The Gene Switch for Regeneration: How ALDH1A2 and Retinoic Acid Made Mice Regrow Lost Tissue

Cut a hole in a rabbit's ear and it quietly rebuilds it — cartilage, skin, blood vessels, even nerves — until the hole is gone. Do the same to a lab mouse and you get what humans get: a scar. For decades that difference looked like a deep, unfixable fact of mammalian biology. In 2025, a team at the National Institute of Biological Sciences in Beijing reported something startling in Science: the gap comes down largely to one missing signal — retinoic acid — and the under-active enzyme that makes it, ALDH1A2. Switch that enzyme back on, and a mouse starts regenerating again. [1][2]

It's one of the cleaner "here is the actual lever" findings in regeneration biology, and — like the one-time gene edit that lowers cholesterol for life — it's the kind of frontier result where the caveats matter as much as the headline. Here's the honest version.

The experiment: rabbits regenerate, mice scar — why?

The researchers used the classic test of mammalian regeneration: a small punch hole through the ear pinna (the external ear flap). Rabbits, goats and African spiny mice close that hole with real, organized tissue. Ordinary mice and rats don't — they seal the edges with fibrotic scar and leave the hole open. [1][5]

To find the difference, the team ran comparative single-cell and spatial transcriptomics — essentially reading, cell by cell and location by location, which genes switch on in a rabbit ear versus a mouse ear in the days after injury. [1] One signal stood out. After wounding, rabbit wound cells ramped up retinoic acid (RA) — the active, hormone-like form of vitamin A. Mouse wound cells barely did. [1][3]

What ALDH1A2 and retinoic acid actually are

Retinoic acid is a morphogen: a diffusible signal that tells cells not just to divide, but what to build and where. Your body makes it from dietary vitamin A in two steps, and the rate-limiting second step — converting retinaldehyde into retinoic acid — is run by an enzyme called ALDH1A2 (also known as RALDH2). [1][4] If ALDH1A2 isn't switched on, the retinoic-acid signal never gets made, no matter how much vitamin A is around.

The Beijing team found the failure in mice is two-sided: [1]

• Too little production — mice fail to induce ALDH1A2 at the wound, so they don't make enough RA.
• Too much destruction — mice simultaneously ramp up RA-degrading machinery (the CYP26 family of enzymes), shredding what little signal there is.

The result is a wound that never reaches the RA threshold needed to flip cells from "patch the hole" to "rebuild the structure." Rabbits cross that threshold; mice don't.

Switching it back on

Here's the part that makes it a landmark rather than just a tidy description. The team showed the block is reversible two different ways: [1][3][4]

1. Supply the signal directly. Giving the wound retinoic acid itself restored regeneration — bypassing the missing enzyme entirely.
2. Restore the switch. Mice under-express ALDH1A2 because several gene-control elements ("enhancers") that should turn it on after injury have been evolutionarily inactivated in mice and rats. The team engineered mice carrying a single rabbit enhancer (called AE1) — one piece of regulatory DNA — and that alone was enough to boost ALDH1A2 after injury and restore regrowth of cartilage and other ear structures. [1]

In other words: the mouse still has the regeneration program. It had only lost the switch that turns it on. Put the switch back — genetically, or by adding the downstream signal — and the program runs.

"Evolution made a trade-off"

If mice carry the machinery and only lost the switch, why would evolution disable it? The honest answer is that we don't fully know, and the work frames it as a trade-off rather than a defect. [1][5] The leading ideas in regeneration biology — and these are hypotheses, not settled fact — are that strong, RA-driven regrowth sits in tension with other priorities of large, long-lived, warm-blooded mammals: fast fibrotic sealing closes a wound quickly against infection, and tightly restrained growth signaling is part of how big animals suppress cancer. A scar may be the fast, low-risk option natural selection favored in our lineage. Interesting, but unproven — treat it as a story, not a result.

This isn't biology's first retinoic-acid regeneration clue

What makes the finding credible is that it rhymes with decades of older work pointing at the same molecule:

• African spiny mice (Acomys). In a landmark 2012 Nature paper, Seifert and colleagues showed these rodents heal skin and ear-punch wounds scar-free, regrowing cartilage, hair follicles, glands, vessels and nerves where ordinary mice scar. [6][7] They were the proof that "mammals can't regenerate" was too strong a claim — and they're one of the regenerators in the 2025 comparison. [10]
• Axolotls and the salamander limb. The amphibians that regrow whole limbs have leaned on retinoic acid for the same job far longer. Classic experiments by Maden in the early 1980s showed retinoic acid re-specifies positional identity in a regenerating limb — bathe a blastema in it and the animal regrows extra, more-proximal limb segments. [8] More recent work shows that controlling RA — including breaking it down via CYP26 — is essential for getting the proportions right. [9]

So the 2025 mouse result lands inside a much older pattern: across animals that regenerate well, retinoic acid is a core "build this structure, here" instruction — and the animals that scar are, in part, the ones that stopped making or keeping it.

The honest caveats (read this part)

This is a genuinely important mechanism. It is not a regeneration therapy, and the distance to one is large:

• It was the ear pinna — in mice. Cartilage, skin, follicles, vessels and small nerves in a thin flap of tissue. That is a very long way from a limb, a heart, a spinal cord, or a whole organ. Nothing here regrows those. [1]
• The tools were genetic engineering or locally delivered RA — not a pill. "Restored regeneration" means a transgenic rabbit enhancer or retinoic acid applied at the wound, not a supplement you can take. [1]
• Retinoic acid is powerful and dangerous. It is a potent morphogen with a narrow safe window and is strongly teratogenic — it causes severe birth defects, which is exactly why the acne drug isotretinoin (Accutane) carries some of medicine's strictest pregnancy controls. [11] More vitamin A does not equal more regeneration in a person; in excess it is toxic.
• "Reactivated" is not "perfected." The transgenic enhancer improved regeneration; this is a first mechanism, not a fully-regrown ear on demand. [1]

In short: a beautiful proof that a single signaling switch gates mammalian regeneration — and a reminder that turning it on safely, in the right place, in a human, is the hard part that hasn't been done.

Where this fits with regenerative peptides

Because we track the research-peptide world, it's worth being precise about what this is and isn't. The compounds people research for wound healing — BPC-157, TB-500, GHK-Cu — are studied for things like angiogenesis, cell migration and tissue remodeling: helping a wound heal better and faster. The ALDH1A2/retinoic-acid finding is a different category — a morphogen positional signal that drives true structural regeneration of an appendage. They are not the same mechanism, and nothing in this study shows any peptide switching on ALDH1A2.

What they share is the theme this site keeps running into: mammalian healing is gated by signaling we're only beginning to map, and the honest evidence rarely matches the marketing. (For a worked example of separating signal from hype, see our reality-check on BPC-157's human evidence.)

Bottom line

A 2025 Science study found that mice fail to regenerate the way rabbits do largely because they don't switch on ALDH1A2 after injury — so they never make enough retinoic acid — and they degrade what little they do make. Reactivate the enzyme with a single rabbit gene switch, or supply retinoic acid directly, and regenerative healing comes back. It reframes a "fixed" limit of mammalian biology as a dormant program with a missing switch. The switch is real. Flipping it safely in humans is the work that hasn't started.

Sources

1. "Reactivation of mammalian regeneration by turning on an evolutionarily disabled genetic switch" (team at the National Institute of Biological Sciences, Beijing), Science (2025), DOI 10.1126/science.adp0176 — https://www.science.org/doi/10.1126/science.adp0176
2. PubMed record (PMID 40570123) — https://pubmed.ncbi.nlm.nih.gov/40570123/
3. Phys.org — "Switching on a silent gene revives tissue regeneration in mice" (June 2025) — https://phys.org/news/2025-06-silent-gene-revives-tissue-regeneration.html
4. Genetic Engineering & Biotechnology News (GEN) — "Mice Regenerate Ear Tissue When Vitamin A Genetic Switch is Flipped" (2025) — https://www.genengnews.com/topics/translational-medicine/mice-regenerate-ear-tissue-when-vitamin-a-genetic-switch-is-flipped/
5. Lifespan.io — "Study Discovers a Mammalian Mechanism of Tissue Regeneration" (July 2025) — https://lifespan.io/news/study-discovers-a-mammalian-mechanism-of-tissue-regeneration/
6. Seifert, A.W. et al. — "Skin shedding and tissue regeneration in African spiny mice (Acomys)," Nature 489:561–565 (2012) — https://www.nature.com/articles/nature11499
7. "Ear wound regeneration in the African spiny mouse Acomys cahirinus," Regeneration (2016), PMC4857749 — https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4857749/
8. Maden, M. — "Vitamin A and pattern formation in the regenerating limb," Nature 295:672–675 (1982) — https://www.nature.com/articles/295672a0
9. "Retinoic acid breakdown is required for proximodistal positional identity during amphibian limb regeneration," Nature Communications (2025) — https://www.nature.com/articles/s41467-025-59497-5
10. "Spiny mouse (Acomys): an emerging research organism for regenerative medicine with applications beyond the skin," npj Regenerative Medicine 5:22 (2020) — https://www.nature.com/articles/s41536-020-00111-1
11. Lammer, E.J. et al. — "Retinoic acid embryopathy," New England Journal of Medicine 313:837–841 (1985) — https://www.nejm.org/doi/10.1056/NEJM198510033131401