🧬 Aging Wiki

R38

8 min read

log/R38.md — Round 38 entries

Sub-file of log — see parent for index.

[2026-05-19] verify — hattori-2004-scf-solar-lentigo

Pages verified: 1

  • studies/hattori-2004-scf-solar-lentigo — full PDF read (10 pages). Corrections: (1) c-KIT RT-PCR assay-table n populated as n=6 (was vague “included in RT-PCR subset”); (2) GROα 1.13-fold figure added to Key Findings prose (omission; NS conclusion unchanged). All R38 priority checks confirmed against source.

Downstream pages to update: main agent to propagate corrected c-KIT n=6 to any entity pages (skin-aging, keratinocytes, melanocytes) that cite this study with c-KIT assay specifics.

[2026-05-19] R38 close — skin-aging cluster substrate (8 tissue + cell-type pages)

Opening round of the R38–R44 skin-aging coverage campaign (drafted same day in ROADMAP.md; transcript: “skin-aging coverage” thread). Substrate layer for the existing verified phenotypes/skin-aging.md hub, which previously dangled wikilinks to [[skin]], [[dermis]], [[epidermis]], [[dermal-fibroblasts]], [[melanocytes]]. 8 atomic pages seeded + verified in single-day round. 16 agent invocations (8 seeders + 8 verifiers) + 1 propagation pass.

Pages added (all 8)

Tissues (3):

  • tissues/skin.md (parent; integumentary system overview)
  • tissues/dermis.md (papillary + reticular; ECM-dominated; fibroblast home)
  • tissues/epidermis.md (stratified squamous keratinizing epithelium; UV frontline)

Cell types (5):

  • cell-types/dermal-fibroblasts.md (CL:0002620; most-studied senescent population in skin; Solé-Boldo 2020 subset framework)
  • cell-types/keratinocytes.md (CL:0000312; ~95% of epidermal cells; Clayton 2007 vs Mascré 2012 IFE stem-cell debate)
  • cell-types/melanocytes.md (CL:0000148; epidermal melanin unit; Hattori 2004 solar lentigo SCF/KIT paradigm)
  • cell-types/langerhans-cells.md (CL:0000453; sole resident dendritic cell of steady-state epidermis; long-lived self-renewing)
  • cell-types/melanocyte-stem-cells.md (no CL term — cell-ontology-id: null + #gap/needs-canonical-id per FAP precedent; Nishimura 2002/2005 + Zhang 2020 + Sun 2023)

Major verifier findings — fabricated/inverted/wrong claims caught

Extremely high yield — every page caught at least one substantive seeder error; total ~45 corrections across the 8-page batch. Continues the seeder-fabricates-outcomes pattern.

Critical biology inversion caught:

  • Inomata 2009 ATM direction was backwards on melanocyte-stem-cells.md: seeder wrote “ectopic differentiation suppressed in Atm-null mice”; paper actually shows Atm−/− mice are more sensitive (3 Gy required vs 5 Gy WT). ATM is a protective stemness checkpoint, not a differentiation driver. Cross-checked hair-greying.md (which cites the broader Nishimura 2005 framework) — does NOT cite Inomata 2009; no propagation needed.

Wholesale primary-source misattributions:

  • Ganceviciene 2012 misattribution on tissues/skin.md: three quantitative anatomy figures (body weight ~15-16%, dermis ~20% thickness loss/decade, melanocyte density ~6-8%/decade) all attributed to this paper, which is actually a therapeutics review (lasers, fillers, BTX, retinoids, HRT) containing none of the cited figures. All demoted to #gap/unsourced.
  • Fisher 2009 YAP/TAZ misattribution on cell-types/dermal-fibroblasts.md: seeder ascribed YAP/TAZ mechanosensing to Fisher 2009, but the 2009 paper predates YAP/TAZ in fibroblast mechano-regulation. Actual Fisher 2009 mechanism is c-Jun/AP-1 + α2β1 integrin axis. Same misattribution propagated to verified phenotypes/skin-aging.md line 61 — fixed during propagation pass: added explicit “(YAP/TAZ is a later-published mechanism not in Fisher 2009 itself)” qualifier.
  • “~1% per year collagen loss” misattributed to Fisher 2009 on both tissues/dermis.md and (pre-existing) phenotypes/skin-aging.md. Figure originates with Shuster 1975, reiterated by Varani 2006. Tagged #gap/unsourced on all three pages with attribution note.

Cell-biology fabrications:

  • Solé-Boldo 2020 subset names ALL wrong on cell-types/dermal-fibroblasts.md: seeder wrote “Papillary-like / Reticular-like / Pro-inflammatory / MPS-like”; paper actually uses Secretory-reticular / Pro-inflammatory / Secretory-papillary / Mesenchymal (Table 1, Fig 4a). Corrected throughout body + frontmatter single-cell-aging-signature field + footnote.
  • Park 2023 fabrication on cell-types/melanocytes.md: melanosome transport defect attributed to KIF5B/KLC2 (kinesins); paper’s actual finding is MYO5A/RAB27A/MLPH tripartite complex downregulation. Confirmed via grep that this fabrication did not propagate to other pages (KIF5B mentions on miro1.md/miro2.md are correct in the mitochondrial transport context).
  • Yu 2025 conflated with Park 2023 on cell-types/melanocytes.md: seeder cross-attributed melanosome transfer impairment to Yu 2025; Yu 2025’s actual finding is elevated MC1R + eumelanin synthesis genes in CDKN1A+ melanocytes. Corrected.
  • Mascré 2012 construct fabrication on cell-types/keratinocytes.md: seeder wrote “K14-rtTA + K10-CreERT” with “K14^low/K14^high” population labels; paper actually uses K14-Cre-ER vs Inv-Cre-ER (involucrin-Cre-ER) with no K14^low/high nomenclature. Corrected.

n-value / cohort corrections:

  • Hattori 2004 n=12 → n=29 total (with per-assay n=7/10/6) — caught by both keratinocytes and melanocytes verifiers (same paper); also fixed page range (122:1256–1265) and added c-KIT-mRNA-unchanged finding.
  • Velarde 2012 n conflation on keratinocytes.md: aged C57BL/6J cohorts (n=6/10/13 at 4/8/24mo) conflated with Sod2-/- experiments (n=6 vs 6 Westerns, n=8 vs 9 histology).
  • Bhushan 2002 site fabrication on langerhans-cells.md: “sun-protected forearm” → “sun-protected buttock” (paper used buttock); density decline “~30-50%” → actual ~28% (1156.3 → 835.7 cells/mm²).
  • Wang 2009 mouse model wrong on langerhans-cells.md: “Langerin-DTA” → “Lang-DTR” (DT receptor knockin, not DT-A toxin chain).

Other corrections:

  • Clayton 2007 IFE division probability ~10% → ~8% (paper’s rλ = 0.088 ± 0.004 per week)
  • Sun 2023 PI attribution: I had briefed “Sun/Hsu” — Mayumi Ito (NYU) is corresponding author; Hsu lab is Zhang 2020 (Harvard). Corrected to Sun/Ito 2023.
  • Sun 2023 81%/83% conflation (two distinct measurements from Fig 1a vs 1g merged)
  • “Stuck McSC” → “stranded McSCs” (paper’s actual term)
  • Inomata 2009 reagent fabrication: “hydroxyurea” → actual reagents busulfan + MMC + H₂O₂
  • Hochberg 1999 DOI resolved (10.1111/j.1751-1097.1999.tb08281.x) + author list corrected (“et al.” → “Hochberg M, Enk CD” — only 2 authors)
  • Lin & Fisher 2007 “30-40 keratinocyte / melanocyte” ratio tagged #gap/unsourced (widely-cited figure not actually in the paper)
  • “10-15 year LC half-life” demoted from langerhans-cells.md (no Merad 2002 statement; only mouse data)
  • HA framing on tissues/dermis.md: “reduced HA content with age” → actual: HA quantity unchanged in intrinsic aging (HA-binding proteins decline); HA increases in photoaged skin (solar elastosis regions). The wiki’s prior claim was factually incorrect; corrected.
  • Versican intrinsic-aging status corrected from “upregulated” → “uncertain” (Shin 2019 Table 1: “Not changed?” intrinsic; increased only in photoaged)
  • Epidermal transit time “30-35 days in older adults” → “>30 days” (Grove 1983 footnote scope; upper bound not in source)

Inter-verifier contradiction surfaced (deferred to R39): Fisher 1996 “~70% AP-1 reduction / 50-80% MMP reduction by tretinoin” — keratinocytes verifier reported these are not explicitly stated in paper text; dermis verifier reported they are confirmed from Figs 4a-4c visual data. Likely both correct in different scopes (figures present in figures, not text). Definitive resolution deferred to R39 when studies/fisher-1996-photoaging-ap1-mmp.md gets a dedicated end-to-end verifier pass.

Propagation work completed

  • phenotypes/skin-aging.md (pre-existing, verified 2026-05-05): updated line 61 collagen-fragmentation paragraph — “~1% per year” attribution to Fisher 2009 demoted to #gap/unsourced with Shuster 1975 / Varani 2006 attribution note; collagen I+III ~80% → ~75% dry weight (Shin 2019); MMP-1 specified as central effector vs MMP-3/MMP-9 as secondary processors; YAP/TAZ qualified as later-published mechanism not in Fisher 2009; Fisher 2009 mechanism correctly noted as c-Jun/AP-1 + α2β1 integrin.
  • processes/advanced-glycation-end-products.md: “skin dermis” → [[dermis|skin dermis]] in intro paragraph + table row.
  • processes/glucosepane.md: “skin dermis” → [[dermis|skin dermis]] in table row.

Propagation work deferred (lower priority; not blocking):

  • New content additions to hallmark MOCs (cellular-senescence.md, stem-cell-exhaustion.md, chronic-inflammation.md, loss-of-proteostasis.md, disabled-adaptive-immunity.md) cross-linking to new R38 substrate — deferred to R39+ when the substrate gets thicker.
  • melanogenesis.md (type: pathway, verified) restating of cell biology already on [[melanocytes]] — leave alone; melanogenesis page is mechanism-focused and the cross-link from melanocytes.md is sufficient.

Schema escalations surfaced

Three separate seeders flagged that type: tissue has no formal frontmatter block in CLAUDE.md despite 6 (now 8) tissue pages existing. Schema convention is established via tissues/skeletal-muscle.md precedent: required fields are type, aliases, parent-system, key-cell-types, key-aging-phenotypes, related-hallmarks, verified/verified-date/verified-by/verified-scope. Action for next CLAUDE.md cleanup pass: formalize type: tissue block using skeletal-muscle.md as canonical specimen.

Also: cell-ontology-id cleanly extended to McSC’s case where no CL term exists (null + #gap/needs-canonical-id per FAP precedent worked as intended).

Cross-page consistency reconciliations

  • Melanocyte density decline figures: previously inconsistent between tissues/skin.md (“~6-8% per decade”, now demoted to #gap/unsourced as Ganceviciene 2012 fabrication) and cell-types/melanocytes.md (“8-20%/decade”, Ortonne 1990 not_oa). Resolution: skin.md figure removed; melanocytes.md retains Ortonne attribution with #gap/no-fulltext-access for primary verification.
  • Epidermal transit time “20→30-35 days” framing: now consistent at “>30 days” across tissues/epidermis.md body + table. Keratinocytes.md retains the gap on the upper bound.

Forward queue surfaced

Items implied by R38 verification that should be considered for downstream rounds:

  • studies/fisher-1996-photoaging-ap1-mmp.md (R39) — definitive resolution of the “70%/50-80% in figures vs text” inter-verifier contradiction.
  • studies/inomata-2009-melanocyte-stem-cell-atm.md (R39 or follow-on) — primary source for the McSC ATM mechanism (download triggered during R38 verification).
  • molecules/proteins/lox.md — promoted from R40 implicit reference to explicit need; dermis.md LOX decline claim is currently #gap/unsourced and needs primary citation.
  • microbiome/skin-microbiome-aging-shifts.md (R44) — surfaced as implicit absence on tissues/skin.md; can defer.
  • molecules/proteins/scf.md / [[kitlg]] (R40) — Hattori 2004 paradigm needs the protein anchor.
  • phenotypes/cutaneous-immunosenescence.md — surfaced by langerhans-cells.md as missing phenotype page; could be a small follow-on.
  • pathways/il-34-csf1r-axis.md — LC niche signaling not currently anchored.

Round disposition

R38 closed cleanly. 8/8 pages flipped to verified: true. ROADMAP.md updated to reflect closure. Ready to dispatch R39 (foundational primary-source studies, 11 pages).

Graph View

Backlinks