Clin Microbiol Infect 2010; 16: 171–178
Clinical isolates of Escherichia coli with reduced susceptibility to oxyimino-cephalosporins and not susceptible to clavulanic acid synergy (n = 402), collected from Norwegian diagnostic laboratories in 2003–2007, were examined for the presence of plasmid-mediated AmpC β-lactamases (PABLs). Antimicrobial susceptibility testing was performed for β-lactam and non-β-lactam antibiotics using Etest and Vitek2, respectively. The AmpC phenotype was confirmed using the boronic acid test. PABL-producing isolates were detected using ampC multiplex-PCR and examined by blaAmpC sequencing, characterization of the blaAmpC genetic environment, phylogenetic grouping, XbaI- pulsed-field gel electrophoresis (PFGE), multi-locus sequence typed (MLST), plasmid profiling and PCR-based replicon typing. For the PABL-positive isolates (n = 38), carrying blaCMY-2 (n = 35), blaCMY-7 (n = 1) and blaDHA-1 (n = 2), from out- (n = 23) and in-patients (n = 15), moderate-high MICs of β-lactams, except cefepime and carbapenems, were determined. All isolates were resistant to trimethoprim-sulphamethoxazole. Multidrug resistance was detected in 58% of the isolates. The genes blaCMY-2 and blaCMY-7 were linked to ISEcp1 upstream in 32 cases and in one case, respectively, and blaDHA-1 was linked to qacEΔ1sul1 upstream and downstream in one case. Twenty isolates were of phylogenetic groups B2 or D. Thirty-three XbaI-PFGE types, including three clusters, were observed. Twenty-five sequence types (ST) were identified, of which ST complexes (STC) 38 (n = 7), STC 448 (n = 5) and ST131 (n = 4) were dominant. Plasmid profiling revealed 1–4 plasmids (50–250 kb) per isolate and 11 different replicons in 37/38 isolates; blaCMY-2 was carried on transferable multiple-replicon plasmids, predominantly of Inc groups I1 (n = 12), FII (n = 10) and A/C (n = 7). Chromosomal integration was observed for blaCMY-2 in ten strains. CMY-2 is the dominant PABL type in Norway and is associated with ISEcp1 and transferable, multiple-replicon IncI1, IncA/C, or IncFII plasmids in nationwide strains of STC 448, STC 38 and ST131.
Plasmid-mediated AmpC β-lactamases (PABLs) have been reported worldwide in various Gram-negative bacteria since their first description in 1989–90 . PABLs are clinically important because they confer transferable resistance to all β-lactams except fourth-generation cephalosporins and carbapenems. In combination with loss of outer membrane porins, they may also mediate resistance to carbapenems . The occurrence of most PABLs appears to be sporadic, but nosocomial outbreaks caused by Klebsiella pneumoniae and Escherichia coli have also been reported [3,4].
PABLs have descended from chromosomal ampC genes of different bacteria, e.g. blaCMY from Citrobacter freundii and Aeromonas spp. and blaDHA from Morganella morganii . Capture and mobilization of blaCMY from the chromosomes of C. freundii and Aeromonas spp. to plasmids is thought to have been mediated by ISEcp1 and class 1 integron-bearing ISCR1 elements [5,6], subsequently spreading to new hosts by transferable IncA/C and IncI1 plasmids . Similarly, the DHA enzymes have also been linked to class 1 integrons-bearing ISCR1 upstream . DHA enzymes are the only inducible PABLs with a functional ampR regulator gene . Association of blaDHA with IncFII type plasmids has so far only been documented in a single study .
Escherichia coli contains a non-inducible chromosomal ampC gene, and phenotypic tests do not distinguish between over-expression of the intrinsic ampC and the presence of PABL. Thus, the prevalence of PABLs in E. coli is unknown as a result of the lack of detailed molecular characterization and population analysis of PABL-producing E. coli. Our recent study on selected Norwegian E. coli strains (from 2003–2005) with an AmpC-phenotype revealed the presence of strains over-expressing the chromosomal ampC gene as well as CMY-producing strains . A single study from France linked CMY-2 expressing strains to phylogenetic group B1 . The aim of this study was to describe the molecular epidemiology of the Norwegian PABL-producing E. coli strains from 2003–2007.
Materials and Methods
Strains and antimicrobial susceptibility
From March 2003 to December 2007, the Norwegian Reference Centre for Detection of Antimicrobial Resistance (K-res) received 402 clinical isolates of E. coli with reduced susceptibility to oxyimino-cephalosporins without clavulanic acid synergy from 21 nationwide diagnostic laboratories. Initial antimicrobial susceptibility testing was performed in each laboratory by routine methods interpreted according to the ‘Norwegian Working Group on Antibiotics‘ guidelines and EUCAST breakpoints. Isolates were examined at K-res using β-lactam Etests (AB Biodisk, Solna, Sweden) according to the manufacturer′s instructions for: ampicillin, amoxicillin-clavulanic acid, piperacillin, piperacillin-tazobactam, cefuroxime, cefoxitin, cefpodoxime, cefotaxime, ceftazidime, cefepime, aztreonam, imipenem and meropenem. Bacterial identification was performed using the Vitek2 ID-GNB system (bioMérieux, Marcy l’Étoile, France) or API ID32E (bioMérieux). Vitek2 ASTN023 (bioMérieux) was used to determine susceptibility to non-β-lactam antibiotics: nitrofurantoin, gentamycin, tobramycin, ciprofloxacin and trimethoprime-sulfamethoxazole. Multidrug resistance was defined as resistance to ≥2 non-β-lactam antibiotics. Synergy tests were performed by ESBL Etests (AB Biodisk) including cefotaxime, ceftazidime and cefepime with clavulanic acid. Phenotypic tests for AmpC β-lactamases were performed using boronic acid with cefoxitin . Selection criteria for analysis of plasmid-mediated ampC-genes were: (i) reduced susceptibility to third-generation cephalosporins (cefotaxime and ceftazidime MICs >1 mg/L) without clavulanic acid synergy, and (ii) positive boronic acid inhibition test. A total of 276 strains fulfilled these criteria.
Detection of bla genes
Isolates were screened for the presence of plasmid-mediated ampC genes using multiplex PCR covering the six families of blaAmpC-genes (blaACC, blaCIT, blaDHA, blaEBC, blaFOX and blaMOX) , and PCR for blaCTX-M . Amplicons were confirmed by bi-directional sequencing.
Genetic environment of blaAmpC-alleles
Population structure analysis
All isolates were classified according to E. coli phylogenetic groups A, B1, B2 and D using multiplex PCR , XbaI-pulsed-field gel electrophoresis (PFGE) typed , and multi-locus sequence typed (MLST) according to the scheme developed by Achtman (http://web.mpiib-berlin.mpg.de) . Purified PCR products were sent to Macrogen (Seoul, Korea) for DNA sequencing. Sequence alignment was performed using BioEdit sequence Alignment Editor  and BURST (Based Upon Related Sequence Types) analysis on allelic profiles online (http://linux.mlst.net/burst.htm).
Plasmids were typed with PCR-based replicon typing (PBRT) . S1-nuclease (Promega, Madison, USA) digested plasmid DNA was examined by PFGE [9,20], blotted onto positively charged nylon membranes (Roche Diagnostics GmbH, Mannheim, Germany) using the VacuGene™ XL Vacuum Blotting System (Amersham and Bioscience, Oslo, Norway) for transfer, and hybridized (68°C, 20 h) with probes of the blaCMY-2 gene and replicon FII, A/C and I1 (PCR DIG Probe Synthesis Kit, Roche Diagnostics).
Transferability of blaCMY
Randomly selected isolates (n = 10) were analysed for the ability to transfer blaCMY to rifampicin-resistant, plasmid free E. coli (J53-2). Conjugation was performed using blaCMY plasmids of both IncI1 and IncA/C types. Transconjugants were selected on Luria–Bertani plates containing cefotaxime (2 μg/ml) and rifampicin (100 μg/ml), confirmed by XbaI-PFGE, S1-nuclease-PFGE and antimicrobial susceptibility to nalidixic acid, cefotaxime, gentamycin, ciprofloxacin and trimethoprime-sulfamethoxazole by disc diffusion (Oxoid, Basingstoke, UK).
A total of 38 (14%) out of 276 E. coli isolates were positive for plasmid-mediated ampC-genes using multiplex-PCR. These strains were obtained from 13 nationwide laboratories, isolated from urine (n = 32), blood (n = 4), skin (n = 1) and abscess (n = 1) samples from in- (n = 15) and out-patients (n = 23). Results from individual isolates are summarized in Table 1.
|Ref. no.||Year of isolation||Hospitala/ sampleb/ patientc||blaAmpC||MLST||PFGE type/cluster||Phylogen. group||Co-resistancee||blaAmpC linked-element upstream||blaAmpC linked-element downstream||Plasmid number||Approximate size blaAmpC plasmid||Plasmid replicon typing||Plasmid transfer|
|ST||STCd||CIP||NIT||SXT||AG||MEC||Confirmed blaCMYf||PCR based|
|K34-41||2006||BD/B/I||DHA-1||23||ST23||XXIX||A||S||S||R||R||S||qacEΔ1 sul1||qacEΔ1 sul1||3||NT||–||I1-FII-FIB-Y||ND|
The minimum inhibitory concentration (MIC) means and ranges for piperacillin, piperacillin-tazobactam, oxyimino-cephalosporins, cefoxitin, cefepime and aztreonam within the different PABL types are presented in Table 2. All PABL-producing strains expressed moderate to high MIC levels for all substrates except cefepime. In all strains, MIC values were higher for cefuroxime and cefpodoxime than cefotaxime or ceftazidime. Co-resistance was seen for trimethoprime-sulfamethoxazole (n = 38; 100%), aminoglycosides gentamycin and/or tobramycin (n = 15; 40%), ciprofloxacin (n = 14; 37%) and nitrofurantoin (n = 12; 32%), with multidrug resistance observed in 23 (58%) strains (Table 1).
|Bla (n)||MIC (mg/L)b|
Characterization of PABLs and the genetic environment
PABL types identified were: blaCMY-2 (n = 35; 92%), blaCMY-7 (n = 1; 3%) and blaDHA-1 (n = 2; 5%). Thirty-two (91%) of blaCMY-2 and the single blaCMY-7 were linked to the ISEcp1 element upstream. A single blaDHA-1 was linked to the 3′CS region of a class 1 integron upstream and downstream as previously observed [6,8] (Table 1).
Isolates were distributed into potentially virulent phylogenetic groups B2 (n = 7; 18%) and D (n = 13; 34%) associated with extraintestinal infections, and groups A (n = 12; 32%) and B1 (n = 6; 16%) of commensal strains and strains associated with enterotoxigenic and enterohaemorrhagic infections.
A total of 33 different PFGE-types were detected among 37 typeable strains, (DNA from a single isolate autodigested, K26-21), including 30 single types and three clusters (>90% similarity) containing two to three isolates each (C1-C3). Clustered isolates were geographically related and of identical phylogenetic grouping in C1 and C3 (Fig. 1).
MLST analysis identified 25 different STs, including five novel STs (ST963, ST976, ST977, ST979 and ST981). A total of 29 isolates were related to 14 previously described ST-complexes (STCs) dominated by 38 (n = 7, 18%) and 448 (n = 5, 13%). Furthermore four isolates were sequence typed as ST131 (11%). Strains within these STs were distributed nationwide, isolated in different periods and included isolates showing diverse PFGE profiles (Table 1).
Isolates related to STC 38 included one isolate of ST38, four single locus variant (SLV) isolates of ST778, one double locus variant (DLV) isolate of ST963 and one SLV isolate ST981 of ST963. All isolates related to STC 38 were CMY-2 producers, grouped into phylogenetic group D and included PFGE C3 (ST778). Isolates related to STC 448 were of ST448, CMY-2 producers and of phylogenetic groups B1 (n = 3) and A (n = 2), including PFGE C2. Isolates related to ST131 were all of phylogenetic group B2 (Fig. 1).
S1-nuclease plasmid profiles were obtained for 36/38 (95%) isolates. DNA from a single isolate autodigested (K26-21) and no plasmids bands were observed for one isolate (K36-28). A total of 1–4 plasmids ranging from 50 to 250 kb were observed per strain. A total of ten single plasmid and 26 multiple plasmid isolates were identified.
PBRT identified 11 different replicons in 37/38 isolates; FII (n = 33; 89%), FIB (n = 33; 89%), I1 (n = 17; 46%), Y (n = 17; 46%), FIA (n = 15; 41%), A/C (n = 11; 27%), K (n = 5; 14%), P (n = 4; 11%), N (n = 2; 5%), B/O (n = 2; 5%) and L/M (n = 1; 3%). None of the 18 replicons screened for were detected in the isolate (K36-28) that showed no plasmids.
Hybridization with the blaCMY-2 probe identified a plasmid location in 25/35 CMY-producing isolates. In the remaining 10/35 isolates, the probe hybridized with a fragment >582 kb, consistent with a possible chromosomal location (data not shown). Chromosomally associated blaCMY were all linked to ISEcp1 upstream, suggesting a possible transposition event. When hybridized on plasmids, blaCMY was observed in five single- and 20 multiple-plasmid isolates. Single-plasmid isolates (n = 5) were positive for six replicon types by PCR, indicating up to five replicon types co-located on the same plasmid: I1 (n = 2), I1-A/C-FII-FIB-Y (n = 1), I1-FII-FIB-N (n = 1) and A/C-FII-FIB (n = 1). Alternatively, the multiple-replicon finding may represent plasmids undetected by the S1-nuclease method or chromosomally integrated plasmid remnants. The location of blaCMY with replicons I1, FII and A/C was confirmed by co-hybridization with the respective probes in all isolates. Multiple-plasmid isolates (n = 20) included ten different replicons with up to six replicon types in single isolates: A/C-FII-FIB-Y (n = 4), I1-FII-FIB (n = 3), FII-FIB-Y-K (n = 2), I1-FII-FIB-Y (n = 1), A/C-FII-FIA-FIB-Y (n = 1), A/C-FII-FIA-FIB-Y-P (n = 1), I1-FIA-FIB-Y (n = 1), I1-FII-FIA-FIB (n = 1), I1-FII-FIA-FIB-P (n = 1), I1-FII-FIA-FIB-Y (n = 1), I1-FII-FIB-N-K (n = 1), I1-FII-FIB-P (n = 1), I1-FII-FIB-P-K-B/O (n = 1) and Y-K-B/O (n = 1). Co-hybridization of blaCMY and replicons I1, A/C and FII probes confirmed blaCMY-linked replicons; I1 (n = 6), A/C-FII (n = 3), I1-FII (n = 2), A/C (n = 2), FII (n = 2) and none of the aforementioned replicons (n = 5) (Table 1).
Transferability of blaCMY
Transfer of blaCMY was observed in six out of ten isolates. Transfer frequencies ranged from 2.6 × 10−6 to 3.4 × 10−3/donor cells. Transconjugants were confirmed for blaCMY-2 (n = 5) and blaCMY-7 (n = 1) on both IncI1- and IncA/C-type plasmids. Transconjugants of blaCMY-7 (IncI1) and blaCMY-2 (IncA/C) donors K2-67 and K5-41, respectively, co-transferred resistance to SXT (Table 1). Transfer was not observed for one donor strain (K05-63) with a chromosomally located blaCMY-2.
We have carried out a broad molecular epidemiological characterization of PABL-producing E. coli collected during 2003–2007 in a nationwide laboratory based study. For 2007, the Norwegian surveillance system for antimicrobial resistance (NORM) calculated the prevalence of resistance to cefotaxime and/or ceftazidime in blood culture E. coli isolates to be 2.7% (n = 32) . The same report verified 1.2% as ESBL producers and only 18 (1.5%) as non-ESBL isolates. Given that most of the non-ESBL isolates have an AmpC-phenotype and that only 14% of our nationwide AmpC-phenotype strains were PABL positive, we infer that PABL-producing E. coli are rarely encountered in Norway.
In our study, we observed a dominance of CMY-2 type plasmid-mediated AmpC (92%), consistent with worldwide observations . Additionally, we identified a low presence of DHA-1 enzyme (5%), which has mostly been reported from Asia . The CMY-2-producing isolates generally expressed high mean MIC values against oxyimino-cephalosporins (except cefepime). Compared with the AmpC-phenotype of PABL-PCR negative E. coli, the mean MICs against oxyimino-cephalosporins were generally higher (data not shown), but a large overlap existed between the two groups. PFGE-analysis revealed that the majority of CMY-2-producing isolates were genetically unrelated, and only three minor clusters were observed. Moreover, most PABL-producing strains were distributed temporally and geographically. Thus, the occurrence of these strains in Norway seems to be sporadic.
BlaCMY-2 was seen on large multi-replicon IncI1 and IncA/C plasmids often co-located with other F-replicons, e.g. FII and/or FIA. Furthermore, IncI1 and IncA/C plasmids were shown to be transferable, co-transferring SXT resistance . The linkage of blaCMY-2 to ISEcp1 has been documented in numerous studies and three different truncated forms of ISEcp1-like elements upstream have been identified . The linkage of ISEcp1 to blaCMY was established in 91% of the isolates, as previously shown in Norwegian blaCMYE. coli isolates , supporting the notion that ISEcp1 is the probable mediator for mobilization and high-level expression of blaCMY as observed for blaCTX-M .
The recently observed worldwide clonal dissemination of E. coli ST131 producing CTX-M-15 encouraged us to investigate PABLs at the strain level. Population analysis of the CMY-2-producing PABL strains revealed that 16/35 (46%) of the isolates were related to internationally disseminated strains of STC 38, STC 448 and ST131. To our knowledge, no MLST analysis of PABL-producing strains has previously been described.
Present data from the MLST database show that strains related to STC 38 consist of globally distributed uropathogenic and enteroaggregative E. coli isolates from Brazil, Germany, India and Nigeria. Furthermore, ST38 strains have recently been identified in Japan among CTX-M-producing isolates . Our STC 38-related strains (n = 7) belonged to the phylogenetic group D and included the SLV ST788 isolates of the PFGE C3. Furthermore, all ST788 isolates (n = 4) and one ST38 isolate were identified with blaCMY-2 linked to an ISEcp1 element upstream. Interestingly, hybridization with blaCMY-2-specific probes identified a possible chromosomal location in these five isolates, and a plasmid location in the two remaining isolates. A possible chromosomal integration of blaCMY-2 in E. coli has previously been suggested [9,24] and reported for Proteus mirabilis .
Strains submitted to the MLST database related to STC 448 are mostly pathogenic E. coli isolated from Germany, Scotland, India, Thailand and Nigeria. Our STC 448-related isolates were all ST448 with blaCMY-2 linked to an ISEcp1 element upstream (n = 5). Three isolates were of phylogenetic group B1 and two of group A. Two isolates clustered in PFGE C2. Four isolates, including the C2 isolates, were also shown to have a possible chromosomal location when hybridized with a blaCMY-2 probe.
Recently, ST131-related E. coli strains have been associated with the global dissemination of CTX-M-15 type ESBL [28,29]. Furthermore, ST131 strains have been described as the major extended spectrum cephalosporin-resistant uropathogenic strains and plasmid-mediated quinolone resistance strains in the UK [30,31] and ciprofloxacin-resistant strain in eight European countries . They have also been isolated from the stools of healthy individuals in France . Taken together, these observations suggest that ST131 had a worldwide dissemination even before the acquisition of ESBL resistance determinants. In this study, a total of four isolates were ST131. They were all CMY-2 producers belonging to phylogenetic group B2 consistent with the documented ST131 reports . However, none of the PABL-producing isolates were positive for blaCTX-M by consensus PCR. Moreover, the observations of low cefepime MICs in combination with a lack of clavulanic acid synergy in ESBL Etests strongly indicate that Norwegian PABL-producing E. coli strains are not associated with ESBL production.
This study documents blaCMY-2 on large broad host-range conjugative plasmids linked to multidrug resistant lineages of E. coli across Norway; STCs 38, 448 and ST131. The presence of multidrug resistant ST131 strains requires special attention and monitoring. These observations support long-term persistence in colonized hosts which facilitates further dissemination even in a country such as Norway with a relatively low usage of antimicrobial agents.
This study has been a national collaborative study coordinated by the Reference Centre for Detection of Antimicrobial Resistance, University Hospital (UH) of North Norway. We thank all Norwegian clinical microbiology laboratories for providing strains.
This study was funded in part by the Northern Norway Regional Health Authority Medical Research Program. The authors declare that they have no conflict of interests.