One of the deepest frustrations in cancer medicine is collateral damage. Chemotherapy drugs are devastating to tumour cells — but they're often devastating to healthy cells too. Nausea, hair loss, immune suppression, organ damage: the side effects arise because the drug has no way to find the tumour and stay there. It floods the whole body.
A new study published on March 17, 2026 in the open-access journal PLOS Biology takes a creative approach to solving this problem — using living bacteria as guided drug-delivery vehicles.
Bacteria That Hunt Tumours
The researchers, based at Shandong University in China, worked with Escherichia coli Nissle 1917 (EcN) — a well-studied probiotic strain of gut bacteria used safely in humans for over a century. The team engineered this benign bacterium to produce Romidepsin (FK228), an FDA-approved cancer drug, but only in conditions characteristic of the tumour microenvironment: low oxygen, specific pH, and molecular signals associated with cancer cells.
In other words: the bacteria seeks out the tumour, recognises it, moves in, and starts manufacturing the drug on-site.
What the Results Showed
In mouse models of breast cancer, the engineered EcN bacteria successfully colonised tumours following injection. Once inside the tumour microenvironment, the bacteria produced Romidepsin at effective concentrations. Tumour growth was inhibited. Healthy tissue, which lacks the triggering conditions, received significantly lower drug exposure — exactly the targeted delivery researchers have long sought.
"The tumour microenvironment is essentially a safe harbour for these bacteria," explained the research team. "The same conditions that make tumours hard to treat — the oxygen depletion, the unusual chemistry — are precisely what we're using to activate the drug production."
Why This Is Exciting
The choice of EcN is deliberate and important. It has a decades-long clinical safety record in humans, so the translational pathway from mouse models to human trials is more accessible than for a novel agent. It also opens the door to oral administration — unlike antibody-drug conjugates or nanoparticle delivery systems, bacteria can in principle be taken as a pill, colonise the gut, migrate to tumour sites, and begin drug production without intravenous infusion.
The study is preclinical — mouse results do not guarantee human success — but the mechanism, the safety profile, and the use of an already-approved drug make this a compelling direction for oncology research.
Sources: Shandong University / PLOS Biology, March 17, 2026; SciTech Daily; Drug Target Review; ecancer.org
