Genetic Characterization of Carbapenem-Resistant Enterobacteriaceae and the Spread of Carbapenem-Resistant Klebsiella pneumonia ST340 at a University Hospital in Thailand

Carbapenem-resistant Enterobacteriaceae (CRE) has increasingly spread worldwide in the past decade. The prevalence and characteristics of CRE in Thailand are unknown. In this study, we conducted a 2-year surveillance of CRE among 12,741 clinical isolates of Enterobacteriaceae at the largest university hospital in Thailand with molecular characterization of beta-lactamase (bla) genes, including carbapenemase genes. The CRE prevalence was 1.4%. bla KPC-13 and bla IMP-14a were the only carbapenemase genes detected among these CRE isolates. bla KPC-13 gene was found in a single isolate of Escherichia coli, Enterobacter cloacae and Citrobacter freundii, and bla IMP-14a was found in four isolates of Klebsiella pneumoniae. Carbapenem-resistant K. pneumoniae (CRKP) isolates were resistant to multiple carbapenems at a higher ratio than other CRE species, and thus were further characterized for resistance phenotypes, bla genotypes and molecular epidemiology. Most CRKP isolates harboured multiple bla genes, especially those related to extended-spectrum beta-lactamases. Seven CRKP isolates were resistant to all tested carbapenems, and showed decreased ompK35 and/or ompK36 porin gene expression. Molecular typing of CRKP based on pulsed-field gel electrophoresis (PFGE) demonstrated several unrelated clones. Multilocus sequence typing (MLST) was partially concordant with PFGE results and revealed that ST340, a member of drug-resistant K. pneumoniae clonal complex 258, was the most predominant clone, followed by ST48, ST11 and ST273. The novel ST1645 was identified from this study. ST340 has neither been shown to be predominated among CRKP from other studies, nor been reported in Thailand. Therefore, it emphases a critical concern to monitor and control the spread of CRKP.


Introduction
Carbapenems are broad-spectrum beta-lactam agents frequently used in treatment of infections caused by multidrug resistant gram-negative bacteria. Due to a rise of extended-spectrum beta-lactamase (ESBL)-producing Enterobacteriaceae in many countries, carbapenems have been increasingly used as a common drug of choice [1]. Enterobacteriaceae that are resistant to carbapenems, known as carbapenem-resistant Enterobacteriaceae (CRE), have gradually emerged and thus strictly limiting options for treatment [2]. This is considered a serious threat because CRE are significantly associated with higher mortality rates [3], [4]. Production of various carbapenemases is a key mechanism mediating the emergence of CRE. Klebsiella pneumoniae carbapenemase (KPC), encoded by bla KPC gene, is a particularly important enzyme related to treatment failure in serious infections and is produced by various species of Enterobacteriaceae [5][6][7][8][9]. Resistance to carbapenems among Enterobacteriaceae could also be mediated by other families of carbapenemases including NDM, IMP, VIM and OXA-48, in which outbreaks of K. pneumoniae producing these enzymes have been described [8], [10][11][12][13]. Other non-enzymatic mechanisms were also related to carbapenem resistance. For example, a decrease in expression of outer membrane proteins (OMPs), especially OmpK35 and OmpK36 porins, has been shown to be associated with increased MICs of both cephalosporins and carbapenems in K. pneumoniae, including outbreak strains [14], [15]. However, it may not be the major component of carbapenem resistance, i.e. only carbapenemase production without defective porin expression could strongly elevate carbapenem MICs [16], [17].
K. pneumoniae is known to be the most common species among CRE, hereafter called carbapenem-resistant K. pneumoniae (CRKP). CRKP has been reported in many countries and their molecular epidemiology based on multilocus sequence typing (MLST) demonstrates that various sequence types (STs) are widespread [18]. For example, K. pneumoniae ST258 is most commonly associated with carbapenem resistance in the United States and Greece [10], [19]. The ST258 clone belongs to the multidrug-resistant clonal complex (CC) 258 of K. pneumoniae that is commonly associated with KPC-producing K. pneumoniae strains. Interestingly, it has been reported that other members of CC258, especially ST11, has become predominant clones in disseminating both bla KPC and bla NDM , or in association with carbapenem resistance, in Brazil and many Asian countries [20][21][22][23][24]. A recent survey in Russia also reported that ST340, another member of CC258, was associated with NDM-1-producing K. pneumoniae isolates [25]. Therefore, a rising trend of K. pneumoniae CC258 in the spread of CRE is worthwhile for further investigation.
In Thailand, an overall prevalence of ESBL-producing K. pneumoniae during 2001-2010 was upraised from 26% to 39% (data from the Ministry of Public Health, Thailand). Probably related to the surge of ESBL producers, a survey of carbapenem use in Thailand during 2010-2013 showed a significant increase from approximately 2.1 to 3.1 million vials per year. However, very little is known about the molecular epidemiology and genetic characteristics of CRE in Thailand. Here we investigated the prevalence and resistance characteristics of CRE at the largest university hospital in Thailand with an emphasis on the characterization and molecular typing of CRKP.

Materials and Methods
Bacterial strains and antimicrobial susceptibility testing A total of 12,741 non-duplicated clinical isolates of Enterobacteriaceae obtained from both inpatients (68.3%) and out-patients (31.7%) at the Faculty of Medicine Siriraj Hospital, a 2300-bed university hospital in Bangkok (Thailand), over a 2-year period (2009-2011) were investigated in this study including K. pneumoniae (38.7%), Escherichia coli (37.2%), Enterobacter cloacae (6.0%), Proteus mirabilis (4.2%), Salmonella spp. (4.1%) and other species (9.8%). All Enterobacteriaceae isolates were determined for their susceptibility to various antimicrobial agents by standard disk diffusion method using Kirby-Bauer disk (Oxoid, UK) according to clinical breakpoints recommended by the Clinical Laboratory Standards Institute (CLSI) [26]. Isolates that were resistant to at least a carbapenem agent were designated as CRE according to the definition recommended by the Centers for Disease Control and Prevention (available at www.cdc.gov, last updated June 29, 2015) and included for further study. In addition, isolates that were resistant to ertapenem, but not other carbapenems, and resistant to extended-spectrum cephalosporins were also included. The study isolates were confirmed by measuring minimal inhibitory concentrations (MICs) of all four carbapenems (ertapenem, imipenem, meropenem or doripenem) as well as ceftazidime and ciprofloxacin by using Etest method (bioMérieux, France) and interpreted based on the CLSI's guideline [26]. Tigecycline was additionally tested against CRKP isolates by using Etest method and was interpreted based on the European Committee on Antimicrobial Susceptibility Testing (EUCAST)'s recommendation of S 1/R >2 μg/mL [27]. MIC 50 and MIC 90 of each tested agent were determined. Study isolates were also phenotypically characterized for the production of ESBL (using combination disk test) and carbapenemase (using modified Hodge test, MHT) by following the CLSI's guideline [26].

Genetic studies of beta-lactamase (bla) genes
Genomic DNA samples from all CRE isolates were sought by PCR sequencing for various bla genes including carbapenemase-related genes (bla KPC , bla IMP , bla VIM , bla NDM and bla OXA-48 ) and other non-carbapenemase-related genes (bla TEM , bla SHV , bla CTX-M , bla OKP , bla VEB , bla OXA and bla AmpC ) using primers and conditions described previously [28][29][30][31]. In addition, CRKP isolates were also determined for bla GES , another member of class A carbapenemase gene, according to a published study [32].

Expression study of outer membrane proteins
Expression of genes encoding outer membrane proteins (OmpK35 and OmpK36 porins) was genetically determined from their mRNA levels for selected CRKP isolates using real-time reverse transcription PCR. A total RNA was extracted and reversely transcribed into cDNA using High Pure Isolation Kit (Roche Diagnostics, USA) and iScript TM Reverse Transcription Supermix (Bio-Rad Laboratories, USA) according to the manufacturers' recommendations. The expression of porin-encoding genes, ompK35 and ompK36, relative to rpoB gene was determined in triplicates by real-time PCR using specific primers and conditions as described previously [33]. Relative expression was analysed by Bio-Rad CFX manager software using 2-Δ (ΔCT) formula, where ΔCT represented the difference of cycle threshold (CT) of gene target and normalizer, and Δ(ΔCT) represented the difference of ΔCT of tested isolate and ΔCT of K. pneumoniae ATCC 13833 which was used as the control strain to determine relative expression level of both genes. The results from isolates resistant to all four carbapenems were compared to isolates resistant to only some (3) carbapenems. The differences of ompK35 and ompK36 gene expression among these two groups were determined using the non-parametric Mann-Whitney test and were considered statistically significant if P 0.05.

Molecular epidemiology study
Molecular typing of CRKP isolates was identified by using both pulsed-field gel electrophoresis (PFGE) and MLST. For PFGE, chromosomal DNA of study isolates were obtained for molecular typing using a CHEF Mapper XA apparatus (Bio-Rad Laboratories) based on a XbaI (New England BioLabs, USA) digestion protocol [34]. A dendogram was generated by both Dice coefficient and Pearson correlations based on the unweighted pair group method using arithmetic averages (UPGMA) to determine the DNA similarities according to the Fingerprinting II Software, version 3.0 (Bio-Rad Laboratories). Clonal relatedness among isolates was defined if the Dice coefficient correlation was over 80% [35]. The clonal relationships of isolates based on MLST were sought by the similarity of an allelic profile of seven housekeeping genes to assign ST using universal primers and PCR conditions stated by the Institut Pasteur MLST (http://www.pasteur.fr/mlst).

Ethics statement
Only bacterial isolates recovered and leftover from routine diagnostic laboratory were used in this study without a direct use of clinical specimens. Patient consents were not required. The study was ethically approved by the Siriraj Institutional Review Board, Faculty of Medicine Siriraj Hospital, Mahidol University (approval number Si 454/2009).

Genetic characterization of bla genes among CRE isolates
The results from the study of bla genes among CRE isolates are summarized in Table 2    were ESBL-positive, and 9 isolates (25%) were MHT-positive. Most CRKP harbored multiple bla genes, especially in the TEM, SHV and CTX-M families (Tables 2 and 3). bla SHV was found in most CRKP isolates (91.7%). Twenty-eight isolates harbored at least one ESBL genes (SHV-2, SHV-12 and/or CTX-M-15 isolates that were resistant to all four carbapenems and isolates that were resistant to some carbapenems, there was no significant difference in ompK35 expression (P = 0.96). However, isolates resistant to all four carbapenems showed statistical significance in ompK36 expression when compared to isolates resistant to some carbapenems (P = 0.01). According to carbapenem MIC values shown in Table 3, changes in ompK35 expression was not significant for resistant isolates comparing to non-resistant isolates for imipenem (P = 0.96), meropenem (P = 0.52) and doripenem (P = 0.52). However, a decrease of ompK36 expression was statistically associated with isolates resistant to imipenem (P = 0.01), meropenem (P = 0.05) and doripenem (P = 0.05).

Molecular epidemiology of CRKP isolates
Clonal relatedness of all CRKP isolates was determined by two typing methods, PFGE and MLST, as shown in Fig 1. According to the PFGE typing, these CRKP isolates were belonged to several clones and no more than two isolates in each PFGE type appeared in the same clone.

Discussion
A worrisome problem of multidrug-resistant bacteria is expanding worldwide, including in the Asia-Pacific region. Previous surveys showed that the rates of ESBL-producing Enterobacteriaceae in Thailand were approximately 40-50% and were relatively high comparing to other countries in the Asia-Pacific region [36], [37]. However, the prevalence of CRE in Thailand is not clearly known. A previously published study demonstrated that carbapenem non-susceptible Enterobacteriaceae in Thailand was approximately 0.4% among isolates from intraabdominal infection, bloodstream infection and nosocomial pneumonia [37]. The overall prevalence of CRE in Thailand remained low (1.4%) during our study period, while the rate of ESBLrelated bla gene carriage among CRE was very high. MHT was determined for the detection of carbapenemase production [26]. Our results, however, showed that MHT was not well correlated with the detection of CRE as the test was positive for 26 isolates (14.4%) and, among these isolates, only three and four isolates were positive for class A (bla KPC ) and class B (bla IMP ) carbapenemases, respectively. Many studies have reported that phenotypic-based MHT often yielded low sensitivity and specificity for detection of carbapenemases, especially for class B carbapenemases [38], [39]. Besides the possibility of false-positive result due to a low reliability of MHT, other mechanisms may play a role in carbapenem resistance among these isolates. For example, a majority of CRE isolates in this study were E. cloacae but only one isolate of carbapenem-resistant E. cloacae was identified to carry a carbapenemase gene. At this center, approximately 16.1% of E. cloacae were identified as CRE, whereas only 0.7% of K. pneumoniae were CRE. E. cloacae has been shown to possess multiple mechanisms other than carbapenemase production that may confer carbapenem resistance, especially AmpC hyperproduction and decreased porin expression [40]. Therefore, MHT has a very limited role in the detection of CRE and its result should be carefully interpreted. Although KPC-producing Enterobacteriaceae have been reported to spread rapidly in the last decade, the prevalence in Thailand remains very low. This study demonstrated that the rate of bla KPC -carrying isolates was only 0.02% among Enterobacteriaceae and 1.7% among CRE isolates. We also reported previously from this survey that it was the first emergence of bla KPC in Thailand [41]. More interestingly, none of them were found in K. pneumoniae isolates, the most common host of bla KPC , and they were all characterized as a novel variant designated bla KPC-13 . The bla IMP-14a has recently been reported among the emerging Enterobacteriaceae isolates carrying New Delhi metallo-beta-lactamase (NDM)-1 in Thailand [42], but has not been reported from other countries. In addition, this gene has only been detected in K. pneumoniae. This may indicate a local spread of bla IMP-14a among K. pneumoniae in Thailand. It is noteworthy that only 3.9% of CRE isolates were detected for important carbapenemase gene families (bla KPC and bla IMP ). Therefore, the dissemination of carbapenemase genes among CRE in Thailand is very low and unique from other countries, and other mechanisms would play a more important role in carbapenem resistance among Enterobacteriaceae.
Most CRKP isolates were highly resistant to ceftazidime and ciprofloxacin. It should even more concerned for susceptibility to tigecycline. Many previous reports showed that tigecycline was active against CRE, but in our study only 47.2% of CRKP isolates remained susceptible to tigecycline. This suggests that CRKP may cause a serious difficulty to find appropriate treatment and a close monitoring for the spread of CRKP is highly recommended. The extensive genetic characterization of bla genes among CRKP demonstrated that most isolates harbored multiple bla genes, especially ESBL-related genes. Among isolates carrying ESBL genes, the bla CTX-M-15 was most commonly found, followed by bla SHV-12 . However, the results of ESBL phenotypic test were not completely correlated with the genotypic study. Similarly, the detection of carbapenemase-related bla gene was not well related to the MHT results and carbapenem MIC values of these CRKP isolates. Therefore, the MHT result has a limitation to predict for the production of carbapenemases, as well as the susceptibility levels of CRKP against carbapenem agents, and the detection of bla carbapenemase genes may only partially explain carbapenem resistance among CRE. Due to a very limited choice of treatment, results of antimicrobial susceptibility testing should be carefully considered as a therapeutic guidance against CRKP.
Multiple mechanisms can contribute to carbapenem resistance among CRKP. Besides the production of carbapenemases, the loss or reduction of porins, OmpK35 and/or OmpK36, has been shown to correlate with increased carbapenem MICs [14][15][16][17]. Seventeen CRKP isolates, including seven isolates resistant to all four carbapnems and ten isolates resistant to some (3) carbapenems, were selected to study the expression of porin proteins. Only one of isolates resistant to all four carbapenems carried a carbapenemase gene, bla IMP-14a . Most isolates, except two, demonstrated a decrease in expression of ompK35 or ompK36 gene, or both, at varying degrees. A reduction of ompK35 expression did not show a significant difference among isolates resistant to all four carbapenems and isolates resistant to some carbapenems, and did not correlate with levels of increased MIC values of each carbapenem among these isolates. However, it was revealed that isolates resistant to all four carbapenems had a significant decrease of ompK36 expression comparing to isolates resistant to some carbapenems. In addition, a decrease of ompK36, but not ompK35, expression was statistically associated with individual carbapenem resistance (imipenem, meropenem and doripenem) among CRKP isolates. This suggests that a decrease of ompK36 expression may be involved in carbapenem resistance with a higher magnitude than does a decrease of ompK35 expression. In addition, it supports previous studies that porin deficiency contributes with a lesser role in carbapenem resistance comparing with the production of beta-lactamases [17], [43].
The PFGE study indicated that various PFGE patterns of CRKP were present and there was no predominated clone. Thus, multiple CRKP clones were disseminated through different hospital services. However, the MLST study revealed that ST340 was the most significant clone among CRKP isolates in this study, with an evidence of its community spread since it was also recovered from ambulatory patients. A half of CRKP isolates were members of the drug-resistant K. pneumoniae CC258 (ST340, 41.7% and ST11, 8.3%). Isolates of CC258 have been shown to be associated with the carriage of KPC gene and international spread in many countries, including North America, Latin America and Asia, and mostly ST11 and ST258 are the common types [20], [22], [24]. ST258 is currently known to be an epidemic clone of KPC-producing K. pneumoniae. Interestingly, KPC gene was not detected in our CRKP isolates and none of isolates were ST258. The epidemiologic information of CRKP in Thailand is unknown. We here report that ST340, a member of CC258, is the most common type among CRKP isolates in Thailand. ST340 clone was sporadically reported, e.g. from Israel and Russia [25], [44], but has never represented the predominant clone. In addition, ST48, the second most common clone in this study, has also rarely been reported among CRKP isolates. Several other ST types were found among CRKP isolates, including the novel ST1645 clone. An isolate of ST101 was found in this study. This clone has recently been reported to be the predominant clone of K. pneumoniae carrying beta-lactamase genes among companion animals in Italy, in which CC258 isolates were also identified from this source [45]. In addition, ST101 clone was associated with NDM-1-producing K. pneumoniae in many regions [46]. This may also suggest a possibility of companion animals as vector to transfer resistant genes to human and further study should be considered. Our study also demonstrated that PFGE and MLST studies were in partial concordance. Isolates in the CC258, ST340 and ST11, were clustered more closely based on PFGE patterns, while non-CC258 members were more distant. These data hence suggest that CRKP isolates in Thailand have a unique epidemiologic characteristic and appear to be widespread in both in-and out-patient services.
In conclusion, we report an extensive survey for the emergence of CRE at the largest university hospital in Thailand. The prevalence of CRE was low but should be closely monitored. Most CRE isolates harboured multiple bla genes. Only a few isolates harboured a carbapenemase gene, bla KPC-13 or bla IMP-14a , which were particularly found in Thailand. A decreased porin expression in selected CRKP isolates was not directly concordant with levels of increased carbapenem MICs. Therefore, further investigation is required to better understand the mechanisms that contribute to carbapenem resistance among CRE isolates. The PFGE and MLST studies revealed that multiple clones of CRKP were widespread. CC258, especially ST340, was the most predominant clone. Our findings represent the novel epidemiologic data of CRKP in Thailand and ones should be an alert for continuous surveillance of CRE and CRKP in this region to promptly control these highly resistant bacteria.