Introduction

Carbapenem-resistant Enterobacteriaceae (CRE) have emerged as a major challenge to global public health. The production of carbapenem-hydrolyzing enzymes (carbapenemases) is the major mechanism mediating resistance to carbapenems in the Enterobacteriaceae. In Escherichia coli, NDM is the most common type of carbapenemase and has a few variants. NDM-4 has an amino acid substation (Met154Leu) compared with NDM-1, which leads to increased activity against carbapenems1. In China, bla NDM-1 and bla NDM-5 are the two most common types of bla NDM variants in the Enterobacteriaceae 2, while bla NDM-4 remains uncommon. During an investigation on the prevalence of carbapenemase genes in carbapenem-resistant Enterobacteriaceae in our hospital, we found a cluster of three E. coli clinical isolates carrying bla NDM-4, which are reported here.

Methods and Materials

Isolates and in vitro susceptibility

The three E. coli isolates were recovered from the rectal swabs of three different patients in 2015 (Table 1). The initial species identification and in vitro antimicrobial susceptibility tests were performed by Vitek II (bioMérieux, Marcy-l'Étoile, France). In addition, MICs of amikacin, aztreonam, ceftazidime, ciprofloxacin, colistin, imipenem, meropenem, piperacillin-tazobactam, tigecycline and trimethoprim-sulfamethoxazole against the isolates were determined using the broth dilution method of the Clinical Laboratory Standards Institute (CLSI)3.

Table 1 Patient demographic data and diseases.
Full size table

Carbapenemase gene screening and phylogenetic group typing

Acquired carbapenemase-encoding genes bla GES, bla KPC, bla IMP, bla NDM, bla OXA-48 and bla VIM were screened using PCR as described previously4,5,6,7. The phylogenetic group for the isolates were determined using PCR as described previously8.

Mating

Filter-based conjugation experiments were performed using the azide-resistant E. coli strain J53 as the recipient and 2 μg/ml meropenem plus 150 μg/ml sodium azide for selecting transconjugants. The presence of bla NDM-4 in transconjugants was confirmed by PCR.

Pulse-field gel electrophoresis (PFGE)

The three isolates were subjected to PFGE following the protocol developed by the Centers for Disease Control and Prevention (Atlanta, GA, USA)9 but with different electrophoresis conditions. Whole-cell DNA from overnight cultures was embedded in 1% InCert agarose plugs, which were digested with 1 mg/L proteinase K and were then restricted with XbaI. PFGE electrophoresis was performed with 1% (w/v) PFGE grade agarose using a CHEF DRII system (Bio-Rad, Hercules, CA, USA) with a 6-V/cm current of 12 h at switch time of 5 to 40 s followed by 8 h at switch time of 3 to 8 s10.

Genome sequencing and analysis

Genomic DNA of the three isolates was prepared using the QIAamp DNA Mini Kit (Qiagen, Hilden, Germany) and was subjected to whole genome sequencing with 150 × coverage using the HiSeq X10 Sequencer (Illumina, San Diego, CA). Reads were trimmed using Trimmomatic11 and were then assembled to contigs using the SPAdes program12 with careful mode turned on. Sequence types were determined using the genomic sequence to query the multi-locus sequence typing database of E. coli (http://enterobase.warwick.ac.uk/species/index/ecoli). Antimicrobial resistance genes were identified from genome sequences using the ABRicate (https://github.com/tseemann/abricate) program. Plasmid replicon types were determined using by the PlasmidFinder tool at http://genomicepidemiology.org/ and the allele types of IncF plasmids were assigned using the IncF replicon typing tool13.

To determine the clonal relatedness of the three isolates, the three genomes were aligned using the Harvest Suite14 with default settings. Single nucleotide polymorphisms (SNPs) on recombination sites were removed by the Gubbins program15.

To facilitate circulating the plasmid sequence, strain WCHEC96200 was also sequenced using the long-read MinION Sequencer (Nanopore, Oxford, UK), which generated 477.161 reads (30.9 GB) and was converted into a single fastq file of 2 GB using poretools16. The assembly of reads were performed using Canu17 with default settings. Circlator18 was then used to locate and circularize complete chromosome and plasmids in the draft assembly. Contigs representing the chromosome and plasmids were subsequently polished using Nanopolish (https://github.com/jts/nanopolish) combined with BWA-MEM19. The polished genome of strain WCH96200 was cured by quality-trimmed Illumina reads using Pilon20 with default settings, to eventually obtain a more accurate assembly.

Nucleotide sequence accession numbers. Draft whole-genome sequences of isolates WCHEC1837, WCHEC96200 and WCHEC99540 have been deposited into GenBank under the accession numbers NGUU00000000, NGUV00000000 and NGUW00000000, respectively. The complete sequences of pNDM4_WCHEC96200 has been deposited into GenBank under the accession number CP022226.

Results and Discussion

The three isolates were all resistant to ampicillin-sulbactam, aztreonam, cefepime, ceftazidime (MIC, >256 μg/ml), ciprofloxacin (MIC, >256 μg/ml), ertapenem, gentamicin, imipenem (MIC, 64 μg/ml), levofloxacin, meropenem (MIC, 64 μg/ml), nitrofurantoin, piperacillin-tazobactam, tobramycin and trimethoprim-sulfamethoxazole but were susceptible to amikacin (MIC, 8 μg/ml for isolate from the first patient or 16 μg/ml for isolates from the other two patients), colistin (MIC, 1 μg/ml) and tigecycline (MIC, < 0.25 μg/ml).

The three isolates had bla NDM only, which was identified as bla NDM-4 by amplifying and sequencing the complete coding sequence of bla NDM using additional primers4. In addition to bla NDM-4, the three isolates had the same 14 intact antimicrobial resistance genes mediating resistance to aminoglycosides (aac(6′)-Ib-cr, aac(3)-IIa, aadA5, strA and strB), β-lactams (bla CTX-M-15 and bla OXA-1), macrolides (mph(A)), phenicol (floR), quinolones (aac(6′)-Ib-cr), tetracycline (tet(A) and tet(B)), sulphonamides (sul1 and sul2) and trimethoprim (dfrA17) in their whole genome sequences (see below).

A total of 4,670,485 to 5,014,495 reads were generated for the three isolates, which were then assembled to 170 to 174 contigs (144 to 147 were ≥1,000 bp in length) with a 50.61 to 50.64% GC content, respectively.

The three isolates belonged to ST405 and phylogenetic group D. ST405 E. coli has a global distribution and is typically associated with extended-spectrum β-lactamases (ESBLs) such as CTX-M-1521, as seen in the three isolates here. Although bla NDM-4 remains uncommon, its association with ST405 E. coli has been previously documented. Six ST405 E. coli carrying bla NDM-4 found in Italy were introduced from India22 and an ST405 E. coli carrying bla NDM-4 was found in a Danish patient who had been previously hospitalized in Vietnam23. Unfortunately, their genome sequences are not available for comparison.

The three isolates had identical PFGE patterns (data not shown). Indeed, there were only 1 or 2 SNPs between the isolates, suggesting very recent acquisition from a common source or recent direct transmission. To investigate this further, the three patients were ordered according to the date on which they provided a positive swab for bla NDM-4-carrying E. coli. All of the three patients were admitted to our hospital in August 2015. The first and second patients were hospitalized in a 50-bed general ICU, while the third was hospitalized in a 30-bed surgical ICU. The hospital stay periods of the three patients were overlapped (Table 1). The first patient was transferred from another local hospital and bla NDM-4-carrying E. coli was detected from the first patient on the same day of his admission to our hospital, suggesting that the isolate was very likely introduced from another hospital. The rectal swab of the second patient that was collected on admission to ICU did not grow bla NDM-4-carrying E. coli, while that collected on the third day of her ICU stay did, suggesting that the bla NDM-4-carrying E. coli was acquired in the ICU. The third patient had not stayed in the local hospital from which the first and second patients were transferred and was admitted to the Liver Surgery Ward in our hospital. Unfortunately, no rectal swabs were collected during his 10-day stay in the surgery ward. The rectal swab collected on admission to the surgical ICU grew bla NDM-4-carrying E. coli. It is possible that the isolate was carried by the patient on admission to our hospital or was acquired during his stay in the Liver Surgery Ward. Nonetheless, the very few SNPs between the isolate of the third patient and those of the other patients suggest that the isolate of the third patient was acquired within our hospital, though the exact route of the acquisition is not clear. No evident epidemiological links could be identified between the third patient and the other two patients. However, as the three patients had overlapped stay in our hospital, there were many possible yet-to-be-identified indirect contacts between the three patients including movement of staff, equipment, or seeding of the wider hospital environment by the strain.

In the three isolates, bla NDM-4 was carried by self-transmissible plasmids. Transconjugants carrying bla NDM-4 were resistant to imipenem and meropenem (MICs, 16 μg/ml). The complete sequence of the plasmid carrying bla NDM-4, designated pNDM4_WCHEC, here in strain WCHEC96200 was therefore also obtained. Sequences of bla NDM-4-carrying plasmids in the other two isolates were circularized by PCR mapping using pNDM4_WCHEC96200 as the template and were actually identical to that of pNDM4_WCHEC96200. pNDM4_WCHEC96200 is 46.5-kb and contains an IncFIA replicon, which belongs to the IncFIA allele 13, and an additional gene encoding a replication initiation protein of the RepB family, which was not assigned an replicon type by the PlasmidFinder. bla NDM-4 was located in a 14.6-kb region bracketed by two copies of IS26 (Fig. 1). In this region, there are Tn5403Δ (a truncated transposon of the Tn3 family), two ISAba125, both of which were interrupted by the insertion of IS26, bla NDM-4 , ble (mediating bleomycin resistance), trpF (encoding the phosphoribosylanthranilate isomerase), dsbC (encoding a tat twin-arginine translocation pathway signal sequence domain protein), cutA1 (encoding a periplasmic divalent cation tolerance protein), groES/groEL (encoding a chaperonin) and ISCR27 (Fig. 1). The two copies of IS26 have the potential to form a composite transposon, which could mobilize the intervening genetic components including bla NDM-4 in this case. The genetic context of bla NDM-4 is highly similar to that of bla NDM-1. It is common that bla NDM-1, ble, trpF, dsbC, cutA1, groES/groEL and ISCR27 are bracketed by two copies of ISAba125, which form a composite transposon termed Tn125 (Fig. 1). It is therefore likely that bla NDM-4 has evolved from bla NDM-1 in such a genetic context.

Figure 1
figure 1

The genetic context of bla NDM-4 on pNDM4_WCHEC96200. The ISAba125-formed composite transposon Tn125 carrying bla NDM-1 is shown for comparison with identical regions being indicated by grey. Δ refers to truncated genes or elements. tnpAΔ and tnpR belong to a transposon of the Tn3 family and sul encodes a sodium-independent anion transporter. repA encodes the replication initiation protein of the IncFIA replicon.

Full size image

bla NDM-4 has been found on plasmids of IncF22,24, IncK25, IncL/M26 or IncX327,28. However, the complete sequence of IncF plasmids carrying bla NDM-4 is not available for comparison. Nonetheless, pNDM4_WCHEC96200 appeared to be a new plasmid that is most closely related (58% coverage and up to 99% identity) to the plasmid tig00001145_pilon (GenBank accession no. CP021881), which contains two replicons, IncFII and IncR, of E. coli AR_0137. pNDM4_WCHEC96200 contains components with significant similarity with several plasmids of different replicon types, suggesting a mosaic composition. A 8.1-kb region containing a gene encoding ATPase is 99% identical to the corresponding region of several plasmids including an IncR plasmid pKPN-041 (GenBank accession no. CP014758). A 7.5-kb region containing a gene encoding sulfate transporter is 99% identical to an IncFII and IncFIA plasmid pCAV1043-51 (GenBank accession no. CP011587).

In conclusion, we identified the in-hospital transmission of an ST405 E. coli strain carrying bla NDM-4, an uncommon variant of bla NDM. The association of bla NDM-4 with a globally distributed clone, ST405 in this case, is worrisome. The self-transmissible IncFIA plasmid carrying bla NDM-4 is a new mosaic plasmid, which could mediate the dissemination of bla NDM-4 into other E. coli strains or even other species of the Enterobacteriaceae. The bla NDM-4 gene was bracketed by two copies of IS26, which have the potential to mobilize bla NDM-4 by hijacking more plasmids as the vehicle to disseminate this gene.