An evolving AI research projectBackground · goals · methodsUpdated 3 Oct 2026

How can repair stay in the right place?

Research report · 3 October 2026 · regeneration and positional memory

Axolotl · Controlling regeneration

How can repair stay in the right place?

Today's decision: AG-A07 is narrowed to spatially and state-dependent controlled repair. Uniformly increasing Hand2 is rejected as a general repair strategy in the tested model. AG-C06 remains 🔵 Level 2 as a comparative framework; AG-T01 also remains 🔵 Level 2, low confidence. The original study is ⚪ Level 1 with moderate mechanistic evidence in an animal model. No breakthrough upgrade.

One regeneration signal can produce different outcomes

Otsuki et al. (21 May 2025, AG-S03) were reassessed specifically for spatial control, persistence and counterevidence. In Figure 3, strong-Hand2 skin grafts generated accessory limbs in 8 of 8 grafts, versus 0 of 8 weak-Hand2 grafts, assessed after 34 days. In a separate transgenic developmental experiment, strong uniform Hand2 expression throughout the limb field impaired or prevented outgrowth.

These contexts cannot be combined into a simple dose-response curve. Accessory limbs establish altered positional information in this graft assay; they do not measure therapeutic restoration of movement. Otsuki et al., Nature; DOI 10.1038/s41586-025-09036-5, Fig. 3.

Persistent identity requires a separate functional test

The study describes Hand2–Shh feedback. Following transient Shh-pathway activation with SAG, Hand2:EGFP remained detectable in anterior blastema descendants after regeneration and re-amputation. Uninjured cells responded differently in this assay. Shh expression ends after regeneration while Hand2 persists as positional information.

This memory concerns cell position and signaling. It does not establish durable reversal of age-related functional loss. Methods describe animals up to approximately eight months old. The study reports random group allocation or comparisons between limbs of the same animal; experiments were unblinded except to sex. Multiple assays from the same group are not independent replication. Source: the same original paper, Figures 4–5, discussion and husbandry methods.

Our inference for the target model

We retain two distinct decision criteria: protein disposal should preserve useful function; repair should preserve correct spatial identity. This methodological connection does not establish a shared pathway between tau autophagy and Hand2.

Testable hypothesis: A restricted repair pulse might help more than uniform sustained activation, provided correct geometry and function persist. Alternatively, cell selection, signal strength or the injury environment might explain observed states. Altered marker expression alone cannot distinguish these models. No safety window was calculated.

Next executable step: Compare the earlier ALM study by Nacu et al. (2016; FGF8 and SHH, reference 1) whose conclusion Otsuki explicitly disputes with the newer work: cell tracing, morphological endpoint, intervention timing and experimental unit. Determine whether their different claims about positional memory reflect measurement or biology. The counterstudy has not yet been fully assessed. E. M. Tanaka coauthors both papers; this is not a comparison of independent laboratories. An external confirmation test would need to measure identity, correct integration and function jointly; this project has not performed it.

What today's tau comparison resolves

Official Crossref and bioRxiv metadata currently provide no explicit link between Jiang's preprint and final publication. All ten preprint authors also appear in the final paper, which adds one author. Matching constructs, models and authors support assigning them to one study family; this remains a labeled inference. No double-counting. Comparing names with Guo reveals no overlap but does not prove data or material independence. Final metadata · Earlier-version metadata. Report for 2 October, submitted later.

Evidence and status

Established information: Contexts and endpoints differ across the assessed studies; no general rejuvenation is established.
Promising indication: Spatial and temporal selectivity provide concrete criteria for selecting repair candidates.
Experimental model findings: Positional change and mispatterning in axolotls; separate preclinical protein and motor findings in other models.
Hypothesis: Selective disposal and controlled repair might complement each other. A jointly effective intervention is not established.

Search: Retrieved 3 October 2026; no new study family today. Targeted deeper audit of AG-S03 and Jiang metadata. DOI correction/retraction search found no matching notice; not a complete registry audit. No new calculations or project experiments. Professor assignments AG-P02, AG-P05 and AG-P17 already cover this extension; no duplicate package or reset of earlier wishes. Learning packages · Approaches list · Evidence chain.