Molecular Epidemiology and Risk Factors of Carbapenemase-Producing Enterobacteriaceae Isolates in Portuguese Hospitals: Results From European Survey on Carbapenemase-Producing Enterobacteriaceae (EuSCAPE)

In Portugal, the epidemiological stage for the spread of carbapenemase-producing Enterobacteriaceae (CPE) increased from sporadic isolates or single hospital clones (2010–2013), to hospital outbreaks, later. Here we report data from a 6-month study performed under the European Survey on Carbapenemase-Producing Enterobacteriaceae (EuSCAPE). During the study period, 67 isolates (61 Klebsiella pneumoniae and 6 Escherichia coli) non-susceptible to carbapenems were identified in participant hospital laboratories. We detected 37 blaKPC–type (including one new variant: blaKPC–21), 1 blaGES–5, and 1 blaGES–6 plus blaKPC–3, alone or in combination with other bla genes. Bioinformatics analysis of the KPC-21-producing E. coli identified the new variant blaKPC–21 in a 12,748 bp length plasmid. The blaKPC–21 gene was harbored on a non-Tn4401 element, presenting upstream a partial ISKpn6 (ΔISKpn6/ΔtraN) with the related left IR (IRL) and downstream a truncated Tn3 transposon. PFGE and MLST analysis showed an important diversity, as isolates belonged to distinct PFGE and STs profiles. In this study, we highlighted the presence of the high-risk clone E. coli sequence-type (ST) 131 clade C/H30. This worldwide disseminated E. coli lineage was already detected in Portugal among other antibiotic resistance reservoirs. This study highlights the intra- and inter-hospital spread and possible intercontinental circulation of CPE isolates.


INTRODUCTION
Carbapenems, a class of β-lactam antibiotics with wide activity, are often the antimicrobials of last resort to treat infections associated to extended-spectrum β-lactamase (ESBL)-or plasmid-mediated AmpC (PMAβ)-producing Enterobacteriaceae isolates (Papp-Wallace et al., 2011;Rodríguez-Baño et al., 2018). Unfortunately, carbapenem non-susceptible Enterobacteriaceae (CNSE) have been reported worldwide mainly because of the acquisition of carbapenemaseencoding genes (Potter et al., 2016;Codjoe and Donkor, 2018). Since the first description of a carbapenemase-producing Enterobacteriaceae (CPE) in Europe in the 1990s, a large variety of carbapenemases has been identified in each of the four Ambler molecular classes, mainly the KPC-type (class A), VIM-, IMP-, and NDM-types (class B), and OXA-48-type (class D) (Grundmann et al., 2017;Logan and Weinstein, 2017). CPE isolates are usually resistant to many other β-lactam and non-β-lactam antibiotics, leading to multi-resistant isolates.
In Portugal, the epidemiological stage for the spread of CPE increased from sporadic isolates or single hospital clones, from April 2010 to February 2013, to sporadic hospital outbreaks later (Albiger et al., 2015;Manageiro et al., 2015b,c). Here we report data from a 6month prevalence study performed under the European Survey on Carbapenemase-Producing Enterobacteriaceae (EuSCAPE) with the collaboration of different Portuguese Laboratories.

Bacterial Isolation, Antibiotic Susceptibility, and Molecular Characterization
This study included a total of 104 clinical isolates (94 Klebsiella pneumoniae and 10 Escherichia coli) collected from November 2013 to April 2014 in 10 Portuguese hospitals. The first ten consecutive and non-replicated CNSE isolates obtained during this period, in each hospital, from blood, lower respiratory tract secretions, urine, puncture fluids, and wound secretions, of single patients, were sent to the National Reference Laboratory, in Lisbon, and were considered. Successive carbapenem-susceptible isolates of the same species were also preserved as controls whenever possible, accordingly to EuSCAPE protocol (Grundmann et al., 2017). Overall, 67 CNSE (61 K. pneumoniae and 6 E. coli) and 37 controls (33 K. pneumoniae and 4 E. coli) were analyzed.
In the context of the EuSCAPE study, all data were anonymized and collected in accordance with the European Parliament and Council decisions on the epidemiological surveillance and control of communicable disease in the European Community (Eur-Lex-31998D2119, 1998Eur-Lex-32000D0096, 2000).

Antibiotic Susceptibility and Molecular Characterization of Antimicrobial Resistance
Antimicrobial susceptibility was performed by disk diffusion method for 15 antibiotics (Table 1), and by broth microdilution method for tigecycline and colistin, using EUCAST guidelines 1 . Clinical isolates with resistance or with decreased susceptibility to ertapenem were considered presumptively CPE. Isolates were considered multidrug resistant when presenting reduced susceptibility to three or more structurally unrelated antibiotics.

Transfer Experiments
Transferability of bla KPC−21 from E. coli UR19829 was performed by broth mating out assays using sodium azide-resistant E. coli J53 as a recipient strain, and by transformation, as previously described (Manageiro et al., 2015b(Manageiro et al., , 2017.
Genomic Characterization of KPC-21-Producing E. coli KPC-21-producing E. coli was genotypically characterized by whole-genome sequencing (WGS) as previously described (Manageiro et al., 2017). The assembled contigs were analyzed and studied for the presence of antibiotic resistance-and virulence-encoding genes, multi-locus sequence types, fim type, serotype, plasmid replicon types, and insertion sequences (ISs) using bioinformatics tools from the Center for Genomic Epidemiology 2 and ISsaga (Varani et al., 2011).
The pUR19829-KPC21 plasmid structure was constructed based on the genetic organization of the closest plasmid sequences obtained by BLASTn, provided by NCBI 3 , followed by contig neighbor's prediction from assembly information.

Statistical Analysis
OpenEpi software, version 3.01 was used for statistical analysis (Sullivan et al., 2009). Fisher exact test was used to assess differences in clinical and epidemiological risk factors for control and CNSE-carrying patients. One-tailed P values of ≤0.05 were considered to be statistically significant. Associations were determined by calculation of odds ratios with 95% confidence intervals.

Nucleotide Sequence Accession Number
The new bla KPC−21 nucleotide sequence was submitted to the NCBI GenBank Database with accession number NG_049254 and the complete plasmid sequence of pUR19829-KPC21 with accession number MH133192.

RESULTS AND DISCUSSION
During the study period, 67 isolates (61 K. pneumoniae and 6 E. coli) CNSE were identified in nine of the 10 Hospital Laboratories, with a non-susceptibility rate for meropenem and imipenem of 64 and 59%, respectively, for K. pneumoniae, 3 http://blast.ncbi.nlm.nih.gov/Blast.cgi and of 100% for E. coli. As expected, when compared with the control isolates, CNSE presented higher level of nonsusceptibility to all antibiotic classes tested ( Table 1). Colistin and tigecycline MIC 50 values for CNSE were similar than those obtained for control isolates. Eleven out of the 104 (16.3%) isolates were colistin resistant, without the presence of the plasmid-mediated mcr-1 or mcr-2 gene. However, MCR-1 determinant was already identified in different reservoirs in Portugal, such as vegetables, animals and humans (Jones-Dias et al., 2016;Beyrouthy et al., 2017;Kieffer et al., 2017).
The new bla KPC−21 gene differed from bla KPC−2 by one point mutation that leads to the amino acid substitution Trp105Arg; this position is involved in the binding and maintaining of the KPC catalytic activity (Papp-Wallace et al., 2010). In silico typing revealed an KPC-21-producing E. coli belonging to ST131 clade C/H30, associated with the fimbriae-encoding fimH allele 30, which become the most dominant lineage since the 2000s  Moreover, bioinformatics analysis of the KPC-21-producing E. coli identified this variant in a 12,748 bp length plasmid, with a mean coverage of 580-fold and GC content of 58.5% (Figure 1). Dissemination of bla KPC has been mainly supported by the horizontal transfer of Tn4401-type transposon, which harbors tnpA encoding a transposase, tnpR encoding resolvase, and two insertion sequence elements (ISKpn7 and ISKpn6) bracketing the bla KPC gene (Cuzon et al., 2011). In this study, the bla KPC−21 gene was harbored on a non-Tn4401 element (Chen et al., 2014), presenting upstream a partial ISKpn6 ( ISKpn6/ traN) with the related left IR (IR L ) and downstream truncated Tn3 transposon downstream (Figure 1). This region has 99.97% of identity with pKP1194a, a plasmid carried by a hospital-associated KPC-2-producing K. pneumoniae isolated in Brazil (Accession number KX756453) (Figure 1gray region I); this suggest an intercontinental circulation of isolates and mobile genetic elements (MGE), and the consequent need of concerted actions against the spreading of antibiotic resistance, at a worldwide level. The pUR19829-KPC21 enclosed also an intact ISPsp7 element, an insertion sequence from IS30 family, firstly described in Pseudomonas spp. (Szuplewska et al., 2014). Furthermore, the pUR19829-KPC21 backbone contained a region coding for plasmid replication (IncQ2 repA, repC), and mobilization (mobA, mobC), showing >99.9% sequence identity to the corresponding regions of pKPSH169, a 7.7 Kbp qnrS2harboring IncQ plasmid identified in municipal wastewater treatment facilities in Israel (Accession number KT896499) (Figure 1gray region II); this similarity highlights the high level of promiscuity of isolates between clinical settings and environment, where both reservoirs play a role in the antibiotic resistance dissemination (Stokes and Gillings, 2011). However, the lack of conjugative elements or an oriT region, associated with the presence of a truncated oriV region (Figure 1) suggests that pUR19829-KPC21 plasmid is nonmobilizable (Smillie et al., 2010). This fact is corroborated by the absence of a successful plasmid conjugation or transformation.
The variables used in the evaluation of risk factors for infection or colonization of patients with CNSE or control isolates are present in Table 2. When compared to the 37 control strains, only ESBL-production and the patient admission at a hospital in the center of Portugal were significantly associated with CNSE isolates in the period of the study. In the era of ESBL-producing Enterobacteriaceae, the antibiotic regimens suggested for severe health-associated infections are necessarily based on carbapenems (Rodríguez-Baño et al., 2018). Unfortunately carbapenem use has being described as a risk factor for CPE acquisition, only preceded by the use of medical devices (van Loon et al., 2018). In addition, the present study attests that Portugal, during the period of the study, has a different CNSE geographical distribution with the center of Portugal significantly associated with carbapenem non-susceptibility. This fact corroborates previous studies which indicated that in Portugal, in 2015, only sporadic isolates or single hospital cases were described (Albiger et al., 2015). PFGE and MLST analysis showed an important diversity, with isolates belonging to distinct PFGE and STs (Supplementary Figure S1). With respect to K. pneumoniae (Supplementary Figure S1A), a total of 10 clusters and 25 unique PFGE profiles were generated using XbaI, indicating the that the circulating clones in that period were genetically diverse. However, carbapenemase-producing K. pneumoniae isolates were more clonal (six PFGE clusters including 69.4% of these isolates) than non-carbapenemase-producing K. pneumoniae (four PFGE clusters including 50.0% of these isolates). As shown in Supplementary Figure S1, both CNSE species showed intra-and inter-hospital spread (e.g., PFGE clusters KpI and KpIX), with some hospital-specific clones (e.g., PFGE clusters KpIV and KpVIII). However, as also showed in Spain in other EuSCAPE study (Esteban-Cantos et al., 2017), the carbapenem-non-susceptible K. pneumoniae population was more clonal than the carbapenem-susceptible population (data not shown). Ten different MLSTs were detected among carbapenemase-producing (ST14, ST15, ST45, ST231, and ST1513) and non-carbapenemase-producing (ST11, ST17, ST348, and ST395) K. pneumoniae isolates. At our knowledge, this is the first description of ST17, ST395, and ST1513 K. pneumoniae in Portugal (Manageiro et al., 2015b;Rodrigues et al., 2016;Vubil et al., 2017). Noteworthy, the GES-5 enzyme was detected in a ST231 K. pneumoniae isolate as previously reported in Portugal, but in the same hospital, which shows its capacity to maintain in clinical settings due to the selection pressure of this environment (Manageiro et al., 2015b). Furthermore, ST45 was recently the cause of a hospitalbased outbreak caused by multidrug-resistant, KPC-3-and MCR-1-producing K. pneumoniae in Portugal (Mendes et al., 2018).
The high-risk clone carbapenemase-positive K. pneumoniae ST258 was not detected in this study or among clinical carbapenemase-producing K. pneumoniae isolates in Portugal (Manageiro et al., 2015b;Rodrigues et al., 2016;Vubil et al., 2017). However, concerning carbapenem-non-susceptible E. coli, besides the six different PFGE unique profiles, the isolates belongs all but two (ST405-fimH27 and ST23-fimH35) to the ST131 clade C/H30 high-risk clone disseminated worldwide (Supplementary Figure S1B) (Woodford et al., 2011;Pitout and DeVinney, 2017). Noteworthy, this clone was already detected in Portugal among other antibiotic resistance reservoirs, such as in an E. coli strain isolated from a dolphin housed at a Zoo Park (Manageiro et al., 2015a); in dogs and cats with urinary tract infection (Marques et al., 2018); and in E. coli strains from wastewater and gulls (Varela et al., 2015). Again, this shows that clinical settings and different environmental compartments may be considered communicating vessels through which bacteria and resistance genes are able to flow (Stokes and Gillings, 2011).
Portugal was one of the EuSCAPE participating countries that presented higher proportions of KPC-positive K. pneumoniae (Grundmann et al., 2017). The percentage of carbapenem nonsusceptible K. pneumoniae was low in invasive infections in the study period [2.4%, EARS-Net 2013] 4 . However, although the consumption of carbapenems has declined by 13.3% between 2012 and 2016 (PPCIRA, 2017), Portugal is reporting since 2013 a significant increasing trend of carbapenem non-susceptible K. pneumoniae [6.4%, EARS-Net 2016] 4 . The number of inter-institutional transmission is also increasing (Glasner et al., 2013;Albiger et al., 2015), being K. pneumoniae the principal cause of bacterial healthassociated infections in Portugal, as in other European countries (ECDC, 2013). Of concern is the fact that KPCproducing organisms cause infections with high morbidity and mortality (Porreca et al., 2018;Rodríguez-Baño et al., 2018). These results reinforces that reducing antibiotic use alone is likely insufficient for reversing resistance (Lopatkin et al., 2017). We strongly believe that the chain of transmission of isolates and genes in clinical settings will be reduced or broken, especially with containment measures rigorously implemented and followed at local level.

AUTHOR CONTRIBUTIONS
VM designed the study, performed the molecular experiments, bioinformatics analysis, analyzed the data, and wrote the manuscript. RR, IBM, and EF performed the microbiological and molecular experiments, and analyzed the data. DAS and LV performed Illumina genome sequencing experiments. The Network EuSCAPE-Portugal participants acquired laboratory data. MC designed the study, wrote and reviewed the manuscript. All authors read and approved the final manuscript.