Frog gut bacterium clears colon tumors in mice with a single injection

Bacteria-based cancer therapy is attractive because certain microbes home in on tumors, but most candidates are genetically engineered strains with safety risks, including mutations that could undo built-in safeguards. This study asks whether naturally occurring, unmodified gut bacteria from unusual hosts could offer potent antitumor activity without engineering. It also builds on the authors' earlier work with tumor-resident bacteria.

The team isolated 45 bacterial strains from the intestines of the Japanese tree frog (Dryophytes japonicus), the Japanese fire belly newt (Cynops pyrrhogaster) and the Japanese grass lizard (Takydromus tachydromoides). Each was given intravenously to BALB/c mice (200 µL at 5 × 10⁹ CFU/mL) as a tolerability screen. Nine acceptable strains, confirmed by 16S rRNA sequencing, moved on to efficacy testing in immunocompetent mice with Colon-26 tumors of about 200 mm³, each receiving a single tail-vein injection and followed for 40 days (n = 3 per group).

Of the nine strains, one (Priestia aryabhattai) had no effect, five significantly slowed tumor growth, and three (E. americana, Citrobacter portucalensis and Enterobacter ludwigii) produced active regression. All three regressors were facultative anaerobes, consistent with preferential colonization of hypoxic tumors. Regression with C. portucalensis and E. ludwigii appeared by day 3 but tumors recurred after day 8, and Rhodococcus qingshengii's early suppression faded after day 5. Only E. americana gave durable results, with 100% survival and a 100% complete response rate. A pilot dose range of 2 × 10⁷ to 1 × 10⁹ CFU identified 1 × 10⁹ CFU as best.

In a head-to-head comparison (n = 5 per group), one E. americana injection was compared with anti-PD-L1 antibody or liposomal doxorubicin, each given at 2.5 mg/kg four times every other day. Both drugs slowed tumor growth, but anti-PD-L1 produced only one complete response and neither drug eradicated tumors consistently. E. americana cleared tumors in all animals (p < 0.0001) and extended survival by at least 30 days versus PBS. In rechallenge experiments, 0 of 10 cured mice grew new tumors versus 10 of 10 naive controls, with immune memory lasting beyond 60 days.

The abstract describes a dual mechanism of direct tumor-cell killing plus activation of T cells, neutrophils and B cells. It also reports minimal pathogenicity at effective doses. In the paper's data, treated mice showed only a slight early weight dip that resolved, with no group differences by day 15. E. americana is generally antibiotic-susceptible, so toxicity could in principle be managed, though it can cause opportunistic infections in neonates and immunocompromised patients.

Caveats are substantial. All data come from one syngeneic mouse colon cancer model with very small groups, and the model's tumor biology differs from human cancer. Intravenous bacteria carry sepsis risk, and long-term toxicity, dosing in larger animals and effects in other tumor types remain untested. Clinical translation is a long way off.