Article Text

Download PDFPDF

The use of whole-genome sequencing for molecular epidemiology and antimicrobial surveillance: identifying the role of IncX3 plasmids and the spread of blaNDM-4-like genes in the Enterobacteriaceae
  1. Björn A Espedido1,2,3,
  2. Borce Dimitrijovski1,2,4,
  3. Sebastiaan J van Hal2,4,
  4. Slade O Jensen1,2,3
  1. 1Molecular Medicine Research Group, School of Medicine, University of Western Sydney, Sydney, New South Wales, Australia
  2. 2Antibiotic Resistance & Mobile Elements Group, Ingham Institute for Applied Medical Research, Sydney, New South Wales, Australia
  3. 3South Western Sydney Clinical School, University of New South Wales, Sydney, New South Wales, Australia
  4. 4Department of Microbiology and Infectious Diseases, Royal Prince Alfred Hospital, Sydney, New South Wales, Australia
  1. Correspondence to Ass Prof. Slade O Jensen, Molecular Medicine Research Group, School of Medicine, University of Western Sydney, Penrith, Sydney, NSW 2751, Australia; s.jensen{at}


Aims To characterise the resistome of a multi-drug resistant Klebsiella pneumoniae (Kp0003) isolated from an Australian traveller who was repatriated to a Sydney Metropolitan Hospital from Myanmar with possible prosthetic aortic valve infective endocarditis.

Methods Kp0003 was recovered from a blood culture of the patient and whole genome sequencing was performed. Read mapping and de novo assembly of reads facilitated in silico multi-locus sequence and plasmid replicon typing as well as the characterisation of antibiotic resistance genes and their genetic context. Conjugation experiments were also performed to assess the plasmid (and resistance gene) transferability and the effect on the antibiotic resistance phenotype.

Results Importantly, and of particular concern, the carbapenem-hydrolysing β-lactamase gene blaNDM-4 was identified on a conjugative IncX3 plasmid (pJEG027). In this respect, the blaNDM-4 genetic context is similar (at least to some extent) to what has previously been identified for blaNDM-1 and blaNDM-4-like variants.

Conclusions This study highlights the potential role that IncX3 plasmids have played in the emergence and dissemination of blaNDM-4-like variants worldwide and emphasises the importance of resistance gene surveillance.

View Full Text

Statistics from


In Australia, carbapenem resistance in the Enterobacteriaceae is rare, as <1% of clinical isolates are not susceptible to meropenem.1 Despite this, there has been a small increase in the number of carbapenem-resistant isolates detected (via surveillance programmes) over the past 5 years. While blaIMP-4 is the dominant metallo-β-lactamase gene,1 blaNDM allelic variants that have previously been identified (and sequenced) in Australia include blaNDM-1 from Enterobacter cloacae,2 Escherichia coli,3 Klebsiella pneumoniae4 and Proteus rettgeri5 and blaNDM-3 from E coli.2 The latter three species represent cases in which there was confirmed intercontinental transfer via patients who had travelled from either Bangladesh3 or India,2 ,3 countries where blaNDM-positive Enterobacteriaceae are widespread. At present, 16 blaNDM allelic variants have been identified (, many of which differ by only a single nucleotide (figure 1A) and are associated with a wide range of plasmids representing different incompatibility groups (eg, IncA/C, IncF-type, IncHII, IncII, IncL/M, IncN and IncX3).6 ,7

Figure 1

Comparison of blaNDM allelic variants and their genetic contexts found on different IncX3 plasmids. The relationship between the known blaNDM allelic variants is shown in an unrooted radial phylogenetic tree (A); nucleotide substitutions with respect to blaNDM-1 and blaNDM-4 are shown (the scale bar shows the expected number of substitutions per nucleotide). Those blaNDM allelic variants confirmed (solid orange line) and predicted (dashed orange line) to be associated with IncX3 plasmids are indicated. Comparisons between these IncX3 plasmid-associated blaNDM­ allelic variants are shown (B); areas of ≥99% sequence identity between plasmids are indicated by the light blue areas. The following selected genes are annotated and represented by coloured arrows: plasmid partitioning and replication genes, red; transposon-related genes, orange; plasmid accessory, maintenance, mobilisation and conjugation genes, yellow; glycopeptide resistance genes, green; β-lactam resistance genes, blue. Dashed arrows represent more than one gene or open reading frame. Insertion sequences (ISs) are represented by orange pentagons; the direction of the IS with respect to the transposase gene is indicated by the point of the pentagon. Inverted repeats associated with ISs and transposons are indicated by vertical orange lines.

Not surprisingly, blaNDM-positive Enterobacteriaceae are often resistant to multiple antimicrobial agents (eg, aminoglycosides, extended-spectrum β-lactams, flouroquinolones),6 as they also contain additional resistance determinants in association with mobile genetic elements. Traditionally, PCRs and Sanger sequencing have been used to gather molecular surveillance data on clinical isolates (ie, sequence types (STs)) and prevalent antibiotic resistance genes. Potential drawbacks in using these methods include the time taken to obtain sequence data and the limitation of detection (particularly in the case of resistance genes) to known targets. However, with the increasing affordability of whole-genome sequencing (WGS), it is likely to supplant traditional PCR-based methods. Moreover, WGS is able to provide a complete snapshot of the genetic data for a particular isolate rather than a few target gene sequences. In this study, we show the use of WGS in characterising the ST, resistome and genetic context (at least in part) of the first known NDM-4-producing carbapenem-resistant K pneumoniae isolate (Kp0003; table 1) identified in Australia.

Table 1

Antibiotic minimum inhibitory concentrations (MICs) and PCR results for Kp0003 and Ec0004

Materials and methods

Bacterial strains, conjugation experiments and antibiotic resistance profiles

Bacterial strains used in this study are listed in table 1. Filter-based conjugation experiments were performed as previously described using a rifampicin-resistant mutant E coli DH5α strain (Ec0002).8 E coli transconjugants were confirmed based on colony morphology and positive indole (bioMérieux, Marcy L’Étoile, France) reactions. Antibiotic resistance profiles were determined on a VITEK 2 AST-N246 card using the global and natural resistance interpretive criteria (bioMérieux; see table 1).

DNA manipulations, sequencing and DNA analysis

Bacterial genomic DNA was extracted using the ISOLATE Genomic DNA Mini Kit (Bioline, London, UK) and sent to The Ramaciotti Centre (University of New South Wales, Sydney, Australia) for sequencing on a MiSeq Desktop Sequencer (Illumina, San Diego, California, USA). Analysis of genomic data was performed using CLC Genomics Workbench 7.0.4 (CLC bio; Aarhus, Denmark). Gaps in plasmid read mappings were closed via PCR amplification using BioTaq (Bioline) and capillary sequencing (Macrogen, Seoul, Korea).

Results and discussion

While travelling in Mynamar, a country which shares borders with Bangladesh and India, an Australian traveller become unwell and presented to a local hospital where a susceptible K pneumoniae was initially recovered from blood cultures; therapy with a broad-spectrum cephalosporin was initiated. However, several days later the patient developed a hospital-acquired pneumonia and a multidrug-resistant K pneumoniae was recovered from both the sputum and the blood culture; therapy was converted to amikacin and colistin. Further investigations, including echocardiography, suggested possible vegetation. Due to a complex cardiac history, and the possibility of a prosthetic aortic valve infective endocarditis, the patient was repatriated back to Australia approximately 14 days after initial presentation and was isolated on arrival.

Subsequently, a multidrug-resistant K pneumoniae (Kp0003) was recovered from a rectal screening swab. The antibiotic susceptibility profile of Kp0003 (table 1) revealed a similar antibiogram to the initial multidrug-resistant blood culture isolate (all attempts to obtain the initial isolate from Myanmar proved unsuccessful). Kp0003 was resistant to all antibiotics tested, including meropenem, with the exception of amikacin (table 1). Although additional investigations, including a transoesophageal echocardiogram and review of the previous results, could not confirm the presence of endocarditis, the patient was treated for a complicated bacteraemia with 4 weeks of parenteral antibiotics and made a full recovery.

To genetically characterise and explain its multidrug-resistant phenotype, WGS was performed on Kp0003 (for sequencing and assembly statistics, see supplementary table 1). Sequence reads (Sequencing Read Archive accession no. SRR1560660) were mapped to the genome of the non-multidrug-resistant K pneumoniae strain NTUH-K2044 (GenBank accession no. AP006725), which facilitated in silico multilocus sequence typing (MLST) and variant analysis, as previously described.8 In terms of locally reported blaNDM-1-positive K pneumoniae isolates, they represented STs 11, 147 and 10684; based on in silico MLST, Kp0003 was an ST11 isolate. Of note, previous K pneumoniae ST11 outbreaks, although of blaNDM-1-positive isolates, have been documented in Greece11 and the United Arab Emirates.12

A de novo assembly of reads that did not map to NTUH-K2044 was used to query an in-house resistance gene database (RGD; containing representative allele sequences of 74 commonly observed antibiotic resistance genes)8 and to identify associated mobile elements by performing BLASTn analyses of contigs using the National Center for Biotechnology Information non-redundant nucleotide database. In summary, contigs comprising the following gene cassette arrays were identified in association with different class 1 integrons (data not shown): arr3-dfrA27; dfrA1-orfC; aac(6’)-Ib-cr-blaOXA-30-catB3. Furthermore, other contigs were identified containing resistance determinants associated with different transposon-related genes were also identified (in conjunction with) and included aacC2 and qnrS1, as well as a multiresistance region consisting of tnpATn2-blaCTX-M-15-ISEcp1-tnpARTn2-blaTEM-1b. Collectively, these genes confer resistance to aminoglycosides (except amikacin), β-lactams (except carbapenems), chloramphenicol, fluoroquinolones, rifampicin, sulfamethoxazole and trimethoprim.

Importantly, querying of our local RGD also revealed the presence of the carbapenem resistance gene blaNDM-4, which has previously only been identified via intercontinental transfer events in association with E coli ST405 isolates; these events involved transfer (via different patients) from India to Italy,13 Vietnam to Denmark14 and Cameroon to France.15 While the plasmids in these E coli isolates were not fully characterised, it was inferred that blaNDM-4 was located on IncF-type plasmids.13 ,15 For Kp0003, in silico plasmid replicon typing9 ,10 identified two plasmids: an IncFII plasmid designated pJEG028 (similar to pKF3–94 GenBank accession no. FJ876826 and carrying the blaCTX-M-15/blaTEM-1b transposon; data not shown) and an IncX3 plasmid designated pJEG027 (GenBank accession no. KM400601; figure 1B) on which the blaNDM-4 gene was determined to be located.

The plasmid backbone of pJEG027 is closely related to that of the blaNDM-5-containing plasmid pNDM_MGR194 (GenBank accession no. KF220657; carried by a K pneumoniae isolate from India) and the blaNDM-1-containing plasmid pNDM-HN380 (carried by a K pneumoniae isolate from Hong Kong; figure 1B),7 both of which are conjugative. Filter-based conjugation experiments demonstrated the transfer of pJEG027, carrying blaNDM-4, between Kp0003 and the plasmid-naïve rifampicin-resistant E coli strain Ec0002. The presence of only pJEG027 and blaNDM-4 in a single transconjugant (Ec0004) was confirmed by PCR (table 1). In comparison with the parent strain (Ec0002), Ec0004 displayed resistance to all β-lactams with the exception of cefepime (tested as intermediate; minimum inhibitory concentration (MIC) was close to the breakpoint of >4 mg/L), while remaining sensitive to other antibiotics tested (table 1).

With respect to blaNDM-1, blaNDM-4 differs by a single nucleotide (A460C; figure 1A), which results in an amino acid substitution (Met154Leu) and increased hydrolytic activity toward carbapenems and several cephalosporins.15 Although high-level carbapenem resistance often requires both the production of a carbapenem-hydrolysing β-lactamase and the mutations in porin genes (which encode the outer-membrane channels that β-lactams use to enter the cell), the increased activity of the A460C mutation is exemplified by the high-level meropenem MICs seen in Kp0003 and its blaNDM-4-positive transconjugant Ec0004 (table 1), both of which have no porin mutations (data not shown). This nucleotide change is shared between other blaNDM-4-like variants (blaNDM-5, blaNDM-7, blaNDM-8, blaNDM-12 and blaNDM-13), which themselves differ by only a single nucleotide (in comparison with blaNDM-4; figure 1A).

Although blaNDM-12 has recently been reported on an IncF plasmid,16 its genetic context, based on available information (GenBank accession no. AB926431), is different to the other blaNDM-4-like variants. In relation to IncX3 plasmids that contain blaNDM alleles, sequences are available for pNDM-HN380 (contains blaNDM-1),7 pJEG027 (contains blaNDM-4; this study), and pNDM_MGR194 (contains blaNDM-5; GenBank accession no. KF220657); only partial sequences are available for the genetic context of blaNDM-7 (pEC1; GenBank accession no. KC567147) and blaNDM-8.17 The similarity in genetic context between these blaNDM alleles suggests that IncX3 plasmids may be playing a significant role in the evolution and spread of blaNDM-4-like variants worldwide. It is possible that pJEG027 could have arisen from a pNDM-HN380-like plasmid ancestor as a result of different IS5 insertion events, an IS26-mediated flanking deletion of cutA1-groL, and acquisition of the A460C mutation in blaNDM-1 (figure 1B). Thus, the spread of pJEG027-like plasmids and acquisition of additional mutations within blaNDM-4 could have led to the emergence of the blaNDM-4-like variants.

Furthermore, plasmid-borne blaNDM-7 has been identified in E coli ST167 and ST599 isolates from patients who travelled from Myanmar to France18 and India to Germany,19 respectively. While IncX typing was not performed on the ST167 isolate,18 blaNDM-7 was reported to be present on an IncX3 plasmid in the ST599 isolate; this plasmid conferred β-lactam resistance (only) in transconjugants generated as part of that study.19 It is important to note that a commonly used plasmid-typing scheme10 does not identify all subtypes of IncX plasmids9 and thus the presence of blaNDM allelic variants on IncX plasmids could be under-reported, especially in studies where untypeable plasmids were detected. A recent study of 225 pathogenic E coli isolates (from animal and human origins) found that 24 isolates (10.7%) harboured IncX plasmids, of which only one isolate harboured an IncX3 plasmid.20 Although there is no dominant plasmid group responsible for the dissemination of blaNDM alleles in general, IncX3 plasmids have been responsible for the spread of blaNDM-17 ,12 and, despite their low prevalence, it appears that IncX3 plasmids may be playing an important role in the evolution, emergence and spread of blaNDM-4-like variants worldwide.

Our data support that greater genomics-based antimicrobial surveillance and molecular epidemiological studies are useful in examining the prevalence of multidrug-resistant isolates particularly of those with novel or uncommon resistance phenotypes. Furthermore, determining the genetic context of clinically relevant resistance determinants is vitally important, as such knowledge informs isolation and infection control policies and facilitates understanding of how antibiotic resistance is disseminated.

Take home messages

  • WGS represents a powerful clinical microbiological tool by providing a complete snapshot of an isolate's genetic information.

  • Compared to traditional PCR-based studies, which target specific genes, WGS facilitates: epidemiological typing, resistance gene identification (and resistance profiling), novel gene discovery and determination of genetic context.

  • Local WGS surveillance and epidemiological data can inform clinicians on isolation and infection control policies.


View Abstract

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.


  • Handling editor Cheok Soon Lee

  • Contributors BAE, SJvH and SOJ planned the experiments and contributed to writing of the manuscript. BAE and BD conducted the experiments.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.