Müeller-Hinton (MH) broth and Müeller-Hinton agar (MHA), phosphate buffered saline (PBS), and all powder-based antimicrobial compounds and reagents, with exception of moxifloxacin, were obtained from Sigma-Aldrich (Ireland). Moxifloxacin was purchased from Santa Cruz Biotechnology (Spain). All the diffusion disk antibiotics were purchased from Oxoid (Thermo Fisher Scientific).

Eleven K. pneumoniae clinical isolates obtained in 2008 from a major tertiary teaching hospital, in Dublin were used in this study. The control isolate Escherichia coli ATCC™25922 was used in antibiotic susceptibility tests and the isolate E. coli ATCC™8739 was used for MALDI-TOF internal control. Similarly, K. pneumoniae MGH 78578 (ATCC™700721) also included as control for MDR K. pneumoniae. Salmonella Typhimurium ATCC™14028 was used as positive control for biofilm formation and RADR morphotype.

Bacterial growth was measured in MH broth. Briefly, overnight cultures of K. pneumoniae were washed twice with PBS and diluted in MH broth to an OD₆₀₀ 0.005. Two hundred microliters of each diluted isolate was pipetted into a 96-well plate and absorbance was measured over a 24 h period, using a Multiskan™ FC Microplate Photometer (Thermo Scientific, IE) at 620 nm, incubated at 37°C. Each isolate was cultured in triplicate with five technical replicates for this assay.

The antibiotic susceptibility of each Klebsiella isolate was determined by disk diffusion (Bauer et al., 1966). In brief, bacterial cultures were grown overnight and diluted in sterile PBS to 0.5 McFarland standard and spread, using a cotton-tipped bud across dried MHA-plates. Disks containing amikacin (AK—30 μg), aztreonam (ATM—30 μg), cefepime (FEP—30 μg), cefotaxime (CTX—5 μg), ceftazidime (CAZ—10 μg), chloramphenicol (C—30 μg), ciprofloxacin (CIP—5 μg), doripenem (DOR—10 μg), doxycycline (DO—30 μg), ertapenem (ETP—10 μg), gentamicin (CN—10 μg), imipenem (IPM—10 μg), levofloxacin (LEV—5 μg), meropenem (MEM—10 μg), minocycline (MH—30 μg), moxifloxacin (MXF—5 μg), nalidixic acid (NA—30 μg), norfloxacin (NOR—10 μg), penicillin (P—10 iu), tetracycline (TET—30 μg), ticarcillin (TIC—75 μg), ticarcillin-clavulanic acid (TIM—85 μg), tigecyline (TGC—15 μg), and trimethoprim-sulfamethoxazole (SXT—25 μg) were included. Inoculated plates, containing the disks above were incubated at 37°C and any zone of inhibition formed thereafter, was measured after 16–18 h. Results were interpreted as susceptible, intermediate or resistant according to breakpoints as defined by CLSI, M100-S23 document (CLSI, 2013). In the case of moxifloxacin and tigecycline the breakpoints used were those from The European Committee on Antimicrobial Susceptibility Testing—EUCAST (2016). This assay was performed in triplicate for each clinical isolate.

MIC values for cefotaxime (CTX), chloramphenicol (CHL), ciprofloxacin (CIP), colistin (CT), gentamicin (CN), imipenem (IMP), kanamycin (KM), moxifloxacin (MXF), nalidixic acid (NAL), polymyxin B (PB), tetracycline (TET) and Benzalkonium chloride (BK) were measured by two-fold broth microdilution in a 96-well microtiter plates. Briefly, overnight cultures were diluted in sterilized PBS to ~10⁵ CFU/mL. Aliquots of 5 μL were then transferred to separate wells in a 96-well plate that contained 100 μL of each compound at varying concentrations prepared from two-fold serial dilutions in MH broth. The concentration ranges used were 1–512 μg/mL, with the exception of moxifloxacin wherein the concentration ranges were 0.25–128 μg/mL. These plates were incubated at 37°C and the MIC values recorded after 16–18 h. Susceptibility or resistance was determined with reference to the CLSI breakpoints (M100-S23). In the case of colistin, polymyxin B, and moxifloxacin the breakpoints used were those specified by The European Committee on Antimicrobial Susceptibility Testing—EUCAST (2016).

Determination of the MBC values for all antimicrobial agents tested above was performed in MH broth media. Five microliters were transferred from the MICs 96-well plates (above) and re-inoculated into fresh sterile 96-well plates containing MH broth, without any of the antimicrobial compounds. Plates were incubated at 37°C and the MBC results recorded after 16–18 h. All assays were carried out three times for each antibiotic.

Genomic DNA was purified from overnight cultures using the UltraClean® Microbial DNA Isolation Kit (MoBio Laboratories, CA) according to manufacturer's instructions. Small plasmid DNA (sizes < 30-kbp) extraction was carried out using the High Pure Plasmid Isolation Kit (Roche, SH). E. coli 39R 861, and E. coli V517 were used as molecular weight controls for plasmid sizing (Wang et al., 2013). Purified DNA was resolved in a 0.8% [w/v] agarose gel, stained with SYBR green.

Large plasmid extraction was carried out using S1-nuclease (Promega, Madison, WI, USA) followed by pulsed-field gel electrophoresis (PFGE; Barton et al., 1995). E. coli 39R 861 and E. coli V517 were included as controls. Briefly, the procedure included a lysis step for the bacterial cells, previously embedded in agarose plugs followed by digestion with 8 U S1-nuclease at 37°C for 45 min. Finally, each plasmid sample was resolved by PFGE in a Chef-Mapper® XA System (Bio-Rad, USA) at 14°C, with a switch time between 1 and 12 s, at 6 V/cm on a 120° angle in 0.5 × TBE buffer for 18 h in a 0.8% [w/v] agarose gel and stained with SYBR green. The approximate molecular mass of plasmids was determined by comparing band molecular sizes with the two control isolates and by using Salmonella Braenderup H9812 digested by XbaI (Wang et al., 2013).

Isolate identification was carried out based on the phenotypic growth pattern for different carbon sources using the GEN III MicroPlate™ platform (Biolog Inc., CA, USA). Three single colonies were collected from an overnight culture in MH-agar and mixed with inoculating fluid B (IF-B) prior to being adjusted to a transmittance of 90–98% using a turbidimeter, as specified in the user guide. For each bacterial isolate, 100 μL of the cell suspension was inoculated into each well of the MicroPlates and incubated at 37°C for a period of 20 h. Results were interpreted using the OmniLog® identification system's software (GEN III database, version 5.2.1).

Bacterial isolate identification was further confirmed with a Matrix Assisted Laser Desorption Ionization Time-of-Flight (MALDI-TOF) system based on the VITEK® MS platform (bioMérieux, Marcy l-Etoile, France). Half of a single colony was picked from an overnight culture in MH agar using a 1 μL sterile loop and placed on the target slide. When dry, the spot was covered with 1 μL α-cyano-4-hydrocinnamic acid (CHCA) matrix (bioMérieux, Marcy l-Etoile, France). E. coli ATCC™8739 was used as an internal control. Results were analyzed using Myla®, software database for microbial identification.

In order to quantitatively assess the efflux ability of the clinical isolates, ethidium bromide was used as described previously (Paixão et al., 2009) with several modifications. Briefly, K. pneumoniae overnight cultures were grown in MH broth until mid-log phase. Bacterial cells were washed twice with PBS and OD_{600 nm} readjusted to 0.6. Cells were incubated in PBS in the presence of EtdBr (50 μM) and CCCP (100 μM) at 25°C for 1 h. After that time, cells were washed with PBS, centrifuged and the contents re-suspended in (i) PBS and (ii) PBS with 50 μM CCCP. The cells were immediately pipetted into a 96-well plate and ethidium bromide fluorescence was measured in a Fluoroskan Ascent FL (Thermo Scientfic, IE) with an excitation and emission wavelengths of 518- and 606-nm, respectively. After 3 min, glucose was added to one of the well containing PBS to a final concentration of 50 mM. Fluorescence emitted was acquired in cycles of 60 s, during 50 min at 37°C. Each bacterial isolate was tested in triplicate for this assay. Results were averaged and normalized based on the fluorescence collected from the respective samples in PBS. Finally, results were plotted in terms of relative fluorescence.

Biofilm formation was examined under defined growth conditions that experimentally approximate different stress conditions. M9 minimal media (containing NH₄Cl [1.9 mM], Na₂HPO₄ [42.3 mM], KH₂PO₄ [22 mM], NaCl [8.56 mM], MgSO₄ [2 mM], CaCl₂ [0.1 mM], and glucose [0.1% w/v]) and MH broth were used for this purpose to enable an assessment of biofilm formation capacity, when cultures were incubated at 25 and 37°C respectively for periods ranging from 8 to 96 h. Initially, overnight cultures were adjusted to OD₆₀₀ 0.3 with fresh media, and 200 μL of this bacterial cell suspension was dispensed across a 96-well microtiter plate. Plates were incubated statically for 8–96 h under the same conditions as described above. Salmonella Typhimurium ATCC™14028, a proven strong biofilm forming isolate, was used as a positive control (Finn et al., 2013).

Crystal violet was used to detect total biofilm biomass formed (Stepanović et al., 2007). Briefly the 96-well plate containing culture medium was discarded after growth and the plate was washed with PBS. Crystal violet (0.4% [v/v]) was added and incubated at room temperature for 15 min. The plate was then washed again with PBS and dried in a sterile air flow hood for 20 min. Once dried, 200 μL 33% [v/v] acetic acid was added to each well and the absorbance was recorded in a Multiskan™ FC Microplate Photometer (Thermo Scientific, Dublin, Ireland) at 570 nm. This assay was performed in triplicate for each isolate with 5 technical replicates for each biological replicate.

Expression of curli fimbriae was assessed by examining colony morphology of K. pneumoniae when grown on Congo red agar. Congo red agar was prepared as described by Römling and Rohde (1999). In brief, MH agar was supplemented with 40 μg/mL Congo red (Sigma-Aldrich, Ireland) and 20 μg/mL Coomassie brilliant blue R-250 (Thermo Fisher Scientific, Waltham, MA). Plates were then inoculated with 3 μL of each overnight bacterial culture grown in MH broth and then incubated for 72 h at 25 and 37°C after which colony morphology was inspected and recorded. Salmonella Typhimurium ATCC™14028 was included as a control for a RDAR phenotype (Römling and Rohde, 1999; Finn et al., 2013).

Calcofluor agar was used to assess these isolates for their ability to produce cellulose (Zogaj et al., 2001). In this case, the plates were prepared by supplementing MH agar with 40 μg/mL calcofluor fluorescent brightener 28 (Sigma-Aldrich). The calcofluor agar plates were inoculated with 3 μL of overnight culture and incubated for 72 h at 25 and 37°C. Salmonella Typhimurium ATCC™14028 was included as a control. The colony morphology was observed and recorded under a 366-nm UV light source to detect the binding of calcofluor to the cellulose produced. Both assays were carried out in triplicate for each isolate.

To examine pellicle formation, 50 μL from an overnight in MH broth was used to inoculate 5 mL MH broth, which was then incubated statically for 96 h at 25 and 37°C after which it was then inspected for the formation of a pellicle at the air-broth interface (Finn et al., 2013). The assay was repeated twice and results were registered as present or absent.

Genomic libraries were prepared commercially (Novogene, Beijing, China) using the Illumina TruSeq library preparation kit for each of the 11 K. pneumoniae and sequenced on the HiSeq platform (Illumina). The quality of the reads was assessed using FastQC (version 0.11.5; Marçais and Kingsford, 2011). Error correction was performed using BFC (version r181; Li, 2015). A relaxed quality trim was performed using Trimmomatic (version 0.36) before the genome was de novo assembled using SPAdes (version 3.9.1; Bankevich et al., 2012; Bolger et al., 2014). The quality of the subsequent assemblies was assessed using Bandage (version 0.8.1) and QUAST (version 4.3; Gurevich et al., 2013; Wick et al., 2015). Assemblies were annotated using Prokka (version 1.11; Seemann, 2014).

Antimicrobial resistance genes were identified using Resfinder 2.1 (version 2016-11-02; https://cge.cbs.dtu.dk/services/ResFinder/; Zankari et al., 2012). Plasmid replicon typing was performed using PlasmidFinder 1.3 (https://cge.cbs.dtu.dk//services/PlasmidFinder/; Carattoli et al., 2014). Antibacterial biocide and metal resistance genes were using BacMet 1.1 (http://bacmet.biomedicine.gu.se; Pal et al., 2014) and virulence genes were determined using the VFDB (http://www.mgc.ac.cn/VFs/; Chen et al., 2016) and BIGSdb (http://bigsdb.web.pasteur.fr/klebsiella/; Bialek-Davenet et al., 2014). Sequences from these databases were identified within the genomes of all isolates using BLAST+ (version 2.5.0; Morgulis et al., 2008) and Biopython (version 1.68; Cock et al., 2009). Multilocus sequence typing (MLST) was performed in silico using the K. pneumoniae database available from Pasteur Institute (http://www.pasteur.fr/mlst). K-loci were identified using Kaptive (Wyres et al., 2016).

Raw sequencing data has been deposited in the SRA under the BioProject PRJNA396729, Table S1 indicates the SRA number for each isolate.

Amino acid substitutions in the quinolone resistant-determining regions (QRDR), master regulators, RND efflux pump acrAB-tolC, outer membrane proteins, biofilm and cellulose genes (Bruchmann et al., 2015) were identified by comparing the genes against a representative of each ST type. Genomes were extracted from Pasteur database (http://bigsdb.pasteur.fr/klebsiella/klebsiella.html) and their ID references are as follows: ST340-ID1558; ST15-ID1698; ST14-ID1526; and ST147-ID3031.

A collection of 11 K. pneumoniae, of clinical origin were cultured in 2008, from a major tertiary teaching hospital in Dublin. These were initially characterized for their growth ability to grow in presence of a rich culture medium, Müeller-Hinton (MH) broth. Based on their growth curve profiles (Figure 1) all were capable of grow approximately at the same rate with generation times ranging between 49 and 51 min with isolate K. pneumoniae CFS0364 showing a less pronounced stationary phase. On occasion, clinical isolates may be misidentified when using traditional laboratory diagnostic methods. In order to confirm the bacterial genus, all of the 11 isolates were inoculated into GEN III MicroPlates™ and incubated in a Biolog platform (Biolog Inc., CA, USA), for identification. These data presented in Table 1 identified 5 of the 11 isolates (45% of the collection) as K. pneumoniae, the remaining isolates could not be reliably identified. The principles underpinning the Biolog identification protocol are based on the ability of bacterial isolates to metabolize a defined set of substrate compounds. Activation of certain metabolic pathways is related to the stress environment in which bacteria can be found. Therefore, clinical isolates may express a phenotype that does not necessarily lend itself to identification, under these conditions.

To further resolve this identification, all isolates were subjected to further analysis by MALDI-TOF mass spectrometry using the VITEK® MS platform. This strategy is faster and more accurate when compared with Biolog, results obtained are presented as in Table 1. These data indicated that all these clinical isolates were identified as K. pneumoniae with 99.9% confidence.

To extend the characterization of these K. pneumoniae their susceptibility to a panel of antimicrobial compounds was studied. These assays were performed using both disk diffusion including a set of 26 antimicrobial compounds containing representatives of each class and broth microdilution. Based on the resistance profiles obtained following disk diffusion, the study collection of K. pneumoniae demonstrated two distinct clusters (Figure 2). All the isolates were found to be resistance to β-lactam-based compounds including, cephalosporins, along with quinolones and fluoroquinolones. Resistance to chloramphenicol accounted for 64% (n = 7) of the collection and trimethoprim-sulfametoxazole resistance was detected in 9 of 11 isolates. One of the two clusters identified (including the isolates K. pneumoniae CFS0363, CFS0369, CFS0365, CFS0370, CFS0366, and CFS0368) was found to be resistance to tetracyclines, aminoglycosides (with exception of the isolate K. pneumoniae CFS0368), and some of the carbapenem compounds tested. The second cluster, comprising of the remaining isolates, was mainly characterized by its susceptibility to aminoglycosides with the exception of K. pneumoniae CFS0364 that was found to be resistant to gentamicin. In contrast to the above, the latter cluster was also notable by its susceptibility to tetracyclines.

In the case of the three carbapenem compounds tested, all isolates were susceptible, with the exception of K. pneumoniae CFS0372 which was found to be resistant to ertapenem and also demonstrated intermediate resistance to meropenem and doripenem.

To further validate the disk diffusion data, broth microdilution was also performed. These data are shown in Table 2. Resistance to ampicillin was a characteristic feature of Enterobacteriaceae and was used as an internal control. All clinical isolates were found to be resistant to cephalosporins, quinolones, and fluoroquinolones whereas only 54 and 45% of the isolates expressed resistance to gentamicin and tetracycline, respectively. The isolates were also tested against a benzalkonium chloride, a common quaternary ammonium compound used as a disinfectant. Results (Table 2) show that the MICs varied from 8 to 32 μg/mL in accordance with previous reports (Abuzaid et al., 2012; Guo et al., 2015).

Results obtained using both of these AST-based protocols, indicate that all eleven bacterial isolates can be defined as multi-drug resistant (MDR), since they were simultaneously resistant to three or more classes of antimicrobial compound. Study collection isolates were found to be resistant to 4 or more and up to 8 classes of dug.

Efflux ability contributes for the increased bacterial resistant phenotype. Isolates were compared for their ability to extrude ethidium bromide (EtdBr) from the cell, revealing thus their efflux potential. In this method cells were saturated with EtdBr in a saline solution using CCCP to facilitate the uptake of EtdBr at a concentration that did not affect the cell viability (Srinivasan and Rajamohan, 2013).

The capacity for cells to extrude EtdBr is improved in the presence of glucose, as the latter energizes the bacterial cell. The over-expression of efflux pumps is indicated by a fast decrease in fluorescence in both the presence or absence of glucose. The results obtained (Figure 3) indicates no clear differences in the efflux activity of these clinical isolates when compared among themselves.

The ability to form biofilms plays a critical role in a bacterium's ability to colonize the host or any indwelling medical devices temporarily, in situ. This phenotype was studied by measuring bacterial growth in the presence of a rich culture media (in this case MH broth; see Figures S1, S2) and in a media designed to stress these isolates, M9 minimal media (see Figures S3, S4). Isolates when inoculated into these culture media, were maintained statically and incubated at 25 and at 37°C for 96 h.

Biofilm formation in MH decreased over time at 25°C, whereas at 37°C it appeared to decrease for the first 48 h and then increased, back to the initial values recorded until 96 h during static growth. In contrast, biofilm formation in M9 minimal media increased over time at both 25 and 37°C, indicating a better adaptation when subjected to a laboratory induced stress condition. The effect of temperature on biofilm formation in MH was stronger at 25°C during the first 48 h, for all isolates with the exception of K. pneumoniae CFS0364 (this isolate was comparatively weaker both at 25 and 37°C). Biofilm maturation was observed to be better in MH at 37°C after 48 h. In M9 minimal media biofilm formation was stronger at 37°C with exception of the isolate K. pneumoniae CFS0363 that exhibited better growth at 25°C. Overall, with the exception of K. pneumoniae CFS0372, the remaining isolates produced more biofilm when cultured in M9 minimal media than when grown on MH at 37°C. Moreover, the effect of stress conditions appeared to promote biofilm formation in the case of isolate K. pneumoniae CFS0364 at 37°C, indicating a potential adaptive survival response.

The expression of curli fimbriae, by bacteria is known to be involved in adhesion of these cells to surfaces, and it is also associated with cell aggregation and biofilm development. Cellulose is an extracellular matrix component of biofilms formed by bacteria of the Enterobacteriaceae family (Zogaj et al., 2001). When cellulose is produced together with these curli fimbriae, both produce a highly inert, hydrophobic extracellular matrix around the bacterium enhancing colonization capacity (Bokranz et al., 2005). The production of these curli fimbriae was determined by observing colony morphology on Congo red and calcofluor agar plates. Morphotypes were categorized according to Zogaj et al. (2001). The growth of a red, dry, and rough (RDAR) colony on Congo red agar plates indicated the presence of curli fimbriae whilst fluorescence detection in the calcofluor plates under UV light indicated cellulose production. Disruption or down-regulation of the corresponding genes, would provide bacteria with a distinct smooth and white (SAW) morphotype. A deficiency in curli fimbrial expression would render the colonies pink dry and rough (PDAR).

The results shown in Tables 3 and 4 provide images of the morphotypes obtained when isolates were grown in MH agar with the corresponding dyes at 37 and 25°C, respectively, as well as the pellicle formation, that occurred under the same conditions. Interestingly, results at 37°C did not show any distinctive morphotypes that would suggest the expression of curli fimbriae. The majority of the isolates showed a smooth-textured red-like morphotype (RAS—red and smooth). The same observations were also reported in Klebsiella species isolated from human gastrointestinal tract (Zogaj et al., 2003).

Production of cellulose was studied when isolates were cultured at 37°C and in this case K. pneumoniae CFS0366, CFS0368, CFS0370, and CFS0373 exhibited a notable expression of cellulose. The formation of cell aggregates at the air-liquid interface, was also determined in MH broth. With exception of K. pneumoniae CFS0365 and CFS0366 all were found to form a pellicle at 37°C.

When isolates were grown at 25°C they all exhibited a similar SAW morphotype indicating the absence of curli fimbriae and production of cellulose. The formation of a pellicle at 25°C was only associated with the isolates K. pneumoniae CFS0363, CFS0364, CFS0367, and CFS0368, all of which belonged to the CG147 and CG15.

In general, the morphotypes observed at 25°C were broadly consistent with the data obtained for biofilm formation. Data obtained when these isolates were cultured at 37°C was inconclusive and it was not possible to correlate, these findings with the biofilm formation.

When the genome sequences were queried, amino acid substitutions were identified in several genes involved in biofilm formation in K. pneumoniae CFS0367 and CFS0368 of the CG15 (see Table S3). The genes included astC, a succinylornithine transaminase, astD a succinylglutamic semialdehyde dehydrogenase, both involved in proline and arginine metabolism. ulaA (a component of the ascorbate-specific PTS system, EIIC), yfiB that codes for an internal membrane protein and yfiN, a diguanylate cyclase, involved in the regulation of type-3 fimbriae and cellulose production. These two isolates show differences when the biofilm formation is compared in the different in vitro conditions (with the exception of Müeller-Hinton at 25°C), possibly due to the substitution found on yfiN of K. pneumoniae CFS0366 that show impaired ability to form biofilm.

All 11 K. pneumoniae were sequenced using the Illumina HiSeq platform and found to have an average genome size of ~5.4-Mbp with the exception of K. pneumoniae CFS0363 that produced a larger genome size of 5.8-Mbp (Table S1 and Figure S5). For the majority of the isolates, an average of 5,000 coding DNA sequences (CDS) were identified, and in the case of the isolate K. pneumoniae CFS0363 this number was higher as a consequence of the larger genome size. An overview of the assembly metrics is presented in Table S1.

In silico MLST was performed for all the isolates (Figure 4) using the seven-marker gene set (gapA, infB, mdh, pgi, phoE, rpoB, tonB), and sequence types (STs) were assigned using the Institute Pasteur scheme. The majority of the isolates (64%) belonged to ST340 (serotype K15) which is part of the widely disseminated clonal group CG258, associated with ESBL- and carbapenem-resistance (Chen et al., 2014). Three of the eleven isolates (K. pneumoniae CFS0364, CFS0367 and CFS0368) belonged to the epidemic clonal group CG15. Further, K. pneumoniae CFS0364 (ST15) had the capsular serotype K24 whilst K. pneumoniae CFS0367 and CFS0368 (ST14) were associated with the K2 capsular serotype. The isolate K. pneumoniae CFS0363 presented the capsular serotype K64 and the sequence type ST147 which belongs the CG147 cluster, and is considered to be an epidemic resistant clonal group (Damjanova et al., 2008; Baraniak et al., 2013).

The presence of acquired antimicrobial resistance-encoding genes was determined using the resistance database Resfinder. The majority of the isolates (82%) harbored the ESBL-encoding gene bla_CTX-M-15 (Table S2). The genes bla_SHV-12 and bla_TEM-1B was also identified in 56 and 45% of the isolates, respectively. A resistance gene for fosfomycin (fosA) was present in all the isolates and the efflux pump oqxAB, responsible for fluoroquinolone resistance, was identified in 10 of the 11 isolates (being absent in isolate K. pneumoniae CFS0364). Moreover, resistance genes to phenicols (catA and catB) and tetracyclines (tet) accounted for 45% of the isolates in both cases. Sulfonamide and trimethoprim resistance-encoding genes were identified in 82% of the isolates.

Interestingly, the isolates from the epidemic group CG15 showed a more susceptible phenotype when compared with the remaining isolates of the collection (Figure 2). A common resistance determinant that was identified in all the isolates of this clonal group was the ESBL resistance gene bla_SHV-28. Additionally, isolates K. pneumoniae CFS0367 and CFS0368 also harbored the AmpC β-lactamase gene bla_ACC-1, which has been reported to be associated with nosocomial outbreaks in Europe (Ohana et al., 2005; Roche et al., 2008). The isolate K. pneumoniae CFS0363 possessed a number of β-lactam-related resistance encoding genes including bla_CMY-4, bla_OXA-1, bla_OXA-9, bla_SHV-12, and bla_TEM-1A. No genes were identified conferring resistance to carbapenems.

The presence of the erythromycin resistance genes ere(A) and ere(B) and the plasmid-mediated quinolone resistance (PMQR) gene qnrB1 was also confirmed in the isolate K. pneumoniae CFS0363. Resistance to quinolones and fluoroquinolones can also be due to target gene mutations in gyrA, gyrB, parC, and parE (Casin et al., 2003; Baucheron et al., 2004; Fu et al., 2013). Table S3 presents a summary of the mutations identified in these proteins and those involved in efflux and regulation. K. pneumoniae CFS0367 and CFS0368 have the same amino acid substitutions in GyrA that were identified at position 83 (F83Y), position 87 (G87D) and a deletion found at position 48. Similarly, K. pneumoniae CFS0364 contained a GyrA amino acid substitution consistent with the positions 645 (N645H) in the polypeptide chain. The K. pneumoniae isolate CFS0367 was found to have one amino acid substitutions in each of the remaining proteins GyrB, ParC, and ParE, at positions 466 (E466D), 84 (G84E), and 460 (Q460P) respectively. In contrast, only one amino acid substitution in the protein sequence of ParC was found in K. pneumoniae CFS0368 at position 84 (G84E). All remaining isolates of the clonal group CG147 and CG258 were devoid of mutations when compared with the respective reference of their ST type. These mutations are often accompanied by the acquisition of additional resistance-encoding genes identified along with the overexpression of efflux pumps oqxAB and kexD known to contribute to fluoroquinolone resistance. KexD efflux pump was identified in 73% (n = 8) of the isolates studied (see Table S2). Remarkably, this pump was absent in those isolates from the clonal group CG15. The efflux pump oqxAB was absent in the isolate K. pneumoniae CFS0364.

Mutations in efflux systems and global regulators could also contribute to the resistance phenotype. Amino acid sequence analysis of the RND efflux pump AcrAB-TolC identified a truncation in the regulator AcrR for the isolate K. pneumoniae CFS0367 and an amino acid substitution of at the position 715 (R715L) for the isolate K. pneumoniae CFS0368. Moreover, one residue was found to be deleted in the regulator RamR in K. pneumoniae CFS0363 and CFS0368. Interestingly, in the isolates K. pneumoniae CFS0364 and CFS0367 the ramRA regulon were not detected.

To gain an insight into the nature of any plasmids, contained in these isolates, replicon typing was performed in silico using PlasmidFinder (Figure 4). All were positive for the colRNAI type, a common replicon associated with small plasmids found in Klebsiella species (Ramirez et al., 2014). The virulence-encoded incompatibility types IncFIB(K) and IncFII were identified in 91% of the isolates whereas IncR and IncFIA accounted for 45% of these isolates. IncF plasmids are often associated with ESBL-resistance encoding genes (Carattoli, 2009; Doumith et al., 2012; Markovska et al., 2014). Moreover, the isolate K. pneumoniae CFS0363 an isolate with four large plasmids also harbored Inc types including IncFIB(pQil), IncHI1B and the IncL/M. These replicon types are widely disseminated in the environment and often associated with MDR non-typhoidal and typhoidal Salmonella species and with carbapenem resistance, respectively (Holt et al., 2011; Zurfluh et al., 2014; Kubasova et al., 2016; Ouertani et al., 2016; Pérez-Vázquez et al., 2016).

The isolate K. pneumoniae CFS0364 was the only bacterium harboring the colpVC and IncX4 replicon types which are known to be associated with bla_CTX-M-mediated resistance both in humans and animals (Johnson et al., 2012; Lo et al., 2014). Recently, these plasmids were also epidemiologically linked with the carriage of the colistin resistance gene mcr-1 (Fernandes et al., 2016). Colistin resistance was not detected among any of the isolates studied. Interestingly, the IncN replicon type was only associated with the K2 serotype isolates K. pneumoniae CFS0367 and CFS0368. The latter replicon type was also associated with the ESLB resistance gene bla_SHV-12 (Markovska et al., 2014) and carbapenamase resistance (Rodrigues et al., 2016; Tijet et al., 2016).

The presence of virulence genes may influence the capacity that these pathogens have on effective host colonization and infection. Thus, it is relevant to gain some understanding of this potential through the identification of such genes, in particular the iron scavenging genes that are known to be associated with invasive disease (Holt et al., 2015). When analyzed, all isolates in this study (Table S2) harbored the enterobactin iron uptake system (ent and fep operon), a ubiquitous system found in K. pneumoniae (Paczosa and Mecsas, 2016). Isolates K. pneumoniae CFS0363 and CFS0364 also presented the yersiniabactin siderophore system (fyuA, irp1, irp2 and ybt operon). The isolates of the clonal group CG15 all presented the kfuabc operon. These iron-scavenging operons have been associated with invasive K. pneumoniae isolates (Ma et al., 2005; Bachman et al., 2012; Holt et al., 2015). One peculiarity of the isolate K. pneumoniae CFS0363 was the presence of the metal-resistance tellurium operon that was identified, in contrast to its complete absence in the remaining isolates.

All the isolates possessed several biocide resistance-encoding genes that were detected in silico using BacMet. The quaternary ammonium resistance-encoding gene qacEΔ1, commonly associated with class 1 integrons, was identified in all isolates but was found to be absent in the K. pneumoniae CFS0364 and CFS0367. The remaining biocide resistance genes is summarized in Table S2.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.