Dissemination of Clonally Related Escherichia coli Strains Expressing Extended-Spectrum β-Lactamase CTX-M-15

E. coli ST131 and ST405 and multidrug-resistant IncFII plasmids may determine spread of this lactamase.

spread of specifi c CTX-M variants. In this article, through analysis of the population biology of CTX-M-15-producing isolates from 7 countries and characterization of their genetic elements, we provide a comprehensive picture of elements involved in international spread of a particularly widespread mechanism of antimicrobial drug resistance.

Clonal Relationships
Clonal relationships were established by pulsed-fi eld gel electrophoresis (PFGE) of XbaI-digested genomic DNA (New England Biolabs, Ipswich, MA, USA) as described (25). Assignment of E. coli phylogenetic groups was conducted by using a multiplex PCR assay described by Clermont et al. (26). All E. coli isolates belonging to phylogroups B2 and D were characterized by multilocus sequence typing (MLST) using the standard 7 housekeeping loci (www.mlst.net). All fumC sequences from E. coli isolates belonging to phylogroup D were analyzed for a C288T single nucleotide polymorphism. This polymorphism is specifi c for a globally disseminated E. coli strain arbitrarily designated as E. coli clonal group A (CgA) that is associated with community-acquired urinary tract infections (27,28).

Transferability and Location of bla CTX-M-15
Transferability was tested by broth and fi lter mating assays using E. coli K12 strain BM21R (resistant to nalidixic acid and rifampin, positive for lactose fermentation, and free of plasmids) as recipient at a 1:2 donor: recipient ratio. Transconjugants were selected on Luria-Bertani agar plates containing cefotaxime (1 mg/L) and rifampin (100 mg/L) and incubated at 37°C for 24-48 h. Transformation was performed for a subset of isolates by using conditions reported (3). Chromosomal or plasmid location of bla CTX-M-15 genes was assessed by hybridization of I-CeuIdigested genomic DNA with bla CTX-M-15 and 16S rDNA probes and electrophoresis (5-25 s for 23 h and 60-120 s for 10 h at 14°C and 6 V/cm 2 ) (25). Transfer and hybridization were performed by using standard procedures. Labeling and detection were conducted by using enhanced chemiluminescence (Amersham Life Sciences, Uppsala, Sweden) following manufacturer's instructions.

Plasmid Characterization
Plasmid DNA was obtained by using different midiprep plasmid purifi cation kits (QIAGEN, Hilden, Germany, and Marlingen Biosciences, Ijamsville, MD, USA). Plasmids were classifi ed according to their incompatibility group by a PCR-based replicon-typing scheme (29). Determination of plasmid size and confi rmation of replicon content was established for transconjugants (or wild-type strains in the absence of transfer) by hybridization of S1 nuclease-digested genomic DNA with probes specifi c for bla CTX-M-15 and for different F replicons (FII, FIA, FIB), which were obtained by PCR as described (19). Relationships among plasmids were determined by comparison of EcoRI and HpaI digested DNA patterns and comparison of repFII sequences. Genescan software (Applied Biosystems, Foster City, CA, USA) was used for collection of gel images. Data of a subset of representative patterns were exported into Fingerprinting II Informatix version 3.0 software (Bio-Rad Laboratories, Hercules, CA, USA) for further interpretation. Cluster analysis was conducted by using the unweighted pair group method with arithmetic averages (optimization 0.5%, tolerance 1.00%).

Epidemiologic Background
Most CTX-M-15-producing E. coli isolates belonged to phylogroups B2 (50%) and D (25%), which are known to be associated with the hospital setting and extraintestinal pathogenic E. coli. Phylogroups A (18%) and B1 (7%), which are associated with animal or human commensal strains, were less frequently represented. All isolates of phylogroups B2, A, and D corresponded to subgroups B2 3 , A 1 , and D 1 , respectively, which are the most common ones within each phylogenetic group (31). The 43 clinical isolates were classifi ed into 32 PFGE types (B2 3 , 13; D 1 , 10; A 1 , 6; and B1, 3). Among B2 3 strains, 10 PFGE types (18 isolates from France, Canada, Spain, Portugal, Kuwait, and Switzerland) were possibly related according to criteria of Tenover et al. (32) (difference <6 bands, >80% similarity) and were assigned to the sequence type (ST) ST131. The 4 unrelated B2 strains were classifi ed within ST695 (1 from France), ST28 (1 from Switzerland), ST354 (1 from Portugal and Spain) and ST405 (1 from Portugal). All isolates of phylogroup D 1 were clonally unrelated by PFGE (difference >6 bands), although MLST studies indicated that 4 PFGE types (5 isolates) from Kuwait, Switzerland, and Spain corresponded to ST405. The fumC sequences of the remaining 6 E. coli D strains were highly diverse (alleles 4, 13, 26, 88, and 132). None of the strains had the C288T single nucleotide polymorphism specifi c for E. coli strain CgA (28). All 3 B1 isolates were found in France. Among B2 E. coli isolates, all but 4 were isolated from urine and all but 2 belonged to ST131. These strains correspond to 2 isolates recovered from wounds and identifi ed as ST28 and ST354 and 2 ST131 isolates from respiratory and fecal samples, respectively.

Location and Transferability of bla CTX-M-15
The bla CTX-M-15 gene was located on plasmids in all but 6 strains and was positively transferred by conjugation or transformation in 37% of the strains tested. In 8 clinical isolates corresponding to 7 PFGE types, the probe for bla CTX-M-15 hybridized in chromosomal bands (2 belonging to B2 3 ST131, 2 to D 1 , 1 to D 1 ST405, and 1 to A 1 ). In 2 other strains, the bla CTX-M-15 probe hybridized both with plasmid and chromosomal bands (1 strain from D ST405 and 1 from phylogroup B1).

Plasmids Encoding CTX-M-15
Plasmids positive for the bla CTX-M-15 gene showed variable sizes (85-160 kb), belonged to the narrow host range incompatibility group IncF, and had replicon FII alone or in association with the FIA or FIB replicons (online Appendix Table, available from www.cdc.gov/EID/content/14/2/ 195-appT.htm). Many restriction fragment length polymorphism (RFLP) patterns were observed, with overrepresentation of 3 profi les corresponding to 3 plasmids arbitrarily designated as plasmid A (85 kb), plasmid B (120 kb), and plasmid C (85 kb). Plasmid A, which was isolated from B2 E. coli strains from 4 countries (India, France, Portugal, and Spain), was associated with different STs (ST131, ST354, or ST405). Plasmid C was also detected in clonally unrelated E. coli of phylogroups B2 and D from Switzerland, Canada and France. Plasmid B, which was only associated with E. coli ST131, was widely disseminated in all countries studied. Sequence analysis of the replicons showed 4 repFII types: repFII(1), which was identical to that of plasmids R100, NR1, or pC15-1a, and was the most represented and identifi ed in 23 plasmids; repFII (2), which had 99%-100% homology with plasmid pRSB107 (Gen-Bank accession no. AJ851089), was identifi ed in 6 plasmids; and repFII(3) and repFII(4), which were detected in 2 and 7 plasmids, respectively, and showed >93% homology with repFII(1). All repFIA and repFIB sequences were 99% and 100% homologous, respectively, with that of pRSB107 (GenBank accession no. AJ851089).
Computer analysis of representative RFLP patterns and repFII sequences grouped CTX-M-15 plasmids within 3 major clusters with similarity >70%. Cluster I comprises most plasmids, including plasmids A and B, most containing repFII(1) and showing variable replicon content. Cluster II comprised only plasmid C derivatives showing slightly different repFII sequences, and cluster III included 2 plasmids carrying repFII(2), FIA, and FIB replicons (Figure).
In the 8 strains with chromosomal location of bla CTX-M-15 , repFII plasmids were identifi ed but these plasmids were negative for the bla CTX-M-15 gene. Several strains that were also positive for additional plasmids and negative for the bla CTX-M-15 gene were assigned to different incompatibility groups or were untypeable by the PCR-based replicon typing scheme used.

Discussion
Our study indicates that current worldwide spread of the bla CTX-M-15 gene is driven mainly by 2 epidemic E. coli strains belonging to phylogroups B2 (ST131) and D (ST405) and by its location on IncF plasmids harboring multiple antimicrobial drug-resistance determinants, including the recently described aac(6′)-Ib-cr gene. The presence of bla CTX-M-15 has previously been associated with E. coli strains of phylogroups B2 and D, and in some instances, with specifi c PFGE types (9)(10)(11)(12)16). We detected an emerging and globally disseminated CTX-M-15 phylogroup B2 E. coli strain corresponding to the ST131 that was responsible for clonal outbreaks in Canada, France, Spain, and Portugal (11,14,16,23). Other CTX-M-15 B2 strains belong to clonal complexes ST695, ST405, ST354, or ST28, which have previously been detected in different geographic areas among isolates that do not express CTX-M-15 (online Appendix Figure, available from www.cdc.gov/EID/content/14/2/195-appG.htm).
Globally disseminated E. coli strains associated with acute, uncomplicated, community-acquired cystitis and pyelonephritis, designated in community patients as clone CgA (ST69), have only been occasionally associated with CTX-M-15 production in Canada (16,27,28). Although the isolates in our study do not belong to clone CgA, they were isolated mainly from urine samples, and an association of ST131 E. coli isolates with urinary tract infections might be inferred. Although most CTX-M-15 isolates studied were recovered from hospitalized patients, these microorganisms are now widely spread in the community setting, including long-term care facilities in the countries from which isolates included in this study originated (2,5,14,33). Our study has increased knowledge of the number of epidemic E. coli clonal complexes causing urinary tract infections.
All plasmids carrying bla CTX-M-15 included in this study corresponded to incompatibility group F, and all had the FII replicon, which was assorted mainly in multireplicon plasmids with additional replicons of the FIA and FIB types. Association of the bla CTX-M-15 gene with IncFII replicons has been described in studies conducted in Canada, France, Spain, and the United Kingdom (5,7,8,17,19). Although we observed intercontinental dissemination of 3 major IncFII plasmid scaffolds (A, B, and C) carrying bla CTX-M-15 , similarity >70% among all variants studied and presence of genes also found in pC15-1a, a CTX-M-15 plasmid (Gen-Bank accession no. AY458016) that has a 28.4-kb multidrug resistance region containing bla TEM-1 , bla OXA-1 , the aac(6′)-Ib-cr gene (aminoglycoside 6′-N-acetyltransferase type Ib-cr variant responsible for reduced susceptibility to both aminoglycosides and certain fl uoroquinolones), and genetic determinants coding for resistance to tetracycline and aminoglycosides (5,30), suggest a common origin or a common particular plasmid scaffold involved in the dissemination of CTX-M-15.
Because IncF plasmids are a heterogeneous and largely diffused family of plasmids in E. coli, they could acquire the bla CTX-M-15 gene. IncF plasmids negative for the bla CTX-M-15 gene in strains with this gene at a chromosomal location also suggest dynamic horizontal exchanges between the chromosome and resident plasmids. Extensive recombination events among IncF plasmids are frequent and may have contributed to their apparent high diversity (variable rep content, plasmid size, transferability, antimicrobial drug-resistance genes), driving their evolution and enabling them to persist in diverse E. coli populations (34,35). Such recombination events among plasmids of the same incompatibility group within the same cell occur frequently (34,35). This hypothesis is supported by the results of Lavollay et al. (17), who described mosaicism in a CTX-M-15 plasmid isolated in France that contained genes from 2 different IncFII plasmids, pC15-1a and pRSB107 (from IncFII plasmids fi rst isolated from persons in Canada and activated sludge bacteria from a wastewater treatment plant in Germany, respectively) (5,36).
Spread and maintenance of conjugative plasmids across bacterial populations have been intensively studied from a theoretical point of view, but data from natural populations are scarce (34,37,38). Recovery of related plasmids from clonally unrelated B2 strains might refl ect effi cient transfer of these elements among different B2 E. coli populations. Sharing the same environment, successive immigrant B2 strains might sweep through the population, enabling plasmid hitchhiking at a high frequency in each selective sweep. However, we lack detailed information on the specifi city and stability of different plasmid groups in specifi c hosts. An evolutionary convergent relationship among B2 genetic background and IncFII plasmids cannot be ruled out and should be studied because it might explain successful dissemination of CTX-M-15 plasmids within this E. coli lineage. In addition, our study is one of the few that have identifi ed bla ESBL genes in the chromosome, which might respond either to plasmid integration or transposition driven by ISEcp1 located upstream from the bla CTX-M-15 gene (25,39,40). In conclusion, worldwide dissemination of bla CTX-M-15 is driven by B2 or D E. coli clones associated mainly with urinary tract infections or IncFII plasmids containing a multiple antimicrobial drug-resistance platform that contributes to spread of CTX-M-15. Further studies to test the stability/variability and fi tness of particular plasmids among different bacterial hosts will be relevant in developing additional strategies to control dissemination of antimicrobial drug resistance. †Transferability of antimicrobial drug resistance to a given drug is indicated in boldface. Drugs in parentheses are not associated with resistance in all isolates. Ac, amoxicillin-clavulanic acid; Gm, gentamicin; Km, kanamycin; Na, nalidixic acid; Ni, nitrofurantoin; Sm, streptomycin; Su, sulfonamide; Tb, tobramycin; Tp, trimethoprim; Tc, tetracycline; Cp, ciprofloxacin; Ak, amikacin; Cm, chloramphenicol. ‡Chromosomal (C) or plasmid (P) location of the blaCTX-M-15 gene was determined by hybridization of probes with I-CeuI-digested genomic DNA. Size of the plasmid in kb is indicated in parentheses. §Plasmid size and content were determined by hybridization of S1 nuclease-digested genomic DNA on transconjugants (or wild type if transfer failed). Successful plasmid transfer by conjugation or transformation is indicated in boldface. ¶Replicon content of plasmids with blaCTX-M-15 genes was determined by PCR, hybridization, and sequencing of replicons. Replicons hybridizing in the same band as that of the blaCTX-M-15 gene are indicated in boldface. Numbers in parentheses represent plasmid sizes in kb. #ND, not done; FE, failed extraction.