https://orcid.org/0000-0001-6668-9179
Abstract
Carbapenemase-producing Enterobacteriaceae (CPE) are a public health threat, and have been found in humans, animals and the environment. Carbapenems are not authorized for use in EU or UK companion animals, and the prevalence of carbapenem-resistant Gram-negative bacilli (CRGNB) in this population is unknown.
We investigated CRGNB isolated from animal specimens received by one diagnostic laboratory from 34 UK veterinary practices (September 2015–December 2016). Any Gram-negative isolates from clinical specimens showing reduced susceptibility to fluoroquinolones and/or aminoglycosides and/or cephalosporins were investigated phenotypically and genotypically for carbapenemases. A complete genome assembly (Illumina/Nanopore) was generated for the single isolate identified to investigate the genetic context for carbapenem resistance.
One ST410 Escherichia coli isolate [(CARB35); 1/191, 0.5%], cultured from a wound in a springer spaniel, harboured a known carbapenem resistance gene (blaNDM-5). The gene was located in the chromosome on an integrated 100 kb IncF plasmid, also harbouring other drug resistance genes (mrx, sul1, ant1 and dfrA). The isolate also contained blaCMY-42 and blaTEM-190 on two separate plasmids (IncI1 and IncFII, respectively) that showed homology with other publicly available plasmid sequences from Italy and Myanmar.
Even though the use of carbapenems in companion animals is restricted, the concurrent presence of blaCMY-42 and other antimicrobial resistance genes could lead to co-selection of carbapenemase genes in this population. Further studies investigating the selection and flow of plasmids carrying important resistance genes amongst humans and companion animals are needed.
Introduction
Carbapenemase-producing Enterobacteriaceae (CPE) are a serious public health problem owing to limited therapeutic options for CPE-associated infections.1 Increased carbapenem use, especially for treating infections caused by ESBL producers, is a significant driver of CPE emergence in human medicine.2 In contrast, carbapenems are not authorized for veterinary use3,4 except for the management of MDR Gram-negative infections under the prescribing cascade (https://www.gov.uk/guidance/the-cascade-prescribing-unauthorised-medicines#special-considerations-for-the-responsible-use-of-antibiotics-under-the-cascade).
Furthermore, there is limited carbapenem resistance testing and no UK national surveillance of CPE prevalence in companion animals. Consequently, the occurrence of carbapenemase-producing bacteria in animals may remain undetected. Here, we report surveillance data from a UK Veterinary Diagnostics Laboratory which introduced carbapenem resistance screening for Gram-negative bacteria. We also describe the molecular characterization of an NDM-5 (New Delhi metallo-β-lactamase-5)-producing Escherichia coli isolate from a wound in a dog.
Materials and methods
Bacterial isolates cultured from clinical specimens submitted during September 2015–December 2016 to one UK diagnostic laboratory were included in this study. Clinical specimens were received from 34 veterinary practices across England (n = 29), Wales (n = 4) and Ireland (n = 1), and included swabs, urine, tissues, sterile fluids, bronchoalveolar lavages, faecal samples (from cats and dogs) and bovine milk samples. To increase detection of all carbapenemase-producing isolates (including OXA-48 producers), any Gram-negative bacteria with reduced susceptibility to fluoroquinolones, aminoglycosides and/or cephalosporins cultured were tested using chromID Carba SMART agar bi-plates (bioMérieux, Basingstoke, UK).
Each half of the bi-plate was inoculated with 10 μL of fresh, pure culture (0.5 MacFarland suspension), and incubated aerobically (37 ± 2°C for 22–24 h). Klebsiella pneumoniae NCTC 13368 [SHV-18 (ESBL)] was used as a negative control, and K. pneumoniae NCTC 13438 (KPC-3), NCTC 13440 (VIM-1) and NCTC 13442 (OXA-48) as positive controls. Isolates exhibiting characteristic growth were identified using either API (20E/20NE, bioMérieux UK Ltd) or MALDI-TOF (Laboklin, Germany).
Susceptibility testing of the carbapenem-resistant isolate identified in this study was performed by broth microdilution (TREK Diagnostic System, West Sussex, UK), interpreted according to the EUCAST testing guidelines (version v7.0, http://www.eucast.org/fileadmin/src/media/PDFs/EUCAST_files/Breakpoint_tables/v_7.1_Breakpoint_Tables.pdf).
Bacterial DNA was extracted by heat lysis and centrifugation, and used to screen for: carbapenemases (blaNDM, blaOXA-48-like, blaVIM, blaIMP and blaKPC); ESBLs (blaCTX-M, blaTEM, blaSHVblaGES, blaPER and blaVEB); plasmid-mediated pAmpC-group genes;5,6 and colistin resistance (blaMCR-1 and blaMCR-2; https://www.eurl-ar.eu/CustomerData/Files/Folders/21-protocols/278_mcr-multiplex-pcr-protocol-v2-oct16.pdf) as previously described.
Whole-genome sequencing and analysis
The carbapenem-resistant E. coli isolate was re-cultured from stock; DNA was extracted using the Qiagen Genomic-tip 100/G kit (Qiagen, Hilden, Germany). Aliquots of the same DNA extract were sequenced on both the Illumina HiSeq 4000 and Oxford Nanopore Technologies’ MinION (library preparation kit: SQK-LSK208, flowcell: FLO-MIN106 R9.4, as in Phan et al.7) Sequence data have been deposited in the NCBI database (BioProject: PRJNA473397).
A hybrid, complete genome assembly was constructed from the two sequencing data sets using Unicycler (v4.1; parameters: –no_correct –min_component_size 500 –min_dead_end_size 500 –verbosity 1 –mode bold). The Unicycler assembly was also compared with a hybridSPAdes (v.3.6; default parameters, ‘careful’ option) assembly to verify the genome structure using a different assembly method.
In silico MLST typing was performed using BLASTn against the PubMLST allele databases (https://pubmlst.org/general.shtml). Plasmid typing, insertion sequence typing and resistance gene characterization were carried out using PlasmidFinder, the ISFinder database and an in-house script (ResistType), as previously described.7 The chromosomally integrated plasmid sequence and MDR region harbouring blaNDM-5 were compared with publicly available plasmid sequences in GenBank using BLASTn, with default settings. Data visualizations were created using the GenomeDiagram module in Biopython.
Results
One hundred and ninety-one Gram-negative isolates from dogs (n = 158), cats (n = 27), cattle (n = 4), a rabbit and a guinea pig were subcultured onto chromID CARBA-SMART bi-plates; of these, 28 isolates generated moderate to heavy growth, where Acinetobacter spp. (n = 4) and Pseudomonas spp. (n = 23) grew on one or both halves, whereas E. coli (1/191) grew on the CARB side only.
No blaESBL/pAmpC or carbapenem resistance genes were identified in cultured Pseudomonas spp. and Acinetobacter spp., which were most probably selected due to their intrinsic resistance (decreased permeability and/or expression of efflux pumps) to the agents included in the CARBA-SMART plates. However, the E. coli isolate (isolate CARB35) harboured blaNDM (confirmed on sequencing to be blaNDM-5), blaCMY and blaTEM. The NDM-producing E. coli was cultured (pure growth) from a foreleg wound on the fifth digit of a 7-year-old English springer spaniel. The dog had a history of foot lacerations, dog bite wounds and urinary tract infections, for which he had received multiple courses of amoxicillin/clavulanate, cefovecin, doxycycline and enrofloxacin in the preceding 6 years.
The NDM-producing E. coli was resistant to ampicillin, cefoxitin, aztreonam, cefazolin, cefepime, cefpodoxime (all >16 mg/L), amoxicillin/clavulanate (>32 mg/L), piperacillin/tazobactam, ticarcillin/clavulanate (>64 mg/L), meropenem, imipenem (4 mg/L), ciprofloxacin (2 mg/L), levofloxacin (>8 mg/L) and trimethoprim/sulfamethoxazole (>4 mg/L). The isolate remained susceptible to gentamicin, amikacin (≤1 and ≤4 mg/L, respectively), tigecycline (≤0.25 mg/L) and colistin/polymyxin B (≤0.25 mg/L).
The two assemblers agreed on a complete assembly for the NDM-5-producing E. coli isolate which included a chromosome (∼4.9 Mb; ST410) and five plasmids [∼3 kb, 4 kb, 59 kb (IncI1), 89 kb (IncFII) and 90 kb (IncY)]. blaNDM-5 was chromosomally integrated into the E. coli genome, flanked by multiple IS elements, in an 18.6 kb configuration harbouring other drug resistance genes (mrx, sul1, ant1 and dfrA) (Figure 1). A similar flanking sequence for blaNDM-5 has been observed in only a handful of publicly available but largely unpublished sequences, including submissions from the USA, China and Germany (Figure S1, available as Supplementary data at JAC Online). This MDR region was nested within a 100 kb region encoding plasmid-associated genes, including IncFII, IncFIA and IncFIB replicons, and flanked by two IS150 insertion sequences in the same orientation, most consistent with it representing an integrated plasmid (Figure 1). This 100 kb region had 99% sequence identity over 86% of its length to the reference plasmid sequences CP024860.1 (172 kb; submitted November 2017 by NIH, USA; isolation source unknown) and KP789020.1 (E. coli WCHEC13-8 plasmid pCTXM15 harbouring blaNDM-1 and blaCMY-42 from Chengdu, China; 56 kb; submitted November 2015; human clinical isolate).
Genomic structure of Escherichia coli CARB35, with the blaNDM-5-containing plasmid-like structure integrated into the chromosomal sequence. Insets show the structures flanking the blaNDM-5 gene in greater detail, highlighting the presence of multiple other co-localized drug resistance genes and insertion sequences. This figure appears in colour in the online version of JAC and in black and white in the printed version of JAC.
The IncI1 plasmid (59 kb), carrying blaCMY-42, had a 99% sequence match to E. coli plasmid tig00001287_pilon (GenBank accession: CP021882.1, 68 kb); the IncFII plasmid (89 kb), carrying blaTEM-190, shared 99% sequence identity over 93% of the sequence query with two E. coli plasmids (GenBank accessions: KY463220.1 and AP018147.1).
Genes/gene mutations encoding resistance to spectinomycin/streptomycin (aadA2, chromosome positions 194104–194895), trimethoprim and trimethoprim/sulfamethoxazole (dfrA12, chromosome positions 193199–193696; sul1, chromosome positions 195400–196239; folP, chromosome positions 680079–680927), macrolides/lincosamide/streptogramins (mphA, chromosome positions 202749–203654; and plasmid pCARB35-2, chromosome positions 83836–84741), nitrofurantoin (in nfsA: chromosome positions 3510946–3511599), fluoroquinolones (gyrA, chromosome positions 1676727–1679354; parC, chromosome positions 841664–843922; parE, chromosome positions 831111–833003), and fimE (chromosome positions 4197047–4197643) and usp (chromosome positions 2077529–2077957, 2620969–2621403 and 3475153–3475581) virulence factors were also identified.
Discussion
ESBL/AmpC-producing Enterobacteriaceae have emerged in food-producing and companion animals over the past two decades.8 Although still rare, carbapenemase producers, mainly NDM-1 Acinetobacter spp., VIM-1 E. coli and Salmonella spp., have been reported worldwide in livestock.9 However, there are very few reports of CPE in companion animals. NDM-1-producing E. coli was first described in dogs and cats from the USA in 2013, only 4 years after the first description of NDM-1-producing bacteria in humans.10 OXA-48-producing E. coli and Klebsiella spp. were first reported in dogs from Germany and subsequently also in clinical canine isolates in the USA, including pandemic strains such as E. coli ST648.11,12 However, a low prevalence (0.6%, n = 160) of CPE, consisting of a single VIM-1-producing K. pneumoniae isolate, was found amongst companion animals from Spain.13 More recently, IMP-4-producing Salmonella Typhimurium was isolated from a cat with persistent haemorrhagic diarrhoea in Australia,14 and OXA-23-producing Acinetobacter baumannii associated with urinary tract infection was detected in a cat from Portugal.15
NDM-5 differs from NDM-1 by two amino acid substitutions and has been described in Enterobacteriaceae from both humans and livestock, mainly in Asian countries, including Myanmar.16 NDM-5-producing E. coli was also recently reported in clinical isolates from Italian patients, one of whom had a history of travel to Thailand,17 and also in Spain in a patient who had not travelled abroad.18 In companion animals, NDM-5-producing E. coli ST1284 was isolated from a rectal swab in a dog from Algeria; molecular characterization suggested that blaNDM-5 was likely to be chromosomally located.19
The NDM-5-producing E. coli isolate in our study was ST410, and harboured blaNDM-5 on a plasmid integrated into the chromosome. ST410 is an emerging clone with worldwide distribution, associated with MDR human infections, including bloodstream infections, and demonstrating potential for nosocomial spread.20,blaTEM-190 was present on an IncFII plasmid highly similar to IncFII blaNDM-5 plasmids found in human clinical isolates from Italy (ST405) and Myanmar (ST410).16,18 Our isolate also harboured blaCMY-42 located on an IncI1 plasmid, similarly present in the Italian NDM-5-producing E. coli isolates,17 suggesting a shared plasmid population amongst which blaNDM-5, blaTEM-190 and blaCMY-42 are circulating.
Although blaNDM-1 is common amongst human carbapenem-resistant isolates in the UK, blaNDM-5 has been reported on only a small number of occasions: once, in 2011, also on an IncFII plasmid in an ST648 E. coli recovered from a patient recently hospitalized in India21 and, in 2014, in four ST410 isolates, for which there are limited additional metadata, and in which the genetic location of blaNDM-5 could not be determined.20 Given the similarity of its genetic background to previously described human isolates, the low prevalence of CPE in animals and the increasing evidence of environmental contamination with CPE by human hospital effluents,8 it is possible that the NDM-5-carrying E. coli isolated in this study might be of human origin. Our study was limited with respect to the sampling frame and lack of available epidemiological data, but suggests that further detailed studies on the selection and flow of important resistance genes, including carbapenemases, amongst humans and animals are needed.
Although the use of carbapenems in companion animals is uncommon, the concurrent presence of blaCMY-42 and blaTEM-190 on common plasmids could lead to rapid co-selection of blaNDM in this population. In addition, the detection of a carbapenemase-producing E. coli ST410, which was recently described as a high-risk MDR clone with increased potential for interspecies transmission,22 in companion animals is concerning. Hence, improved antimicrobial stewardship as well as introducing routine detection of carbapenem resistance in animal isolates is warranted to reduce the risk of zoonotic transmission and will contribute to concerted ‘One Health’ efforts in containing the spread of resistance to last resort antimicrobials.
Funding
This work was supported by the National Institute for Health Research Health Protection Research Unit (NIHR HPRU) in Healthcare Associated Infections and Antimicrobial Resistance at Oxford University in partnership with Public Health England (PHE) (grant HPRU-2012-10041), NIHR Oxford Biomedical Research Centre and Antimicrobial Research Cross Council Initiative supported by the seven research councils (NE/N019989/1). N. S. is funded by a PHE/University of Oxford Clinical Lectureship.
Transparency declarations
None to declare.
Disclaimer
This report presents independent research funded by the National Institute for Health Research and research councils. The views expressed in this publication are those of the authors and not necessarily those of the NHS, the National Institute for Health Research, the Department of Health, the research councils or Public Health England
References
EFSA BIOHAZ Panel (EFSA Panel on Biological Hazards), 2013.
