The source identity is human HGNC:1482, UniProt O15484 (CAN5_HUMAN), a reviewed 640-residue calpain-5 record. CAPN5/NCL3 must be distinguished from the catalytically inactive paralog CAPN6. The normal source seed has 14 GO annotations and no declared alternative products. Every original term, evidence code, reference, qualifier and supporting entity is retained; no new GO assertions are proposed.
CAPN5 is an atypical calpain with a cysteine-protease core, a calpain beta-sandwich region and a C-terminal C2 domain, rather than the penta-EF-hand architecture of classical calpains. Human full-length CAPN5 expressed in SH-SY5Y cells undergoes calcium-dependent autoproteolysis; the C81A active-site substitution and EGTA inhibit that processing PMID:33811937. Both tagged and untagged protein processing were reported. Apparent half-maximal activity in cell lysates was about 80 micromolar calcium when measured by disappearance of the full-length band, or about 200 micromolar when measured by appearance of a fragment. These are assay-dependent estimates, not a native intracellular activation threshold. The isolated recombinant protease core has different calcium requirements and is not a substitute for the complete protein PMID:31968260.
CAPN5 occupies intracellular soluble and membrane-associated pools. C2-domain acidic residues contribute to membrane association, and lipid modification at cysteine residues helps stabilize that association [PMID:33811937; PMID:35643222]. Direct calcium binding to the C2 domain was not established in the 2021 experiments; their authors explicitly discuss electrostatic and structural alternatives to a calcium-binding explanation. The 2022 abstract identifies Cys4 and Cys512 as S-acylated and qualifies Cys507 as likely. No independent calcium-binding, lipid-binding, membrane-trafficking or chaperone molecular function is added on these data.
The original disease study localized tagged wild-type CAPN5 near the cell boundary in permeabilized HEK293T cells PMID:23055945. The later full-length human study measured colocalization with the plasma-membrane marker ATP1A1 PMID:37207905. These observations support plasma-membrane association but do not establish extracellular exposure. The official GO definition of cell surface specifically refers to the external part of the cell wall and/or plasma membrane; plasma membrane includes associated proteins. The existing cell-surface annotation is therefore refined to plasma membrane, preserving its original GOA term and reference in the source fields. Broad cytoplasmic localization remains compatible with soluble, membrane and context-dependent nuclear pools.
The biochemical effects of vitreoretinopathy variants depend on the variant and construct. In full-length human CAPN5 assays, R243L, L244P, K250N, R289W and G376S reduced membrane association, S-acylation and calcium-triggered processing; G267S differed, with modestly increased activity PMID:37207905. Calcium-treated co-immunoprecipitates supported AIRE cleavage with WT and G267S, but not C81A, R243L or R289W. This is evidence from an in-vitro assay, not an established physiological retinal substrate pathway. The proposed dominant-negative mechanism, self-association and structural explanations remain hypotheses requiring further validation. G267S and G376S disease causality was itself qualified in that paper. Blanket statements that every CAPN5 variant is hyperactive are not justified.
The 2026 study supplies independent target-specific support for a CAPN5–STIP1 association: endogenous STIP1 co-immunoprecipitated with WT full-length human CAPN5-3xFLAG and was detected by anti-STIP1 immunoblotting PMID:41597217. STIP1 resisted calcium-induced proteolysis under the tested conditions. Co-complex recovery does not establish a direct binary interface or make CAPN5 a chaperone. Discovery proteomics had three biological replicates; the confirmatory co-IP and cleavage assays were single experiments with extensive controls. The two historical BioPlex binding rows are retained as non-core using this independent corroboration, while their original target-table access limitations remain explicit [PMID:28514442; PMID:33961781]. This follows the ClinGen project rule permitting supported generic protein binding as non-core when no more informative CAPN5 molecular function is established.
The 2024 myoblast work concerns mouse C2C12 cells and murine CAPN5 expressed in human MCF-7 cells for crude-lysate biochemical assays PMID:39357823. Knockdown and rescue support context-specific CAPN5 involvement in differentiation and STAT3 dynamics. The inspected cleavage experiments lack a catalytic-dead CAPN5 control or purified enzyme/substrate reconstitution. These results do not justify a human myogenesis assertion. Likewise, in-vitro cleavage of candidate signaling proteins in the 2026 study does not establish a downstream human cellular signaling response. The historical signal-transduction TAS row remains unresolved rather than being removed.
The complete available normal paper was read for PMID:27152965. Mouse photoreceptor immunolocalization with PSD95 colocalization is distinguished from bovine retinal fractionation and rat brain fractionation. The machine transfer names mouse O08688/ENSMUSP00000048183; O08688 is independently identified as mouse Capn5 in the quarantined normal family export. A mammalian localization observation is not a direct human synapse assay or evidence for a new synaptic-transmission process.
The proposed minimal core is calcium-dependent cysteine-type endopeptidase activity, directly carrying out proteolysis in intracellular soluble and membrane-associated pools. The decisions are six ACCEPT, three KEEP_AS_NON_CORE, one MODIFY and four UNDECIDED; none is PENDING, REMOVE or NEW.
Focal-adhesion and urinary-exosome target identifications remain unresolved because the specific CAPN5 supplementary records were not recovered [PMID:21423176; PMID:19056867; PMID:23533145]. The focal-adhesion paper discusses calpain-2; that mechanism is not transferred to CAPN5. General contamination cautions in the proteomics papers do not demonstrate that CAPN5 is a contaminant. The source HDA assertions are retained, with no claim of absence, misattribution, secretion mechanism or core exosome function.
The 1997 cloning paper covers both CAPN5 and CAPN6 PMID:9339374. Its abstract's predicted catalytic inactivity concerns CAPN6. The CAPN5 activity row is retained with deference to the original TAS curation and corroboration from later direct human CAPN5 assays; the inaccessible historical experiment is not claimed as independently verified. Its broad signaling statement remains UNDECIDED. The accepted IBA activity/proteolysis/cytoplasm assignments are consistent with target evidence and are not evaluated by donor-list length or by treating target self-inclusion as circular.
A normal provider research attempt was made once in isolated storage. Both the configured Falcon service and its Perplexity fallback failed to connect; no provider-generated research file was produced. These manually written notes record the primary-source research instead. No manual text is presented as a provider-generated deep-research artifact.
No new process terms are proposed. The GO-CAM index had no matching CAPN5/O15484/O08688 entry in the inspected local snapshot; this is not evidence of universal absence. Family exports were quarantined and used only for bounded identity checks. All normal GOA, UniProt and publication bytes are protected from modification, including the pre-existing PMID:32900848 cache that is not used to establish a new assertion here.
Which retinal substrates are cleaved by CAPN5 under physiological calcium conditions? How do full-length and isolated catalytic-core constructs alter the effects of individual variants? Can the original BioPlex and proteomics target records be recovered, and can STIP1 coassociation be resolved into a direct or indirect mechanism?
The 2021 experiments positively establish calcium-dependent CAPN5 self-cleavage. Human full-length constructs expressed in SH-SY5Y cells were compared with catalytic C81A; EGTA blocked calcium-dependent fragment production after maitotoxin treatment and cell handling. Untagged processing was also reported. The complete selected Methods and Results and the relevant Discussion were reread PMID:33811937. The authors explicitly leave the physiological relevance of this self-cleavage unresolved. Fragment production does not establish acquisition of a mature functional form, a particular native signaling pathway, an exact cleavage site, or a cis-versus-trans mechanism.
Official ontology definitions distinguish protein autoprocessing, GO:0016540, which descends from protein processing and protein maturation, from calcium-dependent self proteolysis, GO:1990092. The latter describes the observed event without a maturation claim; its parent is self proteolysis and then proteolysis. The existing electronic proteolysis row is therefore MODIFY to GO:1990092 using the new target-specific biochemical support, not because an InterPro domain by itself implies self-cleavage. The broad PAINT proteolysis assertion remains ACCEPT and preserves its phylogenetic scope and source fields. Other substrate cleavage is not excluded. The core lists only the specific demonstrated self-cleavage process, avoiding an ancestor/descendant pair in that synthesis. No NEW descendant is added to an already represented proteolysis process. The local GO-CAM index has no CAPN5/O15484 entry; this absence is not used as evidence for an annotation gap.
The requested signal-transduction downgrade was reassessed against the complete accessible 1997 abstract and primary publisher access. The opening general calpain hypothesis is insufficient positive support for CAPN5 signaling, but the full original text and curator's passage still were not recovered PMID:9339374. It would therefore be inaccurate to claim that the whole historical basis is visible or disproven. The source-specific TAS remains UNDECIDED, with an explicit distinction from the sibling protease assertion that has independent direct human corroboration. Later murine signaling and human coassociation/cleavage studies do not settle the historical human process. No wrong-paralog or source-misattribution claim is introduced.
The focal-adhesion and two exosome identifications retain their actual supplementary-target access limits; the two independently corroborated STIP1 associations retain their original BioPlex access qualifications. These are recoverable evidence questions, not proof of absence and not a reason to convert uncertainty into a biological rejection. The prior dated notes remain an unchanged prefix. The current decisions are five ACCEPT, three KEEP_AS_NON_CORE, two MODIFY and four UNDECIDED, with all 14 machine assertions, 17 reference identities, no declared alternative products, one catalytic core and zero NEW annotations preserved. DRAFT records the four unresolved assertions.