Removed the following annotations from existing_annotations as they were not present in the GOA file:
- GO:0004674 (protein serine/threonine kinase activity) - IDA - PMID:40281343
- GO:0110076 (negative regulation of ferroptosis) - IDA - PMID:40281343
These annotations were based on PMID:40281343 "PSAT1 impairs ferroptosis and reduces immunotherapy efficacy via GPX4 hydroxylation", which shows CaMKII phosphorylates PSAT1 at Ser337 in cancer cells, affecting ferroptosis resistance. While this is valid experimental evidence, these specific annotations with this PMID are not in the current GOA dataset and must be removed to pass validation.
The ferroptosis regulation represents a peripheral, non-neuronal function of CAMK2A discovered in cancer biology context, not central to its primary role in synaptic plasticity.
CAMK2A's principal activity is calcium/calmodulin-regulated serine/threonine phosphorylation. Human alpha-domain structures, biochemical assays and E183V experiments support this activity, with target association and holoenzyme assembly organizing the kinase rather than creating multiple redundant catalytic cores [PMID:20668654; PMID:21884935; PMID:28130356]. Thr286 autophosphorylation can sustain autonomous activity after calcium falls. The single synthesized core joins kinase activity to synaptic potentiation, spine development and regulatory autophosphorylation; it does not equate activation, ATP binding and oligomerization with three independent core functions. Human disease variants can increase or decrease regulatory activity and alter target association; alpha and beta variants are not interchangeable PMID:29100089.
The immutable UniProt record names two actual splice products: Q9UQM7-1 (A, displayed) and Q9UQM7-2 (B, VSP_004766). The standard repository parser restores those literal product records in the candidate. The sequence feature replaces the displayed Lys328 with KKRKSSSSVQLM in B. No functional isoform class, universal compartment difference or new experimental product claim is inferred. The prior absence of a structured products slot did not mean the gene had no alternate products.
PMID:20668654 examines all four human CaMKII kinase domains. Its calmodulin-bound crystal is delta, not alpha; alpha residues 13–301 were among the isolated-domain constructs and weak domain association was measured in solution. PMID:21884935 provides an engineered human-alpha holoenzyme crystal: residues 7–444, a beta7 linker, K42M/D135N/T306V substitutions and bosutinib. The accessible primary Methods also describe chicken calmodulin. These sources support assembly and regulation without requiring every native holoenzyme to adopt the same compact conformation. An erratum (DOI:10.1016/j.cell.2011.10.013) was identified through the primary-author institutional bibliography, but its contents were inaccessible. Fine crystallographic claims are therefore bounded; independent WT/E183V coassembly supports the retained assembly conclusion.
The E183V study reports reduced catalytic/regulatory phosphorylation, changed partner association, neuronal spine targeting and model phenotypes PMID:28130356. Its cache contains the abstract, Introduction and Discussion but not detailed Results/Methods. The CAMK2A/CAMK2B developmental study likewise contains partial narrative, and distinguishes alpha-mutant migration effects from the lack of a migration effect of wild-type alpha overexpression/knockdown at the tested developmental stage PMID:29100089. These limits do not justify rejecting curated experimental annotations from the title or from a paralog's appearance in the paper.
The STAT1 paper includes a real rat-brain alpha/beta kinase control, but its cellular STAT1 interaction experiments chiefly use gamma/delta PMID:11972023. Its alpha-related kinase annotation is therefore supported with the mixed-preparation boundary; the contextual interferon/JAK-STAT annotations defer to the explicit human-Q9UQM7 GO-CAM assignment without claiming an independently verified endogenous human-alpha assay. In contrast, the bacterial-effector study explicitly uses activated CaMKII-alpha T286D/T305A/T306A to rescue an interferon-response phenotype PMID:35568036. The paper's pathogen framing does not invalidate its target experiment. That engineered rescue is still distinct from normal alpha signaling in every cell type.
The PSAT1 study's complete abstract reports CaMKII-dependent PSAT1 phosphorylation upstream of GPX4 stabilization PMID:40281343. Cached GO-CAM 680ad14200007201 explicitly includes CAMK2A kinase activity in the ferroptosis-resistance pathway. Both distinct current GOA relations, acts_upstream_of and involved_in, survive as separate source objects. CAMK2A does kinase work; PSAT1 enzymology, PHD3 hydroxylation and GPX4 peroxidase activity are not transferred to it. This supersedes the older journal entry's historical source-absence rationale: the current authenticated GOA now contains the assertions. That older entry remains unchanged above as history, not current evidence.
Mouse alpha kinase directly phosphorylates DAGLA and modulates 2-AG signaling; selected full Results and Methods establish striatal association, purified recombinant alpha kinase, phosphosite analysis and the limits of phosphomimetic assays PMID:23502535. This resolves the previous uncertainty about negative regulation of hydrolase activity without assigning the lipase's own activity to CAMK2A. Presynaptic alpha deletion and complementary activation-state experiments distinguish activity-dependent release regulation from basal release and from kinase-dependent mechanisms [PMID:12629219; PMID:17660813].
All 92 generic-binding source objects remain separate. Eighty-seven literal partner accessions have exact canonical UniProt IntAct-list matches; the five other curated records are the GRM5/BCL10/CACNB2/LRRC7/SHANK3 source groups. Available assay frameworks and target-specific curator/UniProt evidence support qualified retention. Individual screen target tables were not inspected; there is no claim that each association was independently reproduced in an endogenous human neuron. In accordance with the project's standing instruction, supported generic binding is kept as non-core unless a defined finer binding class is justified. Lack of mechanistic informativeness alone is not grounds for removal. This procedural rule is kept in notes, not biological reasons.
The specific CAMK2D partner in the old PMID:20668654 binding row was incorrectly discussed as calmodulin. It is now explicitly kinase binding. CaMKII/AKT1/PRKCB partners can be refined to serine/threonine kinase binding, HSP90 clients to Hsp90 binding, and GRIN2B/GRM5 to the corresponding ionotropic/metabotropic receptor-binding classes. No partner's catalytic or scaffold function is assigned to CAMK2A. All isoform suffixes remain literal. Three calmodulin source partners remain separate, and recognition of their shared canonical protein sequence does not identify which CALM gene supplied an experimental reagent.
The cached PTHR24347 PAINT table contains positive IBD records for kinase/CaM-binding/PM-localization-regulation at PTN002389359, PSD/plasticity at PTN002389358 and cytoplasm at PTN000686485. Complete tree topology and alignment were not reanalyzed. The exact neuron-projection IBD was not recovered in the inspected table; independent target localization supports that term. Target self-inclusion and a short donor list are not defects. Mouse P11798 source joins were traced separately for the transferred contextual assertions.
Cytosolic Reactome assignments agree with the intracellular kinase, while pathway substrate chemistry is kept separate. The endocytic-vesicle membrane and nucleoplasmic claims remain UNDECIDED because the available event descriptions do not independently settle the kinase's fine location. Sperm-nucleus proteomics supports specialized nuclear detection at its own resolution, not automatic neuronal nucleoplasm localization PMID:21630459. The IRBIT/NHE3 Discussion explicitly leaves direct IRBIT phosphorylation unresolved, despite supporting CaMKII-dependent signaling PMID:20584908. The obsolete positive NF-kappaB term also remains unresolved: the source discusses positive CARMA1 signaling but attenuating BCL10 Ser138 phosphorylation, so a guessed sign-specific replacement would conflate steps PMID:17052756.
The cardiac mitochondrial transfers remain UNDECIDED, with two distinct limits. The original donor paper uses myocardial inhibition/mitoplast preparations and T287D constructs without establishing endogenous human-alpha specificity in the read passages PMID:23051746. Its proposed direct MCU-current mechanism was subsequently challenged in the primary author-institution record, DOI:10.1038/nature13626. That communication reports smaller cardiac mitoplast currents and failure to reproduce direct CaMKII regulation under the compared conditions. The reply DOI:10.1038/nature13627 was not accessed. This is a bounded dispute over direct MCU electrophysiology, not a retraction or proof that every mitochondrial localization/injury result is false. No manual publication cache or unsupported finding_review quote was created. The vascular smooth-muscle G1/S/calcium-transport transfers also retain uncertainty because the accessible abstract does not resolve the original alpha-specific assay PMID:12660151.
The term definitions and relevant parents were checked in official GO/MOD pages. GO:0048168 is a child of GO:0048167. GO:0097106 concerns assembly/arrangement/disassembly of the PSD, not merely CaMKII oligomerization. Existing GO-CAM activity records were checked for the target; their interferon and ferroptosis roles were respected. The comparator evidence is dated and bounded, not an exhaustive negative ontology survey.
The existing Falcon research answer was read as orientation and kept immutable; no provider output was fabricated. All 21 original PMID abstracts, all 10 Reactome summaries, canonical UniProt functional/interaction/product sections and relevant existing primary donor caches were inspected. Full-text flags were not mistaken for complete articles: several HTML caches contain only Introduction/Discussion. The main detailed reads were selected complete Results/Methods in PMID:20668654, PMID:35568036, PMID:23502535 and PMID:23051746, plus available STAT1 Results/Discussion and the accessible primary holoenzyme Methods. Screens were assessed from their actual accessible frameworks and curated target joins, with target tables uninspected. Figures/supplements were not independently read. Web failures and one mistakenly opened unrelated PMC article were excluded from scientific support.
The candidate preserves 170 current machine assertions, one retired historical source record and all three earlier proposal records: 174 rows total. All 175 raw GOA records map to the faithful 170 current keys, including documented byte-identical/date-only duplicates. Eighty-one restored assertions are source restoration, not NEW annotations. The two products are exact output of the existing normal UniProt parser. Source caches, GOA, provider files, history and canonical authored files remain unchanged during TMP preparation. Ten short current evidence anchors are checked against exact canonical text and budgeted in aggregate by source; this appendix introduces no additional quotations. Normal validation and the distinct whole-review peer remain pending at this drafting point.
Initial normal validation correctly flagged the three historical proposals after their actions ceased to be NEW. All three are now explicitly retired from the active proposal set using the existing schema mechanism: one redundant proposal has REMOVE and two unresolved proposals have UNDECIDED. Their original terms/evidence/references remain exact, and all 170 current source objects plus the previously retired source remain unchanged. This preserves the old proposals for audit without presenting them as GOA assertions or active NEW recommendations. GRM5 binding is refined to the official leaf GO:0031802, whose definition and parent were verified directly.
The final candidate passed the normal full term/schema/reference/GOA validation (tool result df076b, exit 0): 72 warnings, comprising 71 supported generic-binding advisories and one unused research-reference advisory. The review remains DRAFT. The complete validator stdout/stderr is retained in an authored execution summary; the original tool transcript, rather than a reconstructed envelope, is the execution evidence. Ten exact primary-source anchors have a maximum aggregate of 17 words per source; these notes add no quotations. Independent whole-review approval and canonical application remain pending. No source, canonical review, provider output, or history file was changed.
The final source check corrected the added UniProt entry title to its immutable header name, KCC2A_HUMAN. It also clarified two ATP-binding rationales: the human kinase-domain structures in PMID:20668654 contain ATP-competitive inhibitors, not ATP. ATP-dependent phosphorylation independently supports ATP use; the nucleotide-binding refinement and the ATP-binding non-core decision are unchanged. No annotation action, machine source field, product, core, or quotation changed.
The exact three-field source correction passed normal validation again (actual 36d32f, exit 0, the same 72 warnings). The completed tool-result envelope is saved as root-source-correction-validation-tool-result.json. The earlier sealed packet is preserved in before-root-source-corrections; all actions, source objects, products, core and quotations are unchanged.
The final serialization preserves the original 100-column YAML style. It is semantically identical to the corrected candidate and passed the final normal check again (actual 03ad5d, exit 0, unchanged 72 warnings); its full tool-result envelope is retained as final-root-correction-validation-tool-result.json.
This entry supersedes the earlier journal statements that retained three withdrawn author proposals as retired annotation rows. The original entries remain above as the chronological record; their temporary-file references and pending-check statements describe earlier checkpoints, not current evidence or outstanding scientific work.
The immutable earlier main review identifies GO:0048167 (regulation of synaptic plasticity), GO:0007611 (learning or memory) and GO:0097106 (postsynaptic density organization) as author-proposed NEW assertions. They are absent from the current GOA. The first two inherited IMP fields and a generated-research-file reference; the third inherited an IEA field without an original reference. Those were proposal metadata, not verified experimental or electronic GOA provenance. The three rows are removed from the annotation list rather than represented as retired source annotations. Their withdrawal reasoning remains:
The separate GO:0007259 JAK-STAT record was already a non-NEW retired historical annotation in the earlier main file. It is retained exactly, including its reviewed evidence and context. The resulting list has 171 records: all 170 current machine assertions and this one historical source record. The retained decisions are 30 ACCEPT, 31 MODIFY, 99 KEEP_AS_NON_CORE and 11 UNDECIDED. All source fields, two products, 50 references and the single kinase core remain unchanged.
Ninety-eight identical propagation-source comments were omitted. Every source identifier, source status, root-cause judgment and annotation rationale is preserved. The seven comments that state a particular PAINT-table observation remain. Omitting repeated prose does not claim new donor, tree or alignment inspection. The description's final sentence now states variant-dependent effects directly; no biological decision changed. All ten existing quotations remain exact and no quotation words were added.