CRISPR system reprograms bacteria to disarm deadly toxin

Antibiotics usually save lives—but against some bacteria, they can make things worse. That’s the case with the Shiga toxin–producing Escherichia coli, where bacterial death releases a flood of a cell-killing toxin. Now scientists have developed a new gene-editing approach that circumvents this problem. Instead of killing the pathogens, it reprograms them to stop producing the toxin and instead make a helpful molecule (Nat. Biomed. Eng. 2025, DOI: 10.1038/s41551-025-01453-1).

Developed by researchers led by Harris Wang at Columbia University, the system, named Bacterial CRISPR–Transposase Reduction of Virulence In Situ (BACTRINS), operates by using a combination of CRISPR-Cas and transposases to neutralize pathogenic E. coli. The Shiga toxin is produced from two genes, stx1 and stx2, and BACTRINS uses guide RNAs to direct Cas proteins to regions in the genes that are less subject to mutations over time. There, the Cas complex recruits a transposase, which inserts new DNA that disrupts the gene, all without cutting the genome.

“We’re essentially converting a pathogenic strain into a nonpathogenic one,” Wang says.

This method prevents toxins from spilling into the gut. But by avoiding double-stranded DNA breaks, it also sidesteps the bacterial repair responses that often introduce mutations at the target site and lead to CRISPR resistance and ineffective treatments.

To make the approach even more potent, the researchers also programmed the inserted DNA to produce nanobodies—tiny antibody-like proteins that are secreted by the engineered bacteria. These nanobodies bind to Tir, a surface protein on pathogenic E. coli that facilitates the bacteria’s attachment to gut cells. Without functional Tir, the bacteria struggle to colonize the gut and establish infection.

To deliver the treatment, the system is first introduced into a harmless bacterium. Once ingested, the engineered microbe acts as a delivery vehicle and transfers the system to harmful bacteria through bacterial conjugation—a natural form of gene sharing between microbes.

The researchers report that in studies of mice that were infected with a Shiga toxin–producing E. coli strain, all untreated mice died. But in mice that received the new treatment, toxin levels dropped by two-thirds and survival rates improved. Although many of the treated mice died, Wang says the bacteria strain used in the study is more virulent in mice than in humans, and that the reduced toxin levels suggest that the strategy could be effective in a clinical setting.

“I think this is a really great study, and has great potential,” says Byeonghwa Jeon, an expert in bacterial pathogenesis at the University of Minnesota who was not involved in the study. But there is a need for further research before BACTRINS can be introduced into a clinical setting, he adds.

Because CRISPR-Cas machinery is widespread in nature, Wang thinks a similar BACTRINS platform could be applied to different bacteria in other areas of the body, or even to soil microbes. The team is now looking into that line of investigation. The researchers also plan to explore ways to use the system in commensal microbes to boost their beneficial effects. “This adds to the genetic toolbox of potential strategies and is hopefully widely applicable to a lot of different situations,” Wang says.