Apolipoprotein F is a liver-specific, secreted plasma glycoprotein that governs where cholesteryl ester goes once cholesteryl ester transfer protein (CETP) has removed it from HDL. It is translated as a 326-residue precursor whose C-terminal 162 residues are released by proprotein-convertase cleavage after Arg-164 and secreted as the mature, N- and O-glycosylated protein; the mature chain is strongly acidic while the propeptide it is cut from is basic, and the protein shares no sequence or architectural relationship with the amphipathic-helix apolipoproteins of the APOA, APOC and APOE families. In plasma it occupies two pools. Most of the protein rides in a roughly 470 kDa, apolipoprotein A-I-containing particle of HDL density in which it is inactive; a minority binds LDL, and only this LDL-bound form inhibits CETP. Binding to LDL, and therefore activation, is set by LDL particle size and surface lipid packing, so the active fraction rises in hypercholesterolaemia and falls in hypertriglyceridaemia. The inhibition is selective rather than global: transfers involving LDL are suppressed most, transfers between VLDL and HDL least, and transfer from VLDL to HDL3 is stimulated, so the protein retargets CETP rather than switching it off. It achieves this by occupying and altering the lipoprotein surface on which CETP must dock, not by binding CETP, so it changes which particles take part in lipid transfer without changing what CETP does to them. The net consequence in an animal that expresses CETP is to limit the movement of HDL-derived cholesteryl ester into LDL; loss of the protein raises LDL cholesterol and lowers HDL cholesterol. Orthologues occur throughout bony vertebrates, including rodents that have no CETP at all, so at least one further, CETP-independent role is expected and has not yet been identified.
Definition: Any process that modulates the frequency, rate or extent of low-density lipoprotein particle remodeling.
Justification: GO carries remodeling terms for all four plasma lipoprotein classes (GO:0034372 VLDL, GO:0034373 IDL, GO:0034374 LDL, GO:0034375 HDL) but regulation children for only one of them, VLDL (GO:0010901 regulation, GO:0010902 positive regulation, GO:0010903 negative regulation). The definition of GO:0034374 names the very reaction apolipoprotein F regulates - "the transfer of cholesterol esters from LDL to a triglyceride-rich lipoprotein particle by cholesteryl ester transfer protein (CETP)" - so the process is in GO but its regulation is not, for the one particle class where this gene acts. Apolipoprotein F is the paradigm gene for the missing term and currently has to be annotated to the coarser GO:0032375 negative regulation of cholesterol transport, which loses the particle specificity that is the whole point of this protein. A matching regulation branch under GO:0034375 high-density lipoprotein particle remodeling is needed for the same reason, since the same protein stimulates VLDL-to-HDL3 transfer.
Parent term: low-density lipoprotein particle remodeling
Supporting Evidence:
Definition: Binding to a lipid particle and thereby stopping, preventing or reducing the transfer of lipid to or from that particle by a lipid transfer protein, without binding the lipid transfer protein itself.
Justification: GO's molecular-function-regulator branch assumes the regulator binds its target: GO:0098772 is defined as regulating "the activity of its target via non-covalent binding", and its descendant GO:0141110 transporter inhibitor activity as "Binds to and stops, prevents, or reduces the activity of a transporter". Apolipoprotein F reduces CETP's transfer activity without binding CETP; it binds the lipoprotein particle that CETP must dock on and alters that surface. No GO term expresses substrate-directed inhibition of a transfer reaction, so the best available molecular function for this protein is the plain binding term GO:0030169, which records the interaction but not the consequence. Applying GO:0141110 or GO:0004857 instead would assert a protein-protein interaction that has never been demonstrated, or a catalytic activity that CETP does not have. The proposal is deliberately phrased around what is established; the physical mechanism by which the bound protein alters the surface is still unknown.
Parent term: molecular function regulator activity
Supporting Evidence:
| GO Term | Evidence | Action | Reason |
|---|---|---|---|
| GO:0005102 signaling receptor binding | TAS PMID:8093033 Purification and molecular cloning of human apolipoprotein F... | REMOVE | Summary: No interaction between apolipoprotein F and any receptor has ever been reported, and the cited paper contains no receptor experiment. This is a legacy ProtInc statement-level annotation dated 2003 whose source is a 1994 purification and cDNA cloning study. Reason: GO:0005102 is defined as "Binding to one or more specific sites on a receptor molecule, a macromolecule that undergoes combination with a hormone, neurotransmitter, drug or intracellular messenger to initiate a change in cell function", i.e. a signalling receptor. Day et al. purified apoF from HDL, cloned its cDNA, ran a Northern blot and mapped the gene to chromosome 12; there is no receptor binding assay, no candidate receptor, and no signalling readout in the paper. Nor is any such interaction reported in the three decades of literature since: the 2020 expert review of this protein describes its mechanism as acting on the lipoprotein surface and states that much of it remains undetermined, and names no receptor. Even on the most charitable reading of what ProtInc may have intended - a lipoprotein receptor rather than a signalling receptor - the appropriate term would be GO:0070325 lipoprotein particle receptor binding, which is equally unsupported. The molecular interaction this protein actually makes is with the LDL particle surface, which is captured by the GO:0030169 row added in this review. Supporting Evidence: PMID:8093033 Apolipoprotein F, with an apparent molecular mass of 29 kilodaltons, was purified from human high density lipoproteins using a modified two dimensional electrophoresis procedure. PMID:32520778 While much remains to be determined about the mechanism of LTIP's action, the data suggest LTIP blocks CETP activity by disrupting and/or preventing the binding of CETP to the lipoprotein surface |
| GO:0005319 lipid carrier activity | TAS PMID:6816269 Isolation and characterization of simple and complex lipopro... | MODIFY | Summary: The essence is right - this protein's molecular function is exercised on lipid moving between lipoproteins - but the polarity is inverted. GO:0005319 is the parent class of the lipid transfer activity that CETP has and that apolipoprotein F exists to suppress. The demonstrated molecular activity is binding the LDL particle. Reason: GO:0005319 is defined as "Directly binding to a specific lipid and delivering it either to an acceptor molecule or to a specific location", and the QuickGO ancestor list for GO:0120013 lipid transfer activity contains GO:0005319, so lipid carrier activity is the superclass of the transfer branch that holds GO:0120020 cholesterol transfer activity - CETP's own IDA-supported molecular function. Assigning it to apoF gives the inhibitor the parent class of its target's activity. No study has shown apoF binding a lipid and delivering it to an acceptor. The cited 1982 paper did not test this either: it isolated three apoF-containing particles by immunosorption and determined their lipid composition by thin-layer chromatography, concluding with an explicit suggestion about the particles, not the protein. What has been measured, repeatedly and in human plasma, is that apoF binds LDL, and that this binding is what the inhibitory activity requires. GO:0030169 low-density lipoprotein particle binding is the molecular function supported by the evidence. Proposed replacements: low-density lipoprotein particle binding Supporting Evidence: PMID:6816269 It is suggested that ApoF-containing lipoproteins may be involved in transport and/or esterification of cholesterol. PMID:6816269 Their lipid composition was determined by thin-layer chromatography. PMID:35016907 Apolipoprotein F (ApoF) modulates lipoprotein metabolism by selectively inhibiting cholesteryl ester transfer protein activity on LDL. This ApoF activity requires that it is bound to LDL. |
| GO:0005515 protein binding | IPI PMID:28514442 Architecture of the human interactome defines protein commun... | MARK AS OVER ANNOTATED | Summary: A BioPlex 2.0 affinity-purification hit with GALNT4, a Golgi polypeptide N-acetylgalactosaminyltransferase. ApoF is an O-GalNAc glycoprotein that transits the Golgi en route to secretion, so this reads as an enzyme-substrate encounter rather than a binding activity of apoF, and the bare term carries no functional information either way. Reason: The partner UniProtKB:Q8N4A0 is GALNT4, polypeptide N-acetylgalactosaminyltransferase 4 (EC 2.4.1.41), a Golgi apparatus membrane enzyme that catalyses the initial transfer of GalNAc to serine or threonine in O-linked glycan biosynthesis. ApoF carries an experimentally mapped O-GalNAc glycan at Thr-274, and its secreted form is maturely glycosylated, so the two proteins are expected to meet in the Golgi lumen with apoF as substrate. A post-translational modification a protein receives is not a binding activity it has, and a substrate is not the agent. The reference is also a whole-proteome projection: counted by full pagination of QuickGO, PMID:28514442 supplies 3731 GO annotations across 2324 distinct gene products, all to GO:0005515. There has been no functional follow-up on an apoF-GALNT4 pair, so there is no informative molecular function to modify to; the row is left in place but flagged. Note that UniProt records NbExp=2 for this pair, which is the same interaction seen in two BioPlex releases, not two independent studies. Supporting Evidence: PMID:28514442 Here we present BioPlex 2.0 (Biophysical Interactions of ORFeome-derived complexes), which uses robust affinity purification-mass spectrometry methodology to elucidate protein interaction networks and co-complexes nucleated by more than 25% of protein-coding genes from the human genome PMID:19008531 Apo F in the media was PNGase F-sensitive but Endo H-resistant, indicating that the secreted forms of Apo F are maturely glycosylated |
| GO:0005515 protein binding | IPI PMID:33961781 Dual proteome-scale networks reveal cell-specific remodeling... | MARK AS OVER ANNOTATED | Summary: The same GALNT4 interaction, re-reported in BioPlex 3.0. Same reading as the BioPlex 2.0 row: a Golgi glycosyltransferase co-purifying with one of its class of secretory-pathway substrates, recorded as an uninformative bare binding term. Reason: This row and the PMID:28514442 row report the same partner, Q8N4A0 GALNT4, from successive releases of the same affinity-purification/mass-spectrometry pipeline; they are not two independent observations. PMID:33961781 is an even larger projection than its predecessor: full pagination of QuickGO gives 9514 annotations across 4757 distinct gene products, of which 9508 are GO:0005515. The biological reading is unchanged - apoF is an O-glycosylated secretory protein and GALNT4 is a Golgi enzyme that initiates O-GalNAc glycosylation, so co-purification is expected on substrate grounds and says nothing about a molecular function of apoF. Marked as over-annotated rather than removed, because the interaction is real and GALNT4 is a plausible agent of the Thr-274 modification. Supporting Evidence: PMID:33961781 Dual proteome-scale networks reveal cell-specific remodeling of the human interactome. PMID:19008531 Apo F in the media was PNGase F-sensitive but Endo H-resistant, indicating that the secreted forms of Apo F are maturely glycosylated |
| GO:0005576 extracellular region | EXP PMID:204339 Isolation and partial characterization of a new acidic apoli... | ACCEPT | Summary: The original isolation of apolipoprotein F, from the high density lipoproteins of human plasma. Extracellular localisation is directly and correctly supported. Reason: ApoF was first obtained as a plasma protein, purified from circulating human HDL; the protein's entire biology takes place in plasma. This reference is one of the three ECO:0000269 citations behind UniProt's SUBCELLULAR LOCATION Secreted line for Q13790. The cached record is abstract-less - this 1978 Biochemistry paper has no abstract in PubMed at all, so the title is the only text available and the quote below is taken from it; the claim it supports, that the protein was isolated from human plasma, is stated in the title itself. The more specific particle assignments this paper implies are captured by the new GO:0034364 row rather than by narrowing this one, since this row's claim - the protein is extracellular - is correct as it stands. Supporting Evidence: PMID:204339 Isolation and partial characterization of a new acidic apolipoprotein (apolipoprotein F) from high density lipoproteins of human plasma. |
| GO:0005576 extracellular region | EXP PMID:8093033 Purification and molecular cloning of human apolipoprotein F... | ACCEPT | Summary: Purification of apoF from human high density lipoproteins, with liver-restricted mRNA expression, establishing a liver-secreted plasma protein. Reason: Day et al. purified the protein from circulating human HDL and showed by Northern blot that its mRNA is confined to liver, which together describe a hepatocyte-synthesised protein exported to plasma. The same paper predicted the proteolytic maturation that the sequence analysis in this review confirms. Extracellular region is correct and directly evidenced. Supporting Evidence: PMID:8093033 Apolipoprotein F, with an apparent molecular mass of 29 kilodaltons, was purified from human high density lipoproteins using a modified two dimensional electrophoresis procedure. PMID:8093033 Northern blot analysis indicates that apolipoprotein F mRNA is detected only in liver for the tissues examined. |
| GO:0005576 extracellular region | IBA GO_REF:0000033 | ACCEPT | Summary: Correct and well-placed family inference. The committed PAINT slice for PTHR15011 shows a single node, PTN002687490 at taxon:117571 (Euteleostomi), whose GO:0005576 IBD is seeded by UniProtKB:Q13790 - the target's own experimentally grounded secretion. The node asserts that secretion is ancestral to the vertebrate apolipoprotein F family rather than a human innovation. Reason: The self-referential WITH/FROM is expected and is a marker of experimental grounding on the target, not circularity: human APOF's own secretion evidence is one of the descendant evidences the PAINT curator used to place the IBD, and the node then adds the claim that the property is inherited. Human APOF sits inside the inheriting clade by definition, and there is no evidence of divergence: the 35-residue signal peptide is present, the protein is recovered from plasma by several independent groups, and the mouse and rat orthologues that make up the rest of the family are likewise secreted plasma proteins. PTHR15011 is a single-subfamily family with no paralogue to confuse, and the slice contains no IRD or IKR anywhere. Secretion is not incidental for this protein - it is a plasma lipid-transfer regulator and has no known intracellular activity - so this is core rather than contextual. Propagation Review Root cause: NO FAILURE CORE Sources checked: PANTHER:PTN002687490 · PTHR15011 apolipoprotein F IBD node (taxon:117571, Euteleostomi) SUPPORTS TRANSFER The family's only PAINT node. The committed slice interpro/panther/PTHR15011/PTHR15011-paint.tsv shows its GO:0005576 IBD seeded by UniProtKB:Q13790, at Euteleostomi. Human APOF is inside the inheriting clade, and the family contains no IRD or IKR and no paralogous subfamily to confuse. UniProtKB:Q13790 · APOF (Homo sapiens), the target itself SUPPORTS TRANSFER The target's own accession, which is expected rather than circular: its experimental secretion evidence (UniProt SUBCELLULAR LOCATION Secreted, ECO:0000269 from PMID:204339, PMID:8093033 and PMID:28935895) is one of the descendant evidences behind the IBD, and its presence here marks experimental grounding on the target. Supporting Evidence: PMID:8093033 Apolipoprotein F, with an apparent molecular mass of 29 kilodaltons, was purified from human high density lipoproteins using a modified two dimensional electrophoresis procedure. file:human/APOF/APOF-bioinformatics/RESULTS.md signal + propeptide + chain = 326 = the full precursor. |
| GO:0005576 extracellular region | IDA PMID:28935895 Absolute quantitation of disease protein biomarkers in a sin... | ACCEPT | Summary: Direct mass-spectrometric detection and absolute quantitation of apoF in human serum and plasma, in the course of developing it as a NAFLD biomarker. Reason: The protein is measured directly in circulating human serum by parallel reaction monitoring, which is as direct a demonstration of extracellular localisation as this term requires. Incidentally, this paper also records that the tryptic peptide DANISQPETTK could not be used for quantitation because it was glycosylated; that peptide maps to precursor positions 265-275, inside the mature chain, and contains both the mature chain's only N-glycosylation sequon and its experimentally mapped O-GalNAc site. Supporting Evidence: PMID:28935895 Human serum and IGNIS prime peptides were digested and the IGNIS assay was used to quantify APO-F in clinical samples. PMID:28935895 We have previously used MRM to detect and quantify APO-F6 and found that the peptide DANISQPETTK was unsuitable since it was glycosylated and so was not included for analysis by PRM. |
| GO:0005576 extracellular region | IEA GO_REF:0000044 | ACCEPT | Summary: A vocabulary translation, not an inference: UniProt's subcellular location term SL-0243 Secreted maps to GO:0005576, and the SUBCELLULAR LOCATION line it comes from on Q13790 carries ECO:0000269 experimental evidence from three publications. Reason: Checked against the UniProt locations REST record, SL-0243 is "Secreted", defined as "Protein located outside the cell membrane(s)", and its declared GO mapping is exactly GO:0005576 extracellular region. The electronic step here therefore introduces no new biological claim; it restates, in GO, a UniProt annotation that is itself experimentally supported by PMID:204339, PMID:8093033 and PMID:28935895 - the same three papers that appear as EXP and IDA rows in this record. Correct and consistent with the rest of the evidence. Propagation Review Root cause: NO FAILURE CORE Sources checked: UniProtKB-SubCell:SL-0243 · Secreted (UniProt subcellular location vocabulary) SUPPORTS TRANSFER The UniProt locations REST record for SL-0243 gives the name 'Secreted' and declares its GO mapping as GO:0005576, so this is a controlled-vocabulary translation rather than an inference, and the UniProt line it translates is itself ECO:0000269-supported. Supporting Evidence: PMID:28935895 Human serum and IGNIS prime peptides were digested and the IGNIS assay was used to quantify APO-F in clinical samples. |
| GO:0006629 lipid metabolic process | TAS PMID:8093033 Purification and molecular cloning of human apolipoprotein F... | KEEP AS NON CORE | Summary: True but uninformative. Lipid metabolic process is a high-level parent that is correct for any apolipoprotein; the specific process this protein participates in is captured by the cholesterol metabolic process rows and by the regulation term proposed on the GO:0006869 row. Reason: There is nothing wrong with this annotation - apoF is unambiguously a participant in lipid metabolism - but it sits far above the level at which this protein's biology is understood, and the paper it is attributed to is a purification and cloning study whose lipid content is the observation that the protein was obtained from HDL. Retained for completeness and marked non-core, since the core statements about this gene are the LDL-selective inhibition of CETP-mediated transfer and the cholesterol metabolism it controls, both of which are recorded on other rows at a useful granularity. Supporting Evidence: PMID:8093033 Apolipoprotein F, with an apparent molecular mass of 29 kilodaltons, was purified from human high density lipoproteins using a modified two dimensional electrophoresis procedure. |
| GO:0006869 lipid transport | TAS PMID:6816269 Isolation and characterization of simple and complex lipopro... | MODIFY | Summary: Apolipoprotein F is a regulator of inter-lipoprotein lipid movement, not an agent of it. It inhibits the transfer protein that performs the movement, so the process term should be a negative regulation term. This row is where the protein's defining activity belongs, and at present it says the opposite. Reason: GO:0006869 asserts that the gene product effects the directed movement of lipids. What apoF does is reduce that movement: it is the lipid transfer inhibitor protein, it equally inhibits CE and TG transfer by CETP, and in vivo knockdown in a CETP-expressing species increases CETP-mediated cholesteryl ester transfer from HDL to LDL up to two-fold and raises LDL cholesterol by 40%. GO already carries the mirror image of this annotation on the target enzyme: CETP holds GO:0032376 positive regulation of cholesterol transport by IDA. The matching negative term, GO:0032375 negative regulation of cholesterol transport, is the correct replacement. Its parent GO:0032369 negative regulation of lipid transport is also true, because the inhibition extends to triglyceride and phospholipid transfer, but the cholesterol child is both more informative and the better evidenced. One caveat is recorded rather than hidden: the inhibition is not uniform in sign across all lipoprotein pairs - transfer from VLDL to HDL3 is stimulated - so the protein is better described as retargeting CETP than as switching it off. The net effect on cholesterol movement in plasma and in vivo is nonetheless inhibitory, which is what the replacement term asserts. Proposed replacements: negative regulation of cholesterol transport Supporting Evidence: PMID:32520778 We identified a novel, acidic protein called lipid transfer inhibitor protein (LTIP) (9) that equally inhibits the transfer of both CE and TG by CETP. PMID:31511396 ApoF knockdown in fat-fed male hamsters created a phenotype in which endogenous CETP-mediated CE transfer from HDL to LDL increased up to 2-fold, LDL cholesterol increased 40%, HDL declined 25%, LDL and HDL lipid compositions were altered, and hepatic LDLR gene expression was decreased. PMID:12907677 In contrast, VLDL to HDL3 transfer was stimulated, resulting in a CETP preference for HDL3 that was 3-fold greater than that for LDL or HDL2. |
| GO:0008203 cholesterol metabolic process | IBA GO_REF:0000033 | ACCEPT | Summary: Correct inference at the right level of generality. The IBD at PTN002687490 (taxon:117571, Euteleostomi) is seeded by mouse Apof, whose own GO:0008203 is an IMP from the knockout: female null mice accumulate hepatic cholesteryl ester. Human APOF sits inside the clade and has independent, direct evidence for a role in cholesterol metabolism. Reason: There is one gene donor on this row, MGI:MGI:104539, which resolves through the UniProt cross-reference search to Swiss-Prot Q91V80 APOF_MOUSE (two hits; the second, A0A0R4J0M4, is TrEMBL), so the singular "the IBD seed" is accurate here. Two features of the transfer are worth stating explicitly rather than objecting to. First, the seed phenotype is CETP-independent, because mice carry no CETP gene at all, and sex-specific; the human function is CETP-dependent. Second, the node spans Euteleostomi, which contains both CETP-less rodents and CETP-bearing fish and primates, so a broad cholesterol-metabolism term is the correct least-common-denominator call across the clade rather than an under-call. This is not a granularity mismatch: with a single gene donor there is no donor agreement to appeal to, and no more specific term is true of the whole clade. Independently of the mouse, human apoF controls the CETP-mediated redistribution of cholesteryl ester among plasma lipoproteins, which is cholesterol metabolism by any reading of the term definition. Core. Propagation Review Root cause: NO FAILURE CORE Sources checked: MGI:MGI:104539 · Apof (Mus musculus) SUPPORTS TRANSFER Resolves to Swiss-Prot Q91V80 APOF_MOUSE by UniProt xref search (two hits; the other, A0A0R4J0M4, is TrEMBL). QuickGO shows its GO:0008203 is IMP from PMID:22363685 with allele MGI:3800154 - female knockouts accumulate hepatic cholesteryl ester. Mice have no CETP, so this seed is a CETP-independent and sex-specific phenotype, and the line is separately known to be Stat2-hypomorphic (PMID:24529150); the human target nonetheless has independent direct evidence for the same broad process. PANTHER:PTN002687490 · PTHR15011 apolipoprotein F IBD node (taxon:117571, Euteleostomi) SUPPORTS TRANSFER One gene donor stands behind this node, so the singular 'the IBD seed' is accurate here. Euteleostomi spans both CETP-less rodents and CETP-bearing fish and primates, so a broad cholesterol-metabolism term is the correct least-common-denominator call for the clade rather than an under-call. Supporting Evidence: PMID:22363685 Female ApoF KO mice had increased liver cholesteryl ester content relative to wild type controls on a chow diet PMID:22363685 This lack of a consistent requirement for plasma CE transfer activity across species suggests that ApoF has been conserved throughout evolution for a purpose distinct from CETP inhibition. PMID:32520778 ApoF preferentially inhibits CETP-mediated lipid transfer specifically involving LDL. |
| GO:0008203 cholesterol metabolic process | TAS PMID:6816269 Isolation and characterization of simple and complex lipopro... | ACCEPT | Summary: The cited statement is a 1982 suggestion about the apoF-containing particles rather than a demonstration about the protein, but the conclusion it reaches is right and is now supported by much stronger evidence. Accepted as core. Reason: Koren et al. isolated three apoF-containing lipoprotein species and found cholesteryl ester and free cholesterol to be their predominant lipids, and closed with a suggestion that such particles may function in cholesterol transport or esterification. On its own that is a statement-level inference about particles. It is nevertheless the correct conclusion: apoF is now known to set how much HDL-derived cholesteryl ester reaches LDL, in vitro, in human plasma and in a CETP-expressing animal model. Cholesterol metabolic process is the right process term and this is a core role of the gene, so the row is accepted, with the weakness of its own evidence recorded here rather than used as grounds to remove a correct annotation. Supporting Evidence: PMID:6816269 It is suggested that ApoF-containing lipoproteins may be involved in transport and/or esterification of cholesterol. PMID:32520778 Therefore, ApoF tailors CETP activity so that less tissue-derived cholesterol traffics from HDL into the LDL compartment. |
| GO:0015485 cholesterol binding | TAS PMID:6816269 Isolation and characterization of simple and complex lipopro... | REMOVE | Summary: No assay of apolipoprotein F binding cholesterol has ever been performed. The cited paper measured the lipid composition of apoF-containing particles by thin-layer chromatography, which reports what the particle contains, not what the protein binds. Reason: Three independent arguments converge. First, the evidence: Koren et al. determined particle lipid composition by TLC, finding cholesteryl ester and free cholesterol to predominate; no binding measurement of any kind appears in the paper, and the annotation converts a property of the particle into a property of one of its passengers. Second, what has actually been measured about apoF binding points elsewhere - the determinant of apoF association with LDL is particle size and surface lipid packing, not cholesterol content, with LDL size the only correlate of binding capacity found in a systematic survey. Third, the protein lacks the structural feature that mediates lipid binding in the apolipoproteins that do have it: apoF is predicted to lack strong amphipathic alpha helices, and the sequence analysis in this review quantifies that, giving 7.6% strongly amphipathic 18-residue windows for mature apoF against 21.3-49.2% for APOA1, APOA2, APOE and APOC3. For contrast, GO:0015485 is carried correctly elsewhere in this same pathway: CETP holds it by IDA from a direct assay. The molecular binding activity that apoF does have is to the LDL particle, which the new GO:0030169 row records. Supporting Evidence: PMID:35016907 Only LDL particle size correlated with ApoF binding capacity. PMID:22363685 In contrast to other apolipoproteins, ApoF is predicted to lack strong amphipathic alpha helices and its true physiological function remains unknown. file:human/APOF/APOF-bioinformatics/RESULTS.md ApoF has the lowest mean moment and a 3- to 6-fold lower proportion of strongly amphipathic windows than every control. |
| GO:0030169 low-density lipoprotein particle binding | IDA PMID:35016907 Apolipoprotein F concentration, activity, and the properties... | NEW | Summary: The molecular function this protein actually has, missing from the record entirely. Apolipoprotein F binds the LDL particle, and that binding is the prerequisite for its inhibition of CETP; the amount bound is set by LDL particle size and is measurable by ELISA in human plasma. Reason: Direct evidence in human plasma and with human protein, from several groups and over two decades. The original identification of LTIP as apoF exploited the tight association of the protein with LDL as a purification step and reported the protein to be associated almost exclusively with LDL, matching the distribution of the activity. Morton and Mihna quantified LDL-bound apoF by ELISA in 200 donors, showed that hypercholesterolaemic LDL carries about twice as much apoF per particle as normolipidaemic LDL while hypertriglyceridaemic LDL carries under 20% of control, and identified LDL particle size as the determinant: enlarging LDL in vitro increased apoF binding capacity and shrinking it again with lipoprotein lipase reduced binding by 90%. Earlier work showed binding can be separated from activity, since oleate blocks activity without reducing binding, which is further evidence that binding is a distinct molecular event rather than a restatement of the inhibition. This is the replacement term proposed on the GO:0005319 and GO:0015485 rows and is added here as a positive annotation in its own right. Supporting Evidence: PMID:35016907 Apolipoprotein F (ApoF) modulates lipoprotein metabolism by selectively inhibiting cholesteryl ester transfer protein activity on LDL. This ApoF activity requires that it is bound to LDL. PMID:35016907 Similarly, when in vitro produced and enlarged LDLs with elevated ApoF binding capacity were incubated with LPL to reduce their size, ApoF binding was reduced by 90%. PMID:9880564 Purification of LTIP, in contrast to other published protocols, took advantage of the tight association of this protein with LDL. PMID:12951364 LTIP binding to LDL was not decreased by oleate. |
| GO:0034362 low-density lipoprotein particle | IDA PMID:9880564 Molecular cloning and expression of lipid transfer inhibitor... | NEW | Summary: The active pool of apolipoprotein F is a component of LDL. This is where the protein exercises its function, and no current row records it; the five GO:0005576 rows only state that the protein is outside the cell. Reason: Wang et al. purified LTIP from ultracentrifugally isolated LDL that had been further cleared of apoE and apoA-I on affinity columns, sequenced it, and found it to be apoF; they reported the protein associated almost exclusively with LDL, matching the distribution of the inhibitory activity. Later work by the same group resolved the apparent conflict with the earlier HDL literature into two pools and established that only the LDL-associated pool is active. Gel filtration independently localises one of the two apoF peaks to LDL. The term is a descendant of GO:0005576 and of GO:0032991, and records component membership of the particle rather than a location in the loose sense. Supporting Evidence: PMID:9880564 In contrast to that previously reported, apoF was shown to be associated almost exclusively with LDL, identical to the distribution of LTIP activity. PMID:18369235 Plasma fractionated by gel filtration chromatography revealed two LTIP protein peaks, one coeluting with LDL, and another of approximately 470 kDa. PMID:32520778 Most importantly, the ApoF associated with LDL is active in regulating CETP, whereas ApoF in the 470kDa complex is inactive (38). |
| GO:0034364 high-density lipoprotein particle | IDA PMID:18369235 Control of cholesteryl ester transfer protein activity by se... | NEW | Summary: Most of the apolipoprotein F in normolipidaemic plasma is not on LDL at all but in a roughly 470 kDa, apolipoprotein A-I-containing particle of HDL density in which it is inactive. This is the pool the 1978, 1982, 1989 and 1994 isolations were working with, and it is not recorded anywhere in the current annotation. Reason: The HDL assignment is not an obsolete claim superseded by the LDL one; the two coexist and differ in activity. He et al. resolved the second apoF peak as a particle of density 1.134 g/ml containing apolipoprotein A-I along with apoC-II, apoD, apoE, apoJ and paraoxonase 1, which places it squarely inside the GO:0034364 definition (density 1.063-1.21 g/ml, contains APOAs, may contain APOCs and APOE), and showed that it does not inhibit CETP. Independent antibody-based fractionation of human lipoproteins put more than 90% of apoF on HDL3, and the earliest activity measurements put around 85% of functional LTIP in the HDL density range. The review by the group that did most of this work states that more than 85% of plasma apoF is in the inactive complex. Annotating only the LDL particle would record the minority pool and omit the majority one. The functional asymmetry between the two pools is stated here and in the GO:0034362 row rather than being resolved by choosing one. Supporting Evidence: PMID:18369235 The 470 kDa LTIP complex had a density of 1.134 g/ml, indicating approximately 50% lipid content, and contained apolipoprotein A-I. PMID:18369235 Unlike LDL-associated LTIP, the 470 kDa LTIP complex does not inhibit CETP activity. PMID:19008531 greater than 90% of the ApoF in human plasma was found on HDL(3), with only a small amount on LDL. PMID:2715721 Assay of ultracentrifugally obtained lipoprotein fractions revealed that approximately 85% of the total functional LTIP activity was in the d 1.063-1.21 g/ml HDL fraction. |
Loading supporting content…
Download this section (compressed HTML)Q: Does apolipoprotein F ever contact CETP directly, even transiently, or is the inhibition entirely substrate-directed? Every published statement of the mechanism is framed as an effect on the lipoprotein surface, and the answer decides whether any existing GO molecular function regulator term can be applied to this protein.
Suggested experts: Richard E. Morton, Yan Liu
Q: What holds apolipoprotein F inactive in the 470 kDa apolipoprotein A-I-containing particle? Is it steric occlusion of the same surface that engages LDL, or a conformational state of the protein?
Suggested experts: Richard E. Morton, Diane J. Greene
Q: What is the function of apolipoprotein F in mice and rats, which have no CETP? The phylogenetic inference behind both of this gene's IBA annotations is seeded at a node spanning bony vertebrates, most of which cannot be using the protein as a CETP inhibitor.
Suggested experts: William R. Lagor, Daniel J. Rader
Q: Is the Asn-267 glycan of the mature protein required for LDL binding or for partition between the active and inactive pools? The mutant was made and shown to lose the upper band of the secreted doublet, but its lipoprotein behaviour was not reported.
Suggested experts: William R. Lagor
Experiment: Measure CETP association with LDL directly, by surface plasmon resonance or by ultracentrifugal re-isolation of labelled CETP, in the presence and absence of saturating apolipoprotein F on the same LDL preparation, under conditions where lipid transfer is abolished. In parallel, test for a direct apoF-CETP interaction with a method sensitive to transient complexes, such as crosslinking mass spectrometry or microscale thermophoresis with purified components in the absence of lipoprotein. A reduction in CETP-LDL association without any detectable apoF-CETP complex would establish the substrate-directed mechanism and justify the new molecular function term proposed here.
Hypothesis: Apolipoprotein F inhibits CETP without binding it, by altering the LDL surface that CETP must engage.
Type: binding and transfer assay with purified components
Experiment: Build chimaeras between human apolipoprotein F and the mouse orthologue, which is about 60% identical over the precursor and functions in an animal with no CETP, and between human apolipoprotein F and progressive N- and C-terminal truncations of the mature chain. Express each in a mammalian secretory line, recover the conditioned medium, and assay LDL binding by ELISA and CETP inhibition in the LDL-only transfer assay. Constructs that bind LDL but do not inhibit would separate the two activities and locate the inhibitory determinant.
Hypothesis: The LDL-binding determinant of apolipoprotein F lies in a discrete region of the mature chain and is separable from whatever confers inhibition.
Type: chimaera and truncation mutagenesis with functional assay
Experiment: Reconstitute the inactive complex from defined components - apolipoprotein A-I, phospholipid, cholesterol and recombinant apolipoprotein F - and test each component for its ability to suppress CETP inhibition. Compare the hydrogen-deuterium exchange profile of apolipoprotein F in the reconstituted complex, free, and bound to LDL. Regions protected in both the complex and on LDL would indicate a shared binding surface and competition; regions that change only in the complex would indicate a conformational mechanism.
Hypothesis: The 470 kDa particle inactivates apolipoprotein F by occluding its LDL-binding surface rather than by changing its conformation.
Type: reconstitution with hydrogen-deuterium exchange mass spectrometry
Experiment: Generate an Apof null allele by CRISPR that leaves the Stat2 promoter region intact, since the existing line is confounded by hypomorphic Stat2 expression from a gene 425 base pairs downstream, and characterise it against the existing line. Profile the hepatic and plasma lipidome in both sexes, measure hepatic cholesteryl ester directly, and determine what apolipoprotein F associates with in CETP-free mouse plasma by immunoaffinity capture followed by mass spectrometry of the co-isolated particle.
Hypothesis: Apolipoprotein F has a CETP-independent function that a clean mouse knockout would reveal.
Type: mouse genetics with lipidomics and particle proteomics
What is not known — curated, literature-grounded statements of the open unknowns (the inverse of core functions).
Gap: How apolipoprotein F, once bound to LDL, prevents CETP from acting on that particle is unknown at the molecular level. There is no structure of apolipoprotein F, no mapped LDL-binding surface, no mapped inhibitory determinant, and no demonstration that the two proteins ever touch.
OPEN BIOLOGYONTOLOGY MF_DARK
What is known: Firmly established: the protein binds LDL; binding is required for inhibition; binding capacity tracks LDL particle size and surface lipid packing and not cholesterol content; inhibition is selective among lipoprotein pairs; and the inhibitor does not change CETP's choice of lipid or its exchange mechanism. Equally firmly established: binding and inhibition are separable, since oleate abolishes activity without reducing binding. What is missing is everything between binding and inhibition.
Significance: This is the mechanism that makes the protein interesting as a therapeutic model. Global pharmacological CETP inhibition has repeatedly failed in clinical trials, whereas this protein inhibits CETP only where LDL is involved, which is the selectivity those drugs lacked. Reproducing that selectivity requires knowing how it is achieved. It is also the reason no molecular function term fits: the inhibition is directed at the substrate particle, and GO's regulator terms all require the regulator to bind the target.
What would resolve it: A structure of apolipoprotein F, alone or on a reconstituted LDL-like particle, and a direct test for a CETP-apoF interaction by a method that would detect a transient one. Failing those, mapping the LDL-binding region by truncation or chimaera with the mouse orthologue, which is only about 60% identical, and asking whether CETP binding to LDL is reduced in the presence of apoF under conditions where transfer is abolished.
Provenance (the field's own admissions):
Gap: What apolipoprotein F does in species that have no CETP is unknown, and since the family is conserved across bony vertebrates while CETP is absent from mice and rats and defective in several other mammals, there is probably a second function that no one has identified.
OPEN BIOLOGY MF_DARK
What is known: Firmly established: apolipoprotein F orthologues are present in mammals and fish, including rodents with no CETP gene; mouse knockouts are viable and fertile with essentially normal plasma lipids, HDL size and HDL composition; the one reproducible lipid phenotype is a female-specific increase in hepatic cholesteryl ester; and the mouse orthologue's molecular function is formally recorded in GO as unknown (GO:0003674 ND). What is missing is any positive account of the CETP-independent activity.
Significance: The entire GO record for this gene, including the phylogenetic inference that spans Euteleostomi, rests on a CETP-dependent function in the species that have CETP and on a single sex-specific hepatic lipid phenotype in a species that does not. If the ancestral function is something else, the family-level inference is describing a consequence rather than the conserved activity.
What would resolve it: Characterisation of the mouse or zebrafish protein on its own terms rather than as a CETP inhibitor: what it binds in a CETP-free plasma, and what changes in the knockout beyond hepatic cholesteryl ester. The existing knockout line is not ideal for this, since its non-lipid phenotypes were traced to hypomorphic expression of Stat2, a gene 425 base pairs downstream, so an independently derived allele would be needed.
Provenance (the field's own admissions):
Gap: Why most of the plasma pool sits inactive in a 470 kDa apolipoprotein A-I-containing particle, what holds it there, and whether that pool has a function of its own rather than being a reservoir, are all unresolved.
OPEN BIOLOGY CC_DARK
What is known: Firmly established: two pools exist; the LDL-bound one is active and the 470 kDa one is not; more than 85% of the protein is in the inactive pool in normolipidaemic plasma and about 50% in hypercholesterolaemia; redistribution between pools happens on incubation of plasma at 37 degrees and requires LCAT or CETP activity; and the trigger is modification of LDL, not release from the complex. What is missing is the composition and organisation of the complex at a level that would explain why the protein is inactive in it.
Significance: Activity of this protein is regulated by partition between two particles rather than by abundance, so plasma concentration is a poor guide to function; this is directly relevant to its use as a biomarker, where total protein is what gets measured. It also means the two cellular component annotations added in this review describe states with opposite functional consequences.
What would resolve it: Reconstitution: rebuild the 470 kDa particle from defined components and ask which of them suppresses activity, and determine whether the suppression is steric occlusion of the CETP-relevant surface or a conformational change in apolipoprotein F.
Provenance (the field's own admissions):
Gap: Whether apolipoprotein F associates with VLDL at all is unsettled, and the disagreement is between a direct negative result and a secondary report of a weak positive.
OPEN CURATION CC_DARK
What is known: Firmly established: the protein is on LDL and on an HDL-density particle. Direct antibody fractionation of human lipoproteins detected none on VLDL. A later lipoproteomic survey, cited only at second hand in the 2020 review, reports VLDL-associated apolipoprotein F at 5-fold lower abundance than on LDL. UniProt's FUNCTION line states that the protein also associates to a lesser degree with VLDL, citing PMID:9880564, whose abstract does not mention VLDL.
Significance: VLDL is the acceptor to which apolipoprotein F redirects HDL-derived cholesteryl ester, so whether the protein is physically present on that acceptor bears on whether it acts only from the LDL side. No GO:0034361 very-low-density lipoprotein particle annotation is proposed in this review for exactly this reason.
What would resolve it: A direct, quantitative measurement of apolipoprotein F on VLDL in human plasma by the same ELISA used for the LDL-bound pool, reported alongside the LDL and HDL pools rather than inferred from proteomic detection.
Provenance (the field's own admissions):
Gap: The direction in which apolipoprotein F affects reverse cholesterol transport is not a fixed property of the protein: knockdown impairs it on one diet and enhances it on another, and no account of the switch exists.
OPEN BIOLOGYCURATION BP_DARK
What is known: Firmly established: in fat-fed hamsters, knockdown impairs the excretion of HDL-derived cholesteryl ester into faeces; in chow-fed hamsters, the same knockdown increases sterol excretion. Both results are from the same study and the same genetic background.
Significance: This is why no reverse cholesterol transport term appears in this review. A process whose sign inverts with diet cannot be annotated with a positive- or negative-regulation term, and the unsigned parent would assert only that the protein is somehow involved. It also bears directly on whether raising apolipoprotein F would be beneficial, which is the translational question the field is asking.
What would resolve it: Identify the diet-dependent variable. The candidates are LDL abundance, LDL composition and hence the active fraction of the protein, and hepatic LDL receptor expression, all of which change with fat feeding; a knockdown series across a graded fat load, with the LDL-bound fraction measured in each arm, would separate them.
Provenance (the field's own admissions):
Gap: The post-translational maturation of this precursor is only partly captured and its functional consequences are untested: the N-glycosylation sites of the mature circulating protein are absent from UniProt, the one site UniProt does annotate lies inside the propeptide and cannot be on the mature protein, and nothing is known about what either the propeptide cleavage or the glycans contribute beyond allowing secretion.
OPEN CURATIONBIOLOGY MF_DARK
What is known: Firmly established: the precursor contains exactly three N-X-S/T sequons, at Asn-118, Asn-139 and Asn-267; the first two are inside the propeptide and the third is the only one in the mature chain; mutagenesis shows the proprotein is glycosylated at Asn-118 and Asn-139 and the mature form variably at Asn-267; mutating Arg-164 abolishes maturation, and the proprotein convertase PC7 processes apolipoprotein F; the mature chain also carries an O-GalNAc glycan at Thr-274; and glycosylation accounts for the gap between the 17.4 kDa polypeptide mass and the 29-33 kDa apparent mass. UniProt Q13790 annotates only Asn-118, by sequence-analysis prediction, and Thr-274.
Significance: The glycans are most of the mass of the mature protein and are responsible for its strongly acidic character, which is the property that named it. Whether they participate in LDL binding or in partition between the active and inactive pools is unknown, and the under-annotation means the question is invisible to anyone reading the feature table.
What would resolve it: Curation: add Asn-267 and Asn-139 to the Q13790 feature table with the mutagenesis evidence, and mark Asn-118 as propeptide-resident. Biology: compare LDL binding and CETP inhibition of the Asn-267 mutant, which is already constructed, against wild type.
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)