Present addresses: Elena Grasselli, Michaela Gutaker and Cinzia Benagli, Dipartimento di Biologia, sezione di Fisiologia Generale e Comparata, Università degli Studi di Genova, Corso Europa 26 V piano, 16132 Genova, Italy. Maruska Convert, Interlifescience, Via San Gottardo 92, 6900 Massagno, Switzerland
One hundred and twenty clinical and commensal Escherichia coli strains isolated in Switzerland from humans and from companion and farm animals were analysed for the prevalence of integrons of classes 1, 2, and 3 and for the characterization of their gene cassettes. The relationships between integron carriage and host category, and between integron carriage and phylogenetic E. coli lineage were also analysed. Integrons were detected in 48 (40%) of the isolates and were thus widely disseminated in the human and animal E. coli strains considered. Moreover, the association between integron carriage and certain animal categories (farm animals) suggests that animals that are raised for economic purposes might be exposed to a major antibiotic pressure. Finally, our data confirm that E. coli commensal strains represent a significant source of antibiotic-resistant determinants.
The intensive use of antibiotics in both human and veterinary medicine, as well as in agriculture, has caused bacteria to develop resistance mechanisms against therapeutic drugs. Antibiotic resistance determinants are carried mostly by mobile genetic elements such as plasmids, transposons, and integrons. Integrons are genetic elements able to recognize and capture mobile gene cassettes carrying antibiotic resistance genes (Stokes & Hall, 1989). An integron includes a strong promoter, the gene for an integrase that catalyzes site-specific recombination, and a nucleotide sequence that acts as recombination site (attI). This specific sequence is adjacent to the integrase gene. Gene cassettes are discrete genetic elements, which are not necessarily part of an integron. They are usually found as linear sequences of 500−1000 bp constituting part of a larger DNA molecule, such as a plasmid or bacterial chromosome. Gene cassettes normally contain only a single gene and an additional short sequence, called the 59 base element (59-be or attC), that functions as a specific recombination site. The genes carried on these elements usually lack a promoter, and once the cassettes are integrated and become part of an integron their genes can be expressed from the upstream integron promoter (Fluit & Schmitz, 2004).
Three types of integrons have been identified and classified on the basis of the intI sequence. Most integrons from clinical isolates belong to class 1, and consist of two conserved segments. The 5′-conserved segment contains the intI gene (intI1), the attI site, and the common promoter Pant, whereas the 3′-conserved segment includes an antiseptic resistance gene (qacEΔ1), a sulphonamide resistance gene (sul1), and an ORF (orf5) of unknown function (Chang et al., 2000a, b). Class 2 integrons are part of the nonreplicative Tn7 transposon. They have a similar organization to class 1 integrons and carry three gene cassettes (dfrA1, sat2, and aadA1) close to an intI2 gene. The class 2 integrase IntI2 is related to the class 1 integrase (46% amino acid identity) (Hansson et al., 2002). Only one example of a class 3 integron has been identified and characterized (Collis et al., 2002).
Integrons are strongly associated with strains isolated from the clinical environment (Martinez-Freijo et al., 1998; Chang et al., 2000b; Yu et al., 2004). Isolates carrying integrons are not limited to pathogens, but have also been found in bacteria originating from healthy hosts and from environmental samples (Rosser & Young, 1999; Petersen et al., 2000). In the present study, we investigated the prevalence of integrons of class 1, 2, and 3 in Escherichia coli strains isolated in a number of Swiss institutes and hospitals from humans and from companion and farm animals. By means of DNA sequencing, we further characterized the inserted gene cassettes and investigated the potential association between integron carriage and host category or bacterial phylogenetic group.
Materials and methods
Bacterial strains and culture
The collection used for this study included 120 E. coli strains isolated from both humans and animals (dogs, cats, swine, cattle and poultry) presenting various pathologies, as well as commensal strains from healthy hosts (Table 2). The strains were provided by a number of institutes and hospitals in Switzerland.
Table 2. Description of the gene cassette arrays and phenotypic antimicrobial profiles of integron-carrying strains of various host origins and pathogenicities
For each strain, antibiotic susceptibility was tested for the following antibiotics: streptomycin, amoxicillin-clavulanic acid, ampicillin, nalidixic acid, trimethoprim/sulfamethoxazole, sulfonamide, tetracycline, gentamicin, nitrofurantoin, chloramphenicol, and trimethoprim. The susceptibility test was performed by disk diffusion on Müller–Hinton agar (Clinical and Laboratory Standards Institute, CLSI, 2006). The strains were classified into resistant and sensitive according to the indications of the CLSI. Intermediate isolates were considered nonsusceptible.
DNA extraction and PCR procedures to detect integron composition
DNA isolation from bacteria was performed with a DNeasy® Tissue kit (QIAGEN AG, Switzerland) according to the manufacturer's instructions. The 120 E. coli strains were screened for integron-associated structures. All isolates were analysed by PCR for the presence of the integrase genes of class 1, 2, and 3 integrons, intI1, intI2, and intI3 (Table 1). In addition, all strains were tested for the presence of sul1. Subsequently, the variable regions of each integron were analysed by additional DNA amplification using various primers (Table 1). Positive and negative controls were included in each assay [strain rl0484 DNA for intI1 and sul1 (Lanz et al., 2003), and recombinant plasmids for intI2 and intI3 (Arakawa et al., 1995)].
Table 1. Primers and annealing temperatures used for integron identification and integron structure investigation
Fragment length (bp)
Annealing temp. (°C)
F, forward; R, reverse; SEQ, sequencing; var, variable.
A triplex PCR was performed in order to differentiate the 120 E. coli strains genetically (Clermont et al., 2000). This PCR is based on the amplification of two genes (chuA and yjiA) and of one genomic DNA fragment (TSPE4.C2). The triplex PCR allows the allocation of the E. coli strains to one of four phylogenetic groups (A, B1, B2, and D), one of which (B2) is known for its potential to cause extra-intestinal infections (Clermont et al., 2000; Johnson & Stell, 2000).
Sequencing of PCR products
The PCR products obtained were purified using Amicon® Microcon®-Centrifugal Filter Devices (Millipore, Switzerland) and stored at 4°C. Cycle sequencing reactions were performed in a total volume of 15 μL using an ABI Prism Big Dye Terminator Cycle Sequencing Kit and analysed on an ABI Prism 310 Genetic Analyzer (Applied Biosystems). DNA sequencing was performed in both directions with the primers described in Table 1. DNA sequences were submitted to GenBank and are available under the accession numbers DQ915900−DQ915939.
DNA sequence analysis
The resulting sequences were handled and stored with the Lasergene sequence editor tool editseq 5.00 (dnastar, Madison, WI) and aligned with megalign (dnastar). The obtained sequences were then compared with those reported in GenBank (NCBI) by a blast search.
Characterization of the various integrons present in a single strain
A cloning strategy was developed in order to distinguish among the various integrons that may be present in a single strain. For this purpose, transformation assays of competent E. coli cells were performed using a TOPO TA cloning kit (Invitrogen AG, Switzerland). The source DNA was the DNA obtained by PCR from bacteria carrying more than one integron. Ten to 50 μL of the solution containing the transformed bacteria was spread on prewarmed selective Petri plates, which were incubated overnight at 37°C. The positive clones obtained were then amplified and sequenced with the primers provided with the kit (Invitrogen AG).
When needed, Fisher's exact test was used, with P<0.05 considered as significant.
Distribution of integrons in E. coli strains
Forty-four out of 120 E. coli strains (37%) carried a class 1 integron, and 41 (i.e. 34.2% of all strains) also carried the sulI gene (Table 2). Three (S10, TI05, TI87) of the intI1-positive strains did not carry the sulI gene. Amplification of the intI2 gene gave a positive PCR product in only four of the 120 strains (3.3%). Finally, no amplification product for intI3 was detected with the primers int3.F/int3.R (Table 1). No strain was found to contain both class 1 and 2 integrons.
Characterization of class 1 integron gene cassettes
The strains carrying the class 1 integrons were further submitted to DNA amplification with primers specific for the variable region of the integrons (Pant-F/qacEΔ1-B). We detected 37 single-band and four multiple-band PCR products from the 41 class 1 integron-positive strains (the three strains that did not carry sulI did not produce an amplicon, either). The DNA of each of the 37 bands was purified and sequenced with primers 5′-CS/3′-CS (Table 1).
Four strains showed multiple bands on the agarose gel. We analysed the DNA bands from two representative isolates (rl0205 and rl0407) by cloning and sequencing. For isolate rl0205, only two of the three fragments could be characterized. One fragment contained two cassettes (dfrA1 and aadA1), while the other had only one cassette (aadA1). For isolate rl0407, only one of the two fragments could be identified. This fragment also carried the dfrA1 and aadA1 gene cassettes.
Further analysis of the sequences showed that the 41 strains carrying class 1 integrons harboured eight distinct cassette arrays, with a total of 11 distinct gene cassettes (Table 2). The most common cassette carried genes encoding aminoglycoside adenylyltransferases genes (aadA), which confer resistance to streptomycin and spectinomycin. These elements represent 55% of all the cassettes we have found (43/78), and are subdivided into aadA1 (86%), aadA5 (12%), and aadA23b (2%). Dfr cassettes carry genes (dfrA1 and dfrA17) that encode dihydrofolate reductases, which confer resistance to trimethoprim. They represent 33% of the detected cassettes (dfrA1: 81%, dfrA17: 19%). Additional cassettes identified in class 1 integrons carried the genes estX and bla-oxa-30, respectively. The first gene encodes an esterase of unknown function (Hall & Partridge, 2005). The second encodes a rare β-lactamase that is closely related to OXA-1 β-lactamase (it differs by one amino acid). OXA-30, like OXA-1 β-lactamase, efficiently hydrolyses penicillins and early cephalosporins (e.g. cefalothin), but not the expanded-spectrum cephalosporins and aztreonam (Hanson et al., 2002). Finally, two strains (F04 and GE50) harboured an aacA4 gene cassette, encoding an aminoglycoside 6′-N-acetyltransferase that confers resistance to aminoglycosides, such as streptomycin and gentamicin. These two strains also carried a catB gene cassette (catB2 in strain F04 and catB3 in strain GE50). CatB encodes a chloramphenicol acetyltransferase.
Strains rl0552, BE10, and rl0110 carried an aadA1 streptomycin resistance gene cassette, but did not show resistance to streptomycin. Similarly, strain LA75 harboured a dfrA17 cassette, without expressing resistance to trimethoprim. F04 had a catB2 cassette, but was sensitive to chloramphenicol. Finally, isolates F04 and GE50 carried an accA4 cassette, but were sensitive to gentamicin.
Characterization of class 2 integron gene cassettes
The four E. coli strains carrying the class 2 integrons were analysed with primers hep51/hep74 (Table 1). An amplicon was obtained for three strains. All 3 strains had the same cassette array, which is usually found in Tn7 transposons: dfrA1, sat2, and aadA1 (Table 2). The sat2 cassette confers resistance to streptothricin by encoding a streptothricin acetyltransferase (Hall & Partridge, 2005).
Relationship between integron carriage and host category
The host origin of the E. coli strains was analysed in relation to the class 1 integron carriage, grouping farm animals (swine, poultry, and cattle) and companion animals (dogs and cats). It was found that 52% (25/48) of the strains isolated from farm animals carried class 1 integrons, whereas only 22% (7/32) and 30% (12/40) of the isolates from companion animals and humans hosted these genetic elements, respectively.
Relationship between integron carriage and phylogenetic group
The integron carriage was analysed in relation to the phylogenetic groups identified in previous E. coli studies (A, B1, B2, D) (Clermont et al., 2000). We observed that integrons were more prevalent in strains of phylogenetic group A than in those of groups B1, B2, and D (P=0.0147), which all show a similar integron distribution (Fig. 1). It is noteworthy that group A includes most commensal and diarrhoeagenic strains, while virulent extra-intestinal strains belong mainly to group B2 and, to a lesser extent, to group D (Clermont et al., 2000).
Integron promoter sequences
The genes integrated in the variable region of class 1 integrons are expressed from a common promoter located in a conserved segment upstream of the cassettes. Sequence analysis has revealed the existence of four versions of the integron promoters (Levesque et al., 1994). The strongest promoter is TTGACAN17TAAACT, found in transposon Tn1696 (GenBank accession number U12388). This version is present only in the class 1 integron of strain GE50. In most of the isolates of the collection we identified two other versions of the promoter: TGGACAN17TAAACT (in nine strains) and TGGACAN17TAAGCT (in 33 strains). In one case, strain F04, a mutation between the two regions of the promoter was observed.
Our results contribute to the observation that integrons are widespread. Thirty seven percent of the total number of isolates (44/120) were found to carry class 1 integrons. Previous reports have shown higher prevalences in other E. coli collections: 64.4% in isolates from swine with diarrhoea (Kang et al., 2005), 62% in intensive-care and surgical-unit isolates from hospitals in nine European countries (Martinez-Freijo et al., 1998), 59% in isolates from calf diarrhoea cases (Du et al., 2005), 52% in various clinical isolates (Chang et al., 2000a), and 45% in urinary-tract isolates (White et al., 2001). The difference between the prevalence value we have found and those from the other reports may be the result of the diverse selection pressures in the various environments considered. In addition, our E. coli collection was not homogeneous, but originated from a variety of hosts and pathologies, including commensal strains.
The fraction of strains carrying a class 2 integron was low, with only four isolates found to harbour it. This, too, is in contrast to other studies, for which higher proportions of class 2 integrons were observed (Goldstein et al., 2001; Grape et al., 2005; Sunde, 2005). As in the case of class 1 integron prevalence, this could be a consequence of the inclusion of E. coli strains originating from a variety of hosts and pathologies.
A number of cassette arrays had been previously detected only in Salmonella enteritidis and Pseudomonas aeruginosa, but not in E. coli (Villa et al., 2002). Here we report the presence in this last species of the cassette arrays aacA4-aadA1-catB2 and aacA4-catB3-dfrA1-orf (Table 2). This study is the first to observe the presence of such cassettes in E. coli.
We observed that a few strains, although they were carrying specific gene cassettes (e.g. aadA1, dfrA1, catB2 and aacA4), did not express the corresponding phenotypical resistance pattern. This may be a result of the distance between the cassette and the promoter that transcribes the cassette genes: when the distance is too large, the gene cassette might not be efficiently transcribed (Collis & Hall, 1995). Another reason could be the presence of a weak promoter version or the mutations we have observed in the region between the two specific sequences of the promoter (−10 and −35 boxes). Using a reporter gene assay such as the promoterless luciferase gene system should confirm the activity of these promoters. Alternatively, mutations upstream or within the gene might hinder its expression.
Figure 1 shows that the so-called nonpathogenic commensal strains (phylogenetic group A) represent an important reservoir of integrons, and consequently of antibiotic resistance gene cassettes. This is in agreement with the hypothesis that virulent strains may acquire these factors from commensal strains and survive in an environment where a high antibiotic pressure is present: interestingly, in such an environment the strains that belong to phylogenetic groups associated with lower intrinsic virulence seems to be favoured (Johnson et al., 2003). In addition, we observed that in phylogenetic groups B2 and D, which include most virulent extra-intestinal strains, the percentage of integron-positive strains is significantly lower.
Currently, in Switzerland animals are treated with antibiotics only upon drug prescription, following an inspection by a veterinarian. Nevertheless, we observed that poultry, cattle, and particularly swine show significant percentages of strains carrying class 1 integrons. In contrast, the corresponding percentages in cats and dogs were smaller. This is probably a result of the fact that farm animals that are raised for economic purposes are more exposed to an antibiotic pressure than the other animals. Indeed, the former are more frequently treated than companion animals. Additional studies with specific E. coli strain collections (including larger samples of E. coli strains isolated from poultry, cattle, or swine) are needed to validate this observation further.
We thank Dr Didier Mazel for the kind gift of E. coli strains carrying the intI2 and intI3 genes. This work was supported by a grant from the Swiss National Science Foundation (NRP 4049-63270) to JCP.