Human factor I supplies the protease step in complement regulation. PMID:21768352(https://pubmed.ncbi.nlm.nih.gov/21768352/) describes the mature serum-derived protein's resting structure. I read its purification, crystal, loop-disorder and ternary-model Methods/Results. The heavy/light chains originate in a single precursor; their processing is not a pair of alternative splice products. The authentic seed has no alternative-product objects, which is preserved.
PMID:28671664(https://pubmed.ncbi.nlm.nih.gov/28671664/), selected complex Results and protein-preparation/cofactor-assay Methods, supplies a human C3b/miniFH/FI complex and separate active-FI cleavage experiments. Its crystallographic FI is S525A; miniFH joins CCP1–4 and 19–20. Those engineered constructs are not described as wild-type full-length factor H cleavage experiments. Cofactor/substrate contacts stabilize the protease, and different cofactors/conditions permit different cleavage products. CFI prevents further convertase assembly while generating receptor-recognized fragments. That supports both negative complement regulation and functional protein processing; CFI's own precursor maturation is a separate event. Mature-sequence numbering in the 2011 paper is distinguished from precursor numbering in the 2017 source.
The original extracellular-region assertions are retained as core location. PMID:6327681(https://pubmed.ncbi.nlm.nih.gov/6327681/) is directly summarized by its authentic abstract; the full original body was unavailable. The original human cDNA reference 2956252 and mouse/comparative cloning reference 8604219 remain accurately scoped to abstracts, with independent human mechanism supporting the correct protease activity. The human-colostrum study 16502470 lacks an independently inspected CFI target table; it is not represented as a verified peptide identification.
All 10 selected Reactome summaries were read. These are provenance for the original TAS location, not new microscopy. Surface-associated substrate/cofactor complexes can transiently attach soluble factor I to the extracellular side of a membrane. GO:0016020 includes attached proteins and complexes. The literal InterPro sources for that electronic row are SRCR/IPR001190 and SRCR-like/IPR036772, not the separate LDL-A domains. Domain presence alone is insufficient, but Reactome R-HSA-977359/R-HSA-977602 independently supplies the limited peripheral-complex context; no integral topology or lipid-binding function is inferred.
The two generic binding annotations remain non-core under the pinned campaign instruction. PMID:16237761(https://pubmed.ncbi.nlm.nih.gov/16237761/), genuine original PDF Methods/Table 1, reports a 98%-similar human factor I library clone with genotype 1b HCV-F bait. The preserved P0C045 accession maps to genotype 1a H77, and the exact CFI insert was not resolved. Its follow-up co-IP is FBP2, not CFI. These limits constrain the interaction without denying the positive library hit. PMID:23864651(https://pubmed.ncbi.nlm.nih.gov/23864651/), original indexed Results/Figures 2–4, explicitly includes CFI in tagged MYTH/CHO co-IP. Exact positive IntAct records independently match that pair and assay context. No endogenous receptor-regulatory or physiological viral mechanism is assigned.
For the urinary-exosome HDA, the authors' original NIH database contains CFI/NP_000195, one peptide, linked by reference 2 to PMID:19056867. This verifies a target identification in that preparation, not luminal topology or a role in vesicle formation. PMID:23533145 full main text Methods/Results/Discussion was read, but its separate EPS-urine target table remains unresolved; this source-specific annotation is UNDECIDED. The paper itself discusses possible soluble and blood-derived material. That is an uncertainty about the exact preparation assignment, not proof of contamination of this specific target or a false original annotation.
CFI-source-evidence.json preserves exact annotation objects, typed ontology data, partner/donor identities, two positive IntAct records, the original-author urinary target row and primary-access limits. Its data keys are explicit evidence anchors. The old normal caches for 23864651/16237761 omit important Methods/Results despite full-text flags; original external access was recorded separately and the cache bytes were preserved. Other failed HTML payloads are not treated as primary papers or spreadsheets. No provider report was fabricated: the genuine Falcon attempt timed out after 90 seconds and the explicit perplexity-lite fallback failed with an insufficient-quota response. Manual primary research supplies this review.
The original 25 assertion objects and 23 reference identities remain exact. Two genuine additional primary caches support mechanism; the source-evidence artifact is declared separately. No NEW annotation is proposed. The negative-regulation and endopeptidase recommendations refine existing assertions, and do not manufacture a new evidence code or silently change WITH fields. Historical PAINT node placement was not reconstructed; a broad or diverse donor set is not treated as weak evidence. The independent bounded membrane/processing consultation agrees with the source-specific choices above.
This entry supersedes the earlier membrane-retention and processing-core recommendations without rewriting those historical notes. The electronic membrane row is now MARK_AS_OVER_ANNOTATED: its actual SRCR/SRCR-like domain sources do not establish a membrane-localized factor-I pool. Reactome R-HSA-977359/R-HSA-977602 still supply legitimate surface-substrate/cofactor encounter context. The conclusion is limited to this broad electronic inference; it neither denies indirect attachment nor imposes a universal direct-lipid-binding or residence-time requirement on GO:0016020. Extracellular location and actual cofactor-dependent cleavage remain supported.
GO:0016485 is specifically framed as maturation toward full functional capacity. The earlier interpretation did not sufficiently distinguish that wording from CFI's regulatory inactivation of C3b/C4b convertase activity. The broad inherited processing assertion is retained non-core because the generated iC3b/C3dg fragments have real receptor-recognition functions (PMIDs:28671664,21768352), but it is removed from the core process list; negative complement regulation remains. This does not reduce cleavage to nonspecific destruction, infer the PAINT ancestral node from a few donor identities, or confuse CFI's own maturation with its substrate processing.
The two exosome rows concern different preparations: the PMID:19056867 author resource supplies one CFI peptide, whereas the PMID:23533145 EPS-urine target row remains unverified. The source-specific endopeptidase refinement is correct but duplicates independently supported specific activity rather than adding coverage. The GLP1R retention cites its positive target records, not the source's artifact caveat as if it were positive evidence; the engineered-assay limits remain. PMID:16237761 remains VERIFIED for the authentic positive library result and citation: genotype/accession mismatch, unresolved insert and absence of CFI follow-up constrain inference, but do not establish paper unsoundness.
All 25 machine objects, all 26 reference identifier/title pairs and the absent alternative-product slot remain exact. No NEW assertion or cache/provider file is added, and no unchanged failed provider attempt is repeated. The complete previous notes remain the byte-exact prefix. The appended source-artifact follow-up records current term definitions and supersession explicitly; normal validation/status/render and aggregate quotation checks accompany the finite patch.
The follow-up to review 5478805606 and comment 6097065192 corrects two propagation classifications while retaining their biological decisions. For GO:0006956, the recommendation to use GO:0045916 identifies factor I's negative regulatory role more precisely. For GO:0008236, the recommendation to use GO:0004252 identifies its demonstrated endopeptidase subtype. Both are term-scoping refinements and now use TERM_SCOPING_PROBLEM, as defined in the review schema and propagation guidance. This supersedes the previous NO_FAILURE_CORE metadata for those two rows; it does not challenge the mouse donor identity or assert a reconstructed PAINT ancestral history.
The retained GO:0016485 protein-processing assertion keeps NO_FAILURE_NON_CORE. The optional suggestion to add a propagation assessment for the membrane IEA row was considered. Its existing reason and source-evidence artifact already record the verified IPR001190 and IPR036772 mappings and the target-localization limitation. No additional source-status confidence is assigned here, and the existing over-annotation decision remains unchanged.
All original source fields, annotation actions, replacement terms, evidence anchors, reference assessments, core functions and product information remain unchanged. This entry introduces no new literature quotations and preserves the complete preceding notes.