Diversity of antibiotic resistance genes increases in urbanized lakes: a multi-tool screening

Antimicrobial resistance (AMR) is a growing global public health threat projected to cause up to 10 million deaths annually by 2050 if no immediate action is taken. While misuse and overuse of antibiotics are the main drivers of increasing AMR, the eco-evolutionary dynamics of AMR in the environment - particularly across the urban-rural continuum - remain poorly understood. Using shotgun sequencing, we investigated urban, farm, and rural water sources in the Berlin-Brandenburg region to explore the distinctness or overlap of their antibiotic resistance gene (ARG) profiles and the potential impact of wastewater treatment plants (WWTP). ARGs were identified using multiple databases and five bioinformatic tools, combining sequence-based alignment and deep learning approaches. This multi-tool approach allowed for the detection of up to 18 AMR classes--more than any single tool alone. The multi-tool screening approach for ARGs, combined with the ABRicate algorithm, was superior to all single ARG tools and databases, detecting more AMR classes, allowing for biocide and metal resistance detection, while less sensitive for detection of aminocoumarin resistance genes. ARG diversity was higher in urban lake sediments, urban waters, and wastewater compared to rural lake sediments and water. Among all environments, urban lake water showed the highest overall ARG abundance, second only to wastewater, and this pattern held across all AMR classes, except for aminoglycoside resistance, which was most prevalent in rural lake sediments. The WWTP was unable to remove the circulating pool of ARGs, despite a decrease in unique ARGs in the outflow.