A team took one of the body's own bone-building proteins, cut out two small pieces of it, and rebuilt those pieces into a short peptide that regrew bone in aged mice and pigs. The design was done largely by software, and the result deliberately avoids the part of the protein that raises cancer risk.
The work was published online June 12 in Nature Biomedical Engineering ↗ by Fanyuan Yu, Ling Ye, and colleagues.
The protein is WNT7B, one of the WNT family that tells stem cells to become bone. WNT signaling is one of the most important controls on skeletal health, which is why drug companies have spent years trying to turn it into an osteoporosis treatment. The problem is that WNT proteins are large, greasy, hard to manufacture, and expensive, and the main pathway they switch on, called canonical Wnt/beta-catenin, is the same pathway that drives several cancers when it stays on too long. A drug that floods that pathway to build bone could also nudge a tumor.
What they built
Instead of using the whole protein, the group asked which parts of WNT7B actually do the useful work. Using AlphaFold, the protein-structure prediction tool, plus computational docking that simulates how molecules fit together, they zeroed in on two regions of the protein they call the thumb and index domains, by analogy to a hand gripping its partner. They reconstructed those two domains into a single small peptide and named it WNT7B RTID, for reconstructed thumb and index domains.
A peptide is a short chain of amino acids, the same building blocks proteins are made of, just far smaller and easier to make and dose than a full protein. Shrinking a bulky protein down to the fragment that carries its activity is the whole point of the exercise.
In aged mice and in pigs with bone loss, two animals whose bones behave more like an older human's than a young lab mouse does, the peptide improved the bone-forming ability of mesenchymal stromal cells, the resident stem cells that turn into bone, and it helped close a critical-sized defect, an engineered gap too large to heal on its own. The repair happened without transplanting any stem cells. The peptide recruited the animal's own resident cells, which the authors tracked with single-cell sequencing and genetic lineage tracing.
Why the pathway matters
The part that makes this more than another bone-growth result is how the peptide works. The authors report that WNT7B RTID drives bone formation through a different, non-canonical route, a calcium-NFAT signal running through two proteins called RECK and GPR124, and that it does so independently of the canonical beta-catenin pathway tied to cancer. In plain terms, they claim to have kept the bone-building signal and dropped the part that worries oncologists. That separation, if it holds up, is the difference between a WNT-based drug that can be developed and one that cannot.
The usual cautions apply. This is animal work, not a human trial, and a peptide that recruits stem cells in a pig still has to clear safety, dosing, and manufacturing before anyone tests it in a person. The cancer-risk claim rests on the mechanism the authors mapped, not on long-term tumor data, which only a longer study can provide.
It also fits a pattern the section has been tracking. This is the latest in a run of peptides that were found or built by software ↗ rather than pulled from nature and tested by hand. The screen that found a KRAS-gripping peptide started from a library and narrowed down. This one started from a known protein and carved a drug out of it. Both lean on structure prediction to do work that used to take a wet lab years, and both still end the same way, with animals, then the long road to people.