Whole-genome sequencing maps resistant E. coli across human, farm and environmental sources in Tanzania
Whole-genome sequencing has exposed how deeply Escherichia coli has moved through the connected worlds of hospitals, farms, aquaculture and the environment in Tanzania. In this cohort, 78 isolates were collected from hospitalised patients, poultry, pigs, aquaculture and environmental sources, then sequenced on an Illumina platform to define their lineages, virulence content and resistance profiles. The result is not a narrow snapshot of one outbreak strain. It is a broad genomic inventory of a bacterium that moves easily across hosts and habitats, carrying both pathogenic potential and antibiotic resistance at the same time.
The distribution of isolates already suggests why a One Health approach is unavoidable. Human samples accounted for 21 of 78 isolates, while 26 came from poultry, 13 from pigs, 12 from aquaculture and 6 from the environment. Such spread matters because E. coli is not a single clinical entity; it is a species complex with commensal, opportunistic and pathogenic lineages that can travel between animals and people through food systems, water and contaminated surroundings. In settings where antibiotic use in farming and aquaculture is increasing, the bacterial population is exposed to selective pressure in more than one niche at once.
Genomic diversity reached across lineages, serotypes and pathotypes
Diversity was striking. The sequencing data captured 36 sequence types, including 5 novel ones, and 59 serotypes, including mixed O-types. Every major phylogroup appeared in the collection, with phylogroup A making up 36 of 78 isolates and phylogroup B1 accounting for 16. Those two groups were the most common, but they did not crowd out the broader genetic spread seen across the collection. This is the kind of pattern that makes surveillance difficult and urgent at the same time: a single species, yet many genomic routes into disease and resistance.
Pathogenic lineages were present in both human and animal sources. Ten isolates belonged to three pathotypes associated with human disease: DAEC, EAEC and EAEC/ETEC. They were found in humans and pigs, a reminder that clinically relevant strains do not respect the boundary between the hospital and the farm. The DAEC pathotype carried the broadest repertoire of genotypic virulence factors, with 23 virulence factors identified. Human isolates carried the heaviest virulence burden overall, with a median of 11 factors compared with 5 in other sources, a difference that was statistically significant. In practical terms, human isolates were not simply more likely to be recovered from symptomatic infection; they were also genomically loaded with traits that can support colonisation, persistence and tissue injury.
Virulence and resistance converged in the same collection.
Resistance genes were nearly universal and extended across 13 antibiotic classes
The resistance profile was more concerning than any single lineage result. Efflux pump-mediated resistance genes were almost universal, with acrF present in 76 of 78 isolates and mdtM in 73. Beyond those transport-related mechanisms, the isolates carried genotypic resistance to 13 classes of antibiotics. Beta-lactams, quinolones and aminoglycosides were among the commonly affected classes, placing several of the most used antimicrobial groups under direct genomic threat.
Such a pattern fits the biology of enteric bacteria under sustained antimicrobial selection. Efflux systems reduce intracellular antibiotic concentrations, allowing cells to survive exposures that might otherwise suppress growth. When those systems coexist with other resistance determinants, the organism gains a layered defense that can blunt treatment response and preserve survival across different environments. The study also found broad multi-drug resistance by genotype in 71 of 78 isolates and by phenotype in 67 of 78. Those proportions leave little room for reassurance. Multidrug resistance was the rule, not the exception, in this collection.
Phenotype and genotype aligned unevenly across drug classes
Concordance between genetic predictions and laboratory susceptibility testing varied substantially by antibiotic class, ranging from 37.2% for aminoglycosides to 89.7% for phenicols. That spread shows why genome data cannot be treated as a simple substitute for phenotypic testing, even when sequencing offers fast and detailed insight into microbial potential. Different drug classes depend on different resistance mechanisms, and the presence of a gene does not always translate into a measured resistant phenotype under the conditions used in the laboratory. The mismatch was most pronounced for aminoglycosides, where genome-based calls captured only part of the resistance picture.
Even with that variability, both methods pointed in the same direction: resistance was widespread and multiple drug classes were affected. This is the clinical consequence that matters at the bedside and in public health laboratories. When a bacterium circulating among people, livestock and environmental reservoirs repeatedly carries resistance determinants, empiric therapy becomes less reliable and routine treatment pathways become more fragile. In low-resource and mixed-exposure settings, that can push clinicians toward broader agents sooner, which in turn can increase selective pressure further.
One Health surveillance needs genomic tools, not just isolated culture results
The central lesson of this work is the value of whole-genome sequencing for One Health surveillance. Culture and susceptibility testing remain essential, but they do not reveal the full structure of spread, lineage mixing or the hidden distribution of virulence and resistance genes across human and animal populations. Sequencing linked isolates from hospitalised patients, poultry, pigs, aquaculture and the environment into one shared genomic framework. That approach is particularly important in Tanzania, where rising demand for fish and meat intersects with increased antibiotic use in farming and aquaculture.
Genomics also helps explain why control efforts that focus on only one sector often fall short. If resistant and potentially pathogenic lineages are moving between humans, livestock, fish systems and environmental reservoirs, then interventions confined to a single setting will leave major transmission pathways untouched. Surveillance has to track the organism where it lives, not just where it causes disease.
This study did not describe clinical outcomes for infected patients, and it was not designed to test whether specific strains were causing particular syndromes at the individual level. Its strength lies elsewhere: it shows that Tanzania’s E. coli population is genetically diverse, virulence-bearing and heavily resistant across connected ecological compartments. The remaining challenge is operational rather than conceptual: turning genomic surveillance into routine public health action before resistant strains become even harder to contain.