The exact normal seed contains 158 distinct source annotations (159 raw GOA rows). The imported seed is still PENDING. UniProt describes apoB100 as the structural lipid carrier and LDL-receptor ligand on VLDL/IDL/LDL, and apoB48 as the intestinal RNA-edited product on chylomicrons. The RNA-editing product is distinct from an alternatively spliced product; the seed has no alternative_products entries or isoform-specific GO rows, and that source structure will be preserved.
PMID:39662503(https://pubmed.ncbi.nlm.nih.gov/39662503/), “The structure of apolipoprotein B100 from human low-density lipoprotein”, DOI 10.1038/s41586-024-08467-w. The original publisher sample-preparation Methods explicitly uses human LDL purified by density-gradient ultracentrifugation and then size-exclusion chromatography. The reconstruction and integrative modelling support a circumferential apoB100 scaffold associated with the particle lipids. Structural resolution varies; the authors describe greater reliance on AlphaFold2 for some C-terminal inserts. This is evidence for structural lipid carriage, not a measured catalytic lipid-transfer reaction. Read scope: complete abstract plus targeted reconstruction Results, sample-preparation Methods and limitations; no exhaustive supplementary review.
PMID:39663455(https://pubmed.ncbi.nlm.nih.gov/39663455/), “Structure of apolipoprotein B100 bound to the low-density lipoprotein receptor”, DOI 10.1038/s41586-024-08223-0. The abstract reports cryo-EM apoB100–LDLR interfaces involving the apoB beta-belt and N-terminal region. This informs the established LDLR-binding role. Read scope is the official complete abstract and publisher preview/data headings; the subscription-limited main Methods and exact reagent provenance have not been verified. Do not infer that every uptake-associated phenotype means apoB performs the receptor's endocytic machinery step.
One ordinary two-source cache-fetch batch was attempted once with a 45-second bound and produced no output files. It was interrupted as a batch; this is not evidence that both individual requests completed and failed. Original caches and source objects remain unchanged. Bounded recovery and actual source closure are pending.
The 29 seeded PMID records were read through their complete available abstracts; PMID:7126555 contains bibliography without an abstract. All 67 normal Reactome summaries were read. A cache being present is not equivalent to having the primary Methods or target-specific supplementary table. The configured deep-research attempt returned no report, so these are manual research notes. Existing normal caches remain unchanged, including abstract-only records for which a targeted original-paper section was subsequently accessible externally.
PMID:24447298(https://febs.onlinelibrary.wiley.com/doi/10.1111/febs.12721), original Results, Table 1, peptide-design/NMR/SPR Methods and limitations were read. Synthetic apoB site A and site B peptides bind recombinant LDLR repeat LR5; a randomized site A control has little binding. The peptide sequences, excluding the engineered WG prefix, exactly match the human P04114 record. This establishes compatible human sequences rather than uniquely human species origin. Calcium and pH alter affinity. The experiment supports LDL-receptor binding, while the authors distinguish isolated-peptide results from intact-particle binding and cellular uptake. It does not demonstrate a cargo-to-coat adaptor function. A small reproducible sequence check is retained in the local research evidence.
PMID:26224785(https://pubmed.ncbi.nlm.nih.gov/26224785/) was read in the cached targeted Methods and Results. COS7 transfections include an explicitly human apoB48 construct and human MTP. Reciprocal apoB17/MTP co-immunoprecipitation is reported, but the apoB17 reagent's species is not independently resolved merely from the apoB48 construct. The D169V MTP defect separates loss of lipid transfer/PDI association from retained apoB17 association. ApoB receives and carries lipid; this does not give it MTP's lipid-transfer chemistry. PMID:25108285(https://pubmed.ncbi.nlm.nih.gov/25108285/) independently reports apoB17/48 association with MTP mutants in the complete abstract; only that scope has been read.
PMID:27138255(https://pubmed.ncbi.nlm.nih.gov/27138255/) targeted cached Results, Figures 6–7, Discussion and co-immunoprecipitation Methods were read. Endogenous apoB in human HepG2/Caco2 cells associates with TANGO1/TALI, and apoB is recruited to ER exit sites. The paper explicitly leaves direct binding versus an intervening adaptor unresolved. Co-immunoprecipitation and cargo recruitment support the reported interactions without assigning apoB the receptor's adaptor activity. Accumulation in autolysosomes after knockout is a perturbation-dependent destination, not a newly inferred normal core activity site. The original Q96PC5-7 partner identifier is retained; it is not an APOB isoform assertion.
PMID:28183703(https://www.researchgate.net/publication/313546565_AUP1_Ancient_Ubiquitous_Protein_1_Is_a_Key_Determinant_of_Hepatic_Very-Low-Density_Lipoprotein_Assembly_and_Secretion), original author-uploaded Results around Figure 5 and the corresponding Discussion were read. HepG2 imaging places some apoB/AUP1 colocalization at lipid-droplet surfaces, particularly with lipid loading and proteasome inhibition. PLA, cross-linking and co-immunoprecipitation corroborate association; these are not a purified binary interface assay. The authors relate this pool to apoB quality control and degradation. This supports a contextual lipid-droplet location without making it the principal site of apoB's circulating carrier activity. Supplementary Methods and all image pixels were not independently inspected.
The actual Human Protein Atlas APOB subcellular table reports HPA049793 staining: vesicles in CACO-2 and vesicles/cytosol in Hep-G2, with cytosol rated supported. This is positive experimental localization evidence, even though the protein is also secreted. No image-based reassessment of antibody specificity is claimed.
The full abstract of PMID:15797858(https://pubmed.ncbi.nlm.nih.gov/15797858/) identifies human R3480P carriers, reduced LDL-receptor affinity and altered VLDL-to-LDL turnover. Original publisher full-text routes remained inaccessible. The current GO:0120020 definition describes extracting cholesterol and carrying it through water in a protected pocket to an acceptor. Whole lipoprotein-particle carriage does not by itself establish that narrower molecular reaction. The IMP assertion therefore needs its own assay-level assessment, not removal from abstract silence. The corresponding PAINT inference also retains an unresolved source/node boundary; the target's own experimental evidence in WITH/FROM is valid, not circular. One correctly configured private PAINT fetch failed at DNS resolution with no node slice; an earlier CLI invocation stopped before networking because its private membership input was absent.
PMID:27477018(https://pubmed.ncbi.nlm.nih.gov/27477018/) remains a complete-abstract read. It reports LDL/TREM2 binding and uptake, while its APOB-specific reagents and assay panels have not been inspected. Primary publisher and author routes did not expose those sections. Other papers' descriptions of that experiment are not treated as a direct read. GO:0050750 describes binding to a low-density lipoprotein receptor; it does not explicitly restrict the receptor to the LDLR gene. The original TREM2 partner must therefore be assessed on its actual assay scope rather than rejected merely because its symbol is not LDLR.
Reactome represents apoB-containing particles during extracellular circulation, receptor engagement, internalization and degradation. An apoB molecule can be present in an endosomal or lysosomal lumen as transported cargo without carrying out the receptor's uptake machinery or the lysosomal hydrolase's reaction. Conversely, apoB's physical lipid-binding scaffold contributes directly to particle assembly and lipid transport; it is not merely an inert substrate in those processes.
The LDLR and chylomicron-remnant summaries specify different recognition routes: apoB100 binds LDLR, whereas remnant uptake can use apoE on apoB48-containing particles. MTP lipidates nascent apoB in the ER, LPL hydrolyzes particle triglyceride, CETP transfers neutral lipid between particles, and LIPA hydrolyzes internalized ester. None of those activities is assigned to apoB simply from membership in the reaction. The late coated-vesicle/uncoating summaries require source-specific care: a general AP2/HSPA8 event does not by itself resolve an APOB plasma-membrane location at that stage. The two coarse cytosol summaries involving chylomicron secretion or retinyl-ester packaging likewise do not establish which side of a membrane the apoB polypeptide occupies.
For the foam-cell and lipid-storage annotations, the available abstracts describe macrophages exposed to LDL, enzymatically modified LDL or infection plus LDL. They do not yet distinguish apoB-specific regulatory work from provision of a lipid-rich particle. These experimental assertions will remain unresolved where the relevant full assay cannot be checked; the evidence is not rejected because lipid delivery is necessary for the outcome. No new biological-process assertion is proposed at this stage.
The initial PENDING and pending-recovery statements above record the earlier research state. All 158 original assertions are now reviewed: 76 ACCEPT, 38 KEEP_AS_NON_CORE, 43 UNDECIDED and one own-source MODIFY. There are no NEW assertions. All machine source objects, the 159-row GOA input and the UniProt record are preserved; no product or isoform record has been invented. Two core functions cover lipid carriage/structural particle assembly and apoB100 receptor recognition. Individual contextual interactions are retained under the user-supplied ActionEnum when positively supported; unread pairs remain uncertain. Genericity alone is not treated as evidence that an interaction is false.
The two additional normal records were recovered once and imported unchanged. PMID:39662503 has normal XML full text; its human LDL reconstruction Results, limitations and extracted Methods were read. PMID:39663455 remains abstract-only, and the normal availability flag stays true even though its published structural claim is informative. No full-paper read is inferred from cache recovery. The original provider produced no report; no provider research file was manually fabricated.
The corrected R-HSA-2512800 decision acknowledges SR-BI endocytosis of oxidized LDL inferred from mouse studies as well as partial endocytic contribution to selective uptake in human cells. This supports a contextual endocytic cargo location. Selective lipid transfer alone would not establish apoB polypeptide uptake; the actual positive whole-particle passage is not omitted. The authored GO:0050750 label is the current “low-density lipoprotein particle receptor binding”; original GOA labels remain untouched.
Two production GO-CAMs were inspected at the actual APOB activity level. Model 632cf60a00000028 records receptor binding and LDL clearance; its RNA-editing chemistry belongs to APOBEC1 and cofactors rather than APOB. Model 6606056e00001418 records human APOB ligand activity activating TREM2 within cellular response to lipoprotein particle. That considered curator judgment is acknowledged while the underlying APOB-specific primary assay remains unresolved in this review. Neither model is treated as a new-annotation gap. The PAINT node/MSA could not be recovered in the single configured attempt; no negative inference is made from donor count or valid self-derived evidence.
An independent all-158 prospective consultation accepted the source-specific map after the endocytosis correction. The schema permits one complex per core-function object: carrier core prose names all supported particle classes, with VLDL as the representative complex and LDL in the receptor-recognition core. This avoids duplicating identical molecular functions merely to enumerate complexes. Final validation and attachment review follow separately.
Full validation and rendering passed. The 15 deliberate advisories comprise seven positively supported contextual protein interactions retained under the explicit user ActionEnum, and eight differences between source-specific decisions for the same term. These differences distinguish inspected cargo events or independently established human biology from uninspected target entities, assay details and receptor contexts; they are not contradictions about a single experiment. The two uncertain IBA rows now carry UNRESOLVED propagation metadata without claiming the node/tree was read. Raw GOA/UniProt and all 98 normal PMID/Reactome dependencies remain unchanged.
This entry supersedes earlier source-access gaps only where the primary entity hierarchy has now been inspected. All 158 original annotation objects, qualifiers and source identifiers remain unchanged; no new annotation is added. The published history is retained verbatim, including its original spacing. This revision receives a new history record.
The seven experimentally reported generic associations remain KEEP_AS_NON_CORE. The explicit user ActionEnum permits that action and defines REMOVE as unlikely correct on combined evidence; these instructions outrank the skill's generic-binding removal default. Neither an uninformative term nor another review's removal precedent establishes that a true association is incorrect. No new specific activity is invented for apoB merely because a partner has a known function. The two unadjudicated pair rows remain UNDECIDED, and the same-source peptide/receptor refinement to GO:0050750 is preserved. This is an explicit policy disagreement.
The published file actually has 20 GO:0005886 rows: four NONCORE and 16 UNDECIDED, rather than the reviewer's 19 and two/seventeen split. All 20 normal summaries were reread. The current GO plasma-membrane term includes associated proteins; it is not restricted to proteins spanning the bilayer. This does not make every receptor ligand an integral membrane protein.
The actual primary component chain was inspected: CLASP proteins:cargo, R-HSA-8867603 includes DAB2,LDLRAP1:LDLR:LDL, R-HSA-8862887, which includes LDL:LDLR, R-HSA-171100. Its LDL component, R-HSA-171112, explicitly contains APOB(28-4563). This confirms a particle-associated cargo identity, not an APOB coat or enzyme function. The last direct click to the APOB100 child page failed; no hidden fields from that page are claimed. The explicit APOB component on the LDL page is sufficient for this identity boundary.
Ten previously uncertain rows now retain NONCORE plasma-membrane association: LDLR adaptor binding (R-HSA-8863471), BAR recruitment (8867754), CLASP/cargo recruitment (8867756), PIK3C2A recruitment (8868071) and reaction (8868072), SNX9/actin machinery recruitment (8868230), dynamin recruitment (8868236), SYNJ lipid hydrolysis (8868648), synaptojanin recruitment (8868651), and the plasma-membrane input to scission (8868661). Their actual hierarchy contains the LDL cargo during the surface-associated phase. Naming APOB in each reaction's short summary is not the criterion. The annotation does not transfer kinase, phosphatase, actin-nucleation or GTPase activity to APOB.
For six later events, the inspected source itself contains a compartment tension: HSPA8 uncoating (8868658), auxilin recruitment (8868659), HSPA8 recruitment (8868660), coat-protein dissociation (8869438), AAK1 displacement (8871193) and RAB5/GAPVD1 recruitment (8871194) have outer participants labeled clathrin-coated endocytic vesicle membrane while the nested LDL:LDLR/LDL entities remain labeled plasma membrane. These six remain UNDECIDED specifically for that location assignment. Their target identity is no longer an access gap, and their readable summaries are acknowledged. The remaining ambiguity concerns the intended nested compartment semantics after scission, rather than a claim that all apoB is absent from the cell surface. A question records this precise issue.
The four earlier NONCORE entries remain supported by their own LDL or apoE-bearing remnant uptake summaries. Remnant uptake does not imply that apoB48 supplies the apoB100 LDLR-binding surface; the apoE-dependent context remains explicit.
The current HPA APOB table reports approved vesicles and supported cytosol, alongside predicted secretion. The existing HPA NONCORE cytosol annotation is retained without reanalyzing antibody images. The statement that APOB can never be cytosolic is not used.
For R-HSA-2187332 and R-HSA-2395764, the source identity is now explicit: nascent chylomicron R-HSA-2395787 contains ApoB48:TG:PL R-HSA-2395777, which contains APOB(28-2179), R-HSA-2395786, mapped to UniProt:P04114 and labeled cytosol. The two assertions retain UNDECIDED because the physical topology of that nascent particle during retinyl-ester loading and secretion is not resolved relative to the established secretory-lumen pool. The separate HPA observation neither proves this particle topology nor justifies dismissing it as impossible.
The official GO:0006629 definition concerns chemical reactions and pathways involving lipids. The broad existing assertion is retained consistently with the experimental GO:0008203 cholesterol-metabolic-process annotation, a descendant of lipid metabolism. The complete PMID:2563166 abstract links a human apoB100 defect to impaired LDL-receptor binding and hypercholesterolemia. This is receptor-dependent cholesterol handling: apoB supplies the particle scaffold and receptor-recognition surface. Neither annotation assigns lipid synthesis or hydrolysis to apoB. The full experimental paper remains uninspected, so retention respects the existing curated assertion with an explicit access limit. The primary ARBA00028591 page exposed only a JavaScript fallback and its REST record was inaccessible; that exact mapping has not been independently reconstructed. Lipid transport and noncovalent particle assembly are not being treated as synonyms or automatic subclasses of metabolism. In particular, the current VLDL assembly hierarchy is under component organization, while GO:0042157 lipoprotein metabolic process concerns covalently lipidated proteins and is not a replacement proposed here.
GO:0043231 is a broad intracellular membrane-bounded organelle term in the current do-not-annotate subset. That status discourages broad new annotations; it does not negate the established ER location of apoB. The existing broad row therefore remains ACCEPT. The full original PMID:7853698 review remains unread; its complete cached abstract concerns RNA editing rather than the location supporting this TAS assertion. Independent PMID:27138255 directly supports the human ER-export pool and remains attached as separate corroboration. It does not convert the original PMID:7853698 evidence into a new ER-specific annotation. Existing precise ER/ER-lumen assertions and core locations are preserved.
The core retains cholesterol transport, lipoprotein transport and VLDL assembly. Its redundant GO:0006869 parent pointer is removed because GO:0030301 is a descendant through sterol transport. The original lipid-transport annotations and the description of triglyceride and other lipid carriage remain unchanged. No extra NEW process is proposed. Repeated generic reference-review prefixes and malformed current GO/PMID identifiers are corrected without changing fetched titles, exact supporting quotations or cache bytes.
Validation and history validation pass. The 15 advisory warnings comprise seven deliberately retained true generic associations under the explicit user action definitions and eight source-specific differences in location or receptor-binding certainty. Those differences preserve distinct event, particle, assay and access scopes. Integrity checks confirm all 158 original source objects, 101 protected published files, 98 fetched titles, 62 verbatim supporting quotations and 84 availability flags; the two cores retain their original molecular functions. The ten plasma-membrane decisions are the only action changes. Final counts are 76 ACCEPT, 48 KEEP_AS_NON_CORE, 33 UNDECIDED and one MODIFY. Rendering also passes.
The retained GO:0005515 decisions implement the user's explicit instruction for this ClinGen task: keep a supported, biologically correct interaction as KEEP_AS_NON_CORE when no defensible, more specific replacement has been established, rather than remove it solely for lacking functional information. This deliberately departs from the annotation-reviewer skill's informational-exclusion recommendation. That recommendation can remove an uninformative annotation without declaring the interaction false; the latest review is correct about that distinction. Earlier references here to the supplied ActionEnum are incomplete as an explanation of authority: the basis is the explicit instruction for this task, not the enum alone, a repository-wide policy change, an exemption granted by the agent, or an observed maintainer sign-off. No external approval comment is claimed.
The seven retained rows remain the contextual NS5A, PCSK9, two MTP, TANGO1, TALI and apoA-II associations (PMID:16203724, PMID:22580899, PMID:25108285, PMID:26224785, PMID:27138255 and PMID:8245722). Their existing reasons distinguish mouse-focused PCSK9 evidence, unresolved apoB17 reagent species, and a direct versus intervening-component interface for TANGO1/TALI. Cargo association does not assign receptor/adaptor activity or MTP lipid-transfer chemistry to apoB. The two unresolved pair assignments remain UNDECIDED, and the same-source LDL-receptor refinement remains MODIFY. This instruction does not convert an uninspected experiment into positive source-specific evidence.
All 158 original source and review objects and both core functions remain unchanged. The ten retained surface-associated Reactome rows and six post-scission compartment ambiguities preserve their prior adjudication and entity-access limits.
This addendum clarifies the authority and exclusion criterion for the existing decisions. It adds no primary-source reading, biological assertion or new annotation adjudication. Earlier journal entries and validation results remain historical; this clarification does not claim reviewer approval or PR completion.
The controlling task instruction is now recorded in the repository's ClinGen curation instructions, with an immutable published copy. It entered main in commit 3eb5f8979787f246e82ecaadcb5fc6e49507ff4c and is already present on the APOB branch at dcbbfe8a47c6688f0314ae4e457d061d322c1080. The project instruction retains supported, biologically correct generic binding as KEEP_AS_NON_CORE when no evidence-backed specific replacement has been established; it requires MODIFY for an established refinement and UNDECIDED when relevant evidence cannot be accessed or adjudicated. It explicitly records the user's project-specific departure from the annotation-reviewer informational-exclusion recommendation. It does not amend the global skill or establish an external maintainer approval.
For APOB this preserves seven contextual associations, two unresolved pairs and the existing LDL-receptor refinement. The source-specific limits remain: the PCSK9 paper is mouse-focused, apoB17 reagent species is not inferred from the human apoB48 construct, and TANGO1/TALI co-association does not resolve a direct interface or transfer adaptor activity to the cargo. The instruction does not turn the two unread pair-specific assays into verified interactions. All 158 source/review objects, both core functions and their evidence remain unchanged.
The phrase “ROOT countersigned the distinct content review” in the earlier history record 2026-10-02T195518Z-codex-4a0f33 refers to an internal coordinating-agent check of the documentation. It was not a GitHub reviewer approval, a maintainer countersignature or an external approval comment. The historical record remains unchanged; this clarification and the accompanying new history record distinguish internal preparation checks from external review. The current changes-requested review is not represented as approval.
The linked Reactome entity pages in the compartment discussion remain external reading evidence rather than repository cache records: R-HSA-171112, R-HSA-171100, R-HSA-8862887, R-HSA-8867603, R-HSA-2395786, R-HSA-2395787 and R-HSA-2395777. This follow-up neither retrieves those pages nor closes their reproducibility limitation. Ten retained surface-associated location decisions and six unresolved post-scission compartment assignments retain their prior source-specific reasoning. No new primary-source reading or biological decision is claimed.