COL7A1 supplies the alpha1 chain of collagen VII anchoring fibrils, which join epithelial basement membranes to underlying connective tissue. Human skin preparations and sequence-defined recombinant experiments support one integrated structural function. NC1 binds matrix ligands, and the C-terminal region participates in antiparallel assembly. These interactions contribute structure directly; collagen VII is not assigned the proteolytic or trafficking activities of the enzymes and receptors that process or transport it. [PMID:18599485; PMID:11274208; PMID:9245798]
The 2008 experiments combine authentic human dermal suprastructures with reconstitution using human full-length 2,944-residue collagen VII and mini-collagens. Collagen I fibrils were made from embryonic chicken tendon. Full-length and an NC1-containing mini-collagen bound those fibrils; a shortened construct lacking the helix-proximal von Willebrand A segment did not. Binding was dose dependent and corroborated by selective cross-linker tag transfer. The native human skin association resisted denaturants, but the authors explicitly leave its covalent chemistry unresolved. Resistance is not reported here as proof of a particular native covalent bond. PMID:18599485
Purified human amnion NC1 also binds laminin 332 and mouse EHS collagen IV, with reciprocal solid-phase, solution co-immunoprecipitation and conformation controls. This matrix-to-matrix attachment does not establish direct integrin binding through a laminin bridge. The separate 2006 abstract reports weaker monomeric collagen I affinity; a monomer assay and a reconstituted fibril assay need not give the same apparent interaction strength. The original 2006 full download was not obtained. [PMID:9245798; PMID:16563355]
One NEW GO:0005518 collagen binding is supported by the direct human-chain experiment, with the tested construct and heterologous collagen-I ligand stated. Current typed ancestry of 5518 does not include the existing 5515 term; the complete ancestor/descendant check has no overlap with original annotation terms. Its collagen-V-specific child is unsupported, and the cell-matrix-adhesion-specific child is not resolved by the matrix-to-matrix assays. Laminin binding does have 5515 ancestry and is not added as another NEW under the strict nonredundancy rule. There is no new biological-process assertion or second core. The GO-CAM index has no target match.
The original human NC2/COL construct contains the 161-residue NC2 region and 186 adjacent collagenous residues. Expressed in human 293 cells, it forms disulfide-linked higher-order assemblies; removal of the collagenous region and specified C-terminal cysteine mutations distinguish trimer formation from antiparallel assembly. These are actual structural participation experiments, not only knockout necessity. PMID:11274208
The same original paper contradicts the automatic Kunitz inhibitor mapping at the tested scope. NC2/COL did not protect NC1 from trypsin or chymotrypsin at 10-fold or 50-fold excess, whereas aprotinin did. Thrombin/plasmin negatives were reported without displayed data, and isolated NC2 was also reported inactive. The original Results and Figure 6 caption differ in the aprotinin ratio; the conclusion depends on the positive control, not selecting one ratio. Other proteases or processing states remain open. The IEA is removed without a new universal NOT annotation or an all-product negative claim. Both actual InterPro mapping records are preserved.
Both official sequences reconstruct exactly: Q02388-1 has 2,944 residues; Q02388-2 has 2,912 residues and deletes displayed residues 1869–1900 through VSP_024026. Both retain NC1 and NC2 architecture, but this is not evidence of equal assembly, binding or tissue deposition. The full-length recombinant study explicitly describes the 2,944-residue chain. Domain constructs have their own experimentally defined boundaries.
The original positive COL7A1–MIA3 IPI source remains non-core generic binding under the campaign exception. Two complete 42-field IntAct records preserve co-immunoprecipitation and pull-down, and their reference-sequence ROGIDs match the official target and partner sequences. The MIA3 45–107 boundary was inferred by curators rather than specified by the paper. The original 2009 full source was not recovered; its positive abstract and literal records remain distinct from the later full structural source. PMID:19269366
The 2023 source identifies the cargo-recognition region as the distinct MOTH domain family and explains why historical SH3 boundaries omitted extensions. This makes a forced SH3-domain-binding refinement unsuitable. Lack of conventional PPII recognition is supporting context, not a condition invented for GO:0017124. The newer collagen-IV MST assay is not assigned to COL7A1. No native-colocalization prerequisite or denial of the original interaction is introduced. PMID:37080980
The original tear XML table explicitly contains CO7A1_HUMAN/COL7A1; this is contextual extracellular detection rather than a target-specific cancer difference or tear activity. Original colon/liver workbook views both identify the same COL7A1 group, including Q02388 and Q02388-2; they are not independent experiments or product-resolved detection. Authentic marrow ECM Methods and S2/S3 target rows were inspected with their donor, disease and threshold columns. The structural classification is supported by the GO collagen convention and independent matrix experiments, not a force measurement in proteomics. [PMID:22664934; PMID:25037231; PMID:28344315]
PMID:28675934 has a partial selected cache despite its machine full-text flag. The purported XLSX and reused HTML responses are challenge pages, not readable target workbooks. Its exact COL7A1 link remains UNVERIFIED. Positive structural and ECM judgments rely on independent target evidence; no newly recovered target row is claimed. The genuine PMID:22261194 abstract describes porcine cardiac ECM experiments. Exact pig COL7A1 F1SKM1 has the relevant HDA/RCA records, but its target supplement was not recovered and no human target experiment is inferred.
The two NAS assertions from PMID:29853175 remain UNDECIDED because its target-specific full passage was not recovered. This is a source-access judgment, not a denial of the established trimer or structural adhesion. The separately inspected original PMID:21421911 Table 1 explicitly lists the alpha1(VII) homotrimer; its selected abstract remains unchanged.
Actual original PMID:8170945 human pedigree and mutation text establishes positive dominant dystrophic-EB association and anchoring-fibril context. It does not demonstrate a COL7A1-performed formation-to-maturity step of epidermis development, so that source-specific process is over-annotated. Actual PMID:23154389 reports COL7A1 transcript upregulation during endoderm induction, but the tested collagen matrices were I, III, IV and V, and the integrin interventions concern other entities. The expression observation is preserved while the target-performed differentiation inference is over-annotated. Neither judgment asserts that COL7A1 can have no developmental role.
All original source identities and qualifiers are preserved. Current exact donor identities and same-term records are distinguished from historical PAINT placement, which was not reconstructed. A short donor list or a target-like donor is not treated as circularity or failure. The rat Col7a1 same-term IMP supports the current assembly premise. ARBA00028691 has actual target-matching InterPro/taxon and FunFam/taxon branches. ARBA00098055 returned 404; that retrieval limit does not invalidate the independently supported homotrimer. SL-0025 is a subcellular vocabulary mapping, not a donor protein.
All 26 authentic Reactome event summaries and input/output names were read. Complete reference-identifier tracing is available only for its bounded recovered portion; a separate set of paths reaches explicit COL7A1 labels in 13 events. Other expansions are incomplete after preserved 522/timeouts and 429 responses. No complete target graph is claimed where unavailable, and no missing graph edge is treated as negative biology. Generic transport summaries do not themselves name COL7A1; the cargo assignment is inherited from GOA and supported by independent secretory-collagen biology, with incomplete graph expansions explicitly retained. Existing secretory cargo locations remain non-core; these do not assign coat, kinase, GTPase, receptor or fusion activities to collagen VII. The ERGIC membrane annotation is refined to the encompassing compartment: the actual membrane definition denotes the bilayer, while the inherited model includes secretory collagen cargo. The current alternative child is TRAPPI complex, not a more appropriate cargo compartment.
The initial normal fetch left 22 Reactome title placeholders. They were filled by the existing deterministic cache-title reader using genuine selected same-ID caches, with the seed archived and an explicit title-change ledger. Four missing caches were produced solely by the unchanged normal Reactome formatter from already authenticated API responses; this is recorded as offline formatting, not a successful fresh network retrieval. No scientific citation prose was fabricated.
The genuine isolated normal gene fetch and PMID fetch completed. Default Falcon reached its configured timeout; the configured perplexity-lite fallback returned HTTP 401 insufficient quota. Both terminal receipts and all genuine outputs are retained. These manual notes do not impersonate a provider report. Protected current-main sources and authentic normal variants remain separate. Two later shared-source composition choices preserve the fixed research inputs: publication uses the existing full PMID:21421911 cache from PR4919/4938 (same identity and exact abstract, no additional full passages claimed as read), and the existing PTHR37456 metadata from PR4934, differing only in its fetch timestamp. Both original variants and exact differences remain pinned. The source index publishes exact record/table projections, definitions, access limits and original identifiers; the full TMP packet retains raw responses and failed attempts.
Independent bounded core and process consultations were consumed with their actual read limits. Whole review and final source/tooling applicability are separate from those consultations. Final ordinary validation, derived-status check and render pass in an isolated publication-composition context. Three advisories are intentional: positive MIA3 generic binding is retained under the campaign exception; broad extracellular-region tissue detection remains non-core alongside structural-location assertions; and the inaccessible NAS homotrimer source remains UNDECIDED alongside independently supported homotrimer rows. The final status is DRAFT. The initial status mismatch and all failed source/retrieval attempts are preserved.
The genuine raw GOA file contains 55 rows; the normal seed collapsed two occurrences of the same MIA3 source assertion into one object, producing 54 preserved originals. This normal deduplication is explicitly reconstructed; neither the raw rows nor either literal interaction record was discarded. The final review adds only the independently supported collagen-binding assertion.
The isolated validator added GO:0005590 to its temporary GO cache through normal lookup; both cache variants and the one-line addition are retained, without resetting runtime context. Final publication composition validates against the shared full PMID:21421911 variant and shared PTHR37456 timestamp variant while the original research inputs remain fixed. No extra full-paper read is claimed.
The eleven extracellular-region Reactome assertions remain ACCEPT because they describe mature collagen VII secretion, matrix contacts and anchoring assemblies. The broader location need not be repeated beside the more specific basement-membrane core location. The tear-fluid HDA assertion remains non-core sample detection, with no separate tear function inferred. Identical GO terms can therefore retain different source-specific core assessments.
Direct collagen binding is explicit in the single anchoring-fibril structural core. The full-chain and NC1 binding assays describe matrix contacts that help make the same extracellular structural connection. GO:0005518 remains a distinct, nonredundant MF assertion; this integration does not claim it is an ontology child of GO:0030020. A second core would repeat this biological activity, and contributes_to_molecular_function would incorrectly suggest a different relation to the directly measured binding.
The external_primary_excerpt entry in the source-evidence index makes the original PMID:11274208 negative assay checkable: it contains one nine-word literal excerpt, the DOI, public author-upload route, construct composition, positive control and exact indexed-text locators. The preserved PMID cache is still abstract-only. External full-text reading does not make that cache full text, while its limited cache coverage does not mean the original was unavailable. The automatic inhibitor inference remains removed at the tested scope; no universal negative annotation is added.
ER and COPII summaries now identify the individual event. Generic transport summaries are explicitly distinguished from events naming procollagen VII, and incomplete participant graphs remain incomplete. The six GO_REF assessments identify their actual evidence methods, with manual and Compara transfers distinguished from PAINT. Original annotation objects, actions, products, references, historical notes and caches are preserved.
Ordinary follow-up validation, status checking and rendering passed with the same three advisories: supported MIA3 generic binding under the campaign exception, source-specific extracellular-region actions, and the unresolved NAS trimer source. No source or action was altered to suppress those advisories.