EGFR couples extracellular growth-factor recognition to intracellular tyrosine phosphorylation. Ligand binding, receptor association, allosteric kinase activation and phosphorylation of docking sites are related but distinct events. The ligand-bound ectodomain study supports ligand recognition; kinase-domain and juxtamembrane studies establish catalytic activation and its regulation. Isolated-domain experiments do not independently demonstrate every step of full-length receptor signaling. The original two overlapping kinase summaries are represented as one integrated epidermal growth factor receptor activity, with EGFR signaling as its biological process. PMID:12297050(https://pubmed.ncbi.nlm.nih.gov/12297050/), PMID:16777603(https://pubmed.ncbi.nlm.nih.gov/16777603/), PMID:19560417(https://pubmed.ncbi.nlm.nih.gov/19560417/), PMID:19563760(https://pubmed.ncbi.nlm.nih.gov/19563760/).
The core includes both plasma-membrane and endosome-membrane locations. The endosomal location is supported by active receptor context, not merely the presence of cargo destined for degradation: hypoxia-associated endosomal EGFR–AGO2 association is accompanied by biochemical support for AGO2 Tyr393 phosphorylation. The localization and kinase experiments are separate observations, so this does not claim that an isolated endosome reaction was reconstituted. In contrast, endosomal-sorting studies provide direct localization but do not necessarily show persistent signaling. EGFR on intraluminal vesicles is assigned to their membrane rather than the enclosed lumen. PMID:23636329(https://pubmed.ncbi.nlm.nih.gov/23636329/), PMID:22719997(https://pubmed.ncbi.nlm.nih.gov/22719997/), PMID:17714434(https://pubmed.ncbi.nlm.nih.gov/17714434/).
The TLR9 study supplies a positive cellular signaling context, with engineered EGFR-deletion localization distinguished from endogenous receptor-dependence experiments. It does not establish a purified EGFR-to-TLR9 Tyr980 reaction, an EGF-driven assay, or equivalent activity of a natural short EGFR product. Nuclear PCNA experiments distinguish commercial GST-EGFR kinase-domain assays from fractionation and cellular observations. PCNA phosphorylation and stability can regulate replication and repair without giving EGFR PCNA's catalytic functions. These contexts remain visible in source-specific annotations without manufacturing separate core activities. PMID:40980882(https://pubmed.ncbi.nlm.nih.gov/40980882/), PMID:17115032(https://pubmed.ncbi.nlm.nih.gov/17115032/).
Four official products are retained at 1,210, 405, 705 and 628 residues. Only the full-length product retains the complete transmembrane and intracellular kinase architecture. Their shared extracellular sequence does not establish equivalent secretion, ligand binding, or receptor modulation. Cancer deletion variants, synthetic kinase fragments and incompletely authenticated experimental inserts are not mapped automatically to natural products. The truncated-product expression report does not by itself establish secretion. PMID:9103388(https://pubmed.ncbi.nlm.nih.gov/9103388/).
The task-specific ClinGen Mendelian project policy retains a supported generic protein-binding assertion as KEEP_AS_NON_CORE when no informative evidence-backed refinement is available. A more specific partner-class term can be supported by a verified association and the partner's independently established function; it does not require a purified binary interface or transfer the partner's catalytic activity to EGFR. Conversely, proximity labeling or colocalization alone does not establish physical binding. Affinity purification, coimmunoprecipitation, MaMTH, BioID and domain–peptide assays are therefore distinguished per source and pair. PMID:24658140(https://pubmed.ncbi.nlm.nih.gov/24658140/), PMID:31980649(https://pubmed.ncbi.nlm.nih.gov/31980649/), PMID:35384245(https://pubmed.ncbi.nlm.nih.gov/35384245/).
Several examples illustrate the limits:
The AGER1 source contains a confirmed partner-identity discrepancy. Its Methods and reciprocal coimmunoprecipitations concern human AGER1/OST48, mapped to DDOST/P39656, rather than AGER/RAGE/Q15109. The imported With/From field remains untouched. A structured source-defect record asks IntAct to check the upstream record; the exact interaction identifier could not be authenticated from the saved current export. This neither deletes the positive AGER1 association nor asserts that EGFR–RAGE binding is impossible. PMID:16954185(https://pubmed.ncbi.nlm.nih.gov/16954185/), UniProt P39656.
The 16 historical/current IBA assertions trace to ten positive ancestral IBD assertions at three ancestors of human EGFR in the examined family tree. No same-term loss was found on the selected target path. Target self-evidence is valid experimental grounding for an ancestral inference; neither donor count nor self-inclusion is used as a circularity objection. Historical and current support sets remain separately represented. The tree proof records target placement and sequence identity.
Reactome evidence was traced through literal event/entity chains to human P00533. This distinguishes an EGFR catalytic active unit from passive membership or receptor cargo and preserves disease, mutant-only, candidate-set and human–mouse chimeric model scope. In particular, the three R-NUL entries do contain human EGFR through their curated complexes; their mixed species are not treated as mouse-only evidence. The 190 source-chain records contain the selected witnesses and official event links. They are database provenance, not a claim that every cited primary body was independently inspected or every other branch exhaustively expanded.
Ortholog evidence retains its experimental species. The bovine donor F1N7F3 is not rat or mouse. Rat neuron experiments and mouse learning experiments in the same paper remain separate, and the Rat1 Shc–EGFR study's human receptor transgene is INSR. HCV postbinding entry dependence supports a positive entry role without proving direct viral-component binding by EGFR. PMID:28833817(https://pubmed.ncbi.nlm.nih.gov/28833817/), PMID:20639532(https://pubmed.ncbi.nlm.nih.gov/20639532/), PMID:7513704(https://pubmed.ncbi.nlm.nih.gov/7513704/), PMID:21516087(https://pubmed.ncbi.nlm.nih.gov/21516087/).
Human Protein Atlas location-specific reliability is used rather than its overall score: plasma membrane, Golgi and cell junctions are supported, primary cilium and basal body are uncertain, and the reported sperm compartments are approved. These are nonexclusive pools, not evidence of universal localization in every cell. Human Protein Atlas.
Forty-one source assertions remain UNDECIDED. This includes inaccessible operative evidence, unresolved target or partner mapping, and positive proximity observations that do not settle binding. One MARK_AS_OVER_ANNOTATED decision retains the positive underlying biology while rejecting excessive claim specificity. The early DNA-nicking report is treated as a positive report with unresolved preparation-specific controls, not dismissed from its abstract or relabeled as indirect signaling. PMID:6325948(https://pubmed.ncbi.nlm.nih.gov/6325948/).
The Expression of Concern for PMID:20473329 concerns Figure 9A phospho-Src and unavailable raw data; it does not name the retained Figures 7 and 9B/C interaction panels. The correction to PMID:21203579 supplies a revised Figure 2A and identifies other affected panels. Neither notice is generalized into wholesale invalidation. The 1997 Grb2 erratum corrects dn/dc to 0.185. The linked 2009 juxtamembrane-study erratum is authenticated but its correction text remains unavailable, so exact kinetics or mutation-role labels are not taken from that source without corroboration. PMID:37264082(https://pubmed.ncbi.nlm.nih.gov/37264082/), PMID:26949965(https://pubmed.ncbi.nlm.nih.gov/26949965/), Grb2 erratum, juxtamembrane-study erratum.
The two inherited NEW process assertions are retained only after checking actual catalytic participation, same-role receptor comparators and typed ontology relations. Neither is an ancestor or descendant of the other or of the 121 distinct terms in the 1,006 imported sources. A shared descendant does not make them ancestor–descendant duplicates. The underlying kinase/fragment evidence is kept distinct from necessity or passive substrate evidence. PMID:27872189(https://pubmed.ncbi.nlm.nih.gov/27872189/).
The review retains all 974 historical objects and appends 34 genuine current imports, for 1,008 total. It contains 289 ACCEPT, 254 KEEP_AS_NON_CORE, 420 MODIFY, 41 UNDECIDED, one REMOVE, one MARK_AS_OVER_ANNOTATED and two inherited NEW decisions. There are no newly fabricated imported assertions.
The exact normal-current GOA snapshot contains 969 raw records and 958 distinct full source contexts. Forty-eight historical contexts are retained with retired: true: 37 absent source identities, eight changed complete support sets, and three changed relations. Their scientific actions, literal sources and evidence remain unchanged. The old enables assertion and current contributes_to assertion are both preserved as different contexts. Retirement means absence of that exact context from this snapshot, not biological disproof. The before/after ledger and old GOA archive preserve the bridge.
The 999 inherited quotation occurrences were explicitly reconciled: six title-only excerpts from the replaced duplicate cores, one model-metadata excerpt and six nuclear title-only excerpts were retired while retaining their reference anchors. A subsequent title-equality audit removed 690 additional optional quotation values, representing 205 distinct reference titles, while retaining every reference anchor. The remaining 296 inherited quotation occurrences are exact and no new quotations were authored. In total, 703 of the original 999 occurrences have been retired with explicit authorization. Publication caches retain their original bytes and flags. A cache marked full text is not treated as proof of complete operative access, and separately recovered original evidence does not turn an abstract-only cache into a full-text cache. The source guide records citation identity and assessment, and the portable evidence README explains the reproducible joins and their limits.
Repeated annotations are compared as claims about EGFR. The 18 term groups raised in review were checked against the actual source rows; GO:0005006 was already homogeneous ACCEPT. The following table records the finite reconciliation. An unresolved source-specific experiment is not a claim that the function is absent. Distinct partner, construct and localization scopes remain visible rather than being replaced by blanket actions.
| Term | Resolution |
|---|---|
| GO:0004713 | Row 257 is refined to GO:0005006 for its ligand-responsive PLC-II evidence. The other existing receptor-context refinements remain. Catalytic-domain/substrate assays retain GO:0004713, including row 478’s PCNA assay; they do not independently demonstrate the complete receptor activity. These are compatible resolutions, not a negative kinase claim. |
| GO:0005006 | The ten published rows already all ACCEPT this receptor activity. The reviewer counted this as mixed, but the parsed source does not. Row 257’s replacement adds evidence for the same activity without creating a new source row. |
| GO:0030296 | Row 265 now accepts the gene-level activator function with reference-only PMID:16777603 asymmetric-interface complementation support. PMID:2176151 remains docking-only evidence. Row 266 retains its distinct EGF-induced ERBB2 pY1248 and heterodimer readouts, not purified allosteric-interface proof. |
| GO:0005737 | Row 287’s own evidence resolves endosomes and is refined accordingly; rows 497 and 537 retain broader measured cytoplasmic pools. A narrower supported location does not negate broader staining. |
| GO:0005768 | Row 21’s official annotation can be refined to endosome membrane; the other sources support endosome localization at their stated resolution. No absence of the broader compartment is asserted. |
| GO:0031901 | Row 334 retains endogenous EGFR limiting-membrane localization; row 995 retains the engineered-deletion/TLR9 signaling context as non-core. The latter is not generalized to every natural receptor product. |
| GO:0043235 | The positive inferred receptor-complex claim is supported; row 341 remains unresolved for the exact partially accessible PMID:23382219 target-complex evidence. This is not a negative statement about receptor complexes. |
| GO:0005576 | Row 19 retains the official secreted-product context; row 996’s specific truncated-transcript paper does not independently establish secretion. Unresolved attribution is not an absence claim. |
| GO:0005634 | Row 997 is aligned with the other positive nuclear-location rows as KEEP_AS_NON_CORE. The bladder-cancer nuclear pool remains positive and context-dependent. |
| GO:0005789 | The official ER trafficking pool remains positive non-core. Row 1000 remains unresolved because the source-specific ER evidence requires unrecovered material; other sources do not reconstruct that experiment. |
| GO:0070374 | Rows 31 and 349 now ACCEPT the same principal ERK-cascade signaling output as row 328. Rule provenance and E2/GPER1 fibroblast scope remain explicit, without assigning EGFR MAP3K chemistry. |
| GO:0008284 | Rows 196 and 493 align with row 327 on EGFR’s principal mitogenic output. The mouse crypt donor and abstract-only source boundaries remain explicit; additional reference-only human EGFR-depletion evidence supplies independent grounding. |
| GO:0016323 | Row 503’s direct human airway basolateral staining remains positive and is classified consistently with row 200 as a polarized-epithelial context. |
| GO:0048471 | Positive mouse-donor perinuclear evidence remains non-core; row 277’s exact source-localization experiment remains unresolved. Neither statement negates the other. |
| GO:0019900 | Row 227 concerns a supported DGKQ lipid-kinase association; row 271’s imported CRK/BCAR3 source-to-partner mapping is unresolved. Different partner assertions are not competing claims about a single interaction. |
| GO:0019903 | Row 500 retains the actual PTPRK association/class. Row 312 remains unresolved for DUSP3 because phosphorylation dependence does not settle that binding assertion. |
| GO:0042802 | The established self-association claim remains accepted. Four source-specific rows remain UNDECIDED for unrecovered or heterotypic/proximity-confounded evidence; this is not an assertion that EGFR cannot self-associate. |
| GO:0005515 | The 574 generic-binding rows describe different partners and methods. Preserve all individual class refinements, positive non-core associations and unresolved cases under the project exception; the consistency table explicitly exempts this term. |
The allosteric activator conclusion is independently grounded in the human EGFR asymmetric-dimer study: complementary kinase-interface mutants and kinase-dead controls distinguish activation of a partner from phosphorylation-dependent recruitment of effectors. The older docking experiment is not claimed to restore the mutant kinase’s activity. The ERBB2 experiment instead supplies a cellular phosphorylation/heterodimer readout. PMID:16777603(https://pubmed.ncbi.nlm.nih.gov/16777603/), PMID:2176151(https://pubmed.ncbi.nlm.nih.gov/2176151/), PMID:11336639(https://pubmed.ncbi.nlm.nih.gov/11336639/).
The MAP3K annotation is removed because EGFR is the upstream receptor tyrosine kinase, not the MAP kinase kinase kinase in the demonstrated pathway. Positive MAPK/ERK regulation is retained. PMID:15542601(https://pubmed.ncbi.nlm.nih.gov/15542601/). No core activity, product boundary, imported context, reference identity or publication-cache content changes in this follow-up.
The current action-consistency rules check selected action pairs, including ACCEPT/REMOVE; they do not classify all ACCEPT/MODIFY or ACCEPT/UNDECIDED combinations as conflicts. No blanket consistency exceptions were added merely to suppress warnings. The nuclear kinase-domain/PCNA versus ligand-responsive receptor distinction is stated directly in its annotation rationale and this table.