Scientists Turned a Probiotic You Can Buy Into a Colon Cancer Hunter

If you've ever looked into probiotics hoping they might do something meaningful for a gut health problem, you've probably learned to be skeptical of the marketing. A new study out of Hunan University complicates that skepticism in an interesting way. Researchers took a probiotic strain you can already buy over the counter and re-engineered it into a drug-delivery vehicle that hunts down colon tumors and disarms them from the inside, using a protein the body normally makes less of as tumors grow. The catch, and it's an important one, is that this happened in mice, patient tissue grown in dishes, and tumors implanted in mice, not in a single human patient taking the bacteria itself.

Key takeaways

• Researchers engineered E. coli Nissle 1917, the probiotic strain sold as Mutaflor, to secrete DKK3, a protein that blocks the tumor-driving Wnt signaling pathway, directly inside colon tumors.
• Given orally twice a week, the engineered bacteria (called EcD) significantly reduced tumor number and area and extended survival in mice, with the same effect replicated across three separate tumor models, including two that mimic inflammatory bowel disease (IBD)-associated colon cancer.
• A second engineered strain detects a stool biomarker called ornithine to flag tumor-bearing mice and colorectal cancer patients, distinct from IBD or normal aging, without a colonoscopy.
• Everything here is preclinical: no human has been dosed with the engineered bacteria, and the study's strongest human-tumor result used a different, injectable bacterial strain rather than the oral probiotic that makes the headline.

Engineered E. coli Nissle 1917 Delivers a Wnt-Blocking Protein Straight Into Colon Tumors

Researchers engineered E. coli Nissle 1917 (EcN), the probiotic bacterium marketed as Mutaflor, to manufacture and secrete a human protein called DKK3 (Dickkopf-3, a natural brake on a cell-growth pathway called Wnt/β-catenin signaling) directly at tumor sites in the gut. In colorectal cancer, Wnt signaling frequently gets stuck switched on, pushing intestinal stem cells to divide out of control. DKK3 is one of the body's own checks on that pathway, and the researchers confirmed it drops off as cancer progresses: using a public tumor-gene-expression database, they found DKK3 significantly reduced in two major colorectal cancer subtypes, and in two independent public patient cohorts, low DKK3 tracked with meaningfully worse survival, with hazard ratios of 0.29 and 0.30 for high versus low expression. To put more DKK3 exactly where it's needed instead of relying on the body to keep making it, the team fused the human DKK3 gene to a bacterial export signal and loaded it onto a plasmid (a small ring of DNA bacteria can carry and express) inside EcN, creating an engineered strain they call EcD.

Colon Cancer and IBD Patients Are the Groups This Research Speaks to Most Directly

Colorectal cancer is one of the most common cancers diagnosed worldwide, and in the United States it now ranks as the leading cause of cancer death in adults under 50, a group where incidence has been climbing even as it falls in older adults. People living with inflammatory bowel disease, Crohn's disease or ulcerative colitis, carry an added, well-documented risk of developing colon cancer on top of their existing disease, which is part of why this study didn't stop at a standard tumor model. The researchers specifically tested EcD in mice with colitis-driven colon cancer, including a model combining a carcinogen with an IBD-mimicking gene knockout that the authors describe as closely mirroring clinical IBD-associated colon cancer. In that model, mice given the untreated parent bacteria developed more severe disease, including rectal prolapse, while EcD-treated mice had significantly fewer and smaller colon tumors.

Five Separate Models, From Mice to Human Tumor Tissue, Point the Same Direction

The core efficacy data comes from Apc-Min mice, a strain that spontaneously develops intestinal tumors, dosed orally with EcD or the unmodified parent bacteria twice a week from week 9 to week 20 of age. Tumor number and total tumor area were significantly lower in the EcD group (10 mice per group), and survival was significantly extended (12 EcN-treated mice versus 13 EcD-treated mice). The effect held in both male and female mice, and held even when treatment was delayed until week 12, after adenomas had already become established, meaning the bacteria worked therapeutically on existing tumors, not just as prevention. The same pattern, significantly fewer and smaller tumors, repeated in a chemically induced colitis-cancer model (6 versus 5 mice) and the IBD-mimicking model (4 mice per group). For human tissue, the team grew tumor organoids (miniature 3D tumor replicas grown from a patient's own cells) from colorectal cancer patients and exposed them to EcD's secretions, which significantly slowed organoid growth; a related engineered strain built on an attenuated Salmonella backbone, delivered intravenously rather than orally, also slowed growth in a patient-derived tumor implanted in mice. Similar EcN engineering for colorectal detection and treatment has been demonstrated by other groups, lending independent support to the broader strategy of using this probiotic chassis against colon tumors. Throughout, mechanistic work traced the effect back to DKK3 physically binding a Wnt co-receptor complex inside tumor cells and shutting down downstream genes tied to cell proliferation, and long-term dosing showed no bacterial spread beyond the gut and no changes in blood counts, organ tissue, or liver and kidney markers in healthy mice.

What This Could Mean if You're Watching Your Own Gut Health

If you take Mutaflor today, this study doesn't apply to what's in the bottle. The commercial probiotic is the unmodified parent strain; every anti-tumor effect described here came from EcD, a genetically engineered version that manufactures and exports DKK3, a modification that does not exist in any product sold to consumers. No person has been dosed with the engineered bacteria described in this paper. What the study does offer is a proof of concept: that a probiotic already known to be safe in humans can be turned into a targeted delivery vehicle for a cancer-relevant protein, tested here across ordinary tumors, colitis-associated tumors, and an IBD-mimicking model without added toxicity in mice. If you have IBD or a family history that raises your colon cancer risk, this isn't something to ask a doctor for, since it doesn't exist as a treatment yet, but standard colonoscopy screening remains the actionable step worth discussing given rising early-onset colorectal cancer rates.

What's Not Yet Known, and Where the Headline Could Get Ahead of the Data

This entire body of work is preclinical: mice, cells, patient-derived organoids, and one implanted patient tumor, with no clinical trial of the bacteria in people. The paper reports its tumor reductions as statistical significance (p-values) and bar charts rather than quoting a specific percentage shrinkage, so any claim of "X% fewer tumors" would be an overstatement of what's actually shown. Group sizes were small throughout, from 4 to 13 animals per group depending on the model, typical for this stage of research but not enough to rule out variability. It's also worth being precise about which bacteria did what: the tumor-reduction results in the patient-derived tumor model and a separate mouse cancer model used the intravenous Salmonella-based strain, not the oral EcN probiotic that anchors the headline, so the strongest human-tumor evidence didn't test the same delivery route being promoted as the accessible, oral option. And the diagnostic stool-sensing strain is a completely separate bacterium from the therapeutic one; detecting a tumor with it wouldn't deliver any treatment on its own.

From Engineered Mice to a Clinical Pipeline: What Would Have to Happen Next

The researchers frame this as a "theranostic" platform, pairing a non-invasive stool-based detector with a targeted oral treatment, and note in their own limitations section that other members of the DKK protein family haven't been tested and that DKK3's effects on the immune system, separate from its direct anti-proliferative action, still need investigation. No clinical trial timeline is given in the paper. The study was funded entirely through Chinese public science funding sources, and the authors report no competing interests and no patents tied to the work, which is a meaningfully cleaner starting position than many bacterial-therapy stories that arrive already wrapped around a company. That also means the next steps, safety testing in larger animals and eventually a first human dosing study, aren't yet backed by a company timeline either.

Common Questions

Can I get any of these effects by taking Mutaflor? No. The commercial Mutaflor product is the unmodified parent bacterium. The anti-tumor effects in this study came exclusively from EcD, a lab-engineered version that manufactures and secretes DKK3, a feature the store-bought strain doesn't have.

Has the engineered bacteria been tested in any human? No. The anti-cancer results in human tissue came from patient-derived tumor organoids grown in dishes and one tumor type implanted in mice (a PDX model), not from a person being dosed with the bacteria.

How would the stool-based detection actually work? A separate engineered strain, mixed briefly with a fresh stool sample, produces a detectable signal when it senses elevated ornithine, an amino acid the study identified as a marker that rose specifically with tumor development and not with IBD or normal aging in early testing on 12 colorectal cancer patients versus 12 healthy controls.

Does this only work for one type of colorectal cancer? No, the therapeutic strain reduced tumors across three different mouse models, spontaneous intestinal tumors, chemically induced colon cancer, and an IBD-mimicking model, plus patient-derived human tumor tissue, suggesting the approach isn't limited to a single subtype, though all of it remains preclinical.

References

1. Yujie Sun, Yu Zhao, Kang Han, et al. (2026). Synthetically engineered probiotic suppresses colorectal cancer via inhibition of Wnt/β-catenin signaling. Cell Reports Medicine. https://doi.org/10.1016/j.xcrm.2026.102913
2. Engineering tumor-colonizing E. coli Nissle 1917 for detection and treatment of colorectal neoplasia. Nature. https://www.nature.com/articles/s41467-024-44776-4