Human LPA encodes apo(a), the distinctive glycoprotein component of lipoprotein(a). The foundational cDNA paper describes a plasminogen-related architecture: "It contains a serine protease domain and two types of plasminogen-like kringle domains, one of which is present in 37 copies." PMID:3670400 That 37-repeat clone is one allele, not a universal protein length. Direct mapping across alleles found that KIV types 3-10 are invariant in count while KIV-2 varies: "Our analysis demonstrates that the number of kringles IV 3-10 is invariable in the human apo(a) gene, suggesting that the 3'domain of Apo(a) is functionally important." PMID:9524278 UniProt likewise states that individuals may encode 2-43 KIV-2 copies and that the P08519 reference allele contains 15 KIV-2 copies. Therefore all residue numbers and the 16 total KIV fragments in the cached P08519 record are reference-allele-specific.
The original biochemical sequence analysis identified the decisive activation-site difference: "Plasminogen is activated by the cleavage of a specific arginine residue by urokinase and tissue plasminogen activator; however, the corresponding site in apo(a) is a serine that would not be cleaved by tissue plasminogen activator or urokinase." It also reported: "Using a plasmin-specific assay, no proteolytic activity could be demonstrated for lipoprotein(a) particles." PMID:3472206
The later recombinant study tested the issue more directly. Restoring arginine permitted tPA cleavage, but "both wildtype r-apo(a) and the mutant, either free or incorporated into r-Lp(a) particles, were uniformly inactive against a variety of chromogenic serine protease tripeptide substrates." The authors concluded that substitutions beyond the activation-site change render apo(a) inactive. PMID:7495809
In tension with those studies, the GOA source paper reports fibronectin fragmentation and says: "The proteolytic activity of Lp(a) was localized to apo(a) and experiments with inhibitors indicated that the proteolytic activity was of serine proteinase-type." PMID:2531657 The cached record is abstract-only even though a PMCID is present, so purification, activation, inhibitor, and contamination controls cannot be reassessed here. This experimental annotation must receive curator deference rather than a claim of curator error, but the conflicting primary evidence and the activation-defective sequence make apo(a) serine endopeptidase/autoproteolytic activity disputed and unsuitable as an unqualified core function.
The local reproducible analysis independently maps an intact H1861/D1904/S1990 charge relay but a noncanonical S1819|I1820 activation junction. Its own limit is explicit: "This sequence result cannot establish absolute inactivity." [file:human/LPA/LPA-bioinformatics/RESULTS.md]
Recombinant particle experiments localized assembly to extracellular material and identified the covalent linkage: "Using site-directed mutagenesis, we demonstrated that Cys4057 in apo(a) is involved in disulfide linkage with apoB-100 in Lp(a) particles." PMID:8366120 An independent HepG2 mutagenesis study found that Cys4057 substitutions yielded free apo(a) without detectable lipoprotein-associated apo(a), supporting an essential disulfide-mediated assembly step. PMID:7505444 Historical residue number Cys4057 is allele-dependent; it denotes the homologous unpaired cysteine in apo(a), not residue 4057 of the shorter P08519 reference sequence.
The canonical strong lysine-binding site is in KIV-10, with weaker sites in other kringles. Direct recombinant-domain measurements found KIV-7 affinity for lysine analogues about tenfold weaker than KIV-10 (epsilon-aminocaproic acid Kd 230 +/- 42 versus 33 +/- 4 micromolar). PMID:11802713 A natural Trp-to-Arg human variant in the historically numbered kringle 4-37 abolished lysine-Sepharose binding, independently tying this kringle site to Lp(a) lysine binding. PMID:7918682
The fibronectin paper directly states: "The binding of Lp(a) was localized to the C-terminal heparin-binding domain of fibronectin." PMID:2531657 This supports fibronectin binding. It does not demonstrate that apo(a) itself binds free heparin; "heparin-binding domain" names the region of fibronectin. The GOA heparin-binding NAS annotation therefore lacks direct support in the cached source.
The KIV-2 yeast two-hybrid screen and coimmunoprecipitation study confirmed APOH interaction in human plasma and recombinant cell supernatants: "Coimmunoprecipitation experiments confirmed that beta-2 glycoprotein I and apo(a)/Lp(a) interact in human plasma and in cell culture supernatants of COS-1 cells, which ectopically expressed apo(a)." PMID:9269765 This is a real interaction but does not by itself define the core function of apo(a).
Surface context matters. One study explicitly reports failure to demonstrate direct tPA inhibition by Lp(a) in solution, but shows that surface-bound Lp(a) binds tPA and reduces plasminogen activation. PMID:1829635 A later recombinant domain-dissection study found strong inhibition on native and degraded fibrin cofactors and showed that deleting the protease domain did not abolish inhibition: "A variant lacking the protease domain also exhibited strong inhibition, indicating that the apo(a)-plasminogen binding interaction mediated by the apo(a) protease domain does not ultimately inhibit plasminogen activation." PMID:12697748 This supports a kringle/template-dependent, noncatalytic antifibrinolytic mechanism rather than direct catalytic protease inhibition.
The transgenic-mouse paper provides in vivo-context corroboration: "We show here that the activation of TGF-beta is inhibited in the aortic wall and serum of mice expressing apolipoprotein(a), as a consequence of apolipoprotein(a) inhibition of plasminogen activation." PMID:8047165 It does not establish generic endopeptidase-inhibitor activity as a molecular function.
All newly fetched decisive primary papers are abstract-only in the local cache, including the PMC-indexed PMID:3472206 and PMID:7505444. Findings in the review use verbatim cached excerpts and set full_text_unavailable: true. PMID:2531657 remains especially important but disputed: the curator may have read assay detail not present in the cache, while independent primary experiments argue against a catalytically active apo(a) protease domain.