The global rise of multidrug-resistant (MDR) *Escherichia coli* (*E. coli*) poses a significant threat to human, animal, and environmental health. While hospitals are recognized hotspots for the emergence and spread of MDR *E. coli*, aquatic ecosystems, impacted by pollution, are increasingly recognized as potential reservoirs. The Baltic Sea, a large brackish sea with slow water exchange and low salinity, is particularly vulnerable to pollutant accumulation, including MDR bacteria and antibiotic residues. Untreated sewage, agricultural runoff containing antibiotics from livestock farming, and wastewater treatment plant discharges all contribute to this pollution. Furthermore, ballast water discharge from ships introduces MDR bacteria and AMR genes from other regions. This study investigates the presence and characteristics of extended-spectrum β-lactamase (ESBL)-producing *E. coli* in the Baltic Sea, considering the interplay between antibiotic selection pressures and the spread of high-risk clonal lineages from a One Health perspective. High-risk clonal *E. coli* lineages, such as ST38, ST58, ST117, ST131, ST410, ST648, ST744, and ST1193, are epidemiologically successful and pose significant health risks due to their multiple AMR and virulence determinants. The study aims to address the gap in understanding the role of environmental reservoirs like the Baltic Sea in the dissemination of MDR *E. coli*.

Existing literature highlights the global burden of antimicrobial resistance (AMR) and the challenges it poses to effective antibiotic therapy. Studies have shown the importance of water bodies as potential reservoirs for MDR bacteria, with the Baltic Sea identified as a particularly vulnerable ecosystem due to its semi-enclosed nature and slow water exchange. Research on antibiotic residues and AMR in the Baltic Sea is limited, with a few studies showing significant concentrations of various antibiotics in sediments and near-bottom waters, primarily in the Pomeranian Bay. While much research focuses on AMR in clinical settings, the role of environmental reservoirs in disseminating MDR *E. coli* remains under-explored. The literature also emphasizes the significance of high-risk clonal lineages of *E. coli* in driving AMR, their global distribution across diverse ecosystems and hosts, and their persistence even in the absence of substantial antimicrobial selection pressures.

Water samples were collected longitudinally from three sites in the Baltic Sea (northeastern Germany) over a year (August 2021-August 2022). Samples were collected weekly from an urban site (Eldena), a peri-urban site near a wastewater treatment plant (WWTP; Ladebow), and a rural site (Riems). Samples were also taken from the WWTP and hospital wastewater for comparison. ESBL-producing *E. coli* were isolated using membrane filtration and CHROMID ESBL and CHROMagar Orientation agar plates. Isolated strains underwent phenotypic and genotypic analyses, including: phenotypic antibiotic resistance and heavy metal/metalloid tolerance testing; biofilm formation assays; and in vivo mortality testing in *Galleria mellonella* larvae. Whole-genome sequencing (Illumina NovaSeq 6000 and ONT Nanopore) and hybrid assembly were performed to characterize genomic properties, AMR determinants, virulence-associated genes (VAGs), phylogenetic relationships, and extra-chromosomal elements. Antibiotic residues (ampicillin, cefotaxime, ciprofloxacin, sulfamethoxazole, and tetracycline) were quantified in water samples positive for ESBL-producing *E. coli* using solid-phase extraction with ultra-high-performance liquid chromatography and mass spectrometry (UHPLC-MS/MS). Phylogenetic analysis used the Clermont method for phylogrouping and the Achtman scheme for multi-locus sequence typing (MLST). Mash distance analysis was used to investigate genomic and plasmid relationships. BRIG was used to visualize plasmid comparisons. Minimum inhibitory concentrations (MICs) for cefiderocol were determined using a broth microdilution kit.

A total of 30 ESBL-producing *E. coli* were isolated, predominantly from Ladebow (60%) and Eldena (36.7%). Genomic analysis revealed diverse phylogroups (A, B1, B2, D, G, F, C), with 20 different STs identified. Several strains belonged to international high-risk clonal lineages, including ST38, ST58, ST117, ST131, ST410, ST744, and ST1193. Mash distance analysis showed a close phylogenetic relationship among three ST117 strains isolated over five months. Comparison with publicly available genomes showed close relationships with clinical strains from various geographical locations, suggesting potential clinical relevance. Plasmid analysis revealed high diversity, with a few clusters of closely related plasmids. ST117 strains had a high plasmid load (10-11 plasmids) with similar sequences containing ESBL and virulence factors. Genomic AMR analysis showed resistance determinants for multiple antibiotic classes (β-lactams, aminoglycosides, diaminopyrimidines, macrolides, phenicols, quinolones, sulfonamides, and tetracyclines). *bla*CTX-M-15 was the predominant ESBL gene. One strain (PBIO3948, ST117) exhibited resistance to cefiderocol. Phenotypic antimicrobial susceptibility testing confirmed the ESBL-producing phenotype and MDR in over half of the strains. Most strains showed tolerance to arsenic, and some showed tolerance to mercury, with discrepancies between genotype and phenotype observed for some mercury-tolerant strains. All strains carried VAGs associated with adherence, biofilm formation, capsule production, iron uptake, invasion, and toxin production. Biofilm formation capacity varied significantly, even within the same ST. The *G. mellonella* infection model showed larval mortality varied regardless of the number of VAGs present. Antibiotic residue analysis revealed variations in antibiotic concentrations, with higher levels in Ladebow (near a WWTP) compared to Eldena and Riems. Wastewater samples showed significantly higher antibiotic concentrations than environmental samples.

The findings demonstrate the presence of ESBL-producing *E. coli*, including high-risk clonal lineages, in Baltic Sea surface water, alongside low concentrations of antibiotic residues. The presence of MDR *E. coli*, including cefiderocol resistance, highlights the potential public health risk. Heavy metal tolerance in many strains suggests co-selection of AMR, potentially through mechanisms like efflux pumps. The high prevalence of virulence factors further underscores the clinical relevance of these strains. Discrepancies between phenotypic and genotypic results for some strains highlight the need for further investigation into the mechanisms of resistance and tolerance. The variable biofilm-forming abilities among strains indicate diverse survival strategies. The observed antibiotic concentrations, while low compared to wastewater, still pose a potential selective pressure contributing to AMR development. Further research should investigate the correlation between antibiotic residues and resistance gene acquisition, focusing on mechanisms beyond direct selection, such as co-selection through heavy metals and the stability of resistance plasmids under non-selective conditions.

This study provides evidence of ESBL-producing, multidrug-resistant *E. coli*, including high-risk clonal lineages, in the Baltic Sea. The occurrence of these bacteria, along with low levels of antibiotic residues, suggests a potential link between environmental contamination and the dissemination of AMR. The findings highlight the need for enhanced monitoring and regulation of antibiotic discharges into aquatic environments to mitigate the risks associated with AMR. Future research should focus on the specific mechanisms driving AMR and heavy metal tolerance in these *E. coli* strains, and on exploring the long-term ecological consequences of antibiotic pollution in the Baltic Sea.

The study is limited by the relatively small number of water samples analyzed. The observed antibiotic concentrations may not represent the entire Baltic Sea, and further sampling is needed for a more comprehensive assessment. The relationship between detected antibiotic residues and the specific resistance genes identified in the *E. coli* strains was not definitively established. The in vivo virulence analysis used a limited number of strains, potentially biasing results. Further research is needed to investigate the observed discrepancies between genotype and phenotype for some strains, and to elucidate the complete mechanisms of resistance, particularly the cefiderocol resistance in ST117 strain PBIO3948.