OXA-163, an OXA-48-Related Class D β-Lactamase with Extended Activity Toward Expanded-Spectrum Cephalosporins

Abstract

Two bla_OXA-48-like-positive isolates (Klebsiella pneumoniae and Enterobacter cloacae) were recovered in Argentina in 2008 as part of a large-scale survey focused on multidrug resistance in Enterobacteriaceae. In both cases, sequencing identified β-lactamase OXA-163, differing from OXA-48 by a single amino substitution and a 4-amino-acid deletion. OXA-163 hydrolyzed penicillins, ceftazidime, and cefotaxime, whereas OXA-48 did not. However, OXA-163 had a much lower ability to hydrolyze carbapenems than OXA-48, therefore barely being considered a carbapenemase. In both isolates, the bla_OXA-163 gene was located on plasmids that differed in structure and size. However, a detailed genetic analysis revealed a similar genetic context in those isolates, with the bla_OXA-163 gene being bracketed by novel transposase genes, making this genetic environment different from that reported for the bla_OXA-48 gene. This study identified the first class D β-lactamase compromising both extended-spectrum cephalosporin and carbapenem activities.

INTRODUCTION

Class D β-lactamases or oxacillinases are widely distributed among clinically relevant Gram-negative species (31). A high degree of diversity is observed at the biochemical and genetic levels among these enzymes exhibiting a narrow or broad spectrum of hydrolysis toward β-lactams (31). Genes encoding oxacillinases are usually embedded as gene cassettes into class 1 integrons, but other potentially mobile genetic vehicles have also been described, including insertion sequences (ISs) (31). Among class D β-lactamases, some enzymes confer the ability to hydrolyze carbapenems and have mostly been identified in Acinetobacter spp. (32) but rarely in Enterobacteriaceae (33). The carbapenem-hydrolyzing class D β-lactamase (CHDL) OXA-48 was initially identified in Turkey, first in Klebsiella pneumoniae and then in other enterobacterial species (1, 5, 6, 15, 23, 29). OXA-48 exhibits a hydrolysis profile that includes penicillins and carbapenems and that spares cephalosporins (29). Lately, the bla_OXA-48 gene has been identified in different countries, such as Lebanon, Tunisia, Israel, Belgium, and France and recently in Senegal and Morocco (3, 9–11, 14, 19, 20–22, 34). Several isolates producing OXA-48 have been involved as a source of nosocomial outbreaks (5, 11). The bla_OXA-48 gene has often been identified on a 70-kb plasmid featured by the replication protein RepP (6). The bla_OXA-48 gene is usually identified in association with two IS1999 insertion sequences, forming the composite transposon Tn1999 (2). Here we report an OXA-48-related CHDL that possesses very peculiar hydrolytic properties and whose gene is associated with novel genetic structures.

MATERIALS AND METHODS

Bacterial strains.

Identification of K. pneumoniae (isolate Kp6299) and Enterobacter cloacae 2185 (isolate Enc2185) was performed by using the API 20E system (bioMérieux, Marcy l'Étoile, France). Escherichia coli TOP10 was used as the host strain for cloning, and E. coli J53 (resistant to azide) was used as the host for conjugation assays.

Antimicrobial agents and MIC determination.

Susceptibility testing was performed by disk diffusion assay (Sanofi-Diagnostic Pasteur, Marnes-la-Coquette, France) as previously described (16). Results were interpreted according to the CLSI guidelines (7). The MICs were determined by Etest (AB bioMérieux, Solna, Sweden) on Mueller-Hinton agar plates at 37°C. Extended-spectrum β-lactamase (ESBL) detection was performed by double-disk synergy tests as described elsewhere (16).

Cloning experiments, PCR, and DNA sequencing.

Whole-cell DNAs were extracted as previously described. PCR screening for carbapenemase-encoding genes of classes A (bla_GES, bla_KPC), B (bla_VIM, bla_IMP), and D (bla_OXA-48) and for ESBL-encoding genes bla_TEM, bla_SHV, bla_CTX-M, bla_PER, bla_VEB, bla_BES, and bla_TLA-1 was performed as described previously (5, 13, 17, 25). PCR combinations were performed as described previously (2) in order to identify the composite transposon Tn1999. In order to express the different bla_OXA-48-like genes in an isogenic background, cloning of the bla_OXA-48 and bla_OXA-163 genes into E. coli TOP10 was performed as described previously (5), using a pCR-BluntII-TOPO cloning kit (Invitrogen, Cergy-Pontoise, France), followed by selection on plates containing 50 μg/ml of ticarcillin and 30 μg/ml of kanamycin, giving rise to E. coli TOP10(pOXA-48) and E. coli TOP10(pOXA-163), respectively. The PCR fragments comprising the entire bla_OXA-48-like sequences and used for cloning were obtained with primers preOXA-48A (5′-TATATTGCATTAAGCAAGGG-3′) and preOXA-48B (5′-CACACAAATACGCGCTAACC-3′). Both strands of the cloned DNA inserts of recombinant plasmids were sequenced by using an Applied Biosystems sequencer (ABI 377; Courtaboeuf, France).

In order to analyze the genetic environment of bla_OXA-163, plasmid DNA of the transformants was extracted using a commercial kit (Qiagen, Courtaboeuf, France). Plasmids were digested by the PstI restriction enzyme and ligated into the PstI-digested pBKCMV vector. It gave rise to recombinant plasmids p2185 and p6299, which were electroporated into E. coli TOP10. Recombinant strains were selected on LB agar plates containing ticarcillin (50 μg/ml) and kanamycin (30 μg/ml). Both strands of the cloned DNA inserts of the recombinant plasmids were sequenced, and deduced protein sequences were analyzed with software available from the NCBI website (http://www.ncbi.nlm.nih.gov/BLAST/).

Plasmid content and conjugation assays.

Plasmid DNAs obtained from isolates Kp6299 and Enc2185 were extracted by using the Kieser method (18). Whole plasmid preparations were electroporated into E. coli TOP10. E. coli NCTC50192, harboring four plasmids of 154, 66, 48, and 7 kb, was used as the size marker for plasmids. Plasmid DNAs were analyzed by agarose gel electrophoresis. Direct transfer of the β-lactam resistance markers into E. coli J53 was attempted by liquid mating-out assays at 37°C. Selection was performed on agar plates supplemented with ticarcillin (50 μg/ml) and sodium azide (100 μg/ml).

Plasmid characterization.

Plasmid DNA was obtained by using a Maxiprep plasmid purification kit (Qiagen). Plasmid incompatibility groups were determined by a PCR-based replicon-typing (PBRT) scheme (4).

β-Lactamase purification.

Cultures of E. coli TOP10 harboring recombinant plasmid pOXA-163 were grown overnight at 37°C in 2 liters of Trypticase soy broth containing 50 μg/ml of ticarcillin and 30 μg/ml of kanamycin. The protein extracts obtained were purified as described previously (29). Briefly, after sonication, the crude extract was ultracentrifuged at 100,000 × g. Isoelectric focusing (IEF) analysis was performed with an Ampholine polyacrylamide gel (pH 3.5 to 9.5), as described previously (24), using a crude β-lactamase extract from a culture of E. coli TOP10(pOXA-163). The focused β-lactamases were detected by overlaying the gel with 1 mM nitrocefin (Oxoid, Dardilly, France) in 100 mM phosphate buffer (pH 7.0). Then, the supernatant was subjected to further purification steps, including fast protein liquid-ion-exchange chromatography with a Q-Sepharose column (GE Healthcare, Vélizy, France) and 20 mM bis-Tris buffer (pH 6.5). The β-lactamase was recovered in the flowthrough. The extract was subsequently dialyzed against a 20 mM diethanolamine-H₂SO₄ (pH 9) buffer and loaded again onto the preequilibrated column. The fractions containing the β-lactamase were eluted with a linear K₂SO₄ gradient. The same procedure was repeated using a 20 mM Tris (pH 8.2) buffer. Finally, the fractions containing the highest β-lactamase activity were pooled and subsequently dialyzed overnight against 100 mM phosphate buffer (pH 7). The β-lactamase activity was determined qualitatively using nitrocefin hydrolysis (Oxoid, Dardilly, France). The protein content was measured using the Bio-Rad DC protein assay (Marnes-la-Coquette, France). The purification factor was measured by comparing the activities of the OXA-163 crude extract and the purified enzyme using 100 μM amoxicillin as substrate.

Kinetic studies.

Purified β-lactamase was used for kinetic measurements performed at 30°C in 100 mM sodium phosphate (pH 7). The k_cat and K_m values were determined by analyzing β-lactam hydrolysis under initial-rate conditions with a UV spectrophotometer using the Eadie-Hoffstee linearization of the Michaelis-Menten equation as previously described (30). The 50% inhibitory concentrations (IC₅₀s) of clavulanic acid, tazobactam, cefoxitin, NaCl, and imipenem were determined (30).

Nucleotide sequence accession number.

The nucleotide sequence reported in the paper has been submitted to the EMBL/GenBank nucleotide sequence database under accession number HQ700343.

RESULTS

Characteristics of K. pneumoniae and E. cloacae isolates.

This study was initiated by the isolation of one K. pneumoniae isolate and one E. cloacae isolate from two different patients in Buenos Aires, Argentina. K. pneumoniae isolate Kp6299 had been recovered from blood cultures, and E. cloacae isolate Ec2185 had been isolated from a peritoneal fluid specimen. They were both resistant to most β-lactams, including expanded-spectrum cephalosporins (cefotaxime, ceftazidime, aztreonam, and cefepime) (Table 1). Although Kp6299 remained susceptible to all carbapenems, it showed some reduced susceptibility to those molecules. E. cloacae 6299 was resistant to ertapenem (MIC, 2 μg/ml) (Table 1), according to the 2010 updated CLSI guidelines (7). Additionally, both isolates were resistant to fluoroquinolones and aminoglycosides.

Table 1.

MICs of β-lactams for clinical isolates Kp6299, Enc2185, their respective transformants in E. coli TOP10 (Tf6299 and Tf2185), recombinant E. coli TOP10(pOXA-48) and E. coli TOP10(pOXA-163), and recipient strain E. coli TOP10

β-Lactam(s)a	MIC (μg/ml)
	Kp6299	Tf6299	Enc2185	Tf2185	E. coli TOP10(pOXA-163)	E. coli TOP10(pOXA-48)	E. coli TOP10
Amoxicillin	>256	>256	>256	>256	>256	>256	4
Amoxicillin + CLA	>256	>256	>256	>256	>256	>256	4
Ticarcillin	>256	>256	>256	>256	>256	>256	4
Ticarcillin + CLA	>256	>256	>256	>256	>256	>256	4
Piperacillin	>256	64	>256	64	>256	64	2
Piperacillin + TZB	>256	32	>256	32	>256	64	2
Cephalothin	>256	>256	>256	>256	>256	8	4
Cefoxitin	8	4	>256	4	4	4	4
Cefotaxime	32	4	>256	4	3	0.25	0.06
Ceftazidime	256	16	>256	16	16	0.12	0.12
Cefepime	32	1.5	>256	1.5	1	0.25	0.06
Aztreonam	16	2	>256	2	2	0.06	0.06
Imipenem	0.5	0.25	0.5	0.25	0.25	0.5	0.12
Ertapenem	0.38	0.03	2	0.03	0.02	0.25	0.004
Meropenem	0.25	0.03	0.5	0.03	0.02	0.1	0.01

^aCLA, clavulanic acid at a fixed concentration of 4 μg/ml; TZB, tazobactam at a fixed concentration of 4 mg/ml.

β-Lactamase identification.

IEF results gave a series of bands that were difficult to interpret, and the number of β-lactamases produced could not be evaluated properly (data not shown). Synergy tests performed with clavulanic acid and ceftazidime or cefotaxime remained negative for both isolates. However, a peculiar synergy image was observed for both isolates between disks containing imipenem on one side and cefotaxime or aztreonam on the other side, with the activity of the last two compounds being potentiated (data not shown). The absence of a known clavulanic acid-inhibited ESBL gene was confirmed by PCR experiments that gave negative results for all ESBL genes. For both isolates, a positive PCR result was obtained for the bla_TEM gene, and sequencing identified the broad-spectrum β-lactamase TEM-1. Screening for bla_OXA-like encoding genes gave a positive result with the primers specific for the class D carbapenemase bla_OXA-48-like gene in both isolates. Sequencing identified a novel gene differing from the bla_OXA-48 gene by a deletion of 12 bp, in addition to a 1-bp substitution, termed bla_OXA-163 (www.lahey.org/Studies). OXA-163 differed from OXA-48 by a 4-amino-acid deletion (corresponding to Arg214, Ile215, Glu216, and Pro217 in OXA-48, DBL numbering [8]), corresponding to a lack of 12 nucleotides in the bla_OXA-48 gene (Fig. 1). In addition, it had a single amino acid substitution (Ser to Asp) at position DBL 220.

Fig. 1.

Amino acid alignment of the OXA-163 and OXA-48 amino acid sequences. Dashes indicate amino acids identical to those in the OXA-48 sequence. Absence of dashes indicates absence of residues at those positions. Amino acid motifs which are well conserved among class D β-lactamases are shaded in gray. Numbering is according to DBL (8).

Characterization of class D β-lactamase OXA-163.

Comparative MIC determinations using isogenic strain backgrounds for expression of OXA-163 and OXA-48 showed that MIC values of E. coli TOP10(pOXA-163) for broad-spectrum cephalosporins were higher than those of E. coli TOP10(pOXA-48) (Table 1). In contrast, MIC values of imipenem, meropenem, ertapenem, and doripenem were much lower than those obtained for OXA-48, being slightly higher than those of the E. coli TOP10 recipient strain (Table 1). Overall, these results suggest that OXA-163 possessed a weaker carbapenemase activity and a much higher rate of hydrolysis toward broad-spectrum cephalosporins (Table 1).

Biochemical properties of β-lactamase OXA-163.

The purification factor of OXA-163 from culture extracts of E. coli TOP10(pOXA-163) was 51. The kinetic parameters of OXA-163 were linear for all substrates, as previously observed for OXA-48 (29). The kinetics of OXA-163 indicated that it hydrolyzed several broad-spectrum cephalosporins at much higher rates than OXA-48, whereas it hydrolyzed carbapenems very slightly (Table 2). OXA-163 can therefore barely be considered a carbapenemase, taking into account MIC values for carbapenems and kinetic parameters. Cefoxitin and oxacillin were not hydrolyzed by OXA-163. Those data are in accordance with the resistance phenotype observed for the recombinant E. coli strains. Furthermore, studies of inhibition, as measured by determination of IC₅₀s, showed that OXA-163 was weakly inhibited by NaCl (275 mM), whereas OXA-48 was very well inhibited by NaCl (7 mM). Interestingly, OXA-163 was significantly inhibited by clavulanic acid (13.4 μM). Surprisingly, tazobactam (0.75 μM), imipenem (0.10 μM), and cefoxitin (0.25 μM) were efficient inhibitors, whereas this is not the case for OXA-48 (29).

Table 2.

Kinetic parameters of purified β-lactamases OXA-163 and OXA-48^a

Substrate	kcat (s−1)		Km (μM)		kcat/Km (mM−1 · s−1)
	OXA-163	OXA-48b	OXA-163	OXA-48	OXA-163	OXA-48
Benzylpenicillin	15	245	40	40	380	6,000
Ampicillin	25	955	320	395	80	2,500
Ticarcillin	1	45	320	55	3	820
Piperacillin	8	75	70	410	110	180
Cephalothin	3	44	80	195	40	230
Cefotaxime	30	>9	850	>900	35	10
Ceftazidime	200	NHc	>2,000	NH	1,000	NH
Cefepime	2	>0.6	350	>550	6	1
Oxacillin	ND	130	ND	95	NDd	1,400
Aztreonam	0.5	NH	230	NH	2	NH
Imipenem	0.03	4.8	530	13	0.05	370
Meropenem	0.07	0.07	2,200	11	0.03	6
Ertapenem	0.01	0.13	150	100	0.1	1

^aData are the means of three independent experiments. Standard deviations were within 15% of the means.

^bPreviously published values by Docquier et al. (12).

^cNH, no detectable hydrolysis (<0.01 s⁻¹), for a maximum amount of 5 μg of purified enzyme and up to 200 nmol of substrate.

^dND, not determined.

Genetic support of bla_OXA-163.

Using isolate Enc2185 as the donor, conjugation experiments allowed E. coli J53 transconjugants to be obtained. These transconjugants exhibited reduced susceptibility toward expanded-spectrum cephalosporins but remained fully susceptible to carbapenems (Table 1) and were found to contain a single 85-kb plasmid harboring the bla_OXA-163 gene (data not shown). In addition, an E. coli transconjugant expressing a broad-spectrum β-lactamase that possessed a 150-kb plasmid harboring the bla_TEM-1 gene was obtained (data not shown).

Mating-out assays using isolate Kp6299 as the donor failed, and transformation experiments were therefore performed using plasmid DNA from Kp6299 and also Enc2185 as a control. E. coli TOP10 transformants exhibiting resistance to expanded-spectrum cephalosporins but remaining susceptible to carbapenems were obtained for both isolates (Tf6299 and Tf2185) (Table 1). The plasmid carrying the bla_OXA-163 gene in isolate Kp6299 was 70 kb in size (Table 1; data not shown). Both OXA-163-producing E. coli transformants obtained from Enc2185 and Kp6299, despite possessing plasmids differing in size, exhibited the same additional markers of resistance to kanamycin and tobramycin.

Attempts to identify the incompatibility group of both bla_OXA-163-carrying plasmids (70 kb and 85 kb) failed using the PBRT method (4). Furthermore, a negative result was obtained by PCR using primers specific for the RepP gene, featuring the 70-kb bla_OXA-48-carrying plasmid described previously (5, 6).

Genetic environment of bla_OXA-163 gene.

PCR mapping based on the known structures encompassing the bla_OXA-48 gene was unsuccessful. Neither the IS1999 insertion sequence nor the tir gene, previously identified to occur in association with the bla_OXA-48 gene, was identified. Sequence analysis of recombinant plasmids pBK-6299 and pBK-2185 identified 8.7-kb and 5.7-kb inserts, respectively (Fig. 2). The close upstream and downstream environments of the bla_OXA-163 gene were almost identical for both recombinant plasmids. The ISEcl4 element was identified upstream of bla_OXA-163. ISEcl4 is an isoform of insertion sequence IS4321, and their respective transposases share 93% amino acid identity; therefore, it belongs to the IS110 family and the IS1111 group. ISEcl4 had truncated a short fragment of a Tn3-type transposase and inserted itself within a DNA fragment originating from Shewanella spp. Downstream of the bla_OXA-163 gene, an IS4-like insertion sequence whose transposase shared 52% amino acid identity with IScrO6 from Citrobacter rhodentium was identified. In pBK-2185, the latter insertion sequence was truncated by another IS6100 element. Further upstream of IS4321 was a truncated IS sharing 60% amino acid identity with IShfr9 from Shewanella frigidimarina, and this truncated IS was interrupted upstream by the presence of another insertion sequence whose transposase shared 57% amino acid identity with ISXc5 from Xanthomonas campestris (Fig. 2).

Fig. 2.

(A) Schematic representation of genetic environment of the bla_OXA-163 gene in isolates Enc2185 and Kp6299; (B) schematic representation of the genetic environment of the bla_OXA-48 gene. The coding regions are shown as horizontal boxes, with an arrow indicating the orientation of transcription. Restriction sites are indicated. Inverted repeats are indicated as vertical boxes. Vertical dashed lines indicate identity between recombinant plasmids p2185 and p6299. Black triangles stand for terminal site duplication after Tn1999 insertion.

Further analysis of the pBK6299 insert showed that a truncated orf4b of unknown function previously described in E. coli was identified on the left end of IS4321. Further upstream, an open reading frame (ORF) encoding a macrolide-phosphotransferase K found in K. pneumoniae was identified and was preceded by another IS6-like insertion sequence previously described in Proteus vulgaris. Downstream of the bla_OXA-163 gene, the IS4-like element was identified, interrupting an ORF of unknown function previously identified in Vibrio tapetis (Fig. 2). Overall, those structures were mosaic, likely as a result of consecutive and unrelated transposition and recombination events.

In silico analysis of the DNA sequences located upstream of the bla_OXA-163 gene identified sequences typical of −10 (TAATAT) and −35 (TAGAAG) enterobacterial promoter regions, separated by an optimal 17-bp spacing. Noteworthy, these promoter sequences likely involved in bla_OXA-163 gene expression were not part of any IS element but, rather, corresponded to the original promoter of the bla_OXA-163 gene, whereas bla_OXA-48 gene expression was demonstrated to be under the control of a promoter provided by IS1999 (29).

DISCUSSION

In this work, we characterized OXA-163, a novel OXA-48 variant identified from two enterobacterial isolates from Argentina. OXA-163 is the first representative of a novel subclass of extended-spectrum oxacillinases that shows a weaker ability to hydrolyze carbapenems than OXA-48. Conversely, expanded-spectrum cephalosporins, and particularly ceftazidime, are significantly hydrolyzed by OXA-163, whereas OXA-48 does not hydrolyze ceftazidime.

According to their definition, oxacillinases were first described to be β-lactamases hydrolyzing oxacillin faster than benzylpenicillin (31). These β-lactamases are usually poorly inhibited by clavulanic acid and tazobactam, whereas their activity may be inhibited in vitro by NaCl. Interestingly, OXA-163 does not share all these characteristics and exhibits peculiar hydrolytic features. In contrast to OXA-48, it is efficiently inactivated by tazobactam, cefoxitin, and imipenem but is inactivated less so by clavulanic acid. In addition, the activity of OXA-163 is not significantly inhibited by NaCl, in contrast to that of most class D β-lactamases, including OXA-48. OXA-163 does not hydrolyze oxacillin, whereas OXA-48 does, but it possesses the ability to hydrolyze cefotaxime, ceftriaxone, and ceftazidime and, to a lesser extent, cefepime and aztreonam. Actually, the hydrolytic properties of OXA-163 make it more similar to an Ambler class A extended-spectrum β-lactamase than to a class D β-lactamase, as previously observed with another class D β-lactamase, OXA-18 (24).

Extension of the substrate profiles toward broad-spectrum cephalosporins of several class D β-lactamases has been previously reported, as for OXA-32 deriving from OXA-2 (26) or for OXA-28 deriving from OXA-10 (27), but none of those variants possessed any carbapenemase activity. Also, and with the exceptions of OXA-18 (24) and OXA-45 (35), which are significantly inhibited by clavulanic acid, OXA-163 is one of the rare examples of a class D β-lactamase susceptible to inhibitor activity.

This study further emphasizes the diversity of class D β-lactamases. Detailed sequence analysis of OXA-163 revealed a lack of 4 amino acids compared to the sequence of OXA-48, corresponding to positions DBL 222 to DBL 225 and amino acids 214 to 217 of the OXA-48 sequence, which are part of or close to the β5 strand, as determined by crystallography by Docquier et al. (12). Molecular dynamic simulations predicted Thr213 and Arg214 to be crucial, since they are part of a cavity at the origin of meropenem hydrolysis. Arg214 was actually shown to be a key element of the OXA-48 active site (12). Therefore, the lack of Arg214 in OXA-163 may significantly disturb the conformation of that site, thus compromising its interactions with carbapenems. However, it is difficult to speculate about the modifications involved in the hydrolysis spectrum extension toward broad-spectrum cephalosporins.

In both isolates, the bla_OXA-163 gene was identified on plasmids, differing in size, that did not possess a backbone similar to that of the bla_OXA-48-bearing plasmids. Even though the two genes are closely related, their genetic environments are different. Indeed, while bla_OXA-48 is part of a Tn1999-like transposon, bla_OXA-163 was surrounded here by mosaic structures made of insertion sequences and truncated mobile elements. Those observations clearly suggest that the mobilization processes leading to the dissemination of both genes were different. Considering that Shewanella oneidensis was identified as the progenitor of bla_OXA-48-like genes (28), it might be hypothesized that two distinct mobilization events from two distinct Shewanella species had occurred.

This study identified a novel OXA variant that possesses unique hydrolysis properties that overall fit with those of a class A ESBL. That feature may suggest that this gene could have disseminated without being recognized as such. Although the bla_OXA-48 gene is increasingly identified in the countries located in the southern and eastern parts of the Mediterranean Sea, the bla_OXA-163 gene may correspond to its equivalent but in South America. Furthers studies will evaluate its prevalence in this part of the world.