From Mar, 2014 to Oct, 2014, a total of 87 carbapenem-resistant Enterobacteriaceae (CRE) isolates causing clinical infections isolated from various specimens of patients at the First Affiliated Hospital of Wenzhou Medical University in Wenzhou, east China, were investigated for carbapenemase genes. The isolates were identified as E. coli by an automated microbiology analyzer (bioMe'rieux, Marcy l'Etoile, France) in accordance with the manufacturer's instructions.

Gram-negative susceptibility (GNS) card on the Vitek system (bioMe'rieux, Marcy l'Etoile, France) was performed initially for antimicrobial susceptibility testing. Disk diffusion method was used for further confirmation and antimicrobial susceptibility results were interpreted according to the criteria recommended by Clinical and Laboratory Standards Institute (CLSI) (CLSI, 2014). The E-test method was used for the determination of minimum inhibitory concentrations (MICs) of imipenem and meropenem for the E. coli isolate and its transconjugant. E. coli ATCC 25922 was used as control strain for antimicrobial susceptibility testing.

The modified Hodge test (MHT) was further performed on a Mueller-Hinton agar plate with ertapenem as substrate for the detection of carbapenemases as described previously (CLSI, 2014). MBLs were determined using a double-disc synergy test (Peleg et al., 2005). ESBLs were tested using the CLSI-recommended confirmatory double disk combination (CLSI, 2014).

The carbapenemase genes responsible for carbapenem resistance, including bla_KPC, bla_GES, bla_SPM, bla_IMP, bla_VIM, bla_SPM, and bla_NDM, were detected using PCR and DNA sequencing as described previously (Queenan and Bush, 2007; Nordmann et al., 2011). ESBLs genes were detected in accordance with the method described previously (Andrade et al., 2010). PCR products were analyzed by electrophoresis in 1% agarose gels and were sequenced on both strands.

In order to determine whether carbapenem resistance was transferable in E.coli DC33 strain, filter mating conjugation was performed using E. coli 600 as the recipient as previously described (Wang et al., 2004). Plasmid DNA of E. coli DC33 strain was extracted with the QIAGENPlasmid Midi kit (Hilden, Germany) according to the manufacturer's instructions. The plasmid extracts were transferred into E. coli DH5α by using chemical transformation and transformants were selected on Luria-Bertani agar plates containing imipenem (0.5 μg/ml).

Multi-Locus Sequence Typing (MLST) was performed on E. coli DC33 using amplification of internal fragments of the seven housekeeping genes including adk, fumC, gyrB, icd, mdh, purA, and recA of E. coli according to MLST website (http://mlst.warwick.ac.uk/mlst/dbs/Ecoli).

The total bacterial DNA of E. coli DC33 was first prepared in agarose plugs, digested with S1 nuclease and further separated by pulsed-field gel electrophoresis (PFGE), as described previously (Chen et al., 2011). Then, the DNA bands were transferred horizontally to a nylon membrane (Millipore). A digoxigenin-labeled bla_NDM-13 probe was used to hybridize with DNA bands and a nitro-blue tetrazolium/5-bromo-4-chloro-3′-indolylphosphate color detection kit (Roche Applied Sciences) was applied to detected hybridization signals.

In order to completely characterize the plasmid from the transconjugant of E. coli DC33 (designated as pNDM13-DC33), pNDM13-DC33 was isolated, purified, and sequenced using the Illumina MiSeq platform. The sequencing reads were de novo assembled, gaps between contigs were closed, open reading frames (ORFs) were predicted, and annotations were performed as described previously (Chen et al., 2013).

Among 87 CRE isolates, 7 were positive for bla_NDM. After sequencing, E. coli strain DC33 was found to harbor bla_NDM-13. E. coli strain DC33 was isolated from a urine culture of a 64-year-old male hospitalized for prostatic hyperplasia in July, 2014. After hospitalized, the patient had the symptom of urinary tract infection. Subsequently, many white cells were found in urine sample under microscope. E. coli strain DC33 was isolated when the patient was hospitalized on day 8. E. coli DC33 was weakly positive for the MHT assay, but β-lactamase activity was inhibited by EDTA, indicating that E. coli DC33 produced a MBL. E. coli DC33 was also positive for CLSI-recommended confirmatory double disk combination test for detecting ESBLs. The results of detection of ESBL genes using PCR showed that E. coli DC33 was also positive for bla_SHV while was negative for other resistance genes tested. After DNA sequencing, bla_SHV was found to bla_SHV-12.

Escherichia coli DC33 exhibited resistance to all antimicrobials tested except tigecycline determined initially by Gram-negative susceptibility (GNS) card on the Vitek system (Table 1), including ampicillin, ampicillin/sulbactam, amikacin, aztreonam, cefotetan, ceftazidme, ceftriaxone, cefepime, ciprofloxacin, ertapenem, gentamicin, imipenem, levofloxacin, nitrofurantoin, piperacillin/tazobactam, tobramycin, and trimethoprim/sulfamethoxazole, The E. coli IOMTU558 carrying bla_NDM-13 located on the chromosome from Nepal was highly resistant to all β lactams tested including ampicillin, ampicillin/sulbactam, cefepime, cefoselis, cefotaxime, cefoxitin, cefpirome, ceftazidime, ceftriaxone, cephradine, doripenem, imipenem, meropenem, and moxalactam (Shrestha et al., 2015). The E. coli IOMTU558 was also resistant to other antibiotics including ciprofloxacin, gentamicin, kanamycin, levofloxacin, and tobramycin, but susceptible to amikacin, colistin, fosfomycin, and minocycline (Shrestha et al., 2015). Tigecycline MICs for E. coli DC33 and E. coli IOMTU558 were 0.05 and 2 μg/ml (Shrestha et al., 2015). The antimicrobial susceptibility pattern of E. coli DC33 was further confirmed by disk diffusion method. The resistance of E. coli DC33 to imipenem and meropenem was further corroborated by E-test method.

MLST result showed E. coli DC33 belonged to ST5138, a single locus variant of ST617. Although ST5138 has been deposited in E. coli MLST database (http://mlst.warwick.ac.uk/mlst/dbs/Ecoli), no study about E. coli ST5138 isolate is published. In our previous study, coexistence of bla_NDM-1 and bla_CMY-42 was found among E. coli ST167 clinical isolates in our hospital (Zhang et al., 2013). As ST5138 was a single-locus variant of S167, we speculate that E. coli DC33 harboring bla_NDM-13 is genetically related to E. coli ST 167 isolates carrying bla_NDM-1 found in our previous study (Zhang et al., 2013). Recently, a Chinese study found an increasing prevalence of E. coli ST167 clinical isolates carrying both bla_NDM-1 and bla_NDM-5 on the conjugative IncX3 plasmid in various parts of China (Huang et al., 2016). Therefore, increasing emergence of bla_NDM variants among E. coli ST167 and ST167 variants clinical isolates should be of concern. Up to now, bla_NDM-13 was only reported in Nepal (Shrestha et al., 2015). The present study is the second report of this novel bla_NDM variant.

S1-PFGE result showed that a ~54-Kb plasmid was found in E. coli DC33 (Figure 1). Subsequently, bla_NDM-13 gene was found to be located on this plasmid, not on chromosome, which was confirmed by Southern-blot (Figure 1). The bla_NDM-13-harboring plasmid of E. coli DC33, designated as pNDM13-DC33, was successfully transferred into recipient E. coli 600 by filter mating conjugation. The antimicrobial resistance patterns of E. coli DC33 and its transconjugant were showed in Table 1. Shrestha et al found that bla_NDM-13 was located within the chromosome (Shrestha et al., 2015). However, bla_NDM-13 was first confirmed to be located on the plasmid in the present study.

Plasmid pNDM13-DC33 is 54.035-bp in length, with an average GC content of 49.03% (Figure 2). BLASTn analysis showed that pNDM13-DC33 is similar to pNDM-HN380, an IncX3-type plasmid carrying bla_NDM-1 among Enterobacteriaceae isolates in China (Ho et al., 2012), with 100% query coverage and >99.9% nucleotide identity (with 8 single nucleotide polymorphisms, SNPs). In China, IncX3-type plasmids carrying bla_NDM variants have been widely found among E. coli clinical isolates with different clones including ST648, ST156, ST131, ST167, and ST3835 clones (Feng Y. et al., 2015; Huang et al., 2016; Wang et al., 2016; Yang et al., 2016). Notably, these similar plasmids have been identified in several hospitals from different geographic regions in China (Wang et al., 2014; Feng J. et al., 2015; Qu et al., 2015), suggesting that pNDM-HN380-like plasmids are common NDM vectors that likely contribute significantly to the dissemination of bla_NDM variants in China.

In accordance with the structure of pNDM-HN380, pNDM13-DC33 consists of a 33-kb backbone encoding plasmid replication (repB), stability partitioning, and transfer (tra, trb, and pil) functions, and a 21-kb antimicrobial resistance region with comparatively high GC content between umuD and mpr genes, suggesting that these two regions were likely acquired and genetically distinct. The resistance region of pNDM13-DC33, containing 16 ORFs sequentially organized as IS26, bla_SHV-12, ygbI, ygbJ, IS26, insE, groL, cutA1, dsbc, ΔtrpF, ble_MBL, bla_NDM-13, ΔISAba125, IS5, ΔISAba125, and Tn3 tnpA, was nearly identical to that of pNDM-HN380, but with the exception that they carry different bla_NDM variants (pNDM13-DC33 with bla_NDM-13 and pNDM-HN380 with bla_NDM-1) (Figure 2). Of note, compared with NDM-1, NDM-13, NDM-3, and NDM-4 had two amino acid substitutions (D95N and M154L), one amino acid substitution (D95N) and one amino acid substitutions (M154L), respectively (Table 2). Although NDM-13 (with two substitutions including the D95N and M154L relative to NDM-1) did not show increased hydrolytic activity against carbapenems, cephalosporins, and penicillins, it increased the affinity of NDM-13 for cefotaxime and affected the catalytic activity of the enzyme against cefotaxime (Shrestha et al., 2015). Our finding that bla_NDM-13-harboring pNDM13-DC33 closely resembles bla_NDM-1-harboring pNDM-HN380 provides evidence that novel bla_NDM variants emerge by sequential mutations of a pNDM-HN380-like plasmid carrying bla_NDM-1.

The chromosomal organization of the bla_NDM-13 gene initially found in the E. coli isolate IOMTU558 from Nepal was similar to that in pNDM13-DC33, except for a 260-bp deletion in ISAba125 of 260 bp (353 to 94 bp upstream bla_NDM-13 start codon). In contrast, the corresponding ISAba125 on pNDM13-DC33 was in full-length (1087 bp), but was interrupted by the insertion of an IS5 (at 265 bp upstream bla_NDM-13 start codon) (Figure 1). A comparison of the chromosomal organization flanking bla_NDM-13 in the E. coli isolate IOMTU558 (Shrestha et al., 2015) with plasmids harboring bla_NDM identified the a set of ordered genes, tnpA-IS30-bla_NDM-13-bleMBL-trpF-dsbC-cutA-groES-groL, that were nearly identical in plasmid pPMK1 from Nepal, plasmidpKPX-1 from Taiwan, plasmid pNDM-MAR from Morocco, and in an Enterobacter hormaechei CCHB10892 plasmid from Brazil (Villa et al., 2012; Huang et al., 2013; Carvalho-Assef et al., 2014; Stoesser et al., 2014; Shrestha et al., 2015). This finding suggests that the chromosomal copy of bla_NDM-13 may be the result of a rare integration event where a region of the plasmid recombined into the E. coli genome.

In conclusion, the present study is the first report of a plasmid-encoded bla_NDM-13 and the complete sequence of a bla_NDM-13-harboring plasmid (pNDM13-DC33). bla_NDM-13 maybe originate from bla_NDM-1 located on a pNDM-HN380-like plasmid by sequential mutations. The emergence of novel plasmid-mediated bla_NDM variants, originating through the mutations in bla_NDM from an epidemic plasmid, poses a concern that NDM variants with different β-lactamases hydrolytic activity will evolve.

The complete nucleotide sequences of plasmid pNDM13-DC33 has been deposited as GenBank accession no. KX094555.

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.