Occurrence and characterization of mcr-1-harbouring Escherichia coli isolated from pigs in Great Britain from 2013 to 2015

Abstract

Objectives: To determine the occurrence of mcr-1-harbouring Escherichia coli in archived pig material originating in Great Britain (GB) from 2013 to 2015 and characterize mcr-1 plasmids.

Methods: Enrichment and selective culture of 387 archived porcine caecal contents and recovery from archive of 1109 E. coli isolates to identify colistin-resistant bacteria by testing for the presence of mcr-1 by PCR and RT–PCR. mcr-1-harbouring E. coli were characterized by WGS and compared with other available mcr-1 WGS.

Results: Using selective isolation following enrichment, the occurrence of mcr-1 E. coli in caeca from healthy pigs at slaughter from unique farms in GB was 0.6% (95% CI 0%–1.5%) in 2015. mcr-1 E. coli were also detected in isolates from two porcine veterinary diagnostic submissions in 2015. All isolates prior to 2015 were negative. WGS analysis of the four mcr-1-positive E. coli indicated no other antimicrobial resistance (AMR) genes were linked to mcr-1-plasmid-bearing contigs, despite all harbouring multiple AMR genes. The sequence similarity between mcr-1-plasmid-bearing contigs identified and those found in GB, Chinese and South African human isolates and Danish, French and Estonian livestock-associated isolates was 90%–99%.

Conclusions: mcr-1-harbouring plasmids were diverse, implying transposable elements are involved in mcr-1 transmission in GB. The low number of mcr-1-positive E. coli isolates identified suggested mcr-1 is currently uncommon in E. coli from pigs within GB. The high sequence similarity between mcr-1 plasmid draft genomes identified in pig E. coli and plasmids found in human and livestock-associated isolates globally requires further investigation to understand the full implications.

Introduction

Colistin is becoming increasingly prescribed to combat MDR bacteria in humans and has been licensed for use in treatment of livestock in the UK since 2004. Occurrence of a plasmid-borne colistin resistance gene, mcr-1, was first reported by Liu et al.¹ in November 2015. Since then, mcr-1-harbouring plasmids have been reported in livestock, food and humans across the globe, with new reports of identification in countries occurring each month.² Despite the recent identification of mcr-1¸ it has subsequently been identified in archived Chinese chicken Escherichia coli isolates from the 1980s; there has also been an increase in reporting of mcr-1 from such isolates from the past 5 years.³ The Animal and Plant Health Agency (APHA) previously reported the characterization of colistin-resistant Salmonella and E. coli isolated from pigs with diarrhoeal disease on a farm in Great Britain (GB). In this study we examined porcine bacterial isolates, frozen pig caecal samples and WGS data of E. coli isolated from pigs, archived from 2013 to 2015, for the presence of mcr-1.

Materials and methods

Archived porcine samples and culture

Details of the samples are given in Table 1 and include their origin, number of pig herds and numbers of E. coli, including colistin-resistant E. coli, examined.

Table 1.

Dates, sources and numbers of samples tested for prevalence of mcr-1 in GB

Date and project type	Sample origin and location	Number of pig herds represented in the sample		Number of archived E. coli isolates	Number of colistin- resistant isolates	Number of mcr-1- positive herds	Herd prevalence (95% CI)
		healthy	diseased/veterinary diagnostic
2013 surveillance	abattoir, UK	115	–	190a	–	0	0% (–)
2014–15 research project	abattoir, England	57	–	556b	63	0g	0% (–)
2015 surveillance	abattoir, GB	313c	–	200d		0e	0% (–)
				NAc	204f	2g	0.6% (0%–1.5%)
2015/16 veterinary diagnostic submissions	veterinary, England and Wales	–	105	163h	4	2g	1.9% (0.0%–4.5%)

NA, not applicable.

^aESBL-producing E. coli recovered from porcine caecal samples from a UK-wide surveillance project²⁰ for which WGS was available.

^bE. coli cultured from randomly selected porcine caecal samples collected from England.

^c387 randomly selected porcine caecal samples originating from 313 different pig herds in GB as part of an EU-wide surveillance programme.

^dE. coli recovered from MacConkey plates without colistin from these porcine caecal samples.

^eIsolates tested by PCR for mcr-1.

^fE. coli recovered on selective MacConkey plates containing 2 mg/L colistin following overnight enrichment.

^gIsolates tested by RT–PCR for mcr-1.

^hE. coli recovered from clinical submissions from diseased animals in England and Wales.

For enrichment of colistin-resistant bacteria from caecal samples, 0.5 g of caecal contents was added to 4.5 mL of buffered peptone water and incubated for 18 h at 37 °C. Plating of enriched contents was on MacConkey agar with/without 2 mg/L colistin.⁴

PCR screening for mcr-1

A sweep of ∼10–20 colistin-resistant lactose-fermenting colonies were picked and PCR was performed on colony lysates as described previously. For RT–PCR, we used the forward primer 5′ → 3′ CCGATCATGCCAATCTACTC and reverse primer 5′ → 3′ CAGGCTTGGTTGCTTGTA in colony lysates under standard RT–PCR conditions.

WGS and analysis of mcr-1-positive isolates

Isolates positive for mcr-1 were sequenced using an Illumina MiSeq 2 × 150 bp run. Sequences were assembled using SPAdes 3.7.0⁵ and annotated using Prokka 1.11.⁶ Contigs with mcr-1 were identified using BLASTN and SeqFinder was used to establish the presence of antimicrobial resistance (AMR) and virulence genes (Table 2). PlasmidFinder was used to determine plasmid compatibility type⁷ and BRIG⁸ was used for plasmid comparisons. WGS data were deposited in the sequence read archive (SRA; Table 2) and used for comparison against all publicly available mcr-1 sequences in the SRA and NCBI (Table S1, available as Supplementary data at JAC Online).

Table 2.

AMR and virulence genes present in four mcr-1-positive isolates from WGS

Isolate and origin	Inc-types present	Estimated size of mcr-1-containing plasmid (kb)a	Similarity to pHNSHP45 (%)	Maximum similarity to publicly available mcr-1 plasmids (%)	Colistin resistance genes	Other AMR genes	Virulence genes
Clinical isolatec O149:H10 2015 veterinary submission	Inc1, IncX4, IncFII(pCoo), IncFIB(AP001918), IncFIC(FII), IncY	32.7	28	99	phoPb, phoQb, pmrAb, etkb, mcr-1	aadA1b, ant3-Ia, dfrA1, folPb, sul2	aec15-19, aec22-27, aec29-32, astA9, cah, eaeH, ecpA-E, ecpR, ehaB, eltA-B, espL4, espR1, faeC-E, faeG-J, fimF-G, hlyA-E, ibeB-C, ItcA, stb1, shf
PO155c –:H56 2015 surveillance study	Incl1, Col8282, pO111, IncX1	91.2	19	90	mcr-1	aac3-IVa, aadA2, ant3-Ia, aph3-Ib, aph4-Ia, aph6-Id, blaTEM-1, cml, dfrA12, inuF, sul2, tetA	aec19, aec32, astA, ecpA-B, ecpD-E, ecpR, espL1, espL4, espX5, fimB-C, fimF-G, fimI, hlyE, iss, iucA, sitA-C
PO169c –:H2 2015 surveillance study	IncX1, IncI2, IncFII(pCoo), IncB/O/K/Z	59.2	90	99	acrRb, phoPb, mcr-1	blaTEM-1, gyrAb, qnrS1, tetA	aec31-32, ecpA-E, ecpR, espR1, fimA-C, fimE-I, hlyE, ibeC, iss, mchF, tia

Bold formatting indicates the Inc type of the mcr-1-harbouring plasmid.

^aEstimated from size of mcr-1-containing contigs present in WGS and plasmid profiling, as described previously.

^bIndicates the presence of non-synonymous SNPs on chromosomal genes that may result in AMR.

^cWGS accession number PRJEB13576.

Results

Laboratory investigations

All E. coli recovered from abattoir samples (2013–15) were tested for the presence of mcr-1 by PCR,¹ RT–PCR or WGS. The 200 E. coli recovered from MacConkey plates from 2015 were negative by PCR, as were the WGS available for the 190 ESBL E. coli from 2013, scanned for the presence of mcr-1 using SeqFinder4 (Table 1). However, two mcr-1-positive isolates were identified following selective culturing and RT–PCR of caecal samples taken from pigs at slaughter in 2015 and originating from two separate anonymized farms (Table 1). Of the 556 E. coli isolates from 2014–15, 63 E. coli showing a colistin-resistant phenotype were tested by RT–PCR, and all 63 were mcr-1-negative. Testing of 163 clinical E. coli identified 4 colistin-resistant isolates, which were mcr-1 positive, originating from two pig farms; 3 were from one farm and 1 (E4), already reported,⁴ was from a second farm. The unadjusted herd prevalence estimates for mcr-1 in E. coli considering these different sample and isolate collections were between 0% and 1.9% (Table 1).

WGS analysis of mcr-1-positive isolates

WGS was performed on five mcr-1-positive isolates (excluding E4), three of which were clones from the same farm. Three isolates, one representative of each farm, were analysed further. Two of these harboured non-synonymous SNPs in other genes associated with colistin resistance, as well as mcr-1 (Table 2). All isolates harboured multiple AMR genes, suggesting MDR, and multiple plasmid Inc types, suggesting multiple plasmids were present, which was confirmed by plasmid profiling (Figure S1). None of the mcr-1-containing plasmids harboured other AMR genes and they varied in size from 32.7 to 91.2 kb. Plasmids of the size predicted from WGS analyses were identified in all isolates by plasmid profiling (Figure S1). A variety of virulence genes were detected, with the E. coli O149:H10 clinical isolate harbouring a heat-stable toxin gene (stb), as well as eaeH and eltA-B, which have been linked with disease in pigs caused by enterotoxigenic E. coli.^9,10 One isolate carried mcr-1 on a pO111-like plasmid, one isolate on an IncI2 plasmid and another on an IncX4 plasmid; the ISApI1 element was only detected in the pO111-like plasmid (Figure S2).

Comparison of mcr-1 plasmid sequences

The three mcr-1 plasmids from E. coli in pigs shared varying degrees of similarity to pHNSHP45,¹ with the highest being 90% in isolate PO169, which had the same plasmid incompatibility group (pPO169; Table 2 and Figure S2). Plasmid pPO169 also shared 99% identity with pS3,⁴ but further comparison indicated pP169 had lost ∼3 kb of its plasmid genome compared with pS3. pPO169 also had 96% identity with a GB human isolate¹¹ and ≥97% identity with isolates from China¹² and Malaysia (Table S1 and Figure S3). The IncX4 plasmid present in E. coli O149:H10 also had ≥96% identity with mcr-1-plasmid-bearing contigs from human isolates from GB,¹¹ the USA, China,¹² Brazil,¹³ Italy¹⁴ and South Africa¹⁵ in addition to Danish¹⁶ and French¹⁷ meat products, Estonian pig slurry and a Swiss environmental isolate (Figure S4).¹⁸ pPO155 showed 99% identity with pE4;⁴ however, pE4 had lost ∼12.7 kb, which probably resulted from recombination events and demonstrates elasticity of the plasmid genome (Figure S5). The second most similar was a meat isolate from Denmark, which showed 90% identity.¹⁶

The area around mcr-1 is conserved in the IncX4 and IncI2 plasmids, but neither harboured the preceding IS element seen in pHNSHP45¹ (Figures S2–S4), and the mcr-1 gene may be stably integrated. Furthermore, the IncI2 mcr-1 plasmid was highly similar to ones found in E. coli, Salmonella enterica and Kluyvera ascorbata, with sizes ranging from ∼57 to 65 kb (Figure S3). However, the pO111-type plasmids, pPO155 and pE4,⁴ both contained the ISApI1 element (Figure S5), so the mcr-1 gene may be unstable and mobilized from these plasmids.

Discussion

The presence of the plasmid-mediated mcr-1 gene was reported in humans in England and Wales and from a pig farm in GB,^4,11 shortly following its first description.¹ This study describes further investigation of a large number of archived samples held at APHA to investigate the occurrence of mcr-1 in E. coli. mcr-1 E. coli was detected on two pig farms in GB through anonymized surveillance of 387 caecal samples collected in 2015 from pigs at slaughter originating from 313 different herds; thus 0.6% of those pig herds sampled were considered to be positive on the basis of this sample set. E. coli positive for mcr-1 were also detected on 2/105 (1.9%) pig farms from which archived E. coli isolates were available from veterinary diagnostic investigations performed in 2015/16. The surveillance of pigs at slaughter was anonymized and therefore it is unknown whether the two herds detected were epidemiologically related to those detected through examination of isolates from veterinary diagnostic submissions. The transferable colistin resistance gene mcr-1 was, however, detected in only a small number of pig herd samples in GB. Although only single or limited numbers of caecal samples were screened from individual herds, detailed studies of two positive pig herds (L. Randall, unpublished results) indicated that a high proportion of animals carried E. coli with mcr-1, albeit as a minor component of the intestinal microbiota, so the estimate of herd prevalence is likely to be realistic, and colistin is not used on the majority of British pig farms.

WGS was used to characterize mcr-1 plasmids from three positive isolates from separate farms. High sequence similarity was found between the three mcr-1-harbouring plasmids in E. coli from GB pigs and other mcr-1 plasmids from within the UK and internationally, including in different bacterial species, and from both livestock and humans.^{11,12,15–17,19} The IncX4 mcr-1 plasmid from this study was highly similar to that found in E. coli, Klebsiella pneumoniae and Salmonella Typhimurium, isolated from humans, meat products and pig slurry between 2012 and 2015 globally, indicating the dissemination of a stable plasmid.^11–18 The variable presence of the IS element preceding the mcr-1 gene possibly indicates that the mcr-1 gene is transmitting in different ways, either via the IS element or through plasmid dissemination. With release of more WGS data from isolates carrying mcr-1, further comparisons between plasmids can be completed, which will facilitate the understanding of mcr-1 plasmid transfer and global dissemination in enteric bacteria from humans, livestock and the food chain.