Textbooks say mammals traded regeneration for speed. Wound a salamander and it rebuilds a limb. Wound a human and you get a scar. The standard explanation: somewhere in the transit from amphibian to mammal, evolution deleted the regeneration code and left fast closure as the consolation prize. Scar tissue is what remains when the program is gone.
Two studies from the past year say the program was never gone.
The order is the instruction
In April 2026, Ken Muneoka's lab at Texas A&M amputated mouse digits and applied two proteins in sequence: first FGF2, a fibroblast growth factor, after the wound had closed; then BMP2, a bone morphogenetic protein, several days later. The mice grew back bone, joint, tendon, ligament, and articular cartilage. Organized multi-tissue architecture, built from ordinary fibroblasts at the wound site, without a single transplanted stem cell.
The proteins are already known. BMP2 has FDA approval for clinical use. FGF2 is in multiple clinical trials. The breakthrough was the sequence.
When Muneoka's team applied both proteins simultaneously instead of sequentially, 20.8 percent of treated digits developed ectopic bone (n = 48), bone growing where bone has no business being. Same cells, same wound, same two molecules. Change the order and the body builds the wrong thing in the wrong place.
"This is really a two-step process," Muneoka said. "You first shift the cells away from scarring, and then you provide the signals that tell them what to build."
FGF2 first: redirect fibroblasts away from the scar pathway, toward a blastema. BMP2 second: instruct the blastema to build. Skip the first step and the constructive signal lands on cells still running the default program. The result is bone shaped by noise instead of pattern.
The switch, not the gene
A year earlier, Weifeng Lin and colleagues at the National Institute of Biological Sciences in Beijing came at the question from the genome. They compared rabbits, which regenerate ear cartilage after injury, with lab mice, which scar. The Aldh1a2 gene, which produces retinoic acid, a molecule central to cellular reprogramming, sits intact in both species. Rabbits and mice diverged roughly 90 million years ago. The gene survived the split. What changed was the wiring: the enhancer sequences that activate Aldh1a2 after injury went silent in the mouse lineage. The hardware is fine. The power switch is off.
Lin's team turned it back on. Transgenic mice carrying a single rabbit enhancer showed partial ear regeneration, regrowing cartilage and other structures. Adding retinoic acid directly to normal mice went further: cartilage, dermis, hair follicles, nerves.
Two labs, two methods, same conclusion. The regenerative program has been sitting in the mammalian genome the whole time, intact and idle. Scarring wins every wound because it is the faster path, the one evolution wired as the default. The fibroblasts that seal your cuts carry the full blueprint for something more elaborate. They never get the signal to run it, and when you do send the signal, it has to arrive in the right order.
First, stop
Larry Suva, a co-author on the Muneoka paper, put it plainly: "The capacity is not absent. It's just obscured."
The part of this that travels beyond biology is the ordering constraint. In a system running a fast default, you cannot layer the better alternative on top. You have to stop the default first. FGF2 before BMP2. Halt the scar before you signal the build. Reverse that and you get bone in the wrong place, which is worse than doing nothing.
The pattern recurs wherever a fast response crowds out a slow one. The body learned this 90 million years ago when it chose speed over fidelity. The researchers learned it last April when they found you could undo that choice, as long as you went in the right order.