Antimicrobial resistance (AMR) is a major global health problem with livestock farms and their surrounding environment highlighted as a potential source of AMR infections¹. The development of AMR in individual bacterial species is dependent not only on antibiotic exposure but also on the presence of other bacteria within their environment, with which they interact². Hence, acknowledging the extent to which different bacteria co-existing within the same environment can co-evolve and share their genome could help the development of more efficient treatments to fight AMR^3,4, specially in settings with high antimicrobial exposure.

We have focused on two important opportunistic pathogens found in livestock, Escherichia coli and Salmonella enterica, that can co-exist in the gut of chickens in livestock farming. These species are commonly associated with food poisoning in humans and display important levels of drug resistance. Moreover, they are a significant cause of diarrheal disease-associated mortality in humans, particularly, in low-to-middle-income countries^5. Previous studies have shown that these species can communicate to increase antibiotic tolerance using bacterial signalling and transfer genetic material via mobile genetic elements (MGEs), a mechanism by which AMR is spread^6,7. 

In this study, we collected 518 E. coli isolates and 143 S. enterica, with an overlapping subset of 113 isolates of each species collected from the same biological samples, from the gut of chickens and surrounding environments, on ten commercial poultry farms and four connected slaughterhouses in three provinces of China over two-and-a-half years. Isolates of both species were collected from chicken faeces, chicken carcasses, chicken feathers, chicken caecal droppings, chicken feed, external soil, the barn environment, wastewater, anal swabs, the abattoir environment, and drinking water. All isolates were subjected to characterisation by antimicrobial susceptibility testing against a panel of up to 28 antimicrobials and whole genome sequencing. 

We employed a data-mining approach enhanced by machine learning, Bayesian divergence analysis, and genome-scale metabolic (GSM) modeling to explore the dissemination and evolution of AMR within E. coli and S. enterica. Our findings indicate that, on a broader scale, distinct variations in the phylogeny and evolution of S. enterica and E. coli were evident. However, on a more detailed level, most isolates within each species co-existing in the gut of chickens and environment shared identical plasmids and MGEs containing clinically significant antibiotic resistance genes (ARGs), which also appear to have co-evolved, which in real-world settings pose a high risk of AMR transfer to humans and the environment⁸. Moreover, these AMR-carrying MGEs could also potentially be a pre-requisite for the bacteria to occupy the same host and environment, as the horizontal gene transfer process may drive the development of host adaptation.

A supervised machine learning framework was used to investigate which genetic determinants (features) were underlying the experimentally determined resistance / susceptibility profiles and if they were in common between the two species. Features underlying the predictions included mutations in core genes, the presence of accessory genes, and mutations in non-coding genome regions. Importantly, approximately 99% of these features were novel antibiotic resistance associations, i.e. not known AMR genes. Genes correlated to resistance by machine learning were found to be enriched for plasmid-located genes in E. coli, but not in S. enterica. To assess the overlap of selected AMR-associated genetic determinants between our co-habiting E. coli and S. enterica isolates we compared the results of our co-habiting Chinese cohort against the same methods applied to not-necessarily co-inhabiting European data collected in other studies. Overall, we found a significantly greater degree of overlap in our cohort (40% compared to 14%).

To understand the mechanistic relationships connecting identified antibiotic resistance genetic signatures, the study integrated machine learning-identified determinants into GSM models. Interestingly, both species exhibited a high number of the same metabolic pathways correlated with antibiotic resistance. However, the phylogenetic patterns of these determinants differed, with prevalent patterns across all metabolic pathways in E. coli, while in S. enterica, the determinants were prevalent in specific serotypes across all pathways.

In conclusion, despite differences in the phylogeny and evolution of S. enterica and E. coli at a larger scale, we found that E. coli and S. enterica, co-existing in the gut of chickens in livestock farming, feature a higher share of AMR-related genetic material, implement more similar resistance and metabolic mechanisms, and are likely the result of a stronger co-evolution pathway. The presence of conserved mobile ARG structures between the two species indicates: i) the potential for easy transfer of resistance, highlighting the challenge of resistance spread within a microbial community; and ii) a recent common evolution. Moreover, the study emphasized the efficacy of integrating whole-genome sequencing and machine learning to identify AMR genes, considering the evolving nature of AMR, while unveiling the potential links between metabolic processes and AMR using a GSM approach, providing insights into the functional overlap of pathways involved in AMR.

Overall, this work has highlighted the need to investigate the interplay of mobilome, resistance, and metabolism in bacteria coexisting in the same gut microbiome in relation to AMR, and has shown the importance and the power, as also suggested by others⁹, of adopting bespoke analytical methods when studying how bacteria coexisting in multi-species communities respond to antibiotics.

 References

1. O'Neill, J. Tackling drug-resistant infections globally: final report and recommendations. The Review on Antimicrobial Resistance (2016)
2. Bottery, M. J., Pitchford, J. W. & Friman, V. P. Ecology and evolution of antimicrobial resistance in bacterial communities. Isme j 15, 939-948, doi:https://10.1038/s41396-020-00832-7 (2021).
3. Partridge, S. R., Kwong, S. M., Firth, N. & Jensen, S. O. Mobile Genetic Elements Associated with Antimicrobial Resistance. Microbiol. Rev. 31, e00088-00017, doi:doi:10.1128/CMR.00088-17 (2018)
4. Davies, N. G., Flasche, S., Jit, M. & Atkins, K. E. Within-host dynamics shape antibiotic resistance in commensal bacteria. Nat Ecol Evol 3, 440-449, doi:10.1038/s41559-018-0786-x (2019)
5. Troeger, C. et al. Estimates of the global, regional, and national morbidity, mortality, and aetiologies of diarrhoea in 195 countries: a systematic analysis for the Global Burden of Disease Study 2016. Lancet Infect. Dis. 18, 1211-1228 (2018)
6. von Wintersdorff, C. J. H. et al. Dissemination of Antimicrobial Resistance in Microbial Ecosystems through Horizontal Gene Transfer. Microbiol. 7, doi:10.3389/fmicb.2016.00173 (2016).
7. Vega, N. M., Allison, K. R., Samuels, A. N., Klempner, M. S. & Collins, J. J. Salmonella typhimurium intercepts Escherichia coli signaling to enhance antibiotic tolerance. Proceedings of the National Academy of Sciences 110, 14420-14425, doi:https://doi.org/10.1073/pnas.1308085110 (2013).
8. Manyi-Loh, C., Mamphweli, S., Meyer, E. & Okoh, A. Antibiotic use in agriculture and its consequential resistance in environmental sources: potential public health implications. Molecules23, 795 (2018).
9. Bottery, M. J., Pitchford, J. W. & Friman, V.-P. Ecology and evolution of antimicrobial resistance in bacterial communities. The ISME Journal 15, 939-948, doi:10.1038/s41396-020-00832-7 (2021).