FN1 is a classic example of developmental/tissue-specific splicing with 17 named isoforms.
The key splicing events involve:
1. EDA domain (Extra Domain A, also called EIIIA) - alternative exon
2. EDB domain (Extra Domain B, also called EIIIB) - alternative exon
3. IIICS region (Variable region, V region) - can be partially or completely included
| Isoform Type | EDA | EDB | Expression | Function |
|---|---|---|---|---|
| Plasma FN | - | - | Hepatocytes, soluble | Wound healing, blood clot |
| Cellular FN | + | +/- | Fibroblasts, solid | ECM assembly, embryonic development |
| Oncofetal FN | + | + | Embryonic, tumors | Angiogenesis, tumor progression |
EDA-containing isoforms:
- Promote cell adhesion more strongly
- Essential for wound healing
- Re-expressed in cancer (oncofetal antigen)
- EDA-null mice have normal development but impaired wound healing
EDB-containing isoforms:
- Strong pro-angiogenic activity
- Expressed during embryonic development
- Re-expressed in tumor angiogenesis
- Target for cancer imaging/therapy (antibody L19)
EDA/EDB double-null mice:
- Cardiovascular defects
- Embryonic lethality or early postnatal death
- Demonstrates functional importance of these domains
UniProt lists 17 isoforms including:
- Isoform 7: Contains EDB domain
- Isoform 8: Lacks EDA domain
- Isoform 9: Lacks both EDA and EDB, uses V64 variant
- Isoform 16: Migration stimulation factor (MSF)
The source guide records primary access, Methods/Results locators, species and construct boundaries. The evidence note explains the scientific decisions, and the context archive preserves the complete earlier review, current original source objects, exact extension projection, product corrections and every inherited quotation association. Earlier journal entries above remain historical records; the current review and this appendix supersede their conflicting conclusions.
The review now represents 234 rows: 223 typed contexts from 222 current raw annotations, ten historical source rows retained despite absence from the current response, and one individually rebuilt authored matrix-organization proposal. Counts are 53 ACCEPT, 113 KEEP_AS_NON_CORE, 42 MODIFY, 22 UNDECIDED, three TOO_GENERAL and one NEW. The 52 inherited generic-binding removals were reassessed individually, alongside distinct current partners and source contexts, under the standing campaign rule. Supported pairs remain non-core or receive an evidence-backed binding subtype; unresolved pairs remain undecided.
Five core activities distinguish structural matrix contribution from integrin, collagen, heparin and proteoglycan recognition. The former RGD-to-every-integrin generalization is removed. The authored fibrinogen-binding and redundant broad-coagulation proposals are retired with their complete original objects retained; fibrin interactions remain supported biology. The matrix-organization proposal is grounded in primary scaffold experiments and explicit participation, comparator, ontology and GO-CAM checks.
All 17 product identities and verified sequences remain. Thirteen product records receive bounded description/metadata corrections, including the displayed sequence's EDA/EDB content, three incomplete VSP lists and a truncated synonym. No product-specific activity is invented. All 149 original quotation associations remain in provenance; 138 active excerpts are unchanged and eleven are individually archived with reasons. No new quotation is introduced.
The original GOA, UniProt and provider snapshots and three old histories are preserved. Current original machine contexts and complete OR/conjunction semantics are published as companion evidence. All 148 normal sources are identified by exact path and SHA256. Six absent publications were genuinely fetched; existing source bytes were retained, including the separately hydrated mouse donor paper. The existing provider report remains background; no new provider-labelled report was manufactured. Current reasoning relies on the primary and official-source reading described in the guide.
Schema, reference-title/excerpt and ontology checks pass with GOA comparison explicitly separated. There are 35 annotation advisories: 34 supported generic-binding non-core decisions retained under the standing user instruction and one deliberate source-dependent extracellular-location distinction. The computed status is DRAFT, consistent with these advisories and the 22 undecided contexts.
Strict GOA validation passes for the exact 223-current-context subset against the fresh source projection. The unchanged historical GOA snapshot produces one missing historical heart-development tuple, nine current/additive tuples absent from that snapshot and two associated evidence-code mismatch notices when checked against the whole augmented review. This diagnostic is retained; the ordinary full-record GOA gate is not reported as passing. Independent conservation checks preserve all 193 original machine-source assertions, 222 current raw objects, all extension branches, 17 products and 148 normal source identities. The source snapshots were not rewritten to suppress the mismatch.
The newly scaffolded history, computed status, private HTML rendering and whitespace checks pass. The PMID helper matched no pathway files and is not claimed to validate these notes. Parsed preview content and 23 local links were checked without claiming a pixel-layout inspection.
The separate diagnostics above identified one unrepresented historical source group: the old heart-development IBA contained a fourth WITH/FROM entry. The current row remains unchanged, and an additional source-backed non-core row now preserves that exact historical group. The review therefore contains 235 rows: 53 ACCEPT, 114 KEEP_AS_NON_CORE, 42 MODIFY, 22 UNDECIDED, three TOO_GENERAL and one retained authored NEW. All prior 234 decisions, products, core functions, references and quotation associations remain unchanged.
The opt-in sibling manifest requires both the original immutable GOA table and an exact copy of the genuine current projection. It checks declared source integrity and target identity before the unchanged strict matcher checks their combined source records. The context archive preserves both byte-exact inputs and the historical row locator. This resolves the representation issue without rewriting a provider snapshot, dropping historical assertions or disabling the ordinary GOA check. Implementation and ordinary validation results will be recorded below after they complete.
Ordinary full FN1 validation and default strict GOA validation now pass. The default loader reads both declared snapshots, verifies their namespace/accession/taxon and SHA256 values, and checks all 445 source records with the unchanged strict matcher. The 235 reviewed rows retain 35 disclosed curation advisories and DRAFT status: 34 supported generic-binding non-core decisions under the standing instruction and one source-dependent location distinction. No GOA bypass is used.
The complete GOA-validator test file passed 95 tests, including negative manifest/source cases and ordinary gene-validator integration. Broader targeted regression execution passed 224 tests; its sole failure was a missing sparse-checkout fixture. Hydrating that one fixture from the exact existing HEAD and rerunning only its test passed, with 12 standard integration deselections retained. A passing pre-existing CLI test rewrote an unrelated publication cache; its generated diff was preserved privately and the exact prior Git bytes were restored. The changed validator passes mypy, and both changed Python files pass ruff. Full repository CI remains required.
Conservation confirms unchanged first 234 reviewed rows and all other biological fields, the exact separately represented historical heart assertion, 138 active inherited quote occurrences and 11 archived originals, all 17 products and five cores. All 148 normal dependencies remain byte-identical after acquisition, and the older fetched GOA/UniProt/provider files and three prior histories remain intact. The new history, computed status, private preview and links were checked.
The mouse Fn1 GO-CAM model describes receptor-ligand signaling during muscle activity, a distinct role from the retained matrix-scaffold proposal. The evidence note now states that distinction rather than claiming the index has no FN1 entry.
Twelve supported microbial adhesion interactions are retained as non-core without an enzyme-binding replacement, because catalytic partner classification does not characterize the adhesion activity measured in those experiments. Six further enzyme-partner interactions are likewise retained as non-core after source-specific reassessment; the eight remaining binding refinements use other informative partner classes. The revised action counts are 53 ACCEPT, 132 KEEP_AS_NON_CORE, 24 MODIFY, 22 UNDECIDED, three TOO_GENERAL and one inherited NEW across 235 source contexts.
Two inherited paper-title quotations do not support the donor-specific muscle-to-liver claim. Their exact originals and former associations remain in provenance; they are removed from active support for that claim. This leaves 136 active inherited quote occurrences and thirteen individually archived associations, with no new quotation.
The peptidase-activator decision retains the actual selected original experiment rather than relying on the abstract-only normal cache. The source guide distinguishes that access from figure-pixel or full-PDF inspection. The snapshot documentation now explains historical versus live source assertions, the unchanged single-file consumers, normal non-force fetch behavior, and a concrete additive-refresh/integrity-recovery procedure.
Cell-shape regulation remains a contextual positive role: fibronectin supplies the receptor-engaging extracellular ligand in directly measured spreading assays, while intracellular effectors implement the morphological response. Human plasma FN on mouse fibroblasts and human wild-type FN10/native FN on engineered human K562 cells are distinguished from the high-affinity antagonist derivative. For angiogenesis, added FN rescues endothelial migration and wound closure; the separate branching rescue is beta-catenin-mediated, not an FN-alone result.
The follow-up passes ordinary schema, reference and ontology validation with 53 disclosed advisories: 52 supported generic-binding non-core decisions and one source-specific extracellular-location action difference. DRAFT remains the computed status. Default strict GOA validation matches both immutable snapshots and their 445 source rows. Exact source-field/order, product, core and quotation conservation checks pass, as do history validation and the private HTML/content and canonical-link checks. No implementation code or tests changed, so the prior targeted code-test results were reused; full checks on the final PR head remain required.
The twelve Reactome FN1–ALK localization records now receive MARK_AS_OVER_ANNOTATED at the unqualified native-FN1 scope, with source-defect checks for possible export-context loss. The fusion model itself is preserved. Apical surface association remains non-core because the official membrane ontology includes peripheral association, supported here by the original mouse-donor abstract. Matrix-organization and ADAMTS1-cofactor judgments retain their already inspected primary evidence and access limits without adding quotations.
The inherited angiogenesis excerpt describes LPP3-induced FN expression, whereas the reviewed function relies on added-FN rescue. Its exact original association is archived; verified rescue evidence remains reference-only. All 149 original associations are retained in provenance, with 135 active and fourteen archived. A fibrin-versus-fibrinogen question records the remaining assay distinction.
Source-only manifests and sibling TSV changes are selected for the affected gene review through the existing strict validation recipe. A committed offline verifier reproduces the dated FN1 projection from all 222 archived original QuickGO objects using the repository exporter. The old and additive TSVs, source files, products, five cores and annotation source fields/order remain unchanged.
The further follow-up passes ordinary schema/reference/ontology validation with 53 disclosed advisories and computed DRAFT status. Default strict GOA validation covers both immutable snapshots and all 445 source rows. All 29 focused CI tests pass, including source-only additions, modifications, deletions, malformed-manifest reachability, deduplication and failure propagation. Lint and the exact 222-object-to-219-row offline projection replay pass. Private corruption checks reject altered source values, target identity, row count, TSV bytes and hash. A private source-defect export contains twelve suspected context checks with no replacement values and no upstream reports sent. Conservation, history and private rendered-content/link checks pass; final-head repository CI remains required.
The endothelial-cell-migration row carried the same LPP3-induced FN-expression excerpt already archived from the angiogenesis row. Its exact original association is now also archived, while the inspected added-FN migration and wound-closure rescue remains reference-only support from PMID:20123964. All 235 annotations keep their order, source fields and actions; five cores and all 17 products remain unchanged. All 149 original quotation associations are conserved individually, with 134 active and fifteen archived, and no new quotation.
The optional full_text_unavailable: true flag for PMID:16061471 records the abstract-only normal cache. Its unchanged reference notes and the source guide continue to identify the separately inspected original article Methods/Results and access limits. The flag does not assert that those selected original passages were unavailable and does not relax quotation checks.
The quotation-alignment follow-up passes ordinary schema/reference/ontology validation with the same 53 disclosed advisories and DRAFT status, and default strict GOA validation against both immutable snapshots and 445 source rows. Conservation independently enumerates all 149 original quotation locations and exact support objects, with 134 active and fifteen archived; the retired string has no remaining active occurrence. History and private rendered-content/link checks pass. Infrastructure code is unchanged, so its prior targeted tests, lint and replay evidence are reused without rerunning them. Final-head repository CI remains required.