This checkpoint covers the normal human CASK source seed. It is a working scientific record, not a final sign-off. The four additional disease/mechanism caches are being obtained through the normal publication workflow; the final review will integrate their actual recovered contents. No source cache or canonical annotation was changed by this authoring step.
The reviewed human protein is CASK, UniProt O14936 / CSKP_HUMAN, HGNC:1497, NCBI Gene 8573. The canonical displayed sequence is 926 residues. Q5JS79 is a different, unreviewed entry and was not substituted. All six normal products, including their exact VSP sequence notes, are preserved. Legacy constructs and transcript labels are not automatically mapped onto those products. In particular, the 921-residue human construct discussed in PMID:9660868 is not a reason to rewrite the current displayed sequence.
The normal GOA file contains 120 raw assertions and the ordinary seed has 118 source objects. Two pairs differ only in assigning group/date, fields omitted from the normal seed projection: GO:0061045 and GO:0090288, both IMP/PMID:18664494. Every source qualifier, partner, evidence code, original reference, isoform/negation field and term is retained. This is documented in the authenticated source projection, not an author-created biological deduplication. No new assertion or guessed family identifier is introduced.
The active N-terminal protein kinase must be separated from the C-terminal guanylate-kinase-like scaffold region. The full experimental account in PMID:18423203(https://pubmed.ncbi.nlm.nih.gov/18423203/) supports ATP-dependent phosphotransfer and identifies neurexin-tail serine phosphorylation. Recruitment by full-length CASK improves substrate phosphorylation. Hypomorphic S24D/V26L experiments, neuronal manipulations and the biochemical assays provide complementary evidence; the rat and mouse cellular experiments are not described as patient enzymology. Calcium/calmodulin dependence is not established by the gene name. A single protein-serine-kinase core summarizes the demonstrated chemistry without creating a second overlapping kinase core.
By contrast, the complete historical PMID:9660868(https://pubmed.ncbi.nlm.nih.gov/9660868/) states that GUK activity had not been tested. The later PMID:12482754(https://pubmed.ncbi.nlm.nih.gov/12482754/) reports no detectable GMP binding or GMP-kinase activity for CASK-SH3GK, with yeast GK as a positive control. Its complete normal abstract and actual publisher-indexed Results/Table I/Figure 1/Discussion were inspected in the independent mechanism consultation. Full Methods, assayed CASK species and figure pixels were not resolved; PSD95-specific folding/binding controls are not relabeled as CASK controls. These observations support rejecting the four GMP-kinase/GMP-metabolism/GDP-metabolism assertions while retaining the independently demonstrated protein kinase. The original GOA source objects remain intact.
The requested later normal references, PMID:20424264 and PMID:36137748, will refine context. The first concerns engineered ATP-pocket substitutions, not additional native chemistry. The second distinguishes ATP binding and liprin-dependent condensate regulation from direct phosphorylation of liprin by CASK. Neither should be used to erase the 2008 neurexin experiments or to claim a newly established physiological substrate without the appropriate assay.
The molecular-adaptor core reflects coordination of partner assemblies, not the mere presence of a list of interactors. In PMID:16688213(https://pubmed.ncbi.nlm.nih.gov/16688213/), the first and second L27 regions are distinguished experimentally. Human CASK bait, canine MDCK complexes, and worm receptor-localization experiments are separate evidence contexts. The curated LIN7/CASK/APBA1 event and neurexin/CASK/protein-4.1 assembly corroborate the conserved organizing role; their summaries are not newly performed localization experiments.
The full human-domain study PMID:21763699(https://pubmed.ncbi.nlm.nih.gov/21763699/) demonstrates CASKIN1 docking and tests APBA1 and TIAM1 motif peptides. A peptide interaction does not establish a physiological kinase substrate or all functions of a full-length complex. PMID:22117215(https://pubmed.ncbi.nlm.nih.gov/22117215/) specifically tests phosphorylated whirlin peptide binding to CASK SH3-GK; the much more extensive DLG experiments are not transferred wholesale to CASK. The source-specific whirlin row is refined to phosphoprotein binding. The neurexin row is refined to the existing neurexin-family-binding term.
Human GPCR interactome Results and Methods in PMID:26617989(https://pubmed.ncbi.nlm.nih.gov/26617989/) explicitly describe CASK copurification with MAS1, CYSLTR2 and ADRA1D. These are retained as qualified non-core physical associations. Detailed SCRIB/syntrophin pharmacology cannot be attributed to CASK, and a direct receptor-CASK interface is not inferred from tandem-affinity proteomics.
The full PMID:39009827(https://pubmed.ncbi.nlm.nih.gov/39009827/) text reports APC W2658L effects in the peptide-selection dataset, supporting bounded APC motif binding. Its CASK–mutant-MAPT example is a different pair and did not validate as a full-length cellular neoassociation. Similarly, the SLC43A2 experiments in PMID:40355756(https://pubmed.ncbi.nlm.nih.gov/40355756/) cannot stand in for the source row naming SLC16A3.
The 23 source assertions from PMID:36115835(https://pubmed.ncbi.nlm.nih.gov/36115835/) remain UNDECIDED at this checkpoint. The complete relevant Methods and Results distinguish isolated PDZ–motif affinity measurements from cell-extract AP-MS capture. In particular, CASK can copurify with viral motifs despite the lack of detectable isolated-PDZ affinity; the authors propose complex-mediated capture through L27 partners. They also explicitly caution that a subthreshold affinity is not evidence of physical impossibility.
The independent bounded consultation joined all 23 partners to 29 motif records in the paper-linked ProfAff SQL repository. That SQL snapshot was uploaded in July 2021, before the 2022 publication, and has not been authenticated as the final supplementary workbook. Its zero composite values and subthreshold DAPF measurements therefore neither confirm 23 direct binding positives nor justify removing the curated experimental assertions. Exact AP-MS versus holdup versus benchmark provenance remains unresolved. Domain-map residues 489–564 do not authenticate cloned construct endpoints. The final workbook, IntAct record and assay controls are the actionable next evidence; no source field was edited.
The official 2022 correction repairs figure labels and panel presentation. The read notice does not announce withdrawal of the supplementary affinity data. Any future visual assignment should use the corrected figures.
The normal cache for PMID:18664494(https://journals.biologists.com/jcs/article/121/16/2705/30173/The-MAGUK-family-protein-CASK-is-targeted-to) is abstract-only, but the complete official publisher scientific prose, Methods and captions were read separately. Human keratinocyte perturbation supports contextual effects on adhesion, proliferation and growth-factor responsiveness. Nuclear, nucleolar and extraction-resistant pools are retained without inventing transcriptional or RNA-processing functions.
The basement-membrane rows remain unresolved and have different relations: source row 67 says colocalizes_with, whereas row 68 says part_of. Spatial coincidence near an extracellular boundary does not require a secreted matrix constituent, and is weaker than membership in the matrix. Figure pixels and the curator's precise interpretation were not independently established. The wound-healing claim is broader than the measured perturbation endpoints and is provisionally marked over-annotated; tissue repair itself was not measured after CASK manipulation. The distinct review will reassess these two boundaries before final sealing.
PAINT rows remain node-based evolutionary judgments. Donor count is not a quality score, and human CASK among its own supporting descendants is expected rather than circular. Conserved receptor binding, membrane/junction localization, partner positioning and sign-neutral neurotransmitter regulation are retained where coherent with the experimentally grounded scaffold. No new PAINT tree reconstruction is claimed.
Specialized donor compartments—ciliary membrane, vesicle and Schaffer collateral–CA1 synapse—need the actual donor experiments. The direct QuickGO donor queries were inaccessible; that technical result does not establish biological absence. The HPA nucleoplasm record and focal-adhesion proteomics also need their exact antibody/image or target-enrichment evidence. A knockdown effect on focal adhesions is not localization of CASK itself.
For unresolved high-throughput interactions, the required next item is the target-level source row with construct, assay and control information, not another broad review abstract. Complete source headers and available abstracts were checked at intake. Missing target evidence does not justify wrong-gene, wrong-paralog, retraction or miscitation claims. The project permits supported non-core protein binding; it does not require declaring every interaction unsupported simply because it is broad, and it does not waive genuine evidence-access limits.
The official disease reports identify the microcephaly/pontocerebellar-hypoplasia and familial intellectual-disability/nystagmus spectrum. Their normal caches are pending at this checkpoint. The Najm correction specifies CINAP/TSPYL2, not TAF9. The Hackett correction changes the family-123 nucleotide notation to c.2755T>C, with p.W919R retained in that publication's coordinate system. Neither historical coordinate nor study transcript labels are silently mapped onto the current six UniProt products. Disease necessity is not used to manufacture new developmental-process annotations.
This preliminary candidate has 118 decisions, 42 references, six unchanged products, two core activities and no NEW annotations. Normal final validation, exact source/preimage audit, completed cache integration and a distinct whole-science review remain required before sealing. Supporting quotations have not been added in this preliminary draft.
The four requested references are now present as immutable publication caches. PMID:19165920 and PMID:20029458 remain abstract-only; PMID:20424264 and PMID:36137748 provide XML-derived full text. Complete unique Results and Discussion were read for both mechanistic papers, and the available Methods and captions were read for the 2022 study. The 2010 supplementary Methods and original figure pixels were not reanalysed. These access statements supersede the pending-cache boundary above.
The 2010 experiments convert CASK to an engineered Mg-stimulated kinase with four ATP-pocket substitutions. That construct does not establish native Mg dependence. Full-length native and engineered CASK produce similar steady-state neurexin phosphorylation in the tested cells; the kinase-domain truncation is a distinct control. The kinase core retains the 2008 direct chemistry and adds the later corroborating source, without adding an overlapping activity.
The 2022 study tests human variants in HEK293T cells and cultured embryonic rat hippocampal neurons. Coassociation, Liprin W981A and CASK E115K perturbations connect the CASK–Liprin interaction to condensate behavior. CASK coexpression reduces Liprin S87 phosphorylation, and S87 substitutions fail to recapitulate the condensation phenotype. These results support a separable interaction mechanism, not a newly established Liprin substrate or a proven disease mechanism. TV3 and TV5 constructs remain unmapped to current UniProt product identifiers.
The two clinical papers support the standalone disease summary. Their official corrections were read and are described above; no obsolete partner name or unqualified historical variant coordinate is imported. Clinical necessity does not manufacture a developmental GO assertion.
Three precision points were resolved after the preliminary read. Adhesion is described as CASK-directed depletion, without claiming that both siRNAs independently replicated that particular assay. PMID:15694377 compares Id1 splice variants, not CASK products. The protein-transport row now records the selected L27-paper Introduction and Methods that became available; the actual complete transport evidence remains unresolved. The wound-healing rationale concerns the direction of a repair mechanism, without requiring a wound-closure assay as the only possible evidence. The independent epithelial consultation supports that distinction and the separate basement-membrane relations.
All 118 actions remain 27 ACCEPT, 26 KEEP_AS_NON_CORE, 3 MODIFY, 4 REMOVE, 57 UNDECIDED and 1 MARK_AS_OVER_ANNOTATED. The six products and all machine assertion fields remain exact. The review contains 46 references, two core activities and no NEW assertions. Two short verbatim anchors support the adaptor and kinase cores; all other supporting citations remain reference-only. DRAFT records unresolved biological or evidence-access questions, not the validator advisories. Nine supported generic-binding advisories follow the project policy; the cytosol advisory reflects broad electronic localization versus contextual Reactome pools, as explained in the row reasons.
One ordinary isolated deep-research attempt with Falcon and its configured Perplexity-lite fallback failed with connection errors. No provider report was produced or fabricated. This is a manual primary-source review with explicit access limits. Final schema/term/reference/GOA checks, status agreement, rendering and the distinct whole-review receipt are recorded in the sealed handoff; the review makes no claim to have resolved inaccessible target tables or donor experiments.
Reassessment retains 25 previously unresolved qualitative associations after checking each exact partner against the immutable CASK UniProt interaction record. The RPH3A paper explicitly reports the CASK GUK and rabphilin C2B domains with yeast two-hybrid, GST pull-down and co-IP confirmation PMID:11377421. The ID1 paper reports yeast two-hybrid, endogenous ECV304 co-IP and mammalian two-hybrid assays PMID:15694377. These establish more than the prior uncertainty wording acknowledged. ID1 splice variants remain distinct from CASK products. Network associations are retained with exact accession provenance and their source-specific experimental limits: NbExp is not a count of independently reverified controls, and an association does not establish phosphorylation, channel regulation or a direct interface. Source-specific uncertainty can coexist with a well-supported qualitative partner association.
All 60 generic-binding rows were reconsidered individually. Supported interactions remain KEEP_AS_NON_CORE when no stronger molecular function is established, following the user's standing ClinGen instruction recorded in the project. That task-specific instruction intentionally overrides the skill's general informational-exclusion REMOVE policy; it does not override genuine evidence-access limits or authorize invented activities. The three existing specific binding refinements remain. All 23 fragmentomics assertions stay UNDECIDED because the published distinction between isolated-domain affinity and cellular copurification makes exact assay attribution material PMID:36115835. The historical ProfAff snapshot is not certified identical to the final supplement, and subthreshold isolated-domain binding does not establish noninteraction. Broad UniProt records do not resolve that particular ambiguity.
The official full publisher article for PMID:18664494 supplies the detailed evidence beyond the abstract-only cache. Results/Figure 1D describe adult human skin; Figures 2 and 5 concern human HaCaT localization and nuclear extraction, including GFP-CASK and nucleolin. Figure 6 establishes the two-siRNA proliferation comparison; Figure 9 supplies adhesion and mouse-wound observations. Newborn-mouse and rat-follicle experiments are separate. The immutable cache flag remains false, and the added cached quote supports only the proliferation statement it actually contains. Full-size image access failed. Both basement-membrane IDA rows now receive the same anatomical assessment without reasoning from the preserved qualifier column: basal cells and cell membranes do not establish extracellular matrix localization, but uninspected image-level attribution cannot be rejected as a demonstrated curation error. The exact mouse donor experiment remains unresolved for the electronic row; no automatic basolateral replacement is made.
The official publisher Results/Figure 1B for PMID:12482754 explicitly test CASK-SH3GK activity against active yeast GK. This is distinct from the PSD95-only mutation result in the cached abstract; the four existing GMP/GDP rejections remain supported. Complete construct Methods and species are not claimed. Short exact cached anchors now support RPH3A, ID1, whirlin phospho-recognition, protein phosphotransfer, GMP binding and keratinocyte proliferation without inserting web text into canonical caches. The IDA protein-kinase row is refined to GO:0106310 for the mapped serine reaction PMID:18423203; the EC-derived broad row remains valid. ATP binding, neurexin recognition and receptor binding remain accepted mechanistic components of the two existing cores, not redundant standalone cores. The ontology does not need to place the broad and specific kinase terms in a direct ancestor relation for this assay-specific refinement.
Current decisions: 26 ACCEPT, 51 KEEP_AS_NON_CORE, 4 MODIFY, 4 REMOVE, 32 UNDECIDED and 1 MARK_AS_OVER_ANNOTATED. DRAFT records unresolved biological evidence, not validator advisories. All 118 source assertions, six alternative products, two cores, prior reference identities and dated notes are preserved; no NEW assertion is added.
The second current-head review of PR #4132 accepts the standing ClinGen generic-binding policy and requests reassessment of the single basement-membrane IEA assertion. The original source remains GO:0005604, GO_REF:0000107, with UniProtKB:O70589 and ensembl:ENSMUSP00000111098. The official GO definition identifies an extracellular matrix compartment under GO:0031012; it is distinct from a cell membrane and from the basal epidermal cell layer. The official transfer method explicitly propagates experimental donor annotations through Ensembl orthology. Its donor therefore has curated evidence; neither electronic provenance alone nor the absence of an inspected donor image is a reason to reject it.
The revised action is MARK_AS_OVER_ANNOTATED, based on the positive localization biology rather than a claimed negative experiment. The separately accessed official publisher Results and Figure 1 caption describe adult human CASK in the cytoplasm and at cell-cell junctions of the basal epidermis. Newborn-mouse nuclear staining, adult-mouse cell-border staining and rat follicle development remain distinct observations. The immutable reviewed UniProt record describes nuclear, cytoplasmic and peripheral cell-membrane pools. Extending this conserved intracellular scaffold distribution to extracellular basement matrix is too specific. This does not establish that extracellular association is impossible, identify the exact donor experiment, or label the donor annotation incorrect. Both experimental GO:0005604 rows retain their exact UNDECIDED reviews and image-access limits. Basolateral plasma membrane is separately supported and is not substituted for this source assertion. No QUALIFIER-column distinction enters this reasoning.
Four bare supported_by entries duplicate an existing quoted entry to the same reference (zero-based rows 1, 3, 16 and 94; PMIDs 12482754, 18423203, 22117215 and 18664494). Remove only those bare redundancies. Every supporting quotation, cited reference identity and scientific decision in these four rows is retained; no quotation words are added. The reviewer listed three examples, and the same exact redundancy check found the fourth.
Current decisions are 26 ACCEPT, 51 KEEP_AS_NON_CORE, four MODIFY, four REMOVE, 31 UNDECIDED and two MARK_AS_OVER_ANNOTATED. DRAFT remains tied to unresolved biological assertions. All 118 machine source objects, 120 raw records, six alternative products, two core functions, 47 reference assessments, 60 binding decisions and 23 unresolved fragmentomics rows are preserved. The only action change is the electronic basement-membrane row. Earlier dated notes remain an exact prefix; this entry supersedes that row's earlier decision to await donor-level evidence. Distinct focused scientific peer and ROOT acceptance are required before application.