Key Takeaways
- Mice can regenerate skin without scarring before birth but lose this ability shortly after, due to new fibroblast populations and nerve interactions.
- A recent study identifies that excessive nerve fibers in wounds prevent regeneration, suggesting that regeneration potential exists but is blocked.
- Removing nerve signals restores the regeneration ability in postnatal wounds, as shown by using botulinum toxin to inhibit nerve activity.
- The research indicates that the ability to regenerate may be present in other organs, raising questions about their regenerative capacity.
- Further studies are needed to translate these findings to humans and explore application for chronic wounds or burns.
Three days before it is born, a mouse can be cut open and healed without a trace. Not healed in the way we usually mean the word (patched over, sealed, scarred) but regenerated, down to the hair follicles and lymphatic vessels and fat cells and pigment cells, restored to something indistinguishable from unwounded tissue. Five days after birth, that same mouse will scar. Somewhere in those eight days, a door closes.
What shuts it has puzzled regenerative biologists for decades. The assumption, a reasonable one, was that the embryo possesses something the newborn lacks: some cocktail of factors, some active regenerative program that switches off with development. If you wanted to restore the ability, the thinking went, you would have to rebuild that program from scratch.
A study published today in Cell, led by Ya-Chieh Hsu’s lab at Harvard, suggests the assumption was wrong. Embryonic skin does not simply possess regeneration in the way you might possess a skill. It is the postnatal skin that acquires something new: a specific population of fibroblast cells that recruits excessive nerve fibers into the wound bed, and those nerves, of all things, are what prevents regeneration from happening. Remove the nerve signal, and the ability to regenerate returns. “Our findings suggest that some organs retain an inherent regenerative potential that is simply held in check,” Hsu said, “and that removing this block may be sufficient to allow regeneration to occur. In other words, regeneration may not need to be built anew, but simply set free.”
Skin is perhaps the most misleadingly named regenerative organ in the body. After an injury, it does reseal itself: epidermal stem cells close the surface, fibroblasts lay down collagen. We call that healing. But skin is not just epidermis. Somewhere between ten and fifty different cell types reside in it, depending on how you count: hair follicles, sweat glands, fat cells, immune sentinels, pigment cells, an elaborate network of blood and lymphatic vessels, and a dense, branching architecture of sensory and sympathetic nerves. After a postnatal wound, most of these cell types simply fail to return. What grows back is a functional patch but not a faithful copy.
Lead author Hannah Tam spent five years working out what goes wrong. She learned to do microsurgery on mouse embryos under a dissection microscope, creating precise full-thickness wounds in skin the size of a thumbnail. One challenge was keeping track of where embryonic wounds had been, because they healed so completely they became invisible, so she marked the sites with fluorescent beads and henna ink. When she compared embryonic and postnatal wounds using single-cell RNA sequencing, a striking difference appeared in the fibroblast compartment. Postnatal wounds contained a large population of fibroblasts entirely absent from embryonic wounds, comprising roughly sixty percent of all fibroblasts after postnatal injury. Tam and her colleagues called these postnatal wound-specific fibroblasts, or PWFs.
PWFs are enriched for secreted signaling molecules, and a screen of nine candidate genes identified three that, when artificially introduced into embryonic wounds, blocked regeneration: Ccl7, Timp1, and Cxcl12. All three induced something the team had initially assumed was a side effect rather than the mechanism: excessive ingrowth of nerve fibers into the wound bed. Hyperinnervation, the paper calls it. “The surprising part is that we identify a block,” said Tam, now a postdoc at Scripps Research in California. “And this block is through fibroblast-nerve interaction. The relationship between those two different cell types has not been the focus in wound healing studies. I feel that this is very helpful to the field, because now we can really consider these two as actual communicators.”
Embryonic skin wounds regenerate all their cell types (hair follicles, fat cells, blood and lymphatic vessels, nerve fibers, pigment cells) because the inhibitory signals that block this process in postnatal skin have not yet appeared. The new Harvard study identifies a specific population of fibroblast cells that emerges only after birth and releases a chemical signal, CXCL12, that draws excessive nerve fibers into wound sites. Those nerves suppress regeneration. Before birth, these fibroblasts and this nerve-recruitment signal are absent, so regeneration proceeds unimpeded.
Botulinum toxin A, the same compound used cosmetically, blocks the release of signaling molecules from nerve terminals. When researchers injected it into postnatal mouse wound sites before healing began, it reduced excessive nerve ingrowth and restored regeneration of diverse skin cell types. The compound has already been used clinically to treat hypertrophic scars after they form, with some success. This study suggests the more powerful application might be earlier: during the initial wound-healing window, before the nerve overgrowth can suppress regeneration.
The team initially suspected immune cells, which are far more numerous in postnatal wounds than embryonic ones. But experiments that dramatically increased immune infiltration in embryonic wounds (including introducing bacterial compounds into the uterine environment) did not block regeneration at all. The key variable is not how many immune cells are present but whether nerve fibers are allowed to overgrow the wound bed. CXCL12, despite being classified as a chemokine that typically attracts immune cells, is acting here primarily on sensory neurons.
In adult mice, deleting the Cxcl12 gene from fibroblasts did improve regeneration compared with normal adult wound healing, and simultaneously targeting all three identified inhibitory genes improved it further still. But regeneration in adults was less complete than in neonatal tissue, which suggests additional age-related factors have not yet been addressed. Whether equivalent improvement would translate to human chronic wounds or burns is unknown; clinical translation would require identifying safe and practical ways to transiently suppress nerve signaling at wound sites in people.
The researchers tested functional connectivity by exposing regenerated mice to cold temperatures and checking for goosebumps. Goosebumps require three different tissue types (sympathetic nerves, arrector pili muscles, and hair follicles) to be present, correctly connected, and capable of coordinating in response to a sensory signal. In embryonically wounded mice, the response was intact in the healed region, confirming that regenerated structures had established genuine working connections rather than simply occupying the right anatomical positions.
The team hit a wall midway through, expecting immune cells to be the culprit. A good guess: postnatal wounds are far more inflamed than embryonic ones, dense with neutrophils and macrophages, and Cxcl12 and Ccl7 are conventionally understood as chemokines that attract immune cells. But when the researchers dramatically elevated immune infiltration in embryonic wounds (introducing bacterial lipopolysaccharide into the uterine environment, or overexpressing powerful immune-recruiting cytokines) regeneration continued unimpeded. Immune cells were not the block. Nerves were.
CXCL12 proved to be the clearest signal. Its receptor, CXCR4, is expressed on sensory neurons, and it is expressed at much higher levels in postnatal than embryonic sensory ganglia. Mice with Cxcl12 genetically deleted from fibroblasts showed dramatically reduced nerve ingrowth after postnatal wounding, and, crucially, robust multilineage regeneration: hair follicles, fat cells, lymphatic capillaries, sensory and sympathetic nerves, melanocytes, all returned. The same result came from a rather different approach: injecting botulinum toxin A, the active ingredient in Botox, into postnatal wound sites before healing began. BoNT/A disrupts the release of signaling molecules from nerve terminals, and wounds treated with it regenerated normally. Selective ablation of sensory neurons at the wound site produced the same effect, confirming that it is the nerve activity itself, not merely the physical presence of the fibers, that suppresses regeneration. Notably, BoNT/A has already shown clinical promise for reducing hypertrophic scars after they form; the paper suggests applying it earlier, during the regenerative window, might be considerably more valuable.
The biological logic of this is, once you see it, almost coherent. Sensory nerves reach the skin around embryonic day 14.5 in mice but continue to mature and densify postnatally. When a wound transects those mature nerves in postnatal skin, they respond by sprouting aggressively, guided in part by CXCL12 from the newly arrived wound fibroblasts. These nascent regenerating axons are hyperexcitable and prone to spontaneous firing. They may physically obstruct the downgrowth of regenerating hair follicles and other dermal structures; they certainly flood the wound bed with neuropeptides and neurotransmitters that appear to interfere with regenerative gene programs. The organism needs this hypersensitive wound response: it alerts you to the injury, promotes protective behavior, wards off further damage. The cost, apparently, is scar.
The regenerated wounds were not merely structurally restored. Tam verified functional connectivity by testing whether cold exposure produced goosebumps in the healed region. Goosebumps require a tri-lineage coordination: sympathetic nerves (neuronal), arrector pili muscles (mesenchymal), and hair follicles (epithelial) must all be present, connected in the correct geometry, and responsive together. In embryonically wounded mice, the cold response was fully intact. In adult mice, the results were more modest: Cxcl12 knockout alone improved regeneration compared with wild-type adults, and a triple knockout targeting all three identified inhibitory genes improved it further, but regeneration was less complete than in neonatal tissue, suggesting other factors accumulate with age that this approach does not yet address.
There are real limits to what this study can tell us about human wound healing. Mouse embryos are not human embryos, and the specific timeline of regenerative window closure will differ. But hyperinnervation of wound beds has been documented in human tissue, too, and the molecular machinery (CXCL12, CXCR4, the general architecture of fibroblast-nerve crosstalk) is conserved. Harvard has filed a provisional patent. Hsu is cautious but not pessimistic about translation: “I didn’t think that we’d have to retract a brake, which actually is good news. It’s a lot easier. It gives me hope that this might be applicable to improving wound healing in humans.” What remains to be worked out is how to time an intervention precisely: early enough to intercept the hyperinnervation before it blocks regeneration, but in a wound environment that has not yet declared what it will do.
The deeper implication may matter more than the immediate application. For decades, regenerative medicine has operated on the assumption that mammalian tissues simply lack the capacity to regenerate organs, and the task was to install that capacity anew. The skin, at least, appears to work differently. The capacity is there, held in reserve, blocked rather than absent. Which raises a question for every other organ system where regeneration seems to have simply stopped: is it gone, or is it waiting?
DOI / Source: 10.1016/j.cell.2026.02.027
Quick Note Before You Read On.
ScienceBlog.com has no paywalls, no sponsored content, and no agenda beyond getting the science right. Every story here is written to inform, not to impress an advertiser or push a point of view.
Good science journalism takes time — reading the papers, checking the claims, finding researchers who can put findings in context. We do that work because we think it matters.
If you find this site useful, consider supporting it with a donation. Even a few dollars a month helps keep the coverage independent and free for everyone.
Related
