This study was performed in the context of one of the largest acute care hospitals in the North of Portugal (705 beds), covering a population of 1.2 million people (hereafter called Hospital A). During a one-year study period (from March 2011 to May 2012) K. pneumoniae clinical isolates showing reduced susceptibility to carbapenems (imipenem or ertapenem or meropenem) were identified as part of routine diagnostics in the hospital Clinical Pathology Service. Isolates were collected from both inpatients admitted to the internal medicine service and from patients admitted to the hospital emergency ward.

An extra-hospital healthcare institution for dependent and old people in the North of Portugal was studied. The LTCF with 54 beds has three different typologies of care: long-term maintenance (LTM, 22 beds), medium-term and rehabilitation (MTR, 22 beds), and palliative care (PC, 10 beds). The institution is located in the same geographic area as Hospital A (distance of 4 Km), and consequently, the circulation of patients between these two healthcare institutions occurs frequently.

Thirty-eight fecal samples from LTCF residents were collected between January and February 2012, suspended in Brain Heart Infusion (BHI; Oxoid, Hampshire, United Kingdom), and incubated overnight at 37°C. The enriched suspensions were then plated onto MacConkey agar plates (Oxoid, Hampshire, United Kingdom) supplemented with meropenem (1mg/l). Isolates that grew in the selective media were re-inoculated in a new plate to exclude any satellite growers (maximum of four random colonies per plate).

The clinical isolates were identified using the Vitek^® 2 automated system (bioMérieux, Marcy l’Étoile, France). Bacteria isolated as part of the intestinal colonization screening were identified using the bacterial identification biochemical galleries API^® 20E and ID^®32GN (bioMérieux).

The antimicrobial susceptibility of clinical isolates was assessed through the determination of the minimum inhibitory concentration (MIC) of different antimicrobial agents, performed using the Vitek^® 2 (bioMérieux) and/or WalkAway (Beckman Coulter, Brea, CA, United States) automated systems. The MICs detected for ampicillin, piperacillin, ticarcillin, amoxicillin + clavulanic acid, piperacillin + tazobactam, ticarcillin + clavulanic acid, cephalothin, cefuroxime, ceftazidime, cefotaxime, cefepime, aztreonam, imipenem, ertapenem, meropenem (β-lactams), gentamicin, tobramycin, amikacin, minocycline, ciprofloxacin, levofloxacin, pefloxacin, nitrofurantoin, trimethoprim + sulfamethoxazole, and rifampicin (non-β-lactams) were interpreted into susceptible, intermediate susceptible or resistant according to the clinical and laboratory standards institute (CLSI) guidelines (Queenan and Bush, 2007; Supplementary Table S1). For isolates collected in the LTCF intestinal colonization screening, antimicrobial susceptibility was determined by disk-diffusion methods; susceptibility to both β-lactams [ampicillin (10μg), amoxicillin + clavulanic acid (20+10μg), ceftazidime (30μg), cefotaxime (30μg), cefepime (30μg), cefoxitin (30μg), aztreonam (30μg), imipenem (10μg), ertapenem (10μg), and meropenem (10μg)] and non-β-lactam antibiotics [streptomycin (10μg), gentamicin (10μg), netilmicin (30μg), tobramycin (10μg), amikacin (30μg), tetracycline (30μg), nalidixic acid (30μg), ciprofloxacin (5μg), nitrofurantoin (300μg), chloramphenicol (30μg), tigecycline (15μg), and trimethoprim + sulfamethoxazole (1.25/23.75μg)] was defined according to the CLSI guidelines (CLSI, 2013) or the EUCAST criteria in the case of tigecycline¹ (Supplementary Table S2).

An initial carbapenemase production screening (MBLs) was performed using the double disk synergism method (DDSM) - IMP (10μg) versus IMP (10μg)+EDTA (0,5M), followed by the confirmatory MBL E-test IP/IPI [(MIC determination; IMP (4-256μg/ml) versus IMP (1-364μg/ml)+EDTA (constant level)] (bioMérieux, Marcy l’Étoile, France). The E-test was considered MBL suggestive when the MIC ratio of imipenem/imipenem plus EDTA was ≥8 and/or when the presence of a phantom zone or deformation of the inhibitory ellipse was observed. The modified hodge test (MHT) was performed in parallel, to screen for non-MBL carbapenemase production. Briefly, an imipenem disk (10μg) was placed at the center of a Müeller-Hinton agar plate (Oxoid, Hampshire, United Kingdom), previously inoculated with E. coli ATCC 25922, and the clinical isolates were streaked heavily from the edge of the disk toward the edge of the plate. The MHT was considered positive when E. coli growth was observed within the usual inhibition zone of the imipenem disk (CLSI, 2013). As a final confirmatory step the biochemical Blue-Carba test was performed as described elsewhere (Pires et al., 2013).

Total DNA was extracted from all isolates via the boiling of single bacterial colony suspensions for 10min, followed by a 5min centrifugation step at 15,000rpm. The supernatant was then collected and stored at 4°C until further use. Relevant beta-lactamase [bla_TEM, bla_OXA, bla_SHV (Dallenne et al., 2010)_, and bla_{CTX-M group 1} (Machado et al., 2005)] and carbapenemase [bla_VIM, bla_IMP, bla_KPC, bla_OXA-48, and bla_NDM (Poirel et al., 2011)] genes were screened using the primers and amplification conditions described in the literature (Machado et al., 2005; Dallenne et al., 2010; Poirel et al., 2011; Goncalves et al., 2016; Teixeira et al., 2016). Whenever relevant, amplicons were sequenced using the ABI-PRISM 3100 automatic genetic analyzer (Thermo Fischer Scientific, Waltham, MA, United States). Sequence analysis and alignment were performed using the National Center for Biotechnology Information tool.² As a final confirmation step, IMP-22 specific primers were used (Pellegrini et al., 2009). A compilation of all primer sequences used in this study can be found in (Goncalves et al., 2016).

The clonal relationships of the K. pneumoniae clinical isolates were studied via pulsed-field gel electrophoresis (PFGE), after total genomic DNA digestion with XbaI (Gautom, 1997). Briefly, carbapenem-resistant clinical isolates were cultured in brain heart infusion (BHI) for 24h at 37°C, then “trapped” into 1.6% agarose plugs. A lysis step was performed at 54°C for 2h (50mm Tris, 50mm EDTA, 1% N-lauryl-sarcosine, 0.1 mg/ml proteinase K, pH 8.0), followed by 2–3 washing cycles, and afterward, the digestion overnight with 30U of XbaI at 37°C. Total DNA digests were separated on 1.0% agarose gels (SeaKem Gold Agarose, Lonza, Basel, Switzerland) via PFGE using the CHEFF DR III system (Bio-Rad Laboratories, Hercules, CA, United States) and the following conditions: electric field strength of 6V/cm² (200V), 14°C, and pulse time of 15s–25s for 16h. After electrophoresis, the gels were stained with ethidium bromide (10μg/ml) for 30min and watched under a UV light (Bio-Rad Laboratories). Data analysis was performed using the BIONUMERICS software, version 8.0 (bioMérieux, Marcy l’Étoile, France); the UPGMA algorithm based on the Dice coefficient (1.0% band tolerance; 1.0% optimization) was applied. The PFGE profiles were defined on the basis of DNA banding patterns in accordance with the criteria defined by Tenover et al. (1995). Isolates with a pattern similarity profile above ≥80% were considered identical.

Conjugation experiments were performed to investigate the transfer of carbapenem resistance determinants. E. coli HB101 (azide resistant, lactose-negative) was used as the recipient strain. Donor and recipient bacterial strains were individually grown overnight in Trypticase Soy Broth (TSB; Oxoid, Hampshire, United Kingdom) and drops of donor and recipient bacterial suspensions were then mixed on the surface of a Müeller-Hinton Agar plate (Oxoid) and re-incubated at 37°C for 24h. The resulting bacterial growth was re-inoculated on Müeller-Hinton medium supplemented with meropenem or ceftazidime (10mg/l) and azide (100μg/ml) and incubated for a maximum of 72h at 37°C. Growing colonies on the selective medium were randomly chosen and inoculated in MacConkey agar (Oxoid) to assess lactose fermentation. Lactose non-fermenters were subjected to antimicrobial susceptibility determination and to genotypic characterization as above stated.

This research was conducted in accordance with the Declaration of Helsinki Ethical Principles. This study was approved by the Ethics Committee of Hospital de Braga, Braga, Portugal. Additionally, human fecal sample collection was performed in accordance with the Good Clinical Practice guidelines; the LTCF direction provided the necessary authorization to conduct this study. Of note, all of the study participants provided written informed consent.

Eight carbapenem-resistant K. pneumoniae showing reduced susceptibility to at least one of the carbapenems tested were isolated from different biological samples of seven distinct hospitalized patients: five inpatients of the internal medicine service of Hospital A and two patients admitted to the emergency ward of the same hospital. The most common type of biological sample from which these bacteria were isolated was sputum (n=5), followed by urine (n=2) and blood (n=1; Table 1). Two of the carbapenem-resistant K. pneumoniae (isolates H13 and H50) were isolated from the same patient (patient 3) in different periods (December 2011 and April 2012, respectively; Table 1). The patients were mostly elderly (median age 74years; range 36–92years) with distinct underlying illnesses, namely urinary tract infection (n=1), endocarditis (n=1), renal insufficiency (n=2), brain tumor (n=1), nosocomial pneumonia (n=1), acute pancreatitis (n=1), respiratory insufficiency (n=2), and bladder tumor (n=1; Table 1). All of the patients had previous hospitalization history in Hospital A, with many of them spending prolonged periods at the internal medicine ward. Of note, patients 3, 6, and 7 received meropenem therapy during their hospitalization period (Table 1).

Four carbapenem-resistant K. pneumoniae isolates (10.53%, 4/38) with reduced susceptibility to imipenem were detected in fecal samples of residents of a LTCF (two different typologies of care; LTM, n=3; MTR, n=1; Table 2). The four residents colonized with carbapenem-resistant bacteria were mostly elderly (median age 72.5years; range 63–82years), with previous history of stroke (n=2), stroke associated with other pathologies (endocarditis and pneumonia; n=1), and chronic renal insufficiency (n=1; Table 2). Three of them had recent hospitalization history in Hospital A: two spent prolonged periods in the internal medicine ward (residents A and D) while the third one was admitted to the orthopedics service (resident B). The fourth resident (C) had hospitalization history in three different hospitals in the same geographic area of Hospital A (one hospital in Braga district and two hospitals in Porto district; Table 2).

All clinical isolates showed reduced susceptibility to meropenem (MIC ≥16; R). Additionally, of the set of clinical isolates analyzed, six presented resistance to Ertapenem, and two to imipenem (three others showed an intermediate phenotype; Table 3). Importantly, most of the clinical isolates also showed resistance to expanded-spectrum cephalosporins, other β-lactams and β-lactam/β-lactamase inhibitor combinations; additionally resistance to gentamycin (n=2), tobramycin (n=5), ciprofloxacin (n=6), norfloxacin (n=1), pefloxacin (n=1), trimethoprim/sulfamethoxazole (n=7), and tetracycline (n=6) was also detected (Table 3).

The four intestinal colonization K. pneumoniae isolates also showed reduced susceptibility to imipenem (Table 4). Two of them also showed reduced susceptibility to ertapenem (isolates 22 and 24), with only one (isolate 22) showing resistance to the three carbapenems tested (imipenem, ertapenem, and meropenem). The majority of the intestinal isolates also showed resistance to expanded-spectrum cephalosporins (n=3), other β-lactams (n=4), β-lactam/β-lactamase inhibitor combinations (n=4) as well as to non-β-lactam antibiotics, including tetracycline (n=3), trimethoprim/sulfamethoxazole (n=3), nalidixic acid (n=4), ciprofloxacin (n=4), chloramphenicol (n=2), and streptomycin (n=1; Table 4).

Of note, all clinical and intestinal colonization K. pneumoniae isolates were defined as multidrug-resistant (MDR) in accordance with the definition proposed by Magiorakos and colleagues (non-susceptible to ≥1 agent in ≥3 antimicrobial categories; Magiorakos et al., 2012).

Interestingly, the production of MBL in the context of all of the K. pneumoniae clinical isolates was initially defined as negative, as per the MBL E-test IP/IPI. Additionally, the results of the MHT were also negative for all of the clinical isolates. However, contrary to these findings, the results of the Blue-Carba test (all positive), suggested the expression of carbapenemases in all clinical isolates. Similarly, one of the intestinal colonization K. pneumoniae isolates was also determined as a carbapenemase producer, as per the results of the Blue-Carba test. Of note, according to the CLSI guidelines, extended-spectrum β-lactamase production was not detected, both in the clinical and intestinal colonization carbapenem-resistant isolates.

Finally, the bla_IMP-22 gene was detected via PCR followed by sequencing in all of the eight K. pneumoniae clinical isolates and in one of the carbapenem-resistant intestinal isolates, part of the commensal intestinal microbiota of one LTCF resident (isolate 22). PCR amplification was further confirmed using IMP-22 specific primers, supporting our sequencing results. No other carbapenemase genes were detected in the set of K. pneumoniae isolates.

For epidemiological purposes, since many different carbapenem-resistant isolates were isolated in a short period of time in Hospital A, next, we investigated the genetic relationships of the eight IMP-22-producing K. pneumoniae clinical isolates using PFGE. Three distinct PFGE profiles (I to III) were revealed. Importantly, six of the clinical isolates (namely H7, H8, H13, H15, H40, and H41) shared the same profile (profile I), indicating that these isolates are genetically identical and have the same origin (Figure 1). The two remaining isolates, H52 and H50 showed two different profiles (profiles II and III, respectively; Figure 1). Of note, the isolates belonging to the single major clone (clone I) were isolated from four patients admitted to the medicine ward (isolates H7, H8, H13, and H15), and two patients admitted to the emergency ward (isolates H40 and H41); however, these two patients had previously been admitted to the same hospital. Interestingly, the two isolates collected from the same patient showed different PFGE patterns: H13, with the predominant profile I, was isolated during a first prolonged stay in Hospital A, while H50, with a unique typing pattern unrelated to profile I, was isolated during a second hospitalization, after a period spent in an LTCF in the same geographic region of Hospital A (the same LTCF from which the IMP-22-producing K. pneumoniae intestinal colonizer was isolated).

Since the findings in the context of the patient from which two isolates were collected suggest the horizontal transfer of bla_IMP-22 (two different PFGE types), we further performed conjugation experiments. Importantly, our results confirmed the above hypothesis; we observed the transference not only of the carbapenem resistance determinant but also of resistance determinants to non-β-lactam antibiotics (Table 3; underlined), in the context of six clinical isolates (H7, H8, H13, H15, H40, and H41), all belonging to the single major clone. On the other hand, no conjugation was achieved in the context of the two remaining clones isolated from clinical samples, as well as of the carbapenem-resistant K. pneumoniae intestinal isolates. Importantly, the presence of the bla_IMP-22 gene in all of the trans-conjugants obtained was confirmed via PCR and sequencing.