Spread of bla_CTX-M-14 Is Driven Mainly by IncK Plasmids Disseminated among Escherichia coli Phylogroups A, B1, and D in Spain^▿

Bacterial strains.

We studied 61 CTX-M-14-producing E. coli isolates from 56 individuals living in a suburban area of Madrid, Spain, from 2000 to 2005. They were mainly community-based persons (n = 40 [71.4%]) comprising 2 healthy volunteers without previous antibiotic or hospital exposure (53), 3 cohabitants of patients with UTIs caused by CTX-M-14 recovered within 2 weeks of the case detection (54), and 35 outpatients. The hospitalized patients were located in medical wards (n = 13 [23.2%]), intensive care units (ICUs; n = 2 [3.6%]), or surgical wards (n = 1 [1.8%]) (Table 1). The sample represents all CTX-M-14-producing isolates recovered in the Hospital Universitario Ramón y Cajal (39 from urine, 1 from blood, 1 from wound, and 2 from feces) and fecal isolates from two surveillance studies performed in the same city during the same time period (53, 54).

One isolate per patient (or more than one per patient, if different antibiotic resistance patterns were detected in isolates from the same sample or if they showed different PFGE patterns) was selected for further analysis. Bacterial identification and antibiotic susceptibility testing were performed with the semiautomated Wider system (Fco. Soria Melguizo, Madrid, Spain). Analysis of the susceptibilities of the isolates to non-ß-lactam antibiotics was performed by the disk diffusion method according to CLSI guidelines (11). The CLSI breakpoints for netilmicin were used for apramycin. Strains showing intermediate susceptibility were considered resistant.

Clonal relatedness.

Clonal relatedness was established by pulsed-field gel electrophoresis (PFGE) of XbaI-digested genomic DNA in a CHEF-DRIII device (Bio-Rad, La Jolla, CA) by using electrophoresis conditions of 14°C, 6 V/cm², 10 to 40 s, and 24 h; standard procedures and criteria were followed (41). The assignation of E. coli phylogenetic groups was carried out by the multiplex PCR assay previously described by Clermont et al. (10). Representative E. coli isolates mainly belonging to phylogroups A and B1 were further characterized by multilocus sequence typing (MLST) by using the standard seven housekeeping loci (http://www.mlst.net). The fumC genes of E. coli isolates belonging to the D phylogroup were analyzed for the presence of a single nucleotide polymorphism (C288T), which is a specific marker for E. coli “clonal group A.” This clone is disseminated worldwide, it belongs to sequence type 69 (ST69), and it is responsible for community UTIs (50).

MLST data were analyzed by using the eBURST program (version 3; http://eburst.mlst.net), which assesses the relationship within clonal complexes. The entire E. coli MLST database (2,255 isolates and 1,065 STs; last update, June 2009) was considered.

Transferability and location of bla_CTX-M-14 gene.

Transfer was tested by broth and filter mating assays with E. coli K-12 strain BM21R (nalidixic acid and rifampin [rifampicin] resistant, lactose fermentation positive, and plasmid free) as the recipient for 6 to 12 h at 30°C and 37°C. Transconjugants were selected on Luria-Bertani agar plates supplemented with cefotaxime (2 mg/liter) and rifampin (100 mg/liter) and were incubated at 37°C for 24 h.

The chromosomal and/or plasmid location of bla_CTX-M-14 genes was assessed by the hybridization of I-CeuI-digested genomic DNA with bla_CTX-M-14 and 16S rRNA gene probes by using the following electrophoresis conditions: 5 to 25 s for 23 h and 60 to 120 s for 10 h, 14°C, and 6 V/cm² (41). Transfer and hybridization were performed by standard procedures. Labeling and detection were carried out with an AlkPhos commercial kit, according to the manufacturer's instructions (Amersham Life Sciences, Uppsala, Sweden).

Plasmid characterization.

Plasmid DNA was obtained by using a QIAgen plasmid midikit (Qiagen, Hilden, Germany). Plasmids were preliminarily classified according to their incompatibility group by using the PCR-based replicon-typing (PBRT) scheme described by Carattoli et al. (8). Plasmid size and content were determined by hybridization of S1 nuclease-digested genomic DNA (41) with probes specific for bla_CTX-M-14. Confirmation of the replicon content was also undertaken by hybridization of PstI-digested plasmid DNA with probes specific for both bla_CTX-M-14 and rep amplified by PCR from different samples. The relationship among plasmids was determined by comparison of their DNA patterns generated after digestion with the EcoRI, PstI, and HpaI enzymes and electrophoresis in 1% agarose at 100 V for 4 h.

Plasmids preliminarily classified as members of the IncI complex (those that yielded an amplification product by using primers specific for incompatibility group I (IncI), IncK, and/or IncB/O by the PBRT method) were further studied (8). We analyzed the sequences corresponding to RNA interference (RNAi; also named inc, which encodes a segment of RNA of approximately 70 bp that regulates replication by repressing the expression of the replicase gene) and also to the replication and the relaxase proteins of a subset of representative isolates. Identification of RNAi was performed by analysis of the amplicons obtained by the PBRT method. Identification of the replication proteins was carried out by analysis of the PCR products obtained with primers RNAI (5′-ATGCGCCATAAGGCATTCA-3′) and REPZA (5′-AGTCGC TTCAGATGGTCAT-3′) and an amplification protocol consisting of 1 cycle at 94°C for 5 min and 35 cycles at 94°C for 1 min, 54°C for 1 min, and 72°C for 2 min, followed by a final step at 72°C for 10 min. Identification of plasmid relaxases was performed by PCR with primers MOB_P12 forward (5′-GCAAAAGATGACACTGAYCCYGTTTT-3′) and MOB_P12 reverse (5′-AGCGATGTGGATGTGAAGGTTATCHGTRTC-3′) and conditions of 30 cycles of 94°C for 30 s, 55°C for 30 s, and 72°C for 30 s. The amplified DNA segments were then sequenced in both orientations (1, 19).

Analysis of bla_CTX-M-14 genetic environment.

As bla_CTX-M-14 has been found to be associated with different genetic environments, we tested for the presence of sequences associated with known genetic platforms: ISEcp1, IS903, intI1, and ISCR1 (2, 17, 39). Oligonucleotides corresponding to different regions of ISEcp1 and IS903 were used in order to detect elements containing ΔISEcp1 and/or ΔIS903, as described previously (17). The link of class 1 integrons containing bla_CTX-M-14 to Tn21-like transposons was screened for by PCR mapping with the primers and conditions previously described (41).

Nucleotide sequence accession numbers.

The sequences corresponding to the plasmids described in this paper were assigned the indicated GenBank (NCBI) accession numbers: pRYC105, GQ892052; pRYC106, GQ892053; pRYC109, GQ892054; pRYC110, GQ892050 and GQ892051.

E. coli isolates producing CTX-M-14 belong to phylogroups A, B1, and D and are mostly associated with UTIs.

The 61 CTX-M-14-producing isolates included in this study were classified into 60 different PFGE patterns and were similarly distributed among phylogenetic groups A, D, and B1 (22, 21, and 18 isolates, respectively, representing 36.1%, 34.4%, and 29.5% of the isolates included in the study, respectively). Most of the CTX-M-14-producing E. coli strains of phylogroups A and B1 belonged to particular clonal complexes; phylogroup A isolates clustered in the ST10 complex (12/18 of the isolates analyzed) and also in the ST168 complex (2/18) or the ST23 complex (2/18), while phylogroup B1 isolates grouped either in the ST359 complex (7/12) or the ST155 complex (5/12). Analysis of the fumC sequences from E. coli isolates belonging to phylogroup D showed a high degree of diversity (Table 1). One strain with the C288T single polymorphism specific for uropathogenic E. coli (UPEC) clonal group A (CGA) was identified as ST106, which fits within the ST69 clonal complex. The other isolates studied corresponded to ST59, ST117, ST393, ST624, ST770, and ST1019 (Fig. 1; Table 1).

The isolates studied were frequently associated with UTIs (n = 39, 64%) and were resistant to nalidixic acid (73.8%; 45/61), tetracycline (72.1%; 44/61), sulfonamides (60.6%; 37/61), trimethoprim (59.0%; 36/61), and streptomycin (57.3%; 35/61) and were less frequently resistant to chloramphenicol (36.1%; 22/61), ciprofloxacin (21.3%; 13/61), spectinomycin (19.7.3%; 12/61), tobramycin (9.8%; 6/61), apramycin (8.2%; 5/61), and gentamicin or amikacin (4.9% each; 5/61). A few isolates of phylogroups A (n = 6), B1 (n = 5), and D (n = 7) exhibited the multiresistant phenotype ACSSuTTp (indicating resistance to ampicillin, chloramphenicol, streptomycin, sulfonamides, tetracycline, and trimethoprim, respectively), which has recently been associated with a genetic island that comprises genes encoding antibiotic resistance and virulence in particular E. coli clones causing UTIs (UPEC strains) (30). Interestingly, phylogroup A and D isolates showing the ACSSuTTp phenotype were also resistant to fluoroquinolones (data not shown).

Our collection contained a large number of community E. coli isolates of phylogroups A and B1, besides those of group D, unlike the population structure of CTX-M-14-producing isolates from other countries, which mainly comprises phylogroups D and B2 (23, 46, 47, 49). It is of note that the majority of the UPEC isolates also belonged to phylogenetic groups B2 and D, with some E. coli isolates recently recognized as being globally disseminated UPEC clones, such as ST131, ST69, ST14, ST73, and ST95 (6, 28, 50). Even though groups A and B1 of E. coli have rarely been involved in extraintestinal infections, they are prevalent as commensal organisms in both animals and humans and are frequently found to be enterotoxigenic, enterohemorrhagic, and, sometimes, enteroaggregative, with each pathotype possessing a unique combination of virulence traits eventually linked to particular successful clones (18, 24, 52, 55). The E. coli ST10 complex is overrepresented among isolates of phylogroup A, being frequently identified among fecal isolates from healthy people, enterotoxigenic strains from community-based persons, and more recently, ESBL producers causing UTIs (29, 44, 52, 55; this study). Similarly, some group B1 E. coli clones are increasingly found to be ST155 or ST359 and have recently been isolated from Spanish, Portuguese, and Brazilian outpatients with UTIs (37, 55; unpublished results). These group A and B1 clones might reflect new UPEC strains that have emerged. These strains are independently associated with the presence of bla_CTX-M-14 and may influence the successful spread of certain ESBL genes. Their high degree of epidemicity is also reflected by their high rate of transmission among household contacts of patients with UTIs (54). The selection and spread of particular STs within these phylogroups might be influenced by the acquisition of specific genetic traits that enhance their colonization ability (24, 30, 50).

The bla_CTX-M-14 gene is located on IncK and IncHI2 plasmids.

The bla_CTX-M-14 gene was located on plasmids exhibiting various transfer properties (bla_CTX-M-14 was transferred by conjugation in 75.4% of the isolates) and hybridizing with probes specific for narrow-host-range incompatibility group IncI and/or IncK (94.9%; 37/39) and to a much lesser extent IncHI2 (5.1%; 2/39).

Plasmids of the IncI complex.

Twenty-seven of the 29 plasmids characterized were similar in size (80 to 90 kb), could be amplified with primers included in the PBRT assay specific for both the IncI and the IncK groups (but rarely with primers specific for the IncB/O group), and hybridized with probes specific for the inc regions of these plasmid groups. They were classified into four types on the basis of their restriction fragment length polymorphism patterns, with three of them (pRYC105, pRYC108, and pRYC109) showing very similar profiles that differed by only a few bands (n = 1 to 3). pRYC105 was the predominant plasmid variant and was recovered from inpatients, outpatients, and healthy volunteers from 2000 to 2005. Nowadays, it is still detected in the Hospital Universitario Ramón y Cajal and other geographical areas (data not shown). pRYC105, pRYC108, and pRYC109 showed 99% to 100% sequence homology to inc_R387, RepA_pSERB1/R387, and NikB_pSERB1/R387 (GenBank accession no. M93063 for plasmid R387 and GenBank accession no. AY686591 for plasmid pSERB1) and were classified as plasmids of the IncK group. The entire 1.2-kb replication region of pRYC105 (inc, repY, repA/repZ, ori) from different isolates was sequenced and showed 99% sequence homology at the nucleotide level with the nucleotide sequence of R387. Nevertheless, analysis of the sequence corresponding to the PCR product obtained with IncI-specific primers showed 99 to 100% homology with the RNAi regions of plasmids pTN38148, pSL476_91, and pCVM29188 (GenBank accession nos. NC_011514, NC_011077, and CP01118, respectively), which indeed belong to the IncI group (Fig. 2 and 3). The probes generated from the PCR products corresponding to inc_R387/pSERB1 and inc_pSL476_91 hybridized with a single band of PstI-digested plasmid DNA from different isolates. Although cross hybridization can occur when these probes are used (13), the presence of two different sequences identical to the inc regions of plasmids belonging to the IncI and IncK groups was demonstrated in the isolates analyzed, indicating the presence of two different inc sequences located close together.

The fourth type, pRYC106, showed a different restriction fragment length polymorphism pattern with sequences identical to inc_pSL476_91, RepA_pSL476_91, and NikB_pSERB1 (GenBank accession no. CP001118 for plasmid pSL476_91 and GenBank accession no. AY686591 for plasmid pSERB1; Fig. 2) and it was considered a mosaic plasmid of the IncI complex. Despite the multiresistance phenotypes of the E. coli hosts, transconjugants containing these plasmids conferred resistance only to cephalosporins (Table 2). The results of the analysis of their DNA sequences corresponding to the RNA antisense molecules involved in the control of replication, the replicases, and the relaxases are shown in Fig. 2 and 3.

The plasmid analytic approach used in this work permits not only the discrimination between IncI and IncK plasmids but also allows the detection of recombinant plasmids containing either the replication or the conjugation systems of these groups, which seem to be frequent in nature. Some examples are recently described plasmids pSC138 (GenBank accession no. NC_006856) and pO113 (GenBank accession no. NC_007365), which contain only conjugation systems related to the IncI group; pCooKm (GenBank accession no. CR942285), which contains a replication region and a conjugative region related to those of the incompatibility plasmid groups I and F, respectively; and p026-Vir (GenBank accession no. NC_012487.1), which contains two replicases homologous to different IncI plasmids and two relaxases highly similar to those of IncK plasmids. Even though the presence of two inc sequences that are located close together has not been observed among plasmids of the IncI complex, plasmids of incompatibility groups F and N may contain two inc/cop sequences within the same replication region (14, 42). It is worthwhile to highlight the stability of the pRYC105 fingerprints among different E. coli lineages over time, which suggests a high degree of adaptation to specific genetic backgrounds. However, the wide distribution of pRYC105 among E. coli isolates from community-based persons (53, 54) and food-borne animals (31) is of concern since it creates a reservoir of highly transmissible plasmids in heterogeneous E. coli populations that colonize different environments which may be able to evolve toward more virulent variants by recombination, as seems to have occurred with p026-Vir (GenBank accession no. NC_012487.1) and pCooKm (GenBank accession no. CR942285).

IncK plasmids have been isolated from patients with UTIs since the early 1970s (1a, 16, 21, 51). A type IV secretion system that contributes both to plasmid conjugation and to adherence to epithelial cells and abiotic surfaces identified in IncK plasmids such as pSERB1 (GenBank accession no. AY686591) (16) may facilitate the ability of E. coli to colonize the intestine and, consequently, enhance the pathogenic profile of specific clones or clonal groups (16, 43). The acquisition of IncK plasmids containing bla_CTX-M-14 by group A and B1 E. coli clones could have enhanced their ability to colonize the urinary tract in patients exposed to antibiotics, as highlighted by the high frequency of fluoroquinolone resistance among CTX-M-14-producing E. coli isolates causing UTIs (65.6%).

Plasmids of the IncHI2 group.

The bla_CTX-M-14 gene was located on a 290-kb transferable IncHI2 plasmid in two isolates. The sequences corresponding to the replication origin and the relaxase showed homology with R478 (GenBank accession no. BX664015.1), the prototype of the IncHI2 plasmids. Despite the low frequency of IncHI2 plasmids in our series, the role of this plasmid group in the spread of bla_CTX-M-14 must not be underestimated, as it has been reported worldwide, suggesting that it is widespread (39).

The bla_CTX-M-14 enzyme is located in different genetic platforms, supporting different mobilization events.

We identified two variants of bla_CTX-M-14 previously designated CTX-M-14a (n = 59/61) and CTX-M-14b (n = 2/61) (GenBank accession nos. AF252622 and DQ359215, respectively). bla_CTX-M-14a is the predominant variant described and encodes the enzyme considered to be the progenitor of the CTX-M-9 cluster also known as CTX-M-18 (4). To date, the bla_CTX-M-14b enzyme (GenBank accession no. EU274579) has been detected only in Spain and Tunisia (39), and it exhibits three nucleotide changes (A372G, G570A, G702A) in relation to the nucleotide sequence of bla_CTX-M-14a. The last two are also found in bla_CTX-M-9 (GenBank accession no. AF174129).

In our study, bla_CTX-M-14a was associated with either ISEcp1 in IncK plasmids (n = 27) or ISCR1 in an IncHI2 plasmid (n = 1) and IncI-like plasmids (n = 3). bla_CTX-M-14b was linked to ISCR1 in an IncHI2 plasmid (n = 1) or to ISEcp1 in an IncK plasmid (pRYC109) (n = 1). A truncated IS903 sequence was identified downstream of ISEcp1-bla sequences in representative isolates corresponding to each plasmid type. ISEcp1 was detected 42 nucleotides upstream of both bla_CTX-M-14a and bla_CTX-M-14b. It is of note that an identical 42-bp region has also been identified upstream of different genes encoding ESBLs of the CTX-M-9 cluster, such as CTX-M-9, -14, -16, -17, -19, -21, and -24 and Toho-2 (GenBank accession nos. AF174129, AF252622, AY033516, AF454633, AF458080, AJ416346, NC_011617, and D89862, respectively), which would indicate a unique origin for these variants, as suggested previously (4). An integron platform similar to that of In60 which contains bla_CTX-M-9 (GenBank accession no. AF174129) and which is not linked to Tn21 was identified among the five isolates harboring ISCR1 and bla_CTX-M-14a or bla_CTX-M-14b within IncI and IncHI2 plasmids (41). The characterization of similar genetic structures for the bla_CTX-M-14 variants in this and other studies probably reflects in-platform β-lactamase gene mutation-based evolution. This kind of presumed local sequence evolution has been demonstrated for CTX-M enzymes belonging to other phylogenetic clusters, such as bla_CTX-M-1 and bla_CTX-M-32 or bla_CTX-M-3 and bla_CTX-M-15 (40).

The location of bla_CTX-M-14a and bla_CTX-M-14b on different genetic platforms linked to ISEcp1 and ISCR1 on plasmids of different incompatibility groups suggests different mobilization events and subsequent recombinatorial processes among plasmids.

Conclusions.

The high frequency of occurrence of the bla_CTX-M-14 enzyme in our geographic area is mainly due to the spread of IncK plasmids among UPEC clones of phylogroups A, B1, and D, with particular STs being overrepresented. The countrywide dissemination of this plasmid may also be suggested on the basis of two Spanish studies which inferred the presence of IncK plasmids among CTX-M-14-producing isolates recovered from different cities by the amplification of specific sequences from a few isolates by the PBRT method (15, 39). However, data about plasmids containing bla_CTX-M-14 from other countries where this ESBL is commonly detected indicate that other incompatibility groups can also be involved in the dissemination of CTX-M-14 (34, 54a). The diversity of genetic platforms containing bla_CTX-M-14 highlights the hypothesis of the existence of different mobilization events for this gene (4) and the influence of the genetic context on its successful dissemination.

The results obtained by the multilevel approach used in the present study offer an example of how different biological and genetic entities should be simultaneously taken into consideration to explain the local spread of antibiotic resistance (3). This particular complex landscape reflects the efficient interplay between clonal epidemicity, plasmid transfer and maintenance, the capture of resistance genes by particular genetic platforms, and, possibly, bacterial virulence traits in determining the evolution of bacteria at the local level (35).