Review date: 2026-10-27. Row numbers below are zero-based positions in the 49-context imported review. These locators describe the material actually inspected; they do not claim unread supplements or unavailable originals were checked. Reference-only support is intentional under the ClinGen campaign's no-new-publication-quotation instruction.
The unchanged UniProt record has FUNCTION at lines 95–99, citing PMID:27164707; SUBUNIT at line 100 is ortholog inference. INTERACTION lines 101–118 record the exact pairs, including target and partner isoforms, with IntAct identifiers. Nuclear localization at line 119 is ECO:0000305. Alternative products at lines 120–125 distinguish FGF12A/P61328-1 from FGF12B/P61328-2; VAR_SEQ lines 278–283 replace residues 1–66 with MESK, giving 181 residues from the 243-residue canonical sequence. This arithmetic does not identify every historical experimental construct.
Rows 2–16 retain the imported experimental interactions from PMID:25416956 and PMID:32296183. Each exact pair matches the curated INTERACTION block (NbExp=3). The screen identities and database provenance were inspected; individual original supplementary clone rows were not independently extracted. The papers' manual assessments remain UNVERIFIED at that finer level. No partner function is transferred to FGF12.
UniProt also cites PMID:14702039 for both products and PMID:15489334 for the short product. Their exact collection clone/sequence rows were not inspected; they are low-relevance sequence provenance, not functional support. The MGC cache contains body text despite false full-text metadata, which was preserved.
PMID:22705208 — the cached original, Figure 3, Table 2, corresponding Results and construct Methods. The solved crystal complex uses FGF13, while independent calorimetry, co-purification and functional experiments use human FGF12B NM_004113 with NaV1.5. Those target-specific results support row 1's refinement and the channel-regulator core; no FGF12 crystal structure is claimed.
PMID:12401812 — author-hosted original, Methods molecular constructs, binding material and abstract. Rat FHF1B AF348446, human NaV1.5 and HEK293 cells are distinct components. The reported hyperpolarizing inactivation result must not be generalized to all FGF12/channel combinations. This supports ISS transfer on rows 41, 44 and 47. The NAS cardiac action-potential context remains non-core: no contracting human cardiomyocyte action potential was recorded here.
PMID:27164707 — original article, indexed Methods, variant identity, zebrafish Results/Figure 4 and normal abstract. The Neuro2A channel experiment is distinct from zebrafish B-type ortholog overexpression. Complete electrophysiology Results were not independently retrieved; the preserved UniProt FUNCTION entry records curation of the same study, not independent replication. The study supports excitability/channel regulation, not a general inference from disease severity to every developmental term.
PMID:12815063 — author-hosted original, Protein Purification, Results Figures 1–3 and IB2 discussion. The human short FHF1b preparation includes C144A. Heparin affinity and binding to a murine IB2 fragment are positive. Figure 2 measures proliferation/antagonism, so absence of mitogenesis cannot alone prove absence of every receptor-binding or signaling event. IB2 binding does not identify a JNK-specific step; the discussion concerns p38 recruitment. Rows 0/38 retain NOT and P61328-2 with UNDECIDED reviews; row 27 is also unresolved, while heparin binding remains non-core.
PMID:32357892 — official full original, Methods, Results Figures 1–3 and Discussion, plus the normal cached full text. Short human FHF1b binds FGFRs in direct assays and produces receptor-dependent signaling and survival responses in recipient cells. The study does not find a mitogenic response in its tested systems. These measurements establish assay-specific receptor activity, but do not by themselves establish physiological cell-cell signaling. They also require reconciliation with the historical negative receptor contexts; absence of proliferation is a different endpoint. Construct and endpoint differences require explicit comparison.
1996 discovery and localization — PMID8790420. The original PNAS article was read from all eight scanned pages, 9850–9857. Human retinal cDNA supplied FHF1; mouse expression mapping and macaque immunohistochemistry are separate experiments. Methods p. 9851 and Results p. 9856/Figure 8 demonstrate nuclear accumulation of expressed human FHF1 in 293 cells and map amino-terminal targeting with artificial deletions and β-galactosidase fusions. Those constructs are not natural splice products. The text calls FHF1 244 residues while Figure 1 labels its sequence 243; no exact current-product sequence match is asserted. Figure 7 compares cell and medium fractions, with secreted growth hormone as a control: FHF1 was not detectably released under those conditions. This does not exclude secretion of another product or in another setting. Figures 4–5 establish mouse neural expression; Figure 6’s macaque antibody staining has acknowledged possible FHF-family cross-reactivity. Discussion p. 9857 explicitly proposes receptor signaling and developmental/growth roles for future functional testing. Nuclear localization is supported; the other historical source claims remain unresolved pending matched independent evidence.
1999 heart-derived cDNA — PMID10049777. The published abstract and identity were verified, but bounded publisher/public searches did not recover the full article. The official U76381.2 sequence record explicitly identifies FGF12b, cites this PMID and records a July 1999 sequence update. Thus, the abstract’s 168-residue and FGF9-homology wording does not establish misattribution. The updated historical CDS is not silently equated with either current natural product. Heart-enriched expression alone establishes neither cardiac development nor growth-factor activity. Extracellular location and signaling assays remain inaccessible; the five source claims retain uncertainty.
2022 nucleolar interactions — PMID36411431. Relevant Methods, Results, Discussion and captions of the full article were read. Methods, “Cells” (cached lines 100–102), specifies tagged a isoforms in human U2OS cells; recombinant-protein Methods use short-chain numbering without a resolved clone-to-current-product sequence join. Preserve row17’s imported P61328-2 context without treating every experiment as isoform-2-specific. Figure 1D–F supports nucleolar localization; Figures 2–3 support FGF12 association with NOLC1 and TCOF1. Endogenous HEK293 PLA, Supplementary Figure S4, is not isoform-resolved. Figure 4 and its Results show loss of NOLC1–TCOF1 PLA after FGF12 depletion. Discussion’s final paragraph leaves ribosome-biogenesis versus DNA-damage-response involvement unresolved. Abstract/Results partner comparisons differ, and Figure 4B’s caption says FGF2 despite surrounding FGF12 text; neither discrepancy erases the consistent FGF12 interaction evidence. Figure pixels and supplementary data were not inspected in this restart read.
2015 platform biology — PMID26399832. The full article, relevant Methods and all ten extended-data captions were read. Human embryonic-stem-cell and Xenopus experiments concern CUL3/KBTBD8 and the NOLC1/TCOF1 platform. No named FGF12 experiment was identified in the cached body; uninspected supplementary tables are not claimed to lack one. Partner functions cannot establish FGF12 neural-crest participation. Positive platform assembly evidence also leaves FGF12’s ribosome-biogenesis attribution unresolved.
2024 galectin regulation — PMID38468333. The full original Methods, Results/Discussion, Conclusions and all captions were read, with Figure 4–5 pixels inspected. Bacterial-protein SPR supports direct FGF12–galectin binding. In the inherited tagged human U2OS model, galectin-1 depletion changes three pairwise PLA signals and increases detected extracellular FGF12. Results’ final subsection/Figure 4 perturbs galectin-1, not FGF12; the earlier FGF12-depletion result is cited from 2022. Figure 5 remains a hypothetical model with unresolved nuclear function. A scaffold could participate without catalysis, but productive ribosome biogenesis is not established by these interaction endpoints. Supplementary PDF retrieval failed; its contents were not assessed.
The compact database evidence records the actual official ARBA, InterPro and PANTHER sources, their acquisition receipts, target-specific projections and a read-only verifier.
The three ARBA outputs are genuine target matches, not guesses from an FGF family name. P61328's FunFam 2.80.10.50:FF:000001 matches ARBA00028183 with Haplorrhini, ARBA00027428 with Homo, and ARBA00093068 with Primates. IPR002209 independently maps the canonical target to growth-factor activity. These traceable mapping mechanisms do not establish all biological outputs by experiment.
The four positive IBD joins have actual target ancestry support: leaf PTN002479516 lies below PTN000160075 and PTN000160077. The canonical alignment matches all 243 residues. The tree download was incomplete, but its received prefix contains the complete target path and leaf; the full MSA response was available. Only compact projections are included with the gene, so independent extraction requires the original captures identified by their hashes and URLs. No full-tree review, rereview of every descendant paper or function of both splice products is claimed.
The electronic nuclear mapping comes from UniProt SL-0191 with ECO:0000305. It is distinct from the experimental localization evidence in PMID36411431: tagged a-form constructs in U2OS cells and complexes in HEK293 cells lacking ectopic FGF12. The shorter product must not be assigned the same localization merely from the gene-level mapping. Cytoplasmic activity and nuclear localization can coexist in different protein pools and contexts.
Cell-cell signaling is assessed against the actual extracellular receptor-response evidence, including PMID32357892, rather than inferred from the absence or presence of a signal peptide. That paper's receptor binding and downstream signaling are distinct from mitogenesis. GO:0008083 specifically requires stimulation of growth or proliferation. With the target-lineage IRD added, the electronic mapping and fully read 1996 hypothesis are flagged as over-annotation; the inaccessible 1999 source retains uncertainty. The two original P61328-2 NOT source objects remain intact; later positive receptor results require comparison of species, sequence, substitutions and assay endpoints before resolving their polarity.
Nervous-system development and neurogenesis remain non-core inherited roles. The target path supports the ancestral judgment; neural expression and altered excitability are not relabeled as independent measurements of neuron generation.
The entire 35-row PTHR11486 PAINT table was read after the initial review omitted its negative half. Seven IRD (Inferred from Rapid Divergence), negated=true assertions occur at the verified target ancestor PTN000160077: GO:0005576, GO:0005104, GO:0008083, GO:0008284, GO:0008543, GO:0030334 and GO:0043410. Each names PTN000160075 as its seed, cancelling the corresponding ancestral positive inference. They are curated evolutionary-loss judgments, not seven independent negative experiments. The export does not identify a PMID or negative experimental donor for these rows, so independence from the historical studies is unestablished.
The loss of growth-factor activity is now reflected in rows 34 and 36 as MARK_AS_OVER_ANNOTATED. Row 35 remains UNDECIDED because its operative 1999 source is inaccessible. Rows 0 and 38 retain their NOT/isoform source fields and unresolved reviews: the later direct binding and receptor-signaling results in PMID:32357892, Figures 1F–H and 2A–E, conflict with treating the negative claim as settled; Figure 3A separately confirms absent proliferation. Physiological secretion is not required for an in-vitro biochemical binding measurement.
For extracellular row 18, PMID:38468333 Methods (cached line 137) and Figure 3B caption (line 167) measure tagged FGF12 recovered from stressed U2OS culture medium, with increased recovery after galectin-1 depletion. This does not establish ordinary physiological secretion or resolve the unread 1999 short-product source, but prevents a universal non-release assertion based on the 1996 assay, the IRD or absence of a signal peptide. The cell-cell-signaling IEA row 28 is now UNDECIDED: actual ARBA rule matching, exogenous-ligand responses and a separate tagged release assay do not establish an endogenous producer-to-recipient signaling chain.
There is no literal GO:0007267 cell-cell-signaling IRD in this TSV. Specific receptor-pathway losses do not by themselves negate every broader signaling process; broader GO-slim NOT summaries are not additional experimental negatives. The compact verifier now classifies every literal family assertion, including the 17 rows off the target path.