Colistin is an antibiotic that has been used for decades in veterinary medicine to control gastrointestinal infections in animals, but also to treat systemic, respiratory or urogenital infections in poultry (Kempf et al., 2013). In France, poultry displayed the greatest exposure to polypeptides, including mainly colistin, measured in ALEA [Animal Level of Exposure to Antimicrobials (Anses, 2022b)], a ratio between the live weight treated and the weight produced (poultry polypeptides ALEA = 0.117), followed by pigs (0.058) and cattle (0.011). In fact, in 2020, colistin consumption was mainly attributed to poultry (45.7%), pigs (30.9%) and cattle (18.4%) (Anses, 2022b). Moreover, polypeptides were the most common class of antibiotics used in broilers and turkeys (representing 30% of the total exposure in ALEA), before mainly penicillins (25%), tetracyclines (20%), sulfonamides (10%) and trimethoprim (8%). After oral administration, the antibiotic is poorly absorbed through the gastrointestinal tract (Mead et al., 2021). Therefore, the intestinal microbiota is subject to selective pressure that can lead to the emergence of colistin resistance. Until 2015, the only known colistin resistance mechanisms were linked to chromosomal mutations affecting the lipopolysaccharide charge, reducing colistin interaction with the outer cell membrane of Gram-negative bacteria. Subsequently, the first mobile colistin resistance gene mcr-1 was discovered in Escherichia coli isolates of various origins, first in China and then worldwide across multiple bacterial species (Liu et al., 2016; Xavier et al., 2016). Since 2015, nine new mcr genes (mcr-2 to mcr-10) and subvariants have been described in Enterobacterales (Wang et al., 2020). In the commensal E. coli strain collection isolated from livestock in France between 2007 to 2014, mcr-1 prevalence average was 5.9% in turkeys and 1.8% in broilers, indicating that mobile colistin resistance was already widespread in farm animals (Perrin-Guyomard et al., 2016). Given the dissemination capabilities of mcr-1, the high usage levels of colistin in animals and the high percentage of multi-drug-resistant (MDR) isolates in strains harbouring mcr-1 (Perrin-Guyomard et al., 2016; Mead et al., 2022), the French national Ecoantibio2 (Ecoantibio2, 2017) plan to tackle the risk of antibiotic resistance in veterinary medicine, set up from 2017, had the objective of a 50% reduction in exposure to colistin in food producing animals within 5 years, from 2014–2015 to 2020. Therefore, there was a need to update prevalence and epidemiology data for colistin resistance in poultry production in the country after the implementation of national policies and new scientific findings. For this purpose, mcr genes and genetic diversity of strains have been investigated in colistin-resistant E. coli originating from broilers and turkeys at slaughterhouses from 2011 to 2020. The correlation between mcr-positive isolate prevalence and colistin usage data was explored to evaluate the efficacy of the stewardship programme in poultry production.

This study aimed to deepen our understanding of the prevalence and genotypic characteristics of colistin-resistant indicator E. coli isolated from broilers and turkeys sampled at slaughterhouses in France, in consideration of sales of colistin for the poultry sector. This analysis was made possible as strains were derived from the European monitoring programme of antibiotic resistance in food producing animals conducted each year, with the same representative sampling of poultry production in the country, e.g., sampling design and sample size, and the same analytical methods. Sales data were collected annually from marketing authorisation holders in a standardised and consistent process (Anses, 2022b).

Our phenotypic results indicate that percentages of colistin resistance are rather low in poultry, but three times higher in turkeys than in broilers. This difference in colistin resistance levels between the two poultry species is also reported in other European countries with ratios ranging from 4 in Italy (5.9% in broilers, 22.9% in turkeys) to 9 in Portugal (3% in broilers, 27% in turkeys) and 15 times more in Poland (0.2% in broilers, 3.1% in turkeys) in 2014 (Alba et al., 2018; Clemente et al., 2019; Zając et al., 2019). An attempt to distinguish between turkeys and broilers exposure from sales data, showed that polymyxins exposure was higher in turkeys than in broilers (Anses, 2022b). This observation was confirmed in parallel by on-farm data collected by professionals (Rousset, 2019; Zając et al., 2019) speculated that the difference in colistin resistance between broilers and turkeys could be related to the longer life span of turkeys, and consequently to a longer length of exposure to colistin treatment, favouring antibiotic resistance selection. During the monitoring period, the number of colistin-resistant E. coli decreased in turkeys, but this decrease was not significant. These findings are consistent with the general trends reported in Europe among food-producing animals, where levels of colistin resistance have decreased since 2014 in isolates from broilers and turkeys to reach median levels of rare in broilers (0%) to low in turkeys (1.2%) in 2020 (European Food Safety Authority and European Centre for Disease Prevention and Control, 2022). In veterinary medicine in France, similar decreasing trends have also been observed from diseased animals since 2007. In 2020, the proportions of non-susceptible E. coli isolates collected from various pathologies reached less than 3% in turkeys and 2% in broilers (Anses, 2022a). The decrease in resistance observed between 2014 and 2020 could be attributed to lower sales volumes observed in the same period. A significant correlation between polymyxin exposure and colistin resistance was observed from our data. A significant relationship has also been reported in Europe from annual surveillance data analysis (European Centre for Disease Prevention and Control (ECDC) European Food Safety Authority (EFSA) and European Medicines Agency (EMA), 2021). In the same way, ban of avoparcin as growth promoter in Denmark in 1995 statistically highly significant declined vancomycin-resistant Enterococcus faecium isolated from broilers during 1995–1998 (Bager et al., 1999). Outside Europe, colistin withdrawal policy after 2015 have also had a significant effect on reducing colistin resistance in both animals and humans in China for 4 years (Wang et al., 2020). The decrease in resistance in our study can therefore be considered an outcome of the stewardship measures adopted in France for reduced colistin usage, which included development of resistance screening method for veterinary diagnostic laboratories (Jouy et al., 2017), publication of scientific opinions in addition to official European (European Medicine Agency (EMA) and European Surveillance of Veterinary Antimicrobial Consumption (ESVAC), 2016) and national reduction targets (Ecoantibio2, 2017). These specific measures were also supported by generic ones applying to all antimicrobial classes, in line with the two Ecoantibio national plans (Urban et al., 2023). Both antimicrobial usage monitoring systems providing information in poultry production in France reported a considerable decrease in polymyxin exposure, −70% between 2011 and 2020 based on poultry sales (Anses, 2022b), −79% in broilers and −57% in turkeys between 2018 and 2020, considering on-farm professional data collection (Rousset et al., 2022). The official national objective of −50% in colistin sales between 2014/2015 and 2020 included in Ecoantibio2 plan (Ecoantibio2, 2017) was achieved and surpassed, with a −68% reduction achieved in poultry. In the same period total antimicrobial consumption in poultry production decreased by 55%. These results highlight engagement and voluntary adhesion of all professionals to the reduction objectives in the absence of specific regulation.

All but one colistin-resistant strains harboured the mobile gene mcr-1. None of the mcr-2 to mcr-9 genes were detected in our study. These findings support the reported data on the prevalence of mcr genes from livestock animals. Moreover, recent meta-analyses including publications concerning mcr-mediated colistin-resistant E. coli have shown that mcr-1 is the most predominant and abundant variant worldwide, especially in poultry isolates (Bastidas-Caldes et al., 2022; Shi et al., 2023). Nevertheless, not all studies have determined the presence of all the reported variants of the mcr genes. This bias in the screening could explain why the other variants of the mcr genes have been less frequently reported.

Most colistin-resistant strains are MDR, with various resistance patterns. These phenotypic antimicrobial resistance patterns are explained by antimicrobial resistance genes, except in a few cases, where MICs of colistin, azithromycin, chloramphenicol or trimethoprim were one two-fold dilution step higher than the Ecoff, which could explain why no gene was found by WGS. In addition to the high diversity in antimicrobial resistance patterns, multiple serotypes, phylotypes and ST types were encountered among isolates of both turkey and broiler origin, demonstrating a wide range of E. coli strains harbouring mcr-plasmids. The distribution of phylotypes encountered in microbiota of animals is usually more diverse than in humans (Smati et al., 2015). High counts of phylogroup A associated with variable proportion of other group are a characteristic of the enterocolitype 3, often host-specific to chicken (Pavlickova et al., 2015; Smati et al., 2015) whereas phylogroup B2 is dominant in commensal human strains in developed countries (Tenaillon et al., 2010). From a one health perspective, it is important to highlight that three mcr-1-positive strains isolated from turkeys in 2014, were characterised as ST131 O25:H4, belonging to the phylogenetic group B2. E. coli sequence type 131 is a worldwide pandemic clone, causing predominantly community onset antimicrobial-resistant infections in humans (Rogers et al., 2011). This clone is an extra-intestinal pathogenic E. coli, initially described amongst E. coli harbouring the bla_CTX-M-15 ESBL gene and also amongst fluoroquinolone-resistant non-ESBL-producing isolates. The ST131 strains isolated in our surveillance programme were MDR, with resistance to CIAs such as fluoroquinolones and/or cephalosporins. These isolates also contained various virulence factors, including traT, ompT, iutA, fimH and kpsM II implicated in bacterial survival, invasiveness or adhesion (Riley, 2014) and are a microbiological hazard for humans through the food chain.

IncI2, IncX4 and IncHI2 accounted for 77% of plasmid types driving the mcr-1 genotype in our poultry isolates. This is consistent with other reported findings on mcr-1 plasmid location for strains of poultry origin. From 80 mcr-1-positive E. coli isolated from food-producing broilers and turkeys in Poland between 2011 and 2016, the mcr-1 gene was detected on the same contig as IncX4 (76.3%) and IncHI2 (6.3%) replicons (Zając et al., 2019). In another study, the mcr-1 gene was located on IncX4 and IncHI2 plasmids in E. coli from broilers in the Netherlands (Veldman et al., 2016). In the environment of German poultry slaughterhouses, Savin et al have demonstrated that E. coli carrying mcr-1 isolated from process waters and wastewater were mostly affiliated with IncX4 (37%), IncI1 (21%) and IncHI2 (21%) plasmids (Savin et al., 2020). IncI plasmids are also described in Europe predominantly in E. coli isolated from poultry sources (Rozwandowicz et al., 2018). In vitro competition assays have demonstrated that mcr-1 bearing IncI2, IncX4 and IncHI2 plasmids were quite stable and would not affect the growth of their bacterial hosts (Wu et al., 2018). This could explain why these plasmids were most often found in colistin-resistant mcr-1-containing E. coli of poultry origin. Moreover, mcr-1 bearing IncHI2-type plasmids are often associated with other antimicrobial resistance genes, as shown in our study. This means that in spite of the reduction of colistin use in poultry, selective pressure with other antimicrobial treatments like ampicillin may help to maintain plasmid-mediated colistin resistance and multidrug resistance spread. Resistance to CIAs like fluoroquinolones in most mcr-positive strains could be a major public health hazard towards therapeutic options. Conversely, mcr-1 bearing plasmids belonging to the IncX4 and IncI2 rarely carry antimicrobial resistance genes other than mcr-1, as demonstrated in our study. Nevertheless, plasmids IncX and IncI are narrow-host range plasmids, mainly isolated from E. coli from animal sources.

In broiler and turkey productions in France, we could detect only the mcr-1 gene in different lineages of colistin resistant E. coli isolated at random and non-selective from caecal content of healthy poultry. This gene is located in different plasmid types, carrying various other antimicrobial resistance genes. Different replicons associated with the mcr-1 gene were identified, including mainly IncHI2/IncI2/X4. Many isolates were MDR, highlighting the potential of colistin resistance to be co-selected by several other antibiotics. Nevertheless, the drop in colistin sales, as well as those of the other main antibiotics used in veterinary medicine, is helping to reduce the selection pressure for colistin resistance. These results, from an ecological design not establishing causal link, reflect the positive impact of the various action plans for the reduction of antibiotic use in the veterinary field and the importance of antibiotic resistance surveillance programmes.

The datasets presented in this study can be found in online repositories. Raw sequence data were submitted to the European Nucleotide Archive under study accession no. PRJEB59826.

APG: Writing – original draft. PH: Writing – review & editing. PL: Writing – review & editing. AF: Writing – review & editing. LD: Writing – review & editing. CC: Writing – original draft. IK: Writing – original draft.