The toxin must bind claudin-4 before it can cause any damage at all
For seventeen years, a common gut bacterium has carried the shadow of cancer without surrendering its secret. Now, researchers at Johns Hopkins have traced the precise handshake by which Bacteroides fragilis toxin seizes hold of colon cells — binding first to a receptor called claudin-4 before dismantling the barrier that keeps the gut intact. The discovery, published in Nature, transforms a long-standing biological mystery into a concrete molecular target, opening a corridor toward therapies that might interrupt the chain between microbial presence and malignancy.
- A toxin carried by bacteria in one in five healthy people has been quietly linked to colorectal cancer for nearly two decades, with no one knowing exactly how it struck.
- The missing mechanism — claudin-4, an unexpected surface protein on colon cells — was uncovered only by disabling thousands of genes one by one until the toxin lost its grip.
- Structural biologists in Barcelona confirmed the toxin and receptor lock together in a tight physical embrace, providing the first hard evidence of the molecular crime scene.
- A molecular decoy — a free-floating imitation of claudin-4 — successfully lured the toxin away from real colon cells in animal models, proving the pathway can be blocked.
- Scientists are now racing to translate this into human therapies, with small molecules and biologics in development that could prevent not only colorectal cancer but also diarrheal illness and bloodstream infections tied to the same toxin.
For seventeen years, scientists knew that Bacteroides fragilis — a bacterium living quietly in the guts of roughly one in five healthy people — could trigger colorectal cancer. What they could not explain was how. The microbe secretes a toxin that erodes the colon's protective lining, but the mechanism by which it found its target remained stubbornly hidden.
In April, a team led by Cynthia Sears at Johns Hopkins Kimmel Cancer Center published the answer in Nature. Using a genome-wide CRISPR screen, M.D./Ph.D. candidate Maxwell White and colleagues systematically disabled genes in colon cells until they found the one whose absence left the toxin powerless: claudin-4, a surface protein no one had expected. The toxin, known as BFT, must bind claudin-4 before it can reach E-cadherin, the protein that holds the colon's barrier together. No other known toxin operates through this receptor, making the finding genuinely novel.
To confirm the physical bond, the Johns Hopkins team partnered with structural biologists in Barcelona, who demonstrated that BFT and claudin-4 form a tight, one-to-one complex — the first direct evidence the two molecules truly lock together. The team then built a molecular decoy: a soluble protein displaying claudin-4 sequences that floated freely in the gut, drawing the toxin away from real colon cells. In animal models, it worked.
The precise three-dimensional structure of the BFT–claudin-4 complex remains uncaptured, even with AI modeling tools like AlphaFold — but researchers describe this as a refinement still to come, not a barrier to progress. White and his colleagues are now developing small molecules and biologics to block the interaction in humans, with applications that may extend beyond cancer prevention to diarrheal disease and bloodstream infections driven by the same toxin. A question that once seemed intractable has become a target.
For seventeen years, scientists have known that a bacterium living in the human gut—Bacteroides fragilis, found in roughly one in five healthy people—can trigger colorectal cancer. What they didn't know was how. The microbe secretes a toxin that damages the protective lining of the colon, but the exact mechanism by which that toxin latched onto its target remained opaque, a missing piece in an otherwise compelling puzzle.
In April, researchers at Johns Hopkins Kimmel Cancer Center published the answer in Nature. A team led by Cynthia Sears, a cancer immunotherapy professor at Johns Hopkins, identified the crucial receptor: claudin-4, a protein sitting on the surface of colon cells. The B. fragilis toxin, known as BFT, must first bind to claudin-4 before it can do any damage at all. Without that binding, the toxin cannot reach its ultimate target—a protein called E-cadherin that holds the colon's protective barrier together. The discovery was unexpected. For years, Sears and her colleagues had assumed the receptor would be a signaling protein, the kind that typically receives chemical messages from outside the cell. Claudin-4 is something different entirely, and a literature review revealed that no other known toxin works this way, making the finding genuinely novel.
The path to this discovery began with a technique called a genome-wide CRISPR screen, led by Maxwell White, an M.D./Ph.D. candidate in Sears' lab, working alongside researchers at Harvard Medical School. By systematically disabling genes in colon cells one by one, White and his collaborators could watch which genetic knockouts prevented the toxin from doing its work. When claudin-4 was removed, the toxin simply could not attach. The result was unmistakable—claudin-4 rose to the top as the clear answer.
To prove that the toxin and receptor were actually binding to each other, the Johns Hopkins team partnered with structural biologists in Barcelona. Using biophysical analysis, they demonstrated that BFT and claudin-4 form a tight, one-to-one complex in a test tube—the first direct physical evidence that the two molecules lock together. The work then moved into living systems. Collaborating with researchers at Harvard, the team created a molecular decoy: a soluble protein displaying claudin-4 sequences but floating freely in the gut rather than anchored to colon cells. The decoy worked. The toxin bound to the decoys instead of the real receptors, and mice treated this way were protected from BFT-induced damage.
The implications are substantial. If the toxin can be blocked from binding claudin-4, then the cascade of inflammation and tumor formation might be prevented entirely. White and his colleagues are now exploring which molecular approaches—small molecules or other biologics with better drug-like properties—might successfully interrupt this interaction in humans. One mystery remains unsolved: the precise three-dimensional structure of how BFT and claudin-4 fit together has not yet been captured, even with advanced AI modeling tools like AlphaFold. But that is a refinement, not a roadblock. The fundamental mechanism is now clear, and the path forward is visible. What began as a question about why a common gut bacterium could cause cancer has become a concrete target for intervention—not just against colorectal cancer, but potentially against other infections and inflammatory conditions tied to this same toxin.
Notable Quotes
Understanding how bacterial toxins work can open doors to new approaches for detection and therapy for associated diseases, including diarrhea, colorectal cancer and bloodstream infections.— Cynthia Sears, senior author and Bloomberg~Kimmel Professor of Cancer Immunotherapy at Johns Hopkins
Once we were able to do the screen, claudin-4 was a clear, resounding top hit. That was an exciting moment.— Maxwell White, M.D./Ph.D. candidate in the Sears lab
The Hearth Conversation Another angle on the story
Why did it take seventeen years to figure out how this toxin actually works?
Because the toxin doesn't go straight to its target. It has to bind a receptor first—claudin-4—and that intermediate step was invisible until they had the right tools to look for it. They were looking for the wrong kind of receptor entirely.
So the bacterium itself is common. Most people have it without getting cancer?
Right. It's in about one in five healthy people. The toxin is the dangerous part, and even then, it seems to require other conditions—inflammation, genetics, maybe other factors—to actually trigger tumors. The bacterium alone isn't destiny.
This decoy they created—how does that actually work as a treatment?
Imagine the toxin as a key looking for a lock. The decoy is a fake lock floating around in the gut. The toxin binds to the decoy instead of the real lock on the colon cells, so the real cells stay protected. In mice, it worked. Now they need to figure out how to deliver that decoy—or something like it—in a way that works in humans.
What's the surprise here, scientifically?
That claudin-4 isn't a signaling protein. Everyone expected the receptor to be something that receives messages from outside the cell. Claudin-4 is a structural protein—it's part of the tight junctions that hold cells together. No other known toxin works this way. It's genuinely unusual.
So this could apply to more than just cancer?
Yes. The same toxin is implicated in severe diarrhea and bloodstream infections. If you can block the toxin from binding claudin-4, you potentially prevent all of those diseases, not just colorectal cancer.