Q8N726 encodes ARF, an unrelated alternative-reading-frame protein of the CDKN2A locus. Its 72 normal seeded annotations and six machine-fetched product descriptions are preserved. The displayed 132-residue p14ARF and internally initiated smARF (residues 48–132) are not collapsed into p16INK4A. The external P42771 product descriptions remain source metadata rather than functional classes in this review. No source isoform tags were invented. The ordinary fetch completed before manual review; the genuine Falcon attempt timed out and the genuine Perplexity fallback returned an authentication/quota failure. No provider research report was fabricated.
Full-length human p14ARF inhibits MDM2 and supports p53 stability, arrest and senescence. The core uses PMID:18560357 human p14ARF comparator experiments, independently supported by PMID:9724636 and the human mammary-cell study PMID:14966292. Nuclear, nucleoplasmic and nucleolar annotations remain coherent with that mechanism. This is distinct from p16 direct CDK inhibition. The new review does not require smARF to share the full-length mechanism.
The standing ClinGen instruction retains supported generic binding as non-core when no justified finer MF is available. Thirteen source-preserving refinements distinguish ligase/TF partner classes, SUMO ligase regulation, a narrower ubiquitin-ligase inhibitor term, and negative regulation of stem-cell self-renewing division. No partner activity is transferred to ARF. Colocalization is not called direct binding; MDM2 rows with limited original assay access use explicitly separate human binding evidence. No NEW annotation is proposed.
The source-specific SUMO refinement uses independent recombinant ARF binding to TRIM28 and enhancement of TRIM28-dependent NPM1 modification (PMID:26055329), without imposing a universal rule about ARF catalysis in every substrate context. The original WRN source remains unchanged. Peptide fibril/disordered-binding annotations remain non-core at their actual human ARF1–14/Hdm2240–254 assay scope; the authentic inventor disclosure supplements an unavailable journal body and is not claimed as independent replication or native cellular amyloid.
Mitochondrial and cytosolic assertions remain non-core with human/mouse, full-length/short-form, transit and mutant distinctions. The sole UNDECIDED concerns mitochondrial matrix precision: the curated Reactome claim is real, but the inspected human fractionation/protection studies do not resolve that compartment. No exclusive mitochondrial localization of smARF is assumed.
The PMID:20208519 editorial Expression of Concern is material and recorded as DISPUTED. It identifies Figures 1e, 3g and 4c and unavailable original data; Figure 2 binding was not specifically named. The association is retained with a separate PMID:27323397 Figure 6B ULF co-IP context; this does not repair or independently validate the questioned images. The article is not called retracted.
Mouse Q64364 donor assertions were checked as actual ortholog evidence, not PAINT/IBA. PMID:15964995 separates Arf from Ink4a knockouts and tests lymphoid phenotypes and neurosphere self-renewal. Its negative regulatory role motivates the stem-cell division refinement, without inventing a human assay. The regulatory enzyme core also has direct human support. No donor-count or circularity argument is used.
Source access, authentic URLs, selected evidence locations and limitations are recorded in CDKN2A__Q8N726-manual-evidence.md. Existing reference caches were selected from exact PR-head or fixed-main blobs; absent records use genuine normal fetches. Fresh full variants were not substituted for existing abstract caches. Twelve initial reference-title placeholders were resolved from genuine normal Reactome caches, without changing any annotation's original reference ID. No authored literature quotations are used in these notes.
This entry supersedes the earlier non-core disposition of amyloid fibril formation, while preserving that historical assessment above. GO:1990000 is now MARK_AS_OVER_ANNOTATED because the demonstrated ARF1–14/HDM2 240–254 peptide system does not establish an unqualified ARF biological fibril-forming process. The primary inventor disclosure records insoluble beta-rich protofibrils at pH 7 and 23°C, followed by long organized bundles after 100 hours at 70°C under acidic conditions. CD/FTIR and EM are positive secondary-structure and morphology evidence; the negative Congo-red/ThT results do not by themselves prove that the material cannot be amyloid. The final PMID:12630860 journal body remains unrecovered, and the patent is identified as supporting primary disclosure. The peptide-binding judgment at GO:0097718 remains non-core and unchanged. The narrower process judgment does not deny the observed peptide assembly or claim that all in-vitro evidence is inadequate.
The unchanged GO:0180016 replacement is verified by the complete official QuickGO response, obtained from https://www.ebi.ac.uk/QuickGO/services/ontology/go/terms/GO:0180016 at 2026-10-10T03:26:32.624005+00:00 (HTTP 200, 338 bytes; SHA-256 7799c9c7854b7ea8c3f6214451a9cef8391aa8be4ebd1d0d2131c506aaecf927). The record explicitly identifies an active molecular-function term named SUMO ligase regulator activity. This provides offline term verification; the mechanistic support remains the separately documented ARF/TRIM28/NPM1 assay, with no new catalytic claim.
The genuine GOA TSV contains 73 records and the normal seed contains 72 annotations. The GO:0005515 / IPI / PMID:20208519 / UniProtKB:Q14669 pair has identical biological assertion fields but different ASSIGNED BY and DATE values: IntAct (20260725) and UniProt (20120824). Those provenance columns are not separate annotation slots in the review schema, so normal seeding collapsed the duplicate without losing a distinct biological assertion. Both original TSV records and all 72 source objects remain intact.