AP4S1 encodes sigma4-adaptin, the single small subunit of adaptor protein complex 4 (AP-4), an obligate heterotetramer that also contains the large adaptins epsilon (AP4E1) and beta4 (AP4B1) and the medium adaptin mu4 (AP4M1). AP-4 assembles a non-clathrin coat on the cytosolic face of the trans-Golgi network, where the small GTPase ARF1 and membrane cargo recruit it cooperatively, and packages a defined set of transmembrane proteins into vesicles that leave the TGN for peripheral and axonal compartments. Its best-characterised client is ATG9A, the lipid scramblase required for autophagosome biogenesis; the amyloid precursor protein, SERINC1 and SERINC3, sortilin, and AMPA and delta-2 glutamate receptors have also been reported as AP-4-dependent, and the accessory proteins tepsin, RUSC1 and RUSC2 link AP-4 vesicles to microtubule motors. Within the complex sigma4 is the architectural subunit: the epsilon trunk wraps around it to build the central core, the two large adaptins cannot associate in its absence, and loss of sigma4 destabilises all four subunits, abolishes complex assembly and abolishes tepsin recruitment to membranes. Unlike the sigma subunits of AP-1, AP-2 and AP-3, sigma4 has no demonstrated role in recognising dileucine sorting signals; in AP-4 the tyrosine-based and non-canonical cargo signals are read by mu4. Bi-allelic loss-of-function variants in AP4S1 cause spastic paraplegia 52, one of four genetically distinct but clinically indistinguishable forms of AP-4 deficiency syndrome, a childhood-onset complex hereditary spastic paraplegia with intellectual disability, progressive lower-limb spasticity and developmental brain malformations. Heterozygous carriers are unaffected.
| GO Term | Evidence | Action | Reason |
|---|---|---|---|
| GO:0005737 cytoplasm | IEA GO_REF:0000117 | ACCEPT | Summary: Correct but maximally coarse, and already entailed by the gene's own IDA complex row: QuickGO's ancestor closure places GO:0005737 above GO:0030124 AP-4 adaptor complex. AP4S1 is a cytosolic subunit of a peripheral membrane coat, present in both a soluble and a membrane-bound pool. Reason: Kept rather than trimmed for two reasons. First it is simply true: UniProt records AP-4 as a peripheral membrane protein of the TGN, and the 1999 characterisation found the complex in both soluble and membrane-associated forms, so a cytoplasmic (non-luminal, non-nuclear) assignment is right. Second it is the same topological fact that makes the endosome lumen row on this gene wrong, so removing it while removing that one would be inconsistent. That second point is what discriminates this row from the other ARBA row on this gene, which is MODIFIED rather than accepted: GO:0012505 is coarse and strictly subsumed by terms the gene already holds, adding nothing, whereas GO:0005737 is coarse but carries independently useful content - it is the topological fact the endosome-lumen removal rests on. Coarse-and-informative is kept; coarse-and-subsumed is sharpened. The annotation does carry a provenance weakness recorded in the propagation review: the ARBA rule's own published condition sets do not fire on AP4S1's three real InterPro matches. That is a reason to distrust the derivation, not the term. Propagation Review Root cause: NO FAILURE NON CORE Sources checked: ARBA:ARBA00026971 · UniProt ARBA rule ARBA00026971 (cytoplasm) SOURCE WEAK OR INFERRED Fetched from https://rest.uniprot.org/arba/ARBA00026971 and intersected against AP4S1's real InterPro matches. The rule publishes 2388 condition sets; exactly one (a set requiring IPR016635 AND IPR022775 AND IPR027156) names any AP4S1 signature, and AP4S1 does not match IPR027156 - checked live against the InterPro API, which returns exactly three entries for Q9Y587: IPR011012, IPR016635 and IPR022775. So this annotation cannot be reproduced from its own rule's published conditions. The term is nevertheless correct, so this is a provenance weakness, not a wrong call. Supporting Evidence: file:human/AP4S1/AP4S1-uniprot.txt SUBCELLULAR LOCATION: Golgi apparatus, trans-Golgi network membrane file:human/AP4S1/AP4S1-uniprot.txt Peripheral membrane protein PMID:10066790 An antibody to beta4 recognized in human cells an approximately 83-kDa polypeptide that exists in both soluble and membrane-associated forms. |
| GO:0005794 Golgi apparatus | IEA GO_REF:0000044 | ACCEPT | Summary: The Golgi apparatus is where AP-4 works. This row is the coarse-grained form of the gene's better-resolved trans-Golgi network annotations, transferred mechanically from the UniProt subcellular-location line. Reason: The compartment is right and is the core location of this protein, so no correction is needed; the TGN refinement is carried separately by the ComplexPortal NAS row and by the two Reactome TAS rows, and nothing is lost by keeping the parent as well. The underlying curator statement is an ECO:0000305 inference from the original immunofluorescence and immunogold work, which placed AP-4 at the TGN and on non-clathrin-coated vesicles there. Propagation Review Root cause: NO FAILURE CORE Sources checked: UniProtKB-SubCell:SL-0132 · UniProt subcellular location SL-0132 (Golgi apparatus) SUPPORTS TRANSFER A faithful, if coarse, restatement of the entry's own curated location line, which reads 'SUBCELLULAR LOCATION: Golgi apparatus, trans-Golgi network membrane' with ECO:0000305 from PubMed:10436028. The mapping drops the TGN refinement, which the gene carries separately by NAS and TAS. Supporting Evidence: file:human/AP4S1/AP4S1-uniprot.txt SUBCELLULAR LOCATION: Golgi apparatus, trans-Golgi network membrane PMID:10066790 Immunofluorescence analyses showed that AP-4 is associated with the trans-Golgi network or an adjacent structure and that this association is sensitive to the drug brefeldin A. PMID:10436028 Immunogold electron microscopy indicates that AP-4 is associated with nonclathrin-coated vesicles in the region of the trans-Golgi network. |
| GO:0005802 trans-Golgi network | NAS PMID:10436028 Characterization of a fourth adaptor-related protein complex... | ACCEPT | Summary: The trans-Golgi network is the compartment at which AP-4 assembles, established in the paper this NAS cites by immunofluorescence, brefeldin A sensitivity and immunogold electron microscopy. Reason: A ComplexPortal author-statement row whose source is the primary characterisation of AP-4, and whose claim that paper makes in its own abstract. This is the gene's core location and the compartment from which its cargo leaves. The NAS evidence code understates how well supported it is, but the term is exactly right and there is no more specific compartment to move to - GO:0032588 trans-Golgi network membrane is already separately annotated by TAS. Supporting Evidence: PMID:10436028 Immunogold electron microscopy indicates that AP-4 is associated with nonclathrin-coated vesicles in the region of the trans-Golgi network. file:human/AP4S1/AP4S1-uniprot.txt SUBCELLULAR LOCATION: Golgi apparatus, trans-Golgi network membrane |
| GO:0006605 protein targeting | IC PMID:10066790 AP-4, a novel protein complex related to clathrin adaptors. | ACCEPT | Summary: A curator inference from AP-4 complex membership, and a sound one: AP-4 selects transmembrane proteins bearing sorting signals and directs them out of the trans-Golgi network to a different compartment, which is what protein targeting means. Reason: GO:0006605 is defined as targeting proteins to particular regions of the cell, usually via an organelle-specific sequence motif, and that is a literal description of what AP-4 does - the motif being the tyrosine-based signal read by AP4M1. AP4S1's participation is real rather than nominal: the complex is obligate, and without sigma4 it does not assemble at all, so the targeting activity is lost. The inference chain runs from a measured fact on this protein (the IDA complex row from the same paper) rather than from a homology guess. Propagation Review Root cause: NO FAILURE CORE Sources checked: GO:0030124 · AP-4 adaptor complex (the cellular-component term the curator reasoned from) SUPPORTS TRANSFER An IC's source is a GO term on the same gene product, not a donor protein. Here that term is GO:0030124, which AP4S1 holds by IDA from the same paper, so the inference chain is complex membership (measured) to participation in what the complex does (inferred). Querying QuickGO by reference=PMID:10066790 returns 13 annotations over exactly 4 entities - the four AP-4 subunits - with GO:0030124 four times and GO:0006605 and GO:0008104 three times each. That is one complex characterisation projected across its subunits, which is legitimate here because the paper identified all four subunits of the complex, but it means the three rows are one observation rather than three. Supporting Evidence: PMID:10436028 The mu4 subunit of the complex specifically interacts with a tyrosine-based sorting signal, indicating that, like the other three AP complexes, AP-4 is involved in the recognition and sorting of cargo proteins with tyrosine-based motifs. PMID:25552650 We show that the premature stop mutations in AP4S1 result in a reduction of all AP-4 subunits and loss of AP-4 complex assembly. |
| GO:0008104 intracellular protein localization | IC PMID:10066790 AP-4, a novel protein complex related to clathrin adaptors. | ACCEPT | Summary: The same curator inference from complex membership, one step more general than the protein targeting row. AP-4 determines where ATG9A, APP and other transmembrane cargoes end up in the cell. Reason: GO:0008104 is 'Any process in which a protein is transported to, or maintained in, a specific location', which AP-4 does by definition and which patient-cell work has since confirmed directly for this gene: bi-allelic AP4S1 loss leaves ATG9A stuck in the trans-Golgi network and absent from the periphery. A parent term alongside a child is not a defect, and this one is independently true, so it is kept rather than merged into the more specific row proposed below. ACCEPT rather than KEEP_AS_NON_CORE despite being the most generic biological-process term on the gene: this is the core function stated coarsely, not a peripheral process. KEEP_AS_NON_CORE is for real-but-secondary biology - developmental or tissue-specific roles of a pleiotropic gene - and using it for a generic parent of the core function would misreport what kind of annotation this is. The nesting is instead made explicit here and in the IC propagation block, which records that the three rows from PMID:10066790 are one observation projected across the complex's subunits. Propagation Review Root cause: NO FAILURE CORE Sources checked: GO:0030124 · AP-4 adaptor complex (the cellular-component term the curator reasoned from) SUPPORTS TRANSFER An IC's source is a GO term on the same gene product, not a donor protein. Here that term is GO:0030124, which AP4S1 holds by IDA from the same paper, so the inference chain is complex membership (measured) to participation in what the complex does (inferred). Querying QuickGO by reference=PMID:10066790 returns 13 annotations over exactly 4 entities - the four AP-4 subunits - with GO:0030124 four times and GO:0006605 and GO:0008104 three times each. That is one complex characterisation projected across its subunits, which is legitimate here because the paper identified all four subunits of the complex, but it means the three rows are one observation rather than three. Supporting Evidence: PMID:31915823 We conclude that bi-allelic loss-of-function variants in any of the AP-4 subunit genes lead to loss of AP-4 function and subsequently to accumulation of ATG9A in the TGN area and depletion from peripheral compartments. PMID:31915823 we generated fibroblast lines from 15 well-characterized patients carrying homozygous or compound-heterozygous variants in AP4B1 (7 lines), AP4M1 (3 lines), AP4E1 (1 line) and AP4S1 (4 lines) |
| GO:0012505 endomembrane system | IEA GO_REF:0000117 | MODIFY | Summary: True but the least informative available statement of a location this gene already carries precisely. AP4S1 sits on the cytosolic face of the trans-Golgi network membrane; 'endomembrane system' is the whole collection of transport membranes in the cell. Reason: This is the textbook too-general case. The gene already holds GO:0005802 trans-Golgi network by NAS and GO:0032588 trans-Golgi network membrane twice by TAS, all of which the endomembrane system term subsumes, and UniProt's own curated location line names the TGN membrane. Replacing it with GO:0032588 keeps the (correct) claim and states it at the grain the evidence actually supports. Note for anyone consuming this: GO:0032588 is already on the gene twice by TAS, so the replacement term is not new and the practical effect is the removal of a redundant parent, not the creation of a third GO:0032588 row. As with the other ARBA row, the rule's published condition sets do not fire on this protein's real InterPro matches, which is recorded in the propagation review; the replacement is proposed on grounds of specificity, not on that provenance issue. Propagation Review Root cause: TERM SCOPING PROBLEM Failure modes: GRANULARITY MISMATCH Sources checked: ARBA:ARBA00028953 · UniProt ARBA rule ARBA00028953 (endomembrane system) SOURCE WEAK OR INFERRED Same audit: the rule publishes 533 condition sets, of which one names an AP4S1 signature but requires IPR010908 AND IPR011012 AND IPR044565 AND Eukaryota; AP4S1 matches only IPR011012 of those three. Not reproducible from the rule's own conditions, and in this case the term is also coarser than what the gene already carries. Proposed replacements: trans-Golgi network membrane Supporting Evidence: file:human/AP4S1/AP4S1-uniprot.txt SUBCELLULAR LOCATION: Golgi apparatus, trans-Golgi network membrane PMID:10436028 Immunogold electron microscopy indicates that AP-4 is associated with nonclathrin-coated vesicles in the region of the trans-Golgi network. |
| GO:0015031 protein transport | IEA GO_REF:0000002 | ACCEPT | Summary: A family-level InterPro2GO transfer that happens to be exactly right: every member of IPR016635 is the small subunit of an AP complex, and every AP complex moves proteins. Reason: AP4S1 really does match IPR016635 - the InterPro API returns exactly three entries for Q9Y587 and this is the family-level one, and it is a DR line in the UniProt flat file - so unlike the two ARBA rows, this annotation is reproducible from its stated source. The term is broad, but it is broad in the way a family-level signature ought to be, and the gene's own evidence supports a protein-transport role directly. No modification needed; the more specific post-Golgi term is proposed as a separate row with its own human loss-of-function evidence rather than as a replacement here. Propagation Review Root cause: NO FAILURE CORE Sources checked: InterPro:IPR016635 · InterPro IPR016635, Adaptor protein complex, sigma subunit SUPPORTS TRANSFER AP4S1 genuinely matches this signature - the InterPro API returns exactly three entries for Q9Y587 and IPR016635 is the family-level one, also present as a DR line in the UniProt flat file. The InterPro2GO mapping of a family whose every member is an AP-complex small subunit to protein transport is a correct family-level statement, and it is correct for this member. Supporting Evidence: file:human/AP4S1/AP4S1-uniprot.txt DR InterPro; IPR016635; AP_complex_ssu. file:human/AP4S1/AP4S1-uniprot.txt SIMILARITY: Belongs to the adaptor complexes small subunit family. PMID:31915823 We conclude that bi-allelic loss-of-function variants in any of the AP-4 subunit genes lead to loss of AP-4 function and subsequently to accumulation of ATG9A in the TGN area and depletion from peripheral compartments. |
| GO:0016192 vesicle-mediated transport | IBA GO_REF:0000033 | ACCEPT | Summary: The single IBA on this gene, inherited from PTN000204281, the one and only annotated PAINT node in PTHR11753. Its eleven gene donors are AP-1, AP-2 and AP-3 sigma subunits from yeast, Candida, fission yeast, worm, fly, rat, mouse and human, and ten of the eleven carry a direct experimental annotation to GO:0016192 or to a term beneath it. AP4S1 is inside the clade and the term is the correct least common ancestor for a donor set that does not agree on anything narrower. Reason: Nothing about this propagation needs arguing with, and the one temptation worth resisting is calling it too general. The donors do not agree on a sub-process: the AP-2 sigmas are annotated to endocytosis, the AP-3 sigmas to Golgi-to-vacuole transport and the AP-1 sigmas to retrograde and endosomal transport. GO:0016192 is the genuine least common ancestor of a heterogeneous set, not a lazy parent, so GRANULARITY_MISMATCH would be the wrong call. The node itself sits at or near the eukaryotic root of the family, before the duplications that produced sigma4, and AP4S1 is a member of that family by HMM (Q9Y587 appears in PTHR11753's own entries.csv), by fold (Longin-like, IPR011012) and by function - AP-4 assembles coats on vesicles leaving the TGN. There is no IRD or IKR anywhere in PTHR11753, so no curator has proposed loss or divergence at any descendant node, and there is no target-specific evidence of either: the human genetics show the opposite, that removing sigma4 removes the complex and its transport function. Q9Y587 does not appear in its own WITH/FROM and has no experimental biological-process annotation of its own, which is the ordinary situation for a gene whose function is inherited rather than lineage-specific. Accepted as core, with the AP4S1-specific refinement proposed as a separate row rather than as a replacement here. Propagation Review Root cause: NO FAILURE CORE Sources checked: CGD:CAL0000182525 · Candida albicans APS3, AP-3 complex sigma subunit (Q59QC5) SUPPORTS TRANSFER Carries GO:0006896 Golgi to vacuole transport by IMP. FB:FBgn0039132 · Drosophila melanogaster AP-1sigma (CG5864) SOURCE WEAK OR INFERRED The one donor with no experimental grounding of its own. Queried through its canonical UniProt accession Q9VCF4, its entire GO:0016192 support is inferred: IEA from InterPro2GO (IPR000804, IPR044733) and ISS from SGD:S000004160 - which is itself another donor in this same WITH/FROM, so that strand adds no independent evidence. Marked SOURCE_WEAK_OR_INFERRED rather than SOURCE_STALE_OR_MISSING or UNRESOLVED because the record is neither missing nor untraceable: it does carry the term, just never experimentally. It remains a legitimate seed of the node, and nothing in the propagation turns on it, since ten of the eleven donors are experimentally grounded. Note separately that this gene seeds both IBD rows at the node while FB:FBgn0043012 currently seeds only the GO:0043231 row despite appearing in this IBA's WITH/FROM; the two IBD rows carry different dates (20260828 and 20260528) and the IBA row predates both (20250902). FB:FBgn0043012 · Drosophila melanogaster AP-2sigma (CG6056) SUPPORTS TRANSFER Queried through Q9VDC3. Its experimental row is the molecular function GO:0035615 clathrin-cargo adaptor activity by IMP, which GO's is_a/part_of closure places beneath GO:0016192 - worth naming explicitly, because counting it as support for a biological-process term only works through that part_of link. MGI:MGI:1098244 · mouse Ap1s1, AP-1 complex subunit sigma-1A (P61967) SUPPORTS TRANSFER Resolved through the UniProt mgi cross-reference (four hits, P61967 the Swiss-Prot one). Carries GO:0042147 retrograde transport, endosome to Golgi by IMP. MGI:MGI:1889383 · mouse Ap1s2, AP-1 complex subunit sigma-2 (Q9DB50) SUPPORTS TRANSFER Resolved through the UniProt mgi cross-reference (five hits, Q9DB50 the Swiss-Prot one). The most directly grounded donor in the set: GO:0016182 synaptic vesicle budding from endosome by IDA and IMP, GO:0036465 synaptic vesicle recycling by IMP, and GO:0016192 itself by IMP. PANTHER:PTN000204281 · PTHR11753 ADAPTOR COMPLEXES SMALL SUBUNIT FAMILY IBD node (seeded by AP-1, AP-2 and AP-3 sigma subunits of fungi, nematode, insect, rodent and human) SUPPORTS TRANSFER The only PAINT node in PTHR11753 that carries any annotation; `just fetch-panther-paint PTHR11753` returns one node with two IBD rows (GO:0043231 and this one), and no IRD or IKR anywhere in the family, so no curator has asserted loss or divergence at any descendant. Its seed set spans the AP-1, AP-2 and AP-3 sigmas of yeast, Candida, fission yeast, worm, fly, rat, mouse and human, which places it at or near the eukaryotic root of the family, before the duplications that produced sigma4. AP4S1 is inside that clade: Q9Y587 is in the family's own entries.csv with gene symbol AP4S1, and AP-4 does exactly what the node asserts, forming coated vesicles at the TGN. This is also the node from which the human AP1S1/AP1S3/AP2S1 reviews in this repository take their GO:0016192 IBA. PomBase:SPAP27G11.06c · Schizosaccharomyces pombe vas2, AP-1 complex subunit sigma-1 (Q9P7N2) SUPPORTS TRANSFER Carries GO:0042147 retrograde transport, endosome to Golgi and GO:0099638 endosome to plasma membrane protein transport, both by IDA. RGD:620188 · rat Ap2s1, AP-2 complex subunit sigma (P62744) SUPPORTS TRANSFER Resolved through the UniProt rgd cross-reference (five hits, P62744 the Swiss-Prot one). Carries GO:0098884 postsynaptic neurotransmitter receptor internalization by EXP, IDA and IMP. SGD:S000003561 · Saccharomyces cerevisiae APS3, AP-3 complex subunit sigma (P47064) SUPPORTS TRANSFER Carries GO:0006896 Golgi to vacuole transport by IMP. With APS1 above, shows the node's fungal grounding is not confined to one AP complex. SGD:S000004160 · Saccharomyces cerevisiae APS1, AP-1 complex subunit sigma-1 (P35181) SUPPORTS TRANSFER Carries GO:0006896 Golgi to vacuole transport by IMP. UniProtKB:P53680 · AP2S1, human AP-2 complex subunit sigma SUPPORTS TRANSFER The closest characterised human relative of sigma4 and the structural reference for the whole sigma family. Its own GO record carries GO:0048488 synaptic vesicle endocytosis by IDA and IMP, i.e. experimental grounding for a descendant of the transferred term. WB:WBGene00000157 · Caenorhabditis elegans aps-2, AP-2 sigma (F02E8.3, Q19123) SUPPORTS TRANSFER Not a UniProt cross-reference key; resolved through the GO API bioentity endpoint to F02E8.3 and thence to Q19123. Carries GO:0072583 clathrin-dependent endocytosis by IMP. Supporting Evidence: file:human/AP4S1/AP4S1-uniprot.txt DR PANTHER; PTHR11753; ADAPTOR COMPLEXES SMALL SUBUNIT FAMILY; 1. file:human/AP4S1/AP4S1-uniprot.txt SIMILARITY: Belongs to the adaptor complexes small subunit family. PMID:10436028 we have searched the expressed sequence tag database and have identified, cloned, and sequenced a new member of each of the four AP subunit families PMID:25552650 In the AP4S1* cell line appeared completely absent, and there was a substantial reduction of all other AP-4 subunits, implying that the loss of σ4 leads to the destabilization of the entire AP-4 complex. |
| GO:0016192 vesicle-mediated transport | NAS PMID:10436028 Characterization of a fourth adaptor-related protein complex... | ACCEPT | Summary: A ComplexPortal author-statement duplicate of the IBA term, cited to the paper that demonstrated AP-4 on non-clathrin-coated vesicles at the TGN. Reason: Duplicating a GO id under a different evidence code is not a problem, and here the NAS row adds something the IBA does not: it ties the claim to direct observation of the human complex on vesicles, rather than to inheritance from an ancestral node. The same reference is also the basis of ComplexPortal's annotation of the AP-4 complex itself (CPX-5151), so this row is the subunit-level restatement of a complex-level assertion, which is the appropriate way to record it for an obligate subunit. Supporting Evidence: PMID:10436028 Immunogold electron microscopy indicates that AP-4 is associated with nonclathrin-coated vesicles in the region of the trans-Golgi network. PMID:10436028 Adaptor protein complexes (APs) function as vesicle coat components in different membrane traffic pathways |
| GO:0030124 AP-4 adaptor complex | IDA PMID:10066790 AP-4, a novel protein complex related to clathrin adaptors. | ACCEPT | Summary: The anchor annotation of the whole record, and the one term whose GO definition names this protein by subunit: GO:0030124 is 'An AP-type membrane coat adaptor complex that consists of beta4, epsilon, mu4 and sigma4 subunits'. Demonstrated by gel filtration, sedimentation velocity and immunoprecipitation, and since confirmed by cryo-EM of the reconstituted complex. Reason: Core, and about as securely established as a complex-membership annotation gets. Two independent 1999 studies identified sigma4 in the complex, by immunoprecipitation and yeast two-hybrid in one and by hydrodynamic and immunoprecipitation analysis in the other; UniProt's SUBUNIT line records the heterotetramer with ECO:0000269 from both; and the 2026 cryo-EM structure was built from recombinant subunits including sigma4 identified explicitly as UniProt Q9Y587, showing the epsilon trunk wrapped around it. The human genetics close the loop from the other direction: when AP4S1 is truncated, the complex does not form. Supporting Evidence: PMID:10066790 Gel filtration, sedimentation velocity, and immunoprecipitation experiments revealed that beta4 is a component of a multisubunit complex (AP-4) that also contains the sigma4 polypeptide and two additional adaptor subunit homologs named mu4 (mu-ARP2) and epsilon. file:human/AP4S1/AP4S1-uniprot.txt (sigma-type AP4S1). {ECO:0000269|PubMed:10066790, PMID:41565640 The N-terminal halves of ε and β4 capture σ4 and the N-terminal domain of μ4 (μ4-NTD), respectively, and wrap around them to form the central structural core of AP-4 PMID:41565640 σ4 (UniProt ID: Q9Y587) |
| GO:0030124 AP-4 adaptor complex | NAS PMID:10436028 Characterization of a fourth adaptor-related protein complex... | ACCEPT | Summary: The ComplexPortal author-statement duplicate of the complex-membership term, cited to the other 1999 paper, which established the same tetramer by co-immunoprecipitation and yeast two-hybrid. Reason: An independent demonstration of the same fact by a different laboratory using different methods, recorded under a weaker evidence code than it deserves. Keeping it preserves the fact that AP-4's composition was established twice, independently, in the same year. No change. Supporting Evidence: PMID:10436028 We have shown by a combination of coimmunoprecipitation and yeast two-hybrid analysis that these four proteins (epsilon, beta4, mu4, and sigma4) are components of a novel adaptor-like heterotetrameric complex, which we are calling AP-4. |
| GO:0031904 endosome lumen | TAS Reactome:R-HSA-5229111 | REMOVE | Summary: Topologically impossible, and traceable to a specific mechanism: Reactome models the reaction's output as the complex AP4:APP sitting in compartment 'endosome lumen', and the Reactome-to-GO mapping gives that compartment to every member of the complex. AP-4 is a cytosolic peripheral membrane coat and never enters a lumen. Reason: GO:0031904 is 'The volume enclosed by the membrane of an endosome'. A coat adaptor cannot be there. Everything known about AP-4's topology puts it on the cytosolic face: UniProt calls it a peripheral membrane protein of the TGN membrane; it is recruited by membrane-anchored ARF1 and binds the cytosolic tails of cargo; and the cryo-EM structure locates the two ARF1 sites on epsilon and on the mu4 C-terminal domain, with the cargo site on mu4 facing the membrane. The Reactome summary itself describes APP being delivered to endosomes, not AP-4 entering their lumen. The reference projection confirms this is a single inherited compartment rather than five findings: QuickGO by reference=Reactome:R-HSA-5229111 returns 11 annotations over 5 entities, the four AP-4 subunits plus APP, with GO:0031904 on five of them. This is a positive biological argument against the term, not an absence of support, which is what REMOVE requires. The same defect sits on the other three AP-4 subunits. Propagation Review Root cause: PROPAGATION BAD Failure modes: COMPARTMENT OR COMPLEX MISMATCH Sources checked: Reactome:R-HSA-5229111 · Reactome reaction R-HSA-5229111, AP4 transports APP from trans-Golgi network to endosome lumen SOURCE BAD Queried through the Reactome ContentService: this is a BlackBoxEvent whose output entity is the complex 'AP4:APP' placed in compartment 'endosome lumen'. Every member of that complex therefore inherits 'endosome lumen' in the Reactome-to-GO cellular-component mapping. Querying QuickGO by reference=Reactome:R-HSA-5229111 returns 11 annotations over 5 entities - the four AP-4 subunits plus APP - with GO:0031904 on five of them. It is one compartment assignment projected onto five proteins, not five observations, and for a cytosolic coat subunit it is topologically impossible. Note that this status is specific to the endosome lumen term: the same reaction is judged separately on the trans-Golgi network membrane row it also sources. Supporting Evidence: file:human/AP4S1/AP4S1-uniprot.txt Peripheral membrane protein file:human/AP4S1/AP4S1-uniprot.txt SUBCELLULAR LOCATION: Golgi apparatus, trans-Golgi network membrane PMID:41565640 the AP-4 core complex contains two ARF1-binding sites in the large subunit ε and the medium subunit μ4-CTD, instead of the large subunit β4, for assembling the membrane-anchored complex. Reactome:R-HSA-5229111 Once bound to AP4M1, APP is transported from the trans-Golgi network (TGN) to endosomes, thereby reducing amyloidogenic processing of the protein. PMID:11707398 This complex consists of four subunits (epsilon, beta4, mu4 and sigma4) and localizes to the cytoplasmic face of the trans-Golgi network (TGN). |
| GO:0032588 trans-Golgi network membrane | TAS Reactome:R-HSA-5229111 | ACCEPT | Summary: The same Reactome projection as the endosome lumen row, but reaching the right answer: the reaction's input complex is placed at the trans-Golgi network membrane, which is where AP-4 assembles. Reason: Core location, at the right grain, and independently established by the primary literature rather than only by Reactome's compartment model: immunogold electron microscopy put AP-4 on non-clathrin-coated vesicles in the TGN region, and UniProt's curated location line reads trans-Golgi network membrane. The contrast with the endosome lumen row from the same reaction is worth stating plainly - one compartment assignment on a Reactome complex propagates to all its members whether or not it is physically possible for them, so the mechanism is not evidence either way and the term has to be judged on the biology. Here the biology supports it. Propagation Review Root cause: NO FAILURE CORE Sources checked: Reactome:R-HSA-5229111 · Reactome reaction R-HSA-5229111, AP4 transports APP from trans-Golgi network to endosome lumen SOURCE WEAK OR INFERRED Judged here with respect to this term, not globally: the same reaction is marked SOURCE_BAD on the endosome lumen row it also sources, because source_status is a statement about a source with respect to one propagated annotation. For trans-Golgi network membrane the compartment Reactome assigns happens to be right, so the source is not bad - but it is weak, because the mechanism that produced it is whole-complex compartment inheritance from a BlackBoxEvent, which would have assigned the same compartment to every member whether or not it were physically possible for them. The term therefore stands on the independent evidence cited in the reason (immunogold electron microscopy and the UniProt location line), not on this projection. Supporting Evidence: file:human/AP4S1/AP4S1-uniprot.txt SUBCELLULAR LOCATION: Golgi apparatus, trans-Golgi network membrane PMID:10436028 Immunogold electron microscopy indicates that AP-4 is associated with nonclathrin-coated vesicles in the region of the trans-Golgi network. |
| GO:0032588 trans-Golgi network membrane | TAS Reactome:R-HSA-5229132 | ACCEPT | Summary: The trans-Golgi network membrane term again, from the companion Reactome reaction in which AP-4 binds APP. Every participant in that reaction is placed at the TGN membrane, which is correct. Reason: A duplicate GO id from a second reaction in the same Reactome pathway, and duplicates are not a defect. This reaction is the better-modelled of the two: its inputs and output all sit at the TGN membrane, and its summary states the mechanism correctly, attributing signal recognition to the mu adaptin rather than to this subunit. Retained as a correct statement of the core location. Propagation Review Root cause: NO FAILURE CORE Sources checked: Reactome:R-HSA-5229132 · Reactome reaction R-HSA-5229132, AP4 binds APP SUPPORTS TRANSFER Queried through the Reactome ContentService: a Reaction whose inputs and output are all placed in compartment 'trans-Golgi network membrane', which is where AP-4 actually is. QuickGO by reference=Reactome:R-HSA-5229132 returns 5 annotations over 5 entities, all GO:0032588. The reaction summary also states the mechanism correctly, attributing signal recognition to AP4M1 rather than to this subunit. Supporting Evidence: Reactome:R-HSA-5229132 The medium (mu) adaptins of all AP complexes can recognise and interact with tyrosine-based (YXXphi) sorting signals found within the cytoplasmic tails of integral membrane proteins file:human/AP4S1/AP4S1-uniprot.txt SUBCELLULAR LOCATION: Golgi apparatus, trans-Golgi network membrane |
| GO:0005198 structural molecule activity | IDA PMID:11409905 Similar subunit interactions contribute to assembly of clath... | NEW | Summary: The molecular function the record is missing. Sigma4 is what holds the AP-4 core together: in a yeast three-hybrid reconstitution the two large adaptins epsilon and beta4 associate with each other only when sigma4 is present, and in patient cells lacking sigma4 every other subunit falls and the complex does not assemble. GO:0005198 is defined as 'The action of a molecule that contributes to the structural integrity of a complex', which is precisely this. Reason: GOA gives AP4S1 no molecular function at all, which for a protein with a solved structure and a named disease is a real gap. The evidence here is specific to this subunit and comes from three independent directions. First, reconstitution: the three-hybrid experiment is a direct assay of sigma4's effect, not a binary interaction test - the readout is whether two other proteins can find each other, and they cannot without it. The same paper shows the identical dependence for AP-1 (gamma and beta1 need sigma1) and for the COPI F subcomplex (gamma-COP and beta-COP need zeta-COP), so this is the conserved architectural role of the small subunit across the superfamily rather than an idiosyncrasy of one assay. Second, human loss of function: in fibroblasts from a patient with compound-heterozygous nonsense and frameshift AP4S1 variants sigma4 is undetectable, all other AP-4 subunits are substantially reduced, hemicomplex and complete-complex assembly both fail, and the accessory protein tepsin loses its membrane recruitment - while AP-1 assembly in the same cells is normal, so this is not a general secretory collapse. Third, structure: the cryo-EM structure of the AP-4 core shows the epsilon N-terminal half wrapping around sigma4 to build the central core, and complex assembly is chaperoned by AAGAB, which stabilises epsilon and sigma4 specifically. IDA rather than IPI because the claim being annotated is an activity (conferring structural integrity) measured by its effect, not a physical interaction with a named partner. Deliberately not annotated as protein binding, which would say nothing. One bound on the claim, from the carrier study cited here: a single functional AP4S1 allele is enough, since 28 heterozygous carriers of a nonsense allele are asymptomatic - so the structural requirement is absolute but not dosage-limiting. Supporting Evidence: PMID:11409905 two large subunits of the AP-4 complex, epsilon-adaptin and beta4-adaptin, are found to interact with each other only in the presence of the small subunit, sigma4 PMID:25552650 In the AP4S1* cell line appeared completely absent, and there was a substantial reduction of all other AP-4 subunits, implying that the loss of σ4 leads to the destabilization of the entire AP-4 complex. PMID:25552650 A control immunoprecipitation using an antibody against γ indicated normal AP-1 complex assembly PMID:25552650 Recruitment of the AP-4 accessory protein tepsin, to the membrane was also abolished. PMID:41565640 The N-terminal halves of ε and β4 capture σ4 and the N-terminal domain of μ4 (μ4-NTD), respectively, and wrap around them to form the central structural core of AP-4 PMID:35976721 binds to and stabilizes the AP-4 ε and σ4 subunits, thus promoting complex assembly PMID:32216065 The discovered variants in AP4S1 lead to reduced AP-4 complex formation in patient-derived fibroblasts. PMID:39865903 The 28 heterozygous carriers of the c.289C>T (p.Arg97Ter) variant in AP4S1, all parents of individuals with SPG52, were asymptomatic and showed no significant neurological impairment. |
| GO:0006892 post-Golgi vesicle-mediated transport | IMP PMID:31915823 Adaptor protein complex 4 deficiency: a paradigm of childhoo... | NEW | Summary: The AP4S1-specific refinement of the IBA. Fibroblasts from four patients with bi-allelic AP4S1 loss-of-function variants retain the AP-4 cargo ATG9A in the trans-Golgi network and lose it from peripheral compartments, and the defect is rescued by restoring AP-4. GO:0006892 is 'The directed movement of substances from the Golgi to other parts of the cell, including organelles and the plasma membrane, mediated by small transport vesicles'. Reason: The IBA correctly gives this gene vesicle-mediated transport at the level the ancestral node supports, and I have accepted it there; this row adds what AP4S1's own human loss-of-function data support, which is narrower. Four AP4S1 patient fibroblast lines were among the fifteen assayed, the ATG9A phenotype was present in all patient lines without exception, and re-expression of an AP-4 subunit redistributed ATG9A, so the missorting is AP-4-dependent rather than incidental to a sick cell. A second study turned the same readout into a quantitative assay and applied it to AP4S1 probands. GO:0006892 is the right term and I checked both halves of that: QuickGO's ancestor closure places it beneath GO:0016192, so it sharpens the IBA rather than contradicting it, and not beneath GO:0015031, so it does not merely duplicate the InterPro protein-transport row. I deliberately did not use GO:0006895 Golgi to endosome transport, whose definition names early sorting endosomes as the destination and clathrin vesicles as the vehicle: AP-4's route is non-clathrin and its ATG9A cargo goes to peripheral and pre-autophagosomal compartments. IMP on human cells, not IDA - the evidence is loss of function in patient-derived fibroblasts. Supporting Evidence: PMID:31915823 we generated fibroblast lines from 15 well-characterized patients carrying homozygous or compound-heterozygous variants in AP4B1 (7 lines), AP4M1 (3 lines), AP4E1 (1 line) and AP4S1 (4 lines) PMID:31915823 We conclude that bi-allelic loss-of-function variants in any of the AP-4 subunit genes lead to loss of AP-4 function and subsequently to accumulation of ATG9A in the TGN area and depletion from peripheral compartments. PMID:31915823 ATG9A was redistributed upon re-expression of AP4B1 arguing that mistrafficking of ATG9A is AP-4-dependent. PMID:34729478 Adaptor protein complex 4-associated hereditary spastic paraplegia is caused by biallelic loss-of-function variants in AP4B1, AP4M1, AP4E1 or AP4S1, which constitute the four subunits of this obligate complex. PMID:41032520 AP-4-mediated TGN export operates independently of clathrin PMID:30262884 We identify three transmembrane cargo proteins, ATG9A, SERINC1 and SERINC3, and two AP-4 accessory proteins, RUSC1 and RUSC2. PMID:11802162 we show that AP-4 can bind different types of cytosolic signals known to mediate basolateral transport in epithelial cells |
Loading supporting content…
Download this section (compressed HTML)Q: For the PAINT curators of PTHR11753: the family carries exactly one annotated node, PTN000204281, whose GO:0016192 IBD is seeded only by AP-1, AP-2 and AP-3 sigmas - no AP-4 sigma from any species is a seed. Now that AP4S1 has human loss-of-function data of its own (ATG9A retention in four patient fibroblast lines) and a cryo-EM structure of the complex, would the AP-4 sigma subclade support a node-level annotation more specific than vesicle-mediated transport, for example post-Golgi vesicle-mediated transport?
Q: For the GO ontology editors: GO:0140312 cargo adaptor activity is defined as binding the structural scaffolding elements of a vesicle coat 'such as clathrin or COPII'. AP-3, AP-4 and AP-5 assemble coats without clathrin, and for AP-4 no scaffold partner is known. Should the definition be broadened, or should a sibling term exist for coat adaptors with no identified scaffold? ComplexPortal already annotates AP-4 (CPX-5151) to GO:0140312 by NAS.
Q: For UniProt: the two ARBA rows on Q9Y587 (ARBA00026971 to cytoplasm, ARBA00028953 to endomembrane system) cannot be reproduced from the condition sets those rules serve - the only sets naming an AP4S1 signature also require IPR027156, IPR010908 or IPR044565, none of which AP4S1 matches. Is the served condition set incomplete, or was annotation run against a different signature snapshot?
Q: For Reactome: R-HSA-5229111 models its output as the complex AP4:APP in compartment 'endosome lumen', which propagates GO:0031904 onto all four AP-4 subunits as well as APP. AP-4 is a cytosolic peripheral coat and cannot occupy a lumen. Could the compartment be carried by the cargo rather than by the whole complex, or the members excluded from the CC mapping for BlackBoxEvent outputs?
Q: For the AP-4 structural community: sigma4 retains Arg15 of the dileucine basic patch, but epsilon carries Thr at the position where gamma1, alphaC and delta all carry the patch arginine (Arg15, Arg21 and Arg26 respectively). Does the AP-4 cryo-EM structure show a pocket at the epsilon-sigma4 interface that is geometrically capable of accepting a dileucine signal, and is Thr44 of epsilon positioned where those arginines sit?
Experiment: Test whether the AP-4 dileucine site is intact but half-built. Substitute Thr44 of epsilon (AP4E1, Q9UPM8) to Arg, the residue that gamma1, alphaC and delta carry at the aligned position, and assay the epsilon(T44R)-sigma4 hemicomplex against the Nef ENTSLL, tyrosinase ERQPLL and LIMP-II ERAPLI signals by yeast three-hybrid, with wild-type epsilon-sigma4 as the negative control and gamma1-sigma1A as the positive. Gain of binding would show that AP-4 lost dileucine recognition by losing one half of a two-part basic patch that sigma4 still supplies; no gain would move the explanation to the surrounding surface or to accessibility of the pocket, and would close off a line of speculation.
Experiment: Ask whether sigma4 has any function outside AP-4. Express BioID2- or TurboID-tagged sigma4 in wild-type and AP4E1-knockout cells, where no complex can assemble, and compare the labelled proteomes. Partners retained in the knockout are candidates for a complex-independent role; a proteome that collapses to nothing would argue that sigma4 is an obligate structural subunit and nothing else, which is the conclusion this review currently draws and which deserves a direct test.
Experiment: Separate the structural function from the transport function on AP4S1 itself. Using the cryo-EM structure of the AP-4 core, design sigma4 substitutions at the epsilon interface that are predicted to weaken but not abolish assembly, complement AP4S1-null cells with each, and score complex assembly by immunoprecipitation alongside the quantitative TGN-to-cytoplasm ATG9A ratio. A graded series relating residual assembly to residual cargo export would convert a binary disease gene into a structure-function map and give missense variant interpretation something to lean on, which it currently lacks.
Experiment: Determine which AP4S1 isoform matters. The clinical literature numbers variants on NM_007077 (Q9Y587-2) while MANE Select and the UniProt canonical sequence are Q9Y587-1, and isoforms 2, 3 and 4 all replace the C-terminal 42 to 46 residues. Establish by isoform-resolved proteomics which forms are made in brain and fibroblast, then complement AP4S1-null cells with each isoform and score complex assembly and the ATG9A ratio. If the isoforms differ, the annotation record needs isoform qualifiers it does not currently have.
What is not known — curated, literature-grounded statements of the open unknowns (the inverse of core functions).
Gap: Sigma4 has no assigned molecular activity outside the AP-4 core, and the one question that could have given it one - whether it helps read a cargo sorting signal, as the sigma subunits of AP-1, AP-2 and AP-3 do - has been answered negatively for the complex but has never been tested by mutating sigma4 itself.
NARROWING BIOLOGY MF_DARK
What is known: Firmly established: epsilon and beta4 cannot associate without sigma4; loss of sigma4 destabilises all four subunits and abolishes assembly; the epsilon trunk wraps around sigma4 in the cryo-EM structure; AAGAB stabilises epsilon and sigma4 during assembly; and the epsilon-sigma4 hemicomplex does not bind three canonical (D/E)XXXL(L/I) signals in the assay where gamma1-sigma1A, alphaC-sigma2 and delta-sigma3A all do. Equally established: sigma4 retains the sigma-side residues of the dileucine site, including the Arg15 of the basic patch, so its sequence does not look like a non-binder. Unknown: whether sigma4 has any partner or activity outside AP-4; whether any AP-4 cargo signal contacts sigma4 at all; and whether the AP-4 negative is explained by the large subunit, since the bioinformatics here shows epsilon carries Thr where all three binding complexes carry the large-subunit Arg of the patch.
Significance: This is essentially the entire molecular-function content of the gene. GOA currently records no molecular function for AP4S1 at all; this review proposes one (structural integrity of the complex), and whether there is a second, recognition-related function determines whether sigma4 is a scaffold or a specificity subunit. It also bears on variant interpretation, since missense alleles are enriched in sigma4 and cannot currently be reasoned about beyond predicted destabilisation.
What would resolve it: Two experiments settle it. For the binding question, substitute Thr44 of epsilon to Arg and assay the epsilon(T44R)-sigma4 hemicomplex against the same three signals by yeast three-hybrid; a gain of binding would show the site is intact but half-built, and no gain would move the explanation to the surrounding surface or to pocket accessibility. For the wider question, proximity labelling from sigma4 in AP4E1-null cells, where no complex can form, would reveal any partner sigma4 has on its own.
Provenance (the field's own admissions):
Gap: The causal step between ATG9A missorting and neurodegeneration in AP-4 deficiency is unidentified, and the obvious candidate has been excluded in the cells where the missorting is measured: bulk autophagic flux is intact in AP-4-deficient patient fibroblasts despite a complete redistribution of ATG9A.
OPEN BIOLOGY RESIDUAL_SUBGAP
What is known: Established: bi-allelic loss of any AP-4 subunit, AP4S1 included, retains ATG9A in the TGN and depletes it from the periphery in patient fibroblasts and iPSC-derived cortical neurons; whole-cell ATG9A rises three- to five-fold; the ratio of TGN to cytoplasmic ATG9A is a quantitative functional readout good enough to be used diagnostically. Also established: autophagic flux in those same fibroblasts is normal under nutrient-rich conditions and on stimulation. Unknown: which downstream event actually causes the axonal pathology, and whether it is autophagy-related at all rather than a consequence of losing one of AP-4's other cargoes.
Significance: It determines whether an autophagy biological-process term belongs on the AP-4 subunits at all. This review does not propose one for AP4S1, precisely because the human evidence stops at cargo missorting and the autophagy phenotypes are spatial, axonal and demonstrated in animal and neuronal models rather than by loss of function in human cells. It is also the gap that any therapeutic programme for AP-4 deficiency has to cross.
What would resolve it: Compartment-resolved rather than bulk autophagy measurements in AP4S1-null human neurons - axonal autophagosome biogenesis rates and autophagosome maturation along the axon, not whole-cell LC3-II - together with a cargo-swap test asking whether forcing ATG9A to the periphery in an AP-4-independent way rescues the axonal phenotype.
Provenance (the field's own admissions):
Gap: Neither of the two ARBA-derived annotations on this gene can be reproduced from its own rule's published condition sets, so there is no recoverable reason why these particular terms were assigned to this particular protein.
OPEN CURATION CC_DARK
What is known: Checked and established: AP4S1 matches exactly three InterPro entries (IPR011012, IPR016635, IPR022775), confirmed live against the InterPro API and consistent with the DR lines in the UniProt flat file. ARBA00026971 (cytoplasm) publishes 2388 condition sets, of which the only one naming an AP4S1 signature additionally requires IPR027156, which AP4S1 does not match. ARBA00028953 (endomembrane system) publishes 533, of which the only candidate requires IPR010908 and IPR044565 as well, neither of which AP4S1 matches. Unknown: whether the rules' served condition sets are incomplete, whether annotation used a different signature snapshot, or whether the mapping is looser than the published conditions imply.
Significance: Both terms happen to be biologically correct here, so nothing is wrong on this gene - which is exactly why it is worth recording. A reviewer cannot tell a correct-by-luck ARBA row from a correct-by-derivation one, and the same opacity would hide a wrong one. This repository already carries a review of another ARBA rule whose annotations could not be reproduced from its condition sets, so the pattern is not specific to these two.
What would resolve it: UniProt would need to serve, per annotation, the condition set that actually fired, or ARBA would need to publish the signature snapshot the rule was evaluated against. Failing that, a systematic audit across many proteins would establish how often served conditions reproduce served annotations.
Provenance (the field's own admissions):
Gap: The clinical and the reference literature on AP4S1 number residues on different isoforms, and nothing in the annotation record says so. Variant nomenclature is standardised on NM_007077 (Q9Y587-2), while UniProt's canonical sequence and the MANE Select transcript are Q9Y587-1.
OPEN CURATIONBIOLOGY RESIDUAL_SUBGAP
What is known: Established: UniProt lists four isoforms, with isoforms 2, 3 and 4 all replacing the C-terminal 42 to 46 residues of the 144-residue canonical form; MANE Select is NM_001128126.3 / NP_001121598.1, which is Q9Y587-1; the founding SPG52 report numbered its variant on NM_007077.3, and the AP-4 diagnostic assay paper drew its AP4S1 domain figure on Q9Y587-2. Unknown: whether the non-canonical isoforms are expressed at protein level in the tissues that matter, and whether any AP-4 function differs between them - no experiment has compared them.
Significance: None of the fourteen GOA rows on this gene carries an isoform qualifier and none of the annotations reviewed here is isoform-specific, so nothing in this review changes. But a published p.Arg97Ter mapped onto the canonical sequence lands in a different place, and the four patient fibroblast lines that supply this gene's only human functional data are described in the other frame. If any isoform-specific function is ever found, the existing record provides no basis for assigning annotations to isoforms.
What would resolve it: Isoform-resolved proteomics or long-read transcriptomics in brain and fibroblast, to establish which AP4S1 isoforms are actually made, followed by complementation of AP4S1-null cells with each isoform scored on the ATG9A readout.
Provenance (the field's own admissions):
Gap: GO has no molecular function that fits the complex-level activity of a non-clathrin AP coat. GO:0140312 cargo adaptor activity, the only term that names the coat-adaptor role, is defined in terms of binding the structural scaffolding elements of a vesicle coat such as clathrin or COPII - and for AP-4 no scaffold partner is known at all.
OPEN ONTOLOGY MF_DARK
What is known: Established: AP-4 does not bind clathrin and assembles a non-clathrin coat; recent vesicle proteomics states plainly that AP-4-mediated TGN export operates independently of clathrin and that the accessory factors responsible for its vesicle biogenesis are uncharacterised, naming PRRC1 and WDR44 as newly found cytosolic regulators. Also established: ComplexPortal nonetheless annotates the human AP-4 complex CPX-5151 to GO:0140312 by NAS. Unknown: what the scaffold of an AP-4 coat is, or whether one exists.
Significance: It forces this review to express AP-4's complex-level activity with the generic parent GO:0060090 molecular adaptor activity rather than the specific child, losing the information that the activity is coat assembly and cargo capture. The same problem applies to AP-3 and AP-5, so a fix would be worth more than one gene. It is raised as an ontology observation rather than a proposed term, because the right repair depends on whether AP-4 turns out to have a scaffold partner.
What would resolve it: Either broaden GO:0140312's definition so that the scaffolding element need not be clathrin or COPII, or create a sibling for coat adaptors that assemble without a known scaffold. Which is correct should follow identification of the AP-4 coat's structural component, if it has one.
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)