Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Genomic evolution of antimicrobial resistance in Escherichia coli


The emergence of antimicrobial resistance (AMR) is one of the biggest health threats globally. In addition, the use of antimicrobial drugs in humans and livestock is considered an important driver of antimicrobial resistance. The commensal microbiota, and especially the intestinal microbiota, has been shown to have an important role in the emergence of AMR. Mobile genetic elements (MGEs) also play a central role in facilitating the acquisition and spread of AMR genes. We isolated Escherichia coli (n = 627) from fecal samples in respectively 25 poultry, 28 swine, and 15 veal calf herds from 6 European countries to investigate the phylogeny of E. coli at country, animal host and farm levels. Furthermore, we examine the evolution of AMR in E. coli genomes including an association with virulence genes, plasmids and MGEs. We compared the abundance metrics retrieved from metagenomic sequencing and whole genome sequenced of E. coli isolates from the same fecal samples and farms. The E. coli isolates in this study indicated no clonality or clustering based on country of origin and genetic markers; AMR, and MGEs. Nonetheless, mobile genetic elements play a role in the acquisition of AMR and virulence genes. Additionally, an abundance of AMR was agreeable between metagenomic and whole genome sequencing analysis for several AMR classes in poultry fecal samples suggesting that metagenomics could be used as an indicator for surveillance of AMR in E. coli isolates and vice versa.


Escherichia coli, is an inhabitant of the gastrointestinal tract and feces of warm-blooded animals and reptiles1,2. Escherichia coli has the interesting characteristic of being both a widespread gut commensal bacteria in vertebrates and a versatile pathogen, resulting in killing more than 2 million humans per year through both intra- and extraintestinal diseases3,4. The genetic structure of commensal E. coli is shaped by multiple host and environmental factors, and genetic determinants involved in the virulence of the bacteria reflect adaptation to commensal habitats5. Selective pressure in the habitats of commensal strains may coincidentally promote the emergence of virulence factors and antimicrobial resistance (AMR), rendering commensal E. coli strain reservoirs of virulence and AMR determinants5. Emergence of antimicrobial resistance is one of the biggest global threat to public health6. In addition to the use of antimicrobials in humans, the use of antimicrobials in livestock is considered an important driver of AMR, subsequently compromising human health7.

Besides AMR in zoonotic pathogens, AMR in commensal bacteria is perturbing due to its ability to spread horizontally to pathogens8. The commensal microbiota, and especially the intestinal microbiota, has shown to play an important role in the emergence of antimicrobial resistance9. The extensive use of antimicrobials in both human and veterinary medicine, trigger the selection of antimicrobial resistance in the commensal microorganisms5. Antimicrobial resistance genes are frequently mobilized and disseminated through the bacterial population through the interplay of various mobile genetic elements (MGEs) of different types10.

In this study, we isolated E. coli from fecal samples in poultry flocks, swine, and veal calve herds from 6 European countries to investigate phylogenetic background and clustering of E. coli at country, animal source and farm levels. Furthermore, this study provided an insight into the evolution of AMR in E. coli genomes including an association with virulence genes, plasmid and mobile genetic elements (MGEs). In addition, the correlation of AMR abundance between metagenomics and WGS analyses was determined.


Escherichia coli phylogeny and genetic diversity

The phylogenetic SNP tree exhibited diverse strains of E. coli with 17,213 total SNPs (Fig. 1). There was no clear clustering either by country of origin or animal host. In addition, within a small cluster from the same animal host, it contained strains from more than one country. Most of E. coli isolates were non-pathogenic strains. There were 39 isolates (6%) identified as EAEC/Enteroaggregative E. coli (n = 15, 2%), ETEC/ Enterotoxigenic E. coli (n = 13, 2%) and STEC/Shiga toxin-producing E. coli (n = 11, 2%). Furthermore, the SNP tree did not show any country and animal reservoir clusters of AMR genes nor chromosomal point mutation conferring to antimicrobial resistance (Figs. S2S4). Looking at SNP tree in a country level (Figs. S5S10), E. coli isolates within the same country could be related to strains from the same or different farms. The 627 E. coli isolates comprised of 182 different sequence types (ST) types with 24 isolates being of unknown ST type, indicating high diversity in the E. coli isolates. The major identified ST type was ST10 (n = 143, 23%) and E. coli isolates seems to be clustered based on ST type rather than geographical location, animal reservoir, E. coli pathotype or AMR (Fig. 1). Similar tree topology of the E. coli isolates was observed in core genome MLST tree (Fig. S11).

Figure 1

SNP tree of 627 E. coli isolates. The tree was constructed from 17,213 SNPs. The legends from inner circle to outer circle are country, host, ST type and pathotype. There are 182 different ST types with 24 isolates characterized as unknown ST type. Only a number of ST type can be showed in figure legend. EAEC Enteroaggregative E. coli, ETEC Enterotoxigenic E. coli and STEC Shiga toxin-producing E. coli.

Average nucleotide identity (ANI) from comparative genomic of the E. coli isolates was summarised on farm and animal host levels (Fig. 2). Mean of overall ANI was 98.3% (red horizontal line). Five out of 10 E. coli isolates from Danish and Spanish farms and five of 15 isolates from Netherlands had ANI lower than average (98.3%). Meanwhile Germany, France and Poland had three out of 17, one out of five and one out of 10 farms with ANI lower than average respectively. Most of the lower average ANI farms were poultry and swine. Figure 2B showed that E. coli isolates from poultry had the lowest mean ANI and broadest range of ANI compared to the ones from swine and veal, reflecting the high strain diversity.

Figure 2

Average nucleotide identity in mean percent identity of E. coli isolates from the same farm (A) and from the same animal host (B). DK Denmark, FR France, DE Germany, NL The Netherlands, PL Poland and ES Spain. Vertical line represents points within 1st/3rd + 1.5 IQR. A red horizontal line indicates mean of overall nucleotide identity (98.34%).

Antimicrobial resistance genes

Prevalence and distribution of resistant E. coli strains per country and host were illustrated in Figs. 3 and 4. Denmark had the lowest abundance of AMR strains, meanwhile Spain had the highest abundance. All countries had a similar distribution of AMR with the exception of some genes that were more abundant in Poland, Spain and France. Poland had higher proportion of fluoroquinolone resistant E. coli (Figs. 3, 4) which were caused by qnrB19 and qnrS1 genes (Fig. S12). Spain and France had higher proportion of sulphonamide resistant E. coli (Figs. 3, 4) associated with sul1, sul2 and sul3 genes (Fig. S12 and Table S1 sub-sheet “Percentage of resistant E. coli”). Regarding the animal reservoir, isolates from veal calves had higher proportion of AMR resistant E. coli strains (Fig. 3), while the proportion of fluoroquinolone resistant E. coli was higher in poultry and swine. Additionally, the proportion of beta-lactam resistant E. coli was higher in poultry which was caused mainly by blaTEM-1B (Fig. S12 and Table S1 sub-sheet “Percentage of resistant E. coli”).

Figure 3

Bar plot showing percentage of resistant E. coli isolates in AMR classes (top) and genes (bottom) stratified by country (left) and host (right).

Figure 4

Distribution of resistance and susceptible E. coli isolates in different AMR class and host. Y-axis is number of isolates. Three columns are number of isolates from poultry, swine and veal. In each column, first bar is number of susceptible isolates and second bar is number of resistant isolates.

The prevalence of AMR was ranked based on percentage of E. coli carrying AMR genes in country of origin and animal host. The most prevalent five AMR genes were revealed (Table S1 sub-sheet “Percentage of resistant E. coli”). They were tet(A), blaTEM-1B, sul2, aadA1 and aph(6)-Id coding for resistance against tetracyclines, aminopenicillins, sulphonamides and aminoglycosides. These five AMR genes were also the top five AMR genes in the strains from France and Poland. Three to Four of them were among the top five AMR genes in other countries in this study. In addition, these five AMR genes were most prevalent AMR genes in the E. coli strains from swine and veal calves. Four of them were among the top five AMR genes in poultry (Table S1 sub-sheet “Percentage of resistant E. coli”).

The most prevalent five AMR genes were analysed for conserved regions on 1 kb upstream and 1 kb downstream of the gene (Fig. S13). Although, upstream and downstream regions from almost all five AMR genes showed variable regions and not any large conserved regions, the upstream region from aph(6)-Id exhibited ~ 600 bp conserved region among the E. coli isolates (Fig. S13, grey colour) suggesting that among the top 5 AMR genes, only aph(6)-Id gene seems to be specifically conserved in E. coli isolates.

Mobile genetic element linked with plasmid, AMR and virulence genes

There was little difference in both total number of MGEs and in the average number of MGEs carried by an E. coli between the different animal hosts (Fig. 5A). However, the composition of MGEs carried by E. coli differed between the hosts. More unity transposon (Tn) were predicted in isolates from poultry and swine than in isolates from veal (Fig. S14). Some insertion sequences (IS) were very prevalent, for instance ISEc1 and IS609 were found in 87.9% and 79.9% of all isolates (Table S2). The predicted IS existed in multiple copies with varying degree of alignment completeness and level of truncation compared to the reference sequence. For all of the predicted IS a majority of the identified copies were highly truncated but had an intact inverted repeat region which could help form composite transposons (ComTn) (Tables S2 and S3). The most frequent transposon was Tn2 (Fig. 5B) which was frequently found on IncX1, Incl1-I(gamma) and IncFIC(FII) plasmids (Fig. 5B). Tn2 is known that E. coil unit-transposon carries blaTEM-1 gene11.

Figure 5

(A) Distribution of mobile genetic elements (MGEs) in E. coli isolates per animal host and country of origin (top). Distribution of MGEs per E. coli isolate in different animal hosts (bottom). Cn XX, IS insertion sequences, MITE miniature inverted-repeat transposable elements, and Tn transposons. (B) Distribution of transposons and their plasmids. (C) Number of MGEs associated with mobile elements per antimicrobial class and animal source. There are three levels of associations, carried on MGEs, if MGE is located within 31 kbp from an AMR and unknown association.

Escherichia coli isolates were clustered on the predicted MGE profile using Jaccard distance and complete linkage for all strains (Fig. S15) and isolates per country (Figs. S16S21). The trees showed the distribution of MGEs within the E. coli population and any relationship with country, animal source, farm ID or ST type. Escherichia coli isolates carried a highly diverse set of MGEs that were seemingly independent of host or ST type. These E. coli isolates carried on average ~ 4 plasmids but a minority of the predicted MGEs were located on them (Fig. S22).

AMR gene was classified based on their location in respect to MGEs. A gene considered to be associated if it was located closer than 31 kbp to an MGE, mobilized if it was an accessory gene within the MGE, and having an unknow association if neither criterion were met. A gene associated with an MGE could be mobilized through formation of ComTn. Beta- lactamase type AMR genes were more frequently mobilized by MGEs more than genes from other AMR classes. In addition, more beta-lactamases were carried on transposons in poultry and swine than in veal calves (Fig. 5C). These carried beta-lactamases were mainly blaTEM-1 carried on transposon Tn2. In addition, three tet(A) genes were mobilized on Tn1721 (Table S4). Some putative ComTn were identified where the most abundant mobilized a variety of AMR genes (Table S5).

The distance between the gene and the MGE its associated with were shown in Fig. 6. MGEs such as IS26, ISVsa3 and Tn2 were associated with several AMR genes where IS26 was associated with the most MGEs (Fig. 6). MGEs tended to be located at a fixed distance from a gene. This could be an indication that they are transferred as a unit. One indication that the AMR genes was mobilized by the MGEs are their tendency to be located arrays of repeated genes with variable length and content. These arrays have MGE located up to 4151 bp downstream of the first AMR (Table S6). However due to the low number of observations these results are rather speculative.

Figure 6

Distance from AMR gene (top) and virulence gene (bottom) to closest MGE.

Abundance of AMR from metagenomic and whole genome sequencing

Fecal samples used for E. coli isolation were sequenced using shotgun metagenomics8 and AMR genes were characterized in the metagenomics data using the ResFinder database. Abundance of AMR genes from metagenomic fecal samples and E. coli whole genome sequencing from the same farm was illustrated as scatter plot in Fig. 7. The AMR abundance in the metagenomic samples were calculated as fragments per kilobase reference per million bacterial fragments (FPKM), while the AMR abundance in E. coli WGS was determined as percentage of E. coli isolates carrying AMR genes. Correlation and p value from the regression analysis in Fig. 7 was shown in Table 1. There was significant correlation between AMR abundance of metagenomics and AMR prevalence in E. coli WGS from poultry in trimethoprim, aminoglycoside, beta-lactam, phenicol and quinolone (5 out of 8 AMR classes present in E. coli WGS from poultry). Correlation between AMR abundance in colistin was found only in swine samples. Significant correlation was not observed in veal samples. This suggested that AMR abundance from poultry feces could potentially represent AMR abundance in single isolation of E. coli from poultry.

Figure 7

Scatter plot between abundance of AMR genes from metagenomic fecal samples and percentage of E. coli isolates carrying AMR genes from the same farm. X-axis is percentage of E. coli isolates carrying resistance genes. Y-axis represents abundance of resistance genes in FPKM (Fragments Per Kilo base per Million fragments) from metagenomic fecal samples.

Table 1 Correlation and p value from regression analysis between abundance of AMR genes in metagenomic fecal samples and percentage of E. coli isolates carrying AMR genes.


This study is a large-scale multinational study of E. coli isolates from several animal sources within the same year. The 627 E. coli exhibited a considerable degree of genetic diversity, differing by a total of 17,213 SNPs and representing 182 different ST types. A recent study on genetic diversification from a large collection of non-clinical E. coli isolates also revealed 442 distinct ST types of which 61% were from a single strain12. Escherichia coli strains isolated from humans, domesticated and wild animals represent a primary habitat of E. coli5. Most of the E. coli from these domesticated animals were not harmful pathogenic strains and 6% were identified as EAEC, ETEC or STEC. The high genetic diversity in the E. coli isolates resulted in phylogeny which showed no clustering based on country of origin, animal reservoir and antimicrobial resistance. Moreover, the strains within the same country could be more related to strains from different farms. The results suggested that the E. coli strains could be randomly spread between the farms in each country. A recent study on E. coli isolates collected from pig surveillance between 2013 and 2017 in the United Kingdom also found no correlation between phylogeny and year of isolation13. This suggests that E. coli isolates could be as well randomly spread in countries and through different years. An alternate explanation would be that the E. coli we have sequenced do not fully represent the extent of the diversity within each animal, farm, and country.

Average nucleotide identity (ANI) is a measure of genomic divergence between genomes. The ANI of E. coli isolates from different countries did not exhibit any clear differences among countries. Nonetheless, E. coli isolates from poultry showed lowest and broadest range of ANI suggesting that the poultry E. coli had higher genetic diversity and were more heterogenous than isolates from swine and veal calves. Previous study concluded that E. coli genotypes and serotypes in pullets and layers are heterogeneous and do not maintain a single clonality within the same bird14. In addition, E. coli do not show definite clonal population structure based on geographical region and age of the host14. In animals, the main environmental force shaping the genetic structure of E. coli gut population is domestication status of the host15 and socioeconomic factors, such as diet16 and hygiene, are presumably the main factors accounting for phylogenetic group distribution rather than geographical or host genetic conditions5.

According to phylogenetic result and distribution of AMR genes in E. coli from different countries, we suggested that AMR seems to spread randomly in all the surveyed countries. Exceptions were higher fluoroquinolone resistance in Polish strains and higher sulphonamide resistant E. coli in Spain and France. Veal calves had a higher proportion of resistant E. coli strains. A previous study on commensal E. coli from a systematic resistance monitoring of primary food production, slaughterhouses and retail, they found that dairy cows had the lowest number of resistant E. coli, whereas veal and pork had higher frequency of resistance strains ranging between 43 and 73%17. In addition, our results showed that fluoroquinolone and beta-lactam resistant E. coli was higher in poultry. Similar results were found in a study in Czech Republic on E. coli from poultry meat, poultry farms and market-weight turkeys. They found that resistance of the poultry isolates to quinolones ranged from 53 to 73%18. Another study on E. coli from broilers, workers who worked in poultry slaughterhouses and clinical specimens in Romania, indicated a very high rate of ESBL and AmpC-producing E. coli in chickens19.

Genes are mobilized through the interplay of several MGEs of different types. Elements with intra-cellular transposition, such as IS and Tn, can mobilize new genes and transport them to conjugative/ mobilizable transposons and plasmids10. AMR genes associated with integrative MGEs (iMGEs) tended to be located at fixed distances from the iMGE, and was often part of an array of AMR genes. Some arrays exhibited greater variability of the genes contained the arrays than other. This indicates that these genes could be part of an integron, which in turn could be mobilized by the associated IS through the formation of a ComTn. Of these arrays the aph(6)-Id associated with IS1133 and mph(A) associated with IS6100 seemed to have less variation in the genes compared to the other arrays. The lack of variation might be caused by the associated MGEs are less prone to form ComTn.

The distribution of MGEs within the E. coli isolates revealed the lack of well-defined clusters and comparably dissimilar profiles suggesting that E. coli has a diverse and flexible mobilome. This could mean that mobile genetic elements are more mobile in E. coli, either due being frequently mobilized on plasmids or higher conjugation rate. Plasmids are important vehicles for the carriage of MGEs and AMR genes10, the result showed that the E. coli carried approximately four plasmids on average but a minority of the MGEs were located on them. However, the number of mobilized MGEs are likely underestimated due to a combination of the limitations in the detection method of PlasmidFinder20 and the difficulty assembling repetitive DNA, and with that plasmids, from short read sequence data. Moreover, a significant number of plasmid sequences in INSDC databases (DDBJ, EMBL-EBI, NCBI) may be misassembled, as evidenced by the presence of fragments of MGEs that are not explained by truncation by other MGEs21. To further study the mobility and distribution of AMRs one would need to construct near complete the assemblies, preferably via hybrid assembly using long read sequencing as a template, and combine it with additional plasmid prediction tools that do bases it predictions on circular features in the data.

When looking at combinations of MGE, distances and AMR genes that were observed more than four times they were primarily found on a single type of plasmid which could explain the observed pattern.

MGEs can influence the expression and/or mobility of nearby genes through a variety of mechanisms. Through the formation of ComTn can an IS, which lacks accessory genes, transpose several gens as unit within the cell. We identified several putative composite transposons carrying AMR genes based on the location and completeness of IS elements. Of these were IS26 based elements both the smallest, only carrying one or two genes, and the one most frequently predicted to carry different genes. This is likely caused by the tendency of IS26 to form arrays of repeated MGEs22. However, further research is needed to verify if the putative ComTn can mobilize these genes.

AMR genes that were found within 31 kbp of a MGE was considered to be associated, meaning that it has the potential to be mobilized. The threshold corresponds to the length of the largest ComTn from Enterobacteriacea in MobileElementFinders database (Tn6108). Association between MGEs and AMR genes showed that most of the AMR genes carried on MGEs were from beta-lactam class especially blaTEM-1 and beta-lactamases were mainly carried on transposons in poultry and swine. In addition, MGEs such as IS26, ISVsa3 and Tn2 were associated with many AMR genes. IS26 has played a crucial role in the dissemination of resistance determinants in Gram-negative bacteria10,22. AMR genes such as blaTEM genes, including extended-spectrum beta-lactamases (ESBL) encoding genes have been known that they have been found within Tn1, Tn2 and Tn3 transposons10. Analysis of the distance from AMR and virulence genes to the nearest MGEs indicated that MGEs tended to be located at a fixed distance from the AMR and virulence genes suggesting that the AMR and virulence genes in the E. coli isolates have been in part mobilized by the associated MGEs.

The predicted MGEs were only of types that are comparably and performs intra cellular transposition. We did not identify any conjugative/mobilizable transposons such as SXT or ICEEc210 suggests that we are missing these classes MGEs which play and important for the mobilization of genes. These elements are likely not being predicted due to the difficulty in assembling these elements with short read data. Like plasmids, are these elements large and contains repetitive sequences and other MGEs which could confuse the assembler resulting in fragmented assemblies. To overcome these shortcomings, one would likely need to employ long read sequencing. While phages play an important role movement of AMR genes they have not been addressed in this work.

Comparing abundance of AMR from metagenomic analysis of fecal samples and whole genome sequencing of E. coli isolated from the same fecal samples, it showed that 5 main AMR classes out of 10 analyzed classes exhibited significant correlation of AMR abundance between metagenomics and WGS results in E. coli poultry isolates. No significant correlation of AMR abundance was found in any AMR classes between metagenomics and WGS analysis in swine (except colistin) and veal calves. This result suggested that metagenomics analysis could be used as a predictor for abundance of AMR from bacteria isolates in poultry.


Commensal E. coli isolates from different farms in different European countries in this study indicated neither clonality nor clustering based on geographical location of countries and genetic markers such as antimicrobial resistance and virulence genes. The E. coli tended to sporadically spread internationally. Nonetheless, mobile genetic elements play a role in the acquisition of antimicrobial resistance and virulence genes. Abundance of AMR was agreeable between metagenomic and whole genome sequencing analysis in several AMR classes in poultry isolates.

Materials and methods

Selection and collection of isolates

As part of the European Union-funded EFFORT project (, fecal samples from animal farms were collected in 2014–2015 from Denmark (10 farms), France (5 farms), Germany (17 farms), The Netherlands (15 farms), Poland (10 farms), and Spain (10 farms) and. From each farm, up to 10 E. coli isolates were isolated and whole genome sequenced from poultry (n = 239 from 25 farms), swine (n = 238 from 28 farms), and veal (n = 150 from 15 farms). Farms were selected based on a ranking of antimicrobial use (AMU). Distribution of number of E. coli isolates and a list of strains including metadata can be found in Fig. S1 and Table S1 respectively. In addition, previously sequenced metagenomic samples (poultry8, swine8 and veal calf [PRJEB39685]) from the same farm and fecal samples that were used for E. coli isolation were included in this study to compare abundance of AMR from metagenomic and whole genome sequence analysis.

DNA extraction and whole genome sequencing

Genomic DNA was extracted from E. coli isolates using PureLink Genomic DNA Mini Kit (Thermo Fisher Scientific), Qiagen, Gentra, or Puregene bacterial kits and DNA concentrations were determined using the Qubit dsDNA BR assay kit (Thermo Fisher Scientific). Sequencing libraries were prepared with the Nextera XT DNA library preparation kit or the Nextera DNA Flex library preparation kit according to the manufacturer’s protocol (Illumina, Inc., San Diego, CA, United States). Paired-end sequencing was performed on the Illumina HiSeq, MiSeq or NextSeq sequencing platform.

Raw sequence data were deposited in the European Nucleotide Archive ( under study accession no.: PRJEB41365. The raw reads were adapter-trimmed, quality filtered using bbduk (part of the suite bbtools version 36.49)23 and de novo assembled using SPAdes version 3.1124 Genomic sequence data including ENA accession numbers is available in the Table S1.

Species identification, conventional seven multilocus sequence typing, screening for AMR genes, chromosomal point mutations, plasmid and mobile genetic elements

Species identification was used to confirm E. coli species using KmerFinder 3.224,25,26. Conventional seven multilocus sequence typing (MLST) was performed using MLST 2.027. Acquired antimicrobial resistance genes as well as chromosomal point mutations causing resistance were determined using ResFinder 4.028. Virulence genes in E. coli were predicted using VirulenceFinder 2.029. Plasmid replicon was identified to infer plasmid presence using PlasmidFinder 2.1. Integrated mobile genetic elements were predicted in the assembled genomes using MobileElementFinder version 1.0 with MGEdb v1.0.230. With these tools we could identify plasmids, miniature inverted repeats, insertion sequences, transposons and conjugative transposons. If two complete copies of the same MGE, or one complete copy and one truncated with preserved IR region was predicted within 31 kbp from one another they was flagged as putative composite transposons. In addition was MGEs located within 31 kbp from an AMR or virulence gene was considered to be associated and to have a theoretical ability to mobilize the associated gene30. The threshold corresponds to the longest composite transposon in Enterobacteriaceae (Tn6108).

Average nucleotide identity (ANI)

Average nucleotide identity (ANI) was determined from assembled genomes of E. coli strains by all-against-all alignment using FastANI31 ( to estimate whole-genome similarity among the strains from different countries, animal hosts and farms.

Single nucleotide polymorphisms tree

Single nucleotide polymorphisms (SNPs) were determined using CSI phylogeny32. Basically, paired-end reads were mapped to the reference genome, Escherichia coli str. K-12 (accession number; NC_000913.3), using Burrows-Wheeler Aligner (BWA) version 0.7.233. SNPs were identified using ‘mpileup’ module in SAMTools version 0.1.1834. Qualified SNPs were chosen using following criteria: (1) a minimum distance of 10 bps between each SNP, (2) a minimum of 10% of the average depth, (3) the mapping quality was above 25, (4) the SNP quality was more than 30, and (5) all INDELs were excluded. The qualified core SNPs were concatenated to a single alignment corresponding to the position of the reference genome. The concatenated SNP sequences were subjected to parsimony tree construction using PhyML35 with the HKY85 substitution model.

cgMLST tree

The cgMLST loci sequences were retrieved from Enterobase36 ( Raw reads were mapped with KMA against the cgMLST sequences to identify the best-matched allele numbers. Sequences were indexed with kmer size 16 and threshold for read mapping were match = 1, mismatch = − 2, gap-opening = − 3, gap-extension = − 1 and no threshold for read depth was used26. ConClave sorting scheme was used to pick the best matching allele for each cgMLST gene26. While all equally well matching alleles were saved in the first iteration, in a second iteration the most likely allele was chosen as the one with the highest number of matches. KMA is freely available at: and The cgMLSTFinder tool, which is based on KMA, can be found at: The pairwise dissimilarities (distances) between genomes were identified based on the allele profile using ‘gower’ distance method from ‘cluster’ package version 2.1.0 in R. The cgMLST trees were constructed from the distance matrix using hierarchical clustering from ‘ape’ package version 5.4-1 in R.


  1. 1.

    Berg, R. D. The indigenous gastrointestinal microflora. Trends Microbiol. 4, 430–435 (1996).

    CAS  Article  Google Scholar 

  2. 2.

    Gordon, D. M. & Cowling, A. The distribution and genetic structure of Escherichia coli in Australian vertebrates: Host and geographic effects. Microbiology 149, 3575–3586 (2003).

    CAS  Article  Google Scholar 

  3. 3.

    Kosek, M., Bern, C. & Guerrant, R. L. The global burden of diarrhoeal disease, as estimated from studies published between 1992 and 2000. Bull. World Health Organ. 81, 197–204 (2003).

    PubMed  PubMed Central  Google Scholar 

  4. 4.

    Russo, T. A. & Johnson, J. R. Medical and economic impact of extraintestinal infections due to Escherichia coli: Focus on an increasingly important endemic problem. Microbes Infect. 5, 449–456 (2003).

    Article  Google Scholar 

  5. 5.

    Tenaillon, O., Skurnik, D., Picard, B. & Denamur, E. The population genetics of commensal Escherichia coli. Nat. Rev. Microbiol. 8, 207–217 (2010).

    CAS  Article  Google Scholar 

  6. 6.

    O’Neill, J. Tackling Drug-Resistant Infections Globally: Final Report and Recommendations. The Review on Antimicrobial Resistance. (2016).

  7. 7.

    Aarestrup, F. M. The livestock reservoir for antimicrobial resistance: A personal view on changing patterns of risks, effects of interventions and the way forward. Philos. Trans. R. Soc. Lond. B. Biol. Sci. 370, 20140085 (2015).

    Article  Google Scholar 

  8. 8.

    Munk, P. et al. Abundance and diversity of the faecal resistome in slaughter pigs and broilers in nine European countries. Nat. Microbiol. 3, 898–908 (2018).

    CAS  Article  Google Scholar 

  9. 9.

    Andremont, A. Commensal flora may play key role in spreading antibiotic resistance. ASM News 63, 601–607 (2003).

    Google Scholar 

  10. 10.

    Partridge, S. R., Kwong, S. M., Firth, N. & Jensen, S. O. Mobile genetic elements associated with antimicrobial resistance. Clin. Microbiol. Rev. 31, e00088-17 (2018).

    Article  Google Scholar 

  11. 11.

    Chen, S. T. & Clowes, R. C. Variations between the nucleotide sequences of Tn1, Tn2, and Tn3 and expression of beta-lactamase in Pseudomonas aeruginosa and Escherichia coli. J. Bacteriol. 169, 913–916 (1987).

    CAS  Article  Google Scholar 

  12. 12.

    Touchon, M. et al. Phylogenetic background and habitat drive the genetic diversification of Escherichia coli. PLoS Genet. 16, e1008866 (2020).

    CAS  Article  Google Scholar 

  13. 13.

    Duggett, N. et al. The importance of using whole genome sequencing and extended spectrum beta-lactamase selective media when monitoring antimicrobial resistance. Sci. Rep. 10, 19880 (2020).

    ADS  CAS  Article  Google Scholar 

  14. 14.

    Paudel, S., Stessl, B., Hess, C., Zloch, A. & Hess, M. High genetic diversity among extraintestinal Escherichia coli isolates in pullets and layers revealed by a longitudinal study. BMC Vet. Res. 12, 221 (2016).

    Article  Google Scholar 

  15. 15.

    Escobar-Páramo, P. et al. Identification of forces shaping the commensal Escherichia coli genetic structure by comparing animal and human isolates. Environ. Microbiol. 8, 1975–1984 (2006).

    Article  Google Scholar 

  16. 16.

    Skarżyńska, M., Leekitcharoenphon, P., Hendriksen, R. S., Aarestrup, F. M. & Wasyl, D. A metagenomic glimpse into the gut of wild and domestic animals: Quantification of antimicrobial resistance and more. PLoS One 15, e0242987 (2020).

    Article  Google Scholar 

  17. 17.

    Kaesbohrer, A. et al. Emerging antimicrobial resistance in commensal Escherichia coli with public health relevance. Zoonoses Public Health 59(Suppl 2), 158–165 (2012).

    Article  Google Scholar 

  18. 18.

    Hricová, K. et al. Quinolone-resistant Escherichia coli in poultry farming. Cent. Eur. J. Public Health 25, 163–167 (2017).

    Article  Google Scholar 

  19. 19.

    Maciuca, I. E. et al. High prevalence of Escherichia coli-producing CTX-M-15 extended-spectrum beta-lactamases in poultry and human clinical isolates in Romania. Microb. Drug Resist. 21, 651–662 (2015).

    CAS  Article  Google Scholar 

  20. 20.

    Arredondo-Alonso, S., Willems, R. J., van Schaik, W. & Schürch, A. C. On the (im)possibility of reconstructing plasmids from whole-genome short-read sequencing data. Microb. Genom. 3, e000128 (2017).

    PubMed  PubMed Central  Google Scholar 

  21. 21.

    Partridge, S. R. & Tsafnat, G. Automated annotation of mobile antibiotic resistance in Gram-negative bacteria: The Multiple Antibiotic Resistance Annotator (MARA) and database. J. Antimicrob. Chemother. 73, 883–890 (2018).

    CAS  Article  Google Scholar 

  22. 22.

    Harmer, C. J., Moran, R. A. & Hall, R. M. Movement of IS26-associated antibiotic resistance genes occurs via a translocatable unit that includes a single IS26 and preferentially inserts adjacent to another IS26. MBio 5, e01801–e01814 (2014).

    CAS  Article  Google Scholar 

  23. 23.

    Bushnell, B., Rood, J. & Singer, E. BBMerge—Accurate paired shotgun read merging via overlap. PLoS One 12, e0185056 (2017).

    Article  Google Scholar 

  24. 24.

    Bankevich, A. et al. SPAdes: A new genome assembly algorithm and its applications to single-cell sequencing. J. Comput. Biol. 19, 455–477 (2012).

    MathSciNet  CAS  Article  Google Scholar 

  25. 25.

    Larsen, M. V. et al. Benchmarking of methods for genomic taxonomy. J. Clin. Microbiol. 52, 1529–1539 (2014).

    Article  Google Scholar 

  26. 26.

    Clausen, P. T. L. C., Aarestrup, F. M. & Lund, O. Rapid and precise alignment of raw reads against redundant databases with KMA. BMC Bioinform. 19, 307 (2018).

    Article  Google Scholar 

  27. 27.

    Larsen, M. V. et al. Multilocus sequence typing of total-genome-sequenced bacteria. J. Clin. Microbiol. 50, 1355–1361 (2012).

    CAS  Article  Google Scholar 

  28. 28.

    Bortolaia, V. et al. ResFinder 4.0 for predictions of phenotypes from genotypes. J. Antimicrob. Chemother. (2020).

    Article  PubMed  PubMed Central  Google Scholar 

  29. 29.

    Joensen, K. G. et al. Real-time whole-genome sequencing for routine typing, surveillance, and outbreak detection of verotoxigenic Escherichia coli. J. Clin. Microbiol. 52, 1501–1510 (2014).

    Article  Google Scholar 

  30. 30.

    Johansson, M. H. K. et al. Detection of mobile genetic elements associated with antibiotic resistance in Salmonella enterica using a newly developed web tool: MobileElementFinder. J. Antimicrob. Chemother. (2020).

    Article  PubMed Central  Google Scholar 

  31. 31.

    Goris, J. et al. DNA–DNA hybridization values and their relationship to whole-genome sequence similarities. Int. J. Syst. Evol. Microbiol. 57, 81–91 (2007).

    CAS  Article  Google Scholar 

  32. 32.

    Kaas, R. S., Leekitcharoenphon, P., Aarestrup, F. M. & Lund, O. Solving the problem of comparing whole bacterial genomes across different sequencing platforms. PLoS One 9, e104984 (2014).

    ADS  Article  Google Scholar 

  33. 33.

    Li, H. & Durbin, R. Fast and accurate short read alignment with Burrows–Wheeler transform. Bioinformatics 25, 1754–1760 (2009).

    CAS  Article  Google Scholar 

  34. 34.

    Li, H. et al. The sequence alignment/map format and SAMtools. Bioinformatics 25, 2078–2079 (2009).

    Article  Google Scholar 

  35. 35.

    Guindon, S. et al. New algorithms and methods to estimate maximum-likelihood phylogenies: Assessing the performance of PhyML 3.0. Syst. Biol. 59, 307–321 (2010).

    CAS  Article  Google Scholar 

  36. 36.

    Zhou, Z. et al. The EnteroBase user’s guide, with case studies on Salmonella transmissions, Yersinia pestis phylogeny, and Escherichia core genomic diversity. Genome Res. 30, 138–152 (2020).

    CAS  Article  Google Scholar 

Download references


This work was supported by the European Union-funded EFFORT project ( (FP7-KBBE-2013-7, Grant Agreement 613754). By the EJP-one health project Full-force (Grant Agreement No. 773830) and by The Novo Nordisk Foundation (NNF16OC0021856: Global Surveillance of Antimicrobial Resistance). In addition, whole-genome sequencing done at National Veterinary Research Institute was possible due to support of Department of Omics Analyses from the Ministry of Science and Higher Education (Grant No. 4477/E-180/SPUB/2018).

Author information





P.L. and M.H.K.J. wrote the main manuscript text and prepared figures. P.M. analysed Metagenomics data. B.M., M.S., K.W., G.M., A.H., K.T.V., A.B., EFFORT Consortium, M.Z., D.W., P.S., B.G.Z., M.S.M.B., J.A.W., D.J.H., D.M., F.M.A. sampled, isolated, sequenced the isolates including quality assurance of the genomic data. F.M.A. supervised the study. All authors reviewed the manuscript.

Corresponding author

Correspondence to Pimlapas Leekitcharoenphon.

Ethics declarations

Competing interests

The authors declare no competing interests.

Additional information

Publisher's note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Leekitcharoenphon, P., Johansson, M.H.K., Munk, P. et al. Genomic evolution of antimicrobial resistance in Escherichia coli. Sci Rep 11, 15108 (2021).

Download citation


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.


Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing