Genomic Evidence of mcr-1.26 IncX4 Plasmid Transmission between Poultry and Humans

ABSTRACT Colistin is still commonly used and misused in animal husbandry driving the evolution and dissemination of transmissible plasmid-mediated colistin resistance (mcr). mcr-1.26 is a rare variant and, so far, has only been detected in Escherichia coli obtained from a hospitalized patient in Germany in 2018. Recently, it was also notified in fecal samples from a pigeon in Lebanon. We report on the presence of 16 colistin-resistant, mcr-1.26-carrying extended-spectrum beta-lactamase (ESBL)-producing and commensal E. coli isolated from poultry samples in Germany, of which retail meat was the most common source. Short- and long-read genome sequencing and bioinformatic analyses revealed the location of mcr-1.26 exclusively on IncX4 plasmids. mcr-1.26 was identified on two different IncX4 plasmid types of 33 and 38 kb and was associated with an IS6-like element. Based on the genetic diversity of E. coli isolates, transmission of the mcr-1.26 resistance determinant is mediated by horizontal transfer of IncX4 plasmids, as confirmed by conjugation experiments. Notably, the 33-kb plasmid is highly similar to the plasmid reported for the human sample. Furthermore, we identified the acquisition of an additional beta-lactam resistance linked to a Tn2 transposon on the mcr-1.26 IncX4 plasmids of three isolates, indicating progressive plasmid evolution. Overall, all described mcr-1.26-carrying plasmids contain a highly conserved core genome necessary for colistin resistance development, transmission, replication, and maintenance. Variations in the plasmid sequences are mainly caused by the acquisition of insertion sequences and alteration in intergenic sequences or genes of unknown function. IMPORTANCE Evolutionary events causing the emergence of new resistances/variants are usually rare and challenging to predict. Conversely, common transmission events of widespread resistance determinants are quantifiable and predictable. One such example is the transmissible plasmid-mediated colistin resistance. The main determinant, mcr-1, has been notified in 2016 but has successfully established itself in multiple plasmid backbones in diverse bacterial species across all One Health sectors. So far, 34 variants of mcr-1 are described, of which some can be used for epidemiological tracing-back analysis to identify the origin and transmission dynamics of these genes. Here, we report the presence of the rare mcr-1.26 gene in E. coli isolated from poultry since 2014. Based on the temporal occurrence and high similarity of the plasmids between poultry and human isolates, our study provides first indications for poultry husbandry as the primary source of mcr-1.26 and its transmission between different niches.

involved in colistin dissemination are attracting further attention. Selected mcr-1-carrying E. coli isolates collected within the annual national monitoring program for zoonoses were analyzed for investigation of the diversity of mcr-carrying plasmids. Thereby, 16 isolates were identified to harbor the mcr-1.26 variant. Notably, all isolates were obtained between 2014 and 2022 from poultry, more specifically from chicken and turkey retail meat as well as chicken and turkey cecal and fecal samples ( Table 1). The isolates originate from 10 different German federal states, and metadata do not indicate an epidemiologic association between isolates. Retail meat was the most common source with 9/16 samples from which mcr-1.26-carrying E. coli were recovered.
Genetic heterogeneity of isolates indicates horizontal transmission of mcr-1. 26. In order to determine a possible clonal spread of mcr-1.26-carrying E. coli or horizontal transmission of the mcr-1.26 resistance determinant, the phylogenetic relationship of 17 isolates, including the human clinical isolate, was analyzed. Three different E. coli phylogroups were present in the analyzed population, of which phylogroup A was the most common followed by phylogroup B1, which included the human isolate 803-18 (Fig. 1A). The E. coli population studied is characterized by 11 different serotypes and 10 different sequence types (STs), reflecting the overall genomic diversity of the isolates. The majority of sequence types have been described to be present in humans and domestic animals and some belong to pandemic lineages (18)(19)(20)(21)(22)(23)(24). Conversely, ST155 of the human clinical isolate 803-18 has been commonly found in poultry samples (19,25). Notably, two clusters contain two closely related isolates each (14-AB02757 and 16-AB01274; 18-AB02652 and 22-AB01721) represented by an identical core genome multilocus sequence type (cgMLST) ( Table 1). Furthermore, these four isolates were obtained from turkey. The single-nucleotide polymorphism (SNP)-based phylogenetic tree using the genome of the human clinical isolate 803-18 as reference showed a substantial diversity of all 17 isolates and covered 75.6% of the reference ( The virulence profiles of all strains are overall comparable, but three strains possessed additional genes encoding adhesive fimbriae (P fimbriae, afa) or a chu locus, which are common in pathogenic E. coli (see Table S2 in the supplemental material) (26,27).
Based on the genetic heterogeneity of the isolates, colistin resistance is likely associated with a mobile genetic element that allows horizontal transmission of the mcr-1.26 gene. In order to investigate the transferability of the colistin resistance, conjugation assays were performed. Intraspecies transmission of the colistin resistance could be confirmed by molecular detection of mcr-1-like genes in transconjugants. The obtained transconjugants showed equal MIC values toward colistin as the corresponding wildtype strains ( Table 2). The mcr-1-like genes were transferred with a frequency of 1.5 Â 10 3 to 2.1 Â 10 5 . The results suggest horizontal gene transfer of mcr-1.26 rather than clonal spread of colistin-resistant isolates.
Characterization of the resistome. Besides mcr-1.26, all strains harbored several additional resistance genes and chromosomal mutations leading to antimicrobial resistance (AMR) and a multidrug resistance phenotype. Resistome analysis revealed 45 different resistance determinants in the E. coli population studied, of which isolates 20-AB00574 (obtained from cecum) and 22-AB00571 (obtained from meat) harbored the most resistance genes (n = 18) (Fig. 1B). Notably, missense mutations in PmrA and PmrB, a two-component system involved in lipid A modification and thus colistin resistance, were found in six different strains (Table 1). However, mutations have not Cecum samples were taken during the slaughter process.
c Feces samples were taken from the animal farm. The genome size was determined using the AQUAMIS pipeline (https://gitlab.com/bfr_bioinformatics/AQUAMIS/). g The total gene count was obtained using the NCBI Prokaryotic Genome Annotation Pipeline (PGAP).
h Polymorphism does not contribute to colistin resistance.
i Reported polymorphism but its association with colistin resistance is unknown and not experimentally confirmed.
j Unknown contribution to colistin resistance.
Similar mcr-1.26-IncX4 Plasmids in Poultry and Humans Microbiology Spectrum been experimentally confirmed to cause colistin resistance and have been reported also in sensitive strains (1,28). Eight different genes conferring resistance to b-lactams were detected in nine isolates, whose ESBL phenotype was confirmed by antimicrobial susceptibility testing (AST). Following mcr-1.26, the chromosomal mutations gyrA_S83L (82.35%) and mdf(A) (70.59%) were the most abundant resistance-associated determinants (Fig. 1C). mcr-1-like genes are encoded on different conjugative IncX4 plasmid types. In silico analysis revealed the presence of at least three different incompatibility (Inc) groups in each strain (see Table S2). The mcr-1.26 genes were located on IncX4 plasmids in all isolates and could be associated with an IS6-like element encoding an IS26 family transposase as previously observed for the human isolate 803-18 (17). We noticed that four transconjugants acquired additional resistance to ampicillin ( Table 2). In silico analyses and molecular investigation revealed the colocalization of mcr-1.26 with a bla TEM gene in three isolates (see Fig. S3 in the supplemental material). This indicated at least two different IncX4 plasmid types carrying mcr-1-like genes, which was confirmed by hybrid assembly of IncX4 plasmids from isolate 14-AB01188 (pEc141188) and 20-AB00574 (pEc200574) ( Table 3). A 33,310-bp plasmid was reconstructed for isolate 14-AB01188, carrying functions for their spread and maintenance (i.e., a hicAB toxin-antitoxin system) (Fig. 2). BLAST analyses revealed high similarity (99.96%; coverage, 99%; harbors mcr-1.1) with a plasmid of similar size (GenBank accession number CP092316.1) obtained from Salmonella enterica (Table 3). In contrast, hybrid assembly generated a 38,265-bp IncX4 plasmid for isolate 20-AB00574. A bla TEM-135 gene exhibiting an identity of 99.84% to 99.88% was found to have been integrated upstream of mcr-1.26 by the insertion of a Tn2 element (Fig. 2). This element integrated at a TATTG sequence in an intergenic region, which resulted in the formation of 5-bp direct repeats flanking the transposon. Sequence comparisons of pEc200574 did not yield a clear result in the NCBI database but predicted similarity to two different plasmids ( Sequence alignments of trimmed raw reads of poultry and human isolates using pEc141188 and pEc200574 as reference confirmed two different types of mcr-carrying IncX4 plasmids within the E. coli population. Plasmid pEc141188 was found in 14 isolates, including the human isolate, whereas pEc200574 was found in three different E. coli isolates obtained from turkey (Table 3). Subsequent analysis of SNPs revealed highly conserved regions in both plasmid types (see Fig. S4 in the supplemental material). All other coding sequences contained synonymous and nonsynonymous SNPs with varying predicted effect on protein function. However, it seems that the SNPs had no obvious impact on the functionality of the plasmid and its transmissibility. Based on the finding that LR882927.1 carried mcr-1.26, we were interested in whether additional mcr-1.26 genes have been reported in public databases (GenBank accessed 13 January 2023). We identified a total of three reports of mcr-1.26 genes, which were located on IncX4 plasmids originating from E. coli and also highly similar to pEc141188 (see Table  S3 in the supplemental material). Interestingly, E. coli was isolated from a human in the Netherlands (GenBank accession number LR882927.1) and from raw retail turkey meat in Czech Republic (GenBank accession numbers MT929276.1 and MT929278.1), both countries neighboring Germany (29).

DISCUSSION
The high prevalence of mcr-carrying Enterobacterales in livestock production and the spillover of resistant bacteria and mcr genes into the human and environmental sectors contribute to the global antimicrobial resistance crisis. The estimated global prevalence of mcr genes in Enterobacterales is 4.7% (47 countries across 6 continents) (13). Animal husbandry is the most significant contributor, with reported prevalences ranging from 1.8% to 63% in poultry and 0.35% to 98.0% in pigs, followed by the human sector with prevalence rates of 0.05% to 4.73% and the environmental sector with prevalence rates ranging from 0.8% to 2.4% (12). In Germany, prevalence rates of mcr-positive Enterobacterales varied from 4.7% to 8.7% for poultry, from 0.4% to 9.9% for pigs, and 1.8% for wild boar. A recent study covering different stages of the meat production chain in Germany determined an overall prevalence of mcr-1-positive Enterobacterales of 18%, with the highest prevalence of 43% at a poultry slaughterhouse (30).
The first mcr-1.26-positive E. coli isolate was isolated from turkey retail meat in 2014, which coincides with the introduction of mandatory susceptibility testing to colistin for bacteria isolated from food-producing animals implemented in the European Union (31). Therefore, it is possible that mcr-1.26-positive E. coli were present in poultry before 2014. Thereafter, mcr-1.26 E. coli were found at biennial intervals, as the monitoring plan specified poultry sampling every 2 years. Notably, mcr-1.26 has also been found in samples outside of Germany, an E. coli isolated from a human in Netherlands and two E. coli obtained from retail meat in Czech Republic, which had been imported from Germany. To date, mcr-1.26 has not been reported in bacterial species other than E. coli or in other animal species. It is worth mentioning that during the BLASTN search, we noticed that other hits had an incomplete start codon, lacking the first two nucleotides, which is crucial to distinguish between mcr-1.26 and mcr-1.1 and eventually leading to an underestimation of mcr-1.26 prevalence (17). The isolates carrying mcr-1.26 in this study comprised a variety of different phylogroups, STs, and serotypes, indicating horizontal gene transfer of mcr-carrying plasmids rather than clonal spread, although these findings are based on a limited number of 16 isolates. Up to present, 15 different Inc-type plasmids associated with mcr-1 have been documented (12). Most plasmids are transferable, of which IncX4, IncHI2, and IncI2 are predominant worldwide (32). In the current study, mcr-1.26 was located exclusively on conjugative IncX4 plasmids. This result is consistent with findings of IncX4 plasmids carrying mcr-1.26 in the human isolates from Germany and Netherlands as well as isolates from turkey meat from Czech Republic. However, we identified two different types of IncX4 plasmids in our E. coli population. mcr-1-harboring IncX4 plasmids with no additional resistance genes seem to disseminate increasingly among Enterobacterales, such as E. coli, Salmonella enterica, and Klebsiella pneumoniae, of animal, human, and environmental origin. Comparative analysis to pEc141188 revealed a ubiquitous presence of highly similar plasmids, indicating its genetic stability as well as its successful global dissemination into different sectors.
The presence of an additional bla TEM-135 gene in pEc200574 suggests that IncX4 plasmids, despite their stability, are also subject to evolution. No similar plasmids could be found in the NCBI database. Interestingly, an IncX4-IncI2 hybrid plasmid of 95,202 bp (MT929289.1) encoded mcr-1.1 and the bla TEM-135 -Tn2 cassette, which was isolated from raw retail meat in Czech Republic originally imported from Germany (29). The fact that pEc200574 was present in E. coli of different phylogroups and ST groups highlights the transmissible nature of this IncX4 plasmid type, thus contributing to the spread of multidrug resistance. It remains subject of future investigations whether this plasmid type will disseminate further in turkey husbandry or even into other ecological niches. Both types of IncX4 plasmids contained a hicAB toxin-antitoxin system, which contributes to plasmid maintenance (33). The presence of hicAB on mcr-1-IncX4 plasmids has been described previously and may be an explanation for the successful dissemination of the plasmid type (34,35). With our study, we have confirmed the transmissible nature of IncX4 plasmids and their role in the spread of mcr-1 genes. Interestingly, mcr-1.26-IncX4 plasmids have disseminated in two different poultry species, chicken and turkey. Furthermore, we show that retail meat is a common reservoir for mcr-1.26-positive E. coli in Germany. This observation is in agreement with previous studies reporting the presence of the mcr-1-carrying IncX4 plasmids in Enterobacterales from retail meat in different countries, pointing out the significance of food-producing animals and retail meat as reservoirs of mcr-1-carrying bacteria and a potential exposure risk to consumers (29,(36)(37)(38)(39)(40)(41)(42)(43)(44).
Spontaneous alteration of the coding sequence that do not affect the protein function or the extensive use of colistin in poultry production in Germany may have been a driver for the evolution of the mcr-1 gene and eventually resulted in the emergence of mcr-1.26 (7,8). Colistin is generally administered to the entire flock for disease treatment or metaphylaxis purposes, which may have ensured the persistence of mcr-1.26 in poultry husbandry in Germany, thus providing a source for mcr-1.26 transmission into the food chain. Additionally, Germany is one of the largest producers and exporters of poultry and meat thereof in the European Union, which may have led to transnational transmission of mcr-1.26 and the detection of mcr-1.26 E. coli on retail meat in Czech Republic (45). Thus, international trade plays a further role in the spread of resistant bacteria.
The study of the transmission routes of colistin-resistant bacteria is complicated by the fact that resistance genes are often transmitted by mobile genetic elements. There is a possibility of transmission of promiscuous IncX4 plasmids containing mcr-1.26 to other bacterial hosts in diverse environmental conditions. Furthermore, mcr-1.26-IncX4 plasmids have been found across distantly related bacterial strains from human and animal origin, which, however, does not exclude a possible zoonotic transmission chain. In line with other research, our results indicate a spillover of mcr-1-like genes into different One Health sectors, highlighting the need to reduce colistin pressure through a One Health approach, both by reducing the use of colistin in poultry husbandry and by applying structural and medical cost-effective alternatives to maintain poultry production.

MATERIALS AND METHODS
Background of isolates. E. coli isolates were collected in the framework of the annual national monitoring program for Germany established by the German Federal Institute for Risk Assessment (BfR), the Federal Office for Consumer Protection and Food Safety (BVL), and the authorities of the federal states to comply with Directive 2003/99/EC and commission implementing decisions (CID) 2013/652/EU and 2020/1729/EU. Selected isolates carrying mcr-1, as confirmed by PCR, were further characterized by whole-genome sequencing (WGS), of which 16 isolates harbored mcr-1.26. The study conducted was not a prevalence analysis. E. coli from this study were isolated from chicken and turkey meat as well as broiler and turkey cecal and fecal samples by regional laboratories and subjected to species confirmation (i.e., growth on indicator media, matrixassisted laser desorption ionization-time of flight [MALDI-ToF]) and antimicrobial susceptibility testing (AST) at the National Reference Laboratory for Antimicrobial Resistance (NRL-AR) hosted at the BfR, as previously described (46).
Whole-genome, Sanger sequencing, and bioinformatics analyses. For WGS, bacterial genomic DNA from liquid overnight cultures was isolated using the PureLink genomic DNA preparation minikit (Invitrogen GmbH, Darmstadt, Germany) according to the manufacturer's instructions. The Nextera DNA Flex library prep kit was used for DNA library preparation as previously published followed by a paired-  Table S1 in the supplemental material). For plasmid analyses, isolates 14-AB01188 and 20-AB00574 were additionally subjected to long-read sequencing and used as templates for reference mapping-based approaches. Therefore, Oxford Nanopore Technology (ONT) sequencing libraries were sequenced on a MinIon Mk1C device followed by a hybrid assembly using Unicycler v0.4.8 as previously described (52). In order to confirm the mcr-1.26 variant, commercial Sanger sequencing of mcr-1 gene amplicons was performed for all isolates (Eurofins Genomics, Ebersberg, Germany) using the primers and PCR protocol as described (17).
Annotation of the bacterial genomes was performed using the automated Prokaryotic Genome Annotation Pipeline (PGAP) (National Center for Biotechnology Information) (53). Phylogenetic analysis of mcr-carrying E. coli was conducted using a reference genome-based, single-nucleotide polymorphism (SNP) strategy with CSIPhylogeny 1.4 under default settings (54). The phylogenetic tree was visualized using the iTOL 6.6 software (55). Isolates were characterized in silico using the web-based tools provided by the Center for Genomic Epidemiology under the default settings (www.genomicepidemiology.org) ( Table 1). The phylogroup was determined in silico using the web-interface ClermonTyper (http://clermontyping.iame-research .center/) (56).
Identification of different IncX4 plasmid types and analyses of SNPs was performed with Geneious Prime 2020.2.2 software under default settings using the hybrid reference assemblies for plasmid genomes pEc141188 (isolate 14-AB01188) and pEc200574 (isolate 20-AB00574). A minimum variant frequency threshold of 50% was used for SNP identification. Plasmid-encoded genes were annotated using PATRIC (The Bacterial and Viral Bioinformatics Resource Center; https://www.bv-brc .org/; accessed November 2022) (57).
In vitro filter mating assays. In order to evaluate the transferability of the colistin resistance, filter mating assays were performed. Briefly, liquid cultures of mcr-carrying donor strains and the SAZ r E. coli J53 recipient strain were mixed in a ratio of 1:2 and centrifuged at 3,500 Â g for 5 min. The supernatant was discarded, and the remaining pellet was resuspended in 100 mL lysogeny broth (LB) and applied on a 0.2 mm pore-size filter on an LB agar plate. Following an incubation of 24 h at 37°C, bacteria were removed from pore-size filters by suspending in 4 mL LB broth. Transconjugants were selected by plating 100 mL aliquots on LB agar plates supplemented with 100 mg/L sodium azide and 2 mg/L colistin. To verify successful conjugation, the transconjugants were subjected to AST and analyzed for the presence of mcr-1 and beta-lactamase (bla TEM ) by PCR.
PCR. The presence of mcr-1.26 and bla TEM was determined by PCR as previously published (11,58), while the proximity of both genes was confirmed as follows. First, genomic DNA of poultry isolates and corresponding transconjugants was analyzed for the presence of bla TEM using the primers and PCR conditions as published resulting in an amplification product of 503 bp (see Fig. S3 in the supplemental material) (58). Subsequently, the intergenic region between mcr-1.26 and bla TEM , including the genes, was amplified with the reverse primers (Mcr-1a REV 59-GGGCATTTTGGAGCATGGTC-39/TEM-R 59-ACCAATGCT TAATCAGTGAG-39; annealing temperature, 55°C; elongation time, 480 s), leading to a 7,317-bp product confirming the close proximity of the resistance genes (17,59).
Data availability. The assembled genomic sequences of poultry isolates were deposited under the BioProject number PRJNA726012 in the NCBI database. The two plasmid sequences were deposited in GenBank (pEc141188, ID2678299, accession number OQ557085; pEC200574, ID2678314, accession number OQ557086). The Lipid A phosphoethanolamine transferase mcr-1.26 is annotated as a misc feature due to the altered start codon.