Early OXA-48-Producing Enterobacterales Isolates Recovered in a Spanish Hospital Reveal a Complex Introduction Dominated by Sequence Type 11 (ST11) and ST405 Klebsiella pneumoniae Clones.

We present results of microbiological analysis of the first Enterobacterales isolates that were isolated in 2012 in our institution expressing OXA-48 carbapenemase. This enzyme confers resistance to carbapenems, an important group of antibiotics widely used in the hospitals. OXA-48 carbapenemase is currently present in many parts of the world, but it is found particularly frequently in the Mediterranean area. It was disseminated at the Ramón y Cajal Hospital and found to be associated with a particular Klebsiella pneumoniae strain, so-called high-risk clone ST11, which was previously found in our institution in association with other enzymes such as CTX-M-15, VIM-1, and KPC-3. This clone might have acquired a plasmid harboring the blaOXA-48 gene. Our results point out the importance of local epidemiology in the dissemination and maintenance of multidrug-resistant bacteria.

harboring the bla OXA-48 gene. Our results point out the importance of local epidemiology in the dissemination and maintenance of multidrug-resistant bacteria. KEYWORDS OXA-48, carbapenemase, VIM-1, CTX-M-15, Enterobacterales, ST11 C arbapenemase-producing Enterobacterales (CPE) isolates have become an important public health concern worldwide. They have been increasingly reported in Europe, particularly in recent years, with current classification of Spain as a country with "interregional spread" of CPE according to the European Centre for Disease Prevention and Control (ECDC) (1)(2)(3)(4)(5)(6).
In our hospital in Madrid (Spain), VIM enzymes were first detected in 2005 (1), KPC enzymes in 2009 (12,13), and OXA-48 enzymes in 2012. It is of note that large OXA-48 outbreaks were described in other hospitals in Madrid (2) and in other Spanish regions (3,14,15) before the emergence in our center. Epidemiology of CPE has been investigated not only locally but also at a national level in the Spanish reference center (16,17). The aim of this work was to characterize the microbiological and epidemiological scenario, through classical and genomic approaches, represented by the first OXA-48-producing Enterobacterales recovered in our hospital (2012 to 2013), where infections by strains producing other carbapenemases (mainly VIM and KPC) had been endemic for several years.

RESULTS
Patient characteristics and carbapenemase characterization. A total of 57 patients (33 males) were infected (n ϭ 41) or colonized (n ϭ 16) with OXA-48-producing Enterobacterales during the studied period. The majority of patients were elderly (median age, 69 years; range, 18 to 92 years). In this period, the prevalence of carbapenemase in our institution was 0.4% and the predominant carbapenemase was OXA-48. Patients were admitted to medical (n ϭ 20; 4%) and surgical (n ϭ 21; 4%) wards and intensive care units (ICU) (n ϭ 4; 7%). It should be noted that 13 isolates were recovered from 13 nonhospitalized patients at the time of sampling, although 5 of them had been previously admitted to our institution ( Table 1). The first OXA-48producing Enterobacterales isolates recovered in our institution were obtained from patients admitted to the urology ward in 2012. In that ward, most of the K. pneumoniae isolates belonged to the same clone (KP-A) identified as sequence type 11 (ST11). Overall, from the 57 patients, we recovered 71 isolates (61 K. pneumoniae, 5 Escherichia coli, 2 Klebsiella aerogenes, and 1 each of Klebsiella oxytoca, Enterobacter cloacae, and Citrobacter amalonaticus). Apart from the OXA-48 enzyme, 51 isolates (48 K. pneumoniae, 1 K. oxytoca, 1 K. aerogenes, and 1 C. amalonaticus) coproduced CTX-M-15, 6 isolates (4 K. pneumoniae and 2 E. coli) coproduced VIM-1 and CTX-M-15, and 1 E. coli isolate also coproduced VIM-1. It is of note that the double-disk synergy test (using EDTA) was negative in 5 of 7 isolates harboring VIM-1.
Transferability and location of bla carbapenemase genes and plasmid typing. The bla OXA-48 gene was located on plasmids (50 to 70 kb), most of them being transferable by conjugation (59/71 isolates). They carried bla OXA-48 , often with bla VIM-1 (7/71, 9.8%) and/or bla CTX-M-15 (57/71, 80.3%). OXA-48-carrying plasmids were categorized as IncL/M according to the PCR-based replicon typing (PBRT) scheme (18). In addition, 2 of 71 isolates presented a larger (ca. 220-kb) plasmid nonrelated to bla genes and typed as IncFIIk by the scheme described previously by Villa et al. (19) ( Table 1). Restriction fragment length polymorphism (RFLP) analysis for OXA-48 plasmids revealed a similar band pattern (Fig. S1). These results indicate a similar backbone of the OXA-48 encoding plasmids, a finding also supported by the results produced by PCR and sequencing of their repA, traU, and parA genes (Table 1). More than 98% identity among all these plasmids and with the pOXA-  (CDS; GC%, 50.8%) and containing two bla genes (bla OXA-48 and bla CTX-M-15 ) as the only antibiotic resistance markers. The bla OXA-48 gene was part of Tn1999.2 inserted at the tir gene, responsible for inhibition of transfer of the plasmid, whereas the bla CTX-M-15 gene was located downstream of ISEcp1 (Fig. 2). The resistance region of the plasmid (including both bla resistance determinants) was delimited by two insertion sequences: ISEcp1 was found to be associated with the bla CTX-M-15 gene and an IS10A-like sequence located upstream of bla OXA-48 (Fig. 2). In silico comparisons with other OXA-48encoding-plasmids revealed a very similar structure (Ͼ99% similarity) with the same backbone as that found in the IncL-OXA-48 plasmids (Fig. 3) (GenBank accession no. JN626286, KC335143, KC354801, and NC_021502).

DISCUSSION
Spread of CPE has been increasingly reported worldwide since first description more than 20 years ago, with predominance of metallo-beta-lactamase (MBL) producers in Asia and Europe, KPC producers in the United States, and OXA-48 producers in the Mediterranean countries (6,9,21,22). These carbapenemases have been increasingly recovered in Spain, and OXA-48 isolates are currently the most prevalent ones (3,6). In our study, the prevalence of carbapenemase in our institution was 0.4%, and the most prevalent carbapenemase was OXA-48 (50.3%). The carbapenemase situation in our institution, which started with VIM producers in 2005 and KPC producers in 2009 (1, 12, 13), can be observed in Fig. S2. In this article, we characterize the early isolates of OXA-48-producing Enterobacterales collected in our hospital; currently, the situation in our institution is one of OXA-48 infection endemicity (11). Due to the difficulties in the detection of OXA-48 carbapenemase producers by the use of carbapenem susceptibility results as a screening test, we implemented measurement of increases in temocillin MIC as a surrogate marker to detect its presence in CarbaNP test-positive isolates with a negative result for MBL or KPC enzymes (23,24). Nevertheless, we recovered isolates with decreased susceptibility to carbapenemase to perform specific PCR for OXA-48, and results were negative (data not shown).
Isolation of OXA-48-producing Enterobacterales in our hospital started in the urology ward in 2012 (Fig. S2). In that ward, most of the K. pneumoniae isolates belonged to the same clone (KP-A) which corresponded to ST11. Interestingly, this ST was previously found in our institution in 2010 in a patient admitted to general surgery who was infected with a KPC-3-producing K. pneumoniae strain and in 2011 in 2 patients admitted to the neurosurgery ICU who were colonized with a VIM-1-producing K. pneumoniae strain. These isolates belonged to the same ST and showed closely related PFGE patterns, suggesting that the circulation of a clonal local pool in our institution plays an important role in the emergence and spread of new resistant variants, as happened with the OXA-48-producing clone. Later, the same clone was detected in other wards in addition to ST405 K. pneumoniae. The latter was first identified at the nearby Hospital La Paz (500 m from our hospital) where such infections had become endemic (2). However, in spite of the close proximity of the two hospitals, these ST405 isolates were found to have different PFGE patterns (data not shown), suggesting independent introduction events. In a study published in 2017 by the national reference laboratory that included data representing K. pneumoniae isolates collected in different Spanish areas, both sequence types were found among the most prevalent ones, highlighting their persistence over time in our country (25).
It should be noted that 7 isolates (4 K. pneumoniae and 3 E. coli) coproduced VIM-1 and that 6 of them (4 K. pneumoniae and 2 E. coli) also coproduced CTX-M-15. The presence of more than one bla gene in these isolates can be explained by the carbapenemase endemic situation involving bla VIM , bla KPC and bla CTX-M-15 genes in our hospital (1,11,13). The presence of both bla VIM-1 and bla OXA-48 enzymes in the same isolate enhances the difficulty of phenotypically detection of OXA-48 producers. We believe that in these cases, the double-disk synergy test performed with EDTA is not entirely reliable to identify VIM-1 producers as in our study we had five out of seven false-negative results.
To our knowledge, there have been only limited previously published descriptions of the coexistence of OXA-48 and VIM-1 (31-34), but the availability of such reports might increase in the future. Interestingly, we observed that bla OXA-48 and bla VIM-1 genes, mostly associated with K. pneumoniae ST11, were located on the same plasmid. The plasmids detected in our study (among different PFGE patterns) showed a high degree of similarity (Ͼ99%) with previously described OXA-48 plasmids, even with those also encoding bla VIM-1 and bla CTX-M- 15 . This indicates that the spread of genes encoding this carbapenemase, which are often linked with other beta-lactamase genes, is associated not only with the appearance of highly adapted clones in the hospital but also with the dissemination of highly transmissible plasmids. The presence of these plasmids in the hospital setting allows acquisition and rapid spread of new resistance determinants such as bla  . The characterized IncL/M plasmid is currently present in our institution (31).
In conclusion, our data showed rapid penetration and spread of bla OXA-48 genes in multiresistant clones of Enterobacterales after its emergence in our institution. The globally spread ST11 K. pneumoniae clone, previously associated in our hospital with CTX-M-15, VIM-1, and KPC-3 enzymes, is now also present coproducing OXA-48 (11). This scenario reinforces the idea of the contribution of high-risk clones in the dissemination and persistence of antibiotic resistance genes. Moreover, emergence of multiresistant ST11 among nonhospitalized patients highlights the difficulties in preventing dissemination of such strains into the community.

MATERIALS AND METHODS
Bacterial isolates. Over a 21-month study period (March 2012 to December 2013), all isolates (n ϭ 71) recovered from clinical specimens (27 urine samples, 9 wounds, 8 sterile fluids, 5 respiratory samples, 2 blood cultures) and surveillance cultures (20 rectal swabs) suspected of carbapenemase production but negative for the phenotypic expression of classes A and B carbapenemases (see below) were investigated for OXA-48-like-producing enzymes. Patients´demographic characteristics were also reviewed. The study was approved by the ethical committee of our institution (reference 251-13). For clonal comparison, we also included in the analysis the CPE clinical isolates associated with contemporary outbreaks due to strains producing MBL or KPC in our institution (1,12,13) and to OXA-48 producers collected in a nearby hospital (Hospital La Paz, 500 m from our hospital) (2).
Bacterial identification, susceptibility testing, and phenotypic assays. Species identification and antibiotic susceptibility testing were performed using matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) (Bruker Daltonics, Leipzig, Germany) and the MicroScan automated system (Beckman Coulter, Brea, CA), respectively. MICs were interpreted using EUCAST criteria (www.eucast.org). Screening for the presence of carbapenemase production was performed with CarbaNP (23). MBL production was investigated by a double-disk synergy test using EDTA plus meropenem and ceftazidime disks as previously described (1,35). KPC production was inferred by an increase of the inhibition zone of an unsupplemented meropenem disk compared with that of a meropenem disk supplemented with 0.3 mg of boronic acid (36). In addition, OXA-48 production was phenotypically inferred by the lack of the inhibition zone of the temocillin disk (37).
Transferability and location of bla carbapenemase genes and plasmid typing. Mating experiments were performed using E. coli strain BM21 (nalidixic acid and rifampin resistant, lactose fermentation positive, and plasmid free) as the recipient (1). The plasmid content (number and size) of each strain was analyzed by S1-digested genomic DNA PFGE, the bla gene location being detected by standard hybridization procedures (40,41). Transconjugants were selected on Luria-Bertani agar plates containing imipenem (0.5 g/ml) and rifampin (100 g/ml) and were incubated at 37°C for 24 h. Plasmid incompatibility groups were inferred by PCR typing schemes based on the presence of genes encoding replication initiator proteins (RIP) (18,19). Further characterization of OXA-48 gene-carrying plasmids was performed for all K. pneumoniae PFGE types by comparison of restriction fragment length polymorphism (RFLP) patterns of DraI-digested plasmid DNA. To further characterize the OXA-48 IncL/M plasmids, the genes associated with replication (repA), conjugation (traU), and maintenance (parA) were amplified and further sequenced (20). ST11 whole-genome sequencing and plasmid characterization. K. pneumoniae ST11 strains associated with production of either VIM-1 (31) or KPC (13) have been prevalent in our hospital since their first detection in 2005. Due to the high transmissibility and persistence of this clone (11,42), we sequenced one ST11 K. pneumoniae isolate, which was selected based on its ESBL-positive phenotype (this isolate coproduced both OXA-48 and CTX-M-15). Whole-genome sequencing (WGS) was performed using Illumina (280ϫ coverage), and DNA fragments were subjected to de novo assembly with Velvet and Spades (43). In addition, the first ST11-OXA-48 isolate found in our hospital (ryc_12106801) was also sequenced and assembled using PacBio and RS_HGAP Assembly 2, respectively. Combining the sequences obtained by Illumina and PacBio, and using SPAdes, the plasmid of this isolate was closed (44)(45)(46)(47). The in silico plasmid typing was carried out using the PlasmidFinder tool (48). This plasmid was compared with other indexed plasmids at the NCBI database (GenBank accession numbers JN626286.1, KC335143.1, KC354801.1, and NC_021502.1) using BRIG (49) (http://brig.sourceforge.net/) and MAUVE (http://darlinglab.org/mauve/mauve.html).
Data availability. Data representing this Whole Genomes Shotgun project have been deposited at DDBJ/ENA/GenBank under accession number VILG00000000. The version described in this paper is version VILG01000000.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.