Apolipoprotein C-IV is a small, liver-secreted plasma apolipoprotein and the fourth and least abundant member of the low-molecular-weight apoC group. It is encoded within the APOE/APOC1/APOC4/APOC2 cluster on chromosome 19, immediately upstream of APOC2 and in the same transcriptional orientation, and is translated as a 127-residue precursor whose 27-residue signal peptide is cleaved to give a 100-residue mature chain carrying an N-glycosylation site at Asn-63. In plasma the protein circulates as two major glycoforms, both variably sialylated. No experimental structure has been determined for apoC-IV in any species, but the sequence has the architecture of an exchangeable apolipoprotein - predicted amphipathic helices that would lie on the phospholipid monolayer of a lipoprotein particle rather than fold into a globular domain - and the rabbit orthologue forms discoidal complexes with phosphatidylcholine in vitro. More than 80% of plasma apoC-IV rides on very-low-density lipoprotein, where it accounts for about 0.7% of total apoVLDL protein, with most of the remainder on high-density lipoprotein; the same distribution is seen in monkey and mouse plasma and in rabbit, where the protein is far more abundant. Its plasma concentration, which is roughly an order of magnitude below that of apoC-I, apoC-II and apoC-III, rises in proportion to plasma triglyceride. Raising apoC-IV experimentally raises triglyceride: mice carrying a human apoC-IV transgene become hypertriglyceridaemic with the excess triglyceride in VLDL, and coding polymorphisms in the human gene associate weakly with plasma triglyceride. The mechanism proposed for this is displacement of apolipoprotein E from the particle surface, slowing hepatic clearance, rather than any effect on lipoprotein lipase, but it has not been measured for apoC-IV directly. APOC4 transcription is induced by the liver X receptors NR1H2 and NR1H3, which act on the APOE/APOC1/APOC2/APOC4 cluster in macrophages, and in hepatocytes by PPAR-gamma/RXR-alpha together with Ku antigen; the transcript rises in steatotic and hepatitis C virus-infected liver. Beyond this the protein is genuinely uncharacterised. There is no loss-of-function model in any species, no measured receptor or enzyme partner, and no reconstituted mechanism, so what apoC-IV contributes to lipoprotein metabolism at its native concentration remains open.
| GO Term | Evidence | Action | Reason |
|---|---|---|---|
| GO:0005319 lipid carrier activity | TAS PMID:8530039 Identification and characterization of a new human gene (APO... | ACCEPT | Summary: Apolipoprotein C-IV binds lipid and rides on plasma lipoprotein particles. The cited 1995 paper is prediction-level - it inferred the activity from two potential amphipathic alpha-helical domains in the translated sequence and from hepatic expression [PMID:8530039] - but the claim has since been made experimentally. Transgenic expression of the human protein gave what its authors called unequivocal evidence that apoC-IV is a lipid-binding protein of the apolipoprotein family [PMID:8827523], the rabbit orthologue forms discoidal micelles with phosphatidylcholine [PMID:8576182], and the human protein was isolated from plasma VLDL by microsequencing [PMID:12700345]. Reason: GO:0005319 is the designated apolipoprotein molecular function. Its definition - 'Directly binding to a specific lipid and delivering it either to an acceptor molecule or to a specific location' - was checked against QuickGO, as was its history: the term was renamed from 'lipid transporter activity' to 'lipid carrier activity' on 2026-02-09 and carries 'apolipoprotein' and 'lipophorin' as narrow synonyms. Its replaces list covers the obsoleted GO:0005321 and GO:0005323 - the old high-density and very-low-density lipoprotein molecular-function terms - and it is the 'consider' target for the obsoleted GO:0005320 'apolipoprotein'. APOE and APOA1 carry it by IDA and IEA respectively. TAS is the honest code for a 1995 author inference, and the inference turned out to be right, so the row stands as written rather than being weakened to a bare lipid-binding term. Independently, a hydrophobic-moment scan of the mature chain places APOC4 inside the exchangeable-apolipoprotein band and well clear of globular controls (file:human/APOC4/APOC4-bioinformatics/RESULTS.md). Supporting Evidence: PMID:8530039 The predicted apoC-IV protein sequence, comprising 127 amino acid residues, contains a putative 25-residue signal peptide and two potential amphipathic alpha-helical domains. PMID:8827523 our analysis of transgenic mice provides unequivocal evidence that human apoC-IV is a lipid-binding protein belonging to the apolipoprotein family and that it has the potential to alter lipoprotein metabolism. PMID:8576182 It contains two potential amphipathic helices characteristic of plasma apolipoproteins and forms discoidal micelles with phosphatidylcholine. file:human/APOC4/APOC4-bioinformatics/RESULTS.md Human apoC-IV sits at 0.0300 — inside the apolipoprotein band, between APOC2/C3 (0.0253) and APOC1 (0.0351), about three times the controls (2.97–3.12×), and differing by 0.0009 from rabbit apoC-IV (0.0309) |
| GO:0005515 protein binding | IPI PMID:32296183 A reference map of the human binary protein interactome. | MARK AS OVER ANNOTATED | Summary: Binary yeast two-hybrid interaction with THBD (Thrombomodulin), reported only in the HuRI reference interactome [PMID:32296183]. UniProt curates THBD to Membrane. Reason: All 23 GO:0005515 rows on APOC4 come from one reference, PMID:32296183 (HuRI), a systematic yeast two-hybrid screen. A full QuickGO export for that reference returns 85,343 annotations across 6,769 distinct gene products, 84,937 of them GO:0005515 - this is a maximal-projection reference, and APOC4's 23 rows are 23 of those. Resolving all 23 partners in UniProt, none has a curated Secreted or Extracellular subcellular location, 10 are curated mitochondrial and 10 carry a transmembrane segment (file:human/APOC4/APOC4-bioinformatics/RESULTS.md). Two points follow, and they are not equally firm. The firm one does not depend on any partner's membrane topology: not one of the 23 is a plasma protein of the kind apoC-IV actually encounters, and every pair was scored by a Gal4 two-hybrid, which reports interactions reconstituted in the yeast nucleus - a compartment APOC4's mature chain cannot occupy at all. The softer one is each partner's own compartment, and it is graded rather than blanket: strongest for the ten curated mitochondrial partners, weakest for the ER- and Golgi-membrane ones, because APOC4 does transit those organelles on its way out of the hepatocyte and the argument then turns on which face of the membrane the partner works on - which is not settled for every partner here. Where that per-partner point is strong it is made in the row itself. HuRI itself notes that most of its pairs were seen in a single screen, and no functional follow-up exists for any APOC4 pair; the NbExp=3 shown against each of these pairs in the UniProt INTERACTION block is HuRI's own screen-and-retest count from this one publication, not three independent studies. Per project policy an experimental row is not removed on this basis; the term is also uninformative, and there is no better MF to MODIFY it to, so it is marked as over-annotated. THBD is the one partner of the 23 with a cleaved signal peptide, and as a single-pass type I membrane protein it does present an extracellular domain to flowing blood, so an encounter with a plasma apolipoprotein is not topologically absurd the way it is for the mitochondrial partners. But that topology is precisely what a Gal4 two-hybrid assay in the yeast nucleus cannot test, and no follow-up experiment has examined an APOC4-THBD interaction in plasma or on endothelium. Supporting Evidence: PMID:32296183 yeast two-hybrid (Y2H) represents the only binary PPI assay that can be operated at sufficient throughput to systematically screen the human proteome for binary PPIs PMID:32296183 the majority of PPIs in HuRI were found in only one screen file:human/APOC4/APOC4-uniprot.txt FT SIGNAL 1..27 |
| GO:0005515 protein binding | IPI PMID:32296183 A reference map of the human binary protein interactome. | MARK AS OVER ANNOTATED | Summary: Binary yeast two-hybrid interaction with IFI27 (Interferon alpha-inducible protein 27, mitochondrial), reported only in the HuRI reference interactome [PMID:32296183]. UniProt curates IFI27 to Mitochondrion membrane; Nucleus inner membrane; Endoplasmic reticulum membrane. Reason: All 23 GO:0005515 rows on APOC4 come from one reference, PMID:32296183 (HuRI), a systematic yeast two-hybrid screen. A full QuickGO export for that reference returns 85,343 annotations across 6,769 distinct gene products, 84,937 of them GO:0005515 - this is a maximal-projection reference, and APOC4's 23 rows are 23 of those. Resolving all 23 partners in UniProt, none has a curated Secreted or Extracellular subcellular location, 10 are curated mitochondrial and 10 carry a transmembrane segment (file:human/APOC4/APOC4-bioinformatics/RESULTS.md). Two points follow, and they are not equally firm. The firm one does not depend on any partner's membrane topology: not one of the 23 is a plasma protein of the kind apoC-IV actually encounters, and every pair was scored by a Gal4 two-hybrid, which reports interactions reconstituted in the yeast nucleus - a compartment APOC4's mature chain cannot occupy at all. The softer one is each partner's own compartment, and it is graded rather than blanket: strongest for the ten curated mitochondrial partners, weakest for the ER- and Golgi-membrane ones, because APOC4 does transit those organelles on its way out of the hepatocyte and the argument then turns on which face of the membrane the partner works on - which is not settled for every partner here. Where that per-partner point is strong it is made in the row itself. HuRI itself notes that most of its pairs were seen in a single screen, and no functional follow-up exists for any APOC4 pair; the NbExp=3 shown against each of these pairs in the UniProt INTERACTION block is HuRI's own screen-and-retest count from this one publication, not three independent studies. Per project policy an experimental row is not removed on this basis; the term is also uninformative, and there is no better MF to MODIFY it to, so it is marked as over-annotated. Supporting Evidence: PMID:32296183 yeast two-hybrid (Y2H) represents the only binary PPI assay that can be operated at sufficient throughput to systematically screen the human proteome for binary PPIs PMID:32296183 the majority of PPIs in HuRI were found in only one screen file:human/APOC4/APOC4-uniprot.txt FT SIGNAL 1..27 |
| GO:0005515 protein binding | IPI PMID:32296183 A reference map of the human binary protein interactome. | MARK AS OVER ANNOTATED | Summary: Binary yeast two-hybrid interaction with GAD2 (Glutamate decarboxylase 2), reported only in the HuRI reference interactome [PMID:32296183]. UniProt curates GAD2 to Cytoplasm, cytosol; Cytoplasmic vesicle; Presynaptic cell membrane; Golgi apparatus membrane. Reason: All 23 GO:0005515 rows on APOC4 come from one reference, PMID:32296183 (HuRI), a systematic yeast two-hybrid screen. A full QuickGO export for that reference returns 85,343 annotations across 6,769 distinct gene products, 84,937 of them GO:0005515 - this is a maximal-projection reference, and APOC4's 23 rows are 23 of those. Resolving all 23 partners in UniProt, none has a curated Secreted or Extracellular subcellular location, 10 are curated mitochondrial and 10 carry a transmembrane segment (file:human/APOC4/APOC4-bioinformatics/RESULTS.md). Two points follow, and they are not equally firm. The firm one does not depend on any partner's membrane topology: not one of the 23 is a plasma protein of the kind apoC-IV actually encounters, and every pair was scored by a Gal4 two-hybrid, which reports interactions reconstituted in the yeast nucleus - a compartment APOC4's mature chain cannot occupy at all. The softer one is each partner's own compartment, and it is graded rather than blanket: strongest for the ten curated mitochondrial partners, weakest for the ER- and Golgi-membrane ones, because APOC4 does transit those organelles on its way out of the hepatocyte and the argument then turns on which face of the membrane the partner works on - which is not settled for every partner here. Where that per-partner point is strong it is made in the row itself. HuRI itself notes that most of its pairs were seen in a single screen, and no functional follow-up exists for any APOC4 pair; the NbExp=3 shown against each of these pairs in the UniProt INTERACTION block is HuRI's own screen-and-retest count from this one publication, not three independent studies. Per project policy an experimental row is not removed on this basis; the term is also uninformative, and there is no better MF to MODIFY it to, so it is marked as over-annotated. Supporting Evidence: PMID:32296183 yeast two-hybrid (Y2H) represents the only binary PPI assay that can be operated at sufficient throughput to systematically screen the human proteome for binary PPIs PMID:32296183 the majority of PPIs in HuRI were found in only one screen file:human/APOC4/APOC4-uniprot.txt FT SIGNAL 1..27 |
| GO:0005515 protein binding | IPI PMID:32296183 A reference map of the human binary protein interactome. | MARK AS OVER ANNOTATED | Summary: Binary yeast two-hybrid interaction with SYT16 (Synaptotagmin-16), reported only in the HuRI reference interactome [PMID:32296183]. UniProt curates SYT16 to no curated subcellular location. Reason: All 23 GO:0005515 rows on APOC4 come from one reference, PMID:32296183 (HuRI), a systematic yeast two-hybrid screen. A full QuickGO export for that reference returns 85,343 annotations across 6,769 distinct gene products, 84,937 of them GO:0005515 - this is a maximal-projection reference, and APOC4's 23 rows are 23 of those. Resolving all 23 partners in UniProt, none has a curated Secreted or Extracellular subcellular location, 10 are curated mitochondrial and 10 carry a transmembrane segment (file:human/APOC4/APOC4-bioinformatics/RESULTS.md). Two points follow, and they are not equally firm. The firm one does not depend on any partner's membrane topology: not one of the 23 is a plasma protein of the kind apoC-IV actually encounters, and every pair was scored by a Gal4 two-hybrid, which reports interactions reconstituted in the yeast nucleus - a compartment APOC4's mature chain cannot occupy at all. The softer one is each partner's own compartment, and it is graded rather than blanket: strongest for the ten curated mitochondrial partners, weakest for the ER- and Golgi-membrane ones, because APOC4 does transit those organelles on its way out of the hepatocyte and the argument then turns on which face of the membrane the partner works on - which is not settled for every partner here. Where that per-partner point is strong it is made in the row itself. HuRI itself notes that most of its pairs were seen in a single screen, and no functional follow-up exists for any APOC4 pair; the NbExp=3 shown against each of these pairs in the UniProt INTERACTION block is HuRI's own screen-and-retest count from this one publication, not three independent studies. Per project policy an experimental row is not removed on this basis; the term is also uninformative, and there is no better MF to MODIFY it to, so it is marked as over-annotated. SYT16 has no curated subcellular location at all in UniProt, so the pair cannot even be assessed for compartment compatibility. Supporting Evidence: PMID:32296183 yeast two-hybrid (Y2H) represents the only binary PPI assay that can be operated at sufficient throughput to systematically screen the human proteome for binary PPIs PMID:32296183 the majority of PPIs in HuRI were found in only one screen file:human/APOC4/APOC4-uniprot.txt FT SIGNAL 1..27 |
| GO:0005515 protein binding | IPI PMID:32296183 A reference map of the human binary protein interactome. | MARK AS OVER ANNOTATED | Summary: Binary yeast two-hybrid interaction with SNAP47 (Synaptosomal-associated protein 47), reported only in the HuRI reference interactome [PMID:32296183]. UniProt curates SNAP47 to Endomembrane system; Cytoplasm, perinuclear region. Reason: All 23 GO:0005515 rows on APOC4 come from one reference, PMID:32296183 (HuRI), a systematic yeast two-hybrid screen. A full QuickGO export for that reference returns 85,343 annotations across 6,769 distinct gene products, 84,937 of them GO:0005515 - this is a maximal-projection reference, and APOC4's 23 rows are 23 of those. Resolving all 23 partners in UniProt, none has a curated Secreted or Extracellular subcellular location, 10 are curated mitochondrial and 10 carry a transmembrane segment (file:human/APOC4/APOC4-bioinformatics/RESULTS.md). Two points follow, and they are not equally firm. The firm one does not depend on any partner's membrane topology: not one of the 23 is a plasma protein of the kind apoC-IV actually encounters, and every pair was scored by a Gal4 two-hybrid, which reports interactions reconstituted in the yeast nucleus - a compartment APOC4's mature chain cannot occupy at all. The softer one is each partner's own compartment, and it is graded rather than blanket: strongest for the ten curated mitochondrial partners, weakest for the ER- and Golgi-membrane ones, because APOC4 does transit those organelles on its way out of the hepatocyte and the argument then turns on which face of the membrane the partner works on - which is not settled for every partner here. Where that per-partner point is strong it is made in the row itself. HuRI itself notes that most of its pairs were seen in a single screen, and no functional follow-up exists for any APOC4 pair; the NbExp=3 shown against each of these pairs in the UniProt INTERACTION block is HuRI's own screen-and-retest count from this one publication, not three independent studies. Per project policy an experimental row is not removed on this basis; the term is also uninformative, and there is no better MF to MODIFY it to, so it is marked as over-annotated. Supporting Evidence: PMID:32296183 yeast two-hybrid (Y2H) represents the only binary PPI assay that can be operated at sufficient throughput to systematically screen the human proteome for binary PPIs PMID:32296183 the majority of PPIs in HuRI were found in only one screen file:human/APOC4/APOC4-uniprot.txt FT SIGNAL 1..27 |
| GO:0005515 protein binding | IPI PMID:32296183 A reference map of the human binary protein interactome. | MARK AS OVER ANNOTATED | Summary: Binary yeast two-hybrid interaction with MICOS10 (MICOS complex subunit MIC10), reported only in the HuRI reference interactome [PMID:32296183]. UniProt curates MICOS10 to Mitochondrion inner membrane. Reason: All 23 GO:0005515 rows on APOC4 come from one reference, PMID:32296183 (HuRI), a systematic yeast two-hybrid screen. A full QuickGO export for that reference returns 85,343 annotations across 6,769 distinct gene products, 84,937 of them GO:0005515 - this is a maximal-projection reference, and APOC4's 23 rows are 23 of those. Resolving all 23 partners in UniProt, none has a curated Secreted or Extracellular subcellular location, 10 are curated mitochondrial and 10 carry a transmembrane segment (file:human/APOC4/APOC4-bioinformatics/RESULTS.md). Two points follow, and they are not equally firm. The firm one does not depend on any partner's membrane topology: not one of the 23 is a plasma protein of the kind apoC-IV actually encounters, and every pair was scored by a Gal4 two-hybrid, which reports interactions reconstituted in the yeast nucleus - a compartment APOC4's mature chain cannot occupy at all. The softer one is each partner's own compartment, and it is graded rather than blanket: strongest for the ten curated mitochondrial partners, weakest for the ER- and Golgi-membrane ones, because APOC4 does transit those organelles on its way out of the hepatocyte and the argument then turns on which face of the membrane the partner works on - which is not settled for every partner here. Where that per-partner point is strong it is made in the row itself. HuRI itself notes that most of its pairs were seen in a single screen, and no functional follow-up exists for any APOC4 pair; the NbExp=3 shown against each of these pairs in the UniProt INTERACTION block is HuRI's own screen-and-retest count from this one publication, not three independent studies. Per project policy an experimental row is not removed on this basis; the term is also uninformative, and there is no better MF to MODIFY it to, so it is marked as over-annotated. MICOS10 is curated exclusively to the mitochondrion, a compartment with no connection to the secretory pathway at all, so not even the transient lumenal contact a secreted protein does make in the ER and Golgi is available here. Supporting Evidence: PMID:32296183 yeast two-hybrid (Y2H) represents the only binary PPI assay that can be operated at sufficient throughput to systematically screen the human proteome for binary PPIs PMID:32296183 the majority of PPIs in HuRI were found in only one screen file:human/APOC4/APOC4-uniprot.txt FT SIGNAL 1..27 |
| GO:0005515 protein binding | IPI PMID:32296183 A reference map of the human binary protein interactome. | MARK AS OVER ANNOTATED | Summary: Binary yeast two-hybrid interaction with MICOS13 (MICOS complex subunit MIC13), reported only in the HuRI reference interactome [PMID:32296183]. UniProt curates MICOS13 to Mitochondrion inner membrane. Reason: All 23 GO:0005515 rows on APOC4 come from one reference, PMID:32296183 (HuRI), a systematic yeast two-hybrid screen. A full QuickGO export for that reference returns 85,343 annotations across 6,769 distinct gene products, 84,937 of them GO:0005515 - this is a maximal-projection reference, and APOC4's 23 rows are 23 of those. Resolving all 23 partners in UniProt, none has a curated Secreted or Extracellular subcellular location, 10 are curated mitochondrial and 10 carry a transmembrane segment (file:human/APOC4/APOC4-bioinformatics/RESULTS.md). Two points follow, and they are not equally firm. The firm one does not depend on any partner's membrane topology: not one of the 23 is a plasma protein of the kind apoC-IV actually encounters, and every pair was scored by a Gal4 two-hybrid, which reports interactions reconstituted in the yeast nucleus - a compartment APOC4's mature chain cannot occupy at all. The softer one is each partner's own compartment, and it is graded rather than blanket: strongest for the ten curated mitochondrial partners, weakest for the ER- and Golgi-membrane ones, because APOC4 does transit those organelles on its way out of the hepatocyte and the argument then turns on which face of the membrane the partner works on - which is not settled for every partner here. Where that per-partner point is strong it is made in the row itself. HuRI itself notes that most of its pairs were seen in a single screen, and no functional follow-up exists for any APOC4 pair; the NbExp=3 shown against each of these pairs in the UniProt INTERACTION block is HuRI's own screen-and-retest count from this one publication, not three independent studies. Per project policy an experimental row is not removed on this basis; the term is also uninformative, and there is no better MF to MODIFY it to, so it is marked as over-annotated. MICOS13 is curated exclusively to the mitochondrion, a compartment with no connection to the secretory pathway at all, so not even the transient lumenal contact a secreted protein does make in the ER and Golgi is available here. Supporting Evidence: PMID:32296183 yeast two-hybrid (Y2H) represents the only binary PPI assay that can be operated at sufficient throughput to systematically screen the human proteome for binary PPIs PMID:32296183 the majority of PPIs in HuRI were found in only one screen file:human/APOC4/APOC4-uniprot.txt FT SIGNAL 1..27 |
| GO:0005515 protein binding | IPI PMID:32296183 A reference map of the human binary protein interactome. | MARK AS OVER ANNOTATED | Summary: Binary yeast two-hybrid interaction with LETMD1 (LETM1 domain-containing protein 1), reported only in the HuRI reference interactome [PMID:32296183]. UniProt curates LETMD1 to Mitochondrion outer membrane; Nucleus; Mitochondrion inner membrane. Reason: All 23 GO:0005515 rows on APOC4 come from one reference, PMID:32296183 (HuRI), a systematic yeast two-hybrid screen. A full QuickGO export for that reference returns 85,343 annotations across 6,769 distinct gene products, 84,937 of them GO:0005515 - this is a maximal-projection reference, and APOC4's 23 rows are 23 of those. Resolving all 23 partners in UniProt, none has a curated Secreted or Extracellular subcellular location, 10 are curated mitochondrial and 10 carry a transmembrane segment (file:human/APOC4/APOC4-bioinformatics/RESULTS.md). Two points follow, and they are not equally firm. The firm one does not depend on any partner's membrane topology: not one of the 23 is a plasma protein of the kind apoC-IV actually encounters, and every pair was scored by a Gal4 two-hybrid, which reports interactions reconstituted in the yeast nucleus - a compartment APOC4's mature chain cannot occupy at all. The softer one is each partner's own compartment, and it is graded rather than blanket: strongest for the ten curated mitochondrial partners, weakest for the ER- and Golgi-membrane ones, because APOC4 does transit those organelles on its way out of the hepatocyte and the argument then turns on which face of the membrane the partner works on - which is not settled for every partner here. Where that per-partner point is strong it is made in the row itself. HuRI itself notes that most of its pairs were seen in a single screen, and no functional follow-up exists for any APOC4 pair; the NbExp=3 shown against each of these pairs in the UniProt INTERACTION block is HuRI's own screen-and-retest count from this one publication, not three independent studies. Per project policy an experimental row is not removed on this basis; the term is also uninformative, and there is no better MF to MODIFY it to, so it is marked as over-annotated. Supporting Evidence: PMID:32296183 yeast two-hybrid (Y2H) represents the only binary PPI assay that can be operated at sufficient throughput to systematically screen the human proteome for binary PPIs PMID:32296183 the majority of PPIs in HuRI were found in only one screen file:human/APOC4/APOC4-uniprot.txt FT SIGNAL 1..27 |
| GO:0005515 protein binding | IPI PMID:32296183 A reference map of the human binary protein interactome. | MARK AS OVER ANNOTATED | Summary: Binary yeast two-hybrid interaction with TPRG1 (Tumor protein p63-regulated gene 1 protein), reported only in the HuRI reference interactome [PMID:32296183]. UniProt curates TPRG1 to Cytoplasm. Reason: All 23 GO:0005515 rows on APOC4 come from one reference, PMID:32296183 (HuRI), a systematic yeast two-hybrid screen. A full QuickGO export for that reference returns 85,343 annotations across 6,769 distinct gene products, 84,937 of them GO:0005515 - this is a maximal-projection reference, and APOC4's 23 rows are 23 of those. Resolving all 23 partners in UniProt, none has a curated Secreted or Extracellular subcellular location, 10 are curated mitochondrial and 10 carry a transmembrane segment (file:human/APOC4/APOC4-bioinformatics/RESULTS.md). Two points follow, and they are not equally firm. The firm one does not depend on any partner's membrane topology: not one of the 23 is a plasma protein of the kind apoC-IV actually encounters, and every pair was scored by a Gal4 two-hybrid, which reports interactions reconstituted in the yeast nucleus - a compartment APOC4's mature chain cannot occupy at all. The softer one is each partner's own compartment, and it is graded rather than blanket: strongest for the ten curated mitochondrial partners, weakest for the ER- and Golgi-membrane ones, because APOC4 does transit those organelles on its way out of the hepatocyte and the argument then turns on which face of the membrane the partner works on - which is not settled for every partner here. Where that per-partner point is strong it is made in the row itself. HuRI itself notes that most of its pairs were seen in a single screen, and no functional follow-up exists for any APOC4 pair; the NbExp=3 shown against each of these pairs in the UniProt INTERACTION block is HuRI's own screen-and-retest count from this one publication, not three independent studies. Per project policy an experimental row is not removed on this basis; the term is also uninformative, and there is no better MF to MODIFY it to, so it is marked as over-annotated. Supporting Evidence: PMID:32296183 yeast two-hybrid (Y2H) represents the only binary PPI assay that can be operated at sufficient throughput to systematically screen the human proteome for binary PPIs PMID:32296183 the majority of PPIs in HuRI were found in only one screen file:human/APOC4/APOC4-uniprot.txt FT SIGNAL 1..27 |
| GO:0005515 protein binding | IPI PMID:32296183 A reference map of the human binary protein interactome. | MARK AS OVER ANNOTATED | Summary: Binary yeast two-hybrid interaction with MAIP1 (m-AAA protease-interacting protein 1, mitochondrial), reported only in the HuRI reference interactome [PMID:32296183]. UniProt curates MAIP1 to Mitochondrion matrix. Reason: All 23 GO:0005515 rows on APOC4 come from one reference, PMID:32296183 (HuRI), a systematic yeast two-hybrid screen. A full QuickGO export for that reference returns 85,343 annotations across 6,769 distinct gene products, 84,937 of them GO:0005515 - this is a maximal-projection reference, and APOC4's 23 rows are 23 of those. Resolving all 23 partners in UniProt, none has a curated Secreted or Extracellular subcellular location, 10 are curated mitochondrial and 10 carry a transmembrane segment (file:human/APOC4/APOC4-bioinformatics/RESULTS.md). Two points follow, and they are not equally firm. The firm one does not depend on any partner's membrane topology: not one of the 23 is a plasma protein of the kind apoC-IV actually encounters, and every pair was scored by a Gal4 two-hybrid, which reports interactions reconstituted in the yeast nucleus - a compartment APOC4's mature chain cannot occupy at all. The softer one is each partner's own compartment, and it is graded rather than blanket: strongest for the ten curated mitochondrial partners, weakest for the ER- and Golgi-membrane ones, because APOC4 does transit those organelles on its way out of the hepatocyte and the argument then turns on which face of the membrane the partner works on - which is not settled for every partner here. Where that per-partner point is strong it is made in the row itself. HuRI itself notes that most of its pairs were seen in a single screen, and no functional follow-up exists for any APOC4 pair; the NbExp=3 shown against each of these pairs in the UniProt INTERACTION block is HuRI's own screen-and-retest count from this one publication, not three independent studies. Per project policy an experimental row is not removed on this basis; the term is also uninformative, and there is no better MF to MODIFY it to, so it is marked as over-annotated. MAIP1 is curated exclusively to the mitochondrion, a compartment with no connection to the secretory pathway at all, so not even the transient lumenal contact a secreted protein does make in the ER and Golgi is available here. Supporting Evidence: PMID:32296183 yeast two-hybrid (Y2H) represents the only binary PPI assay that can be operated at sufficient throughput to systematically screen the human proteome for binary PPIs PMID:32296183 the majority of PPIs in HuRI were found in only one screen file:human/APOC4/APOC4-uniprot.txt FT SIGNAL 1..27 |
| GO:0005515 protein binding | IPI PMID:32296183 A reference map of the human binary protein interactome. | MARK AS OVER ANNOTATED | Summary: Binary yeast two-hybrid interaction with RSAD2 (S-adenosylmethionine-dependent nucleotide dehydratase RSAD2), reported only in the HuRI reference interactome [PMID:32296183]. UniProt curates RSAD2 to Endoplasmic reticulum membrane; Golgi apparatus; Endoplasmic reticulum; Lipid droplet; Mitochondrion; Mitochondrion inner membrane; Mitochondrion outer membrane. Reason: All 23 GO:0005515 rows on APOC4 come from one reference, PMID:32296183 (HuRI), a systematic yeast two-hybrid screen. A full QuickGO export for that reference returns 85,343 annotations across 6,769 distinct gene products, 84,937 of them GO:0005515 - this is a maximal-projection reference, and APOC4's 23 rows are 23 of those. Resolving all 23 partners in UniProt, none has a curated Secreted or Extracellular subcellular location, 10 are curated mitochondrial and 10 carry a transmembrane segment (file:human/APOC4/APOC4-bioinformatics/RESULTS.md). Two points follow, and they are not equally firm. The firm one does not depend on any partner's membrane topology: not one of the 23 is a plasma protein of the kind apoC-IV actually encounters, and every pair was scored by a Gal4 two-hybrid, which reports interactions reconstituted in the yeast nucleus - a compartment APOC4's mature chain cannot occupy at all. The softer one is each partner's own compartment, and it is graded rather than blanket: strongest for the ten curated mitochondrial partners, weakest for the ER- and Golgi-membrane ones, because APOC4 does transit those organelles on its way out of the hepatocyte and the argument then turns on which face of the membrane the partner works on - which is not settled for every partner here. Where that per-partner point is strong it is made in the row itself. HuRI itself notes that most of its pairs were seen in a single screen, and no functional follow-up exists for any APOC4 pair; the NbExp=3 shown against each of these pairs in the UniProt INTERACTION block is HuRI's own screen-and-retest count from this one publication, not three independent studies. Per project policy an experimental row is not removed on this basis; the term is also uninformative, and there is no better MF to MODIFY it to, so it is marked as over-annotated. Supporting Evidence: PMID:32296183 yeast two-hybrid (Y2H) represents the only binary PPI assay that can be operated at sufficient throughput to systematically screen the human proteome for binary PPIs PMID:32296183 the majority of PPIs in HuRI were found in only one screen file:human/APOC4/APOC4-uniprot.txt FT SIGNAL 1..27 |
| GO:0005515 protein binding | IPI PMID:32296183 A reference map of the human binary protein interactome. | MARK AS OVER ANNOTATED | Summary: Binary yeast two-hybrid interaction with BCL2L2 (Bcl-2-like protein 2), reported only in the HuRI reference interactome [PMID:32296183]. UniProt curates BCL2L2 to Mitochondrion membrane. Reason: All 23 GO:0005515 rows on APOC4 come from one reference, PMID:32296183 (HuRI), a systematic yeast two-hybrid screen. A full QuickGO export for that reference returns 85,343 annotations across 6,769 distinct gene products, 84,937 of them GO:0005515 - this is a maximal-projection reference, and APOC4's 23 rows are 23 of those. Resolving all 23 partners in UniProt, none has a curated Secreted or Extracellular subcellular location, 10 are curated mitochondrial and 10 carry a transmembrane segment (file:human/APOC4/APOC4-bioinformatics/RESULTS.md). Two points follow, and they are not equally firm. The firm one does not depend on any partner's membrane topology: not one of the 23 is a plasma protein of the kind apoC-IV actually encounters, and every pair was scored by a Gal4 two-hybrid, which reports interactions reconstituted in the yeast nucleus - a compartment APOC4's mature chain cannot occupy at all. The softer one is each partner's own compartment, and it is graded rather than blanket: strongest for the ten curated mitochondrial partners, weakest for the ER- and Golgi-membrane ones, because APOC4 does transit those organelles on its way out of the hepatocyte and the argument then turns on which face of the membrane the partner works on - which is not settled for every partner here. Where that per-partner point is strong it is made in the row itself. HuRI itself notes that most of its pairs were seen in a single screen, and no functional follow-up exists for any APOC4 pair; the NbExp=3 shown against each of these pairs in the UniProt INTERACTION block is HuRI's own screen-and-retest count from this one publication, not three independent studies. Per project policy an experimental row is not removed on this basis; the term is also uninformative, and there is no better MF to MODIFY it to, so it is marked as over-annotated. BCL2L2 is curated exclusively to the mitochondrion, a compartment with no connection to the secretory pathway at all, so not even the transient lumenal contact a secreted protein does make in the ER and Golgi is available here. Supporting Evidence: PMID:32296183 yeast two-hybrid (Y2H) represents the only binary PPI assay that can be operated at sufficient throughput to systematically screen the human proteome for binary PPIs PMID:32296183 the majority of PPIs in HuRI were found in only one screen file:human/APOC4/APOC4-uniprot.txt FT SIGNAL 1..27 |
| GO:0005515 protein binding | IPI PMID:32296183 A reference map of the human binary protein interactome. | MARK AS OVER ANNOTATED | Summary: Binary yeast two-hybrid interaction with TBC1D20 (TBC1 domain family member 20), reported only in the HuRI reference interactome [PMID:32296183]. UniProt curates TBC1D20 to Membrane. Reason: All 23 GO:0005515 rows on APOC4 come from one reference, PMID:32296183 (HuRI), a systematic yeast two-hybrid screen. A full QuickGO export for that reference returns 85,343 annotations across 6,769 distinct gene products, 84,937 of them GO:0005515 - this is a maximal-projection reference, and APOC4's 23 rows are 23 of those. Resolving all 23 partners in UniProt, none has a curated Secreted or Extracellular subcellular location, 10 are curated mitochondrial and 10 carry a transmembrane segment (file:human/APOC4/APOC4-bioinformatics/RESULTS.md). Two points follow, and they are not equally firm. The firm one does not depend on any partner's membrane topology: not one of the 23 is a plasma protein of the kind apoC-IV actually encounters, and every pair was scored by a Gal4 two-hybrid, which reports interactions reconstituted in the yeast nucleus - a compartment APOC4's mature chain cannot occupy at all. The softer one is each partner's own compartment, and it is graded rather than blanket: strongest for the ten curated mitochondrial partners, weakest for the ER- and Golgi-membrane ones, because APOC4 does transit those organelles on its way out of the hepatocyte and the argument then turns on which face of the membrane the partner works on - which is not settled for every partner here. Where that per-partner point is strong it is made in the row itself. HuRI itself notes that most of its pairs were seen in a single screen, and no functional follow-up exists for any APOC4 pair; the NbExp=3 shown against each of these pairs in the UniProt INTERACTION block is HuRI's own screen-and-retest count from this one publication, not three independent studies. Per project policy an experimental row is not removed on this basis; the term is also uninformative, and there is no better MF to MODIFY it to, so it is marked as over-annotated. Supporting Evidence: PMID:32296183 yeast two-hybrid (Y2H) represents the only binary PPI assay that can be operated at sufficient throughput to systematically screen the human proteome for binary PPIs PMID:32296183 the majority of PPIs in HuRI were found in only one screen file:human/APOC4/APOC4-uniprot.txt FT SIGNAL 1..27 |
| GO:0005515 protein binding | IPI PMID:32296183 A reference map of the human binary protein interactome. | MARK AS OVER ANNOTATED | Summary: Binary yeast two-hybrid interaction with GSDMA (Gasdermin-A), reported only in the HuRI reference interactome [PMID:32296183]. UniProt curates GSDMA to Cytoplasm, perinuclear region; Cytoplasm, cytosol; Cell membrane. Reason: All 23 GO:0005515 rows on APOC4 come from one reference, PMID:32296183 (HuRI), a systematic yeast two-hybrid screen. A full QuickGO export for that reference returns 85,343 annotations across 6,769 distinct gene products, 84,937 of them GO:0005515 - this is a maximal-projection reference, and APOC4's 23 rows are 23 of those. Resolving all 23 partners in UniProt, none has a curated Secreted or Extracellular subcellular location, 10 are curated mitochondrial and 10 carry a transmembrane segment (file:human/APOC4/APOC4-bioinformatics/RESULTS.md). Two points follow, and they are not equally firm. The firm one does not depend on any partner's membrane topology: not one of the 23 is a plasma protein of the kind apoC-IV actually encounters, and every pair was scored by a Gal4 two-hybrid, which reports interactions reconstituted in the yeast nucleus - a compartment APOC4's mature chain cannot occupy at all. The softer one is each partner's own compartment, and it is graded rather than blanket: strongest for the ten curated mitochondrial partners, weakest for the ER- and Golgi-membrane ones, because APOC4 does transit those organelles on its way out of the hepatocyte and the argument then turns on which face of the membrane the partner works on - which is not settled for every partner here. Where that per-partner point is strong it is made in the row itself. HuRI itself notes that most of its pairs were seen in a single screen, and no functional follow-up exists for any APOC4 pair; the NbExp=3 shown against each of these pairs in the UniProt INTERACTION block is HuRI's own screen-and-retest count from this one publication, not three independent studies. Per project policy an experimental row is not removed on this basis; the term is also uninformative, and there is no better MF to MODIFY it to, so it is marked as over-annotated. Supporting Evidence: PMID:32296183 yeast two-hybrid (Y2H) represents the only binary PPI assay that can be operated at sufficient throughput to systematically screen the human proteome for binary PPIs PMID:32296183 the majority of PPIs in HuRI were found in only one screen file:human/APOC4/APOC4-uniprot.txt FT SIGNAL 1..27 |
| GO:0005515 protein binding | IPI PMID:32296183 A reference map of the human binary protein interactome. | MARK AS OVER ANNOTATED | Summary: Binary yeast two-hybrid interaction with APOL2 (Apolipoprotein L2), reported only in the HuRI reference interactome [PMID:32296183]. UniProt curates APOL2 to Cytoplasm. Reason: All 23 GO:0005515 rows on APOC4 come from one reference, PMID:32296183 (HuRI), a systematic yeast two-hybrid screen. A full QuickGO export for that reference returns 85,343 annotations across 6,769 distinct gene products, 84,937 of them GO:0005515 - this is a maximal-projection reference, and APOC4's 23 rows are 23 of those. Resolving all 23 partners in UniProt, none has a curated Secreted or Extracellular subcellular location, 10 are curated mitochondrial and 10 carry a transmembrane segment (file:human/APOC4/APOC4-bioinformatics/RESULTS.md). Two points follow, and they are not equally firm. The firm one does not depend on any partner's membrane topology: not one of the 23 is a plasma protein of the kind apoC-IV actually encounters, and every pair was scored by a Gal4 two-hybrid, which reports interactions reconstituted in the yeast nucleus - a compartment APOC4's mature chain cannot occupy at all. The softer one is each partner's own compartment, and it is graded rather than blanket: strongest for the ten curated mitochondrial partners, weakest for the ER- and Golgi-membrane ones, because APOC4 does transit those organelles on its way out of the hepatocyte and the argument then turns on which face of the membrane the partner works on - which is not settled for every partner here. Where that per-partner point is strong it is made in the row itself. HuRI itself notes that most of its pairs were seen in a single screen, and no functional follow-up exists for any APOC4 pair; the NbExp=3 shown against each of these pairs in the UniProt INTERACTION block is HuRI's own screen-and-retest count from this one publication, not three independent studies. Per project policy an experimental row is not removed on this basis; the term is also uninformative, and there is no better MF to MODIFY it to, so it is marked as over-annotated. Supporting Evidence: PMID:32296183 yeast two-hybrid (Y2H) represents the only binary PPI assay that can be operated at sufficient throughput to systematically screen the human proteome for binary PPIs PMID:32296183 the majority of PPIs in HuRI were found in only one screen file:human/APOC4/APOC4-uniprot.txt FT SIGNAL 1..27 |
| GO:0005515 protein binding | IPI PMID:32296183 A reference map of the human binary protein interactome. | MARK AS OVER ANNOTATED | Summary: Binary yeast two-hybrid interaction with PORCN (Protein-serine O-palmitoleoyltransferase porcupine), reported only in the HuRI reference interactome [PMID:32296183]. UniProt curates PORCN to Endoplasmic reticulum membrane. Reason: All 23 GO:0005515 rows on APOC4 come from one reference, PMID:32296183 (HuRI), a systematic yeast two-hybrid screen. A full QuickGO export for that reference returns 85,343 annotations across 6,769 distinct gene products, 84,937 of them GO:0005515 - this is a maximal-projection reference, and APOC4's 23 rows are 23 of those. Resolving all 23 partners in UniProt, none has a curated Secreted or Extracellular subcellular location, 10 are curated mitochondrial and 10 carry a transmembrane segment (file:human/APOC4/APOC4-bioinformatics/RESULTS.md). Two points follow, and they are not equally firm. The firm one does not depend on any partner's membrane topology: not one of the 23 is a plasma protein of the kind apoC-IV actually encounters, and every pair was scored by a Gal4 two-hybrid, which reports interactions reconstituted in the yeast nucleus - a compartment APOC4's mature chain cannot occupy at all. The softer one is each partner's own compartment, and it is graded rather than blanket: strongest for the ten curated mitochondrial partners, weakest for the ER- and Golgi-membrane ones, because APOC4 does transit those organelles on its way out of the hepatocyte and the argument then turns on which face of the membrane the partner works on - which is not settled for every partner here. Where that per-partner point is strong it is made in the row itself. HuRI itself notes that most of its pairs were seen in a single screen, and no functional follow-up exists for any APOC4 pair; the NbExp=3 shown against each of these pairs in the UniProt INTERACTION block is HuRI's own screen-and-retest count from this one publication, not three independent studies. Per project policy an experimental row is not removed on this basis; the term is also uninformative, and there is no better MF to MODIFY it to, so it is marked as over-annotated. Supporting Evidence: PMID:32296183 yeast two-hybrid (Y2H) represents the only binary PPI assay that can be operated at sufficient throughput to systematically screen the human proteome for binary PPIs PMID:32296183 the majority of PPIs in HuRI were found in only one screen file:human/APOC4/APOC4-uniprot.txt FT SIGNAL 1..27 |
| GO:0005515 protein binding | IPI PMID:32296183 A reference map of the human binary protein interactome. | MARK AS OVER ANNOTATED | Summary: Binary yeast two-hybrid interaction with SENP2 (Sentrin-specific protease 2), reported only in the HuRI reference interactome [PMID:32296183]. UniProt curates SENP2 to Nucleus, nuclear pore complex; Nucleus membrane; Cytoplasm. Reason: All 23 GO:0005515 rows on APOC4 come from one reference, PMID:32296183 (HuRI), a systematic yeast two-hybrid screen. A full QuickGO export for that reference returns 85,343 annotations across 6,769 distinct gene products, 84,937 of them GO:0005515 - this is a maximal-projection reference, and APOC4's 23 rows are 23 of those. Resolving all 23 partners in UniProt, none has a curated Secreted or Extracellular subcellular location, 10 are curated mitochondrial and 10 carry a transmembrane segment (file:human/APOC4/APOC4-bioinformatics/RESULTS.md). Two points follow, and they are not equally firm. The firm one does not depend on any partner's membrane topology: not one of the 23 is a plasma protein of the kind apoC-IV actually encounters, and every pair was scored by a Gal4 two-hybrid, which reports interactions reconstituted in the yeast nucleus - a compartment APOC4's mature chain cannot occupy at all. The softer one is each partner's own compartment, and it is graded rather than blanket: strongest for the ten curated mitochondrial partners, weakest for the ER- and Golgi-membrane ones, because APOC4 does transit those organelles on its way out of the hepatocyte and the argument then turns on which face of the membrane the partner works on - which is not settled for every partner here. Where that per-partner point is strong it is made in the row itself. HuRI itself notes that most of its pairs were seen in a single screen, and no functional follow-up exists for any APOC4 pair; the NbExp=3 shown against each of these pairs in the UniProt INTERACTION block is HuRI's own screen-and-retest count from this one publication, not three independent studies. Per project policy an experimental row is not removed on this basis; the term is also uninformative, and there is no better MF to MODIFY it to, so it is marked as over-annotated. Supporting Evidence: PMID:32296183 yeast two-hybrid (Y2H) represents the only binary PPI assay that can be operated at sufficient throughput to systematically screen the human proteome for binary PPIs PMID:32296183 the majority of PPIs in HuRI were found in only one screen file:human/APOC4/APOC4-uniprot.txt FT SIGNAL 1..27 |
| GO:0005515 protein binding | IPI PMID:32296183 A reference map of the human binary protein interactome. | MARK AS OVER ANNOTATED | Summary: Binary yeast two-hybrid interaction with TIMMDC1 (Complex I assembly factor TIMMDC1, mitochondrial), reported only in the HuRI reference interactome [PMID:32296183]. UniProt curates TIMMDC1 to Mitochondrion membrane. Reason: All 23 GO:0005515 rows on APOC4 come from one reference, PMID:32296183 (HuRI), a systematic yeast two-hybrid screen. A full QuickGO export for that reference returns 85,343 annotations across 6,769 distinct gene products, 84,937 of them GO:0005515 - this is a maximal-projection reference, and APOC4's 23 rows are 23 of those. Resolving all 23 partners in UniProt, none has a curated Secreted or Extracellular subcellular location, 10 are curated mitochondrial and 10 carry a transmembrane segment (file:human/APOC4/APOC4-bioinformatics/RESULTS.md). Two points follow, and they are not equally firm. The firm one does not depend on any partner's membrane topology: not one of the 23 is a plasma protein of the kind apoC-IV actually encounters, and every pair was scored by a Gal4 two-hybrid, which reports interactions reconstituted in the yeast nucleus - a compartment APOC4's mature chain cannot occupy at all. The softer one is each partner's own compartment, and it is graded rather than blanket: strongest for the ten curated mitochondrial partners, weakest for the ER- and Golgi-membrane ones, because APOC4 does transit those organelles on its way out of the hepatocyte and the argument then turns on which face of the membrane the partner works on - which is not settled for every partner here. Where that per-partner point is strong it is made in the row itself. HuRI itself notes that most of its pairs were seen in a single screen, and no functional follow-up exists for any APOC4 pair; the NbExp=3 shown against each of these pairs in the UniProt INTERACTION block is HuRI's own screen-and-retest count from this one publication, not three independent studies. Per project policy an experimental row is not removed on this basis; the term is also uninformative, and there is no better MF to MODIFY it to, so it is marked as over-annotated. TIMMDC1 is curated exclusively to the mitochondrion, a compartment with no connection to the secretory pathway at all, so not even the transient lumenal contact a secreted protein does make in the ER and Golgi is available here. Supporting Evidence: PMID:32296183 yeast two-hybrid (Y2H) represents the only binary PPI assay that can be operated at sufficient throughput to systematically screen the human proteome for binary PPIs PMID:32296183 the majority of PPIs in HuRI were found in only one screen file:human/APOC4/APOC4-uniprot.txt FT SIGNAL 1..27 |
| GO:0005515 protein binding | IPI PMID:32296183 A reference map of the human binary protein interactome. | MARK AS OVER ANNOTATED | Summary: Binary yeast two-hybrid interaction with DIABLO (Diablo IAP-binding mitochondrial protein), reported only in the HuRI reference interactome [PMID:32296183]. UniProt curates DIABLO to Mitochondrion; Cytoplasm, cytosol. Reason: All 23 GO:0005515 rows on APOC4 come from one reference, PMID:32296183 (HuRI), a systematic yeast two-hybrid screen. A full QuickGO export for that reference returns 85,343 annotations across 6,769 distinct gene products, 84,937 of them GO:0005515 - this is a maximal-projection reference, and APOC4's 23 rows are 23 of those. Resolving all 23 partners in UniProt, none has a curated Secreted or Extracellular subcellular location, 10 are curated mitochondrial and 10 carry a transmembrane segment (file:human/APOC4/APOC4-bioinformatics/RESULTS.md). Two points follow, and they are not equally firm. The firm one does not depend on any partner's membrane topology: not one of the 23 is a plasma protein of the kind apoC-IV actually encounters, and every pair was scored by a Gal4 two-hybrid, which reports interactions reconstituted in the yeast nucleus - a compartment APOC4's mature chain cannot occupy at all. The softer one is each partner's own compartment, and it is graded rather than blanket: strongest for the ten curated mitochondrial partners, weakest for the ER- and Golgi-membrane ones, because APOC4 does transit those organelles on its way out of the hepatocyte and the argument then turns on which face of the membrane the partner works on - which is not settled for every partner here. Where that per-partner point is strong it is made in the row itself. HuRI itself notes that most of its pairs were seen in a single screen, and no functional follow-up exists for any APOC4 pair; the NbExp=3 shown against each of these pairs in the UniProt INTERACTION block is HuRI's own screen-and-retest count from this one publication, not three independent studies. Per project policy an experimental row is not removed on this basis; the term is also uninformative, and there is no better MF to MODIFY it to, so it is marked as over-annotated. Supporting Evidence: PMID:32296183 yeast two-hybrid (Y2H) represents the only binary PPI assay that can be operated at sufficient throughput to systematically screen the human proteome for binary PPIs PMID:32296183 the majority of PPIs in HuRI were found in only one screen file:human/APOC4/APOC4-uniprot.txt FT SIGNAL 1..27 |
| GO:0005515 protein binding | IPI PMID:32296183 A reference map of the human binary protein interactome. | MARK AS OVER ANNOTATED | Summary: Binary yeast two-hybrid interaction with TMEM14B (Transmembrane protein 14B), reported only in the HuRI reference interactome [PMID:32296183]. UniProt curates TMEM14B to Membrane. Reason: All 23 GO:0005515 rows on APOC4 come from one reference, PMID:32296183 (HuRI), a systematic yeast two-hybrid screen. A full QuickGO export for that reference returns 85,343 annotations across 6,769 distinct gene products, 84,937 of them GO:0005515 - this is a maximal-projection reference, and APOC4's 23 rows are 23 of those. Resolving all 23 partners in UniProt, none has a curated Secreted or Extracellular subcellular location, 10 are curated mitochondrial and 10 carry a transmembrane segment (file:human/APOC4/APOC4-bioinformatics/RESULTS.md). Two points follow, and they are not equally firm. The firm one does not depend on any partner's membrane topology: not one of the 23 is a plasma protein of the kind apoC-IV actually encounters, and every pair was scored by a Gal4 two-hybrid, which reports interactions reconstituted in the yeast nucleus - a compartment APOC4's mature chain cannot occupy at all. The softer one is each partner's own compartment, and it is graded rather than blanket: strongest for the ten curated mitochondrial partners, weakest for the ER- and Golgi-membrane ones, because APOC4 does transit those organelles on its way out of the hepatocyte and the argument then turns on which face of the membrane the partner works on - which is not settled for every partner here. Where that per-partner point is strong it is made in the row itself. HuRI itself notes that most of its pairs were seen in a single screen, and no functional follow-up exists for any APOC4 pair; the NbExp=3 shown against each of these pairs in the UniProt INTERACTION block is HuRI's own screen-and-retest count from this one publication, not three independent studies. Per project policy an experimental row is not removed on this basis; the term is also uninformative, and there is no better MF to MODIFY it to, so it is marked as over-annotated. Supporting Evidence: PMID:32296183 yeast two-hybrid (Y2H) represents the only binary PPI assay that can be operated at sufficient throughput to systematically screen the human proteome for binary PPIs PMID:32296183 the majority of PPIs in HuRI were found in only one screen file:human/APOC4/APOC4-uniprot.txt FT SIGNAL 1..27 |
| GO:0005515 protein binding | IPI PMID:32296183 A reference map of the human binary protein interactome. | MARK AS OVER ANNOTATED | Summary: Binary yeast two-hybrid interaction with UBQLN2 (Ubiquilin-2), reported only in the HuRI reference interactome [PMID:32296183]. UniProt curates UBQLN2 to Cytoplasm; Nucleus; Membrane; Cytoplasmic vesicle, autophagosome. Reason: All 23 GO:0005515 rows on APOC4 come from one reference, PMID:32296183 (HuRI), a systematic yeast two-hybrid screen. A full QuickGO export for that reference returns 85,343 annotations across 6,769 distinct gene products, 84,937 of them GO:0005515 - this is a maximal-projection reference, and APOC4's 23 rows are 23 of those. Resolving all 23 partners in UniProt, none has a curated Secreted or Extracellular subcellular location, 10 are curated mitochondrial and 10 carry a transmembrane segment (file:human/APOC4/APOC4-bioinformatics/RESULTS.md). Two points follow, and they are not equally firm. The firm one does not depend on any partner's membrane topology: not one of the 23 is a plasma protein of the kind apoC-IV actually encounters, and every pair was scored by a Gal4 two-hybrid, which reports interactions reconstituted in the yeast nucleus - a compartment APOC4's mature chain cannot occupy at all. The softer one is each partner's own compartment, and it is graded rather than blanket: strongest for the ten curated mitochondrial partners, weakest for the ER- and Golgi-membrane ones, because APOC4 does transit those organelles on its way out of the hepatocyte and the argument then turns on which face of the membrane the partner works on - which is not settled for every partner here. Where that per-partner point is strong it is made in the row itself. HuRI itself notes that most of its pairs were seen in a single screen, and no functional follow-up exists for any APOC4 pair; the NbExp=3 shown against each of these pairs in the UniProt INTERACTION block is HuRI's own screen-and-retest count from this one publication, not three independent studies. Per project policy an experimental row is not removed on this basis; the term is also uninformative, and there is no better MF to MODIFY it to, so it is marked as over-annotated. Supporting Evidence: PMID:32296183 yeast two-hybrid (Y2H) represents the only binary PPI assay that can be operated at sufficient throughput to systematically screen the human proteome for binary PPIs PMID:32296183 the majority of PPIs in HuRI were found in only one screen file:human/APOC4/APOC4-uniprot.txt FT SIGNAL 1..27 |
| GO:0005515 protein binding | IPI PMID:32296183 A reference map of the human binary protein interactome. | MARK AS OVER ANNOTATED | Summary: Binary yeast two-hybrid interaction with NFU1 (NFU1 iron-sulfur cluster scaffold homolog, mitochondrial), reported only in the HuRI reference interactome [PMID:32296183]. UniProt curates NFU1 to Mitochondrion; Cytoplasm, cytosol. Reason: All 23 GO:0005515 rows on APOC4 come from one reference, PMID:32296183 (HuRI), a systematic yeast two-hybrid screen. A full QuickGO export for that reference returns 85,343 annotations across 6,769 distinct gene products, 84,937 of them GO:0005515 - this is a maximal-projection reference, and APOC4's 23 rows are 23 of those. Resolving all 23 partners in UniProt, none has a curated Secreted or Extracellular subcellular location, 10 are curated mitochondrial and 10 carry a transmembrane segment (file:human/APOC4/APOC4-bioinformatics/RESULTS.md). Two points follow, and they are not equally firm. The firm one does not depend on any partner's membrane topology: not one of the 23 is a plasma protein of the kind apoC-IV actually encounters, and every pair was scored by a Gal4 two-hybrid, which reports interactions reconstituted in the yeast nucleus - a compartment APOC4's mature chain cannot occupy at all. The softer one is each partner's own compartment, and it is graded rather than blanket: strongest for the ten curated mitochondrial partners, weakest for the ER- and Golgi-membrane ones, because APOC4 does transit those organelles on its way out of the hepatocyte and the argument then turns on which face of the membrane the partner works on - which is not settled for every partner here. Where that per-partner point is strong it is made in the row itself. HuRI itself notes that most of its pairs were seen in a single screen, and no functional follow-up exists for any APOC4 pair; the NbExp=3 shown against each of these pairs in the UniProt INTERACTION block is HuRI's own screen-and-retest count from this one publication, not three independent studies. Per project policy an experimental row is not removed on this basis; the term is also uninformative, and there is no better MF to MODIFY it to, so it is marked as over-annotated. Supporting Evidence: PMID:32296183 yeast two-hybrid (Y2H) represents the only binary PPI assay that can be operated at sufficient throughput to systematically screen the human proteome for binary PPIs PMID:32296183 the majority of PPIs in HuRI were found in only one screen file:human/APOC4/APOC4-uniprot.txt FT SIGNAL 1..27 |
| GO:0005515 protein binding | IPI PMID:32296183 A reference map of the human binary protein interactome. | MARK AS OVER ANNOTATED | Summary: Binary yeast two-hybrid interaction with UBQLN1 (Ubiquilin-1), reported only in the HuRI reference interactome [PMID:32296183]. UniProt curates UBQLN1 to Cytoplasm; Nucleus; Endoplasmic reticulum; Cytoplasmic vesicle, autophagosome; Cell membrane. Reason: All 23 GO:0005515 rows on APOC4 come from one reference, PMID:32296183 (HuRI), a systematic yeast two-hybrid screen. A full QuickGO export for that reference returns 85,343 annotations across 6,769 distinct gene products, 84,937 of them GO:0005515 - this is a maximal-projection reference, and APOC4's 23 rows are 23 of those. Resolving all 23 partners in UniProt, none has a curated Secreted or Extracellular subcellular location, 10 are curated mitochondrial and 10 carry a transmembrane segment (file:human/APOC4/APOC4-bioinformatics/RESULTS.md). Two points follow, and they are not equally firm. The firm one does not depend on any partner's membrane topology: not one of the 23 is a plasma protein of the kind apoC-IV actually encounters, and every pair was scored by a Gal4 two-hybrid, which reports interactions reconstituted in the yeast nucleus - a compartment APOC4's mature chain cannot occupy at all. The softer one is each partner's own compartment, and it is graded rather than blanket: strongest for the ten curated mitochondrial partners, weakest for the ER- and Golgi-membrane ones, because APOC4 does transit those organelles on its way out of the hepatocyte and the argument then turns on which face of the membrane the partner works on - which is not settled for every partner here. Where that per-partner point is strong it is made in the row itself. HuRI itself notes that most of its pairs were seen in a single screen, and no functional follow-up exists for any APOC4 pair; the NbExp=3 shown against each of these pairs in the UniProt INTERACTION block is HuRI's own screen-and-retest count from this one publication, not three independent studies. Per project policy an experimental row is not removed on this basis; the term is also uninformative, and there is no better MF to MODIFY it to, so it is marked as over-annotated. Supporting Evidence: PMID:32296183 yeast two-hybrid (Y2H) represents the only binary PPI assay that can be operated at sufficient throughput to systematically screen the human proteome for binary PPIs PMID:32296183 the majority of PPIs in HuRI were found in only one screen file:human/APOC4/APOC4-uniprot.txt FT SIGNAL 1..27 |
| GO:0005576 extracellular region | IEA GO_REF:0000044 | ACCEPT | Summary: Automatic mapping of the UniProt subcellular-location term SL-0243 ('Secreted') to GO:0005576. APOC4's UniProt record carries that location, and it is well grounded: the protein was isolated from human plasma VLDL [PMID:12700345], and the rabbit orthologue is demonstrably secreted from transfected mammalian cells [PMID:8576182]. Reason: The keyword-to-GO mapping is correct and the underlying keyword is correct. GO:0005576 is defined as 'The space external to the outermost structure of a cell', which is where a liver-secreted plasma apolipoprotein acts. The term is broad, but for a protein whose compartment is blood plasma there is no more specific GO cellular-component term that would be better; the informative localisation is captured separately by the VLDL and HDL particle rows. Propagation Review Root cause: NO FAILURE CORE Sources checked: UniProtKB-SubCell:SL-0243 · Secreted (UniProt subcellular location vocabulary, cellular component) SUPPORTS TRANSFER SL-0243 resolves to 'Secreted' in the UniProt controlled vocabulary (rest.uniprot.org/locations/SL-0243), and APOC4's own record asserts that location and adds that the protein is expressed by the liver and secreted in plasma. The keyword is supported by direct isolation of the protein from human plasma and by demonstrated secretion of the rabbit orthologue, so both the source assertion and its mapping to GO:0005576 hold. Supporting Evidence: file:human/APOC4/APOC4-uniprot.txt CC -!- SUBCELLULAR LOCATION: Secreted. file:human/APOC4/APOC4-uniprot.txt CC -!- TISSUE SPECIFICITY: Expressed by the liver and secreted in plasma. PMID:8576182 Sialylated apolipoprotein C-IV is secreted from transfected mammalian cells. |
| GO:0005576 extracellular region | TAS Reactome:R-HSA-8854408 | ACCEPT | Summary: Reactome traceable-author statement from 'APOBR dimer binds VLDLs' (Reactome:R-HSA-8854408) placing the secreted APOC4 protein in the extracellular region. APOC4 appears in this Reactome reaction as a component of the VLDL particle that the macrophage apolipoprotein B receptor binds. The extracellular-region assignment follows from the particle being in blood. Reason: Correct and consistent with the directly established biology: APOC4 is made by the liver, secreted, and recovered from blood plasma. Four Reactome reactions each generate one such row, which is duplication rather than four independent claims, but duplicate rows for the same term are not a defect and none of them is wrong. All four cached Reactome summaries are APOC4-specific and two cite the same primary isolation paper that supports the IDA rows, so this is not a generic pathway projection. Supporting Evidence: file:human/APOC4/APOC4-uniprot.txt CC -!- SUBCELLULAR LOCATION: Secreted. PMID:12700345 Human apoC-IV, isolated from apo VLDL by DEAE-cellulose chromatography and two-dimensional electrophoresis, was identified by microsequencing four tryptic peptides. |
| GO:0005576 extracellular region | TAS Reactome:R-HSA-8854462 | ACCEPT | Summary: Reactome traceable-author statement from 'VLDLR binds VLDL' (Reactome:R-HSA-8854462) placing the secreted APOC4 protein in the extracellular region. APOC4 appears here as a constituent of the VLDL particle bound by VLDLR; the Reactome summary cites Kotite et al. 2003 for apoC-IV being a minor VLDL constituent, which is the same evidence as the IDA row. Reason: Correct and consistent with the directly established biology: APOC4 is made by the liver, secreted, and recovered from blood plasma. Four Reactome reactions each generate one such row, which is duplication rather than four independent claims, but duplicate rows for the same term are not a defect and none of them is wrong. All four cached Reactome summaries are APOC4-specific and two cite the same primary isolation paper that supports the IDA rows, so this is not a generic pathway projection. Supporting Evidence: file:human/APOC4/APOC4-uniprot.txt CC -!- SUBCELLULAR LOCATION: Secreted. PMID:12700345 Human apoC-IV, isolated from apo VLDL by DEAE-cellulose chromatography and two-dimensional electrophoresis, was identified by microsequencing four tryptic peptides. |
| GO:0005576 extracellular region | TAS Reactome:R-HSA-8866321 | ACCEPT | Summary: Reactome traceable-author statement from 'VLDL binds APOC1 and APOC4' (Reactome:R-HSA-8866321) placing the secreted APOC4 protein in the extracellular region. This is the reaction in which APOC4 is the participant rather than a bystander: newly secreted VLDL acquiring apoC-I and apoC-IV in the circulation, again citing Kotite et al. 2003. Reason: Correct and consistent with the directly established biology: APOC4 is made by the liver, secreted, and recovered from blood plasma. Four Reactome reactions each generate one such row, which is duplication rather than four independent claims, but duplicate rows for the same term are not a defect and none of them is wrong. All four cached Reactome summaries are APOC4-specific and two cite the same primary isolation paper that supports the IDA rows, so this is not a generic pathway projection. Supporting Evidence: file:human/APOC4/APOC4-uniprot.txt CC -!- SUBCELLULAR LOCATION: Secreted. PMID:12700345 Human apoC-IV, isolated from apo VLDL by DEAE-cellulose chromatography and two-dimensional electrophoresis, was identified by microsequencing four tryptic peptides. |
| GO:0005576 extracellular region | TAS Reactome:R-HSA-9035279 | ACCEPT | Summary: Reactome traceable-author statement from 'Expression of APOC4 regulated by NR1H2 or NR1H3' (Reactome:R-HSA-9035279) placing the secreted APOC4 protein in the extracellular region. A gene-expression event - liver X receptors inducing the APOE/APOC1/APOC2/APOC4 cluster in macrophages - in which the APOC4 protein is the product. The extracellular-region assignment is for the secreted product. Reason: Correct and consistent with the directly established biology: APOC4 is made by the liver, secreted, and recovered from blood plasma. Four Reactome reactions each generate one such row, which is duplication rather than four independent claims, but duplicate rows for the same term are not a defect and none of them is wrong. All four cached Reactome summaries are APOC4-specific and two cite the same primary isolation paper that supports the IDA rows, so this is not a generic pathway projection. Supporting Evidence: file:human/APOC4/APOC4-uniprot.txt CC -!- SUBCELLULAR LOCATION: Secreted. PMID:12700345 Human apoC-IV, isolated from apo VLDL by DEAE-cellulose chromatography and two-dimensional electrophoresis, was identified by microsequencing four tryptic peptides. |
| GO:0006629 lipid metabolic process | TAS PMID:8530039 Identification and characterization of a new human gene (APO... | KEEP AS NON CORE | Summary: Author statement, from the gene-identification paper, that APOC4 belongs to the apolipoprotein gene family on the basis of its hepatic expression and predicted structure [PMID:8530039]. The inference to lipid metabolic process is sound but generic. Reason: GO:0006629 lipid metabolic process is true of APOC4 but says almost nothing: its definition spans every reaction involving any lipid. It is worth being exact about why it is uninformative, because the obvious justification is wrong. This is not a broad parent sitting above more specific children already on the gene: the two particle rows are cellular components and so cannot be descendants of a biological-process term at all, and neither of APOC4's other process terms is under it either - GO:0070328 triglyceride homeostasis descends through lipid homeostasis (GO:0055088) and GO:0019915 lipid storage through cellular process, with GO:0006629 in neither ancestry (checked against the QuickGO ancestors endpoint). So the row is not redundant with anything; it is simply the most general true statement that can be made about the gene, and it is the only metabolic-process term APOC4 carries. Retained because it is not incorrect and removing a true term buys nothing, but it carries no information a reader could act on and is not a core function statement. That a generic parent is still the gene's only metabolic-process annotation is a fact about the evidence rather than about the curation: the affinage deep-research record, working independently from the literature, reaches the same conclusion about how little has been established. Supporting Evidence: PMID:8530039 Since its hepatic expression and predicted protein structure are characteristic of the other genes in this cluster, we propose that the APOC4 gene is a member of the apolipoprotein gene family. PMID:8530039 RNase protection analysis indicated relatively low apoC-IV mRNA levels in human liver, compared to apoC-II mRNA levels. file:human/APOC4/APOC4-deep-research-affinage.md Beyond this molecular characterization and its association with lipid traits, no enzymatic activity, receptor interaction, or reconstituted mechanism for apoC-IV has been characterized in the available corpus. |
| GO:0019915 lipid storage | IBA GO_REF:0000033 | KEEP AS NON CORE | Summary: Phylogenetic propagation of lipid storage from the PTHR32288 apolipoprotein C-IV node. The PAINT slice shows this IBD carries a single seed, UniProtKB:P55056 - the human gene itself - and is dated 2025-06-24, seven months after the BHF-UCL IMP (2024-11-01) it rests on. That IMP is one experiment: transient transfection of the APOC4 cDNA into Huh-7 hepatoma cells produced cytoplasmic lipid droplets and about a 2.7-fold rise in triglyceride [PMID:18809223]. Reason: The node placement is not the problem - the family is eutherian and human APOC4 is squarely inside the clade. The problem is the depth of the assertion relative to its evidence. The whole IBD, and the eleven IBA rows that follow from it, rest on one ectopic gain-of-function in a transformed hepatoma line, with no loss-of-function for APOC4 in any species and no independent replication in any other eutherian. It is worth separating this row from the IMP that seeds it, because the two are graded the same way for different reasons: the IMP is a real, conservatively coded experiment and is non-core because one ectopic overexpression does not establish a physiological role, whereas this IBA is a propagation whose reach - eleven eutherian species - is what over-extends. On the propagation considered alone MARK_AS_OVER_ANNOTATED would be arguable. KEEP_AS_NON_CORE is used instead because the term is not wrong for the target, because the two rows carrying this term should carry one action, and because the over-reach is recorded where it belongs - in propagation_review as SOURCE_WEAK_OR_INFERRED - rather than by downgrading a claim the evidence does support at the level of the target itself. Intracellular lipid storage is also at a tangent to everything else established about apoC-IV, which is a secreted plasma apolipoprotein: in that paper APOC4 is mostly the regulated readout, with HCV core protein, Ku antigen and PPARgamma/RXRalpha as the agents, and the in vivo arm is explicitly a correlation between transcript level and hepatic triglyceride. Kept, because an overexpression phenotype is real evidence and the claim is not contradicted, but not as a core function. Propagation Review Root cause: SOURCE WEAK OR INFERRED Failure modes: SOURCE EVIDENCE WEAK Sources checked: PANTHER:PTN000792756 · PTHR32288 APOLIPOPROTEIN C-IV IBD node at taxon:9347 (Eutheria) SOURCE WEAK OR INFERRED Node placement is sound - see the VLDL row for the taxonomic argument - but for this term the node has exactly one seed, read directly from interpro/panther/PTHR32288/PTHR32288-paint.tsv, and that seed's evidence is a single cell-culture overexpression. The IBD nonetheless asserts that lipid storage is an ancestral eutherian property of apoC-IV and propagates it to all eleven descendants. That is a large taxonomic reach for one transfection experiment, which is what makes the source weak rather than the propagation wrong. UniProtKB:P55056 · APOC4 (human, the target) SUPPORTS TRANSFER The target's own accession among the sources is expected and not circular: human APOC4's experimental annotations are among the descendant evidences the PAINT curator used to place the IBD, so it legitimately appears among the sources of the IBA it later receives. Here that evidence is the BHF-UCL IMP from PMID:18809223 and it is the only seed the IBD has for this term. It supports the transfer in the sense that the descendant evidence is genuine and experimental; what it does not support is treating lipid storage as a core function, which is why the row is kept as non-core rather than accepted. Supporting Evidence: PMID:18809223 Considering the transfection efficiency of about 40%, the triglyceride content in ApoC-IV transfected cells was approximately 2.7-fold higher than that in vector transfected cells. PMID:18809223 we identified host transcription factors regulating the ApoC-IV gene expression PMID:37226733 Endogenous APOC4 is thought to increase plasma triglyceride levels because overexpression of human APOC4 at supraphysiological levels in mice causes elevated levels of VLDL. |
| GO:0019915 lipid storage | IMP PMID:18809223 Expression of apolipoprotein C-IV is regulated by Ku antigen... | KEEP AS NON CORE | Summary: Transient transfection of the APOC4 cDNA into Huh-7 hepatoma cells produced lipid droplets in the cytoplasm and perinuclear space and raised cell triglyceride roughly 2.7-fold after correcting for a 40% transfection efficiency [PMID:18809223]. In the same study APOC4 transcript level correlated with hepatic triglyceride in HCV-infected livers. Reason: This is a genuine, statistically supported gain-of-function result and BHF-UCL coded it conservatively. It is retained. But it is a single ectopic overexpression in a transformed line: there is no knockdown of APOC4 itself anywhere in the paper (the silencing experiments target Ku70), no loss-of-function model for APOC4 in any species, and the in vivo evidence is a correlation. The bulk of the paper establishes what controls APOC4 expression rather than what APOC4 does. Intracellular lipid droplet accumulation is also hard to reconcile mechanistically with a protein whose mature form is secreted into plasma, which is a further reason to treat it as peripheral until a physiological-dose experiment exists. Supporting Evidence: PMID:18809223 cells transfected with the ApoC-IV cDNA displayed lipid droplets of granule-like structure in the cytoplasm and perinuclear space, distinct from the weak staining and homogenous distribution observed in cells transfected with the empty vector PMID:18809223 Considering the transfection efficiency of about 40%, the triglyceride content in ApoC-IV transfected cells was approximately 2.7-fold higher than that in vector transfected cells. PMID:18809223 the ApoC-IV transcript levels correlated with the concentration of intracellular triglycerides in the HCV infected livers as revealed by a coefficient of determination (R2 = 0.78) file:human/APOC4/APOC4-uniprot.txt CC -!- SUBCELLULAR LOCATION: Secreted. |
| GO:0034361 very-low-density lipoprotein particle | IBA GO_REF:0000033 | ACCEPT | Summary: Phylogenetic propagation of very-low-density lipoprotein particle membership from the PTHR32288 apolipoprotein C-IV node. Two gene-level donors: MGI:MGI:87878, which resolves through the UniProt cross-reference to Q61268 (APOC4_MOUSE, Swiss-Prot), and the human target itself. This is the best-supported of APOC4's four IBAs and it agrees with direct evidence on the human protein. Reason: GO:0034361 is defined as a triglyceride-rich particle 'typically composed of APOB100, APOE and APOCs' - apoC-IV is an apoC, and more than 80% of plasma apoC-IV is on VLDL [PMID:12700345]. The node placement survives every check available: PTHR32288 is a single-subfamily, apoC-IV-only family, the IBD sits at Eutheria, and 185 of 189 family members are eutherian, matching the published origin of the gene at that node [PMID:31722659]. The human protein carries the same annotation independently by IDA from two unrelated methods and by IMP from the transgenic model, so nothing about the transfer is in tension with target-specific data. This is the core cellular-component assignment for APOC4. Propagation Review Root cause: NO FAILURE CORE Sources checked: MGI:MGI:87878 · Apoc4 (Mus musculus); UniProtKB:Q61268, APOC4_MOUSE SUPPORTS TRANSFER Resolved through the UniProt cross-reference search (query=xref:mgi-87878), which returns the single reviewed entry Q61268, APOC4_MOUSE - the mouse orthologue, not a paralog. Worth stating precisely: Q61268's only experimental GO annotation in QuickGO is GO:0034361 IDA from PMID:12700345, which is the mouse-plasma clause of the same study that gives the human target its IDA. So the row's two gene-level donors are two species observed in one publication, not two independent studies. That does not weaken the claim - it is a direct isolation, and human VLDL proteomics (PMID:17154273) replicates it by a different method - but it is not the extra independence a second donor usually implies. PANTHER:PTN000792756 · PTHR32288 APOLIPOPROTEIN C-IV IBD node at taxon:9347 (Eutheria) SUPPORTS TRANSFER The IBD sits at taxon:9347 (Eutheria) in PTHR32288, a single-subfamily family (SF0) whose members are all named Apolipoprotein C-IV, so there is no apoC-I/C-II/C-III biology in this node to leak. The family is effectively co-extensive with the node's clade: 185 of 189 InterPro-assigned members and 146 of 149 source organisms are inside Eutheria, the only non-eutherian apoC-IV sequences being three marsupial ones (file:human/APOC4/APOC4-bioinformatics/RESULTS.md), which matches the published phylogeny placing the gene's origin at Eutheria (PMID:31722659). Human APOC4 is therefore inside the inheriting clade by construction, and the cached PAINT GAF shows the node propagating all four of its terms to exactly 11 eutherian gene products with no IRD or IKR anywhere in the slice. UniProtKB:P55056 · APOC4 (human, the target) SUPPORTS TRANSFER The target's own accession among the sources is expected and not circular: human APOC4's experimental annotations are among the descendant evidences the PAINT curator used to place the IBD, so it legitimately appears among the sources of the IBA it later receives. For this term the target's descendant evidence is unusually strong: two independent IDAs (plasma isolation with tryptic-peptide microsequencing, PMID:12700345; 2-DE/MALDI-TOF-TOF VLDL proteomics, PMID:17154273) and an IMP from the transgenic model (PMID:8827523). Supporting Evidence: PMID:12700345 In normolipidemic plasma, greater than 80% of the protein is in VLDL (0.7% of total apo VLDL), with most of the remainder in HDL. PMID:12700345 We show that apoC-IV is a basic protein in human, monkey, and mouse plasma, present as a minor apoC component of VLDL. PMID:17154273 Several VLDL-associated apolipoproteins were identified, including five isoforms of apoE, three isoforms of apoC-IV, and one isoform each of apoC-III, apoM, apoA-I, and apoA-IV. file:human/APOC4/APOC4-bioinformatics/RESULTS.md 98% of the family is eutherian. The only credible extension beyond Eutheria is three marsupial (Metatheria) sequences, so the family spans at most Theria |
| GO:0034361 very-low-density lipoprotein particle | IDA PMID:12700345 Human apoC-IV: isolation, characterization, and immunochemic... | ACCEPT | Summary: Human apoC-IV was purified from apoVLDL by DEAE-cellulose chromatography and two-dimensional electrophoresis and identified by microsequencing four tryptic peptides; in normolipidaemic plasma more than 80% of the protein is in VLDL, where it accounts for 0.7% of total apoVLDL [PMID:12700345]. Reason: This is the strongest single piece of evidence about APOC4 and the anchor for the whole review. It is a direct assay on the native human protein in native particles, not a recombinant or transgenic system, and the identification is by peptide sequencing. It also supersedes the earlier immunoblot report that apoC-IV could not be detected in normal adult human plasma [PMID:8827523], which was a sensitivity limit rather than an absence - the measured concentration range, 0.1-1.9 mg/dl, is about an order of magnitude below the other apoCs [PMID:37226733]. VLDL membership is the core localisation of this protein. Supporting Evidence: PMID:12700345 Human apoC-IV, isolated from apo VLDL by DEAE-cellulose chromatography and two-dimensional electrophoresis, was identified by microsequencing four tryptic peptides. PMID:12700345 In normolipidemic plasma, greater than 80% of the protein is in VLDL (0.7% of total apo VLDL), with most of the remainder in HDL. PMID:12700345 The concentration of apoC-IV in the plasma and lipoproteins of rho < 1.21 g/ml is closely related to plasma triglyceride concentration up to 1,770 mg/dl, varying from 0.1-1.9 mg/dl. PMID:37226733 plasma levels of APOC4 are at least an order of magnitude lower than those of APOC1, APOC2 and APOC3 |
| GO:0034361 very-low-density lipoprotein particle | IDA PMID:17154273 Proteomic analysis of human very low-density lipoprotein by ... | ACCEPT | Summary: A two-dimensional gel / MALDI-TOF-TOF protein map of human VLDL identified three isoforms of apoC-IV among the VLDL-associated apolipoproteins [PMID:17154273]. Reason: Independent methodological confirmation of the same cellular component by a different laboratory using a different technique, on human VLDL. Detection of three isoforms is also consistent with the glycoform and sialoform heterogeneity described for the purified protein [PMID:12700345] and with the N-glycosylation site UniProt maps at Asn-63 from plasma glycoproteomics, so the identification is not a spurious spot assignment. Supporting Evidence: PMID:17154273 Several VLDL-associated apolipoproteins were identified, including five isoforms of apoE, three isoforms of apoC-IV, and one isoform each of apoC-III, apoM, apoA-I, and apoA-IV. PMID:12700345 The protein exhibits two major isoforms; one is N-glycosylated, and both are variably sialylated. file:human/APOC4/APOC4-uniprot.txt FT CARBOHYD 63 |
| GO:0034361 very-low-density lipoprotein particle | IEA GO_REF:0000107 | ACCEPT | Summary: Ensembl Compara orthology transfer of VLDL particle membership from mouse Apoc4 to human APOC4. The claim is correct, but the donor's own GO:0034361 is an IDA from PMID:12700345 - the same publication that already gives the human target a direct IDA for this term. Reason: The biology is right and the orthology is right: Q61268 is the genuine mouse orthologue and Kotite et al. reported apoC-IV as a minor apoC component of VLDL in mouse plasma as well as human. What the row adds evidentially is nothing. It re-derives, by a round trip through the mouse, a claim the human gene already holds by direct experiment from the very same study, and the reciprocal transfer also runs in the other direction (mouse Q61268 carries the same term by GO_REF:0000107 from human ENSP00000468236). The term is kept because duplicate rows for a correct term are not a defect and because VLDL membership is APOC4's core localisation; the redundancy is recorded rather than acted on. Propagation Review Root cause: EVIDENCE CIRCULAR OR REDUNDANT Sources checked: UniProtKB:Q61268 · Apoc4 (Mus musculus), APOC4_MOUSE CIRCULAR OR REDUNDANT The donor is the correct mouse orthologue and its annotation is sound: QuickGO shows Q61268's only experimental GO annotation is GO:0034361 IDA from PMID:12700345. That is the same publication behind the human target's own IDA, so the orthology transfer returns a claim the target already holds from the same experiment. This is the 'target already has stronger direct evidence' sense of redundancy, not a broken chain: the mouse gene's three other GO:0034361 rows (IBA, ISO by GO_REF:0000119, and IEA by GO_REF:0000107) are all propagations that name human P55056 among their sources, so the IDA is the one place independent mouse evidence enters - and it enters from the human target's own paper. ensembl:ENSMUSP00000003071 · Ensembl protein for the mouse Apoc4 gene CIRCULAR OR REDUNDANT This is the Compara protein identifier for the same mouse gene product as UniProtKB:Q61268, not a second donor. The row therefore has one donor under two identifiers; no independent source is added by the second entry. Supporting Evidence: PMID:12700345 We show that apoC-IV is a basic protein in human, monkey, and mouse plasma, present as a minor apoC component of VLDL. PMID:12700345 In normolipidemic plasma, greater than 80% of the protein is in VLDL (0.7% of total apo VLDL), with most of the remainder in HDL. |
| GO:0034361 very-low-density lipoprotein particle | IMP PMID:8827523 Expression of a novel human apolipoprotein (apoC-IV) causes ... | ACCEPT | Summary: Human apoC-IV expressed from a cDNA transgene in mice was recovered with plasma lipoproteins of d < 1.21 g/ml, mainly with VLDL, in higher-molecular-mass glycosylated and sialylated forms [PMID:8827523]. Reason: The particle assignment agrees with the direct evidence on native human protein [PMID:12700345, PMID:17154273], so the conclusion is not in doubt. The grading deserves a note: this is heterologous expression of the human protein in mouse at a level the authors and later reviewers describe as supraphysiological [PMID:37226733], so it is gain-of-function, ectopic evidence - IMP-grade, as BHF-UCL coded it, rather than IDA on the native protein. Accepted as a correct and corroborating row, with the native-plasma IDA carrying the weight. Supporting Evidence: PMID:8827523 Human apoC-IV was found associated with plasma lipoproteins (d < 1.21 g/ml), mainly in very low density lipoproteins (VLDL), and higher molecular mass isoforms were present, due to N-linked glycosylation and variable sialylation of apoC-IV. PMID:8827523 our analysis of transgenic mice provides unequivocal evidence that human apoC-IV is a lipid-binding protein belonging to the apolipoprotein family and that it has the potential to alter lipoprotein metabolism. PMID:37226733 Endogenous APOC4 is thought to increase plasma triglyceride levels because overexpression of human APOC4 at supraphysiological levels in mice causes elevated levels of VLDL. |
| GO:0034364 high-density lipoprotein particle | IBA GO_REF:0000033 | KEEP AS NON CORE | Summary: Phylogenetic propagation of high-density lipoprotein particle membership from the PTHR32288 apolipoprotein C-IV node. The PAINT slice shows a single seed for this term, UniProtKB:P55056, whose descendant evidence is one clause of the human plasma isolation: with more than 80% of apoC-IV in VLDL, 'most of the remainder' is in HDL [PMID:12700345]. Reason: The claim is right and independently corroborated since the IBD was made - apoC-IV is recovered as an HDL-bound apolipoprotein in two later human HDL proteomic studies [PMID:36173828, PMID:32844720], and a 2023 review states plainly that APOC4 is found on circulating triglyceride-rich lipoproteins and HDL like the other apoCs [PMID:37226733]. The definition fits: GO:0034364 describes a particle that 'contains APOAs and may contain APOCs and APOE'. It is non-core rather than core because it is quantitatively the minor pool - under 20% of a protein that is itself an order of magnitude less abundant than the other apoCs - and because exchangeable apoCs redistributing between VLDL and HDL is expected class behaviour rather than a distinguishing feature of apoC-IV. The node placement raises no concern; see the VLDL row for the taxonomic argument. Propagation Review Root cause: NO FAILURE NON CORE Sources checked: PANTHER:PTN000792756 · PTHR32288 APOLIPOPROTEIN C-IV IBD node at taxon:9347 (Eutheria) SUPPORTS TRANSFER The IBD sits at taxon:9347 (Eutheria) in PTHR32288, a single-subfamily family (SF0) whose members are all named Apolipoprotein C-IV, so there is no apoC-I/C-II/C-III biology in this node to leak. The family is effectively co-extensive with the node's clade: 185 of 189 InterPro-assigned members and 146 of 149 source organisms are inside Eutheria, the only non-eutherian apoC-IV sequences being three marsupial ones (file:human/APOC4/APOC4-bioinformatics/RESULTS.md), which matches the published phylogeny placing the gene's origin at Eutheria (PMID:31722659). Human APOC4 is therefore inside the inheriting clade by construction, and the cached PAINT GAF shows the node propagating all four of its terms to exactly 11 eutherian gene products with no IRD or IKR anywhere in the slice. For this term specifically the IBD carries one seed rather than two, read from interpro/panther/PTHR32288/PTHR32288-paint.tsv. A single seed is not weak support in itself - the curator's claim is about where the function arose, made with the alignment and tree in view - and the rabbit orthologue, which was shown biochemically to associate with plasma VLDL and HDL (PMID:8576182), could not be a seed because its GO record contains no experimental annotation at all. UniProtKB:P55056 · APOC4 (human, the target) SUPPORTS TRANSFER The target's own accession among the sources is expected and not circular: human APOC4's experimental annotations are among the descendant evidences the PAINT curator used to place the IBD, so it legitimately appears among the sources of the IBA it later receives. The descendant evidence here is the BHF-UCL IDA from PMID:12700345. It is a direct measurement on native human protein and it has since been reproduced in two independent HDL proteomic datasets, so the seed is sound; the reservation is about the size of the pool, not the reality of it. Supporting Evidence: PMID:12700345 In normolipidemic plasma, greater than 80% of the protein is in VLDL (0.7% of total apo VLDL), with most of the remainder in HDL. PMID:36173828 our work identified 5 HDL-bound apolipoproteins that were significantly decreased in patients with NAFLD and advanced fibrosis: ApoC-I, ApoC-IV, ApoM, LCAT, and SAA4. PMID:32844720 Changes in nine proteins (APOE, APOF, APOL1, APMAP, APOC4, APOM, LPA, PCYOX1, PON1) significantly correlated PMID:37226733 APOC4 is found on circulating TRLs and HDL, like the other APOCs. |
| GO:0034364 high-density lipoprotein particle | IDA PMID:12700345 Human apoC-IV: isolation, characterization, and immunochemic... | KEEP AS NON CORE | Summary: In the same plasma isolation that established VLDL as apoC-IV's main carrier, the remainder of the protein was found mostly in HDL [PMID:12700345]. Reason: A direct observation on native human plasma, and therefore retained without qualification as to its correctness; two later human HDL proteomic studies independently detect apoC-IV as an HDL-bound apolipoprotein [PMID:36173828, PMID:32844720]. Marked non-core because this is the minority fraction of an already scarce protein, and the HDL pool of an exchangeable apoC is best read as the exchange reservoir rather than the site of action. Worth noting for curators: PMID:36173828 states it found no prior reports on apoC-IV levels in liver disease, which is a measure of how thin the HDL literature on this protein still is. Supporting Evidence: PMID:12700345 In normolipidemic plasma, greater than 80% of the protein is in VLDL (0.7% of total apo VLDL), with most of the remainder in HDL. PMID:36173828 our work identified 5 HDL-bound apolipoproteins that were significantly decreased in patients with NAFLD and advanced fibrosis: ApoC-I, ApoC-IV, ApoM, LCAT, and SAA4. PMID:36173828 We found no prior reports on SAA4 or ApoC-IV levels in liver diseases. |
| GO:0070328 triglyceride homeostasis | IBA GO_REF:0000033 | KEEP AS NON CORE | Summary: Phylogenetic propagation of triglyceride homeostasis from the PTHR32288 apolipoprotein C-IV node, which for this term carries a single seed, UniProtKB:P55056. The seed's descendant evidence is the transgenic-mouse phenotype: human apoC-IV transgenic mice were hypertriglyceridaemic, with the accumulated triglyceride in VLDL [PMID:8827523]. Reason: Triglyceride is the one functional axis on which apoC-IV has ever produced a signal, and it produces it repeatedly and in a consistent direction: transgenic overexpression raises plasma triglyceride [PMID:8827523], plasma apoC-IV concentration tracks plasma triglyceride across a wide range [PMID:12700345], and two coding polymorphisms associate with triglyceride levels in women [PMID:10996355]. The transfer itself is unobjectionable: the node is correctly placed and the human gene is inside the clade. It is non-core because every strand is gain-of-function or correlative - there is no apoC-IV loss-of-function model in any species, the transgene was expressed at supraphysiological levels, and the genetic contribution to plasma triglyceride was about 2% [PMID:37226733]. What endogenous apoC-IV contributes at its native concentration is unestablished, and the review literature says so. Propagation Review Root cause: NO FAILURE NON CORE Sources checked: PANTHER:PTN000792756 · PTHR32288 APOLIPOPROTEIN C-IV IBD node at taxon:9347 (Eutheria) SUPPORTS TRANSFER The IBD sits at taxon:9347 (Eutheria) in PTHR32288, a single-subfamily family (SF0) whose members are all named Apolipoprotein C-IV, so there is no apoC-I/C-II/C-III biology in this node to leak. The family is effectively co-extensive with the node's clade: 185 of 189 InterPro-assigned members and 146 of 149 source organisms are inside Eutheria, the only non-eutherian apoC-IV sequences being three marsupial ones (file:human/APOC4/APOC4-bioinformatics/RESULTS.md), which matches the published phylogeny placing the gene's origin at Eutheria (PMID:31722659). Human APOC4 is therefore inside the inheriting clade by construction, and the cached PAINT GAF shows the node propagating all four of its terms to exactly 11 eutherian gene products with no IRD or IKR anywhere in the slice. The IBD carries one seed for this term, read from the PAINT slice. Its placement at Eutheria is defensible for a process term of this kind: the plasma triglyceride-transport system the gene participates in is itself conserved across the clade, and nothing suggests human apoC-IV has diverged from its orthologues. UniProtKB:P55056 · APOC4 (human, the target) SUPPORTS TRANSFER The target's own accession among the sources is expected and not circular: human APOC4's experimental annotations are among the descendant evidences the PAINT curator used to place the IBD, so it legitimately appears among the sources of the IBA it later receives. The descendant evidence is the BHF-UCL IMP from PMID:8827523 - an in vivo phenotype with non-transgenic controls, which is stronger than the cell-culture seed behind the lipid-storage IBD and is why that row is graded SOURCE_WEAK_OR_INFERRED while this one is not. The limitation it does carry is dose: the human transgene was expressed at supraphysiological levels in a species whose own apoC-IV is present only in trace amounts. Supporting Evidence: PMID:8827523 Human apoC-IV transgenic mice were hypertriglyceridemic compared to nontransgenic controls; the accumulated plasma triglycerides were present mainly in VLDL. PMID:12700345 The concentration of apoC-IV in the plasma and lipoproteins of rho < 1.21 g/ml is closely related to plasma triglyceride concentration up to 1,770 mg/dl, varying from 0.1-1.9 mg/dl. PMID:10996355 A general linear model analysis reveled a significant association of the Leu36Pro and the Leu96Arg polymorphisms with triglyceride levels in women. PMID:37226733 The biological functions of endogenous APOC4 as well as its potential role in diabetes and diabetes complications need further study. |
| GO:0070328 triglyceride homeostasis | IMP PMID:8827523 Expression of a novel human apolipoprotein (apoC-IV) causes ... | KEEP AS NON CORE | Summary: Mice carrying a human apoC-IV cDNA transgene were hypertriglyceridaemic relative to non-transgenic controls, with the accumulated plasma triglyceride mainly in VLDL and cholesterol redistributed towards VLDL and larger particles [PMID:8827523]. Reason: A clean in vivo gain-of-function result with the proper control, and the only whole-animal functional experiment on this protein, so it is retained. It is peripheral rather than core for two reasons. First, direction of effect is established but physiological necessity is not: there is no knockout, and the transgene was supraphysiological [PMID:37226733]. Second, the mechanism that has been proposed for it is not an apoC-IV activity at all - the elevated VLDL is attributed to displacement of apoE and consequently reduced hepatic clearance, explicitly not to an effect on lipase activity [PMID:37226733] - and that displacement has never been measured for apoC-IV directly. So the row records a real organismal consequence of having more apoC-IV, without identifying what apoC-IV does to produce it. Supporting Evidence: PMID:8827523 Human apoC-IV transgenic mice were hypertriglyceridemic compared to nontransgenic controls; the accumulated plasma triglycerides were present mainly in VLDL. PMID:8827523 There was little change in plasma cholesterol levels, although apoC-IV expression redistributed cholesterol to VLDL and larger particles in low density lipoprotein/large high density lipoprotein fractions. PMID:37226733 The mechanism of the elevated VLDL levels in APOC4 transgenic mice was suggested to be due to displacement of APOE, resulting in reduced hepatic clearance of VLDL, rather than to a reduction in lipase activity |
| GO:0005543 phospholipid binding | ISS PMID:8576182 Identification, characterization, cloning, and expression of... | NEW | Summary: Proposed new molecular function. After review, APOC4's only molecular-function row that survives as informative is GO:0005319 lipid carrier activity, which states the role but not the binding step that underlies it. The binding step has been measured, on the rabbit orthologue: apoC-IV 'contains two potential amphipathic helices characteristic of plasma apolipoproteins and forms discoidal micelles with phosphatidylcholine' [PMID:8576182]. Discoidal-complex formation with phosphatidylcholine is the classical demonstration that an exchangeable apolipoprotein binds and solubilises phospholipid. The rabbit and human proteins are the same PANTHER family and subfamily (PTHR32288:SF0), the gene is 'highly conserved during mammalian evolution' [PMID:8576182], and a hydrophobic-moment scan places the human mature chain 0.0009 segments per residue from the rabbit protein and 0.0021 from APOA1, inside an exchangeable-apolipoprotein band that does not overlap the globular controls (file:human/APOC4/APOC4-bioinformatics/RESULTS.md). Reason: Coded ISS with the rabbit orthologue as the supporting entity, deliberately not IDA or IMP: the discoidal-micelle measurement is on P55057, not on the human protein, and the sequence analysis done here is a statistic rather than an assay. GO:0005543 'Binding to a phospholipid' is the right level - phosphatidylcholine binding (GO:0031210) would over-specify from a single lipid used in one reconstitution, and lipid binding (GO:0008289) would lose the information that the ligand is the particle's phospholipid monolayer. APOC3 carries GO:0005543 by IDA and IBA and APOA1 by IDA and IBA, so this places apoC-IV at the same resolution as its characterised relatives. The annotation also fills a real gap: without it the mechanism by which apoC-IV stays on a lipoprotein surface is unrepresented in GO. One ordering caveat, stated because it affects how the row should be actioned: under GOA's ISS convention the WITH/FROM source is expected to carry the transferred term experimentally, and P55057 currently carries no experimental GO annotation of any kind. This row is therefore contingent on PMID:8576182 first being curated to P55057 - the curation request recorded here as a CURATION knowledge gap and as a suggested question. The measurement exists; it is the GO record of it that does not yet. Propagation Review Root cause: NO FAILURE CORE Sources checked: UniProtKB:P55057 · APOC4 (Oryctolagus cuniculus), APOC4_RABIT SUPPORTS TRANSFER The rabbit orthologue is the only member of this family in which lipid binding has been measured directly, by discoidal-micelle formation with phosphatidylcholine (PMID:8576182); it is also where apoC-IV is abundant enough to purify readily, and where liver-specific expression and secretion from transfected cells were shown. It is a true one-to-one orthologue in PTHR32288 rather than a paralog, and human APOC4's UniProt record derives its own signal-peptide boundary from this entry by similarity. Worth flagging to curators: P55057 carries no experimental GO annotation of any kind - all five of its GO rows are IEA - which is why PAINT could not use the best-characterised member of the family as an IBD seed. The transfer is supported by conservation of the amphipathic-helix architecture that does the binding, tested here rather than assumed. Supporting Evidence: PMID:8576182 It contains two potential amphipathic helices characteristic of plasma apolipoproteins and forms discoidal micelles with phosphatidylcholine. PMID:8576182 The mature protein of 97 amino acids, designated apolipoprotein C-IV, is associated with the lipoproteins of blood plasma, primarily very low density and high density lipoproteins. PMID:8576182 the apolipoprotein C4 gene has been highly conserved during mammalian evolution file:human/APOC4/APOC4-bioinformatics/RESULTS.md Human apoC-IV sits at 0.0300 — inside the apolipoprotein band, between APOC2/C3 (0.0253) and APOC1 (0.0351), about three times the controls (2.97–3.12×), and differing by 0.0009 from rabbit apoC-IV (0.0309) file:human/APOC4/APOC4-uniprot.txt DR PANTHER; PTHR32288; APOLIPOPROTEIN C-IV; 1. |
Loading supporting content…
Download this section (compressed HTML)Q: For the PAINT curators of PTHR32288: the GO:0019915 lipid storage IBD at PTN000792756, added 2025-06-24, has a single seed - the human gene's own IMP from one transient transfection of Huh-7 cells (PMID:18809223) - and it now propagates lipid storage to eleven eutherian orthologues of a secreted plasma apolipoprotein on the strength of cytoplasmic lipid droplets in that one transfected line. Is that the intended reach for an overexpression phenotype in a transformed line, given that the same paper's main finding is what regulates APOC4 rather than what APOC4 does?
Q: For GOC and MGI jointly: human APOC4 carries GO:0034361 twice over by orthology (GO_REF:0000107 from mouse Q61268) while mouse Apoc4 carries the reciprocal transfer from human ENSP00000468236, and the single experimental annotation underlying both is the same IDA from PMID:12700345. Is there value in a Compara transfer whose donor's evidence is the same publication as the recipient's own direct annotation, and should such reciprocal pairs be suppressed?
Q: For UniProt and GOA: rabbit APOC4 (P55057) has five GO annotations, all IEA, yet it is the species where apoC-IV was purified, shown to form discoidal micelles with phosphatidylcholine, localised to VLDL and HDL, and shown to be secreted (PMID:8576182). Could this paper be curated to P55057? It would give PTN000792756 its only non-human, non-overexpression descendant evidence.
Q: For IntAct: all 23 GO:0005515 rows on APOC4 come from HuRI, and none of the 23 partners has a curated secreted or extracellular location while ten are mitochondrial. For a protein whose mature form exists only outside the cell, is there a case for withholding GO protein-binding annotations from nuclear-compartment two-hybrid data unless an orthogonal assay in a compatible compartment supports them?
Q: For the GO ontology editors: GO:0005319 was renamed to 'lipid carrier activity' in February 2026 and now carries 'apolipoprotein' as a narrow synonym, having absorbed several obsoleted apolipoprotein terms. Is a single term the right granularity for a class that spans apoB (structural, non-exchangeable), the exchangeable apoCs, and intracellular lipid-transfer proteins, or would a child term for exchangeable apolipoprotein surface residence be useful?
Experiment: Make the loss-of-function measurement that does not exist. Generate an Apoc4 knockout mouse, or an AAV8-delivered hepatocyte-specific knockdown in adult mice, and measure fasting and post-prandial plasma triglyceride, VLDL triglyceride secretion rate by Poloxamer-407 block, and VLDL turnover by labelled-particle clearance. The prediction from every existing gain-of-function result is lower plasma triglyceride and faster VLDL clearance; a null result at endogenous expression would be equally informative and would argue that the current process annotations describe overexpressed protein.
Experiment: Test the apoE-displacement mechanism directly. Reconstitute triglyceride-rich particles of defined composition, titrate purified recombinant apoC-IV onto them, and measure apoE occupancy by immunoblot and particle binding to immobilised LDLR and LRP1 ectodomains by surface plasmon resonance. Run apoC-I as the positive control, since the displacement model was established there. This distinguishes a specific molecular function from non-specific surface crowding, and only the former would justify a receptor-binding-regulator annotation.
Experiment: Convert the proposed ISS phospholipid-binding annotation into an IDA on the human protein. Express and purify the mature human apoC-IV chain, in both the glycosylated and unglycosylated forms, and run the standard exchangeable-apolipoprotein panel: DMPC clearance kinetics, discoidal complex formation assayed by native gel and negative-stain electron microscopy, and circular dichroism before and after lipid addition to measure helix induction. Rabbit apoC-IV is the built-in positive control, since the assay has already been run on it.
Experiment: Settle whether the human-versus-rabbit abundance difference is transcriptional. Clone the human and rabbit APOC4 promoters, including the rabbit purine-rich repeat and a version with it deleted, into matched luciferase reporters and assay them in the same human and rabbit hepatocyte lines. Pair this with quantitative measurement of apoC-IV protein and mRNA in matched liver samples from both species, so that transcription rate, transcript abundance and plasma protein concentration can be compared on the same axis.
Experiment: Replace the 23 two-hybrid interactions with partners that exist in plasma. Immunoprecipitate native apoC-IV from human plasma, and separately from isolated VLDL and HDL fractions, using an antibody raised against the mature chain, and identify co-eluting proteins by mass spectrometry in normolipidaemic and hypertriglyceridaemic donors. The other apoCs and apoE serve as the internal positive control, since they are known to share the same particles, and any reproducible non-apolipoprotein partner would be the first physiologically plausible interaction for this protein.
What is not known — curated, literature-grounded statements of the open unknowns (the inverse of core functions).
Gap: What apolipoprotein C-IV does at its endogenous concentration is unknown. Every functional claim about the protein comes from adding more of it - a supraphysiological transgene in mouse, or a transient transfection of a hepatoma line - and no loss-of-function experiment has ever been performed on APOC4 in any species.
OPEN BIOLOGY BP_DARK
What is known: Firmly established: apoC-IV is made by the liver, secreted, and recovered from human plasma, where more than 80% of it is on VLDL and most of the rest on HDL; its plasma concentration (0.1-1.9 mg/dl) rises with plasma triglyceride; and raising it experimentally raises plasma triglyceride with the excess in VLDL. Equally firmly established: it is present at roughly a tenth the concentration of the other apoCs, and no knockout, knockdown or naturally occurring null allele has been reported.
Significance: The direction of the gain-of-function effect is what all four of APOC4's IBAs ultimately encode, and it is what propagates to ten other eutherian orthologues. If endogenous apoC-IV is functionally negligible at a tenth the abundance of its relatives, then triglyceride homeostasis and lipid storage are properties of overexpressed apoC-IV rather than of the gene, and the whole node's process assertions are describing an artefact of dose.
What would resolve it: An Apoc4 knockout or conditional hepatic knockdown in mouse, phenotyped for plasma triglyceride, VLDL secretion rate and VLDL clearance, would settle it in one experiment. A knock-in expressing human APOC4 at its own promoter, rather than from a strong transgene, would separate dose from function.
Provenance (the field's own admissions):
Gap: The mechanism by which apoC-IV raises plasma triglyceride has never been measured for apoC-IV. The apoE-displacement model is asserted for APOC1, APOC3 and APOC4 as a group, and the experimental work behind it is on the relatives, not on apoC-IV.
OPEN BIOLOGYCURATION MF_DARK
What is known: Established: transgenic apoC-IV overexpression elevates VLDL triglyceride, and the elevation is attributed to reduced hepatic clearance rather than to reduced lipase activity. Not established: that apoC-IV displaces apoE from the particle, that it interferes with LDLR or LRP1 binding, or that it binds any receptor or enzyme at all. No receptor, lipase or other protein partner of apoC-IV has been demonstrated in a physiologically relevant assay - the 23 GO:0005515 rows on the gene all come from a single yeast two-hybrid screen, scored in the yeast nucleus, and not one of its 23 partners is a plasma protein of the kind apoC-IV encounters.
Significance: This is the difference between apoC-IV having a molecular function in lipoprotein clearance and apoC-IV being a passive surface passenger whose presence displaces other apolipoproteins by mass action. GO currently expresses neither, and cannot until the experiment exists - which is the correct outcome, but it means the gene's molecular-function record stops at 'binds lipid and rides on particles'.
What would resolve it: Surface plasmon resonance or competition experiments measuring apoE displacement from reconstituted VLDL-like particles by apoC-IV, and VLDL clearance assays in Ldlr-/- and Lrp1-deficient backgrounds, would test the proposed mechanism directly. A positive result would support a receptor-binding-regulator molecular function; a negative one would make the case that apoC-IV's effect is non-specific surface occupancy.
Provenance (the field's own admissions):
Gap: The only direct lipid-binding measurement anywhere in the apolipoprotein C-IV family is uncurated in GO, so the family's experimental anchor is invisible to every inference pipeline that uses GO as input.
OPEN CURATION MF_DARK
What is known: Established in the literature: rabbit apoC-IV forms discoidal micelles with phosphatidylcholine, associates with plasma VLDL and HDL, is transcribed only in liver, and is secreted from transfected mammalian cells. Established about the GO record: UniProtKB:P55057 carries five GO annotations and all five are IEA - four from the GO_REF:0000118 PANTHER ancestral-node pipeline and one from the subcellular-location mapping. The rabbit protein therefore could not be, and was not, an IBD seed for PTN000792756.
Significance: Three of the four IBDs at this node rest on a single human seed, and two of those seeds are overexpression experiments. A curated rabbit IDA for phospholipid binding and for VLDL and HDL membership would give the node independent, biochemical, non-human descendant evidence, which is exactly what it currently lacks - and it would turn the ISS row proposed in this review into something better grounded than a similarity inference.
What would resolve it: Curation, not new experiments: annotate PMID:8576182 to UniProtKB:P55057 for GO:0005543 (IDA), GO:0034361 and GO:0034364 (IDA), and GO:0005576 (IDA), then re-examine whether the PTN000792756 IBDs should be re-seeded.
Provenance (the field's own admissions):
Gap: Why apoC-IV is abundant in rabbit plasma and a trace protein in human plasma is unexplained, and so is the function of the conserved N-glycan and sialoform heterogeneity the protein carries in both species.
OPEN BIOLOGY BP_DARK
What is known: Established: the gene is highly conserved across mammals and arose in Eutheria; the human protein exists in two major glycoforms, one N-glycosylated at Asn-63 and both variably sialylated, and the rabbit protein in up to five sialoforms; human apoC-IV mRNA is low in liver relative to apoC-II. A promoter explanation has been proposed - neither gene has a TATA box, and the rabbit 5' UTR carries a GAGA-box-like purine-rich repeat absent from the human gene - but it has not been tested.
Significance: A large species difference in abundance in an otherwise conserved gene - a major apoC component of rabbit plasma, a trace protein in human plasma - is either a regulatory accident or a real difference in how much the protein matters, and which it is determines how far rabbit biochemistry can be transferred to the human protein - which is precisely what the proposed ISS annotation does.
What would resolve it: Reporter assays comparing the human and rabbit APOC4 promoters in the same hepatocyte background, with and without the purine-rich repeat, would test the promoter model. Enzymatic deglycosylation and desialylation of purified apoC-IV followed by particle-binding and clearance assays would test whether the glycan does anything.
Provenance (the field's own admissions):
Loading supporting content…
Download this section (compressed HTML)Loading supporting content…
Download this section (compressed HTML)Loading supporting content…
Download this section (compressed HTML)Loading supporting content…
Download this section (compressed HTML)