An Escherichia coli isolate from a patient in Denmark and another from river water in the Czech Republic were within 20 single-nucleotide differences in the study’s genomic comparison. The pair was among the closest cross-source matches in a global study of an E. coli lineage associated with urinary tract and bloodstream infections.1 The researchers could quantify how closely the two isolates were related. The harder question was what, if anything, connected the person and the river.
That distinction matters for One Health surveillance, which looks for connections among people, animals, and the environment. Investigators tracking antimicrobial resistance (AMR) might want to know where a bacterial lineage circulates, how a resistance gene appears in different strains, or whether a particular exposure carried bacteria to a new host. Those questions can start with the same set of genomes, but they require quite different evidence to answer.
What a Global Lineage Can Reveal
Sukkar and colleagues examined 6,907 genomes of E. coli sequence type 69 (ST69), combining 160 isolates from the Czech Republic with 6,747 publicly available genomes.1 In the assembled collection, 26.2% carried genes encoding extended-spectrum β-lactamases (ESBLs), with blaCTX-M-15 the most common. That figure describes the genomes the researchers assembled from different studies and sources, but estimating prevalence in any particular host population would require sampling designed for that purpose.2
A sequence type groups bacteria by a set of genetic markers. Within that group, comparisons across more of the genome can distinguish near matches from distant relatives. In the Czech collection alone, pairs of ST69 isolates ranged from zero to 10,394 single-nucleotide polymorphism (SNP) differences. For their global network analysis, the researchers used a threshold of 100 SNPs to identify clusters and examined pairs with much smaller genetic distances. The largest European cluster contained 131 genomes from humans, wastewater, and wildlife. Its membership described the range of sources in which related ST69 bacteria had been recovered.
The 100-SNP threshold determined which genetic links the researchers counted in the network. The accompanying records determined which isolates could contribute to the analysis in the first place. Source information was missing for 1,672 genomes, and geographic information was missing for 1,212. The researchers used the 4,932 genomes with both fields available and analyzed them in four geographic subsets. The African subset yielded no cross-source pairs as close as some of those found elsewhere, but it also contained the fewest genomes. With fewer genomes available, there were fewer opportunities to find a close match. A mixed-source cluster can identify where to look more closely. Its size and composition also reflect which sources were sampled and which genomes had enough information to be included.
What Can Move Across Sectors
An E. coli cell can carry resistance with it when the bacterium moves between hosts or environments. Resistance can also move between bacteria on a plasmid, a DNA molecule separate from the chromosome. Within a plasmid, a resistance gene may sit on a transposon, a segment of DNA that can move between genetic locations. That means a gene and the DNA immediately surrounding it can recur in bacteria whose chromosomes and plasmids differ. A comparison of bacterial genomes, a comparison of complete plasmids, and a comparison of the DNA around a resistance gene each track a different unit.
Jiang and colleagues followed the DNA surrounding blaCTX-M in a Hong Kong collection of 445 CTX-M-producing E. coli isolates.3 The human samples yielded 48 bloodstream isolates. Another 397 isolates came from raw meat, ready-to-eat food, animal feces, and food-processing surfaces. Most isolates grouped by sector, and close strain sharing between human and nonhuman sources was rare. However, the researchers found related transposons carrying blaCTX-M across sectors. Their genomic comparisons pointed to transposons as a more frequent explanation for the observed gene sharing than the movement of intact plasmids.
That inference depends on being able to see where a gene sits. Short sequencing reads can leave a resistance gene and parts of its surrounding DNA on separate fragments. The researchers used sequence alignments to identify the elements it could resolve and acknowledged that the method could miss unfamiliar mobile elements. The repeated arrangements supported an explanation for how the gene had spread within the sampled collection.
Xu and colleagues added a functional test. Working with 1,016 E. coli isolates from human-associated, animal-associated, and environmental aquatic samples in Hong Kong, they identified 142 clonal sharing pairs linking human-associated samples with environmental waters.4 They also identified 195 plasmids in groups represented across all three sectors. Some closely matching plasmids occurred in bacteria that were genetically distinct. The team then brought selected donor and recipient isolates together in conjugation assays and demonstrated that representative resistance plasmids could transfer between bacteria recovered from different sectors. The assays established what those plasmids could do under laboratory conditions. The field samples established where the bacteria carrying them had been recovered.
What the Two Hong Kong Studies Show
The two studies used different windows into the human population. Jiang’s human isolates came from people with bloodstream infections, and the comparison focused chiefly on meat and animal feces.3 Xu used aquatic sampling, including wastewater as a proxy for human-associated bacteria.4 A match between wastewater and environmental water shows overlap between the bacteria recovered at those sampling points. Its unit of observation is a sample from a water system, so it cannot connect the match to a particular patient. Conversely, limited overlap with bloodstream isolates leaves open what researchers might find in wastewater or among people carrying E. coli without a bloodstream infection.
One resistance gene makes the difference tangible. In Xu’s collection, blaCTX-M-55 featured in 135 of the 142 clonal sharing pairs between human-associated samples and environmental waters. There, the gene appeared in comparisons where the bacteria themselves were closely related. Jiang found the same variant across sectors and frequently associated with two transposon types, even though closely related strains rarely linked the human and nonhuman samples. In that collection, looking at the DNA around the gene revealed connections that a comparison of bacterial strains alone would have missed. Establishing how often either process leads to human infection would require sampling that connects those genetic findings to the people and exposures in question.
What Linked Sampling Adds
Musicha and colleagues built those kinds of connections into a prospective household study in Malawi and Uganda.5 They enrolled 259 households in Malawi and 92 in Uganda and returned for follow-up visits. They sampled human and animal stool, food, water, and surfaces that people touch. When two isolates were closely related, the investigators could compare their collection dates and locations and determine whether they came from the same household. That context gave a genomic match a more specific epidemiological setting than a source label in a global collection could provide.
Using a five-SNP threshold, the researchers inferred 463 links between human and environmental isolates, 146 between human and animal isolates, and 142 between animal and environmental isolates. The threshold served a different purpose from the 100-SNP rule in the ST69 study. Musicha’s team used it to identify close links in a prospective cohort, while Sukkar’s team used its threshold to build networks within a global lineage. The counts from the household study describe the links the researchers inferred among sampled isolates. Comparing the frequency of transmission between compartments would also require accounting for how each was sampled. In Uganda, for instance, the study recovered only three environmental genomes, limiting the comparisons available there.
Repeated visits also leave gaps. If investigators recover a strain from water on one visit and from a household member on the next, the collection dates establish the order of detection. To infer that the person acquired the strain from the water, investigators would need to know whether the person used that water during the relevant interval, what earlier samples showed, and whether another exposure could explain both detections.
What It Would Take to Trace a Route
The World Health Organization and its three One Health partners, the Food and Agriculture Organization of the United Nations, the United Nations Environment Programme, and the World Organisation for Animal Health, recommend defining what a surveillance program needs to find out before deciding what to sample, which records to collect, and which laboratory methods to use.2 That objective shapes what investigators sample, which records they collect, and what laboratory methods they use. For the Danish human isolate and the Czech river-water isolate, the original question was where closely related ST69 bacteria appeared. A route investigation would start with a narrower question. Did the person have contact with the river water, or could a common source have exposed the person and seeded the river?
To pursue that question, investigators would need the collection dates, the precise river sampling location, and any relevant exposure history for the person. If those records pointed to a specific connection, investigators could sample along it and check other potential sources for the strain. Finding it at several sites with no evident connection to the person would give the original pair a different interpretation from recovering it repeatedly along a route the person could have encountered. The genomic distance would be the same in both cases. The exposure records and the pattern of additional detections would determine how much weight the proposed route could bear.
If the question concerned resistance specifically, investigators would also need to establish what the isolates shared beyond their chromosomal similarity. A claim about a shared resistance plasmid would require enough sequence information to compare the plasmids themselves and determine where the resistance gene sits.
From a Suspected Route to an Intervention
In Nairobi, Kim and colleagues examined a possible exposure close to home, the water people stored for drinking.6 The researchers collected 321 samples from 50 households and compared E. coli from people, poultry, dogs, soil, and stored drinking water. Within households, people shared strains with stored water more often than they shared strains with animals. Strain sharing among household members was also more frequent in households whose stored water contained E. coli. In a separate analysis, sample types with more strain sharing within households tended to have more similar resistance gene profiles. That association suggests that the spread of whole strains could help explain why resistance genes overlapped across household sources.
The study also compared two communities, one with access to chlorinated water and one without. Human–water strain sharing was lower in the community with access to chlorinated water, but the difference was not statistically significant. The comparison was observational, and access to chlorinated water was a characteristic of the community rather than an intervention assigned by the study. The study therefore left open whether chlorination reduced strain sharing. The comparison measured strain sharing itself, so any effect on resistance transmission would need to be evaluated separately.
The next question is whether stored water carried E. coli to household members, whether people or their surroundings contaminated the water, or whether the water and the people acquired the strain from a shared source. Those possibilities put the point of intervention in different places. Sampling water when households collect it and again after storage, alongside repeated sampling of household members, could help establish when the strain first appears in each. Evaluating a treatment would then require measuring whether it changes the outcome along the route under investigation. The Nairobi findings give investigators a specific place to start with the water between collection and consumption, where the household’s exposure can be examined directly.