exposure time/image (min)	5
Space group	P62
Unit cell dimensions
a, b, c (Å)	144.08, 144.08, 53.13
α, β, γ (°)	90.00, 90.00, 120.00
Resolution (Å)	17.2–2.21
Rmerge	0.134 (0.527)
I/σ 〈I〉	13.8 (4.1)
% completeness	97.6 (90.5)
Multiplicity	7.0 (5.7)
Refinement statistics
Resolution range (Å)	17.2–2.21
No. of reflections
Unique	30,389
Working set	30,389 (2,284)
Test set	1,534 (102)
Rwork/Rfree	0.172/0.208 (0.343/0.404)
Overall B factor from Wilson plot of the Wilson B factor (Å2)	24.3
Cruickshank’s DPI for coordinate error (Å)	0.179
No. of nonhydrogen atoms	4,551
Protein	4,014
Water	471
Ligand/ions	66
Avg B (Å2)
All atoms	34.72
Protein	33.63 (31.01 [main chain]/36.11 [side chain])
Water	42.85
Ligand/ions	46.37
Root mean square deviations
Bond lengths (Å)	0.01
Bond angles (°)	1.03
Ramachandran plot (%)
Most favored	98.2
Allowed	1.8

^aI, the intensity of a reflection; 〈I〉, average intensity; DPI, diffraction-component precision index.

^bValues in parentheses are for the outer shell (2.21 to 2.36 Å).

Molecular modeling.

Molecular modeling was performed to evaluate the effects of the mutations on the OXA-232 β-lactamase. Substitution of amino acids was performed by mutating the OXA-232 (PDB accession number 5HFO) and OXA-48 (PDB accession number 6P97) structures in silico using the Dunbrack rotamer library (swapaa command), which is a part of UCSF Chimera software (13, 26). The Dunbrack rotamer library predicts the conformation of the amino acid side chain based on the global energy minimum of the protein. The identification of interatomic clashes based on Van der Waals radii (27) was performed with UCSF Chimera software (26). Three-dimensional structures of the β-lactam ligands were generated using the Corina (version 3.60) program (Molecular Networks GmbH, Erlangen, Germany). Molecular docking calculations were performed using the Gold program (Cambridge Crystallographic Data Centre, Cambridge, UK) (28) and the GoldScore scoring function. The binding site, defined as a 20-Å-radius sphere, was centered on the OG oxygen atom of Ser70. All other parameters had default values. The receptor-ligand complex images were produced using UCSF Chimera software (26).

Data availability.

The OXA-232 structure atomic coordinates were deposited in the Protein Data Bank under accession number 5HFO.