AP4M1 encodes mu4-adaptin, the medium (mu) subunit of adaptor protein complex 4 (AP-4), a heterotetramer of epsilon (AP4E1), beta-4 (AP4B1), mu-4 (AP4M1) and sigma-4 (AP4S1) that forms a non-clathrin coat on the cytosolic face of the trans-Golgi network. Within the complex, mu4 is the cargo-recognition subunit: its C-terminal mu homology domain (residues 184-452) binds short sorting motifs in the cytosolic tails of transmembrane proteins, with a signal preference of its own rather than the canonical YXXPhi specificity of the mu subunits of AP-1, AP-2 and AP-3, and with a binding surface whose location on the domain differs from theirs. Recruitment of AP-4 to trans-Golgi network membranes is ARF-dependent and brefeldin A-sensitive, and mu4 itself binds ARF1 directly and independently of its nucleotide state. The physiological cargo established by both targeted and unbiased approaches is ATG9A, the only multispanning membrane protein of the core autophagy machinery: AP-4 exports ATG9A from the trans-Golgi network to the peripheral cytoplasm, and when AP-4 is lost ATG9A is retained at the Golgi, autophagosome formation is impaired, and axons accumulate swellings and aggregates. Additional AP-4-dependent cargoes include the amyloid precursor protein, whose mu4-bound YKFFE motif routes it from the trans-Golgi network to endosomes and so limits amyloidogenic processing, and SERINC1/SERINC3. Accessory factors tepsin, RUSC1/2 and the FTS-Hook-FHIP complex act with the coat; mu4 binds the Hook1 and Hook2 coiled-coil domains directly, coupling AP-4 and its ATG9A cargo to dynein-dependent perinuclear positioning. In neurons AP-4 restricts cargo, including AMPA receptor-TARP complexes, to the somatodendritic domain, and in polarized epithelia it participates in basolateral sorting. Biallelic loss-of-function variants in AP4M1 cause spastic paraplegia 50, a childhood-onset complicated hereditary spastic paraplegia with intellectual disability, microcephaly and progressive spasticity, one of the four genetically equivalent forms of AP-4 deficiency syndrome.
Definition: Binding to a tyrosine-based sorting signal, a short linear motif of the form YXXPhi (where Phi is a bulky hydrophobic residue) or a related tyrosine-containing motif, located within the cytosolic domain of a transmembrane protein and read by an adaptor to direct that protein into a transport vesicle.
Justification: Recognition of tyrosine-based sorting signals is the defining molecular activity of the mu subunits of all four AP complexes, yet GO offers no term for it that is not tied to clathrin-mediated endocytosis. The one available term, GO:0089710 endocytic targeting sequence binding, is defined as binding a signal which directs internalization by clathrin-coated pits, so it can be used for mu2 but not for mu1, mu3 or mu4, whose signals act in post-Golgi and endosomal sorting. The effect is that the best-characterized activity of AP4M1 - a specificity profile measured by combinatorial peptide screening, a dissociation constant measured by surface plasmon resonance, and a co-crystal with the APP signal - can only be recorded as GO:0019904 protein domain specific binding, which is both too general and technically inapt, since a sorting motif is a short linear element rather than a folded domain. A general parent for GO:0089710 would let the mu subunits, and other signal-reading adaptors, be annotated at the resolution the experiments actually reach.
Parent term: protein binding
Supporting Evidence:
| GO Term | Evidence | Action | Reason |
|---|---|---|---|
| GO:0000045 autophagosome assembly | IMP PMID:29180427 AP-4 mediates export of ATG9A from the trans-Golgi network t... | ACCEPT | Summary: Core. Loss of AP-4 traps ATG9A at the trans-Golgi network and impairs LC3B lipidation and maturation of preautophagosomal structures; the same ATG9A redistribution is seen in fibroblasts from patients whose mutation is in the mu4 subunit itself. Reason: This is the function that finally explained AP-4 after nearly twenty years without an assigned pathway, and it is supported from three independent directions: the targeted study that identified ATG9A as an AP-4 cargo, unbiased organellar proteomics in AP-4 knockout and patient cells, and an AP-4 knockout mouse. AP4M1 is an obligate subunit, so the complex-level perturbation is evidence about this gene product; the AP4M1-specific patient-fibroblast result removes any residual doubt that the phenotype requires mu4. Supporting Evidence: PMID:29180427 AP-4 promotes signal-mediated export of ATG9A from the trans-Golgi network to the peripheral cytoplasm, contributing to lipidation of the autophagy protein LC3B and maturation of preautophagosomal structures. PMID:29698489 Importantly, we found that ATG9A was more concentrated at the TGN and depleted from the peripheral cytoplasm in both skin fibroblasts from patients with mutations in AP-4 μ4 and neurons from AP-4 ε KO mice, as compared to their normal counterparts. PMID:30262884 We demonstrate that AP-4 deficiency causes missorting of ATG9A in diverse cell types, including patient-derived cells, as well as dysregulation of autophagy. |
| GO:0005515 protein binding | IPI PMID:25416956 A proteome-scale map of the human interactome network. | MARK AS OVER ANNOTATED | Summary: Partner resolved as USP47, a deubiquitinase, from a proteome-scale binary yeast two-hybrid map. No follow-up anywhere in the AP-4 literature and no functional interpretation available. Reason: Bare protein binding from a single high-throughput binary screen. USP47 appears in none of the AP-4 mechanistic literature - not in the coat composition papers, not in the AP-4 affinity-purification datasets, and not among the accessory factors (tepsin, RUSC1/2, FHF) that unbiased proteomics did recover. The edge may be real but it supports no informative molecular function, so there is nothing to modify it to. Supporting Evidence: PMID:25416956 high-quality human binary protein-protein interactions |
| GO:0005515 protein binding | IPI PMID:26496610 A human interactome in three quantitative dimensions organiz... | MARK AS OVER ANNOTATED | Summary: Partner resolved as TEPSIN, the only known AP-4 accessory protein. The association is genuine but it is with the complex, not with mu4: tepsin's two conserved motifs bind the C-terminal ear domains of beta-4 and epsilon. Reason: A GFP-tagged affinity-purification survey reports co-purification of AP4M1 and TEPSIN, which follows from both being part of the AP-4 coat. The binding surfaces were subsequently mapped and they are on the other two subunits, so this row does not describe a mu4 molecular function. What it does record - membership of the same coat - is already asserted by the GO:0030124 rows. Supporting Evidence: PMID:26542808 which interact with the C-terminal ear (or appendage) domains of the β4 and ϵ subunits of AP-4, respectively. PMID:26496610 Here, we have generated a library of HeLa cell lines expressing 1,125 GFP-tagged proteins |
| GO:0005515 protein binding | IPI PMID:26544806 Association between Rare Variants in AP4E1, a Component of I... | MODIFY | Summary: Partner resolved as NAGPA. This is a genuine mu4-specific interaction with the cytosolic region of a candidate cargo, and the same paper already supports a more informative term on this gene. Reason: The partner is resolved and the interaction is mu4-specific rather than complex-level, so the bare binding term should be replaced by the informative one. GOA itself records GO:0019904 from this same reference; making the two rows agree states the fact once, at the level of signal recognition by the mu homology domain, rather than leaving one of them as uninformative protein binding. Proposed replacements: protein domain specific binding Supporting Evidence: PMID:26544806 We found that the μ4 subunit of AP-4 interacts with NAGPA, an enzyme involved in the synthesis of the mannose 6-phosphate signal that targets acid hydrolases to the lysosome file:human/AP4M1/AP4M1-uniprot.txt Interacts with tyrosine-based sorting signals on the cytoplasmic tail |
| GO:0005515 protein binding | IPI PMID:29180427 AP-4 mediates export of ATG9A from the trans-Golgi network t... | MODIFY | Summary: Partner resolved as ATG9A, the physiological AP-4 cargo. This is signal-mediated cargo selection by the coat, which has a molecular function term of its own. Reason: Recording the central cargo-recognition event of this complex as bare protein binding loses everything that matters about it. The interaction is what makes AP-4 a coat adaptor for ATG9A, so cargo adaptor activity is the informative term; on a subunit of an obligate heterotetramer it is a contributed rather than an independently enabled activity, which is how it is represented in core_functions. The narrower child GO:0035615 is not available here because it commits to clathrin and to endocytic vesicles, neither of which applies to AP-4. Proposed replacements: cargo adaptor activity Supporting Evidence: PMID:29180427 Here we report the identification of ATG9A, the only multispanning membrane component of the core autophagy machinery, as a specific AP-4 cargo. PMID:29180427 AP-4 promotes signal-mediated export of ATG9A from the trans-Golgi network to the peripheral cytoplasm, contributing to lipidation of the autophagy protein LC3B and maturation of preautophagosomal structures. |
| GO:0005515 protein binding | IPI PMID:32073997 The FTS-Hook-FHIP (FHF) complex interacts with AP-4 to media... | MARK AS OVER ANNOTATED | Summary: Partner resolved as TEPSIN again, here from affinity purification of the AP-4 complex. Same verdict as the other two TEPSIN rows: co-membership of the coat, not a mu4 binding surface. Reason: This paper's affinity purification was performed on AP-4 and recovered its known accessory protein along with the new one, FHF. Of the two hits, only the FHF subunits were shown to bind mu4 directly; tepsin's binding sites are on the beta-4 and epsilon ears. The row is not wrong about the coat but it is not a mu4 molecular function. Supporting Evidence: PMID:26542808 which interact with the C-terminal ear (or appendage) domains of the β4 and ϵ subunits of AP-4, respectively. PMID:32073997 To identify additional proteins that cooperate with AP-4 in ATG9A trafficking, we performed affinity purification-mass spectrometry followed by validation of the hits by biochemical and functional analyses. |
| GO:0005515 protein binding | IPI PMID:32073997 The FTS-Hook-FHIP (FHF) complex interacts with AP-4 to media... | MODIFY | Summary: Partner resolved as HOOK1. The paper maps the interaction to direct binding of mu4 to the Hook1 and Hook2 coiled-coil domains, with the functional consequence that AP-4 and ATG9A lose their perinuclear positioning when FHF is depleted. Reason: This is the one protein-binding row on AP4M1 that is unambiguously a mu4 surface and unambiguously domain-directed: the element bound is a named coiled-coil domain, which is what GO:0019904 is for. The bare term should be replaced. Note that GOA carries two identical rows for this pair differing only in date; the seeder correctly collapses them into this single entry. Proposed replacements: protein domain specific binding Supporting Evidence: PMID:32073997 We found that the AP-4-FHF interaction is mediated by direct binding of the AP-4 μ4 subunit to coiled-coil domains in the Hook1 and Hook2 subunits of FHF. PMID:32073997 Knockdown of FHF subunits resulted in dispersal of AP-4 and ATG9A from the perinuclear region of the cell |
| GO:0005515 protein binding | IPI PMID:32296183 A reference map of the human binary protein interactome. | MARK AS OVER ANNOTATED | Summary: Partner resolved as FNTA, the shared alpha subunit of protein farnesyltransferase and geranylgeranyltransferase type 1, from the HuRI binary interactome. No functional follow-up. Reason: A single binary two-hybrid edge with no supporting cell biology. Prenylation has no established role in AP-4 assembly or ARF-dependent recruitment, and FNTA does not appear in the AP-4 affinity-purification or organellar-proteomics datasets. Bare protein binding with nothing informative to replace it with. Supporting Evidence: PMID:32296183 Here we present a human 'all-by-all' reference interactome map of human binary protein interactions, or 'HuRI'. |
| GO:0005515 protein binding | IPI PMID:32296183 A reference map of the human binary protein interactome. | MARK AS OVER ANNOTATED | Summary: Partner resolved as isoform 3 of USP47, recovering in HuRI the same protein that the earlier binary map reported. Still no functional follow-up. Reason: Two independent binary two-hybrid screens agreeing on an edge raises confidence that the edge is reproducible, but reproducibility is not function: neither screen says what mu4 and USP47 do together, and no targeted study has followed it up. This is the pattern the over-annotation action exists for. Supporting Evidence: PMID:32296183 Here we present a human 'all-by-all' reference interactome map of human binary protein interactions, or 'HuRI'. |
| GO:0005515 protein binding | IPI PMID:33961781 Dual proteome-scale networks reveal cell-specific remodeling... | MARK AS OVER ANNOTATED | Summary: Partner resolved as TEPSIN for the third time, here from BioPlex affinity purification. Same reading as the other two. Reason: Three separate proteome-scale datasets recover the AP4M1-TEPSIN pair because both proteins are in the AP-4 coat and co-purify. The targeted mapping study places tepsin's binding sites on beta-4 and epsilon, so none of the three rows is evidence of a mu4 binding activity. Supporting Evidence: PMID:33961781 Through affinity-purification mass spectrometry, we have created two proteome-scale, cell-line-specific interaction networks. PMID:26542808 which interact with the C-terminal ear (or appendage) domains of the β4 and ϵ subunits of AP-4, respectively. |
| GO:0005737 cytoplasm | IEA GO_REF:0000117 | KEEP AS NON CORE | Summary: True but maximally general. AP-4 is a peripheral membrane coat on the cytosolic face of the trans-Golgi network and also has a soluble cytosolic pool, so cytoplasm holds; it conveys nothing that the trans-Golgi network rows do not. Reason: Retained as a correct high-level parent. It should not be treated as informative about localization when the same record carries trans-Golgi network and trans-Golgi network membrane with direct experimental support. Propagation Review Root cause: NO FAILURE NON CORE Sources checked: ARBA:ARBA00026971 · ARBA rule asserting GO:0005737 SUPPORTS TRANSFER Rule fetched from rest.uniprot.org/arba. It has 2388 condition sets over FunFam, InterPro, PANTHER and taxon conditions; two mention an AP4M1 signature and none is fully satisfied by this protein's signature set plus the human lineage, so which set fires here was not reproduced. The conclusion is independently correct, so this is a note on checkability, not a challenge to the row. Supporting Evidence: file:human/AP4M1/AP4M1-uniprot.txt AP-4 forms a non clathrin-associated coat on vesicles departing the file:human/AP4M1/AP4M1-bioinformatics/RESULTS.md The cytoplasm rule does not reproduce - of its 2388 condition sets, two mention an AP4M1 signature and none is fully satisfied. |
| GO:0005769 early endosome | EXP PMID:20230749 Sorting of the Alzheimer's disease amyloid precursor protein... | KEEP AS NON CORE | Summary: A genuine secondary station. The APP study places AP-4 at early endosomes as well as the trans-Golgi network, and UniProt records early endosome as an experimental subcellular location. Reason: Real but peripheral. Every mechanistic account of AP-4 - ARF1-dependent recruitment, coat assembly with tepsin, cargo capture, vesicle export - is trans-Golgi network-centred, and the endosome is where AP-4-dependent cargo arrives rather than where the coat works. Keeping it as non-core preserves the observation without letting it compete with the trans-Golgi network for the core location. Supporting Evidence: PMID:20230749 These findings demonstrate that APP and AP-4 engage in a distinct type of signal-adaptor interaction that mediates transport of APP from the trans-Golgi network (TGN) to endosomes, thereby reducing amyloidogenic processing of the protein. file:human/AP4M1/AP4M1-uniprot.txt {ECO:0000305|PubMed:32073997}. Early endosome |
| GO:0005769 early endosome | IEA GO_REF:0000044 | KEEP AS NON CORE | Summary: Mechanical mapping of the UniProt SUBCELLULAR LOCATION keyword for early endosome, which the record carries with experimental evidence from the APP study. Reason: The mapping is faithful to a UniProt statement that is itself experimentally supported, so there is nothing to correct. It is graded non-core for the same reason as the EXP row: the endosome is a destination of AP-4 traffic rather than the compartment where the coat assembles and functions. Propagation Review Root cause: NO FAILURE NON CORE Sources checked: UniProtKB-SubCell:SL-0094 · UniProt controlled vocabulary term for early endosome SUPPORTS TRANSFER The vocabulary term is asserted on O00189 with ECO:0000269|PubMed:20230749, so the automatic mapping rests on an experimental UniProt statement rather than on a prediction. Supporting Evidence: file:human/AP4M1/AP4M1-uniprot.txt {ECO:0000305|PubMed:32073997}. Early endosome |
| GO:0005769 early endosome | ISS GO_REF:0000024 | KEEP AS NON CORE | Summary: Curator transfer from dog AP4M1. The donor's own early-endosome annotation is an IDA from the MDCK study, and the human protein has independent experimental support for the same location. Reason: A well-grounded transfer that happens to be redundant with the human EXP row. Dog is the correct species for the donor evidence because MDCK cells are canine, so this is not a case of an ortholog annotation floating free of an experiment. Graded non-core for the same reason as the other early-endosome rows. Propagation Review Root cause: NO FAILURE NON CORE Sources checked: UniProtKB:E2RED8 · AP4M1_CANLF, dog mu4 SUPPORTS TRANSFER QuickGO shows E2RED8 carrying GO:0005769 IDA from PMID:11802162, the MDCK study, and MDCK is a canine cell line - so the donor annotation and the experiment are in the same species. Dog and human mu4 are 94.9% identical over a global alignment. Supporting Evidence: PMID:11802162 Furthermore, in MDCK cells with depleted mu 4 protein levels, several basolateral proteins are mis-sorted to the apical surface, showing that AP-4 participates in basolateral sorting in epithelial cells. file:human/AP4M1/AP4M1-bioinformatics/RESULTS.md F255 and R283 are present in mouse, rat and dog mu4 at the same native positions. |
| GO:0005794 Golgi apparatus | IEA GO_REF:0000044 | KEEP AS NON CORE | Summary: The parent of the informative location. UniProt states Golgi apparatus, trans-Golgi network membrane; the mapping emits the Golgi apparatus parent alongside. Reason: Correct but subsumed. The trans-Golgi network and trans-Golgi network membrane rows on this gene carry the same information at the resolution the biology actually has, and AP-4 is not distributed through the Golgi stack generally. Propagation Review Root cause: NO FAILURE NON CORE Sources checked: UniProtKB-SubCell:SL-0132 · UniProt controlled vocabulary term for Golgi apparatus SUPPORTS TRANSFER Derived from the same SUBCELLULAR LOCATION line that yields the trans-Golgi network membrane assignment, which carries ECO:0000269 evidence from three papers. Faithful, just coarser than the record supports. Supporting Evidence: file:human/AP4M1/AP4M1-uniprot.txt -!- SUBCELLULAR LOCATION: Golgi apparatus, trans-Golgi network membrane |
| GO:0005802 trans-Golgi network | IBA GO_REF:0000033 | ACCEPT | Summary: Sound and core. The IBD sits on the deep pan-mu node PTN000055849, and trans-Golgi network localization is one of the few properties that genuinely is ancestral across the mu subunits; AP4M1 itself is among the node's seeds and has three direct IDA rows for the same term. Reason: Reviewing this IBA means asking whether AP4M1 sits inside the clade that inherited the localization and whether it shows evidence of having lost it. Both answers favour the annotation. The node's seeds span plant AP-1 and AP-4 mu, yeast Apm2, dog mu4 and human mu4, each with its own IDA for trans-Golgi network in QuickGO, and the PAINT curator's judgment is visibly discriminating here: the same family slice carries an IRD at the AP-2 mu node PTN000242370 that rejects this exact term, which is correct for a plasma-membrane adaptor. AP4M1 is not in that rejecting clade. Propagation Review Root cause: NO FAILURE CORE Sources checked: AGI_LocusCode:AT1G60780 · O22715, AP1M2_ARATH, Arabidopsis AP-1 complex subunit mu-2 SUPPORTS TRANSFER Resolved through the UniProt cross-reference for this AGI locus code. QuickGO shows GO:0005802 IDA from PMID:23733933 on O22715. A plant AP-1 mu, so it supports the deep node rather than an AP-4-specific claim. AGI_LocusCode:AT4G24550 · Q9SB50, AP4M_ARATH, Arabidopsis AP-4 complex subunit mu SUPPORTS TRANSFER Resolved through the UniProt cross-reference (3 hits: one Swiss-Prot entry plus two TrEMBL isoform entries; the reviewed entry is Q9SB50). GO:0005802 IDA from PMID:26546666. The only plant seed that is itself an AP-4 subunit. PANTHER:PTN000055849 · PTHR10529 pan-mu-adaptin IBD node, seeded by plant AP-1/AP-4 mu, yeast Apm2, dog mu4 and human mu4 SUPPORTS TRANSFER The node, not a donor. Ancestral to the mu subunits of all four AP complexes. For trans-Golgi network the placement is defensible at this depth because the seed list reaches into the AP-4 clade itself, and the sibling AP-2 node carries an explicit IRD against the term. SGD:S000001011 · P38700, APM2_YEAST, adaptin medium chain homolog APM2 SUPPORTS TRANSFER Resolved through the UniProt cross-reference for this SGD identifier. QuickGO shows GO:0005802 IDA from PMID:26658609. UniProtKB:E2RED8 · AP4M1_CANLF, dog mu4 SUPPORTS TRANSFER GO:0005802 IDA from PMID:11802162, the MDCK study. An AP-4 mu subunit donor inside the clade of interest. UniProtKB:O00189 · AP4M1, the target itself SUPPORTS TRANSFER The target's own three IDA rows for this term are among the descendant evidences behind the IBD. Expected for a gene with direct experimental grounding, and a marker that the localization is inherited rather than lineage-specific. Supporting Evidence: 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: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:0005802 trans-Golgi network | IDA PMID:20230749 Sorting of the Alzheimer's disease amyloid precursor protein... | ACCEPT | Summary: Direct localization of AP-4 to the trans-Golgi network in the APP sorting study, which is also where the mu4-dependent export of APP begins. Reason: The core compartment for this gene product, supported here and in two further IDA rows from independent papers, and recorded by UniProt as an experimental subcellular location. Supporting Evidence: PMID:20230749 These findings demonstrate that APP and AP-4 engage in a distinct type of signal-adaptor interaction that mediates transport of APP from the trans-Golgi network (TGN) to endosomes, thereby reducing amyloidogenic processing of the protein. |
| GO:0005802 trans-Golgi network | IDA PMID:29180427 AP-4 mediates export of ATG9A from the trans-Golgi network t... | ACCEPT | Summary: Direct localization in the ATG9A study, the compartment from which AP-4 exports its cargo. Reason: Core. This is the compartment in which the AP-4 coat assembles and captures cargo, and it is the compartment where ATG9A accumulates when AP-4 is lost, so the localization and the function are the same observation seen twice. Supporting Evidence: PMID:29180427 AP-4 promotes signal-mediated export of ATG9A from the trans-Golgi network to the peripheral cytoplasm, contributing to lipidation of the autophagy protein LC3B and maturation of preautophagosomal structures. |
| GO:0005802 trans-Golgi network | IDA PMID:32073997 The FTS-Hook-FHIP (FHF) complex interacts with AP-4 to media... | ACCEPT | Summary: Direct localization in the FHF study, which additionally shows that the perinuclear concentration of AP-4 depends on the FHF complex binding mu4. Reason: Core, and the third independent IDA for the same compartment. UniProt records trans-Golgi network membrane with ECO:0000269 evidence from this paper among others, and adds that the association is peripheral rather than integral. Supporting Evidence: PMID:32073997 The heterotetrameric adaptor protein complex 4 (AP-4) is a component of a protein coat associated with the trans-Golgi network (TGN). file:human/AP4M1/AP4M1-uniprot.txt -!- SUBCELLULAR LOCATION: Golgi apparatus, trans-Golgi network membrane |
| GO:0005802 trans-Golgi network | NAS PMID:10436028 Characterization of a fourth adaptor-related protein complex... | ACCEPT | Summary: Author statement from the founding characterization, where the localization was established by immunofluorescence and immunogold electron microscopy. Reason: Correct, and redundant with three IDA rows for the same term, so nothing turns on the weaker evidence code. The underlying observation in this paper is in fact a direct one - discrete perinuclear puncta, brefeldin A-sensitive, with AP-4 on nonclathrin-coated vesicles in the trans-Golgi network region by immunogold EM. 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. |
| GO:0005829 cytosol | IEA GO_REF:0000120 | KEEP AS NON CORE | Summary: Inferred from two biological processes already on the gene. AP-4 does have a real soluble cytosolic pool, so the conclusion is right even though the route to it is indirect. Reason: A cytosolic pool of AP-4 was described in the founding papers - the coat cycles on and off trans-Golgi network membranes under ARF control, and brefeldin A drives it into the cytosol. The term is therefore correct, but it describes the unbound state of a membrane coat rather than where it acts, so it is not a core location. Propagation Review Root cause: NO FAILURE NON CORE Sources checked: GO:0006895 · Golgi to endosome transport, a process term already on this gene SUPPORTS TRANSFER A GO term, not a gene product: this row is an inter-ontology inference from processes the gene is annotated to, not a transfer from a donor protein. The human GO:0006895 IMP row it draws on is itself sound. GO:0090160 · Golgi to lysosome transport, a process term already on this gene SUPPORTS SOURCE BUT NOT TARGET The weaker of the two inputs. This review grades the GO:0090160 rows as non-core because they rest on a reporter chimera rather than an endogenous lysosomal cargo, so an inference built on it should not be read as strong support either. The cytosol conclusion survives on the independent evidence of a soluble AP-4 pool. Supporting Evidence: PMID:10066790 An antibody to beta4 recognized in human cells an approximately 83-kDa polypeptide that exists in both soluble and membrane-associated forms. PMID:11707398 In agreement with these observations, treatment of HeLa cells with 5 µg/ml BFA for 10 min caused redistribution of AP-4 from the TGN to the cytosol |
| GO:0006605 protein targeting | IBA GO_REF:0000033 | ACCEPT | Summary: Sound and core. The IBD is placed on PTN000242612, the AP-4 mu clade, and all three gene-level seeds are AP-4 mu subunits - mouse, dog and human - each with its own experimental annotation for targeting. Reason: This is what a well-placed IBA looks like. The node is the AP-4 mu node rather than the deep pan-mu node, so nothing about AP-1, AP-2 or AP-3 biology is being carried across; the seeds are orthologs in the strict sense; and the target is inside the clade by construction. The term is general, but generality is the right call for a node whose descendants target different cargoes in different tissues, and the gene carries more specific process terms alongside it. Propagation Review Root cause: NO FAILURE CORE Sources checked: MGI:MGI:1337063 · Q9JKC7, AP4M1_MOUSE, mouse Ap4m1 SUPPORTS TRANSFER Resolved through the UniProt cross-reference for this MGI identifier (4 hits: Q9JKC7 Swiss-Prot plus three TrEMBL entries; the reviewed entry is the donor). QuickGO shows GO:0006605 IMP from PMID:18341993, the AP-4 knockout study in neurons. PANTHER:PTN000242612 · PTHR10529 AP-4 mu IBD node, seeded by mouse, dog and human AP4M1 SUPPORTS TRANSFER The node. This is the AP-4-specific node of the family, also carrying the GO:0030124 and GO:0090160 IBDs, so the assertion is scoped to the AP-4 clade and the target sits inside it. UniProtKB:E2RED8 · AP4M1_CANLF, dog mu4 SUPPORTS TRANSFER QuickGO shows GO:0006605 IDA from PMID:11802162. A second independent experimental grounding for the node, in a third species. UniProtKB:O00189 · AP4M1, the target itself SUPPORTS TRANSFER The target's own GO:0006605 IDA from PMID:20230749 is one of the descendant evidences behind the IBD. Expected and not circular. Supporting Evidence: PMID:18341993 These results indicate that AP-4 may regulate proper somatodendritic-specific distribution of its cargo proteins, including AMPA receptor-TARP complexes and the autophagic pathway in neurons. file:human/AP4M1/AP4M1-uniprot.txt (By similarity). Within AP-4, the mu-type subunit AP4M1 is directly |
| GO:0006605 protein targeting | IDA PMID:20230749 Sorting of the Alzheimer's disease amyloid precursor protein... | ACCEPT | Summary: Direct evidence that AP-4 targets APP through a signal recognized by mu4, with the structural basis solved in the same paper. Reason: Core, and the best-characterized instance of the activity: the signal is identified (YKFFE), the binding site is crystallized, point mutations in mu4 abolish the interaction, and disrupting it changes where APP goes and how it is processed. Supporting Evidence: PMID:20230749 Herein, we report the interaction of an YKFFE sequence from the cytosolic tail of the Alzheimer's disease amyloid precursor protein (APP) with the mu4 subunit of AP-4. file:human/AP4M1/AP4M1-uniprot.txt F->A: Abolishes interaction with APP. |
| GO:0006605 protein targeting | IEA GO_REF:0000120 | ACCEPT | Summary: Automatic route to a term the gene already holds by IDA and IBA. Both of its components check out: the ARBA rule reproduces on AP4M1's own signatures, and the Ensembl projection is from mouse Ap4m1, which has an IMP for the term. Reason: There is nothing to correct. The conclusion agrees with direct human evidence, the mouse donor is a true ortholog with its own experimental annotation, and unlike the cytoplasm rule this ARBA rule was reproducible: of its 16 condition sets exactly one is fully satisfied by AP4M1's FunFam and taxonomic conditions. Propagation Review Root cause: NO FAILURE CORE Sources checked: ARBA:ARBA00028630 · ARBA rule asserting GO:0006605 SUPPORTS TRANSFER Fetched from rest.uniprot.org/arba. 16 condition sets over FunFam and taxon conditions; exactly one is fully satisfied by AP4M1, so the rule genuinely fires on this protein rather than reaching it by an unreproducible route. UniProtKB:Q9JKC7 · AP4M1_MOUSE, mouse Ap4m1 SUPPORTS TRANSFER The orthology-projection donor. QuickGO shows GO:0006605 IMP from PMID:18341993 on this entry, so the projected term has experimental grounding on the donor side. ensembl:ENSMUSP00000019662 · Ensembl mouse protein identifier used by the Compara projection SUPPORTS TRANSFER Not a lookup key in any local family index - Ensembl protein identifiers are not accessions - but it is the same mouse gene product as the Q9JKC7 entry listed alongside it, which is what the projection transfers from. Supporting Evidence: file:human/AP4M1/AP4M1-bioinformatics/RESULTS.md The two protein-trafficking rules reproduce cleanly: each has exactly one condition set satisfied by AP4M1, so those IEA rows follow from the rule as published. |
| GO:0006622 protein targeting to lysosome | IDA PMID:11139587 Signal-binding specificity of the mu4 subunit of the adaptor... | KEEP AS NON CORE | Summary: Rests on a reporter construct: a Tac chimera carrying a signal selected by mu4 was delivered to the endosomal-lysosomal system without passing through the plasma membrane. It shows what a mu4-read signal can do, not that AP-4 targets an endogenous lysosomal protein. Reason: The experiment is a genuine direct assay and the curator read the full text, so this is not a row to remove. But the cargo is engineered, and twenty years later no endogenous lysosomal cargo of AP-4 has been established: the unbiased organellar survey of AP-4-dependent proteins returned ATG9A, SERINC1 and SERINC3, and the LAMP-2 signal that mu4 binds in vitro has not been shown to require AP-4 in cells. Non-core is the honest grading. Supporting Evidence: PMID:11139587 we constructed a Tac chimera bearing a mu4-specific YXXphi signal. This chimera was targeted to the endosomal-lysosomal system without being internalized from the plasma membrane. PMID:30262884 We identify three transmembrane cargo proteins, ATG9A, SERINC1 and SERINC3, and two AP-4 accessory proteins, RUSC1 and RUSC2. |
| GO:0006886 intracellular protein transport | IEA GO_REF:0000002 | ACCEPT | Summary: A correct interpro2go transfer. Both signatures are pan-mu-subunit, and intracellular protein transport is a property that really does hold for every mu subunit, AP-4's included. Reason: This is the well-behaved member of the three terms these two signatures carry. Unlike clathrin adaptor complex, which is true only of the AP-1 and AP-2 branches, intracellular protein transport is family-wide, so a family-level signature is exactly the right instrument for it. The term is general but not wrong, and human IMP rows support it independently. Propagation Review Root cause: NO FAILURE CORE Sources checked: InterPro:IPR001392 · Clathrin adaptor, mu subunit SUPPORTS TRANSFER Family-level signature matching every AP complex mu subunit. Its interpro2go mapping carries GO:0006886, GO:0016192 and GO:0030131 (checked at the InterPro API); this is the term of the three that is a genuine family property. InterPro:IPR018240 · Clathrin adaptor, mu subunit, conserved site SUPPORTS TRANSFER A conserved-site signature within the same family, with the same three-term mapping. Same reading. Supporting Evidence: file:human/AP4M1/AP4M1-uniprot.txt -!- SIMILARITY: Belongs to the adaptor complexes medium subunit family. PMID:10436028 Adaptor protein complexes (APs) function as vesicle coat components in different membrane traffic pathways; however, there are a number of pathways for which there is still no candidate coat. |
| GO:0006886 intracellular protein transport | IMP PMID:20230749 Sorting of the Alzheimer's disease amyloid precursor protein... | ACCEPT | Summary: Perturbing the AP-4-APP interaction changes where APP goes, which is intracellular protein transport at the general level. Reason: Correct and directly supported in human cells. The term is a broad parent of the more specific Golgi to endosome transport row from the same paper, and both are legitimately on the gene. Supporting Evidence: PMID:20230749 Disruption of the APP-AP-4 interaction decreases localization of APP to endosomes and enhances gamma-secretase-catalyzed cleavage of APP to the pathogenic amyloid-beta peptide. |
| GO:0006886 intracellular protein transport | IMP PMID:29180427 AP-4 mediates export of ATG9A from the trans-Golgi network t... | ACCEPT | Summary: Loss of AP-4 changes where ATG9A is, from the peripheral cytoplasm to the trans-Golgi network. Intracellular protein transport at the general level. Reason: Correct and core, though stated generally. The specific content of this row - that the cargo is ATG9A and the step is trans-Golgi network export - is carried by the autophagosome assembly and Golgi to endosome transport rows. Supporting Evidence: PMID:29180427 AP-4 promotes signal-mediated export of ATG9A from the trans-Golgi network to the peripheral cytoplasm, contributing to lipidation of the autophagy protein LC3B and maturation of preautophagosomal structures. |
| GO:0006892 post-Golgi vesicle-mediated transport | TAS Reactome:R-HSA-199992 | ACCEPT | Summary: A correct, broad process term attached through a Reactome pathway whose own narrative is about clathrin coats nucleated by AP-1 rather than about AP-4. Reason: The term itself is right: AP-4 is a post-Golgi coat and everything it does is post-Golgi vesicle-mediated transport. The provenance is imperfect - the cached pathway summary describes AP-1-triggered clathrin-coated vesicles - but that is an argument about which pathway AP4M1 should be placed in, not about whether the GO term holds, and it does. Supporting Evidence: PMID:10066790 We propose that, like the related AP-1, AP-2, and AP-3 complexes, AP-4 plays a role in signal-mediated trafficking of integral membrane proteins in mammalian cells. PMID:33084855 AP-4 associates with the trans-Golgi network (TGN) through interaction with small GTPases of the ARF family and recognizes transmembrane proteins (i.e. cargos) having specific sorting signals in their cytosolic domains. |
| GO:0006895 Golgi to endosome transport | IMP PMID:20230749 Sorting of the Alzheimer's disease amyloid precursor protein... | ACCEPT | Summary: The specific route established for APP: mu4 binds its YKFFE signal and AP-4 carries it from the trans-Golgi network to endosomes, which limits amyloidogenic processing. Reason: Core, and the most precisely characterized AP-4 itinerary. The evidence is a perturbation in human cells with a defined molecular lesion - mutations in the mu4 binding site and in the APP signal - and a measured consequence for both localization and processing. Supporting Evidence: PMID:20230749 These findings demonstrate that APP and AP-4 engage in a distinct type of signal-adaptor interaction that mediates transport of APP from the trans-Golgi network (TGN) to endosomes, thereby reducing amyloidogenic processing of the protein. file:human/AP4M1/AP4M1-uniprot.txt R->D: Strongly reduced interaction with APP. |
| GO:0008104 intracellular protein localization | IEA GO_REF:0000120 | KEEP AS NON CORE | Summary: Automatic route to a very general process term, from a reproducible ARBA rule plus orthology projection from mouse Ap4m1, which holds the term by IMP. Reason: The inference is mechanically sound on both legs and the conclusion is true, but intracellular protein localization is a near-root process term that says only that the protein is involved in putting proteins somewhere. Kept as a correct parent rather than treated as informative. Propagation Review Root cause: NO FAILURE NON CORE Sources checked: ARBA:ARBA00028253 · ARBA rule asserting GO:0008104 SUPPORTS TRANSFER Fetched from rest.uniprot.org/arba. 126 condition sets over FunFam, InterPro, PANTHER and taxon conditions; exactly one is fully satisfied by AP4M1's signatures, so the rule reproduces on this protein. UniProtKB:Q9JKC7 · AP4M1_MOUSE, mouse Ap4m1 SUPPORTS TRANSFER The orthology donor. QuickGO shows GO:0008104 IMP on Q9JKC7 from PMID:18341993, the same neuronal AP-4 study that grounds its GO:0006605. ensembl:ENSMUSP00000019662 · Ensembl mouse protein identifier used by the Compara projection SUPPORTS TRANSFER The projection's internal handle for the same mouse gene product; not an accession and so not resolvable in any local family index, but co-listed with Q9JKC7, which is. Supporting Evidence: file:human/AP4M1/AP4M1-bioinformatics/RESULTS.md The two protein-trafficking rules reproduce cleanly: each has exactly one condition set satisfied by AP4M1, so those IEA rows follow from the rule as published. |
| GO:0008104 intracellular protein localization | ISS GO_REF:0000024 | KEEP AS NON CORE | Summary: Curator transfer from mouse Ap4m1, whose own annotation is an IMP from the AP-4 knockout study showing mislocalization of AMPA receptor-TARP complexes into axons. Reason: A sound transfer between true orthologs, 93.8% identical over a global alignment, and the right evidence code for a mouse-derived claim on a human gene - no human IMP should be manufactured for the somatodendritic-sorting result. The term itself is too general to be a core function, so it is kept as a correct parent. Propagation Review Root cause: NO FAILURE NON CORE Sources checked: UniProtKB:Q9JKC7 · AP4M1_MOUSE, mouse Ap4m1 SUPPORTS TRANSFER QuickGO shows GO:0008104 IMP from PMID:18341993 on this entry. The gene disrupted in that paper was the AP-4 beta subunit, but the same study separately disrupted the AP-4-TARP interaction, and AP-4 is an obligate heterotetramer, so the MGI curator's assignment to Ap4m1 is not a mis-attribution. Supporting Evidence: PMID:18341993 AP-4 indirectly associated with the AMPA receptor via TARPs, and the specific disruption of the interaction between AP-4 and TARPs caused the mislocalization of endogenous AMPA receptors in axons of wild-type neurons. |
| GO:0008320 transmembrane protein transporter activity | TAS Reactome:R-HSA-5229111 | REMOVE | Summary: Asserts that AP4M1 moves a protein from one side of a membrane to the other. It does not. AP-4 selects cargo into vesicles; nothing crosses a bilayer. Reason: The GO definition is explicit - enables the transfer of a protein from one side of a membrane to the other - and that is a translocase activity of the kind performed by the Sec61 or TOM complexes. AP-4 is a peripheral membrane coat on the cytosolic face of the trans-Golgi network; its cargo stays in the membrane throughout, and the transport it mediates is vesicular. The cached Reactome summary for this event describes exactly that - signal recognition by mu4 followed by transport of APP to endosomes - with no translocation step anywhere in it. The molecular function is an artefact of Reactome's reaction being classed as transport, and it is additionally the upstream cause of the spurious GO:0071806 row. Supporting Evidence: Reactome:R-HSA-5229111 AP-4 complex subunit mu-1 (AP4M1) can recognise and interact with tyrosine-based (YXXphi) sorting signals found within the cytoplasmic tails of integral membrane proteins such as the amyloid precursor protein (APP), implicated in Alzheimer's disease. file:human/AP4M1/AP4M1-uniprot.txt AP-4 forms a non clathrin-associated coat on vesicles departing the |
| GO:0016192 vesicle-mediated transport | IEA GO_REF:0000002 | KEEP AS NON CORE | Summary: The second of the three terms these pan-mu signatures carry, and like the first it is a genuine family-wide property. Broad, correct, uninformative. Reason: Every AP complex mu subunit participates in vesicle-mediated transport, so a family-level signature is an appropriate instrument and the transfer is safe. It is graded non-core because it is a high-level parent of the post-Golgi and Golgi-to-endosome terms that describe what AP-4 specifically does. Propagation Review Root cause: NO FAILURE NON CORE Sources checked: InterPro:IPR001392 · Clathrin adaptor, mu subunit SUPPORTS TRANSFER Family-level signature. This term is one of the two in its mapping that hold for the whole family rather than only for the clathrin-associated branches. InterPro:IPR018240 · Clathrin adaptor, mu subunit, conserved site SUPPORTS TRANSFER Same family, same three-term mapping, same reading. Supporting Evidence: PMID:10436028 Adaptor protein complexes (APs) function as vesicle coat components in different membrane traffic pathways; however, there are a number of pathways for which there is still no candidate coat. |
| GO:0016192 vesicle-mediated transport | NAS PMID:10436028 Characterization of a fourth adaptor-related protein complex... | KEEP AS NON CORE | Summary: Author statement from the founding paper that AP complexes function as vesicle coat components. Correct, general. Reason: Accurate but at the level of the whole adaptor family. Kept as a parent term; the informative content for AP-4 sits in the trans-Golgi-network-export rows. Supporting Evidence: PMID:10436028 Adaptor protein complexes (APs) function as vesicle coat components in different membrane traffic pathways; however, there are a number of pathways for which there is still no candidate coat. |
| GO:0019904 protein domain specific binding | IDA PMID:11139587 Signal-binding specificity of the mu4 subunit of the adaptor... | ACCEPT | Summary: The defining molecular activity of this subunit: mu4 reads short sorting motifs in cytosolic tails, with a preference profile of its own and micromolar affinity. Reason: Core. Of the terms GO currently offers, this is the closest fit for recognition of a sorting motif in a cargo tail, and the curator's choice should stand. It is not a perfect fit - a YXXPhi motif is a short linear signal rather than a folded domain - but the only alternative GO provides, GO:0089710 endocytic targeting sequence binding, is defined for signals that direct internalization by clathrin-coated pits, which is the wrong pathway for AP-4. That gap is recorded in proposed_new_terms rather than papered over with a worse term. Supporting Evidence: PMID:11139587 Statistical analyses of the results revealed that mu4 prefers aspartic acid at position Y+1, proline or arginine at Y+2, and phenylalanine at Y-1 and Y+3 (phi). PMID:11139587 These experiments showed that mu4 recognized the tyrosine signal from the human lysosomal protein LAMP-2, HTGYEQF. PMID:11139587 Using surface plasmon resonance measurements, we determined the apparent dissociation constant for the mu4-YXXphi interaction to be in the micromolar range. |
| GO:0019904 protein domain specific binding | IDA PMID:26544806 Association between Rare Variants in AP4E1, a Component of I... | ACCEPT | Summary: The same activity demonstrated on a second cargo, NAGPA, in the stuttering-genetics paper. Reason: Core, and a good illustration of why a title should not decide a review: the paper is titled and framed around AP4E1 genetics, but it contains a mu4-specific biochemical experiment, which is what this row records. The curator read the full text; the abstract alone already states the mu4-NAGPA result. Supporting Evidence: PMID:26544806 We found that the μ4 subunit of AP-4 interacts with NAGPA, an enzyme involved in the synthesis of the mannose 6-phosphate signal that targets acid hydrolases to the lysosome |
| GO:0030124 AP-4 adaptor complex | IBA GO_REF:0000033 | ACCEPT | Summary: Sound and core. Placed on the AP-4 mu node with two gene-level seeds, Arabidopsis AP4M and human AP4M1, both with their own experimental complex-membership annotations. Reason: The node is AP-4-specific, so no other AP complex's membership is being carried across, and the seed list reaches from plants to humans, which is the right span for a complex whose subunits are conserved across eukaryotes. The two gene-level seeds are a short list, but donor count is not a measure of strength: what matters is that the assertion is placed at the node where the AP-4 complex arose and that the target is a descendant of it. Four independent human IDA rows for the same term sit alongside. Propagation Review Root cause: NO FAILURE CORE Sources checked: AGI_LocusCode:AT4G24550 · Q9SB50, AP4M_ARATH, Arabidopsis AP-4 complex subunit mu SUPPORTS TRANSFER Resolved through the UniProt cross-reference for this AGI locus code (3 hits; the reviewed entry is Q9SB50). QuickGO shows GO:0030124 IPI from PMID:26546666 with three Arabidopsis partners, so plant AP-4 complex membership is itself experimentally established rather than inferred. PANTHER:PTN000242612 · PTHR10529 AP-4 mu IBD node, seeded by Arabidopsis AP4M and human AP4M1 SUPPORTS TRANSFER The node. The same node carries the GO:0006605 and GO:0090160 IBDs for this family, and the family slice shows sibling nodes carrying the AP-2 (PTN000242370) and AP-3 (PTN002237676) complex terms, so the four complexes are kept apart in the tree. UniProtKB:O00189 · AP4M1, the target itself SUPPORTS TRANSFER The target's own four IDA rows for AP-4 adaptor complex are among the descendant evidences behind the IBD. Expected for the best-characterized member of the clade. 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. file:human/AP4M1/AP4M1-uniprot.txt -!- SUBUNIT: Adaptor protein complex 4 (AP-4) is a heterotetramer composed |
| GO:0030124 AP-4 adaptor complex | IDA PMID:10066790 AP-4, a novel protein complex related to clathrin adaptors. | ACCEPT | Summary: One of the two founding demonstrations that mu4 is a subunit of a four-member adaptor complex, by gel filtration, sedimentation velocity and immunoprecipitation. Reason: Core, and the defining cellular-component statement for this gene. The complex was identified biochemically rather than inferred, and the subunit list matches the one UniProt still records. 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. |
| GO:0030124 AP-4 adaptor complex | IDA PMID:11802162 AP-4 binds basolateral signals and participates in basolater... | ACCEPT | Summary: Complex membership confirmed in the MDCK basolateral-sorting study, which also enumerates the subunit composition of all four AP complexes. Reason: Core, and independent of the two 1999 papers. Note the species: this work is in canine MDCK cells, which is why the dog entry E2RED8 is the donor for the human ISS rows that derive from it. Supporting Evidence: PMID:11802162 Four adaptors, AP-1 (beta 1, gamma, mu 1A or mu 1B, sigma 1), AP-2 (beta 2, alpha, mu 2, sigma 2), AP-3 (beta 3 , delta, mu 3, sigma 3) or AP-4 (beta 4, epsilon, mu 4, sigma 4), have been characterized. |
| GO:0030124 AP-4 adaptor complex | IDA PMID:29180427 AP-4 mediates export of ATG9A from the trans-Golgi network t... | ACCEPT | Summary: Complex membership in the ATG9A study, which worked with AP-4 as a unit and established its cargo. Reason: Core. This paper's framing - AP-4 as a member of the heterotetrameric adaptor family whose four subunit genes all cause the same disease - is exactly the obligate complex that makes subunit-level and complex-level evidence interchangeable for this gene. Supporting Evidence: PMID:29180427 AP-4 is a member of the heterotetrameric adaptor protein (AP) complex family involved in protein sorting in the endomembrane system of eukaryotic cells. |
| GO:0030124 AP-4 adaptor complex | NAS PMID:10436028 Characterization of a fourth adaptor-related protein complex... | ACCEPT | Summary: Author statement from the second founding paper, where the four-subunit composition was shown by co-immunoprecipitation and yeast two-hybrid. Reason: Correct and redundant with four other rows for the same term; the weaker evidence code changes nothing, and the underlying experiments in this paper are direct. 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:0030131 clathrin adaptor complex | IEA GO_REF:0000002 | REMOVE | Summary: The third term carried by the pan-mu InterPro signatures, and the one that does not transfer. AP-4 is a non-clathrin coat, and GO itself places AP-4 outside the clathrin adaptor complex branch. Reason: This is a demonstrably wrong interpro2go mapping for this protein, wrong on two independent checks. Biologically, the founding characterization put AP-4 on nonclathrin-coated vesicles by immunogold electron microscopy, and the field has described the AP-4 coat as non-clathrin ever since. Ontologically, GO:0030131 has exactly two children, GO:0030121 AP-1 adaptor complex and GO:0030122 AP-2 adaptor complex, and GO:0030124 AP-4 adaptor complex is not among its descendants - AP-4's ancestry runs through GO:0030119 AP-type membrane coat adaptor complex instead. So the annotation contradicts the ontology's own placement of the complex this protein belongs to. The signatures are family-level and match every mu subunit; the error is in assuming that a property of the AP-1 and AP-2 branches is a property of the family. Propagation Review Root cause: PROPAGATION BAD Failure modes: COMPARTMENT OR COMPLEX MISMATCH Sources checked: InterPro:IPR001392 · Clathrin adaptor, mu subunit SUPPORTS SOURCE BUT NOT TARGET The signature correctly identifies AP4M1 as an AP complex mu subunit - UniProt agrees the protein belongs to the adaptor complexes medium subunit family. What does not follow is the clathrin-specific complex term: the signature spans mu1, mu2, mu3 and mu4, and only the first two are in clathrin adaptor complexes. InterPro:IPR018240 · Clathrin adaptor, mu subunit, conserved site SUPPORTS SOURCE BUT NOT TARGET Same family, same mapping, same failure. Its other two mapped terms, GO:0006886 and GO:0016192, are retained on this gene; only the complex term is removed. 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:26542808 The heterotetrameric (ϵ-β4-μ4-σ4) complex adaptor protein 4 (AP-4) is a component of a non-clathrin coat involved in protein sorting at the trans-Golgi network (TGN). file:human/AP4M1/AP4M1-uniprot.txt AP-4 forms a non clathrin-associated coat on vesicles departing the |
| GO:0031904 endosome lumen | TAS Reactome:R-HSA-5229111 | REMOVE | Summary: Places the adaptor inside the endosome. AP4M1 is a cytosolic, peripheral-membrane coat subunit; the lumen is where its cargo ends up in Reactome's model, not where the protein is. Reason: GO:0031904 is the volume enclosed by the endosome membrane. Nothing puts AP4M1 there: UniProt localizes it to the trans-Golgi network membrane as a peripheral membrane protein and to early endosomes, and the entire mechanism - ARF1-dependent recruitment, cargo tail binding, coat assembly - happens on the cytosolic face. The row is an artefact of the Reactome event being named for the destination compartment of the APP cargo, and it should not survive into GO as a location of the adaptor. Supporting Evidence: 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). file:human/AP4M1/AP4M1-uniprot.txt {ECO:0000305|PubMed:32073997}. Early endosome |
| GO:0032588 trans-Golgi network membrane | TAS Reactome:R-HSA-5229111 | ACCEPT | Summary: The membrane-level refinement of the core localization, and the most precise cellular-component term on this gene. Reason: Core, and better than its parent GO:0005802: UniProt's subcellular location line is specifically trans-Golgi network membrane with peripheral membrane protein, which is exactly what a coat adaptor recruited by ARF1 to the cytosolic face should be. Unlike the other two terms this Reactome event carries, this one is right. Supporting Evidence: file:human/AP4M1/AP4M1-uniprot.txt -!- SUBCELLULAR LOCATION: Golgi apparatus, trans-Golgi network membrane 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-5229132 | ACCEPT | Summary: The same membrane localization from the binding reaction rather than the transport reaction. Correct. Reason: Core, and the cached summary for this Reactome event is an accurate statement of the biology - the four subunits, mu adaptins reading tyrosine-based signals, and APP as a cargo. This is the one of the two AP-4/APP Reactome events whose GO output is entirely sound. Supporting Evidence: Reactome:R-HSA-5229132 AP4 consists of four subunits; two large adaptins (AP4E1 and AP4B1), a medium adaptin (AP4M1) and a small adaptin (AP4S1). file:human/AP4M1/AP4M1-uniprot.txt -!- SUBCELLULAR LOCATION: Golgi apparatus, trans-Golgi network membrane |
| GO:0035615 clathrin-cargo adaptor activity | IBA GO_REF:0000033 | MODIFY | Summary: The one IBA on this gene that does not survive. The IBD sits on the deep pan-mu node PTN000055849 and every one of its seeds is a clathrin-adaptor mu subunit - Drosophila AP-1mu and AP-2mu, Dictyostelium AP-1 mu, and in the current family slice human AP1M2. AP-4 works without clathrin and outside endocytosis, which is what this term's definition requires. Reason: The failure is in the node placement rather than in the source annotations. Each seed has its own experimental grounding for the term in QuickGO - AP1M2 IDA PMID:10338135, AP2M1 IDA PMID:23676497, Drosophila AP-1mu IMP PMID:22389401, AP-2mu IMP PMID:20226669, Dictyostelium apm1 IDA PMID:12802059 - and every one of them is a genuine clathrin adaptor. Placing the assertion at the node ancestral to all four AP complexes then carries it into the AP-3 and AP-4 clades, which are non-clathrin coats. AP4M1 has GO:0035615 by IBA only, with no experimental row for it in any organism, so nothing on the target side rescues it. The term's definition is doubly disqualifying: it requires bringing together a cargo protein with clathrin, responsible for the formation of endocytic vesicles, which commits both to clathrin, that AP-4 demonstrably lacks, and to endocytic vesicles, when AP-4 is a trans-Golgi network export coat. The term is current, not obsolete - it was renamed from clathrin adaptor activity in December 2025 and its clathrin definition dates from 2019. The correction is a generalisation rather than a deletion, because the cargo-adaptor essence does transfer even though the clathrin specialisation does not: GO:0035615's only parent, GO:0140312 cargo adaptor activity, states the coat-cargo bridging without the clathrin or endocytic commitments, and mu4 supplies exactly that function within the AP-4 coat. On an obligate-heterotetramer subunit it is a contributed activity, which is how core_functions records it. Propagation Review Root cause: PROPAGATION BAD Failure modes: COMPARTMENT OR COMPLEX MISMATCH Sources checked: FB:FBgn0024833 · O62531, Drosophila AP-1mu SUPPORTS SOURCE BUT NOT TARGET Resolved through the UniProt cross-reference for this FlyBase identifier (1 hit, TrEMBL). QuickGO shows GO:0035615 IMP from PMID:22389401. A true clathrin adaptor mu subunit; the term is right for it and wrong for AP-4. FB:FBgn0263351 · O62530, Drosophila AP-2mu SUPPORTS SOURCE BUT NOT TARGET Resolved through the UniProt cross-reference (1 hit, TrEMBL). QuickGO shows GO:0035615 IMP from PMID:20226669. The endocytic adaptor mu subunit, i.e. the protein the term's endocytic-vesicle clause was written for. PANTHER:PTN000055849 · PTHR10529 pan-mu-adaptin IBD node SUPPORTS SOURCE BUT NOT TARGET The node, and the locus of the error. It is ancestral to the mu subunits of all four AP complexes, so an assertion placed here reaches AP-3 and AP-4 as well as AP-1 and AP-2. The same node's GO:0005802 IBD is fine because its seed list includes AP-4 subunits; this one's does not include a single one. dictyBase:DDB_G0289247 · Q54HS9, AP1M_DICDI, Dictyostelium AP-1 complex subunit mu SUPPORTS SOURCE BUT NOT TARGET Resolved through the UniProt cross-reference for this dictyBase identifier (1 hit, Swiss-Prot). QuickGO shows GO:0035615 IDA from PMID:12802059. A third clathrin-adaptor donor; the seed list contains no AP-4 subunit at all. Proposed replacements: cargo adaptor activity 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:26542808 The heterotetrameric (ϵ-β4-μ4-σ4) complex adaptor protein 4 (AP-4) is a component of a non-clathrin coat involved in protein sorting at the trans-Golgi network (TGN). PMID:20230749 Biochemical and X-ray crystallographic analyses reveal that the properties of the APP sequence and the location of the binding site on mu4 are distinct from those of other signal-adaptor interactions. file:human/AP4M1/AP4M1-bioinformatics/RESULTS.md F255 has a phenylalanine counterpart in every mu subunit tested, including all five clathrin-adaptor paralogs, so it is a family-wide feature and carries no mu4-specific information. |
| GO:0070062 extracellular exosome | HDA PMID:19056867 Large-scale proteomics and phosphoproteomics of urinary exos... | MARK AS OVER ANNOTATED | Summary: Detection of AP4M1 peptides in shotgun proteomics of urinary exosomes. Reproducible as a detection, uninformative as a localization. Reason: A cytosolic coat subunit appearing in an exosome preparation says more about the preparation than about the protein. Nothing in the AP-4 literature places the complex in the extracellular space or assigns it a function there, and the annotation competes with three experimentally supported locations on the same gene. Kept on the record as a high-throughput observation, marked as the over-annotation it is. Supporting Evidence: PMID:19056867 Here, we used LC-MS/MS to profile the proteome of human urinary exosomes. |
| GO:0071806 protein transmembrane transport | IEA GO_REF:0000108 | REMOVE | Summary: A mechanically correct inter-ontology inference applied to a molecular function that AP4M1 should not have. The process inherits the error in the source. Reason: The logical link from transmembrane protein transporter activity to protein transmembrane transport is not in question - a protein that translocates other proteins across a membrane is by definition involved in that process. What is in question is the premise. AP-4 is a coat adaptor: it selects cargo into vesicles budding from the trans-Golgi network, and its cargo never crosses a bilayer. With the GO:0008320 row removed, this one has no basis, so it goes with it. Propagation Review Root cause: SOURCE BAD Failure modes: ROLE CONFLATION Sources checked: GO:0008320 · transmembrane protein transporter activity, the source annotation on this gene SOURCE BAD Not a gene product but the GO term this inference reads from. Its own row on AP4M1 comes from Reactome:R-HSA-5229111 and is rejected in this review: a cargo-selecting coat subunit has been cast as the translocase that moves the protein, which is the conflation of a specificity subunit with the core machinery. Remove the premise and the conclusion has no support. Supporting Evidence: file:human/AP4M1/AP4M1-uniprot.txt AP-4 forms a non clathrin-associated coat on vesicles departing the PMID:33084855 AP-4 associates with the trans-Golgi network (TGN) through interaction with small GTPases of the ARF family and recognizes transmembrane proteins (i.e. cargos) having specific sorting signals in their cytosolic domains. |
| GO:0090160 Golgi to lysosome transport | IBA GO_REF:0000033 | KEEP AS NON CORE | Summary: Placed on the AP-4 mu node and seeded by AP4M1 alone, so this IBA is a propagation of the human Tac-chimera result out to the AP-4 mu clade. The node placement is right; the underlying evidence is thinner than the term suggests. Reason: The propagation is mechanically correct - the node is AP-4-specific and the target is trivially inside it, being the sole seed - so there is no failure to report in the phylogeny. The reservation is about the strength of what is being propagated: the human IDA behind it is an engineered Tac chimera bearing a mu4-selected signal, and no endogenous lysosomal cargo of AP-4 has been identified since. A single-seed node is not weak because it has one donor; it is weak here because the one donor's evidence is a reporter assay. Graded non-core rather than removed, because the chimera result is real and the term is not contradicted by anything. Propagation Review Root cause: NO FAILURE NON CORE Sources checked: PANTHER:PTN000242612 · PTHR10529 AP-4 mu IBD node, seeded by human AP4M1 SUPPORTS TRANSFER The node. AP-4-specific, the same node that carries the well-supported GO:0030124 and GO:0006605 IBDs. The family slice shows this IBD has a single seed, so the phrase one of the IBD seeds would be wrong here: AP4M1 is the only one. UniProtKB:O00189 · AP4M1, the target itself and the sole seed of this IBD SUPPORTS TRANSFER The target's own GO:0090160 IDA from PMID:11139587 is the descendant evidence behind the node. Self-seeding is expected and not circular; the caveat here is the nature of that IDA - a Tac reporter chimera - not the fact that the target appears in its own WITH/FROM. Supporting Evidence: PMID:11139587 we constructed a Tac chimera bearing a mu4-specific YXXphi signal. This chimera was targeted to the endosomal-lysosomal system without being internalized from the plasma membrane. PMID:30262884 We identify three transmembrane cargo proteins, ATG9A, SERINC1 and SERINC3, and two AP-4 accessory proteins, RUSC1 and RUSC2. |
| GO:0090160 Golgi to lysosome transport | IDA PMID:11139587 Signal-binding specificity of the mu4 subunit of the adaptor... | KEEP AS NON CORE | Summary: The human experiment that seeds the IBA above: a Tac chimera carrying a mu4-binding signal reaches the endosomal-lysosomal system from the biosynthetic route. Reason: A real direct assay, and the curator read the full text, so it stays. It is graded non-core for the same reason as the paired GO:0006622 row: the cargo is engineered, the signal was selected by mu4 rather than found on a validated AP-4 cargo, and the route that later work established for AP-4 runs to the peripheral cytoplasm and autophagosome-associated vesicles rather than to lysosomes. Supporting Evidence: PMID:11139587 we constructed a Tac chimera bearing a mu4-specific YXXphi signal. This chimera was targeted to the endosomal-lysosomal system without being internalized from the plasma membrane. PMID:11139587 To gain insight into a possible role of AP-4 in intracellular trafficking, we constructed a Tac chimera bearing a mu4-specific YXXphi signal. |
| GO:1903361 protein localization to basolateral plasma membrane | ISS GO_REF:0000024 | KEEP AS NON CORE | Summary: Curator transfer from dog AP4M1, whose own annotation is an IMP from the MDCK knockdown experiment - and MDCK cells are canine, so donor species and experiment species agree. Reason: A textbook-clean ISS: the donor carries GO:1903361 IMP from PMID:11802162, the experiment was a mu4-specific knockdown in the donor's own species, and dog and human mu4 are 94.9% identical with the cargo-binding residues F255 and R283 present at the same positions in both. It is graded non-core rather than core because epithelial basolateral polarity is one context in which AP-4 sorting has been assayed, not the function AP-4 exists for; the ubiquitously required activity is trans-Golgi network export of ATG9A. Note also that the human protein has no independent basolateral-sorting experiment of its own. Propagation Review Root cause: NO FAILURE NON CORE Sources checked: UniProtKB:E2RED8 · AP4M1_CANLF, dog mu4 SUPPORTS TRANSFER QuickGO shows GO:1903361 IMP from PMID:11802162 on this entry. The paper's experiment is depletion of mu 4 protein in MDCK cells, which are canine, so the annotation sits on the species that was actually assayed rather than being itself a transfer. Supporting Evidence: PMID:11802162 Furthermore, in MDCK cells with depleted mu 4 protein levels, several basolateral proteins are mis-sorted to the apical surface, showing that AP-4 participates in basolateral sorting in epithelial cells. file:human/AP4M1/AP4M1-bioinformatics/RESULTS.md F255 and R283 are present in mouse, rat and dog mu4 at the same native positions. |
| GO:0031267 small GTPase binding | IPI PMID:11707398 Functional and physical interactions of the adaptor protein ... | NEW | Summary: Proposed. mu4 binds ARF1 directly, through the same signal-binding domain that reads cargo tails, and does so independently of ARF1's nucleotide state. Nothing in the 53-row GOA set for this gene records any ARF or GTPase interaction. Reason: This is a real gap rather than a refinement. ARF-dependent recruitment to the trans-Golgi network is the mechanism that puts the AP-4 coat on a membrane at all - it is why the localization is brefeldin A-sensitive, a fact both founding papers reported - and the study that worked it out showed that two subunits contact ARF1 directly, epsilon and mu4, with different nucleotide requirements. mu4's is the constitutive, nucleotide-independent contact. GO:0031267, binding to a small monomeric GTPase, is the right level: GO has no ARF-specific binding term, and GO:0030742 GTP-dependent protein binding would be positively wrong for mu4, since nucleotide independence is the paper's explicit finding, however apt it would be for the epsilon subunit. Coded IPI with the human ARF1 entry P84077 as the partner; the experiments are direct binding assays with recombinant proteins, not a perturbation. Supporting Evidence: PMID:11707398 In addition, we demonstrate a direct interaction of the epsilon and mu4 subunits of AP-4 with ARF1. PMID:11707398 epsilon binds only to ARF1-GTP and requires residues in the switch I and switch II regions of ARF1. In contrast, mu4 binds equally well to the GTP- and GDP-bound forms of ARF1 and is less dependent on switch I and switch II residues. PMID:11707398 We map the interacting regions on the AP-4 subunits to the trunk region of ε and the signal-binding domain of µ4. |
| GO:0030120 vesicle coat | IDA PMID:10436028 Characterization of a fourth adaptor-related protein complex... | NEW | Summary: Proposed. Immunogold electron microscopy in the founding characterization places AP-4 on coated vesicles in the trans-Golgi network region, and later work isolated AP-4-derived vesicles carrying ATG9A. The gene has no vesicle coat term. Reason: AP4M1 already carries GO:0030124, which sits under GO:0030117 membrane coat, but GO:0030120 vesicle coat is a separate specialization and is not implied by AP-4 adaptor complex - checked against the QuickGO ancestor list for GO:0030124, which contains GO:0030117 and GO:0030119 but not GO:0030120. The direct observation supporting it is the immunogold EM, and it is corroborated by the later isolation of AP-4-derived, ATG9A-positive vesicles. Adding it also states positively what the removed GO:0030131 row got wrong: AP-4 is on a vesicle coat, just not a clathrin one. 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:30262884 RUSC2 facilitates the transport of AP-4-derived, ATG9A-positive vesicles from the trans-Golgi network to the cell periphery. |
Loading supporting content…
Download this section (compressed HTML)Q: For GO_Central PAINT curators of PTHR10529: should the GO:0035615 IBD at PTN000055849 move down to the AP-1/AP-2 clades, or gain an IRD at the AP-4 node PTN000242612? All four of its seeds are AP-1 or AP-2 mu subunits, the term's definition commits to clathrin and to endocytic vesicles, and the same leak will reach every AP-4 and AP-3 mu orthologue in the family, not just the human gene. The node already demonstrates the mechanism: its GO:0005802 IBD is correctly blocked at the AP-2 clade by an IRD at PTN000242370.
Q: For InterPro: should the interpro2go mappings IPR001392 and IPR018240 to GO:0030131 clathrin adaptor complex exist? Both signatures match the mu subunits of all four AP complexes, but GO:0030131 has only AP-1 and AP-2 among its descendants, so the mapping is guaranteed to over-call on every mu3 and mu4 protein. The other two terms these signatures carry, GO:0006886 and GO:0016192, are genuine family properties.
Q: For Reactome: the event "AP4 transports APP from trans-Golgi network to endosome lumen" yields GO:0008320 transmembrane protein transporter activity and GO:0031904 endosome lumen on AP4M1. Can the reaction be re-typed as vesicular transport, so that a coat adaptor is not recorded as a translocase and is not placed inside the compartment its cargo is delivered to?
Q: For the Bonifacino group or whoever next works on AP-4: is the mu4-ARF1 contact, which is nucleotide-independent and maps to the signal-binding domain, on the same surface that binds cargo signals? If so, ARF1 and cargo would compete, which would be a different model of coat assembly from the AP-1/AP-2 paradigm.
Q: For GO ontology editors: would a general parent for GO:0089710 endocytic targeting sequence binding be accepted, so that recognition of a tyrosine-based sorting signal outside the endocytic pathway has a term? Three of the four AP mu subunits, and the GGAs, currently have nowhere to go but GO:0019904.
Experiment: Alanine-scan the ATG9A cytosolic determinant against recombinant mu4 by isothermal titration calorimetry or surface plasmon resonance, then test the mapped motif for AP-4-dependent export in an AP4M1-null line. This is the experiment that would connect the measured mu4 signal preference to the one cargo that matters for the disease, and its absence is the main knowledge gap recorded here.
Experiment: Reconstitute AP-4 with mu4 carrying F255A or R283D in AP4M1-null cells and score ATG9A distribution, LC3B lipidation and APP processing in parallel. This separates cargo-specific from coat-wide effects, and would show whether the crystallographically defined APP site is used for ATG9A as well.
Experiment: Perform dynamic organellar mapping in isogenic AP4M1-null neurons and fibroblasts derived from the same iPSC line, and ask whether the AP-4-dependent proteome is cell-type-specific. The published unbiased surveys were done in HeLa cells and fibroblasts, so a neuronal cargo set would speak directly to why the disease is neurological.
Experiment: Mutate the mu4 surface that contacts ARF1 within the signal-binding domain and ask whether AP-4 still becomes membrane-associated and brefeldin A-sensitive, and whether it still binds cargo. A single mutant that loses ARF1 binding while retaining cargo binding, or the converse, would settle whether the two activities share a surface.
Experiment: Determine the structure of the AP-4 core bound to a Hook coiled-coil fragment by crystallography or cryo-EM, to establish whether the FHF site on mu4 overlaps the cargo-signal site and therefore whether coat positioning and cargo capture can occur simultaneously.
What is not known — curated, literature-grounded statements of the open unknowns (the inverse of core functions).
Gap: The sorting signal in ATG9A that mu4 reads has not been defined, so the connection between AP4M1's measured binding specificity and its only established physiological cargo is still missing.
OPEN BIOLOGY MF_DARK
What is known: Firmly established: ATG9A is an AP-4 cargo, identified independently by a targeted search and by unbiased organellar proteomics, and its export from the trans-Golgi network is signal-mediated. Equally firmly established: mu4 has a measurable and idiosyncratic signal preference, obtained from a combinatorial peptide library, and at least one natural motif it binds - the APP YKFFE - has been co-crystallized with the domain, with two required residues identified by mutagenesis. What has not been published is the ATG9A motif itself, or a demonstration that the mu4 residues required for APP binding are also required for ATG9A export.
Significance: Every functional annotation on this gene downstream of cargo selection - autophagosome assembly, trans-Golgi network export, the disease mechanism - assumes that mu4's signal-binding activity is what recognizes ATG9A. If the ATG9A determinant turned out to be read by a different subunit, or not to be a mu4-type motif at all, the molecular function currently credited to AP4M1 would be supporting a process it does not directly select for. It is also the experiment that would justify a specific signal-binding molecular function term rather than the generic one now used.
What would resolve it: Map the ATG9A cytosolic determinant by truncation and alanine scanning, test it against recombinant mu4 in a direct binding assay, and ask whether the F255A and R283D mutants that lose APP binding also fail to export ATG9A in an AP4M1-null background.
Provenance (the field's own admissions):
Gap: Whether AP-4 has an endogenous lysosomal cargo is unresolved, and two GO annotations on this gene depend on the answer.
OPEN BIOLOGYCURATION BP_DARK
What is known: Established: mu4 binds the LAMP-2 tyrosine signal in vitro, and a Tac reporter carrying a mu4-selected signal reaches the endosomal-lysosomal system by the biosynthetic route rather than by endocytosis. Also established: an unbiased survey of proteins whose localization depends on AP-4 returned ATG9A, SERINC1 and SERINC3, with no lysosomal membrane protein among them, and AP-4-deficient cells have not been reported to missort LAMP-2. So the capability is demonstrated and the physiological instance is not.
Significance: GO:0006622 protein targeting to lysosome and GO:0090160 Golgi to lysosome transport are both on this gene, the latter also propagated as an IBA through the AP-4 mu node, and both rest on the same reporter experiment. If no endogenous lysosomal cargo exists, these are annotations of an assay rather than of a function, and the IBA carries that across the whole AP-4 clade. This review grades them non-core for exactly this reason.
What would resolve it: Quantitative comparison of lysosomal membrane protein delivery in AP4M1-null versus wild-type cells, and a test of whether LAMP-2 localization or the composition of the lysosomal membrane changes in patient fibroblasts, which are already available.
Provenance (the field's own admissions):
Gap: How a defect in a ubiquitously expressed trans-Golgi network coat produces a disease that is almost entirely neurological is unexplained.
NARROWING BIOLOGY RESIDUAL_SUBGAP
What is known: Established: AP-4 is expressed in all cell types examined; loss of any one of its four subunits causes the same syndrome; ATG9A missorting is seen in patient fibroblasts as well as in neurons; and in neurons the consequences include impaired axonal autophagosome generation, axonal swellings and aggregate accumulation. Also established: restoring AP4M1 alone rescues AP-4 function in patient fibroblasts and gives dose- and age-dependent benefit in knockout mice. What is not established is why non-neuronal tissues tolerate the same molecular defect.
Significance: The answer determines whether AP-4's core function is best described as a general trans-Golgi network export step that neurons happen to be most dependent on, or as a neuron-specific sorting activity that the standard cell-culture assays only approximate. It also bears on the tissue targeting of the gene therapies now in preclinical and early clinical testing.
What would resolve it: Side-by-side quantification of ATG9A flux, autophagic capacity and cargo missorting in isogenic AP4M1-null neurons, astrocytes and fibroblasts, with attention to whether a parallel export route substitutes for AP-4 in non-neuronal cells; and a test of whether the axon-specific deficit reflects transport distance rather than a distinct sorting requirement.
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)