The complete direct decision table preserves each source annotation and its zero-based intake index. The binding table adds verified partner identities and assay methods. Evidence-supported BMP binding and fibronectin binding replace six generic labels; other positive bare binding remains non-core under the project instruction. These terms describe the measured binding, without adding a cellular-component or partner-derived catalytic function. The product evidence separates two preserved splice records from the two applied processed-product classes.
Zhang et al., 1994, Methods p. 856 and Figures 5–9, distinguishes bovine microfibril immunogold from human fetal immunostaining. The 2003 coassembly study, Figures 5/6/9/10, independently images human FBN2 microfibrils. Elasticity remains a positive matrix-level inference, prioritized below the broader structural role that also covers nonelastic matrix. No isolated-protein recoil measurement is claimed.
The osteoblast study uses mouse loss-of-function cultures and recombinant FBN2–LTBP assays (Figure 4). Matrix sequestration regulates signaling; even double-fibrillin-deficient cultures can mineralize. Neither a generic calcium-phosphate scaffold requirement nor human in-vivo perturbation is inferred.
Human tissue localization is positive but spatially bounded: 15277214, Results and Figure 3B, directly assays FBN2 in renal GBM despite its FBN1-focused title; 24899048, Figure 2, emphasizes the central elastic layer of Bruch membrane, which does not itself prove basal-lamina incorporation. Its generic basement-membrane decision instead relies on the independent renal and other ocular localization evidence; 25406291, Figures 2–5, shows stage-specific ocular deposition including Descemet membrane and anterior lens capsule. Eye-process decisions retain a contextual structural/IEP inference and curator deference, not a demonstrated human morphogenetic perturbation. The two eye terms have no verified strict ancestor path. FBN1/2/3 in the eye paper are paralogs, not FBN2 splice isoforms.
Stable original locators are 12429738, Figures 8–9; 15131124, Figures 2–4; 18339631, Figure 6 and Table 3; 25034023, Figure 3; and 26601954, Figure 6. These experiments test recombinant FBN2 regions, with bovine MAGP partners explicitly retained. BMP binding uses prodomains: Figure 6 and Table 3 report rF86 affinities of 8, 25, 15, 13 and 24 nM for BMP2, BMP4, BMP7, BMP10 and GDF5/BMP14, respectively. GO:0036122 names binding to a BMP-family member without a mature-domain restriction. Figure 3B of PMID25034023 independently supports GO:0001968 fibronectin binding. Neither refinement establishes mature growth-factor activity, the sign of downstream signaling or an additional core function. B7ZLY3 identifies human LTBP1, without evidence that this exact transcript product was separately tested.
23658023, Table 1, reports FBN2 in human HUES1-derived ECM separately from mouse feeder data. 28675934, original supplementary workbook pr7b00191_si_002.xlsx, has sole-FBN2/P35556 rows B557,D5844,D6245,D7043; mixed descriptions remain separate. Extraction/MS identifies the matrix compartment, not intact molecular assembly or natural product boundaries.
28327460, main Table I, swaps fibrillin/fibulin names beside FBN/FBLN symbols. The exact FBN2 supplemental accession join remains unresolved. Independent scaffold evidence supports its generic RCA structural claim, but does not reconstruct the particular HDA identification. This is the reason that HDA remains undecided; no misattribution is asserted. MatrisomeDB 2 describes database methodology, but its FBN2 target page was unavailable. The interstitial-matrix decision instead retains the matching generic claim using direct human connective-tissue imaging and curator deference, without claiming a recovered database hit.
32329225, Results/Figures 2–5 and expression/IP-MS Methods, supports a released human C-terminal product and recombinant biological activity. Methods use 2780–2912, whereas Figure 2 caption says 2779–2912. Reviewed UniProt uses 2780–2912; these are not newly mapped endogenous termini. UniProt P35556 explicitly records linked secretion followed by furin separation with ECO:0000269/PMID32329225. That curated model and the positive released-product evidence are retained. The primary article identifies the site by PiTou prediction (Figure 1), reports furin colocalization (Figure EV1), and uses digested IP-MS for secreted products (Figure EV3 and Methods); these do not independently map the intact endogenous terminus or isolate furin necessity. Mouse hepatocyte metabolic responses and human trophoblast invasion are distinct from serum pregnancy associations. The truncated 1,473-aa splice product lacks the entire mapped hormone region.
The normal publication caches remain unchanged, including their abstract-only flags. Original text recovered separately can support a manual assessment even when the committed cache contains only an abstract. Actual read scopes and unresolved evidence are stated in the complete reference assessments; no quotations were added.
Reviewed UniProt P35556, FUNCTION and FT DOMAIN, supports the early elastic-fiber assembly role and identifies the target calcium-binding EGF-like architecture (for example, residues 276–317 and 318–359; PROSITE-ProRule PRU00076). The assembly FUNCTION sentence uses ortholog evidence ECO:0000250/Q61555. PMID8120105, target sequence analysis and Discussion p. 862, is primary target-architecture evidence; its localization supports a microfibrillar scaffold but is not a direct force or calcium-affinity measurement. The structural core includes elastic-fiber assembly and independently documented nonelastic matrix roles.
UniProt DISEASE records congenital contractural arachnodactyly (CCA; MIM121050). PMID24899048, the family Results and Discussion, identifies a segregating rare variant in early-onset macular degeneration and distinguishes that phenotype from CCA. These contexts are described without assigning a single loss-of-function mechanism, new developmental process or placensin mechanism to all FBN2 variants.