Scientists Crack 20-Year DNA Repair Puzzle in Human Cells

Your DNA breaks every day. Sunlight, chemicals and the ordinary chemistry of living cells snap its strands thousands of times daily, and the body stays healthy because DNA repair crews find each break and fix it. For more than twenty years, one of those crews stayed out of reach: five proteins known as the RAD51 paralogs, long linked to cancer but too unstable to study in the lab.

That blind spot is closing. Researchers at the Francis Crick Institute in London have published a detailed picture of how the paralogs contribute to DNA repair, in the journal Science. The team, led by Luke Greenhough, combined AI-based structure prediction, cryo-electron microscopy and single-molecule imaging to study the proteins directly. According to the researchers, the five proteins organize into two distinct complexes, each built from four subunits, and each complex has its own job. One starts the assembly of RAD51 filaments, the scaffolds that carry out the actual repair. The other caps the ends of those filaments, holding them stable while the repair runs its course.

The result corrects a simpler picture of the DNA repair machinery and ends a search that began in the early 2000s, when Stephen West's group at the Crick first showed that the paralogs interact to form protein complexes. After that, progress stalled. The complexes clumped together or fell apart in the test tube before anyone could pin them down, and the proteins became effectively off limits.

The builder and the cap

RAD51 itself is the central protein in DNA repair by homologous recombination, the process that fixes double-strand breaks by copying the matching sequence from a sister chromosome. Its paralogs, RAD51B, RAD51C, RAD51D, XRCC2 and XRCC3, support that work, but the details were missing. The new paper, which builds on a 2023 study from the same group, shows a division of labor that had been impossible to see: as the team wrote in the paper, one complex centered on RAD51B promotes the nucleation and growth of RAD51 filaments on damaged DNA, while a second complex built around XRCC3 sits on the end of a filament and stops it unraveling.

Watching the two complexes behave changed how the researchers understood them. Rather than a fixed set of parts, the DNA repair proteins assemble and disassemble as the job requires, first laying down the filament and then guarding its exposed ends. That dynamic picture emerged only because the team could finally keep the proteins intact long enough to observe them, something earlier methods never allowed.

"It was a real gap in the field," West said. "It was impossible to investigate their biochemistry."

Three tools opened the door

The advance was not a single discovery but three technologies arriving together. An AlphaFold3 prediction of the proteins' shapes gave the researchers a starting point they might not otherwise have considered. Cryo-electron microscopy, which passed through a resolution revolution in the 2010s, let them see the protein clusters in fine molecular detail. Single-molecule imaging allowed them to watch individual proteins interact with DNA in real time. Together, the methods supplied both the structure and the behavior of the complexes: what they look like and what they do.

Greenhough said the combination did what no earlier method could, giving the team the chance to "see the machinery in action." The paper is published as Luke A. Greenhough et al., "Cryo–electron microscopic visualization of RAD51 filament assembly and end-capping by XRCC3-RAD51C-RAD51D-XRCC2," Science (2026), doi:10.1126/science.aea1546. A summary of the findings is also available from the researchers' institutional reporting at phys.org. The result arrives as biology leans harder on AI-guided discovery, including a recent virus-enzyme find with potential for gene editing (GenZ NewZ).

The study also helps explain a long-standing puzzle about why the paralogs were so hard to work with. Because the complexes are modular and constantly rearranging, they resist the stable, frozen states that structural biology prefers. The team's methods met the proteins on their own terms, catching transient shapes that purification-heavy approaches had been washing away.

What it means for cancer

The finding reaches well beyond structural biology. The RAD51 paralogs operate in the same biological pathway as BRCA1 and BRCA2, the well-known DNA repair genes whose mutations raise the risk of breast and ovarian cancer. "The RAD51 paralogs operate in the same biological pathway as well-known DNA repair genes BRCA1 and BRCA2," West said. Mutations in the paralog genes themselves are also linked to those cancers and to Fanconi anemia, a rare inherited disorder that leaves patients prone to cancer. The cancer connection is not abstract: the World Health Organization recently set out a strategy to close the global gap in childhood cancer medicine, as reported here (GenZ NewZ).

That shared pathway is the reason some cancers driven by mutations in these genes respond to drugs called PARP inhibitors. When a tumor's DNA repair machinery is already broken, blocking a second repair route can push the cancer cells past the point of survival while sparing healthy tissue. Understanding exactly how the paralogs assemble the repair filament makes it easier to map which mutations matter most and how they weaken the system.

Greenhough said mapping the structural effects of paralog mutations "helps explain why they lead to disease," opening a route to understanding which variants carry the greatest risk. West described the result as the solution to a "20-year puzzle," crediting recent advances in imaging technology and Greenhough's persistence for uncovering "this missing piece of information."

Nobody involved is claiming new treatments yet. The work is basic research: a clearer map of the DNA repair machinery and of how specific mutations break it. For cancer genetics, that map is the starting point everything else builds on, and it came from proteins that spent years defying study.