CNTN2 is a 1040-residue human GPI-anchored neural adhesion protein (Q02246). The genuine seed contains 38 assertions, no alternative-product field and eight reference pairs. The raw QuickGO response and normal TSV each contain 41 rows: four detailed IDA ECO records project to one postsynaptic-membrane IDA assertion. Every raw record remains mapped; this is not a deletion of biological source evidence.
The default Falcon attempt timed out and the recorded fallback returned an insufficient-quota response. No provider report was invented. These manual notes and the source index record the actual primary work.
Human ectodomain structures, binding assays and surface-controlled cell-aggregation mutants establish homophilic adhesion. The 2024 study uses soluble residues29–1004 and defined fragments, with cross-linking for structural particles; the 2025 purified protein spans29–1012, while its tagged cellular constructs test aggregation separately. Neither soluble construct is an intact membrane-tethered protein. The Ig3–Ig6 contact evidence is stronger than the earlier Ig1–Ig4 lattice-contact models, and concentration-dependent association does not impose one universal native stoichiometry. One cell-cell adhesion mediator core captures the performed work. [PMID:38968938; PMID:39702996; PMID:17766378]
The original human cloning abstract explicitly supports GPI attachment and surface localization in COS1 and myeloma host cells. Its broad adhesion framing is supplemented by the later direct assays, rather than presented as an adhesion measurement in that abstract. PMID:8425542
The 2003 papers directly place TAG-1 with CASPR2 and Kv1 channels at juxtaparanodes. The tau-lacZ staining of nodes and myelin in one paper is a reporter of expression, not direct TAG-1 protein localization. The two nodal annotations are refined to the directly resolved juxtaparanodal compartment; this does not categorically exclude transient developmental staining. The broader localization process is preferred over directed establishment because recruitment and retention were not separated experimentally. [PMID:12963709; PMID:12975355]
Glial rat TAG-1 expression rescues CNS channel clustering in mouse Tag1-null animals, with direct glial surface staining and myelin fractionation. Peripheral ectopic paranodal expression does not give the same rescue. These findings support an adhesion-based organizing role, not a potassium-channel pore or a universal axon–glia complex geometry. PMID:20962216
Human TAG-1-GFP constructs localize on postsynaptic shafts and spines in rat cultures and restore Caspr2-Fc binding in mouse knockout cultures. The source focused on anti-Caspr2 autoantibodies nevertheless directly tests CNTN2. Similarly, original Fig.3D of the CNTNAP2 genetics paper directly detects rat TAG-1 in synaptic plasma-membrane fractions; its selected cache remains abstract-only and the authentic external PMC source supplies the checked target experiment. Cofractionation is not treated as proof of physical binding. [PMID:26217189; PMID:18179895]
Retinal-axon and glial morphology findings come from mouse tissues, with separate human TAG-1 substrate and TAX-expressing-cell experiments. AIS protein staining in the mapped source actually uses rat hippocampal cultures, despite the current mouse annotation provenance. Developing neuronal-cell-body surface staining is explicit in the older mouse abstract. [PMID:18650339; PMID:18509034; PMID:8089271]
The single generic protein-binding assertion has three positive APP–CNTN2 two-hybrid records from the neurodegenerative-disease interaction screen. APP is a verified adhesion molecule, permitting a partner-class refinement. The APP mutation feature, bait/prey versus neutral roles, and assay context are preserved; no amyloid-processing or native APP–CNTN2 adhesion mechanism follows from the screen. [PMID:32814053; PMID:25122912]
Human TAX-1 in CHO lysates is specifically recovered by Lewis-x conjugate with protein and carbohydrate controls, supporting trisaccharide binding. This secondary capability does not constitute a separate core or a purified-affinity measurement. PMID:19458237
Current donor identities and exact-term records are recorded for every inherited assertion. Historical PAINT trees and node placement were not reconstructed. Turtle and DSCAM self-avoidance experiments are genuine donor evidence but are not direct CNTN2 repulsion assays. Existing self-avoidance and synapse-organization inferences remain non-core, without alleging evolutionary failure from donor count, identity or target self-inclusion. [PMID:19783736; PMID:19945391; GO_REF:0000033]
Mouse Tag1-null diet experiments measure whole-body adiposity and food consumption. They do not assay adipocyte differentiation, so that electronic specificity is over-annotated. The precise reduction-of-feeding step after dietary-excess detection remains undecided; the broader diet phenotype is retained outside the core. PMID:22752552
The clinical CNTN2 paper supplies positive human segregation evidence. An authentic author reprint was read through complete Methods/Results and the first Discussion section; the remaining discussion and supplement were not inspected. No presynaptic terminal-assembly experiment was established in those sections, so that original IMP assertion stays undecided. This is not a wrong-gene or false-paper claim. PMID:23518707
No NEW annotation is proposed. Existing process coverage is retained, the GO-CAM index has no target hit, and clinical necessity is not used to manufacture a process assertion. Primary quotations are limited to three short anchors in the YAML; the notes and source index contain paraphrases and source locators, not duplicate quotation arrays. The source index is a reproducibility aid and is not cited as independent biological evidence in annotation support.