AP3M1 (mu3A, mu-adaptin 3A) is the medium subunit of the ubiquitously expressed AP-3A adaptor protein complex, a heterotetrameric coat that sorts transmembrane proteins out of tubular early endosomes and the trans-Golgi region and delivers them to lysosomes and to lysosome-related organelles such as melanosomes and platelet dense granules. The complex comprises two large adaptins (delta, encoded by AP3D1, and beta3A or beta3B), one medium adaptin (mu3A or the neuron-enriched mu3B/AP3M2) and one small adaptin (sigma3A or sigma3B); AP3M1 is the medium subunit of the ubiquitous, beta3A-containing variants. Within the complex the mu subunit is the cargo reader: its C-terminal mu homology domain (residues 176-417) carries the pocket that binds tyrosine-based YxxPhi sorting signals in the cytoplasmic tails of membrane cargo such as LAMP1, TGN38 and the 46 kDa mannose-6-phosphate receptor, with sequence preferences that differ from those of the AP-1 and AP-2 medium subunits. Unlike AP-1 and AP-2, AP-3 sits in a constitutively open conformation, is recruited to membranes by the small GTPase Arf1 through its large subunits, and polymerises into tubular carriers without a clathrin lattice. Cargo capture by the mu3A pocket is the step that rigidifies the complex and licenses coat assembly. Loss of AP-3 function causes lysosomal membrane proteins to accumulate at the cell surface and underlies Hermansky-Pudlak syndrome, although the human disease alleles lie in the beta3A and delta subunits rather than in AP3M1. In neurons the protein is additionally reported to be upregulated during homeostatic synaptic scaling, redistributing with AMPA receptors into recycling endosomes.
Definition: A heterotetrameric AP-type membrane coat adaptor complex composed of the delta subunit, the ubiquitously expressed beta3A subunit, the ubiquitously expressed mu3A medium subunit and a sigma3 small subunit; it is the form of AP-3 present in non-neuronal cells and sorts transmembrane cargo from endosomes to lysosomes and lysosome-related organelles.
Justification: GO:0030123 has no children, so nothing in the ontology distinguishes the ubiquitous complex that contains AP3M1 and AP3B1 from the neuronal complex that contains AP3M2 and AP3B2, even though the two have been shown genetically to have distinct and in places opposite functions in brain and are separately modelled by ComplexPortal (CPX-5051 and CPX-5052 versus CPX-5053 and CPX-5055). Without the distinction, neuronal AP-3 phenotypes obtained with the shared delta subunit propagate indiscriminately onto the ubiquitous subunits.
Parent term: AP-3 adaptor complex
Mappings:
Supporting Evidence:
Definition: Binding to a tyrosine-based sorting signal conforming to the YxxPhi motif, where x is any amino acid and Phi is a bulky hydrophobic residue, in the cytoplasmic domain of a transmembrane protein.
Justification: The medium subunits of the AP complexes, and stonins, all read YxxPhi motifs, and their differing residue preferences at the x and Phi positions are the basis of cargo selectivity between AP-1, AP-2 and AP-3. GO currently forces this onto GO:0005048 'signal sequence receptor activity', whose four existing children are all organelle-import targeting-sequence receptors (ER, mitochondrial, chloroplast, peroxisomal); a sibling for cytoplasmic-tail sorting motifs would let the adaptor-subunit activity be stated precisely and would separate it from the acidic dileucine motif activity that resides on the sigma and large subunits. Note on naming: GO:0005048's current primary label in GO is 'signal sequence receptor activity'; this document writes its exact synonym 'signal sequence binding' in machine-readable term slots because that is the label in the repository's committed ontology cache, and the current GO label should win on export.
Parent term: signal sequence binding
Supporting Evidence:
Definition: Formation of a macromolecular complex between proteins of the AP-3 adaptor complex and the proteins and/or lipoproteins that are to be transported by the resulting vesicular or tubular carrier.
Justification: The existing AP-3 term, GO:0035654, is a child of GO:0035652 'clathrin-coated vesicle cargo loading' and therefore commits to a clathrin coat that AP-3 has now been shown not to need: reconstituted AP3:ARF1 builds coated tubular carriers with no clathrin lattice. Its own definition already hedges the point in prose. A clathrin-neutral child of GO:0035459 would let AP-3 cargo loading be annotated at full specificity without asserting the coat type, which is currently only achievable by generalising all the way to the grandparent.
Parent term: vesicle cargo loading
Supporting Evidence:
| GO Term | Evidence | Action | Reason |
|---|---|---|---|
| GO:0005515 protein binding | IPI PMID:19116314 Localization to mature melanosomes by virtue of cytoplasmic ... | MARK AS OVER ANNOTATED | Summary: Protein binding to OCA2 (Q04671, the P protein), and the only low-throughput entry among the twenty-three GO:0005515 rows - assigned by UniProt rather than IntAct, from a study of how OCA2 reaches melanosomes. In GOA the reference annotates four gene products: AP1G1, AP2A1, OCA2 and AP3M1, so it is a small, adaptor-focused projection rather than a screen. The motifs the paper assays are acidic dileucine signals in the OCA2 cytoplasmic N-terminus, and it reports that the two signals interact differentially with cytoplasmic adaptors known to traffic other proteins to melanosomes. Reason: Biologically the most interesting of the binding rows, and still not convertible into an informative molecular function on the evidence available. Acidic dileucine signals are read by the sigma3/delta hemicomplex of AP-3; the mu subunit's site is the tyrosine YxxPhi pocket, and the two are distinct binding sites on distinct subunits. The cached record for this paper is abstract-only, so I cannot see which AP-3 construct was used or whether an additional tyrosine-motif experiment sits in the full text, and I am not going to assert a mis-attribution I cannot check - the curator read the whole paper. What I can say is that converting this row into a cargo-recognition MF would attribute to mu3A a motif specificity the abstract assigns to dileucine signals. Kept, marked over-annotated, with the AP-3 role in melanosomal cargo delivery captured instead by the melanosome assembly row. Supporting Evidence: PMID:19116314 Steady-state melanosomal localization requires a conserved consensus acidic dileucine-based sorting motif within the cytoplasmic N-terminal region of OCA2 PMID:19116314 The two dileucine signals physically interact in a differential manner with cytoplasmic adaptors known to function in trafficking other proteins to melanosomes PMID:39705307 This closed form of the complex occludes the two known binding sites for transmembrane cargo: tyrosine-based YxxΦ motifs (where x is any amino acid and Φ is a bulky, hydrophobic amino acid) on the μ subunit (12, 13) |
| GO:0005515 protein binding | IPI PMID:25416956 A proteome-scale map of the human interactome network. | MARK AS OVER ANNOTATED | Summary: Bare protein binding to CRMP1 (Q14194), from a systematic yeast two-hybrid map of ~14,000 human binary interactions (HI-II-14). A high-throughput screening hit curated into IntAct, with no AP-3-specific follow-up. Reason: GO:0005515 says nothing about what AP3M1 does, and this partner does not let a more informative molecular function be substituted: it is not an AP-3 subunit, not a known AP-3 cargo whose sorting was traced, and there is no follow-up study naming AP3M1 and this protein together. Two of the informative alternatives are ruled out specifically - the mu3A cargo site reads YxxPhi motifs in cytoplasmic tails, which is proposed separately as a GO:0005048 row from the peptide-binding and structural literature, and the small-GTPase rows on this gene come from a different, complex-level experiment. Where UniProt records an NbExp value for one of these pairs it counts replicates within these same datasets, not independent studies. Marked over-annotated: reproducible, uncharacterised, and not removable, since nothing contradicts the pairing. Supporting Evidence: PMID:25416956 Here, we describe a systematic map of ?14,000 high-quality human binary protein-protein interactions |
| GO:0005515 protein binding | IPI PMID:25416956 A proteome-scale map of the human interactome network. | MARK AS OVER ANNOTATED | Summary: Bare protein binding to AGTRAP (Q6RW13), from a systematic yeast two-hybrid map of ~14,000 human binary interactions (HI-II-14). A high-throughput screening hit curated into IntAct, with no AP-3-specific follow-up. Reason: GO:0005515 says nothing about what AP3M1 does, and this partner does not let a more informative molecular function be substituted: it is not an AP-3 subunit, not a known AP-3 cargo whose sorting was traced, and there is no follow-up study naming AP3M1 and this protein together. Two of the informative alternatives are ruled out specifically - the mu3A cargo site reads YxxPhi motifs in cytoplasmic tails, which is proposed separately as a GO:0005048 row from the peptide-binding and structural literature, and the small-GTPase rows on this gene come from a different, complex-level experiment. Where UniProt records an NbExp value for one of these pairs it counts replicates within these same datasets, not independent studies. Marked over-annotated: reproducible, uncharacterised, and not removable, since nothing contradicts the pairing. Supporting Evidence: PMID:25416956 Here, we describe a systematic map of ?14,000 high-quality human binary protein-protein interactions |
| GO:0005515 protein binding | IPI PMID:25416956 A proteome-scale map of the human interactome network. | MARK AS OVER ANNOTATED | Summary: Bare protein binding to TRIM9 (Q9C026), from a systematic yeast two-hybrid map of ~14,000 human binary interactions (HI-II-14). A high-throughput screening hit curated into IntAct, with no AP-3-specific follow-up. Reason: GO:0005515 says nothing about what AP3M1 does, and this partner does not let a more informative molecular function be substituted: it is not an AP-3 subunit, not a known AP-3 cargo whose sorting was traced, and there is no follow-up study naming AP3M1 and this protein together. Two of the informative alternatives are ruled out specifically - the mu3A cargo site reads YxxPhi motifs in cytoplasmic tails, which is proposed separately as a GO:0005048 row from the peptide-binding and structural literature, and the small-GTPase rows on this gene come from a different, complex-level experiment. Where UniProt records an NbExp value for one of these pairs it counts replicates within these same datasets, not independent studies. Marked over-annotated: reproducible, uncharacterised, and not removable, since nothing contradicts the pairing. Supporting Evidence: PMID:25416956 Here, we describe a systematic map of ?14,000 high-quality human binary protein-protein interactions |
| GO:0005515 protein binding | IPI PMID:25416956 A proteome-scale map of the human interactome network. | MARK AS OVER ANNOTATED | Summary: Bare protein binding to SIPA1L2 (isoform 2) (Q9P2F8-2), from a systematic yeast two-hybrid map of ~14,000 human binary interactions (HI-II-14). A high-throughput screening hit curated into IntAct, with no AP-3-specific follow-up. Reason: GO:0005515 says nothing about what AP3M1 does, and this partner does not let a more informative molecular function be substituted: it is not an AP-3 subunit, not a known AP-3 cargo whose sorting was traced, and there is no follow-up study naming AP3M1 and this protein together. Two of the informative alternatives are ruled out specifically - the mu3A cargo site reads YxxPhi motifs in cytoplasmic tails, which is proposed separately as a GO:0005048 row from the peptide-binding and structural literature, and the small-GTPase rows on this gene come from a different, complex-level experiment. Where UniProt records an NbExp value for one of these pairs it counts replicates within these same datasets, not independent studies. Marked over-annotated: reproducible, uncharacterised, and not removable, since nothing contradicts the pairing. Supporting Evidence: PMID:25416956 Here, we describe a systematic map of ?14,000 high-quality human binary protein-protein interactions |
| GO:0005515 protein binding | IPI PMID:25416956 A proteome-scale map of the human interactome network. | MARK AS OVER ANNOTATED | Summary: Bare protein binding to RSPH14 (Q9UHP6), from a systematic yeast two-hybrid map of ~14,000 human binary interactions (HI-II-14). A high-throughput screening hit curated into IntAct, with no AP-3-specific follow-up. Reason: GO:0005515 says nothing about what AP3M1 does, and this partner does not let a more informative molecular function be substituted: it is not an AP-3 subunit, not a known AP-3 cargo whose sorting was traced, and there is no follow-up study naming AP3M1 and this protein together. Two of the informative alternatives are ruled out specifically - the mu3A cargo site reads YxxPhi motifs in cytoplasmic tails, which is proposed separately as a GO:0005048 row from the peptide-binding and structural literature, and the small-GTPase rows on this gene come from a different, complex-level experiment. Where UniProt records an NbExp value for one of these pairs it counts replicates within these same datasets, not independent studies. Marked over-annotated: reproducible, uncharacterised, and not removable, since nothing contradicts the pairing. Supporting Evidence: PMID:25416956 Here, we describe a systematic map of ?14,000 high-quality human binary protein-protein interactions |
| GO:0005515 protein binding | IPI PMID:28514442 Architecture of the human interactome defines protein commun... | MARK AS OVER ANNOTATED | Summary: Bare protein binding to HLA-DMB (P28068), from BioPlex 2.0 affinity-purification mass spectrometry. A high-throughput screening hit curated into IntAct, with no AP-3-specific follow-up. Reason: GO:0005515 says nothing about what AP3M1 does, and this partner does not let a more informative molecular function be substituted: it is not an AP-3 subunit, not a known AP-3 cargo whose sorting was traced, and there is no follow-up study naming AP3M1 and this protein together. Two of the informative alternatives are ruled out specifically - the mu3A cargo site reads YxxPhi motifs in cytoplasmic tails, which is proposed separately as a GO:0005048 row from the peptide-binding and structural literature, and the small-GTPase rows on this gene come from a different, complex-level experiment. Where UniProt records an NbExp value for one of these pairs it counts replicates within these same datasets, not independent studies. Marked over-annotated: reproducible, uncharacterised, and not removable, since nothing contradicts the pairing. Supporting Evidence: PMID:28514442 Here we present BioPlex 2.0 (Biophysical Interactions of ORFeome-derived complexes), which uses robust affinity purification-mass spectrometry methodology to elucidate protein interaction networks and co-complexes nucleated by more than 25% of protein-coding genes from the human genome |
| GO:0005515 protein binding | IPI PMID:28514442 Architecture of the human interactome defines protein commun... | MARK AS OVER ANNOTATED | Summary: Bare protein binding to FRMD6 (isoform 2) (Q96NE9-2), from BioPlex 2.0 affinity-purification mass spectrometry. A high-throughput screening hit curated into IntAct, with no AP-3-specific follow-up. Reason: GO:0005515 says nothing about what AP3M1 does, and this partner does not let a more informative molecular function be substituted: it is not an AP-3 subunit, not a known AP-3 cargo whose sorting was traced, and there is no follow-up study naming AP3M1 and this protein together. Two of the informative alternatives are ruled out specifically - the mu3A cargo site reads YxxPhi motifs in cytoplasmic tails, which is proposed separately as a GO:0005048 row from the peptide-binding and structural literature, and the small-GTPase rows on this gene come from a different, complex-level experiment. Where UniProt records an NbExp value for one of these pairs it counts replicates within these same datasets, not independent studies. Marked over-annotated: reproducible, uncharacterised, and not removable, since nothing contradicts the pairing. Supporting Evidence: PMID:28514442 Here we present BioPlex 2.0 (Biophysical Interactions of ORFeome-derived complexes), which uses robust affinity purification-mass spectrometry methodology to elucidate protein interaction networks and co-complexes nucleated by more than 25% of protein-coding genes from the human genome |
| GO:0005515 protein binding | IPI PMID:28514442 Architecture of the human interactome defines protein commun... | MARK AS OVER ANNOTATED | Summary: Bare protein binding to RSPH14 (Q9UHP6), from BioPlex 2.0 affinity-purification mass spectrometry. A high-throughput screening hit curated into IntAct, with no AP-3-specific follow-up. Reason: GO:0005515 says nothing about what AP3M1 does, and this partner does not let a more informative molecular function be substituted: it is not an AP-3 subunit, not a known AP-3 cargo whose sorting was traced, and there is no follow-up study naming AP3M1 and this protein together. Two of the informative alternatives are ruled out specifically - the mu3A cargo site reads YxxPhi motifs in cytoplasmic tails, which is proposed separately as a GO:0005048 row from the peptide-binding and structural literature, and the small-GTPase rows on this gene come from a different, complex-level experiment. Where UniProt records an NbExp value for one of these pairs it counts replicates within these same datasets, not independent studies. Marked over-annotated: reproducible, uncharacterised, and not removable, since nothing contradicts the pairing. Supporting Evidence: PMID:28514442 Here we present BioPlex 2.0 (Biophysical Interactions of ORFeome-derived complexes), which uses robust affinity purification-mass spectrometry methodology to elucidate protein interaction networks and co-complexes nucleated by more than 25% of protein-coding genes from the human genome |
| GO:0005515 protein binding | IPI PMID:32296183 A reference map of the human binary protein interactome. | MARK AS OVER ANNOTATED | Summary: Bare protein binding to TRIM23 (P36406), from the HuRI all-by-all yeast two-hybrid reference interactome. A high-throughput screening hit curated into IntAct, with no AP-3-specific follow-up. Reason: GO:0005515 says nothing about what AP3M1 does, and this partner does not let a more informative molecular function be substituted: it is not an AP-3 subunit, not a known AP-3 cargo whose sorting was traced, and there is no follow-up study naming AP3M1 and this protein together. Two of the informative alternatives are ruled out specifically - the mu3A cargo site reads YxxPhi motifs in cytoplasmic tails, which is proposed separately as a GO:0005048 row from the peptide-binding and structural literature, and the small-GTPase rows on this gene come from a different, complex-level experiment. Where UniProt records an NbExp value for one of these pairs it counts replicates within these same datasets, not independent studies. Marked over-annotated: reproducible, uncharacterised, and not removable, since nothing contradicts the pairing. 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: Bare protein binding to CRMP1 (Q14194), from the HuRI all-by-all yeast two-hybrid reference interactome. A high-throughput screening hit curated into IntAct, with no AP-3-specific follow-up. Reason: GO:0005515 says nothing about what AP3M1 does, and this partner does not let a more informative molecular function be substituted: it is not an AP-3 subunit, not a known AP-3 cargo whose sorting was traced, and there is no follow-up study naming AP3M1 and this protein together. Two of the informative alternatives are ruled out specifically - the mu3A cargo site reads YxxPhi motifs in cytoplasmic tails, which is proposed separately as a GO:0005048 row from the peptide-binding and structural literature, and the small-GTPase rows on this gene come from a different, complex-level experiment. Where UniProt records an NbExp value for one of these pairs it counts replicates within these same datasets, not independent studies. Marked over-annotated: reproducible, uncharacterised, and not removable, since nothing contradicts the pairing. 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: Bare protein binding to AGTRAP (isoform 2) (Q6RW13-2), from the HuRI all-by-all yeast two-hybrid reference interactome. A high-throughput screening hit curated into IntAct, with no AP-3-specific follow-up. Reason: GO:0005515 says nothing about what AP3M1 does, and this partner does not let a more informative molecular function be substituted: it is not an AP-3 subunit, not a known AP-3 cargo whose sorting was traced, and there is no follow-up study naming AP3M1 and this protein together. Two of the informative alternatives are ruled out specifically - the mu3A cargo site reads YxxPhi motifs in cytoplasmic tails, which is proposed separately as a GO:0005048 row from the peptide-binding and structural literature, and the small-GTPase rows on this gene come from a different, complex-level experiment. Where UniProt records an NbExp value for one of these pairs it counts replicates within these same datasets, not independent studies. Marked over-annotated: reproducible, uncharacterised, and not removable, since nothing contradicts the pairing. 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: Bare protein binding to SPAG16 (isoform 4) (Q8N0X2-4), from the HuRI all-by-all yeast two-hybrid reference interactome. A high-throughput screening hit curated into IntAct, with no AP-3-specific follow-up. Reason: GO:0005515 says nothing about what AP3M1 does, and this partner does not let a more informative molecular function be substituted: it is not an AP-3 subunit, not a known AP-3 cargo whose sorting was traced, and there is no follow-up study naming AP3M1 and this protein together. Two of the informative alternatives are ruled out specifically - the mu3A cargo site reads YxxPhi motifs in cytoplasmic tails, which is proposed separately as a GO:0005048 row from the peptide-binding and structural literature, and the small-GTPase rows on this gene come from a different, complex-level experiment. Where UniProt records an NbExp value for one of these pairs it counts replicates within these same datasets, not independent studies. Marked over-annotated: reproducible, uncharacterised, and not removable, since nothing contradicts the pairing. 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: Bare protein binding to DVL3 (Q92997), from the HuRI all-by-all yeast two-hybrid reference interactome. A high-throughput screening hit curated into IntAct, with no AP-3-specific follow-up. Reason: GO:0005515 says nothing about what AP3M1 does, and this partner does not let a more informative molecular function be substituted: it is not an AP-3 subunit, not a known AP-3 cargo whose sorting was traced, and there is no follow-up study naming AP3M1 and this protein together. Two of the informative alternatives are ruled out specifically - the mu3A cargo site reads YxxPhi motifs in cytoplasmic tails, which is proposed separately as a GO:0005048 row from the peptide-binding and structural literature, and the small-GTPase rows on this gene come from a different, complex-level experiment. Where UniProt records an NbExp value for one of these pairs it counts replicates within these same datasets, not independent studies. Marked over-annotated: reproducible, uncharacterised, and not removable, since nothing contradicts the pairing. 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: Bare protein binding to ARNT2 (Q9HBZ2), from the HuRI all-by-all yeast two-hybrid reference interactome. A high-throughput screening hit curated into IntAct, with no AP-3-specific follow-up. Reason: GO:0005515 says nothing about what AP3M1 does, and this partner does not let a more informative molecular function be substituted: it is not an AP-3 subunit, not a known AP-3 cargo whose sorting was traced, and there is no follow-up study naming AP3M1 and this protein together. Two of the informative alternatives are ruled out specifically - the mu3A cargo site reads YxxPhi motifs in cytoplasmic tails, which is proposed separately as a GO:0005048 row from the peptide-binding and structural literature, and the small-GTPase rows on this gene come from a different, complex-level experiment. Where UniProt records an NbExp value for one of these pairs it counts replicates within these same datasets, not independent studies. Marked over-annotated: reproducible, uncharacterised, and not removable, since nothing contradicts the pairing. 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: Bare protein binding to SIPA1L2 (isoform 2) (Q9P2F8-2), from the HuRI all-by-all yeast two-hybrid reference interactome. A high-throughput screening hit curated into IntAct, with no AP-3-specific follow-up. Reason: GO:0005515 says nothing about what AP3M1 does, and this partner does not let a more informative molecular function be substituted: it is not an AP-3 subunit, not a known AP-3 cargo whose sorting was traced, and there is no follow-up study naming AP3M1 and this protein together. Two of the informative alternatives are ruled out specifically - the mu3A cargo site reads YxxPhi motifs in cytoplasmic tails, which is proposed separately as a GO:0005048 row from the peptide-binding and structural literature, and the small-GTPase rows on this gene come from a different, complex-level experiment. Where UniProt records an NbExp value for one of these pairs it counts replicates within these same datasets, not independent studies. Marked over-annotated: reproducible, uncharacterised, and not removable, since nothing contradicts the pairing. 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: Bare protein binding to IL36RN (Q9UBH0), from the HuRI all-by-all yeast two-hybrid reference interactome. A high-throughput screening hit curated into IntAct, with no AP-3-specific follow-up. Reason: GO:0005515 says nothing about what AP3M1 does, and this partner does not let a more informative molecular function be substituted: it is not an AP-3 subunit, not a known AP-3 cargo whose sorting was traced, and there is no follow-up study naming AP3M1 and this protein together. Two of the informative alternatives are ruled out specifically - the mu3A cargo site reads YxxPhi motifs in cytoplasmic tails, which is proposed separately as a GO:0005048 row from the peptide-binding and structural literature, and the small-GTPase rows on this gene come from a different, complex-level experiment. Where UniProt records an NbExp value for one of these pairs it counts replicates within these same datasets, not independent studies. Marked over-annotated: reproducible, uncharacterised, and not removable, since nothing contradicts the pairing. 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: Bare protein binding to RSPH14 (Q9UHP6), from the HuRI all-by-all yeast two-hybrid reference interactome. A high-throughput screening hit curated into IntAct, with no AP-3-specific follow-up. Reason: GO:0005515 says nothing about what AP3M1 does, and this partner does not let a more informative molecular function be substituted: it is not an AP-3 subunit, not a known AP-3 cargo whose sorting was traced, and there is no follow-up study naming AP3M1 and this protein together. Two of the informative alternatives are ruled out specifically - the mu3A cargo site reads YxxPhi motifs in cytoplasmic tails, which is proposed separately as a GO:0005048 row from the peptide-binding and structural literature, and the small-GTPase rows on this gene come from a different, complex-level experiment. Where UniProt records an NbExp value for one of these pairs it counts replicates within these same datasets, not independent studies. Marked over-annotated: reproducible, uncharacterised, and not removable, since nothing contradicts the pairing. 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: Bare protein binding to SLC12A4 (Q9UP95), from the HuRI all-by-all yeast two-hybrid reference interactome. A high-throughput screening hit curated into IntAct, with no AP-3-specific follow-up. Reason: GO:0005515 says nothing about what AP3M1 does, and this partner does not let a more informative molecular function be substituted: it is not an AP-3 subunit, not a known AP-3 cargo whose sorting was traced, and there is no follow-up study naming AP3M1 and this protein together. Two of the informative alternatives are ruled out specifically - the mu3A cargo site reads YxxPhi motifs in cytoplasmic tails, which is proposed separately as a GO:0005048 row from the peptide-binding and structural literature, and the small-GTPase rows on this gene come from a different, complex-level experiment. Where UniProt records an NbExp value for one of these pairs it counts replicates within these same datasets, not independent studies. Marked over-annotated: reproducible, uncharacterised, and not removable, since nothing contradicts the pairing. 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:32814053 Interactome Mapping Provides a Network of Neurodegenerative ... | MARK AS OVER ANNOTATED | Summary: Bare protein binding to HTT (P42858), from a yeast two-hybrid screen of ~500 neurodegeneration-related baits. A high-throughput screening hit curated into IntAct, with no AP-3-specific follow-up. Reason: GO:0005515 says nothing about what AP3M1 does, and this partner does not let a more informative molecular function be substituted: it is not an AP-3 subunit, not a known AP-3 cargo whose sorting was traced, and there is no follow-up study naming AP3M1 and this protein together. Two of the informative alternatives are ruled out specifically - the mu3A cargo site reads YxxPhi motifs in cytoplasmic tails, which is proposed separately as a GO:0005048 row from the peptide-binding and structural literature, and the small-GTPase rows on this gene come from a different, complex-level experiment. Where UniProt records an NbExp value for one of these pairs it counts replicates within these same datasets, not independent studies. Marked over-annotated: reproducible, uncharacterised, and not removable, since nothing contradicts the pairing. Supporting Evidence: PMID:32814053 Here, we report on an interactome map that focuses on neurodegenerative disease (ND), connects βΌ5,000 human proteins via βΌ30,000 candidate interactions and is generated by systematic yeast two-hybrid interaction screening of βΌ500 ND-related proteins and integration of literature interactions |
| GO:0005515 protein binding | IPI PMID:33961781 Dual proteome-scale networks reveal cell-specific remodeling... | MARK AS OVER ANNOTATED | Summary: Bare protein binding to HLA-DMB (P28068), from BioPlex 3.0 affinity-purification mass spectrometry in two cell lines. A high-throughput screening hit curated into IntAct, with no AP-3-specific follow-up. Reason: GO:0005515 says nothing about what AP3M1 does, and this partner does not let a more informative molecular function be substituted: it is not an AP-3 subunit, not a known AP-3 cargo whose sorting was traced, and there is no follow-up study naming AP3M1 and this protein together. Two of the informative alternatives are ruled out specifically - the mu3A cargo site reads YxxPhi motifs in cytoplasmic tails, which is proposed separately as a GO:0005048 row from the peptide-binding and structural literature, and the small-GTPase rows on this gene come from a different, complex-level experiment. Where UniProt records an NbExp value for one of these pairs it counts replicates within these same datasets, not independent studies. Marked over-annotated: reproducible, uncharacterised, and not removable, since nothing contradicts the pairing. Supporting Evidence: PMID:33961781 Through affinity-purification mass spectrometry, we have created two proteome-scale, cell-line-specific interaction networks |
| GO:0005515 protein binding | IPI PMID:33961781 Dual proteome-scale networks reveal cell-specific remodeling... | MARK AS OVER ANNOTATED | Summary: Bare protein binding to RSPH14 (Q9UHP6), from BioPlex 3.0 affinity-purification mass spectrometry in two cell lines. A high-throughput screening hit curated into IntAct, with no AP-3-specific follow-up. Reason: GO:0005515 says nothing about what AP3M1 does, and this partner does not let a more informative molecular function be substituted: it is not an AP-3 subunit, not a known AP-3 cargo whose sorting was traced, and there is no follow-up study naming AP3M1 and this protein together. Two of the informative alternatives are ruled out specifically - the mu3A cargo site reads YxxPhi motifs in cytoplasmic tails, which is proposed separately as a GO:0005048 row from the peptide-binding and structural literature, and the small-GTPase rows on this gene come from a different, complex-level experiment. Where UniProt records an NbExp value for one of these pairs it counts replicates within these same datasets, not independent studies. Marked over-annotated: reproducible, uncharacterised, and not removable, since nothing contradicts the pairing. Supporting Evidence: PMID:33961781 Through affinity-purification mass spectrometry, we have created two proteome-scale, cell-line-specific interaction networks |
| GO:0005515 protein binding | IPI PMID:40355756 The solute carrier superfamily interactome. | MARK AS OVER ANNOTATED | Summary: Bare protein binding to SLC12A4 (Q9UP95), from an interaction-proteomics survey of 396 solute carriers. A high-throughput screening hit curated into IntAct, with no AP-3-specific follow-up. Reason: GO:0005515 says nothing about what AP3M1 does, and this partner does not let a more informative molecular function be substituted: it is not an AP-3 subunit, not a known AP-3 cargo whose sorting was traced, and there is no follow-up study naming AP3M1 and this protein together. Two of the informative alternatives are ruled out specifically - the mu3A cargo site reads YxxPhi motifs in cytoplasmic tails, which is proposed separately as a GO:0005048 row from the peptide-binding and structural literature, and the small-GTPase rows on this gene come from a different, complex-level experiment. Where UniProt records an NbExp value for one of these pairs it counts replicates within these same datasets, not independent studies. Marked over-annotated: reproducible, uncharacterised, and not removable, since nothing contradicts the pairing. Supporting Evidence: PMID:40355756 We aimed at obtaining a global survey of the SLC-protein interaction landscape and mapped the protein-protein interactions of 396 SLCs by interaction proteomics |
| GO:0005737 cytoplasm | IEA GO_REF:0000117 | KEEP AS NON CORE | Summary: Cytoplasm, from the UniProt ARBA rule ARBA00026971. AP-3 is a soluble heterotetramer that cycles between cytosol and membranes, so the term is true. The rule itself is worth recording: fetching https://rest.uniprot.org/arba/ARBA00026971 returns 2388 OR-ed condition sets and a single output annotation, GO:0005737. Only two of those condition sets mention any signature AP3M1 carries, and neither can fire for this protein - one additionally requires taxon Saccharomyces, the other additionally requires IPR016635 and IPR027156, which are absent from AP3M1's complete seven-entry InterPro match list. Reason: Keep: the statement is true and harmless, and cytoplasmic residence of the unassembled coat is part of how AP-3 works - the complex is recruited from cytosol onto membranes by Arf1. The reproducibility problem with the rule is recorded rather than used as grounds for removal, since the served rule definition may not be everything the production pipeline uses; this repository has already documented the same right-branch-rule pattern for ARBA00027853. Non-core because 'cytoplasm' says nothing the specific compartment terms on this gene do not say better. Propagation Review Root cause: NO FAILURE NON CORE Sources checked: ARBA:ARBA00026971 Β· UniProt ARBA rule emitting GO:0005737 from 2388 OR-ed condition sets SOURCE WEAK OR INFERRED The rule emits only GO:0005737 and does so from 2388 condition sets. Of those, just two cite an InterPro entry AP3M1 has, and both carry a further condition AP3M1 fails (taxon Saccharomyces in one; IPR016635 plus IPR027156 in the other), so the annotation cannot be reproduced from the rule as published. The conclusion happens to be correct, which is why the row is kept. Supporting Evidence: PMID:39705307 the primary Arf1 binding site on AP-3 is on Ξ΄, in contrast to AP-1, which has been reported to interact more strongly with Arf1 through Ξ²1 file:human/AP3M1/AP3M1-uniprot.txt SUBCELLULAR LOCATION: Golgi apparatus. Cytoplasmic vesicle membrane |
| GO:0005764 lysosome | TAS PMID:10024875 Altered trafficking of lysosomal proteins in Hermansky-Pudla... | MARK AS OVER ANNOTATED | Summary: Lysosome as a location, the second traceable author statement from the same paper. What the paper shows is that AP-3 is required for lysosomal membrane proteins to reach lysosomes instead of the plasma membrane; it does not show that AP-3 itself resides in the lysosome. Reason: The evidence supports a process, not a residence. AP-3 is a cytosolic heterotetramer recruited onto tubular early-endosomal and Golgi-region membranes; the destination of the cargo it sorts is not its own compartment. UniProt records Golgi apparatus and cytoplasmic vesicle membrane, not lysosome. The gene already carries a separate, better-scoped lysosomal-membrane row from proteomics (GO:0005765, HDA) which captures whatever association exists. Marked over-annotated rather than removed: a coat acting at the endolysosomal boundary will legitimately be detected on those membranes, so the term is not false, only mis-scoped relative to the cited evidence. Supporting Evidence: PMID:10024875 Our results suggest that AP-3 functions in protein sorting to lysosomes file:human/AP3M1/AP3M1-uniprot.txt SUBCELLULAR LOCATION: Golgi apparatus. Cytoplasmic vesicle membrane PMID:42139345 is recruited onto early/tubular endosomal compartments from where it sorts transmembrane protein cargo to endolysosomes, lysosomes, and a diverse group of cell typeβspecific compartments termed lysosome-related organelles (LROs) that include melanosomes |
| GO:0005765 lysosomal membrane | HDA PMID:17897319 Integral and associated lysosomal membrane proteins. | KEEP AS NON CORE | Summary: Lysosomal membrane, from a high-throughput proteomic survey of purified placental lysosomal membranes. The GOA projection from this reference covers 242 distinct gene products, and the paper's own framing of the extra hits is 'significantly enriched in the lysosomal membrane fraction' rather than integral lysosomal proteins - it explicitly separates out proteins with signalling and targeting functions that are only partially associated with lysosomes. Reason: Plausible and worth keeping, but not a core location. AP-3 is a peripheral coat that delivers cargo to lysosomes, so co-purifying with a lysosomal membrane fraction is expected without implying residence. Consistent with this, the mu3 immunostaining in neurons includes a LAMP1-positive pool. Kept as non-core, in the same bracket as the Golgi and TGN pools. Supporting Evidence: PMID:17897319 We report on additional 86 proteins that were significantly enriched in the lysosomal membrane fraction PMID:17897319 Finally, our results identified a particular set of proteins with known functions in signaling and targeting to be at least partially associated with lysosomes PMID:27568566 As expected, immunohistochemistry localized ΞΌ3 to a number of endosomal compartments within pyramidal neurons |
| GO:0005769 early endosome | NAS PMID:23247405 Cell type-specific Rab32 and Rab38 cooperate with the ubiqui... | ACCEPT | Summary: Early endosome, asserted by ComplexPortal for the AP-3 complexes and cited to the Rab32/Rab38 review. This is the compartment where the mammalian complex is now placed: tyrosinase and TYRP1 are packaged at early/recycling endosome tubules in an AP-1- and AP-3-dependent way, and the 2026 reconstitution describes AP-3 as recruited onto early/tubular endosomal compartments. Reason: Core location, and the one that matters functionally. Together with the cytoplasmic vesicle membrane row this is where AP-3 selects cargo; the Golgi and TGN rows record an additional, less active pool. Supporting Evidence: PMID:23247405 Packaging of the tyrosinases into transport vesicles at early/recycling endosome-associated tubules is dependent on ubiquitous adaptor protein complex (AP)-1 and AP-3, and biogenesis of lysosome-related organelles complex (BLOC)-1 and BLOC-2 PMID:42139345 is recruited onto early/tubular endosomal compartments from where it sorts transmembrane protein cargo to endolysosomes, lysosomes, and a diverse group of cell typeβspecific compartments termed lysosome-related organelles (LROs) that include melanosomes PMID:39705307 Adaptor protein complex-3 (AP-3) mediates cargo sorting from endosomes to lysosomes and lysosome-related organelles |
| GO:0005794 Golgi apparatus | IEA GO_REF:0000044 | KEEP AS NON CORE | Summary: Golgi apparatus, from the UniProt subcellular-location vocabulary mapping of SL-0132. The underlying UniProt statement is 'SUBCELLULAR LOCATION: Golgi apparatus. Cytoplasmic vesicle membrane', which traces to the founding localisations of AP-3 in the Golgi region and at the TGN. Reason: Same call as the trans-Golgi network IBA, for the same reason: a Golgi-region pool of AP-3 is documented and is what UniProt records, but the compartment where mammalian AP-3 is now placed functionally is the tubular early endosome. Kept as an observed location, not promoted to the site of action. Propagation Review Root cause: NO FAILURE NON CORE Sources checked: UniProtKB-SubCell:SL-0132 Β· UniProt subcellular location SL-0132 (Golgi apparatus) SUPPORTS TRANSFER A controlled-vocabulary term mapped one-to-one onto GO:0005794; it faithfully reproduces the UniProt SUBCELLULAR LOCATION line for Q9Y2T2 and adds no evidence beyond it. Supporting Evidence: PMID:9151686 Immunofluorescence using anti-delta antibodies reveals that the AP-3 complex is associated with the Golgi region of the cell as well as with more peripheral structures PMID:42139345 is recruited onto early/tubular endosomal compartments from where it sorts transmembrane protein cargo to endolysosomes, lysosomes, and a diverse group of cell typeβspecific compartments termed lysosome-related organelles (LROs) that include melanosomes file:human/AP3M1/AP3M1-uniprot.txt SUBCELLULAR LOCATION: Golgi apparatus. Cytoplasmic vesicle membrane |
| GO:0005802 trans-Golgi network | IBA GO_REF:0000033 | KEEP AS NON CORE | Summary: Trans-Golgi network localisation, inherited from the family-level node PTN000055849. None of the five gene-level donors is an AP-3 subunit: two Arabidopsis proteins (AT1G60780 = AP1M2, AT4G24550 = AP4M), yeast Apm2, and the dog and human AP4M1 orthologues. Human AP-3 does have independent TGN evidence of its own - both founding papers placed the complex in the Golgi region and at the TGN - but the modern mammalian picture puts the functional pool on tubular early endosomes. Reason: Real but peripheral. Simpson et al. localised endogenous AP-3 to the Golgi region and more peripheral structures, and Dell'Angelica et al. found the sigma3 complex at the TGN and in peripheral structures, so a TGN pool is documented and UniProt records Golgi apparatus as a subcellular location. What has changed is where the work is thought to happen: the 2026 cryo-ET study describes mammalian AP-3 as recruited onto early/tubular endosomal compartments. Keeping the term as non-core records the observed pool without promoting it to the site of action; GO:0005769 early endosome, already on this gene, carries the functional compartment. Propagation Review Root cause: NO FAILURE NON CORE Sources checked: AGI_LocusCode:AT1G60780 Β· AP1M2 / O22715, Arabidopsis thaliana AP-1 mu-2 subunit SUPPORTS SOURCE BUT NOT TARGET Arabidopsis AP1M2 (O22715), an AP-1 medium subunit. AP-1 is a bona fide TGN coat, so the donor annotation is sound; it is the step to an AP-3 subunit that is loose. AGI_LocusCode:AT4G24550 Β· AP4M / Q9SB50, Arabidopsis thaliana AP-4 mu subunit SUPPORTS SOURCE BUT NOT TARGET Arabidopsis AP4M (Q9SB50). Same situation: correct for AP-4 at the plant TGN, not evidence about AP-3. PANTHER:PTN000055849 Β· PTHR10529 medium-subunit family node (pan-eukaryotic; its annotations reach AP3M1 and AP4M1 alike) SUPPORTS TRANSFER The family-level node. The PAINT curators did discriminate here - PTN000242370, the AP-2 mu clade, carries an IRD that negates this same term - so the node placement reflects a judgement that everything except AP-2 acts at the TGN. For human AP-3 that judgement is defensible but is not where the complex does most of its work. SGD:S000001011 Β· APM2 / P38700, Saccharomyces cerevisiae adaptin medium chain homolog Apm2 SUPPORTS SOURCE BUT NOT TARGET Yeast Apm2, again not an AP-3 subunit. UniProtKB:E2RED8 Β· AP4M1, dog AP-4 mu subunit SUPPORTS SOURCE BUT NOT TARGET Dog AP4M1; with O00189 this is one protein in two species, not two independent lines of evidence. UniProtKB:O00189 Β· AP4M1, human AP-4 mu subunit SUPPORTS SOURCE BUT NOT TARGET Human AP4M1. AP-4 is a TGN-associated coat, so the donor is right about itself; the AP-3 pool at the TGN is supported by this gene's own literature rather than by this donor. Supporting Evidence: PMID:9151686 Immunofluorescence using anti-delta antibodies reveals that the AP-3 complex is associated with the Golgi region of the cell as well as with more peripheral structures PMID:9118953 Immunofluorescence microscopy analyses show that the sigma3-containing complex is present both in the area of the TGN and in peripheral structures, some of which contain the transferrin receptor PMID:42139345 is recruited onto early/tubular endosomal compartments from where it sorts transmembrane protein cargo to endolysosomes, lysosomes, and a diverse group of cell typeβspecific compartments termed lysosome-related organelles (LROs) that include melanosomes |
| GO:0006622 protein targeting to lysosome | TAS PMID:10024875 Altered trafficking of lysosomal proteins in Hermansky-Pudla... | ACCEPT | Summary: Protein targeting to lysosome, a traceable author statement from the paper that identified AP3B1 mutations in Hermansky-Pudlak syndrome. Loss of AP-3 in patient fibroblasts sends CD63, LAMP1 and LAMP2 to the cell surface while leaving non-lysosomal proteins alone, and the authors attribute the selectivity to the mu3A subunit's preference for tyrosine-based lysosomal targeting signals - which is why the curator placed the row on AP3M1. In GOA this reference annotates one gene product only, AP3M1. Reason: Core. This is AP-3's defining cellular job and the paper names mu3A as the reason the cargo selection is specific. Convergent modern evidence: the cryo-EM structure resolves a LAMP1 tyrosine motif in the mu3 pocket, the same LAMP1/lgp120 YQTI motif that was shown by two-hybrid to bind mu3A almost exclusively. Supporting Evidence: PMID:10024875 The AP-3 deficiency results in increased surface expression of the lysosomal membrane proteins CD63, lamp-1, and lamp-2, but not of nonlysosomal proteins PMID:10024875 These differential effects are consistent with the preferential interaction of the AP-3 mu 3A subunit with tyrosine-based signals involved in lysosomal targeting PMID:9794796 we show that the cytosolic domain of lgp120 interacts almost exclusively with mu3A PMID:39705307 There is also clear density for the LAMP1 cargo motif in the ΞΌ3-CTD tyrosine cargo-binding pocket, unambiguously showing that this complex represents the cargo-engaged state of AP-3 |
| GO:0006886 intracellular protein transport | IEA GO_REF:0000002 | KEEP AS NON CORE | Summary: The generic half of the same interpro2go mapping: the clathrin-adaptor mu family signature also yields 'intracellular protein transport'. AP-3 does move transmembrane proteins between intracellular compartments, so this is true of AP3M1 at a high level. Reason: True but uninformative. The specific statements on this gene - protein targeting to lysosome, Golgi to vacuole transport, and the proposed endosome to lysosome transport row - already say where the transport goes. Kept rather than removed because, unlike GO:0030131, nothing in the mapping is wrong. Propagation Review Root cause: NO FAILURE NON CORE Sources checked: InterPro:IPR001392 Β· InterPro family 'Clathrin adaptor, mu subunit' (covers every AP medium subunit) SUPPORTS TRANSFER A family signature whose members are all vesicular-transport adaptor subunits, so a generic transport term transfers safely even though the complex-level term from the same entry does not. InterPro:IPR018240 Β· InterPro conserved site 'Clathrin adaptor, mu subunit, conserved site' SUPPORTS TRANSFER The conserved-site companion signature; same scope, same verdict. Supporting Evidence: PMID:9545220 A heterotetrameric complex termed AP-3 is involved in signal-mediated protein sorting to endosomal-lysosomal organelles file:human/AP3M1/AP3M1-uniprot.txt not clathrin-associated. The complex is associated with the Golgi |
| GO:0006896 Golgi to vacuole transport | IBA GO_REF:0000033 | ACCEPT | Summary: Post-Golgi delivery of transmembrane cargo to the vacuole/lysosome is what AP-3 is for. The IBD is at PTN000055849, the family-level node, with four gene-level donors: yeast Apm3 (the AP-3 mu subunit, which drives the ALP pathway to the vacuole), yeast Apm2 and Apm1, and human AP4M1. GO:0006896 'Golgi to vacuole transport' is the least common ancestor that covers both the fungal vacuole and the animal lysosome, since GO:0005764 lysosome is_a GO:0005773 vacuole and the human-specific child GO:0090160 'Golgi to lysosome transport' is_a GO:0006896. Reason: Core. In human cells the machinery is the same and the destination is the lysosome and its related organelles: loss of AP-3 redistributes LAMP1, LAMP2 and CD63 to the cell surface (PMID:10024875). One nuance worth recording rather than acting on: the mammalian step is now placed on tubular early endosomes rather than at the Golgi itself (PMID:42139345), so the 'Golgi to' framing describes the route's origin rather than the membrane AP-3 is recruited to. The parent term remains the right LCA for a node whose donors span fungi and animals, so this is not a granularity failure - the donors do not all agree on a single more specific child. Propagation Review Root cause: NO FAILURE CORE Sources checked: PANTHER:PTN000055849 Β· PTHR10529 medium-subunit family node (pan-eukaryotic; its annotations reach AP3M1 and AP4M1 alike) SUPPORTS TRANSFER A deep node ancestral to every AP medium subunit. For a transport process shared by AP-1, AP-3 and AP-4 that placement is defensible, and the PAINT curators fenced off the one clade where it is not: PTN002575694, the stonins, carries an IRD (negated) for this same term. SGD:S000000492 Β· APM3 / P38153, Saccharomyces cerevisiae AP-3 mu subunit SUPPORTS TRANSFER Yeast Apm3, the AP-3 mu subunit - the donor closest in role to the target, and the one whose vacuolar (ALP) pathway is the fungal counterpart of the human AP-3 route to lysosomes. SGD:S000001011 Β· APM2 / P38700, Saccharomyces cerevisiae adaptin medium chain homolog Apm2 SUPPORTS TRANSFER Yeast Apm2. A co-donor from a different AP complex; it broadens the node but the AP-3 specific support comes from Apm3. SGD:S000006180 Β· APM1 / Q00776, Saccharomyces cerevisiae AP-1 mu subunit SUPPORTS TRANSFER Yeast Apm1, the AP-1 mu subunit. Same comment: it argues for the process at family level, not for AP-3 in particular. UniProtKB:O00189 Β· AP4M1, human AP-4 mu subunit SUPPORTS TRANSFER Human AP4M1. Its own PAINT record carries the more specific child GO:0090160 'Golgi to lysosome transport' at PTN000242612, which is why the parent, not the child, is the term that reaches the whole family. Supporting Evidence: PMID:10024875 The AP-3 deficiency results in increased surface expression of the lysosomal membrane proteins CD63, lamp-1, and lamp-2, but not of nonlysosomal proteins PMID:10024875 Our results suggest that AP-3 functions in protein sorting to lysosomes PMID:42139345 The AP3 complex mediates cargo sorting and carrier assembly for the trafficking of transmembrane proteins from endosomes to lysosomes |
| GO:0008089 anterograde axonal transport | IEA GO_REF:0000107 | KEEP AS NON CORE | Summary: Transferred from mouse Ap3m1 (Q9JKC8), whose own row for this term is an IMP citing PMID:21998198. Reading that paper: all of its AP-3 genetics use the mocha allele Ap3d1mh/mh, a delta-subunit null, and the phenotype is failure of the AP-3 cargo PI4KIIalpha to reach neurites from the cell body. Searching the cached full text, the strings 'Ap3m1', 'mu3A' and 'AP3M' do not occur at all, and 'mu3' occurs twice, both inside the introductory sentence that names the heterotetramer's four subunits. So mu3A itself is never perturbed or assayed. The term itself is also a stretch for a coat adaptor: GO:0008089 is defined as directed movement along microtubules toward the cell periphery, and AP-3 is a sorting coat that loads cargo into carriers rather than a motor that moves them. Reason: Keep, but not as a core AP3M1 function, and with the source-side limitation on record. Two separate dilutions sit between the experiment and this row. First, the evidence is complex-level: a delta null removes the whole coat, so the phenotype belongs to AP-3 rather than to its medium subunit. Second, delta is shared by the ubiquitous and the neuronal complexes (ComplexPortal CPX-5051/CPX-5052 versus CPX-5053/CPX-5055), so a mocha phenotype cannot even be assigned to the AP-3A complex that contains AP3M1. The genetics that do separate the two point the other way: beta3B-containing complexes are the ones preferentially targeted to neuronal processes and required for synaptic-vesicle targeting of ZnT3 and ClC-3, while beta3A-deficient brain has more of those cargoes, not less. A subunit of the ubiquitous complex is a poor candidate for the core agent of synaptic vesicle delivery. Propagation Review Root cause: SOURCE WEAK OR INFERRED Failure modes: SOURCE EVIDENCE WEAK Sources checked: UniProtKB:Q9JKC8 Β· Ap3m1, mouse AP-3 complex subunit mu-1 SOURCE WEAK OR INFERRED Mouse Ap3m1 is a true 1:1 orthologue and the cargo pocket is identical to human at all 14 tested columns, so orthology is not the problem. The problem is upstream of the transfer: the donor's own IMP rests on the Ap3d1 mocha allele, so it is evidence about the AP-3 complex and not about this subunit. ensembl:ENSMUSP00000117346 Β· Ensembl Compara protein model for mouse Ap3m1 SOURCE WEAK OR INFERRED The Ensembl Compara protein model through which the orthology call was made; it carries no evidence of its own beyond the Q9JKC8 record. Supporting Evidence: PMID:21998198 the AP-3βdeficient allele mocha (Ap3d1mh/mh PMID:21998198 PI4KIIΞ± was targeted to processes in wild-type primary cultured cortical neurons and PC12 cells but failed to reach neurites in cells lacking either AP-3 or BLOC-1 PMID:15537701 beta3B-containing AP-3 complexes were preferentially targeted to neuronal processes PMID:15537701 Surprisingly, despite the lack of neurological symptoms, beta3A-deficient mouse brain possessed significantly increased synaptic zinc stores and synaptic vesicle content of ZnT3 and ClC-3 |
| GO:0008089 anterograde axonal transport | ISS GO_REF:0000024 | KEEP AS NON CORE | Summary: Transferred from mouse Ap3m1 (Q9JKC8), whose own row for this term is an IMP citing PMID:21998198. Reading that paper: all of its AP-3 genetics use the mocha allele Ap3d1mh/mh, a delta-subunit null, and the phenotype is failure of the AP-3 cargo PI4KIIalpha to reach neurites from the cell body. Searching the cached full text, the strings 'Ap3m1', 'mu3A' and 'AP3M' do not occur at all, and 'mu3' occurs twice, both inside the introductory sentence that names the heterotetramer's four subunits. So mu3A itself is never perturbed or assayed. The term itself is also a stretch for a coat adaptor: GO:0008089 is defined as directed movement along microtubules toward the cell periphery, and AP-3 is a sorting coat that loads cargo into carriers rather than a motor that moves them. Reason: Keep, but not as a core AP3M1 function, and with the source-side limitation on record. Two separate dilutions sit between the experiment and this row. First, the evidence is complex-level: a delta null removes the whole coat, so the phenotype belongs to AP-3 rather than to its medium subunit. Second, delta is shared by the ubiquitous and the neuronal complexes (ComplexPortal CPX-5051/CPX-5052 versus CPX-5053/CPX-5055), so a mocha phenotype cannot even be assigned to the AP-3A complex that contains AP3M1. The genetics that do separate the two point the other way: beta3B-containing complexes are the ones preferentially targeted to neuronal processes and required for synaptic-vesicle targeting of ZnT3 and ClC-3, while beta3A-deficient brain has more of those cargoes, not less. A subunit of the ubiquitous complex is a poor candidate for the core agent of synaptic vesicle delivery. Propagation Review Root cause: SOURCE WEAK OR INFERRED Failure modes: SOURCE EVIDENCE WEAK Sources checked: UniProtKB:Q9JKC8 Β· Ap3m1, mouse AP-3 complex subunit mu-1 SOURCE WEAK OR INFERRED Mouse Ap3m1, the ISS donor. Orthology is unambiguous and the cargo-binding site is identical between the two species, but the donor annotation is an IMP obtained with the Ap3d1 mocha allele rather than by perturbing Ap3m1. Supporting Evidence: PMID:21998198 the AP-3βdeficient allele mocha (Ap3d1mh/mh PMID:21998198 PI4KIIΞ± was targeted to processes in wild-type primary cultured cortical neurons and PC12 cells but failed to reach neurites in cells lacking either AP-3 or BLOC-1 PMID:15537701 beta3B-containing AP-3 complexes were preferentially targeted to neuronal processes PMID:15537701 Surprisingly, despite the lack of neurological symptoms, beta3A-deficient mouse brain possessed significantly increased synaptic zinc stores and synaptic vesicle content of ZnT3 and ClC-3 |
| GO:0016192 vesicle-mediated transport | IEA GO_REF:0000002 | KEEP AS NON CORE | Summary: The other generic interpro2go output: vesicle-mediated transport. AP-3 is a vesicle coat, so the parent term holds. Reason: Correct and very general; retained as background. The informative descendants are already on the gene. There is a duplicate of this term on the same gene from ComplexPortal (NAS, PMID:23247405), which is fine - duplicate terms under different evidence are normal. Propagation Review Root cause: NO FAILURE NON CORE Sources checked: InterPro:IPR001392 Β· InterPro family 'Clathrin adaptor, mu subunit' (covers every AP medium subunit) SUPPORTS TRANSFER Every member of this family is a medium subunit of a vesicle coat adaptor, so the generic vesicle-transport term is safe at family level. InterPro:IPR018240 Β· InterPro conserved site 'Clathrin adaptor, mu subunit, conserved site' SUPPORTS TRANSFER Conserved-site signature of the same family; no independent content. Supporting Evidence: PMID:42139345 The AP3 complex mediates cargo sorting and carrier assembly for the trafficking of transmembrane proteins from endosomes to lysosomes file:human/AP3M1/AP3M1-uniprot.txt not clathrin-associated. The complex is associated with the Golgi |
| GO:0016192 vesicle-mediated transport | NAS PMID:23247405 Cell type-specific Rab32 and Rab38 cooperate with the ubiqui... | KEEP AS NON CORE | Summary: The generic vesicle-mediated transport statement from ComplexPortal, duplicating the interpro2go row for the same term. Reason: True and very general. Retained as background; the specific routes are covered by the lysosome-targeting and Golgi-to-vacuole rows. Supporting Evidence: PMID:23247405 The biogenesis of LROs is known to utilize much of the common protein machinery used in the transport of integral membrane proteins to lysosomes PMID:42139345 The AP3 complex mediates cargo sorting and carrier assembly for the trafficking of transmembrane proteins from endosomes to lysosomes |
| GO:0030123 AP-3 adaptor complex | IBA GO_REF:0000033 | ACCEPT | Summary: AP3M1 is the medium (mu) subunit of the AP-3 heterotetramer, and this IBA sits on PTN002237676, the AP-3 mu clade node of PTHR10529. Both seeds are genuine AP-3 mu subunits - yeast Apm3 (SGD:S000000492 = P38153) and Dictyostelium apm3 (dictyBase:DDB_G0277901 = Q9GPF1) - so the node is not a family-wide catch-all but the AP-3-specific one, and the human AP-3 mu subunit sits squarely inside its clade. The complex itself was defined biochemically by co-immunoprecipitation of the four subunits (PMID:9151686) and human AP-3 containing AP3M1 has since been reconstituted and solved by cryo-EM (PMID:39705307, PDB 9C58-9C5C, chain M = AP3M1 1-418). Reason: Core. Complex membership is the least disputable fact about this protein: UniProt states the composition, ComplexPortal models AP3M1 into CPX-5051 and CPX-5052, and the 2024 cryo-EM series resolves it as chain M of the reconstituted human complex. The only imprecision is at the ontology's end rather than the annotation's: GO:0030123 has no children, so it cannot express that AP3M1 belongs specifically to the ubiquitous AP-3A variants and not to the neuronal AP-3B complexes, which use AP3M2 and AP3B2. That is recorded as a proposed new term rather than as a defect in this row. Propagation Review Root cause: NO FAILURE CORE Sources checked: PANTHER:PTN002237676 Β· PTHR10529 AP-3 mu clade IBD node (seeded by yeast Apm3 and Dictyostelium apm3) SUPPORTS TRANSFER The IBD is at the AP-3 mu clade node, not at the family root PTN000055849 that carries the other three IBAs on this gene. AP3M1 is the human member of that clade, so the transfer is the intended one. SGD:S000000492 Β· APM3 / P38153, Saccharomyces cerevisiae AP-3 mu subunit SUPPORTS TRANSFER Yeast Apm3 is the AP-3 mu subunit of the fungal complex; a same-role, same-complex donor rather than a paralogue. dictyBase:DDB_G0277901 Β· apm3 / Q9GPF1, Dictyostelium discoideum AP-3 mu subunit SUPPORTS TRANSFER Dictyostelium apm3 is likewise the AP-3 mu subunit. Two donors from widely separated eukaryotic lineages both occupying the AP-3 mu slot is what places the node deep rather than in one lineage. Supporting Evidence: PMID:9151686 Immunoprecipitation experiments indicated that the complex consists not only of Ξ²-NAP and p47, but also of two other proteins of βΌ160 (p160) and βΌ25 kD (p25), making it a heterotetramer like an adaptor complex PMID:39705307 In this state, the CTD of the ΞΌ3 subunit has been ejected from the bowl formed by the two large subunits file:human/AP3M1/AP3M1-uniprot.txt AP3B1 or AP3B2), a medium adaptin (mu-type subunit AP3M1 or AP3M2) and file:human/AP3M1/AP3M1-uniprot.txt ComplexPortal; CPX-5051; Ubiquitous AP-3 Adaptor complex, sigma3a variant. |
| GO:0030123 AP-3 adaptor complex | NAS PMID:9151686 Characterization of the adaptor-related protein complex, AP-... | ACCEPT | Summary: ComplexPortal's statement-level assertion of the same membership, citing the paper that defined AP-3. Simpson et al. raised antibodies against the delta and sigma3 candidates and showed by co-immunoprecipitation that they are the missing subunits of the complex already known to contain p47 (mu3) and beta-NAP (beta3B). An independent, non-mechanistic corroboration comes from proteomics: AP3B1 depletion lowers AP3M1 protein abundance, the behaviour expected of an obligate subunit whose stability depends on assembly. That stoichiometric observation is the entire content of the affinage record for this gene. Reason: Core, and correctly cited. The reference projects over 14 entities in GOA, which is the AP-3 subunits and their orthologues rather than a mass projection. NAS is the right code for a ComplexPortal curation statement, and it is consistent with the IBA row for the same term. Supporting Evidence: PMID:9151686 Immunoprecipitation experiments indicated that the complex consists not only of Ξ²-NAP and p47, but also of two other proteins of βΌ160 (p160) and βΌ25 kD (p25), making it a heterotetramer like an adaptor complex PMID:9151686 p47 exists as two isoforms: p47A, which is expressed ubiquitously, and p47B, which is specifically expressed in neuronal tissues PMID:29032074 We identified 48 such rate-limiting interactions and experimentally confirmed our predictions on the interactions of AP3B1 with AP3M1 and GTF2E2 with GTF2E1 file:human/AP3M1/AP3M1-deep-research-affinage.md Within this complex, AP3B1 acts as a rate-limiting assembly factor: depletion of AP3B1 reduces AP3M1 protein abundance, indicating that AP3M1 stability and stoichiometry depend on co-regulation with AP3B1 |
| GO:0030131 clathrin adaptor complex | IEA GO_REF:0000002 | REMOVE | Summary: InterPro2GO maps IPR001392 'Clathrin adaptor, mu subunit' and its conserved site IPR018240 onto GO:0030131 'clathrin adaptor complex', defined as 'A membrane coat adaptor complex that links clathrin to a membrane'. IPR001392 is a family signature that matches every AP medium subunit - AP1M1, AP1M2, AP2M1, AP3M1, AP3M2 and AP4M1 are all in PTHR10529 and all carry it - so the mapping cannot distinguish the two AP complexes that link clathrin from the two that do not. Reason: A demonstrably wrong domain-to-complex mapping, which is the licensed use of REMOVE on an IEA row. GO itself places AP-3 outside this class: GO:0030131 has exactly two children, GO:0030121 (AP-1) and GO:0030122 (AP-2), and the ancestors of GO:0030123 (AP-3) are GO:0030119, GO:0030117, GO:0048475, GO:0098796 and GO:0005737 - GO:0030131 is not among them. The definition of GO:0030123 says the same in prose: AP-3 does not appear to associate with clathrin in all organisms. Biologically, reconstituted AP3:ARF1 builds coated tubular carriers with no clathrin lattice, and UniProt's own FUNCTION line calls AP-3 an adaptor-related complex which is not clathrin-associated. The residual clathrin link that does exist is a beta3 hinge interaction (PMID:9545220), not a property of the mu subunit and not sufficient to make AP-3 a clathrin adaptor complex. The same InterPro pair also yields GO:0006886 and GO:0016192 on this gene; those are generically true and are kept. Propagation Review Root cause: PROPAGATION BAD Failure modes: COMPARTMENT OR COMPLEX MISMATCH Sources checked: InterPro:IPR001392 Β· InterPro family 'Clathrin adaptor, mu subunit' (covers every AP medium subunit) SOURCE BAD The signature is correct for AP3M1 - InterPro returns it among the seven entries matching Q9Y2T2 - but its name embeds a complex assignment the family does not license. The entry covers AP-3 and AP-4 medium subunits, neither of whose complexes is a GO:0030131 descendant, so the interpro2go mapping is wrong for a subset of the entry's own members. InterPro:IPR018240 Β· InterPro conserved site 'Clathrin adaptor, mu subunit, conserved site' SOURCE BAD A conserved-site signature within the same family, inheriting the same problem; it adds no independent evidence about clathrin. Supporting Evidence: PMID:42139345 By demonstrating that AP3:ARF1 can generate carriers without using a clathrin lattice, we explain the clathrin independence of AP3-mediated trafficking PMID:9151686 The AP-3 complex is not clathrin associated file:human/AP3M1/AP3M1-uniprot.txt not clathrin-associated. The complex is associated with the Golgi file:human/AP3M1/AP3M1-uniprot.txt PANTHER; PTHR10529; AP COMPLEX SUBUNIT MU; 1. |
| GO:0030659 cytoplasmic vesicle membrane | IEA GO_REF:0000044 | ACCEPT | Summary: Cytoplasmic vesicle membrane, from SL-0089. This is the coat's working location: UniProt adds 'Peripheral membrane protein; Cytoplasmic side' and notes that AP-3 is a component of the coat surrounding the cytoplasmic face of coated vesicles at the Golgi complex. The 2026 cryo-ET reconstitution visualises exactly this - AP3:ARF1 arches spiralling on the cytoplasmic face of tubulated membranes. Reason: Core location. A coat adaptor subunit is, by definition, on the cytoplasmic face of a transport vesicle membrane when it is doing its job, and that state is now directly visualised for AP-3 with cargo bound. Propagation Review Root cause: NO FAILURE CORE Sources checked: UniProtKB-SubCell:SL-0089 Β· UniProt subcellular location SL-0089 (Cytoplasmic vesicle membrane) SUPPORTS TRANSFER Faithful mapping of the UniProt SUBCELLULAR LOCATION line, and independently confirmed by the membrane-bound AP-3 structures, in which the mu3 C-terminal domain's cargo site sits adjacent to the bilayer. Supporting Evidence: file:human/AP3M1/AP3M1-uniprot.txt SUBCELLULAR LOCATION: Golgi apparatus. Cytoplasmic vesicle membrane PMID:42139345 The known cargo binding sites on C-ΞΌ3 and Ο3/Ξ΄ are adjacent to the membrane PMID:39705307 There is also clear density for the LAMP1 cargo motif in the ΞΌ3-CTD tyrosine cargo-binding pocket, unambiguously showing that this complex represents the cargo-engaged state of AP-3 |
| GO:0031267 small GTPase binding | IPI PMID:22511774 BLOC-2, AP-3, and AP-1 proteins function in concert with Rab... | KEEP AS NON CORE | Summary: Small GTPase binding, with RAB38 as the partner. The experiment is a co-immunoprecipitation from MNT-1 melanocytic cell extracts in which BLOC-2, AP-3 and AP-1 all came down with Rab38 and Rab32, together with colocalisation and a cargo-trafficking phenotype for tyrosinase and TYRP1 in Rab32/Rab38-deficient cells. The GOA projection from this reference covers seven gene products - AP3D1, AP3B1, AP1G1, HPS5, RAB38, RAB32 and AP3M1 - i.e. the complexes and the Rabs, not a mass projection. Reason: Keep. The interaction is real and biologically meaningful - it is the mechanism by which cell-type-specific Rabs recruit the ubiquitous sorting machinery to the maturing melanosome - but it was detected at complex level in one specialised cell type, and nothing identifies mu3A as the subunit that contacts the Rab. The known GTPase interface of AP-3 is on the large subunits: the cryo-EM work places the primary Arf1 site on delta with a second on beta3, and reports no nucleotide-GTPase contact for mu3. Non-core on both counts. Worth stating why this is kept as non-core while GO:0035651 'AP-3 adaptor complex binding', argued from the same complex-level premise, is marked over-annotated instead: there the term names the protein's own complex, so it restates membership already asserted as GO:0030123 part_of and adds nothing; here the partner is a genuinely external protein and the row records a real physical association of AP-3 with a Rab, which no other term on this gene captures. The weakness is subunit attribution, not the existence of the interaction, and the contributes_to-style hedge that would express that is not something this review adds or edits. Supporting Evidence: PMID:22511774 BLOC-2, AP-3, and AP-1 coimmunoprecipitated with Rab38 and Rab32 from MNT-1 melanocytic cell extracts PMID:22511774 Rab38- and Rab32-deficient MNT-1 cells displayed abnormal trafficking and steady state levels of known cargoes of the BLOC-2, AP-3, and AP-1 pathways, the melanin-synthesizing enzymes tyrosinase and tyrosinase-related protein-1 PMID:39705307 the primary Arf1 binding site on AP-3 is on Ξ΄, in contrast to AP-1, which has been reported to interact more strongly with Arf1 through Ξ²1 |
| GO:0031267 small GTPase binding | IPI PMID:22511774 BLOC-2, AP-3, and AP-1 proteins function in concert with Rab... | KEEP AS NON CORE | Summary: Small GTPase binding, with RAB32 as the partner. The experiment is a co-immunoprecipitation from MNT-1 melanocytic cell extracts in which BLOC-2, AP-3 and AP-1 all came down with Rab38 and Rab32, together with colocalisation and a cargo-trafficking phenotype for tyrosinase and TYRP1 in Rab32/Rab38-deficient cells. The GOA projection from this reference covers seven gene products - AP3D1, AP3B1, AP1G1, HPS5, RAB38, RAB32 and AP3M1 - i.e. the complexes and the Rabs, not a mass projection. Reason: Keep. The interaction is real and biologically meaningful - it is the mechanism by which cell-type-specific Rabs recruit the ubiquitous sorting machinery to the maturing melanosome - but it was detected at complex level in one specialised cell type, and nothing identifies mu3A as the subunit that contacts the Rab. The known GTPase interface of AP-3 is on the large subunits: the cryo-EM work places the primary Arf1 site on delta with a second on beta3, and reports no nucleotide-GTPase contact for mu3. Non-core on both counts. Worth stating why this is kept as non-core while GO:0035651 'AP-3 adaptor complex binding', argued from the same complex-level premise, is marked over-annotated instead: there the term names the protein's own complex, so it restates membership already asserted as GO:0030123 part_of and adds nothing; here the partner is a genuinely external protein and the row records a real physical association of AP-3 with a Rab, which no other term on this gene captures. The weakness is subunit attribution, not the existence of the interaction, and the contributes_to-style hedge that would express that is not something this review adds or edits. Supporting Evidence: PMID:22511774 BLOC-2, AP-3, and AP-1 coimmunoprecipitated with Rab38 and Rab32 from MNT-1 melanocytic cell extracts PMID:22511774 Rab38- and Rab32-deficient MNT-1 cells displayed abnormal trafficking and steady state levels of known cargoes of the BLOC-2, AP-3, and AP-1 pathways, the melanin-synthesizing enzymes tyrosinase and tyrosinase-related protein-1 PMID:39705307 the primary Arf1 binding site on AP-3 is on Ξ΄, in contrast to AP-1, which has been reported to interact more strongly with Arf1 through Ξ²1 |
| GO:0035615 clathrin-cargo adaptor activity | IBA GO_REF:0000033 | MODIFY | Summary: This is the AP-1/AP-2-specific term leaking onto AP-3 through the shared medium-subunit family. GO:0035615 is defined as 'Bringing together a cargo protein with clathrin, responsible for the formation of endocytic vesicles', and every gene-level donor on the row is an AP-1 or AP-2 medium subunit: Drosophila AP-1mu (FBgn0024833 = O62531), Drosophila AP-2mu (FBgn0263351 = O62530) and Dictyostelium apm1 (DDB_G0289247 = Q54HS9); the current PAINT slice for PTN000055849 adds human AP1M2 (Q9Y6Q5) as a fourth. Both halves of the definition fail for AP-3. AP-3 builds carriers without a clathrin lattice, and it acts on the biosynthetic/endosomal route to lysosomes rather than in endocytosis. What is true, and what the donors really share with AP3M1, is the underlying cargo-adaptor activity: the mu subunit reads YxxPhi sorting signals and bridges them to the coat. Reason: Retain the adaptor essence, drop the clathrin and endocytic commitments. GO:0140312 'cargo adaptor activity' is the immediate parent of GO:0035615 and says what the family node actually supports. Three independent grounds for the change. (i) The ontology already separates the two coats: GO:0030131 'clathrin adaptor complex' has exactly two children, AP-1 and AP-2, and GO:0030123 AP-3 is not among its descendants. (ii) The biology: reconstituted AP3:ARF1 forms coated tubular carriers with no clathrin lattice. (iii) Direction of traffic: AP-3 sorts from endosomes to lysosomes, not into endocytic vesicles. The cargo-adaptor half of the term is, by contrast, directly demonstrated for this subunit - the cryo-EM structure resolves a LAMP1 YxxPhi peptide in the mu3 pocket, and a contact analysis of that structure (file:human/AP3M1/AP3M1-bioinformatics/RESULTS.md) shows the pocket is the same structural site as AP-2's, retained in AP3M1 but chemically divergent from it. Propagation Review Root cause: TERM SCOPING PROBLEM Failure modes: GRANULARITY MISMATCH Sources checked: FB:FBgn0024833 · AP-1mu / O62531, Drosophila melanogaster AP-1 mu subunit SUPPORTS SOURCE BUT NOT TARGET Drosophila AP-1mu. AP-1 is a clathrin adaptor, so the donor annotation is correct about the donor; it is the clathrin half that must not travel. FB:FBgn0263351 · AP-2mu / O62530, Drosophila melanogaster AP-2 mu subunit SUPPORTS SOURCE BUT NOT TARGET Drosophila AP-2mu. AP-2 is the endocytic clathrin adaptor, which is the source of the 'formation of endocytic vesicles' clause in the term definition - the clause that is furthest from AP-3's biology. PANTHER:PTN000055849 · PTHR10529 medium-subunit family node (pan-eukaryotic; its annotations reach AP3M1 and AP4M1 alike) SUPPORTS SOURCE BUT NOT TARGET The family-level node, ancestral to AP-1, AP-2, AP-3 and AP-4 medium subunits alike. AP3M1 is genuinely a descendant, so the propagation is mechanically correct; the error is that a clathrin- and endocytosis-specific child was placed at a node two of whose descendant coats are neither. Human AP4M1 (O00189) receives the same IBA from this node and has no clathrin activity of its own either. dictyBase:DDB_G0289247 · apm1 / Q54HS9, Dictyostelium discoideum AP-1 mu subunit SUPPORTS SOURCE BUT NOT TARGET Dictyostelium apm1, an AP-1 medium subunit. With the two fly donors this makes the seed set 100% AP-1/AP-2, which is the whole argument. Proposed replacements: cargo adaptor activity Supporting Evidence: PMID:42139345 By demonstrating that AP3:ARF1 can generate carriers without using a clathrin lattice, we explain the clathrin independence of AP3-mediated trafficking PMID:39705307 This closed form of the complex occludes the two known binding sites for transmembrane cargo: tyrosine-based YxxΦ motifs (where x is any amino acid and Φ is a bulky, hydrophobic amino acid) on the μ subunit (12, 13) PMID:39705307 There is also clear density for the LAMP1 cargo motif in the μ3-CTD tyrosine cargo-binding pocket, unambiguously showing that this complex represents the cargo-engaged state of AP-3 PMID:9151686 The AP-3 complex is not clathrin associated file:human/AP3M1/AP3M1-uniprot.txt not clathrin-associated. The complex is associated with the Golgi file:human/AP3M1/AP3M1-bioinformatics/RESULTS.md It is the *same* site as the AP-2 one: 9 of those 10 residues fall on alignment columns that are also AP2M1 cargo-pocket positions in the mu2/TGN38 crystal structure. |
| GO:0035651 AP-3 adaptor complex binding | IEA GO_REF:0000107 | MARK AS OVER ANNOTATED | Summary: 'AP-3 adaptor complex binding' transferred from mouse Ap3m1, whose own row is an MGI IDA citing PMID:19010779. That paper cross-linked cells in vivo, affinity-purified AP-3 complexes and identified the associated proteins by mass spectrometry. Finding mu3A in a purified AP-3 preparation demonstrates that mu3A is in AP-3 - which is complex membership, and is already asserted on this gene as GO:0030123 part_of, twice. Reason: Not wrong, but the wrong kind of statement. AP3M1 is an obligate subunit of AP-3, not an external ligand of it; typing that relationship as the molecular function 'binding to an AP-3 adaptor complex' recasts machinery as one of its own interactors. The informative content is fully carried by the two existing GO:0030123 part_of rows. GO:0035651 is useful for genuine AP-3 partners - BLOC-1 subunits, PI4KIIalpha, cargo - and AP3M2 carries the same IEA, so this is a pattern in the mu subfamily rather than a one-off. Marked over-annotated rather than removed because the underlying mouse IDA is a real observation and MGI curators read the full paper. Propagation Review Root cause: TERM SCOPING PROBLEM Sources checked: UniProtKB:Q9JKC8 Β· Ap3m1, mouse AP-3 complex subunit mu-1 SOURCE WEAK OR INFERRED Mouse Ap3m1's MGI IDA for this term comes from affinity purification of the AP-3 complex itself, so what was observed is that the subunit is in the complex. The orthology step is fine; it is the term's role that does not fit an obligate subunit. ensembl:ENSMUSP00000117346 Β· Ensembl Compara protein model for mouse Ap3m1 SOURCE WEAK OR INFERRED The Ensembl Compara protein model through which the orthology call was made; it carries no evidence of its own beyond the Q9JKC8 record. Supporting Evidence: PMID:19010779 AP-3 was co-isolated with BLOC-1, BLOC-2, and homotypic fusion and vacuole protein sorting complex subunits; clathrin; and phosphatidylinositol-4-kinase type II alpha (PI4KIIalpha) file:human/AP3M1/AP3M1-uniprot.txt AP3B1 or AP3B2), a medium adaptin (mu-type subunit AP3M1 or AP3M2) and |
| GO:0035654 clathrin-coated vesicle cargo loading, AP-3-mediated | NAS PMID:9545220 Association of the AP-3 adaptor complex with clathrin. | MODIFY | Summary: 'Clathrin-coated vesicle cargo loading, AP-3-mediated', asserted by ComplexPortal and cited to the 1998 Science report that mammalian AP-3 binds clathrin through the beta3 appendage. The AP-3-mediated cargo-loading half of the term is exactly right for this protein - it is what the mu subunit does. The clathrin-coated-vesicle half is the problem, and the term's own definition already hedges it ('in some organisms, links clathrin'). Reason: Generalise to GO:0035459 'vesicle cargo loading', the grandparent that makes the same assertion without committing to a clathrin coat. The cited paper is a real result and its own conclusion is hedged - AP-3 function 'may depend on clathrin' - and the interaction it reports is with beta3, not with mu3A. Against it: AP-3 was originally characterised as a non-clathrin coat, the 2024 review of the field calls the question open with only partial colocalisation in vivo, and the 2026 cryo-ET reconstitution shows AP3:ARF1 generating carriers with no clathrin lattice at all. Since the intermediate term GO:0035652 is also clathrin-specific, GO:0035459 is the nearest ancestor free of the commitment. The AP-3 specificity that is lost by generalising is worth recovering with a new clathrin-neutral child term, which is proposed separately. Proposed replacements: vesicle cargo loading Supporting Evidence: PMID:9545220 In vitro binding assays showed that mammalian AP-3 did associate with clathrin by interaction of the appendage domain of its beta3 subunit with the amino-terminal domain of the clathrin heavy chain PMID:9545220 Thus, AP-3 function in protein sorting may depend on clathrin PMID:42139345 By demonstrating that AP3:ARF1 can generate carriers without using a clathrin lattice, we explain the clathrin independence of AP3-mediated trafficking PMID:39705307 AP-3 has been reported to have both clathrin-dependent (21, 22) and -independent functions (23β25) and is found to only partially colocalize with clathrin in vivo (26, 27) |
| GO:0048490 anterograde synaptic vesicle transport | IEA GO_REF:0000107 | KEEP AS NON CORE | Summary: Transferred from mouse Ap3m1 (Q9JKC8), whose own row for this term is an IMP citing PMID:21998198. Reading that paper: all of its AP-3 genetics use the mocha allele Ap3d1mh/mh, a delta-subunit null, and the phenotype is failure of the AP-3 cargo PI4KIIalpha to reach neurites from the cell body. Searching the cached full text, the strings 'Ap3m1', 'mu3A' and 'AP3M' do not occur at all, and 'mu3' occurs twice, both inside the introductory sentence that names the heterotetramer's four subunits. So mu3A itself is never perturbed or assayed. Anterograde synaptic vesicle transport is the synaptic-vesicle-specific version of the same claim, from the same mouse IMP and the same mocha experiment. Reason: Keep, but not as a core AP3M1 function, and with the source-side limitation on record. Two separate dilutions sit between the experiment and this row. First, the evidence is complex-level: a delta null removes the whole coat, so the phenotype belongs to AP-3 rather than to its medium subunit. Second, delta is shared by the ubiquitous and the neuronal complexes (ComplexPortal CPX-5051/CPX-5052 versus CPX-5053/CPX-5055), so a mocha phenotype cannot even be assigned to the AP-3A complex that contains AP3M1. The genetics that do separate the two point the other way: beta3B-containing complexes are the ones preferentially targeted to neuronal processes and required for synaptic-vesicle targeting of ZnT3 and ClC-3, while beta3A-deficient brain has more of those cargoes, not less. A subunit of the ubiquitous complex is a poor candidate for the core agent of synaptic vesicle delivery. This row is the one where that matters most, because synaptic vesicle cargo targeting is precisely the job the beta3B/neuronal complex was shown to own. Propagation Review Root cause: SOURCE WEAK OR INFERRED Failure modes: SOURCE EVIDENCE WEAK Sources checked: UniProtKB:Q9JKC8 Β· Ap3m1, mouse AP-3 complex subunit mu-1 SOURCE WEAK OR INFERRED Mouse Ap3m1 is a true 1:1 orthologue and the cargo pocket is identical to human at all 14 tested columns, so orthology is not the problem. The problem is upstream of the transfer: the donor's own IMP rests on the Ap3d1 mocha allele, so it is evidence about the AP-3 complex and not about this subunit. ensembl:ENSMUSP00000117346 Β· Ensembl Compara protein model for mouse Ap3m1 SOURCE WEAK OR INFERRED The Ensembl Compara protein model through which the orthology call was made; it carries no evidence of its own beyond the Q9JKC8 record. Supporting Evidence: PMID:21998198 the AP-3βdeficient allele mocha (Ap3d1mh/mh PMID:15537701 Consistently, beta3B deficiency compromised synaptic zinc stores assessed by Timm's staining and the synaptic vesicle targeting of membrane proteins involved in zinc uptake (ZnT3 and ClC-3) PMID:15537701 These observations indicate that the functions of beta3A- and beta3B-containing complexes are distinct and divergent |
| GO:0048490 anterograde synaptic vesicle transport | ISS GO_REF:0000024 | KEEP AS NON CORE | Summary: Transferred from mouse Ap3m1 (Q9JKC8), whose own row for this term is an IMP citing PMID:21998198. Reading that paper: all of its AP-3 genetics use the mocha allele Ap3d1mh/mh, a delta-subunit null, and the phenotype is failure of the AP-3 cargo PI4KIIalpha to reach neurites from the cell body. Searching the cached full text, the strings 'Ap3m1', 'mu3A' and 'AP3M' do not occur at all, and 'mu3' occurs twice, both inside the introductory sentence that names the heterotetramer's four subunits. So mu3A itself is never perturbed or assayed. Anterograde synaptic vesicle transport is the synaptic-vesicle-specific version of the same claim, from the same mouse IMP and the same mocha experiment. Reason: Keep, but not as a core AP3M1 function, and with the source-side limitation on record. Two separate dilutions sit between the experiment and this row. First, the evidence is complex-level: a delta null removes the whole coat, so the phenotype belongs to AP-3 rather than to its medium subunit. Second, delta is shared by the ubiquitous and the neuronal complexes (ComplexPortal CPX-5051/CPX-5052 versus CPX-5053/CPX-5055), so a mocha phenotype cannot even be assigned to the AP-3A complex that contains AP3M1. The genetics that do separate the two point the other way: beta3B-containing complexes are the ones preferentially targeted to neuronal processes and required for synaptic-vesicle targeting of ZnT3 and ClC-3, while beta3A-deficient brain has more of those cargoes, not less. A subunit of the ubiquitous complex is a poor candidate for the core agent of synaptic vesicle delivery. This row is the one where that matters most, because synaptic vesicle cargo targeting is precisely the job the beta3B/neuronal complex was shown to own. Propagation Review Root cause: SOURCE WEAK OR INFERRED Failure modes: SOURCE EVIDENCE WEAK Sources checked: UniProtKB:Q9JKC8 Β· Ap3m1, mouse AP-3 complex subunit mu-1 SOURCE WEAK OR INFERRED Mouse Ap3m1, the ISS donor. Orthology is unambiguous and the cargo-binding site is identical between the two species, but the donor annotation is an IMP obtained with the Ap3d1 mocha allele rather than by perturbing Ap3m1. Supporting Evidence: PMID:21998198 the AP-3βdeficient allele mocha (Ap3d1mh/mh PMID:15537701 Consistently, beta3B deficiency compromised synaptic zinc stores assessed by Timm's staining and the synaptic vesicle targeting of membrane proteins involved in zinc uptake (ZnT3 and ClC-3) PMID:15537701 These observations indicate that the functions of beta3A- and beta3B-containing complexes are distinct and divergent |
| GO:0060155 platelet dense granule organization | NAS PMID:23247405 Cell type-specific Rab32 and Rab38 cooperate with the ubiqui... | KEEP AS NON CORE | Summary: Platelet dense granule organization, asserted by ComplexPortal for AP-3. Dense granules are lysosome-related organelles, and their defective formation is what makes AP-3 loss a bleeding disorder: Hermansky-Pudlak syndrome type 2 is caused by AP3B1 mutations. The cited review covers the shared LRO machinery rather than platelets specifically. Reason: Real but cell-type-specific and complex-level. AP3M1 is the medium subunit of the ubiquitous complex that carries out this sorting wherever the specialised organelle is made; the annotation belongs to the complex's activity in megakaryocytes, not to a platelet-restricted property of this subunit. No AP3M1-specific platelet experiment exists. Kept, since the underlying biology is genuine, but as context rather than core. Supporting Evidence: PMID:23247405 The biogenesis of LROs is known to utilize much of the common protein machinery used in the transport of integral membrane proteins to lysosomes PMID:10024875 Our results suggest that AP-3 functions in protein sorting to lysosomes PMID:31898847 These features result from defects in lysosome-related organelles (LROs), such as melanosomes in melanocytes and delta granules in platelets PMID:42139345 is recruited onto early/tubular endosomal compartments from where it sorts transmembrane protein cargo to endolysosomes, lysosomes, and a diverse group of cell typeβspecific compartments termed lysosome-related organelles (LROs) that include melanosomes |
| GO:0098837 postsynaptic recycling endosome | IEA GO_REF:0000107 | KEEP AS NON CORE | Summary: Postsynaptic recycling endosome, from three mouse SynGO IDA rows on the same paper. The underlying observation is immunohistochemical: mu3 was localised to several endosomal compartments in pyramidal neurons, including transferrin-receptor-positive recycling endosomes, and activity blockade shifted it further into that compartment. Reason: A real but conditional and cell-type-restricted location. The paper's own framing is that the recycling-endosome pool is a redirection away from the protein's normal lysosomal route, induced by activity blockade - so this is not where AP3M1 sits by default, even in neurons. Kept as non-core; the constitutive compartments for this gene are the cytoplasmic vesicle membrane and the early endosome. Propagation Review Root cause: NO FAILURE NON CORE Sources checked: UniProtKB:Q9JKC8 Β· Ap3m1, mouse AP-3 complex subunit mu-1 SUPPORTS TRANSFER Mouse Ap3m1, with SynGO IDA rows from a study that stained mu3 directly in mouse cortical pyramidal neurons; orthology and subject matter both line up. ensembl:ENSMUSP00000117346 Β· Ensembl Compara protein model for mouse Ap3m1 SUPPORTS TRANSFER The Ensembl Compara protein model through which the orthology call was made; it carries no evidence of its own beyond the Q9JKC8 record. Supporting Evidence: PMID:27568566 As expected, immunohistochemistry localized ΞΌ3 to a number of endosomal compartments within pyramidal neurons PMID:27568566 AP-3 is classically known for sorting membrane proteins into the lysosomal pathway for degradation (Bonafacino and Traub, 2003), so it was surprising to find that TTX treatment re-routed ΞΌ3A from lysosomes to recycling endosomes (RE) |
| GO:0098884 postsynaptic neurotransmitter receptor internalization | IBA GO_REF:0000033 | KEEP AS NON CORE | Summary: PTN002575745 is a vertebrate node (taxon:117571) whose single seed is MGI:MGI:1929212 = mouse Ap3m1, the 1:1 orthologue of the target. The seed's own evidence is three SynGO IDA rows from PMID:27568566, in which mu3A is upregulated in mouse visual cortex pyramidal neurons during synaptic scaling and redistributes with AMPAR to recycling endosomes. Note that this is the AP-3 family's separate, AP-2-flavoured GO:0098884 node - PTN000242374, seeded by mouse and rat Ap2m1 - and that AP3M1 does not descend from it; the row reaching AP3M1 is the mu3A-specific one. Reason: Keep. The target is inside the clade and the seed is its own mouse orthologue, so the transfer is sound. Non-core for two reasons. First, this is a neuron-restricted, activity-dependent behaviour of a ubiquitously expressed coat subunit, not what AP3M1 does in most cells. Second, the source paper's own conclusion pulls away from the complex: the sufficiency experiments are overexpression, including of a truncated mu3A that cannot join AP-3, and the authors read the effect as AP-3-independent. I am not second-guessing the SynGO IDA - the curators read the full text and the paper is squarely about mu3A - only declining to treat it as a core function. Propagation Review Root cause: NO FAILURE NON CORE Sources checked: MGI:MGI:1929212 Β· Ap3m1 / Q9JKC8, mouse AP-3 mu-1 (1:1 orthologue of the target) SUPPORTS TRANSFER Mouse Ap3m1 (Q9JKC8), the target's 1:1 orthologue, carrying three SynGO IDA rows for this term from PMID:27568566. A single well-characterised MOD donor is not weak support; it is the reason the node is placed at vertebrates rather than deeper. PANTHER:PTN002575745 Β· PTHR10529 vertebrate AP3M1 node (taxon 117571, seeded by mouse Ap3m1) SUPPORTS TRANSFER The vertebrate AP3M1 node (taxon:117571). Human AP3M1 is inside the inheriting clade, and the cargo-binding site is identical between human and mouse at all 14 tested pocket columns (file:human/AP3M1/AP3M1-bioinformatics/RESULTS.md), so there is no target-specific evidence of divergence to argue against the node. Supporting Evidence: PMID:27568566 Synaptic scaling increased ΞΌ3A (but not other AP-3 subunits) in pyramidal neurons and redistributed dendritic ΞΌ3A and AMPAR to recycling endosomes (REs) PMID:27568566 Knockdown of ΞΌ3A prevented PMID:27568566 AP-3 is classically known for sorting membrane proteins into the lysosomal pathway for degradation (Bonafacino and Traub, 2003), so it was surprising to find that TTX treatment re-routed ΞΌ3A from lysosomes to recycling endosomes (RE) |
| GO:0098884 postsynaptic neurotransmitter receptor internalization | IEA GO_REF:0000107 | KEEP AS NON CORE | Summary: The Ensembl Compara twin of the GO:0098884 IBA row, transferring the same three mouse SynGO IDA annotations from PMID:27568566 by 1:1 orthology instead of through the PANTHER node. Reason: Same verdict as the IBA row for this term: sound transfer, neuron-restricted and activity-dependent biology, and a source paper whose own reading is that the effect is AP-3-independent. Duplicate terms under different evidence codes are normal and neither row needs to be removed in favour of the other. Propagation Review Root cause: NO FAILURE NON CORE Sources checked: UniProtKB:Q9JKC8 Β· Ap3m1, mouse AP-3 complex subunit mu-1 SUPPORTS TRANSFER Mouse Ap3m1 carries three SynGO IDA rows for this term. This is the one place where the mouse evidence is genuinely about mu3A rather than about the complex, since PMID:27568566 knocks down and overexpresses mu3A itself. ensembl:ENSMUSP00000117346 Β· Ensembl Compara protein model for mouse Ap3m1 SUPPORTS TRANSFER The Ensembl Compara protein model through which the orthology call was made; it carries no evidence of its own beyond the Q9JKC8 record. Supporting Evidence: PMID:27568566 Synaptic scaling increased ΞΌ3A (but not other AP-3 subunits) in pyramidal neurons and redistributed dendritic ΞΌ3A and AMPAR to recycling endosomes (REs) PMID:27568566 AP-3 is classically known for sorting membrane proteins into the lysosomal pathway for degradation (Bonafacino and Traub, 2003), so it was surprising to find that TTX treatment re-routed ΞΌ3A from lysosomes to recycling endosomes (RE) |
| GO:1903232 melanosome assembly | NAS PMID:23247405 Cell type-specific Rab32 and Rab38 cooperate with the ubiqui... | KEEP AS NON CORE | Summary: Melanosome assembly, from the same ComplexPortal statement. The cited review is explicit that the ubiquitous machinery - BLOC-2, AP-1 and AP-3 - is redirected by the melanocyte-specific Rab32 and Rab38 to carry tyrosinase and TYRP1 to the maturing melanosome. Reason: Same reasoning as the platelet dense granule row. The melanosome is the best-studied lysosome-related organelle and AP-3 is genuinely required for cargo delivery to it, but this is the ubiquitous complex doing its ordinary job in a specialised cell, under the direction of cell-type-specific Rabs. Non-core for AP3M1; the core statement is delivery of transmembrane cargo to lysosomes and lysosome-related organelles generally. Supporting Evidence: PMID:23247405 Two small GTPases, Rab32 and Rab38, are key proteins in the biogenesis of melanosomes and were recently shown to redirect the ubiquitous machinery-BLOC-2, AP-1 and AP-3-to traffic specialized cargoes to melanosomes in melanocytes PMID:22511774 Rab38- and Rab32-deficient MNT-1 cells displayed abnormal trafficking and steady state levels of known cargoes of the BLOC-2, AP-3, and AP-1 pathways, the melanin-synthesizing enzymes tyrosinase and tyrosinase-related protein-1 PMID:42139345 is recruited onto early/tubular endosomal compartments from where it sorts transmembrane protein cargo to endolysosomes, lysosomes, and a diverse group of cell typeβspecific compartments termed lysosome-related organelles (LROs) that include melanosomes |
| GO:1904115 axon cytoplasm | IEA GO_REF:0000108 | KEEP AS NON CORE | Summary: An inter-ontology logical inference: anything involved in GO:0008089 'anterograde axonal transport' is placed in GO:1904115 'axon cytoplasm'. The row therefore carries no evidence of its own and is exactly as strong as the GO:0008089 row it is computed from, which in turn is an Ensembl/ISS transfer from a mouse IMP obtained with the mocha Ap3d1 allele. Reason: Kept for consistency with its parent row rather than on its own merits. Since GO:0008089 is retained as non-core, removing its logical consequence would leave the two inconsistent. But the chain is worth flagging: a subunit-level location in axonal cytoplasm has never been shown for AP3M1, and the mouse experiment behind the process term concerns cargo export from cell bodies into neurites in a delta-null background. Propagation Review Root cause: NO FAILURE NON CORE Sources checked: GO:0008089 Β· the anterograde axonal transport row on this gene, from which this location was logically inferred SOURCE WEAK OR INFERRED Not a gene product but the GO term this location was inferred from. Its own row on AP3M1 is an Ensembl Compara transfer of a mouse IMP whose AP-3 genetics are the Ap3d1 mocha allele, so the inference inherits that weakness: it is a location computed from a process that was measured on a different subunit. Supporting Evidence: PMID:21998198 PI4KIIΞ± was targeted to processes in wild-type primary cultured cortical neurons and PC12 cells but failed to reach neurites in cells lacking either AP-3 or BLOC-1 PMID:21998198 the AP-3βdeficient allele mocha (Ap3d1mh/mh |
| GO:0005048 signal sequence binding | IDA PMID:39705307 A structure-based mechanism for initiation of AP-3 coated ve... | NEW | Summary: The molecular function of a medium subunit is to read the tyrosine-based (YxxPhi) sorting signal in a cargo protein's cytoplasmic tail, and for mu3A this is directly demonstrated. In the human AP-3 cryo-EM series a LAMP1 cytoplasmic-tail peptide carrying the YQTI motif is resolved in the mu3 C-terminal domain's tyrosine cargo-binding pocket; the same structures show that in the closed AP-1/AP-2 state that pocket is the occluded cargo site and that in AP-3 it is constitutively exposed. Four independent earlier lines say the same thing for mu3A specifically: the ubiquitous p47A chain binds the TGN38 YQRL signal; a combinatorial XXXYXXO two-hybrid library gives mu3A its own residue preferences distinct from mu1 and mu2; the lgp120 (LAMP1) tail binds mu3A almost exclusively among the medium chains, and the very motif involved is YQTI; and the MPR46 cytoplasmic tail binds both mu2 and mu3A. A fifth comes from virology, where the human respiratory syncytial virus matrix protein was shown to engage AP-3 specifically through mu3A, dependent on the tyrosine of a YXXL motif. Reason: GOA gives AP3M1 no molecular function of its own: the only MF rows are a clathrin-specific IBA seeded entirely from AP-1/AP-2 donors, two small-GTPase IPIs from a complex-level co-immunoprecipitation, and twenty-three bare protein-binding rows. The one activity that is directly measured for this protein is therefore missing. GO:0005048 is the term GO already uses for this activity on the sister subunit - human AP2M1 (Q96CW1) carries it with IDA evidence from PMID:8918456, the mu-chain tyrosine-signal work - so this row brings AP3M1 into line with existing practice rather than inventing a precedent. The structural site was checked rather than assumed: a contact analysis of PDB 9C5B and 1BXX (file:human/AP3M1/AP3M1-bioinformatics/RESULTS.md) finds ten AP3M1 residues within 4.5 A of the LAMP1 peptide, nine of them on alignment columns that are also AP2M1 cargo-pocket positions, with AP3M1 identical to AP2M1 at only 5 of 14 of those columns - the site is present and used, and its chemistry differs, which is exactly what the combinatorial specificity data predict. A dedicated child term for tyrosine-based sorting-signal binding would be better still and is proposed separately. Supporting Evidence: PMID:39705307 There is also clear density for the LAMP1 cargo motif in the μ3-CTD tyrosine cargo-binding pocket, unambiguously showing that this complex represents the cargo-engaged state of AP-3 PMID:39705307 This closed form of the complex occludes the two known binding sites for transmembrane cargo: tyrosine-based YxxΦ motifs (where x is any amino acid and Φ is a bulky, hydrophobic amino acid) on the μ subunit (12, 13) PMID:8918456 We recently determined that fusion proteins containing tyrosine-based endocytic signals bind to the mu 2 subunit of AP-2 PMID:8918456 Here we analyze the selectivity of peptide recognition by mu 2 and by AP-2 using combinatorial selection methods and surface plasmon resonance PMID:9118953 Like other members of the medium-chain family, the p47A chain is capable of interacting with the tyrosine-based sorting signal YQRL from TGN38 PMID:9748267 We have analyzed the selectivity of interaction between YXXO signals and the mu1, mu2, and mu3 (A or B) subunits of the AP-1, AP-2, and AP-3 complexes, respectively, by screening a combinatorial XXXYXXO library using the yeast two-hybrid system PMID:9748267 each medium subunit favored specific sets of residues at the X and O positions; these preferences were consistent with the proposed roles of the different adaptor complexes in rapid endocytosis and lysosomal targeting PMID:9794796 we show that the cytosolic domain of lgp120 interacts almost exclusively with mu3A PMID:11071885 Both mu2 and mu3A bind specifically to the MPR46-CT PMID:29028839 Here, we show that HRSV Matrix (M) protein interacts with the cellular adaptor protein complex 3 specifically via its medium subunit (AP-3Mu3A) PMID:29028839 Analysis of point-mutated HRSV M derivatives indicated that AP-3Mu3A- mediated trafficking is contingent on the presence of the tyrosine residue within the YXXL sorting sequence at amino acids 197-200 of the M protein file:human/AP3M1/AP3M1-bioinformatics/RESULTS.md E178 (4.44 A), Y180 (2.83 A), F181 (3.7 A), V389 (3.56 A), L392 (3.89 A), F402 (3.71 A), K403 (3.1 A), G404 (3.29 A), V405 (2.8 A), K406 (3.37 A) file:human/AP3M1/AP3M1-bioinformatics/RESULTS.md The site is structurally conserved but chemically divergent: AP3M1 is identical to AP2M1 at only 5 of 14 pocket positions while aligning to a residue at 14 of 14, which is the structural basis for the different YxxPhi preferences of the mu subunits. |
| GO:0008333 endosome to lysosome transport | TAS PMID:42139345 Architecture of clathrin-independent AP3:ARF1-coated carrier... | NEW | Summary: Where AP-3 acts has shifted since the terms currently on this gene were assigned. GOA describes the route as Golgi to vacuole (IBA) and as protein targeting to lysosome (TAS), but both recent structural studies state the step as endosome to lysosome: AP-3 is recruited onto early/tubular endosomal compartments and sorts transmembrane cargo from there to endolysosomes, lysosomes and lysosome-related organelles. The melanocyte literature agrees independently, packaging tyrosinase and TYRP1 at early/recycling endosome-associated tubules under AP-1 and AP-3 control, and GOA already carries early endosome as a location for this gene. Reason: Adds the compartment step that the existing process rows leave unstated, using the evidence code that matches the evidence. TAS rather than IMP or IDA is deliberate: the cell experiments that anchor these statements perturb the delta subunit and the AP3:AP3 lattice, not AP3M1, so this is a traceable author statement about the complex of which AP3M1 is a constitutive subunit, on the same footing as the existing GO:0006622 TAS row. The alternative considered was IC with GO:0030123 in the with/from field, which would make the reasoning - subunit inherits the complex's process - explicit rather than implicit. TAS was chosen for consistency with the row already on this gene that makes the same kind of inference from the same kind of source, so that a downstream submitter sees one convention rather than two; an IC row with GO:0030123 as the with/from would be an acceptable substitute. It complements rather than replaces GO:0006896: that term names the origin of the biosynthetic route, this one names the step AP-3 itself performs. Supporting Evidence: PMID:42139345 The AP3 complex mediates cargo sorting and carrier assembly for the trafficking of transmembrane proteins from endosomes to lysosomes PMID:42139345 is recruited onto early/tubular endosomal compartments from where it sorts transmembrane protein cargo to endolysosomes, lysosomes, and a diverse group of cell typeβspecific compartments termed lysosome-related organelles (LROs) that include melanosomes PMID:39705307 Adaptor protein complex-3 (AP-3) mediates cargo sorting from endosomes to lysosomes and lysosome-related organelles PMID:23247405 Packaging of the tyrosinases into transport vesicles at early/recycling endosome-associated tubules is dependent on ubiquitous adaptor protein complex (AP)-1 and AP-3, and biogenesis of lysosome-related organelles complex (BLOC)-1 and BLOC-2 |
Loading supporting contentβ¦
Download this section (compressed HTML)Q: In human cells, does AP3M1 loss phenocopy AP3B1 loss for lysosomal membrane protein traffic, or does AP3M2 substitute? The answer determines whether AP3M1 should carry the lysosome-targeting process terms in its own right or only as a complex subunit.
Q: Is mu3A ever found in a beta3B-containing complex in neurons? If not, the anterograde axonal and synaptic vesicle transport rows now propagated onto AP3M1 from the mocha delta null belong to AP3M2, and AP3M1 should lose them.
Q: Which cargo classes does the mu3A pocket select that mu1 and mu2 do not? The combinatorial library gave mu3A distinct residue preferences at the x and Phi positions, but the physiological cargo repertoire that follows from those preferences has never been enumerated.
Q: Does the clathrin interaction of AP-3 have any function in cells, given that AP3:ARF1 builds carriers without a clathrin lattice? A negative answer would justify revising the definition of GO:0035654 rather than working around it.
Q: Should GO:0030123 be split into ubiquitous and neuronal AP-3 complex children, and if so, should the split be defined by the beta3 isoform, the mu3 isoform, or both?
Experiment: Make AP3M1, AP3M2 and AP3M1/AP3M2 double knockouts in a pigmented human cell line (MNT-1) and in HeLa, and measure surface versus lysosomal distribution of a matched cargo panel by flow cytometry and microscopy: LAMP1 and CD63, whose tails carry YxxPhi motifs, against tyrosinase and LIMP-II, whose AP-3 signals are acidic dileucine. Rescue the AP3M1 knockout with wild type mu3A and with pocket mutants at the measured LAMP1-contact positions (Y180, K403, V405) to test whether the cargo defect maps to the tyrosine site. The prediction that distinguishes this from a general complex-destabilisation phenotype is selectivity: pocket mutants should assemble into AP-3 normally, by blue-native gel or by co-purification with delta, while failing only the YxxPhi cargoes.
Hypothesis: AP3M1 is the YxxPhi-reading subunit of the ubiquitous AP-3 complex, so an AP3M1-null human cell should mislocalise tyrosine-motif lysosomal cargo while leaving dileucine-motif cargo comparatively intact.
Type: genetic perturbation with cargo-class-resolved readout
Experiment: In mouse cortical neurons, compare conditional Ap3m1 and Ap3m2 deletions against the mocha Ap3d1 null for the cargo readouts that defined the AP-3 neuronal phenotype - PI4KIIalpha, VAMP7 and ZnT3 content of synaptosomes, and PI4KIIalpha export from cell bodies into neurites. Immunoprecipitate beta3A and beta3B separately from the same brain lysates and blot for mu3A and mu3B to establish which beta each mu partners with in vivo. If the mocha phenotype is reproduced by Ap3m2 deletion and not by Ap3m1 deletion, the ISS and Ensembl rows for anterograde axonal and synaptic vesicle transport should move off AP3M1.
Hypothesis: The neuronal AP-3 phenotypes currently attributed to AP3M1 belong to mu3B-containing complexes.
Type: subunit-resolved mouse genetics with biochemical complex mapping
Experiment: Run a proteome-scale peptide-array or yeast display selection of human cytoplasmic-tail YxxPhi peptides against purified mu3A, mu3B, mu1A and mu2 C-terminal domains, calibrating against the known binders (LAMP1 YQTI, TGN38 YQRL, MPR46) and the known non-binder behaviour of mu4. Cross the resulting mu3A-selective set against surface proteomics of AP3M1-null cells from the first experiment; the informative output is the overlap, which would convert the in vitro specificity into a defensible in vivo cargo list and give the proposed tyrosine-based sorting signal binding term an annotatable set of substrates.
Hypothesis: mu3A's measured YxxPhi preferences predict a definable cargo repertoire distinct from mu1 and mu2.
Type: in vitro binding selection cross-validated by cell-surface proteomics
Experiment: Express clathrin-box mutants of beta3A in AP3B1-null cells and assay AP-3-dependent cargo traffic and carrier morphology alongside the AP3:AP3 and AP3:ARF1 lattice mutants already shown to disrupt lysosomal cargo trafficking. A clean separation - lattice mutants defective, clathrin-box mutants normal - would settle the term definition question and support replacing GO:0035654 with a clathrin-neutral AP-3-specific child of GO:0035459.
Hypothesis: AP-3's clathrin interaction is dispensable for cargo delivery in cells.
Type: structure-guided separation-of-function mutagenesis
What is not known β curated, literature-grounded statements of the open unknowns (the inverse of core functions).
Gap: No experiment has perturbed AP3M1 itself in a human cell. Every process and localisation statement on this gene traces either to the complex - beta3A mutations in Hermansky-Pudlak syndrome fibroblasts, the mouse mocha Ap3d1 null, melanocyte co-immunoprecipitations - or to overexpression and knockdown of mu3A in mouse neurons. What a human AP3M1 loss of function does to lysosomal membrane protein traffic, and whether AP3M2 can cover for it, is unknown.
OPEN BIOLOGY
What is known: Established: AP3M1 is a constitutive subunit of the ubiquitous AP-3A complex (ComplexPortal CPX-5051/CPX-5052, cryo-EM chain M of reconstituted human AP-3); its mu homology domain binds YxxPhi sorting signals including the LAMP1 YQTI motif, shown by two-hybrid and by structure; AP-3 loss sends LAMP1, LAMP2 and CD63 to the plasma membrane. Not established: any AP3M1-specific loss-of-function phenotype in any human cell, any human disease allele (a Europe PMC search over 143 hits for AP3M1 with mutation, variant or patient terms returned no Mendelian AP3M1 disorder, while HPS2 maps to AP3B1 and HPS10 to AP3D1), and the degree of redundancy with AP3M2. One route by which a human loss could arise has been described but never observed clinically: the mu3A gene lies head-to-head with adenosine kinase across a shared bidirectional CpG-island promoter, and large ADK-region deletions in cultured hamster cells removed the whole mu3A gene.
Significance: Because the evidence is complex-level, the review can only record most process rows as non-core. A subunit-resolved knockout would decide whether AP3M1 is the specificity-determining subunit for a particular cargo class or an interchangeable structural component.
Provenance (the field's own admissions):
Gap: Whether mu3A and mu3B partition with beta3A and beta3B respectively, or whether the subunits mix and match, has been an open question since AP-3 was characterised, and I found nothing settling it for human cells. ComplexPortal models strict pairing; the genetics that dissect AP-3 function in brain used the beta3 isoforms, not the mu ones.
OPEN BIOLOGYCURATION
What is known: Established: beta3A- and beta3B-containing complexes have distinct and in places opposite functions in mouse brain, with beta3B complexes preferentially targeted to neuronal processes; mu3A itself is upregulated in mouse cortical pyramidal neurons during synaptic scaling, so it is present and regulated in neurons. Not established: which beta partner mu3A associates with in neurons, and whether the neuronal AP-3 rows currently propagated onto AP3M1 belong to a mu3A-containing complex at all.
Significance: Five rows on this gene - two anterograde transport terms, the postsynaptic recycling endosome, and two postsynaptic receptor internalization rows - are neuronal, and how much of that belongs to AP3M1 rather than to AP3M2 turns on this question.
Provenance (the field's own admissions):
Gap: The ontology cannot express three distinctions this gene needs: the ubiquitous versus neuronal AP-3 complexes, tyrosine-motif sorting-signal binding as distinct from organelle-import signal receptors, and AP-3 cargo loading without a clathrin commitment. Each forces a curation compromise here - a generic complex term, a parent-level molecular function, and a generalisation from GO:0035654 to GO:0035459.
OPEN ONTOLOGY
What is known: Established: GO:0030123 has no children; GO:0005048's four children are all organelle-targeting-sequence receptors; GO:0035654's only route to a clathrin-neutral ancestor is GO:0035459, two levels up. Not established: whether GO would accept the complex-variant terms, since they turn on isoform composition rather than on a distinct activity.
Significance: Without the AP-3A/AP-3B split, phylogenetic and orthology propagation cannot be stopped at the right subunit, which is the mechanism behind several of the non-core calls in this review.
Provenance (the field's own admissions):
Gap: AP-3's residual clathrin association has never been given a function. AP-3 carries a clathrin-binding motif in the beta3 hinge and partially colocalises with clathrin, yet carrier formation demonstrably does not require a clathrin lattice, and clathrin knockdown does not fully disrupt AP-3-dependent cargo traffic.
NARROWING BIOLOGYONTOLOGY
What is known: Established: the beta3 appendage binds the clathrin heavy chain N-terminal domain in vitro; AP3:ARF1 generates coated tubular carriers without clathrin. Not established: whether the clathrin interaction does anything in cells, or is a vestige of the shared adaptor architecture.
Significance: Three GOA rows on this gene assert clathrin - one removed here, one modified, one generalised - so the unresolved biology is directly responsible for the annotation noise.
Provenance (the field's own admissions):
Loading supporting contentβ¦
Download this section (compressed HTML)Loading supporting contentβ¦
Download this section (compressed HTML)Loading supporting contentβ¦
Download this section (compressed HTML)Loading supporting contentβ¦
Download this section (compressed HTML)