Crohn’s disease is a form of inflammatory bowel disease that can inflame any part of the digestive tract, most often the small intestine and colon. It can cause diarrhea, abdominal pain, weight loss, and deep fatigue, sometimes for years at a stretch.
In the United States, inflammatory bowel diseases affect an estimated 2.4 to 3.1 million adults, roughly a little over one percent of the population. Many of them hear the same thing from their doctors. Genes, environment, microbes, all mixed together. Important, but vague.
Back in 2001, scientists identified a gene called NOD2 that clearly raises the risk of Crohn’s. Yet no one could agree on what the faulty gene was actually doing to the gut. That uncertainty has lingered in exam rooms and research labs for more than two decades.
AI maps the gut’s “peacekeepers”
The new work starts with two kinds of immune cells that patrol the intestinal wall. These cells, called macrophages, are specialized white blood cells.
One group attacks invading microbes. The other repairs tissue and helps calm inflammation once the danger has passed. When that balance is off, the intestine can stay inflamed long after any threat is gone.
To understand what pushes those cells toward attack or repair, the team used machine learning to sift through thousands of gene activity profiles from macrophages in colon samples taken from patients with inflammatory bowel disease and from healthy volunteers.
The algorithms flagged a “signature” of 53 genes that reliably separated inflammatory macrophages from tissue-healing ones.
That is the sort of pattern no human could spot by eye. But once the software highlighted it, biologists could begin to test what those genes were doing in real cells.
NOD2, girdin, and a broken partnership
Among the 53 genes, one stood out. It encodes a protein called girdin, also known as GIV. Further experiments showed that in non-inflammatory, tissue repairing macrophages, a specific region of the NOD2 protein physically binds to girdin.
That partnership helps keep inflammation under control, clears harmful microbes, and lets damaged tissue heal.
Senior author Pradipta Ghosh explained it simply in a university release. “NOD2 functions as the body’s infection surveillance system. When bound to girdin, it detects invading pathogens and maintains gut immune balance by swiftly neutralizing them.”
Here is the catch. The most common Crohn’s linked mutation in NOD2 deletes the very segment where girdin would normally attach. Without that docking site, the surveillance system falters. Macrophages tilt toward the inflammatory team, and the gut can turn into a long-term battleground rather than a place that heals between insults.
To test the idea, the scientists turned to mice engineered to lack girdin. Those animals developed an imbalanced gut microbiome and inflammation in the small intestine. Many died from sepsis, an overwhelming immune reaction that can damage vital organs.
Co-first author Gajanan D. Katkar summed it up with a metaphor that anyone who has rushed to a restroom can probably picture. “The gut is a battlefield, and macrophages are the peacekeepers. For the first time, AI has allowed us to clearly define and track the players on two opposing teams.”
What this could mean for patients
The study does not deliver a new drug today. Crohn’s disease still involves many overlapping factors, from other genes to diet, smoking, and the trillions of bacteria that share our intestines.
Yet by clarifying exactly how a key mutation in NOD2 derails the gut’s peacekeeping cells, the work gives researchers a concrete target. Future therapies might try to mimic the missing girdin interaction, restore the surveillance pathway, or gently push macrophages back toward a healing state.
For people watching their diet, planning life around flare ups, and worrying about the next hospital bill, that kind of precision could eventually mean treatments that feel less like trial and error and more like fixing a known broken part.
The study was published in the The Journal of Clinical Investigation.
