|Home | About | Journals | Submit | Contact Us | Français|
We determined the complete nucleotide sequences of three plasmids that encode CTX-M extended-spectrum β-lactamases (ESBLs) in pulsed-field gel electrophoresis-defined United Kingdom variants (strains A, C, and D) of the internationally prevalent Escherichia coli O25:H4-ST131 clone. Plasmid pEK499 (strain A; 117,536 bp) was a fusion of type FII and FIA replicons and harbored the following 10 antibiotic resistance genes conferring resistance to eight antibiotic classes: blaCTX-M-15, blaOXA-1, blaTEM-1, aac6′-Ib-cr, mph(A), catB4, tet(A), and the integron-borne dfrA7, aadA5, and sulI genes. pEK516 (strain D; 64,471 bp) belonged to incompatibility group IncFII and carried seven antibiotic resistance genes: blaCTX-M-15, blaOXA-1, blaTEM-1, aac6′-Ib-cr, catB4, and tet(A), all as in pEK499. It also carried aac3-IIa, conferring gentamicin resistance, and was highly related to pC15-1a, a plasmid encoding the CTX-M-15 enzyme in Canada. By contrast, pEK204 (strain C; 93,732 bp) belonged to incompatibility group IncI1 and carried only two resistance genes, blaCTX-M-3 and blaTEM-1. It probably arose by the transposition of Tn3 and ISEcp1-blaCTX-M-3 elements into a pCOLIb-P9-like plasmid. We conclude that (i) United Kingdom variants of the successful E. coli ST131 clone have acquired different plasmids encoding CTX-M ESBLs on separate occasions, (ii) the blaCTX-M-3 and blaCTX-M-15 genes on pEK204 and pEK499/pEK516 represent separate escape events, and (iii) IncFII plasmids harboring blaCTX-M-15 have played a crucial role in the global spread of CTX-M-15 ESBLs in E. coli.
Escherichia coli strains producing CTX-M extended-spectrum β-lactamases (ESBLs) have emerged as major global pathogens, primarily associated with urinary tract infections, sometimes with contingent bacteremia (3, 25). Internationally disseminated clones have been recognized among these ESBL-producing E. coli strains through the application of multilocus sequence typing. These include E. coli lineages of sequence types ST131 and ST405, which are of particular public health concern (6, 27).
ST131 was first defined in 2008, though it is now known to have been in circulation at least from 2003, and has been reported widely across Europe, North America, and the Far East, often with the CTX-M-15 ESBL encoded by related IncFII plasmids that also encode OXA-1 and TEM-1 β-lactamases and the aminoglycoside/fluoroquinolone acetyltransferase AAC(6′)-Ib-cr (6, 16, 27). It belongs to phylogroup B2, is uropathogenic, and includes many variants with various complements of virulence determinants and divergent pulsed-field gel electrophoresis (PFGE) profiles, albeit generally related at ≥65% banding pattern similarity (15, 21, 23, 27, 27).
Allowing for its PFGE diversity, ST131 E. coli may be more widespread than is currently realized and may be prone to acquire prevalent resistance plasmids that are circulating in a particular area (35). Consistent with this view, ST131 variants in Japan have plasmids that encode group 2 or group 9 CTX-M enzymes, which are the dominant Asian types, rather than CTX-M-15 (31), whereas some United Kingdom members have the blaCTX-M-3 enzyme with different flanking sequences from the common blaCTX-M-15 gene. The prevalence of antibiotic-susceptible members of this clone is largely unknown, although one recent study identified carriage in 7% of healthy subjects in the Paris area (23).
In the United Kingdom, E. coli isolates with the CTX-M-15 ESBL have become prevalent since 2003. They include five PFGE-defined strains, A to E, which all belong to ST131 (21, 34), along with PFGE-diverse isolates, some of which also belong to ST131. Strain A, which is the most prevalent variant, is locally dominant (e.g., in Hampshire, Shropshire, and parts of Lancashire); strains B, C, and E are nationally scattered, whereas strain D is local to one center (34). The CTX-M-3 enzyme is associated with strain C isolates from Belfast: producers are less multiresistant than representatives of strain C with the CTX-M-15 enzyme from elsewhere in the United Kingdom.
We report here the complete nucleotide sequences of the blaCTX-M-15-harboring plasmids pEK499 and pEK516, from representative isolates of United Kingdom strains A and D, respectively, and of the blaCTX-M-3-harboring plasmid pEK204, from a Belfast representative of strain C.
E. coli isolates EO499, H041280204, and EO516 all belong to the O25:H4-ST131 lineage (27, 35) and were selected to represent United Kingdom PFGE-defined strains A, C, and D, respectively (15, 16, 34). Isolates EO499 and EO516 produce CTX-M-15 ESBL, whereas H041280204 produces its close relative, CTX-M-3; their ESBL-encoding plasmids were designated pEK499, pEK516, and pEK204, respectively.
Plasmids were transferred by conjugation or electroporation into E. coli strain DH5α or J53 (Table (Table1)1) and their sequences were determined by a shotgun cloning method (MWG, Planegg-Martinsried, Germany). Briefly, randomly sheared plasmid fragments of 2 to 3 kb were cloned into the pGEM-T Easy vector and then transformed into E. coli DH10b. Inserts were sequenced by BigDye Terminator chemistry. Sequences were assembled using the Staden package. Combinatorial PCRs, directed PCRs, and walking reads were used to assemble the contigs and to fill in gaps.
The plasmid multilocus sequence typing (pMLST) scheme for IncI1 plasmids (9) was used for pEK204. The relevant fragments of the pilL (254 bp), sogS (254 bp), ardA (343 bp), and repI1 (104 bp) genes and an 812-bp tnbA-pndC region were compared with known allelic variants (http://pubmlst.org/perl/mlstdbnet/mlstdbnet.pl?file=incI1_profiles.xml&page=oneseq).
Open reading frames were predicted and annotated using the Bacterial Annotation System (BASys; http://wishart.biology.ualberta.ca/basys/cgi/submit.pl) (33) and confirmed with DNAMAN 5.2.10 software (Lynnon BioSoft, Lynnon Corporation; http://www.lynnon.com). Each predicted protein was compared against an all-protein database using BlastP (http://blast.ncbi.nlm.nih.gov/Blast.cgi) with a minimum cutoff of 30% identity over 80% length coverage, checking at least two best hits among the COG, KEGG, and nonredundant protein databases. Gene sequences were further compared and aligned with GenBank data using BLAST (http://blast.ncbi.nlm.nih.gov/Blast.cgi) and CLUSTAL W (http://www.ebi.ac.uk/clustalw) and with reference plasmids by two sequence alignments using the Blastnt-Blast2 algorithm (http://blast.ncbi.nlm.nih.gov/Blast.cgi). The IncI1 plasmid R64 (GenBank accession no. NC_005014.1) was used as a reference for annotating pEK204, whereas IncF plasmids pU302L (NC_006816), pIP1206 (AM886293), pRSB107 (AJ851089), and pC15-1a (NC_005327) were used to annotate pEK499 and pEK516. Data files were compiled using Sequin (http://www.ncbi.nlm.nih.gov/Sequin/). The gross structures of whole plasmids were compared with WebACT (http://www.webact.org/WebACT/home) (1), and plasmid maps were prepared using the CGView server (http://stothard.afns.ualberta.ca/cgview_server/) (11).
The GenBank accession numbers for the three plasmids are EU935738 (pEK516), EU935739 (pEK499), and EU935740 (pEK204).
Plasmid pEK499 was a circular molecule of 117,536 bp, belonged to incompatibility group F, and represented a fusion of two replicons of types FII and FIA. It harbored 185 predicted genes (Table (Table2),2), including 10 conferring resistance to eight antibiotic classes; pEK499 also included various modules that have been identified on other IncF plasmids and which are composed of resistance genes and insertion sequences. With the exception of blaTEM-1, all the antibiotic resistance genes were clustered in a 25-kb region. They included blaCTX-M-15 and blaOXA-1 as well as genes conferring resistance to aminoglycosides and ciprofloxacin (aac6′-Ib-cr), macrolides [mph(A)], chloramphenicol (catB4), and tetracycline [tet(A)]. A 1.8-kb class I integron was present within this multiresistance region and carried dfrA7 and aadA5, encoding trimethoprim and streptomycin resistances, respectively, and sulI, encoding sulfonamide resistance.
The structure of pEK499 was compared with the following two fully sequenced plasmids carrying the IncFII-FIA replicons: pRSB107 from a bacterium collected in a sewage treatment plant (32) and pIP1206 from an E. coli clinical isolate in a Belgian hospital (28). Neither of these encodes an ESBL (Fig. (Fig.1).1). pRSB107 encodes an aerobactin iron acquisition system and resistance to penicillins (blaTEM-1), aminoglycosides [aph(3) and strA-strB], sulfonamides (sulII), macrolides [mph(A)], chloramphenicol (catA), and tetracyclines [tet(A)]; it also carried the class 1 integron-borne trimethoprim resistance gene cassette dhfR (32). In common with pRSB107, pIP1206 also carried catA, tet(A), and blaTEM-1, but it also harbored (i) two class 1 integrons with the aadA4-dhfr17 and qepA gene cassettes, respectively (the latter gene confers resistance to hydrophilic fluoroquinolones), and (ii) rmtB, which encodes an 16S rRNA m7G methyltransferase that confers resistance to all aminoglycosides (28).
The pEK499 scaffold encoded no fewer than the following five systems to ensure stable plasmid inheritance and postsegregation killing: (i) the postsegregation killing protein Hok and its modulator Mok, located near the parB gene; (ii) the toxin-antitoxin system pemI-pemK, flanking the region of the replicon FII; (iii and iv) two copies of the vagC-vagD virulence-associated genes; and (v) one copy of the toxin-antitoxin system ccdA-ccdB, located in the region of the FIA replicon. This represents the largest number of addiction systems yet described on any IncF plasmid. For comparison, pRSB107 and pIP1206 carry four (two vagC-vagG, ccdA-ccdB, and pemI-pemK) and three (vagC-vagG, ccdA-ccdB, and pemI-pemK) systems, respectively (28, 32). These features seem likely to ensure that pEK499 is maintained in the absence of any antibiotic selective pressure. Plasmids pEK499 and pRSB107 also shared a region with two copies of the module that encodes permeases and ATP binding proteins of the ABC transporter family, and which is also partially present on pIP1206. The contribution of these transporters to virulence and plasmid maintenance has not yet been established.
pEK499 contained an incomplete transfer region composed of the genes traM to traC, but not traW to traX, and thus lacked functional conjugation machinery. Consistent with this observation, pEK499 was not transferable by conjugation in vitro; it was transformed prior to plasmid sequencing. pEK499 also lacked the aerobactin iron acquisition system, and the type I DNA restriction, raffinose and arginine deaminase clusters that have been described in the IncF plasmids pRSK107 and pIP1206, respectively (28, 32).
Plasmid pEK516 was a circular molecule of 64,471 bp, harboring 103 predicted genes (Table (Table3)3) and belonging to incompatibility group IncFII. It carried the following seven genes encoding antibiotic resistance clustered in a 22-kb region: blaCTX-M-15, blaOXA-1, blaTEM-1, aac6′-Ib-cr, aac3-IIa, catB4, and tet(A). An ISEcp1 element was located 48 bp upstream of blaCTX-M-15. pEK516 shared 75% of its DNA sequence with pEK499, albeit with considerable rearrangements (Fig. (Fig.1);1); notably, both plasmids carried the region containing the FII replicon and the hok-mok and parB genes. However, pEK516 was 53 kb (45%) smaller than pEK499 but carried the type I partitioning locus (parM and stbB), ensuring stable plasmid inheritance. Moreover, its Tn3::blaTEM-1 module flanked the FII replicon, whereas this module was located close to the deleted transfer region on pEK499 (Fig. (Fig.1,1, red line). The resistance region of pEK516 encoded an AAC(3) enzyme, absent from pEK499 and conferring resistance to gentamicin, but lacked the class 1 integron with genes conferring resistance to trimethoprim, streptomycin, and sulfonamides, also the macrolide resistance cluster. Consequently, it was similar to pC15-1a (2), from a widespread Canadian strain of E. coli with the CTX-M-15 ESBL and probably also belonging to the ST131 clone. In fact, three genetic events potentially accounted for all of the differences observed between pC15-1a and pEK516, namely, (i) the partial deletion of the tra region, (ii) the inversion of the region between the FII replicon and the parM gene, and (iii) the acquisition of catB4 close to blaOXA-1 within the resistance region of pEK516.
Both pEK516 and pC15-1a carried only two addiction systems, pemI-pemK and hok-mok, and resembled the IncFII portion of pEK499. pEK516 showed a greater deletion of the transfer region than pEK499, since the latter contained traC, which was missing in pEK516. Despite this, pEK516 was transferred by conjugation in vitro (16); so it seems likely that the tra deletion(s) occurred during subsequent storage and prior to plasmid sequencing.
In summary, pEK516 represented a highly related variant of the previously identified plasmid, pC15-1a (2). Plasmids of the FII group harboring blaCTX-M-15 have wide geographic scatter (4, 6) and have played a crucial role in the global spread of CTX-M-15 ESBLs in E. coli.
Plasmid pEK204 was a circular molecule of 93,732 bp, harboring 112 predicted genes (Table (Table4),4), and could be transferred by conjugation in vitro. It belonged to incompatibility group IncI1, which is characterized by the presence of a gene cluster encoding a thin, type IV pilus required for liquid matings (18) and the RepZ replicase gene. As such, it was distinct from two other fully sequenced plasmids encoding the CTX-M-3 enzyme, pCTX-M3, which is widespread in Poland (10, 26), and pK29 (5). pEK204 was assigned to a new IncI1 pMLST type (9), ST16 (I1, A5, S8, P6, and T10), with four unique alleles (pil6, sogS8, ardA5, and trbA-pndC10). However, its gross structure revealed strong similarity to IncI1 plasmid pCOLIb-P9 (GenBank AB021078), although the region that includes the colicin 1b gene (pCOLIb-P9 nucleotides 8310 to 20275) was absent, and to R64, which is the reference plasmid for the IncI1 group (GenBank accession no. AP005147; Fig. Fig.2).2). In comparison with those of R64, pEK204 lacked the arsenic, tetracycline, and streptomycin resistance genes and the addiction systems mck-kor and parA-parB; this deleted region was substituted by its own resistance region. The colinearity between the pEK204 and R64 plasmid scaffolds was therefore well maintained in the transfer region, with the exception of the shufflons—a characteristic feature of IncI1 plasmids that may act as a biological switch (12, 19, 20)—which appeared to be rearranged in pEK204 with respect to R64 and also by the insertion of an IS66 element.
In contrast to multiresistance plasmids pEK499 and pEK516, pEK204 carried only two known resistance genes, blaCTX-M-3 and blaTEM-1 (Fig. (Fig.3).3). The blaCTX-M-3 gene, which encodes an ESBL that differs from CTX-M-15 by only a Asp240→Gly substitution, has previously been detected on plasmids belonging to different incompatibility groups and with broad host ranges (4), including IncL/M (pCTX-M-3; ca. 90 kb) in Poland (10, 26) and IncHI2 (pK29; ca. 270 kb) in Taiwan (5), as well as the IncA/C, IncFII, and IncN types (4). Large (>90-kb) IncI1 plasmids encoding the CTX-M-3 enzyme have been reported previously in diverse members of the Enterobacteriaceae in a university hospital in Taiwan (24). Moreover, IncI1 plasmids have been reported to encode myriad other β-lactamases besides the CTX-M-3 enzyme, including CTX-M-1, CTX-M-2, CTX-M-9, CTX-M-14, CTX-M-15, and CTX-M-24, as well as TEM-1 (as here), TEM-52, SHV-12, several CMY (acquired AmpC) enzymes, and the metallocarbapenemase VIM-1 (4, 5, 9).
The sequence data suggest that pEK204 arose by the transposition of Tn3 and ISEcp1-blaCTX-M-3 elements into a pCOLIb-P9-like plasmid. This Tn3 element inserted after the position equivalent to nucleotide 8269 of pCOLIb-P9 and was flanked by 5-bp direct repeats of TTTTC. Both of the Tn3 terminal inverted repeats, IRL and IRR, were intact, but the tnpA gene (encoding the transposase) was disrupted by ISEcp1-blaCTX-M-3 (Fig. (Fig.3).3). The ISEcp1 element was located 128 bp upstream of blaCTX-M-3, and the linking sequence was identical to that of pCTX-M-3 (10), though different from that (48 bp) between ISEcp1 and blaCTX-M-15 on pEK516 or the remnant of this IS element on pEK499. This underscores the point that, although they differ only by one nucleotide, the blaCTX-M-3 and blaCTX-M-15 genes in the United Kingdom plasmids represent separate escape events from Kluyvera spp.
The ISEcp1-blaCTX-M-3 element of pEK204 was flanked by 5-bp direct repeats of TATTG, consistent with ISEcp1-mediated transposition, which prefers AT-rich target sequences (29). Despite disruption, the Tn3 transposase (tnpA) gene on pEK204 was still predicted to encode a protein of 999 amino acids, comprising 970 (ca. 97%) “genuine” amino acids plus 29 “new” C-terminal amino acids resulting from the read-through of the in-frame ISEcp1-blaCTX-M-3 insertion. The genuine C-terminal residues of the Tn3 transposase were encoded by a ΔtnpA remnant located between IRL and ISEcp1 (Fig. (Fig.3).3). The disruption of transposons by other transposable elements is a feature common to many bacterial genomes and plasmids, including all of those reported here. The resistance region of pEK204 represents a simple example of this but raises the question of whether this plasmid evolved via one or two transposition events. If the predicted transposase of Tn3 on pEK204 remains functional, it would be expected to mediate the transposition of a new element, still defined by the Tn3 terminal inverted repeats IRL and IRR but encoding CTX-M-3 ESBL in addition to TEM-1 penicillinase (Fig. (Fig.3).3). The potential for the simultaneous transposition of two distinct β-lactamase genes from pEK204-like plasmids has public health importance and will be investigated further. It is relevant in context that Tn3 played a major role in the huge dissemination of TEM-1 β-lactamase in the 1960s and 1970s.
We have determined the complete sequences of the CTX-M ESBL-encoding plasmids found in three United Kingdom variants (A, C, and D) of the pathogenic, multiresistant ST131 E. coli lineage. Strain A (pEK499), which is widespread and locally dominant in the United Kingdom, has been found in Austria (8) and, more recently, in Bolzano, northern Italy (R. Aschbacher, D. M. Livermore, and N. Woodford, unpublished data). Strain C is also widely found in the United Kingdom: some Belfast isolates of the strain have the CTX-M-3 enzyme encoded by pEK204, as here, but others, including many from the United Kingdom mainland, have CTX-M-15, possibly encoded by a different plasmid. In contrast to these widely disseminated strains, the third ST131 variant studied, strain D (pEK516), is prevalent only in Shropshire, a county on the English-Welsh border.
Clearly, differences exist among variants of the ST131 clone and these probably have significant impact on their relative success and prevalence. Success is likely to reflect a complex combination of characteristics, both intrinsic to the PFGE-defined variant or acquired on plasmids and mobile genetic elements, interacting with local pressures and opportunities relating to antibiotic usage and patient types. It is clear, however, that these strains and their multiresistance plasmids have been moving beyond the hospital environment in the United Kingdom for some years (34). A recent study identified ESBL-producing E. coli in the bowel flora of 40% of residents of long-term-care facilities in Belfast; almost half were colonized by strain A (30) and most of the remainder by diverse ST131 variants harboring pEK204-like plasmids (7). The multitude of addiction systems present on pEK499 will, in particular, ensure that it is maintained even in the absence of antibiotic selection.
Many patients with E. coli producing CTX-M ESBLs, most of them elderly, present with community-onset infections in increasing numbers, providing evidence of true community acquisition (25). This age distribution may change over time, leading to new problems. A study from Hong Kong found significant rates of ESBL production (particularly of the CTX-M-14 enzyme) among urinary isolates of E. coli from women of all ages; the prevalence was 7.3% among community-onset infections in the 18-to-35 age group (13). Another study, from Canada, of community-onset infections caused by ESBL-producing E. coli recorded a much broader age distribution among patients deemed to have community-acquired infections versus those considered to have health care-associated infections (22). Rising rates of E. coli with CTX-M ESBLs in the genitourinary tracts of sexually active women raise the alarming possibility that these enzymes might “escape” into sexually transmitted bacterial pathogens, specifically Neisseria gonorrhoeae. Oral and intramuscular oxyimino-cephalosporins, such as cefixime and ceftriaxone, are widely used as a first-line treatment for uncomplicated gonorrhea, and any evolution of ESBL-producing gonococci would be a catastrophic development.
In summary, United Kingdom E. coli strains A, C, and D belonging to the internationally disseminated O25:H4-ST131 clone (21, 27, 34) have acquired different plasmids encoding CTX-M ESBLs on separate occasions.
We thank AstraZeneca for supporting this work.
Published ahead of print on 17 August 2009.