New Delhi Metallo-Beta-Lactamase-Producing Enterobacteriaceae in South Korea Between 2010 and 2015

This study was carried out to investigate the epidemiological time-course of New Delhi metallo-beta-lactamase- (NDM-) mediated carbapenem resistance in Enterobacteriaceae in South Korea. A total of 146 non-duplicate NDM-producing Enterobacteriaceae recovered between 2010 and 2015 were voluntarily collected from 33 general hospitals and confirmed by PCR. The species were identified by sequences of the 16S rDNA. Antimicrobial susceptibility was determined either by the disk diffusion method or by broth microdilution, and the carbapenem MICs were determined by agar dilution. Then, multilocus sequence typing and PCR-based replicon typing was carried out. Co-carried genes for drug resistance were identified by PCR and sequencing. The entire genomes of eight random selected NDM producers were sequenced. A total of 69 Klebsiella pneumoniae of 12 sequence types (STs), 34 Escherichia coli of 15 STs, 28 Enterobacter spp. (including one Enterobacter aerogenes), nine Citrobacter freundii, four Raoultella spp., and two Klebsiella oxytoca isolates produced either NDM-1 (n = 126), NDM-5 (n = 18), or NDM-7 (n = 2). The isolates co-produced CTX-M-type ESBL (52.1%), AmpCs (27.4%), additional carbapenemases (7.1%), and/or 16S rRNA methyltransferases (4.8%), resulting in multidrug-resistance (47.9%) or extensively drug-resistance (52.1%). Among plasmids harboring blaNDM, IncX3 was predominant (77.4%), followed by the IncFII type (5.8%). Genome analysis revealed inter-species and inter-strain horizontal gene transfer of the plasmid. Both clonal dissemination and plasmid transfer contributed to the wide dissemination of NDM producers in South Korea.


INTRODUCTION
The first New Delhi metallo-beta-lactamase (MBL), NDM-1, was identified in 2008 in an extensively drug-resistant (XDR) Klebsiella pneumoniae clinical isolate recovered from a urinary specimen of a patient (Yong et al., 2009). The enzyme had an astonishing range of betalactam substrates, including penicillins, late-generation cephalosporins, and carbapenems with the exception of monobactams (Yong et al., 2009). The coding gene was found to be located in a ∼180-kb plasmid that is readily transferable by conjugation (Yong et al., 2009). Further epidemiology revealed that the gene had already spread to various species of bacteria including K. pneumoniae, Escherichia coli, Enterobacter spp., Morganella morganii, and Acinetobacter baumannii throughout the six continents (Johnson and Woodford, 2013), implying that the plasmid encoding NDM has an enormous ability to spread.
Since the identification of the first NDM variant (NDM-2) in A. baumannii (Kaase et al., 2011), 17 subtypes of NDM variants have been functionally characterized (Liu et al., 2017). The NDM subtypes, which differ from the NDM-1 prototype by one to five amino-acid substitutions, exhibit different levels of hydrolyzing activity against carbapenems and other betalactam substrates, depending on the substitutions. Among the first seven subtypes, NDM-7 containing two substitutions of Asp130Asn and Met154Leu has the greatest carbapenemhydrolyzing activity, followed by NDM-5 with Val88Leu and Met154Leu, NDM-6 with Ala233Val, and NDM-1 (Rahman et al., 2014).
The bla NDM gene has been found in plasmids of diverse replicon types, such as IncF, IncX3, IncL/M, IncH, and IncA/C, and has an extensive host range. In plasmids, the gene is frequently associated with mobile genetic elements such as transposons and insertion sequences (ISs). The complete form of the transposable element of the bla NDM gene Tn125 was first identified in A. baumannii, suggesting that Acinetobacter spp. are the reservoir of the gene . Tn125 is bracketed by two copies of ISAba125. One copy of ISAba125 is located upstream of the bla NDM gene, and the ble MBL gene encoding resistance to the anti-tumor glycopeptide bleomycin is at 3 ′ end of the gene, followed by eight ORFs and the 3 ′ copy of ISAba125. The composite form of the transposon is common in A. baumannii, and at least in this species, the mobility of the bla NDM gene is mainly attributed to the transposon . Unlike in A. baumannii, the Tn125 transposon of Enterobacteriaceae is in a truncated form that is no longer mobile. This indicates that plasmids are important for bla NDM mobility among Enterobacteriaceae (Partridge and Iredell, 2012).
To monitor the spread of carbapenemase-producing Enterobacteriaceae (CPE), the Korea Center for Disease Prevention and Control (KCDC) had run a national surveillance system since 2010. In this study, the NDM producers collected from this surveillance system between 2010 and 2015 were investigated for the epidemiological time-course of NDMmediated carbapenem resistance by analyzing the clones, plasmids, and genetic contexts of the bla NDM gene. In order to evaluate the inter-species and inter-strain horizontal gene transfer, WGS was performed to characterize the main mobile genetic elements responsible for NDM dissemination.

Bacterial Strains
During the period between 2010 and 2015, 146 non-duplicate NDM producers were voluntarily collected from 33 general hospitals composing the National Laboratory Surveillance System for carbapenemase-producing Enterobacteriaceae (CPE) of the Korea Center for Disease Prevention and Control. Further testing including the species identification by using Bruker MALDI MS instrument (Bruker, Billerica, MA, USA) and by 16S rDNA sequencing was performed in a reference laboratory.

Antimicrobial Susceptibility Testing and Resistance Genotyping
Antimicrobial susceptibility to eight drugs (aztreonam, cefotaxime, ceftazidime, cefoxitin, gentamicin, amikacin, ciprofloxacin, and tigecycline) was evaluated by the disk diffusion method on Mueller-Hinton (MH) agar (Difco Laboratories, Detroit, MI, USA) in accordance with CLSI guidelines (CLSI, 2015b). The MICs for colistin were assessed by the broth microdilution method with MH broth (Difco Laboratories) in accordance with the recommendations of The joint CLSI-EUCAST (2016). The MICs for imipenem and meropenem were determined by the agar dilution method in accordance with CLSI guidelines (CLSI, 2015a). Both E. coli ATCC 25922 and Pseudomonas aeruginosa ATCC 27853 were used for quality control. The bla NDM gene was detected by PCR with gene-specific primers and direct sequencing was carried out for NDM subtyping . Genetic contexts for the vicinity of the bla NDM gene were explored by PCR mapping using primers in the supplemental figure. The confirmed NDM producers were further evaluated by PCR and by direct sequencing to identify the following additional resistance determinants: bla OXA−48−like , bla IMP , bla VIM , and bla GES for carbapenemases ; bla CTX−M−1−like , bla CTX−M−9−like , and bla CTX−M−25−like for CTX-M-type extended-spectrum-beta-lactamases (ESBLs); bla ACT , bla ACC , bla CMY , and bla DHA for plasmid-mediated AmpCs; and armA, rmtA, and rmtB for 16S ribosomal methyltransferases (Pérez-Pérez and Hanson, 2002;Ryoo et al., 2005).

MLST
PCR and sequencing were carried out for seven housekeeping genes per species: gapA, infB, mdh, pgi, phoE, rpoB, and tonB for K. pneumoniae (Diancourt et al., 2005) and adk, fumC, gyrB, icd, mdh, purA, and recA for E. coli (Wirth et al., 2006) and the sequences were compared in the multilocus sequence typing (MLST) database, so that allelic numbers and sequence types (STs) could be determined.

Plasmid Transfer by Bacterial Conjugation and Electroporation
For bacterial conjugation, spontaneous rifampin-resistant mutants from E. coli J53 were used as recipients. Equal amounts of exponential cultures of the donor and recipient isolates were mixed, incubated in MH broth for 12 h, and spread on Brain Heart Infusion (BHI) agar (Difco Laboratories) containing rifampin (30 mg/L), sodium azide (100 mg/L), and imipenem (1 mg/L). Each colony was tested by disk diffusion test and confirmed by PCR. The plasmid transfer frequency was calculated as the number of transconjugants per donor. The non-conjugative plasmids were electrotransferred into One Shot TM TOP10 Electrocomp TM E. coli (ThermoFisher Scientific, Waltham, MA, USA) following the manufacturer's instruction and selected on BHI supplemented with imipenem 4 mg/L.

Whole Genome Sequencing
Eight NDM producers were entirely sequenced: one Enterobacter sp. with an IncX3 plasmid encoding NDM-1; four E. coli isolates with IncX3 plasmids encoding any of NDM-1, −5, and −7; one Citrobacter freundii with IncX3 plasmid encoding NDM-5; two K. pneumoniae strains co-producing OXA-181 and NDM-5 which the bla NDM−5 gene was carried by the IncFII plasmid. Bacterial whole genomes were sequenced with single-molecule real-time sequencing on a PacBio RSII instrument (Pacific Biosciences, Menlo Park, CA, USA) as previously described (Hong et al., 2016). De novo genome assembly was performed with the PacBio SMRT analysis software suite (version 2.3.0) and the annotations of coding sequences, tRNA sequences, and rRNA sequences were performed using the NCBI Prokaryotic Genome Annotation Pipeline.

Occurrence of NDM Producers Between 2010 and 2015
In 2010, four NDM-1-producing K. pneumoniae ST340 isolates were identified in a single hospital outbreak in Seoul (Kim et al., 2012). In 2011, five isolates of the same clone were successively identified at the same hospital and one and three NDM-1producing E. coli ST101 isolates were identified at another hospital in Seoul and at a hospital in the southeastern region of the Korean peninsula, respectively. The NDM-1-producing K. pneumoniae ST340 was identified in the central part of the peninsula in 2012 and in 2013, diverse NDM-producing bacterial pathogens such as C. freundii and Enterobacter spp. were identified in a broader region, including the southern peninsula. By 2014, the number of NDM producers had explosively increased from eight to 22, with even more diverse bacterial pathogens, including Klebsiella oxytoca and Raoultella planticola. Finally, in 2015, among the 102 NDM producers identified over the peninsula, NDM-5 and NDM-7 producers were identified. Three of the 18 NDM-5 producers were previously reported as NDM-5 and OXA-181 co-producer (Sun et al., 2016).
Since the genetic environment of the bla NDM gene is diverse in terms of size and Tn125 truncation, the vicinity of the gene was evaluated by PCR mapping. For the upstream region of the gene, all plasmids had a promoter region located at the 72-bp extremity of ISAba125. One exceptional case having imipenem and meropenem MICs of 16 and 32 mg/L, respectively, carried a relatively short IS125 remnant at the upstream of the bla NDM gene, remaining the promoter sequences. Downstream of the gene, 17 cases of truncated Tn125 ended with the groEL gene (the 6th ORF from bla NDM among eight) and thus the most preserved, whereas one case ended with ble MBL next to the bla NDM gene (the most shrunken case). Around half of the cases (n = 66, 45.2%) ended with the tat gene, the 4th ORF from the bla NDM gene, while 34.2% (n = 50) terminated with trpF, the 2nd ORF from the gene. The Tn125 transposons carrying bla NDM−1 were truncated at heterogeneous sites along the transposon, while those carrying bla NDM−7 ceased at the tat gene and those carrying bla NDM−5 were truncated at the end of either tat or trpF. No obvious correlation was observed between carbapenem susceptibility and the Tn125 truncation locus.
IncX3 Plasmids Carrying Three Subtypes of bla NDM and IncFII Plasmids Carrying bla  As the IncX3 plasmids were the most common among NDM producers, regardless of the NDM subtype or bacterial host, we completely sequenced a panel of six NDM producers that possessed IncX3 plasmids carrying any of three NDM subtypes (Figure 1, Table 4). Two OXA-181 and NDM-5 co-producers in which the bla NDM−5 gene was carried by the IncFII plasmid were added to the panel for comparative analysis. The eight isolates exhibited MDR or XDR resistance phenotypes (Figure 1,  Table 4). The Enterobacter sp. CRENT-193 was an isolate in 2013, while K. pneumoniae CRKP-1215 was in 2014, and the remaining six were in 2015 (Table 4). Three isolates (Enterobacter sp. CRENT-193, K. pneumoniae CRKP-1215, and CRKP-2297) were from Seoul, among which both K. pneumoniae isolates coproducing OXA-181 and NDM-5 were from one hospital with  one-year interval. Two E. coli isolates (CREC-544 and CREC-591) and C. freundii CRCB-101 were from Incheon, a suburb of Seoul, among which E. coli CREC-591 and C. freundii CRCB-101 were from a single hospital outbreak. The remaining E. coli CREC-532 and CREC-629 isolates were recovered in the southwestern and southeastern peninsula, respectively. The IncX3 plasmids harboring bla NDM−5 or bla NDM−7 had consistent sizes of 46,137 and 46,161 bp, respectively, whereas the two plasmids harboring the bla NDM−1 gene were 44,962-and 51,455-bp long. The six IncX3 plasmids had conserved backbones including elements of conjugal transfer, the replication origin, and other accessory genes. The differing lengths of the plasmids were due to the excision of a 1,175-bp IS5 copy upstream of the truncated Tn125 in pCRENT-193_3 and the insertion of a 5,294-bp fragment including bla SHV−12 in pCREC-544_2 (Figure 1). Of note, the plasmids pCREC-591_4 and pCRCB-101_1, which were identified from different pathogenic species (E. coli ST101 and C. freundii) in a single hospital outbreak, were identical, clarifying the inter-species transfer of the plasmid. The six isolates possessed various resistance determinants in their chromosomes. Interestingly, CTX-M-type ESBLs and plasmidmediated AmpCs, which are known to be more common in plasmids than in the chromosome, were frequently encoded in the chromosome (Table 4). Further, except for C. freundii CRCB-101, all isolates co-carried additional plasmids with multiple genes for drug resistance ( Table 4).
Both IncFII plasmids (pCRKP-1215_2 and pCRKP-2297_2) from K. pneumoniae ST147 isolates recovered at a hospital were identical, even they were identified one year apart. The coproduced OXA-181 was encoded in the chromosome by two copies of bla OXA−181 and the chromosomes of the two isolates were very similar to each other, sharing 99% nt identity on 100% coverage. Taken together, these results clearly demonstrate the persistence of the clone in the clinical setting. The IncFII plasmid itself conferred MDR, due to the aadA3, dfrA23, and sul1 genes associated with the class 1 integron, the bla TEM−1B accompanied by Tn3, and the erm(B), mph(A), and rmtB genes (Figure 1 and Table 4). A set of three plasmids (two different IncFII and one non-typeable) was carried together by the isolates.

DISCUSSION
The epidemiology of NDM-mediated carbapenem resistance is complex because the dissemination involves multiple factors, such as the clonal spread of resistant strains and the interstrain and inter-species horizontal transfer of the resistance determinants. The global NDM producers include diverse species and genera within Gram-negative bacteria and a wide range of clones within individual species, as indicated by ST. The NDMencoding plasmids and the mobile elements of the bla NDM gene are also complicated in South Korea depend on the bacterial species and the strains. The mobility of the plasmid is critical for the dissemination of bla NDM . NDM-producing clones possessing a transferable plasmid with bla NDM have been demonstrated to spread bla NDM efficiently, while a non-conjugative plasmid is reliable for clonal spread (Kumarasamy et al., 2010).
Until 2012, the occurrence of NDM producers in South Korea was regionally restricted and the constituents were simple. Beginning in 2013, the NDM producers became diverse at the species level and in 2014, a range of varied clones began to be identified within individual species. Finally, in 2015, two additional NDM variants were recognized within even more diverse bacterial hosts. Thus, the population of NDM producers in South Korea achieved the epidemiological complexity and diversity as worldwide NDM population. The IncX3 plasmid carrying the bla NDM gene, which is either transferable or nonmobile, was likely responsible for the dissemination of NDM producers over the peninsula.
The six circularized IncX3 plasmids had a relatively simple structure composed of conjugal transfer elements, accessory FIGURE 1 | Schematic representation of the bla NDM -harboring IncX3 and IncFII plasmids. Name of the plasmids and the size in parenthesis are denoted for the plasmids, and the district of isolation are indicated on the map of Korean peninsula in front of the plasmid name. Name of the plasmids from E. coli are indicated in black, from Enterobacter spp. in blue, and from K. pneumoniae in red. Open arrows indicate coding sequences for conjugal transfer (light green), drug resistance (red), replication origin (blue), toxin/antitoxin system (black), transposase (yellow), and integrase (green). Insertion sequences are indicated in yellow boxes. The Tn125 is marked with a red-bordered open box, and the class 1 integron is indicated with a green-bordered open box. Genetic variants of each plasmid referred by pCREC-629_3 are indicated on the right side of the schematic representation of the plasmid for locating the point of substitution: arrowheads, deletion; square bracket, insertion. Together with the nt substitutions, details of insertion are indicated with +, and that of deletion with in the end of the schematic presentation of the plasmid. genes, and a single resistance determinant (bla NDM ). Only one IncX3 plasmid possessed one bla SHV−12 gene in addition to the bla NDM gene. The IncX3 plasmids identified in this study were identical to those from the Middle East, which is a recently added reservoir of NDM producers (Zowawi et al., 2013). This simple plasmid, in terms of resistance gene carriages, has a broad host range, which greatly enhances its inter-genera and inter-species dissemination. Actually, in our collection of isolates, more than threequarters of the bla NDM -possessing plasmids were IncX3 type and were carried by varied clones of various species of bacteria.
All NDM producers were resistant to more than three classes of antimicrobial drugs. Further, the NDM producers, especially those possessing the IncX3-type bla NDM plasmid, were vested in MDR, either due to supplemental plasmid(s) containing several genes for antimicrobial resistance, as we observed in this study, or due to the hybridization of plasmids with additional resistance genes (Sun et al., 2016). The IncA/C-and IncFIItype plasmids were kept within E. coli and K. pneumoniae bacterial hosts, though they are known to have broad host ranges; thus, their spreads were limited both temporally and regionally.
The NDM-5 and NDM-7 producers, even with relatively small numbers of isolates, had higher carbapenem MICs than the NDM-1 producers ( Table 1). These subtypes share a substitution of Methionine to Leucine at position 154, identical to that in NDM-4 , and have greater MICs than NDM-1, not only for carbapenems, but also for 3rd-and 4thgeneration cephalosporins such as cefotaxime, ceftazidime, and cefpirome (Hornsey et al., 2011). Additionally, as shown by both entirely sequenced co-producers, co-production of OXA-181 with NDM-5, could contribute to elevate the carbapenem MICs. Notably, the bla NDM−5 gene, which was observed in both IncX3 and IncFII plasmids, was initially identified in an IncF-type plasmid possessed by E. coli ST648 (Hornsey et al., 2011). And the bla NDM−7 gene, which was identified only in the IncX3 plasmid, was first identified in the IncX3 plasmid (Göttig et al., 2013).
It is noteworthy that a quite systematized alliance with other antibiotic resistance determinants has been observed in almost all NDM producers. The additional antimicrobial resistance genes for plasmid-mediated AmpC cephalosporinases,  CTX-M-type ESBLs, and 16S RNA methyltransferases have enhanced the MDR of NDM producers. This strategy has facilitated the persistence of pathogens that have encountered combination antimicrobial therapies in clinical settings. Clonal dissemination played a major role in the dissemination of NDM producers in South Korea and the IncX3 plasmid possessing the bla NDM gene collaborated for the gene spread, at least among the entirely sequenced strains. NDM producers pose a major healthcare risk since the isolates are usually MDR, and the plasmids are frequently transferable to a wide range of Gram-negative bacteria. This time-course observation of the genomic epidemiology of NDM has warned us of the great public health threat of NDM producers, raising the importance of national policy to combat drug resistance.

ETHICS STATEMENT
The research, which has no involvement of human subject but the clinical isolates, does meet the exempt category without approval from Ethics Committee on Human Research of the Health Ministry in South Korea and the study design has not been reviewed by the committee.

AUTHOR CONTRIBUTIONS
KL and SJ conceived the project. E-JY and SJ wrote the manuscript. E-JY, DaK, and JY performed the experiment. E-JY, SJ, and KL analyzed and interpreted the data. JY, DoK, HL, and KL maintained the collections.