Scientists Crack How the Immune System Aims Its DNA Mutator

Your immune system has a secret superpower, and it involves deliberately scrambling its own DNA. Every day, B cells — the white blood cells that make antibodies — deploy an enzyme called AID that mutates antibody genes on purpose, sharpening antibodies so they latch onto invaders with ever-greater precision. For more than two decades, scientists could not explain how the immune system keeps that weapon aimed only at antibody genes instead of letting it loose across the whole genome. Now, in a study published October 1, 2026 in the journal Nature, researchers at Montreal's IRCM say they have found the aiming mechanism: two proteins that herd the enzyme into droplet-like condensates exactly where it is supposed to act.

AID, short for activation-induced cytidine deaminase, is what makes your immune memory work. When a vaccine or an infection trains your body, AID drives two upgrades inside B cells: class-switch recombination, which swaps an antibody's "effector" machinery, and somatic hypermutation, which introduces point mutations that fine-tune how tightly the antibody grips its target. But a DNA-mutating enzyme that misses its mark can wreck the genome — and AID is a known accomplice in cancers of the B cell, including lymphomas. Javier M. Di Noia, the Université de Montréal professor who led the IRCM team, put the stakes plainly: "AID is both essential and potentially dangerous… if this enzyme acts in the wrong place in the genome, it can damage important genes, cause chromosomal rearrangements and contribute to the development of B-cell cancers," according to MedicalXpress, which covered the paper on October 1, 2026.

The new work identifies the gatekeepers: MLLT1 and MLLT3, proteins better known to cancer researchers as ENL and AF9. Both are part of the super elongation complex, and each carries a YEATS domain — a "histone reader" module that latches onto specific chemical marks on histones, parking the proteins at active, acetylated stretches of DNA such as enhancers and super-enhancers. Where those readers cluster, the new study shows, AID follows, ScienMag reported in its summary of the Nature paper.

How the immune system keeps AID on target

The most dramatic evidence came from mouse B cells engineered to lack both proteins. With the readers gone, every AID-dependent mutagenic process ground to a halt: class-switch recombination was abolished and somatic hypermutation disappeared entirely — even though AID itself was still present at normal levels. In mice, MLLT1 does most of the heavy lifting. Genome mapping then showed that the regions AID mutates are marked by unusually high occupancy of the two readers, a pattern that held in human B cells, in human lymphoma cell lines, and across a catalogue of 610 lymphoma cases, where the most frequently mutated loci correlated with the readers' enrichment.

The mechanism is physical, not just chemical. MLLT1 has a long, floppy, intrinsically disordered region that drives phase separation — the same phenomenon that lets oil bead up in water — forming droplet-like condensates inside the cell nucleus. Live-cell imaging and test-tube experiments showed AID preferentially partitioning into those droplets, effectively concentrating the enzyme where the readers have gathered. And in the experiment that clinched it, the team fused AID directly to either reader and restored class-switch recombination and mutagenesis in cells missing both proteins — proof, the authors argue, that local concentration is the license that lets AID act.

What the discovery could mean for cancer

Because misdirected AID activity feeds the chromosomal translocations behind some lymphomas, the researchers tested whether blocking the gatekeepers could slow that damage. Molecules that inhibit the two proteins reduced both AID-driven mutations and cancer-linked translocations in experimental models, according to MedicalXpress. The therapeutic angle is not purely theoretical: inhibitors of MLLT1 are already in clinical development for certain forms of leukemia, and repurposing them against diseases where AID drives tumor progression "could eventually be considered," as the coverage put it.

There is a real counterweight, though. The same machinery that cancers exploit is the machinery your body uses to diversify antibodies after vaccination — so a drug that breaks up these condensates could also dull the immune memory that vaccines build. And the results so far come from experimental models, not patients; no one is proposing a treatment yet. Still, the finding closes a gap that has puzzled immunologists since AID's targeting behavior was first described: the work reveals a previously unknown layer of control, showing how evolution managed to tolerate such a dangerous enzyme without broadly compromising genome integrity.

The paper is also the payoff of a long campaign. Di Noia's IRCM laboratory has chased AID's regulation for more than a decade, including a 2014 study linking a new AID function to a rare immunodeficiency syndrome and earlier work identifying Hsp90 as a protein that stabilizes the enzyme, according to background reporting from McGill's health news service. Compared with the blunt tools of older cancer therapies — and even with precision approaches like the CRISPR cholesterol therapy now cutting bad fat in half, covered in our science coverage — this discovery targets the geography of mutation rather than any single gene.

For young readers, the takeaway is concrete: the vaccines you have taken rely on exactly this enzyme doing its job correctly, and understanding its targeting is the first step toward drugs that stop it when it goes rogue. The next moves belong to translational researchers, who will need to show that blocking the readers works in living systems without wrecking normal antibody diversification. If they can, a twenty-year mystery could become a genuinely new class of cancer therapy. For more on the beat, see our science topic page, and our look at how five infections drive millions of cancer cases a year.