Conjugative IncFI plasmids carrying CTX-M-15 among Escherichia coli ESBL producing isolates at a University hospital in Germany

Background Multi-drug-resistant, extended-spectrum β-lactamase (ESBL)-producing Enterobacteriaceae, constitute an emerging public-health concern. Little data on the molecular epidemiology of ESBL producing Escherichia coli is available in Germany. Here we describe the prevalence and molecular epidemiology of ESBL producing-Escherichia coli isolates at a German University hospital. Methods We analysed 63 non-duplicate clinical ESBL isolates obtained over an 8-month period using PCR and sequence-based ESBL allele typing, plasmid replicon typing, phylogenetic group typing. Pulsed-field gel electrophoresis (PFGE) based genotyping and plasmid profiling was performed, as well as confirmatory DNA-based hybridization assays. Results Examination of the 63 Escherichia coli isolates revealed an almost equal distribution among the E. coli phylogenetic groups A, B1, B2 and D. High prevalence (36/63) of the CTX-M-15 gene was observed and an analysis of PFGE-based patterns revealed the presence of this CTX-M allele in multiple clones. Resistance to cefotaxime was a transferable trait and a commonly occurring 145.5 kb conjugative IncFI plasmid was detected in 65% of E. coli carrying the CTX-M-15 allele. The rate of transferable antibiotic resistances for GM, SXT, TET, GM-SXT-TET, SXT-TET and GM-TET was 33%, 61%, 61%, 27%, 44% and 11%, respectively. The remaining strains did not have a common IncFI plasmid but harboured transferable IncFI plasmids with sizes that ranged from 97 to 242.5 kb. Conclusion Our data demonstrate the presence of IncFI plasmids within the prevailing E. coli population in a hospital setting and suggest that the dissemination of CTX-M-15 allele is associated to lateral transfer of these well-adapted, conjugative IncFI plasmids among various E. coli genotypes.

enterobacterial isolates express enzyme variants that are derived from TEM-1 and SHV-1 by mutations. Recently, different types of ESBL such as CTX-M have been detected. These enzymes hydrolyze cefepime with high efficiency and cefotaxime more efficiently than ceftazidime [3]. In clinical strains, CTX-M-encoding genes have commonly been located on plasmids which vary in size from 7 -200 kb [3][4][5]. A number of studies have established that most of these plasmids have either replicons that belong to the incompatibility group (Inc) FII, or multireplicons of Inc FII associated with Inc FIA and FIB [4][5][6][7][8]. The presence of an Inc FI plasmid replicon harbouring CTX-M-15 was reported in a single isolate from Turkey [8]. Associations with plasmids of other Inc groups such as Inc I1 and Inc N have also been reported [7]. Many of these plasmids are conjugative and have transfer frequencies ranging from 10 -2 -10 -7 . Additionally, they also encode multiple resistance genes for different antibiotics as described for various Inc FII plasmids isolated in UK and Canada [9]. To date, more than 60 different CTX-M ESBLs belonging to 5 evolutionary groups have been described. In most clinical isolates CTX-M-15 is the most frequent CTX-M type, and has been reported in Enterobacteriaceae isolates from Poland, Canada, France, UK, Russia, Cameroon, India, Bulgaria and Japan [3][4][5]. In Germany isolates harbouring CTX-M-1, -3, -9 have previously been described [3].
In the present study, we prospectively examined E. coli ESBL-producing isolates at a German University Hospital with respect to their phylogenetic type, PFGE pattern, plasmid incompatibility groups and their antibiotic susceptibility profiles in order to shed light into the epidemiology of these isolates in our clinical setting. We report here on the emergence of a cluster of CTX-M-15 producers in Germany exclusively associated with a large conjugative IncFI replicon-type plasmid.
All isolates resistant to multiple cephalosporins were confirmed for ESBL production using the double disk synergy method (disk approximation method) [10]. The MIC for cefepime and tigecycline in all isolates producing CTX-M alleles was determined using E-tests ranging from 0.016 to 256 μg/ml (AB Biodisk, Sweden) according to the manufacturer's instructions and the recommendations of the CLSI. Quality of media, antibiotic disks and E-test strips were controlled with the E. coli ATCC 25922 isolate. Isolates with a MIC of ≥ 8 μg/ml for cefepime and a MIC of ≥ 2 μg/ml for tigecycline were considered resistant according to the CLSI [10,11].

Pulsed-field Gel Electrophoresis (PFGE) and phylogenetic group typing
PFGE was performed according to the Pulse Net protocol of the Centers for Disease Control and Prevention, Atlanta, USA http://www.cdc.gov/pulsenet/proto cols.htm. The agarose-embedded DNA was digested with the restriction endonuclease XbaI (New England Biolabs, USA) at 37°C for 16 hrs. Electrophoresis was conducted using a CHEF Drive II (Bio-Rad, UK); conditions were 6 V, with 2.2 s-54 s pulses for 20 hrs. Strain differentiation by PFGE analysis was achieved by comparison of band patterns using Gelcompar II (Applied Maths, Belgium). Patterns were normalized using the molecular weight marker (PFGE Lambda Marker, Fermentas, Germany). Dendograms were generated to visualize relationships among the isolates. The similarity coefficient (SAB) of sample pairs was calculated based on band positions by using DICE metric [14,15]. A similarity coefficient SAB of 0.80 was set as a threshold for defining clusters of genetically similar isolates.

Plasmid analysis, DNA hybridization and PCR-based replicon typing (PBRT)
Plasmids from transconjugants and clinical isolates were detected using PFGE as described previously [4,5,17]. Genomic DNA was prepared as described above. A single block was incubated at 55°C for 45 minutes with 1 unit of S1 nuclease (Invitrogen, Germany) in 200 μl of 50 mM NaCl, 30 mM sodium acetate and 5 mM ZnSO 4 . Electrophoresis was done under the following conditions 6 V, 5 s-50 s for 20 h. DNA fragments were transferred to the polyvinyl-based membrane using overnight capillary transfer followed by hybridization with digoxygenin (DIG)-labeled CTX-M-15 FIA, and FIB amplicon probes prepared according to the manufacturers instruction (DIG High Prime DNA labelling and Detection Starter Kit II, Roche, Germany) [5].
PCR-based replicon typing was carried out as described by Carattoli et al [5,7]. DNA was extracted using a DNeasy tissue extraction kit (Qiagen, Germany) and 5 μl of template DNA (50 ng/μl) was used in simplex PCR to detect FIA, FIB, FII, I1 and N groups. PCR conditions were: initial denaturation at 94°C for 5 minutes followed by 30 amplification cycles comprising of 30 seconds denaturation at 94°C, 30 seconds annealing at 58°C, 60 seconds exten-sion at 72°C, followed by a final extension step at 72°C for 5 minutes. Escherichia coli strains ATCC 25922 and R 100 were used as negative and positive controls, respectively.
Sequencing PCR products of all ESBL genes detected as well as 6 randomly selected FIA-and FIB-amplicons were sequenced. PCR products were purified using Invitek purification kit (Invitek, Germany) following the manufacturer's instructions. Sequence reactions were done using the corresponding primers used for amplification, and sequencing was performed using the automated sequencer ABI Prism 3100 (Applied Biosystems, USA). DNA sequence was analysed using Lasergene software (DNASTAR, USA) followed by homology searches using the NCBI BLAST algorithm (Altschul SF et al. 1997).

Transfer of antibiotic-resistance genes
Conjugation experiments were performed using E. coli CC118 (Rif r , Str r , Lac-, plasmid-free) as a recipient strain [18] and 18 randomly selected clinical isolates of E. coli, representing different PFGE-based clusters as donor strains. Strains were mixed at the ratio of 1:2 (donor/ recipient) on LB agar followed by overnight incubation at 37°C. Transconjugants were selected by suspending the growth in 1 ml of PBS and 0.1 ml portions of 10 -1 to 10 -4 dilutions were plated on LB agar containing 300 μg/ml rifampicin and 30 μg/ml cefotaxime. The transconjugants were tested for ESBL production using disk approximation method followed by PCR amplification of ESBL genes and replicon typing.

Isolates and types of ESBL genes
A total of 63 non-duplicate clinically relevant isolates of E. coli were confirmed as ESBL-producing strains using the double disk assay. Thirty five of these isolates were recovered from urine (55.5%), 6 from blood culture (9.5%), 5 from sputum (7.9%) and 17 were from swabs (wound, eye, cervix, 27%). Twenty-nine of the isolates originated from general medical wards (46%) and 7 were from intensive care units (11.1%). Among the 63 phenotypically

Antibiotic resistance profiles
All isolates were sensitive to imipenem/cilastin and meropenem. Additionally, we tested for tigecycline and obtained MICs ranging from 0.5 μg/ml to 1 μg/ml with a median of 0.75 μg/ml indicating that all isolates were sensitive to tigecycline. All isolates carrying the CTX-M-15 allele were resistant to cefepime (MIC ≥ 8 μg/ml). The majority of isolates carrying CTX-M-15 were resistant to gentamicin, trimethoprim/sulfamethoxazole (SXT), tetracycline and ciprofloxacin when compared to other alleles (p = 0.006, chi-square test). There was a significant association (p = 0.00067, Fisher exact test) between resistance to cefepime and the presence of the CTX-M-15 allele, when compared to other CTX-M alleles data not shown.

Conjugation
The 18 randomly selected CTX-M-15-carrying E. coli isolates were all capable of transferring plasmids by conjugation with transfer frequencies ranging from 10 -4 -10 -9 per donor cell. Genes for tetracycline and SXT resistance were co-transferable in 61% and gentamicin in 33% of isolates. No transferable ciprofloxacin resistance was observed ( Table 1).

Characterization of isolates using PFGE and phylogenetic grouping
All of the 63 E. coli isolates were subjected to PFGE analysis. Examination of the PFGE patterns revealed 13 PFGE types (Figure 1). Using a similarity level SAB of 0.80 (line X in Figure 1), the 63 strains could be assigned to 6 clusters (clusters 1-6 in Figure 1) indicating the overall close relationship of the isolates. PCR-based phylogenetic analysis revealed that 28 (44.4%) of all isolates belonged to the B2 phylogenetic group, 20 were classified as group A (31.7%), 10 (15.8%) as D and 5 (7.9%) as B1. CTX-M-15 was found in 61% of group B1 isolates, in 50% of group D isolates, and in 60% of group A and B2, respectively. Thirteen (65%) of the group A isolates were obtained from urine samples. Phylogenetic group B2 formed the majority of the PFGE clusters 2 (15/21), 3 (3/4) and 5 (5/ 7) (Figure 1).

Plasmid analysis and replicon typing
Plasmids of variable sizes were detected, and the majority of transconjugants had plasmids ranging from 145.5 kb to 194 kb (Figure 2). A 145.5 kb plasmid was detected in 65% of the isolates tested and when present it was always found to hybridize with CTX-M-15, FIA as well as FIB probes, '[see Additional file 2]'. In one isolate the CTX-Mgene was located on a 242.5 kb IncF1 plasmid. All clinical isolates and transconjugants were found to harbour IncFI group plasmids. The majority of isolates (71.4%) were positive for both FIA and FIB. FIA and FIB alone were found in 10 (15.8%) and 8 (12.7%) isolates respectively. Plasmids harbouring Inc FII, IncN and IncI1 replicon types were not detected.

Discussion
This study provides molecular-epidemiological data on ESBL-carrying Escherichia coli in the clinical setting of a  CTX-M-15 ESBL producing isolates displayed a significantly (p = 0.0064) higher level of resistance to ciprofloxacin, gentamicin, tetracycline and trimethoprim/sulfamethoxazole (SXT) as compared to other ESBL alleles. Other studies have also reported co-resistance to tetracycline, aminoglycosides, fluoroquinolones in ESBL-producing organisms [9,12,18,19]. In this study gentamicin, tetracycline and SXT resistance were transferable by conjugation in more than 30% of isolates tested. Isolates harbouring CTX-M-15 were resistant to cefepime with most of them exhibiting a MIC of greater than 32 μg/ml which is significantly higher than that reported for other CTX-alleles (p = 0.00067). It has been demonstrated that CTX-M ESBLs hydrolyze cefepime with higher efficiency compared to other ESBL types [13]; this finding was confirmed in our study. Our data furthermore suggests that among the CTX-M-alleles, CTX-M-15 confers a higher ability to hydrolyze cefepime as compared to other CTX-M-types. In addition to carbapenems, all our ESBL isolates habouring CTX-M types were found to be sensitive to tigecycline with a MIC ≤ 2 μg/ml. This suggests that tigecycline could be a therapeutic alternative to carbapenems in cases of systemic infection due to ESBL producing organisms [11,20].
All phylogenetic groups of E. coli were observed in our study with group B2 representing the majority of our isolates [21]. Twenty isolates in our study were classified as phylogenetic group A. Among these 13 (65%) were from urine samples and 60% of them carried the CTX-M-15 allele. Bacteria from the phylogenetic group A are considered as commensals, but they have occasionally been associated with serious nosocomial infections. In this study, one group A CTX-M-15 positive isolate was recovered from a blood culture. In contrast to data reported in other studies, we cannot detect a significant bias in the distribution of CTX-M-15 among the different phylogenetic groups [22].
Analyses of PFGE banding patterns indicate a total of 55 genotypes among the isolates collected (Figure 1), suggesting that the spread of ESBLs is most probably linked to mobile genetic elements. Using a similarity coefficient (SAB) of 0.80, we were able to establish 6 different clusters (1-6) in our isolates. Cluster 2 was the largest cluster and was mainly formed by the B2 phylogenetic group with 67% of the isolates harbouring the CTX-M-15 allele. All isolates were typed for plasmid incompatibility groups and could be classified as FIA-and FIB-replicon types. Unlike most previously published studies [4,5,8], where an association of FII together with FIA and/or FIB was observed, no FII-replicon type could be detected in our isolates. To date there is only a single report from Turkey in which CTX-M-15 of an E. coli isolate was associated with an IncFI plasmid [8]. Plasmid analysis revealed that the majority of our transconjugants harboured large plasmids ranging from 145.5 to 194 kb. In 65% of the tested ESBL isolates, a common IncFI plasmid of about 145.5 kb could be demonstrated. One isolate carried a 242.5 kb IncFI plasmid harbouring CTX-M-15, which is the largest IncFI plasmid described in Escherichia coli to be associated with CTX-M-15. Most previous studies have found plasmids ranging from 7 to 200 kb in association with CTX-M-15 [3][4][5]. These findings suggest that in our setting, the CTX-M-15 allele is carried on large conjugative plasmids which are welladapted and constantly exchanged by lateral gene transfer among the E. coli isolates. The detection of other Enterobacteriaceae with the same ESBL type, as demonstrated in the above-mentioned cases of polymicrobial infections, furthermore suggest a species-overlapping transfer of CTX-M-15 (data not shown). Indeed all E. coli tested were able to transfer antibiotic resistance by conjugation with a frequency ranging from 10 -4 to 10 -9 per donor cell, indicating in several cases an at least 100-fold higher frequency compared to previous reports from the UK [18]. In addition CTX-M ESBL genes are often associated with an ISEcp1 element which facilitates its transfer. In this study the ISEcpl element was found in all isolates tested.
Agarose gel showing S1 nuclease PFGE-based sizing of large plasmids for 8 isolates

Conclusion
This study demonstrates the predominant presence of CTX-M-15 ESBL producing E. coli, commonly associated with a 145.5 kb IncFI plasmid, in our setting. This is the first report of a CTX-M-15 occurrence in Germany associated with large conjugative plasmids with exclusive IncFI replicon-type. This study furthermore illustrates that the prevalence of CTX-M-15 is not due to spread of a single clonal type but is associated with the spread of related E. coli isolates. Dissemination of the ESBL phenotype is linked to the lateral transfer of highly adapted IncF1 conjugative plasmid. Based on these findings, larger multicenter studies to determine the molecular epidemiology of E. coli ESBL isolates, the distribution of CTX-M ESBL as well as the presence of conjugative plasmids among Enterobacteriaceae in hospital populations are warranted.