|Home | About | Journals | Submit | Contact Us | Français|
TEM-24 remains one of the most widespread TEM-type extended-spectrum β-lactamases (ESBLs) among Enterobacteriaceae. To analyze the reasons influencing its spread and persistence, a multilevel population genetics study was carried out on 28 representative TEM-24 producers from Belgium, France, Portugal, and Spain (13 Enterobacter aerogenes isolates, 6 Escherichia coli isolates, 6 Klebsiella pneumoniae isolates, 2 Proteus mirabilis isolates, and 1 Klebsiella oxytoca isolate, from 1998 to 2004). Clonal relatedness (XbaI pulsed-field gel electrophoresis [PFGE] and E. coli phylogroups) and antibiotic susceptibility were determined by standard procedures. Plasmid analysis included determination of the incompatibility group (by PCR, hybridization, and/or sequencing) and comparison of restriction fragment length polymorphism (RFLP) patterns. Characterization of genetic elements conferring antibiotic resistance included integrons (classes 1, 2, and 3) and transposons (Tn3, Tn21, and Tn402). Similar PFGE patterns were identified among E. aerogenes, K. pneumoniae, and P. mirabilis isolates, while E. coli strains were diverse (phylogenetic groups A, B2, and D). Highly related 180-kb IncA/C2 plasmids conferring resistance to kanamycin, tobramycin, chloramphenicol, trimethoprim, and sulfonamides were identified. Each plasmid contained defective In0-Tn402 (dfrA1-aadA1, aacA4, or aacA4-aacC1-orfE-aadA2-cmlA1) and In4-Tn402 (aacA4 or dfrA1-aadA1) variants. These integrons were located within Tn21, Tn1696, or hybrids of these transposons, with IS5075 interrupting their IRtnp and IRmer. In all cases, blaTEM-24 was part of an IS5075-ΔTn1 transposon within tnp1696, mimicking other genetic elements containing blaTEM-2 and blaTEM-3 variants. The international dissemination of TEM-24 is fuelled by an IncA/C2 plasmid acquired by different enterobacterial clones which seem to evolve by gaining diverse genetic elements. This work highlights the risks of a confluence between highly penetrating clones and highly promiscuous plasmids in the spread of antibiotic resistance, and it contributes to the elucidation of the origin and evolution of TEM-2 ESBL derivatives.
Class A extended-spectrum β-lactamases (ESBLs) have been increasingly reported since their first description in 1983. Until recently, those of the TEM family were among the most frequently isolated from both nosocomial and community settings (14, 50). Almost 180 variants of these enzymes, which differ by specific point mutations responsible for extending the spectrum of activity, have already been identified (http://www.lahey.org/studies). Some of them, such as TEM-3, TEM-4, TEM-10, TEM-12, TEM-21, TEM-24, TEM-26, and TEM-52, are widespread in particular geographic regions (5, 6, 12, 14, 15, 17, 19, 31).
TEM-type ESBLs are derivatives of TEM-1 or TEM-2 penicillinases, the first plasmid-mediated β-lactamases from Gram-negative bacteria described, which were characterized in the late 1960s (50). The blaTEM genes are located on different Tn3-like derivatives depending on the ancestor gene from which they evolved (Tn3 [blaTEM-1a], Tn2 [blaTEM-1b], and Tn1 or Tn801 [blaTEM-2]) (Table (Table1)1) (4, 45, 53). The heterogeneous prevalence and distribution of TEM-1 and TEM-2 variants might be related to the distinct genetic platforms in which they were initially located. While blaTEM-1 was found in plasmids of many incompatibility groups, blaTEM-2 was distributed among only a few plasmid types (IncFI, IncA/C, and IncP) in a highly diverse number of Enterobacteriaceae and non-Enterobacteriaceae species (Table (Table1)1) (27, 37). The wide distribution of the genes encoding certain TEM ESBLs seems to have been greatly influenced by their location in conjugative epidemic plasmids (5, 10, 12, 14, 15, 23, 30). The complete genetic environment of these genes has been studied in only a few cases (12, 19, 30).
TEM-24 was first identified in a Klebsiella pneumoniae isolate in France in 1988 and was subsequently detected in multiple Enterobacteriaceae species (Enterobacter aerogenes, Enterobacter cloacae, Escherichia coli, Klebsiella pneumoniae, Klebsiella oxytoca, Citrobacter freundii, Proteus mirabilis, Proteus vulgaris, Providencia stuartii, Providencia rettgeri, Morganella morganii, and Serratia marcescens), Aeromonas spp., and Pseudomonas aeruginosa (1, 2, 3, 17, 24, 34-36, 41, 42). Nevertheless, the successful spread of this enzyme in different European countries has been associated mostly with the expansion of an E. aerogenes clone isolated during outbreak and nonoutbreak situations (1, 6, 17, 18, 20, 31, 42). Large conjugative plasmids carrying blaTEM-24 and conferring resistance to multiple non-β-lactam antibiotics have also been identified, although they were not fully characterized in any case (10, 23, 24, 36, 42). A recent study including TEM-24 producers from France has suggested that blaTEM-24 is located on IncA/C plasmids, based only on the positive amplification of a sequence corresponding to the replication protein of this plasmid type (33).
To better understand the recent and rapid spread of TEM-24-producing Enterobacteriaceae and identify the genetic elements participating in the dissemination and persistence of this ESBL over time, we analyzed representative TEM-24-producing Enterobacteriaceae isolates recovered in different European countries.
Twenty-eight TEM-24-producing isolates from four European countries collected between 1998 and 2004 were studied. They included 13 E. aerogenes isolates, 6 E. coli isolates, 6 K. pneumoniae isolates, 2 P. mirabilis isolates, and 1 Klebsiella oxytoca isolate recovered from Spain (E. aerogenes, n = 7; E. coli, n = 4), Portugal (K. pneumoniae, n = 6; E. aerogenes, n = 3; E. coli, n = 2; P. mirabilis, n = 2; K. oxytoca, n = 1), and France/Belgium (E. aerogenes, n = 3) in outbreak and nonoutbreak situations (Table (Table22 ). Representatives of the epidemic E. aerogenes clone associated with nosocomial outbreaks in France, Belgium, Spain, and Portugal were included (6, 18, 23, 31). Isolates were obtained from 26 patients located at medical wards (44%), intensive care units (32%), and surgical wards (12%), and 12% of patients were outpatients. Species identification, determination of patterns of susceptibility to 12 non-β-lactam antibiotics, and ESBL characterization were performed by standard procedures (15). Relationships among isolates were established by pulsed-field gel electrophoresis (PFGE) as previously described (6, 15, 43). The E. coli isolates were also characterized by multilocus sequence typing (MLST) and classified according to the phylogenetic groups identified by the multiplex PCR assay described by Clermont et al. (11; see also http://www.mlst.net).
Conjugative transfer was tested by broth and/or filter mating methods using E. coli K-12 strain BM21R (Nalr, Rifr, Lac+, plasmid free) as recipient at a 1:2 donor-to-recipient ratio and transconjugants were selected in plates supplemented with ceftazidime (2 μg/ml) and rifampin (100 μg/ml). Plasmid characterization was accomplished by determination of plasmid content and size, identification of the incompatibility group according to a PCR-based replicon typing scheme, sequencing, and further hybridization with specific probes and comparison of plasmid DNA patterns after digestion with EcoRI, PstI, and HpaI restriction enzymes (8, 43). Restriction fragment length polymorphism (RFLP) profiles corresponding to plasmids identified in this study were compared with those of other sequenced plasmids virtually generated by using the NEBcutter V2.0 tool (http://tools.neb.com/NEBcutter2/index.php).
Screening for sequences of widely disseminated genetic elements generally associated with antibiotic resistance was achieved by PCR. They included sequences related to class 1, 2, and 3 integrons (intI1, intI2, and intI3), transposons (tnpR3, tnpA3, tnpR21, tnpM21, tni402, and Tn5393), IS1326, IS1353, IS6100, merA, and other antibiotic resistance genes, such as sul2 (32, 43). Transposable elements were further characterized in selected isolates by a PCR mapping assay based on known sequences of Tn402, Tn3, Tn21, and Tn1696 (43). Oligonucleotide sequences used in this study appear in Table Table33.
Different PCR mapping experiments were performed using primers described in Table Table33 and the following amplification conditions. (i) The first set of conditions comprised 1.5 mM MgCl2, 1× reaction buffer, 0.2 mM each deoxynucleoside triphosphate, 1 μM each primer, and 1.5 units of Taq DNA polymerase (AmpliTaq Gold; Applied Biosystems, Foster City, CA) with amplification programs of 12 min at 94°C and 35 cycles of 1 min at 94°C, 1 to 2 min at 55°C to 65°C, and 1 to 3 min at 72°C, with a final extension step of 10 min at 72°C for standard PCR assays (PCR products < 3 kb). (ii) The other set of conditions comprised 2.5 mM MgCl2, 1× reaction buffer, 0.4 mM each deoxynucleoside triphosphate, 1 μM each primer, 5% dimethyl sulfoxide (when necessary), and 2.5 units of LA Taq polymerase (Takara Bio Inc., Shiga, Japan) with amplification conditions of 1 min at 94°C and 35 cycles of a denaturation step of 20 s at 96°C, annealing at 52°C to 68°C for 1 min, and an extension step at 72°C for 3 to 4 min, followed by a final step of 10 min at 72°C, for long PCRs (products > 3 kb). Southern blot DNA transfer and hybridization assays were performed by standard procedures (43). The blaTEM-24 and repA/C probes used in the hybridization assays were generated by PCR using well-known positive controls as template DNA. Labeling and detection were carried out using the Gene Images Alkphos direct labeling system kit, following the manufacturer's instructions (Amersham Life Sciences, Uppsala, Sweden). PFGE was performed as described previously, using the following conditions: 14°C, 6 V/cm2, 5- to 25-s pulses for 6 h followed by 30- to 45-s pulses for 18 h (S1 nuclease), and 10- to 40-s pulses for 24 h (XbaI) (43).
GenBank searches were performed using the NCBI BLASTN alignment tool. Open reading frame (ORF) identification and annotation were facilitated by the use of ARTEMIS (version 10.1) (http://www.sanger.ac.uk/Software).
The sequences shown in Fig. Fig.11 have been submitted to the GenBank database and correspond to plasmid variants pRYC103T24.1 (GQ293498 and GQ293499) and pRYC103T24.2 (GQ293500 and GQ293501).
We identified similar PFGE patterns among all E. aerogenes, K. pneumoniae, and P. mirabilis TEM-24-producing isolates, in contrast to the clonal diversity found among E. coli strains (6 isolates/6 PFGE types), which belonged to phylogenetic groups B2 (n = 2), D (n = 2), and A (n = 2) (Table (Table2).2). The isolates exhibited resistance to β-lactams, chloramphenicol, sulfonamides, and trimethoprim, and most of them also exhibited resistance to certain aminoglycosides (amikacin, kanamycin, netilmicin, streptomycin, and tobramycin) except gentamicin (only 3.6%), while resistance to ciprofloxacin (71%) or tetracycline (61%) was variable (Table (Table2).2). This phenotype was variably expressed in the transconjugants obtained.
The E. aerogenes strain corresponded to the multidrug-resistant clone responsible for the dissemination of TEM-24 in hospitals in Belgium, France, Portugal, and Spain (6, 18, 23, 31). This clone was initially identified in the early 1990s and, since then, has been extensively detected in European hospitals and health care facilities associated either with the international spread of TEM-24 or with the local spread of TEM-3 in France (1, 6, 17, 18, 20, 23). A recent study has also identified isolates of this clone carrying plasmids encoding different ESBLs (blaSHV-12, blaSHV-5, and blaTEM-20) and metallo-β-lactamases (blaIMP-1 and blaVIM-2), which highlights its role as a highly efficient vehicle of multiresistant conjugative plasmids (2). We cannot establish if the K. pneumoniae and P. mirabilis clones identified were related to other widespread clones, since MLST analysis was not performed for these species. However, some E. coli isolates corresponded to the B2-ST131, A-ST10, A-ST23, and D-ST405 clonal complexes, which are currently widespread in the community carrying plasmids encoding different ESBLs (14). These results indicate that the surveillance and identification of highly penetrating clones is becoming a critical issue for controlling antibiotic resistance (14). The term “penetration” is used here, by analogy with the term used in security engineering, for elements having a high attack profile regarding existing networks, in our case, the human microbiota (16).
In all species analyzed, blaTEM-24 was located on conjugative plasmids of approximately 180 kb, containing replication and relaxase proteins (data not shown) identical to those associated with incompatibility group A/C2 (22). All studied plasmids showed similar restriction patterns, arbitrarily designated pRYC103T24 plasmids (variants differing in the regions encoding antibiotic multiresistance characterized in this study were represented by subtypes designated pRYC103T24.1-3 plasmids, with two of them selected for full characterization of genetic elements) (Fig. (Fig.1)1) . Comparison of RFLP profiles corresponding to blaTEM-24-containing plasmids included in this study and known IncA/C plasmids (GenBank accession numbers CP000602, CP000603, CP000604, and FJ705807) virtually generated by using NEBcutter V2.0 tool (data not shown) showed common bands among them, suggesting recent evolution of this plasmid lineage (22, 38, 57; this study). The recovery of pRYC103T24 from different species (E. coli and E. aerogenes) from the same patient suggests the possibility of an efficient in vivo transfer, previously demonstrated for other large plasmids encoding TEM-24 (34-36, 41, 42), although environmental transfer cannot be excluded.
The broad-host-range IncA/C group of plasmids has been detected in multidrug-resistant human and animal pathogens of Enterobacteriaeceae species (E. coli, Klebsiella sp., Salmonella enterica, P. mirabilis, Providencia sp., S. marcescens, Yersinia pestis, Yersinia ruckeri, Edwarsiella ictaluri), Photobacterium damselae subsp. piscicida (Pasteurella piscicida), Vibrio cholerae El Tor 013, Aeromonas salmonicida, and Pseudomonas spp. on all continents since the first description of pRA1 in the late 1960s (22, 56, 57). Plasmids belonging to the IncA/C group have recently facilitated the spread of genes conferring antibiotic resistance to β-lactams such as blaTEM-3, blaTEM-21, blaTEM-24, blaSHV-2, blaCTX-M-2, blaCTX-M-3, blaCTX-M-14, blaCTX-M-15, blaCMY-2, blaCMY-4, blaIMP-4, and blaVIM-4 or conferring a high level of resistance to aminoglycosides such as those encoded by armA and rmtB (7, 13, 33, 38, 57; this study). Acquisition of different antibiotic resistance transposons and/or genetic islands seems to have occurred multiple times, as described for other particular plasmid groups (Table (Table1)1) (22, 46, 53, 57).
Two Tn402 variants belonging to the In0 and In4 lineages were found in each IncA/C2 plasmid included in this study. They were associated with fully characterized mercury resistance transposons from two E. coli transconjugant strains (Fig. (Fig.1).1). The analyzed IncA/C2 plasmids were designated pRYC103T24.1 and pRYC103T24.2.
The two In0-Tn402 constructs analyzed differed in their 5′ conserved sequence (CS) regions, containing an intI1 gene with a weak (TGGACA-17 bp-TAAGCT) or strong (TTGACA-17 bp-TAAACT) P1/Pant promoter, and in the identity and/or number of gene cassettes (aacA4, aacA4-aacC1-orfE-aadA2-cmlA1, or dfrA1-aadA1) (48). Nevertheless, they exhibited a common tni402 deletion, which was identical to the case for the formerly described prototype in plasmid pVS1. The genetic platforms described here constitute one of the few complete In0-like integrons available in the GenBank sequence database (accession numbers U49101, DQ125241, and CP001232).
The two In4-Tn402 types identified also differed in the 5′ CS region (containing intI1 with either a weak or strong P1/Pant promoter) and the identity and/or number of gene cassettes (aacA4 or dfrA1-aadA1). In addition, a 19-bp duplication identical to that found in In4 in Tn1696 (GenBank accession number U12338) was detected within the attI integration site of the aacA4-containing integrons, creating an in-frame ATG codon which might direct the expression of the aacA4 cassette (Fig. (Fig.1)1) (48). Similar to what was observed for In0-Tn402, these two Tn402 types showed a common tni module. In this case, it represents a new In4-like configuration, differing from that of the prototype In4 (GenBank accession number U12338) by a 26-bp deletion of the 3′ CS region and the absence of the duplication of the last 321 bp of IS6100. Among the diverse In4 elements found in the GenBank database, In1 (GenBank accession number AY046276) has a 42-bp deletion of the same region, suggesting independent evolutionary events. In4-type integrons have been frequently detected among contemporary antibiotic-resistant bacteria and differ in several lengths of deletions of regions flanking IS6100 or in insertions of IS26 at different sites of the 5′ CS or 3′ CS regions, which seem to direct the evolution of members of this family (39, 49; this study). The high diversity of genetic platforms linked to In4 integrons further highlights their high degree of plasticity (29, 39, 49, 54, 55).
The gene cassettes of the integrons identified explain most of the multiresistance phenotypes observed, affecting in most cases trimethoprim (dfrA1), chloramphenicol (cmlA1), streptomycin and spectinomycin (aadA1), gentamicin (aacC1), and other different aminoglycosides, such as amikacin, kanamycin, tobramycin, and netilmicin (aacA4). Variations in the gene cassette content and expression level might explain slight differences in the resistance patterns observed in different strains and transconjugants (Table (Table2).2). The sul2 gene was part of a different module which has been identified within the common 100-kb backbone shared by all known IncA/C2 plasmids (22, 57).
In0-Tn402 and In4-Tn402 were identified within Tn21 and Tn1696, respectively, in pRYC103T24.1. However, they were linked to mosaic transposons comprising tnp21-mer1696 or tnp1696-mer21 in pRYC103T24.2. IRtnp21, IRmer21, and IRtnp1696 were interrupted by IS5075 in different orientations, but we could not detect either IRmer1696 or an IS5075 copy adjacent to mer1696 (Fig. (Fig.1).1). These insertion sequences (ISs) belong to the IS1111 family, which typically shows target site specificity and has been found interrupting the terminal ends of members of the Tn21-like subgroup of transposons (GenBank accession numbers AY333434, AL513383, DQ310703, X64523, CP000971, CP000650, and FJ223605) (13, 21, 46, 47, 57). The presence of similar Tn21-like elements with IRtnp and IRmer interrupted by IS5075 in different plasmid types highlights the possible role of these ISs in the mobilization of these transposons to distinct genetic contexts. Although the precise locations of the transposons described in this work are not provided, we cannot exclude the possibility of insertion in a region of complex modular structure, as was already observed in other multiresistance plasmids (46). However, mercury resistance transposons identified in other IncA/C plasmids are flanked by direct repeats such as TTGTA in pCC416 and pSN254 and TACAA in pIP1202, indicating different acquisitions in distinct specific target sites (Fig. (Fig.2)2) (22, 57). Homologous and/or site-specific recombination events occurring within the plasmid or the transposon backbones seem to have played a key role in the evolution of this IncA/C plasmid group (45, 53, 58; this study). Although apparently diverse, the multiresistance regions described in this study and other studies consist of an assembly of a few highly modular genetic elements clustered in specific plasmid regions with potential for recombination (22, 44, 45, 46, 58; see below). They have also been observed in widely disseminated plasmids of the incompatibility groups FII, HI1, HI2, P1-α, P1-β, and W (GenBank accession numbers AP000342, AY458016, AJ851089, AF550679, AL513383, BX664015, NC_006352, and NC_004840) (44, 53).
In all isolates analyzed (3 E. coli isolates, 2 K. pneumoniae isolates, and 1 E. aerogenes isolate), blaTEM-24 was identified in a defective Tn1 containing the right inverted repeat (IR), tnpR, and a tnpA gene truncated at bp 1965. This transposon was inserted 465 nucleotides apart from the start codon of tnpA1696 (Fig. (Fig.1).1). A related configuration, IS5075-ΔTn1(blaTEM-3)-ΔtnpA1696, showing a complete deletion of tnpA1 was found in the IncL/M plasmid pCFF04 (GenBank accession number X64523) in which the Tn1 (blaTEM-3) was identified in the same nucleotide position of tnpA of Tn3926 or Tn5036 from which Tn1696 was presumptively generated after the insertion of In4 (30, 48). This structure might have eventually arisen by insertion of Tn1 in a tnp1696 module followed by further independent deletions on tnpA1. However, other hypotheses, like one involving a one-ended transposition event of an incomplete copy of Tn1 or a transposase complementation, have also been suggested (30). Interestingly, sequences associated with Tn1::Tn1696-like structures were also identified in other related plasmids from French clinical isolates producing TEM β-lactamases (TEM-7 and TEM-8) recovered during the 1980s (30). These and other variants (TEM-16, TEM-24, and TEM-66) are TEM-2 derivatives encoded by plasmids of similar sizes and RFLP patterns, conferring resistance to common antibiotics (kanamycin, tobramycin, amikacin, netilmicin, tetracyclines, trimethoprim, and sulfonamides) and showing identical genetic surroundings containing blaTEM genes (Tn1 derivatives and/or presence of aac(6)′-Ib) in different species (Table (Table1)1) (3, 10, 30). The blaTEM-24 gene identified in our study corresponded to the blaTEM-24b variant which was 99% homologous (100% identity at amino acid level) to that first identified in 1992 by Chanal et al. and initially designated blaCAZ-6 (GenBank accession number X65253) (9). It has been proposed that this gene arose by recombination between blaTEM-5 and blaTEM-8 variants (25), whereas the blaTEM-24a variant (blaCAZ-6) would have resulted from a recombination event between blaTEM-5 and blaTEM-3 (9). It is thus tempting to suggest a possible common origin for all these β-lactamases from a single insertion event of Tn1 containing blaTEM-2 and/or blaTEM-3 and further evolution by mutation and/or recombination of a variety of blaTEM genes in a given clinical setting (30).
Tn3 elements are often identified in identical genetic surroundings such as tnpA (this case) or mer (as described in pRMH760, p1658/97, and TnSF1) in plasmids of the incompatibility groups A/C2, L/M, and FII. The identical boundaries identified for Tn2-merT (pRMH760, p1658/97, and TnSF1) and Tn1-tnpA suggest single insertion events creating structures that have subsequently spread (22, 30, 46, 57; this study).
Our study indicates that the international dissemination of TEM-24 is driven by the confluence of highly penetrating clones of Enterobacteriaceae and a highly transmissible IncA/C2 plasmid, confirming the contribution of IncA/C plasmids in the global spread of antibiotic resistance. It further demonstrates the evolution of this plasmid lineage by acquisition of new different mobile genetic elements as described for other particular incompatibility groups (22, 46, 53).
The variability of Tn402 and Tn21 within the pRYC103T24 group illustrates the plasticity of these genetic platforms, and it highlights the need for full multilevel characterization of the genetic environments linked to antibiotic resistance genes in order to understand how they spread. This paper contributes to extending the knowledge on the spread, distribution, and evolution of Tn3-like elements, helping to elucidate the origin and evolution of ESBLs derived from TEM-2 β-lactamase.
Ângela Novais is supported by CIBERESP (Network Center for Biomedical Research in Epidemiology and Public Health) of Instituto Carlos III, Ministerio de Ciencia e Innovación of Spain. Elisabete Machado was supported by a fellowship from Fundação para a Ciência e Tecnologia de Portugal (SFRH/BD/11304/2002). This study was funded by research grants from the European Commission (LSHM-CT-2008-223031, KBBE-2008-2B-227258, and LSHM-CT-2005-018705) and the CIBERESP Network for Biomedical Research in Epidemiology and Public Health (Instituto Carlos III, Ministerio de Ciencia e Innovación of Spain, reference number CB06/02/0053).
We thank Catherine Branger (INSERM Université Denis Diderot, France) for supplying isolates corresponding to the French-Belgian TEM-24-producing E. aerogenes clone, Fernando de la Cruz and María del Pilar Garcillán-Barcia (Universidad Cantabria, Spain) for continuous support to the classification of plasmid conjugation modules, and Juan Imperial (E.T.S. Ingenieros Agrónomos y Centro de Biotecnologia y Genómica de Plantas, Universidad Politécnica de Madrid) for technical support with ARTEMIS software. We also are grateful to the Spanish Network for the Study of Plasmids and Extrachromosomal Elements (REDEXX) for encouraging and funding cooperation among Spanish microbiologists working on the biology of mobile genetic elements (Ministerio de Ciencia e Innovación of Spain, reference number BFU 2008-0079-E/BMC).
Published ahead of print on 7 December 2009.