The study was approved by the Ethical Review Committee of the Institut de Recerca de l’Hospital de la Santa Creu i Sant Pau.

In a prospective study in 2012 (January to December) involving 12 hospitals in Catalonia, K. pneumoniae isolates resistant to β-lactams were collected, excluding those with a natural resistance pattern. Epidemiological data on patient gender, age, chronic diseases and treatment were collected in parallel. Only one strain per patient was included. The selected strains were resistant to any of the following β-lactams: cephalotin, cefoxitin, cefuroxime, cefotaxime, ceftazidime, cefepime, ertapenem, imipenem, aztreonam, amoxicillin/clavulanic acid and/or piperacillin/tazobactam.

Each hospital carried out identification and antibiotic susceptibility tests and provided epidemiological patient data. The participating hospitals were: Hospital Municipal de Badalona (HMB), Hospital de Barcelona (HB), Corporació de Salut del Maresme I la Selva (HC), Hospital General de Granollers (HGG), Hospital General Universitari de Catalunya at Sant Cugat del Vallès (HGC), Hospital General de L’Hospitalet (HGH), Hospital Sant Joan de Déu at Manresa (HSJDD), Hospital Sant Joan Martorell (HMLL), Hospital de Mataró (HM), Hospital Universitari de Vic (HGV), Hospital Universitari de Sant Joan de Reus (HUSJR) and Hospital de la Santa Creu i Sant Pau (HSCSP) (covering a population of approximately 250,000 inhabitants).

The patients were classified in three categories: nosocomial (the infection occurring after 48h of hospital admission), healthcare centre (resident in a healthcare centre) and community (no recent contact with a medical environment).

K. pneumoniae identification and antimicrobial susceptibility tests were performed in each hospital following routine laboratory methods, either manual or automated [MicroScan WalkAway (Siemens) and Vitek system (bioMérieux, Marcy l’Etoile, France)]. The susceptibility pattern to β-lactams, quinolones and aminoglycosides was obtained by the disc-diffusion method (Rosco Diagnostica A/S, Taastrup, Denmark) following Clinical and Laboratory Standards Institute (CLSI) criteria, as used routinely in the laboratory.11 The antibiotics used were: ampicillin (AMP), piperacillin (PIP), amoxicillin/clavulanic acid (AMC), piperacillin/tazobactam (TZP), cephalotin (CEF), cefoxitin (FOX), cefuroxime (CXM), cefotaxime (CTX), ceftazidime (CAZ), aztreonam (ATM), cefepime (FEP), ertapenem (ERT), imipenem (IMP), kanamycin (K), gentamicin (G), tobramycin (T), amikacin (A), netilmicin (Nt), neomycin (Nm), nalidixic acid (NAL), ciprofloxacin (CIP), and cotrimoxazole (SXT).

The then recommended Modified Hodge Test (MHT) was performed to detect carbapenemase activity, using imipenem according to CLSI criteria.11 The strains selected for analysis were resistant to any of the studied β-lactams. Strains with positive or weakly positive MHT results were included for the molecular characterization of carbapenemase resistance mechanisms.

The polymerase chain reaction (PCR) was used to detect the following genes in all studied strains according to their resistance phenotype: carbapenemases (blaOXA-48, blaVIM, blaSPM, blaIMP, blaGIM, blaSME, blaNMC, blaKPC, blaIMI and blaGES, blaNDM), ESBLs (blaTEM, blaSHV, blaCTX-M), acquired AmpC genes (acAmpC) (blaACC, blaCIT, blaEBC, blaDHA, blaFOX, blaMOX) and the penicillinase blaOXA-1, quinolones (qepA and qnrA, qnrB, qnrS, qnrC, qnrD) and aminoglycoside modifying enzymes (AME) (aac(3′)-IIa, aac(6)-Ib, aph(3′)-Ia, ant(2′)-Ia and aac(2′)-Ia).4,12,13

The blaOXA-48 genetic platform was determined by PCR using specific primers for IS1999, lysR and IS1R of Tn1999 and Tn1999.26,14 in all blaOXA-48-carrying strains.

All amplicon sequences were obtained by Sanger sequencing in an external genome service enterprise (Macrogen Europe, Amsterdam, The Netherlands).

Pulsed-field gel electrophoresis (PFGE) with genomic DNA digestion by the enzyme XbaI was performed with the CHEF-DRIII system (Bio-Rad, Hemel Hempstead, UK) and analyzed by BioNumerics 6.6 (Applied-Maths, Ghent, Belgium).2,12,15 The relatedness was calculated by the unweighted-pair group method using an average linkage (UPGMA) algorithm, with band similarity calculated using the Dice coefficient with a 2% optimization value and 1% tolerance. PFGE patterns were interpreted as previously described.11,12 MultiLocus Sequence Typing (MLST) was performed according to the Pasteur Institute website (www.pasteur.fr/recherche/genopole/P8/mlst)15 in 11 strains selected according to their PFGE pattern (A-E). For cluster E, we selected the majority and minority patterns consulting the resistance pattern.

Conjugation was performed in the aforementioned 11 isolates. Rifampicin-resistant-GFP E. coli VA6190 was used as a recipient,11 which also expresses a green fluorescent protein (GFP) marker. Luria-Bertani broth (LB) was used for the conjugation or, if LB results were negative, solid mating with HA 45μm-pore-size filters (Millipore, Billerica, MA). Putative transconjugants were selected on LB agar supplemented with ceftazidime (10mg/L) and rifampicin (100mg/L), and exposed to UV illumination to check for GFP fluorescence. The transfer frequency was expressed as the ratio of transconjugants to total recipient cells. The presence of blaOXA-48 in the transconjugants was confirmed by PCR.

Plasmid Identification was performed in 37 strains by a homemade PCR-based replicon typing (PBRT-PCR) method based on Carattoli et al. (2009).16 The studied strains represented each PFGE subcluster, adding some strains with the same PFGE pattern but different resistance patterns. Poirel et al. (2012)17 observed that PBRT-PCR did not detect a blaOXA-48-carrying IncL/M replicon. Carattoli et al. (2015)18 described specific primers to detect the blaOXA-48-carrying IncL plasmid, which distinguished IncL from IncM. We used these primers to describe the plasmid in our OXA-48-producing K. pneumoniae strains.

The plasmid localization of the bla genes was determined in the same 11 strains by PFGE using S1 nuclease and Southern blot hybridization methods, as previously described.12 The probes were obtained by the DIG Probe Synthesis Kit (Sigma–Aldrich) using primers corresponding to the bla and replicon genes.

Categorical variables were compared by the χ2 test and continuous variables by the Student's t-test or Mann–Whitney test. P-values of <0.005 were considered statistically significant. The software GraphPad Prism (GraphPad Software, Inc., CA, USA) was used for the analyses.

Between January and December 2012, we collected a total of 3901 K. pneumoniae isolates resistant to β-lactams, not including those with a natural resistance pattern. Among these, 171 (4.38%) gave positive MHT results, including strains with only a weak growth around the streak. And among these171 strains, 98 (57.3%) were resistant to ertapenem and 22 (12.8%) to imipenem. After the PCR and sequencing, 85 strains gave positive results for blaOXA-48, nine of them from faecal carriers. These 85 strains were from eight out of the 12 participating hospitals. Thus, the prevalence of OXA-48-producing K. pneumoniae strains in Catalonia was 1.9% (subtracting the nine faecal carriers). No other targeted carbapenemases were detected. The 85 OXA-48-producing strains were resistant to ampicillin, piperacillin and the association amoxicillin/clavulanic acid. Only 23.5% were resistant to imipenem, whereas 95.3% were resistant to ertapenem, which is the most sensitive drug for OXA-48-producing strain detection. Most strains were resistant to cephalosporins, in some cases remaining susceptible to cefoxitin and cefepime (Table 1). These different phenotypes were due to the presence of an ESBL, as 89.4% co-expressed the CTX-M-15, together with the penicillinases OXA-1 (94.1%) and TEM-1 (85.4%). All 85 isolates also expressed the chromosomal penicillinase SHV characteristic of this species: in 77.64% we detected blaSHV-76, 20% blaSHV-1, 1.1% blaSHV-11 and 1.1% blaSHV-42.

These strains were also resistant to aminoglycosides (90.5%), cotrimoxazole (89.4%) and ciprofloxacin (83.5%). Only four strains remained susceptible to all aminoglycosides and quinolones tested. Regarding aminoglycoside resistance, we found nine different phenotype patterns: KTG (64.7%), KTGN (14.1%), KTGNm (4.7%), KTA (1.1%), G (2.3%), KTAGN (1.1%), KTAN (3.5%), KT (2.3%) and KTN (1.1%). The AMEs present in these strains were AAC(3′)-IIa (83.5%), AAC(6′)-Ib (81.2%), APH(3′)-Ia (3.5%), AAC(2′)-Ia (2.4%) and ANT (2′)-Ia (1.2%). Finally, 88.2% carried QnrB and in one case QnrS (Table 2).

As mentioned above, 86 out of 171 positive MHT strains (50.3%) were non-carbapenemase-producing. Among these isolates, 17 (19.7%) were resistant to ertapenem and 1 (1%) to imipenem. To explain the positive MHT results, we checked for the presence of ESBLs and acAmpC. The PCR results revealed that 47 strains (54.6%) carried an acAmpC (80.8% DHA, 10.6% ACC and 8.6% CMY), 21 (24.4%) carried an ESBL (CTX-M-1-type) and 14 (16.3%) co-expressed both ESBL and acAmpC (64.3% CTX-M-1-type+DHA, 14.3% CTX-M-9-type+DHA, 14.3% CTX-M-1-type+ACC and 7.1% CTX-M-1-type with CMY). Finally, four strains (4.6%) did not show any studied acAmpC or ESBL. The 86 strains without carbapenemase production were excluded from further studies.

The origin of the 85 OXA-48-producing strains was: 47 (55.3%) from urine samples, 13 (15.3%) from respiratory tract samples, six (7%) from blood, four (4.7%) from surgical wounds, six (7%) from other samples (vaginal discharge, peritoneal fluid, sore, cellulitis, bile, post-operative ulcer) and nine (10.6%) were isolated from faeces (Table 3).

The demographic data collected from the 85 patients with OXA-48-producing K. pneumoniae infection showed that 52 were female (61.2%) of an average age of 76 years (range 26–98 years). There was no significant difference between genders (P>0.005). In 41 (48.2%) cases the infection had a nosocomial origin, 30 (35.3%) were related to a healthcare centre, and the remaining 14 (16.5%) were acquired in the community (Table 3).

The PFGE analysis revealed five well-defined clusters (A–E), which were subdivided in multiple subclusters, with 76% homology between the main clusters. Cluster A, B, C, D and E represented 18.8%, 1.1%, 1.1%, 1.1% and 77.6% of all isolates, respectively. Clusters A and E showed subclusters A1-3 and E1-18. The cluster distribution is depicted in Fig. 1.

Fig. 1.

Distribution of the different PFGE clusters by hospital. HC (Consorci de Salut del Maresme), HGC (Hospital General Universitari de Catalunya), HGG (Hospital General de Granollers), HGH (Hospital General de l’Hospitalet), HGV (Consorci hospitalari de Vic), HM (Hospital de Mataró), HSCSP (Hospital de la Santa Creu i Sant Pau) and HSJDD (Hospital Sant Joan de Déu de Manresa).

(0,07MB).

By MLST, the five PFGE clusters corresponded to five different K. pneumoniae sequence types (STs) as follows: cluster A belonged to ST101 (n=16); B to ST17 (n=1); C to ST1233 (n=1); D to ST14 (n=1) and E to ST405 (n=66). ST1233 is described here for the first time and is a single-locus variant (SLV) of ST540 in the gapA gene. Cluster E, which corresponds with ST405, was isolated in six of the eight OXA-48-detecting hospitals; cluster A, all ST101, was present in three of the eight hospitals. The minor clusters B (ST17), C (ST1233) and D (ST14), with one strain each, were from HGG, HM and HGV, and were isolated in different hospitals together with isolates belonging to major clones.

We observed that all OXA-48-producing K. pneumoniae strains, except those belonging to ST14, ST17 and ST1233, were resistant to multiple antimicrobial drugs (mainly aminoglycosides and quinolones) by acquiring resistance genes. blaOXA-48 seemed to be associated with blaOXA-1, aac(3′)-IIa, aac(6′)-Ib and qnrB, and in most cases with blaCTX-M-15 and blaTEM-1 (Table 2).

The PBRT-PCR data show that the 37 selected strains carried a plasmid with an IncL plasmid, 96% out of the 37 a plasmid replicon FIIK and 33.3% a plasmid replicon ColE (Table 4). Two cluster A-ST101 strains also carried one plasmid replicon R and the strain of cluster E16-ST405 showed the plasmid replicon FIA. The strains belonging to ST17 and ST14 were the only ones without plasmid replicons other than IncL.

PFGE and Southern blot studies revealed that that blaOXA-48 was present in all 85 OXA-48-producing strains in an approx. 62kb plasmid belonging to the IncL incompatibility group. These data were confirmed by conjugation assays. In all 11 conjugated strains we found OXA-48-producing E. coli transconjugants with a conjugation frequency between 1.3×10−5 and 5×10−7. All transconjugants with blaOXA-48 were resistant to penicillins, remaining susceptible to cefotaxime and ceftazidime and with a reduction in their susceptibility to ertapenem. In all cases the carbapenemase had transferred alone, as confirmed by PCR.

Southern blot experiments revealed that blaCTX-M-15 and blaOXA-1 were carried in FIIK plasmids of approx. 240Kb in ST101strains and approx. 290Kb in ST405.

The blaOXA-48-carrying genetic platform was related with the transposon Tn1999 (Table 4). Seventy-eight strains (91.8%) showed Tn1999.2 and seven the intact Tn1999. We sequenced the blaOXA-48 genetic surroundings in eleven randomly selected strains (Fig. 2) (GenBank accession numbers: KT265174, KT265175, KT265176, KT265177, KT265178, KT265179, KT265180, KT265181, KT265182, KT265183 and KT265173). In 10 cases a unique sequence between the IS1R element, upstream of blaOXA-48 and lysR, was found, with 100% homology with the Tn1999.2 sequence from pKpn-E1.Nr7 (KM406491.1). The remaining strain had 100% homology with K. pneumoniae E71T (KC335143.1) and pKPoxa-48N1 (KC757416.2). The only difference between the two groups was due to a transversion T-G at position 911, where the K. pneumoniae E71T (KC335143.1) and pKPoxa-48N1 (KC757416.2) sequences have a guanine.

Fig. 2.

Genetic backgrounds of the blaOXA-48 gene in our K. pneumoniae carbapenemase-producing strains (Genbank Access Numbers of fragment sequenced: KT265173, KT265174, KT265175, KT265176, KT265177, KT265178, KT265179, KT265180, KT265181, KT265182 and KT265183).

(0,09MB).