The imported normal seed has 34 assertions, four UniProt products and 18 references. All machine assertions, source qualifiers and products are preserved. The proposed decisions are eight ACCEPT, 24 KEEP_AS_NON_CORE, one MODIFY and one UNDECIDED, with no NEW assertion. The completed review remains DRAFT because the validator reports 19 generic-binding advisories. These are acknowledged under the project's instruction to preserve positive interaction evidence as non-core when no supported specific function follows from an individual row.
CCM2 is an intracellular noncatalytic adaptor. The strongest molecular evidence combines PTB-mediated KRIT1 binding and localization effects with a KRIT1–CCM2–MEKK3 ternary complex (PMID:16037064(https://pubmed.ncbi.nlm.nih.gov/16037064/)). The source generic binding row is refined to GO:0035591 signaling adaptor activity, whose definition requires coordination of multiple signaling participants. Direct binding alone would not establish that function; the ternary-complex and localization results supply the additional rationale. The narrower MAP kinase scaffold term is not selected solely because MEKK3 is a partner. No kinase, phosphatase, GTPase or ubiquitin-ligase activity is assigned to CCM2.
The best human process experiment is the CCM2-specific versus scrambled-control shRNA comparison in HUVEC fibrin-bead assays: visible lumens decline while branched cord length, migration, proliferation and size are retained (PMID:19151727(https://pubmed.ncbi.nlm.nih.gov/19151727/)). The original selected Results and Methods were read, not every figure or supplement. Mouse tagged CCM2/KRIT1 constructs used for separate biochemical assays in that paper remain explicitly distinct from its human endothelial experiments. HEG1 recruitment through KRIT1 does not establish direct CCM2–HEG1 contact. The existing endothelial-tube-morphogenesis annotation supplies the core process; broader development terms are not added redundantly.
The CCM3 structural paper contains relevant CCM2 experiments despite its title (PMID:20489202(https://pubmed.ncbi.nlm.nih.gov/20489202/)). Its broad CCM2 51–251 construct includes the region after the narrower PTB fold. In the CCM2 HHD study, full-length CCM2 binds CCM3 while the isolated HHD does not; the relevant linker lies after the PTB domain (PMID:23266514(https://pubmed.ncbi.nlm.nih.gov/23266514/)). Weak crystal packing and solution monomer measurements do not justify a stable physiological homodimer claim. The separate CCM2–KRIT1 structural study supports the PTB interface (PMID:25525273(https://pubmed.ncbi.nlm.nih.gov/25525273/)); the domain name alone does not establish a phosphorylation requirement.
The original PMID:16037064 full text was unavailable. Its abstract draws an analogy involving ICAP1 and integrin interactions; it does not independently settle CCM2 participation in integrin-initiated signaling. That TAS row is UNDECIDED. The abstract-only human genetics paper PMID:14740320 likewise is not relabeled as a direct vasculogenesis experiment; its curator-supported assertion is retained because the biology is independently supported, without accusing the curator of an error on incomplete evidence.
The 19 remaining generic binding rows preserve exact partner accessions. Binary-interaction evidence in PMID:32296183 is distinguished from cell-specific network association in PMID:33961781 and the multimodal resource PMID:40205054. Study-wide methods do not identify every individual edge's physiological role. Target-specific supplementary tables were not inspected. In particular, the phosphotyrosine screen PMID:25814554 does not automatically make every CCM2 edge phosphorylation dependent. The PMID:17657516 abstract's STK25 and FAP-1 enzymatic claims concern CCM3, not CCM2.
The official Human Protein Atlas CCM2 subcellular page reports enhanced mitochondrial localization with HPA020273 and HPA065815. The existing IDA location is retained as non-core, with no respiratory enzyme, obligate mitochondrial function or respiratory-complex membership inferred. Cytoplasm remains the broad core location. Cardiovascular developmental and stress-MAPK annotations are contextualized rather than treated as additional intrinsic molecular activities.
All 13 PMID cache headers and available scientific abstracts were read. Canonical full-text flags are retained: 14740320, 16037064, 17657516, 23007647 and 25525273 are abstract-only; other cache bodies are available but that does not constitute a claim that every full body was read. Selected target-specific primary Results and Methods were additionally inspected for 19151727, 20489202 and 23266514. No reference is marked VERIFIED merely from an automated report. The richer recovered copy of 23007647 does not overwrite its existing canonical cache; the existing 25910212 and 33961781 caches are also preserved.
Normal deep research was attempted through just deep-research human CCM2 --provider perplexity but failed because the offline environment lacked the required deep-research-client dependency. No provider-named research file was authored as a substitute. Manual reading and access limits are documented here. Additional pathway-oriented papers were used for orientation only; no annotation or cached citation was manufactured from an inaccessible article.
The first staged validation failed solely because the immutable GOA companion had not yet been copied beside the draft. The companion GOA and UniProt records were then staged byte-for-byte from the imported normal sources before the normal checks were repeated. No canonical review, source or history file is edited by this proposal.
The five broad vasculogenesis/endothelium-development assertions are retained as non-core developmental context. The single core retains signaling adaptor activity and the more specific human endothelial tube-morphogenesis context. All34 original source objects and four products are unchanged; decisions are3 ACCEPT,29 KEEP_AS_NON_CORE,one MODIFY and one UNDECIDED, with no NEW assertion. The original integrin-specific source remains unresolved: an ICAP1 analogy in its abstract does not by itself settle the original TAS mapping.
The additional mechanisms are distinguished by experimental system. PMID:20308363(https://pmc.ncbi.nlm.nih.gov/articles/PMC2856024/) tests interaction-dependent KRIT1 localization and barrier rescue in human HUVEC, with separate mouse permeability experiments. PMID:19318350(https://pmc.ncbi.nlm.nih.gov/articles/PMC2679429/) links CCM2–SMURF1 localization to RhoA abundance in mouse bEND.3 and transfection experiments. The former did not observe a total-RhoA change in primary endothelial cells. The results support context-dependent regulation rather than a universal degradation mechanism.
PMID:27027284(https://pmc.ncbi.nlm.nih.gov/articles/PMC4864035/) connects MEKK3–KLF2/4 gain to early lesions in neonatal endothelial Krit1-deletion mice and separately tests a disease-causing human CCM2 mutant that loses MEKK3 interaction while preserving CCM-complex association. The published correction supplies an omitted Renz et al. citation; it does not report changed data. These findings inform the adaptor description without adding kinase activity or new biological-process rows.
The original OSM paper includes both MEKK3 and MKK3, correcting the earlier overly narrow account. Addgene12191 identifies the deposited OSM construct as mouse Ccm2. The conserved osmotic-stress scaffold context is retained separately from the human endothelial mechanism. The human KRIT1 interaction row remains a refinement to signaling adaptor activity. Both ternary-complex anchors now include the CCM1 subject.
The official HPA record supports retaining the mitochondrial assertion as non-core. Its enhanced localization summary and antibody/cell-line table were inspected, without independent pixel or genetic-control reanalysis. This is compatible with other intracellular pools and does not establish mitochondrial catalysis.
The core's process synthesis combines a defined signaling interaction mechanism with the directly measured human lumen-formation phenotype. The original GOA qualifier remains acts_upstream_of_or_within; it is not rewritten to claim a different experiment. Broad developmental outcomes are treated consistently as non-core. No substrate is assigned the enzyme activity that acts on it.
Four reference identities are added from normal cached metadata. Existing caches remain unchanged, including the abstract-only27027284 source. Reference assessments distinguish checked identities and selected primary passages from uninspected full studies. New sources were obtained through the unchanged normal fetcher; no manually fabricated cache or provider-named research report is used. Normal validation and peer review must be completed before canonical application.