CAV3 evidence review

Identity and source boundary

CAV3 is human caveolin-3, reviewed UniProtKB P56539 (CAV3_HUMAN), HGNC:1529, with a 151-amino-acid canonical sequence. A8K777 and Q3T1A4 are secondary accessions. No alternative products are declared in the normal source record. The note about uncertain initiation at the first or second methionine is not a basis for inventing product objects. Caveolin-3 must be distinguished from calcium-channel names such as CaV3.2.

The normal source acquisition returned 117 raw GOA rows and 116 distinct seeded assertions. The two raw copies behind the RSV matrix-protein interaction differ in assignment/date provenance; their common machine assertion is represented once, exactly as seeded. The review preserves every machine term, evidence code, original reference, donor/partner list and source relationship. Ordinary relationship qualifiers are not used to decide biology; NOT and contributes_to retain their special meaning.

The frozen ClinGen cohort contains autosomal dominant caveolinopathy as Definitive and autosomal dominant long QT syndrome as Limited. Those original associations and dates remain unchanged. Historical channel papers and UniProt disease language do not upgrade the Limited long-QT association. Molecular evidence and clinical gene–disease validity are separate questions.

An isolated normal deep-research attempt failed in the configured provider path; its actual log and failure are retained in TMP. No provider-branded research file was authored. This review instead uses the immutable normal caches, bounded primary-source access, official ontology definitions and a distinct sodium-current consultation. Canonical application, standard CREATE history, final rendering and publication require normal checks, distinct whole-science review and ROOT acceptance.

Caveolar organization and muscle disease

The muscle membrane role is supported by CAV3-dependent caveola abundance, human sarcolemmal localization and cavin interactions. The cardiac overexpression study uses mouse transgenics and rodent cardiomyocytes, with blinded ultrastructural measurements; its caveola measurements are separate from infarct protection, Akt/GSK signaling and pharmacological inhibition. The human GOA source evidence code remains untouched, while the experimental species is explicit. Increased expression is not assumed to be uniformly beneficial PMID:18936328.

The original dominant muscle-disease report describes eight patients from two families with severe muscle CAV3 deficiency. Its abstract proposes altered oligomerization/caveola formation as a mechanism; that proposal is not presented as a completed oligomer-structure experiment PMID:9537420. The rippling-muscle-disease family report supplies clinical context but its accessible abstract does not resolve the exact contraction mechanism PMID:17524427. Human myopathy-associated CAV3 constructs show variant-dependent surface/Golgi localization and different effects on growth-factor signaling, with an important distinction between wild type and retained mutants PMID:20472890.

The PTRF study establishes secondary caveolin deficiency in human muscle and uses sequence-verified human CAV3 in reciprocal association experiments. PTRF disease is not attributed to CAV3 merely because CAV3 protein is affected PMID:19726876. CAVIN4 interaction and trafficking experiments likewise separate binding, localization dependency and downstream ERK/hypertrophy effects; a deletion can preserve CAV3 association while changing its trafficking [PMID:24567387; PMID:26497963].

The broad adaptor rows are refined to protein-macromolecule adaptor activity because the hERG/NEDD4-2 experiments establish coordinated recruitment of protein partners. This is not a claim that every pairwise interaction is an adaptor mechanism. A stable caveolar structural scaffold is conceptually distinct; an additional scaffold annotation is left as a question pending a complete target-specific and same-role comparator assessment. No NEW assertion is introduced merely to fill a core-coverage advisory.

Channel mechanisms and physiological limits

The calcium-channel study uses human CAV3 and human CaV3.1/CaV3.2 constructs in HEK293 cells and mouse cardiomyocytes. Both channels associate with CAV3, but the functional decrease is selective for CaV3.2 in the tested conditions. Surface abundance was not reduced, and the N-terminal CAV3 region participates in association. CAV3 depletion removes cAMP-dependent augmentation without simply lowering basal T-type current. These results support calcium-channel regulator activity, without equating CAV3 with the pore, a calcium sensor or an inhibitor of every calcium channel PMID:21084288.

The potassium-channel study combines heterologous hERG experiments with native rat/rabbit association and neonatal rat depletion. CAV3 recruits NEDD4-2 and promotes turnover of mature surface hERG; channel PY-motif mutations and ligase depletion limit the proposed mechanism. Current changes reflect surface-channel abundance rather than a demonstrated direct pore block or altered voltage-dependent gating. CAV3 is not the ubiquitin ligase or a protease. The inhibitor term permits a binding-dependent reduction in channel function and does not require physical pore occlusion PMID:22879586.

The sodium studies were evaluated using the canonical caches and the accessible original primary article. Human CAV3 and human SCN5A/Nav1.5 were tested in HEK293 cells. F97C/S141R increased persistent late current in the first study, but wild-type CAV3 did not measurably alter current versus vector. The second study used tagged human CAV3 variants with SCN5A Q1077del; the variant increases were significant, the smaller wild-type increase was not, and peak-current/tested gating differences were not significant. Controlled coimmunoprecipitation supports association, with mutants retaining it; it does not prove a purified direct interface [PMID:17060380; PMID:17275750].

Human right-ventricular sections in the first study directly support sarcolemmal localization. Its heart-rate discussion, however, extrapolates an untested T78M variant from a separate SCN5A pacemaking mechanism. That specific heart-rate extension is marked over-annotated. This does not dispute the measured late-current result or assert that CAV3 can never influence heart rate PMID:17060380.

Ventricular repolarization, depolarization and contraction-linked annotations remain undecided at the source-specific physiological level. A native cardiomyocyte or whole-heart assay is not an absolute prerequisite: a sufficiently demonstrated mechanism could support a process inference. Here the exact wild-type/native mechanism and the connection from mutant late current to those particular processes remain unresolved. The anonymous SIDS cohort and the clinical limitations do not erase the molecular result, but neither do they establish every downstream physiological process [PMID:17060380; PMID:17275750].

The sodium-current consultation was provided by a distinct reviewer and disclosed as a bounded evidence contribution. The author independently read the relevant original Methods, Results and Discussion; figure pixels and supplementary material were not reanalysed. No verified formal correction was located in the bounded search, and search completeness is not claimed. Manuscript typography is not treated as a formal correction or permission to rewrite machine identifiers.

Dystrophin association and membrane repair

The beta-dystroglycan study supports association with the cytoplasmic tail and competition involving dystrophin. A separate biochemical study finds that residual CAV3 separates from the stable purified dystrophin complex, and that CAV3 sarcolemmal localization persists independently of dystrophin/sarcoglycan defects. Stable integral membership and transient association are different claims. The current GO complex definition explicitly includes caveolin-3; the existing component and complex-binding annotations are retained non-core with the distinction stated, rather than treating the purification result as universal absence of interaction [PMID:10988290; PMID:9607328].

Dysferlin association is independently reported in normal human muscle biopsies, with abnormal localization in affected muscle. The historical WW-domain/motif proposal is not accepted as verified CAV3 architecture PMID:11532985. The complete repair paper uses mouse Cav3 constructs, C2C12 cells and mouse muscle. Controlled association, MG53/dysferlin mislocalization, transgenic and acute P104L perturbations, and R26Q/C71W comparisons support coordinated trafficking and repair effects. MG53 is the injury-responsive recruiter in those experiments; CAV3 is not thereby shown to catalyze membrane fusion or sense oxidation. Wild-type overexpression also has effects, so mutant comparisons do not justify a simple universal dosage rule PMID:19380584. The complete Reactome event provides a human pathway context for the DYSF–CAV3–TRIM72 assembly, with its underlying experimental species kept explicit [Reactome:R-HSA-5263633].

Binding policy, source access and unresolved assertions

The standing project instruction retains supported generic binding non-core or refines it when a specific molecular activity is demonstrated. It does not permit blanket removal because GO:0005515 is uninformative. Exact partners, including the RSV matrix protein and screen-derived associations, remain preserved. Reviewed-UniProt corroboration supports qualified associations without being mistaken for independently read pair-specific control tables [PMID:28514442; PMID:32296183; PMID:32814053; PMID:33961781].

The existing RSV cache explicitly discusses all three caveolins as matrix interactors, while functional replication effects focus on CAV1/CAV2. The review does not infer a CAV3 viral-process role from those paralog effects. A normal HTTP429 preserved the old cache; only the seeded placeholder title is corrected in the authored reference list PMID:25556234. The TRIM50 paper's title is not evidence of a wrong-gene citation: the exact CAV3 endoplasmic-reticulum observation was not located and remains undecided PMID:22792322.

Canonical PMID:19726876 and PMID:24567387 remain abstract-only although fuller authenticated artifact bodies supplied selected primary reading. PMID:33961781 and PMID:25556234 have body-available flags but only partial cached prose; the review does not call that complete Methods or target-table access. All such distinctions are stated in the reference assessments. Sources were never edited to harmonize flags or replace an existing cache.

The PAINT assertions represent ancestral-node judgments, not donor votes or pairwise similarity transfer. A target in WITH/FROM is legitimate descendant evidence. No node is rejected because its donor list is short or contains CAV1. Several developmental, metabolic and location transfers remain undecided because their exact donor assay or ancestral scope was not verified. The official MGI comparative graphs supplied provenance leads; those leads are not claimed as independently read donor experiments.

The audit table below preserves every undecided row and identifies its exact source/partner for follow-up. An IEP-derived lead is not intrinsically invalid, and an abstract that foregrounds another gene is not proof of misattribution.

2026-10-04: first external-review follow-up

This follow-up reassesses the requested channel, dystrophin-complex, differentiation and donor-source issues individually. The standing ClinGen project instruction in projects/CLINGEN_MENDELIAN.md retains supported generic protein binding as non-core when a more informative activity has not been established. It takes precedence over the general removal guidance cited in the external review. It does not verify inaccessible target experiments, require a quota of retained interactions, or permit an activity to be invented from a partner's function. The machine annotation objects, exact partners, source identifiers and existing source bytes remain unchanged.

The DAG1 interaction is refined to the existing protein-containing complex binding term, consistent with the same-source curated complex-binding assertion. The reported beta-dystroglycan tail interaction provides the physical link to that assembly; no stable obligatory membership, solved adaptor bridge or verified WW-domain architecture is added. The other ten generic-binding decisions remain qualified, source-specific non-core associations. In particular, BioPlex/HuRI database corroboration is not described as an independently inspected target-control table, and the RSV study's CAV2 functional emphasis does not erase its reported CAV3 interaction.

The two sodium-channel regulator rows are retained non-core and the sodium core is removed. Human CAV3 variants change late SCN5A current in HEK293 systems, but wild-type baseline effects were not significant in those studies. This retains the molecular observation while separating it from a demonstrated normal physiological CAV3 role. The external full primary source used for the first sodium study is PMID:17060380, Circulation author PDF; selected Methods, Results, Discussion and captions were read, without image reanalysis. The cached abstract-only flag remains unchanged. The frozen ClinGen long-QT Limited classification is also unchanged.

The potassium inhibitor reassessment uses the actual GO:0019870 definition, which does not require pore occlusion. Its regulator and inhibitor ancestors (GO:0098772 and GO:0140678) specify noncovalent regulation without resulting covalent target modification. The mechanism demonstrated in PMID:22879586 recruits NEDD4-2, causes mature surface hERG ubiquitination/internalization, loses its effect after disrupting the hERG PY motif, and reduces surface-channel abundance. That covalent-modification-dependent turnover mechanism supports adaptor and trafficking regulation rather than the inhibitor MF. The two transferred inhibitor MF assertions are therefore over-annotated, without denying the current phenotype or asserting that unchanged gating excludes every inhibitory mechanism. This supersedes the earlier notes' broader inhibitor interpretation.

The dystrophin-complex decision remains non-core because the official GO:0016010 definition explicitly lists caveolin-3 among components. PMID:10988290 supplies partner association, whereas PMID:9607328 separates CAV3 from the stable purified complex and demonstrates independent sarcolemmal localization. These findings constrain stable integral membership rather than refuting every association. The decision rests on that biological scope and definition; the machine relationship column is ignored, as required by the review skill.

The differentiation split is source-specific. The IBA records a PAINT ancestral judgment and is retained with curator deference, without reconstructing or inventing its node placement. The rat IEA/ISS assertions trace to PMID:16135403, an expression study during induced astrocyte-like differentiation of C6 glioma cells. Its complete abstract is accessible, but the full experiments remain unavailable. These two transfers remain undecided; an expression annotation is not rejected merely because it uses expression evidence, and it is not silently recast as muscle differentiation.

The bounded donor investigation resolves several previously uninspected experiments. PMID:16480946 explicitly tests adeno-Cav3 against GFP in rat vascular smooth muscle cells and measures polymerized tubulin; the existing two transfers become non-core with overexpression limits. Its abstract's peptide and knockdown claims are not separately attributed to CAV3 without the missing construct details. PMID:21508095 directly depletes CAV3 in primary rat folliculostellate cells on laminin and measures proliferation. PMID:15725944 reports an ionomycin-induced CAV3/caveolar response in rat cardiomyocytes. Neither becomes a core muscle function or a new human experimental assertion.

PMID:15314230 supports mouse glucose and cholesterol homeostasis phenotypes. Its circulating triglyceride measurement is distinguished from triglyceride chemical reactions; the latter process is marked over-annotated rather than inferred from concentration alone. No new metabolic activity is proposed.

The description and first core now emphasize muscle caveolar coat organization and recruitment, including dysferlin/MG53 repair context. The existing normal references PMID:11532985 and PMID:9537420 supply short, cache-verbatim anchors for human muscle association and muscle caveolin deficiency. The calcium-channel core's anchor now reports the positive CaV3.2 result, replacing a negative CaV3.1 comparison. Informative anchors are budgeted per source; a quotation is not treated as a substitute for construct, species or assay interpretation.

All five donor references are now assessed from their exact normal caches; each cache remains abstract-only. External primary body access is recorded separately for each source, without changing those flags. The final decisions are 22 ACCEPT, 58 KEEP_AS_NON_CORE, 28 UNDECIDED, four MODIFY and four MARK_AS_OVER_ANNOTATED. The review retains two core functions and proposes no NEW assertions. Normal validation, status reconciliation, rendering, distinct scientific review and ROOT acceptance govern canonical application. A new standard EDIT history records the accepted change; prior notes and histories remain append-only.